{ "best_metric": null, "best_model_checkpoint": null, "epoch": 99.91646586345381, "eval_steps": 500, "global_step": 15500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00642570281124498, "grad_norm": 6.289779186248779, "learning_rate": 4.999677419354839e-05, "loss": 1.0311, "step": 1 }, { "epoch": 0.01285140562248996, "grad_norm": 10.10856819152832, "learning_rate": 4.999354838709678e-05, "loss": 0.5526, "step": 2 }, { "epoch": 0.01927710843373494, "grad_norm": 7.540353298187256, "learning_rate": 4.9990322580645166e-05, "loss": 0.9129, "step": 3 }, { "epoch": 0.02570281124497992, "grad_norm": 5.922585964202881, "learning_rate": 4.9987096774193545e-05, "loss": 0.4286, "step": 4 }, { "epoch": 0.0321285140562249, "grad_norm": 4.353242874145508, "learning_rate": 4.998387096774194e-05, "loss": 0.5559, "step": 5 }, { "epoch": 0.03855421686746988, "grad_norm": 4.786715984344482, "learning_rate": 4.9980645161290323e-05, "loss": 0.5169, "step": 6 }, { "epoch": 0.04497991967871486, "grad_norm": 7.102758884429932, "learning_rate": 4.997741935483871e-05, "loss": 0.7124, "step": 7 }, { "epoch": 0.05140562248995984, "grad_norm": 6.3182806968688965, "learning_rate": 4.99741935483871e-05, "loss": 0.4885, "step": 8 }, { "epoch": 0.05783132530120482, "grad_norm": 7.925058841705322, "learning_rate": 4.997096774193549e-05, "loss": 0.5508, "step": 9 }, { "epoch": 0.0642570281124498, "grad_norm": 6.677995204925537, "learning_rate": 4.996774193548387e-05, "loss": 0.5849, "step": 10 }, { "epoch": 0.07068273092369477, "grad_norm": 3.8706135749816895, "learning_rate": 4.9964516129032266e-05, "loss": 0.5568, "step": 11 }, { "epoch": 0.07710843373493977, "grad_norm": 6.342950344085693, "learning_rate": 4.9961290322580644e-05, "loss": 0.5713, "step": 12 }, { "epoch": 0.08353413654618475, "grad_norm": 5.465851783752441, "learning_rate": 4.995806451612903e-05, "loss": 0.651, "step": 13 }, { "epoch": 0.08995983935742972, "grad_norm": 8.341426849365234, "learning_rate": 4.995483870967742e-05, "loss": 0.5937, "step": 14 }, { "epoch": 0.0963855421686747, "grad_norm": 8.490134239196777, "learning_rate": 4.995161290322581e-05, "loss": 0.8078, "step": 15 }, { "epoch": 0.10281124497991968, "grad_norm": 6.18711519241333, "learning_rate": 4.9948387096774194e-05, "loss": 0.8606, "step": 16 }, { "epoch": 0.10923694779116466, "grad_norm": 4.206506252288818, "learning_rate": 4.9945161290322587e-05, "loss": 0.2513, "step": 17 }, { "epoch": 0.11566265060240964, "grad_norm": 6.628728866577148, "learning_rate": 4.994193548387097e-05, "loss": 0.4567, "step": 18 }, { "epoch": 0.12208835341365462, "grad_norm": 7.694202423095703, "learning_rate": 4.993870967741936e-05, "loss": 0.6095, "step": 19 }, { "epoch": 0.1285140562248996, "grad_norm": 6.6055402755737305, "learning_rate": 4.9935483870967744e-05, "loss": 0.4893, "step": 20 }, { "epoch": 0.13493975903614458, "grad_norm": 5.736470699310303, "learning_rate": 4.993225806451613e-05, "loss": 0.5958, "step": 21 }, { "epoch": 0.14136546184738955, "grad_norm": 4.132664680480957, "learning_rate": 4.9929032258064515e-05, "loss": 0.6889, "step": 22 }, { "epoch": 0.14779116465863454, "grad_norm": 4.619383811950684, "learning_rate": 4.992580645161291e-05, "loss": 0.3995, "step": 23 }, { "epoch": 0.15421686746987953, "grad_norm": 8.579829216003418, "learning_rate": 4.992258064516129e-05, "loss": 0.7418, "step": 24 }, { "epoch": 0.1606425702811245, "grad_norm": 2.837878942489624, "learning_rate": 4.991935483870968e-05, "loss": 0.2709, "step": 25 }, { "epoch": 0.1670682730923695, "grad_norm": 6.2830915451049805, "learning_rate": 4.991612903225807e-05, "loss": 0.4178, "step": 26 }, { "epoch": 0.17349397590361446, "grad_norm": 6.224307060241699, "learning_rate": 4.991290322580646e-05, "loss": 0.3698, "step": 27 }, { "epoch": 0.17991967871485945, "grad_norm": 8.073163986206055, "learning_rate": 4.9909677419354836e-05, "loss": 0.9302, "step": 28 }, { "epoch": 0.18634538152610441, "grad_norm": 7.330682754516602, "learning_rate": 4.990645161290323e-05, "loss": 0.7699, "step": 29 }, { "epoch": 0.1927710843373494, "grad_norm": 8.107137680053711, "learning_rate": 4.9903225806451614e-05, "loss": 0.5792, "step": 30 }, { "epoch": 0.19919678714859437, "grad_norm": 7.307425498962402, "learning_rate": 4.99e-05, "loss": 0.9953, "step": 31 }, { "epoch": 0.20562248995983937, "grad_norm": 11.467477798461914, "learning_rate": 4.989677419354839e-05, "loss": 0.6243, "step": 32 }, { "epoch": 0.21204819277108433, "grad_norm": 6.759311676025391, "learning_rate": 4.989354838709678e-05, "loss": 0.7024, "step": 33 }, { "epoch": 0.21847389558232932, "grad_norm": 6.139105796813965, "learning_rate": 4.9890322580645164e-05, "loss": 0.6084, "step": 34 }, { "epoch": 0.2248995983935743, "grad_norm": 6.263670444488525, "learning_rate": 4.9887096774193556e-05, "loss": 0.5911, "step": 35 }, { "epoch": 0.23132530120481928, "grad_norm": 6.1808671951293945, "learning_rate": 4.9883870967741935e-05, "loss": 0.5718, "step": 36 }, { "epoch": 0.23775100401606425, "grad_norm": 4.708409786224365, "learning_rate": 4.988064516129032e-05, "loss": 0.2719, "step": 37 }, { "epoch": 0.24417670682730924, "grad_norm": 6.317741394042969, "learning_rate": 4.9877419354838713e-05, "loss": 0.6962, "step": 38 }, { "epoch": 0.25060240963855424, "grad_norm": 5.216880798339844, "learning_rate": 4.98741935483871e-05, "loss": 0.3076, "step": 39 }, { "epoch": 0.2570281124497992, "grad_norm": 5.757438659667969, "learning_rate": 4.9870967741935485e-05, "loss": 0.364, "step": 40 }, { "epoch": 0.26345381526104417, "grad_norm": 17.03708267211914, "learning_rate": 4.986774193548388e-05, "loss": 0.5741, "step": 41 }, { "epoch": 0.26987951807228916, "grad_norm": 6.495184421539307, "learning_rate": 4.986451612903226e-05, "loss": 0.4203, "step": 42 }, { "epoch": 0.27630522088353415, "grad_norm": 8.956806182861328, "learning_rate": 4.986129032258064e-05, "loss": 0.6933, "step": 43 }, { "epoch": 0.2827309236947791, "grad_norm": 10.098581314086914, "learning_rate": 4.9858064516129034e-05, "loss": 0.3642, "step": 44 }, { "epoch": 0.2891566265060241, "grad_norm": 5.262434005737305, "learning_rate": 4.985483870967742e-05, "loss": 0.4563, "step": 45 }, { "epoch": 0.2955823293172691, "grad_norm": 9.325684547424316, "learning_rate": 4.9851612903225806e-05, "loss": 0.5514, "step": 46 }, { "epoch": 0.30200803212851407, "grad_norm": 4.861095905303955, "learning_rate": 4.98483870967742e-05, "loss": 0.1199, "step": 47 }, { "epoch": 0.30843373493975906, "grad_norm": 8.600152969360352, "learning_rate": 4.9845161290322584e-05, "loss": 0.4661, "step": 48 }, { "epoch": 0.314859437751004, "grad_norm": 8.110428810119629, "learning_rate": 4.984193548387097e-05, "loss": 0.5575, "step": 49 }, { "epoch": 0.321285140562249, "grad_norm": 7.853572368621826, "learning_rate": 4.983870967741936e-05, "loss": 0.3802, "step": 50 }, { "epoch": 0.327710843373494, "grad_norm": 4.605618953704834, "learning_rate": 4.983548387096774e-05, "loss": 0.4178, "step": 51 }, { "epoch": 0.334136546184739, "grad_norm": 5.590431213378906, "learning_rate": 4.983225806451613e-05, "loss": 0.5285, "step": 52 }, { "epoch": 0.3405622489959839, "grad_norm": 3.3249497413635254, "learning_rate": 4.982903225806452e-05, "loss": 0.2209, "step": 53 }, { "epoch": 0.3469879518072289, "grad_norm": 8.909160614013672, "learning_rate": 4.9825806451612905e-05, "loss": 0.4056, "step": 54 }, { "epoch": 0.3534136546184739, "grad_norm": 6.78776216506958, "learning_rate": 4.982258064516129e-05, "loss": 0.6553, "step": 55 }, { "epoch": 0.3598393574297189, "grad_norm": 5.1622395515441895, "learning_rate": 4.981935483870968e-05, "loss": 0.5461, "step": 56 }, { "epoch": 0.36626506024096384, "grad_norm": 8.728926658630371, "learning_rate": 4.981612903225807e-05, "loss": 0.6644, "step": 57 }, { "epoch": 0.37269076305220883, "grad_norm": 9.814579010009766, "learning_rate": 4.9812903225806455e-05, "loss": 0.6733, "step": 58 }, { "epoch": 0.3791164658634538, "grad_norm": 9.20268726348877, "learning_rate": 4.980967741935484e-05, "loss": 0.6709, "step": 59 }, { "epoch": 0.3855421686746988, "grad_norm": 7.0428547859191895, "learning_rate": 4.9806451612903226e-05, "loss": 0.4534, "step": 60 }, { "epoch": 0.39196787148594375, "grad_norm": 10.683008193969727, "learning_rate": 4.980322580645161e-05, "loss": 0.5942, "step": 61 }, { "epoch": 0.39839357429718875, "grad_norm": 6.303813934326172, "learning_rate": 4.9800000000000004e-05, "loss": 0.5876, "step": 62 }, { "epoch": 0.40481927710843374, "grad_norm": 6.049997329711914, "learning_rate": 4.979677419354839e-05, "loss": 0.4098, "step": 63 }, { "epoch": 0.41124497991967873, "grad_norm": 7.559628963470459, "learning_rate": 4.9793548387096776e-05, "loss": 0.2796, "step": 64 }, { "epoch": 0.41767068273092367, "grad_norm": 10.972475051879883, "learning_rate": 4.979032258064517e-05, "loss": 0.657, "step": 65 }, { "epoch": 0.42409638554216866, "grad_norm": 5.260735988616943, "learning_rate": 4.9787096774193554e-05, "loss": 0.5913, "step": 66 }, { "epoch": 0.43052208835341366, "grad_norm": 7.60853385925293, "learning_rate": 4.978387096774193e-05, "loss": 0.4829, "step": 67 }, { "epoch": 0.43694779116465865, "grad_norm": 3.407289981842041, "learning_rate": 4.9780645161290325e-05, "loss": 0.348, "step": 68 }, { "epoch": 0.4433734939759036, "grad_norm": 3.9688470363616943, "learning_rate": 4.977741935483871e-05, "loss": 0.2899, "step": 69 }, { "epoch": 0.4497991967871486, "grad_norm": 8.830917358398438, "learning_rate": 4.97741935483871e-05, "loss": 0.4925, "step": 70 }, { "epoch": 0.4562248995983936, "grad_norm": 4.9900617599487305, "learning_rate": 4.977096774193549e-05, "loss": 0.3153, "step": 71 }, { "epoch": 0.46265060240963857, "grad_norm": 6.262260437011719, "learning_rate": 4.9767741935483875e-05, "loss": 0.5321, "step": 72 }, { "epoch": 0.46907630522088356, "grad_norm": 5.420540809631348, "learning_rate": 4.976451612903226e-05, "loss": 0.3285, "step": 73 }, { "epoch": 0.4755020080321285, "grad_norm": 7.803525924682617, "learning_rate": 4.9761290322580646e-05, "loss": 0.613, "step": 74 }, { "epoch": 0.4819277108433735, "grad_norm": 9.742189407348633, "learning_rate": 4.975806451612903e-05, "loss": 0.6011, "step": 75 }, { "epoch": 0.4883534136546185, "grad_norm": 6.771279335021973, "learning_rate": 4.975483870967742e-05, "loss": 0.5958, "step": 76 }, { "epoch": 0.4947791164658635, "grad_norm": 7.624481678009033, "learning_rate": 4.975161290322581e-05, "loss": 0.7747, "step": 77 }, { "epoch": 0.5012048192771085, "grad_norm": 3.8246915340423584, "learning_rate": 4.9748387096774196e-05, "loss": 0.4128, "step": 78 }, { "epoch": 0.5076305220883535, "grad_norm": 5.4229350090026855, "learning_rate": 4.974516129032258e-05, "loss": 0.3719, "step": 79 }, { "epoch": 0.5140562248995983, "grad_norm": 7.140689373016357, "learning_rate": 4.9741935483870974e-05, "loss": 0.6277, "step": 80 }, { "epoch": 0.5204819277108433, "grad_norm": 6.446412086486816, "learning_rate": 4.973870967741936e-05, "loss": 0.5256, "step": 81 }, { "epoch": 0.5269076305220883, "grad_norm": 4.682521343231201, "learning_rate": 4.973548387096774e-05, "loss": 0.3443, "step": 82 }, { "epoch": 0.5333333333333333, "grad_norm": 11.275835990905762, "learning_rate": 4.973225806451613e-05, "loss": 0.6797, "step": 83 }, { "epoch": 0.5397590361445783, "grad_norm": 4.924052715301514, "learning_rate": 4.972903225806452e-05, "loss": 0.3249, "step": 84 }, { "epoch": 0.5461847389558233, "grad_norm": 5.751348495483398, "learning_rate": 4.97258064516129e-05, "loss": 0.4494, "step": 85 }, { "epoch": 0.5526104417670683, "grad_norm": 5.235645771026611, "learning_rate": 4.9722580645161295e-05, "loss": 0.3698, "step": 86 }, { "epoch": 0.5590361445783133, "grad_norm": 6.026668548583984, "learning_rate": 4.971935483870968e-05, "loss": 0.3609, "step": 87 }, { "epoch": 0.5654618473895582, "grad_norm": 4.437653064727783, "learning_rate": 4.9716129032258067e-05, "loss": 0.312, "step": 88 }, { "epoch": 0.5718875502008032, "grad_norm": 5.244723796844482, "learning_rate": 4.971290322580646e-05, "loss": 0.4375, "step": 89 }, { "epoch": 0.5783132530120482, "grad_norm": 2.9434661865234375, "learning_rate": 4.970967741935484e-05, "loss": 0.3319, "step": 90 }, { "epoch": 0.5847389558232932, "grad_norm": 5.709009647369385, "learning_rate": 4.9706451612903224e-05, "loss": 0.4071, "step": 91 }, { "epoch": 0.5911646586345382, "grad_norm": 5.898602485656738, "learning_rate": 4.9703225806451616e-05, "loss": 0.4753, "step": 92 }, { "epoch": 0.5975903614457831, "grad_norm": 6.049220561981201, "learning_rate": 4.97e-05, "loss": 0.5586, "step": 93 }, { "epoch": 0.6040160642570281, "grad_norm": 8.429190635681152, "learning_rate": 4.969677419354839e-05, "loss": 0.8489, "step": 94 }, { "epoch": 0.6104417670682731, "grad_norm": 4.625291347503662, "learning_rate": 4.969354838709678e-05, "loss": 0.28, "step": 95 }, { "epoch": 0.6168674698795181, "grad_norm": 4.896864891052246, "learning_rate": 4.9690322580645166e-05, "loss": 0.5365, "step": 96 }, { "epoch": 0.623293172690763, "grad_norm": 6.790309429168701, "learning_rate": 4.968709677419355e-05, "loss": 0.4837, "step": 97 }, { "epoch": 0.629718875502008, "grad_norm": 11.160213470458984, "learning_rate": 4.968387096774194e-05, "loss": 0.9259, "step": 98 }, { "epoch": 0.636144578313253, "grad_norm": 5.817671298980713, "learning_rate": 4.968064516129032e-05, "loss": 0.4321, "step": 99 }, { "epoch": 0.642570281124498, "grad_norm": 7.265375137329102, "learning_rate": 4.967741935483871e-05, "loss": 0.4667, "step": 100 }, { "epoch": 0.648995983935743, "grad_norm": 7.887307167053223, "learning_rate": 4.96741935483871e-05, "loss": 0.8003, "step": 101 }, { "epoch": 0.655421686746988, "grad_norm": 7.852207183837891, "learning_rate": 4.967096774193549e-05, "loss": 0.5069, "step": 102 }, { "epoch": 0.661847389558233, "grad_norm": 9.376118659973145, "learning_rate": 4.966774193548387e-05, "loss": 0.4703, "step": 103 }, { "epoch": 0.668273092369478, "grad_norm": 12.027544975280762, "learning_rate": 4.9664516129032265e-05, "loss": 0.8166, "step": 104 }, { "epoch": 0.6746987951807228, "grad_norm": 9.515453338623047, "learning_rate": 4.966129032258065e-05, "loss": 0.6511, "step": 105 }, { "epoch": 0.6811244979919678, "grad_norm": 7.603589057922363, "learning_rate": 4.965806451612903e-05, "loss": 0.3767, "step": 106 }, { "epoch": 0.6875502008032128, "grad_norm": 17.284399032592773, "learning_rate": 4.965483870967742e-05, "loss": 0.5665, "step": 107 }, { "epoch": 0.6939759036144578, "grad_norm": 6.044131755828857, "learning_rate": 4.965161290322581e-05, "loss": 0.5214, "step": 108 }, { "epoch": 0.7004016064257028, "grad_norm": 7.697197914123535, "learning_rate": 4.9648387096774194e-05, "loss": 0.5046, "step": 109 }, { "epoch": 0.7068273092369478, "grad_norm": 4.851754665374756, "learning_rate": 4.9645161290322586e-05, "loss": 0.3687, "step": 110 }, { "epoch": 0.7132530120481928, "grad_norm": 12.41080093383789, "learning_rate": 4.964193548387097e-05, "loss": 0.5301, "step": 111 }, { "epoch": 0.7196787148594378, "grad_norm": 4.711777210235596, "learning_rate": 4.963870967741936e-05, "loss": 0.4257, "step": 112 }, { "epoch": 0.7261044176706827, "grad_norm": 6.908281326293945, "learning_rate": 4.963548387096774e-05, "loss": 0.4149, "step": 113 }, { "epoch": 0.7325301204819277, "grad_norm": 7.849958419799805, "learning_rate": 4.963225806451613e-05, "loss": 0.7514, "step": 114 }, { "epoch": 0.7389558232931727, "grad_norm": 9.594267845153809, "learning_rate": 4.9629032258064515e-05, "loss": 0.5442, "step": 115 }, { "epoch": 0.7453815261044177, "grad_norm": 9.368268013000488, "learning_rate": 4.962580645161291e-05, "loss": 0.4031, "step": 116 }, { "epoch": 0.7518072289156627, "grad_norm": 7.328243255615234, "learning_rate": 4.962258064516129e-05, "loss": 0.4136, "step": 117 }, { "epoch": 0.7582329317269076, "grad_norm": 5.8368940353393555, "learning_rate": 4.961935483870968e-05, "loss": 0.375, "step": 118 }, { "epoch": 0.7646586345381526, "grad_norm": 5.233092784881592, "learning_rate": 4.961612903225807e-05, "loss": 0.2446, "step": 119 }, { "epoch": 0.7710843373493976, "grad_norm": 5.5411176681518555, "learning_rate": 4.961290322580646e-05, "loss": 0.4667, "step": 120 }, { "epoch": 0.7775100401606426, "grad_norm": 5.817262172698975, "learning_rate": 4.9609677419354836e-05, "loss": 0.3257, "step": 121 }, { "epoch": 0.7839357429718875, "grad_norm": 9.074666023254395, "learning_rate": 4.960645161290323e-05, "loss": 0.5812, "step": 122 }, { "epoch": 0.7903614457831325, "grad_norm": 7.789498805999756, "learning_rate": 4.9603225806451614e-05, "loss": 0.3473, "step": 123 }, { "epoch": 0.7967871485943775, "grad_norm": 4.843997955322266, "learning_rate": 4.96e-05, "loss": 0.3968, "step": 124 }, { "epoch": 0.8032128514056225, "grad_norm": 6.839101314544678, "learning_rate": 4.959677419354839e-05, "loss": 0.705, "step": 125 }, { "epoch": 0.8096385542168675, "grad_norm": 8.926077842712402, "learning_rate": 4.959354838709678e-05, "loss": 0.3352, "step": 126 }, { "epoch": 0.8160642570281125, "grad_norm": 7.389978885650635, "learning_rate": 4.959032258064516e-05, "loss": 0.3403, "step": 127 }, { "epoch": 0.8224899598393575, "grad_norm": 5.588038921356201, "learning_rate": 4.9587096774193556e-05, "loss": 0.4785, "step": 128 }, { "epoch": 0.8289156626506025, "grad_norm": 4.812056541442871, "learning_rate": 4.9583870967741935e-05, "loss": 0.1267, "step": 129 }, { "epoch": 0.8353413654618473, "grad_norm": 6.8123674392700195, "learning_rate": 4.958064516129032e-05, "loss": 0.61, "step": 130 }, { "epoch": 0.8417670682730923, "grad_norm": 6.3771653175354, "learning_rate": 4.957741935483871e-05, "loss": 0.7435, "step": 131 }, { "epoch": 0.8481927710843373, "grad_norm": 5.261626720428467, "learning_rate": 4.95741935483871e-05, "loss": 0.5302, "step": 132 }, { "epoch": 0.8546184738955823, "grad_norm": 7.3179612159729, "learning_rate": 4.9570967741935484e-05, "loss": 0.6297, "step": 133 }, { "epoch": 0.8610441767068273, "grad_norm": 8.682855606079102, "learning_rate": 4.956774193548388e-05, "loss": 0.5769, "step": 134 }, { "epoch": 0.8674698795180723, "grad_norm": 7.2942914962768555, "learning_rate": 4.956451612903226e-05, "loss": 0.433, "step": 135 }, { "epoch": 0.8738955823293173, "grad_norm": 10.636357307434082, "learning_rate": 4.956129032258065e-05, "loss": 0.5635, "step": 136 }, { "epoch": 0.8803212851405623, "grad_norm": 7.024284362792969, "learning_rate": 4.9558064516129034e-05, "loss": 0.5376, "step": 137 }, { "epoch": 0.8867469879518072, "grad_norm": 5.720835208892822, "learning_rate": 4.955483870967742e-05, "loss": 0.4509, "step": 138 }, { "epoch": 0.8931726907630522, "grad_norm": 10.392807006835938, "learning_rate": 4.9551612903225805e-05, "loss": 0.8249, "step": 139 }, { "epoch": 0.8995983935742972, "grad_norm": 7.5771965980529785, "learning_rate": 4.95483870967742e-05, "loss": 0.6843, "step": 140 }, { "epoch": 0.9060240963855422, "grad_norm": 8.911270141601562, "learning_rate": 4.9545161290322584e-05, "loss": 0.3044, "step": 141 }, { "epoch": 0.9124497991967871, "grad_norm": 5.429070949554443, "learning_rate": 4.954193548387097e-05, "loss": 0.3979, "step": 142 }, { "epoch": 0.9188755020080321, "grad_norm": 3.708488702774048, "learning_rate": 4.953870967741936e-05, "loss": 0.2968, "step": 143 }, { "epoch": 0.9253012048192771, "grad_norm": 3.9545493125915527, "learning_rate": 4.953548387096774e-05, "loss": 0.1416, "step": 144 }, { "epoch": 0.9317269076305221, "grad_norm": 4.794806480407715, "learning_rate": 4.9532258064516126e-05, "loss": 0.2526, "step": 145 }, { "epoch": 0.9381526104417671, "grad_norm": 5.724381446838379, "learning_rate": 4.952903225806452e-05, "loss": 0.5869, "step": 146 }, { "epoch": 0.944578313253012, "grad_norm": 7.3134565353393555, "learning_rate": 4.9525806451612905e-05, "loss": 0.4408, "step": 147 }, { "epoch": 0.951004016064257, "grad_norm": 4.378663063049316, "learning_rate": 4.952258064516129e-05, "loss": 0.3107, "step": 148 }, { "epoch": 0.957429718875502, "grad_norm": 10.66454792022705, "learning_rate": 4.951935483870968e-05, "loss": 0.6065, "step": 149 }, { "epoch": 0.963855421686747, "grad_norm": 7.237403392791748, "learning_rate": 4.951612903225807e-05, "loss": 0.4172, "step": 150 }, { "epoch": 0.970281124497992, "grad_norm": 8.880029678344727, "learning_rate": 4.9512903225806454e-05, "loss": 0.3938, "step": 151 }, { "epoch": 0.976706827309237, "grad_norm": 9.227506637573242, "learning_rate": 4.950967741935484e-05, "loss": 0.5816, "step": 152 }, { "epoch": 0.983132530120482, "grad_norm": 6.70538330078125, "learning_rate": 4.9506451612903226e-05, "loss": 0.4738, "step": 153 }, { "epoch": 0.989558232931727, "grad_norm": 11.700044631958008, "learning_rate": 4.950322580645161e-05, "loss": 0.4368, "step": 154 }, { "epoch": 0.9959839357429718, "grad_norm": 8.549652099609375, "learning_rate": 4.9500000000000004e-05, "loss": 0.3987, "step": 155 }, { "epoch": 1.0056224899598394, "grad_norm": 10.459319114685059, "learning_rate": 4.949677419354839e-05, "loss": 0.3795, "step": 156 }, { "epoch": 1.0120481927710843, "grad_norm": 19.970252990722656, "learning_rate": 4.9493548387096775e-05, "loss": 0.5301, "step": 157 }, { "epoch": 1.0184738955823294, "grad_norm": 12.03763484954834, "learning_rate": 4.949032258064517e-05, "loss": 0.3755, "step": 158 }, { "epoch": 1.0248995983935743, "grad_norm": 5.899304389953613, "learning_rate": 4.9487096774193553e-05, "loss": 0.459, "step": 159 }, { "epoch": 1.0313253012048194, "grad_norm": 11.486488342285156, "learning_rate": 4.948387096774193e-05, "loss": 0.4484, "step": 160 }, { "epoch": 1.0377510040160642, "grad_norm": 10.212610244750977, "learning_rate": 4.9480645161290325e-05, "loss": 0.3834, "step": 161 }, { "epoch": 1.0441767068273093, "grad_norm": 17.82301139831543, "learning_rate": 4.947741935483871e-05, "loss": 0.2802, "step": 162 }, { "epoch": 1.0506024096385542, "grad_norm": 3.6150712966918945, "learning_rate": 4.9474193548387096e-05, "loss": 0.5105, "step": 163 }, { "epoch": 1.057028112449799, "grad_norm": 10.530637741088867, "learning_rate": 4.947096774193549e-05, "loss": 0.495, "step": 164 }, { "epoch": 1.0634538152610442, "grad_norm": 3.974712610244751, "learning_rate": 4.9467741935483874e-05, "loss": 0.2001, "step": 165 }, { "epoch": 1.069879518072289, "grad_norm": 13.041728019714355, "learning_rate": 4.946451612903226e-05, "loss": 0.5966, "step": 166 }, { "epoch": 1.0763052208835342, "grad_norm": 10.850910186767578, "learning_rate": 4.946129032258065e-05, "loss": 0.8043, "step": 167 }, { "epoch": 1.082730923694779, "grad_norm": 4.858514308929443, "learning_rate": 4.945806451612903e-05, "loss": 0.277, "step": 168 }, { "epoch": 1.0891566265060242, "grad_norm": 7.906742095947266, "learning_rate": 4.945483870967742e-05, "loss": 0.4847, "step": 169 }, { "epoch": 1.095582329317269, "grad_norm": 4.2577314376831055, "learning_rate": 4.945161290322581e-05, "loss": 0.4953, "step": 170 }, { "epoch": 1.1020080321285142, "grad_norm": 4.778835773468018, "learning_rate": 4.9448387096774195e-05, "loss": 0.288, "step": 171 }, { "epoch": 1.108433734939759, "grad_norm": 24.172943115234375, "learning_rate": 4.944516129032258e-05, "loss": 0.8241, "step": 172 }, { "epoch": 1.114859437751004, "grad_norm": 7.6387104988098145, "learning_rate": 4.9441935483870974e-05, "loss": 0.2074, "step": 173 }, { "epoch": 1.121285140562249, "grad_norm": 6.893314361572266, "learning_rate": 4.943870967741936e-05, "loss": 0.3755, "step": 174 }, { "epoch": 1.127710843373494, "grad_norm": 15.285599708557129, "learning_rate": 4.9435483870967745e-05, "loss": 0.3059, "step": 175 }, { "epoch": 1.134136546184739, "grad_norm": 4.075883865356445, "learning_rate": 4.943225806451613e-05, "loss": 0.1908, "step": 176 }, { "epoch": 1.140562248995984, "grad_norm": 7.8378472328186035, "learning_rate": 4.9429032258064516e-05, "loss": 0.2968, "step": 177 }, { "epoch": 1.146987951807229, "grad_norm": 12.767370223999023, "learning_rate": 4.94258064516129e-05, "loss": 0.4631, "step": 178 }, { "epoch": 1.153413654618474, "grad_norm": 21.71901512145996, "learning_rate": 4.9422580645161295e-05, "loss": 0.9801, "step": 179 }, { "epoch": 1.159839357429719, "grad_norm": 7.210814476013184, "learning_rate": 4.941935483870968e-05, "loss": 0.5443, "step": 180 }, { "epoch": 1.1662650602409639, "grad_norm": 9.688201904296875, "learning_rate": 4.9416129032258066e-05, "loss": 0.6457, "step": 181 }, { "epoch": 1.1726907630522088, "grad_norm": 11.822343826293945, "learning_rate": 4.941290322580646e-05, "loss": 0.326, "step": 182 }, { "epoch": 1.1791164658634539, "grad_norm": 10.104959487915039, "learning_rate": 4.940967741935484e-05, "loss": 0.3448, "step": 183 }, { "epoch": 1.1855421686746987, "grad_norm": 8.575824737548828, "learning_rate": 4.940645161290322e-05, "loss": 0.3071, "step": 184 }, { "epoch": 1.1919678714859439, "grad_norm": 8.379708290100098, "learning_rate": 4.9403225806451616e-05, "loss": 0.4869, "step": 185 }, { "epoch": 1.1983935742971887, "grad_norm": 8.680233001708984, "learning_rate": 4.94e-05, "loss": 0.5604, "step": 186 }, { "epoch": 1.2048192771084336, "grad_norm": 13.174912452697754, "learning_rate": 4.939677419354839e-05, "loss": 0.7011, "step": 187 }, { "epoch": 1.2112449799196787, "grad_norm": 18.390470504760742, "learning_rate": 4.939354838709678e-05, "loss": 0.269, "step": 188 }, { "epoch": 1.2176706827309236, "grad_norm": 7.794299602508545, "learning_rate": 4.9390322580645165e-05, "loss": 0.2131, "step": 189 }, { "epoch": 1.2240963855421687, "grad_norm": 6.92564058303833, "learning_rate": 4.938709677419355e-05, "loss": 0.4683, "step": 190 }, { "epoch": 1.2305220883534136, "grad_norm": 22.07716941833496, "learning_rate": 4.938387096774194e-05, "loss": 0.2329, "step": 191 }, { "epoch": 1.2369477911646587, "grad_norm": 2.0575404167175293, "learning_rate": 4.938064516129032e-05, "loss": 0.4743, "step": 192 }, { "epoch": 1.2433734939759036, "grad_norm": 18.24085235595703, "learning_rate": 4.937741935483871e-05, "loss": 0.3731, "step": 193 }, { "epoch": 1.2497991967871487, "grad_norm": 19.2037410736084, "learning_rate": 4.93741935483871e-05, "loss": 0.3613, "step": 194 }, { "epoch": 1.2562248995983936, "grad_norm": 17.32529640197754, "learning_rate": 4.9370967741935486e-05, "loss": 0.4111, "step": 195 }, { "epoch": 1.2626506024096384, "grad_norm": 5.337353706359863, "learning_rate": 4.936774193548387e-05, "loss": 0.489, "step": 196 }, { "epoch": 1.2690763052208835, "grad_norm": 4.8129472732543945, "learning_rate": 4.9364516129032264e-05, "loss": 0.3701, "step": 197 }, { "epoch": 1.2755020080321284, "grad_norm": 3.077430248260498, "learning_rate": 4.936129032258065e-05, "loss": 0.4234, "step": 198 }, { "epoch": 1.2819277108433735, "grad_norm": 16.346431732177734, "learning_rate": 4.935806451612903e-05, "loss": 0.385, "step": 199 }, { "epoch": 1.2883534136546184, "grad_norm": 18.818262100219727, "learning_rate": 4.935483870967742e-05, "loss": 0.7003, "step": 200 }, { "epoch": 1.2947791164658635, "grad_norm": 13.500885009765625, "learning_rate": 4.935161290322581e-05, "loss": 0.494, "step": 201 }, { "epoch": 1.3012048192771084, "grad_norm": 4.7407402992248535, "learning_rate": 4.934838709677419e-05, "loss": 0.6118, "step": 202 }, { "epoch": 1.3076305220883535, "grad_norm": 3.48532772064209, "learning_rate": 4.9345161290322585e-05, "loss": 0.329, "step": 203 }, { "epoch": 1.3140562248995984, "grad_norm": 21.803159713745117, "learning_rate": 4.934193548387097e-05, "loss": 0.5667, "step": 204 }, { "epoch": 1.3204819277108433, "grad_norm": 9.640645980834961, "learning_rate": 4.933870967741936e-05, "loss": 0.3899, "step": 205 }, { "epoch": 1.3269076305220884, "grad_norm": 20.76266098022461, "learning_rate": 4.933548387096775e-05, "loss": 0.5509, "step": 206 }, { "epoch": 1.3333333333333333, "grad_norm": 10.639326095581055, "learning_rate": 4.933225806451613e-05, "loss": 0.4539, "step": 207 }, { "epoch": 1.3397590361445784, "grad_norm": 5.910966396331787, "learning_rate": 4.9329032258064514e-05, "loss": 0.3765, "step": 208 }, { "epoch": 1.3461847389558232, "grad_norm": 4.081665992736816, "learning_rate": 4.9325806451612907e-05, "loss": 0.412, "step": 209 }, { "epoch": 1.3526104417670683, "grad_norm": 4.758931636810303, "learning_rate": 4.932258064516129e-05, "loss": 0.3489, "step": 210 }, { "epoch": 1.3590361445783132, "grad_norm": 6.497827529907227, "learning_rate": 4.931935483870968e-05, "loss": 0.458, "step": 211 }, { "epoch": 1.3654618473895583, "grad_norm": 5.7895827293396, "learning_rate": 4.931612903225807e-05, "loss": 0.5023, "step": 212 }, { "epoch": 1.3718875502008032, "grad_norm": 4.9118242263793945, "learning_rate": 4.9312903225806456e-05, "loss": 0.4272, "step": 213 }, { "epoch": 1.378313253012048, "grad_norm": 11.30607795715332, "learning_rate": 4.9309677419354835e-05, "loss": 0.3193, "step": 214 }, { "epoch": 1.3847389558232932, "grad_norm": 10.456920623779297, "learning_rate": 4.930645161290323e-05, "loss": 0.3064, "step": 215 }, { "epoch": 1.391164658634538, "grad_norm": 43.47055435180664, "learning_rate": 4.930322580645161e-05, "loss": 1.0927, "step": 216 }, { "epoch": 1.3975903614457832, "grad_norm": 16.35538673400879, "learning_rate": 4.93e-05, "loss": 0.4783, "step": 217 }, { "epoch": 1.404016064257028, "grad_norm": 5.165719032287598, "learning_rate": 4.929677419354839e-05, "loss": 0.2396, "step": 218 }, { "epoch": 1.410441767068273, "grad_norm": 22.234893798828125, "learning_rate": 4.929354838709678e-05, "loss": 0.5887, "step": 219 }, { "epoch": 1.416867469879518, "grad_norm": 12.268518447875977, "learning_rate": 4.929032258064516e-05, "loss": 0.4012, "step": 220 }, { "epoch": 1.4232931726907632, "grad_norm": 14.700037002563477, "learning_rate": 4.9287096774193555e-05, "loss": 0.439, "step": 221 }, { "epoch": 1.429718875502008, "grad_norm": 3.185882091522217, "learning_rate": 4.9283870967741934e-05, "loss": 0.2489, "step": 222 }, { "epoch": 1.436144578313253, "grad_norm": 13.600610733032227, "learning_rate": 4.928064516129032e-05, "loss": 0.602, "step": 223 }, { "epoch": 1.442570281124498, "grad_norm": 5.31561803817749, "learning_rate": 4.927741935483871e-05, "loss": 0.4974, "step": 224 }, { "epoch": 1.448995983935743, "grad_norm": 7.039394855499268, "learning_rate": 4.92741935483871e-05, "loss": 0.3629, "step": 225 }, { "epoch": 1.455421686746988, "grad_norm": 6.997218132019043, "learning_rate": 4.9270967741935484e-05, "loss": 0.5656, "step": 226 }, { "epoch": 1.461847389558233, "grad_norm": 10.066534042358398, "learning_rate": 4.9267741935483876e-05, "loss": 0.5311, "step": 227 }, { "epoch": 1.4682730923694778, "grad_norm": 17.435487747192383, "learning_rate": 4.926451612903226e-05, "loss": 0.4987, "step": 228 }, { "epoch": 1.4746987951807229, "grad_norm": 6.782506942749023, "learning_rate": 4.926129032258065e-05, "loss": 0.3472, "step": 229 }, { "epoch": 1.481124497991968, "grad_norm": 8.608692169189453, "learning_rate": 4.9258064516129033e-05, "loss": 0.3215, "step": 230 }, { "epoch": 1.4875502008032129, "grad_norm": 22.98235321044922, "learning_rate": 4.925483870967742e-05, "loss": 0.4471, "step": 231 }, { "epoch": 1.4939759036144578, "grad_norm": 9.524588584899902, "learning_rate": 4.9251612903225805e-05, "loss": 0.2746, "step": 232 }, { "epoch": 1.5004016064257029, "grad_norm": 17.61592674255371, "learning_rate": 4.92483870967742e-05, "loss": 0.4455, "step": 233 }, { "epoch": 1.5068273092369477, "grad_norm": 15.149947166442871, "learning_rate": 4.924516129032258e-05, "loss": 0.7759, "step": 234 }, { "epoch": 1.5132530120481928, "grad_norm": 3.8440117835998535, "learning_rate": 4.924193548387097e-05, "loss": 0.1763, "step": 235 }, { "epoch": 1.5196787148594377, "grad_norm": 15.00955867767334, "learning_rate": 4.923870967741936e-05, "loss": 0.6599, "step": 236 }, { "epoch": 1.5261044176706826, "grad_norm": 22.11161994934082, "learning_rate": 4.923548387096775e-05, "loss": 0.6145, "step": 237 }, { "epoch": 1.5325301204819277, "grad_norm": 10.049532890319824, "learning_rate": 4.9232258064516126e-05, "loss": 0.5136, "step": 238 }, { "epoch": 1.5389558232931728, "grad_norm": 13.604654312133789, "learning_rate": 4.922903225806452e-05, "loss": 0.4681, "step": 239 }, { "epoch": 1.5453815261044177, "grad_norm": 6.657008647918701, "learning_rate": 4.9225806451612904e-05, "loss": 0.2875, "step": 240 }, { "epoch": 1.5518072289156626, "grad_norm": 24.565288543701172, "learning_rate": 4.922258064516129e-05, "loss": 0.5404, "step": 241 }, { "epoch": 1.5582329317269075, "grad_norm": 5.924808025360107, "learning_rate": 4.921935483870968e-05, "loss": 0.4975, "step": 242 }, { "epoch": 1.5646586345381526, "grad_norm": 14.483229637145996, "learning_rate": 4.921612903225807e-05, "loss": 0.4323, "step": 243 }, { "epoch": 1.5710843373493977, "grad_norm": 10.955862998962402, "learning_rate": 4.9212903225806454e-05, "loss": 0.5639, "step": 244 }, { "epoch": 1.5775100401606426, "grad_norm": 14.401949882507324, "learning_rate": 4.9209677419354846e-05, "loss": 0.4369, "step": 245 }, { "epoch": 1.5839357429718874, "grad_norm": 11.313241004943848, "learning_rate": 4.9206451612903225e-05, "loss": 0.4466, "step": 246 }, { "epoch": 1.5903614457831325, "grad_norm": 14.544839859008789, "learning_rate": 4.920322580645161e-05, "loss": 0.5407, "step": 247 }, { "epoch": 1.5967871485943776, "grad_norm": 10.995196342468262, "learning_rate": 4.92e-05, "loss": 0.4554, "step": 248 }, { "epoch": 1.6032128514056225, "grad_norm": 1.2909692525863647, "learning_rate": 4.919677419354839e-05, "loss": 0.4533, "step": 249 }, { "epoch": 1.6096385542168674, "grad_norm": 3.138383150100708, "learning_rate": 4.9193548387096775e-05, "loss": 0.3042, "step": 250 }, { "epoch": 1.6160642570281123, "grad_norm": 2.2950780391693115, "learning_rate": 4.919032258064517e-05, "loss": 0.5429, "step": 251 }, { "epoch": 1.6224899598393574, "grad_norm": 7.203291893005371, "learning_rate": 4.918709677419355e-05, "loss": 0.444, "step": 252 }, { "epoch": 1.6289156626506025, "grad_norm": 12.096460342407227, "learning_rate": 4.918387096774193e-05, "loss": 0.2325, "step": 253 }, { "epoch": 1.6353413654618474, "grad_norm": 6.157529354095459, "learning_rate": 4.9180645161290324e-05, "loss": 0.2522, "step": 254 }, { "epoch": 1.6417670682730923, "grad_norm": 24.771076202392578, "learning_rate": 4.917741935483871e-05, "loss": 0.5092, "step": 255 }, { "epoch": 1.6481927710843374, "grad_norm": 8.816523551940918, "learning_rate": 4.9174193548387096e-05, "loss": 0.56, "step": 256 }, { "epoch": 1.6546184738955825, "grad_norm": 25.578123092651367, "learning_rate": 4.917096774193549e-05, "loss": 0.7826, "step": 257 }, { "epoch": 1.6610441767068274, "grad_norm": 8.93865966796875, "learning_rate": 4.9167741935483874e-05, "loss": 0.2065, "step": 258 }, { "epoch": 1.6674698795180722, "grad_norm": 12.507147789001465, "learning_rate": 4.916451612903226e-05, "loss": 0.4563, "step": 259 }, { "epoch": 1.6738955823293171, "grad_norm": 4.556493759155273, "learning_rate": 4.916129032258065e-05, "loss": 0.6578, "step": 260 }, { "epoch": 1.6803212851405622, "grad_norm": 10.73857593536377, "learning_rate": 4.915806451612903e-05, "loss": 0.4954, "step": 261 }, { "epoch": 1.6867469879518073, "grad_norm": 14.427401542663574, "learning_rate": 4.915483870967742e-05, "loss": 0.5695, "step": 262 }, { "epoch": 1.6931726907630522, "grad_norm": 17.593708038330078, "learning_rate": 4.915161290322581e-05, "loss": 0.7959, "step": 263 }, { "epoch": 1.699598393574297, "grad_norm": 15.56449031829834, "learning_rate": 4.9148387096774195e-05, "loss": 0.3595, "step": 264 }, { "epoch": 1.7060240963855422, "grad_norm": 4.910727024078369, "learning_rate": 4.914516129032258e-05, "loss": 0.6815, "step": 265 }, { "epoch": 1.7124497991967873, "grad_norm": 2.633993625640869, "learning_rate": 4.914193548387097e-05, "loss": 0.3437, "step": 266 }, { "epoch": 1.7188755020080322, "grad_norm": 34.44636535644531, "learning_rate": 4.913870967741936e-05, "loss": 0.6393, "step": 267 }, { "epoch": 1.725301204819277, "grad_norm": 8.390472412109375, "learning_rate": 4.9135483870967745e-05, "loss": 0.7066, "step": 268 }, { "epoch": 1.731726907630522, "grad_norm": 4.172723770141602, "learning_rate": 4.913225806451613e-05, "loss": 0.2418, "step": 269 }, { "epoch": 1.738152610441767, "grad_norm": 16.086894989013672, "learning_rate": 4.9129032258064516e-05, "loss": 0.5735, "step": 270 }, { "epoch": 1.7445783132530122, "grad_norm": 8.542539596557617, "learning_rate": 4.91258064516129e-05, "loss": 0.3221, "step": 271 }, { "epoch": 1.751004016064257, "grad_norm": 18.947513580322266, "learning_rate": 4.9122580645161294e-05, "loss": 0.4245, "step": 272 }, { "epoch": 1.757429718875502, "grad_norm": 14.130194664001465, "learning_rate": 4.911935483870968e-05, "loss": 0.4432, "step": 273 }, { "epoch": 1.763855421686747, "grad_norm": 15.626811027526855, "learning_rate": 4.9116129032258066e-05, "loss": 0.4496, "step": 274 }, { "epoch": 1.7702811244979921, "grad_norm": 2.8836112022399902, "learning_rate": 4.911290322580646e-05, "loss": 0.251, "step": 275 }, { "epoch": 1.776706827309237, "grad_norm": 7.918999195098877, "learning_rate": 4.9109677419354844e-05, "loss": 0.683, "step": 276 }, { "epoch": 1.783132530120482, "grad_norm": 25.020790100097656, "learning_rate": 4.910645161290322e-05, "loss": 0.622, "step": 277 }, { "epoch": 1.7895582329317268, "grad_norm": 14.058879852294922, "learning_rate": 4.9103225806451615e-05, "loss": 0.4381, "step": 278 }, { "epoch": 1.7959839357429719, "grad_norm": 16.027381896972656, "learning_rate": 4.91e-05, "loss": 0.5556, "step": 279 }, { "epoch": 1.802409638554217, "grad_norm": 2.9135172367095947, "learning_rate": 4.9096774193548387e-05, "loss": 0.2383, "step": 280 }, { "epoch": 1.8088353413654619, "grad_norm": 22.196666717529297, "learning_rate": 4.909354838709678e-05, "loss": 0.5208, "step": 281 }, { "epoch": 1.8152610441767068, "grad_norm": 7.097161293029785, "learning_rate": 4.9090322580645165e-05, "loss": 0.4194, "step": 282 }, { "epoch": 1.8216867469879519, "grad_norm": 13.041504859924316, "learning_rate": 4.908709677419355e-05, "loss": 0.6518, "step": 283 }, { "epoch": 1.8281124497991967, "grad_norm": 16.859344482421875, "learning_rate": 4.908387096774194e-05, "loss": 0.4174, "step": 284 }, { "epoch": 1.8345381526104418, "grad_norm": 11.86681842803955, "learning_rate": 4.908064516129032e-05, "loss": 0.4665, "step": 285 }, { "epoch": 1.8409638554216867, "grad_norm": 8.721968650817871, "learning_rate": 4.907741935483871e-05, "loss": 0.6055, "step": 286 }, { "epoch": 1.8473895582329316, "grad_norm": 7.334936618804932, "learning_rate": 4.90741935483871e-05, "loss": 0.2537, "step": 287 }, { "epoch": 1.8538152610441767, "grad_norm": 6.172140598297119, "learning_rate": 4.9070967741935486e-05, "loss": 0.5268, "step": 288 }, { "epoch": 1.8602409638554218, "grad_norm": 2.4647939205169678, "learning_rate": 4.906774193548387e-05, "loss": 0.148, "step": 289 }, { "epoch": 1.8666666666666667, "grad_norm": 20.839069366455078, "learning_rate": 4.9064516129032264e-05, "loss": 0.6027, "step": 290 }, { "epoch": 1.8730923694779116, "grad_norm": 7.962982654571533, "learning_rate": 4.906129032258065e-05, "loss": 0.3753, "step": 291 }, { "epoch": 1.8795180722891565, "grad_norm": 8.607810974121094, "learning_rate": 4.9058064516129035e-05, "loss": 0.4818, "step": 292 }, { "epoch": 1.8859437751004016, "grad_norm": 6.273110866546631, "learning_rate": 4.905483870967742e-05, "loss": 0.4916, "step": 293 }, { "epoch": 1.8923694779116467, "grad_norm": 11.917095184326172, "learning_rate": 4.905161290322581e-05, "loss": 0.7474, "step": 294 }, { "epoch": 1.8987951807228916, "grad_norm": 5.84059476852417, "learning_rate": 4.904838709677419e-05, "loss": 0.4977, "step": 295 }, { "epoch": 1.9052208835341364, "grad_norm": 5.6499762535095215, "learning_rate": 4.9045161290322585e-05, "loss": 0.4719, "step": 296 }, { "epoch": 1.9116465863453815, "grad_norm": 6.646381855010986, "learning_rate": 4.904193548387097e-05, "loss": 0.4146, "step": 297 }, { "epoch": 1.9180722891566266, "grad_norm": 1.887205719947815, "learning_rate": 4.9038709677419356e-05, "loss": 0.2135, "step": 298 }, { "epoch": 1.9244979919678715, "grad_norm": 3.491243362426758, "learning_rate": 4.903548387096775e-05, "loss": 0.2947, "step": 299 }, { "epoch": 1.9309236947791164, "grad_norm": 8.173507690429688, "learning_rate": 4.903225806451613e-05, "loss": 0.781, "step": 300 }, { "epoch": 1.9373493975903613, "grad_norm": 11.32041072845459, "learning_rate": 4.9029032258064514e-05, "loss": 0.3044, "step": 301 }, { "epoch": 1.9437751004016064, "grad_norm": 9.811552047729492, "learning_rate": 4.9025806451612906e-05, "loss": 0.4103, "step": 302 }, { "epoch": 1.9502008032128515, "grad_norm": 15.316747665405273, "learning_rate": 4.902258064516129e-05, "loss": 0.5024, "step": 303 }, { "epoch": 1.9566265060240964, "grad_norm": 9.213105201721191, "learning_rate": 4.901935483870968e-05, "loss": 0.7844, "step": 304 }, { "epoch": 1.9630522088353413, "grad_norm": 16.751731872558594, "learning_rate": 4.901612903225807e-05, "loss": 0.5876, "step": 305 }, { "epoch": 1.9694779116465864, "grad_norm": 9.24449634552002, "learning_rate": 4.9012903225806456e-05, "loss": 0.3848, "step": 306 }, { "epoch": 1.9759036144578315, "grad_norm": 14.611693382263184, "learning_rate": 4.900967741935484e-05, "loss": 0.373, "step": 307 }, { "epoch": 1.9823293172690764, "grad_norm": 5.673837184906006, "learning_rate": 4.900645161290323e-05, "loss": 0.4228, "step": 308 }, { "epoch": 1.9887550200803212, "grad_norm": 4.408359050750732, "learning_rate": 4.900322580645161e-05, "loss": 0.317, "step": 309 }, { "epoch": 1.9951807228915661, "grad_norm": 14.352408409118652, "learning_rate": 4.9e-05, "loss": 0.5554, "step": 310 }, { "epoch": 2.004819277108434, "grad_norm": 6.351159572601318, "learning_rate": 4.899677419354839e-05, "loss": 0.4952, "step": 311 }, { "epoch": 2.0112449799196788, "grad_norm": 16.67235565185547, "learning_rate": 4.8993548387096777e-05, "loss": 0.7014, "step": 312 }, { "epoch": 2.0176706827309236, "grad_norm": 8.022540092468262, "learning_rate": 4.899032258064516e-05, "loss": 0.3891, "step": 313 }, { "epoch": 2.0240963855421685, "grad_norm": 3.879352331161499, "learning_rate": 4.8987096774193555e-05, "loss": 0.4804, "step": 314 }, { "epoch": 2.030522088353414, "grad_norm": 16.29292869567871, "learning_rate": 4.898387096774194e-05, "loss": 0.4678, "step": 315 }, { "epoch": 2.0369477911646587, "grad_norm": 31.95669937133789, "learning_rate": 4.898064516129032e-05, "loss": 0.4545, "step": 316 }, { "epoch": 2.0433734939759036, "grad_norm": 8.443345069885254, "learning_rate": 4.897741935483871e-05, "loss": 0.7159, "step": 317 }, { "epoch": 2.0497991967871485, "grad_norm": 7.071547985076904, "learning_rate": 4.89741935483871e-05, "loss": 0.5925, "step": 318 }, { "epoch": 2.0562248995983934, "grad_norm": 5.981539726257324, "learning_rate": 4.897096774193548e-05, "loss": 0.2946, "step": 319 }, { "epoch": 2.0626506024096387, "grad_norm": 17.490751266479492, "learning_rate": 4.8967741935483876e-05, "loss": 0.3179, "step": 320 }, { "epoch": 2.0690763052208836, "grad_norm": 7.483447074890137, "learning_rate": 4.896451612903226e-05, "loss": 0.3135, "step": 321 }, { "epoch": 2.0755020080321285, "grad_norm": 22.21017074584961, "learning_rate": 4.896129032258065e-05, "loss": 0.8873, "step": 322 }, { "epoch": 2.0819277108433734, "grad_norm": 5.313900947570801, "learning_rate": 4.895806451612903e-05, "loss": 0.5735, "step": 323 }, { "epoch": 2.0883534136546187, "grad_norm": 7.3915886878967285, "learning_rate": 4.895483870967742e-05, "loss": 0.3577, "step": 324 }, { "epoch": 2.0947791164658636, "grad_norm": 4.42692232131958, "learning_rate": 4.8951612903225804e-05, "loss": 0.2308, "step": 325 }, { "epoch": 2.1012048192771084, "grad_norm": 11.525419235229492, "learning_rate": 4.89483870967742e-05, "loss": 0.5847, "step": 326 }, { "epoch": 2.1076305220883533, "grad_norm": 34.85057067871094, "learning_rate": 4.894516129032258e-05, "loss": 0.3937, "step": 327 }, { "epoch": 2.114056224899598, "grad_norm": 10.172204971313477, "learning_rate": 4.894193548387097e-05, "loss": 0.516, "step": 328 }, { "epoch": 2.1204819277108435, "grad_norm": 13.956056594848633, "learning_rate": 4.893870967741936e-05, "loss": 0.7459, "step": 329 }, { "epoch": 2.1269076305220884, "grad_norm": 12.076092720031738, "learning_rate": 4.8935483870967746e-05, "loss": 0.4057, "step": 330 }, { "epoch": 2.1333333333333333, "grad_norm": 6.308197498321533, "learning_rate": 4.893225806451613e-05, "loss": 0.1869, "step": 331 }, { "epoch": 2.139759036144578, "grad_norm": 6.908775329589844, "learning_rate": 4.892903225806452e-05, "loss": 0.4106, "step": 332 }, { "epoch": 2.1461847389558235, "grad_norm": 6.7748589515686035, "learning_rate": 4.8925806451612904e-05, "loss": 0.2467, "step": 333 }, { "epoch": 2.1526104417670684, "grad_norm": 16.169015884399414, "learning_rate": 4.892258064516129e-05, "loss": 0.7584, "step": 334 }, { "epoch": 2.1590361445783133, "grad_norm": 1.7853831052780151, "learning_rate": 4.891935483870968e-05, "loss": 0.3354, "step": 335 }, { "epoch": 2.165461847389558, "grad_norm": 17.208454132080078, "learning_rate": 4.891612903225807e-05, "loss": 0.6562, "step": 336 }, { "epoch": 2.171887550200803, "grad_norm": 6.221571922302246, "learning_rate": 4.891290322580645e-05, "loss": 0.3783, "step": 337 }, { "epoch": 2.1783132530120484, "grad_norm": 14.36730670928955, "learning_rate": 4.8909677419354846e-05, "loss": 0.4372, "step": 338 }, { "epoch": 2.1847389558232932, "grad_norm": 14.665215492248535, "learning_rate": 4.8906451612903225e-05, "loss": 0.4764, "step": 339 }, { "epoch": 2.191164658634538, "grad_norm": 20.386966705322266, "learning_rate": 4.890322580645161e-05, "loss": 0.8589, "step": 340 }, { "epoch": 2.197590361445783, "grad_norm": 14.547446250915527, "learning_rate": 4.89e-05, "loss": 0.3582, "step": 341 }, { "epoch": 2.2040160642570283, "grad_norm": 16.743844985961914, "learning_rate": 4.889677419354839e-05, "loss": 0.4246, "step": 342 }, { "epoch": 2.2104417670682732, "grad_norm": 13.118790626525879, "learning_rate": 4.8893548387096774e-05, "loss": 0.3535, "step": 343 }, { "epoch": 2.216867469879518, "grad_norm": 32.728187561035156, "learning_rate": 4.889032258064517e-05, "loss": 0.5535, "step": 344 }, { "epoch": 2.223293172690763, "grad_norm": 3.614194869995117, "learning_rate": 4.888709677419355e-05, "loss": 0.3667, "step": 345 }, { "epoch": 2.229718875502008, "grad_norm": 9.299104690551758, "learning_rate": 4.888387096774194e-05, "loss": 0.3806, "step": 346 }, { "epoch": 2.236144578313253, "grad_norm": 3.511805295944214, "learning_rate": 4.8880645161290324e-05, "loss": 0.2592, "step": 347 }, { "epoch": 2.242570281124498, "grad_norm": 11.291939735412598, "learning_rate": 4.887741935483871e-05, "loss": 0.6101, "step": 348 }, { "epoch": 2.248995983935743, "grad_norm": 5.1154303550720215, "learning_rate": 4.8874193548387095e-05, "loss": 0.4683, "step": 349 }, { "epoch": 2.255421686746988, "grad_norm": 2.0693275928497314, "learning_rate": 4.887096774193549e-05, "loss": 0.4991, "step": 350 }, { "epoch": 2.261847389558233, "grad_norm": 5.340583324432373, "learning_rate": 4.886774193548387e-05, "loss": 0.4371, "step": 351 }, { "epoch": 2.268273092369478, "grad_norm": 20.421960830688477, "learning_rate": 4.886451612903226e-05, "loss": 0.8064, "step": 352 }, { "epoch": 2.274698795180723, "grad_norm": 2.3375906944274902, "learning_rate": 4.886129032258065e-05, "loss": 0.3873, "step": 353 }, { "epoch": 2.281124497991968, "grad_norm": 11.518815040588379, "learning_rate": 4.885806451612904e-05, "loss": 0.5747, "step": 354 }, { "epoch": 2.2875502008032127, "grad_norm": 13.740503311157227, "learning_rate": 4.885483870967742e-05, "loss": 0.5884, "step": 355 }, { "epoch": 2.293975903614458, "grad_norm": 14.489356994628906, "learning_rate": 4.885161290322581e-05, "loss": 0.4256, "step": 356 }, { "epoch": 2.300401606425703, "grad_norm": 13.74975299835205, "learning_rate": 4.8848387096774194e-05, "loss": 0.607, "step": 357 }, { "epoch": 2.306827309236948, "grad_norm": 10.78126335144043, "learning_rate": 4.884516129032258e-05, "loss": 0.6392, "step": 358 }, { "epoch": 2.3132530120481927, "grad_norm": 6.021533012390137, "learning_rate": 4.884193548387097e-05, "loss": 0.6007, "step": 359 }, { "epoch": 2.319678714859438, "grad_norm": 12.449559211730957, "learning_rate": 4.883870967741936e-05, "loss": 0.7541, "step": 360 }, { "epoch": 2.326104417670683, "grad_norm": 5.090046405792236, "learning_rate": 4.8835483870967744e-05, "loss": 0.527, "step": 361 }, { "epoch": 2.3325301204819278, "grad_norm": 5.470488548278809, "learning_rate": 4.883225806451613e-05, "loss": 0.4885, "step": 362 }, { "epoch": 2.3389558232931726, "grad_norm": 12.449920654296875, "learning_rate": 4.8829032258064515e-05, "loss": 0.6658, "step": 363 }, { "epoch": 2.3453815261044175, "grad_norm": 8.215715408325195, "learning_rate": 4.88258064516129e-05, "loss": 0.3854, "step": 364 }, { "epoch": 2.3518072289156624, "grad_norm": 14.366580963134766, "learning_rate": 4.8822580645161294e-05, "loss": 0.7023, "step": 365 }, { "epoch": 2.3582329317269077, "grad_norm": 14.731744766235352, "learning_rate": 4.881935483870968e-05, "loss": 0.3573, "step": 366 }, { "epoch": 2.3646586345381526, "grad_norm": 12.46597671508789, "learning_rate": 4.8816129032258065e-05, "loss": 0.6551, "step": 367 }, { "epoch": 2.3710843373493975, "grad_norm": 8.746798515319824, "learning_rate": 4.881290322580646e-05, "loss": 0.7609, "step": 368 }, { "epoch": 2.3775100401606424, "grad_norm": 20.84713363647461, "learning_rate": 4.880967741935484e-05, "loss": 0.4698, "step": 369 }, { "epoch": 2.3839357429718877, "grad_norm": 2.750516414642334, "learning_rate": 4.880645161290323e-05, "loss": 0.3008, "step": 370 }, { "epoch": 2.3903614457831326, "grad_norm": 2.013629913330078, "learning_rate": 4.8803225806451615e-05, "loss": 0.6595, "step": 371 }, { "epoch": 2.3967871485943775, "grad_norm": 24.612573623657227, "learning_rate": 4.88e-05, "loss": 0.4471, "step": 372 }, { "epoch": 2.4032128514056224, "grad_norm": 2.9059982299804688, "learning_rate": 4.8796774193548386e-05, "loss": 0.2235, "step": 373 }, { "epoch": 2.4096385542168672, "grad_norm": 13.482975959777832, "learning_rate": 4.879354838709678e-05, "loss": 0.5299, "step": 374 }, { "epoch": 2.4160642570281126, "grad_norm": 12.513169288635254, "learning_rate": 4.8790322580645164e-05, "loss": 0.4757, "step": 375 }, { "epoch": 2.4224899598393574, "grad_norm": 5.080874443054199, "learning_rate": 4.878709677419355e-05, "loss": 0.8612, "step": 376 }, { "epoch": 2.4289156626506023, "grad_norm": 7.49531364440918, "learning_rate": 4.878387096774194e-05, "loss": 0.2984, "step": 377 }, { "epoch": 2.435341365461847, "grad_norm": 6.403908729553223, "learning_rate": 4.878064516129032e-05, "loss": 0.4932, "step": 378 }, { "epoch": 2.4417670682730925, "grad_norm": 6.011518955230713, "learning_rate": 4.877741935483871e-05, "loss": 0.4583, "step": 379 }, { "epoch": 2.4481927710843374, "grad_norm": 4.414745807647705, "learning_rate": 4.87741935483871e-05, "loss": 0.3563, "step": 380 }, { "epoch": 2.4546184738955823, "grad_norm": 6.991732120513916, "learning_rate": 4.8770967741935485e-05, "loss": 0.5714, "step": 381 }, { "epoch": 2.461044176706827, "grad_norm": 5.321561813354492, "learning_rate": 4.876774193548387e-05, "loss": 0.5287, "step": 382 }, { "epoch": 2.467469879518072, "grad_norm": 8.740898132324219, "learning_rate": 4.8764516129032263e-05, "loss": 0.6465, "step": 383 }, { "epoch": 2.4738955823293174, "grad_norm": 1.8816094398498535, "learning_rate": 4.876129032258065e-05, "loss": 0.584, "step": 384 }, { "epoch": 2.4803212851405623, "grad_norm": 6.791060447692871, "learning_rate": 4.8758064516129035e-05, "loss": 0.3995, "step": 385 }, { "epoch": 2.486746987951807, "grad_norm": 8.387206077575684, "learning_rate": 4.875483870967742e-05, "loss": 0.3447, "step": 386 }, { "epoch": 2.493172690763052, "grad_norm": 8.550765991210938, "learning_rate": 4.8751612903225806e-05, "loss": 0.6371, "step": 387 }, { "epoch": 2.4995983935742974, "grad_norm": 8.890618324279785, "learning_rate": 4.874838709677419e-05, "loss": 0.7069, "step": 388 }, { "epoch": 2.5060240963855422, "grad_norm": 1.8137555122375488, "learning_rate": 4.8745161290322584e-05, "loss": 0.4034, "step": 389 }, { "epoch": 2.512449799196787, "grad_norm": 7.715780258178711, "learning_rate": 4.874193548387097e-05, "loss": 0.5108, "step": 390 }, { "epoch": 2.518875502008032, "grad_norm": 2.620068311691284, "learning_rate": 4.8738709677419356e-05, "loss": 0.3307, "step": 391 }, { "epoch": 2.525301204819277, "grad_norm": 15.666391372680664, "learning_rate": 4.873548387096775e-05, "loss": 0.4989, "step": 392 }, { "epoch": 2.531726907630522, "grad_norm": 7.824061393737793, "learning_rate": 4.873225806451613e-05, "loss": 0.4065, "step": 393 }, { "epoch": 2.538152610441767, "grad_norm": 8.207316398620605, "learning_rate": 4.872903225806452e-05, "loss": 0.5798, "step": 394 }, { "epoch": 2.544578313253012, "grad_norm": 17.222562789916992, "learning_rate": 4.8725806451612905e-05, "loss": 0.5453, "step": 395 }, { "epoch": 2.551004016064257, "grad_norm": 10.560887336730957, "learning_rate": 4.872258064516129e-05, "loss": 0.5786, "step": 396 }, { "epoch": 2.557429718875502, "grad_norm": 26.903820037841797, "learning_rate": 4.871935483870968e-05, "loss": 0.3916, "step": 397 }, { "epoch": 2.563855421686747, "grad_norm": 6.306243896484375, "learning_rate": 4.871612903225807e-05, "loss": 0.2331, "step": 398 }, { "epoch": 2.570281124497992, "grad_norm": 34.15444564819336, "learning_rate": 4.8712903225806455e-05, "loss": 0.563, "step": 399 }, { "epoch": 2.576706827309237, "grad_norm": 8.776156425476074, "learning_rate": 4.870967741935484e-05, "loss": 0.5481, "step": 400 }, { "epoch": 2.5831325301204817, "grad_norm": 7.624814033508301, "learning_rate": 4.8706451612903226e-05, "loss": 0.2782, "step": 401 }, { "epoch": 2.589558232931727, "grad_norm": 4.0539422035217285, "learning_rate": 4.870322580645161e-05, "loss": 0.3276, "step": 402 }, { "epoch": 2.595983935742972, "grad_norm": 4.275896072387695, "learning_rate": 4.87e-05, "loss": 0.5248, "step": 403 }, { "epoch": 2.602409638554217, "grad_norm": 17.009023666381836, "learning_rate": 4.869677419354839e-05, "loss": 0.5927, "step": 404 }, { "epoch": 2.6088353413654617, "grad_norm": 10.069941520690918, "learning_rate": 4.8693548387096776e-05, "loss": 0.5237, "step": 405 }, { "epoch": 2.615261044176707, "grad_norm": 3.64692759513855, "learning_rate": 4.869032258064516e-05, "loss": 0.3194, "step": 406 }, { "epoch": 2.621686746987952, "grad_norm": 25.63099479675293, "learning_rate": 4.8687096774193554e-05, "loss": 0.605, "step": 407 }, { "epoch": 2.628112449799197, "grad_norm": 20.53011131286621, "learning_rate": 4.868387096774194e-05, "loss": 0.1775, "step": 408 }, { "epoch": 2.6345381526104417, "grad_norm": 4.56806755065918, "learning_rate": 4.8680645161290326e-05, "loss": 0.165, "step": 409 }, { "epoch": 2.6409638554216865, "grad_norm": 5.896443843841553, "learning_rate": 4.867741935483871e-05, "loss": 0.4417, "step": 410 }, { "epoch": 2.647389558232932, "grad_norm": 9.775384902954102, "learning_rate": 4.86741935483871e-05, "loss": 0.3289, "step": 411 }, { "epoch": 2.6538152610441768, "grad_norm": 9.258476257324219, "learning_rate": 4.867096774193548e-05, "loss": 0.5758, "step": 412 }, { "epoch": 2.6602409638554216, "grad_norm": 3.2956924438476562, "learning_rate": 4.8667741935483875e-05, "loss": 0.3894, "step": 413 }, { "epoch": 2.6666666666666665, "grad_norm": 4.302741050720215, "learning_rate": 4.866451612903226e-05, "loss": 0.5758, "step": 414 }, { "epoch": 2.673092369477912, "grad_norm": 10.264629364013672, "learning_rate": 4.866129032258065e-05, "loss": 0.6454, "step": 415 }, { "epoch": 2.6795180722891567, "grad_norm": 33.70549774169922, "learning_rate": 4.865806451612904e-05, "loss": 0.3498, "step": 416 }, { "epoch": 2.6859437751004016, "grad_norm": 9.362329483032227, "learning_rate": 4.865483870967742e-05, "loss": 0.4627, "step": 417 }, { "epoch": 2.6923694779116465, "grad_norm": 4.578156471252441, "learning_rate": 4.865161290322581e-05, "loss": 0.376, "step": 418 }, { "epoch": 2.6987951807228914, "grad_norm": 20.416845321655273, "learning_rate": 4.8648387096774196e-05, "loss": 0.6402, "step": 419 }, { "epoch": 2.7052208835341367, "grad_norm": 20.018945693969727, "learning_rate": 4.864516129032258e-05, "loss": 0.3857, "step": 420 }, { "epoch": 2.7116465863453816, "grad_norm": 7.791701793670654, "learning_rate": 4.864193548387097e-05, "loss": 0.408, "step": 421 }, { "epoch": 2.7180722891566265, "grad_norm": 3.432651996612549, "learning_rate": 4.863870967741936e-05, "loss": 0.412, "step": 422 }, { "epoch": 2.7244979919678713, "grad_norm": 1.9815020561218262, "learning_rate": 4.8635483870967746e-05, "loss": 0.3019, "step": 423 }, { "epoch": 2.7309236947791167, "grad_norm": 5.5898284912109375, "learning_rate": 4.863225806451613e-05, "loss": 0.4115, "step": 424 }, { "epoch": 2.7373493975903616, "grad_norm": 7.170601844787598, "learning_rate": 4.862903225806452e-05, "loss": 0.9146, "step": 425 }, { "epoch": 2.7437751004016064, "grad_norm": 9.127796173095703, "learning_rate": 4.86258064516129e-05, "loss": 0.4653, "step": 426 }, { "epoch": 2.7502008032128513, "grad_norm": 67.5741958618164, "learning_rate": 4.862258064516129e-05, "loss": 0.5334, "step": 427 }, { "epoch": 2.756626506024096, "grad_norm": 16.35715103149414, "learning_rate": 4.861935483870968e-05, "loss": 0.398, "step": 428 }, { "epoch": 2.7630522088353415, "grad_norm": 6.73328971862793, "learning_rate": 4.861612903225807e-05, "loss": 0.6422, "step": 429 }, { "epoch": 2.7694779116465864, "grad_norm": 14.288458824157715, "learning_rate": 4.861290322580645e-05, "loss": 0.4259, "step": 430 }, { "epoch": 2.7759036144578313, "grad_norm": 1.9930909872055054, "learning_rate": 4.8609677419354845e-05, "loss": 0.3315, "step": 431 }, { "epoch": 2.782329317269076, "grad_norm": 51.40473175048828, "learning_rate": 4.8606451612903224e-05, "loss": 0.8764, "step": 432 }, { "epoch": 2.7887550200803215, "grad_norm": 13.921738624572754, "learning_rate": 4.8603225806451617e-05, "loss": 0.701, "step": 433 }, { "epoch": 2.7951807228915664, "grad_norm": 12.915340423583984, "learning_rate": 4.86e-05, "loss": 0.5316, "step": 434 }, { "epoch": 2.8016064257028113, "grad_norm": 8.66297721862793, "learning_rate": 4.859677419354839e-05, "loss": 0.4282, "step": 435 }, { "epoch": 2.808032128514056, "grad_norm": 7.898273468017578, "learning_rate": 4.8593548387096774e-05, "loss": 0.4403, "step": 436 }, { "epoch": 2.814457831325301, "grad_norm": 8.397368431091309, "learning_rate": 4.8590322580645166e-05, "loss": 0.7394, "step": 437 }, { "epoch": 2.820883534136546, "grad_norm": 7.648226737976074, "learning_rate": 4.858709677419355e-05, "loss": 0.4991, "step": 438 }, { "epoch": 2.8273092369477912, "grad_norm": 5.623685836791992, "learning_rate": 4.858387096774194e-05, "loss": 0.4022, "step": 439 }, { "epoch": 2.833734939759036, "grad_norm": 12.498046875, "learning_rate": 4.858064516129032e-05, "loss": 0.6475, "step": 440 }, { "epoch": 2.840160642570281, "grad_norm": 20.600244522094727, "learning_rate": 4.857741935483871e-05, "loss": 0.621, "step": 441 }, { "epoch": 2.8465863453815263, "grad_norm": 9.307296752929688, "learning_rate": 4.8574193548387095e-05, "loss": 0.4617, "step": 442 }, { "epoch": 2.853012048192771, "grad_norm": 10.652766227722168, "learning_rate": 4.857096774193549e-05, "loss": 0.651, "step": 443 }, { "epoch": 2.859437751004016, "grad_norm": 8.533763885498047, "learning_rate": 4.856774193548387e-05, "loss": 0.2352, "step": 444 }, { "epoch": 2.865863453815261, "grad_norm": 14.623491287231445, "learning_rate": 4.856451612903226e-05, "loss": 0.4681, "step": 445 }, { "epoch": 2.872289156626506, "grad_norm": 10.192721366882324, "learning_rate": 4.856129032258065e-05, "loss": 0.7164, "step": 446 }, { "epoch": 2.8787148594377507, "grad_norm": 10.455683708190918, "learning_rate": 4.855806451612904e-05, "loss": 0.7475, "step": 447 }, { "epoch": 2.885140562248996, "grad_norm": 2.763063907623291, "learning_rate": 4.855483870967742e-05, "loss": 0.1782, "step": 448 }, { "epoch": 2.891566265060241, "grad_norm": 18.533191680908203, "learning_rate": 4.855161290322581e-05, "loss": 0.4251, "step": 449 }, { "epoch": 2.897991967871486, "grad_norm": 5.099410057067871, "learning_rate": 4.8548387096774194e-05, "loss": 0.5289, "step": 450 }, { "epoch": 2.904417670682731, "grad_norm": 5.630273818969727, "learning_rate": 4.854516129032258e-05, "loss": 0.523, "step": 451 }, { "epoch": 2.910843373493976, "grad_norm": 15.771490097045898, "learning_rate": 4.854193548387097e-05, "loss": 0.4006, "step": 452 }, { "epoch": 2.917269076305221, "grad_norm": 13.709071159362793, "learning_rate": 4.853870967741936e-05, "loss": 0.5591, "step": 453 }, { "epoch": 2.923694779116466, "grad_norm": 27.137903213500977, "learning_rate": 4.8535483870967743e-05, "loss": 0.3166, "step": 454 }, { "epoch": 2.9301204819277107, "grad_norm": 7.0183796882629395, "learning_rate": 4.8532258064516136e-05, "loss": 0.4704, "step": 455 }, { "epoch": 2.9365461847389556, "grad_norm": 6.372150421142578, "learning_rate": 4.8529032258064515e-05, "loss": 0.3968, "step": 456 }, { "epoch": 2.942971887550201, "grad_norm": 3.3378500938415527, "learning_rate": 4.852580645161291e-05, "loss": 0.5814, "step": 457 }, { "epoch": 2.9493975903614458, "grad_norm": 12.45730209350586, "learning_rate": 4.852258064516129e-05, "loss": 0.6558, "step": 458 }, { "epoch": 2.9558232931726907, "grad_norm": 10.92990493774414, "learning_rate": 4.851935483870968e-05, "loss": 0.5588, "step": 459 }, { "epoch": 2.962248995983936, "grad_norm": 15.077335357666016, "learning_rate": 4.8516129032258065e-05, "loss": 0.4129, "step": 460 }, { "epoch": 2.968674698795181, "grad_norm": 3.0889978408813477, "learning_rate": 4.851290322580646e-05, "loss": 0.5012, "step": 461 }, { "epoch": 2.9751004016064257, "grad_norm": 4.295764446258545, "learning_rate": 4.850967741935484e-05, "loss": 0.506, "step": 462 }, { "epoch": 2.9815261044176706, "grad_norm": 22.0218448638916, "learning_rate": 4.850645161290323e-05, "loss": 0.4264, "step": 463 }, { "epoch": 2.9879518072289155, "grad_norm": 3.9791531562805176, "learning_rate": 4.8503225806451614e-05, "loss": 0.4415, "step": 464 }, { "epoch": 2.9943775100401604, "grad_norm": 12.948200225830078, "learning_rate": 4.85e-05, "loss": 1.0042, "step": 465 }, { "epoch": 3.004016064257028, "grad_norm": 15.94398307800293, "learning_rate": 4.8496774193548386e-05, "loss": 0.5203, "step": 466 }, { "epoch": 3.010441767068273, "grad_norm": 13.07730484008789, "learning_rate": 4.849354838709678e-05, "loss": 0.5827, "step": 467 }, { "epoch": 3.016867469879518, "grad_norm": 10.98026180267334, "learning_rate": 4.8490322580645164e-05, "loss": 0.5059, "step": 468 }, { "epoch": 3.0232931726907633, "grad_norm": 7.850858211517334, "learning_rate": 4.848709677419355e-05, "loss": 0.3007, "step": 469 }, { "epoch": 3.029718875502008, "grad_norm": 18.120473861694336, "learning_rate": 4.848387096774194e-05, "loss": 0.6419, "step": 470 }, { "epoch": 3.036144578313253, "grad_norm": 18.985387802124023, "learning_rate": 4.848064516129032e-05, "loss": 0.481, "step": 471 }, { "epoch": 3.042570281124498, "grad_norm": 8.193123817443848, "learning_rate": 4.847741935483871e-05, "loss": 0.4511, "step": 472 }, { "epoch": 3.048995983935743, "grad_norm": 9.259217262268066, "learning_rate": 4.84741935483871e-05, "loss": 0.6253, "step": 473 }, { "epoch": 3.055421686746988, "grad_norm": 19.84555435180664, "learning_rate": 4.8470967741935485e-05, "loss": 0.2375, "step": 474 }, { "epoch": 3.061847389558233, "grad_norm": 3.182964563369751, "learning_rate": 4.846774193548387e-05, "loss": 0.4418, "step": 475 }, { "epoch": 3.068273092369478, "grad_norm": 8.964004516601562, "learning_rate": 4.846451612903226e-05, "loss": 0.5136, "step": 476 }, { "epoch": 3.0746987951807228, "grad_norm": 17.29197883605957, "learning_rate": 4.846129032258065e-05, "loss": 0.5367, "step": 477 }, { "epoch": 3.081124497991968, "grad_norm": 5.304012298583984, "learning_rate": 4.8458064516129034e-05, "loss": 0.5114, "step": 478 }, { "epoch": 3.087550200803213, "grad_norm": 8.187363624572754, "learning_rate": 4.845483870967742e-05, "loss": 0.7325, "step": 479 }, { "epoch": 3.093975903614458, "grad_norm": 5.264475345611572, "learning_rate": 4.8451612903225806e-05, "loss": 0.4822, "step": 480 }, { "epoch": 3.1004016064257027, "grad_norm": 10.95467758178711, "learning_rate": 4.84483870967742e-05, "loss": 0.577, "step": 481 }, { "epoch": 3.1068273092369476, "grad_norm": 5.726831436157227, "learning_rate": 4.8445161290322584e-05, "loss": 0.6941, "step": 482 }, { "epoch": 3.113253012048193, "grad_norm": 6.738022327423096, "learning_rate": 4.844193548387097e-05, "loss": 0.2973, "step": 483 }, { "epoch": 3.119678714859438, "grad_norm": 10.215533256530762, "learning_rate": 4.8438709677419355e-05, "loss": 0.8186, "step": 484 }, { "epoch": 3.1261044176706827, "grad_norm": 15.263949394226074, "learning_rate": 4.843548387096775e-05, "loss": 0.5617, "step": 485 }, { "epoch": 3.1325301204819276, "grad_norm": 4.4117841720581055, "learning_rate": 4.8432258064516134e-05, "loss": 0.3511, "step": 486 }, { "epoch": 3.1389558232931725, "grad_norm": 5.299454689025879, "learning_rate": 4.842903225806452e-05, "loss": 0.5059, "step": 487 }, { "epoch": 3.145381526104418, "grad_norm": 5.67463493347168, "learning_rate": 4.8425806451612905e-05, "loss": 0.434, "step": 488 }, { "epoch": 3.1518072289156627, "grad_norm": 5.993772983551025, "learning_rate": 4.842258064516129e-05, "loss": 0.3853, "step": 489 }, { "epoch": 3.1582329317269076, "grad_norm": 20.375728607177734, "learning_rate": 4.8419354838709676e-05, "loss": 0.6868, "step": 490 }, { "epoch": 3.1646586345381524, "grad_norm": 18.89373207092285, "learning_rate": 4.841612903225807e-05, "loss": 0.5022, "step": 491 }, { "epoch": 3.1710843373493978, "grad_norm": 2.6128909587860107, "learning_rate": 4.8412903225806455e-05, "loss": 0.4144, "step": 492 }, { "epoch": 3.1775100401606426, "grad_norm": 14.53881549835205, "learning_rate": 4.840967741935484e-05, "loss": 0.3426, "step": 493 }, { "epoch": 3.1839357429718875, "grad_norm": 7.790198802947998, "learning_rate": 4.840645161290323e-05, "loss": 0.4309, "step": 494 }, { "epoch": 3.1903614457831324, "grad_norm": 9.81341552734375, "learning_rate": 4.840322580645161e-05, "loss": 0.477, "step": 495 }, { "epoch": 3.1967871485943773, "grad_norm": 12.055136680603027, "learning_rate": 4.8400000000000004e-05, "loss": 0.4857, "step": 496 }, { "epoch": 3.2032128514056226, "grad_norm": 7.5992112159729, "learning_rate": 4.839677419354839e-05, "loss": 0.3475, "step": 497 }, { "epoch": 3.2096385542168675, "grad_norm": 5.409062385559082, "learning_rate": 4.8393548387096776e-05, "loss": 0.58, "step": 498 }, { "epoch": 3.2160642570281124, "grad_norm": 5.764282703399658, "learning_rate": 4.839032258064516e-05, "loss": 0.598, "step": 499 }, { "epoch": 3.2224899598393573, "grad_norm": 8.959898948669434, "learning_rate": 4.8387096774193554e-05, "loss": 0.6535, "step": 500 }, { "epoch": 3.2289156626506026, "grad_norm": 6.959432601928711, "learning_rate": 4.838387096774194e-05, "loss": 0.3986, "step": 501 }, { "epoch": 3.2353413654618475, "grad_norm": 11.26677417755127, "learning_rate": 4.8380645161290325e-05, "loss": 0.783, "step": 502 }, { "epoch": 3.2417670682730924, "grad_norm": 13.44477367401123, "learning_rate": 4.837741935483871e-05, "loss": 0.4541, "step": 503 }, { "epoch": 3.2481927710843372, "grad_norm": 10.80493450164795, "learning_rate": 4.8374193548387097e-05, "loss": 0.4686, "step": 504 }, { "epoch": 3.254618473895582, "grad_norm": 3.565406322479248, "learning_rate": 4.837096774193548e-05, "loss": 0.431, "step": 505 }, { "epoch": 3.2610441767068274, "grad_norm": 15.313773155212402, "learning_rate": 4.8367741935483875e-05, "loss": 0.547, "step": 506 }, { "epoch": 3.2674698795180723, "grad_norm": 3.426693916320801, "learning_rate": 4.836451612903226e-05, "loss": 0.3476, "step": 507 }, { "epoch": 3.273895582329317, "grad_norm": 6.360467910766602, "learning_rate": 4.8361290322580646e-05, "loss": 0.4399, "step": 508 }, { "epoch": 3.280321285140562, "grad_norm": 12.212074279785156, "learning_rate": 4.835806451612904e-05, "loss": 0.588, "step": 509 }, { "epoch": 3.2867469879518074, "grad_norm": 20.115161895751953, "learning_rate": 4.835483870967742e-05, "loss": 0.5232, "step": 510 }, { "epoch": 3.2931726907630523, "grad_norm": 10.932350158691406, "learning_rate": 4.835161290322581e-05, "loss": 0.4352, "step": 511 }, { "epoch": 3.299598393574297, "grad_norm": 4.497518062591553, "learning_rate": 4.8348387096774196e-05, "loss": 0.3399, "step": 512 }, { "epoch": 3.306024096385542, "grad_norm": 14.720540046691895, "learning_rate": 4.834516129032258e-05, "loss": 0.3405, "step": 513 }, { "epoch": 3.312449799196787, "grad_norm": 12.249598503112793, "learning_rate": 4.834193548387097e-05, "loss": 0.4385, "step": 514 }, { "epoch": 3.3188755020080323, "grad_norm": 7.282612323760986, "learning_rate": 4.833870967741936e-05, "loss": 0.4501, "step": 515 }, { "epoch": 3.325301204819277, "grad_norm": 21.23067855834961, "learning_rate": 4.8335483870967745e-05, "loss": 0.8596, "step": 516 }, { "epoch": 3.331726907630522, "grad_norm": 3.4805657863616943, "learning_rate": 4.833225806451613e-05, "loss": 0.4496, "step": 517 }, { "epoch": 3.338152610441767, "grad_norm": 5.133415222167969, "learning_rate": 4.832903225806452e-05, "loss": 0.6887, "step": 518 }, { "epoch": 3.3445783132530122, "grad_norm": 12.120992660522461, "learning_rate": 4.83258064516129e-05, "loss": 0.6874, "step": 519 }, { "epoch": 3.351004016064257, "grad_norm": 18.62766456604004, "learning_rate": 4.8322580645161295e-05, "loss": 0.5358, "step": 520 }, { "epoch": 3.357429718875502, "grad_norm": 8.973854064941406, "learning_rate": 4.831935483870968e-05, "loss": 0.3919, "step": 521 }, { "epoch": 3.363855421686747, "grad_norm": 9.449162483215332, "learning_rate": 4.8316129032258066e-05, "loss": 0.3885, "step": 522 }, { "epoch": 3.3702811244979918, "grad_norm": 7.867352485656738, "learning_rate": 4.831290322580645e-05, "loss": 0.2964, "step": 523 }, { "epoch": 3.376706827309237, "grad_norm": 3.784162998199463, "learning_rate": 4.8309677419354845e-05, "loss": 0.4152, "step": 524 }, { "epoch": 3.383132530120482, "grad_norm": 4.036539077758789, "learning_rate": 4.830645161290323e-05, "loss": 0.6874, "step": 525 }, { "epoch": 3.389558232931727, "grad_norm": 3.986520767211914, "learning_rate": 4.8303225806451616e-05, "loss": 0.4374, "step": 526 }, { "epoch": 3.3959839357429717, "grad_norm": 17.574466705322266, "learning_rate": 4.83e-05, "loss": 0.2951, "step": 527 }, { "epoch": 3.402409638554217, "grad_norm": 5.089045524597168, "learning_rate": 4.829677419354839e-05, "loss": 0.504, "step": 528 }, { "epoch": 3.408835341365462, "grad_norm": 19.12257194519043, "learning_rate": 4.829354838709677e-05, "loss": 0.7643, "step": 529 }, { "epoch": 3.415261044176707, "grad_norm": 15.910208702087402, "learning_rate": 4.8290322580645166e-05, "loss": 0.5684, "step": 530 }, { "epoch": 3.4216867469879517, "grad_norm": 15.252849578857422, "learning_rate": 4.828709677419355e-05, "loss": 0.4638, "step": 531 }, { "epoch": 3.4281124497991966, "grad_norm": 13.639739036560059, "learning_rate": 4.828387096774194e-05, "loss": 0.293, "step": 532 }, { "epoch": 3.434538152610442, "grad_norm": 4.984840393066406, "learning_rate": 4.828064516129033e-05, "loss": 0.4896, "step": 533 }, { "epoch": 3.440963855421687, "grad_norm": 23.236557006835938, "learning_rate": 4.827741935483871e-05, "loss": 0.5465, "step": 534 }, { "epoch": 3.4473895582329317, "grad_norm": 9.729069709777832, "learning_rate": 4.82741935483871e-05, "loss": 0.3789, "step": 535 }, { "epoch": 3.4538152610441766, "grad_norm": 4.266533374786377, "learning_rate": 4.827096774193549e-05, "loss": 0.566, "step": 536 }, { "epoch": 3.460240963855422, "grad_norm": 17.500843048095703, "learning_rate": 4.826774193548387e-05, "loss": 0.7504, "step": 537 }, { "epoch": 3.466666666666667, "grad_norm": 4.834299087524414, "learning_rate": 4.826451612903226e-05, "loss": 0.2663, "step": 538 }, { "epoch": 3.4730923694779117, "grad_norm": 10.180542945861816, "learning_rate": 4.826129032258065e-05, "loss": 0.3957, "step": 539 }, { "epoch": 3.4795180722891565, "grad_norm": 4.1700544357299805, "learning_rate": 4.8258064516129036e-05, "loss": 0.5526, "step": 540 }, { "epoch": 3.4859437751004014, "grad_norm": 3.8274662494659424, "learning_rate": 4.825483870967742e-05, "loss": 0.686, "step": 541 }, { "epoch": 3.4923694779116468, "grad_norm": 4.170368671417236, "learning_rate": 4.825161290322581e-05, "loss": 0.2843, "step": 542 }, { "epoch": 3.4987951807228916, "grad_norm": 11.972061157226562, "learning_rate": 4.824838709677419e-05, "loss": 0.3365, "step": 543 }, { "epoch": 3.5052208835341365, "grad_norm": 14.85086441040039, "learning_rate": 4.8245161290322586e-05, "loss": 0.4143, "step": 544 }, { "epoch": 3.5116465863453814, "grad_norm": 19.364662170410156, "learning_rate": 4.824193548387097e-05, "loss": 0.4053, "step": 545 }, { "epoch": 3.5180722891566267, "grad_norm": 4.576084136962891, "learning_rate": 4.823870967741936e-05, "loss": 0.2501, "step": 546 }, { "epoch": 3.5244979919678716, "grad_norm": 6.699336528778076, "learning_rate": 4.823548387096774e-05, "loss": 0.4602, "step": 547 }, { "epoch": 3.5309236947791165, "grad_norm": 4.937297344207764, "learning_rate": 4.8232258064516135e-05, "loss": 0.4274, "step": 548 }, { "epoch": 3.5373493975903614, "grad_norm": 6.980901718139648, "learning_rate": 4.8229032258064514e-05, "loss": 0.4047, "step": 549 }, { "epoch": 3.5437751004016063, "grad_norm": 10.80716323852539, "learning_rate": 4.822580645161291e-05, "loss": 0.6237, "step": 550 }, { "epoch": 3.550200803212851, "grad_norm": 10.880539894104004, "learning_rate": 4.822258064516129e-05, "loss": 0.7267, "step": 551 }, { "epoch": 3.5566265060240965, "grad_norm": 12.19912338256836, "learning_rate": 4.821935483870968e-05, "loss": 0.2744, "step": 552 }, { "epoch": 3.5630522088353414, "grad_norm": 13.106900215148926, "learning_rate": 4.8216129032258064e-05, "loss": 0.3996, "step": 553 }, { "epoch": 3.5694779116465862, "grad_norm": 12.939781188964844, "learning_rate": 4.8212903225806456e-05, "loss": 0.7133, "step": 554 }, { "epoch": 3.5759036144578316, "grad_norm": 29.936439514160156, "learning_rate": 4.820967741935484e-05, "loss": 0.5625, "step": 555 }, { "epoch": 3.5823293172690764, "grad_norm": 13.32860279083252, "learning_rate": 4.820645161290323e-05, "loss": 0.66, "step": 556 }, { "epoch": 3.5887550200803213, "grad_norm": 12.884637832641602, "learning_rate": 4.8203225806451614e-05, "loss": 0.3715, "step": 557 }, { "epoch": 3.595180722891566, "grad_norm": 9.013917922973633, "learning_rate": 4.82e-05, "loss": 0.5217, "step": 558 }, { "epoch": 3.601606425702811, "grad_norm": 15.11435604095459, "learning_rate": 4.819677419354839e-05, "loss": 0.3078, "step": 559 }, { "epoch": 3.608032128514056, "grad_norm": 6.433659076690674, "learning_rate": 4.819354838709678e-05, "loss": 0.6698, "step": 560 }, { "epoch": 3.6144578313253013, "grad_norm": 5.484065055847168, "learning_rate": 4.819032258064516e-05, "loss": 0.634, "step": 561 }, { "epoch": 3.620883534136546, "grad_norm": 8.794512748718262, "learning_rate": 4.818709677419355e-05, "loss": 0.317, "step": 562 }, { "epoch": 3.627309236947791, "grad_norm": 16.98649024963379, "learning_rate": 4.818387096774194e-05, "loss": 0.4517, "step": 563 }, { "epoch": 3.6337349397590364, "grad_norm": 11.016792297363281, "learning_rate": 4.818064516129033e-05, "loss": 0.566, "step": 564 }, { "epoch": 3.6401606425702813, "grad_norm": 7.287659645080566, "learning_rate": 4.817741935483871e-05, "loss": 0.194, "step": 565 }, { "epoch": 3.646586345381526, "grad_norm": 9.350982666015625, "learning_rate": 4.81741935483871e-05, "loss": 0.4516, "step": 566 }, { "epoch": 3.653012048192771, "grad_norm": 9.69450855255127, "learning_rate": 4.8170967741935484e-05, "loss": 0.2974, "step": 567 }, { "epoch": 3.659437751004016, "grad_norm": 2.2422008514404297, "learning_rate": 4.816774193548387e-05, "loss": 0.3761, "step": 568 }, { "epoch": 3.665863453815261, "grad_norm": 10.900657653808594, "learning_rate": 4.816451612903226e-05, "loss": 0.3684, "step": 569 }, { "epoch": 3.672289156626506, "grad_norm": 13.074374198913574, "learning_rate": 4.816129032258065e-05, "loss": 0.4989, "step": 570 }, { "epoch": 3.678714859437751, "grad_norm": 5.309755325317383, "learning_rate": 4.8158064516129034e-05, "loss": 0.3954, "step": 571 }, { "epoch": 3.685140562248996, "grad_norm": 7.151030540466309, "learning_rate": 4.815483870967742e-05, "loss": 0.5132, "step": 572 }, { "epoch": 3.691566265060241, "grad_norm": 5.496908664703369, "learning_rate": 4.8151612903225805e-05, "loss": 0.5134, "step": 573 }, { "epoch": 3.697991967871486, "grad_norm": 18.125402450561523, "learning_rate": 4.81483870967742e-05, "loss": 0.5301, "step": 574 }, { "epoch": 3.704417670682731, "grad_norm": 5.772481918334961, "learning_rate": 4.8145161290322583e-05, "loss": 0.3766, "step": 575 }, { "epoch": 3.710843373493976, "grad_norm": 7.02168083190918, "learning_rate": 4.814193548387097e-05, "loss": 0.6194, "step": 576 }, { "epoch": 3.7172690763052207, "grad_norm": 3.694047212600708, "learning_rate": 4.8138709677419355e-05, "loss": 0.3564, "step": 577 }, { "epoch": 3.7236947791164656, "grad_norm": 6.89149284362793, "learning_rate": 4.813548387096775e-05, "loss": 0.4786, "step": 578 }, { "epoch": 3.730120481927711, "grad_norm": 16.726665496826172, "learning_rate": 4.813225806451613e-05, "loss": 0.4161, "step": 579 }, { "epoch": 3.736546184738956, "grad_norm": 4.311234951019287, "learning_rate": 4.812903225806452e-05, "loss": 0.2054, "step": 580 }, { "epoch": 3.7429718875502007, "grad_norm": 7.306130886077881, "learning_rate": 4.8125806451612904e-05, "loss": 0.3249, "step": 581 }, { "epoch": 3.749397590361446, "grad_norm": 6.864406108856201, "learning_rate": 4.812258064516129e-05, "loss": 0.4286, "step": 582 }, { "epoch": 3.755823293172691, "grad_norm": 23.583162307739258, "learning_rate": 4.811935483870968e-05, "loss": 0.6071, "step": 583 }, { "epoch": 3.762248995983936, "grad_norm": 6.634088516235352, "learning_rate": 4.811612903225807e-05, "loss": 0.7599, "step": 584 }, { "epoch": 3.7686746987951807, "grad_norm": 15.062341690063477, "learning_rate": 4.8112903225806454e-05, "loss": 0.2995, "step": 585 }, { "epoch": 3.7751004016064256, "grad_norm": 18.564910888671875, "learning_rate": 4.810967741935484e-05, "loss": 0.6092, "step": 586 }, { "epoch": 3.7815261044176705, "grad_norm": 11.457507133483887, "learning_rate": 4.810645161290323e-05, "loss": 0.5707, "step": 587 }, { "epoch": 3.787951807228916, "grad_norm": 9.531595230102539, "learning_rate": 4.810322580645161e-05, "loss": 0.2352, "step": 588 }, { "epoch": 3.7943775100401607, "grad_norm": 11.689484596252441, "learning_rate": 4.8100000000000004e-05, "loss": 0.6154, "step": 589 }, { "epoch": 3.8008032128514055, "grad_norm": 3.113664388656616, "learning_rate": 4.809677419354839e-05, "loss": 0.2315, "step": 590 }, { "epoch": 3.807228915662651, "grad_norm": 14.61591625213623, "learning_rate": 4.8093548387096775e-05, "loss": 0.6109, "step": 591 }, { "epoch": 3.8136546184738958, "grad_norm": 4.544252872467041, "learning_rate": 4.809032258064516e-05, "loss": 0.2938, "step": 592 }, { "epoch": 3.8200803212851406, "grad_norm": 9.976329803466797, "learning_rate": 4.808709677419355e-05, "loss": 0.2748, "step": 593 }, { "epoch": 3.8265060240963855, "grad_norm": 13.571074485778809, "learning_rate": 4.808387096774194e-05, "loss": 0.5422, "step": 594 }, { "epoch": 3.8329317269076304, "grad_norm": 6.591293811798096, "learning_rate": 4.8080645161290325e-05, "loss": 0.2598, "step": 595 }, { "epoch": 3.8393574297188753, "grad_norm": 6.24371337890625, "learning_rate": 4.807741935483871e-05, "loss": 0.6909, "step": 596 }, { "epoch": 3.8457831325301206, "grad_norm": 10.927971839904785, "learning_rate": 4.8074193548387096e-05, "loss": 0.2782, "step": 597 }, { "epoch": 3.8522088353413655, "grad_norm": 20.36640739440918, "learning_rate": 4.807096774193549e-05, "loss": 0.4348, "step": 598 }, { "epoch": 3.8586345381526104, "grad_norm": 1.4812424182891846, "learning_rate": 4.8067741935483874e-05, "loss": 0.2667, "step": 599 }, { "epoch": 3.8650602409638557, "grad_norm": 12.45275592803955, "learning_rate": 4.806451612903226e-05, "loss": 0.4224, "step": 600 }, { "epoch": 3.8714859437751006, "grad_norm": 8.374836921691895, "learning_rate": 4.8061290322580646e-05, "loss": 0.418, "step": 601 }, { "epoch": 3.8779116465863455, "grad_norm": 8.561665534973145, "learning_rate": 4.805806451612904e-05, "loss": 0.4993, "step": 602 }, { "epoch": 3.8843373493975903, "grad_norm": 13.309924125671387, "learning_rate": 4.8054838709677424e-05, "loss": 0.2904, "step": 603 }, { "epoch": 3.8907630522088352, "grad_norm": 5.6209588050842285, "learning_rate": 4.805161290322581e-05, "loss": 0.3614, "step": 604 }, { "epoch": 3.89718875502008, "grad_norm": 8.994816780090332, "learning_rate": 4.8048387096774195e-05, "loss": 0.5039, "step": 605 }, { "epoch": 3.9036144578313254, "grad_norm": 4.624064922332764, "learning_rate": 4.804516129032258e-05, "loss": 0.3827, "step": 606 }, { "epoch": 3.9100401606425703, "grad_norm": 8.400717735290527, "learning_rate": 4.8041935483870973e-05, "loss": 0.3891, "step": 607 }, { "epoch": 3.916465863453815, "grad_norm": 10.918342590332031, "learning_rate": 4.803870967741936e-05, "loss": 0.3926, "step": 608 }, { "epoch": 3.92289156626506, "grad_norm": 4.447176933288574, "learning_rate": 4.8035483870967745e-05, "loss": 0.2899, "step": 609 }, { "epoch": 3.9293172690763054, "grad_norm": 4.550628185272217, "learning_rate": 4.803225806451613e-05, "loss": 0.4613, "step": 610 }, { "epoch": 3.9357429718875503, "grad_norm": 10.448762893676758, "learning_rate": 4.8029032258064516e-05, "loss": 0.2903, "step": 611 }, { "epoch": 3.942168674698795, "grad_norm": 7.2309441566467285, "learning_rate": 4.80258064516129e-05, "loss": 0.5958, "step": 612 }, { "epoch": 3.94859437751004, "grad_norm": 4.779754638671875, "learning_rate": 4.8022580645161294e-05, "loss": 0.4608, "step": 613 }, { "epoch": 3.955020080321285, "grad_norm": 4.393620014190674, "learning_rate": 4.801935483870968e-05, "loss": 0.4658, "step": 614 }, { "epoch": 3.9614457831325303, "grad_norm": 8.011616706848145, "learning_rate": 4.8016129032258066e-05, "loss": 0.3069, "step": 615 }, { "epoch": 3.967871485943775, "grad_norm": 16.70987319946289, "learning_rate": 4.801290322580645e-05, "loss": 0.8174, "step": 616 }, { "epoch": 3.97429718875502, "grad_norm": 8.025530815124512, "learning_rate": 4.8009677419354844e-05, "loss": 0.4567, "step": 617 }, { "epoch": 3.980722891566265, "grad_norm": 11.639288902282715, "learning_rate": 4.800645161290323e-05, "loss": 0.3994, "step": 618 }, { "epoch": 3.9871485943775102, "grad_norm": 11.577380180358887, "learning_rate": 4.8003225806451616e-05, "loss": 0.5536, "step": 619 }, { "epoch": 3.993574297188755, "grad_norm": 13.689465522766113, "learning_rate": 4.8e-05, "loss": 0.6288, "step": 620 }, { "epoch": 4.003212851405623, "grad_norm": 3.870349407196045, "learning_rate": 4.799677419354839e-05, "loss": 0.7633, "step": 621 }, { "epoch": 4.009638554216868, "grad_norm": 11.144816398620605, "learning_rate": 4.799354838709678e-05, "loss": 0.5398, "step": 622 }, { "epoch": 4.016064257028113, "grad_norm": 6.496993541717529, "learning_rate": 4.7990322580645165e-05, "loss": 0.3577, "step": 623 }, { "epoch": 4.0224899598393575, "grad_norm": 5.640164852142334, "learning_rate": 4.798709677419355e-05, "loss": 0.4761, "step": 624 }, { "epoch": 4.028915662650602, "grad_norm": 12.483172416687012, "learning_rate": 4.7983870967741937e-05, "loss": 0.5299, "step": 625 }, { "epoch": 4.035341365461847, "grad_norm": 2.7476346492767334, "learning_rate": 4.798064516129033e-05, "loss": 0.2172, "step": 626 }, { "epoch": 4.041767068273092, "grad_norm": 3.175377130508423, "learning_rate": 4.797741935483871e-05, "loss": 0.4358, "step": 627 }, { "epoch": 4.048192771084337, "grad_norm": 32.252227783203125, "learning_rate": 4.79741935483871e-05, "loss": 0.4108, "step": 628 }, { "epoch": 4.054618473895582, "grad_norm": 5.927119731903076, "learning_rate": 4.7970967741935486e-05, "loss": 0.8238, "step": 629 }, { "epoch": 4.061044176706828, "grad_norm": 42.43552017211914, "learning_rate": 4.796774193548387e-05, "loss": 0.4417, "step": 630 }, { "epoch": 4.067469879518073, "grad_norm": 9.666000366210938, "learning_rate": 4.796451612903226e-05, "loss": 0.4248, "step": 631 }, { "epoch": 4.0738955823293175, "grad_norm": 14.674446105957031, "learning_rate": 4.796129032258065e-05, "loss": 0.4072, "step": 632 }, { "epoch": 4.080321285140562, "grad_norm": 17.21763038635254, "learning_rate": 4.7958064516129036e-05, "loss": 0.6733, "step": 633 }, { "epoch": 4.086746987951807, "grad_norm": 5.074434280395508, "learning_rate": 4.795483870967742e-05, "loss": 0.4566, "step": 634 }, { "epoch": 4.093172690763052, "grad_norm": 10.715412139892578, "learning_rate": 4.795161290322581e-05, "loss": 0.7678, "step": 635 }, { "epoch": 4.099598393574297, "grad_norm": 5.506896495819092, "learning_rate": 4.794838709677419e-05, "loss": 0.3437, "step": 636 }, { "epoch": 4.106024096385542, "grad_norm": 8.938709259033203, "learning_rate": 4.7945161290322585e-05, "loss": 0.384, "step": 637 }, { "epoch": 4.112449799196787, "grad_norm": 8.512730598449707, "learning_rate": 4.794193548387097e-05, "loss": 0.512, "step": 638 }, { "epoch": 4.1188755020080325, "grad_norm": 8.629199028015137, "learning_rate": 4.793870967741936e-05, "loss": 0.4105, "step": 639 }, { "epoch": 4.125301204819277, "grad_norm": 17.819759368896484, "learning_rate": 4.793548387096774e-05, "loss": 0.5033, "step": 640 }, { "epoch": 4.131726907630522, "grad_norm": 55.150352478027344, "learning_rate": 4.7932258064516135e-05, "loss": 0.7097, "step": 641 }, { "epoch": 4.138152610441767, "grad_norm": 8.826456069946289, "learning_rate": 4.7929032258064514e-05, "loss": 0.4927, "step": 642 }, { "epoch": 4.144578313253012, "grad_norm": 10.312532424926758, "learning_rate": 4.7925806451612906e-05, "loss": 0.6528, "step": 643 }, { "epoch": 4.151004016064257, "grad_norm": 16.01117515563965, "learning_rate": 4.792258064516129e-05, "loss": 0.45, "step": 644 }, { "epoch": 4.157429718875502, "grad_norm": 26.726533889770508, "learning_rate": 4.791935483870968e-05, "loss": 0.7813, "step": 645 }, { "epoch": 4.163855421686747, "grad_norm": 10.589249610900879, "learning_rate": 4.791612903225807e-05, "loss": 0.7028, "step": 646 }, { "epoch": 4.170281124497992, "grad_norm": 2.8673229217529297, "learning_rate": 4.7912903225806456e-05, "loss": 0.4221, "step": 647 }, { "epoch": 4.176706827309237, "grad_norm": 8.271647453308105, "learning_rate": 4.790967741935484e-05, "loss": 0.7253, "step": 648 }, { "epoch": 4.183132530120482, "grad_norm": 3.5805323123931885, "learning_rate": 4.790645161290323e-05, "loss": 0.3315, "step": 649 }, { "epoch": 4.189558232931727, "grad_norm": 15.175993919372559, "learning_rate": 4.790322580645161e-05, "loss": 0.3119, "step": 650 }, { "epoch": 4.195983935742972, "grad_norm": 7.317105293273926, "learning_rate": 4.79e-05, "loss": 0.3723, "step": 651 }, { "epoch": 4.202409638554217, "grad_norm": 7.007072925567627, "learning_rate": 4.789677419354839e-05, "loss": 0.4421, "step": 652 }, { "epoch": 4.208835341365462, "grad_norm": 4.628463268280029, "learning_rate": 4.789354838709678e-05, "loss": 0.5041, "step": 653 }, { "epoch": 4.215261044176707, "grad_norm": 6.247984409332275, "learning_rate": 4.789032258064516e-05, "loss": 0.3323, "step": 654 }, { "epoch": 4.2216867469879515, "grad_norm": 5.438474178314209, "learning_rate": 4.788709677419355e-05, "loss": 0.3555, "step": 655 }, { "epoch": 4.228112449799196, "grad_norm": 11.056044578552246, "learning_rate": 4.788387096774194e-05, "loss": 0.6031, "step": 656 }, { "epoch": 4.234538152610442, "grad_norm": 7.9885358810424805, "learning_rate": 4.7880645161290327e-05, "loss": 0.3932, "step": 657 }, { "epoch": 4.240963855421687, "grad_norm": 4.98292875289917, "learning_rate": 4.787741935483871e-05, "loss": 0.4875, "step": 658 }, { "epoch": 4.247389558232932, "grad_norm": 7.822724342346191, "learning_rate": 4.78741935483871e-05, "loss": 0.3215, "step": 659 }, { "epoch": 4.253815261044177, "grad_norm": 7.334625244140625, "learning_rate": 4.7870967741935484e-05, "loss": 0.3466, "step": 660 }, { "epoch": 4.260240963855422, "grad_norm": 4.715963363647461, "learning_rate": 4.7867741935483876e-05, "loss": 0.742, "step": 661 }, { "epoch": 4.266666666666667, "grad_norm": 21.673288345336914, "learning_rate": 4.786451612903226e-05, "loss": 0.4672, "step": 662 }, { "epoch": 4.2730923694779115, "grad_norm": 21.027612686157227, "learning_rate": 4.786129032258065e-05, "loss": 0.6224, "step": 663 }, { "epoch": 4.279518072289156, "grad_norm": 16.69331169128418, "learning_rate": 4.785806451612903e-05, "loss": 0.4211, "step": 664 }, { "epoch": 4.285943775100401, "grad_norm": 10.119441986083984, "learning_rate": 4.7854838709677426e-05, "loss": 0.5405, "step": 665 }, { "epoch": 4.292369477911647, "grad_norm": 6.5939555168151855, "learning_rate": 4.7851612903225805e-05, "loss": 0.3346, "step": 666 }, { "epoch": 4.298795180722892, "grad_norm": 10.222407341003418, "learning_rate": 4.78483870967742e-05, "loss": 0.4943, "step": 667 }, { "epoch": 4.305220883534137, "grad_norm": 9.23044490814209, "learning_rate": 4.784516129032258e-05, "loss": 0.6729, "step": 668 }, { "epoch": 4.311646586345382, "grad_norm": 11.06701946258545, "learning_rate": 4.784193548387097e-05, "loss": 0.5674, "step": 669 }, { "epoch": 4.3180722891566266, "grad_norm": 10.504576683044434, "learning_rate": 4.783870967741936e-05, "loss": 0.7073, "step": 670 }, { "epoch": 4.324497991967871, "grad_norm": 4.876540660858154, "learning_rate": 4.783548387096775e-05, "loss": 0.4504, "step": 671 }, { "epoch": 4.330923694779116, "grad_norm": 11.053621292114258, "learning_rate": 4.783225806451613e-05, "loss": 0.5028, "step": 672 }, { "epoch": 4.337349397590361, "grad_norm": 3.075951337814331, "learning_rate": 4.782903225806452e-05, "loss": 0.269, "step": 673 }, { "epoch": 4.343775100401606, "grad_norm": 9.535651206970215, "learning_rate": 4.7825806451612904e-05, "loss": 0.5343, "step": 674 }, { "epoch": 4.350200803212852, "grad_norm": 10.3850736618042, "learning_rate": 4.782258064516129e-05, "loss": 0.645, "step": 675 }, { "epoch": 4.356626506024097, "grad_norm": 4.773569107055664, "learning_rate": 4.781935483870968e-05, "loss": 0.2747, "step": 676 }, { "epoch": 4.363052208835342, "grad_norm": 4.665687561035156, "learning_rate": 4.781612903225807e-05, "loss": 0.5409, "step": 677 }, { "epoch": 4.3694779116465865, "grad_norm": 4.157199382781982, "learning_rate": 4.7812903225806454e-05, "loss": 0.3473, "step": 678 }, { "epoch": 4.375903614457831, "grad_norm": 6.212765216827393, "learning_rate": 4.780967741935484e-05, "loss": 0.4109, "step": 679 }, { "epoch": 4.382329317269076, "grad_norm": 7.177492618560791, "learning_rate": 4.780645161290323e-05, "loss": 0.6135, "step": 680 }, { "epoch": 4.388755020080321, "grad_norm": 5.17397403717041, "learning_rate": 4.780322580645161e-05, "loss": 0.2962, "step": 681 }, { "epoch": 4.395180722891566, "grad_norm": 4.837353229522705, "learning_rate": 4.78e-05, "loss": 0.3838, "step": 682 }, { "epoch": 4.401606425702811, "grad_norm": 9.989047050476074, "learning_rate": 4.779677419354839e-05, "loss": 0.5119, "step": 683 }, { "epoch": 4.408032128514057, "grad_norm": 12.210887908935547, "learning_rate": 4.7793548387096775e-05, "loss": 0.5448, "step": 684 }, { "epoch": 4.414457831325302, "grad_norm": 7.123885154724121, "learning_rate": 4.779032258064517e-05, "loss": 0.5687, "step": 685 }, { "epoch": 4.4208835341365464, "grad_norm": 7.7840728759765625, "learning_rate": 4.778709677419355e-05, "loss": 0.4466, "step": 686 }, { "epoch": 4.427309236947791, "grad_norm": 3.956212043762207, "learning_rate": 4.778387096774194e-05, "loss": 0.2821, "step": 687 }, { "epoch": 4.433734939759036, "grad_norm": 3.782926082611084, "learning_rate": 4.7780645161290324e-05, "loss": 0.179, "step": 688 }, { "epoch": 4.440160642570281, "grad_norm": 4.9404425621032715, "learning_rate": 4.777741935483871e-05, "loss": 0.7819, "step": 689 }, { "epoch": 4.446586345381526, "grad_norm": 12.790063858032227, "learning_rate": 4.7774193548387096e-05, "loss": 0.3568, "step": 690 }, { "epoch": 4.453012048192771, "grad_norm": 8.637455940246582, "learning_rate": 4.777096774193549e-05, "loss": 0.5632, "step": 691 }, { "epoch": 4.459437751004016, "grad_norm": 3.5888426303863525, "learning_rate": 4.7767741935483874e-05, "loss": 0.2952, "step": 692 }, { "epoch": 4.4658634538152615, "grad_norm": 6.2133469581604, "learning_rate": 4.776451612903226e-05, "loss": 0.6765, "step": 693 }, { "epoch": 4.472289156626506, "grad_norm": 3.9659972190856934, "learning_rate": 4.7761290322580645e-05, "loss": 0.6321, "step": 694 }, { "epoch": 4.478714859437751, "grad_norm": 10.719700813293457, "learning_rate": 4.775806451612904e-05, "loss": 0.449, "step": 695 }, { "epoch": 4.485140562248996, "grad_norm": 6.2206573486328125, "learning_rate": 4.775483870967742e-05, "loss": 0.4294, "step": 696 }, { "epoch": 4.491566265060241, "grad_norm": 4.782667636871338, "learning_rate": 4.775161290322581e-05, "loss": 0.3822, "step": 697 }, { "epoch": 4.497991967871486, "grad_norm": 5.623037815093994, "learning_rate": 4.7748387096774195e-05, "loss": 0.7417, "step": 698 }, { "epoch": 4.504417670682731, "grad_norm": 5.29359769821167, "learning_rate": 4.774516129032258e-05, "loss": 0.4268, "step": 699 }, { "epoch": 4.510843373493976, "grad_norm": 4.258016586303711, "learning_rate": 4.774193548387097e-05, "loss": 0.4322, "step": 700 }, { "epoch": 4.517269076305221, "grad_norm": 9.697691917419434, "learning_rate": 4.773870967741936e-05, "loss": 0.607, "step": 701 }, { "epoch": 4.523694779116466, "grad_norm": 4.906437873840332, "learning_rate": 4.7735483870967744e-05, "loss": 0.4585, "step": 702 }, { "epoch": 4.530120481927711, "grad_norm": 8.29771614074707, "learning_rate": 4.773225806451613e-05, "loss": 0.5586, "step": 703 }, { "epoch": 4.536546184738956, "grad_norm": 10.474533081054688, "learning_rate": 4.772903225806452e-05, "loss": 0.6721, "step": 704 }, { "epoch": 4.542971887550201, "grad_norm": 5.467524528503418, "learning_rate": 4.77258064516129e-05, "loss": 0.9071, "step": 705 }, { "epoch": 4.549397590361446, "grad_norm": 10.158985137939453, "learning_rate": 4.7722580645161294e-05, "loss": 0.3564, "step": 706 }, { "epoch": 4.555823293172691, "grad_norm": 23.390087127685547, "learning_rate": 4.771935483870968e-05, "loss": 0.6102, "step": 707 }, { "epoch": 4.562248995983936, "grad_norm": 23.29156494140625, "learning_rate": 4.7716129032258065e-05, "loss": 0.4093, "step": 708 }, { "epoch": 4.5686746987951805, "grad_norm": 9.0564546585083, "learning_rate": 4.771290322580646e-05, "loss": 0.441, "step": 709 }, { "epoch": 4.575100401606425, "grad_norm": 7.1021952629089355, "learning_rate": 4.7709677419354844e-05, "loss": 0.5683, "step": 710 }, { "epoch": 4.581526104417671, "grad_norm": 4.7914509773254395, "learning_rate": 4.770645161290323e-05, "loss": 0.6594, "step": 711 }, { "epoch": 4.587951807228916, "grad_norm": 11.284749984741211, "learning_rate": 4.7703225806451615e-05, "loss": 0.8638, "step": 712 }, { "epoch": 4.594377510040161, "grad_norm": 3.5816609859466553, "learning_rate": 4.77e-05, "loss": 0.4538, "step": 713 }, { "epoch": 4.600803212851406, "grad_norm": 3.9365551471710205, "learning_rate": 4.7696774193548386e-05, "loss": 0.2228, "step": 714 }, { "epoch": 4.607228915662651, "grad_norm": 7.476469993591309, "learning_rate": 4.769354838709678e-05, "loss": 0.4893, "step": 715 }, { "epoch": 4.613654618473896, "grad_norm": 7.488080978393555, "learning_rate": 4.7690322580645165e-05, "loss": 0.2489, "step": 716 }, { "epoch": 4.6200803212851405, "grad_norm": 11.419556617736816, "learning_rate": 4.768709677419355e-05, "loss": 0.4121, "step": 717 }, { "epoch": 4.626506024096385, "grad_norm": 10.621696472167969, "learning_rate": 4.7683870967741936e-05, "loss": 0.4923, "step": 718 }, { "epoch": 4.63293172690763, "grad_norm": 9.7233247756958, "learning_rate": 4.768064516129033e-05, "loss": 0.4922, "step": 719 }, { "epoch": 4.639357429718876, "grad_norm": 4.0977067947387695, "learning_rate": 4.767741935483871e-05, "loss": 0.274, "step": 720 }, { "epoch": 4.64578313253012, "grad_norm": 14.30873966217041, "learning_rate": 4.76741935483871e-05, "loss": 0.5705, "step": 721 }, { "epoch": 4.652208835341366, "grad_norm": 7.426396369934082, "learning_rate": 4.7670967741935486e-05, "loss": 0.4691, "step": 722 }, { "epoch": 4.658634538152611, "grad_norm": 12.113080024719238, "learning_rate": 4.766774193548387e-05, "loss": 0.3075, "step": 723 }, { "epoch": 4.6650602409638555, "grad_norm": 11.997623443603516, "learning_rate": 4.7664516129032264e-05, "loss": 0.4329, "step": 724 }, { "epoch": 4.6714859437751, "grad_norm": 6.530179500579834, "learning_rate": 4.766129032258065e-05, "loss": 0.4332, "step": 725 }, { "epoch": 4.677911646586345, "grad_norm": 4.42352294921875, "learning_rate": 4.7658064516129035e-05, "loss": 0.3209, "step": 726 }, { "epoch": 4.68433734939759, "grad_norm": 7.649783611297607, "learning_rate": 4.765483870967742e-05, "loss": 0.4141, "step": 727 }, { "epoch": 4.690763052208835, "grad_norm": 4.199863433837891, "learning_rate": 4.765161290322581e-05, "loss": 0.4137, "step": 728 }, { "epoch": 4.697188755020081, "grad_norm": 8.59699535369873, "learning_rate": 4.764838709677419e-05, "loss": 0.4354, "step": 729 }, { "epoch": 4.703614457831325, "grad_norm": 12.023775100708008, "learning_rate": 4.7645161290322585e-05, "loss": 0.5124, "step": 730 }, { "epoch": 4.710040160642571, "grad_norm": 7.023086071014404, "learning_rate": 4.764193548387097e-05, "loss": 0.4365, "step": 731 }, { "epoch": 4.7164658634538155, "grad_norm": 3.879354476928711, "learning_rate": 4.7638709677419356e-05, "loss": 0.4275, "step": 732 }, { "epoch": 4.72289156626506, "grad_norm": 6.329866409301758, "learning_rate": 4.763548387096775e-05, "loss": 0.4622, "step": 733 }, { "epoch": 4.729317269076305, "grad_norm": 8.134129524230957, "learning_rate": 4.7632258064516134e-05, "loss": 0.407, "step": 734 }, { "epoch": 4.73574297188755, "grad_norm": 4.258336067199707, "learning_rate": 4.762903225806452e-05, "loss": 0.2631, "step": 735 }, { "epoch": 4.742168674698795, "grad_norm": 5.381927490234375, "learning_rate": 4.7625806451612906e-05, "loss": 0.3874, "step": 736 }, { "epoch": 4.74859437751004, "grad_norm": 9.265498161315918, "learning_rate": 4.762258064516129e-05, "loss": 0.4652, "step": 737 }, { "epoch": 4.755020080321285, "grad_norm": 6.923249244689941, "learning_rate": 4.761935483870968e-05, "loss": 0.6259, "step": 738 }, { "epoch": 4.76144578313253, "grad_norm": 4.511381149291992, "learning_rate": 4.761612903225807e-05, "loss": 0.448, "step": 739 }, { "epoch": 4.767871485943775, "grad_norm": 5.585071086883545, "learning_rate": 4.7612903225806455e-05, "loss": 0.4907, "step": 740 }, { "epoch": 4.77429718875502, "grad_norm": 6.669870853424072, "learning_rate": 4.760967741935484e-05, "loss": 0.3809, "step": 741 }, { "epoch": 4.780722891566265, "grad_norm": 6.19124174118042, "learning_rate": 4.760645161290323e-05, "loss": 0.3477, "step": 742 }, { "epoch": 4.78714859437751, "grad_norm": 7.211981296539307, "learning_rate": 4.760322580645162e-05, "loss": 0.6321, "step": 743 }, { "epoch": 4.793574297188755, "grad_norm": 14.683920860290527, "learning_rate": 4.76e-05, "loss": 0.6182, "step": 744 }, { "epoch": 4.8, "grad_norm": 5.158877849578857, "learning_rate": 4.759677419354839e-05, "loss": 0.6266, "step": 745 }, { "epoch": 4.806425702811245, "grad_norm": 8.119921684265137, "learning_rate": 4.7593548387096776e-05, "loss": 0.4327, "step": 746 }, { "epoch": 4.81285140562249, "grad_norm": 8.1658296585083, "learning_rate": 4.759032258064516e-05, "loss": 0.5043, "step": 747 }, { "epoch": 4.8192771084337345, "grad_norm": 13.220368385314941, "learning_rate": 4.7587096774193555e-05, "loss": 0.6635, "step": 748 }, { "epoch": 4.82570281124498, "grad_norm": 5.631162166595459, "learning_rate": 4.758387096774194e-05, "loss": 0.5244, "step": 749 }, { "epoch": 4.832128514056225, "grad_norm": 2.4847538471221924, "learning_rate": 4.7580645161290326e-05, "loss": 0.2747, "step": 750 }, { "epoch": 4.83855421686747, "grad_norm": 7.292346954345703, "learning_rate": 4.757741935483871e-05, "loss": 0.5648, "step": 751 }, { "epoch": 4.844979919678715, "grad_norm": 5.091940402984619, "learning_rate": 4.75741935483871e-05, "loss": 0.4061, "step": 752 }, { "epoch": 4.85140562248996, "grad_norm": 10.187068939208984, "learning_rate": 4.757096774193548e-05, "loss": 0.5837, "step": 753 }, { "epoch": 4.857831325301205, "grad_norm": 6.984403610229492, "learning_rate": 4.7567741935483876e-05, "loss": 0.4804, "step": 754 }, { "epoch": 4.8642570281124495, "grad_norm": 10.445493698120117, "learning_rate": 4.756451612903226e-05, "loss": 0.5231, "step": 755 }, { "epoch": 4.870682730923694, "grad_norm": 8.145037651062012, "learning_rate": 4.756129032258065e-05, "loss": 0.4176, "step": 756 }, { "epoch": 4.877108433734939, "grad_norm": 2.6305932998657227, "learning_rate": 4.755806451612903e-05, "loss": 0.3706, "step": 757 }, { "epoch": 4.883534136546185, "grad_norm": 2.604595184326172, "learning_rate": 4.7554838709677425e-05, "loss": 0.4757, "step": 758 }, { "epoch": 4.88995983935743, "grad_norm": 9.240584373474121, "learning_rate": 4.7551612903225804e-05, "loss": 0.6, "step": 759 }, { "epoch": 4.896385542168675, "grad_norm": 15.203973770141602, "learning_rate": 4.75483870967742e-05, "loss": 0.7904, "step": 760 }, { "epoch": 4.90281124497992, "grad_norm": 43.883304595947266, "learning_rate": 4.754516129032258e-05, "loss": 0.3016, "step": 761 }, { "epoch": 4.909236947791165, "grad_norm": 12.352065086364746, "learning_rate": 4.754193548387097e-05, "loss": 0.4252, "step": 762 }, { "epoch": 4.9156626506024095, "grad_norm": 8.400856018066406, "learning_rate": 4.753870967741936e-05, "loss": 0.3543, "step": 763 }, { "epoch": 4.922088353413654, "grad_norm": 45.41724395751953, "learning_rate": 4.7535483870967746e-05, "loss": 0.329, "step": 764 }, { "epoch": 4.928514056224899, "grad_norm": 5.35670280456543, "learning_rate": 4.753225806451613e-05, "loss": 0.3277, "step": 765 }, { "epoch": 4.934939759036144, "grad_norm": 4.409581184387207, "learning_rate": 4.752903225806452e-05, "loss": 0.4708, "step": 766 }, { "epoch": 4.94136546184739, "grad_norm": 10.429356575012207, "learning_rate": 4.7525806451612903e-05, "loss": 0.4894, "step": 767 }, { "epoch": 4.947791164658635, "grad_norm": 11.448781967163086, "learning_rate": 4.752258064516129e-05, "loss": 0.3273, "step": 768 }, { "epoch": 4.95421686746988, "grad_norm": 3.5856409072875977, "learning_rate": 4.751935483870968e-05, "loss": 0.2957, "step": 769 }, { "epoch": 4.9606425702811245, "grad_norm": 9.229110717773438, "learning_rate": 4.751612903225807e-05, "loss": 0.5575, "step": 770 }, { "epoch": 4.967068273092369, "grad_norm": 9.371766090393066, "learning_rate": 4.751290322580645e-05, "loss": 0.3852, "step": 771 }, { "epoch": 4.973493975903614, "grad_norm": 24.980548858642578, "learning_rate": 4.7509677419354845e-05, "loss": 0.2573, "step": 772 }, { "epoch": 4.979919678714859, "grad_norm": 5.101824760437012, "learning_rate": 4.750645161290323e-05, "loss": 0.4696, "step": 773 }, { "epoch": 4.986345381526104, "grad_norm": 4.832958221435547, "learning_rate": 4.750322580645162e-05, "loss": 0.6944, "step": 774 }, { "epoch": 4.992771084337349, "grad_norm": 10.263834953308105, "learning_rate": 4.75e-05, "loss": 0.4637, "step": 775 }, { "epoch": 5.002409638554217, "grad_norm": 11.621912002563477, "learning_rate": 4.749677419354839e-05, "loss": 0.4759, "step": 776 }, { "epoch": 5.008835341365462, "grad_norm": 6.316365718841553, "learning_rate": 4.7493548387096774e-05, "loss": 0.3571, "step": 777 }, { "epoch": 5.0152610441767065, "grad_norm": 2.7929089069366455, "learning_rate": 4.7490322580645167e-05, "loss": 0.2484, "step": 778 }, { "epoch": 5.021686746987951, "grad_norm": 16.906497955322266, "learning_rate": 4.748709677419355e-05, "loss": 0.5243, "step": 779 }, { "epoch": 5.028112449799197, "grad_norm": 13.36916446685791, "learning_rate": 4.748387096774194e-05, "loss": 0.3595, "step": 780 }, { "epoch": 5.034538152610442, "grad_norm": 9.29102897644043, "learning_rate": 4.7480645161290324e-05, "loss": 0.6799, "step": 781 }, { "epoch": 5.040963855421687, "grad_norm": 3.1384546756744385, "learning_rate": 4.7477419354838716e-05, "loss": 0.2357, "step": 782 }, { "epoch": 5.047389558232932, "grad_norm": 14.120749473571777, "learning_rate": 4.7474193548387095e-05, "loss": 0.4666, "step": 783 }, { "epoch": 5.053815261044177, "grad_norm": 6.532683372497559, "learning_rate": 4.747096774193549e-05, "loss": 0.5875, "step": 784 }, { "epoch": 5.0602409638554215, "grad_norm": 10.505824089050293, "learning_rate": 4.746774193548387e-05, "loss": 0.6376, "step": 785 }, { "epoch": 5.066666666666666, "grad_norm": 5.0455780029296875, "learning_rate": 4.746451612903226e-05, "loss": 0.3848, "step": 786 }, { "epoch": 5.073092369477911, "grad_norm": 8.407227516174316, "learning_rate": 4.746129032258065e-05, "loss": 0.5461, "step": 787 }, { "epoch": 5.079518072289156, "grad_norm": 4.463189125061035, "learning_rate": 4.745806451612904e-05, "loss": 0.3865, "step": 788 }, { "epoch": 5.085943775100402, "grad_norm": 4.0331573486328125, "learning_rate": 4.745483870967742e-05, "loss": 0.3705, "step": 789 }, { "epoch": 5.092369477911647, "grad_norm": 13.018644332885742, "learning_rate": 4.745161290322581e-05, "loss": 0.4851, "step": 790 }, { "epoch": 5.098795180722892, "grad_norm": 5.112215518951416, "learning_rate": 4.7448387096774194e-05, "loss": 0.4592, "step": 791 }, { "epoch": 5.105220883534137, "grad_norm": 4.959601879119873, "learning_rate": 4.744516129032258e-05, "loss": 0.3137, "step": 792 }, { "epoch": 5.1116465863453815, "grad_norm": 6.828153610229492, "learning_rate": 4.744193548387097e-05, "loss": 0.7251, "step": 793 }, { "epoch": 5.118072289156626, "grad_norm": 9.985513687133789, "learning_rate": 4.743870967741936e-05, "loss": 0.5082, "step": 794 }, { "epoch": 5.124497991967871, "grad_norm": 3.5421009063720703, "learning_rate": 4.7435483870967744e-05, "loss": 0.5046, "step": 795 }, { "epoch": 5.130923694779116, "grad_norm": 5.231625556945801, "learning_rate": 4.7432258064516136e-05, "loss": 0.6277, "step": 796 }, { "epoch": 5.137349397590361, "grad_norm": 5.82951545715332, "learning_rate": 4.742903225806452e-05, "loss": 0.5744, "step": 797 }, { "epoch": 5.143775100401607, "grad_norm": 4.716634273529053, "learning_rate": 4.74258064516129e-05, "loss": 0.4407, "step": 798 }, { "epoch": 5.150200803212852, "grad_norm": 4.66391134262085, "learning_rate": 4.7422580645161293e-05, "loss": 0.4936, "step": 799 }, { "epoch": 5.156626506024097, "grad_norm": 9.070045471191406, "learning_rate": 4.741935483870968e-05, "loss": 0.4666, "step": 800 }, { "epoch": 5.163052208835341, "grad_norm": 6.133974552154541, "learning_rate": 4.7416129032258065e-05, "loss": 0.5727, "step": 801 }, { "epoch": 5.169477911646586, "grad_norm": 8.140664100646973, "learning_rate": 4.741290322580646e-05, "loss": 1.057, "step": 802 }, { "epoch": 5.175903614457831, "grad_norm": 5.54121732711792, "learning_rate": 4.740967741935484e-05, "loss": 0.3428, "step": 803 }, { "epoch": 5.182329317269076, "grad_norm": 8.94742202758789, "learning_rate": 4.740645161290323e-05, "loss": 0.2863, "step": 804 }, { "epoch": 5.188755020080321, "grad_norm": 13.67180347442627, "learning_rate": 4.7403225806451614e-05, "loss": 0.5751, "step": 805 }, { "epoch": 5.195180722891566, "grad_norm": 17.720603942871094, "learning_rate": 4.74e-05, "loss": 0.5635, "step": 806 }, { "epoch": 5.201606425702812, "grad_norm": 3.3942339420318604, "learning_rate": 4.7396774193548386e-05, "loss": 0.1354, "step": 807 }, { "epoch": 5.2080321285140565, "grad_norm": 11.61681079864502, "learning_rate": 4.739354838709678e-05, "loss": 0.3028, "step": 808 }, { "epoch": 5.214457831325301, "grad_norm": 6.402846813201904, "learning_rate": 4.7390322580645164e-05, "loss": 0.2716, "step": 809 }, { "epoch": 5.220883534136546, "grad_norm": 7.85732364654541, "learning_rate": 4.738709677419355e-05, "loss": 0.4797, "step": 810 }, { "epoch": 5.227309236947791, "grad_norm": 11.546478271484375, "learning_rate": 4.738387096774194e-05, "loss": 0.8253, "step": 811 }, { "epoch": 5.233734939759036, "grad_norm": 7.58867883682251, "learning_rate": 4.738064516129033e-05, "loss": 0.7888, "step": 812 }, { "epoch": 5.240160642570281, "grad_norm": 18.896442413330078, "learning_rate": 4.7377419354838714e-05, "loss": 0.5325, "step": 813 }, { "epoch": 5.246586345381526, "grad_norm": 8.332119941711426, "learning_rate": 4.73741935483871e-05, "loss": 0.3691, "step": 814 }, { "epoch": 5.253012048192771, "grad_norm": 19.79954719543457, "learning_rate": 4.7370967741935485e-05, "loss": 0.6935, "step": 815 }, { "epoch": 5.2594377510040164, "grad_norm": 6.768062114715576, "learning_rate": 4.736774193548387e-05, "loss": 0.5623, "step": 816 }, { "epoch": 5.265863453815261, "grad_norm": 4.124804973602295, "learning_rate": 4.736451612903226e-05, "loss": 0.428, "step": 817 }, { "epoch": 5.272289156626506, "grad_norm": 7.031583309173584, "learning_rate": 4.736129032258065e-05, "loss": 0.4287, "step": 818 }, { "epoch": 5.278714859437751, "grad_norm": 4.083448886871338, "learning_rate": 4.7358064516129035e-05, "loss": 0.2582, "step": 819 }, { "epoch": 5.285140562248996, "grad_norm": 7.178497314453125, "learning_rate": 4.735483870967742e-05, "loss": 0.5578, "step": 820 }, { "epoch": 5.291566265060241, "grad_norm": 6.743470668792725, "learning_rate": 4.7351612903225806e-05, "loss": 0.3384, "step": 821 }, { "epoch": 5.297991967871486, "grad_norm": 3.5494067668914795, "learning_rate": 4.734838709677419e-05, "loss": 0.3352, "step": 822 }, { "epoch": 5.304417670682731, "grad_norm": 7.412623882293701, "learning_rate": 4.7345161290322584e-05, "loss": 0.3143, "step": 823 }, { "epoch": 5.3108433734939755, "grad_norm": 5.802249431610107, "learning_rate": 4.734193548387097e-05, "loss": 0.4113, "step": 824 }, { "epoch": 5.317269076305221, "grad_norm": 11.241743087768555, "learning_rate": 4.7338709677419356e-05, "loss": 0.5484, "step": 825 }, { "epoch": 5.323694779116466, "grad_norm": 6.283975601196289, "learning_rate": 4.733548387096775e-05, "loss": 0.4677, "step": 826 }, { "epoch": 5.330120481927711, "grad_norm": 5.8127546310424805, "learning_rate": 4.7332258064516134e-05, "loss": 0.4141, "step": 827 }, { "epoch": 5.336546184738956, "grad_norm": 4.531114101409912, "learning_rate": 4.732903225806452e-05, "loss": 0.4054, "step": 828 }, { "epoch": 5.342971887550201, "grad_norm": 9.385262489318848, "learning_rate": 4.7325806451612905e-05, "loss": 0.6489, "step": 829 }, { "epoch": 5.349397590361446, "grad_norm": 2.8236758708953857, "learning_rate": 4.732258064516129e-05, "loss": 0.3448, "step": 830 }, { "epoch": 5.355823293172691, "grad_norm": 13.766401290893555, "learning_rate": 4.731935483870968e-05, "loss": 0.4869, "step": 831 }, { "epoch": 5.3622489959839355, "grad_norm": 29.310304641723633, "learning_rate": 4.731612903225807e-05, "loss": 0.5041, "step": 832 }, { "epoch": 5.36867469879518, "grad_norm": 3.9884181022644043, "learning_rate": 4.7312903225806455e-05, "loss": 0.5861, "step": 833 }, { "epoch": 5.375100401606426, "grad_norm": 5.222535610198975, "learning_rate": 4.730967741935484e-05, "loss": 0.3899, "step": 834 }, { "epoch": 5.381526104417671, "grad_norm": 9.090160369873047, "learning_rate": 4.730645161290323e-05, "loss": 0.4685, "step": 835 }, { "epoch": 5.387951807228916, "grad_norm": 4.162207126617432, "learning_rate": 4.730322580645162e-05, "loss": 0.1477, "step": 836 }, { "epoch": 5.394377510040161, "grad_norm": 18.31096649169922, "learning_rate": 4.73e-05, "loss": 0.6018, "step": 837 }, { "epoch": 5.400803212851406, "grad_norm": 8.401172637939453, "learning_rate": 4.729677419354839e-05, "loss": 0.4318, "step": 838 }, { "epoch": 5.4072289156626505, "grad_norm": 6.509830951690674, "learning_rate": 4.7293548387096776e-05, "loss": 0.5339, "step": 839 }, { "epoch": 5.413654618473895, "grad_norm": 8.435197830200195, "learning_rate": 4.729032258064516e-05, "loss": 0.2389, "step": 840 }, { "epoch": 5.42008032128514, "grad_norm": 3.0672147274017334, "learning_rate": 4.7287096774193554e-05, "loss": 0.2222, "step": 841 }, { "epoch": 5.426506024096385, "grad_norm": 6.204017639160156, "learning_rate": 4.728387096774194e-05, "loss": 0.3179, "step": 842 }, { "epoch": 5.432931726907631, "grad_norm": 8.441490173339844, "learning_rate": 4.7280645161290326e-05, "loss": 0.5466, "step": 843 }, { "epoch": 5.439357429718876, "grad_norm": 1.9725416898727417, "learning_rate": 4.727741935483871e-05, "loss": 0.3153, "step": 844 }, { "epoch": 5.445783132530121, "grad_norm": 7.0545172691345215, "learning_rate": 4.72741935483871e-05, "loss": 0.4651, "step": 845 }, { "epoch": 5.452208835341366, "grad_norm": 7.69012451171875, "learning_rate": 4.727096774193548e-05, "loss": 0.3852, "step": 846 }, { "epoch": 5.4586345381526105, "grad_norm": 2.6767897605895996, "learning_rate": 4.7267741935483875e-05, "loss": 0.1369, "step": 847 }, { "epoch": 5.465060240963855, "grad_norm": 4.420098304748535, "learning_rate": 4.726451612903226e-05, "loss": 0.2187, "step": 848 }, { "epoch": 5.4714859437751, "grad_norm": 5.156297206878662, "learning_rate": 4.7261290322580647e-05, "loss": 0.452, "step": 849 }, { "epoch": 5.477911646586345, "grad_norm": 13.7605619430542, "learning_rate": 4.725806451612904e-05, "loss": 0.3525, "step": 850 }, { "epoch": 5.48433734939759, "grad_norm": 6.319297790527344, "learning_rate": 4.7254838709677425e-05, "loss": 0.528, "step": 851 }, { "epoch": 5.490763052208836, "grad_norm": 7.62947416305542, "learning_rate": 4.725161290322581e-05, "loss": 0.2921, "step": 852 }, { "epoch": 5.497188755020081, "grad_norm": 3.092348098754883, "learning_rate": 4.7248387096774196e-05, "loss": 0.2052, "step": 853 }, { "epoch": 5.5036144578313255, "grad_norm": 18.200510025024414, "learning_rate": 4.724516129032258e-05, "loss": 0.4992, "step": 854 }, { "epoch": 5.51004016064257, "grad_norm": 17.004533767700195, "learning_rate": 4.724193548387097e-05, "loss": 0.4469, "step": 855 }, { "epoch": 5.516465863453815, "grad_norm": 8.701032638549805, "learning_rate": 4.723870967741936e-05, "loss": 0.4193, "step": 856 }, { "epoch": 5.52289156626506, "grad_norm": 2.50925874710083, "learning_rate": 4.7235483870967746e-05, "loss": 0.2105, "step": 857 }, { "epoch": 5.529317269076305, "grad_norm": 6.941011905670166, "learning_rate": 4.723225806451613e-05, "loss": 0.4001, "step": 858 }, { "epoch": 5.53574297188755, "grad_norm": 10.448933601379395, "learning_rate": 4.7229032258064524e-05, "loss": 0.377, "step": 859 }, { "epoch": 5.542168674698795, "grad_norm": 2.9185099601745605, "learning_rate": 4.72258064516129e-05, "loss": 0.2914, "step": 860 }, { "epoch": 5.548594377510041, "grad_norm": 4.849759101867676, "learning_rate": 4.722258064516129e-05, "loss": 0.3551, "step": 861 }, { "epoch": 5.5550200803212855, "grad_norm": 5.835464000701904, "learning_rate": 4.721935483870968e-05, "loss": 0.3684, "step": 862 }, { "epoch": 5.56144578313253, "grad_norm": 4.032254219055176, "learning_rate": 4.721612903225807e-05, "loss": 0.5637, "step": 863 }, { "epoch": 5.567871485943775, "grad_norm": 3.740051507949829, "learning_rate": 4.721290322580645e-05, "loss": 0.5545, "step": 864 }, { "epoch": 5.57429718875502, "grad_norm": 5.980421543121338, "learning_rate": 4.7209677419354845e-05, "loss": 0.7182, "step": 865 }, { "epoch": 5.580722891566265, "grad_norm": 2.8677709102630615, "learning_rate": 4.720645161290323e-05, "loss": 0.4873, "step": 866 }, { "epoch": 5.58714859437751, "grad_norm": 9.128663063049316, "learning_rate": 4.7203225806451616e-05, "loss": 0.8128, "step": 867 }, { "epoch": 5.593574297188755, "grad_norm": 2.0417189598083496, "learning_rate": 4.72e-05, "loss": 0.2309, "step": 868 }, { "epoch": 5.6, "grad_norm": 13.9517183303833, "learning_rate": 4.719677419354839e-05, "loss": 0.6353, "step": 869 }, { "epoch": 5.606425702811245, "grad_norm": 11.9827880859375, "learning_rate": 4.7193548387096774e-05, "loss": 0.3674, "step": 870 }, { "epoch": 5.61285140562249, "grad_norm": 3.312407970428467, "learning_rate": 4.7190322580645166e-05, "loss": 0.3027, "step": 871 }, { "epoch": 5.619277108433735, "grad_norm": 4.492595195770264, "learning_rate": 4.718709677419355e-05, "loss": 0.3966, "step": 872 }, { "epoch": 5.62570281124498, "grad_norm": 18.010316848754883, "learning_rate": 4.718387096774194e-05, "loss": 0.3498, "step": 873 }, { "epoch": 5.632128514056225, "grad_norm": 10.969120025634766, "learning_rate": 4.718064516129033e-05, "loss": 0.3353, "step": 874 }, { "epoch": 5.63855421686747, "grad_norm": 6.885847091674805, "learning_rate": 4.7177419354838716e-05, "loss": 0.4479, "step": 875 }, { "epoch": 5.644979919678715, "grad_norm": 20.689260482788086, "learning_rate": 4.7174193548387095e-05, "loss": 0.6653, "step": 876 }, { "epoch": 5.65140562248996, "grad_norm": 6.850219249725342, "learning_rate": 4.717096774193549e-05, "loss": 0.3944, "step": 877 }, { "epoch": 5.6578313253012045, "grad_norm": 9.957620620727539, "learning_rate": 4.716774193548387e-05, "loss": 0.5292, "step": 878 }, { "epoch": 5.66425702811245, "grad_norm": 15.594244003295898, "learning_rate": 4.716451612903226e-05, "loss": 0.3957, "step": 879 }, { "epoch": 5.670682730923695, "grad_norm": 10.753636360168457, "learning_rate": 4.716129032258065e-05, "loss": 0.3939, "step": 880 }, { "epoch": 5.67710843373494, "grad_norm": 4.753134250640869, "learning_rate": 4.715806451612904e-05, "loss": 0.2819, "step": 881 }, { "epoch": 5.683534136546185, "grad_norm": 10.871026039123535, "learning_rate": 4.715483870967742e-05, "loss": 0.5669, "step": 882 }, { "epoch": 5.68995983935743, "grad_norm": 5.971273899078369, "learning_rate": 4.715161290322581e-05, "loss": 0.5393, "step": 883 }, { "epoch": 5.696385542168675, "grad_norm": 7.386717796325684, "learning_rate": 4.7148387096774194e-05, "loss": 0.5029, "step": 884 }, { "epoch": 5.7028112449799195, "grad_norm": 6.962315559387207, "learning_rate": 4.714516129032258e-05, "loss": 0.7043, "step": 885 }, { "epoch": 5.709236947791164, "grad_norm": 6.089379787445068, "learning_rate": 4.714193548387097e-05, "loss": 0.4, "step": 886 }, { "epoch": 5.715662650602409, "grad_norm": 8.105810165405273, "learning_rate": 4.713870967741936e-05, "loss": 0.7043, "step": 887 }, { "epoch": 5.722088353413655, "grad_norm": 11.646936416625977, "learning_rate": 4.713548387096774e-05, "loss": 0.6368, "step": 888 }, { "epoch": 5.7285140562249, "grad_norm": 6.4409565925598145, "learning_rate": 4.7132258064516136e-05, "loss": 0.2672, "step": 889 }, { "epoch": 5.734939759036145, "grad_norm": 10.417530059814453, "learning_rate": 4.712903225806452e-05, "loss": 0.7379, "step": 890 }, { "epoch": 5.74136546184739, "grad_norm": 3.6160659790039062, "learning_rate": 4.71258064516129e-05, "loss": 0.3538, "step": 891 }, { "epoch": 5.747791164658635, "grad_norm": 6.4334821701049805, "learning_rate": 4.712258064516129e-05, "loss": 0.3327, "step": 892 }, { "epoch": 5.7542168674698795, "grad_norm": 6.179240703582764, "learning_rate": 4.711935483870968e-05, "loss": 0.7792, "step": 893 }, { "epoch": 5.760642570281124, "grad_norm": 11.68407917022705, "learning_rate": 4.7116129032258064e-05, "loss": 0.2851, "step": 894 }, { "epoch": 5.767068273092369, "grad_norm": 8.155572891235352, "learning_rate": 4.711290322580646e-05, "loss": 0.3703, "step": 895 }, { "epoch": 5.773493975903614, "grad_norm": 8.687248229980469, "learning_rate": 4.710967741935484e-05, "loss": 0.3977, "step": 896 }, { "epoch": 5.77991967871486, "grad_norm": 4.225980281829834, "learning_rate": 4.710645161290323e-05, "loss": 0.3192, "step": 897 }, { "epoch": 5.786345381526105, "grad_norm": 5.765499114990234, "learning_rate": 4.710322580645162e-05, "loss": 0.4615, "step": 898 }, { "epoch": 5.79277108433735, "grad_norm": 15.121464729309082, "learning_rate": 4.71e-05, "loss": 0.7794, "step": 899 }, { "epoch": 5.7991967871485945, "grad_norm": 6.661050796508789, "learning_rate": 4.7096774193548385e-05, "loss": 0.4517, "step": 900 }, { "epoch": 5.805622489959839, "grad_norm": 4.005463600158691, "learning_rate": 4.709354838709678e-05, "loss": 0.6086, "step": 901 }, { "epoch": 5.812048192771084, "grad_norm": 5.839357376098633, "learning_rate": 4.7090322580645164e-05, "loss": 0.4066, "step": 902 }, { "epoch": 5.818473895582329, "grad_norm": 5.256525039672852, "learning_rate": 4.708709677419355e-05, "loss": 0.5349, "step": 903 }, { "epoch": 5.824899598393574, "grad_norm": 6.33106803894043, "learning_rate": 4.708387096774194e-05, "loss": 0.7762, "step": 904 }, { "epoch": 5.831325301204819, "grad_norm": 10.9049711227417, "learning_rate": 4.708064516129033e-05, "loss": 0.4665, "step": 905 }, { "epoch": 5.837751004016065, "grad_norm": 2.8971431255340576, "learning_rate": 4.707741935483871e-05, "loss": 0.4216, "step": 906 }, { "epoch": 5.84417670682731, "grad_norm": 17.43858528137207, "learning_rate": 4.70741935483871e-05, "loss": 0.4204, "step": 907 }, { "epoch": 5.8506024096385545, "grad_norm": 1.9089080095291138, "learning_rate": 4.7070967741935485e-05, "loss": 0.281, "step": 908 }, { "epoch": 5.857028112449799, "grad_norm": 8.499836921691895, "learning_rate": 4.706774193548387e-05, "loss": 0.4826, "step": 909 }, { "epoch": 5.863453815261044, "grad_norm": 2.1953296661376953, "learning_rate": 4.706451612903226e-05, "loss": 0.1043, "step": 910 }, { "epoch": 5.869879518072289, "grad_norm": 8.153531074523926, "learning_rate": 4.706129032258065e-05, "loss": 0.6049, "step": 911 }, { "epoch": 5.876305220883534, "grad_norm": 5.204280376434326, "learning_rate": 4.7058064516129034e-05, "loss": 0.2992, "step": 912 }, { "epoch": 5.882730923694779, "grad_norm": 4.9666571617126465, "learning_rate": 4.705483870967743e-05, "loss": 0.6585, "step": 913 }, { "epoch": 5.889156626506024, "grad_norm": 21.247814178466797, "learning_rate": 4.705161290322581e-05, "loss": 0.5443, "step": 914 }, { "epoch": 5.8955823293172696, "grad_norm": 6.933648586273193, "learning_rate": 4.704838709677419e-05, "loss": 0.3298, "step": 915 }, { "epoch": 5.902008032128514, "grad_norm": 4.69050931930542, "learning_rate": 4.7045161290322584e-05, "loss": 0.4885, "step": 916 }, { "epoch": 5.908433734939759, "grad_norm": 5.189001560211182, "learning_rate": 4.704193548387097e-05, "loss": 0.5022, "step": 917 }, { "epoch": 5.914859437751004, "grad_norm": 4.256651878356934, "learning_rate": 4.7038709677419355e-05, "loss": 0.5115, "step": 918 }, { "epoch": 5.921285140562249, "grad_norm": 3.6839983463287354, "learning_rate": 4.703548387096775e-05, "loss": 0.2676, "step": 919 }, { "epoch": 5.927710843373494, "grad_norm": 5.558050155639648, "learning_rate": 4.7032258064516133e-05, "loss": 0.4305, "step": 920 }, { "epoch": 5.934136546184739, "grad_norm": 8.91173267364502, "learning_rate": 4.702903225806452e-05, "loss": 0.8351, "step": 921 }, { "epoch": 5.940562248995984, "grad_norm": 9.738186836242676, "learning_rate": 4.702580645161291e-05, "loss": 0.5328, "step": 922 }, { "epoch": 5.946987951807229, "grad_norm": 16.151620864868164, "learning_rate": 4.702258064516129e-05, "loss": 0.4896, "step": 923 }, { "epoch": 5.953413654618474, "grad_norm": 8.791516304016113, "learning_rate": 4.7019354838709676e-05, "loss": 0.4891, "step": 924 }, { "epoch": 5.959839357429718, "grad_norm": 8.275433540344238, "learning_rate": 4.701612903225807e-05, "loss": 0.7505, "step": 925 }, { "epoch": 5.966265060240964, "grad_norm": 19.452316284179688, "learning_rate": 4.7012903225806454e-05, "loss": 0.2562, "step": 926 }, { "epoch": 5.972690763052209, "grad_norm": 5.532136917114258, "learning_rate": 4.700967741935484e-05, "loss": 0.1689, "step": 927 }, { "epoch": 5.979116465863454, "grad_norm": 8.861352920532227, "learning_rate": 4.700645161290323e-05, "loss": 0.5456, "step": 928 }, { "epoch": 5.985542168674699, "grad_norm": 1.9066940546035767, "learning_rate": 4.700322580645162e-05, "loss": 0.2642, "step": 929 }, { "epoch": 5.991967871485944, "grad_norm": 7.1684136390686035, "learning_rate": 4.7e-05, "loss": 0.6897, "step": 930 }, { "epoch": 6.001606425702811, "grad_norm": 4.541660785675049, "learning_rate": 4.699677419354839e-05, "loss": 0.4485, "step": 931 }, { "epoch": 6.008032128514056, "grad_norm": 6.385036468505859, "learning_rate": 4.6993548387096775e-05, "loss": 0.2039, "step": 932 }, { "epoch": 6.014457831325301, "grad_norm": 8.743837356567383, "learning_rate": 4.699032258064516e-05, "loss": 0.613, "step": 933 }, { "epoch": 6.020883534136546, "grad_norm": 6.882359981536865, "learning_rate": 4.6987096774193554e-05, "loss": 0.4705, "step": 934 }, { "epoch": 6.027309236947791, "grad_norm": 6.436438083648682, "learning_rate": 4.698387096774194e-05, "loss": 0.2737, "step": 935 }, { "epoch": 6.033734939759036, "grad_norm": 5.858416557312012, "learning_rate": 4.6980645161290325e-05, "loss": 0.4367, "step": 936 }, { "epoch": 6.040160642570281, "grad_norm": 7.120092391967773, "learning_rate": 4.697741935483872e-05, "loss": 0.3926, "step": 937 }, { "epoch": 6.0465863453815265, "grad_norm": 8.477090835571289, "learning_rate": 4.6974193548387096e-05, "loss": 0.4078, "step": 938 }, { "epoch": 6.053012048192771, "grad_norm": 25.34640121459961, "learning_rate": 4.697096774193548e-05, "loss": 0.5334, "step": 939 }, { "epoch": 6.059437751004016, "grad_norm": 3.0527548789978027, "learning_rate": 4.6967741935483875e-05, "loss": 0.3106, "step": 940 }, { "epoch": 6.065863453815261, "grad_norm": 5.7898359298706055, "learning_rate": 4.696451612903226e-05, "loss": 0.4305, "step": 941 }, { "epoch": 6.072289156626506, "grad_norm": 8.923791885375977, "learning_rate": 4.6961290322580646e-05, "loss": 0.6892, "step": 942 }, { "epoch": 6.078714859437751, "grad_norm": 7.60736083984375, "learning_rate": 4.695806451612904e-05, "loss": 0.3454, "step": 943 }, { "epoch": 6.085140562248996, "grad_norm": 6.43980598449707, "learning_rate": 4.6954838709677424e-05, "loss": 0.38, "step": 944 }, { "epoch": 6.091566265060241, "grad_norm": 10.66915512084961, "learning_rate": 4.695161290322581e-05, "loss": 0.3395, "step": 945 }, { "epoch": 6.097991967871486, "grad_norm": 6.191562652587891, "learning_rate": 4.6948387096774196e-05, "loss": 0.6921, "step": 946 }, { "epoch": 6.104417670682731, "grad_norm": 3.2287933826446533, "learning_rate": 4.694516129032258e-05, "loss": 0.2927, "step": 947 }, { "epoch": 6.110843373493976, "grad_norm": 7.01705265045166, "learning_rate": 4.694193548387097e-05, "loss": 0.8613, "step": 948 }, { "epoch": 6.117269076305221, "grad_norm": 23.323759078979492, "learning_rate": 4.693870967741936e-05, "loss": 0.5477, "step": 949 }, { "epoch": 6.123694779116466, "grad_norm": 4.150506973266602, "learning_rate": 4.6935483870967745e-05, "loss": 0.3387, "step": 950 }, { "epoch": 6.130120481927711, "grad_norm": 7.5457987785339355, "learning_rate": 4.693225806451613e-05, "loss": 0.5408, "step": 951 }, { "epoch": 6.136546184738956, "grad_norm": 5.390160083770752, "learning_rate": 4.6929032258064523e-05, "loss": 0.4261, "step": 952 }, { "epoch": 6.142971887550201, "grad_norm": 6.432540416717529, "learning_rate": 4.692580645161291e-05, "loss": 0.4016, "step": 953 }, { "epoch": 6.1493975903614455, "grad_norm": 7.387927055358887, "learning_rate": 4.692258064516129e-05, "loss": 0.51, "step": 954 }, { "epoch": 6.15582329317269, "grad_norm": 10.530190467834473, "learning_rate": 4.691935483870968e-05, "loss": 0.3405, "step": 955 }, { "epoch": 6.162248995983936, "grad_norm": 6.014406204223633, "learning_rate": 4.6916129032258066e-05, "loss": 0.5283, "step": 956 }, { "epoch": 6.168674698795181, "grad_norm": 8.80288028717041, "learning_rate": 4.691290322580645e-05, "loss": 0.4916, "step": 957 }, { "epoch": 6.175100401606426, "grad_norm": 4.807860851287842, "learning_rate": 4.6909677419354844e-05, "loss": 0.4923, "step": 958 }, { "epoch": 6.181526104417671, "grad_norm": 8.26623249053955, "learning_rate": 4.690645161290323e-05, "loss": 0.5361, "step": 959 }, { "epoch": 6.187951807228916, "grad_norm": 6.505654811859131, "learning_rate": 4.6903225806451616e-05, "loss": 0.3791, "step": 960 }, { "epoch": 6.194377510040161, "grad_norm": 5.2137956619262695, "learning_rate": 4.69e-05, "loss": 0.5786, "step": 961 }, { "epoch": 6.2008032128514055, "grad_norm": 11.285112380981445, "learning_rate": 4.689677419354839e-05, "loss": 0.564, "step": 962 }, { "epoch": 6.20722891566265, "grad_norm": 6.610538959503174, "learning_rate": 4.689354838709677e-05, "loss": 0.3055, "step": 963 }, { "epoch": 6.213654618473895, "grad_norm": 2.936208963394165, "learning_rate": 4.6890322580645165e-05, "loss": 0.2712, "step": 964 }, { "epoch": 6.220080321285141, "grad_norm": 5.737492084503174, "learning_rate": 4.688709677419355e-05, "loss": 0.1646, "step": 965 }, { "epoch": 6.226506024096386, "grad_norm": 9.992554664611816, "learning_rate": 4.688387096774194e-05, "loss": 0.5376, "step": 966 }, { "epoch": 6.232931726907631, "grad_norm": 6.226168632507324, "learning_rate": 4.688064516129033e-05, "loss": 0.3108, "step": 967 }, { "epoch": 6.239357429718876, "grad_norm": 17.37044334411621, "learning_rate": 4.6877419354838715e-05, "loss": 0.6049, "step": 968 }, { "epoch": 6.2457831325301205, "grad_norm": 3.4814095497131348, "learning_rate": 4.6874193548387094e-05, "loss": 0.2372, "step": 969 }, { "epoch": 6.252208835341365, "grad_norm": 12.45607852935791, "learning_rate": 4.6870967741935486e-05, "loss": 0.6604, "step": 970 }, { "epoch": 6.25863453815261, "grad_norm": 10.102653503417969, "learning_rate": 4.686774193548387e-05, "loss": 0.5063, "step": 971 }, { "epoch": 6.265060240963855, "grad_norm": 10.707001686096191, "learning_rate": 4.686451612903226e-05, "loss": 0.5478, "step": 972 }, { "epoch": 6.2714859437751, "grad_norm": 3.9059677124023438, "learning_rate": 4.686129032258065e-05, "loss": 0.2532, "step": 973 }, { "epoch": 6.277911646586345, "grad_norm": 2.215906858444214, "learning_rate": 4.6858064516129036e-05, "loss": 0.4424, "step": 974 }, { "epoch": 6.284337349397591, "grad_norm": 4.491418838500977, "learning_rate": 4.685483870967742e-05, "loss": 0.3089, "step": 975 }, { "epoch": 6.290763052208836, "grad_norm": 7.417610168457031, "learning_rate": 4.6851612903225814e-05, "loss": 0.6399, "step": 976 }, { "epoch": 6.2971887550200805, "grad_norm": 3.3065080642700195, "learning_rate": 4.684838709677419e-05, "loss": 0.2983, "step": 977 }, { "epoch": 6.303614457831325, "grad_norm": 6.595003128051758, "learning_rate": 4.684516129032258e-05, "loss": 0.3846, "step": 978 }, { "epoch": 6.31004016064257, "grad_norm": 10.493443489074707, "learning_rate": 4.684193548387097e-05, "loss": 0.6435, "step": 979 }, { "epoch": 6.316465863453815, "grad_norm": 10.391754150390625, "learning_rate": 4.683870967741936e-05, "loss": 0.4284, "step": 980 }, { "epoch": 6.32289156626506, "grad_norm": 8.359827995300293, "learning_rate": 4.683548387096774e-05, "loss": 0.3477, "step": 981 }, { "epoch": 6.329317269076305, "grad_norm": 6.454802989959717, "learning_rate": 4.6832258064516135e-05, "loss": 0.21, "step": 982 }, { "epoch": 6.33574297188755, "grad_norm": 4.683472156524658, "learning_rate": 4.682903225806452e-05, "loss": 0.2489, "step": 983 }, { "epoch": 6.3421686746987955, "grad_norm": 14.835593223571777, "learning_rate": 4.682580645161291e-05, "loss": 0.3735, "step": 984 }, { "epoch": 6.34859437751004, "grad_norm": 5.319607257843018, "learning_rate": 4.682258064516129e-05, "loss": 0.2332, "step": 985 }, { "epoch": 6.355020080321285, "grad_norm": 4.495523929595947, "learning_rate": 4.681935483870968e-05, "loss": 0.2103, "step": 986 }, { "epoch": 6.36144578313253, "grad_norm": 7.7772297859191895, "learning_rate": 4.6816129032258064e-05, "loss": 0.366, "step": 987 }, { "epoch": 6.367871485943775, "grad_norm": 14.004790306091309, "learning_rate": 4.6812903225806456e-05, "loss": 0.8213, "step": 988 }, { "epoch": 6.37429718875502, "grad_norm": 4.859986305236816, "learning_rate": 4.680967741935484e-05, "loss": 0.3258, "step": 989 }, { "epoch": 6.380722891566265, "grad_norm": 20.611553192138672, "learning_rate": 4.680645161290323e-05, "loss": 0.6412, "step": 990 }, { "epoch": 6.38714859437751, "grad_norm": 12.723191261291504, "learning_rate": 4.680322580645162e-05, "loss": 0.4994, "step": 991 }, { "epoch": 6.393574297188755, "grad_norm": 4.839332103729248, "learning_rate": 4.6800000000000006e-05, "loss": 0.4466, "step": 992 }, { "epoch": 6.4, "grad_norm": 62.19983673095703, "learning_rate": 4.6796774193548385e-05, "loss": 0.5929, "step": 993 }, { "epoch": 6.406425702811245, "grad_norm": 7.658316135406494, "learning_rate": 4.679354838709678e-05, "loss": 0.4513, "step": 994 }, { "epoch": 6.41285140562249, "grad_norm": 3.9261932373046875, "learning_rate": 4.679032258064516e-05, "loss": 0.3556, "step": 995 }, { "epoch": 6.419277108433735, "grad_norm": 11.208858489990234, "learning_rate": 4.678709677419355e-05, "loss": 0.6029, "step": 996 }, { "epoch": 6.42570281124498, "grad_norm": 7.28602409362793, "learning_rate": 4.678387096774194e-05, "loss": 0.2764, "step": 997 }, { "epoch": 6.432128514056225, "grad_norm": 6.6600661277771, "learning_rate": 4.678064516129033e-05, "loss": 0.2834, "step": 998 }, { "epoch": 6.43855421686747, "grad_norm": 8.775915145874023, "learning_rate": 4.677741935483871e-05, "loss": 0.3007, "step": 999 }, { "epoch": 6.4449799196787145, "grad_norm": 10.695310592651367, "learning_rate": 4.67741935483871e-05, "loss": 0.5368, "step": 1000 }, { "epoch": 6.451405622489959, "grad_norm": 2.206958055496216, "learning_rate": 4.6770967741935484e-05, "loss": 0.256, "step": 1001 }, { "epoch": 6.457831325301205, "grad_norm": 15.210186958312988, "learning_rate": 4.676774193548387e-05, "loss": 0.6637, "step": 1002 }, { "epoch": 6.46425702811245, "grad_norm": 3.785717248916626, "learning_rate": 4.676451612903226e-05, "loss": 0.2452, "step": 1003 }, { "epoch": 6.470682730923695, "grad_norm": 5.268436431884766, "learning_rate": 4.676129032258065e-05, "loss": 0.5519, "step": 1004 }, { "epoch": 6.47710843373494, "grad_norm": 6.823965072631836, "learning_rate": 4.6758064516129034e-05, "loss": 0.3084, "step": 1005 }, { "epoch": 6.483534136546185, "grad_norm": 3.5109055042266846, "learning_rate": 4.6754838709677426e-05, "loss": 0.3229, "step": 1006 }, { "epoch": 6.48995983935743, "grad_norm": 2.6853702068328857, "learning_rate": 4.675161290322581e-05, "loss": 0.3491, "step": 1007 }, { "epoch": 6.4963855421686745, "grad_norm": 6.260295391082764, "learning_rate": 4.674838709677419e-05, "loss": 0.7304, "step": 1008 }, { "epoch": 6.502811244979919, "grad_norm": 6.1738996505737305, "learning_rate": 4.674516129032258e-05, "loss": 0.4278, "step": 1009 }, { "epoch": 6.509236947791164, "grad_norm": 19.49614143371582, "learning_rate": 4.674193548387097e-05, "loss": 0.3708, "step": 1010 }, { "epoch": 6.51566265060241, "grad_norm": 7.938289642333984, "learning_rate": 4.6738709677419355e-05, "loss": 0.7103, "step": 1011 }, { "epoch": 6.522088353413655, "grad_norm": 4.537919998168945, "learning_rate": 4.673548387096775e-05, "loss": 0.3266, "step": 1012 }, { "epoch": 6.5285140562249, "grad_norm": 13.904438972473145, "learning_rate": 4.673225806451613e-05, "loss": 0.6043, "step": 1013 }, { "epoch": 6.534939759036145, "grad_norm": 7.527066230773926, "learning_rate": 4.672903225806452e-05, "loss": 0.6599, "step": 1014 }, { "epoch": 6.5413654618473895, "grad_norm": 9.43061637878418, "learning_rate": 4.672580645161291e-05, "loss": 0.5124, "step": 1015 }, { "epoch": 6.547791164658634, "grad_norm": 2.978546619415283, "learning_rate": 4.672258064516129e-05, "loss": 0.1902, "step": 1016 }, { "epoch": 6.554216867469879, "grad_norm": 4.991593837738037, "learning_rate": 4.6719354838709676e-05, "loss": 0.2876, "step": 1017 }, { "epoch": 6.560642570281124, "grad_norm": 8.588459968566895, "learning_rate": 4.671612903225807e-05, "loss": 0.3742, "step": 1018 }, { "epoch": 6.567068273092369, "grad_norm": 8.027508735656738, "learning_rate": 4.6712903225806454e-05, "loss": 0.7659, "step": 1019 }, { "epoch": 6.573493975903615, "grad_norm": 6.723647594451904, "learning_rate": 4.670967741935484e-05, "loss": 0.5119, "step": 1020 }, { "epoch": 6.57991967871486, "grad_norm": 9.30924129486084, "learning_rate": 4.670645161290323e-05, "loss": 0.324, "step": 1021 }, { "epoch": 6.586345381526105, "grad_norm": 8.29932975769043, "learning_rate": 4.670322580645162e-05, "loss": 0.3018, "step": 1022 }, { "epoch": 6.5927710843373495, "grad_norm": 8.177258491516113, "learning_rate": 4.6700000000000003e-05, "loss": 0.4603, "step": 1023 }, { "epoch": 6.599196787148594, "grad_norm": 4.7083740234375, "learning_rate": 4.669677419354839e-05, "loss": 0.5588, "step": 1024 }, { "epoch": 6.605622489959839, "grad_norm": 10.913573265075684, "learning_rate": 4.6693548387096775e-05, "loss": 0.525, "step": 1025 }, { "epoch": 6.612048192771084, "grad_norm": 11.553202629089355, "learning_rate": 4.669032258064516e-05, "loss": 0.6717, "step": 1026 }, { "epoch": 6.618473895582329, "grad_norm": 3.523061752319336, "learning_rate": 4.668709677419355e-05, "loss": 0.2734, "step": 1027 }, { "epoch": 6.624899598393574, "grad_norm": 29.32855796813965, "learning_rate": 4.668387096774194e-05, "loss": 0.4355, "step": 1028 }, { "epoch": 6.63132530120482, "grad_norm": 11.372054100036621, "learning_rate": 4.6680645161290325e-05, "loss": 0.4166, "step": 1029 }, { "epoch": 6.6377510040160645, "grad_norm": 9.630682945251465, "learning_rate": 4.667741935483872e-05, "loss": 0.3287, "step": 1030 }, { "epoch": 6.644176706827309, "grad_norm": 8.668755531311035, "learning_rate": 4.6674193548387096e-05, "loss": 0.7417, "step": 1031 }, { "epoch": 6.650602409638554, "grad_norm": 7.627446174621582, "learning_rate": 4.667096774193548e-05, "loss": 0.4883, "step": 1032 }, { "epoch": 6.657028112449799, "grad_norm": 8.874289512634277, "learning_rate": 4.6667741935483874e-05, "loss": 0.465, "step": 1033 }, { "epoch": 6.663453815261044, "grad_norm": 7.731517314910889, "learning_rate": 4.666451612903226e-05, "loss": 0.5261, "step": 1034 }, { "epoch": 6.669879518072289, "grad_norm": 13.181906700134277, "learning_rate": 4.6661290322580646e-05, "loss": 0.4697, "step": 1035 }, { "epoch": 6.676305220883534, "grad_norm": 2.938978672027588, "learning_rate": 4.665806451612904e-05, "loss": 0.3883, "step": 1036 }, { "epoch": 6.682730923694779, "grad_norm": 7.73545503616333, "learning_rate": 4.6654838709677424e-05, "loss": 0.3645, "step": 1037 }, { "epoch": 6.6891566265060245, "grad_norm": 3.9010915756225586, "learning_rate": 4.665161290322581e-05, "loss": 0.3323, "step": 1038 }, { "epoch": 6.695582329317269, "grad_norm": 9.885717391967773, "learning_rate": 4.6648387096774195e-05, "loss": 0.433, "step": 1039 }, { "epoch": 6.702008032128514, "grad_norm": 9.496582984924316, "learning_rate": 4.664516129032258e-05, "loss": 0.4446, "step": 1040 }, { "epoch": 6.708433734939759, "grad_norm": 15.628626823425293, "learning_rate": 4.6641935483870967e-05, "loss": 0.4554, "step": 1041 }, { "epoch": 6.714859437751004, "grad_norm": 4.472988128662109, "learning_rate": 4.663870967741936e-05, "loss": 0.3588, "step": 1042 }, { "epoch": 6.721285140562249, "grad_norm": 6.1796183586120605, "learning_rate": 4.6635483870967745e-05, "loss": 0.4434, "step": 1043 }, { "epoch": 6.727710843373494, "grad_norm": 10.517698287963867, "learning_rate": 4.663225806451613e-05, "loss": 0.8033, "step": 1044 }, { "epoch": 6.734136546184739, "grad_norm": 7.258074760437012, "learning_rate": 4.662903225806452e-05, "loss": 0.5739, "step": 1045 }, { "epoch": 6.7405622489959836, "grad_norm": 7.971994400024414, "learning_rate": 4.662580645161291e-05, "loss": 0.3717, "step": 1046 }, { "epoch": 6.746987951807229, "grad_norm": 6.883732318878174, "learning_rate": 4.662258064516129e-05, "loss": 0.4565, "step": 1047 }, { "epoch": 6.753413654618474, "grad_norm": 3.5414679050445557, "learning_rate": 4.661935483870968e-05, "loss": 0.1344, "step": 1048 }, { "epoch": 6.759839357429719, "grad_norm": 5.476131439208984, "learning_rate": 4.6616129032258066e-05, "loss": 0.3314, "step": 1049 }, { "epoch": 6.766265060240964, "grad_norm": 2.5914087295532227, "learning_rate": 4.661290322580645e-05, "loss": 0.3207, "step": 1050 }, { "epoch": 6.772690763052209, "grad_norm": 4.104128360748291, "learning_rate": 4.6609677419354844e-05, "loss": 0.4807, "step": 1051 }, { "epoch": 6.779116465863454, "grad_norm": 4.449959754943848, "learning_rate": 4.660645161290323e-05, "loss": 0.2987, "step": 1052 }, { "epoch": 6.785542168674699, "grad_norm": 6.429729461669922, "learning_rate": 4.6603225806451615e-05, "loss": 0.2033, "step": 1053 }, { "epoch": 6.7919678714859435, "grad_norm": 2.9653823375701904, "learning_rate": 4.660000000000001e-05, "loss": 0.3259, "step": 1054 }, { "epoch": 6.798393574297188, "grad_norm": 9.707965850830078, "learning_rate": 4.659677419354839e-05, "loss": 0.2533, "step": 1055 }, { "epoch": 6.804819277108434, "grad_norm": 8.539097785949707, "learning_rate": 4.659354838709677e-05, "loss": 0.4626, "step": 1056 }, { "epoch": 6.811244979919679, "grad_norm": 8.934945106506348, "learning_rate": 4.6590322580645165e-05, "loss": 0.6283, "step": 1057 }, { "epoch": 6.817670682730924, "grad_norm": 3.8745059967041016, "learning_rate": 4.658709677419355e-05, "loss": 0.348, "step": 1058 }, { "epoch": 6.824096385542169, "grad_norm": 4.945444107055664, "learning_rate": 4.6583870967741936e-05, "loss": 0.525, "step": 1059 }, { "epoch": 6.830522088353414, "grad_norm": 6.051918983459473, "learning_rate": 4.658064516129033e-05, "loss": 0.4028, "step": 1060 }, { "epoch": 6.836947791164659, "grad_norm": 10.36018180847168, "learning_rate": 4.6577419354838715e-05, "loss": 0.518, "step": 1061 }, { "epoch": 6.843373493975903, "grad_norm": 15.900336265563965, "learning_rate": 4.65741935483871e-05, "loss": 0.3871, "step": 1062 }, { "epoch": 6.849799196787148, "grad_norm": 7.424167633056641, "learning_rate": 4.6570967741935486e-05, "loss": 0.3092, "step": 1063 }, { "epoch": 6.856224899598393, "grad_norm": 3.1919147968292236, "learning_rate": 4.656774193548387e-05, "loss": 0.3463, "step": 1064 }, { "epoch": 6.862650602409639, "grad_norm": 16.880382537841797, "learning_rate": 4.656451612903226e-05, "loss": 0.312, "step": 1065 }, { "epoch": 6.869076305220884, "grad_norm": 7.6045026779174805, "learning_rate": 4.656129032258065e-05, "loss": 0.5768, "step": 1066 }, { "epoch": 6.875502008032129, "grad_norm": 3.5094964504241943, "learning_rate": 4.6558064516129036e-05, "loss": 0.4669, "step": 1067 }, { "epoch": 6.881927710843374, "grad_norm": 11.786995887756348, "learning_rate": 4.655483870967742e-05, "loss": 0.5681, "step": 1068 }, { "epoch": 6.8883534136546185, "grad_norm": 10.820655822753906, "learning_rate": 4.6551612903225814e-05, "loss": 0.5397, "step": 1069 }, { "epoch": 6.894779116465863, "grad_norm": 9.350234031677246, "learning_rate": 4.654838709677419e-05, "loss": 0.3532, "step": 1070 }, { "epoch": 6.901204819277108, "grad_norm": 9.472579002380371, "learning_rate": 4.654516129032258e-05, "loss": 0.459, "step": 1071 }, { "epoch": 6.907630522088353, "grad_norm": 19.02714729309082, "learning_rate": 4.654193548387097e-05, "loss": 0.7792, "step": 1072 }, { "epoch": 6.914056224899598, "grad_norm": 7.094696521759033, "learning_rate": 4.6538709677419357e-05, "loss": 0.4234, "step": 1073 }, { "epoch": 6.920481927710844, "grad_norm": 6.166698932647705, "learning_rate": 4.653548387096774e-05, "loss": 0.5148, "step": 1074 }, { "epoch": 6.926907630522089, "grad_norm": 8.42577838897705, "learning_rate": 4.6532258064516135e-05, "loss": 0.5099, "step": 1075 }, { "epoch": 6.933333333333334, "grad_norm": 9.507230758666992, "learning_rate": 4.652903225806452e-05, "loss": 0.3384, "step": 1076 }, { "epoch": 6.9397590361445785, "grad_norm": 2.8513498306274414, "learning_rate": 4.6525806451612906e-05, "loss": 0.1906, "step": 1077 }, { "epoch": 6.946184738955823, "grad_norm": 9.658355712890625, "learning_rate": 4.652258064516129e-05, "loss": 0.5966, "step": 1078 }, { "epoch": 6.952610441767068, "grad_norm": 13.663691520690918, "learning_rate": 4.651935483870968e-05, "loss": 0.4725, "step": 1079 }, { "epoch": 6.959036144578313, "grad_norm": 12.93767261505127, "learning_rate": 4.651612903225806e-05, "loss": 0.3578, "step": 1080 }, { "epoch": 6.965461847389558, "grad_norm": 6.920025825500488, "learning_rate": 4.6512903225806456e-05, "loss": 0.527, "step": 1081 }, { "epoch": 6.971887550200803, "grad_norm": 3.0349254608154297, "learning_rate": 4.650967741935484e-05, "loss": 0.2879, "step": 1082 }, { "epoch": 6.978313253012049, "grad_norm": 7.0320940017700195, "learning_rate": 4.650645161290323e-05, "loss": 0.4132, "step": 1083 }, { "epoch": 6.9847389558232935, "grad_norm": 8.968367576599121, "learning_rate": 4.650322580645162e-05, "loss": 0.5483, "step": 1084 }, { "epoch": 6.991164658634538, "grad_norm": 4.239584445953369, "learning_rate": 4.6500000000000005e-05, "loss": 0.4621, "step": 1085 }, { "epoch": 7.000803212851405, "grad_norm": 7.233964443206787, "learning_rate": 4.6496774193548384e-05, "loss": 0.3883, "step": 1086 }, { "epoch": 7.00722891566265, "grad_norm": 4.706003665924072, "learning_rate": 4.649354838709678e-05, "loss": 0.3928, "step": 1087 }, { "epoch": 7.013654618473896, "grad_norm": 6.897171974182129, "learning_rate": 4.649032258064516e-05, "loss": 0.2917, "step": 1088 }, { "epoch": 7.020080321285141, "grad_norm": 8.440574645996094, "learning_rate": 4.648709677419355e-05, "loss": 0.4468, "step": 1089 }, { "epoch": 7.026506024096386, "grad_norm": 3.9294273853302, "learning_rate": 4.648387096774194e-05, "loss": 0.2359, "step": 1090 }, { "epoch": 7.032931726907631, "grad_norm": 3.9698150157928467, "learning_rate": 4.6480645161290326e-05, "loss": 0.5022, "step": 1091 }, { "epoch": 7.0393574297188755, "grad_norm": 7.47318172454834, "learning_rate": 4.647741935483871e-05, "loss": 0.4941, "step": 1092 }, { "epoch": 7.04578313253012, "grad_norm": 3.644160270690918, "learning_rate": 4.6474193548387105e-05, "loss": 0.3633, "step": 1093 }, { "epoch": 7.052208835341365, "grad_norm": 5.961205005645752, "learning_rate": 4.6470967741935484e-05, "loss": 0.3438, "step": 1094 }, { "epoch": 7.05863453815261, "grad_norm": 8.771951675415039, "learning_rate": 4.646774193548387e-05, "loss": 0.3825, "step": 1095 }, { "epoch": 7.065060240963855, "grad_norm": 9.846116065979004, "learning_rate": 4.646451612903226e-05, "loss": 0.6573, "step": 1096 }, { "epoch": 7.071485943775101, "grad_norm": 24.203649520874023, "learning_rate": 4.646129032258065e-05, "loss": 0.3452, "step": 1097 }, { "epoch": 7.077911646586346, "grad_norm": 6.259042263031006, "learning_rate": 4.645806451612903e-05, "loss": 0.6739, "step": 1098 }, { "epoch": 7.0843373493975905, "grad_norm": 3.9003167152404785, "learning_rate": 4.6454838709677426e-05, "loss": 0.4828, "step": 1099 }, { "epoch": 7.090763052208835, "grad_norm": 4.587352275848389, "learning_rate": 4.645161290322581e-05, "loss": 0.2538, "step": 1100 }, { "epoch": 7.09718875502008, "grad_norm": 7.111338138580322, "learning_rate": 4.64483870967742e-05, "loss": 0.2317, "step": 1101 }, { "epoch": 7.103614457831325, "grad_norm": 17.217247009277344, "learning_rate": 4.644516129032258e-05, "loss": 0.6343, "step": 1102 }, { "epoch": 7.11004016064257, "grad_norm": 4.591298580169678, "learning_rate": 4.644193548387097e-05, "loss": 0.456, "step": 1103 }, { "epoch": 7.116465863453815, "grad_norm": 4.395002841949463, "learning_rate": 4.6438709677419354e-05, "loss": 0.3879, "step": 1104 }, { "epoch": 7.12289156626506, "grad_norm": 3.390869617462158, "learning_rate": 4.643548387096775e-05, "loss": 0.2421, "step": 1105 }, { "epoch": 7.129317269076306, "grad_norm": 8.694073677062988, "learning_rate": 4.643225806451613e-05, "loss": 0.5397, "step": 1106 }, { "epoch": 7.1357429718875505, "grad_norm": 9.128288269042969, "learning_rate": 4.642903225806452e-05, "loss": 0.7247, "step": 1107 }, { "epoch": 7.142168674698795, "grad_norm": 26.379074096679688, "learning_rate": 4.642580645161291e-05, "loss": 0.4328, "step": 1108 }, { "epoch": 7.14859437751004, "grad_norm": 7.248779296875, "learning_rate": 4.642258064516129e-05, "loss": 0.8468, "step": 1109 }, { "epoch": 7.155020080321285, "grad_norm": 6.93614387512207, "learning_rate": 4.6419354838709675e-05, "loss": 0.4712, "step": 1110 }, { "epoch": 7.16144578313253, "grad_norm": 9.812421798706055, "learning_rate": 4.641612903225807e-05, "loss": 0.3959, "step": 1111 }, { "epoch": 7.167871485943775, "grad_norm": 9.61568546295166, "learning_rate": 4.641290322580645e-05, "loss": 0.4407, "step": 1112 }, { "epoch": 7.17429718875502, "grad_norm": 8.680220603942871, "learning_rate": 4.640967741935484e-05, "loss": 0.6585, "step": 1113 }, { "epoch": 7.180722891566265, "grad_norm": 4.828248977661133, "learning_rate": 4.640645161290323e-05, "loss": 0.4791, "step": 1114 }, { "epoch": 7.18714859437751, "grad_norm": 15.6604585647583, "learning_rate": 4.640322580645162e-05, "loss": 0.5467, "step": 1115 }, { "epoch": 7.193574297188755, "grad_norm": 6.178389072418213, "learning_rate": 4.64e-05, "loss": 0.3843, "step": 1116 }, { "epoch": 7.2, "grad_norm": 6.957866668701172, "learning_rate": 4.639677419354839e-05, "loss": 0.2411, "step": 1117 }, { "epoch": 7.206425702811245, "grad_norm": 10.999950408935547, "learning_rate": 4.6393548387096774e-05, "loss": 0.4665, "step": 1118 }, { "epoch": 7.21285140562249, "grad_norm": 4.237268447875977, "learning_rate": 4.639032258064516e-05, "loss": 0.307, "step": 1119 }, { "epoch": 7.219277108433735, "grad_norm": 6.015897750854492, "learning_rate": 4.638709677419355e-05, "loss": 0.2734, "step": 1120 }, { "epoch": 7.22570281124498, "grad_norm": 4.051245212554932, "learning_rate": 4.638387096774194e-05, "loss": 0.4025, "step": 1121 }, { "epoch": 7.232128514056225, "grad_norm": 16.532875061035156, "learning_rate": 4.6380645161290324e-05, "loss": 0.6042, "step": 1122 }, { "epoch": 7.2385542168674695, "grad_norm": 6.217335224151611, "learning_rate": 4.6377419354838716e-05, "loss": 0.4253, "step": 1123 }, { "epoch": 7.244979919678715, "grad_norm": 5.38313627243042, "learning_rate": 4.63741935483871e-05, "loss": 0.3932, "step": 1124 }, { "epoch": 7.25140562248996, "grad_norm": 5.4221930503845215, "learning_rate": 4.637096774193548e-05, "loss": 0.4268, "step": 1125 }, { "epoch": 7.257831325301205, "grad_norm": 4.879687786102295, "learning_rate": 4.6367741935483874e-05, "loss": 0.5022, "step": 1126 }, { "epoch": 7.26425702811245, "grad_norm": 4.992167949676514, "learning_rate": 4.636451612903226e-05, "loss": 0.4709, "step": 1127 }, { "epoch": 7.270682730923695, "grad_norm": 16.95472526550293, "learning_rate": 4.6361290322580645e-05, "loss": 0.3881, "step": 1128 }, { "epoch": 7.27710843373494, "grad_norm": 12.96505069732666, "learning_rate": 4.635806451612904e-05, "loss": 0.2071, "step": 1129 }, { "epoch": 7.2835341365461845, "grad_norm": 6.190305233001709, "learning_rate": 4.635483870967742e-05, "loss": 0.4575, "step": 1130 }, { "epoch": 7.289959839357429, "grad_norm": 8.844825744628906, "learning_rate": 4.635161290322581e-05, "loss": 0.3988, "step": 1131 }, { "epoch": 7.296385542168674, "grad_norm": 4.66742467880249, "learning_rate": 4.63483870967742e-05, "loss": 0.3021, "step": 1132 }, { "epoch": 7.30281124497992, "grad_norm": 10.526039123535156, "learning_rate": 4.634516129032258e-05, "loss": 0.3214, "step": 1133 }, { "epoch": 7.309236947791165, "grad_norm": 3.291598320007324, "learning_rate": 4.6341935483870966e-05, "loss": 0.2919, "step": 1134 }, { "epoch": 7.31566265060241, "grad_norm": 34.323184967041016, "learning_rate": 4.633870967741936e-05, "loss": 0.7362, "step": 1135 }, { "epoch": 7.322088353413655, "grad_norm": 29.868328094482422, "learning_rate": 4.6335483870967744e-05, "loss": 0.4545, "step": 1136 }, { "epoch": 7.3285140562249, "grad_norm": 9.935047149658203, "learning_rate": 4.633225806451613e-05, "loss": 0.6392, "step": 1137 }, { "epoch": 7.3349397590361445, "grad_norm": 14.933503150939941, "learning_rate": 4.632903225806452e-05, "loss": 0.415, "step": 1138 }, { "epoch": 7.341365461847389, "grad_norm": 6.670083522796631, "learning_rate": 4.632580645161291e-05, "loss": 0.4044, "step": 1139 }, { "epoch": 7.347791164658634, "grad_norm": 3.497728109359741, "learning_rate": 4.632258064516129e-05, "loss": 0.4061, "step": 1140 }, { "epoch": 7.354216867469879, "grad_norm": 6.722110271453857, "learning_rate": 4.631935483870968e-05, "loss": 0.4172, "step": 1141 }, { "epoch": 7.360642570281125, "grad_norm": 2.372227191925049, "learning_rate": 4.6316129032258065e-05, "loss": 0.2495, "step": 1142 }, { "epoch": 7.36706827309237, "grad_norm": 9.053668022155762, "learning_rate": 4.631290322580645e-05, "loss": 0.6152, "step": 1143 }, { "epoch": 7.373493975903615, "grad_norm": 8.403499603271484, "learning_rate": 4.6309677419354843e-05, "loss": 0.5549, "step": 1144 }, { "epoch": 7.3799196787148595, "grad_norm": 7.351271629333496, "learning_rate": 4.630645161290323e-05, "loss": 0.1999, "step": 1145 }, { "epoch": 7.386345381526104, "grad_norm": 7.084466457366943, "learning_rate": 4.6303225806451615e-05, "loss": 0.4077, "step": 1146 }, { "epoch": 7.392771084337349, "grad_norm": 5.684420585632324, "learning_rate": 4.630000000000001e-05, "loss": 0.4243, "step": 1147 }, { "epoch": 7.399196787148594, "grad_norm": 6.040533542633057, "learning_rate": 4.6296774193548386e-05, "loss": 0.443, "step": 1148 }, { "epoch": 7.405622489959839, "grad_norm": 3.446445941925049, "learning_rate": 4.629354838709677e-05, "loss": 0.3255, "step": 1149 }, { "epoch": 7.412048192771084, "grad_norm": 8.890243530273438, "learning_rate": 4.6290322580645164e-05, "loss": 0.605, "step": 1150 }, { "epoch": 7.41847389558233, "grad_norm": 3.7256531715393066, "learning_rate": 4.628709677419355e-05, "loss": 0.2635, "step": 1151 }, { "epoch": 7.424899598393575, "grad_norm": 4.371268272399902, "learning_rate": 4.6283870967741936e-05, "loss": 0.3811, "step": 1152 }, { "epoch": 7.4313253012048195, "grad_norm": 11.229676246643066, "learning_rate": 4.628064516129033e-05, "loss": 0.39, "step": 1153 }, { "epoch": 7.437751004016064, "grad_norm": 3.5051867961883545, "learning_rate": 4.6277419354838714e-05, "loss": 0.2957, "step": 1154 }, { "epoch": 7.444176706827309, "grad_norm": 7.055939197540283, "learning_rate": 4.62741935483871e-05, "loss": 0.3858, "step": 1155 }, { "epoch": 7.450602409638554, "grad_norm": 18.99593734741211, "learning_rate": 4.6270967741935485e-05, "loss": 0.3961, "step": 1156 }, { "epoch": 7.457028112449799, "grad_norm": 7.08853816986084, "learning_rate": 4.626774193548387e-05, "loss": 0.4739, "step": 1157 }, { "epoch": 7.463453815261044, "grad_norm": 6.375276565551758, "learning_rate": 4.626451612903226e-05, "loss": 0.3682, "step": 1158 }, { "epoch": 7.469879518072289, "grad_norm": 9.315054893493652, "learning_rate": 4.626129032258065e-05, "loss": 0.3063, "step": 1159 }, { "epoch": 7.4763052208835346, "grad_norm": 7.323049545288086, "learning_rate": 4.6258064516129035e-05, "loss": 0.4567, "step": 1160 }, { "epoch": 7.482730923694779, "grad_norm": 10.215336799621582, "learning_rate": 4.625483870967742e-05, "loss": 0.5076, "step": 1161 }, { "epoch": 7.489156626506024, "grad_norm": 9.233339309692383, "learning_rate": 4.625161290322581e-05, "loss": 0.4261, "step": 1162 }, { "epoch": 7.495582329317269, "grad_norm": 12.066170692443848, "learning_rate": 4.62483870967742e-05, "loss": 0.3272, "step": 1163 }, { "epoch": 7.502008032128514, "grad_norm": 9.68665599822998, "learning_rate": 4.624516129032258e-05, "loss": 0.2732, "step": 1164 }, { "epoch": 7.508433734939759, "grad_norm": 4.877880573272705, "learning_rate": 4.624193548387097e-05, "loss": 0.2715, "step": 1165 }, { "epoch": 7.514859437751004, "grad_norm": 11.005605697631836, "learning_rate": 4.6238709677419356e-05, "loss": 0.4112, "step": 1166 }, { "epoch": 7.521285140562249, "grad_norm": 6.651071071624756, "learning_rate": 4.623548387096774e-05, "loss": 0.4016, "step": 1167 }, { "epoch": 7.527710843373494, "grad_norm": 4.185391902923584, "learning_rate": 4.6232258064516134e-05, "loss": 0.1924, "step": 1168 }, { "epoch": 7.5341365461847385, "grad_norm": 4.305217742919922, "learning_rate": 4.622903225806452e-05, "loss": 0.341, "step": 1169 }, { "epoch": 7.540562248995984, "grad_norm": 7.722776412963867, "learning_rate": 4.6225806451612906e-05, "loss": 0.5951, "step": 1170 }, { "epoch": 7.546987951807229, "grad_norm": 5.7658538818359375, "learning_rate": 4.62225806451613e-05, "loss": 0.5685, "step": 1171 }, { "epoch": 7.553413654618474, "grad_norm": 8.381380081176758, "learning_rate": 4.621935483870968e-05, "loss": 0.4727, "step": 1172 }, { "epoch": 7.559839357429719, "grad_norm": 2.2390329837799072, "learning_rate": 4.621612903225806e-05, "loss": 0.2014, "step": 1173 }, { "epoch": 7.566265060240964, "grad_norm": 3.7566192150115967, "learning_rate": 4.6212903225806455e-05, "loss": 0.2543, "step": 1174 }, { "epoch": 7.572690763052209, "grad_norm": 7.7184834480285645, "learning_rate": 4.620967741935484e-05, "loss": 0.3157, "step": 1175 }, { "epoch": 7.5791164658634536, "grad_norm": 9.794537544250488, "learning_rate": 4.620645161290323e-05, "loss": 0.5094, "step": 1176 }, { "epoch": 7.585542168674698, "grad_norm": 11.921385765075684, "learning_rate": 4.620322580645162e-05, "loss": 0.6878, "step": 1177 }, { "epoch": 7.591967871485943, "grad_norm": 15.691027641296387, "learning_rate": 4.6200000000000005e-05, "loss": 0.6585, "step": 1178 }, { "epoch": 7.598393574297189, "grad_norm": 4.384514808654785, "learning_rate": 4.6196774193548384e-05, "loss": 0.2666, "step": 1179 }, { "epoch": 7.604819277108434, "grad_norm": 6.324769020080566, "learning_rate": 4.6193548387096776e-05, "loss": 0.6137, "step": 1180 }, { "epoch": 7.611244979919679, "grad_norm": 5.035000801086426, "learning_rate": 4.619032258064516e-05, "loss": 0.259, "step": 1181 }, { "epoch": 7.617670682730924, "grad_norm": 5.407453536987305, "learning_rate": 4.618709677419355e-05, "loss": 0.3279, "step": 1182 }, { "epoch": 7.624096385542169, "grad_norm": 6.755465507507324, "learning_rate": 4.618387096774194e-05, "loss": 0.3228, "step": 1183 }, { "epoch": 7.6305220883534135, "grad_norm": 7.035127639770508, "learning_rate": 4.6180645161290326e-05, "loss": 0.3264, "step": 1184 }, { "epoch": 7.636947791164658, "grad_norm": 5.653645992279053, "learning_rate": 4.617741935483871e-05, "loss": 0.2819, "step": 1185 }, { "epoch": 7.643373493975903, "grad_norm": 8.604130744934082, "learning_rate": 4.6174193548387104e-05, "loss": 0.3978, "step": 1186 }, { "epoch": 7.649799196787148, "grad_norm": 12.577999114990234, "learning_rate": 4.617096774193548e-05, "loss": 0.5967, "step": 1187 }, { "epoch": 7.656224899598394, "grad_norm": 3.652315855026245, "learning_rate": 4.616774193548387e-05, "loss": 0.2234, "step": 1188 }, { "epoch": 7.662650602409639, "grad_norm": 2.520179271697998, "learning_rate": 4.616451612903226e-05, "loss": 0.148, "step": 1189 }, { "epoch": 7.669076305220884, "grad_norm": 15.360176086425781, "learning_rate": 4.616129032258065e-05, "loss": 0.5621, "step": 1190 }, { "epoch": 7.675502008032129, "grad_norm": 3.9702022075653076, "learning_rate": 4.615806451612903e-05, "loss": 0.3616, "step": 1191 }, { "epoch": 7.6819277108433734, "grad_norm": 10.91743278503418, "learning_rate": 4.6154838709677425e-05, "loss": 0.3203, "step": 1192 }, { "epoch": 7.688353413654618, "grad_norm": 6.751670837402344, "learning_rate": 4.615161290322581e-05, "loss": 0.5365, "step": 1193 }, { "epoch": 7.694779116465863, "grad_norm": 9.194317817687988, "learning_rate": 4.6148387096774197e-05, "loss": 0.3531, "step": 1194 }, { "epoch": 7.701204819277108, "grad_norm": 8.677982330322266, "learning_rate": 4.614516129032258e-05, "loss": 0.3548, "step": 1195 }, { "epoch": 7.707630522088353, "grad_norm": 7.670979976654053, "learning_rate": 4.614193548387097e-05, "loss": 0.2892, "step": 1196 }, { "epoch": 7.714056224899599, "grad_norm": 8.336075782775879, "learning_rate": 4.6138709677419354e-05, "loss": 0.6934, "step": 1197 }, { "epoch": 7.720481927710844, "grad_norm": 18.237112045288086, "learning_rate": 4.6135483870967746e-05, "loss": 0.7352, "step": 1198 }, { "epoch": 7.7269076305220885, "grad_norm": 6.726197242736816, "learning_rate": 4.613225806451613e-05, "loss": 0.5193, "step": 1199 }, { "epoch": 7.733333333333333, "grad_norm": 2.579115152359009, "learning_rate": 4.612903225806452e-05, "loss": 0.3413, "step": 1200 }, { "epoch": 7.739759036144578, "grad_norm": 8.623457908630371, "learning_rate": 4.612580645161291e-05, "loss": 0.5431, "step": 1201 }, { "epoch": 7.746184738955823, "grad_norm": 3.432537317276001, "learning_rate": 4.6122580645161296e-05, "loss": 0.3655, "step": 1202 }, { "epoch": 7.752610441767068, "grad_norm": 10.696565628051758, "learning_rate": 4.6119354838709675e-05, "loss": 0.3512, "step": 1203 }, { "epoch": 7.759036144578313, "grad_norm": 8.83281135559082, "learning_rate": 4.611612903225807e-05, "loss": 0.6247, "step": 1204 }, { "epoch": 7.765461847389558, "grad_norm": 12.036530494689941, "learning_rate": 4.611290322580645e-05, "loss": 0.4762, "step": 1205 }, { "epoch": 7.771887550200804, "grad_norm": 4.885997772216797, "learning_rate": 4.610967741935484e-05, "loss": 0.3113, "step": 1206 }, { "epoch": 7.7783132530120485, "grad_norm": 3.9661948680877686, "learning_rate": 4.610645161290323e-05, "loss": 0.4166, "step": 1207 }, { "epoch": 7.784738955823293, "grad_norm": 7.176928997039795, "learning_rate": 4.610322580645162e-05, "loss": 0.4328, "step": 1208 }, { "epoch": 7.791164658634538, "grad_norm": 3.306053638458252, "learning_rate": 4.61e-05, "loss": 0.3373, "step": 1209 }, { "epoch": 7.797590361445783, "grad_norm": 5.613463878631592, "learning_rate": 4.609677419354839e-05, "loss": 0.4727, "step": 1210 }, { "epoch": 7.804016064257028, "grad_norm": 6.018500328063965, "learning_rate": 4.6093548387096774e-05, "loss": 0.5377, "step": 1211 }, { "epoch": 7.810441767068273, "grad_norm": 7.374053001403809, "learning_rate": 4.609032258064516e-05, "loss": 0.5384, "step": 1212 }, { "epoch": 7.816867469879518, "grad_norm": 8.9775972366333, "learning_rate": 4.608709677419355e-05, "loss": 0.4711, "step": 1213 }, { "epoch": 7.823293172690763, "grad_norm": 7.9424591064453125, "learning_rate": 4.608387096774194e-05, "loss": 0.5335, "step": 1214 }, { "epoch": 7.829718875502008, "grad_norm": 8.559467315673828, "learning_rate": 4.6080645161290323e-05, "loss": 0.3129, "step": 1215 }, { "epoch": 7.836144578313253, "grad_norm": 14.018755912780762, "learning_rate": 4.6077419354838716e-05, "loss": 0.3386, "step": 1216 }, { "epoch": 7.842570281124498, "grad_norm": 5.339531898498535, "learning_rate": 4.60741935483871e-05, "loss": 0.2216, "step": 1217 }, { "epoch": 7.848995983935743, "grad_norm": 8.51940631866455, "learning_rate": 4.607096774193548e-05, "loss": 0.2969, "step": 1218 }, { "epoch": 7.855421686746988, "grad_norm": 4.12825870513916, "learning_rate": 4.606774193548387e-05, "loss": 0.2634, "step": 1219 }, { "epoch": 7.861847389558233, "grad_norm": 11.128599166870117, "learning_rate": 4.606451612903226e-05, "loss": 0.4167, "step": 1220 }, { "epoch": 7.868273092369478, "grad_norm": 6.970978260040283, "learning_rate": 4.6061290322580645e-05, "loss": 0.6322, "step": 1221 }, { "epoch": 7.874698795180723, "grad_norm": 9.371246337890625, "learning_rate": 4.605806451612904e-05, "loss": 0.4812, "step": 1222 }, { "epoch": 7.8811244979919675, "grad_norm": 12.079713821411133, "learning_rate": 4.605483870967742e-05, "loss": 0.494, "step": 1223 }, { "epoch": 7.887550200803213, "grad_norm": 6.290570259094238, "learning_rate": 4.605161290322581e-05, "loss": 0.4398, "step": 1224 }, { "epoch": 7.893975903614458, "grad_norm": 6.563447952270508, "learning_rate": 4.60483870967742e-05, "loss": 0.4416, "step": 1225 }, { "epoch": 7.900401606425703, "grad_norm": 6.902187347412109, "learning_rate": 4.604516129032258e-05, "loss": 0.432, "step": 1226 }, { "epoch": 7.906827309236948, "grad_norm": 6.8189616203308105, "learning_rate": 4.6041935483870966e-05, "loss": 0.4282, "step": 1227 }, { "epoch": 7.913253012048193, "grad_norm": 9.758851051330566, "learning_rate": 4.603870967741936e-05, "loss": 0.4417, "step": 1228 }, { "epoch": 7.919678714859438, "grad_norm": 7.671939373016357, "learning_rate": 4.6035483870967744e-05, "loss": 0.3633, "step": 1229 }, { "epoch": 7.9261044176706825, "grad_norm": 5.742655277252197, "learning_rate": 4.603225806451613e-05, "loss": 0.2928, "step": 1230 }, { "epoch": 7.932530120481927, "grad_norm": 6.565049171447754, "learning_rate": 4.602903225806452e-05, "loss": 0.459, "step": 1231 }, { "epoch": 7.938955823293172, "grad_norm": 8.890172958374023, "learning_rate": 4.602580645161291e-05, "loss": 0.453, "step": 1232 }, { "epoch": 7.945381526104418, "grad_norm": 13.321788787841797, "learning_rate": 4.602258064516129e-05, "loss": 0.7027, "step": 1233 }, { "epoch": 7.951807228915663, "grad_norm": 6.707500457763672, "learning_rate": 4.601935483870968e-05, "loss": 0.5691, "step": 1234 }, { "epoch": 7.958232931726908, "grad_norm": 5.967440605163574, "learning_rate": 4.6016129032258065e-05, "loss": 0.5527, "step": 1235 }, { "epoch": 7.964658634538153, "grad_norm": 4.261924743652344, "learning_rate": 4.601290322580645e-05, "loss": 0.6115, "step": 1236 }, { "epoch": 7.971084337349398, "grad_norm": 14.593262672424316, "learning_rate": 4.600967741935484e-05, "loss": 0.5452, "step": 1237 }, { "epoch": 7.9775100401606425, "grad_norm": 6.6620283126831055, "learning_rate": 4.600645161290323e-05, "loss": 0.4301, "step": 1238 }, { "epoch": 7.983935742971887, "grad_norm": 10.47166919708252, "learning_rate": 4.6003225806451614e-05, "loss": 0.7208, "step": 1239 }, { "epoch": 7.990361445783132, "grad_norm": 13.44771957397461, "learning_rate": 4.600000000000001e-05, "loss": 0.3743, "step": 1240 }, { "epoch": 7.996787148594377, "grad_norm": 8.337272644042969, "learning_rate": 4.599677419354839e-05, "loss": 0.8407, "step": 1241 }, { "epoch": 8.006425702811246, "grad_norm": 6.8136982917785645, "learning_rate": 4.599354838709677e-05, "loss": 0.631, "step": 1242 }, { "epoch": 8.01285140562249, "grad_norm": 6.435269832611084, "learning_rate": 4.5990322580645164e-05, "loss": 0.5282, "step": 1243 }, { "epoch": 8.019277108433736, "grad_norm": 10.392305374145508, "learning_rate": 4.598709677419355e-05, "loss": 0.3085, "step": 1244 }, { "epoch": 8.02570281124498, "grad_norm": 6.412435531616211, "learning_rate": 4.5983870967741935e-05, "loss": 0.5082, "step": 1245 }, { "epoch": 8.032128514056225, "grad_norm": 5.042918682098389, "learning_rate": 4.598064516129033e-05, "loss": 0.3827, "step": 1246 }, { "epoch": 8.03855421686747, "grad_norm": 23.113771438598633, "learning_rate": 4.5977419354838714e-05, "loss": 0.456, "step": 1247 }, { "epoch": 8.044979919678715, "grad_norm": 9.140106201171875, "learning_rate": 4.59741935483871e-05, "loss": 0.3833, "step": 1248 }, { "epoch": 8.051405622489959, "grad_norm": 8.036559104919434, "learning_rate": 4.5970967741935485e-05, "loss": 0.5454, "step": 1249 }, { "epoch": 8.057831325301205, "grad_norm": 4.019773960113525, "learning_rate": 4.596774193548387e-05, "loss": 0.318, "step": 1250 }, { "epoch": 8.06425702811245, "grad_norm": 5.608523368835449, "learning_rate": 4.5964516129032256e-05, "loss": 0.149, "step": 1251 }, { "epoch": 8.070682730923695, "grad_norm": 9.093694686889648, "learning_rate": 4.596129032258065e-05, "loss": 0.6189, "step": 1252 }, { "epoch": 8.07710843373494, "grad_norm": 4.162722587585449, "learning_rate": 4.5958064516129035e-05, "loss": 0.2752, "step": 1253 }, { "epoch": 8.083534136546184, "grad_norm": 5.376933574676514, "learning_rate": 4.595483870967742e-05, "loss": 0.4295, "step": 1254 }, { "epoch": 8.08995983935743, "grad_norm": 4.711907863616943, "learning_rate": 4.595161290322581e-05, "loss": 0.3545, "step": 1255 }, { "epoch": 8.096385542168674, "grad_norm": 7.7055134773254395, "learning_rate": 4.59483870967742e-05, "loss": 0.4158, "step": 1256 }, { "epoch": 8.10281124497992, "grad_norm": 8.613548278808594, "learning_rate": 4.594516129032258e-05, "loss": 0.4198, "step": 1257 }, { "epoch": 8.109236947791164, "grad_norm": 8.822214126586914, "learning_rate": 4.594193548387097e-05, "loss": 0.1897, "step": 1258 }, { "epoch": 8.11566265060241, "grad_norm": 8.909083366394043, "learning_rate": 4.5938709677419356e-05, "loss": 0.5412, "step": 1259 }, { "epoch": 8.122088353413655, "grad_norm": 4.460502624511719, "learning_rate": 4.593548387096774e-05, "loss": 0.2773, "step": 1260 }, { "epoch": 8.1285140562249, "grad_norm": 5.338690757751465, "learning_rate": 4.5932258064516134e-05, "loss": 0.4567, "step": 1261 }, { "epoch": 8.134939759036145, "grad_norm": 3.1669321060180664, "learning_rate": 4.592903225806452e-05, "loss": 0.1992, "step": 1262 }, { "epoch": 8.14136546184739, "grad_norm": 9.329297065734863, "learning_rate": 4.5925806451612905e-05, "loss": 0.5537, "step": 1263 }, { "epoch": 8.147791164658635, "grad_norm": 6.4765119552612305, "learning_rate": 4.59225806451613e-05, "loss": 0.3396, "step": 1264 }, { "epoch": 8.154216867469879, "grad_norm": 5.011570453643799, "learning_rate": 4.5919354838709677e-05, "loss": 0.3884, "step": 1265 }, { "epoch": 8.160642570281125, "grad_norm": 8.092144966125488, "learning_rate": 4.591612903225806e-05, "loss": 0.2535, "step": 1266 }, { "epoch": 8.167068273092369, "grad_norm": 5.739036560058594, "learning_rate": 4.5912903225806455e-05, "loss": 0.3661, "step": 1267 }, { "epoch": 8.173493975903614, "grad_norm": 5.2736592292785645, "learning_rate": 4.590967741935484e-05, "loss": 0.5255, "step": 1268 }, { "epoch": 8.17991967871486, "grad_norm": 2.692713737487793, "learning_rate": 4.5906451612903226e-05, "loss": 0.209, "step": 1269 }, { "epoch": 8.186345381526104, "grad_norm": 11.65398120880127, "learning_rate": 4.590322580645162e-05, "loss": 0.5024, "step": 1270 }, { "epoch": 8.19277108433735, "grad_norm": 7.567448616027832, "learning_rate": 4.5900000000000004e-05, "loss": 0.4494, "step": 1271 }, { "epoch": 8.199196787148594, "grad_norm": 3.614149332046509, "learning_rate": 4.589677419354839e-05, "loss": 0.4137, "step": 1272 }, { "epoch": 8.20562248995984, "grad_norm": 4.092833042144775, "learning_rate": 4.5893548387096776e-05, "loss": 0.4403, "step": 1273 }, { "epoch": 8.212048192771084, "grad_norm": 7.710834980010986, "learning_rate": 4.589032258064516e-05, "loss": 0.3051, "step": 1274 }, { "epoch": 8.21847389558233, "grad_norm": 7.573477745056152, "learning_rate": 4.588709677419355e-05, "loss": 0.3154, "step": 1275 }, { "epoch": 8.224899598393574, "grad_norm": 2.388866901397705, "learning_rate": 4.588387096774194e-05, "loss": 0.2134, "step": 1276 }, { "epoch": 8.23132530120482, "grad_norm": 6.252061367034912, "learning_rate": 4.5880645161290325e-05, "loss": 0.2603, "step": 1277 }, { "epoch": 8.237751004016065, "grad_norm": 3.619215488433838, "learning_rate": 4.587741935483871e-05, "loss": 0.223, "step": 1278 }, { "epoch": 8.244176706827309, "grad_norm": 10.691193580627441, "learning_rate": 4.5874193548387104e-05, "loss": 0.6835, "step": 1279 }, { "epoch": 8.250602409638555, "grad_norm": 6.848972320556641, "learning_rate": 4.587096774193548e-05, "loss": 0.3248, "step": 1280 }, { "epoch": 8.257028112449799, "grad_norm": 4.87488317489624, "learning_rate": 4.586774193548387e-05, "loss": 0.3099, "step": 1281 }, { "epoch": 8.263453815261045, "grad_norm": 5.386133193969727, "learning_rate": 4.586451612903226e-05, "loss": 0.5114, "step": 1282 }, { "epoch": 8.269879518072289, "grad_norm": 7.593181133270264, "learning_rate": 4.5861290322580646e-05, "loss": 0.2405, "step": 1283 }, { "epoch": 8.276305220883534, "grad_norm": 7.469789981842041, "learning_rate": 4.585806451612903e-05, "loss": 0.4597, "step": 1284 }, { "epoch": 8.282730923694778, "grad_norm": 5.570747375488281, "learning_rate": 4.5854838709677425e-05, "loss": 0.3572, "step": 1285 }, { "epoch": 8.289156626506024, "grad_norm": 6.763558387756348, "learning_rate": 4.585161290322581e-05, "loss": 0.3443, "step": 1286 }, { "epoch": 8.29558232931727, "grad_norm": 4.042703151702881, "learning_rate": 4.5848387096774196e-05, "loss": 0.4754, "step": 1287 }, { "epoch": 8.302008032128514, "grad_norm": 11.53266716003418, "learning_rate": 4.584516129032258e-05, "loss": 0.6591, "step": 1288 }, { "epoch": 8.30843373493976, "grad_norm": 4.3847150802612305, "learning_rate": 4.584193548387097e-05, "loss": 0.2938, "step": 1289 }, { "epoch": 8.314859437751004, "grad_norm": 6.255500793457031, "learning_rate": 4.583870967741935e-05, "loss": 0.9771, "step": 1290 }, { "epoch": 8.32128514056225, "grad_norm": 7.528818607330322, "learning_rate": 4.5835483870967746e-05, "loss": 0.4456, "step": 1291 }, { "epoch": 8.327710843373493, "grad_norm": 3.6183345317840576, "learning_rate": 4.583225806451613e-05, "loss": 0.2257, "step": 1292 }, { "epoch": 8.33413654618474, "grad_norm": 7.131880760192871, "learning_rate": 4.582903225806452e-05, "loss": 0.5795, "step": 1293 }, { "epoch": 8.340562248995983, "grad_norm": 4.239846229553223, "learning_rate": 4.582580645161291e-05, "loss": 0.2638, "step": 1294 }, { "epoch": 8.346987951807229, "grad_norm": 4.688156604766846, "learning_rate": 4.5822580645161295e-05, "loss": 0.4396, "step": 1295 }, { "epoch": 8.353413654618475, "grad_norm": 11.026789665222168, "learning_rate": 4.5819354838709674e-05, "loss": 0.5168, "step": 1296 }, { "epoch": 8.359839357429719, "grad_norm": 8.635170936584473, "learning_rate": 4.581612903225807e-05, "loss": 0.581, "step": 1297 }, { "epoch": 8.366265060240965, "grad_norm": 5.219493865966797, "learning_rate": 4.581290322580645e-05, "loss": 0.3505, "step": 1298 }, { "epoch": 8.372690763052208, "grad_norm": 4.942891597747803, "learning_rate": 4.580967741935484e-05, "loss": 0.3237, "step": 1299 }, { "epoch": 8.379116465863454, "grad_norm": 5.364743232727051, "learning_rate": 4.580645161290323e-05, "loss": 0.2892, "step": 1300 }, { "epoch": 8.385542168674698, "grad_norm": 5.62924861907959, "learning_rate": 4.5803225806451616e-05, "loss": 0.3914, "step": 1301 }, { "epoch": 8.391967871485944, "grad_norm": 4.160019874572754, "learning_rate": 4.58e-05, "loss": 0.3993, "step": 1302 }, { "epoch": 8.398393574297188, "grad_norm": 16.582983016967773, "learning_rate": 4.5796774193548394e-05, "loss": 0.3553, "step": 1303 }, { "epoch": 8.404819277108434, "grad_norm": 6.817555904388428, "learning_rate": 4.579354838709677e-05, "loss": 0.2402, "step": 1304 }, { "epoch": 8.41124497991968, "grad_norm": 8.465091705322266, "learning_rate": 4.579032258064516e-05, "loss": 0.4466, "step": 1305 }, { "epoch": 8.417670682730924, "grad_norm": 7.0495758056640625, "learning_rate": 4.578709677419355e-05, "loss": 0.5975, "step": 1306 }, { "epoch": 8.42409638554217, "grad_norm": 5.077824592590332, "learning_rate": 4.578387096774194e-05, "loss": 0.3249, "step": 1307 }, { "epoch": 8.430522088353413, "grad_norm": 7.974123001098633, "learning_rate": 4.578064516129032e-05, "loss": 0.3517, "step": 1308 }, { "epoch": 8.436947791164659, "grad_norm": 3.556340456008911, "learning_rate": 4.5777419354838715e-05, "loss": 0.1254, "step": 1309 }, { "epoch": 8.443373493975903, "grad_norm": 9.841462135314941, "learning_rate": 4.57741935483871e-05, "loss": 0.5666, "step": 1310 }, { "epoch": 8.449799196787149, "grad_norm": 6.23788595199585, "learning_rate": 4.577096774193549e-05, "loss": 0.2597, "step": 1311 }, { "epoch": 8.456224899598393, "grad_norm": 13.83651065826416, "learning_rate": 4.576774193548387e-05, "loss": 0.4215, "step": 1312 }, { "epoch": 8.462650602409639, "grad_norm": 8.424846649169922, "learning_rate": 4.576451612903226e-05, "loss": 0.3405, "step": 1313 }, { "epoch": 8.469076305220884, "grad_norm": 5.275803565979004, "learning_rate": 4.5761290322580644e-05, "loss": 0.5494, "step": 1314 }, { "epoch": 8.475502008032128, "grad_norm": 9.443286895751953, "learning_rate": 4.5758064516129036e-05, "loss": 0.399, "step": 1315 }, { "epoch": 8.481927710843374, "grad_norm": 8.782391548156738, "learning_rate": 4.575483870967742e-05, "loss": 0.4672, "step": 1316 }, { "epoch": 8.488353413654618, "grad_norm": 5.9457550048828125, "learning_rate": 4.575161290322581e-05, "loss": 0.4588, "step": 1317 }, { "epoch": 8.494779116465864, "grad_norm": 7.296711444854736, "learning_rate": 4.57483870967742e-05, "loss": 0.5228, "step": 1318 }, { "epoch": 8.501204819277108, "grad_norm": 13.110780715942383, "learning_rate": 4.574516129032258e-05, "loss": 0.6295, "step": 1319 }, { "epoch": 8.507630522088354, "grad_norm": 6.948070526123047, "learning_rate": 4.5741935483870965e-05, "loss": 0.4896, "step": 1320 }, { "epoch": 8.514056224899598, "grad_norm": 5.229643821716309, "learning_rate": 4.573870967741936e-05, "loss": 0.346, "step": 1321 }, { "epoch": 8.520481927710843, "grad_norm": 12.890914916992188, "learning_rate": 4.573548387096774e-05, "loss": 0.5191, "step": 1322 }, { "epoch": 8.52690763052209, "grad_norm": 4.484952449798584, "learning_rate": 4.573225806451613e-05, "loss": 0.4345, "step": 1323 }, { "epoch": 8.533333333333333, "grad_norm": 3.367780923843384, "learning_rate": 4.572903225806452e-05, "loss": 0.1766, "step": 1324 }, { "epoch": 8.539759036144579, "grad_norm": 11.291997909545898, "learning_rate": 4.572580645161291e-05, "loss": 0.6672, "step": 1325 }, { "epoch": 8.546184738955823, "grad_norm": 7.2898406982421875, "learning_rate": 4.572258064516129e-05, "loss": 0.2436, "step": 1326 }, { "epoch": 8.552610441767069, "grad_norm": 7.985048770904541, "learning_rate": 4.571935483870968e-05, "loss": 0.4995, "step": 1327 }, { "epoch": 8.559036144578313, "grad_norm": 4.264366149902344, "learning_rate": 4.5716129032258064e-05, "loss": 0.564, "step": 1328 }, { "epoch": 8.565461847389559, "grad_norm": 5.963706016540527, "learning_rate": 4.571290322580645e-05, "loss": 0.6515, "step": 1329 }, { "epoch": 8.571887550200803, "grad_norm": 8.371935844421387, "learning_rate": 4.570967741935484e-05, "loss": 0.4514, "step": 1330 }, { "epoch": 8.578313253012048, "grad_norm": 8.411486625671387, "learning_rate": 4.570645161290323e-05, "loss": 0.4048, "step": 1331 }, { "epoch": 8.584738955823294, "grad_norm": 5.376402378082275, "learning_rate": 4.5703225806451614e-05, "loss": 0.2632, "step": 1332 }, { "epoch": 8.591164658634538, "grad_norm": 4.147646427154541, "learning_rate": 4.5700000000000006e-05, "loss": 0.4636, "step": 1333 }, { "epoch": 8.597590361445784, "grad_norm": 12.988231658935547, "learning_rate": 4.569677419354839e-05, "loss": 0.6504, "step": 1334 }, { "epoch": 8.604016064257028, "grad_norm": 8.679765701293945, "learning_rate": 4.569354838709677e-05, "loss": 0.5836, "step": 1335 }, { "epoch": 8.610441767068274, "grad_norm": 7.938448429107666, "learning_rate": 4.5690322580645163e-05, "loss": 0.5115, "step": 1336 }, { "epoch": 8.616867469879518, "grad_norm": 10.852405548095703, "learning_rate": 4.568709677419355e-05, "loss": 0.3914, "step": 1337 }, { "epoch": 8.623293172690763, "grad_norm": 16.549203872680664, "learning_rate": 4.5683870967741935e-05, "loss": 0.8266, "step": 1338 }, { "epoch": 8.629718875502007, "grad_norm": 5.642305850982666, "learning_rate": 4.568064516129033e-05, "loss": 0.3582, "step": 1339 }, { "epoch": 8.636144578313253, "grad_norm": 15.60041332244873, "learning_rate": 4.567741935483871e-05, "loss": 0.8917, "step": 1340 }, { "epoch": 8.642570281124499, "grad_norm": 8.512679100036621, "learning_rate": 4.56741935483871e-05, "loss": 0.4689, "step": 1341 }, { "epoch": 8.648995983935743, "grad_norm": 4.858104228973389, "learning_rate": 4.567096774193549e-05, "loss": 0.2192, "step": 1342 }, { "epoch": 8.655421686746989, "grad_norm": 8.511239051818848, "learning_rate": 4.566774193548387e-05, "loss": 0.5867, "step": 1343 }, { "epoch": 8.661847389558233, "grad_norm": 7.781827449798584, "learning_rate": 4.5664516129032256e-05, "loss": 0.3366, "step": 1344 }, { "epoch": 8.668273092369478, "grad_norm": 10.318175315856934, "learning_rate": 4.566129032258065e-05, "loss": 0.4359, "step": 1345 }, { "epoch": 8.674698795180722, "grad_norm": 6.1180620193481445, "learning_rate": 4.5658064516129034e-05, "loss": 0.575, "step": 1346 }, { "epoch": 8.681124497991968, "grad_norm": 3.30202054977417, "learning_rate": 4.565483870967742e-05, "loss": 0.1418, "step": 1347 }, { "epoch": 8.687550200803212, "grad_norm": 4.921051979064941, "learning_rate": 4.565161290322581e-05, "loss": 0.1849, "step": 1348 }, { "epoch": 8.693975903614458, "grad_norm": 8.656163215637207, "learning_rate": 4.56483870967742e-05, "loss": 0.6562, "step": 1349 }, { "epoch": 8.700401606425704, "grad_norm": 5.05127477645874, "learning_rate": 4.5645161290322584e-05, "loss": 0.3084, "step": 1350 }, { "epoch": 8.706827309236948, "grad_norm": 10.924267768859863, "learning_rate": 4.564193548387097e-05, "loss": 0.6478, "step": 1351 }, { "epoch": 8.713253012048193, "grad_norm": 11.963493347167969, "learning_rate": 4.5638709677419355e-05, "loss": 0.64, "step": 1352 }, { "epoch": 8.719678714859437, "grad_norm": 6.2881035804748535, "learning_rate": 4.563548387096774e-05, "loss": 0.3408, "step": 1353 }, { "epoch": 8.726104417670683, "grad_norm": 11.18084716796875, "learning_rate": 4.563225806451613e-05, "loss": 0.5047, "step": 1354 }, { "epoch": 8.732530120481927, "grad_norm": 7.131741046905518, "learning_rate": 4.562903225806452e-05, "loss": 0.5075, "step": 1355 }, { "epoch": 8.738955823293173, "grad_norm": 7.09971284866333, "learning_rate": 4.5625806451612905e-05, "loss": 0.1961, "step": 1356 }, { "epoch": 8.745381526104417, "grad_norm": 10.181737899780273, "learning_rate": 4.56225806451613e-05, "loss": 0.6394, "step": 1357 }, { "epoch": 8.751807228915663, "grad_norm": 4.7584028244018555, "learning_rate": 4.5619354838709676e-05, "loss": 0.286, "step": 1358 }, { "epoch": 8.758232931726909, "grad_norm": 4.792050361633301, "learning_rate": 4.561612903225806e-05, "loss": 0.3544, "step": 1359 }, { "epoch": 8.764658634538153, "grad_norm": 5.042547225952148, "learning_rate": 4.5612903225806454e-05, "loss": 0.285, "step": 1360 }, { "epoch": 8.771084337349398, "grad_norm": 6.173068523406982, "learning_rate": 4.560967741935484e-05, "loss": 0.4666, "step": 1361 }, { "epoch": 8.777510040160642, "grad_norm": 7.480648040771484, "learning_rate": 4.5606451612903226e-05, "loss": 0.39, "step": 1362 }, { "epoch": 8.783935742971888, "grad_norm": 4.661794662475586, "learning_rate": 4.560322580645162e-05, "loss": 0.2857, "step": 1363 }, { "epoch": 8.790361445783132, "grad_norm": 5.550902366638184, "learning_rate": 4.5600000000000004e-05, "loss": 0.345, "step": 1364 }, { "epoch": 8.796787148594378, "grad_norm": 7.190758228302002, "learning_rate": 4.559677419354839e-05, "loss": 0.3002, "step": 1365 }, { "epoch": 8.803212851405622, "grad_norm": 20.205678939819336, "learning_rate": 4.5593548387096775e-05, "loss": 0.5025, "step": 1366 }, { "epoch": 8.809638554216868, "grad_norm": 6.496103286743164, "learning_rate": 4.559032258064516e-05, "loss": 0.3814, "step": 1367 }, { "epoch": 8.816064257028113, "grad_norm": 6.2158589363098145, "learning_rate": 4.558709677419355e-05, "loss": 0.2216, "step": 1368 }, { "epoch": 8.822489959839357, "grad_norm": 10.625479698181152, "learning_rate": 4.558387096774194e-05, "loss": 0.4985, "step": 1369 }, { "epoch": 8.828915662650603, "grad_norm": 5.845732688903809, "learning_rate": 4.5580645161290325e-05, "loss": 0.3285, "step": 1370 }, { "epoch": 8.835341365461847, "grad_norm": 7.395769119262695, "learning_rate": 4.557741935483871e-05, "loss": 0.4855, "step": 1371 }, { "epoch": 8.841767068273093, "grad_norm": 4.568187236785889, "learning_rate": 4.55741935483871e-05, "loss": 0.3849, "step": 1372 }, { "epoch": 8.848192771084337, "grad_norm": 4.4611358642578125, "learning_rate": 4.557096774193549e-05, "loss": 0.2568, "step": 1373 }, { "epoch": 8.854618473895583, "grad_norm": 4.3644819259643555, "learning_rate": 4.556774193548387e-05, "loss": 0.2808, "step": 1374 }, { "epoch": 8.861044176706827, "grad_norm": 7.493407726287842, "learning_rate": 4.556451612903226e-05, "loss": 0.7846, "step": 1375 }, { "epoch": 8.867469879518072, "grad_norm": 3.36728835105896, "learning_rate": 4.5561290322580646e-05, "loss": 0.1986, "step": 1376 }, { "epoch": 8.873895582329318, "grad_norm": 5.342142105102539, "learning_rate": 4.555806451612903e-05, "loss": 0.5716, "step": 1377 }, { "epoch": 8.880321285140562, "grad_norm": 8.292991638183594, "learning_rate": 4.5554838709677424e-05, "loss": 0.4529, "step": 1378 }, { "epoch": 8.886746987951808, "grad_norm": 3.6188204288482666, "learning_rate": 4.555161290322581e-05, "loss": 0.1646, "step": 1379 }, { "epoch": 8.893172690763052, "grad_norm": 7.484078884124756, "learning_rate": 4.5548387096774196e-05, "loss": 0.4131, "step": 1380 }, { "epoch": 8.899598393574298, "grad_norm": 6.831184387207031, "learning_rate": 4.554516129032259e-05, "loss": 0.3266, "step": 1381 }, { "epoch": 8.906024096385542, "grad_norm": 4.717890739440918, "learning_rate": 4.554193548387097e-05, "loss": 0.3974, "step": 1382 }, { "epoch": 8.912449799196787, "grad_norm": 7.997795104980469, "learning_rate": 4.553870967741935e-05, "loss": 0.2223, "step": 1383 }, { "epoch": 8.918875502008031, "grad_norm": 8.702207565307617, "learning_rate": 4.5535483870967745e-05, "loss": 0.7058, "step": 1384 }, { "epoch": 8.925301204819277, "grad_norm": 7.458806037902832, "learning_rate": 4.553225806451613e-05, "loss": 0.3556, "step": 1385 }, { "epoch": 8.931726907630523, "grad_norm": 4.79680061340332, "learning_rate": 4.5529032258064517e-05, "loss": 0.3479, "step": 1386 }, { "epoch": 8.938152610441767, "grad_norm": 10.583337783813477, "learning_rate": 4.552580645161291e-05, "loss": 0.4089, "step": 1387 }, { "epoch": 8.944578313253013, "grad_norm": 4.816627502441406, "learning_rate": 4.5522580645161295e-05, "loss": 0.2753, "step": 1388 }, { "epoch": 8.951004016064257, "grad_norm": 5.8597846031188965, "learning_rate": 4.5519354838709674e-05, "loss": 0.2724, "step": 1389 }, { "epoch": 8.957429718875503, "grad_norm": 6.713744163513184, "learning_rate": 4.5516129032258066e-05, "loss": 0.4668, "step": 1390 }, { "epoch": 8.963855421686747, "grad_norm": 4.525312423706055, "learning_rate": 4.551290322580645e-05, "loss": 0.3257, "step": 1391 }, { "epoch": 8.970281124497992, "grad_norm": 13.91582202911377, "learning_rate": 4.550967741935484e-05, "loss": 0.5058, "step": 1392 }, { "epoch": 8.976706827309236, "grad_norm": 6.611674785614014, "learning_rate": 4.550645161290323e-05, "loss": 0.5051, "step": 1393 }, { "epoch": 8.983132530120482, "grad_norm": 5.346808910369873, "learning_rate": 4.5503225806451616e-05, "loss": 0.3981, "step": 1394 }, { "epoch": 8.989558232931728, "grad_norm": 9.2650146484375, "learning_rate": 4.55e-05, "loss": 0.6203, "step": 1395 }, { "epoch": 8.995983935742972, "grad_norm": 4.470982551574707, "learning_rate": 4.5496774193548394e-05, "loss": 0.2106, "step": 1396 }, { "epoch": 9.005622489959839, "grad_norm": 1.9410216808319092, "learning_rate": 4.549354838709677e-05, "loss": 0.2095, "step": 1397 }, { "epoch": 9.012048192771084, "grad_norm": 7.938120365142822, "learning_rate": 4.549032258064516e-05, "loss": 0.2936, "step": 1398 }, { "epoch": 9.018473895582328, "grad_norm": 8.77293872833252, "learning_rate": 4.548709677419355e-05, "loss": 0.5181, "step": 1399 }, { "epoch": 9.024899598393574, "grad_norm": 9.190606117248535, "learning_rate": 4.548387096774194e-05, "loss": 0.4141, "step": 1400 }, { "epoch": 9.03132530120482, "grad_norm": 8.53229808807373, "learning_rate": 4.548064516129032e-05, "loss": 0.4626, "step": 1401 }, { "epoch": 9.037751004016064, "grad_norm": 1.8463751077651978, "learning_rate": 4.5477419354838715e-05, "loss": 0.3022, "step": 1402 }, { "epoch": 9.04417670682731, "grad_norm": 8.152097702026367, "learning_rate": 4.54741935483871e-05, "loss": 0.2773, "step": 1403 }, { "epoch": 9.050602409638554, "grad_norm": 14.88347053527832, "learning_rate": 4.5470967741935486e-05, "loss": 0.2806, "step": 1404 }, { "epoch": 9.0570281124498, "grad_norm": 3.7201943397521973, "learning_rate": 4.546774193548387e-05, "loss": 0.3196, "step": 1405 }, { "epoch": 9.063453815261044, "grad_norm": 10.704480171203613, "learning_rate": 4.546451612903226e-05, "loss": 0.3609, "step": 1406 }, { "epoch": 9.06987951807229, "grad_norm": 14.864017486572266, "learning_rate": 4.5461290322580643e-05, "loss": 0.2436, "step": 1407 }, { "epoch": 9.076305220883533, "grad_norm": 4.512131214141846, "learning_rate": 4.5458064516129036e-05, "loss": 0.4736, "step": 1408 }, { "epoch": 9.082730923694779, "grad_norm": 2.009955644607544, "learning_rate": 4.545483870967742e-05, "loss": 0.2442, "step": 1409 }, { "epoch": 9.089156626506025, "grad_norm": 11.114818572998047, "learning_rate": 4.545161290322581e-05, "loss": 0.5403, "step": 1410 }, { "epoch": 9.095582329317269, "grad_norm": 32.16618347167969, "learning_rate": 4.54483870967742e-05, "loss": 0.5536, "step": 1411 }, { "epoch": 9.102008032128515, "grad_norm": 32.25920867919922, "learning_rate": 4.5445161290322586e-05, "loss": 0.6179, "step": 1412 }, { "epoch": 9.108433734939759, "grad_norm": 9.933538436889648, "learning_rate": 4.5441935483870964e-05, "loss": 0.5515, "step": 1413 }, { "epoch": 9.114859437751004, "grad_norm": 5.334659099578857, "learning_rate": 4.543870967741936e-05, "loss": 0.0618, "step": 1414 }, { "epoch": 9.121285140562248, "grad_norm": 18.491975784301758, "learning_rate": 4.543548387096774e-05, "loss": 0.4416, "step": 1415 }, { "epoch": 9.127710843373494, "grad_norm": 7.175950527191162, "learning_rate": 4.543225806451613e-05, "loss": 0.8164, "step": 1416 }, { "epoch": 9.134136546184738, "grad_norm": 37.60507583618164, "learning_rate": 4.542903225806452e-05, "loss": 0.3413, "step": 1417 }, { "epoch": 9.140562248995984, "grad_norm": 5.38280725479126, "learning_rate": 4.5425806451612907e-05, "loss": 0.2487, "step": 1418 }, { "epoch": 9.14698795180723, "grad_norm": 33.432804107666016, "learning_rate": 4.542258064516129e-05, "loss": 0.5312, "step": 1419 }, { "epoch": 9.153413654618474, "grad_norm": 5.003303050994873, "learning_rate": 4.5419354838709685e-05, "loss": 0.2569, "step": 1420 }, { "epoch": 9.15983935742972, "grad_norm": 33.64423751831055, "learning_rate": 4.5416129032258064e-05, "loss": 0.4825, "step": 1421 }, { "epoch": 9.166265060240963, "grad_norm": 19.414777755737305, "learning_rate": 4.541290322580645e-05, "loss": 0.5995, "step": 1422 }, { "epoch": 9.17269076305221, "grad_norm": 13.412867546081543, "learning_rate": 4.540967741935484e-05, "loss": 0.5242, "step": 1423 }, { "epoch": 9.179116465863453, "grad_norm": 8.119392395019531, "learning_rate": 4.540645161290323e-05, "loss": 0.2552, "step": 1424 }, { "epoch": 9.185542168674699, "grad_norm": 5.35421895980835, "learning_rate": 4.540322580645161e-05, "loss": 0.4537, "step": 1425 }, { "epoch": 9.191967871485943, "grad_norm": 27.406110763549805, "learning_rate": 4.5400000000000006e-05, "loss": 0.5339, "step": 1426 }, { "epoch": 9.198393574297189, "grad_norm": 13.932709693908691, "learning_rate": 4.539677419354839e-05, "loss": 0.5072, "step": 1427 }, { "epoch": 9.204819277108435, "grad_norm": 0.7756669521331787, "learning_rate": 4.539354838709677e-05, "loss": 0.12, "step": 1428 }, { "epoch": 9.211244979919678, "grad_norm": 19.701335906982422, "learning_rate": 4.539032258064516e-05, "loss": 0.6143, "step": 1429 }, { "epoch": 9.217670682730924, "grad_norm": 15.741942405700684, "learning_rate": 4.538709677419355e-05, "loss": 0.3669, "step": 1430 }, { "epoch": 9.224096385542168, "grad_norm": 3.6649937629699707, "learning_rate": 4.5383870967741934e-05, "loss": 0.4748, "step": 1431 }, { "epoch": 9.230522088353414, "grad_norm": 11.162946701049805, "learning_rate": 4.538064516129033e-05, "loss": 0.4833, "step": 1432 }, { "epoch": 9.236947791164658, "grad_norm": 24.04448127746582, "learning_rate": 4.537741935483871e-05, "loss": 0.4426, "step": 1433 }, { "epoch": 9.243373493975904, "grad_norm": 10.35733699798584, "learning_rate": 4.53741935483871e-05, "loss": 0.4045, "step": 1434 }, { "epoch": 9.249799196787148, "grad_norm": 28.79336929321289, "learning_rate": 4.537096774193549e-05, "loss": 0.3725, "step": 1435 }, { "epoch": 9.256224899598394, "grad_norm": 7.7517218589782715, "learning_rate": 4.536774193548387e-05, "loss": 0.3884, "step": 1436 }, { "epoch": 9.26265060240964, "grad_norm": 13.574085235595703, "learning_rate": 4.5364516129032255e-05, "loss": 0.3909, "step": 1437 }, { "epoch": 9.269076305220883, "grad_norm": 5.496007442474365, "learning_rate": 4.536129032258065e-05, "loss": 0.5945, "step": 1438 }, { "epoch": 9.275502008032129, "grad_norm": 18.014028549194336, "learning_rate": 4.5358064516129034e-05, "loss": 0.2869, "step": 1439 }, { "epoch": 9.281927710843373, "grad_norm": 10.969353675842285, "learning_rate": 4.535483870967742e-05, "loss": 0.4361, "step": 1440 }, { "epoch": 9.288353413654619, "grad_norm": 10.547163963317871, "learning_rate": 4.535161290322581e-05, "loss": 0.5421, "step": 1441 }, { "epoch": 9.294779116465863, "grad_norm": 16.08407974243164, "learning_rate": 4.53483870967742e-05, "loss": 0.2742, "step": 1442 }, { "epoch": 9.301204819277109, "grad_norm": 7.539339542388916, "learning_rate": 4.534516129032258e-05, "loss": 0.4507, "step": 1443 }, { "epoch": 9.307630522088353, "grad_norm": 12.213764190673828, "learning_rate": 4.534193548387097e-05, "loss": 0.3263, "step": 1444 }, { "epoch": 9.314056224899598, "grad_norm": 10.641971588134766, "learning_rate": 4.5338709677419355e-05, "loss": 0.4767, "step": 1445 }, { "epoch": 9.320481927710844, "grad_norm": 45.33008575439453, "learning_rate": 4.533548387096774e-05, "loss": 0.9708, "step": 1446 }, { "epoch": 9.326907630522088, "grad_norm": 10.348326683044434, "learning_rate": 4.533225806451613e-05, "loss": 0.4217, "step": 1447 }, { "epoch": 9.333333333333334, "grad_norm": 24.776920318603516, "learning_rate": 4.532903225806452e-05, "loss": 0.2593, "step": 1448 }, { "epoch": 9.339759036144578, "grad_norm": 9.360572814941406, "learning_rate": 4.5325806451612904e-05, "loss": 0.2895, "step": 1449 }, { "epoch": 9.346184738955824, "grad_norm": 6.996976375579834, "learning_rate": 4.53225806451613e-05, "loss": 0.5049, "step": 1450 }, { "epoch": 9.352610441767068, "grad_norm": 8.362908363342285, "learning_rate": 4.531935483870968e-05, "loss": 0.3066, "step": 1451 }, { "epoch": 9.359036144578313, "grad_norm": 16.614484786987305, "learning_rate": 4.531612903225806e-05, "loss": 0.3964, "step": 1452 }, { "epoch": 9.365461847389557, "grad_norm": 21.746784210205078, "learning_rate": 4.5312903225806454e-05, "loss": 0.2555, "step": 1453 }, { "epoch": 9.371887550200803, "grad_norm": 12.890558242797852, "learning_rate": 4.530967741935484e-05, "loss": 0.5909, "step": 1454 }, { "epoch": 9.378313253012049, "grad_norm": 6.1142377853393555, "learning_rate": 4.5306451612903225e-05, "loss": 0.47, "step": 1455 }, { "epoch": 9.384738955823293, "grad_norm": 14.484336853027344, "learning_rate": 4.530322580645162e-05, "loss": 0.4949, "step": 1456 }, { "epoch": 9.391164658634539, "grad_norm": 3.5389761924743652, "learning_rate": 4.53e-05, "loss": 0.318, "step": 1457 }, { "epoch": 9.397590361445783, "grad_norm": 8.59897518157959, "learning_rate": 4.529677419354839e-05, "loss": 0.3083, "step": 1458 }, { "epoch": 9.404016064257029, "grad_norm": 10.655777931213379, "learning_rate": 4.5293548387096775e-05, "loss": 0.3825, "step": 1459 }, { "epoch": 9.410441767068273, "grad_norm": 12.249585151672363, "learning_rate": 4.529032258064516e-05, "loss": 0.3095, "step": 1460 }, { "epoch": 9.416867469879518, "grad_norm": 4.206648349761963, "learning_rate": 4.5287096774193546e-05, "loss": 0.3586, "step": 1461 }, { "epoch": 9.423293172690762, "grad_norm": 17.976228713989258, "learning_rate": 4.528387096774194e-05, "loss": 0.4433, "step": 1462 }, { "epoch": 9.429718875502008, "grad_norm": 6.7105817794799805, "learning_rate": 4.5280645161290324e-05, "loss": 0.3914, "step": 1463 }, { "epoch": 9.436144578313254, "grad_norm": 3.310899257659912, "learning_rate": 4.527741935483871e-05, "loss": 0.3529, "step": 1464 }, { "epoch": 9.442570281124498, "grad_norm": 17.005163192749023, "learning_rate": 4.52741935483871e-05, "loss": 0.492, "step": 1465 }, { "epoch": 9.448995983935744, "grad_norm": 35.71643829345703, "learning_rate": 4.527096774193549e-05, "loss": 0.8791, "step": 1466 }, { "epoch": 9.455421686746988, "grad_norm": 8.366581916809082, "learning_rate": 4.526774193548387e-05, "loss": 0.3732, "step": 1467 }, { "epoch": 9.461847389558233, "grad_norm": 7.859811782836914, "learning_rate": 4.526451612903226e-05, "loss": 0.3897, "step": 1468 }, { "epoch": 9.468273092369477, "grad_norm": 24.72102928161621, "learning_rate": 4.5261290322580645e-05, "loss": 0.3959, "step": 1469 }, { "epoch": 9.474698795180723, "grad_norm": 17.30596351623535, "learning_rate": 4.525806451612903e-05, "loss": 0.6015, "step": 1470 }, { "epoch": 9.481124497991967, "grad_norm": 12.554073333740234, "learning_rate": 4.5254838709677424e-05, "loss": 0.3994, "step": 1471 }, { "epoch": 9.487550200803213, "grad_norm": 18.320695877075195, "learning_rate": 4.525161290322581e-05, "loss": 0.3959, "step": 1472 }, { "epoch": 9.493975903614459, "grad_norm": 16.06900405883789, "learning_rate": 4.5248387096774195e-05, "loss": 0.353, "step": 1473 }, { "epoch": 9.500401606425703, "grad_norm": 7.745049953460693, "learning_rate": 4.524516129032259e-05, "loss": 0.4258, "step": 1474 }, { "epoch": 9.506827309236948, "grad_norm": 6.286248207092285, "learning_rate": 4.5241935483870966e-05, "loss": 0.2644, "step": 1475 }, { "epoch": 9.513253012048192, "grad_norm": 7.657230854034424, "learning_rate": 4.523870967741935e-05, "loss": 0.3751, "step": 1476 }, { "epoch": 9.519678714859438, "grad_norm": 8.163783073425293, "learning_rate": 4.5235483870967745e-05, "loss": 0.3556, "step": 1477 }, { "epoch": 9.526104417670682, "grad_norm": 7.423781394958496, "learning_rate": 4.523225806451613e-05, "loss": 0.4998, "step": 1478 }, { "epoch": 9.532530120481928, "grad_norm": 19.012521743774414, "learning_rate": 4.5229032258064516e-05, "loss": 0.4135, "step": 1479 }, { "epoch": 9.538955823293172, "grad_norm": 14.886188507080078, "learning_rate": 4.522580645161291e-05, "loss": 0.4044, "step": 1480 }, { "epoch": 9.545381526104418, "grad_norm": 4.470116138458252, "learning_rate": 4.5222580645161294e-05, "loss": 0.4792, "step": 1481 }, { "epoch": 9.551807228915663, "grad_norm": 12.418229103088379, "learning_rate": 4.521935483870968e-05, "loss": 0.2616, "step": 1482 }, { "epoch": 9.558232931726907, "grad_norm": 7.379891872406006, "learning_rate": 4.5216129032258066e-05, "loss": 0.3872, "step": 1483 }, { "epoch": 9.564658634538153, "grad_norm": 11.238941192626953, "learning_rate": 4.521290322580645e-05, "loss": 0.2418, "step": 1484 }, { "epoch": 9.571084337349397, "grad_norm": 9.3504638671875, "learning_rate": 4.520967741935484e-05, "loss": 0.7679, "step": 1485 }, { "epoch": 9.577510040160643, "grad_norm": 11.91142463684082, "learning_rate": 4.520645161290323e-05, "loss": 0.3536, "step": 1486 }, { "epoch": 9.583935742971887, "grad_norm": 3.932453155517578, "learning_rate": 4.5203225806451615e-05, "loss": 0.1503, "step": 1487 }, { "epoch": 9.590361445783133, "grad_norm": 12.319634437561035, "learning_rate": 4.52e-05, "loss": 0.3891, "step": 1488 }, { "epoch": 9.596787148594377, "grad_norm": 9.653397560119629, "learning_rate": 4.5196774193548393e-05, "loss": 0.4675, "step": 1489 }, { "epoch": 9.603212851405623, "grad_norm": 41.05164337158203, "learning_rate": 4.519354838709678e-05, "loss": 0.5645, "step": 1490 }, { "epoch": 9.609638554216868, "grad_norm": 9.346763610839844, "learning_rate": 4.519032258064516e-05, "loss": 0.4786, "step": 1491 }, { "epoch": 9.616064257028112, "grad_norm": 6.372498989105225, "learning_rate": 4.518709677419355e-05, "loss": 0.2862, "step": 1492 }, { "epoch": 9.622489959839358, "grad_norm": 11.160929679870605, "learning_rate": 4.5183870967741936e-05, "loss": 0.7186, "step": 1493 }, { "epoch": 9.628915662650602, "grad_norm": 8.773771286010742, "learning_rate": 4.518064516129032e-05, "loss": 0.2738, "step": 1494 }, { "epoch": 9.635341365461848, "grad_norm": 8.48831558227539, "learning_rate": 4.5177419354838714e-05, "loss": 0.5258, "step": 1495 }, { "epoch": 9.641767068273092, "grad_norm": 6.790502071380615, "learning_rate": 4.51741935483871e-05, "loss": 0.3827, "step": 1496 }, { "epoch": 9.648192771084338, "grad_norm": 29.611186981201172, "learning_rate": 4.5170967741935486e-05, "loss": 0.6451, "step": 1497 }, { "epoch": 9.654618473895582, "grad_norm": 11.71569538116455, "learning_rate": 4.516774193548387e-05, "loss": 0.3467, "step": 1498 }, { "epoch": 9.661044176706827, "grad_norm": 9.01624584197998, "learning_rate": 4.516451612903226e-05, "loss": 0.3589, "step": 1499 }, { "epoch": 9.667469879518073, "grad_norm": 17.688636779785156, "learning_rate": 4.516129032258064e-05, "loss": 0.4505, "step": 1500 }, { "epoch": 9.673895582329317, "grad_norm": 29.656827926635742, "learning_rate": 4.5158064516129035e-05, "loss": 0.686, "step": 1501 }, { "epoch": 9.680321285140563, "grad_norm": 44.269325256347656, "learning_rate": 4.515483870967742e-05, "loss": 0.4059, "step": 1502 }, { "epoch": 9.686746987951807, "grad_norm": 5.8450188636779785, "learning_rate": 4.515161290322581e-05, "loss": 0.7359, "step": 1503 }, { "epoch": 9.693172690763053, "grad_norm": 1.6970384120941162, "learning_rate": 4.51483870967742e-05, "loss": 0.3278, "step": 1504 }, { "epoch": 9.699598393574297, "grad_norm": 3.48824143409729, "learning_rate": 4.5145161290322585e-05, "loss": 0.3072, "step": 1505 }, { "epoch": 9.706024096385542, "grad_norm": 8.775885581970215, "learning_rate": 4.514193548387097e-05, "loss": 0.7572, "step": 1506 }, { "epoch": 9.712449799196786, "grad_norm": 13.283108711242676, "learning_rate": 4.5138709677419356e-05, "loss": 0.5321, "step": 1507 }, { "epoch": 9.718875502008032, "grad_norm": 19.482812881469727, "learning_rate": 4.513548387096774e-05, "loss": 0.3812, "step": 1508 }, { "epoch": 9.725301204819278, "grad_norm": 18.184694290161133, "learning_rate": 4.513225806451613e-05, "loss": 0.3281, "step": 1509 }, { "epoch": 9.731726907630522, "grad_norm": 7.270438194274902, "learning_rate": 4.512903225806452e-05, "loss": 0.5539, "step": 1510 }, { "epoch": 9.738152610441768, "grad_norm": 4.379840850830078, "learning_rate": 4.5125806451612906e-05, "loss": 0.4606, "step": 1511 }, { "epoch": 9.744578313253012, "grad_norm": 10.527506828308105, "learning_rate": 4.512258064516129e-05, "loss": 0.2912, "step": 1512 }, { "epoch": 9.751004016064257, "grad_norm": 12.502981185913086, "learning_rate": 4.5119354838709684e-05, "loss": 0.4716, "step": 1513 }, { "epoch": 9.757429718875501, "grad_norm": 9.790298461914062, "learning_rate": 4.511612903225806e-05, "loss": 0.3807, "step": 1514 }, { "epoch": 9.763855421686747, "grad_norm": 4.580970287322998, "learning_rate": 4.511290322580645e-05, "loss": 0.2146, "step": 1515 }, { "epoch": 9.770281124497991, "grad_norm": 5.978707313537598, "learning_rate": 4.510967741935484e-05, "loss": 0.3623, "step": 1516 }, { "epoch": 9.776706827309237, "grad_norm": 66.40177917480469, "learning_rate": 4.510645161290323e-05, "loss": 0.38, "step": 1517 }, { "epoch": 9.783132530120483, "grad_norm": 5.668610095977783, "learning_rate": 4.510322580645161e-05, "loss": 0.469, "step": 1518 }, { "epoch": 9.789558232931727, "grad_norm": 3.8342978954315186, "learning_rate": 4.5100000000000005e-05, "loss": 0.3987, "step": 1519 }, { "epoch": 9.795983935742973, "grad_norm": 25.99637794494629, "learning_rate": 4.509677419354839e-05, "loss": 0.4895, "step": 1520 }, { "epoch": 9.802409638554217, "grad_norm": 14.886703491210938, "learning_rate": 4.509354838709678e-05, "loss": 0.8216, "step": 1521 }, { "epoch": 9.808835341365462, "grad_norm": 3.530921697616577, "learning_rate": 4.509032258064516e-05, "loss": 0.316, "step": 1522 }, { "epoch": 9.815261044176706, "grad_norm": 10.960981369018555, "learning_rate": 4.508709677419355e-05, "loss": 0.5246, "step": 1523 }, { "epoch": 9.821686746987952, "grad_norm": 4.918986797332764, "learning_rate": 4.5083870967741934e-05, "loss": 0.3783, "step": 1524 }, { "epoch": 9.828112449799196, "grad_norm": 13.406768798828125, "learning_rate": 4.5080645161290326e-05, "loss": 0.6157, "step": 1525 }, { "epoch": 9.834538152610442, "grad_norm": 10.468243598937988, "learning_rate": 4.507741935483871e-05, "loss": 0.3428, "step": 1526 }, { "epoch": 9.840963855421688, "grad_norm": 2.818289041519165, "learning_rate": 4.50741935483871e-05, "loss": 0.1801, "step": 1527 }, { "epoch": 9.847389558232932, "grad_norm": 15.121238708496094, "learning_rate": 4.507096774193549e-05, "loss": 0.5732, "step": 1528 }, { "epoch": 9.853815261044177, "grad_norm": 13.7129545211792, "learning_rate": 4.506774193548387e-05, "loss": 0.4836, "step": 1529 }, { "epoch": 9.860240963855421, "grad_norm": 31.439464569091797, "learning_rate": 4.5064516129032255e-05, "loss": 0.4868, "step": 1530 }, { "epoch": 9.866666666666667, "grad_norm": 8.375990867614746, "learning_rate": 4.506129032258065e-05, "loss": 0.2676, "step": 1531 }, { "epoch": 9.873092369477911, "grad_norm": 20.29941749572754, "learning_rate": 4.505806451612903e-05, "loss": 0.3903, "step": 1532 }, { "epoch": 9.879518072289157, "grad_norm": 33.0708122253418, "learning_rate": 4.505483870967742e-05, "loss": 0.6485, "step": 1533 }, { "epoch": 9.885943775100401, "grad_norm": 10.073287963867188, "learning_rate": 4.505161290322581e-05, "loss": 0.3592, "step": 1534 }, { "epoch": 9.892369477911647, "grad_norm": 16.438465118408203, "learning_rate": 4.50483870967742e-05, "loss": 0.3547, "step": 1535 }, { "epoch": 9.898795180722892, "grad_norm": 3.2824082374572754, "learning_rate": 4.504516129032258e-05, "loss": 0.3545, "step": 1536 }, { "epoch": 9.905220883534136, "grad_norm": 2.031538963317871, "learning_rate": 4.504193548387097e-05, "loss": 0.2928, "step": 1537 }, { "epoch": 9.911646586345382, "grad_norm": 13.096834182739258, "learning_rate": 4.5038709677419354e-05, "loss": 0.6059, "step": 1538 }, { "epoch": 9.918072289156626, "grad_norm": 3.0029542446136475, "learning_rate": 4.503548387096774e-05, "loss": 0.3311, "step": 1539 }, { "epoch": 9.924497991967872, "grad_norm": 3.29683256149292, "learning_rate": 4.503225806451613e-05, "loss": 0.3339, "step": 1540 }, { "epoch": 9.930923694779116, "grad_norm": 17.15999984741211, "learning_rate": 4.502903225806452e-05, "loss": 0.2604, "step": 1541 }, { "epoch": 9.937349397590362, "grad_norm": 2.5838394165039062, "learning_rate": 4.5025806451612904e-05, "loss": 0.4298, "step": 1542 }, { "epoch": 9.943775100401606, "grad_norm": 18.798046112060547, "learning_rate": 4.5022580645161296e-05, "loss": 0.7753, "step": 1543 }, { "epoch": 9.950200803212851, "grad_norm": 23.990774154663086, "learning_rate": 4.501935483870968e-05, "loss": 0.3644, "step": 1544 }, { "epoch": 9.956626506024097, "grad_norm": 14.848284721374512, "learning_rate": 4.501612903225807e-05, "loss": 0.5193, "step": 1545 }, { "epoch": 9.963052208835341, "grad_norm": 14.8323335647583, "learning_rate": 4.501290322580645e-05, "loss": 0.5107, "step": 1546 }, { "epoch": 9.969477911646587, "grad_norm": 14.363834381103516, "learning_rate": 4.500967741935484e-05, "loss": 0.6711, "step": 1547 }, { "epoch": 9.975903614457831, "grad_norm": 4.327536582946777, "learning_rate": 4.5006451612903225e-05, "loss": 0.1834, "step": 1548 }, { "epoch": 9.982329317269077, "grad_norm": 10.951763153076172, "learning_rate": 4.500322580645162e-05, "loss": 0.4192, "step": 1549 }, { "epoch": 9.98875502008032, "grad_norm": 5.590513229370117, "learning_rate": 4.5e-05, "loss": 0.2513, "step": 1550 }, { "epoch": 9.995180722891567, "grad_norm": 9.35864543914795, "learning_rate": 4.499677419354839e-05, "loss": 0.3885, "step": 1551 }, { "epoch": 10.004819277108433, "grad_norm": 20.261734008789062, "learning_rate": 4.499354838709678e-05, "loss": 0.2927, "step": 1552 }, { "epoch": 10.01124497991968, "grad_norm": 3.7440407276153564, "learning_rate": 4.499032258064516e-05, "loss": 0.6654, "step": 1553 }, { "epoch": 10.017670682730923, "grad_norm": 8.796720504760742, "learning_rate": 4.4987096774193546e-05, "loss": 0.2531, "step": 1554 }, { "epoch": 10.024096385542169, "grad_norm": 15.8095703125, "learning_rate": 4.498387096774194e-05, "loss": 1.0022, "step": 1555 }, { "epoch": 10.030522088353413, "grad_norm": 9.554750442504883, "learning_rate": 4.4980645161290324e-05, "loss": 0.3392, "step": 1556 }, { "epoch": 10.036947791164659, "grad_norm": 20.865909576416016, "learning_rate": 4.497741935483871e-05, "loss": 0.464, "step": 1557 }, { "epoch": 10.043373493975903, "grad_norm": 7.241203308105469, "learning_rate": 4.49741935483871e-05, "loss": 0.4688, "step": 1558 }, { "epoch": 10.049799196787149, "grad_norm": 6.12443208694458, "learning_rate": 4.497096774193549e-05, "loss": 0.1515, "step": 1559 }, { "epoch": 10.056224899598394, "grad_norm": 13.04537582397461, "learning_rate": 4.4967741935483873e-05, "loss": 0.4072, "step": 1560 }, { "epoch": 10.062650602409638, "grad_norm": 4.18139123916626, "learning_rate": 4.496451612903226e-05, "loss": 0.3654, "step": 1561 }, { "epoch": 10.069076305220884, "grad_norm": 6.116156101226807, "learning_rate": 4.4961290322580645e-05, "loss": 0.4127, "step": 1562 }, { "epoch": 10.075502008032128, "grad_norm": 7.457000255584717, "learning_rate": 4.495806451612903e-05, "loss": 0.3828, "step": 1563 }, { "epoch": 10.081927710843374, "grad_norm": 18.627914428710938, "learning_rate": 4.495483870967742e-05, "loss": 0.4954, "step": 1564 }, { "epoch": 10.088353413654618, "grad_norm": 7.246151447296143, "learning_rate": 4.495161290322581e-05, "loss": 0.4008, "step": 1565 }, { "epoch": 10.094779116465864, "grad_norm": 9.027616500854492, "learning_rate": 4.4948387096774194e-05, "loss": 0.4146, "step": 1566 }, { "epoch": 10.101204819277108, "grad_norm": 3.326887845993042, "learning_rate": 4.494516129032259e-05, "loss": 0.2997, "step": 1567 }, { "epoch": 10.107630522088353, "grad_norm": 13.716141700744629, "learning_rate": 4.4941935483870966e-05, "loss": 0.2595, "step": 1568 }, { "epoch": 10.114056224899599, "grad_norm": 22.61565589904785, "learning_rate": 4.493870967741936e-05, "loss": 0.4051, "step": 1569 }, { "epoch": 10.120481927710843, "grad_norm": 5.078835487365723, "learning_rate": 4.4935483870967744e-05, "loss": 0.6409, "step": 1570 }, { "epoch": 10.126907630522089, "grad_norm": 2.565654754638672, "learning_rate": 4.493225806451613e-05, "loss": 0.2897, "step": 1571 }, { "epoch": 10.133333333333333, "grad_norm": 7.139062404632568, "learning_rate": 4.4929032258064515e-05, "loss": 0.3779, "step": 1572 }, { "epoch": 10.139759036144579, "grad_norm": 5.513024806976318, "learning_rate": 4.492580645161291e-05, "loss": 0.6627, "step": 1573 }, { "epoch": 10.146184738955823, "grad_norm": 5.7382893562316895, "learning_rate": 4.4922580645161294e-05, "loss": 0.3219, "step": 1574 }, { "epoch": 10.152610441767068, "grad_norm": 8.356813430786133, "learning_rate": 4.491935483870968e-05, "loss": 0.5747, "step": 1575 }, { "epoch": 10.159036144578312, "grad_norm": 2.866269588470459, "learning_rate": 4.4916129032258065e-05, "loss": 0.2021, "step": 1576 }, { "epoch": 10.165461847389558, "grad_norm": 24.36916160583496, "learning_rate": 4.491290322580645e-05, "loss": 0.5021, "step": 1577 }, { "epoch": 10.171887550200804, "grad_norm": 13.80521297454834, "learning_rate": 4.4909677419354837e-05, "loss": 0.6623, "step": 1578 }, { "epoch": 10.178313253012048, "grad_norm": 13.848732948303223, "learning_rate": 4.490645161290323e-05, "loss": 0.2186, "step": 1579 }, { "epoch": 10.184738955823294, "grad_norm": 14.559137344360352, "learning_rate": 4.4903225806451615e-05, "loss": 0.3145, "step": 1580 }, { "epoch": 10.191164658634538, "grad_norm": 10.860401153564453, "learning_rate": 4.49e-05, "loss": 0.5418, "step": 1581 }, { "epoch": 10.197590361445783, "grad_norm": 11.266544342041016, "learning_rate": 4.489677419354839e-05, "loss": 0.1455, "step": 1582 }, { "epoch": 10.204016064257027, "grad_norm": 24.687971115112305, "learning_rate": 4.489354838709678e-05, "loss": 0.5474, "step": 1583 }, { "epoch": 10.210441767068273, "grad_norm": 3.714075803756714, "learning_rate": 4.4890322580645164e-05, "loss": 0.5952, "step": 1584 }, { "epoch": 10.216867469879517, "grad_norm": 16.78169059753418, "learning_rate": 4.488709677419355e-05, "loss": 0.2861, "step": 1585 }, { "epoch": 10.223293172690763, "grad_norm": 3.9853317737579346, "learning_rate": 4.4883870967741936e-05, "loss": 0.2574, "step": 1586 }, { "epoch": 10.229718875502009, "grad_norm": 6.682840824127197, "learning_rate": 4.488064516129032e-05, "loss": 0.4805, "step": 1587 }, { "epoch": 10.236144578313253, "grad_norm": 15.894432067871094, "learning_rate": 4.4877419354838714e-05, "loss": 0.7153, "step": 1588 }, { "epoch": 10.242570281124499, "grad_norm": 5.846157073974609, "learning_rate": 4.48741935483871e-05, "loss": 0.3724, "step": 1589 }, { "epoch": 10.248995983935743, "grad_norm": 29.074466705322266, "learning_rate": 4.4870967741935485e-05, "loss": 0.4059, "step": 1590 }, { "epoch": 10.255421686746988, "grad_norm": 9.427406311035156, "learning_rate": 4.486774193548388e-05, "loss": 0.2752, "step": 1591 }, { "epoch": 10.261847389558232, "grad_norm": 8.930350303649902, "learning_rate": 4.486451612903226e-05, "loss": 0.4657, "step": 1592 }, { "epoch": 10.268273092369478, "grad_norm": 11.829649925231934, "learning_rate": 4.486129032258064e-05, "loss": 0.4832, "step": 1593 }, { "epoch": 10.274698795180722, "grad_norm": 25.366666793823242, "learning_rate": 4.4858064516129035e-05, "loss": 0.7207, "step": 1594 }, { "epoch": 10.281124497991968, "grad_norm": 10.63737964630127, "learning_rate": 4.485483870967742e-05, "loss": 0.3761, "step": 1595 }, { "epoch": 10.287550200803214, "grad_norm": 21.89575958251953, "learning_rate": 4.4851612903225806e-05, "loss": 0.7294, "step": 1596 }, { "epoch": 10.293975903614458, "grad_norm": 10.132829666137695, "learning_rate": 4.48483870967742e-05, "loss": 0.2409, "step": 1597 }, { "epoch": 10.300401606425703, "grad_norm": 5.012129306793213, "learning_rate": 4.4845161290322585e-05, "loss": 0.4153, "step": 1598 }, { "epoch": 10.306827309236947, "grad_norm": 12.669698715209961, "learning_rate": 4.484193548387097e-05, "loss": 0.5521, "step": 1599 }, { "epoch": 10.313253012048193, "grad_norm": 14.228965759277344, "learning_rate": 4.4838709677419356e-05, "loss": 0.3903, "step": 1600 }, { "epoch": 10.319678714859437, "grad_norm": 7.016195297241211, "learning_rate": 4.483548387096774e-05, "loss": 0.4213, "step": 1601 }, { "epoch": 10.326104417670683, "grad_norm": 7.683666706085205, "learning_rate": 4.483225806451613e-05, "loss": 0.3425, "step": 1602 }, { "epoch": 10.332530120481927, "grad_norm": 3.91682505607605, "learning_rate": 4.482903225806452e-05, "loss": 0.1622, "step": 1603 }, { "epoch": 10.338955823293173, "grad_norm": 17.365541458129883, "learning_rate": 4.4825806451612906e-05, "loss": 0.692, "step": 1604 }, { "epoch": 10.345381526104418, "grad_norm": 12.625564575195312, "learning_rate": 4.482258064516129e-05, "loss": 0.3481, "step": 1605 }, { "epoch": 10.351807228915662, "grad_norm": 18.640127182006836, "learning_rate": 4.4819354838709684e-05, "loss": 0.7323, "step": 1606 }, { "epoch": 10.358232931726908, "grad_norm": 21.861000061035156, "learning_rate": 4.481612903225806e-05, "loss": 1.0859, "step": 1607 }, { "epoch": 10.364658634538152, "grad_norm": 9.204906463623047, "learning_rate": 4.4812903225806455e-05, "loss": 0.3397, "step": 1608 }, { "epoch": 10.371084337349398, "grad_norm": 10.598442077636719, "learning_rate": 4.480967741935484e-05, "loss": 0.3277, "step": 1609 }, { "epoch": 10.377510040160642, "grad_norm": 10.526311874389648, "learning_rate": 4.4806451612903227e-05, "loss": 0.3555, "step": 1610 }, { "epoch": 10.383935742971888, "grad_norm": 16.193456649780273, "learning_rate": 4.480322580645161e-05, "loss": 0.3821, "step": 1611 }, { "epoch": 10.390361445783132, "grad_norm": 4.088657855987549, "learning_rate": 4.4800000000000005e-05, "loss": 0.5856, "step": 1612 }, { "epoch": 10.396787148594377, "grad_norm": 37.08736801147461, "learning_rate": 4.479677419354839e-05, "loss": 0.5044, "step": 1613 }, { "epoch": 10.403212851405623, "grad_norm": 30.210372924804688, "learning_rate": 4.4793548387096776e-05, "loss": 0.4985, "step": 1614 }, { "epoch": 10.409638554216867, "grad_norm": 8.32634162902832, "learning_rate": 4.479032258064516e-05, "loss": 0.4005, "step": 1615 }, { "epoch": 10.416064257028113, "grad_norm": 4.184170722961426, "learning_rate": 4.478709677419355e-05, "loss": 0.2558, "step": 1616 }, { "epoch": 10.422489959839357, "grad_norm": 4.5763654708862305, "learning_rate": 4.478387096774193e-05, "loss": 0.3026, "step": 1617 }, { "epoch": 10.428915662650603, "grad_norm": 12.34216594696045, "learning_rate": 4.4780645161290326e-05, "loss": 0.4875, "step": 1618 }, { "epoch": 10.435341365461847, "grad_norm": 5.62372350692749, "learning_rate": 4.477741935483871e-05, "loss": 0.5355, "step": 1619 }, { "epoch": 10.441767068273093, "grad_norm": 23.508426666259766, "learning_rate": 4.47741935483871e-05, "loss": 0.3391, "step": 1620 }, { "epoch": 10.448192771084337, "grad_norm": 23.689680099487305, "learning_rate": 4.477096774193549e-05, "loss": 0.4444, "step": 1621 }, { "epoch": 10.454618473895582, "grad_norm": 24.302614212036133, "learning_rate": 4.4767741935483875e-05, "loss": 0.3223, "step": 1622 }, { "epoch": 10.461044176706828, "grad_norm": 7.765223503112793, "learning_rate": 4.476451612903226e-05, "loss": 0.4529, "step": 1623 }, { "epoch": 10.467469879518072, "grad_norm": 11.072317123413086, "learning_rate": 4.476129032258065e-05, "loss": 0.5267, "step": 1624 }, { "epoch": 10.473895582329318, "grad_norm": 8.244447708129883, "learning_rate": 4.475806451612903e-05, "loss": 0.3738, "step": 1625 }, { "epoch": 10.480321285140562, "grad_norm": 5.246201515197754, "learning_rate": 4.475483870967742e-05, "loss": 0.5845, "step": 1626 }, { "epoch": 10.486746987951808, "grad_norm": 9.81362533569336, "learning_rate": 4.475161290322581e-05, "loss": 0.5602, "step": 1627 }, { "epoch": 10.493172690763052, "grad_norm": 2.986774206161499, "learning_rate": 4.4748387096774196e-05, "loss": 0.3277, "step": 1628 }, { "epoch": 10.499598393574297, "grad_norm": 16.94302749633789, "learning_rate": 4.474516129032258e-05, "loss": 0.6463, "step": 1629 }, { "epoch": 10.506024096385541, "grad_norm": 9.586493492126465, "learning_rate": 4.4741935483870975e-05, "loss": 0.3059, "step": 1630 }, { "epoch": 10.512449799196787, "grad_norm": 14.42587947845459, "learning_rate": 4.4738709677419354e-05, "loss": 0.369, "step": 1631 }, { "epoch": 10.518875502008033, "grad_norm": 5.100485801696777, "learning_rate": 4.4735483870967746e-05, "loss": 0.5492, "step": 1632 }, { "epoch": 10.525301204819277, "grad_norm": 20.091331481933594, "learning_rate": 4.473225806451613e-05, "loss": 0.5176, "step": 1633 }, { "epoch": 10.531726907630523, "grad_norm": 15.228740692138672, "learning_rate": 4.472903225806452e-05, "loss": 0.6893, "step": 1634 }, { "epoch": 10.538152610441767, "grad_norm": 14.828134536743164, "learning_rate": 4.47258064516129e-05, "loss": 0.2033, "step": 1635 }, { "epoch": 10.544578313253012, "grad_norm": 14.897771835327148, "learning_rate": 4.4722580645161296e-05, "loss": 0.4459, "step": 1636 }, { "epoch": 10.551004016064256, "grad_norm": 19.34054946899414, "learning_rate": 4.471935483870968e-05, "loss": 0.5744, "step": 1637 }, { "epoch": 10.557429718875502, "grad_norm": 17.27313804626465, "learning_rate": 4.471612903225807e-05, "loss": 0.6083, "step": 1638 }, { "epoch": 10.563855421686746, "grad_norm": 19.050573348999023, "learning_rate": 4.471290322580645e-05, "loss": 0.2265, "step": 1639 }, { "epoch": 10.570281124497992, "grad_norm": 20.12076187133789, "learning_rate": 4.470967741935484e-05, "loss": 0.4572, "step": 1640 }, { "epoch": 10.576706827309238, "grad_norm": 6.260252952575684, "learning_rate": 4.4706451612903224e-05, "loss": 0.3288, "step": 1641 }, { "epoch": 10.583132530120482, "grad_norm": 12.18903636932373, "learning_rate": 4.4703225806451617e-05, "loss": 0.282, "step": 1642 }, { "epoch": 10.589558232931727, "grad_norm": 4.765219688415527, "learning_rate": 4.47e-05, "loss": 0.2999, "step": 1643 }, { "epoch": 10.595983935742971, "grad_norm": 15.261346817016602, "learning_rate": 4.469677419354839e-05, "loss": 0.6338, "step": 1644 }, { "epoch": 10.602409638554217, "grad_norm": 10.75886344909668, "learning_rate": 4.469354838709678e-05, "loss": 0.6434, "step": 1645 }, { "epoch": 10.608835341365461, "grad_norm": 12.972408294677734, "learning_rate": 4.469032258064516e-05, "loss": 0.7483, "step": 1646 }, { "epoch": 10.615261044176707, "grad_norm": 8.999165534973145, "learning_rate": 4.468709677419355e-05, "loss": 0.5216, "step": 1647 }, { "epoch": 10.621686746987951, "grad_norm": 2.5601673126220703, "learning_rate": 4.468387096774194e-05, "loss": 0.2506, "step": 1648 }, { "epoch": 10.628112449799197, "grad_norm": 2.9364402294158936, "learning_rate": 4.468064516129032e-05, "loss": 0.2726, "step": 1649 }, { "epoch": 10.634538152610443, "grad_norm": 8.84613037109375, "learning_rate": 4.467741935483871e-05, "loss": 0.4157, "step": 1650 }, { "epoch": 10.640963855421687, "grad_norm": 4.388571739196777, "learning_rate": 4.46741935483871e-05, "loss": 0.1936, "step": 1651 }, { "epoch": 10.647389558232932, "grad_norm": 11.654314994812012, "learning_rate": 4.467096774193549e-05, "loss": 0.3866, "step": 1652 }, { "epoch": 10.653815261044176, "grad_norm": 15.50309944152832, "learning_rate": 4.466774193548387e-05, "loss": 0.3346, "step": 1653 }, { "epoch": 10.660240963855422, "grad_norm": 20.49254035949707, "learning_rate": 4.466451612903226e-05, "loss": 0.4787, "step": 1654 }, { "epoch": 10.666666666666666, "grad_norm": 14.674749374389648, "learning_rate": 4.4661290322580644e-05, "loss": 0.4554, "step": 1655 }, { "epoch": 10.673092369477912, "grad_norm": 11.307194709777832, "learning_rate": 4.465806451612903e-05, "loss": 0.5293, "step": 1656 }, { "epoch": 10.679518072289156, "grad_norm": 4.174642562866211, "learning_rate": 4.465483870967742e-05, "loss": 0.3619, "step": 1657 }, { "epoch": 10.685943775100402, "grad_norm": 12.543534278869629, "learning_rate": 4.465161290322581e-05, "loss": 0.5766, "step": 1658 }, { "epoch": 10.692369477911647, "grad_norm": 7.299986839294434, "learning_rate": 4.4648387096774194e-05, "loss": 0.233, "step": 1659 }, { "epoch": 10.698795180722891, "grad_norm": 2.143785238265991, "learning_rate": 4.4645161290322586e-05, "loss": 0.57, "step": 1660 }, { "epoch": 10.705220883534137, "grad_norm": 6.10801362991333, "learning_rate": 4.464193548387097e-05, "loss": 0.421, "step": 1661 }, { "epoch": 10.711646586345381, "grad_norm": 6.074007034301758, "learning_rate": 4.463870967741936e-05, "loss": 0.5913, "step": 1662 }, { "epoch": 10.718072289156627, "grad_norm": 4.897146224975586, "learning_rate": 4.4635483870967744e-05, "loss": 0.4555, "step": 1663 }, { "epoch": 10.724497991967871, "grad_norm": 14.596113204956055, "learning_rate": 4.463225806451613e-05, "loss": 0.334, "step": 1664 }, { "epoch": 10.730923694779117, "grad_norm": 10.900601387023926, "learning_rate": 4.4629032258064515e-05, "loss": 0.466, "step": 1665 }, { "epoch": 10.73734939759036, "grad_norm": 27.12323760986328, "learning_rate": 4.462580645161291e-05, "loss": 0.5425, "step": 1666 }, { "epoch": 10.743775100401606, "grad_norm": 8.415623664855957, "learning_rate": 4.462258064516129e-05, "loss": 0.4389, "step": 1667 }, { "epoch": 10.750200803212852, "grad_norm": 3.5860776901245117, "learning_rate": 4.461935483870968e-05, "loss": 0.2754, "step": 1668 }, { "epoch": 10.756626506024096, "grad_norm": 11.235335350036621, "learning_rate": 4.461612903225807e-05, "loss": 0.4123, "step": 1669 }, { "epoch": 10.763052208835342, "grad_norm": 19.016387939453125, "learning_rate": 4.461290322580645e-05, "loss": 0.7891, "step": 1670 }, { "epoch": 10.769477911646586, "grad_norm": 4.618960380554199, "learning_rate": 4.460967741935484e-05, "loss": 0.323, "step": 1671 }, { "epoch": 10.775903614457832, "grad_norm": 26.493022918701172, "learning_rate": 4.460645161290323e-05, "loss": 0.2457, "step": 1672 }, { "epoch": 10.782329317269076, "grad_norm": 14.864387512207031, "learning_rate": 4.4603225806451614e-05, "loss": 0.7217, "step": 1673 }, { "epoch": 10.788755020080322, "grad_norm": 4.76592493057251, "learning_rate": 4.46e-05, "loss": 0.4932, "step": 1674 }, { "epoch": 10.795180722891565, "grad_norm": 13.654341697692871, "learning_rate": 4.459677419354839e-05, "loss": 0.4407, "step": 1675 }, { "epoch": 10.801606425702811, "grad_norm": 6.560559272766113, "learning_rate": 4.459354838709678e-05, "loss": 0.6744, "step": 1676 }, { "epoch": 10.808032128514057, "grad_norm": 11.370556831359863, "learning_rate": 4.4590322580645164e-05, "loss": 0.6721, "step": 1677 }, { "epoch": 10.814457831325301, "grad_norm": 8.849052429199219, "learning_rate": 4.458709677419355e-05, "loss": 0.3445, "step": 1678 }, { "epoch": 10.820883534136547, "grad_norm": 21.940696716308594, "learning_rate": 4.4583870967741935e-05, "loss": 0.5562, "step": 1679 }, { "epoch": 10.82730923694779, "grad_norm": 2.9612255096435547, "learning_rate": 4.458064516129032e-05, "loss": 0.3177, "step": 1680 }, { "epoch": 10.833734939759037, "grad_norm": 20.10212516784668, "learning_rate": 4.457741935483871e-05, "loss": 0.5578, "step": 1681 }, { "epoch": 10.84016064257028, "grad_norm": 13.140884399414062, "learning_rate": 4.45741935483871e-05, "loss": 0.4755, "step": 1682 }, { "epoch": 10.846586345381526, "grad_norm": 39.90276336669922, "learning_rate": 4.4570967741935485e-05, "loss": 0.4899, "step": 1683 }, { "epoch": 10.85301204819277, "grad_norm": 8.231800079345703, "learning_rate": 4.456774193548388e-05, "loss": 0.3318, "step": 1684 }, { "epoch": 10.859437751004016, "grad_norm": 6.851098537445068, "learning_rate": 4.4564516129032256e-05, "loss": 0.1989, "step": 1685 }, { "epoch": 10.865863453815262, "grad_norm": 18.486303329467773, "learning_rate": 4.456129032258065e-05, "loss": 0.3842, "step": 1686 }, { "epoch": 10.872289156626506, "grad_norm": 10.718703269958496, "learning_rate": 4.4558064516129034e-05, "loss": 0.5154, "step": 1687 }, { "epoch": 10.878714859437752, "grad_norm": 11.23798656463623, "learning_rate": 4.455483870967742e-05, "loss": 0.4695, "step": 1688 }, { "epoch": 10.885140562248996, "grad_norm": 15.679261207580566, "learning_rate": 4.4551612903225806e-05, "loss": 0.5812, "step": 1689 }, { "epoch": 10.891566265060241, "grad_norm": 4.956465244293213, "learning_rate": 4.45483870967742e-05, "loss": 0.4437, "step": 1690 }, { "epoch": 10.897991967871485, "grad_norm": 27.868927001953125, "learning_rate": 4.4545161290322584e-05, "loss": 0.6381, "step": 1691 }, { "epoch": 10.904417670682731, "grad_norm": 7.564282417297363, "learning_rate": 4.454193548387097e-05, "loss": 0.8707, "step": 1692 }, { "epoch": 10.910843373493975, "grad_norm": 9.692399024963379, "learning_rate": 4.4538709677419355e-05, "loss": 0.5303, "step": 1693 }, { "epoch": 10.917269076305221, "grad_norm": 8.592081069946289, "learning_rate": 4.453548387096774e-05, "loss": 0.2016, "step": 1694 }, { "epoch": 10.923694779116467, "grad_norm": 11.425627708435059, "learning_rate": 4.4532258064516134e-05, "loss": 0.4889, "step": 1695 }, { "epoch": 10.93012048192771, "grad_norm": 9.292033195495605, "learning_rate": 4.452903225806452e-05, "loss": 0.6824, "step": 1696 }, { "epoch": 10.936546184738956, "grad_norm": 8.81248664855957, "learning_rate": 4.4525806451612905e-05, "loss": 0.3928, "step": 1697 }, { "epoch": 10.9429718875502, "grad_norm": 6.588391304016113, "learning_rate": 4.452258064516129e-05, "loss": 0.3636, "step": 1698 }, { "epoch": 10.949397590361446, "grad_norm": 11.97680950164795, "learning_rate": 4.451935483870968e-05, "loss": 0.3597, "step": 1699 }, { "epoch": 10.95582329317269, "grad_norm": 40.10157012939453, "learning_rate": 4.451612903225807e-05, "loss": 0.4629, "step": 1700 }, { "epoch": 10.962248995983936, "grad_norm": 3.3649492263793945, "learning_rate": 4.4512903225806455e-05, "loss": 0.1859, "step": 1701 }, { "epoch": 10.96867469879518, "grad_norm": 9.014705657958984, "learning_rate": 4.450967741935484e-05, "loss": 0.3913, "step": 1702 }, { "epoch": 10.975100401606426, "grad_norm": 39.9904899597168, "learning_rate": 4.4506451612903226e-05, "loss": 0.4602, "step": 1703 }, { "epoch": 10.981526104417672, "grad_norm": 11.499724388122559, "learning_rate": 4.450322580645161e-05, "loss": 0.5553, "step": 1704 }, { "epoch": 10.987951807228916, "grad_norm": 5.6467132568359375, "learning_rate": 4.4500000000000004e-05, "loss": 0.272, "step": 1705 }, { "epoch": 10.994377510040161, "grad_norm": 4.188947677612305, "learning_rate": 4.449677419354839e-05, "loss": 0.1779, "step": 1706 }, { "epoch": 11.004016064257028, "grad_norm": 17.50454330444336, "learning_rate": 4.4493548387096776e-05, "loss": 0.547, "step": 1707 }, { "epoch": 11.010441767068274, "grad_norm": 3.6488513946533203, "learning_rate": 4.449032258064516e-05, "loss": 0.3999, "step": 1708 }, { "epoch": 11.016867469879518, "grad_norm": 9.904997825622559, "learning_rate": 4.448709677419355e-05, "loss": 0.3156, "step": 1709 }, { "epoch": 11.023293172690764, "grad_norm": 5.348606109619141, "learning_rate": 4.448387096774194e-05, "loss": 0.2319, "step": 1710 }, { "epoch": 11.029718875502008, "grad_norm": 26.132320404052734, "learning_rate": 4.4480645161290325e-05, "loss": 0.5046, "step": 1711 }, { "epoch": 11.036144578313253, "grad_norm": 24.13222885131836, "learning_rate": 4.447741935483871e-05, "loss": 0.2587, "step": 1712 }, { "epoch": 11.042570281124497, "grad_norm": 5.5433173179626465, "learning_rate": 4.44741935483871e-05, "loss": 0.4478, "step": 1713 }, { "epoch": 11.048995983935743, "grad_norm": 32.923927307128906, "learning_rate": 4.447096774193549e-05, "loss": 0.4621, "step": 1714 }, { "epoch": 11.055421686746987, "grad_norm": 33.47465515136719, "learning_rate": 4.4467741935483875e-05, "loss": 0.4524, "step": 1715 }, { "epoch": 11.061847389558233, "grad_norm": 11.40574836730957, "learning_rate": 4.446451612903226e-05, "loss": 0.5715, "step": 1716 }, { "epoch": 11.068273092369479, "grad_norm": 21.927227020263672, "learning_rate": 4.4461290322580646e-05, "loss": 0.3012, "step": 1717 }, { "epoch": 11.074698795180723, "grad_norm": 16.87617301940918, "learning_rate": 4.445806451612903e-05, "loss": 0.3102, "step": 1718 }, { "epoch": 11.081124497991969, "grad_norm": 11.0452241897583, "learning_rate": 4.445483870967742e-05, "loss": 0.4368, "step": 1719 }, { "epoch": 11.087550200803213, "grad_norm": 4.261916160583496, "learning_rate": 4.445161290322581e-05, "loss": 0.3916, "step": 1720 }, { "epoch": 11.093975903614458, "grad_norm": 11.981135368347168, "learning_rate": 4.4448387096774196e-05, "loss": 0.3471, "step": 1721 }, { "epoch": 11.100401606425702, "grad_norm": 7.075514793395996, "learning_rate": 4.444516129032258e-05, "loss": 0.3927, "step": 1722 }, { "epoch": 11.106827309236948, "grad_norm": 2.223888635635376, "learning_rate": 4.4441935483870974e-05, "loss": 0.2503, "step": 1723 }, { "epoch": 11.113253012048192, "grad_norm": 20.507734298706055, "learning_rate": 4.443870967741935e-05, "loss": 0.5847, "step": 1724 }, { "epoch": 11.119678714859438, "grad_norm": 4.71273136138916, "learning_rate": 4.4435483870967745e-05, "loss": 0.4572, "step": 1725 }, { "epoch": 11.126104417670684, "grad_norm": 6.136356830596924, "learning_rate": 4.443225806451613e-05, "loss": 0.2855, "step": 1726 }, { "epoch": 11.132530120481928, "grad_norm": 6.983724594116211, "learning_rate": 4.442903225806452e-05, "loss": 0.2875, "step": 1727 }, { "epoch": 11.138955823293173, "grad_norm": 4.887809753417969, "learning_rate": 4.44258064516129e-05, "loss": 0.254, "step": 1728 }, { "epoch": 11.145381526104417, "grad_norm": 10.7438325881958, "learning_rate": 4.4422580645161295e-05, "loss": 0.7232, "step": 1729 }, { "epoch": 11.151807228915663, "grad_norm": 17.85144805908203, "learning_rate": 4.441935483870968e-05, "loss": 0.5377, "step": 1730 }, { "epoch": 11.158232931726907, "grad_norm": 11.64863109588623, "learning_rate": 4.4416129032258066e-05, "loss": 0.2341, "step": 1731 }, { "epoch": 11.164658634538153, "grad_norm": 5.254570007324219, "learning_rate": 4.441290322580645e-05, "loss": 0.4199, "step": 1732 }, { "epoch": 11.171084337349397, "grad_norm": 2.904249429702759, "learning_rate": 4.440967741935484e-05, "loss": 0.0914, "step": 1733 }, { "epoch": 11.177510040160643, "grad_norm": 6.988886833190918, "learning_rate": 4.440645161290323e-05, "loss": 0.34, "step": 1734 }, { "epoch": 11.183935742971888, "grad_norm": 6.8980793952941895, "learning_rate": 4.4403225806451616e-05, "loss": 0.3386, "step": 1735 }, { "epoch": 11.190361445783132, "grad_norm": 14.90070915222168, "learning_rate": 4.44e-05, "loss": 0.4086, "step": 1736 }, { "epoch": 11.196787148594378, "grad_norm": 3.2431960105895996, "learning_rate": 4.439677419354839e-05, "loss": 0.5309, "step": 1737 }, { "epoch": 11.203212851405622, "grad_norm": 8.289189338684082, "learning_rate": 4.439354838709678e-05, "loss": 0.7035, "step": 1738 }, { "epoch": 11.209638554216868, "grad_norm": 14.731082916259766, "learning_rate": 4.4390322580645166e-05, "loss": 0.4518, "step": 1739 }, { "epoch": 11.216064257028112, "grad_norm": 15.212279319763184, "learning_rate": 4.438709677419355e-05, "loss": 0.5239, "step": 1740 }, { "epoch": 11.222489959839358, "grad_norm": 9.642303466796875, "learning_rate": 4.438387096774194e-05, "loss": 0.51, "step": 1741 }, { "epoch": 11.228915662650602, "grad_norm": 5.2568840980529785, "learning_rate": 4.438064516129032e-05, "loss": 0.2812, "step": 1742 }, { "epoch": 11.235341365461847, "grad_norm": 11.618955612182617, "learning_rate": 4.437741935483871e-05, "loss": 0.438, "step": 1743 }, { "epoch": 11.241767068273091, "grad_norm": 8.071309089660645, "learning_rate": 4.43741935483871e-05, "loss": 0.3959, "step": 1744 }, { "epoch": 11.248192771084337, "grad_norm": 22.37892723083496, "learning_rate": 4.437096774193549e-05, "loss": 0.4916, "step": 1745 }, { "epoch": 11.254618473895583, "grad_norm": 35.806697845458984, "learning_rate": 4.436774193548387e-05, "loss": 0.51, "step": 1746 }, { "epoch": 11.261044176706827, "grad_norm": 11.844949722290039, "learning_rate": 4.436451612903226e-05, "loss": 0.297, "step": 1747 }, { "epoch": 11.267469879518073, "grad_norm": 3.8206493854522705, "learning_rate": 4.4361290322580644e-05, "loss": 0.3775, "step": 1748 }, { "epoch": 11.273895582329317, "grad_norm": 7.955654144287109, "learning_rate": 4.4358064516129036e-05, "loss": 0.5372, "step": 1749 }, { "epoch": 11.280321285140563, "grad_norm": 2.7302544116973877, "learning_rate": 4.435483870967742e-05, "loss": 0.2554, "step": 1750 }, { "epoch": 11.286746987951807, "grad_norm": 8.11384105682373, "learning_rate": 4.435161290322581e-05, "loss": 0.4401, "step": 1751 }, { "epoch": 11.293172690763052, "grad_norm": 5.545860290527344, "learning_rate": 4.4348387096774193e-05, "loss": 0.4107, "step": 1752 }, { "epoch": 11.299598393574296, "grad_norm": 11.797832489013672, "learning_rate": 4.4345161290322586e-05, "loss": 0.3651, "step": 1753 }, { "epoch": 11.306024096385542, "grad_norm": 16.167909622192383, "learning_rate": 4.434193548387097e-05, "loss": 0.439, "step": 1754 }, { "epoch": 11.312449799196788, "grad_norm": 25.5878963470459, "learning_rate": 4.433870967741936e-05, "loss": 0.3009, "step": 1755 }, { "epoch": 11.318875502008032, "grad_norm": 4.192933082580566, "learning_rate": 4.433548387096774e-05, "loss": 0.592, "step": 1756 }, { "epoch": 11.325301204819278, "grad_norm": 3.48460054397583, "learning_rate": 4.433225806451613e-05, "loss": 0.4633, "step": 1757 }, { "epoch": 11.331726907630522, "grad_norm": 7.289281845092773, "learning_rate": 4.432903225806452e-05, "loss": 0.3167, "step": 1758 }, { "epoch": 11.338152610441767, "grad_norm": 7.784161567687988, "learning_rate": 4.432580645161291e-05, "loss": 0.4541, "step": 1759 }, { "epoch": 11.344578313253011, "grad_norm": 25.923221588134766, "learning_rate": 4.432258064516129e-05, "loss": 0.5821, "step": 1760 }, { "epoch": 11.351004016064257, "grad_norm": 12.170945167541504, "learning_rate": 4.431935483870968e-05, "loss": 0.3573, "step": 1761 }, { "epoch": 11.357429718875501, "grad_norm": 9.928370475769043, "learning_rate": 4.431612903225807e-05, "loss": 0.6382, "step": 1762 }, { "epoch": 11.363855421686747, "grad_norm": 12.637680053710938, "learning_rate": 4.431290322580645e-05, "loss": 0.6918, "step": 1763 }, { "epoch": 11.370281124497993, "grad_norm": 17.84426498413086, "learning_rate": 4.430967741935484e-05, "loss": 0.602, "step": 1764 }, { "epoch": 11.376706827309237, "grad_norm": 6.647047519683838, "learning_rate": 4.430645161290323e-05, "loss": 0.415, "step": 1765 }, { "epoch": 11.383132530120482, "grad_norm": 13.386226654052734, "learning_rate": 4.4303225806451614e-05, "loss": 0.5968, "step": 1766 }, { "epoch": 11.389558232931726, "grad_norm": 6.164967060089111, "learning_rate": 4.43e-05, "loss": 0.3456, "step": 1767 }, { "epoch": 11.395983935742972, "grad_norm": 5.480262279510498, "learning_rate": 4.429677419354839e-05, "loss": 0.5539, "step": 1768 }, { "epoch": 11.402409638554216, "grad_norm": 17.196455001831055, "learning_rate": 4.429354838709678e-05, "loss": 0.4301, "step": 1769 }, { "epoch": 11.408835341365462, "grad_norm": 28.540760040283203, "learning_rate": 4.429032258064516e-05, "loss": 0.3231, "step": 1770 }, { "epoch": 11.415261044176706, "grad_norm": 18.254234313964844, "learning_rate": 4.428709677419355e-05, "loss": 0.275, "step": 1771 }, { "epoch": 11.421686746987952, "grad_norm": 11.93423080444336, "learning_rate": 4.4283870967741935e-05, "loss": 0.3356, "step": 1772 }, { "epoch": 11.428112449799197, "grad_norm": 5.730836391448975, "learning_rate": 4.428064516129033e-05, "loss": 0.2856, "step": 1773 }, { "epoch": 11.434538152610441, "grad_norm": 10.764653205871582, "learning_rate": 4.427741935483871e-05, "loss": 0.3233, "step": 1774 }, { "epoch": 11.440963855421687, "grad_norm": 8.105271339416504, "learning_rate": 4.42741935483871e-05, "loss": 0.1934, "step": 1775 }, { "epoch": 11.447389558232931, "grad_norm": 5.137781143188477, "learning_rate": 4.4270967741935484e-05, "loss": 0.2891, "step": 1776 }, { "epoch": 11.453815261044177, "grad_norm": 1.8426623344421387, "learning_rate": 4.426774193548388e-05, "loss": 0.2082, "step": 1777 }, { "epoch": 11.460240963855421, "grad_norm": 29.303457260131836, "learning_rate": 4.4264516129032256e-05, "loss": 0.7018, "step": 1778 }, { "epoch": 11.466666666666667, "grad_norm": 5.160272598266602, "learning_rate": 4.426129032258065e-05, "loss": 0.4923, "step": 1779 }, { "epoch": 11.47309236947791, "grad_norm": 21.41295051574707, "learning_rate": 4.4258064516129034e-05, "loss": 0.3121, "step": 1780 }, { "epoch": 11.479518072289157, "grad_norm": 8.212661743164062, "learning_rate": 4.425483870967742e-05, "loss": 0.4413, "step": 1781 }, { "epoch": 11.485943775100402, "grad_norm": 7.490704536437988, "learning_rate": 4.4251612903225805e-05, "loss": 0.5126, "step": 1782 }, { "epoch": 11.492369477911646, "grad_norm": 21.9736385345459, "learning_rate": 4.42483870967742e-05, "loss": 0.4281, "step": 1783 }, { "epoch": 11.498795180722892, "grad_norm": 7.3630242347717285, "learning_rate": 4.4245161290322583e-05, "loss": 0.2672, "step": 1784 }, { "epoch": 11.505220883534136, "grad_norm": 18.775880813598633, "learning_rate": 4.424193548387097e-05, "loss": 0.4196, "step": 1785 }, { "epoch": 11.511646586345382, "grad_norm": 5.104211330413818, "learning_rate": 4.4238709677419355e-05, "loss": 0.2894, "step": 1786 }, { "epoch": 11.518072289156626, "grad_norm": 10.76314926147461, "learning_rate": 4.423548387096774e-05, "loss": 0.3995, "step": 1787 }, { "epoch": 11.524497991967872, "grad_norm": 23.35580062866211, "learning_rate": 4.423225806451613e-05, "loss": 0.7741, "step": 1788 }, { "epoch": 11.530923694779116, "grad_norm": 8.649627685546875, "learning_rate": 4.422903225806452e-05, "loss": 0.4026, "step": 1789 }, { "epoch": 11.537349397590361, "grad_norm": 9.189298629760742, "learning_rate": 4.4225806451612905e-05, "loss": 0.3643, "step": 1790 }, { "epoch": 11.543775100401607, "grad_norm": 6.555013179779053, "learning_rate": 4.422258064516129e-05, "loss": 0.495, "step": 1791 }, { "epoch": 11.550200803212851, "grad_norm": 9.710036277770996, "learning_rate": 4.421935483870968e-05, "loss": 0.5718, "step": 1792 }, { "epoch": 11.556626506024097, "grad_norm": 20.631954193115234, "learning_rate": 4.421612903225807e-05, "loss": 0.5101, "step": 1793 }, { "epoch": 11.563052208835341, "grad_norm": 7.366153717041016, "learning_rate": 4.4212903225806454e-05, "loss": 0.4173, "step": 1794 }, { "epoch": 11.569477911646587, "grad_norm": 13.999110221862793, "learning_rate": 4.420967741935484e-05, "loss": 0.3291, "step": 1795 }, { "epoch": 11.57590361445783, "grad_norm": 11.77374267578125, "learning_rate": 4.4206451612903226e-05, "loss": 0.248, "step": 1796 }, { "epoch": 11.582329317269076, "grad_norm": 18.378063201904297, "learning_rate": 4.420322580645162e-05, "loss": 0.2605, "step": 1797 }, { "epoch": 11.58875502008032, "grad_norm": 13.08482551574707, "learning_rate": 4.4200000000000004e-05, "loss": 0.3104, "step": 1798 }, { "epoch": 11.595180722891566, "grad_norm": 7.943434238433838, "learning_rate": 4.419677419354839e-05, "loss": 0.4745, "step": 1799 }, { "epoch": 11.601606425702812, "grad_norm": 12.99714469909668, "learning_rate": 4.4193548387096775e-05, "loss": 0.7296, "step": 1800 }, { "epoch": 11.608032128514056, "grad_norm": 4.018809795379639, "learning_rate": 4.419032258064517e-05, "loss": 0.2841, "step": 1801 }, { "epoch": 11.614457831325302, "grad_norm": 7.5128350257873535, "learning_rate": 4.4187096774193547e-05, "loss": 0.7237, "step": 1802 }, { "epoch": 11.620883534136546, "grad_norm": 3.460815191268921, "learning_rate": 4.418387096774194e-05, "loss": 0.4571, "step": 1803 }, { "epoch": 11.627309236947792, "grad_norm": 9.373250961303711, "learning_rate": 4.4180645161290325e-05, "loss": 0.2952, "step": 1804 }, { "epoch": 11.633734939759035, "grad_norm": 9.338949203491211, "learning_rate": 4.417741935483871e-05, "loss": 0.4781, "step": 1805 }, { "epoch": 11.640160642570281, "grad_norm": 7.575860023498535, "learning_rate": 4.4174193548387096e-05, "loss": 0.4394, "step": 1806 }, { "epoch": 11.646586345381525, "grad_norm": 8.705543518066406, "learning_rate": 4.417096774193549e-05, "loss": 0.3494, "step": 1807 }, { "epoch": 11.653012048192771, "grad_norm": 17.256664276123047, "learning_rate": 4.4167741935483874e-05, "loss": 0.3337, "step": 1808 }, { "epoch": 11.659437751004017, "grad_norm": 24.074846267700195, "learning_rate": 4.416451612903226e-05, "loss": 0.5407, "step": 1809 }, { "epoch": 11.66586345381526, "grad_norm": 16.091175079345703, "learning_rate": 4.4161290322580646e-05, "loss": 0.4349, "step": 1810 }, { "epoch": 11.672289156626507, "grad_norm": 25.256818771362305, "learning_rate": 4.415806451612903e-05, "loss": 0.4757, "step": 1811 }, { "epoch": 11.67871485943775, "grad_norm": 11.098151206970215, "learning_rate": 4.4154838709677424e-05, "loss": 0.4941, "step": 1812 }, { "epoch": 11.685140562248996, "grad_norm": 4.3648176193237305, "learning_rate": 4.415161290322581e-05, "loss": 0.4278, "step": 1813 }, { "epoch": 11.69156626506024, "grad_norm": 7.652712821960449, "learning_rate": 4.4148387096774195e-05, "loss": 0.53, "step": 1814 }, { "epoch": 11.697991967871486, "grad_norm": 14.600770950317383, "learning_rate": 4.414516129032258e-05, "loss": 0.6173, "step": 1815 }, { "epoch": 11.70441767068273, "grad_norm": 19.51537322998047, "learning_rate": 4.4141935483870974e-05, "loss": 0.592, "step": 1816 }, { "epoch": 11.710843373493976, "grad_norm": 7.4120612144470215, "learning_rate": 4.413870967741935e-05, "loss": 0.5146, "step": 1817 }, { "epoch": 11.717269076305222, "grad_norm": 19.96170997619629, "learning_rate": 4.4135483870967745e-05, "loss": 0.4286, "step": 1818 }, { "epoch": 11.723694779116466, "grad_norm": 6.1219706535339355, "learning_rate": 4.413225806451613e-05, "loss": 0.3538, "step": 1819 }, { "epoch": 11.730120481927711, "grad_norm": 5.169804096221924, "learning_rate": 4.4129032258064516e-05, "loss": 0.1609, "step": 1820 }, { "epoch": 11.736546184738955, "grad_norm": 23.375261306762695, "learning_rate": 4.412580645161291e-05, "loss": 0.3037, "step": 1821 }, { "epoch": 11.742971887550201, "grad_norm": 5.673272609710693, "learning_rate": 4.4122580645161295e-05, "loss": 0.2929, "step": 1822 }, { "epoch": 11.749397590361445, "grad_norm": 11.93902587890625, "learning_rate": 4.411935483870968e-05, "loss": 0.3466, "step": 1823 }, { "epoch": 11.755823293172691, "grad_norm": 13.442099571228027, "learning_rate": 4.4116129032258066e-05, "loss": 0.3889, "step": 1824 }, { "epoch": 11.762248995983935, "grad_norm": 10.0659818649292, "learning_rate": 4.411290322580645e-05, "loss": 0.331, "step": 1825 }, { "epoch": 11.76867469879518, "grad_norm": 35.612606048583984, "learning_rate": 4.410967741935484e-05, "loss": 0.4908, "step": 1826 }, { "epoch": 11.775100401606426, "grad_norm": 13.118001937866211, "learning_rate": 4.410645161290323e-05, "loss": 0.3361, "step": 1827 }, { "epoch": 11.78152610441767, "grad_norm": 5.213441848754883, "learning_rate": 4.4103225806451616e-05, "loss": 0.2508, "step": 1828 }, { "epoch": 11.787951807228916, "grad_norm": 64.32823181152344, "learning_rate": 4.41e-05, "loss": 0.5482, "step": 1829 }, { "epoch": 11.79437751004016, "grad_norm": 8.632515907287598, "learning_rate": 4.409677419354839e-05, "loss": 0.4642, "step": 1830 }, { "epoch": 11.800803212851406, "grad_norm": 19.30299186706543, "learning_rate": 4.409354838709678e-05, "loss": 0.4977, "step": 1831 }, { "epoch": 11.80722891566265, "grad_norm": 4.633859634399414, "learning_rate": 4.4090322580645165e-05, "loss": 0.1925, "step": 1832 }, { "epoch": 11.813654618473896, "grad_norm": 21.072059631347656, "learning_rate": 4.408709677419355e-05, "loss": 0.5947, "step": 1833 }, { "epoch": 11.82008032128514, "grad_norm": 7.631752014160156, "learning_rate": 4.4083870967741937e-05, "loss": 0.5981, "step": 1834 }, { "epoch": 11.826506024096386, "grad_norm": 8.043185234069824, "learning_rate": 4.408064516129032e-05, "loss": 0.3017, "step": 1835 }, { "epoch": 11.832931726907631, "grad_norm": 18.591846466064453, "learning_rate": 4.4077419354838715e-05, "loss": 0.5579, "step": 1836 }, { "epoch": 11.839357429718875, "grad_norm": 7.123997211456299, "learning_rate": 4.40741935483871e-05, "loss": 0.4324, "step": 1837 }, { "epoch": 11.845783132530121, "grad_norm": 3.826538324356079, "learning_rate": 4.4070967741935486e-05, "loss": 0.1812, "step": 1838 }, { "epoch": 11.852208835341365, "grad_norm": 3.811849355697632, "learning_rate": 4.406774193548387e-05, "loss": 0.3079, "step": 1839 }, { "epoch": 11.85863453815261, "grad_norm": 6.1769208908081055, "learning_rate": 4.4064516129032264e-05, "loss": 0.2737, "step": 1840 }, { "epoch": 11.865060240963855, "grad_norm": 13.517488479614258, "learning_rate": 4.406129032258064e-05, "loss": 0.4947, "step": 1841 }, { "epoch": 11.8714859437751, "grad_norm": 10.735214233398438, "learning_rate": 4.4058064516129036e-05, "loss": 0.2608, "step": 1842 }, { "epoch": 11.877911646586345, "grad_norm": 4.499989032745361, "learning_rate": 4.405483870967742e-05, "loss": 0.5368, "step": 1843 }, { "epoch": 11.88433734939759, "grad_norm": 5.297222137451172, "learning_rate": 4.405161290322581e-05, "loss": 0.2814, "step": 1844 }, { "epoch": 11.890763052208836, "grad_norm": 20.671180725097656, "learning_rate": 4.404838709677419e-05, "loss": 0.3946, "step": 1845 }, { "epoch": 11.89718875502008, "grad_norm": 10.58663558959961, "learning_rate": 4.4045161290322585e-05, "loss": 0.7357, "step": 1846 }, { "epoch": 11.903614457831326, "grad_norm": 3.6158268451690674, "learning_rate": 4.404193548387097e-05, "loss": 0.2786, "step": 1847 }, { "epoch": 11.91004016064257, "grad_norm": 14.923593521118164, "learning_rate": 4.403870967741936e-05, "loss": 0.4267, "step": 1848 }, { "epoch": 11.916465863453816, "grad_norm": 17.173784255981445, "learning_rate": 4.403548387096774e-05, "loss": 0.6622, "step": 1849 }, { "epoch": 11.92289156626506, "grad_norm": 18.69308853149414, "learning_rate": 4.403225806451613e-05, "loss": 0.3216, "step": 1850 }, { "epoch": 11.929317269076305, "grad_norm": 7.912259101867676, "learning_rate": 4.402903225806452e-05, "loss": 0.3884, "step": 1851 }, { "epoch": 11.93574297188755, "grad_norm": 5.544404029846191, "learning_rate": 4.4025806451612906e-05, "loss": 0.3749, "step": 1852 }, { "epoch": 11.942168674698795, "grad_norm": 6.7178449630737305, "learning_rate": 4.402258064516129e-05, "loss": 0.3337, "step": 1853 }, { "epoch": 11.948594377510041, "grad_norm": 4.901413917541504, "learning_rate": 4.401935483870968e-05, "loss": 0.4674, "step": 1854 }, { "epoch": 11.955020080321285, "grad_norm": 21.034652709960938, "learning_rate": 4.401612903225807e-05, "loss": 0.263, "step": 1855 }, { "epoch": 11.96144578313253, "grad_norm": 15.909843444824219, "learning_rate": 4.401290322580645e-05, "loss": 0.3121, "step": 1856 }, { "epoch": 11.967871485943775, "grad_norm": 10.070402145385742, "learning_rate": 4.400967741935484e-05, "loss": 0.4445, "step": 1857 }, { "epoch": 11.97429718875502, "grad_norm": 30.198211669921875, "learning_rate": 4.400645161290323e-05, "loss": 0.585, "step": 1858 }, { "epoch": 11.980722891566264, "grad_norm": 6.548479080200195, "learning_rate": 4.400322580645161e-05, "loss": 0.2985, "step": 1859 }, { "epoch": 11.98714859437751, "grad_norm": 15.730144500732422, "learning_rate": 4.4000000000000006e-05, "loss": 0.2193, "step": 1860 }, { "epoch": 11.993574297188754, "grad_norm": 15.697107315063477, "learning_rate": 4.399677419354839e-05, "loss": 0.418, "step": 1861 }, { "epoch": 12.003212851405623, "grad_norm": 11.025388717651367, "learning_rate": 4.399354838709678e-05, "loss": 0.4445, "step": 1862 }, { "epoch": 12.009638554216867, "grad_norm": 14.748648643493652, "learning_rate": 4.399032258064516e-05, "loss": 0.4281, "step": 1863 }, { "epoch": 12.016064257028113, "grad_norm": 14.413525581359863, "learning_rate": 4.398709677419355e-05, "loss": 0.8462, "step": 1864 }, { "epoch": 12.022489959839357, "grad_norm": 6.475235462188721, "learning_rate": 4.3983870967741934e-05, "loss": 0.3144, "step": 1865 }, { "epoch": 12.028915662650602, "grad_norm": 19.339435577392578, "learning_rate": 4.398064516129033e-05, "loss": 0.5725, "step": 1866 }, { "epoch": 12.035341365461848, "grad_norm": 13.711463928222656, "learning_rate": 4.397741935483871e-05, "loss": 0.2672, "step": 1867 }, { "epoch": 12.041767068273092, "grad_norm": 7.532972812652588, "learning_rate": 4.39741935483871e-05, "loss": 0.2718, "step": 1868 }, { "epoch": 12.048192771084338, "grad_norm": 12.275171279907227, "learning_rate": 4.3970967741935484e-05, "loss": 0.3772, "step": 1869 }, { "epoch": 12.054618473895582, "grad_norm": 4.019687652587891, "learning_rate": 4.3967741935483876e-05, "loss": 0.5067, "step": 1870 }, { "epoch": 12.061044176706828, "grad_norm": 9.655720710754395, "learning_rate": 4.396451612903226e-05, "loss": 0.7498, "step": 1871 }, { "epoch": 12.067469879518072, "grad_norm": 18.672374725341797, "learning_rate": 4.396129032258065e-05, "loss": 0.7495, "step": 1872 }, { "epoch": 12.073895582329317, "grad_norm": 13.33669376373291, "learning_rate": 4.395806451612903e-05, "loss": 0.3091, "step": 1873 }, { "epoch": 12.080321285140561, "grad_norm": 5.078612327575684, "learning_rate": 4.395483870967742e-05, "loss": 0.2406, "step": 1874 }, { "epoch": 12.086746987951807, "grad_norm": 10.968573570251465, "learning_rate": 4.395161290322581e-05, "loss": 0.2994, "step": 1875 }, { "epoch": 12.093172690763053, "grad_norm": 6.738646030426025, "learning_rate": 4.39483870967742e-05, "loss": 0.6805, "step": 1876 }, { "epoch": 12.099598393574297, "grad_norm": 12.426583290100098, "learning_rate": 4.394516129032258e-05, "loss": 0.3955, "step": 1877 }, { "epoch": 12.106024096385543, "grad_norm": 13.31438159942627, "learning_rate": 4.394193548387097e-05, "loss": 0.2478, "step": 1878 }, { "epoch": 12.112449799196787, "grad_norm": 23.637319564819336, "learning_rate": 4.393870967741936e-05, "loss": 0.4879, "step": 1879 }, { "epoch": 12.118875502008033, "grad_norm": 10.917484283447266, "learning_rate": 4.393548387096774e-05, "loss": 0.401, "step": 1880 }, { "epoch": 12.125301204819277, "grad_norm": 18.501352310180664, "learning_rate": 4.393225806451613e-05, "loss": 0.5588, "step": 1881 }, { "epoch": 12.131726907630522, "grad_norm": 13.336381912231445, "learning_rate": 4.392903225806452e-05, "loss": 0.5983, "step": 1882 }, { "epoch": 12.138152610441766, "grad_norm": 10.60734748840332, "learning_rate": 4.3925806451612904e-05, "loss": 0.5885, "step": 1883 }, { "epoch": 12.144578313253012, "grad_norm": 13.768803596496582, "learning_rate": 4.3922580645161296e-05, "loss": 0.3981, "step": 1884 }, { "epoch": 12.151004016064258, "grad_norm": 14.383221626281738, "learning_rate": 4.391935483870968e-05, "loss": 0.4188, "step": 1885 }, { "epoch": 12.157429718875502, "grad_norm": 16.541139602661133, "learning_rate": 4.391612903225807e-05, "loss": 0.3615, "step": 1886 }, { "epoch": 12.163855421686748, "grad_norm": 18.252681732177734, "learning_rate": 4.3912903225806454e-05, "loss": 0.3783, "step": 1887 }, { "epoch": 12.170281124497992, "grad_norm": 9.211971282958984, "learning_rate": 4.390967741935484e-05, "loss": 0.4826, "step": 1888 }, { "epoch": 12.176706827309237, "grad_norm": 7.858119964599609, "learning_rate": 4.3906451612903225e-05, "loss": 0.3442, "step": 1889 }, { "epoch": 12.183132530120481, "grad_norm": 19.39546775817871, "learning_rate": 4.390322580645162e-05, "loss": 0.4157, "step": 1890 }, { "epoch": 12.189558232931727, "grad_norm": 5.45915412902832, "learning_rate": 4.39e-05, "loss": 0.3983, "step": 1891 }, { "epoch": 12.195983935742971, "grad_norm": 5.02386999130249, "learning_rate": 4.389677419354839e-05, "loss": 0.1863, "step": 1892 }, { "epoch": 12.202409638554217, "grad_norm": 20.828580856323242, "learning_rate": 4.3893548387096775e-05, "loss": 0.4502, "step": 1893 }, { "epoch": 12.208835341365463, "grad_norm": 10.179596900939941, "learning_rate": 4.389032258064517e-05, "loss": 0.458, "step": 1894 }, { "epoch": 12.215261044176707, "grad_norm": 17.42799949645996, "learning_rate": 4.3887096774193546e-05, "loss": 0.3526, "step": 1895 }, { "epoch": 12.221686746987952, "grad_norm": 6.020331382751465, "learning_rate": 4.388387096774194e-05, "loss": 0.1801, "step": 1896 }, { "epoch": 12.228112449799196, "grad_norm": 10.082193374633789, "learning_rate": 4.3880645161290324e-05, "loss": 0.4231, "step": 1897 }, { "epoch": 12.234538152610442, "grad_norm": 3.1572794914245605, "learning_rate": 4.387741935483871e-05, "loss": 0.4279, "step": 1898 }, { "epoch": 12.240963855421686, "grad_norm": 6.43044900894165, "learning_rate": 4.38741935483871e-05, "loss": 0.4331, "step": 1899 }, { "epoch": 12.247389558232932, "grad_norm": 2.73519229888916, "learning_rate": 4.387096774193549e-05, "loss": 0.2152, "step": 1900 }, { "epoch": 12.253815261044176, "grad_norm": 6.788663864135742, "learning_rate": 4.3867741935483874e-05, "loss": 0.2967, "step": 1901 }, { "epoch": 12.260240963855422, "grad_norm": 23.919042587280273, "learning_rate": 4.386451612903226e-05, "loss": 0.5305, "step": 1902 }, { "epoch": 12.266666666666667, "grad_norm": 8.212907791137695, "learning_rate": 4.3861290322580645e-05, "loss": 0.4001, "step": 1903 }, { "epoch": 12.273092369477911, "grad_norm": 10.54776382446289, "learning_rate": 4.385806451612903e-05, "loss": 0.2843, "step": 1904 }, { "epoch": 12.279518072289157, "grad_norm": 45.87351608276367, "learning_rate": 4.3854838709677423e-05, "loss": 0.4269, "step": 1905 }, { "epoch": 12.285943775100401, "grad_norm": 27.827186584472656, "learning_rate": 4.385161290322581e-05, "loss": 0.411, "step": 1906 }, { "epoch": 12.292369477911647, "grad_norm": 9.999151229858398, "learning_rate": 4.3848387096774195e-05, "loss": 0.3416, "step": 1907 }, { "epoch": 12.298795180722891, "grad_norm": 28.245712280273438, "learning_rate": 4.384516129032258e-05, "loss": 0.6196, "step": 1908 }, { "epoch": 12.305220883534137, "grad_norm": 8.860243797302246, "learning_rate": 4.384193548387097e-05, "loss": 0.2074, "step": 1909 }, { "epoch": 12.31164658634538, "grad_norm": 20.634519577026367, "learning_rate": 4.383870967741936e-05, "loss": 0.3542, "step": 1910 }, { "epoch": 12.318072289156627, "grad_norm": 6.42080020904541, "learning_rate": 4.3835483870967744e-05, "loss": 0.3452, "step": 1911 }, { "epoch": 12.324497991967872, "grad_norm": 7.866924285888672, "learning_rate": 4.383225806451613e-05, "loss": 0.4646, "step": 1912 }, { "epoch": 12.330923694779116, "grad_norm": 19.13068389892578, "learning_rate": 4.3829032258064516e-05, "loss": 0.5519, "step": 1913 }, { "epoch": 12.337349397590362, "grad_norm": 24.803163528442383, "learning_rate": 4.382580645161291e-05, "loss": 0.3902, "step": 1914 }, { "epoch": 12.343775100401606, "grad_norm": 10.938776016235352, "learning_rate": 4.3822580645161294e-05, "loss": 0.3467, "step": 1915 }, { "epoch": 12.350200803212852, "grad_norm": 5.133962154388428, "learning_rate": 4.381935483870968e-05, "loss": 0.3024, "step": 1916 }, { "epoch": 12.356626506024096, "grad_norm": 15.267156600952148, "learning_rate": 4.3816129032258065e-05, "loss": 0.4224, "step": 1917 }, { "epoch": 12.363052208835342, "grad_norm": 10.430817604064941, "learning_rate": 4.381290322580646e-05, "loss": 0.4004, "step": 1918 }, { "epoch": 12.369477911646586, "grad_norm": 14.079564094543457, "learning_rate": 4.380967741935484e-05, "loss": 0.308, "step": 1919 }, { "epoch": 12.375903614457831, "grad_norm": 22.706151962280273, "learning_rate": 4.380645161290323e-05, "loss": 0.8072, "step": 1920 }, { "epoch": 12.382329317269077, "grad_norm": 4.451821327209473, "learning_rate": 4.3803225806451615e-05, "loss": 0.7027, "step": 1921 }, { "epoch": 12.388755020080321, "grad_norm": 4.040282726287842, "learning_rate": 4.38e-05, "loss": 0.1512, "step": 1922 }, { "epoch": 12.395180722891567, "grad_norm": 6.872121334075928, "learning_rate": 4.379677419354839e-05, "loss": 0.5167, "step": 1923 }, { "epoch": 12.401606425702811, "grad_norm": 17.7385196685791, "learning_rate": 4.379354838709678e-05, "loss": 0.4167, "step": 1924 }, { "epoch": 12.408032128514057, "grad_norm": 7.282138347625732, "learning_rate": 4.3790322580645165e-05, "loss": 0.4518, "step": 1925 }, { "epoch": 12.4144578313253, "grad_norm": 7.593144416809082, "learning_rate": 4.378709677419355e-05, "loss": 0.5236, "step": 1926 }, { "epoch": 12.420883534136546, "grad_norm": 11.531991958618164, "learning_rate": 4.3783870967741936e-05, "loss": 0.4284, "step": 1927 }, { "epoch": 12.42730923694779, "grad_norm": 18.775745391845703, "learning_rate": 4.378064516129032e-05, "loss": 0.4703, "step": 1928 }, { "epoch": 12.433734939759036, "grad_norm": 5.863308429718018, "learning_rate": 4.3777419354838714e-05, "loss": 0.1402, "step": 1929 }, { "epoch": 12.440160642570282, "grad_norm": 18.292892456054688, "learning_rate": 4.37741935483871e-05, "loss": 0.697, "step": 1930 }, { "epoch": 12.446586345381526, "grad_norm": 18.722850799560547, "learning_rate": 4.3770967741935486e-05, "loss": 0.3887, "step": 1931 }, { "epoch": 12.453012048192772, "grad_norm": 11.864858627319336, "learning_rate": 4.376774193548387e-05, "loss": 0.4983, "step": 1932 }, { "epoch": 12.459437751004016, "grad_norm": 10.599403381347656, "learning_rate": 4.3764516129032264e-05, "loss": 0.1904, "step": 1933 }, { "epoch": 12.465863453815262, "grad_norm": 7.619869232177734, "learning_rate": 4.376129032258064e-05, "loss": 0.6143, "step": 1934 }, { "epoch": 12.472289156626506, "grad_norm": 9.572484016418457, "learning_rate": 4.3758064516129035e-05, "loss": 0.9354, "step": 1935 }, { "epoch": 12.478714859437751, "grad_norm": 9.52208137512207, "learning_rate": 4.375483870967742e-05, "loss": 0.2716, "step": 1936 }, { "epoch": 12.485140562248995, "grad_norm": 13.319890022277832, "learning_rate": 4.375161290322581e-05, "loss": 0.5322, "step": 1937 }, { "epoch": 12.491566265060241, "grad_norm": 12.738389015197754, "learning_rate": 4.37483870967742e-05, "loss": 0.4307, "step": 1938 }, { "epoch": 12.497991967871485, "grad_norm": 5.79206657409668, "learning_rate": 4.3745161290322585e-05, "loss": 0.399, "step": 1939 }, { "epoch": 12.50441767068273, "grad_norm": 18.611858367919922, "learning_rate": 4.374193548387097e-05, "loss": 0.4528, "step": 1940 }, { "epoch": 12.510843373493977, "grad_norm": 7.409040927886963, "learning_rate": 4.3738709677419356e-05, "loss": 0.2914, "step": 1941 }, { "epoch": 12.51726907630522, "grad_norm": 14.21033000946045, "learning_rate": 4.373548387096774e-05, "loss": 0.4186, "step": 1942 }, { "epoch": 12.523694779116466, "grad_norm": 20.67784881591797, "learning_rate": 4.373225806451613e-05, "loss": 0.4617, "step": 1943 }, { "epoch": 12.53012048192771, "grad_norm": 8.538625717163086, "learning_rate": 4.372903225806452e-05, "loss": 0.3642, "step": 1944 }, { "epoch": 12.536546184738956, "grad_norm": 12.15466594696045, "learning_rate": 4.3725806451612906e-05, "loss": 0.5103, "step": 1945 }, { "epoch": 12.5429718875502, "grad_norm": 12.172327041625977, "learning_rate": 4.372258064516129e-05, "loss": 0.5348, "step": 1946 }, { "epoch": 12.549397590361446, "grad_norm": 20.54591941833496, "learning_rate": 4.3719354838709684e-05, "loss": 0.6289, "step": 1947 }, { "epoch": 12.55582329317269, "grad_norm": 12.405083656311035, "learning_rate": 4.371612903225807e-05, "loss": 0.2441, "step": 1948 }, { "epoch": 12.562248995983936, "grad_norm": 7.735447883605957, "learning_rate": 4.3712903225806456e-05, "loss": 0.5335, "step": 1949 }, { "epoch": 12.568674698795181, "grad_norm": 38.11739730834961, "learning_rate": 4.370967741935484e-05, "loss": 0.3862, "step": 1950 }, { "epoch": 12.575100401606425, "grad_norm": 13.798151969909668, "learning_rate": 4.370645161290323e-05, "loss": 0.4696, "step": 1951 }, { "epoch": 12.581526104417671, "grad_norm": 22.9470272064209, "learning_rate": 4.370322580645161e-05, "loss": 0.4617, "step": 1952 }, { "epoch": 12.587951807228915, "grad_norm": 6.076642036437988, "learning_rate": 4.3700000000000005e-05, "loss": 0.2838, "step": 1953 }, { "epoch": 12.594377510040161, "grad_norm": 8.804183959960938, "learning_rate": 4.369677419354839e-05, "loss": 0.4278, "step": 1954 }, { "epoch": 12.600803212851405, "grad_norm": 24.704307556152344, "learning_rate": 4.3693548387096777e-05, "loss": 0.6514, "step": 1955 }, { "epoch": 12.60722891566265, "grad_norm": 11.493000984191895, "learning_rate": 4.369032258064516e-05, "loss": 0.6005, "step": 1956 }, { "epoch": 12.613654618473895, "grad_norm": 3.9955849647521973, "learning_rate": 4.368709677419355e-05, "loss": 0.3987, "step": 1957 }, { "epoch": 12.62008032128514, "grad_norm": 8.46557903289795, "learning_rate": 4.3683870967741934e-05, "loss": 0.4784, "step": 1958 }, { "epoch": 12.626506024096386, "grad_norm": 22.86375617980957, "learning_rate": 4.3680645161290326e-05, "loss": 0.268, "step": 1959 }, { "epoch": 12.63293172690763, "grad_norm": 19.060571670532227, "learning_rate": 4.367741935483871e-05, "loss": 0.1798, "step": 1960 }, { "epoch": 12.639357429718876, "grad_norm": 21.86528968811035, "learning_rate": 4.36741935483871e-05, "loss": 0.4195, "step": 1961 }, { "epoch": 12.64578313253012, "grad_norm": 5.293360233306885, "learning_rate": 4.367096774193549e-05, "loss": 0.2157, "step": 1962 }, { "epoch": 12.652208835341366, "grad_norm": 8.862184524536133, "learning_rate": 4.3667741935483876e-05, "loss": 0.4884, "step": 1963 }, { "epoch": 12.65863453815261, "grad_norm": 8.081490516662598, "learning_rate": 4.366451612903226e-05, "loss": 0.3187, "step": 1964 }, { "epoch": 12.665060240963856, "grad_norm": 7.916811943054199, "learning_rate": 4.366129032258065e-05, "loss": 0.4569, "step": 1965 }, { "epoch": 12.6714859437751, "grad_norm": 6.624551296234131, "learning_rate": 4.365806451612903e-05, "loss": 0.4725, "step": 1966 }, { "epoch": 12.677911646586345, "grad_norm": 30.63180923461914, "learning_rate": 4.365483870967742e-05, "loss": 0.5586, "step": 1967 }, { "epoch": 12.684337349397591, "grad_norm": 9.888032913208008, "learning_rate": 4.365161290322581e-05, "loss": 0.607, "step": 1968 }, { "epoch": 12.690763052208835, "grad_norm": 8.808504104614258, "learning_rate": 4.36483870967742e-05, "loss": 0.5299, "step": 1969 }, { "epoch": 12.69718875502008, "grad_norm": 11.97283935546875, "learning_rate": 4.364516129032258e-05, "loss": 0.2846, "step": 1970 }, { "epoch": 12.703614457831325, "grad_norm": 55.10110092163086, "learning_rate": 4.364193548387097e-05, "loss": 0.3401, "step": 1971 }, { "epoch": 12.71004016064257, "grad_norm": 26.15562629699707, "learning_rate": 4.363870967741936e-05, "loss": 0.4042, "step": 1972 }, { "epoch": 12.716465863453815, "grad_norm": 13.758146286010742, "learning_rate": 4.363548387096774e-05, "loss": 0.3001, "step": 1973 }, { "epoch": 12.72289156626506, "grad_norm": 7.509040832519531, "learning_rate": 4.363225806451613e-05, "loss": 0.4652, "step": 1974 }, { "epoch": 12.729317269076304, "grad_norm": 9.690409660339355, "learning_rate": 4.362903225806452e-05, "loss": 0.534, "step": 1975 }, { "epoch": 12.73574297188755, "grad_norm": 12.69223403930664, "learning_rate": 4.3625806451612903e-05, "loss": 0.5952, "step": 1976 }, { "epoch": 12.742168674698796, "grad_norm": 13.033437728881836, "learning_rate": 4.3622580645161296e-05, "loss": 0.3392, "step": 1977 }, { "epoch": 12.74859437751004, "grad_norm": 6.386074542999268, "learning_rate": 4.361935483870968e-05, "loss": 0.6032, "step": 1978 }, { "epoch": 12.755020080321286, "grad_norm": 7.364192485809326, "learning_rate": 4.361612903225807e-05, "loss": 0.3923, "step": 1979 }, { "epoch": 12.76144578313253, "grad_norm": 13.011807441711426, "learning_rate": 4.361290322580645e-05, "loss": 0.477, "step": 1980 }, { "epoch": 12.767871485943775, "grad_norm": 7.176928520202637, "learning_rate": 4.360967741935484e-05, "loss": 0.3101, "step": 1981 }, { "epoch": 12.77429718875502, "grad_norm": 19.326263427734375, "learning_rate": 4.3606451612903224e-05, "loss": 0.3558, "step": 1982 }, { "epoch": 12.780722891566265, "grad_norm": 5.59330940246582, "learning_rate": 4.360322580645162e-05, "loss": 0.24, "step": 1983 }, { "epoch": 12.78714859437751, "grad_norm": 8.092485427856445, "learning_rate": 4.36e-05, "loss": 0.2911, "step": 1984 }, { "epoch": 12.793574297188755, "grad_norm": 15.36271858215332, "learning_rate": 4.359677419354839e-05, "loss": 0.8447, "step": 1985 }, { "epoch": 12.8, "grad_norm": 22.198326110839844, "learning_rate": 4.359354838709678e-05, "loss": 0.625, "step": 1986 }, { "epoch": 12.806425702811245, "grad_norm": 7.536934852600098, "learning_rate": 4.3590322580645167e-05, "loss": 0.3072, "step": 1987 }, { "epoch": 12.81285140562249, "grad_norm": 25.006420135498047, "learning_rate": 4.358709677419355e-05, "loss": 0.2971, "step": 1988 }, { "epoch": 12.819277108433734, "grad_norm": 8.593048095703125, "learning_rate": 4.358387096774194e-05, "loss": 0.2886, "step": 1989 }, { "epoch": 12.82570281124498, "grad_norm": 6.386456489562988, "learning_rate": 4.3580645161290324e-05, "loss": 0.3955, "step": 1990 }, { "epoch": 12.832128514056224, "grad_norm": 6.317018032073975, "learning_rate": 4.357741935483871e-05, "loss": 0.2286, "step": 1991 }, { "epoch": 12.83855421686747, "grad_norm": 30.78919219970703, "learning_rate": 4.35741935483871e-05, "loss": 0.424, "step": 1992 }, { "epoch": 12.844979919678714, "grad_norm": 18.011011123657227, "learning_rate": 4.357096774193549e-05, "loss": 0.4692, "step": 1993 }, { "epoch": 12.85140562248996, "grad_norm": 6.694151878356934, "learning_rate": 4.356774193548387e-05, "loss": 0.5357, "step": 1994 }, { "epoch": 12.857831325301206, "grad_norm": 14.890731811523438, "learning_rate": 4.356451612903226e-05, "loss": 0.4498, "step": 1995 }, { "epoch": 12.86425702811245, "grad_norm": 14.337540626525879, "learning_rate": 4.3561290322580645e-05, "loss": 0.4064, "step": 1996 }, { "epoch": 12.870682730923695, "grad_norm": 9.769471168518066, "learning_rate": 4.355806451612903e-05, "loss": 0.4912, "step": 1997 }, { "epoch": 12.87710843373494, "grad_norm": 12.22624683380127, "learning_rate": 4.355483870967742e-05, "loss": 0.3521, "step": 1998 }, { "epoch": 12.883534136546185, "grad_norm": 12.04348373413086, "learning_rate": 4.355161290322581e-05, "loss": 0.6449, "step": 1999 }, { "epoch": 12.889959839357429, "grad_norm": 10.87618350982666, "learning_rate": 4.3548387096774194e-05, "loss": 0.6624, "step": 2000 }, { "epoch": 12.896385542168675, "grad_norm": 21.211523056030273, "learning_rate": 4.354516129032259e-05, "loss": 0.4704, "step": 2001 }, { "epoch": 12.902811244979919, "grad_norm": 19.98280906677246, "learning_rate": 4.354193548387097e-05, "loss": 0.3453, "step": 2002 }, { "epoch": 12.909236947791165, "grad_norm": 9.722183227539062, "learning_rate": 4.353870967741936e-05, "loss": 0.4222, "step": 2003 }, { "epoch": 12.91566265060241, "grad_norm": 4.576184272766113, "learning_rate": 4.3535483870967744e-05, "loss": 0.2308, "step": 2004 }, { "epoch": 12.922088353413654, "grad_norm": 20.767866134643555, "learning_rate": 4.353225806451613e-05, "loss": 0.3585, "step": 2005 }, { "epoch": 12.9285140562249, "grad_norm": 9.56954574584961, "learning_rate": 4.3529032258064515e-05, "loss": 0.4164, "step": 2006 }, { "epoch": 12.934939759036144, "grad_norm": 9.101359367370605, "learning_rate": 4.352580645161291e-05, "loss": 0.3701, "step": 2007 }, { "epoch": 12.94136546184739, "grad_norm": 19.939361572265625, "learning_rate": 4.3522580645161294e-05, "loss": 0.5722, "step": 2008 }, { "epoch": 12.947791164658634, "grad_norm": 10.152958869934082, "learning_rate": 4.351935483870968e-05, "loss": 0.5206, "step": 2009 }, { "epoch": 12.95421686746988, "grad_norm": 9.762393951416016, "learning_rate": 4.351612903225807e-05, "loss": 0.6129, "step": 2010 }, { "epoch": 12.960642570281124, "grad_norm": 26.417129516601562, "learning_rate": 4.351290322580646e-05, "loss": 0.416, "step": 2011 }, { "epoch": 12.96706827309237, "grad_norm": 21.15757942199707, "learning_rate": 4.3509677419354836e-05, "loss": 0.6232, "step": 2012 }, { "epoch": 12.973493975903615, "grad_norm": 6.581351280212402, "learning_rate": 4.350645161290323e-05, "loss": 0.1459, "step": 2013 }, { "epoch": 12.97991967871486, "grad_norm": 9.825797080993652, "learning_rate": 4.3503225806451615e-05, "loss": 0.5795, "step": 2014 }, { "epoch": 12.986345381526105, "grad_norm": 8.63624095916748, "learning_rate": 4.35e-05, "loss": 0.3764, "step": 2015 }, { "epoch": 12.992771084337349, "grad_norm": 12.208728790283203, "learning_rate": 4.349677419354839e-05, "loss": 0.6704, "step": 2016 }, { "epoch": 13.002409638554218, "grad_norm": 6.297462463378906, "learning_rate": 4.349354838709678e-05, "loss": 0.4101, "step": 2017 }, { "epoch": 13.008835341365462, "grad_norm": 4.97869873046875, "learning_rate": 4.3490322580645164e-05, "loss": 0.208, "step": 2018 }, { "epoch": 13.015261044176707, "grad_norm": 11.74799633026123, "learning_rate": 4.348709677419355e-05, "loss": 0.2836, "step": 2019 }, { "epoch": 13.021686746987951, "grad_norm": 47.028587341308594, "learning_rate": 4.3483870967741936e-05, "loss": 0.6382, "step": 2020 }, { "epoch": 13.028112449799197, "grad_norm": 3.3930890560150146, "learning_rate": 4.348064516129032e-05, "loss": 0.3892, "step": 2021 }, { "epoch": 13.034538152610441, "grad_norm": 3.1190922260284424, "learning_rate": 4.3477419354838714e-05, "loss": 0.2103, "step": 2022 }, { "epoch": 13.040963855421687, "grad_norm": 20.13909912109375, "learning_rate": 4.34741935483871e-05, "loss": 0.3054, "step": 2023 }, { "epoch": 13.047389558232931, "grad_norm": 18.87809181213379, "learning_rate": 4.3470967741935485e-05, "loss": 0.6525, "step": 2024 }, { "epoch": 13.053815261044177, "grad_norm": 10.801161766052246, "learning_rate": 4.346774193548388e-05, "loss": 0.4759, "step": 2025 }, { "epoch": 13.060240963855422, "grad_norm": 7.565611362457275, "learning_rate": 4.346451612903226e-05, "loss": 0.5177, "step": 2026 }, { "epoch": 13.066666666666666, "grad_norm": 4.793542861938477, "learning_rate": 4.346129032258064e-05, "loss": 0.3287, "step": 2027 }, { "epoch": 13.073092369477912, "grad_norm": 4.063715934753418, "learning_rate": 4.3458064516129035e-05, "loss": 0.2213, "step": 2028 }, { "epoch": 13.079518072289156, "grad_norm": 9.621849060058594, "learning_rate": 4.345483870967742e-05, "loss": 0.3359, "step": 2029 }, { "epoch": 13.085943775100402, "grad_norm": 9.056136131286621, "learning_rate": 4.3451612903225806e-05, "loss": 0.3142, "step": 2030 }, { "epoch": 13.092369477911646, "grad_norm": 18.02399253845215, "learning_rate": 4.34483870967742e-05, "loss": 0.4769, "step": 2031 }, { "epoch": 13.098795180722892, "grad_norm": 5.254321098327637, "learning_rate": 4.3445161290322584e-05, "loss": 0.3897, "step": 2032 }, { "epoch": 13.105220883534136, "grad_norm": 8.19043254852295, "learning_rate": 4.344193548387097e-05, "loss": 0.3035, "step": 2033 }, { "epoch": 13.111646586345381, "grad_norm": 11.534531593322754, "learning_rate": 4.3438709677419356e-05, "loss": 0.6342, "step": 2034 }, { "epoch": 13.118072289156627, "grad_norm": 8.977028846740723, "learning_rate": 4.343548387096774e-05, "loss": 0.4882, "step": 2035 }, { "epoch": 13.124497991967871, "grad_norm": 9.814233779907227, "learning_rate": 4.343225806451613e-05, "loss": 0.5407, "step": 2036 }, { "epoch": 13.130923694779117, "grad_norm": 7.077760219573975, "learning_rate": 4.342903225806452e-05, "loss": 0.4284, "step": 2037 }, { "epoch": 13.137349397590361, "grad_norm": 19.34929847717285, "learning_rate": 4.3425806451612905e-05, "loss": 0.569, "step": 2038 }, { "epoch": 13.143775100401607, "grad_norm": 21.02012825012207, "learning_rate": 4.342258064516129e-05, "loss": 0.3447, "step": 2039 }, { "epoch": 13.15020080321285, "grad_norm": 11.805845260620117, "learning_rate": 4.3419354838709684e-05, "loss": 0.2587, "step": 2040 }, { "epoch": 13.156626506024097, "grad_norm": 5.89279317855835, "learning_rate": 4.341612903225807e-05, "loss": 0.3909, "step": 2041 }, { "epoch": 13.16305220883534, "grad_norm": 7.947062015533447, "learning_rate": 4.3412903225806455e-05, "loss": 0.5332, "step": 2042 }, { "epoch": 13.169477911646586, "grad_norm": 10.785028457641602, "learning_rate": 4.340967741935484e-05, "loss": 0.2319, "step": 2043 }, { "epoch": 13.175903614457832, "grad_norm": 15.476936340332031, "learning_rate": 4.3406451612903226e-05, "loss": 0.8248, "step": 2044 }, { "epoch": 13.182329317269076, "grad_norm": 6.271817207336426, "learning_rate": 4.340322580645161e-05, "loss": 0.5048, "step": 2045 }, { "epoch": 13.188755020080322, "grad_norm": 8.659668922424316, "learning_rate": 4.3400000000000005e-05, "loss": 0.7348, "step": 2046 }, { "epoch": 13.195180722891566, "grad_norm": 8.019754409790039, "learning_rate": 4.339677419354839e-05, "loss": 0.5608, "step": 2047 }, { "epoch": 13.201606425702812, "grad_norm": 9.193039894104004, "learning_rate": 4.3393548387096776e-05, "loss": 0.3543, "step": 2048 }, { "epoch": 13.208032128514056, "grad_norm": 2.406906843185425, "learning_rate": 4.339032258064517e-05, "loss": 0.163, "step": 2049 }, { "epoch": 13.214457831325301, "grad_norm": 8.761470794677734, "learning_rate": 4.3387096774193554e-05, "loss": 0.4592, "step": 2050 }, { "epoch": 13.220883534136545, "grad_norm": 9.867949485778809, "learning_rate": 4.338387096774193e-05, "loss": 0.5505, "step": 2051 }, { "epoch": 13.227309236947791, "grad_norm": 9.009355545043945, "learning_rate": 4.3380645161290326e-05, "loss": 0.3044, "step": 2052 }, { "epoch": 13.233734939759037, "grad_norm": 24.494218826293945, "learning_rate": 4.337741935483871e-05, "loss": 0.4286, "step": 2053 }, { "epoch": 13.240160642570281, "grad_norm": 7.369952201843262, "learning_rate": 4.33741935483871e-05, "loss": 0.1752, "step": 2054 }, { "epoch": 13.246586345381527, "grad_norm": 14.874293327331543, "learning_rate": 4.337096774193549e-05, "loss": 0.5578, "step": 2055 }, { "epoch": 13.25301204819277, "grad_norm": 5.844465255737305, "learning_rate": 4.3367741935483875e-05, "loss": 0.264, "step": 2056 }, { "epoch": 13.259437751004016, "grad_norm": 8.585806846618652, "learning_rate": 4.336451612903226e-05, "loss": 0.6226, "step": 2057 }, { "epoch": 13.26586345381526, "grad_norm": 9.093049049377441, "learning_rate": 4.336129032258065e-05, "loss": 0.4506, "step": 2058 }, { "epoch": 13.272289156626506, "grad_norm": 15.04317855834961, "learning_rate": 4.335806451612903e-05, "loss": 0.4571, "step": 2059 }, { "epoch": 13.27871485943775, "grad_norm": 11.796319961547852, "learning_rate": 4.335483870967742e-05, "loss": 0.3144, "step": 2060 }, { "epoch": 13.285140562248996, "grad_norm": 4.74921989440918, "learning_rate": 4.335161290322581e-05, "loss": 0.2715, "step": 2061 }, { "epoch": 13.291566265060242, "grad_norm": 14.349250793457031, "learning_rate": 4.3348387096774196e-05, "loss": 0.5443, "step": 2062 }, { "epoch": 13.297991967871486, "grad_norm": 5.49673318862915, "learning_rate": 4.334516129032258e-05, "loss": 0.5013, "step": 2063 }, { "epoch": 13.304417670682732, "grad_norm": 4.8104352951049805, "learning_rate": 4.3341935483870974e-05, "loss": 0.237, "step": 2064 }, { "epoch": 13.310843373493976, "grad_norm": 18.14663314819336, "learning_rate": 4.333870967741936e-05, "loss": 0.244, "step": 2065 }, { "epoch": 13.317269076305221, "grad_norm": 8.504544258117676, "learning_rate": 4.333548387096774e-05, "loss": 0.5213, "step": 2066 }, { "epoch": 13.323694779116465, "grad_norm": 11.265205383300781, "learning_rate": 4.333225806451613e-05, "loss": 0.5569, "step": 2067 }, { "epoch": 13.330120481927711, "grad_norm": 15.284006118774414, "learning_rate": 4.332903225806452e-05, "loss": 0.5333, "step": 2068 }, { "epoch": 13.336546184738955, "grad_norm": 5.612270832061768, "learning_rate": 4.33258064516129e-05, "loss": 0.3854, "step": 2069 }, { "epoch": 13.3429718875502, "grad_norm": 14.438019752502441, "learning_rate": 4.3322580645161295e-05, "loss": 0.5642, "step": 2070 }, { "epoch": 13.349397590361447, "grad_norm": 10.048796653747559, "learning_rate": 4.331935483870968e-05, "loss": 0.2407, "step": 2071 }, { "epoch": 13.35582329317269, "grad_norm": 13.09814167022705, "learning_rate": 4.331612903225807e-05, "loss": 0.481, "step": 2072 }, { "epoch": 13.362248995983936, "grad_norm": 24.51125717163086, "learning_rate": 4.331290322580646e-05, "loss": 0.2649, "step": 2073 }, { "epoch": 13.36867469879518, "grad_norm": 17.903667449951172, "learning_rate": 4.330967741935484e-05, "loss": 0.3888, "step": 2074 }, { "epoch": 13.375100401606426, "grad_norm": 17.786928176879883, "learning_rate": 4.3306451612903224e-05, "loss": 0.3641, "step": 2075 }, { "epoch": 13.38152610441767, "grad_norm": 7.753917694091797, "learning_rate": 4.3303225806451616e-05, "loss": 0.4574, "step": 2076 }, { "epoch": 13.387951807228916, "grad_norm": 4.629144191741943, "learning_rate": 4.33e-05, "loss": 0.2077, "step": 2077 }, { "epoch": 13.39437751004016, "grad_norm": 17.33765411376953, "learning_rate": 4.329677419354839e-05, "loss": 0.7684, "step": 2078 }, { "epoch": 13.400803212851406, "grad_norm": 3.6721932888031006, "learning_rate": 4.329354838709678e-05, "loss": 0.1993, "step": 2079 }, { "epoch": 13.407228915662651, "grad_norm": 14.255911827087402, "learning_rate": 4.3290322580645166e-05, "loss": 0.5265, "step": 2080 }, { "epoch": 13.413654618473895, "grad_norm": 15.248600006103516, "learning_rate": 4.328709677419355e-05, "loss": 0.5417, "step": 2081 }, { "epoch": 13.420080321285141, "grad_norm": 10.005661010742188, "learning_rate": 4.328387096774194e-05, "loss": 0.4241, "step": 2082 }, { "epoch": 13.426506024096385, "grad_norm": 7.688076972961426, "learning_rate": 4.328064516129032e-05, "loss": 0.6676, "step": 2083 }, { "epoch": 13.432931726907631, "grad_norm": 8.157419204711914, "learning_rate": 4.327741935483871e-05, "loss": 0.4281, "step": 2084 }, { "epoch": 13.439357429718875, "grad_norm": 10.88434886932373, "learning_rate": 4.32741935483871e-05, "loss": 0.5018, "step": 2085 }, { "epoch": 13.44578313253012, "grad_norm": 21.582809448242188, "learning_rate": 4.327096774193549e-05, "loss": 0.5869, "step": 2086 }, { "epoch": 13.452208835341365, "grad_norm": 13.455031394958496, "learning_rate": 4.326774193548387e-05, "loss": 0.2742, "step": 2087 }, { "epoch": 13.45863453815261, "grad_norm": 24.08326530456543, "learning_rate": 4.3264516129032265e-05, "loss": 0.3375, "step": 2088 }, { "epoch": 13.465060240963856, "grad_norm": 22.21592140197754, "learning_rate": 4.326129032258065e-05, "loss": 0.5213, "step": 2089 }, { "epoch": 13.4714859437751, "grad_norm": 15.444645881652832, "learning_rate": 4.325806451612903e-05, "loss": 0.6307, "step": 2090 }, { "epoch": 13.477911646586346, "grad_norm": 27.19178581237793, "learning_rate": 4.325483870967742e-05, "loss": 0.463, "step": 2091 }, { "epoch": 13.48433734939759, "grad_norm": 10.712075233459473, "learning_rate": 4.325161290322581e-05, "loss": 0.2703, "step": 2092 }, { "epoch": 13.490763052208836, "grad_norm": 7.315849781036377, "learning_rate": 4.3248387096774194e-05, "loss": 0.4019, "step": 2093 }, { "epoch": 13.49718875502008, "grad_norm": 8.24238109588623, "learning_rate": 4.3245161290322586e-05, "loss": 0.1719, "step": 2094 }, { "epoch": 13.503614457831326, "grad_norm": 9.058215141296387, "learning_rate": 4.324193548387097e-05, "loss": 0.473, "step": 2095 }, { "epoch": 13.51004016064257, "grad_norm": 5.855128765106201, "learning_rate": 4.323870967741936e-05, "loss": 0.5313, "step": 2096 }, { "epoch": 13.516465863453815, "grad_norm": 16.581077575683594, "learning_rate": 4.3235483870967743e-05, "loss": 0.4558, "step": 2097 }, { "epoch": 13.522891566265061, "grad_norm": 18.78167724609375, "learning_rate": 4.323225806451613e-05, "loss": 0.2758, "step": 2098 }, { "epoch": 13.529317269076305, "grad_norm": 18.9476261138916, "learning_rate": 4.3229032258064515e-05, "loss": 0.3446, "step": 2099 }, { "epoch": 13.53574297188755, "grad_norm": 9.317782402038574, "learning_rate": 4.322580645161291e-05, "loss": 0.6221, "step": 2100 }, { "epoch": 13.542168674698795, "grad_norm": 27.18820571899414, "learning_rate": 4.322258064516129e-05, "loss": 0.4641, "step": 2101 }, { "epoch": 13.54859437751004, "grad_norm": 6.461639881134033, "learning_rate": 4.321935483870968e-05, "loss": 0.3869, "step": 2102 }, { "epoch": 13.555020080321285, "grad_norm": 7.483946323394775, "learning_rate": 4.321612903225807e-05, "loss": 0.3186, "step": 2103 }, { "epoch": 13.56144578313253, "grad_norm": 10.922788619995117, "learning_rate": 4.321290322580646e-05, "loss": 0.3415, "step": 2104 }, { "epoch": 13.567871485943774, "grad_norm": 11.755668640136719, "learning_rate": 4.3209677419354836e-05, "loss": 0.3872, "step": 2105 }, { "epoch": 13.57429718875502, "grad_norm": 11.763670921325684, "learning_rate": 4.320645161290323e-05, "loss": 0.3755, "step": 2106 }, { "epoch": 13.580722891566266, "grad_norm": 15.350937843322754, "learning_rate": 4.3203225806451614e-05, "loss": 0.4843, "step": 2107 }, { "epoch": 13.58714859437751, "grad_norm": 10.545384407043457, "learning_rate": 4.32e-05, "loss": 0.2921, "step": 2108 }, { "epoch": 13.593574297188756, "grad_norm": 17.110740661621094, "learning_rate": 4.319677419354839e-05, "loss": 0.8793, "step": 2109 }, { "epoch": 13.6, "grad_norm": 17.888587951660156, "learning_rate": 4.319354838709678e-05, "loss": 0.6692, "step": 2110 }, { "epoch": 13.606425702811245, "grad_norm": 13.247289657592773, "learning_rate": 4.3190322580645164e-05, "loss": 0.3147, "step": 2111 }, { "epoch": 13.61285140562249, "grad_norm": 22.597658157348633, "learning_rate": 4.3187096774193556e-05, "loss": 0.3579, "step": 2112 }, { "epoch": 13.619277108433735, "grad_norm": 11.18848991394043, "learning_rate": 4.3183870967741935e-05, "loss": 0.4924, "step": 2113 }, { "epoch": 13.62570281124498, "grad_norm": 7.643991470336914, "learning_rate": 4.318064516129032e-05, "loss": 0.2834, "step": 2114 }, { "epoch": 13.632128514056225, "grad_norm": 7.371681213378906, "learning_rate": 4.317741935483871e-05, "loss": 0.3697, "step": 2115 }, { "epoch": 13.638554216867469, "grad_norm": 24.710132598876953, "learning_rate": 4.31741935483871e-05, "loss": 0.6122, "step": 2116 }, { "epoch": 13.644979919678715, "grad_norm": 7.9063334465026855, "learning_rate": 4.3170967741935485e-05, "loss": 0.1758, "step": 2117 }, { "epoch": 13.65140562248996, "grad_norm": 11.96016788482666, "learning_rate": 4.316774193548388e-05, "loss": 0.7193, "step": 2118 }, { "epoch": 13.657831325301204, "grad_norm": 15.131418228149414, "learning_rate": 4.316451612903226e-05, "loss": 1.016, "step": 2119 }, { "epoch": 13.66425702811245, "grad_norm": 21.787765502929688, "learning_rate": 4.316129032258065e-05, "loss": 0.3189, "step": 2120 }, { "epoch": 13.670682730923694, "grad_norm": 14.042656898498535, "learning_rate": 4.3158064516129034e-05, "loss": 0.4044, "step": 2121 }, { "epoch": 13.67710843373494, "grad_norm": 11.505581855773926, "learning_rate": 4.315483870967742e-05, "loss": 0.3477, "step": 2122 }, { "epoch": 13.683534136546184, "grad_norm": 6.966708660125732, "learning_rate": 4.3151612903225806e-05, "loss": 0.4793, "step": 2123 }, { "epoch": 13.68995983935743, "grad_norm": 19.8608341217041, "learning_rate": 4.31483870967742e-05, "loss": 0.5467, "step": 2124 }, { "epoch": 13.696385542168674, "grad_norm": 6.136839866638184, "learning_rate": 4.3145161290322584e-05, "loss": 0.2892, "step": 2125 }, { "epoch": 13.70281124497992, "grad_norm": 12.115589141845703, "learning_rate": 4.314193548387097e-05, "loss": 0.5804, "step": 2126 }, { "epoch": 13.709236947791165, "grad_norm": 10.95750617980957, "learning_rate": 4.313870967741936e-05, "loss": 0.6034, "step": 2127 }, { "epoch": 13.71566265060241, "grad_norm": 14.691605567932129, "learning_rate": 4.313548387096775e-05, "loss": 0.3418, "step": 2128 }, { "epoch": 13.722088353413655, "grad_norm": 6.156759738922119, "learning_rate": 4.313225806451613e-05, "loss": 0.5706, "step": 2129 }, { "epoch": 13.728514056224899, "grad_norm": 9.186192512512207, "learning_rate": 4.312903225806452e-05, "loss": 0.3924, "step": 2130 }, { "epoch": 13.734939759036145, "grad_norm": 7.155511379241943, "learning_rate": 4.3125806451612905e-05, "loss": 0.2201, "step": 2131 }, { "epoch": 13.741365461847389, "grad_norm": 9.675512313842773, "learning_rate": 4.312258064516129e-05, "loss": 0.5128, "step": 2132 }, { "epoch": 13.747791164658635, "grad_norm": 18.765960693359375, "learning_rate": 4.311935483870968e-05, "loss": 0.4614, "step": 2133 }, { "epoch": 13.754216867469879, "grad_norm": 13.045021057128906, "learning_rate": 4.311612903225807e-05, "loss": 0.3127, "step": 2134 }, { "epoch": 13.760642570281124, "grad_norm": 7.326337814331055, "learning_rate": 4.3112903225806454e-05, "loss": 0.5657, "step": 2135 }, { "epoch": 13.76706827309237, "grad_norm": 27.003263473510742, "learning_rate": 4.310967741935484e-05, "loss": 0.6631, "step": 2136 }, { "epoch": 13.773493975903614, "grad_norm": 8.735246658325195, "learning_rate": 4.3106451612903226e-05, "loss": 0.4198, "step": 2137 }, { "epoch": 13.77991967871486, "grad_norm": 5.652395725250244, "learning_rate": 4.310322580645161e-05, "loss": 0.5821, "step": 2138 }, { "epoch": 13.786345381526104, "grad_norm": 4.716536045074463, "learning_rate": 4.3100000000000004e-05, "loss": 0.4525, "step": 2139 }, { "epoch": 13.79277108433735, "grad_norm": 14.454534530639648, "learning_rate": 4.309677419354839e-05, "loss": 0.4659, "step": 2140 }, { "epoch": 13.799196787148594, "grad_norm": 6.90958833694458, "learning_rate": 4.3093548387096775e-05, "loss": 0.3442, "step": 2141 }, { "epoch": 13.80562248995984, "grad_norm": 8.413738250732422, "learning_rate": 4.309032258064517e-05, "loss": 0.6501, "step": 2142 }, { "epoch": 13.812048192771083, "grad_norm": 27.207365036010742, "learning_rate": 4.3087096774193554e-05, "loss": 0.2814, "step": 2143 }, { "epoch": 13.81847389558233, "grad_norm": 9.861042022705078, "learning_rate": 4.308387096774193e-05, "loss": 0.3409, "step": 2144 }, { "epoch": 13.824899598393575, "grad_norm": 13.491822242736816, "learning_rate": 4.3080645161290325e-05, "loss": 0.4779, "step": 2145 }, { "epoch": 13.831325301204819, "grad_norm": 18.689119338989258, "learning_rate": 4.307741935483871e-05, "loss": 0.8425, "step": 2146 }, { "epoch": 13.837751004016065, "grad_norm": 17.33590316772461, "learning_rate": 4.3074193548387097e-05, "loss": 0.5463, "step": 2147 }, { "epoch": 13.844176706827309, "grad_norm": 16.110225677490234, "learning_rate": 4.307096774193549e-05, "loss": 0.8183, "step": 2148 }, { "epoch": 13.850602409638554, "grad_norm": 13.13866138458252, "learning_rate": 4.3067741935483875e-05, "loss": 0.2755, "step": 2149 }, { "epoch": 13.857028112449798, "grad_norm": 11.515828132629395, "learning_rate": 4.306451612903226e-05, "loss": 0.4242, "step": 2150 }, { "epoch": 13.863453815261044, "grad_norm": 9.367462158203125, "learning_rate": 4.306129032258065e-05, "loss": 0.4432, "step": 2151 }, { "epoch": 13.869879518072288, "grad_norm": 4.966350078582764, "learning_rate": 4.305806451612903e-05, "loss": 0.2673, "step": 2152 }, { "epoch": 13.876305220883534, "grad_norm": 3.6131398677825928, "learning_rate": 4.305483870967742e-05, "loss": 0.1827, "step": 2153 }, { "epoch": 13.88273092369478, "grad_norm": 20.878753662109375, "learning_rate": 4.305161290322581e-05, "loss": 0.6365, "step": 2154 }, { "epoch": 13.889156626506024, "grad_norm": 5.543981075286865, "learning_rate": 4.3048387096774196e-05, "loss": 0.1977, "step": 2155 }, { "epoch": 13.89558232931727, "grad_norm": 5.685655117034912, "learning_rate": 4.304516129032258e-05, "loss": 0.5925, "step": 2156 }, { "epoch": 13.902008032128514, "grad_norm": 2.7066659927368164, "learning_rate": 4.3041935483870974e-05, "loss": 0.1393, "step": 2157 }, { "epoch": 13.90843373493976, "grad_norm": 4.238872528076172, "learning_rate": 4.303870967741936e-05, "loss": 0.3546, "step": 2158 }, { "epoch": 13.914859437751003, "grad_norm": 4.295796871185303, "learning_rate": 4.3035483870967745e-05, "loss": 0.4387, "step": 2159 }, { "epoch": 13.921285140562249, "grad_norm": 7.672455787658691, "learning_rate": 4.303225806451613e-05, "loss": 0.3955, "step": 2160 }, { "epoch": 13.927710843373493, "grad_norm": 9.79590892791748, "learning_rate": 4.302903225806452e-05, "loss": 0.2532, "step": 2161 }, { "epoch": 13.934136546184739, "grad_norm": 24.96980857849121, "learning_rate": 4.30258064516129e-05, "loss": 0.502, "step": 2162 }, { "epoch": 13.940562248995985, "grad_norm": 31.640775680541992, "learning_rate": 4.3022580645161295e-05, "loss": 0.3179, "step": 2163 }, { "epoch": 13.946987951807229, "grad_norm": 5.720144748687744, "learning_rate": 4.301935483870968e-05, "loss": 0.3907, "step": 2164 }, { "epoch": 13.953413654618474, "grad_norm": 7.890456676483154, "learning_rate": 4.3016129032258066e-05, "loss": 0.2834, "step": 2165 }, { "epoch": 13.959839357429718, "grad_norm": 9.607283592224121, "learning_rate": 4.301290322580646e-05, "loss": 0.4693, "step": 2166 }, { "epoch": 13.966265060240964, "grad_norm": 8.64065933227539, "learning_rate": 4.3009677419354845e-05, "loss": 0.208, "step": 2167 }, { "epoch": 13.972690763052208, "grad_norm": 26.561403274536133, "learning_rate": 4.3006451612903223e-05, "loss": 0.5173, "step": 2168 }, { "epoch": 13.979116465863454, "grad_norm": 4.9514641761779785, "learning_rate": 4.3003225806451616e-05, "loss": 0.6988, "step": 2169 }, { "epoch": 13.985542168674698, "grad_norm": 6.445315361022949, "learning_rate": 4.3e-05, "loss": 0.2835, "step": 2170 }, { "epoch": 13.991967871485944, "grad_norm": 4.645116806030273, "learning_rate": 4.299677419354839e-05, "loss": 0.2028, "step": 2171 }, { "epoch": 14.00160642570281, "grad_norm": 32.90705871582031, "learning_rate": 4.299354838709678e-05, "loss": 0.3958, "step": 2172 }, { "epoch": 14.008032128514056, "grad_norm": 172.37010192871094, "learning_rate": 4.2990322580645166e-05, "loss": 0.4252, "step": 2173 }, { "epoch": 14.0144578313253, "grad_norm": 3.3374509811401367, "learning_rate": 4.298709677419355e-05, "loss": 0.3238, "step": 2174 }, { "epoch": 14.020883534136546, "grad_norm": 4.271337032318115, "learning_rate": 4.298387096774194e-05, "loss": 0.4775, "step": 2175 }, { "epoch": 14.027309236947792, "grad_norm": 8.883110046386719, "learning_rate": 4.298064516129032e-05, "loss": 0.4481, "step": 2176 }, { "epoch": 14.033734939759036, "grad_norm": 9.999893188476562, "learning_rate": 4.297741935483871e-05, "loss": 0.5354, "step": 2177 }, { "epoch": 14.040160642570282, "grad_norm": 9.588628768920898, "learning_rate": 4.29741935483871e-05, "loss": 0.4897, "step": 2178 }, { "epoch": 14.046586345381526, "grad_norm": 14.148740768432617, "learning_rate": 4.2970967741935487e-05, "loss": 0.5085, "step": 2179 }, { "epoch": 14.053012048192771, "grad_norm": 4.555102825164795, "learning_rate": 4.296774193548387e-05, "loss": 0.4331, "step": 2180 }, { "epoch": 14.059437751004015, "grad_norm": 53.92068862915039, "learning_rate": 4.2964516129032265e-05, "loss": 0.496, "step": 2181 }, { "epoch": 14.065863453815261, "grad_norm": 19.68284797668457, "learning_rate": 4.296129032258065e-05, "loss": 0.4871, "step": 2182 }, { "epoch": 14.072289156626505, "grad_norm": 8.435906410217285, "learning_rate": 4.295806451612903e-05, "loss": 0.4032, "step": 2183 }, { "epoch": 14.078714859437751, "grad_norm": 15.35134220123291, "learning_rate": 4.295483870967742e-05, "loss": 0.3792, "step": 2184 }, { "epoch": 14.085140562248997, "grad_norm": 9.422470092773438, "learning_rate": 4.295161290322581e-05, "loss": 0.2981, "step": 2185 }, { "epoch": 14.09156626506024, "grad_norm": 10.609326362609863, "learning_rate": 4.294838709677419e-05, "loss": 0.4102, "step": 2186 }, { "epoch": 14.097991967871486, "grad_norm": 8.573257446289062, "learning_rate": 4.2945161290322586e-05, "loss": 0.4028, "step": 2187 }, { "epoch": 14.10441767068273, "grad_norm": 23.066505432128906, "learning_rate": 4.294193548387097e-05, "loss": 0.5294, "step": 2188 }, { "epoch": 14.110843373493976, "grad_norm": 11.50471305847168, "learning_rate": 4.293870967741936e-05, "loss": 0.4947, "step": 2189 }, { "epoch": 14.11726907630522, "grad_norm": 4.7530694007873535, "learning_rate": 4.293548387096775e-05, "loss": 0.177, "step": 2190 }, { "epoch": 14.123694779116466, "grad_norm": 11.500153541564941, "learning_rate": 4.293225806451613e-05, "loss": 0.7365, "step": 2191 }, { "epoch": 14.13012048192771, "grad_norm": 9.959073066711426, "learning_rate": 4.2929032258064514e-05, "loss": 0.3999, "step": 2192 }, { "epoch": 14.136546184738956, "grad_norm": 21.566478729248047, "learning_rate": 4.292580645161291e-05, "loss": 0.8315, "step": 2193 }, { "epoch": 14.142971887550202, "grad_norm": 4.301250457763672, "learning_rate": 4.292258064516129e-05, "loss": 0.3575, "step": 2194 }, { "epoch": 14.149397590361446, "grad_norm": 16.019065856933594, "learning_rate": 4.291935483870968e-05, "loss": 0.6874, "step": 2195 }, { "epoch": 14.155823293172691, "grad_norm": 82.73240661621094, "learning_rate": 4.291612903225807e-05, "loss": 0.4887, "step": 2196 }, { "epoch": 14.162248995983935, "grad_norm": 4.909401893615723, "learning_rate": 4.2912903225806456e-05, "loss": 0.2817, "step": 2197 }, { "epoch": 14.168674698795181, "grad_norm": 18.374080657958984, "learning_rate": 4.290967741935484e-05, "loss": 0.4499, "step": 2198 }, { "epoch": 14.175100401606425, "grad_norm": 16.792816162109375, "learning_rate": 4.290645161290323e-05, "loss": 0.4689, "step": 2199 }, { "epoch": 14.18152610441767, "grad_norm": 19.149505615234375, "learning_rate": 4.2903225806451614e-05, "loss": 0.3923, "step": 2200 }, { "epoch": 14.187951807228915, "grad_norm": 2.9501359462738037, "learning_rate": 4.29e-05, "loss": 0.1638, "step": 2201 }, { "epoch": 14.19437751004016, "grad_norm": 7.911105155944824, "learning_rate": 4.289677419354839e-05, "loss": 0.3725, "step": 2202 }, { "epoch": 14.200803212851406, "grad_norm": 17.5534610748291, "learning_rate": 4.289354838709678e-05, "loss": 0.516, "step": 2203 }, { "epoch": 14.20722891566265, "grad_norm": 5.004327297210693, "learning_rate": 4.289032258064516e-05, "loss": 0.5552, "step": 2204 }, { "epoch": 14.213654618473896, "grad_norm": 7.614180564880371, "learning_rate": 4.2887096774193556e-05, "loss": 0.5078, "step": 2205 }, { "epoch": 14.22008032128514, "grad_norm": 4.328590393066406, "learning_rate": 4.2883870967741935e-05, "loss": 0.1543, "step": 2206 }, { "epoch": 14.226506024096386, "grad_norm": 14.108028411865234, "learning_rate": 4.288064516129032e-05, "loss": 0.4627, "step": 2207 }, { "epoch": 14.23293172690763, "grad_norm": 8.44379711151123, "learning_rate": 4.287741935483871e-05, "loss": 0.3657, "step": 2208 }, { "epoch": 14.239357429718876, "grad_norm": 9.573346138000488, "learning_rate": 4.28741935483871e-05, "loss": 0.228, "step": 2209 }, { "epoch": 14.24578313253012, "grad_norm": 10.40972900390625, "learning_rate": 4.2870967741935484e-05, "loss": 0.5075, "step": 2210 }, { "epoch": 14.252208835341365, "grad_norm": 11.527359008789062, "learning_rate": 4.286774193548388e-05, "loss": 0.4965, "step": 2211 }, { "epoch": 14.258634538152611, "grad_norm": 7.513031959533691, "learning_rate": 4.286451612903226e-05, "loss": 0.2466, "step": 2212 }, { "epoch": 14.265060240963855, "grad_norm": 13.474590301513672, "learning_rate": 4.286129032258065e-05, "loss": 0.2225, "step": 2213 }, { "epoch": 14.271485943775101, "grad_norm": 43.17463684082031, "learning_rate": 4.2858064516129034e-05, "loss": 0.4016, "step": 2214 }, { "epoch": 14.277911646586345, "grad_norm": 16.078514099121094, "learning_rate": 4.285483870967742e-05, "loss": 0.6008, "step": 2215 }, { "epoch": 14.28433734939759, "grad_norm": 12.661422729492188, "learning_rate": 4.2851612903225805e-05, "loss": 0.2819, "step": 2216 }, { "epoch": 14.290763052208835, "grad_norm": 10.922041893005371, "learning_rate": 4.28483870967742e-05, "loss": 0.5501, "step": 2217 }, { "epoch": 14.29718875502008, "grad_norm": 5.812463283538818, "learning_rate": 4.284516129032258e-05, "loss": 0.3172, "step": 2218 }, { "epoch": 14.303614457831324, "grad_norm": 19.435827255249023, "learning_rate": 4.284193548387097e-05, "loss": 0.3533, "step": 2219 }, { "epoch": 14.31004016064257, "grad_norm": 11.36717414855957, "learning_rate": 4.283870967741936e-05, "loss": 0.3284, "step": 2220 }, { "epoch": 14.316465863453816, "grad_norm": 57.26483917236328, "learning_rate": 4.283548387096775e-05, "loss": 0.399, "step": 2221 }, { "epoch": 14.32289156626506, "grad_norm": 6.572181224822998, "learning_rate": 4.2832258064516126e-05, "loss": 0.2401, "step": 2222 }, { "epoch": 14.329317269076306, "grad_norm": 13.714420318603516, "learning_rate": 4.282903225806452e-05, "loss": 0.7304, "step": 2223 }, { "epoch": 14.33574297188755, "grad_norm": 7.221745491027832, "learning_rate": 4.2825806451612904e-05, "loss": 0.4066, "step": 2224 }, { "epoch": 14.342168674698796, "grad_norm": 22.35626983642578, "learning_rate": 4.282258064516129e-05, "loss": 0.3626, "step": 2225 }, { "epoch": 14.34859437751004, "grad_norm": 7.587913990020752, "learning_rate": 4.281935483870968e-05, "loss": 0.4482, "step": 2226 }, { "epoch": 14.355020080321285, "grad_norm": 34.92123031616211, "learning_rate": 4.281612903225807e-05, "loss": 0.4283, "step": 2227 }, { "epoch": 14.36144578313253, "grad_norm": 17.428253173828125, "learning_rate": 4.2812903225806454e-05, "loss": 0.54, "step": 2228 }, { "epoch": 14.367871485943775, "grad_norm": 6.389767169952393, "learning_rate": 4.2809677419354846e-05, "loss": 0.4156, "step": 2229 }, { "epoch": 14.37429718875502, "grad_norm": 11.415511131286621, "learning_rate": 4.2806451612903225e-05, "loss": 0.4609, "step": 2230 }, { "epoch": 14.380722891566265, "grad_norm": 11.183442115783691, "learning_rate": 4.280322580645161e-05, "loss": 0.51, "step": 2231 }, { "epoch": 14.38714859437751, "grad_norm": 13.032974243164062, "learning_rate": 4.2800000000000004e-05, "loss": 0.2545, "step": 2232 }, { "epoch": 14.393574297188755, "grad_norm": 13.33880615234375, "learning_rate": 4.279677419354839e-05, "loss": 0.6092, "step": 2233 }, { "epoch": 14.4, "grad_norm": 7.962555885314941, "learning_rate": 4.2793548387096775e-05, "loss": 0.6192, "step": 2234 }, { "epoch": 14.406425702811244, "grad_norm": 8.412079811096191, "learning_rate": 4.279032258064517e-05, "loss": 0.4969, "step": 2235 }, { "epoch": 14.41285140562249, "grad_norm": 12.147238731384277, "learning_rate": 4.278709677419355e-05, "loss": 0.29, "step": 2236 }, { "epoch": 14.419277108433734, "grad_norm": 9.255400657653809, "learning_rate": 4.278387096774194e-05, "loss": 0.5203, "step": 2237 }, { "epoch": 14.42570281124498, "grad_norm": 4.5343804359436035, "learning_rate": 4.2780645161290325e-05, "loss": 0.2579, "step": 2238 }, { "epoch": 14.432128514056226, "grad_norm": 13.82616138458252, "learning_rate": 4.277741935483871e-05, "loss": 0.3845, "step": 2239 }, { "epoch": 14.43855421686747, "grad_norm": 8.172707557678223, "learning_rate": 4.2774193548387096e-05, "loss": 0.3326, "step": 2240 }, { "epoch": 14.444979919678715, "grad_norm": 6.097843170166016, "learning_rate": 4.277096774193549e-05, "loss": 0.2734, "step": 2241 }, { "epoch": 14.45140562248996, "grad_norm": 9.07337474822998, "learning_rate": 4.2767741935483874e-05, "loss": 0.2418, "step": 2242 }, { "epoch": 14.457831325301205, "grad_norm": 3.995266914367676, "learning_rate": 4.276451612903226e-05, "loss": 0.2573, "step": 2243 }, { "epoch": 14.46425702811245, "grad_norm": 9.79601764678955, "learning_rate": 4.276129032258065e-05, "loss": 0.4666, "step": 2244 }, { "epoch": 14.470682730923695, "grad_norm": 21.447107315063477, "learning_rate": 4.275806451612903e-05, "loss": 0.5542, "step": 2245 }, { "epoch": 14.477108433734939, "grad_norm": 19.016109466552734, "learning_rate": 4.275483870967742e-05, "loss": 0.2855, "step": 2246 }, { "epoch": 14.483534136546185, "grad_norm": 19.665634155273438, "learning_rate": 4.275161290322581e-05, "loss": 0.4377, "step": 2247 }, { "epoch": 14.48995983935743, "grad_norm": 6.019944190979004, "learning_rate": 4.2748387096774195e-05, "loss": 0.4374, "step": 2248 }, { "epoch": 14.496385542168674, "grad_norm": 5.201944351196289, "learning_rate": 4.274516129032258e-05, "loss": 0.279, "step": 2249 }, { "epoch": 14.50281124497992, "grad_norm": 19.268646240234375, "learning_rate": 4.2741935483870973e-05, "loss": 0.5269, "step": 2250 }, { "epoch": 14.509236947791164, "grad_norm": 10.330641746520996, "learning_rate": 4.273870967741936e-05, "loss": 0.1305, "step": 2251 }, { "epoch": 14.51566265060241, "grad_norm": 5.003353595733643, "learning_rate": 4.2735483870967745e-05, "loss": 0.4707, "step": 2252 }, { "epoch": 14.522088353413654, "grad_norm": 2.8915271759033203, "learning_rate": 4.273225806451613e-05, "loss": 0.1088, "step": 2253 }, { "epoch": 14.5285140562249, "grad_norm": 4.428996562957764, "learning_rate": 4.2729032258064516e-05, "loss": 0.487, "step": 2254 }, { "epoch": 14.534939759036144, "grad_norm": 13.411375045776367, "learning_rate": 4.27258064516129e-05, "loss": 0.6311, "step": 2255 }, { "epoch": 14.54136546184739, "grad_norm": 12.256490707397461, "learning_rate": 4.2722580645161294e-05, "loss": 0.4847, "step": 2256 }, { "epoch": 14.547791164658635, "grad_norm": 7.639462947845459, "learning_rate": 4.271935483870968e-05, "loss": 0.289, "step": 2257 }, { "epoch": 14.55421686746988, "grad_norm": 21.818166732788086, "learning_rate": 4.2716129032258066e-05, "loss": 0.5579, "step": 2258 }, { "epoch": 14.560642570281125, "grad_norm": 8.454188346862793, "learning_rate": 4.271290322580646e-05, "loss": 0.6877, "step": 2259 }, { "epoch": 14.567068273092369, "grad_norm": 12.683578491210938, "learning_rate": 4.2709677419354844e-05, "loss": 0.5824, "step": 2260 }, { "epoch": 14.573493975903615, "grad_norm": 7.082840442657471, "learning_rate": 4.270645161290322e-05, "loss": 0.5472, "step": 2261 }, { "epoch": 14.579919678714859, "grad_norm": 4.755046367645264, "learning_rate": 4.2703225806451615e-05, "loss": 0.4571, "step": 2262 }, { "epoch": 14.586345381526105, "grad_norm": 5.460975170135498, "learning_rate": 4.27e-05, "loss": 0.3247, "step": 2263 }, { "epoch": 14.592771084337349, "grad_norm": 7.2642364501953125, "learning_rate": 4.269677419354839e-05, "loss": 0.6618, "step": 2264 }, { "epoch": 14.599196787148594, "grad_norm": 12.900131225585938, "learning_rate": 4.269354838709678e-05, "loss": 0.4606, "step": 2265 }, { "epoch": 14.60562248995984, "grad_norm": 8.548208236694336, "learning_rate": 4.2690322580645165e-05, "loss": 0.3714, "step": 2266 }, { "epoch": 14.612048192771084, "grad_norm": 22.528215408325195, "learning_rate": 4.268709677419355e-05, "loss": 0.3911, "step": 2267 }, { "epoch": 14.61847389558233, "grad_norm": 9.786056518554688, "learning_rate": 4.268387096774194e-05, "loss": 0.4145, "step": 2268 }, { "epoch": 14.624899598393574, "grad_norm": 16.712791442871094, "learning_rate": 4.268064516129032e-05, "loss": 0.3111, "step": 2269 }, { "epoch": 14.63132530120482, "grad_norm": 10.781570434570312, "learning_rate": 4.267741935483871e-05, "loss": 0.2965, "step": 2270 }, { "epoch": 14.637751004016064, "grad_norm": 46.900325775146484, "learning_rate": 4.26741935483871e-05, "loss": 0.4731, "step": 2271 }, { "epoch": 14.64417670682731, "grad_norm": 10.822519302368164, "learning_rate": 4.2670967741935486e-05, "loss": 0.2737, "step": 2272 }, { "epoch": 14.650602409638553, "grad_norm": 2.1632041931152344, "learning_rate": 4.266774193548387e-05, "loss": 0.2807, "step": 2273 }, { "epoch": 14.6570281124498, "grad_norm": 4.741670608520508, "learning_rate": 4.2664516129032264e-05, "loss": 0.2925, "step": 2274 }, { "epoch": 14.663453815261045, "grad_norm": 12.811636924743652, "learning_rate": 4.266129032258065e-05, "loss": 0.7272, "step": 2275 }, { "epoch": 14.669879518072289, "grad_norm": 9.923003196716309, "learning_rate": 4.265806451612903e-05, "loss": 0.4475, "step": 2276 }, { "epoch": 14.676305220883535, "grad_norm": 3.488581895828247, "learning_rate": 4.265483870967742e-05, "loss": 0.2495, "step": 2277 }, { "epoch": 14.682730923694779, "grad_norm": 14.62210464477539, "learning_rate": 4.265161290322581e-05, "loss": 0.5064, "step": 2278 }, { "epoch": 14.689156626506024, "grad_norm": 6.6959147453308105, "learning_rate": 4.264838709677419e-05, "loss": 0.3628, "step": 2279 }, { "epoch": 14.695582329317268, "grad_norm": 9.656224250793457, "learning_rate": 4.2645161290322585e-05, "loss": 0.6889, "step": 2280 }, { "epoch": 14.702008032128514, "grad_norm": 8.42806625366211, "learning_rate": 4.264193548387097e-05, "loss": 0.4121, "step": 2281 }, { "epoch": 14.708433734939758, "grad_norm": 12.292628288269043, "learning_rate": 4.263870967741936e-05, "loss": 0.8033, "step": 2282 }, { "epoch": 14.714859437751004, "grad_norm": 8.447422981262207, "learning_rate": 4.263548387096775e-05, "loss": 0.4014, "step": 2283 }, { "epoch": 14.72128514056225, "grad_norm": 5.14626407623291, "learning_rate": 4.263225806451613e-05, "loss": 0.2997, "step": 2284 }, { "epoch": 14.727710843373494, "grad_norm": 9.496203422546387, "learning_rate": 4.2629032258064514e-05, "loss": 0.465, "step": 2285 }, { "epoch": 14.73413654618474, "grad_norm": 9.31380844116211, "learning_rate": 4.2625806451612906e-05, "loss": 0.3576, "step": 2286 }, { "epoch": 14.740562248995984, "grad_norm": 3.0964207649230957, "learning_rate": 4.262258064516129e-05, "loss": 0.2139, "step": 2287 }, { "epoch": 14.74698795180723, "grad_norm": 5.2041802406311035, "learning_rate": 4.261935483870968e-05, "loss": 0.3566, "step": 2288 }, { "epoch": 14.753413654618473, "grad_norm": 9.16838550567627, "learning_rate": 4.261612903225807e-05, "loss": 0.5269, "step": 2289 }, { "epoch": 14.759839357429719, "grad_norm": 21.235172271728516, "learning_rate": 4.2612903225806456e-05, "loss": 0.3674, "step": 2290 }, { "epoch": 14.766265060240963, "grad_norm": 28.936214447021484, "learning_rate": 4.260967741935484e-05, "loss": 0.4102, "step": 2291 }, { "epoch": 14.772690763052209, "grad_norm": 7.71586275100708, "learning_rate": 4.260645161290323e-05, "loss": 0.3199, "step": 2292 }, { "epoch": 14.779116465863455, "grad_norm": 14.06052303314209, "learning_rate": 4.260322580645161e-05, "loss": 0.6959, "step": 2293 }, { "epoch": 14.785542168674699, "grad_norm": 10.477877616882324, "learning_rate": 4.26e-05, "loss": 0.3952, "step": 2294 }, { "epoch": 14.791967871485944, "grad_norm": 17.14105224609375, "learning_rate": 4.259677419354839e-05, "loss": 0.7349, "step": 2295 }, { "epoch": 14.798393574297188, "grad_norm": 53.238529205322266, "learning_rate": 4.259354838709678e-05, "loss": 0.3701, "step": 2296 }, { "epoch": 14.804819277108434, "grad_norm": 11.340794563293457, "learning_rate": 4.259032258064516e-05, "loss": 0.6425, "step": 2297 }, { "epoch": 14.811244979919678, "grad_norm": 14.097909927368164, "learning_rate": 4.2587096774193555e-05, "loss": 0.327, "step": 2298 }, { "epoch": 14.817670682730924, "grad_norm": 5.937420845031738, "learning_rate": 4.258387096774194e-05, "loss": 0.4007, "step": 2299 }, { "epoch": 14.824096385542168, "grad_norm": 6.921764850616455, "learning_rate": 4.258064516129032e-05, "loss": 0.3666, "step": 2300 }, { "epoch": 14.830522088353414, "grad_norm": 16.933229446411133, "learning_rate": 4.257741935483871e-05, "loss": 0.369, "step": 2301 }, { "epoch": 14.83694779116466, "grad_norm": 13.46768856048584, "learning_rate": 4.25741935483871e-05, "loss": 0.3727, "step": 2302 }, { "epoch": 14.843373493975903, "grad_norm": 11.494064331054688, "learning_rate": 4.2570967741935484e-05, "loss": 0.7692, "step": 2303 }, { "epoch": 14.84979919678715, "grad_norm": 7.866414546966553, "learning_rate": 4.2567741935483876e-05, "loss": 0.5159, "step": 2304 }, { "epoch": 14.856224899598393, "grad_norm": 2.689148187637329, "learning_rate": 4.256451612903226e-05, "loss": 0.3468, "step": 2305 }, { "epoch": 14.862650602409639, "grad_norm": 7.903514862060547, "learning_rate": 4.256129032258065e-05, "loss": 0.1836, "step": 2306 }, { "epoch": 14.869076305220883, "grad_norm": 8.638612747192383, "learning_rate": 4.255806451612904e-05, "loss": 0.3431, "step": 2307 }, { "epoch": 14.875502008032129, "grad_norm": 6.731717109680176, "learning_rate": 4.255483870967742e-05, "loss": 0.6971, "step": 2308 }, { "epoch": 14.881927710843373, "grad_norm": 6.183466911315918, "learning_rate": 4.2551612903225805e-05, "loss": 0.5359, "step": 2309 }, { "epoch": 14.888353413654619, "grad_norm": 8.725085258483887, "learning_rate": 4.25483870967742e-05, "loss": 0.2451, "step": 2310 }, { "epoch": 14.894779116465864, "grad_norm": 4.154870510101318, "learning_rate": 4.254516129032258e-05, "loss": 0.2397, "step": 2311 }, { "epoch": 14.901204819277108, "grad_norm": 21.312528610229492, "learning_rate": 4.254193548387097e-05, "loss": 0.6181, "step": 2312 }, { "epoch": 14.907630522088354, "grad_norm": 15.94996452331543, "learning_rate": 4.253870967741936e-05, "loss": 0.3004, "step": 2313 }, { "epoch": 14.914056224899598, "grad_norm": 3.907435417175293, "learning_rate": 4.253548387096775e-05, "loss": 0.101, "step": 2314 }, { "epoch": 14.920481927710844, "grad_norm": 10.299281120300293, "learning_rate": 4.2532258064516126e-05, "loss": 0.3466, "step": 2315 }, { "epoch": 14.926907630522088, "grad_norm": 13.438175201416016, "learning_rate": 4.252903225806452e-05, "loss": 0.2399, "step": 2316 }, { "epoch": 14.933333333333334, "grad_norm": 5.117659091949463, "learning_rate": 4.2525806451612904e-05, "loss": 0.4755, "step": 2317 }, { "epoch": 14.939759036144578, "grad_norm": 7.2984843254089355, "learning_rate": 4.252258064516129e-05, "loss": 0.4944, "step": 2318 }, { "epoch": 14.946184738955823, "grad_norm": 11.88187026977539, "learning_rate": 4.251935483870968e-05, "loss": 0.2383, "step": 2319 }, { "epoch": 14.952610441767069, "grad_norm": 20.08172607421875, "learning_rate": 4.251612903225807e-05, "loss": 0.4777, "step": 2320 }, { "epoch": 14.959036144578313, "grad_norm": 3.411022663116455, "learning_rate": 4.2512903225806453e-05, "loss": 0.3281, "step": 2321 }, { "epoch": 14.965461847389559, "grad_norm": 12.479717254638672, "learning_rate": 4.2509677419354846e-05, "loss": 0.7007, "step": 2322 }, { "epoch": 14.971887550200803, "grad_norm": 8.110184669494629, "learning_rate": 4.2506451612903225e-05, "loss": 0.3827, "step": 2323 }, { "epoch": 14.978313253012049, "grad_norm": 7.016851425170898, "learning_rate": 4.250322580645161e-05, "loss": 0.1353, "step": 2324 }, { "epoch": 14.984738955823293, "grad_norm": 9.004602432250977, "learning_rate": 4.25e-05, "loss": 0.6334, "step": 2325 }, { "epoch": 14.991164658634538, "grad_norm": 10.884005546569824, "learning_rate": 4.249677419354839e-05, "loss": 0.5858, "step": 2326 }, { "epoch": 15.000803212851405, "grad_norm": 5.6766557693481445, "learning_rate": 4.2493548387096774e-05, "loss": 0.1686, "step": 2327 }, { "epoch": 15.007228915662651, "grad_norm": 9.604116439819336, "learning_rate": 4.249032258064517e-05, "loss": 0.846, "step": 2328 }, { "epoch": 15.013654618473895, "grad_norm": 12.83220100402832, "learning_rate": 4.248709677419355e-05, "loss": 0.4497, "step": 2329 }, { "epoch": 15.02008032128514, "grad_norm": 7.023367881774902, "learning_rate": 4.248387096774194e-05, "loss": 0.539, "step": 2330 }, { "epoch": 15.026506024096385, "grad_norm": 10.42199420928955, "learning_rate": 4.2480645161290324e-05, "loss": 0.542, "step": 2331 }, { "epoch": 15.03293172690763, "grad_norm": 7.316269397735596, "learning_rate": 4.247741935483871e-05, "loss": 0.2268, "step": 2332 }, { "epoch": 15.039357429718876, "grad_norm": 6.688636779785156, "learning_rate": 4.2474193548387095e-05, "loss": 0.3289, "step": 2333 }, { "epoch": 15.04578313253012, "grad_norm": 15.670218467712402, "learning_rate": 4.247096774193549e-05, "loss": 0.255, "step": 2334 }, { "epoch": 15.052208835341366, "grad_norm": 7.564036846160889, "learning_rate": 4.2467741935483874e-05, "loss": 0.3692, "step": 2335 }, { "epoch": 15.05863453815261, "grad_norm": 10.932005882263184, "learning_rate": 4.246451612903226e-05, "loss": 0.5852, "step": 2336 }, { "epoch": 15.065060240963856, "grad_norm": 10.231724739074707, "learning_rate": 4.246129032258065e-05, "loss": 0.785, "step": 2337 }, { "epoch": 15.0714859437751, "grad_norm": 4.292795658111572, "learning_rate": 4.245806451612904e-05, "loss": 0.203, "step": 2338 }, { "epoch": 15.077911646586346, "grad_norm": 8.123945236206055, "learning_rate": 4.2454838709677417e-05, "loss": 0.6201, "step": 2339 }, { "epoch": 15.08433734939759, "grad_norm": 6.585451126098633, "learning_rate": 4.245161290322581e-05, "loss": 0.5083, "step": 2340 }, { "epoch": 15.090763052208835, "grad_norm": 7.1013007164001465, "learning_rate": 4.2448387096774195e-05, "loss": 0.4748, "step": 2341 }, { "epoch": 15.097188755020081, "grad_norm": 16.2506160736084, "learning_rate": 4.244516129032258e-05, "loss": 0.5761, "step": 2342 }, { "epoch": 15.103614457831325, "grad_norm": 13.429302215576172, "learning_rate": 4.244193548387097e-05, "loss": 0.3952, "step": 2343 }, { "epoch": 15.110040160642571, "grad_norm": 6.341864109039307, "learning_rate": 4.243870967741936e-05, "loss": 0.3006, "step": 2344 }, { "epoch": 15.116465863453815, "grad_norm": 23.40614128112793, "learning_rate": 4.2435483870967744e-05, "loss": 0.7563, "step": 2345 }, { "epoch": 15.12289156626506, "grad_norm": 6.352750301361084, "learning_rate": 4.243225806451614e-05, "loss": 0.5566, "step": 2346 }, { "epoch": 15.129317269076305, "grad_norm": 17.941740036010742, "learning_rate": 4.2429032258064516e-05, "loss": 0.5245, "step": 2347 }, { "epoch": 15.13574297188755, "grad_norm": 5.664742469787598, "learning_rate": 4.24258064516129e-05, "loss": 0.2246, "step": 2348 }, { "epoch": 15.142168674698794, "grad_norm": 5.506618976593018, "learning_rate": 4.2422580645161294e-05, "loss": 0.3106, "step": 2349 }, { "epoch": 15.14859437751004, "grad_norm": 8.194740295410156, "learning_rate": 4.241935483870968e-05, "loss": 0.4411, "step": 2350 }, { "epoch": 15.155020080321286, "grad_norm": 5.2654242515563965, "learning_rate": 4.2416129032258065e-05, "loss": 0.2925, "step": 2351 }, { "epoch": 15.16144578313253, "grad_norm": 13.671700477600098, "learning_rate": 4.241290322580646e-05, "loss": 0.4075, "step": 2352 }, { "epoch": 15.167871485943776, "grad_norm": 5.349864959716797, "learning_rate": 4.2409677419354844e-05, "loss": 0.2209, "step": 2353 }, { "epoch": 15.17429718875502, "grad_norm": 13.429641723632812, "learning_rate": 4.240645161290322e-05, "loss": 0.7626, "step": 2354 }, { "epoch": 15.180722891566266, "grad_norm": 4.637446403503418, "learning_rate": 4.2403225806451615e-05, "loss": 0.3384, "step": 2355 }, { "epoch": 15.18714859437751, "grad_norm": 6.7236857414245605, "learning_rate": 4.24e-05, "loss": 0.5406, "step": 2356 }, { "epoch": 15.193574297188755, "grad_norm": 12.331396102905273, "learning_rate": 4.2396774193548386e-05, "loss": 0.3182, "step": 2357 }, { "epoch": 15.2, "grad_norm": 8.327754020690918, "learning_rate": 4.239354838709678e-05, "loss": 0.2931, "step": 2358 }, { "epoch": 15.206425702811245, "grad_norm": 13.528026580810547, "learning_rate": 4.2390322580645165e-05, "loss": 0.2844, "step": 2359 }, { "epoch": 15.21285140562249, "grad_norm": 4.614267826080322, "learning_rate": 4.238709677419355e-05, "loss": 0.2331, "step": 2360 }, { "epoch": 15.219277108433735, "grad_norm": 4.037598609924316, "learning_rate": 4.238387096774194e-05, "loss": 0.2468, "step": 2361 }, { "epoch": 15.22570281124498, "grad_norm": 16.225753784179688, "learning_rate": 4.238064516129032e-05, "loss": 0.5224, "step": 2362 }, { "epoch": 15.232128514056225, "grad_norm": 7.665651321411133, "learning_rate": 4.237741935483871e-05, "loss": 0.3444, "step": 2363 }, { "epoch": 15.23855421686747, "grad_norm": 7.163440704345703, "learning_rate": 4.23741935483871e-05, "loss": 0.4842, "step": 2364 }, { "epoch": 15.244979919678714, "grad_norm": 21.759172439575195, "learning_rate": 4.2370967741935486e-05, "loss": 0.6801, "step": 2365 }, { "epoch": 15.25140562248996, "grad_norm": 6.257438659667969, "learning_rate": 4.236774193548387e-05, "loss": 0.3818, "step": 2366 }, { "epoch": 15.257831325301204, "grad_norm": 8.949494361877441, "learning_rate": 4.2364516129032264e-05, "loss": 0.4398, "step": 2367 }, { "epoch": 15.26425702811245, "grad_norm": 7.535526752471924, "learning_rate": 4.236129032258065e-05, "loss": 0.6624, "step": 2368 }, { "epoch": 15.270682730923696, "grad_norm": 8.554759979248047, "learning_rate": 4.2358064516129035e-05, "loss": 0.3715, "step": 2369 }, { "epoch": 15.27710843373494, "grad_norm": 10.126842498779297, "learning_rate": 4.235483870967742e-05, "loss": 0.2388, "step": 2370 }, { "epoch": 15.283534136546185, "grad_norm": 7.700638294219971, "learning_rate": 4.2351612903225807e-05, "loss": 0.5747, "step": 2371 }, { "epoch": 15.28995983935743, "grad_norm": 8.150126457214355, "learning_rate": 4.234838709677419e-05, "loss": 0.6105, "step": 2372 }, { "epoch": 15.296385542168675, "grad_norm": 5.3803911209106445, "learning_rate": 4.2345161290322585e-05, "loss": 0.3442, "step": 2373 }, { "epoch": 15.30281124497992, "grad_norm": 15.103699684143066, "learning_rate": 4.234193548387097e-05, "loss": 0.6876, "step": 2374 }, { "epoch": 15.309236947791165, "grad_norm": 5.1493024826049805, "learning_rate": 4.2338709677419356e-05, "loss": 0.2261, "step": 2375 }, { "epoch": 15.315662650602409, "grad_norm": 6.218255996704102, "learning_rate": 4.233548387096775e-05, "loss": 0.5332, "step": 2376 }, { "epoch": 15.322088353413655, "grad_norm": 6.921578884124756, "learning_rate": 4.2332258064516134e-05, "loss": 0.4208, "step": 2377 }, { "epoch": 15.3285140562249, "grad_norm": 7.697537422180176, "learning_rate": 4.232903225806451e-05, "loss": 0.4342, "step": 2378 }, { "epoch": 15.334939759036144, "grad_norm": 11.077205657958984, "learning_rate": 4.2325806451612906e-05, "loss": 0.6999, "step": 2379 }, { "epoch": 15.34136546184739, "grad_norm": 6.246143341064453, "learning_rate": 4.232258064516129e-05, "loss": 0.3805, "step": 2380 }, { "epoch": 15.347791164658634, "grad_norm": 11.630885124206543, "learning_rate": 4.231935483870968e-05, "loss": 0.3247, "step": 2381 }, { "epoch": 15.35421686746988, "grad_norm": 3.998335123062134, "learning_rate": 4.231612903225807e-05, "loss": 0.1482, "step": 2382 }, { "epoch": 15.360642570281124, "grad_norm": 11.962170600891113, "learning_rate": 4.2312903225806455e-05, "loss": 0.4717, "step": 2383 }, { "epoch": 15.36706827309237, "grad_norm": 9.89771842956543, "learning_rate": 4.230967741935484e-05, "loss": 0.5396, "step": 2384 }, { "epoch": 15.373493975903614, "grad_norm": 10.393823623657227, "learning_rate": 4.230645161290323e-05, "loss": 0.2836, "step": 2385 }, { "epoch": 15.37991967871486, "grad_norm": 11.674910545349121, "learning_rate": 4.230322580645161e-05, "loss": 0.5744, "step": 2386 }, { "epoch": 15.386345381526105, "grad_norm": 6.841465473175049, "learning_rate": 4.23e-05, "loss": 0.148, "step": 2387 }, { "epoch": 15.39277108433735, "grad_norm": 22.442556381225586, "learning_rate": 4.229677419354839e-05, "loss": 0.4327, "step": 2388 }, { "epoch": 15.399196787148595, "grad_norm": 5.058197021484375, "learning_rate": 4.2293548387096776e-05, "loss": 0.1871, "step": 2389 }, { "epoch": 15.405622489959839, "grad_norm": 11.201704025268555, "learning_rate": 4.229032258064516e-05, "loss": 0.416, "step": 2390 }, { "epoch": 15.412048192771085, "grad_norm": 23.17746925354004, "learning_rate": 4.2287096774193555e-05, "loss": 0.5232, "step": 2391 }, { "epoch": 15.418473895582329, "grad_norm": 7.128472328186035, "learning_rate": 4.228387096774194e-05, "loss": 0.3064, "step": 2392 }, { "epoch": 15.424899598393575, "grad_norm": 6.4573163986206055, "learning_rate": 4.228064516129032e-05, "loss": 0.3219, "step": 2393 }, { "epoch": 15.431325301204819, "grad_norm": 10.64971923828125, "learning_rate": 4.227741935483871e-05, "loss": 0.5138, "step": 2394 }, { "epoch": 15.437751004016064, "grad_norm": 10.380655288696289, "learning_rate": 4.22741935483871e-05, "loss": 0.7361, "step": 2395 }, { "epoch": 15.44417670682731, "grad_norm": 8.105147361755371, "learning_rate": 4.227096774193548e-05, "loss": 0.2916, "step": 2396 }, { "epoch": 15.450602409638554, "grad_norm": 18.957441329956055, "learning_rate": 4.2267741935483876e-05, "loss": 0.3936, "step": 2397 }, { "epoch": 15.4570281124498, "grad_norm": 24.41665267944336, "learning_rate": 4.226451612903226e-05, "loss": 0.6303, "step": 2398 }, { "epoch": 15.463453815261044, "grad_norm": 10.326897621154785, "learning_rate": 4.226129032258065e-05, "loss": 0.5535, "step": 2399 }, { "epoch": 15.46987951807229, "grad_norm": 9.082820892333984, "learning_rate": 4.225806451612904e-05, "loss": 0.5376, "step": 2400 }, { "epoch": 15.476305220883534, "grad_norm": 8.6959228515625, "learning_rate": 4.225483870967742e-05, "loss": 0.4995, "step": 2401 }, { "epoch": 15.48273092369478, "grad_norm": 11.085436820983887, "learning_rate": 4.2251612903225804e-05, "loss": 0.3361, "step": 2402 }, { "epoch": 15.489156626506023, "grad_norm": 8.150581359863281, "learning_rate": 4.2248387096774197e-05, "loss": 0.5386, "step": 2403 }, { "epoch": 15.49558232931727, "grad_norm": 8.621581077575684, "learning_rate": 4.224516129032258e-05, "loss": 0.2418, "step": 2404 }, { "epoch": 15.502008032128515, "grad_norm": 3.729879140853882, "learning_rate": 4.224193548387097e-05, "loss": 0.2075, "step": 2405 }, { "epoch": 15.508433734939759, "grad_norm": 6.421565532684326, "learning_rate": 4.223870967741936e-05, "loss": 0.3378, "step": 2406 }, { "epoch": 15.514859437751005, "grad_norm": 16.344449996948242, "learning_rate": 4.2235483870967746e-05, "loss": 0.5771, "step": 2407 }, { "epoch": 15.521285140562249, "grad_norm": 9.539776802062988, "learning_rate": 4.223225806451613e-05, "loss": 0.3051, "step": 2408 }, { "epoch": 15.527710843373494, "grad_norm": 5.387951850891113, "learning_rate": 4.222903225806452e-05, "loss": 0.3396, "step": 2409 }, { "epoch": 15.534136546184738, "grad_norm": 7.315929889678955, "learning_rate": 4.22258064516129e-05, "loss": 0.3505, "step": 2410 }, { "epoch": 15.540562248995984, "grad_norm": 9.300087928771973, "learning_rate": 4.222258064516129e-05, "loss": 0.3803, "step": 2411 }, { "epoch": 15.546987951807228, "grad_norm": 6.548305988311768, "learning_rate": 4.221935483870968e-05, "loss": 0.5799, "step": 2412 }, { "epoch": 15.553413654618474, "grad_norm": 3.2415425777435303, "learning_rate": 4.221612903225807e-05, "loss": 0.2229, "step": 2413 }, { "epoch": 15.55983935742972, "grad_norm": 8.635759353637695, "learning_rate": 4.221290322580645e-05, "loss": 0.4636, "step": 2414 }, { "epoch": 15.566265060240964, "grad_norm": 19.10055160522461, "learning_rate": 4.2209677419354845e-05, "loss": 0.3917, "step": 2415 }, { "epoch": 15.57269076305221, "grad_norm": 6.912172794342041, "learning_rate": 4.220645161290323e-05, "loss": 0.1577, "step": 2416 }, { "epoch": 15.579116465863454, "grad_norm": 15.726428031921387, "learning_rate": 4.220322580645161e-05, "loss": 0.3027, "step": 2417 }, { "epoch": 15.5855421686747, "grad_norm": 9.092350959777832, "learning_rate": 4.22e-05, "loss": 0.2341, "step": 2418 }, { "epoch": 15.591967871485943, "grad_norm": 16.23231315612793, "learning_rate": 4.219677419354839e-05, "loss": 0.5635, "step": 2419 }, { "epoch": 15.598393574297189, "grad_norm": 28.883005142211914, "learning_rate": 4.2193548387096774e-05, "loss": 0.3983, "step": 2420 }, { "epoch": 15.604819277108433, "grad_norm": 11.856239318847656, "learning_rate": 4.2190322580645166e-05, "loss": 0.214, "step": 2421 }, { "epoch": 15.611244979919679, "grad_norm": 4.76497220993042, "learning_rate": 4.218709677419355e-05, "loss": 0.1793, "step": 2422 }, { "epoch": 15.617670682730925, "grad_norm": 7.177992820739746, "learning_rate": 4.218387096774194e-05, "loss": 0.5813, "step": 2423 }, { "epoch": 15.624096385542169, "grad_norm": 7.799806594848633, "learning_rate": 4.2180645161290324e-05, "loss": 0.3958, "step": 2424 }, { "epoch": 15.630522088353414, "grad_norm": 4.211315155029297, "learning_rate": 4.217741935483871e-05, "loss": 0.5312, "step": 2425 }, { "epoch": 15.636947791164658, "grad_norm": 22.47519302368164, "learning_rate": 4.2174193548387095e-05, "loss": 0.7257, "step": 2426 }, { "epoch": 15.643373493975904, "grad_norm": 7.41736364364624, "learning_rate": 4.217096774193549e-05, "loss": 0.4272, "step": 2427 }, { "epoch": 15.649799196787148, "grad_norm": 8.000711441040039, "learning_rate": 4.216774193548387e-05, "loss": 0.2626, "step": 2428 }, { "epoch": 15.656224899598394, "grad_norm": 8.802648544311523, "learning_rate": 4.216451612903226e-05, "loss": 0.1952, "step": 2429 }, { "epoch": 15.662650602409638, "grad_norm": 6.999892234802246, "learning_rate": 4.216129032258065e-05, "loss": 0.3721, "step": 2430 }, { "epoch": 15.669076305220884, "grad_norm": 5.79235315322876, "learning_rate": 4.215806451612904e-05, "loss": 0.4001, "step": 2431 }, { "epoch": 15.67550200803213, "grad_norm": 14.65339469909668, "learning_rate": 4.2154838709677416e-05, "loss": 0.4598, "step": 2432 }, { "epoch": 15.681927710843373, "grad_norm": 9.302274703979492, "learning_rate": 4.215161290322581e-05, "loss": 0.3374, "step": 2433 }, { "epoch": 15.68835341365462, "grad_norm": 14.027687072753906, "learning_rate": 4.2148387096774194e-05, "loss": 0.5235, "step": 2434 }, { "epoch": 15.694779116465863, "grad_norm": 11.379298210144043, "learning_rate": 4.214516129032258e-05, "loss": 0.5125, "step": 2435 }, { "epoch": 15.701204819277109, "grad_norm": 8.660789489746094, "learning_rate": 4.214193548387097e-05, "loss": 0.2986, "step": 2436 }, { "epoch": 15.707630522088353, "grad_norm": 14.046642303466797, "learning_rate": 4.213870967741936e-05, "loss": 0.6541, "step": 2437 }, { "epoch": 15.714056224899599, "grad_norm": 18.61618995666504, "learning_rate": 4.2135483870967744e-05, "loss": 0.2582, "step": 2438 }, { "epoch": 15.720481927710843, "grad_norm": 8.111678123474121, "learning_rate": 4.2132258064516136e-05, "loss": 0.2551, "step": 2439 }, { "epoch": 15.726907630522089, "grad_norm": 7.861298084259033, "learning_rate": 4.2129032258064515e-05, "loss": 0.51, "step": 2440 }, { "epoch": 15.733333333333333, "grad_norm": 23.703166961669922, "learning_rate": 4.21258064516129e-05, "loss": 0.3835, "step": 2441 }, { "epoch": 15.739759036144578, "grad_norm": 8.618806838989258, "learning_rate": 4.212258064516129e-05, "loss": 0.3612, "step": 2442 }, { "epoch": 15.746184738955824, "grad_norm": 15.769638061523438, "learning_rate": 4.211935483870968e-05, "loss": 0.5362, "step": 2443 }, { "epoch": 15.752610441767068, "grad_norm": 12.987544059753418, "learning_rate": 4.2116129032258065e-05, "loss": 0.3868, "step": 2444 }, { "epoch": 15.759036144578314, "grad_norm": 5.869516849517822, "learning_rate": 4.211290322580646e-05, "loss": 0.3049, "step": 2445 }, { "epoch": 15.765461847389558, "grad_norm": 6.138005256652832, "learning_rate": 4.210967741935484e-05, "loss": 0.358, "step": 2446 }, { "epoch": 15.771887550200804, "grad_norm": 22.154016494750977, "learning_rate": 4.210645161290323e-05, "loss": 0.5794, "step": 2447 }, { "epoch": 15.778313253012048, "grad_norm": 11.385793685913086, "learning_rate": 4.2103225806451614e-05, "loss": 0.3475, "step": 2448 }, { "epoch": 15.784738955823293, "grad_norm": 7.66956901550293, "learning_rate": 4.21e-05, "loss": 0.5226, "step": 2449 }, { "epoch": 15.791164658634537, "grad_norm": 7.792828559875488, "learning_rate": 4.2096774193548386e-05, "loss": 0.4788, "step": 2450 }, { "epoch": 15.797590361445783, "grad_norm": 9.780460357666016, "learning_rate": 4.209354838709678e-05, "loss": 0.2589, "step": 2451 }, { "epoch": 15.804016064257029, "grad_norm": 11.493890762329102, "learning_rate": 4.2090322580645164e-05, "loss": 0.7022, "step": 2452 }, { "epoch": 15.810441767068273, "grad_norm": 5.09968376159668, "learning_rate": 4.208709677419355e-05, "loss": 0.272, "step": 2453 }, { "epoch": 15.816867469879519, "grad_norm": 8.356493949890137, "learning_rate": 4.208387096774194e-05, "loss": 0.4225, "step": 2454 }, { "epoch": 15.823293172690763, "grad_norm": 11.386983871459961, "learning_rate": 4.208064516129032e-05, "loss": 0.4971, "step": 2455 }, { "epoch": 15.829718875502008, "grad_norm": 12.787942886352539, "learning_rate": 4.207741935483871e-05, "loss": 0.5527, "step": 2456 }, { "epoch": 15.836144578313252, "grad_norm": 21.36832046508789, "learning_rate": 4.20741935483871e-05, "loss": 0.1868, "step": 2457 }, { "epoch": 15.842570281124498, "grad_norm": 9.613000869750977, "learning_rate": 4.2070967741935485e-05, "loss": 0.6888, "step": 2458 }, { "epoch": 15.848995983935742, "grad_norm": 6.858872413635254, "learning_rate": 4.206774193548387e-05, "loss": 0.3091, "step": 2459 }, { "epoch": 15.855421686746988, "grad_norm": 11.21135425567627, "learning_rate": 4.206451612903226e-05, "loss": 0.6445, "step": 2460 }, { "epoch": 15.861847389558234, "grad_norm": 12.832311630249023, "learning_rate": 4.206129032258065e-05, "loss": 0.4134, "step": 2461 }, { "epoch": 15.868273092369478, "grad_norm": 4.364022731781006, "learning_rate": 4.2058064516129035e-05, "loss": 0.4136, "step": 2462 }, { "epoch": 15.874698795180723, "grad_norm": 5.41567325592041, "learning_rate": 4.205483870967742e-05, "loss": 0.3075, "step": 2463 }, { "epoch": 15.881124497991967, "grad_norm": 12.690702438354492, "learning_rate": 4.2051612903225806e-05, "loss": 0.2453, "step": 2464 }, { "epoch": 15.887550200803213, "grad_norm": 5.416585922241211, "learning_rate": 4.204838709677419e-05, "loss": 0.2304, "step": 2465 }, { "epoch": 15.893975903614457, "grad_norm": 33.945838928222656, "learning_rate": 4.2045161290322584e-05, "loss": 0.5071, "step": 2466 }, { "epoch": 15.900401606425703, "grad_norm": 5.813387393951416, "learning_rate": 4.204193548387097e-05, "loss": 0.3455, "step": 2467 }, { "epoch": 15.906827309236947, "grad_norm": 11.287200927734375, "learning_rate": 4.2038709677419356e-05, "loss": 0.4175, "step": 2468 }, { "epoch": 15.913253012048193, "grad_norm": 8.13749885559082, "learning_rate": 4.203548387096775e-05, "loss": 0.6886, "step": 2469 }, { "epoch": 15.919678714859439, "grad_norm": 15.819361686706543, "learning_rate": 4.2032258064516134e-05, "loss": 0.2478, "step": 2470 }, { "epoch": 15.926104417670683, "grad_norm": 7.4973320960998535, "learning_rate": 4.202903225806451e-05, "loss": 0.6646, "step": 2471 }, { "epoch": 15.932530120481928, "grad_norm": 4.193480014801025, "learning_rate": 4.2025806451612905e-05, "loss": 0.1969, "step": 2472 }, { "epoch": 15.938955823293172, "grad_norm": 5.9604973793029785, "learning_rate": 4.202258064516129e-05, "loss": 0.2073, "step": 2473 }, { "epoch": 15.945381526104418, "grad_norm": 9.048995971679688, "learning_rate": 4.201935483870968e-05, "loss": 0.5072, "step": 2474 }, { "epoch": 15.951807228915662, "grad_norm": 9.399669647216797, "learning_rate": 4.201612903225807e-05, "loss": 0.596, "step": 2475 }, { "epoch": 15.958232931726908, "grad_norm": 25.430660247802734, "learning_rate": 4.2012903225806455e-05, "loss": 0.4204, "step": 2476 }, { "epoch": 15.964658634538152, "grad_norm": 11.863367080688477, "learning_rate": 4.200967741935484e-05, "loss": 0.6038, "step": 2477 }, { "epoch": 15.971084337349398, "grad_norm": 11.709099769592285, "learning_rate": 4.200645161290323e-05, "loss": 0.3799, "step": 2478 }, { "epoch": 15.977510040160643, "grad_norm": 14.189543724060059, "learning_rate": 4.200322580645161e-05, "loss": 0.526, "step": 2479 }, { "epoch": 15.983935742971887, "grad_norm": 10.41120433807373, "learning_rate": 4.2e-05, "loss": 0.5951, "step": 2480 }, { "epoch": 15.990361445783133, "grad_norm": 17.792394638061523, "learning_rate": 4.199677419354839e-05, "loss": 0.5343, "step": 2481 }, { "epoch": 15.996787148594377, "grad_norm": 5.363988876342773, "learning_rate": 4.1993548387096776e-05, "loss": 0.3613, "step": 2482 }, { "epoch": 16.006425702811246, "grad_norm": 10.922323226928711, "learning_rate": 4.199032258064516e-05, "loss": 0.679, "step": 2483 }, { "epoch": 16.01285140562249, "grad_norm": 21.02955436706543, "learning_rate": 4.1987096774193554e-05, "loss": 0.605, "step": 2484 }, { "epoch": 16.019277108433734, "grad_norm": 6.587525367736816, "learning_rate": 4.198387096774194e-05, "loss": 0.5079, "step": 2485 }, { "epoch": 16.02570281124498, "grad_norm": 7.5422163009643555, "learning_rate": 4.1980645161290325e-05, "loss": 0.5695, "step": 2486 }, { "epoch": 16.032128514056225, "grad_norm": 5.503248691558838, "learning_rate": 4.197741935483871e-05, "loss": 0.2598, "step": 2487 }, { "epoch": 16.03855421686747, "grad_norm": 11.73767375946045, "learning_rate": 4.19741935483871e-05, "loss": 0.4055, "step": 2488 }, { "epoch": 16.044979919678713, "grad_norm": 3.7047104835510254, "learning_rate": 4.197096774193548e-05, "loss": 0.3998, "step": 2489 }, { "epoch": 16.05140562248996, "grad_norm": 9.086389541625977, "learning_rate": 4.1967741935483875e-05, "loss": 0.3888, "step": 2490 }, { "epoch": 16.057831325301205, "grad_norm": 6.980482578277588, "learning_rate": 4.196451612903226e-05, "loss": 0.3929, "step": 2491 }, { "epoch": 16.06425702811245, "grad_norm": 24.300260543823242, "learning_rate": 4.1961290322580646e-05, "loss": 0.5018, "step": 2492 }, { "epoch": 16.070682730923696, "grad_norm": 4.919691562652588, "learning_rate": 4.195806451612904e-05, "loss": 0.3961, "step": 2493 }, { "epoch": 16.07710843373494, "grad_norm": 16.63945960998535, "learning_rate": 4.195483870967742e-05, "loss": 0.2591, "step": 2494 }, { "epoch": 16.083534136546184, "grad_norm": 5.398129463195801, "learning_rate": 4.1951612903225804e-05, "loss": 0.175, "step": 2495 }, { "epoch": 16.08995983935743, "grad_norm": 10.15347957611084, "learning_rate": 4.1948387096774196e-05, "loss": 0.2895, "step": 2496 }, { "epoch": 16.096385542168676, "grad_norm": 10.801280975341797, "learning_rate": 4.194516129032258e-05, "loss": 0.526, "step": 2497 }, { "epoch": 16.102811244979918, "grad_norm": 18.209121704101562, "learning_rate": 4.194193548387097e-05, "loss": 0.507, "step": 2498 }, { "epoch": 16.109236947791164, "grad_norm": 3.79247784614563, "learning_rate": 4.193870967741936e-05, "loss": 0.1687, "step": 2499 }, { "epoch": 16.11566265060241, "grad_norm": 6.301097869873047, "learning_rate": 4.1935483870967746e-05, "loss": 0.1359, "step": 2500 }, { "epoch": 16.122088353413655, "grad_norm": 10.006635665893555, "learning_rate": 4.193225806451613e-05, "loss": 0.4024, "step": 2501 }, { "epoch": 16.1285140562249, "grad_norm": 15.957002639770508, "learning_rate": 4.192903225806452e-05, "loss": 0.5519, "step": 2502 }, { "epoch": 16.134939759036143, "grad_norm": 6.932527542114258, "learning_rate": 4.19258064516129e-05, "loss": 0.2835, "step": 2503 }, { "epoch": 16.14136546184739, "grad_norm": 6.624276161193848, "learning_rate": 4.192258064516129e-05, "loss": 0.3905, "step": 2504 }, { "epoch": 16.147791164658635, "grad_norm": 15.273351669311523, "learning_rate": 4.191935483870968e-05, "loss": 0.4939, "step": 2505 }, { "epoch": 16.15421686746988, "grad_norm": 6.8521223068237305, "learning_rate": 4.191612903225807e-05, "loss": 0.2682, "step": 2506 }, { "epoch": 16.160642570281123, "grad_norm": 10.261390686035156, "learning_rate": 4.191290322580645e-05, "loss": 0.3515, "step": 2507 }, { "epoch": 16.16706827309237, "grad_norm": 14.454829216003418, "learning_rate": 4.1909677419354845e-05, "loss": 0.4982, "step": 2508 }, { "epoch": 16.173493975903614, "grad_norm": 6.200360298156738, "learning_rate": 4.190645161290323e-05, "loss": 0.2002, "step": 2509 }, { "epoch": 16.17991967871486, "grad_norm": 20.304306030273438, "learning_rate": 4.190322580645161e-05, "loss": 0.4126, "step": 2510 }, { "epoch": 16.186345381526106, "grad_norm": 15.576193809509277, "learning_rate": 4.19e-05, "loss": 0.5978, "step": 2511 }, { "epoch": 16.19277108433735, "grad_norm": 19.06662940979004, "learning_rate": 4.189677419354839e-05, "loss": 0.4856, "step": 2512 }, { "epoch": 16.199196787148594, "grad_norm": 11.529547691345215, "learning_rate": 4.1893548387096773e-05, "loss": 0.6269, "step": 2513 }, { "epoch": 16.20562248995984, "grad_norm": 14.315959930419922, "learning_rate": 4.1890322580645166e-05, "loss": 0.4785, "step": 2514 }, { "epoch": 16.212048192771086, "grad_norm": 12.084338188171387, "learning_rate": 4.188709677419355e-05, "loss": 0.5358, "step": 2515 }, { "epoch": 16.218473895582328, "grad_norm": 12.462569236755371, "learning_rate": 4.188387096774194e-05, "loss": 0.4449, "step": 2516 }, { "epoch": 16.224899598393574, "grad_norm": 26.547805786132812, "learning_rate": 4.188064516129033e-05, "loss": 0.4203, "step": 2517 }, { "epoch": 16.23132530120482, "grad_norm": 25.03313636779785, "learning_rate": 4.187741935483871e-05, "loss": 0.8473, "step": 2518 }, { "epoch": 16.237751004016065, "grad_norm": 4.830761909484863, "learning_rate": 4.1874193548387094e-05, "loss": 0.2307, "step": 2519 }, { "epoch": 16.24417670682731, "grad_norm": 11.077620506286621, "learning_rate": 4.187096774193549e-05, "loss": 0.4774, "step": 2520 }, { "epoch": 16.250602409638553, "grad_norm": 8.113081932067871, "learning_rate": 4.186774193548387e-05, "loss": 0.3855, "step": 2521 }, { "epoch": 16.2570281124498, "grad_norm": 5.04317045211792, "learning_rate": 4.186451612903226e-05, "loss": 0.33, "step": 2522 }, { "epoch": 16.263453815261045, "grad_norm": 4.896555423736572, "learning_rate": 4.186129032258065e-05, "loss": 0.1965, "step": 2523 }, { "epoch": 16.26987951807229, "grad_norm": 7.186720371246338, "learning_rate": 4.1858064516129037e-05, "loss": 0.2854, "step": 2524 }, { "epoch": 16.276305220883533, "grad_norm": 6.59860897064209, "learning_rate": 4.1854838709677415e-05, "loss": 0.4275, "step": 2525 }, { "epoch": 16.28273092369478, "grad_norm": 13.542860984802246, "learning_rate": 4.185161290322581e-05, "loss": 0.69, "step": 2526 }, { "epoch": 16.289156626506024, "grad_norm": 8.679874420166016, "learning_rate": 4.1848387096774194e-05, "loss": 0.4224, "step": 2527 }, { "epoch": 16.29558232931727, "grad_norm": 12.366410255432129, "learning_rate": 4.184516129032258e-05, "loss": 0.4309, "step": 2528 }, { "epoch": 16.302008032128516, "grad_norm": 9.276639938354492, "learning_rate": 4.184193548387097e-05, "loss": 0.5469, "step": 2529 }, { "epoch": 16.308433734939758, "grad_norm": 18.13412857055664, "learning_rate": 4.183870967741936e-05, "loss": 0.8089, "step": 2530 }, { "epoch": 16.314859437751004, "grad_norm": 9.749701499938965, "learning_rate": 4.183548387096774e-05, "loss": 0.5667, "step": 2531 }, { "epoch": 16.32128514056225, "grad_norm": 9.753225326538086, "learning_rate": 4.1832258064516136e-05, "loss": 0.5993, "step": 2532 }, { "epoch": 16.327710843373495, "grad_norm": 21.965673446655273, "learning_rate": 4.1829032258064515e-05, "loss": 0.5742, "step": 2533 }, { "epoch": 16.334136546184737, "grad_norm": 14.266467094421387, "learning_rate": 4.18258064516129e-05, "loss": 0.5548, "step": 2534 }, { "epoch": 16.340562248995983, "grad_norm": 7.019999027252197, "learning_rate": 4.182258064516129e-05, "loss": 0.6112, "step": 2535 }, { "epoch": 16.34698795180723, "grad_norm": 4.195750713348389, "learning_rate": 4.181935483870968e-05, "loss": 0.3022, "step": 2536 }, { "epoch": 16.353413654618475, "grad_norm": 9.974512100219727, "learning_rate": 4.1816129032258064e-05, "loss": 0.2358, "step": 2537 }, { "epoch": 16.35983935742972, "grad_norm": 8.582579612731934, "learning_rate": 4.181290322580646e-05, "loss": 0.3298, "step": 2538 }, { "epoch": 16.366265060240963, "grad_norm": 15.433536529541016, "learning_rate": 4.180967741935484e-05, "loss": 0.4654, "step": 2539 }, { "epoch": 16.37269076305221, "grad_norm": 21.441438674926758, "learning_rate": 4.180645161290323e-05, "loss": 0.5066, "step": 2540 }, { "epoch": 16.379116465863454, "grad_norm": 7.705052852630615, "learning_rate": 4.1803225806451614e-05, "loss": 0.2897, "step": 2541 }, { "epoch": 16.3855421686747, "grad_norm": 11.694377899169922, "learning_rate": 4.18e-05, "loss": 0.6281, "step": 2542 }, { "epoch": 16.391967871485942, "grad_norm": 5.699876308441162, "learning_rate": 4.1796774193548385e-05, "loss": 0.5513, "step": 2543 }, { "epoch": 16.398393574297188, "grad_norm": 11.819549560546875, "learning_rate": 4.179354838709678e-05, "loss": 0.401, "step": 2544 }, { "epoch": 16.404819277108434, "grad_norm": 3.481398344039917, "learning_rate": 4.1790322580645163e-05, "loss": 0.2264, "step": 2545 }, { "epoch": 16.41124497991968, "grad_norm": 8.998744010925293, "learning_rate": 4.178709677419355e-05, "loss": 0.2912, "step": 2546 }, { "epoch": 16.417670682730925, "grad_norm": 19.367372512817383, "learning_rate": 4.178387096774194e-05, "loss": 0.4756, "step": 2547 }, { "epoch": 16.424096385542168, "grad_norm": 7.797924518585205, "learning_rate": 4.178064516129033e-05, "loss": 0.3226, "step": 2548 }, { "epoch": 16.430522088353413, "grad_norm": 8.291269302368164, "learning_rate": 4.1777419354838706e-05, "loss": 0.319, "step": 2549 }, { "epoch": 16.43694779116466, "grad_norm": 4.64478063583374, "learning_rate": 4.17741935483871e-05, "loss": 0.2174, "step": 2550 }, { "epoch": 16.443373493975905, "grad_norm": 3.2188265323638916, "learning_rate": 4.1770967741935485e-05, "loss": 0.1655, "step": 2551 }, { "epoch": 16.449799196787147, "grad_norm": 22.46759796142578, "learning_rate": 4.176774193548387e-05, "loss": 0.4873, "step": 2552 }, { "epoch": 16.456224899598393, "grad_norm": 2.524566411972046, "learning_rate": 4.176451612903226e-05, "loss": 0.1012, "step": 2553 }, { "epoch": 16.46265060240964, "grad_norm": 6.30817985534668, "learning_rate": 4.176129032258065e-05, "loss": 0.5207, "step": 2554 }, { "epoch": 16.469076305220884, "grad_norm": 10.875990867614746, "learning_rate": 4.1758064516129034e-05, "loss": 0.2495, "step": 2555 }, { "epoch": 16.47550200803213, "grad_norm": 8.351128578186035, "learning_rate": 4.1754838709677427e-05, "loss": 0.4951, "step": 2556 }, { "epoch": 16.481927710843372, "grad_norm": 3.9690868854522705, "learning_rate": 4.1751612903225806e-05, "loss": 0.2185, "step": 2557 }, { "epoch": 16.488353413654618, "grad_norm": 9.042223930358887, "learning_rate": 4.174838709677419e-05, "loss": 0.532, "step": 2558 }, { "epoch": 16.494779116465864, "grad_norm": 12.564483642578125, "learning_rate": 4.1745161290322584e-05, "loss": 0.5589, "step": 2559 }, { "epoch": 16.50120481927711, "grad_norm": 6.566292762756348, "learning_rate": 4.174193548387097e-05, "loss": 0.355, "step": 2560 }, { "epoch": 16.507630522088352, "grad_norm": 14.049600601196289, "learning_rate": 4.1738709677419355e-05, "loss": 0.6803, "step": 2561 }, { "epoch": 16.514056224899598, "grad_norm": 14.912120819091797, "learning_rate": 4.173548387096775e-05, "loss": 0.3754, "step": 2562 }, { "epoch": 16.520481927710843, "grad_norm": 12.84584903717041, "learning_rate": 4.173225806451613e-05, "loss": 0.4272, "step": 2563 }, { "epoch": 16.52690763052209, "grad_norm": 13.2744722366333, "learning_rate": 4.172903225806451e-05, "loss": 0.6541, "step": 2564 }, { "epoch": 16.533333333333335, "grad_norm": 26.453330993652344, "learning_rate": 4.1725806451612905e-05, "loss": 0.6566, "step": 2565 }, { "epoch": 16.539759036144577, "grad_norm": 11.408154487609863, "learning_rate": 4.172258064516129e-05, "loss": 0.4979, "step": 2566 }, { "epoch": 16.546184738955823, "grad_norm": 6.628384590148926, "learning_rate": 4.1719354838709676e-05, "loss": 0.417, "step": 2567 }, { "epoch": 16.55261044176707, "grad_norm": 11.205554962158203, "learning_rate": 4.171612903225807e-05, "loss": 0.4747, "step": 2568 }, { "epoch": 16.559036144578315, "grad_norm": 9.518960952758789, "learning_rate": 4.1712903225806454e-05, "loss": 0.3321, "step": 2569 }, { "epoch": 16.565461847389557, "grad_norm": 7.09514856338501, "learning_rate": 4.170967741935484e-05, "loss": 0.6438, "step": 2570 }, { "epoch": 16.571887550200803, "grad_norm": 8.224778175354004, "learning_rate": 4.170645161290323e-05, "loss": 0.3387, "step": 2571 }, { "epoch": 16.57831325301205, "grad_norm": 12.651840209960938, "learning_rate": 4.170322580645161e-05, "loss": 0.3458, "step": 2572 }, { "epoch": 16.584738955823294, "grad_norm": 15.52864933013916, "learning_rate": 4.17e-05, "loss": 0.4849, "step": 2573 }, { "epoch": 16.59116465863454, "grad_norm": 12.565882682800293, "learning_rate": 4.169677419354839e-05, "loss": 0.7831, "step": 2574 }, { "epoch": 16.597590361445782, "grad_norm": 9.6824951171875, "learning_rate": 4.1693548387096775e-05, "loss": 0.286, "step": 2575 }, { "epoch": 16.604016064257028, "grad_norm": 2.8116066455841064, "learning_rate": 4.169032258064516e-05, "loss": 0.3186, "step": 2576 }, { "epoch": 16.610441767068274, "grad_norm": 9.079751014709473, "learning_rate": 4.1687096774193554e-05, "loss": 0.283, "step": 2577 }, { "epoch": 16.61686746987952, "grad_norm": 19.743940353393555, "learning_rate": 4.168387096774194e-05, "loss": 0.3926, "step": 2578 }, { "epoch": 16.62329317269076, "grad_norm": 9.291316986083984, "learning_rate": 4.1680645161290325e-05, "loss": 0.2492, "step": 2579 }, { "epoch": 16.629718875502007, "grad_norm": 18.53618812561035, "learning_rate": 4.167741935483871e-05, "loss": 0.5305, "step": 2580 }, { "epoch": 16.636144578313253, "grad_norm": 13.459234237670898, "learning_rate": 4.1674193548387096e-05, "loss": 0.4314, "step": 2581 }, { "epoch": 16.6425702811245, "grad_norm": 12.954390525817871, "learning_rate": 4.167096774193548e-05, "loss": 0.4882, "step": 2582 }, { "epoch": 16.648995983935745, "grad_norm": 7.83958101272583, "learning_rate": 4.1667741935483875e-05, "loss": 0.2753, "step": 2583 }, { "epoch": 16.655421686746987, "grad_norm": 10.061188697814941, "learning_rate": 4.166451612903226e-05, "loss": 0.3803, "step": 2584 }, { "epoch": 16.661847389558233, "grad_norm": 17.928831100463867, "learning_rate": 4.1661290322580646e-05, "loss": 0.244, "step": 2585 }, { "epoch": 16.66827309236948, "grad_norm": 10.885594367980957, "learning_rate": 4.165806451612904e-05, "loss": 0.3208, "step": 2586 }, { "epoch": 16.674698795180724, "grad_norm": 13.820005416870117, "learning_rate": 4.1654838709677424e-05, "loss": 0.3593, "step": 2587 }, { "epoch": 16.681124497991966, "grad_norm": 6.300478458404541, "learning_rate": 4.16516129032258e-05, "loss": 0.1258, "step": 2588 }, { "epoch": 16.687550200803212, "grad_norm": 14.85417366027832, "learning_rate": 4.1648387096774196e-05, "loss": 0.5567, "step": 2589 }, { "epoch": 16.693975903614458, "grad_norm": 20.13079833984375, "learning_rate": 4.164516129032258e-05, "loss": 0.2938, "step": 2590 }, { "epoch": 16.700401606425704, "grad_norm": 11.423604965209961, "learning_rate": 4.164193548387097e-05, "loss": 0.5788, "step": 2591 }, { "epoch": 16.70682730923695, "grad_norm": 4.438379764556885, "learning_rate": 4.163870967741936e-05, "loss": 0.1983, "step": 2592 }, { "epoch": 16.71325301204819, "grad_norm": 7.144856929779053, "learning_rate": 4.1635483870967745e-05, "loss": 0.5733, "step": 2593 }, { "epoch": 16.719678714859437, "grad_norm": 8.555366516113281, "learning_rate": 4.163225806451613e-05, "loss": 0.223, "step": 2594 }, { "epoch": 16.726104417670683, "grad_norm": 18.245319366455078, "learning_rate": 4.162903225806452e-05, "loss": 0.5174, "step": 2595 }, { "epoch": 16.73253012048193, "grad_norm": 8.045618057250977, "learning_rate": 4.16258064516129e-05, "loss": 0.3586, "step": 2596 }, { "epoch": 16.73895582329317, "grad_norm": 8.142438888549805, "learning_rate": 4.162258064516129e-05, "loss": 0.275, "step": 2597 }, { "epoch": 16.745381526104417, "grad_norm": 28.078388214111328, "learning_rate": 4.161935483870968e-05, "loss": 0.45, "step": 2598 }, { "epoch": 16.751807228915663, "grad_norm": 7.498473644256592, "learning_rate": 4.1616129032258066e-05, "loss": 0.2548, "step": 2599 }, { "epoch": 16.75823293172691, "grad_norm": 15.171133995056152, "learning_rate": 4.161290322580645e-05, "loss": 0.649, "step": 2600 }, { "epoch": 16.764658634538154, "grad_norm": 6.21567440032959, "learning_rate": 4.1609677419354844e-05, "loss": 0.3502, "step": 2601 }, { "epoch": 16.771084337349397, "grad_norm": 6.7450385093688965, "learning_rate": 4.160645161290323e-05, "loss": 0.3692, "step": 2602 }, { "epoch": 16.777510040160642, "grad_norm": 5.035745620727539, "learning_rate": 4.160322580645161e-05, "loss": 0.2321, "step": 2603 }, { "epoch": 16.783935742971888, "grad_norm": 7.907656669616699, "learning_rate": 4.16e-05, "loss": 0.4752, "step": 2604 }, { "epoch": 16.790361445783134, "grad_norm": 10.20055103302002, "learning_rate": 4.159677419354839e-05, "loss": 0.4582, "step": 2605 }, { "epoch": 16.796787148594376, "grad_norm": 11.16786003112793, "learning_rate": 4.159354838709677e-05, "loss": 0.3619, "step": 2606 }, { "epoch": 16.803212851405622, "grad_norm": 11.28504467010498, "learning_rate": 4.1590322580645165e-05, "loss": 0.5141, "step": 2607 }, { "epoch": 16.809638554216868, "grad_norm": 9.221504211425781, "learning_rate": 4.158709677419355e-05, "loss": 0.2737, "step": 2608 }, { "epoch": 16.816064257028113, "grad_norm": 25.59369468688965, "learning_rate": 4.158387096774194e-05, "loss": 0.7239, "step": 2609 }, { "epoch": 16.82248995983936, "grad_norm": 5.739678382873535, "learning_rate": 4.158064516129033e-05, "loss": 0.3557, "step": 2610 }, { "epoch": 16.8289156626506, "grad_norm": 11.031641960144043, "learning_rate": 4.157741935483871e-05, "loss": 0.5177, "step": 2611 }, { "epoch": 16.835341365461847, "grad_norm": 8.364300727844238, "learning_rate": 4.1574193548387094e-05, "loss": 0.3223, "step": 2612 }, { "epoch": 16.841767068273093, "grad_norm": 16.48232650756836, "learning_rate": 4.1570967741935486e-05, "loss": 0.2572, "step": 2613 }, { "epoch": 16.84819277108434, "grad_norm": 4.395567417144775, "learning_rate": 4.156774193548387e-05, "loss": 0.1535, "step": 2614 }, { "epoch": 16.85461847389558, "grad_norm": 15.014358520507812, "learning_rate": 4.156451612903226e-05, "loss": 0.555, "step": 2615 }, { "epoch": 16.861044176706827, "grad_norm": 9.324984550476074, "learning_rate": 4.156129032258065e-05, "loss": 0.5242, "step": 2616 }, { "epoch": 16.867469879518072, "grad_norm": 13.013428688049316, "learning_rate": 4.1558064516129036e-05, "loss": 0.288, "step": 2617 }, { "epoch": 16.873895582329318, "grad_norm": 7.020426273345947, "learning_rate": 4.155483870967742e-05, "loss": 0.3903, "step": 2618 }, { "epoch": 16.880321285140564, "grad_norm": 4.8528571128845215, "learning_rate": 4.155161290322581e-05, "loss": 0.3706, "step": 2619 }, { "epoch": 16.886746987951806, "grad_norm": 9.064408302307129, "learning_rate": 4.154838709677419e-05, "loss": 0.4566, "step": 2620 }, { "epoch": 16.893172690763052, "grad_norm": 11.16069507598877, "learning_rate": 4.154516129032258e-05, "loss": 0.3107, "step": 2621 }, { "epoch": 16.899598393574298, "grad_norm": 11.25145435333252, "learning_rate": 4.154193548387097e-05, "loss": 0.6881, "step": 2622 }, { "epoch": 16.906024096385543, "grad_norm": 31.709436416625977, "learning_rate": 4.153870967741936e-05, "loss": 0.5638, "step": 2623 }, { "epoch": 16.912449799196786, "grad_norm": 23.372833251953125, "learning_rate": 4.153548387096774e-05, "loss": 0.6911, "step": 2624 }, { "epoch": 16.91887550200803, "grad_norm": 6.5749945640563965, "learning_rate": 4.1532258064516135e-05, "loss": 0.4634, "step": 2625 }, { "epoch": 16.925301204819277, "grad_norm": 6.785894870758057, "learning_rate": 4.152903225806452e-05, "loss": 0.2932, "step": 2626 }, { "epoch": 16.931726907630523, "grad_norm": 13.939692497253418, "learning_rate": 4.15258064516129e-05, "loss": 0.7956, "step": 2627 }, { "epoch": 16.93815261044177, "grad_norm": 23.49359130859375, "learning_rate": 4.152258064516129e-05, "loss": 0.362, "step": 2628 }, { "epoch": 16.94457831325301, "grad_norm": 16.195402145385742, "learning_rate": 4.151935483870968e-05, "loss": 0.3527, "step": 2629 }, { "epoch": 16.951004016064257, "grad_norm": 5.385386943817139, "learning_rate": 4.1516129032258064e-05, "loss": 0.2997, "step": 2630 }, { "epoch": 16.957429718875503, "grad_norm": 9.277475357055664, "learning_rate": 4.1512903225806456e-05, "loss": 0.383, "step": 2631 }, { "epoch": 16.96385542168675, "grad_norm": 9.078250885009766, "learning_rate": 4.150967741935484e-05, "loss": 0.389, "step": 2632 }, { "epoch": 16.97028112449799, "grad_norm": 42.94200897216797, "learning_rate": 4.150645161290323e-05, "loss": 0.3608, "step": 2633 }, { "epoch": 16.976706827309236, "grad_norm": 11.873924255371094, "learning_rate": 4.150322580645161e-05, "loss": 0.2707, "step": 2634 }, { "epoch": 16.983132530120482, "grad_norm": 19.808740615844727, "learning_rate": 4.15e-05, "loss": 0.4301, "step": 2635 }, { "epoch": 16.989558232931728, "grad_norm": 33.16215515136719, "learning_rate": 4.1496774193548385e-05, "loss": 0.2562, "step": 2636 }, { "epoch": 16.99598393574297, "grad_norm": 17.434877395629883, "learning_rate": 4.149354838709678e-05, "loss": 0.3642, "step": 2637 }, { "epoch": 17.00562248995984, "grad_norm": 9.532145500183105, "learning_rate": 4.149032258064516e-05, "loss": 0.2003, "step": 2638 }, { "epoch": 17.012048192771083, "grad_norm": 14.264655113220215, "learning_rate": 4.148709677419355e-05, "loss": 0.6212, "step": 2639 }, { "epoch": 17.01847389558233, "grad_norm": 67.9417953491211, "learning_rate": 4.148387096774194e-05, "loss": 0.5743, "step": 2640 }, { "epoch": 17.024899598393574, "grad_norm": 11.336649894714355, "learning_rate": 4.148064516129033e-05, "loss": 0.326, "step": 2641 }, { "epoch": 17.03132530120482, "grad_norm": 4.143169403076172, "learning_rate": 4.1477419354838706e-05, "loss": 0.5878, "step": 2642 }, { "epoch": 17.037751004016066, "grad_norm": 5.358041763305664, "learning_rate": 4.14741935483871e-05, "loss": 0.408, "step": 2643 }, { "epoch": 17.044176706827308, "grad_norm": 20.754802703857422, "learning_rate": 4.1470967741935484e-05, "loss": 0.4138, "step": 2644 }, { "epoch": 17.050602409638554, "grad_norm": 28.02033233642578, "learning_rate": 4.146774193548387e-05, "loss": 0.5139, "step": 2645 }, { "epoch": 17.0570281124498, "grad_norm": 3.678636074066162, "learning_rate": 4.146451612903226e-05, "loss": 0.3605, "step": 2646 }, { "epoch": 17.063453815261045, "grad_norm": 1.5670864582061768, "learning_rate": 4.146129032258065e-05, "loss": 0.3375, "step": 2647 }, { "epoch": 17.069879518072288, "grad_norm": 20.052352905273438, "learning_rate": 4.1458064516129034e-05, "loss": 0.5684, "step": 2648 }, { "epoch": 17.076305220883533, "grad_norm": 4.805728912353516, "learning_rate": 4.1454838709677426e-05, "loss": 0.4487, "step": 2649 }, { "epoch": 17.08273092369478, "grad_norm": 15.258405685424805, "learning_rate": 4.1451612903225805e-05, "loss": 0.2857, "step": 2650 }, { "epoch": 17.089156626506025, "grad_norm": 19.36154556274414, "learning_rate": 4.144838709677419e-05, "loss": 0.5192, "step": 2651 }, { "epoch": 17.09558232931727, "grad_norm": 12.96336555480957, "learning_rate": 4.144516129032258e-05, "loss": 0.3431, "step": 2652 }, { "epoch": 17.102008032128513, "grad_norm": 1.3033356666564941, "learning_rate": 4.144193548387097e-05, "loss": 0.5024, "step": 2653 }, { "epoch": 17.10843373493976, "grad_norm": 28.440523147583008, "learning_rate": 4.1438709677419355e-05, "loss": 0.357, "step": 2654 }, { "epoch": 17.114859437751004, "grad_norm": 16.34256935119629, "learning_rate": 4.143548387096775e-05, "loss": 0.4884, "step": 2655 }, { "epoch": 17.12128514056225, "grad_norm": 7.546705722808838, "learning_rate": 4.143225806451613e-05, "loss": 0.4341, "step": 2656 }, { "epoch": 17.127710843373492, "grad_norm": 11.876081466674805, "learning_rate": 4.142903225806452e-05, "loss": 0.4792, "step": 2657 }, { "epoch": 17.134136546184738, "grad_norm": 21.622127532958984, "learning_rate": 4.1425806451612904e-05, "loss": 0.5542, "step": 2658 }, { "epoch": 17.140562248995984, "grad_norm": 5.888617038726807, "learning_rate": 4.142258064516129e-05, "loss": 0.3692, "step": 2659 }, { "epoch": 17.14698795180723, "grad_norm": 12.651172637939453, "learning_rate": 4.1419354838709676e-05, "loss": 0.5901, "step": 2660 }, { "epoch": 17.153413654618475, "grad_norm": 17.347734451293945, "learning_rate": 4.141612903225807e-05, "loss": 0.4859, "step": 2661 }, { "epoch": 17.159839357429718, "grad_norm": 14.83842658996582, "learning_rate": 4.1412903225806454e-05, "loss": 0.3003, "step": 2662 }, { "epoch": 17.166265060240963, "grad_norm": 6.58375358581543, "learning_rate": 4.140967741935484e-05, "loss": 0.2947, "step": 2663 }, { "epoch": 17.17269076305221, "grad_norm": 25.25887680053711, "learning_rate": 4.140645161290323e-05, "loss": 0.4602, "step": 2664 }, { "epoch": 17.179116465863455, "grad_norm": 21.24820899963379, "learning_rate": 4.140322580645162e-05, "loss": 0.3299, "step": 2665 }, { "epoch": 17.185542168674697, "grad_norm": 13.40915298461914, "learning_rate": 4.14e-05, "loss": 0.4398, "step": 2666 }, { "epoch": 17.191967871485943, "grad_norm": 1.306782603263855, "learning_rate": 4.139677419354839e-05, "loss": 0.2386, "step": 2667 }, { "epoch": 17.19839357429719, "grad_norm": 5.570834636688232, "learning_rate": 4.1393548387096775e-05, "loss": 0.2361, "step": 2668 }, { "epoch": 17.204819277108435, "grad_norm": 33.03346633911133, "learning_rate": 4.139032258064516e-05, "loss": 0.3571, "step": 2669 }, { "epoch": 17.21124497991968, "grad_norm": 2.8058555126190186, "learning_rate": 4.138709677419355e-05, "loss": 0.3737, "step": 2670 }, { "epoch": 17.217670682730922, "grad_norm": 11.23360824584961, "learning_rate": 4.138387096774194e-05, "loss": 0.32, "step": 2671 }, { "epoch": 17.22409638554217, "grad_norm": 5.704238414764404, "learning_rate": 4.1380645161290324e-05, "loss": 0.5888, "step": 2672 }, { "epoch": 17.230522088353414, "grad_norm": 12.849963188171387, "learning_rate": 4.137741935483871e-05, "loss": 0.7145, "step": 2673 }, { "epoch": 17.23694779116466, "grad_norm": 11.129168510437012, "learning_rate": 4.1374193548387096e-05, "loss": 0.2378, "step": 2674 }, { "epoch": 17.243373493975902, "grad_norm": 6.828073024749756, "learning_rate": 4.137096774193548e-05, "loss": 0.4416, "step": 2675 }, { "epoch": 17.249799196787148, "grad_norm": 4.767469882965088, "learning_rate": 4.1367741935483874e-05, "loss": 0.1216, "step": 2676 }, { "epoch": 17.256224899598394, "grad_norm": 8.054969787597656, "learning_rate": 4.136451612903226e-05, "loss": 0.2224, "step": 2677 }, { "epoch": 17.26265060240964, "grad_norm": 11.425410270690918, "learning_rate": 4.1361290322580645e-05, "loss": 0.4173, "step": 2678 }, { "epoch": 17.269076305220885, "grad_norm": 4.795685768127441, "learning_rate": 4.135806451612904e-05, "loss": 0.4281, "step": 2679 }, { "epoch": 17.275502008032127, "grad_norm": 19.52126693725586, "learning_rate": 4.1354838709677424e-05, "loss": 0.3081, "step": 2680 }, { "epoch": 17.281927710843373, "grad_norm": 22.563114166259766, "learning_rate": 4.13516129032258e-05, "loss": 0.501, "step": 2681 }, { "epoch": 17.28835341365462, "grad_norm": 10.61418342590332, "learning_rate": 4.1348387096774195e-05, "loss": 0.6056, "step": 2682 }, { "epoch": 17.294779116465865, "grad_norm": 9.821906089782715, "learning_rate": 4.134516129032258e-05, "loss": 0.5849, "step": 2683 }, { "epoch": 17.301204819277107, "grad_norm": 10.059077262878418, "learning_rate": 4.1341935483870966e-05, "loss": 0.542, "step": 2684 }, { "epoch": 17.307630522088353, "grad_norm": 26.670513153076172, "learning_rate": 4.133870967741936e-05, "loss": 0.4726, "step": 2685 }, { "epoch": 17.3140562248996, "grad_norm": 13.13899040222168, "learning_rate": 4.1335483870967745e-05, "loss": 0.4399, "step": 2686 }, { "epoch": 17.320481927710844, "grad_norm": 6.089256286621094, "learning_rate": 4.133225806451613e-05, "loss": 0.3365, "step": 2687 }, { "epoch": 17.32690763052209, "grad_norm": 12.525248527526855, "learning_rate": 4.132903225806452e-05, "loss": 0.5659, "step": 2688 }, { "epoch": 17.333333333333332, "grad_norm": 22.797527313232422, "learning_rate": 4.13258064516129e-05, "loss": 0.6057, "step": 2689 }, { "epoch": 17.339759036144578, "grad_norm": 6.480545997619629, "learning_rate": 4.132258064516129e-05, "loss": 0.4021, "step": 2690 }, { "epoch": 17.346184738955824, "grad_norm": 22.226076126098633, "learning_rate": 4.131935483870968e-05, "loss": 0.2421, "step": 2691 }, { "epoch": 17.35261044176707, "grad_norm": 15.551674842834473, "learning_rate": 4.1316129032258066e-05, "loss": 0.6838, "step": 2692 }, { "epoch": 17.35903614457831, "grad_norm": 12.819093704223633, "learning_rate": 4.131290322580645e-05, "loss": 0.2916, "step": 2693 }, { "epoch": 17.365461847389557, "grad_norm": 9.421598434448242, "learning_rate": 4.1309677419354844e-05, "loss": 0.2729, "step": 2694 }, { "epoch": 17.371887550200803, "grad_norm": 11.598146438598633, "learning_rate": 4.130645161290323e-05, "loss": 0.4278, "step": 2695 }, { "epoch": 17.37831325301205, "grad_norm": 6.7154951095581055, "learning_rate": 4.1303225806451615e-05, "loss": 0.4323, "step": 2696 }, { "epoch": 17.384738955823295, "grad_norm": 4.674041748046875, "learning_rate": 4.13e-05, "loss": 0.5586, "step": 2697 }, { "epoch": 17.391164658634537, "grad_norm": 3.211707353591919, "learning_rate": 4.129677419354839e-05, "loss": 0.2759, "step": 2698 }, { "epoch": 17.397590361445783, "grad_norm": 31.45516586303711, "learning_rate": 4.129354838709677e-05, "loss": 0.3216, "step": 2699 }, { "epoch": 17.40401606425703, "grad_norm": 14.697972297668457, "learning_rate": 4.1290322580645165e-05, "loss": 0.5017, "step": 2700 }, { "epoch": 17.410441767068274, "grad_norm": 3.275519371032715, "learning_rate": 4.128709677419355e-05, "loss": 0.3786, "step": 2701 }, { "epoch": 17.416867469879517, "grad_norm": 10.362958908081055, "learning_rate": 4.1283870967741936e-05, "loss": 0.4351, "step": 2702 }, { "epoch": 17.423293172690762, "grad_norm": 26.546260833740234, "learning_rate": 4.128064516129033e-05, "loss": 0.614, "step": 2703 }, { "epoch": 17.429718875502008, "grad_norm": 22.549501419067383, "learning_rate": 4.127741935483871e-05, "loss": 0.7848, "step": 2704 }, { "epoch": 17.436144578313254, "grad_norm": 11.353501319885254, "learning_rate": 4.1274193548387093e-05, "loss": 0.3872, "step": 2705 }, { "epoch": 17.4425702811245, "grad_norm": 9.216573715209961, "learning_rate": 4.1270967741935486e-05, "loss": 0.3585, "step": 2706 }, { "epoch": 17.448995983935742, "grad_norm": 4.760696887969971, "learning_rate": 4.126774193548387e-05, "loss": 0.2705, "step": 2707 }, { "epoch": 17.455421686746988, "grad_norm": 45.42566680908203, "learning_rate": 4.126451612903226e-05, "loss": 0.6852, "step": 2708 }, { "epoch": 17.461847389558233, "grad_norm": 3.6249544620513916, "learning_rate": 4.126129032258065e-05, "loss": 0.5639, "step": 2709 }, { "epoch": 17.46827309236948, "grad_norm": 13.202280044555664, "learning_rate": 4.1258064516129036e-05, "loss": 0.2334, "step": 2710 }, { "epoch": 17.47469879518072, "grad_norm": 8.313739776611328, "learning_rate": 4.125483870967742e-05, "loss": 0.6285, "step": 2711 }, { "epoch": 17.481124497991967, "grad_norm": 2.3562002182006836, "learning_rate": 4.125161290322581e-05, "loss": 0.3594, "step": 2712 }, { "epoch": 17.487550200803213, "grad_norm": 8.349257469177246, "learning_rate": 4.124838709677419e-05, "loss": 0.3139, "step": 2713 }, { "epoch": 17.49397590361446, "grad_norm": 7.059516429901123, "learning_rate": 4.124516129032258e-05, "loss": 0.3844, "step": 2714 }, { "epoch": 17.500401606425704, "grad_norm": 12.54814624786377, "learning_rate": 4.124193548387097e-05, "loss": 0.2143, "step": 2715 }, { "epoch": 17.506827309236947, "grad_norm": 10.803960800170898, "learning_rate": 4.1238709677419357e-05, "loss": 0.1819, "step": 2716 }, { "epoch": 17.513253012048192, "grad_norm": 20.778902053833008, "learning_rate": 4.123548387096774e-05, "loss": 0.4573, "step": 2717 }, { "epoch": 17.519678714859438, "grad_norm": 20.81266212463379, "learning_rate": 4.1232258064516135e-05, "loss": 0.4986, "step": 2718 }, { "epoch": 17.526104417670684, "grad_norm": 17.81420135498047, "learning_rate": 4.122903225806452e-05, "loss": 0.2161, "step": 2719 }, { "epoch": 17.532530120481926, "grad_norm": 13.966682434082031, "learning_rate": 4.1225806451612906e-05, "loss": 0.3764, "step": 2720 }, { "epoch": 17.538955823293172, "grad_norm": 4.634941577911377, "learning_rate": 4.122258064516129e-05, "loss": 0.3247, "step": 2721 }, { "epoch": 17.545381526104418, "grad_norm": 16.41790771484375, "learning_rate": 4.121935483870968e-05, "loss": 0.2062, "step": 2722 }, { "epoch": 17.551807228915663, "grad_norm": 3.981204032897949, "learning_rate": 4.121612903225806e-05, "loss": 0.5068, "step": 2723 }, { "epoch": 17.55823293172691, "grad_norm": 12.680961608886719, "learning_rate": 4.1212903225806456e-05, "loss": 0.5032, "step": 2724 }, { "epoch": 17.56465863453815, "grad_norm": 14.982961654663086, "learning_rate": 4.120967741935484e-05, "loss": 0.4882, "step": 2725 }, { "epoch": 17.571084337349397, "grad_norm": 8.640175819396973, "learning_rate": 4.120645161290323e-05, "loss": 0.1863, "step": 2726 }, { "epoch": 17.577510040160643, "grad_norm": 8.027534484863281, "learning_rate": 4.120322580645162e-05, "loss": 0.4835, "step": 2727 }, { "epoch": 17.58393574297189, "grad_norm": 20.20481300354004, "learning_rate": 4.12e-05, "loss": 0.5683, "step": 2728 }, { "epoch": 17.59036144578313, "grad_norm": 17.497600555419922, "learning_rate": 4.1196774193548384e-05, "loss": 0.464, "step": 2729 }, { "epoch": 17.596787148594377, "grad_norm": 9.8389253616333, "learning_rate": 4.119354838709678e-05, "loss": 0.3431, "step": 2730 }, { "epoch": 17.603212851405623, "grad_norm": 20.896329879760742, "learning_rate": 4.119032258064516e-05, "loss": 0.4247, "step": 2731 }, { "epoch": 17.60963855421687, "grad_norm": 10.361952781677246, "learning_rate": 4.118709677419355e-05, "loss": 0.2864, "step": 2732 }, { "epoch": 17.616064257028114, "grad_norm": 13.948866844177246, "learning_rate": 4.118387096774194e-05, "loss": 0.4393, "step": 2733 }, { "epoch": 17.622489959839356, "grad_norm": 7.284894943237305, "learning_rate": 4.1180645161290326e-05, "loss": 0.3386, "step": 2734 }, { "epoch": 17.628915662650602, "grad_norm": 13.080557823181152, "learning_rate": 4.117741935483871e-05, "loss": 0.2939, "step": 2735 }, { "epoch": 17.635341365461848, "grad_norm": 10.109557151794434, "learning_rate": 4.11741935483871e-05, "loss": 0.4051, "step": 2736 }, { "epoch": 17.641767068273094, "grad_norm": 6.927098751068115, "learning_rate": 4.1170967741935483e-05, "loss": 0.3048, "step": 2737 }, { "epoch": 17.648192771084336, "grad_norm": 10.310840606689453, "learning_rate": 4.116774193548387e-05, "loss": 0.3743, "step": 2738 }, { "epoch": 17.65461847389558, "grad_norm": 2.76889967918396, "learning_rate": 4.116451612903226e-05, "loss": 0.34, "step": 2739 }, { "epoch": 17.661044176706827, "grad_norm": 9.178025245666504, "learning_rate": 4.116129032258065e-05, "loss": 0.3215, "step": 2740 }, { "epoch": 17.667469879518073, "grad_norm": 12.581595420837402, "learning_rate": 4.115806451612903e-05, "loss": 0.4272, "step": 2741 }, { "epoch": 17.67389558232932, "grad_norm": 11.315430641174316, "learning_rate": 4.1154838709677426e-05, "loss": 0.3303, "step": 2742 }, { "epoch": 17.68032128514056, "grad_norm": 11.048087120056152, "learning_rate": 4.1151612903225804e-05, "loss": 0.4711, "step": 2743 }, { "epoch": 17.686746987951807, "grad_norm": 12.49985122680664, "learning_rate": 4.114838709677419e-05, "loss": 0.438, "step": 2744 }, { "epoch": 17.693172690763053, "grad_norm": 19.322021484375, "learning_rate": 4.114516129032258e-05, "loss": 0.3888, "step": 2745 }, { "epoch": 17.6995983935743, "grad_norm": 40.160091400146484, "learning_rate": 4.114193548387097e-05, "loss": 0.536, "step": 2746 }, { "epoch": 17.70602409638554, "grad_norm": 13.758234977722168, "learning_rate": 4.1138709677419354e-05, "loss": 0.3635, "step": 2747 }, { "epoch": 17.712449799196786, "grad_norm": 23.76569938659668, "learning_rate": 4.1135483870967747e-05, "loss": 0.4208, "step": 2748 }, { "epoch": 17.718875502008032, "grad_norm": 5.453919410705566, "learning_rate": 4.113225806451613e-05, "loss": 0.2655, "step": 2749 }, { "epoch": 17.725301204819278, "grad_norm": 18.891605377197266, "learning_rate": 4.112903225806452e-05, "loss": 0.3354, "step": 2750 }, { "epoch": 17.731726907630524, "grad_norm": 4.558624267578125, "learning_rate": 4.1125806451612904e-05, "loss": 0.3623, "step": 2751 }, { "epoch": 17.738152610441766, "grad_norm": 30.650962829589844, "learning_rate": 4.112258064516129e-05, "loss": 0.4335, "step": 2752 }, { "epoch": 17.74457831325301, "grad_norm": 3.7015879154205322, "learning_rate": 4.1119354838709675e-05, "loss": 0.4129, "step": 2753 }, { "epoch": 17.751004016064257, "grad_norm": 25.490760803222656, "learning_rate": 4.111612903225807e-05, "loss": 0.3251, "step": 2754 }, { "epoch": 17.757429718875503, "grad_norm": 9.601323127746582, "learning_rate": 4.111290322580645e-05, "loss": 0.4956, "step": 2755 }, { "epoch": 17.763855421686745, "grad_norm": 1.2637052536010742, "learning_rate": 4.110967741935484e-05, "loss": 0.2224, "step": 2756 }, { "epoch": 17.77028112449799, "grad_norm": 21.393892288208008, "learning_rate": 4.110645161290323e-05, "loss": 0.174, "step": 2757 }, { "epoch": 17.776706827309237, "grad_norm": 10.220739364624023, "learning_rate": 4.110322580645162e-05, "loss": 0.3284, "step": 2758 }, { "epoch": 17.783132530120483, "grad_norm": 24.614961624145508, "learning_rate": 4.11e-05, "loss": 0.6726, "step": 2759 }, { "epoch": 17.78955823293173, "grad_norm": 16.768606185913086, "learning_rate": 4.109677419354839e-05, "loss": 0.3065, "step": 2760 }, { "epoch": 17.79598393574297, "grad_norm": 24.46707534790039, "learning_rate": 4.1093548387096774e-05, "loss": 0.4379, "step": 2761 }, { "epoch": 17.802409638554217, "grad_norm": 3.917388677597046, "learning_rate": 4.109032258064516e-05, "loss": 0.4933, "step": 2762 }, { "epoch": 17.808835341365462, "grad_norm": 15.467706680297852, "learning_rate": 4.108709677419355e-05, "loss": 0.3382, "step": 2763 }, { "epoch": 17.815261044176708, "grad_norm": 19.45978355407715, "learning_rate": 4.108387096774194e-05, "loss": 0.3626, "step": 2764 }, { "epoch": 17.82168674698795, "grad_norm": 19.852205276489258, "learning_rate": 4.1080645161290324e-05, "loss": 0.1943, "step": 2765 }, { "epoch": 17.828112449799196, "grad_norm": 20.499408721923828, "learning_rate": 4.1077419354838716e-05, "loss": 0.751, "step": 2766 }, { "epoch": 17.834538152610442, "grad_norm": 65.06178283691406, "learning_rate": 4.1074193548387095e-05, "loss": 0.3279, "step": 2767 }, { "epoch": 17.840963855421688, "grad_norm": 8.687735557556152, "learning_rate": 4.107096774193548e-05, "loss": 0.4368, "step": 2768 }, { "epoch": 17.847389558232933, "grad_norm": 36.795188903808594, "learning_rate": 4.1067741935483874e-05, "loss": 0.4026, "step": 2769 }, { "epoch": 17.853815261044176, "grad_norm": 25.60209083557129, "learning_rate": 4.106451612903226e-05, "loss": 0.8861, "step": 2770 }, { "epoch": 17.86024096385542, "grad_norm": 10.773643493652344, "learning_rate": 4.1061290322580645e-05, "loss": 0.4017, "step": 2771 }, { "epoch": 17.866666666666667, "grad_norm": 26.240562438964844, "learning_rate": 4.105806451612904e-05, "loss": 0.7086, "step": 2772 }, { "epoch": 17.873092369477913, "grad_norm": 5.202223300933838, "learning_rate": 4.105483870967742e-05, "loss": 0.4424, "step": 2773 }, { "epoch": 17.879518072289155, "grad_norm": 21.569732666015625, "learning_rate": 4.105161290322581e-05, "loss": 0.3699, "step": 2774 }, { "epoch": 17.8859437751004, "grad_norm": 2.9916632175445557, "learning_rate": 4.1048387096774195e-05, "loss": 0.2589, "step": 2775 }, { "epoch": 17.892369477911647, "grad_norm": 39.90477752685547, "learning_rate": 4.104516129032258e-05, "loss": 0.4994, "step": 2776 }, { "epoch": 17.898795180722892, "grad_norm": 8.424264907836914, "learning_rate": 4.1041935483870966e-05, "loss": 0.4248, "step": 2777 }, { "epoch": 17.905220883534138, "grad_norm": 15.044805526733398, "learning_rate": 4.103870967741936e-05, "loss": 0.3734, "step": 2778 }, { "epoch": 17.91164658634538, "grad_norm": 11.351819038391113, "learning_rate": 4.1035483870967744e-05, "loss": 0.3774, "step": 2779 }, { "epoch": 17.918072289156626, "grad_norm": 2.9256186485290527, "learning_rate": 4.103225806451613e-05, "loss": 0.233, "step": 2780 }, { "epoch": 17.924497991967872, "grad_norm": 7.186807632446289, "learning_rate": 4.102903225806452e-05, "loss": 0.3092, "step": 2781 }, { "epoch": 17.930923694779118, "grad_norm": 10.538321495056152, "learning_rate": 4.10258064516129e-05, "loss": 0.4285, "step": 2782 }, { "epoch": 17.93734939759036, "grad_norm": 34.02754211425781, "learning_rate": 4.1022580645161294e-05, "loss": 0.5299, "step": 2783 }, { "epoch": 17.943775100401606, "grad_norm": 10.006622314453125, "learning_rate": 4.101935483870968e-05, "loss": 0.2829, "step": 2784 }, { "epoch": 17.95020080321285, "grad_norm": 26.87874984741211, "learning_rate": 4.1016129032258065e-05, "loss": 0.3858, "step": 2785 }, { "epoch": 17.956626506024097, "grad_norm": 21.330533981323242, "learning_rate": 4.101290322580645e-05, "loss": 0.4099, "step": 2786 }, { "epoch": 17.963052208835343, "grad_norm": 12.099080085754395, "learning_rate": 4.100967741935484e-05, "loss": 0.6132, "step": 2787 }, { "epoch": 17.969477911646585, "grad_norm": 17.984983444213867, "learning_rate": 4.100645161290323e-05, "loss": 0.3341, "step": 2788 }, { "epoch": 17.97590361445783, "grad_norm": 5.2432379722595215, "learning_rate": 4.1003225806451615e-05, "loss": 0.43, "step": 2789 }, { "epoch": 17.982329317269077, "grad_norm": 11.000969886779785, "learning_rate": 4.1e-05, "loss": 0.4946, "step": 2790 }, { "epoch": 17.988755020080323, "grad_norm": 16.179134368896484, "learning_rate": 4.0996774193548386e-05, "loss": 0.6119, "step": 2791 }, { "epoch": 17.995180722891565, "grad_norm": 25.41037368774414, "learning_rate": 4.099354838709677e-05, "loss": 0.515, "step": 2792 }, { "epoch": 18.004819277108435, "grad_norm": 24.008615493774414, "learning_rate": 4.0990322580645164e-05, "loss": 0.708, "step": 2793 }, { "epoch": 18.011244979919677, "grad_norm": 6.619835376739502, "learning_rate": 4.098709677419355e-05, "loss": 0.2843, "step": 2794 }, { "epoch": 18.017670682730923, "grad_norm": 7.739819526672363, "learning_rate": 4.0983870967741936e-05, "loss": 0.28, "step": 2795 }, { "epoch": 18.02409638554217, "grad_norm": 10.73713207244873, "learning_rate": 4.098064516129033e-05, "loss": 0.5518, "step": 2796 }, { "epoch": 18.030522088353415, "grad_norm": 7.776186943054199, "learning_rate": 4.0977419354838714e-05, "loss": 0.2983, "step": 2797 }, { "epoch": 18.036947791164657, "grad_norm": 8.13533878326416, "learning_rate": 4.09741935483871e-05, "loss": 0.6266, "step": 2798 }, { "epoch": 18.043373493975903, "grad_norm": 11.173669815063477, "learning_rate": 4.0970967741935485e-05, "loss": 0.2928, "step": 2799 }, { "epoch": 18.04979919678715, "grad_norm": 10.929129600524902, "learning_rate": 4.096774193548387e-05, "loss": 0.369, "step": 2800 }, { "epoch": 18.056224899598394, "grad_norm": 8.129709243774414, "learning_rate": 4.096451612903226e-05, "loss": 0.367, "step": 2801 }, { "epoch": 18.06265060240964, "grad_norm": 17.0223331451416, "learning_rate": 4.096129032258065e-05, "loss": 0.9299, "step": 2802 }, { "epoch": 18.069076305220882, "grad_norm": 21.217023849487305, "learning_rate": 4.0958064516129035e-05, "loss": 0.8579, "step": 2803 }, { "epoch": 18.075502008032128, "grad_norm": 10.526494979858398, "learning_rate": 4.095483870967742e-05, "loss": 0.2863, "step": 2804 }, { "epoch": 18.081927710843374, "grad_norm": 33.29640579223633, "learning_rate": 4.095161290322581e-05, "loss": 0.4185, "step": 2805 }, { "epoch": 18.08835341365462, "grad_norm": 5.792956352233887, "learning_rate": 4.094838709677419e-05, "loss": 0.3719, "step": 2806 }, { "epoch": 18.094779116465862, "grad_norm": 7.026912689208984, "learning_rate": 4.094516129032258e-05, "loss": 0.4866, "step": 2807 }, { "epoch": 18.101204819277108, "grad_norm": 7.889013290405273, "learning_rate": 4.094193548387097e-05, "loss": 0.1306, "step": 2808 }, { "epoch": 18.107630522088353, "grad_norm": 13.920941352844238, "learning_rate": 4.0938709677419356e-05, "loss": 0.7834, "step": 2809 }, { "epoch": 18.1140562248996, "grad_norm": 14.969334602355957, "learning_rate": 4.093548387096774e-05, "loss": 0.5836, "step": 2810 }, { "epoch": 18.120481927710845, "grad_norm": 37.54371643066406, "learning_rate": 4.0932258064516134e-05, "loss": 0.4956, "step": 2811 }, { "epoch": 18.126907630522087, "grad_norm": 12.680665016174316, "learning_rate": 4.092903225806452e-05, "loss": 0.3499, "step": 2812 }, { "epoch": 18.133333333333333, "grad_norm": 26.196718215942383, "learning_rate": 4.0925806451612906e-05, "loss": 0.3909, "step": 2813 }, { "epoch": 18.13975903614458, "grad_norm": 23.855018615722656, "learning_rate": 4.092258064516129e-05, "loss": 0.4589, "step": 2814 }, { "epoch": 18.146184738955824, "grad_norm": 6.835495471954346, "learning_rate": 4.091935483870968e-05, "loss": 0.2319, "step": 2815 }, { "epoch": 18.152610441767067, "grad_norm": 15.208220481872559, "learning_rate": 4.091612903225806e-05, "loss": 0.3996, "step": 2816 }, { "epoch": 18.159036144578312, "grad_norm": 10.397377014160156, "learning_rate": 4.0912903225806455e-05, "loss": 0.4022, "step": 2817 }, { "epoch": 18.165461847389558, "grad_norm": 10.407573699951172, "learning_rate": 4.090967741935484e-05, "loss": 0.323, "step": 2818 }, { "epoch": 18.171887550200804, "grad_norm": 11.798656463623047, "learning_rate": 4.090645161290323e-05, "loss": 0.5143, "step": 2819 }, { "epoch": 18.17831325301205, "grad_norm": 10.407705307006836, "learning_rate": 4.090322580645162e-05, "loss": 0.3182, "step": 2820 }, { "epoch": 18.184738955823292, "grad_norm": 6.5133819580078125, "learning_rate": 4.09e-05, "loss": 0.2481, "step": 2821 }, { "epoch": 18.191164658634538, "grad_norm": 6.723387718200684, "learning_rate": 4.089677419354839e-05, "loss": 0.2132, "step": 2822 }, { "epoch": 18.197590361445783, "grad_norm": 23.5011043548584, "learning_rate": 4.0893548387096776e-05, "loss": 0.4912, "step": 2823 }, { "epoch": 18.20401606425703, "grad_norm": 20.043760299682617, "learning_rate": 4.089032258064516e-05, "loss": 0.421, "step": 2824 }, { "epoch": 18.21044176706827, "grad_norm": 10.243109703063965, "learning_rate": 4.088709677419355e-05, "loss": 0.3579, "step": 2825 }, { "epoch": 18.216867469879517, "grad_norm": 24.86797523498535, "learning_rate": 4.088387096774194e-05, "loss": 0.4203, "step": 2826 }, { "epoch": 18.223293172690763, "grad_norm": 15.575715065002441, "learning_rate": 4.0880645161290326e-05, "loss": 0.2373, "step": 2827 }, { "epoch": 18.22971887550201, "grad_norm": 12.491861343383789, "learning_rate": 4.087741935483871e-05, "loss": 0.4688, "step": 2828 }, { "epoch": 18.236144578313255, "grad_norm": 8.847874641418457, "learning_rate": 4.08741935483871e-05, "loss": 0.5006, "step": 2829 }, { "epoch": 18.242570281124497, "grad_norm": 9.715044021606445, "learning_rate": 4.087096774193548e-05, "loss": 0.2639, "step": 2830 }, { "epoch": 18.248995983935743, "grad_norm": 9.15538215637207, "learning_rate": 4.086774193548387e-05, "loss": 0.2914, "step": 2831 }, { "epoch": 18.25542168674699, "grad_norm": 18.109806060791016, "learning_rate": 4.086451612903226e-05, "loss": 0.5955, "step": 2832 }, { "epoch": 18.261847389558234, "grad_norm": 18.230369567871094, "learning_rate": 4.086129032258065e-05, "loss": 0.2607, "step": 2833 }, { "epoch": 18.268273092369476, "grad_norm": 9.237996101379395, "learning_rate": 4.085806451612903e-05, "loss": 0.3715, "step": 2834 }, { "epoch": 18.274698795180722, "grad_norm": 10.276531219482422, "learning_rate": 4.0854838709677425e-05, "loss": 0.5192, "step": 2835 }, { "epoch": 18.281124497991968, "grad_norm": 7.535825252532959, "learning_rate": 4.085161290322581e-05, "loss": 0.3688, "step": 2836 }, { "epoch": 18.287550200803214, "grad_norm": 7.825775146484375, "learning_rate": 4.0848387096774196e-05, "loss": 0.2105, "step": 2837 }, { "epoch": 18.29397590361446, "grad_norm": 6.900887489318848, "learning_rate": 4.084516129032258e-05, "loss": 0.2688, "step": 2838 }, { "epoch": 18.3004016064257, "grad_norm": 6.4309563636779785, "learning_rate": 4.084193548387097e-05, "loss": 0.325, "step": 2839 }, { "epoch": 18.306827309236947, "grad_norm": 13.308982849121094, "learning_rate": 4.0838709677419354e-05, "loss": 0.5851, "step": 2840 }, { "epoch": 18.313253012048193, "grad_norm": 4.096290588378906, "learning_rate": 4.0835483870967746e-05, "loss": 0.4012, "step": 2841 }, { "epoch": 18.31967871485944, "grad_norm": 11.207386016845703, "learning_rate": 4.083225806451613e-05, "loss": 0.5284, "step": 2842 }, { "epoch": 18.32610441767068, "grad_norm": 5.263863563537598, "learning_rate": 4.082903225806452e-05, "loss": 0.4122, "step": 2843 }, { "epoch": 18.332530120481927, "grad_norm": 5.185812473297119, "learning_rate": 4.082580645161291e-05, "loss": 0.2822, "step": 2844 }, { "epoch": 18.338955823293173, "grad_norm": 10.269573211669922, "learning_rate": 4.082258064516129e-05, "loss": 0.3635, "step": 2845 }, { "epoch": 18.34538152610442, "grad_norm": 11.250258445739746, "learning_rate": 4.081935483870968e-05, "loss": 0.2753, "step": 2846 }, { "epoch": 18.351807228915664, "grad_norm": 2.8153607845306396, "learning_rate": 4.081612903225807e-05, "loss": 0.5458, "step": 2847 }, { "epoch": 18.358232931726906, "grad_norm": 6.190710544586182, "learning_rate": 4.081290322580645e-05, "loss": 0.3783, "step": 2848 }, { "epoch": 18.364658634538152, "grad_norm": 18.83096694946289, "learning_rate": 4.080967741935484e-05, "loss": 0.2974, "step": 2849 }, { "epoch": 18.371084337349398, "grad_norm": 7.207024574279785, "learning_rate": 4.080645161290323e-05, "loss": 0.2679, "step": 2850 }, { "epoch": 18.377510040160644, "grad_norm": 8.890618324279785, "learning_rate": 4.080322580645162e-05, "loss": 0.2091, "step": 2851 }, { "epoch": 18.383935742971886, "grad_norm": 23.62449836730957, "learning_rate": 4.08e-05, "loss": 0.7564, "step": 2852 }, { "epoch": 18.39036144578313, "grad_norm": 5.2813720703125, "learning_rate": 4.079677419354839e-05, "loss": 0.3174, "step": 2853 }, { "epoch": 18.396787148594377, "grad_norm": 9.134225845336914, "learning_rate": 4.0793548387096774e-05, "loss": 0.4855, "step": 2854 }, { "epoch": 18.403212851405623, "grad_norm": 10.566487312316895, "learning_rate": 4.079032258064516e-05, "loss": 0.3078, "step": 2855 }, { "epoch": 18.40963855421687, "grad_norm": 17.937196731567383, "learning_rate": 4.078709677419355e-05, "loss": 0.4623, "step": 2856 }, { "epoch": 18.41606425702811, "grad_norm": 7.390125274658203, "learning_rate": 4.078387096774194e-05, "loss": 0.2734, "step": 2857 }, { "epoch": 18.422489959839357, "grad_norm": 2.123666763305664, "learning_rate": 4.0780645161290323e-05, "loss": 0.3715, "step": 2858 }, { "epoch": 18.428915662650603, "grad_norm": 3.9106411933898926, "learning_rate": 4.0777419354838716e-05, "loss": 0.4794, "step": 2859 }, { "epoch": 18.43534136546185, "grad_norm": 3.260589361190796, "learning_rate": 4.0774193548387095e-05, "loss": 0.1501, "step": 2860 }, { "epoch": 18.44176706827309, "grad_norm": 18.326608657836914, "learning_rate": 4.077096774193549e-05, "loss": 0.4446, "step": 2861 }, { "epoch": 18.448192771084337, "grad_norm": 20.01215362548828, "learning_rate": 4.076774193548387e-05, "loss": 0.4062, "step": 2862 }, { "epoch": 18.454618473895582, "grad_norm": 6.60083532333374, "learning_rate": 4.076451612903226e-05, "loss": 0.167, "step": 2863 }, { "epoch": 18.461044176706828, "grad_norm": 9.704972267150879, "learning_rate": 4.0761290322580644e-05, "loss": 0.5023, "step": 2864 }, { "epoch": 18.467469879518074, "grad_norm": 8.37039852142334, "learning_rate": 4.075806451612904e-05, "loss": 0.4586, "step": 2865 }, { "epoch": 18.473895582329316, "grad_norm": 9.591119766235352, "learning_rate": 4.075483870967742e-05, "loss": 0.6958, "step": 2866 }, { "epoch": 18.480321285140562, "grad_norm": 5.147849082946777, "learning_rate": 4.075161290322581e-05, "loss": 0.2302, "step": 2867 }, { "epoch": 18.486746987951808, "grad_norm": 14.645471572875977, "learning_rate": 4.0748387096774194e-05, "loss": 0.3443, "step": 2868 }, { "epoch": 18.493172690763053, "grad_norm": 11.087594985961914, "learning_rate": 4.074516129032258e-05, "loss": 0.3861, "step": 2869 }, { "epoch": 18.499598393574296, "grad_norm": 18.991071701049805, "learning_rate": 4.0741935483870965e-05, "loss": 0.4643, "step": 2870 }, { "epoch": 18.50602409638554, "grad_norm": 6.463563919067383, "learning_rate": 4.073870967741936e-05, "loss": 0.1885, "step": 2871 }, { "epoch": 18.512449799196787, "grad_norm": 13.878045082092285, "learning_rate": 4.0735483870967744e-05, "loss": 0.2186, "step": 2872 }, { "epoch": 18.518875502008033, "grad_norm": 14.017963409423828, "learning_rate": 4.073225806451613e-05, "loss": 0.5076, "step": 2873 }, { "epoch": 18.52530120481928, "grad_norm": 18.639280319213867, "learning_rate": 4.072903225806452e-05, "loss": 0.6679, "step": 2874 }, { "epoch": 18.53172690763052, "grad_norm": 11.485676765441895, "learning_rate": 4.072580645161291e-05, "loss": 0.6157, "step": 2875 }, { "epoch": 18.538152610441767, "grad_norm": 6.916253566741943, "learning_rate": 4.072258064516129e-05, "loss": 0.138, "step": 2876 }, { "epoch": 18.544578313253012, "grad_norm": 52.63363265991211, "learning_rate": 4.071935483870968e-05, "loss": 0.4482, "step": 2877 }, { "epoch": 18.551004016064258, "grad_norm": 7.923621654510498, "learning_rate": 4.0716129032258065e-05, "loss": 0.3051, "step": 2878 }, { "epoch": 18.5574297188755, "grad_norm": 38.3310546875, "learning_rate": 4.071290322580645e-05, "loss": 0.6413, "step": 2879 }, { "epoch": 18.563855421686746, "grad_norm": 16.935016632080078, "learning_rate": 4.070967741935484e-05, "loss": 0.6026, "step": 2880 }, { "epoch": 18.570281124497992, "grad_norm": 15.804218292236328, "learning_rate": 4.070645161290323e-05, "loss": 0.4587, "step": 2881 }, { "epoch": 18.576706827309238, "grad_norm": 13.550766944885254, "learning_rate": 4.0703225806451614e-05, "loss": 0.4333, "step": 2882 }, { "epoch": 18.583132530120483, "grad_norm": 19.820039749145508, "learning_rate": 4.07e-05, "loss": 0.5611, "step": 2883 }, { "epoch": 18.589558232931726, "grad_norm": 20.845661163330078, "learning_rate": 4.0696774193548386e-05, "loss": 0.4194, "step": 2884 }, { "epoch": 18.59598393574297, "grad_norm": 5.1049723625183105, "learning_rate": 4.069354838709678e-05, "loss": 0.28, "step": 2885 }, { "epoch": 18.602409638554217, "grad_norm": 35.48162078857422, "learning_rate": 4.0690322580645164e-05, "loss": 0.4345, "step": 2886 }, { "epoch": 18.608835341365463, "grad_norm": 9.677090644836426, "learning_rate": 4.068709677419355e-05, "loss": 0.3144, "step": 2887 }, { "epoch": 18.615261044176705, "grad_norm": 33.8408203125, "learning_rate": 4.0683870967741935e-05, "loss": 0.5418, "step": 2888 }, { "epoch": 18.62168674698795, "grad_norm": 21.283370971679688, "learning_rate": 4.068064516129033e-05, "loss": 0.5275, "step": 2889 }, { "epoch": 18.628112449799197, "grad_norm": 14.475394248962402, "learning_rate": 4.0677419354838713e-05, "loss": 0.4275, "step": 2890 }, { "epoch": 18.634538152610443, "grad_norm": 14.18741512298584, "learning_rate": 4.06741935483871e-05, "loss": 0.3524, "step": 2891 }, { "epoch": 18.64096385542169, "grad_norm": 30.30306053161621, "learning_rate": 4.0670967741935485e-05, "loss": 0.5329, "step": 2892 }, { "epoch": 18.64738955823293, "grad_norm": 10.42248249053955, "learning_rate": 4.066774193548387e-05, "loss": 0.1673, "step": 2893 }, { "epoch": 18.653815261044176, "grad_norm": 15.950190544128418, "learning_rate": 4.0664516129032256e-05, "loss": 0.2438, "step": 2894 }, { "epoch": 18.660240963855422, "grad_norm": 19.170442581176758, "learning_rate": 4.066129032258065e-05, "loss": 0.6441, "step": 2895 }, { "epoch": 18.666666666666668, "grad_norm": 6.71483039855957, "learning_rate": 4.0658064516129034e-05, "loss": 0.178, "step": 2896 }, { "epoch": 18.67309236947791, "grad_norm": 19.470497131347656, "learning_rate": 4.065483870967742e-05, "loss": 0.3717, "step": 2897 }, { "epoch": 18.679518072289156, "grad_norm": 13.553448677062988, "learning_rate": 4.065161290322581e-05, "loss": 0.3896, "step": 2898 }, { "epoch": 18.6859437751004, "grad_norm": 8.70510196685791, "learning_rate": 4.064838709677419e-05, "loss": 0.3078, "step": 2899 }, { "epoch": 18.692369477911647, "grad_norm": 18.398427963256836, "learning_rate": 4.0645161290322584e-05, "loss": 0.2978, "step": 2900 }, { "epoch": 18.698795180722893, "grad_norm": 21.31379508972168, "learning_rate": 4.064193548387097e-05, "loss": 0.4418, "step": 2901 }, { "epoch": 18.705220883534135, "grad_norm": 12.783194541931152, "learning_rate": 4.0638709677419355e-05, "loss": 0.4196, "step": 2902 }, { "epoch": 18.71164658634538, "grad_norm": 4.590249538421631, "learning_rate": 4.063548387096774e-05, "loss": 0.3204, "step": 2903 }, { "epoch": 18.718072289156627, "grad_norm": 91.50357818603516, "learning_rate": 4.0632258064516134e-05, "loss": 0.2065, "step": 2904 }, { "epoch": 18.724497991967873, "grad_norm": 7.7415971755981445, "learning_rate": 4.062903225806452e-05, "loss": 0.2576, "step": 2905 }, { "epoch": 18.730923694779115, "grad_norm": 21.890443801879883, "learning_rate": 4.0625806451612905e-05, "loss": 0.5875, "step": 2906 }, { "epoch": 18.73734939759036, "grad_norm": 10.595995903015137, "learning_rate": 4.062258064516129e-05, "loss": 0.5308, "step": 2907 }, { "epoch": 18.743775100401606, "grad_norm": 12.281490325927734, "learning_rate": 4.0619354838709677e-05, "loss": 0.2641, "step": 2908 }, { "epoch": 18.750200803212852, "grad_norm": 14.059441566467285, "learning_rate": 4.061612903225807e-05, "loss": 0.5449, "step": 2909 }, { "epoch": 18.756626506024098, "grad_norm": 3.0055274963378906, "learning_rate": 4.0612903225806455e-05, "loss": 0.2065, "step": 2910 }, { "epoch": 18.76305220883534, "grad_norm": 7.564716339111328, "learning_rate": 4.060967741935484e-05, "loss": 0.3697, "step": 2911 }, { "epoch": 18.769477911646586, "grad_norm": 17.84307289123535, "learning_rate": 4.0606451612903226e-05, "loss": 0.3037, "step": 2912 }, { "epoch": 18.77590361445783, "grad_norm": 16.759395599365234, "learning_rate": 4.060322580645162e-05, "loss": 0.6875, "step": 2913 }, { "epoch": 18.782329317269078, "grad_norm": 14.413251876831055, "learning_rate": 4.0600000000000004e-05, "loss": 0.403, "step": 2914 }, { "epoch": 18.78875502008032, "grad_norm": 4.201951503753662, "learning_rate": 4.059677419354839e-05, "loss": 0.5584, "step": 2915 }, { "epoch": 18.795180722891565, "grad_norm": 27.58136749267578, "learning_rate": 4.0593548387096776e-05, "loss": 0.7339, "step": 2916 }, { "epoch": 18.80160642570281, "grad_norm": 2.171137809753418, "learning_rate": 4.059032258064516e-05, "loss": 0.1629, "step": 2917 }, { "epoch": 18.808032128514057, "grad_norm": 7.74688196182251, "learning_rate": 4.058709677419355e-05, "loss": 0.3151, "step": 2918 }, { "epoch": 18.814457831325303, "grad_norm": 10.223246574401855, "learning_rate": 4.058387096774194e-05, "loss": 0.3522, "step": 2919 }, { "epoch": 18.820883534136545, "grad_norm": 8.619744300842285, "learning_rate": 4.0580645161290325e-05, "loss": 0.3747, "step": 2920 }, { "epoch": 18.82730923694779, "grad_norm": 14.040658950805664, "learning_rate": 4.057741935483871e-05, "loss": 0.4434, "step": 2921 }, { "epoch": 18.833734939759037, "grad_norm": 11.112103462219238, "learning_rate": 4.05741935483871e-05, "loss": 0.377, "step": 2922 }, { "epoch": 18.840160642570282, "grad_norm": 15.445022583007812, "learning_rate": 4.057096774193548e-05, "loss": 0.2969, "step": 2923 }, { "epoch": 18.846586345381525, "grad_norm": 53.7274284362793, "learning_rate": 4.0567741935483875e-05, "loss": 0.6148, "step": 2924 }, { "epoch": 18.85301204819277, "grad_norm": 8.281532287597656, "learning_rate": 4.056451612903226e-05, "loss": 0.4847, "step": 2925 }, { "epoch": 18.859437751004016, "grad_norm": 4.5510149002075195, "learning_rate": 4.0561290322580646e-05, "loss": 0.2952, "step": 2926 }, { "epoch": 18.865863453815262, "grad_norm": 8.58899974822998, "learning_rate": 4.055806451612903e-05, "loss": 0.459, "step": 2927 }, { "epoch": 18.872289156626508, "grad_norm": 17.99184226989746, "learning_rate": 4.0554838709677425e-05, "loss": 0.4997, "step": 2928 }, { "epoch": 18.87871485943775, "grad_norm": 17.792564392089844, "learning_rate": 4.055161290322581e-05, "loss": 0.7733, "step": 2929 }, { "epoch": 18.885140562248996, "grad_norm": 8.547128677368164, "learning_rate": 4.0548387096774196e-05, "loss": 0.3606, "step": 2930 }, { "epoch": 18.89156626506024, "grad_norm": 8.40954875946045, "learning_rate": 4.054516129032258e-05, "loss": 0.3383, "step": 2931 }, { "epoch": 18.897991967871487, "grad_norm": 18.436542510986328, "learning_rate": 4.054193548387097e-05, "loss": 0.4518, "step": 2932 }, { "epoch": 18.90441767068273, "grad_norm": 6.33981990814209, "learning_rate": 4.053870967741935e-05, "loss": 0.5889, "step": 2933 }, { "epoch": 18.910843373493975, "grad_norm": 30.173288345336914, "learning_rate": 4.0535483870967746e-05, "loss": 0.6683, "step": 2934 }, { "epoch": 18.91726907630522, "grad_norm": 7.81049108505249, "learning_rate": 4.053225806451613e-05, "loss": 0.3528, "step": 2935 }, { "epoch": 18.923694779116467, "grad_norm": 12.611742973327637, "learning_rate": 4.052903225806452e-05, "loss": 0.4205, "step": 2936 }, { "epoch": 18.930120481927712, "grad_norm": 7.236326694488525, "learning_rate": 4.052580645161291e-05, "loss": 0.262, "step": 2937 }, { "epoch": 18.936546184738955, "grad_norm": 24.1197566986084, "learning_rate": 4.052258064516129e-05, "loss": 0.4778, "step": 2938 }, { "epoch": 18.9429718875502, "grad_norm": 30.63934326171875, "learning_rate": 4.051935483870968e-05, "loss": 0.4101, "step": 2939 }, { "epoch": 18.949397590361446, "grad_norm": 3.3496577739715576, "learning_rate": 4.0516129032258067e-05, "loss": 0.2235, "step": 2940 }, { "epoch": 18.955823293172692, "grad_norm": 11.064098358154297, "learning_rate": 4.051290322580645e-05, "loss": 0.4799, "step": 2941 }, { "epoch": 18.962248995983934, "grad_norm": 4.768432140350342, "learning_rate": 4.050967741935484e-05, "loss": 0.491, "step": 2942 }, { "epoch": 18.96867469879518, "grad_norm": 4.454521656036377, "learning_rate": 4.050645161290323e-05, "loss": 0.3371, "step": 2943 }, { "epoch": 18.975100401606426, "grad_norm": 13.42793083190918, "learning_rate": 4.0503225806451616e-05, "loss": 0.3357, "step": 2944 }, { "epoch": 18.98152610441767, "grad_norm": 7.740208148956299, "learning_rate": 4.05e-05, "loss": 0.5761, "step": 2945 }, { "epoch": 18.987951807228917, "grad_norm": 3.588958263397217, "learning_rate": 4.049677419354839e-05, "loss": 0.1931, "step": 2946 }, { "epoch": 18.99437751004016, "grad_norm": 12.179488182067871, "learning_rate": 4.049354838709677e-05, "loss": 0.5135, "step": 2947 }, { "epoch": 19.004016064257026, "grad_norm": 11.395217895507812, "learning_rate": 4.0490322580645166e-05, "loss": 0.4258, "step": 2948 }, { "epoch": 19.010441767068272, "grad_norm": 18.3454647064209, "learning_rate": 4.048709677419355e-05, "loss": 0.344, "step": 2949 }, { "epoch": 19.016867469879518, "grad_norm": 11.138448715209961, "learning_rate": 4.048387096774194e-05, "loss": 0.5515, "step": 2950 }, { "epoch": 19.023293172690764, "grad_norm": 6.047088623046875, "learning_rate": 4.048064516129032e-05, "loss": 0.497, "step": 2951 }, { "epoch": 19.02971887550201, "grad_norm": 19.9931583404541, "learning_rate": 4.0477419354838715e-05, "loss": 0.4201, "step": 2952 }, { "epoch": 19.03614457831325, "grad_norm": 33.79189682006836, "learning_rate": 4.0474193548387094e-05, "loss": 0.4968, "step": 2953 }, { "epoch": 19.042570281124497, "grad_norm": 5.043968200683594, "learning_rate": 4.047096774193549e-05, "loss": 0.3912, "step": 2954 }, { "epoch": 19.048995983935743, "grad_norm": 17.437631607055664, "learning_rate": 4.046774193548387e-05, "loss": 0.2616, "step": 2955 }, { "epoch": 19.05542168674699, "grad_norm": 3.631974697113037, "learning_rate": 4.046451612903226e-05, "loss": 0.4054, "step": 2956 }, { "epoch": 19.06184738955823, "grad_norm": 9.794013023376465, "learning_rate": 4.0461290322580644e-05, "loss": 0.5385, "step": 2957 }, { "epoch": 19.068273092369477, "grad_norm": 7.259638786315918, "learning_rate": 4.0458064516129036e-05, "loss": 0.3067, "step": 2958 }, { "epoch": 19.074698795180723, "grad_norm": 32.992916107177734, "learning_rate": 4.045483870967742e-05, "loss": 0.3962, "step": 2959 }, { "epoch": 19.08112449799197, "grad_norm": 8.443679809570312, "learning_rate": 4.045161290322581e-05, "loss": 0.3757, "step": 2960 }, { "epoch": 19.087550200803214, "grad_norm": 20.421598434448242, "learning_rate": 4.0448387096774194e-05, "loss": 0.4719, "step": 2961 }, { "epoch": 19.093975903614457, "grad_norm": 10.989665031433105, "learning_rate": 4.044516129032258e-05, "loss": 0.3563, "step": 2962 }, { "epoch": 19.100401606425702, "grad_norm": 11.528019905090332, "learning_rate": 4.044193548387097e-05, "loss": 0.519, "step": 2963 }, { "epoch": 19.106827309236948, "grad_norm": 8.83297348022461, "learning_rate": 4.043870967741936e-05, "loss": 0.3603, "step": 2964 }, { "epoch": 19.113253012048194, "grad_norm": 4.9485039710998535, "learning_rate": 4.043548387096774e-05, "loss": 0.3494, "step": 2965 }, { "epoch": 19.119678714859436, "grad_norm": 12.328977584838867, "learning_rate": 4.043225806451613e-05, "loss": 0.4528, "step": 2966 }, { "epoch": 19.126104417670682, "grad_norm": 9.518362998962402, "learning_rate": 4.042903225806452e-05, "loss": 0.256, "step": 2967 }, { "epoch": 19.132530120481928, "grad_norm": 17.873714447021484, "learning_rate": 4.042580645161291e-05, "loss": 0.4004, "step": 2968 }, { "epoch": 19.138955823293173, "grad_norm": 23.859912872314453, "learning_rate": 4.042258064516129e-05, "loss": 0.8873, "step": 2969 }, { "epoch": 19.14538152610442, "grad_norm": 14.30458927154541, "learning_rate": 4.041935483870968e-05, "loss": 0.4676, "step": 2970 }, { "epoch": 19.15180722891566, "grad_norm": 17.25421714782715, "learning_rate": 4.0416129032258064e-05, "loss": 0.3376, "step": 2971 }, { "epoch": 19.158232931726907, "grad_norm": 7.673294544219971, "learning_rate": 4.041290322580646e-05, "loss": 0.4225, "step": 2972 }, { "epoch": 19.164658634538153, "grad_norm": 20.945241928100586, "learning_rate": 4.040967741935484e-05, "loss": 0.4485, "step": 2973 }, { "epoch": 19.1710843373494, "grad_norm": 8.47669506072998, "learning_rate": 4.040645161290323e-05, "loss": 0.4716, "step": 2974 }, { "epoch": 19.17751004016064, "grad_norm": 6.982848167419434, "learning_rate": 4.0403225806451614e-05, "loss": 0.3565, "step": 2975 }, { "epoch": 19.183935742971887, "grad_norm": 14.027816772460938, "learning_rate": 4.0400000000000006e-05, "loss": 0.4175, "step": 2976 }, { "epoch": 19.190361445783132, "grad_norm": 18.09636688232422, "learning_rate": 4.0396774193548385e-05, "loss": 0.4022, "step": 2977 }, { "epoch": 19.196787148594378, "grad_norm": 40.82854080200195, "learning_rate": 4.039354838709678e-05, "loss": 0.4525, "step": 2978 }, { "epoch": 19.203212851405624, "grad_norm": 4.921298503875732, "learning_rate": 4.039032258064516e-05, "loss": 0.2901, "step": 2979 }, { "epoch": 19.209638554216866, "grad_norm": 3.161543130874634, "learning_rate": 4.038709677419355e-05, "loss": 0.4786, "step": 2980 }, { "epoch": 19.216064257028112, "grad_norm": 12.628063201904297, "learning_rate": 4.0383870967741935e-05, "loss": 0.309, "step": 2981 }, { "epoch": 19.222489959839358, "grad_norm": 16.65206527709961, "learning_rate": 4.038064516129033e-05, "loss": 0.4772, "step": 2982 }, { "epoch": 19.228915662650603, "grad_norm": 16.017011642456055, "learning_rate": 4.037741935483871e-05, "loss": 0.3174, "step": 2983 }, { "epoch": 19.235341365461846, "grad_norm": 5.699028491973877, "learning_rate": 4.03741935483871e-05, "loss": 0.2556, "step": 2984 }, { "epoch": 19.24176706827309, "grad_norm": 46.67585754394531, "learning_rate": 4.0370967741935484e-05, "loss": 0.2725, "step": 2985 }, { "epoch": 19.248192771084337, "grad_norm": 29.496417999267578, "learning_rate": 4.036774193548387e-05, "loss": 0.2642, "step": 2986 }, { "epoch": 19.254618473895583, "grad_norm": 9.638434410095215, "learning_rate": 4.036451612903226e-05, "loss": 0.5858, "step": 2987 }, { "epoch": 19.26104417670683, "grad_norm": 39.084381103515625, "learning_rate": 4.036129032258065e-05, "loss": 0.7401, "step": 2988 }, { "epoch": 19.26746987951807, "grad_norm": 12.71834945678711, "learning_rate": 4.0358064516129034e-05, "loss": 0.2333, "step": 2989 }, { "epoch": 19.273895582329317, "grad_norm": 19.456270217895508, "learning_rate": 4.035483870967742e-05, "loss": 0.5523, "step": 2990 }, { "epoch": 19.280321285140563, "grad_norm": 4.589573860168457, "learning_rate": 4.035161290322581e-05, "loss": 0.2946, "step": 2991 }, { "epoch": 19.28674698795181, "grad_norm": 7.962978363037109, "learning_rate": 4.034838709677419e-05, "loss": 0.1766, "step": 2992 }, { "epoch": 19.29317269076305, "grad_norm": 3.336972713470459, "learning_rate": 4.0345161290322584e-05, "loss": 0.1776, "step": 2993 }, { "epoch": 19.299598393574296, "grad_norm": 5.658544063568115, "learning_rate": 4.034193548387097e-05, "loss": 0.6134, "step": 2994 }, { "epoch": 19.306024096385542, "grad_norm": 4.219685077667236, "learning_rate": 4.0338709677419355e-05, "loss": 0.6838, "step": 2995 }, { "epoch": 19.312449799196788, "grad_norm": 7.5239105224609375, "learning_rate": 4.033548387096774e-05, "loss": 0.2884, "step": 2996 }, { "epoch": 19.318875502008034, "grad_norm": 13.647181510925293, "learning_rate": 4.033225806451613e-05, "loss": 0.2839, "step": 2997 }, { "epoch": 19.325301204819276, "grad_norm": 16.969541549682617, "learning_rate": 4.032903225806452e-05, "loss": 0.258, "step": 2998 }, { "epoch": 19.33172690763052, "grad_norm": 9.15221881866455, "learning_rate": 4.0325806451612905e-05, "loss": 0.3986, "step": 2999 }, { "epoch": 19.338152610441767, "grad_norm": 15.631939888000488, "learning_rate": 4.032258064516129e-05, "loss": 0.9395, "step": 3000 }, { "epoch": 19.344578313253013, "grad_norm": 13.410357475280762, "learning_rate": 4.0319354838709676e-05, "loss": 0.4242, "step": 3001 }, { "epoch": 19.351004016064255, "grad_norm": 13.059131622314453, "learning_rate": 4.031612903225807e-05, "loss": 0.4578, "step": 3002 }, { "epoch": 19.3574297188755, "grad_norm": 7.682995319366455, "learning_rate": 4.0312903225806454e-05, "loss": 0.5803, "step": 3003 }, { "epoch": 19.363855421686747, "grad_norm": 9.783590316772461, "learning_rate": 4.030967741935484e-05, "loss": 0.4853, "step": 3004 }, { "epoch": 19.370281124497993, "grad_norm": 7.345393180847168, "learning_rate": 4.0306451612903226e-05, "loss": 0.238, "step": 3005 }, { "epoch": 19.37670682730924, "grad_norm": 16.572765350341797, "learning_rate": 4.030322580645162e-05, "loss": 0.425, "step": 3006 }, { "epoch": 19.38313253012048, "grad_norm": 3.0071706771850586, "learning_rate": 4.0300000000000004e-05, "loss": 0.2063, "step": 3007 }, { "epoch": 19.389558232931726, "grad_norm": 12.636817932128906, "learning_rate": 4.029677419354839e-05, "loss": 0.5449, "step": 3008 }, { "epoch": 19.395983935742972, "grad_norm": 12.1331205368042, "learning_rate": 4.0293548387096775e-05, "loss": 0.3183, "step": 3009 }, { "epoch": 19.402409638554218, "grad_norm": 26.995763778686523, "learning_rate": 4.029032258064516e-05, "loss": 0.1226, "step": 3010 }, { "epoch": 19.40883534136546, "grad_norm": 25.437753677368164, "learning_rate": 4.0287096774193553e-05, "loss": 0.4742, "step": 3011 }, { "epoch": 19.415261044176706, "grad_norm": 30.411590576171875, "learning_rate": 4.028387096774194e-05, "loss": 0.3243, "step": 3012 }, { "epoch": 19.42168674698795, "grad_norm": 4.107144832611084, "learning_rate": 4.0280645161290325e-05, "loss": 0.169, "step": 3013 }, { "epoch": 19.428112449799197, "grad_norm": 4.2838897705078125, "learning_rate": 4.027741935483871e-05, "loss": 0.2615, "step": 3014 }, { "epoch": 19.434538152610443, "grad_norm": 3.012359857559204, "learning_rate": 4.02741935483871e-05, "loss": 0.3302, "step": 3015 }, { "epoch": 19.440963855421685, "grad_norm": 5.989187240600586, "learning_rate": 4.027096774193548e-05, "loss": 0.303, "step": 3016 }, { "epoch": 19.44738955823293, "grad_norm": 6.388246536254883, "learning_rate": 4.0267741935483874e-05, "loss": 0.2335, "step": 3017 }, { "epoch": 19.453815261044177, "grad_norm": 6.217385768890381, "learning_rate": 4.026451612903226e-05, "loss": 0.2628, "step": 3018 }, { "epoch": 19.460240963855423, "grad_norm": 13.785588264465332, "learning_rate": 4.0261290322580646e-05, "loss": 0.7163, "step": 3019 }, { "epoch": 19.466666666666665, "grad_norm": 40.476497650146484, "learning_rate": 4.025806451612903e-05, "loss": 0.2464, "step": 3020 }, { "epoch": 19.47309236947791, "grad_norm": 11.102096557617188, "learning_rate": 4.0254838709677424e-05, "loss": 0.5876, "step": 3021 }, { "epoch": 19.479518072289157, "grad_norm": 7.318199634552002, "learning_rate": 4.025161290322581e-05, "loss": 0.2994, "step": 3022 }, { "epoch": 19.485943775100402, "grad_norm": 25.944568634033203, "learning_rate": 4.0248387096774195e-05, "loss": 0.5908, "step": 3023 }, { "epoch": 19.492369477911648, "grad_norm": 13.677558898925781, "learning_rate": 4.024516129032258e-05, "loss": 0.3279, "step": 3024 }, { "epoch": 19.49879518072289, "grad_norm": 20.030559539794922, "learning_rate": 4.024193548387097e-05, "loss": 0.4495, "step": 3025 }, { "epoch": 19.505220883534136, "grad_norm": 5.108152866363525, "learning_rate": 4.023870967741936e-05, "loss": 0.2464, "step": 3026 }, { "epoch": 19.511646586345382, "grad_norm": 9.088566780090332, "learning_rate": 4.0235483870967745e-05, "loss": 0.5306, "step": 3027 }, { "epoch": 19.518072289156628, "grad_norm": 12.62138843536377, "learning_rate": 4.023225806451613e-05, "loss": 0.4218, "step": 3028 }, { "epoch": 19.52449799196787, "grad_norm": 48.35542297363281, "learning_rate": 4.0229032258064516e-05, "loss": 0.6403, "step": 3029 }, { "epoch": 19.530923694779116, "grad_norm": 10.865435600280762, "learning_rate": 4.022580645161291e-05, "loss": 0.4545, "step": 3030 }, { "epoch": 19.53734939759036, "grad_norm": 11.470409393310547, "learning_rate": 4.022258064516129e-05, "loss": 0.6298, "step": 3031 }, { "epoch": 19.543775100401607, "grad_norm": 5.964349746704102, "learning_rate": 4.021935483870968e-05, "loss": 0.2107, "step": 3032 }, { "epoch": 19.550200803212853, "grad_norm": 26.2325439453125, "learning_rate": 4.0216129032258066e-05, "loss": 0.4845, "step": 3033 }, { "epoch": 19.556626506024095, "grad_norm": 15.673453330993652, "learning_rate": 4.021290322580645e-05, "loss": 0.4355, "step": 3034 }, { "epoch": 19.56305220883534, "grad_norm": 16.027515411376953, "learning_rate": 4.0209677419354844e-05, "loss": 0.5075, "step": 3035 }, { "epoch": 19.569477911646587, "grad_norm": 4.97487211227417, "learning_rate": 4.020645161290323e-05, "loss": 0.29, "step": 3036 }, { "epoch": 19.575903614457832, "grad_norm": 17.049232482910156, "learning_rate": 4.0203225806451616e-05, "loss": 0.473, "step": 3037 }, { "epoch": 19.582329317269075, "grad_norm": 15.840766906738281, "learning_rate": 4.02e-05, "loss": 0.4875, "step": 3038 }, { "epoch": 19.58875502008032, "grad_norm": 30.5295467376709, "learning_rate": 4.019677419354839e-05, "loss": 0.8966, "step": 3039 }, { "epoch": 19.595180722891566, "grad_norm": 6.506639003753662, "learning_rate": 4.019354838709677e-05, "loss": 0.4299, "step": 3040 }, { "epoch": 19.601606425702812, "grad_norm": 10.732118606567383, "learning_rate": 4.0190322580645165e-05, "loss": 0.5014, "step": 3041 }, { "epoch": 19.608032128514058, "grad_norm": 8.884928703308105, "learning_rate": 4.018709677419355e-05, "loss": 0.406, "step": 3042 }, { "epoch": 19.6144578313253, "grad_norm": 12.509607315063477, "learning_rate": 4.018387096774194e-05, "loss": 0.624, "step": 3043 }, { "epoch": 19.620883534136546, "grad_norm": 9.61300277709961, "learning_rate": 4.018064516129032e-05, "loss": 0.4269, "step": 3044 }, { "epoch": 19.62730923694779, "grad_norm": 18.119537353515625, "learning_rate": 4.0177419354838715e-05, "loss": 0.4966, "step": 3045 }, { "epoch": 19.633734939759037, "grad_norm": 11.243851661682129, "learning_rate": 4.01741935483871e-05, "loss": 0.4371, "step": 3046 }, { "epoch": 19.64016064257028, "grad_norm": 5.951083183288574, "learning_rate": 4.0170967741935486e-05, "loss": 0.4673, "step": 3047 }, { "epoch": 19.646586345381525, "grad_norm": 7.586277008056641, "learning_rate": 4.016774193548387e-05, "loss": 0.5508, "step": 3048 }, { "epoch": 19.65301204819277, "grad_norm": 8.419976234436035, "learning_rate": 4.016451612903226e-05, "loss": 0.6633, "step": 3049 }, { "epoch": 19.659437751004017, "grad_norm": 63.20789337158203, "learning_rate": 4.016129032258065e-05, "loss": 0.2667, "step": 3050 }, { "epoch": 19.665863453815263, "grad_norm": 4.228516578674316, "learning_rate": 4.0158064516129036e-05, "loss": 0.4475, "step": 3051 }, { "epoch": 19.672289156626505, "grad_norm": 30.88907814025879, "learning_rate": 4.015483870967742e-05, "loss": 0.7645, "step": 3052 }, { "epoch": 19.67871485943775, "grad_norm": 15.586522102355957, "learning_rate": 4.015161290322581e-05, "loss": 0.4247, "step": 3053 }, { "epoch": 19.685140562248996, "grad_norm": 11.804200172424316, "learning_rate": 4.01483870967742e-05, "loss": 0.594, "step": 3054 }, { "epoch": 19.691566265060242, "grad_norm": 23.651710510253906, "learning_rate": 4.014516129032258e-05, "loss": 0.4172, "step": 3055 }, { "epoch": 19.697991967871484, "grad_norm": 13.645153045654297, "learning_rate": 4.014193548387097e-05, "loss": 0.4802, "step": 3056 }, { "epoch": 19.70441767068273, "grad_norm": 13.822563171386719, "learning_rate": 4.013870967741936e-05, "loss": 0.5515, "step": 3057 }, { "epoch": 19.710843373493976, "grad_norm": 18.154550552368164, "learning_rate": 4.013548387096774e-05, "loss": 0.3311, "step": 3058 }, { "epoch": 19.71726907630522, "grad_norm": 12.063791275024414, "learning_rate": 4.013225806451613e-05, "loss": 0.3438, "step": 3059 }, { "epoch": 19.723694779116467, "grad_norm": 6.730207920074463, "learning_rate": 4.012903225806452e-05, "loss": 0.4662, "step": 3060 }, { "epoch": 19.73012048192771, "grad_norm": 34.89542770385742, "learning_rate": 4.0125806451612906e-05, "loss": 0.6668, "step": 3061 }, { "epoch": 19.736546184738955, "grad_norm": 4.29250955581665, "learning_rate": 4.012258064516129e-05, "loss": 0.4257, "step": 3062 }, { "epoch": 19.7429718875502, "grad_norm": 1.7694456577301025, "learning_rate": 4.011935483870968e-05, "loss": 0.181, "step": 3063 }, { "epoch": 19.749397590361447, "grad_norm": 21.83246421813965, "learning_rate": 4.0116129032258064e-05, "loss": 0.4313, "step": 3064 }, { "epoch": 19.75582329317269, "grad_norm": 13.297554016113281, "learning_rate": 4.0112903225806456e-05, "loss": 0.4219, "step": 3065 }, { "epoch": 19.762248995983935, "grad_norm": 7.641295909881592, "learning_rate": 4.010967741935484e-05, "loss": 0.1916, "step": 3066 }, { "epoch": 19.76867469879518, "grad_norm": 25.299360275268555, "learning_rate": 4.010645161290323e-05, "loss": 0.5455, "step": 3067 }, { "epoch": 19.775100401606426, "grad_norm": 6.605160713195801, "learning_rate": 4.010322580645161e-05, "loss": 0.2777, "step": 3068 }, { "epoch": 19.781526104417672, "grad_norm": 12.083555221557617, "learning_rate": 4.0100000000000006e-05, "loss": 0.4913, "step": 3069 }, { "epoch": 19.787951807228914, "grad_norm": 5.139948844909668, "learning_rate": 4.0096774193548385e-05, "loss": 0.2326, "step": 3070 }, { "epoch": 19.79437751004016, "grad_norm": 16.149700164794922, "learning_rate": 4.009354838709678e-05, "loss": 0.653, "step": 3071 }, { "epoch": 19.800803212851406, "grad_norm": 14.691129684448242, "learning_rate": 4.009032258064516e-05, "loss": 0.6176, "step": 3072 }, { "epoch": 19.80722891566265, "grad_norm": 14.653351783752441, "learning_rate": 4.008709677419355e-05, "loss": 0.2914, "step": 3073 }, { "epoch": 19.813654618473894, "grad_norm": 10.073997497558594, "learning_rate": 4.008387096774194e-05, "loss": 0.4268, "step": 3074 }, { "epoch": 19.82008032128514, "grad_norm": 7.1353373527526855, "learning_rate": 4.008064516129033e-05, "loss": 0.2904, "step": 3075 }, { "epoch": 19.826506024096386, "grad_norm": 51.647972106933594, "learning_rate": 4.007741935483871e-05, "loss": 0.3118, "step": 3076 }, { "epoch": 19.83293172690763, "grad_norm": 6.784672260284424, "learning_rate": 4.00741935483871e-05, "loss": 0.3517, "step": 3077 }, { "epoch": 19.839357429718877, "grad_norm": 3.36034893989563, "learning_rate": 4.0070967741935484e-05, "loss": 0.4567, "step": 3078 }, { "epoch": 19.84578313253012, "grad_norm": 20.757204055786133, "learning_rate": 4.006774193548387e-05, "loss": 0.2378, "step": 3079 }, { "epoch": 19.852208835341365, "grad_norm": 25.94438934326172, "learning_rate": 4.006451612903226e-05, "loss": 0.7385, "step": 3080 }, { "epoch": 19.85863453815261, "grad_norm": 33.617919921875, "learning_rate": 4.006129032258065e-05, "loss": 0.3956, "step": 3081 }, { "epoch": 19.865060240963857, "grad_norm": 26.149995803833008, "learning_rate": 4.0058064516129033e-05, "loss": 0.3927, "step": 3082 }, { "epoch": 19.8714859437751, "grad_norm": 14.704350471496582, "learning_rate": 4.005483870967742e-05, "loss": 0.4891, "step": 3083 }, { "epoch": 19.877911646586345, "grad_norm": 19.552209854125977, "learning_rate": 4.005161290322581e-05, "loss": 0.349, "step": 3084 }, { "epoch": 19.88433734939759, "grad_norm": 8.267333030700684, "learning_rate": 4.00483870967742e-05, "loss": 0.3328, "step": 3085 }, { "epoch": 19.890763052208836, "grad_norm": 10.238679885864258, "learning_rate": 4.004516129032258e-05, "loss": 0.4249, "step": 3086 }, { "epoch": 19.897188755020082, "grad_norm": 10.242851257324219, "learning_rate": 4.004193548387097e-05, "loss": 0.4734, "step": 3087 }, { "epoch": 19.903614457831324, "grad_norm": 10.006227493286133, "learning_rate": 4.0038709677419354e-05, "loss": 0.5493, "step": 3088 }, { "epoch": 19.91004016064257, "grad_norm": 12.220799446105957, "learning_rate": 4.003548387096775e-05, "loss": 0.4663, "step": 3089 }, { "epoch": 19.916465863453816, "grad_norm": 6.694653034210205, "learning_rate": 4.003225806451613e-05, "loss": 0.3861, "step": 3090 }, { "epoch": 19.92289156626506, "grad_norm": 6.991049289703369, "learning_rate": 4.002903225806452e-05, "loss": 0.3719, "step": 3091 }, { "epoch": 19.929317269076304, "grad_norm": 2.898287057876587, "learning_rate": 4.0025806451612904e-05, "loss": 0.3747, "step": 3092 }, { "epoch": 19.93574297188755, "grad_norm": 2.395418882369995, "learning_rate": 4.002258064516129e-05, "loss": 0.1898, "step": 3093 }, { "epoch": 19.942168674698795, "grad_norm": 13.498381614685059, "learning_rate": 4.0019354838709675e-05, "loss": 0.4, "step": 3094 }, { "epoch": 19.94859437751004, "grad_norm": 9.378887176513672, "learning_rate": 4.001612903225807e-05, "loss": 0.2739, "step": 3095 }, { "epoch": 19.955020080321287, "grad_norm": 51.058197021484375, "learning_rate": 4.0012903225806454e-05, "loss": 0.2521, "step": 3096 }, { "epoch": 19.96144578313253, "grad_norm": 23.612102508544922, "learning_rate": 4.000967741935484e-05, "loss": 0.4678, "step": 3097 }, { "epoch": 19.967871485943775, "grad_norm": 5.709471225738525, "learning_rate": 4.000645161290323e-05, "loss": 0.4893, "step": 3098 }, { "epoch": 19.97429718875502, "grad_norm": 6.311336994171143, "learning_rate": 4.000322580645162e-05, "loss": 0.4576, "step": 3099 }, { "epoch": 19.980722891566266, "grad_norm": 10.75546646118164, "learning_rate": 4e-05, "loss": 0.4951, "step": 3100 }, { "epoch": 19.98714859437751, "grad_norm": 24.645435333251953, "learning_rate": 3.999677419354839e-05, "loss": 0.8388, "step": 3101 }, { "epoch": 19.993574297188754, "grad_norm": 9.715960502624512, "learning_rate": 3.9993548387096775e-05, "loss": 0.3584, "step": 3102 }, { "epoch": 20.00321285140562, "grad_norm": 8.57751750946045, "learning_rate": 3.999032258064516e-05, "loss": 0.505, "step": 3103 }, { "epoch": 20.009638554216867, "grad_norm": 18.990327835083008, "learning_rate": 3.998709677419355e-05, "loss": 0.6868, "step": 3104 }, { "epoch": 20.016064257028113, "grad_norm": 12.236085891723633, "learning_rate": 3.998387096774194e-05, "loss": 0.4668, "step": 3105 }, { "epoch": 20.02248995983936, "grad_norm": 3.154386043548584, "learning_rate": 3.9980645161290324e-05, "loss": 0.4112, "step": 3106 }, { "epoch": 20.0289156626506, "grad_norm": 5.087803840637207, "learning_rate": 3.997741935483871e-05, "loss": 0.2388, "step": 3107 }, { "epoch": 20.035341365461846, "grad_norm": 26.446393966674805, "learning_rate": 3.99741935483871e-05, "loss": 0.4779, "step": 3108 }, { "epoch": 20.041767068273092, "grad_norm": 8.4571533203125, "learning_rate": 3.997096774193548e-05, "loss": 0.3713, "step": 3109 }, { "epoch": 20.048192771084338, "grad_norm": 11.641880989074707, "learning_rate": 3.9967741935483874e-05, "loss": 0.2518, "step": 3110 }, { "epoch": 20.054618473895584, "grad_norm": 51.36195755004883, "learning_rate": 3.996451612903226e-05, "loss": 0.7402, "step": 3111 }, { "epoch": 20.061044176706826, "grad_norm": 9.939443588256836, "learning_rate": 3.9961290322580645e-05, "loss": 0.3564, "step": 3112 }, { "epoch": 20.06746987951807, "grad_norm": 11.242743492126465, "learning_rate": 3.995806451612904e-05, "loss": 0.7341, "step": 3113 }, { "epoch": 20.073895582329317, "grad_norm": 7.096105098724365, "learning_rate": 3.9954838709677423e-05, "loss": 0.2853, "step": 3114 }, { "epoch": 20.080321285140563, "grad_norm": 6.1082963943481445, "learning_rate": 3.995161290322581e-05, "loss": 0.164, "step": 3115 }, { "epoch": 20.086746987951805, "grad_norm": 11.849237442016602, "learning_rate": 3.9948387096774195e-05, "loss": 0.2748, "step": 3116 }, { "epoch": 20.09317269076305, "grad_norm": 6.588474750518799, "learning_rate": 3.994516129032258e-05, "loss": 0.229, "step": 3117 }, { "epoch": 20.099598393574297, "grad_norm": 16.725271224975586, "learning_rate": 3.9941935483870966e-05, "loss": 0.3121, "step": 3118 }, { "epoch": 20.106024096385543, "grad_norm": 31.763195037841797, "learning_rate": 3.993870967741936e-05, "loss": 0.3879, "step": 3119 }, { "epoch": 20.11244979919679, "grad_norm": 12.295952796936035, "learning_rate": 3.9935483870967745e-05, "loss": 0.453, "step": 3120 }, { "epoch": 20.11887550200803, "grad_norm": 6.484204292297363, "learning_rate": 3.993225806451613e-05, "loss": 0.2103, "step": 3121 }, { "epoch": 20.125301204819277, "grad_norm": 15.521539688110352, "learning_rate": 3.9929032258064516e-05, "loss": 0.527, "step": 3122 }, { "epoch": 20.131726907630522, "grad_norm": 15.156124114990234, "learning_rate": 3.992580645161291e-05, "loss": 0.284, "step": 3123 }, { "epoch": 20.138152610441768, "grad_norm": 15.247665405273438, "learning_rate": 3.9922580645161294e-05, "loss": 0.3486, "step": 3124 }, { "epoch": 20.14457831325301, "grad_norm": 16.77501678466797, "learning_rate": 3.991935483870968e-05, "loss": 0.3413, "step": 3125 }, { "epoch": 20.151004016064256, "grad_norm": 8.31629753112793, "learning_rate": 3.9916129032258066e-05, "loss": 0.4436, "step": 3126 }, { "epoch": 20.157429718875502, "grad_norm": 10.626673698425293, "learning_rate": 3.991290322580645e-05, "loss": 0.4059, "step": 3127 }, { "epoch": 20.163855421686748, "grad_norm": 16.099260330200195, "learning_rate": 3.9909677419354844e-05, "loss": 0.451, "step": 3128 }, { "epoch": 20.170281124497993, "grad_norm": 10.860342025756836, "learning_rate": 3.990645161290323e-05, "loss": 0.3294, "step": 3129 }, { "epoch": 20.176706827309236, "grad_norm": 4.921207427978516, "learning_rate": 3.9903225806451615e-05, "loss": 0.4662, "step": 3130 }, { "epoch": 20.18313253012048, "grad_norm": 16.46937370300293, "learning_rate": 3.99e-05, "loss": 0.4279, "step": 3131 }, { "epoch": 20.189558232931727, "grad_norm": 12.531882286071777, "learning_rate": 3.9896774193548387e-05, "loss": 0.5544, "step": 3132 }, { "epoch": 20.195983935742973, "grad_norm": 6.6427388191223145, "learning_rate": 3.989354838709677e-05, "loss": 0.5736, "step": 3133 }, { "epoch": 20.202409638554215, "grad_norm": 15.464712142944336, "learning_rate": 3.9890322580645165e-05, "loss": 0.4165, "step": 3134 }, { "epoch": 20.20883534136546, "grad_norm": 22.719650268554688, "learning_rate": 3.988709677419355e-05, "loss": 0.637, "step": 3135 }, { "epoch": 20.215261044176707, "grad_norm": 5.9624342918396, "learning_rate": 3.9883870967741936e-05, "loss": 0.4069, "step": 3136 }, { "epoch": 20.221686746987952, "grad_norm": 20.803627014160156, "learning_rate": 3.988064516129033e-05, "loss": 0.3004, "step": 3137 }, { "epoch": 20.228112449799198, "grad_norm": 5.512121677398682, "learning_rate": 3.9877419354838714e-05, "loss": 0.4796, "step": 3138 }, { "epoch": 20.23453815261044, "grad_norm": 14.196781158447266, "learning_rate": 3.98741935483871e-05, "loss": 0.3, "step": 3139 }, { "epoch": 20.240963855421686, "grad_norm": 45.03133010864258, "learning_rate": 3.9870967741935486e-05, "loss": 0.4776, "step": 3140 }, { "epoch": 20.247389558232932, "grad_norm": 7.270020961761475, "learning_rate": 3.986774193548387e-05, "loss": 0.2702, "step": 3141 }, { "epoch": 20.253815261044178, "grad_norm": 12.837575912475586, "learning_rate": 3.986451612903226e-05, "loss": 0.3409, "step": 3142 }, { "epoch": 20.26024096385542, "grad_norm": 7.744220733642578, "learning_rate": 3.986129032258065e-05, "loss": 0.2043, "step": 3143 }, { "epoch": 20.266666666666666, "grad_norm": 7.044261455535889, "learning_rate": 3.9858064516129035e-05, "loss": 0.2647, "step": 3144 }, { "epoch": 20.27309236947791, "grad_norm": 7.888002395629883, "learning_rate": 3.985483870967742e-05, "loss": 0.3183, "step": 3145 }, { "epoch": 20.279518072289157, "grad_norm": 17.962242126464844, "learning_rate": 3.985161290322581e-05, "loss": 0.5707, "step": 3146 }, { "epoch": 20.285943775100403, "grad_norm": 8.379111289978027, "learning_rate": 3.98483870967742e-05, "loss": 0.2818, "step": 3147 }, { "epoch": 20.292369477911645, "grad_norm": 8.636109352111816, "learning_rate": 3.984516129032258e-05, "loss": 0.439, "step": 3148 }, { "epoch": 20.29879518072289, "grad_norm": 13.755574226379395, "learning_rate": 3.984193548387097e-05, "loss": 0.3976, "step": 3149 }, { "epoch": 20.305220883534137, "grad_norm": 11.736943244934082, "learning_rate": 3.9838709677419356e-05, "loss": 0.4356, "step": 3150 }, { "epoch": 20.311646586345383, "grad_norm": 23.729312896728516, "learning_rate": 3.983548387096774e-05, "loss": 0.3571, "step": 3151 }, { "epoch": 20.318072289156625, "grad_norm": 9.236698150634766, "learning_rate": 3.9832258064516135e-05, "loss": 0.3778, "step": 3152 }, { "epoch": 20.32449799196787, "grad_norm": 11.911749839782715, "learning_rate": 3.982903225806452e-05, "loss": 0.5099, "step": 3153 }, { "epoch": 20.330923694779116, "grad_norm": 10.03878402709961, "learning_rate": 3.9825806451612906e-05, "loss": 0.4162, "step": 3154 }, { "epoch": 20.337349397590362, "grad_norm": 11.2142915725708, "learning_rate": 3.982258064516129e-05, "loss": 0.5801, "step": 3155 }, { "epoch": 20.343775100401608, "grad_norm": 14.825469970703125, "learning_rate": 3.981935483870968e-05, "loss": 0.3987, "step": 3156 }, { "epoch": 20.35020080321285, "grad_norm": 31.96288299560547, "learning_rate": 3.981612903225806e-05, "loss": 0.4055, "step": 3157 }, { "epoch": 20.356626506024096, "grad_norm": 2.6822640895843506, "learning_rate": 3.9812903225806456e-05, "loss": 0.199, "step": 3158 }, { "epoch": 20.36305220883534, "grad_norm": 6.083547115325928, "learning_rate": 3.980967741935484e-05, "loss": 0.1695, "step": 3159 }, { "epoch": 20.369477911646587, "grad_norm": 12.399429321289062, "learning_rate": 3.980645161290323e-05, "loss": 0.5073, "step": 3160 }, { "epoch": 20.37590361445783, "grad_norm": 36.87381362915039, "learning_rate": 3.980322580645162e-05, "loss": 0.575, "step": 3161 }, { "epoch": 20.382329317269075, "grad_norm": 2.689652442932129, "learning_rate": 3.9800000000000005e-05, "loss": 0.2021, "step": 3162 }, { "epoch": 20.38875502008032, "grad_norm": 6.741870880126953, "learning_rate": 3.979677419354839e-05, "loss": 0.226, "step": 3163 }, { "epoch": 20.395180722891567, "grad_norm": 9.699628829956055, "learning_rate": 3.9793548387096777e-05, "loss": 0.215, "step": 3164 }, { "epoch": 20.401606425702813, "grad_norm": 3.0290160179138184, "learning_rate": 3.979032258064516e-05, "loss": 0.1969, "step": 3165 }, { "epoch": 20.408032128514055, "grad_norm": 7.955510139465332, "learning_rate": 3.978709677419355e-05, "loss": 0.1832, "step": 3166 }, { "epoch": 20.4144578313253, "grad_norm": 6.786160469055176, "learning_rate": 3.978387096774194e-05, "loss": 0.4636, "step": 3167 }, { "epoch": 20.420883534136546, "grad_norm": 11.79191780090332, "learning_rate": 3.9780645161290326e-05, "loss": 0.4624, "step": 3168 }, { "epoch": 20.427309236947792, "grad_norm": 13.065023422241211, "learning_rate": 3.977741935483871e-05, "loss": 0.2984, "step": 3169 }, { "epoch": 20.433734939759034, "grad_norm": 7.67827033996582, "learning_rate": 3.97741935483871e-05, "loss": 0.4514, "step": 3170 }, { "epoch": 20.44016064257028, "grad_norm": 12.24150276184082, "learning_rate": 3.977096774193548e-05, "loss": 0.4399, "step": 3171 }, { "epoch": 20.446586345381526, "grad_norm": 19.87950897216797, "learning_rate": 3.976774193548387e-05, "loss": 0.3782, "step": 3172 }, { "epoch": 20.45301204819277, "grad_norm": 9.310997009277344, "learning_rate": 3.976451612903226e-05, "loss": 0.656, "step": 3173 }, { "epoch": 20.459437751004018, "grad_norm": 7.544458866119385, "learning_rate": 3.976129032258065e-05, "loss": 0.5182, "step": 3174 }, { "epoch": 20.46586345381526, "grad_norm": 36.32183837890625, "learning_rate": 3.975806451612903e-05, "loss": 0.4812, "step": 3175 }, { "epoch": 20.472289156626506, "grad_norm": 5.700336456298828, "learning_rate": 3.9754838709677425e-05, "loss": 0.3418, "step": 3176 }, { "epoch": 20.47871485943775, "grad_norm": 9.837760925292969, "learning_rate": 3.975161290322581e-05, "loss": 0.378, "step": 3177 }, { "epoch": 20.485140562248997, "grad_norm": 1.8975811004638672, "learning_rate": 3.97483870967742e-05, "loss": 0.3216, "step": 3178 }, { "epoch": 20.49156626506024, "grad_norm": 10.181573867797852, "learning_rate": 3.974516129032258e-05, "loss": 0.218, "step": 3179 }, { "epoch": 20.497991967871485, "grad_norm": 10.475750923156738, "learning_rate": 3.974193548387097e-05, "loss": 0.3338, "step": 3180 }, { "epoch": 20.50441767068273, "grad_norm": 11.8452730178833, "learning_rate": 3.9738709677419354e-05, "loss": 0.4702, "step": 3181 }, { "epoch": 20.510843373493977, "grad_norm": 4.0189127922058105, "learning_rate": 3.9735483870967746e-05, "loss": 0.3227, "step": 3182 }, { "epoch": 20.517269076305222, "grad_norm": 7.703409671783447, "learning_rate": 3.973225806451613e-05, "loss": 0.2433, "step": 3183 }, { "epoch": 20.523694779116465, "grad_norm": 15.483936309814453, "learning_rate": 3.972903225806452e-05, "loss": 0.6565, "step": 3184 }, { "epoch": 20.53012048192771, "grad_norm": 6.082330703735352, "learning_rate": 3.9725806451612904e-05, "loss": 0.4986, "step": 3185 }, { "epoch": 20.536546184738956, "grad_norm": 15.524839401245117, "learning_rate": 3.9722580645161296e-05, "loss": 0.4565, "step": 3186 }, { "epoch": 20.542971887550202, "grad_norm": 3.1867988109588623, "learning_rate": 3.9719354838709675e-05, "loss": 0.1367, "step": 3187 }, { "epoch": 20.549397590361444, "grad_norm": 9.939740180969238, "learning_rate": 3.971612903225807e-05, "loss": 0.6351, "step": 3188 }, { "epoch": 20.55582329317269, "grad_norm": 9.802570343017578, "learning_rate": 3.971290322580645e-05, "loss": 0.3548, "step": 3189 }, { "epoch": 20.562248995983936, "grad_norm": 4.829046249389648, "learning_rate": 3.970967741935484e-05, "loss": 0.2791, "step": 3190 }, { "epoch": 20.56867469879518, "grad_norm": 16.05450439453125, "learning_rate": 3.970645161290323e-05, "loss": 0.374, "step": 3191 }, { "epoch": 20.575100401606427, "grad_norm": 12.071246147155762, "learning_rate": 3.970322580645162e-05, "loss": 0.424, "step": 3192 }, { "epoch": 20.58152610441767, "grad_norm": 19.783517837524414, "learning_rate": 3.97e-05, "loss": 0.5598, "step": 3193 }, { "epoch": 20.587951807228915, "grad_norm": 8.180109024047852, "learning_rate": 3.969677419354839e-05, "loss": 0.2535, "step": 3194 }, { "epoch": 20.59437751004016, "grad_norm": 33.1153678894043, "learning_rate": 3.9693548387096774e-05, "loss": 0.4774, "step": 3195 }, { "epoch": 20.600803212851407, "grad_norm": 4.812952041625977, "learning_rate": 3.969032258064516e-05, "loss": 0.4185, "step": 3196 }, { "epoch": 20.60722891566265, "grad_norm": 18.447425842285156, "learning_rate": 3.968709677419355e-05, "loss": 0.357, "step": 3197 }, { "epoch": 20.613654618473895, "grad_norm": 14.401050567626953, "learning_rate": 3.968387096774194e-05, "loss": 0.3811, "step": 3198 }, { "epoch": 20.62008032128514, "grad_norm": 6.476546764373779, "learning_rate": 3.9680645161290324e-05, "loss": 0.5014, "step": 3199 }, { "epoch": 20.626506024096386, "grad_norm": 10.380890846252441, "learning_rate": 3.9677419354838716e-05, "loss": 0.2995, "step": 3200 }, { "epoch": 20.632931726907632, "grad_norm": 22.42681312561035, "learning_rate": 3.96741935483871e-05, "loss": 0.4742, "step": 3201 }, { "epoch": 20.639357429718874, "grad_norm": 7.827770233154297, "learning_rate": 3.967096774193548e-05, "loss": 0.2038, "step": 3202 }, { "epoch": 20.64578313253012, "grad_norm": 4.6466569900512695, "learning_rate": 3.966774193548387e-05, "loss": 0.2936, "step": 3203 }, { "epoch": 20.652208835341366, "grad_norm": 12.854907035827637, "learning_rate": 3.966451612903226e-05, "loss": 0.2704, "step": 3204 }, { "epoch": 20.65863453815261, "grad_norm": 12.438070297241211, "learning_rate": 3.9661290322580645e-05, "loss": 0.5586, "step": 3205 }, { "epoch": 20.665060240963854, "grad_norm": 7.0358781814575195, "learning_rate": 3.965806451612904e-05, "loss": 0.3823, "step": 3206 }, { "epoch": 20.6714859437751, "grad_norm": 19.132020950317383, "learning_rate": 3.965483870967742e-05, "loss": 0.3277, "step": 3207 }, { "epoch": 20.677911646586345, "grad_norm": 8.082879066467285, "learning_rate": 3.965161290322581e-05, "loss": 0.342, "step": 3208 }, { "epoch": 20.68433734939759, "grad_norm": 5.381679058074951, "learning_rate": 3.9648387096774194e-05, "loss": 0.2319, "step": 3209 }, { "epoch": 20.690763052208837, "grad_norm": 6.899646282196045, "learning_rate": 3.964516129032258e-05, "loss": 0.6255, "step": 3210 }, { "epoch": 20.69718875502008, "grad_norm": 15.97201919555664, "learning_rate": 3.9641935483870966e-05, "loss": 0.7228, "step": 3211 }, { "epoch": 20.703614457831325, "grad_norm": 15.321749687194824, "learning_rate": 3.963870967741936e-05, "loss": 0.3489, "step": 3212 }, { "epoch": 20.71004016064257, "grad_norm": 24.65039825439453, "learning_rate": 3.9635483870967744e-05, "loss": 0.4394, "step": 3213 }, { "epoch": 20.716465863453816, "grad_norm": 8.443490982055664, "learning_rate": 3.963225806451613e-05, "loss": 0.1958, "step": 3214 }, { "epoch": 20.72289156626506, "grad_norm": 17.284378051757812, "learning_rate": 3.962903225806452e-05, "loss": 0.7378, "step": 3215 }, { "epoch": 20.729317269076304, "grad_norm": 18.772438049316406, "learning_rate": 3.962580645161291e-05, "loss": 0.3827, "step": 3216 }, { "epoch": 20.73574297188755, "grad_norm": 16.5590763092041, "learning_rate": 3.9622580645161294e-05, "loss": 0.3918, "step": 3217 }, { "epoch": 20.742168674698796, "grad_norm": 20.500560760498047, "learning_rate": 3.961935483870968e-05, "loss": 0.6432, "step": 3218 }, { "epoch": 20.74859437751004, "grad_norm": 11.56241512298584, "learning_rate": 3.9616129032258065e-05, "loss": 0.4055, "step": 3219 }, { "epoch": 20.755020080321284, "grad_norm": 32.438838958740234, "learning_rate": 3.961290322580645e-05, "loss": 0.5646, "step": 3220 }, { "epoch": 20.76144578313253, "grad_norm": 33.5250129699707, "learning_rate": 3.960967741935484e-05, "loss": 0.3254, "step": 3221 }, { "epoch": 20.767871485943775, "grad_norm": 30.266979217529297, "learning_rate": 3.960645161290323e-05, "loss": 0.3501, "step": 3222 }, { "epoch": 20.77429718875502, "grad_norm": 21.388938903808594, "learning_rate": 3.9603225806451615e-05, "loss": 0.3684, "step": 3223 }, { "epoch": 20.780722891566263, "grad_norm": 14.910436630249023, "learning_rate": 3.960000000000001e-05, "loss": 0.3144, "step": 3224 }, { "epoch": 20.78714859437751, "grad_norm": 3.1050431728363037, "learning_rate": 3.959677419354839e-05, "loss": 0.3224, "step": 3225 }, { "epoch": 20.793574297188755, "grad_norm": 14.641000747680664, "learning_rate": 3.959354838709677e-05, "loss": 0.1992, "step": 3226 }, { "epoch": 20.8, "grad_norm": 23.94977378845215, "learning_rate": 3.9590322580645164e-05, "loss": 0.5735, "step": 3227 }, { "epoch": 20.806425702811246, "grad_norm": 36.59243392944336, "learning_rate": 3.958709677419355e-05, "loss": 0.6749, "step": 3228 }, { "epoch": 20.81285140562249, "grad_norm": 14.113014221191406, "learning_rate": 3.9583870967741936e-05, "loss": 0.3762, "step": 3229 }, { "epoch": 20.819277108433734, "grad_norm": 28.983781814575195, "learning_rate": 3.958064516129033e-05, "loss": 0.1982, "step": 3230 }, { "epoch": 20.82570281124498, "grad_norm": 19.159744262695312, "learning_rate": 3.9577419354838714e-05, "loss": 0.6134, "step": 3231 }, { "epoch": 20.832128514056226, "grad_norm": 8.015813827514648, "learning_rate": 3.95741935483871e-05, "loss": 0.2609, "step": 3232 }, { "epoch": 20.83855421686747, "grad_norm": 5.135195732116699, "learning_rate": 3.9570967741935485e-05, "loss": 0.4093, "step": 3233 }, { "epoch": 20.844979919678714, "grad_norm": 8.554088592529297, "learning_rate": 3.956774193548387e-05, "loss": 0.5348, "step": 3234 }, { "epoch": 20.85140562248996, "grad_norm": 15.320512771606445, "learning_rate": 3.956451612903226e-05, "loss": 0.4637, "step": 3235 }, { "epoch": 20.857831325301206, "grad_norm": 16.105722427368164, "learning_rate": 3.956129032258065e-05, "loss": 0.3868, "step": 3236 }, { "epoch": 20.86425702811245, "grad_norm": 27.695186614990234, "learning_rate": 3.9558064516129035e-05, "loss": 0.4978, "step": 3237 }, { "epoch": 20.870682730923694, "grad_norm": 11.595693588256836, "learning_rate": 3.955483870967742e-05, "loss": 0.4288, "step": 3238 }, { "epoch": 20.87710843373494, "grad_norm": 10.136857986450195, "learning_rate": 3.955161290322581e-05, "loss": 0.2726, "step": 3239 }, { "epoch": 20.883534136546185, "grad_norm": 21.229101181030273, "learning_rate": 3.95483870967742e-05, "loss": 0.2133, "step": 3240 }, { "epoch": 20.88995983935743, "grad_norm": 12.110089302062988, "learning_rate": 3.954516129032258e-05, "loss": 0.2849, "step": 3241 }, { "epoch": 20.896385542168673, "grad_norm": 9.551129341125488, "learning_rate": 3.954193548387097e-05, "loss": 0.4104, "step": 3242 }, { "epoch": 20.90281124497992, "grad_norm": 4.153226852416992, "learning_rate": 3.9538709677419356e-05, "loss": 0.2323, "step": 3243 }, { "epoch": 20.909236947791165, "grad_norm": 13.79879379272461, "learning_rate": 3.953548387096774e-05, "loss": 0.4712, "step": 3244 }, { "epoch": 20.91566265060241, "grad_norm": 23.4422607421875, "learning_rate": 3.9532258064516134e-05, "loss": 0.269, "step": 3245 }, { "epoch": 20.922088353413656, "grad_norm": 14.710312843322754, "learning_rate": 3.952903225806452e-05, "loss": 0.3447, "step": 3246 }, { "epoch": 20.9285140562249, "grad_norm": 19.513545989990234, "learning_rate": 3.9525806451612905e-05, "loss": 0.6498, "step": 3247 }, { "epoch": 20.934939759036144, "grad_norm": 19.593677520751953, "learning_rate": 3.952258064516129e-05, "loss": 0.4831, "step": 3248 }, { "epoch": 20.94136546184739, "grad_norm": 14.136308670043945, "learning_rate": 3.951935483870968e-05, "loss": 0.3394, "step": 3249 }, { "epoch": 20.947791164658636, "grad_norm": 11.361215591430664, "learning_rate": 3.951612903225806e-05, "loss": 0.4783, "step": 3250 }, { "epoch": 20.954216867469878, "grad_norm": 5.259056091308594, "learning_rate": 3.9512903225806455e-05, "loss": 0.3858, "step": 3251 }, { "epoch": 20.960642570281124, "grad_norm": 14.427826881408691, "learning_rate": 3.950967741935484e-05, "loss": 0.3518, "step": 3252 }, { "epoch": 20.96706827309237, "grad_norm": 13.938740730285645, "learning_rate": 3.9506451612903226e-05, "loss": 0.4073, "step": 3253 }, { "epoch": 20.973493975903615, "grad_norm": 15.128973007202148, "learning_rate": 3.950322580645162e-05, "loss": 0.3626, "step": 3254 }, { "epoch": 20.97991967871486, "grad_norm": 16.532655715942383, "learning_rate": 3.9500000000000005e-05, "loss": 0.5457, "step": 3255 }, { "epoch": 20.986345381526103, "grad_norm": 16.875240325927734, "learning_rate": 3.949677419354839e-05, "loss": 0.5545, "step": 3256 }, { "epoch": 20.99277108433735, "grad_norm": 24.185932159423828, "learning_rate": 3.9493548387096776e-05, "loss": 0.3303, "step": 3257 }, { "epoch": 21.002409638554216, "grad_norm": 15.55116081237793, "learning_rate": 3.949032258064516e-05, "loss": 0.3803, "step": 3258 }, { "epoch": 21.00883534136546, "grad_norm": 21.823287963867188, "learning_rate": 3.948709677419355e-05, "loss": 0.5549, "step": 3259 }, { "epoch": 21.015261044176707, "grad_norm": 54.658447265625, "learning_rate": 3.948387096774194e-05, "loss": 0.5932, "step": 3260 }, { "epoch": 21.021686746987953, "grad_norm": 26.676250457763672, "learning_rate": 3.9480645161290326e-05, "loss": 0.4787, "step": 3261 }, { "epoch": 21.028112449799195, "grad_norm": 12.337543487548828, "learning_rate": 3.947741935483871e-05, "loss": 0.2458, "step": 3262 }, { "epoch": 21.03453815261044, "grad_norm": 16.222564697265625, "learning_rate": 3.9474193548387104e-05, "loss": 0.533, "step": 3263 }, { "epoch": 21.040963855421687, "grad_norm": 5.2115936279296875, "learning_rate": 3.947096774193549e-05, "loss": 0.3839, "step": 3264 }, { "epoch": 21.047389558232933, "grad_norm": 6.702060222625732, "learning_rate": 3.946774193548387e-05, "loss": 0.209, "step": 3265 }, { "epoch": 21.05381526104418, "grad_norm": 9.972187042236328, "learning_rate": 3.946451612903226e-05, "loss": 0.6371, "step": 3266 }, { "epoch": 21.06024096385542, "grad_norm": 6.135876655578613, "learning_rate": 3.946129032258065e-05, "loss": 0.2182, "step": 3267 }, { "epoch": 21.066666666666666, "grad_norm": 8.519782066345215, "learning_rate": 3.945806451612903e-05, "loss": 0.464, "step": 3268 }, { "epoch": 21.073092369477912, "grad_norm": 17.259891510009766, "learning_rate": 3.9454838709677425e-05, "loss": 0.3177, "step": 3269 }, { "epoch": 21.079518072289158, "grad_norm": 23.043458938598633, "learning_rate": 3.945161290322581e-05, "loss": 0.5963, "step": 3270 }, { "epoch": 21.0859437751004, "grad_norm": 10.011810302734375, "learning_rate": 3.9448387096774196e-05, "loss": 0.3299, "step": 3271 }, { "epoch": 21.092369477911646, "grad_norm": 27.97663688659668, "learning_rate": 3.944516129032258e-05, "loss": 0.2143, "step": 3272 }, { "epoch": 21.09879518072289, "grad_norm": 24.731510162353516, "learning_rate": 3.944193548387097e-05, "loss": 0.4349, "step": 3273 }, { "epoch": 21.105220883534137, "grad_norm": 10.268356323242188, "learning_rate": 3.9438709677419353e-05, "loss": 0.4165, "step": 3274 }, { "epoch": 21.111646586345383, "grad_norm": 12.688679695129395, "learning_rate": 3.9435483870967746e-05, "loss": 0.2686, "step": 3275 }, { "epoch": 21.118072289156625, "grad_norm": 9.647183418273926, "learning_rate": 3.943225806451613e-05, "loss": 0.4824, "step": 3276 }, { "epoch": 21.12449799196787, "grad_norm": 12.882625579833984, "learning_rate": 3.942903225806452e-05, "loss": 0.4981, "step": 3277 }, { "epoch": 21.130923694779117, "grad_norm": 15.819168090820312, "learning_rate": 3.942580645161291e-05, "loss": 0.4431, "step": 3278 }, { "epoch": 21.137349397590363, "grad_norm": 11.707969665527344, "learning_rate": 3.9422580645161296e-05, "loss": 0.5502, "step": 3279 }, { "epoch": 21.143775100401605, "grad_norm": 10.413691520690918, "learning_rate": 3.9419354838709674e-05, "loss": 0.3308, "step": 3280 }, { "epoch": 21.15020080321285, "grad_norm": 18.008195877075195, "learning_rate": 3.941612903225807e-05, "loss": 0.2864, "step": 3281 }, { "epoch": 21.156626506024097, "grad_norm": 4.77959680557251, "learning_rate": 3.941290322580645e-05, "loss": 0.1482, "step": 3282 }, { "epoch": 21.163052208835342, "grad_norm": 7.313478469848633, "learning_rate": 3.940967741935484e-05, "loss": 0.4195, "step": 3283 }, { "epoch": 21.169477911646588, "grad_norm": 14.01305866241455, "learning_rate": 3.940645161290323e-05, "loss": 0.4124, "step": 3284 }, { "epoch": 21.17590361445783, "grad_norm": 13.594583511352539, "learning_rate": 3.9403225806451617e-05, "loss": 0.3719, "step": 3285 }, { "epoch": 21.182329317269076, "grad_norm": 11.037776947021484, "learning_rate": 3.94e-05, "loss": 0.3767, "step": 3286 }, { "epoch": 21.188755020080322, "grad_norm": 7.8656463623046875, "learning_rate": 3.9396774193548395e-05, "loss": 0.3148, "step": 3287 }, { "epoch": 21.195180722891568, "grad_norm": 15.411916732788086, "learning_rate": 3.9393548387096774e-05, "loss": 0.4014, "step": 3288 }, { "epoch": 21.20160642570281, "grad_norm": 33.99292755126953, "learning_rate": 3.939032258064516e-05, "loss": 0.5029, "step": 3289 }, { "epoch": 21.208032128514056, "grad_norm": 17.86922264099121, "learning_rate": 3.938709677419355e-05, "loss": 0.3073, "step": 3290 }, { "epoch": 21.2144578313253, "grad_norm": 7.113110065460205, "learning_rate": 3.938387096774194e-05, "loss": 0.4192, "step": 3291 }, { "epoch": 21.220883534136547, "grad_norm": 11.09314250946045, "learning_rate": 3.938064516129032e-05, "loss": 0.5976, "step": 3292 }, { "epoch": 21.227309236947793, "grad_norm": 25.796661376953125, "learning_rate": 3.9377419354838716e-05, "loss": 0.4927, "step": 3293 }, { "epoch": 21.233734939759035, "grad_norm": 17.291580200195312, "learning_rate": 3.93741935483871e-05, "loss": 0.5121, "step": 3294 }, { "epoch": 21.24016064257028, "grad_norm": 7.275795936584473, "learning_rate": 3.937096774193549e-05, "loss": 0.3744, "step": 3295 }, { "epoch": 21.246586345381527, "grad_norm": 12.245406150817871, "learning_rate": 3.936774193548387e-05, "loss": 0.5045, "step": 3296 }, { "epoch": 21.253012048192772, "grad_norm": 11.148791313171387, "learning_rate": 3.936451612903226e-05, "loss": 0.2962, "step": 3297 }, { "epoch": 21.259437751004015, "grad_norm": 34.844635009765625, "learning_rate": 3.9361290322580644e-05, "loss": 0.7065, "step": 3298 }, { "epoch": 21.26586345381526, "grad_norm": 20.344532012939453, "learning_rate": 3.935806451612904e-05, "loss": 0.576, "step": 3299 }, { "epoch": 21.272289156626506, "grad_norm": 14.511658668518066, "learning_rate": 3.935483870967742e-05, "loss": 0.677, "step": 3300 }, { "epoch": 21.278714859437752, "grad_norm": 11.593645095825195, "learning_rate": 3.935161290322581e-05, "loss": 0.5155, "step": 3301 }, { "epoch": 21.285140562248998, "grad_norm": 13.156426429748535, "learning_rate": 3.93483870967742e-05, "loss": 0.2925, "step": 3302 }, { "epoch": 21.29156626506024, "grad_norm": 8.785014152526855, "learning_rate": 3.9345161290322586e-05, "loss": 0.2631, "step": 3303 }, { "epoch": 21.297991967871486, "grad_norm": 6.294126510620117, "learning_rate": 3.9341935483870965e-05, "loss": 0.1933, "step": 3304 }, { "epoch": 21.30441767068273, "grad_norm": 10.330317497253418, "learning_rate": 3.933870967741936e-05, "loss": 0.409, "step": 3305 }, { "epoch": 21.310843373493977, "grad_norm": 16.883766174316406, "learning_rate": 3.9335483870967743e-05, "loss": 0.3994, "step": 3306 }, { "epoch": 21.31726907630522, "grad_norm": 9.949392318725586, "learning_rate": 3.933225806451613e-05, "loss": 0.3405, "step": 3307 }, { "epoch": 21.323694779116465, "grad_norm": 10.716243743896484, "learning_rate": 3.932903225806452e-05, "loss": 0.2331, "step": 3308 }, { "epoch": 21.33012048192771, "grad_norm": 4.820130825042725, "learning_rate": 3.932580645161291e-05, "loss": 0.1363, "step": 3309 }, { "epoch": 21.336546184738957, "grad_norm": 11.722029685974121, "learning_rate": 3.932258064516129e-05, "loss": 0.4026, "step": 3310 }, { "epoch": 21.3429718875502, "grad_norm": 12.37242603302002, "learning_rate": 3.931935483870968e-05, "loss": 0.5407, "step": 3311 }, { "epoch": 21.349397590361445, "grad_norm": 9.049507141113281, "learning_rate": 3.9316129032258064e-05, "loss": 0.3827, "step": 3312 }, { "epoch": 21.35582329317269, "grad_norm": 12.354798316955566, "learning_rate": 3.931290322580645e-05, "loss": 0.2882, "step": 3313 }, { "epoch": 21.362248995983936, "grad_norm": 7.69230318069458, "learning_rate": 3.930967741935484e-05, "loss": 0.4534, "step": 3314 }, { "epoch": 21.368674698795182, "grad_norm": 15.46497631072998, "learning_rate": 3.930645161290323e-05, "loss": 0.2436, "step": 3315 }, { "epoch": 21.375100401606424, "grad_norm": 19.562610626220703, "learning_rate": 3.9303225806451614e-05, "loss": 0.3063, "step": 3316 }, { "epoch": 21.38152610441767, "grad_norm": 7.918242454528809, "learning_rate": 3.9300000000000007e-05, "loss": 0.4863, "step": 3317 }, { "epoch": 21.387951807228916, "grad_norm": 3.099595546722412, "learning_rate": 3.929677419354839e-05, "loss": 0.1856, "step": 3318 }, { "epoch": 21.39437751004016, "grad_norm": 11.68242073059082, "learning_rate": 3.929354838709677e-05, "loss": 0.534, "step": 3319 }, { "epoch": 21.400803212851407, "grad_norm": 8.17670726776123, "learning_rate": 3.9290322580645164e-05, "loss": 0.3595, "step": 3320 }, { "epoch": 21.40722891566265, "grad_norm": 10.884778022766113, "learning_rate": 3.928709677419355e-05, "loss": 0.4025, "step": 3321 }, { "epoch": 21.413654618473895, "grad_norm": 23.127758026123047, "learning_rate": 3.9283870967741935e-05, "loss": 0.4076, "step": 3322 }, { "epoch": 21.42008032128514, "grad_norm": 14.900663375854492, "learning_rate": 3.928064516129033e-05, "loss": 0.5984, "step": 3323 }, { "epoch": 21.426506024096387, "grad_norm": 8.257502555847168, "learning_rate": 3.927741935483871e-05, "loss": 0.4886, "step": 3324 }, { "epoch": 21.43293172690763, "grad_norm": 9.801166534423828, "learning_rate": 3.92741935483871e-05, "loss": 0.4113, "step": 3325 }, { "epoch": 21.439357429718875, "grad_norm": 5.036794662475586, "learning_rate": 3.927096774193549e-05, "loss": 0.1445, "step": 3326 }, { "epoch": 21.44578313253012, "grad_norm": 9.33691120147705, "learning_rate": 3.926774193548387e-05, "loss": 0.3644, "step": 3327 }, { "epoch": 21.452208835341366, "grad_norm": 4.357815265655518, "learning_rate": 3.9264516129032256e-05, "loss": 0.3611, "step": 3328 }, { "epoch": 21.45863453815261, "grad_norm": 7.499757766723633, "learning_rate": 3.926129032258065e-05, "loss": 0.3821, "step": 3329 }, { "epoch": 21.465060240963854, "grad_norm": 47.06676483154297, "learning_rate": 3.9258064516129034e-05, "loss": 0.8164, "step": 3330 }, { "epoch": 21.4714859437751, "grad_norm": 16.68478775024414, "learning_rate": 3.925483870967742e-05, "loss": 0.307, "step": 3331 }, { "epoch": 21.477911646586346, "grad_norm": 19.115440368652344, "learning_rate": 3.925161290322581e-05, "loss": 0.569, "step": 3332 }, { "epoch": 21.48433734939759, "grad_norm": 8.344329833984375, "learning_rate": 3.92483870967742e-05, "loss": 0.3677, "step": 3333 }, { "epoch": 21.490763052208834, "grad_norm": 7.911175727844238, "learning_rate": 3.9245161290322584e-05, "loss": 0.204, "step": 3334 }, { "epoch": 21.49718875502008, "grad_norm": 11.656858444213867, "learning_rate": 3.924193548387097e-05, "loss": 0.2073, "step": 3335 }, { "epoch": 21.503614457831326, "grad_norm": 8.326302528381348, "learning_rate": 3.9238709677419355e-05, "loss": 0.3366, "step": 3336 }, { "epoch": 21.51004016064257, "grad_norm": 14.44531536102295, "learning_rate": 3.923548387096774e-05, "loss": 0.3799, "step": 3337 }, { "epoch": 21.516465863453817, "grad_norm": 11.358368873596191, "learning_rate": 3.9232258064516134e-05, "loss": 0.2787, "step": 3338 }, { "epoch": 21.52289156626506, "grad_norm": 13.594155311584473, "learning_rate": 3.922903225806452e-05, "loss": 0.2957, "step": 3339 }, { "epoch": 21.529317269076305, "grad_norm": 13.143461227416992, "learning_rate": 3.9225806451612905e-05, "loss": 0.4152, "step": 3340 }, { "epoch": 21.53574297188755, "grad_norm": 6.313676357269287, "learning_rate": 3.92225806451613e-05, "loss": 0.239, "step": 3341 }, { "epoch": 21.542168674698797, "grad_norm": 47.04666519165039, "learning_rate": 3.9219354838709676e-05, "loss": 0.4113, "step": 3342 }, { "epoch": 21.54859437751004, "grad_norm": 24.10997200012207, "learning_rate": 3.921612903225806e-05, "loss": 0.3282, "step": 3343 }, { "epoch": 21.555020080321285, "grad_norm": 20.13530158996582, "learning_rate": 3.9212903225806455e-05, "loss": 0.6643, "step": 3344 }, { "epoch": 21.56144578313253, "grad_norm": 13.398016929626465, "learning_rate": 3.920967741935484e-05, "loss": 0.2293, "step": 3345 }, { "epoch": 21.567871485943776, "grad_norm": 14.050085067749023, "learning_rate": 3.9206451612903226e-05, "loss": 0.2153, "step": 3346 }, { "epoch": 21.57429718875502, "grad_norm": 13.915634155273438, "learning_rate": 3.920322580645162e-05, "loss": 0.5198, "step": 3347 }, { "epoch": 21.580722891566264, "grad_norm": 2.7417540550231934, "learning_rate": 3.9200000000000004e-05, "loss": 0.2434, "step": 3348 }, { "epoch": 21.58714859437751, "grad_norm": 5.685986042022705, "learning_rate": 3.919677419354839e-05, "loss": 0.3019, "step": 3349 }, { "epoch": 21.593574297188756, "grad_norm": 9.099749565124512, "learning_rate": 3.9193548387096776e-05, "loss": 0.4589, "step": 3350 }, { "epoch": 21.6, "grad_norm": 17.122713088989258, "learning_rate": 3.919032258064516e-05, "loss": 0.3427, "step": 3351 }, { "epoch": 21.606425702811244, "grad_norm": 6.848479747772217, "learning_rate": 3.918709677419355e-05, "loss": 0.3931, "step": 3352 }, { "epoch": 21.61285140562249, "grad_norm": 17.867021560668945, "learning_rate": 3.918387096774194e-05, "loss": 0.3961, "step": 3353 }, { "epoch": 21.619277108433735, "grad_norm": 9.010595321655273, "learning_rate": 3.9180645161290325e-05, "loss": 0.4247, "step": 3354 }, { "epoch": 21.62570281124498, "grad_norm": 9.195846557617188, "learning_rate": 3.917741935483871e-05, "loss": 0.3943, "step": 3355 }, { "epoch": 21.632128514056227, "grad_norm": 6.210819721221924, "learning_rate": 3.91741935483871e-05, "loss": 0.3093, "step": 3356 }, { "epoch": 21.63855421686747, "grad_norm": 25.195466995239258, "learning_rate": 3.917096774193549e-05, "loss": 0.3157, "step": 3357 }, { "epoch": 21.644979919678715, "grad_norm": 17.564594268798828, "learning_rate": 3.916774193548387e-05, "loss": 0.5094, "step": 3358 }, { "epoch": 21.65140562248996, "grad_norm": 15.166448593139648, "learning_rate": 3.916451612903226e-05, "loss": 0.2235, "step": 3359 }, { "epoch": 21.657831325301206, "grad_norm": 7.728507041931152, "learning_rate": 3.9161290322580646e-05, "loss": 0.4356, "step": 3360 }, { "epoch": 21.66425702811245, "grad_norm": 29.268306732177734, "learning_rate": 3.915806451612903e-05, "loss": 0.5545, "step": 3361 }, { "epoch": 21.670682730923694, "grad_norm": 21.426301956176758, "learning_rate": 3.9154838709677424e-05, "loss": 0.4543, "step": 3362 }, { "epoch": 21.67710843373494, "grad_norm": 24.110851287841797, "learning_rate": 3.915161290322581e-05, "loss": 0.4419, "step": 3363 }, { "epoch": 21.683534136546186, "grad_norm": 7.715743541717529, "learning_rate": 3.9148387096774196e-05, "loss": 0.2166, "step": 3364 }, { "epoch": 21.689959839357428, "grad_norm": 13.642264366149902, "learning_rate": 3.914516129032259e-05, "loss": 0.4744, "step": 3365 }, { "epoch": 21.696385542168674, "grad_norm": 19.471742630004883, "learning_rate": 3.914193548387097e-05, "loss": 0.2848, "step": 3366 }, { "epoch": 21.70281124497992, "grad_norm": 12.41714096069336, "learning_rate": 3.913870967741935e-05, "loss": 0.3466, "step": 3367 }, { "epoch": 21.709236947791165, "grad_norm": 9.602289199829102, "learning_rate": 3.9135483870967745e-05, "loss": 0.3693, "step": 3368 }, { "epoch": 21.71566265060241, "grad_norm": 5.038344383239746, "learning_rate": 3.913225806451613e-05, "loss": 0.2269, "step": 3369 }, { "epoch": 21.722088353413653, "grad_norm": 21.80316925048828, "learning_rate": 3.912903225806452e-05, "loss": 0.5808, "step": 3370 }, { "epoch": 21.7285140562249, "grad_norm": 21.82911491394043, "learning_rate": 3.912580645161291e-05, "loss": 0.5409, "step": 3371 }, { "epoch": 21.734939759036145, "grad_norm": 4.171671390533447, "learning_rate": 3.9122580645161295e-05, "loss": 0.4444, "step": 3372 }, { "epoch": 21.74136546184739, "grad_norm": 25.97250747680664, "learning_rate": 3.911935483870968e-05, "loss": 0.3596, "step": 3373 }, { "epoch": 21.747791164658633, "grad_norm": 13.385181427001953, "learning_rate": 3.9116129032258066e-05, "loss": 0.4788, "step": 3374 }, { "epoch": 21.75421686746988, "grad_norm": 43.726722717285156, "learning_rate": 3.911290322580645e-05, "loss": 0.2796, "step": 3375 }, { "epoch": 21.760642570281124, "grad_norm": 6.620734214782715, "learning_rate": 3.910967741935484e-05, "loss": 0.3649, "step": 3376 }, { "epoch": 21.76706827309237, "grad_norm": 4.311135768890381, "learning_rate": 3.910645161290323e-05, "loss": 0.3444, "step": 3377 }, { "epoch": 21.773493975903616, "grad_norm": 8.818968772888184, "learning_rate": 3.9103225806451616e-05, "loss": 0.2762, "step": 3378 }, { "epoch": 21.779919678714858, "grad_norm": 3.388427972793579, "learning_rate": 3.91e-05, "loss": 0.5292, "step": 3379 }, { "epoch": 21.786345381526104, "grad_norm": 7.089523792266846, "learning_rate": 3.9096774193548394e-05, "loss": 0.22, "step": 3380 }, { "epoch": 21.79277108433735, "grad_norm": 16.205833435058594, "learning_rate": 3.909354838709677e-05, "loss": 0.2447, "step": 3381 }, { "epoch": 21.799196787148595, "grad_norm": 6.807889461517334, "learning_rate": 3.909032258064516e-05, "loss": 0.4568, "step": 3382 }, { "epoch": 21.805622489959838, "grad_norm": 21.874935150146484, "learning_rate": 3.908709677419355e-05, "loss": 0.4977, "step": 3383 }, { "epoch": 21.812048192771083, "grad_norm": 4.3249592781066895, "learning_rate": 3.908387096774194e-05, "loss": 0.2194, "step": 3384 }, { "epoch": 21.81847389558233, "grad_norm": 19.86972999572754, "learning_rate": 3.908064516129032e-05, "loss": 0.6191, "step": 3385 }, { "epoch": 21.824899598393575, "grad_norm": 4.430800914764404, "learning_rate": 3.9077419354838715e-05, "loss": 0.2153, "step": 3386 }, { "epoch": 21.83132530120482, "grad_norm": 3.969601631164551, "learning_rate": 3.90741935483871e-05, "loss": 0.3152, "step": 3387 }, { "epoch": 21.837751004016063, "grad_norm": 19.294036865234375, "learning_rate": 3.907096774193549e-05, "loss": 0.3774, "step": 3388 }, { "epoch": 21.84417670682731, "grad_norm": 28.42940902709961, "learning_rate": 3.906774193548387e-05, "loss": 0.5257, "step": 3389 }, { "epoch": 21.850602409638554, "grad_norm": 24.329242706298828, "learning_rate": 3.906451612903226e-05, "loss": 0.2794, "step": 3390 }, { "epoch": 21.8570281124498, "grad_norm": 17.449298858642578, "learning_rate": 3.9061290322580644e-05, "loss": 0.4941, "step": 3391 }, { "epoch": 21.863453815261042, "grad_norm": 17.397005081176758, "learning_rate": 3.9058064516129036e-05, "loss": 0.6267, "step": 3392 }, { "epoch": 21.86987951807229, "grad_norm": 11.253728866577148, "learning_rate": 3.905483870967742e-05, "loss": 0.1836, "step": 3393 }, { "epoch": 21.876305220883534, "grad_norm": 6.964405536651611, "learning_rate": 3.905161290322581e-05, "loss": 0.3164, "step": 3394 }, { "epoch": 21.88273092369478, "grad_norm": 19.57056427001953, "learning_rate": 3.90483870967742e-05, "loss": 0.6021, "step": 3395 }, { "epoch": 21.889156626506026, "grad_norm": 12.770813941955566, "learning_rate": 3.9045161290322586e-05, "loss": 0.3764, "step": 3396 }, { "epoch": 21.895582329317268, "grad_norm": 19.942670822143555, "learning_rate": 3.9041935483870965e-05, "loss": 0.3976, "step": 3397 }, { "epoch": 21.902008032128514, "grad_norm": 5.759477615356445, "learning_rate": 3.903870967741936e-05, "loss": 0.2357, "step": 3398 }, { "epoch": 21.90843373493976, "grad_norm": 16.37567901611328, "learning_rate": 3.903548387096774e-05, "loss": 0.5994, "step": 3399 }, { "epoch": 21.914859437751005, "grad_norm": 16.80613136291504, "learning_rate": 3.903225806451613e-05, "loss": 0.5798, "step": 3400 }, { "epoch": 21.921285140562247, "grad_norm": 5.196872711181641, "learning_rate": 3.902903225806452e-05, "loss": 0.3487, "step": 3401 }, { "epoch": 21.927710843373493, "grad_norm": 14.86912727355957, "learning_rate": 3.902580645161291e-05, "loss": 0.3688, "step": 3402 }, { "epoch": 21.93413654618474, "grad_norm": 9.324958801269531, "learning_rate": 3.902258064516129e-05, "loss": 0.491, "step": 3403 }, { "epoch": 21.940562248995985, "grad_norm": 13.839170455932617, "learning_rate": 3.9019354838709685e-05, "loss": 0.7227, "step": 3404 }, { "epoch": 21.94698795180723, "grad_norm": 9.382929801940918, "learning_rate": 3.9016129032258064e-05, "loss": 0.2577, "step": 3405 }, { "epoch": 21.953413654618473, "grad_norm": 16.627300262451172, "learning_rate": 3.901290322580645e-05, "loss": 0.4302, "step": 3406 }, { "epoch": 21.95983935742972, "grad_norm": 51.25236511230469, "learning_rate": 3.900967741935484e-05, "loss": 0.4239, "step": 3407 }, { "epoch": 21.966265060240964, "grad_norm": 9.315829277038574, "learning_rate": 3.900645161290323e-05, "loss": 0.2884, "step": 3408 }, { "epoch": 21.97269076305221, "grad_norm": 14.209221839904785, "learning_rate": 3.9003225806451614e-05, "loss": 0.3958, "step": 3409 }, { "epoch": 21.979116465863452, "grad_norm": 11.577492713928223, "learning_rate": 3.9000000000000006e-05, "loss": 0.572, "step": 3410 }, { "epoch": 21.985542168674698, "grad_norm": 20.547321319580078, "learning_rate": 3.899677419354839e-05, "loss": 0.6025, "step": 3411 }, { "epoch": 21.991967871485944, "grad_norm": 17.687421798706055, "learning_rate": 3.899354838709678e-05, "loss": 0.384, "step": 3412 }, { "epoch": 22.00160642570281, "grad_norm": 11.912986755371094, "learning_rate": 3.899032258064516e-05, "loss": 0.4779, "step": 3413 }, { "epoch": 22.008032128514056, "grad_norm": 9.149157524108887, "learning_rate": 3.898709677419355e-05, "loss": 0.3874, "step": 3414 }, { "epoch": 22.014457831325302, "grad_norm": 10.853452682495117, "learning_rate": 3.8983870967741935e-05, "loss": 0.2772, "step": 3415 }, { "epoch": 22.020883534136548, "grad_norm": 6.5138840675354, "learning_rate": 3.898064516129033e-05, "loss": 0.1556, "step": 3416 }, { "epoch": 22.02730923694779, "grad_norm": 18.643423080444336, "learning_rate": 3.897741935483871e-05, "loss": 0.4128, "step": 3417 }, { "epoch": 22.033734939759036, "grad_norm": 19.123737335205078, "learning_rate": 3.89741935483871e-05, "loss": 0.5021, "step": 3418 }, { "epoch": 22.04016064257028, "grad_norm": 12.558156967163086, "learning_rate": 3.897096774193549e-05, "loss": 0.4884, "step": 3419 }, { "epoch": 22.046586345381527, "grad_norm": 6.652852535247803, "learning_rate": 3.896774193548387e-05, "loss": 0.228, "step": 3420 }, { "epoch": 22.05301204819277, "grad_norm": 19.06941032409668, "learning_rate": 3.8964516129032256e-05, "loss": 0.3262, "step": 3421 }, { "epoch": 22.059437751004015, "grad_norm": 17.685291290283203, "learning_rate": 3.896129032258065e-05, "loss": 0.5774, "step": 3422 }, { "epoch": 22.06586345381526, "grad_norm": 6.405605792999268, "learning_rate": 3.8958064516129034e-05, "loss": 0.1601, "step": 3423 }, { "epoch": 22.072289156626507, "grad_norm": 13.19540023803711, "learning_rate": 3.895483870967742e-05, "loss": 0.1704, "step": 3424 }, { "epoch": 22.078714859437753, "grad_norm": 10.326749801635742, "learning_rate": 3.895161290322581e-05, "loss": 0.3728, "step": 3425 }, { "epoch": 22.085140562248995, "grad_norm": 15.558708190917969, "learning_rate": 3.89483870967742e-05, "loss": 0.4904, "step": 3426 }, { "epoch": 22.09156626506024, "grad_norm": 25.93842315673828, "learning_rate": 3.8945161290322583e-05, "loss": 0.5984, "step": 3427 }, { "epoch": 22.097991967871486, "grad_norm": 11.106996536254883, "learning_rate": 3.894193548387097e-05, "loss": 0.3429, "step": 3428 }, { "epoch": 22.104417670682732, "grad_norm": 11.699763298034668, "learning_rate": 3.8938709677419355e-05, "loss": 0.4331, "step": 3429 }, { "epoch": 22.110843373493974, "grad_norm": 10.913141250610352, "learning_rate": 3.893548387096774e-05, "loss": 0.4258, "step": 3430 }, { "epoch": 22.11726907630522, "grad_norm": 7.945165634155273, "learning_rate": 3.893225806451613e-05, "loss": 0.4856, "step": 3431 }, { "epoch": 22.123694779116466, "grad_norm": 16.148517608642578, "learning_rate": 3.892903225806452e-05, "loss": 0.3438, "step": 3432 }, { "epoch": 22.13012048192771, "grad_norm": 4.483079433441162, "learning_rate": 3.8925806451612904e-05, "loss": 0.2704, "step": 3433 }, { "epoch": 22.136546184738958, "grad_norm": 8.194324493408203, "learning_rate": 3.89225806451613e-05, "loss": 0.4708, "step": 3434 }, { "epoch": 22.1429718875502, "grad_norm": 9.038599967956543, "learning_rate": 3.891935483870968e-05, "loss": 0.4025, "step": 3435 }, { "epoch": 22.149397590361446, "grad_norm": 7.526703357696533, "learning_rate": 3.891612903225806e-05, "loss": 0.1687, "step": 3436 }, { "epoch": 22.15582329317269, "grad_norm": 33.856163024902344, "learning_rate": 3.8912903225806454e-05, "loss": 0.5678, "step": 3437 }, { "epoch": 22.162248995983937, "grad_norm": 10.747069358825684, "learning_rate": 3.890967741935484e-05, "loss": 0.3986, "step": 3438 }, { "epoch": 22.16867469879518, "grad_norm": 29.67582893371582, "learning_rate": 3.8906451612903225e-05, "loss": 0.3873, "step": 3439 }, { "epoch": 22.175100401606425, "grad_norm": 28.754898071289062, "learning_rate": 3.890322580645162e-05, "loss": 0.2293, "step": 3440 }, { "epoch": 22.18152610441767, "grad_norm": 10.808884620666504, "learning_rate": 3.8900000000000004e-05, "loss": 0.7567, "step": 3441 }, { "epoch": 22.187951807228917, "grad_norm": 5.943881988525391, "learning_rate": 3.889677419354839e-05, "loss": 0.2433, "step": 3442 }, { "epoch": 22.194377510040162, "grad_norm": 7.576895713806152, "learning_rate": 3.889354838709678e-05, "loss": 0.4944, "step": 3443 }, { "epoch": 22.200803212851405, "grad_norm": 5.386629581451416, "learning_rate": 3.889032258064516e-05, "loss": 0.2499, "step": 3444 }, { "epoch": 22.20722891566265, "grad_norm": 13.450294494628906, "learning_rate": 3.8887096774193546e-05, "loss": 0.5559, "step": 3445 }, { "epoch": 22.213654618473896, "grad_norm": 6.777096271514893, "learning_rate": 3.888387096774194e-05, "loss": 0.3894, "step": 3446 }, { "epoch": 22.220080321285142, "grad_norm": 10.1998929977417, "learning_rate": 3.8880645161290325e-05, "loss": 0.3494, "step": 3447 }, { "epoch": 22.226506024096384, "grad_norm": 10.223339080810547, "learning_rate": 3.887741935483871e-05, "loss": 0.2714, "step": 3448 }, { "epoch": 22.23293172690763, "grad_norm": 16.35748291015625, "learning_rate": 3.88741935483871e-05, "loss": 0.4446, "step": 3449 }, { "epoch": 22.239357429718876, "grad_norm": 24.721927642822266, "learning_rate": 3.887096774193549e-05, "loss": 0.5488, "step": 3450 }, { "epoch": 22.24578313253012, "grad_norm": 15.108840942382812, "learning_rate": 3.886774193548387e-05, "loss": 0.3446, "step": 3451 }, { "epoch": 22.252208835341367, "grad_norm": 24.03217887878418, "learning_rate": 3.886451612903226e-05, "loss": 0.1959, "step": 3452 }, { "epoch": 22.25863453815261, "grad_norm": 10.955098152160645, "learning_rate": 3.8861290322580646e-05, "loss": 0.5988, "step": 3453 }, { "epoch": 22.265060240963855, "grad_norm": 41.987022399902344, "learning_rate": 3.885806451612903e-05, "loss": 0.2296, "step": 3454 }, { "epoch": 22.2714859437751, "grad_norm": 8.643657684326172, "learning_rate": 3.8854838709677424e-05, "loss": 0.4065, "step": 3455 }, { "epoch": 22.277911646586347, "grad_norm": 8.820880889892578, "learning_rate": 3.885161290322581e-05, "loss": 0.4689, "step": 3456 }, { "epoch": 22.28433734939759, "grad_norm": 8.342575073242188, "learning_rate": 3.8848387096774195e-05, "loss": 0.4387, "step": 3457 }, { "epoch": 22.290763052208835, "grad_norm": 18.150728225708008, "learning_rate": 3.884516129032259e-05, "loss": 0.4228, "step": 3458 }, { "epoch": 22.29718875502008, "grad_norm": 6.932944297790527, "learning_rate": 3.884193548387097e-05, "loss": 0.3635, "step": 3459 }, { "epoch": 22.303614457831326, "grad_norm": 9.953007698059082, "learning_rate": 3.883870967741935e-05, "loss": 0.4326, "step": 3460 }, { "epoch": 22.310040160642572, "grad_norm": 22.949935913085938, "learning_rate": 3.8835483870967745e-05, "loss": 0.7989, "step": 3461 }, { "epoch": 22.316465863453814, "grad_norm": 13.469207763671875, "learning_rate": 3.883225806451613e-05, "loss": 0.1491, "step": 3462 }, { "epoch": 22.32289156626506, "grad_norm": 18.96015167236328, "learning_rate": 3.8829032258064516e-05, "loss": 0.4524, "step": 3463 }, { "epoch": 22.329317269076306, "grad_norm": 28.904592514038086, "learning_rate": 3.882580645161291e-05, "loss": 0.5925, "step": 3464 }, { "epoch": 22.33574297188755, "grad_norm": 16.10734748840332, "learning_rate": 3.8822580645161294e-05, "loss": 0.6168, "step": 3465 }, { "epoch": 22.342168674698794, "grad_norm": 13.698968887329102, "learning_rate": 3.881935483870968e-05, "loss": 0.2694, "step": 3466 }, { "epoch": 22.34859437751004, "grad_norm": 5.495956897735596, "learning_rate": 3.8816129032258066e-05, "loss": 0.2675, "step": 3467 }, { "epoch": 22.355020080321285, "grad_norm": 4.606384754180908, "learning_rate": 3.881290322580645e-05, "loss": 0.2806, "step": 3468 }, { "epoch": 22.36144578313253, "grad_norm": 20.065814971923828, "learning_rate": 3.880967741935484e-05, "loss": 0.6829, "step": 3469 }, { "epoch": 22.367871485943777, "grad_norm": 16.055360794067383, "learning_rate": 3.880645161290323e-05, "loss": 0.3837, "step": 3470 }, { "epoch": 22.37429718875502, "grad_norm": 74.27485656738281, "learning_rate": 3.8803225806451616e-05, "loss": 0.4075, "step": 3471 }, { "epoch": 22.380722891566265, "grad_norm": 10.18300724029541, "learning_rate": 3.88e-05, "loss": 0.5108, "step": 3472 }, { "epoch": 22.38714859437751, "grad_norm": 9.254419326782227, "learning_rate": 3.8796774193548394e-05, "loss": 0.6468, "step": 3473 }, { "epoch": 22.393574297188756, "grad_norm": 6.668877601623535, "learning_rate": 3.879354838709678e-05, "loss": 0.4552, "step": 3474 }, { "epoch": 22.4, "grad_norm": 19.433406829833984, "learning_rate": 3.879032258064516e-05, "loss": 0.4962, "step": 3475 }, { "epoch": 22.406425702811244, "grad_norm": 18.203201293945312, "learning_rate": 3.878709677419355e-05, "loss": 0.6052, "step": 3476 }, { "epoch": 22.41285140562249, "grad_norm": 13.17719554901123, "learning_rate": 3.8783870967741937e-05, "loss": 0.5519, "step": 3477 }, { "epoch": 22.419277108433736, "grad_norm": 5.477985858917236, "learning_rate": 3.878064516129032e-05, "loss": 0.232, "step": 3478 }, { "epoch": 22.42570281124498, "grad_norm": 17.547405242919922, "learning_rate": 3.8777419354838715e-05, "loss": 0.3381, "step": 3479 }, { "epoch": 22.432128514056224, "grad_norm": 14.516705513000488, "learning_rate": 3.87741935483871e-05, "loss": 0.661, "step": 3480 }, { "epoch": 22.43855421686747, "grad_norm": 12.247808456420898, "learning_rate": 3.8770967741935486e-05, "loss": 0.4088, "step": 3481 }, { "epoch": 22.444979919678715, "grad_norm": 8.079507827758789, "learning_rate": 3.876774193548388e-05, "loss": 0.3583, "step": 3482 }, { "epoch": 22.45140562248996, "grad_norm": 8.495936393737793, "learning_rate": 3.876451612903226e-05, "loss": 0.5995, "step": 3483 }, { "epoch": 22.457831325301203, "grad_norm": 2.9898154735565186, "learning_rate": 3.876129032258064e-05, "loss": 0.1705, "step": 3484 }, { "epoch": 22.46425702811245, "grad_norm": 24.33673858642578, "learning_rate": 3.8758064516129036e-05, "loss": 0.4126, "step": 3485 }, { "epoch": 22.470682730923695, "grad_norm": 11.33456802368164, "learning_rate": 3.875483870967742e-05, "loss": 0.32, "step": 3486 }, { "epoch": 22.47710843373494, "grad_norm": 16.498271942138672, "learning_rate": 3.875161290322581e-05, "loss": 0.3338, "step": 3487 }, { "epoch": 22.483534136546183, "grad_norm": 5.490726470947266, "learning_rate": 3.87483870967742e-05, "loss": 0.4296, "step": 3488 }, { "epoch": 22.48995983935743, "grad_norm": 26.155305862426758, "learning_rate": 3.8745161290322585e-05, "loss": 0.5676, "step": 3489 }, { "epoch": 22.496385542168674, "grad_norm": 16.11275291442871, "learning_rate": 3.8741935483870964e-05, "loss": 0.3625, "step": 3490 }, { "epoch": 22.50281124497992, "grad_norm": 10.485849380493164, "learning_rate": 3.873870967741936e-05, "loss": 0.4575, "step": 3491 }, { "epoch": 22.509236947791166, "grad_norm": 13.05556869506836, "learning_rate": 3.873548387096774e-05, "loss": 0.3387, "step": 3492 }, { "epoch": 22.51566265060241, "grad_norm": 8.607215881347656, "learning_rate": 3.873225806451613e-05, "loss": 0.3104, "step": 3493 }, { "epoch": 22.522088353413654, "grad_norm": 8.87901496887207, "learning_rate": 3.872903225806452e-05, "loss": 0.2807, "step": 3494 }, { "epoch": 22.5285140562249, "grad_norm": 7.94420862197876, "learning_rate": 3.8725806451612906e-05, "loss": 0.3217, "step": 3495 }, { "epoch": 22.534939759036146, "grad_norm": 11.935037612915039, "learning_rate": 3.872258064516129e-05, "loss": 0.519, "step": 3496 }, { "epoch": 22.54136546184739, "grad_norm": 15.944056510925293, "learning_rate": 3.8719354838709685e-05, "loss": 0.3049, "step": 3497 }, { "epoch": 22.547791164658634, "grad_norm": 15.100893020629883, "learning_rate": 3.8716129032258063e-05, "loss": 0.4392, "step": 3498 }, { "epoch": 22.55421686746988, "grad_norm": 8.937443733215332, "learning_rate": 3.871290322580645e-05, "loss": 0.264, "step": 3499 }, { "epoch": 22.560642570281125, "grad_norm": 5.563431262969971, "learning_rate": 3.870967741935484e-05, "loss": 0.4046, "step": 3500 }, { "epoch": 22.56706827309237, "grad_norm": 7.107187747955322, "learning_rate": 3.870645161290323e-05, "loss": 0.6144, "step": 3501 }, { "epoch": 22.573493975903613, "grad_norm": 14.660552024841309, "learning_rate": 3.870322580645161e-05, "loss": 0.5857, "step": 3502 }, { "epoch": 22.57991967871486, "grad_norm": 6.319762229919434, "learning_rate": 3.8700000000000006e-05, "loss": 0.4933, "step": 3503 }, { "epoch": 22.586345381526105, "grad_norm": 9.310393333435059, "learning_rate": 3.869677419354839e-05, "loss": 0.3005, "step": 3504 }, { "epoch": 22.59277108433735, "grad_norm": 14.431374549865723, "learning_rate": 3.869354838709678e-05, "loss": 0.455, "step": 3505 }, { "epoch": 22.599196787148593, "grad_norm": 8.249683380126953, "learning_rate": 3.869032258064516e-05, "loss": 0.6313, "step": 3506 }, { "epoch": 22.60562248995984, "grad_norm": 11.094064712524414, "learning_rate": 3.868709677419355e-05, "loss": 0.6728, "step": 3507 }, { "epoch": 22.612048192771084, "grad_norm": 22.361764907836914, "learning_rate": 3.8683870967741934e-05, "loss": 0.5215, "step": 3508 }, { "epoch": 22.61847389558233, "grad_norm": 17.250682830810547, "learning_rate": 3.8680645161290327e-05, "loss": 0.2567, "step": 3509 }, { "epoch": 22.624899598393576, "grad_norm": 5.164998531341553, "learning_rate": 3.867741935483871e-05, "loss": 0.4255, "step": 3510 }, { "epoch": 22.631325301204818, "grad_norm": 12.177507400512695, "learning_rate": 3.86741935483871e-05, "loss": 0.2855, "step": 3511 }, { "epoch": 22.637751004016064, "grad_norm": 33.543983459472656, "learning_rate": 3.867096774193549e-05, "loss": 0.3938, "step": 3512 }, { "epoch": 22.64417670682731, "grad_norm": 4.302333354949951, "learning_rate": 3.8667741935483876e-05, "loss": 0.3103, "step": 3513 }, { "epoch": 22.650602409638555, "grad_norm": 7.390701770782471, "learning_rate": 3.8664516129032255e-05, "loss": 0.233, "step": 3514 }, { "epoch": 22.6570281124498, "grad_norm": 15.049347877502441, "learning_rate": 3.866129032258065e-05, "loss": 0.5024, "step": 3515 }, { "epoch": 22.663453815261043, "grad_norm": 23.573986053466797, "learning_rate": 3.865806451612903e-05, "loss": 0.4903, "step": 3516 }, { "epoch": 22.66987951807229, "grad_norm": 3.0933690071105957, "learning_rate": 3.865483870967742e-05, "loss": 0.1696, "step": 3517 }, { "epoch": 22.676305220883535, "grad_norm": 12.558716773986816, "learning_rate": 3.865161290322581e-05, "loss": 0.6342, "step": 3518 }, { "epoch": 22.68273092369478, "grad_norm": 13.018028259277344, "learning_rate": 3.86483870967742e-05, "loss": 0.273, "step": 3519 }, { "epoch": 22.689156626506023, "grad_norm": 8.763374328613281, "learning_rate": 3.864516129032258e-05, "loss": 0.4755, "step": 3520 }, { "epoch": 22.69558232931727, "grad_norm": 8.874258041381836, "learning_rate": 3.864193548387097e-05, "loss": 0.3781, "step": 3521 }, { "epoch": 22.702008032128514, "grad_norm": 22.032123565673828, "learning_rate": 3.8638709677419354e-05, "loss": 0.5462, "step": 3522 }, { "epoch": 22.70843373493976, "grad_norm": 17.33104133605957, "learning_rate": 3.863548387096774e-05, "loss": 0.6615, "step": 3523 }, { "epoch": 22.714859437751002, "grad_norm": 20.48407745361328, "learning_rate": 3.863225806451613e-05, "loss": 0.4297, "step": 3524 }, { "epoch": 22.721285140562248, "grad_norm": 10.808270454406738, "learning_rate": 3.862903225806452e-05, "loss": 0.414, "step": 3525 }, { "epoch": 22.727710843373494, "grad_norm": 5.407540321350098, "learning_rate": 3.8625806451612904e-05, "loss": 0.3122, "step": 3526 }, { "epoch": 22.73413654618474, "grad_norm": 23.538175582885742, "learning_rate": 3.8622580645161296e-05, "loss": 0.395, "step": 3527 }, { "epoch": 22.740562248995985, "grad_norm": 11.83050537109375, "learning_rate": 3.861935483870968e-05, "loss": 0.5061, "step": 3528 }, { "epoch": 22.746987951807228, "grad_norm": 15.14716911315918, "learning_rate": 3.861612903225806e-05, "loss": 0.4386, "step": 3529 }, { "epoch": 22.753413654618473, "grad_norm": 12.274280548095703, "learning_rate": 3.8612903225806454e-05, "loss": 0.45, "step": 3530 }, { "epoch": 22.75983935742972, "grad_norm": 10.277016639709473, "learning_rate": 3.860967741935484e-05, "loss": 0.2049, "step": 3531 }, { "epoch": 22.766265060240965, "grad_norm": 4.980801582336426, "learning_rate": 3.8606451612903225e-05, "loss": 0.4301, "step": 3532 }, { "epoch": 22.77269076305221, "grad_norm": 15.733292579650879, "learning_rate": 3.860322580645162e-05, "loss": 0.3488, "step": 3533 }, { "epoch": 22.779116465863453, "grad_norm": 5.468967914581299, "learning_rate": 3.86e-05, "loss": 0.5484, "step": 3534 }, { "epoch": 22.7855421686747, "grad_norm": 12.286474227905273, "learning_rate": 3.859677419354839e-05, "loss": 0.4444, "step": 3535 }, { "epoch": 22.791967871485944, "grad_norm": 10.083943367004395, "learning_rate": 3.859354838709678e-05, "loss": 0.514, "step": 3536 }, { "epoch": 22.79839357429719, "grad_norm": 22.790388107299805, "learning_rate": 3.859032258064516e-05, "loss": 0.3961, "step": 3537 }, { "epoch": 22.804819277108432, "grad_norm": 14.09924030303955, "learning_rate": 3.8587096774193546e-05, "loss": 0.4072, "step": 3538 }, { "epoch": 22.811244979919678, "grad_norm": 12.129369735717773, "learning_rate": 3.858387096774194e-05, "loss": 0.4019, "step": 3539 }, { "epoch": 22.817670682730924, "grad_norm": 15.642332077026367, "learning_rate": 3.8580645161290324e-05, "loss": 0.5055, "step": 3540 }, { "epoch": 22.82409638554217, "grad_norm": 7.763985633850098, "learning_rate": 3.857741935483871e-05, "loss": 0.3136, "step": 3541 }, { "epoch": 22.830522088353412, "grad_norm": 22.471027374267578, "learning_rate": 3.85741935483871e-05, "loss": 0.547, "step": 3542 }, { "epoch": 22.836947791164658, "grad_norm": 9.794718742370605, "learning_rate": 3.857096774193549e-05, "loss": 0.2235, "step": 3543 }, { "epoch": 22.843373493975903, "grad_norm": 16.430152893066406, "learning_rate": 3.8567741935483874e-05, "loss": 0.544, "step": 3544 }, { "epoch": 22.84979919678715, "grad_norm": 18.500123977661133, "learning_rate": 3.856451612903226e-05, "loss": 0.6306, "step": 3545 }, { "epoch": 22.856224899598395, "grad_norm": 5.54966926574707, "learning_rate": 3.8561290322580645e-05, "loss": 0.2521, "step": 3546 }, { "epoch": 22.862650602409637, "grad_norm": 14.633441925048828, "learning_rate": 3.855806451612903e-05, "loss": 0.4718, "step": 3547 }, { "epoch": 22.869076305220883, "grad_norm": 24.348718643188477, "learning_rate": 3.855483870967742e-05, "loss": 0.4929, "step": 3548 }, { "epoch": 22.87550200803213, "grad_norm": 10.415568351745605, "learning_rate": 3.855161290322581e-05, "loss": 0.3118, "step": 3549 }, { "epoch": 22.881927710843375, "grad_norm": 7.3126020431518555, "learning_rate": 3.8548387096774195e-05, "loss": 0.2499, "step": 3550 }, { "epoch": 22.88835341365462, "grad_norm": 8.59892463684082, "learning_rate": 3.854516129032259e-05, "loss": 0.5905, "step": 3551 }, { "epoch": 22.894779116465863, "grad_norm": 6.015598297119141, "learning_rate": 3.854193548387097e-05, "loss": 0.1709, "step": 3552 }, { "epoch": 22.90120481927711, "grad_norm": 4.160667896270752, "learning_rate": 3.853870967741935e-05, "loss": 0.3963, "step": 3553 }, { "epoch": 22.907630522088354, "grad_norm": 3.9820590019226074, "learning_rate": 3.8535483870967744e-05, "loss": 0.2656, "step": 3554 }, { "epoch": 22.9140562248996, "grad_norm": 48.22384262084961, "learning_rate": 3.853225806451613e-05, "loss": 0.4625, "step": 3555 }, { "epoch": 22.920481927710842, "grad_norm": 18.11779022216797, "learning_rate": 3.8529032258064516e-05, "loss": 0.6638, "step": 3556 }, { "epoch": 22.926907630522088, "grad_norm": 19.604692459106445, "learning_rate": 3.852580645161291e-05, "loss": 0.5329, "step": 3557 }, { "epoch": 22.933333333333334, "grad_norm": 23.222625732421875, "learning_rate": 3.8522580645161294e-05, "loss": 0.6438, "step": 3558 }, { "epoch": 22.93975903614458, "grad_norm": 11.833881378173828, "learning_rate": 3.851935483870968e-05, "loss": 0.4445, "step": 3559 }, { "epoch": 22.94618473895582, "grad_norm": 10.715567588806152, "learning_rate": 3.8516129032258065e-05, "loss": 0.2321, "step": 3560 }, { "epoch": 22.952610441767067, "grad_norm": 4.450475692749023, "learning_rate": 3.851290322580645e-05, "loss": 0.3269, "step": 3561 }, { "epoch": 22.959036144578313, "grad_norm": 15.808937072753906, "learning_rate": 3.850967741935484e-05, "loss": 0.7116, "step": 3562 }, { "epoch": 22.96546184738956, "grad_norm": 25.405071258544922, "learning_rate": 3.850645161290323e-05, "loss": 0.3658, "step": 3563 }, { "epoch": 22.971887550200805, "grad_norm": 6.906757831573486, "learning_rate": 3.8503225806451615e-05, "loss": 0.4393, "step": 3564 }, { "epoch": 22.978313253012047, "grad_norm": 9.183610916137695, "learning_rate": 3.85e-05, "loss": 0.3683, "step": 3565 }, { "epoch": 22.984738955823293, "grad_norm": 13.637290954589844, "learning_rate": 3.849677419354839e-05, "loss": 0.4278, "step": 3566 }, { "epoch": 22.99116465863454, "grad_norm": 7.444928169250488, "learning_rate": 3.849354838709678e-05, "loss": 0.4563, "step": 3567 }, { "epoch": 23.000803212851405, "grad_norm": 5.888779640197754, "learning_rate": 3.849032258064516e-05, "loss": 0.3793, "step": 3568 }, { "epoch": 23.00722891566265, "grad_norm": 9.354326248168945, "learning_rate": 3.848709677419355e-05, "loss": 0.4146, "step": 3569 }, { "epoch": 23.013654618473897, "grad_norm": 19.218198776245117, "learning_rate": 3.8483870967741936e-05, "loss": 0.4765, "step": 3570 }, { "epoch": 23.02008032128514, "grad_norm": 13.348125457763672, "learning_rate": 3.848064516129032e-05, "loss": 0.445, "step": 3571 }, { "epoch": 23.026506024096385, "grad_norm": 8.7850341796875, "learning_rate": 3.8477419354838714e-05, "loss": 0.446, "step": 3572 }, { "epoch": 23.03293172690763, "grad_norm": 9.505046844482422, "learning_rate": 3.84741935483871e-05, "loss": 0.3564, "step": 3573 }, { "epoch": 23.039357429718876, "grad_norm": 8.132492065429688, "learning_rate": 3.8470967741935486e-05, "loss": 0.375, "step": 3574 }, { "epoch": 23.045783132530122, "grad_norm": 8.675435066223145, "learning_rate": 3.846774193548388e-05, "loss": 0.2668, "step": 3575 }, { "epoch": 23.052208835341364, "grad_norm": 8.628107070922852, "learning_rate": 3.846451612903226e-05, "loss": 0.5967, "step": 3576 }, { "epoch": 23.05863453815261, "grad_norm": 15.959792137145996, "learning_rate": 3.846129032258064e-05, "loss": 0.3788, "step": 3577 }, { "epoch": 23.065060240963856, "grad_norm": 7.844707489013672, "learning_rate": 3.8458064516129035e-05, "loss": 0.5364, "step": 3578 }, { "epoch": 23.0714859437751, "grad_norm": 20.886014938354492, "learning_rate": 3.845483870967742e-05, "loss": 0.6571, "step": 3579 }, { "epoch": 23.077911646586344, "grad_norm": 13.266661643981934, "learning_rate": 3.845161290322581e-05, "loss": 0.3975, "step": 3580 }, { "epoch": 23.08433734939759, "grad_norm": 11.832317352294922, "learning_rate": 3.84483870967742e-05, "loss": 0.4358, "step": 3581 }, { "epoch": 23.090763052208835, "grad_norm": 12.30675983428955, "learning_rate": 3.8445161290322585e-05, "loss": 0.5757, "step": 3582 }, { "epoch": 23.09718875502008, "grad_norm": 24.695199966430664, "learning_rate": 3.844193548387097e-05, "loss": 0.7142, "step": 3583 }, { "epoch": 23.103614457831327, "grad_norm": 14.874889373779297, "learning_rate": 3.8438709677419356e-05, "loss": 0.3219, "step": 3584 }, { "epoch": 23.11004016064257, "grad_norm": 9.843557357788086, "learning_rate": 3.843548387096774e-05, "loss": 0.6423, "step": 3585 }, { "epoch": 23.116465863453815, "grad_norm": 12.146288871765137, "learning_rate": 3.843225806451613e-05, "loss": 0.2675, "step": 3586 }, { "epoch": 23.12289156626506, "grad_norm": 20.158342361450195, "learning_rate": 3.842903225806452e-05, "loss": 1.0122, "step": 3587 }, { "epoch": 23.129317269076306, "grad_norm": 2.9812216758728027, "learning_rate": 3.8425806451612906e-05, "loss": 0.2052, "step": 3588 }, { "epoch": 23.13574297188755, "grad_norm": 25.613374710083008, "learning_rate": 3.842258064516129e-05, "loss": 0.5692, "step": 3589 }, { "epoch": 23.142168674698794, "grad_norm": 25.153892517089844, "learning_rate": 3.8419354838709684e-05, "loss": 0.696, "step": 3590 }, { "epoch": 23.14859437751004, "grad_norm": 21.323598861694336, "learning_rate": 3.841612903225806e-05, "loss": 0.5051, "step": 3591 }, { "epoch": 23.155020080321286, "grad_norm": 16.405181884765625, "learning_rate": 3.841290322580645e-05, "loss": 0.5213, "step": 3592 }, { "epoch": 23.16144578313253, "grad_norm": 23.799846649169922, "learning_rate": 3.840967741935484e-05, "loss": 0.2617, "step": 3593 }, { "epoch": 23.167871485943774, "grad_norm": 7.757800102233887, "learning_rate": 3.840645161290323e-05, "loss": 0.5235, "step": 3594 }, { "epoch": 23.17429718875502, "grad_norm": 4.529571533203125, "learning_rate": 3.840322580645161e-05, "loss": 0.2031, "step": 3595 }, { "epoch": 23.180722891566266, "grad_norm": 8.17531681060791, "learning_rate": 3.8400000000000005e-05, "loss": 0.506, "step": 3596 }, { "epoch": 23.18714859437751, "grad_norm": 10.569864273071289, "learning_rate": 3.839677419354839e-05, "loss": 0.4603, "step": 3597 }, { "epoch": 23.193574297188754, "grad_norm": 14.314037322998047, "learning_rate": 3.8393548387096776e-05, "loss": 0.4027, "step": 3598 }, { "epoch": 23.2, "grad_norm": 14.239825248718262, "learning_rate": 3.839032258064516e-05, "loss": 0.4141, "step": 3599 }, { "epoch": 23.206425702811245, "grad_norm": 9.606986045837402, "learning_rate": 3.838709677419355e-05, "loss": 0.3123, "step": 3600 }, { "epoch": 23.21285140562249, "grad_norm": 15.769445419311523, "learning_rate": 3.8383870967741934e-05, "loss": 0.3783, "step": 3601 }, { "epoch": 23.219277108433737, "grad_norm": 19.382102966308594, "learning_rate": 3.8380645161290326e-05, "loss": 0.4044, "step": 3602 }, { "epoch": 23.22570281124498, "grad_norm": 22.678041458129883, "learning_rate": 3.837741935483871e-05, "loss": 0.2523, "step": 3603 }, { "epoch": 23.232128514056225, "grad_norm": 11.240117073059082, "learning_rate": 3.83741935483871e-05, "loss": 0.8522, "step": 3604 }, { "epoch": 23.23855421686747, "grad_norm": 14.12989330291748, "learning_rate": 3.837096774193549e-05, "loss": 0.5009, "step": 3605 }, { "epoch": 23.244979919678716, "grad_norm": 5.8354668617248535, "learning_rate": 3.8367741935483876e-05, "loss": 0.3725, "step": 3606 }, { "epoch": 23.25140562248996, "grad_norm": 6.278074264526367, "learning_rate": 3.8364516129032255e-05, "loss": 0.3428, "step": 3607 }, { "epoch": 23.257831325301204, "grad_norm": 9.154566764831543, "learning_rate": 3.836129032258065e-05, "loss": 0.1789, "step": 3608 }, { "epoch": 23.26425702811245, "grad_norm": 10.572389602661133, "learning_rate": 3.835806451612903e-05, "loss": 0.4065, "step": 3609 }, { "epoch": 23.270682730923696, "grad_norm": 14.42238998413086, "learning_rate": 3.835483870967742e-05, "loss": 0.4188, "step": 3610 }, { "epoch": 23.27710843373494, "grad_norm": 12.559586524963379, "learning_rate": 3.835161290322581e-05, "loss": 0.2206, "step": 3611 }, { "epoch": 23.283534136546184, "grad_norm": 11.710147857666016, "learning_rate": 3.83483870967742e-05, "loss": 0.7047, "step": 3612 }, { "epoch": 23.28995983935743, "grad_norm": 21.30648422241211, "learning_rate": 3.834516129032258e-05, "loss": 0.7727, "step": 3613 }, { "epoch": 23.296385542168675, "grad_norm": 12.783341407775879, "learning_rate": 3.8341935483870975e-05, "loss": 0.5261, "step": 3614 }, { "epoch": 23.30281124497992, "grad_norm": 4.555001735687256, "learning_rate": 3.8338709677419354e-05, "loss": 0.3239, "step": 3615 }, { "epoch": 23.309236947791163, "grad_norm": 8.965538024902344, "learning_rate": 3.833548387096774e-05, "loss": 0.2823, "step": 3616 }, { "epoch": 23.31566265060241, "grad_norm": 14.111420631408691, "learning_rate": 3.833225806451613e-05, "loss": 0.4998, "step": 3617 }, { "epoch": 23.322088353413655, "grad_norm": 21.77393341064453, "learning_rate": 3.832903225806452e-05, "loss": 0.5531, "step": 3618 }, { "epoch": 23.3285140562249, "grad_norm": 15.33271598815918, "learning_rate": 3.8325806451612903e-05, "loss": 0.6781, "step": 3619 }, { "epoch": 23.334939759036146, "grad_norm": 10.945493698120117, "learning_rate": 3.8322580645161296e-05, "loss": 0.3966, "step": 3620 }, { "epoch": 23.34136546184739, "grad_norm": 4.519842624664307, "learning_rate": 3.831935483870968e-05, "loss": 0.2733, "step": 3621 }, { "epoch": 23.347791164658634, "grad_norm": 10.521261215209961, "learning_rate": 3.831612903225807e-05, "loss": 0.3996, "step": 3622 }, { "epoch": 23.35421686746988, "grad_norm": 11.083212852478027, "learning_rate": 3.831290322580645e-05, "loss": 0.6766, "step": 3623 }, { "epoch": 23.360642570281126, "grad_norm": 8.541653633117676, "learning_rate": 3.830967741935484e-05, "loss": 0.1523, "step": 3624 }, { "epoch": 23.367068273092368, "grad_norm": 8.209257125854492, "learning_rate": 3.8306451612903224e-05, "loss": 0.1927, "step": 3625 }, { "epoch": 23.373493975903614, "grad_norm": 15.018196105957031, "learning_rate": 3.830322580645162e-05, "loss": 0.281, "step": 3626 }, { "epoch": 23.37991967871486, "grad_norm": 9.954232215881348, "learning_rate": 3.83e-05, "loss": 0.3061, "step": 3627 }, { "epoch": 23.386345381526105, "grad_norm": 7.598156452178955, "learning_rate": 3.829677419354839e-05, "loss": 0.4212, "step": 3628 }, { "epoch": 23.39277108433735, "grad_norm": 6.177085876464844, "learning_rate": 3.829354838709678e-05, "loss": 0.216, "step": 3629 }, { "epoch": 23.399196787148593, "grad_norm": 17.09168815612793, "learning_rate": 3.829032258064516e-05, "loss": 0.5938, "step": 3630 }, { "epoch": 23.40562248995984, "grad_norm": 12.520232200622559, "learning_rate": 3.8287096774193545e-05, "loss": 0.5417, "step": 3631 }, { "epoch": 23.412048192771085, "grad_norm": 6.592396259307861, "learning_rate": 3.828387096774194e-05, "loss": 0.3709, "step": 3632 }, { "epoch": 23.41847389558233, "grad_norm": 10.826860427856445, "learning_rate": 3.8280645161290324e-05, "loss": 0.4649, "step": 3633 }, { "epoch": 23.424899598393573, "grad_norm": 15.03710651397705, "learning_rate": 3.827741935483871e-05, "loss": 0.3156, "step": 3634 }, { "epoch": 23.43132530120482, "grad_norm": 12.173885345458984, "learning_rate": 3.82741935483871e-05, "loss": 0.3851, "step": 3635 }, { "epoch": 23.437751004016064, "grad_norm": 32.048431396484375, "learning_rate": 3.827096774193549e-05, "loss": 0.4175, "step": 3636 }, { "epoch": 23.44417670682731, "grad_norm": 7.118498802185059, "learning_rate": 3.826774193548387e-05, "loss": 0.3439, "step": 3637 }, { "epoch": 23.450602409638556, "grad_norm": 8.568367958068848, "learning_rate": 3.826451612903226e-05, "loss": 0.3097, "step": 3638 }, { "epoch": 23.457028112449798, "grad_norm": 9.556573867797852, "learning_rate": 3.8261290322580645e-05, "loss": 0.1856, "step": 3639 }, { "epoch": 23.463453815261044, "grad_norm": 13.988377571105957, "learning_rate": 3.825806451612903e-05, "loss": 0.5092, "step": 3640 }, { "epoch": 23.46987951807229, "grad_norm": 29.153249740600586, "learning_rate": 3.825483870967742e-05, "loss": 0.717, "step": 3641 }, { "epoch": 23.476305220883535, "grad_norm": 7.3266682624816895, "learning_rate": 3.825161290322581e-05, "loss": 0.2678, "step": 3642 }, { "epoch": 23.482730923694778, "grad_norm": 8.721044540405273, "learning_rate": 3.8248387096774194e-05, "loss": 0.4667, "step": 3643 }, { "epoch": 23.489156626506023, "grad_norm": 5.286614894866943, "learning_rate": 3.824516129032259e-05, "loss": 0.3945, "step": 3644 }, { "epoch": 23.49558232931727, "grad_norm": 14.114830017089844, "learning_rate": 3.824193548387097e-05, "loss": 0.5398, "step": 3645 }, { "epoch": 23.502008032128515, "grad_norm": 13.245737075805664, "learning_rate": 3.823870967741935e-05, "loss": 0.607, "step": 3646 }, { "epoch": 23.50843373493976, "grad_norm": 18.035322189331055, "learning_rate": 3.8235483870967744e-05, "loss": 0.2332, "step": 3647 }, { "epoch": 23.514859437751003, "grad_norm": 9.51993179321289, "learning_rate": 3.823225806451613e-05, "loss": 0.3571, "step": 3648 }, { "epoch": 23.52128514056225, "grad_norm": 12.369452476501465, "learning_rate": 3.8229032258064515e-05, "loss": 0.3536, "step": 3649 }, { "epoch": 23.527710843373494, "grad_norm": 9.353987693786621, "learning_rate": 3.822580645161291e-05, "loss": 0.3161, "step": 3650 }, { "epoch": 23.53413654618474, "grad_norm": 14.557169914245605, "learning_rate": 3.8222580645161293e-05, "loss": 0.5462, "step": 3651 }, { "epoch": 23.540562248995982, "grad_norm": 7.5732598304748535, "learning_rate": 3.821935483870968e-05, "loss": 0.226, "step": 3652 }, { "epoch": 23.54698795180723, "grad_norm": 21.249874114990234, "learning_rate": 3.821612903225807e-05, "loss": 0.5551, "step": 3653 }, { "epoch": 23.553413654618474, "grad_norm": 21.295185089111328, "learning_rate": 3.821290322580645e-05, "loss": 0.4137, "step": 3654 }, { "epoch": 23.55983935742972, "grad_norm": 6.369776725769043, "learning_rate": 3.8209677419354836e-05, "loss": 0.3055, "step": 3655 }, { "epoch": 23.566265060240966, "grad_norm": 20.303625106811523, "learning_rate": 3.820645161290323e-05, "loss": 0.4212, "step": 3656 }, { "epoch": 23.572690763052208, "grad_norm": 8.563718795776367, "learning_rate": 3.8203225806451614e-05, "loss": 0.4761, "step": 3657 }, { "epoch": 23.579116465863454, "grad_norm": 6.492434978485107, "learning_rate": 3.82e-05, "loss": 0.1402, "step": 3658 }, { "epoch": 23.5855421686747, "grad_norm": 24.678295135498047, "learning_rate": 3.819677419354839e-05, "loss": 0.4886, "step": 3659 }, { "epoch": 23.591967871485945, "grad_norm": 10.656782150268555, "learning_rate": 3.819354838709678e-05, "loss": 0.5249, "step": 3660 }, { "epoch": 23.598393574297187, "grad_norm": 11.240211486816406, "learning_rate": 3.8190322580645164e-05, "loss": 0.4487, "step": 3661 }, { "epoch": 23.604819277108433, "grad_norm": 11.257104873657227, "learning_rate": 3.818709677419355e-05, "loss": 0.2958, "step": 3662 }, { "epoch": 23.61124497991968, "grad_norm": 7.8342485427856445, "learning_rate": 3.8183870967741935e-05, "loss": 0.466, "step": 3663 }, { "epoch": 23.617670682730925, "grad_norm": 9.243145942687988, "learning_rate": 3.818064516129032e-05, "loss": 0.3206, "step": 3664 }, { "epoch": 23.62409638554217, "grad_norm": 13.4445219039917, "learning_rate": 3.8177419354838714e-05, "loss": 0.5136, "step": 3665 }, { "epoch": 23.630522088353413, "grad_norm": 8.297382354736328, "learning_rate": 3.81741935483871e-05, "loss": 0.2114, "step": 3666 }, { "epoch": 23.63694779116466, "grad_norm": 12.70518684387207, "learning_rate": 3.8170967741935485e-05, "loss": 0.6135, "step": 3667 }, { "epoch": 23.643373493975904, "grad_norm": 16.995834350585938, "learning_rate": 3.816774193548388e-05, "loss": 0.4426, "step": 3668 }, { "epoch": 23.64979919678715, "grad_norm": 11.29144287109375, "learning_rate": 3.8164516129032257e-05, "loss": 0.4999, "step": 3669 }, { "epoch": 23.656224899598392, "grad_norm": 12.457321166992188, "learning_rate": 3.816129032258064e-05, "loss": 0.63, "step": 3670 }, { "epoch": 23.662650602409638, "grad_norm": 5.800262928009033, "learning_rate": 3.8158064516129035e-05, "loss": 0.3803, "step": 3671 }, { "epoch": 23.669076305220884, "grad_norm": 21.317691802978516, "learning_rate": 3.815483870967742e-05, "loss": 0.5071, "step": 3672 }, { "epoch": 23.67550200803213, "grad_norm": 12.238798141479492, "learning_rate": 3.8151612903225806e-05, "loss": 0.4566, "step": 3673 }, { "epoch": 23.681927710843375, "grad_norm": 27.78399658203125, "learning_rate": 3.81483870967742e-05, "loss": 0.5733, "step": 3674 }, { "epoch": 23.688353413654617, "grad_norm": 17.651514053344727, "learning_rate": 3.8145161290322584e-05, "loss": 0.6375, "step": 3675 }, { "epoch": 23.694779116465863, "grad_norm": 5.826592922210693, "learning_rate": 3.814193548387097e-05, "loss": 0.2409, "step": 3676 }, { "epoch": 23.70120481927711, "grad_norm": 7.676506042480469, "learning_rate": 3.8138709677419356e-05, "loss": 0.5002, "step": 3677 }, { "epoch": 23.707630522088355, "grad_norm": 13.666747093200684, "learning_rate": 3.813548387096774e-05, "loss": 0.9002, "step": 3678 }, { "epoch": 23.714056224899597, "grad_norm": 3.98343563079834, "learning_rate": 3.813225806451613e-05, "loss": 0.2176, "step": 3679 }, { "epoch": 23.720481927710843, "grad_norm": 24.278417587280273, "learning_rate": 3.812903225806452e-05, "loss": 0.6039, "step": 3680 }, { "epoch": 23.72690763052209, "grad_norm": 3.4620394706726074, "learning_rate": 3.8125806451612905e-05, "loss": 0.1691, "step": 3681 }, { "epoch": 23.733333333333334, "grad_norm": 15.59699535369873, "learning_rate": 3.812258064516129e-05, "loss": 0.4227, "step": 3682 }, { "epoch": 23.739759036144576, "grad_norm": 10.248353958129883, "learning_rate": 3.8119354838709684e-05, "loss": 0.4933, "step": 3683 }, { "epoch": 23.746184738955822, "grad_norm": 5.109351634979248, "learning_rate": 3.811612903225807e-05, "loss": 0.2296, "step": 3684 }, { "epoch": 23.752610441767068, "grad_norm": 32.04221725463867, "learning_rate": 3.811290322580645e-05, "loss": 0.3586, "step": 3685 }, { "epoch": 23.759036144578314, "grad_norm": 18.091646194458008, "learning_rate": 3.810967741935484e-05, "loss": 0.5859, "step": 3686 }, { "epoch": 23.76546184738956, "grad_norm": 11.488927841186523, "learning_rate": 3.8106451612903226e-05, "loss": 0.4019, "step": 3687 }, { "epoch": 23.771887550200802, "grad_norm": 6.664236068725586, "learning_rate": 3.810322580645161e-05, "loss": 0.3063, "step": 3688 }, { "epoch": 23.778313253012048, "grad_norm": 8.372321128845215, "learning_rate": 3.8100000000000005e-05, "loss": 0.4584, "step": 3689 }, { "epoch": 23.784738955823293, "grad_norm": 16.472808837890625, "learning_rate": 3.809677419354839e-05, "loss": 0.4043, "step": 3690 }, { "epoch": 23.79116465863454, "grad_norm": 9.758806228637695, "learning_rate": 3.8093548387096776e-05, "loss": 0.4298, "step": 3691 }, { "epoch": 23.797590361445785, "grad_norm": 10.800868034362793, "learning_rate": 3.809032258064517e-05, "loss": 0.4031, "step": 3692 }, { "epoch": 23.804016064257027, "grad_norm": 5.929788589477539, "learning_rate": 3.808709677419355e-05, "loss": 0.3307, "step": 3693 }, { "epoch": 23.810441767068273, "grad_norm": 14.096739768981934, "learning_rate": 3.808387096774193e-05, "loss": 0.3937, "step": 3694 }, { "epoch": 23.81686746987952, "grad_norm": 53.801536560058594, "learning_rate": 3.8080645161290326e-05, "loss": 0.4393, "step": 3695 }, { "epoch": 23.823293172690764, "grad_norm": 10.665088653564453, "learning_rate": 3.807741935483871e-05, "loss": 0.2818, "step": 3696 }, { "epoch": 23.829718875502007, "grad_norm": 8.765824317932129, "learning_rate": 3.80741935483871e-05, "loss": 0.3659, "step": 3697 }, { "epoch": 23.836144578313252, "grad_norm": 19.62428855895996, "learning_rate": 3.807096774193549e-05, "loss": 0.4869, "step": 3698 }, { "epoch": 23.842570281124498, "grad_norm": 8.991098403930664, "learning_rate": 3.8067741935483875e-05, "loss": 0.7239, "step": 3699 }, { "epoch": 23.848995983935744, "grad_norm": 8.216568946838379, "learning_rate": 3.8064516129032254e-05, "loss": 0.4089, "step": 3700 }, { "epoch": 23.855421686746986, "grad_norm": 8.0570068359375, "learning_rate": 3.8061290322580647e-05, "loss": 0.501, "step": 3701 }, { "epoch": 23.861847389558232, "grad_norm": 8.282992362976074, "learning_rate": 3.805806451612903e-05, "loss": 0.4743, "step": 3702 }, { "epoch": 23.868273092369478, "grad_norm": 18.629030227661133, "learning_rate": 3.805483870967742e-05, "loss": 0.3317, "step": 3703 }, { "epoch": 23.874698795180723, "grad_norm": 4.104436874389648, "learning_rate": 3.805161290322581e-05, "loss": 0.2483, "step": 3704 }, { "epoch": 23.88112449799197, "grad_norm": 7.5443902015686035, "learning_rate": 3.8048387096774196e-05, "loss": 0.6211, "step": 3705 }, { "epoch": 23.88755020080321, "grad_norm": 9.143326759338379, "learning_rate": 3.804516129032258e-05, "loss": 0.3999, "step": 3706 }, { "epoch": 23.893975903614457, "grad_norm": 13.968475341796875, "learning_rate": 3.8041935483870974e-05, "loss": 0.4179, "step": 3707 }, { "epoch": 23.900401606425703, "grad_norm": 5.0189528465271, "learning_rate": 3.803870967741935e-05, "loss": 0.3253, "step": 3708 }, { "epoch": 23.90682730923695, "grad_norm": 13.285310745239258, "learning_rate": 3.803548387096774e-05, "loss": 0.3453, "step": 3709 }, { "epoch": 23.913253012048195, "grad_norm": 12.097421646118164, "learning_rate": 3.803225806451613e-05, "loss": 0.3714, "step": 3710 }, { "epoch": 23.919678714859437, "grad_norm": 12.903337478637695, "learning_rate": 3.802903225806452e-05, "loss": 0.3508, "step": 3711 }, { "epoch": 23.926104417670683, "grad_norm": 18.25607681274414, "learning_rate": 3.80258064516129e-05, "loss": 0.4236, "step": 3712 }, { "epoch": 23.93253012048193, "grad_norm": 20.429763793945312, "learning_rate": 3.8022580645161295e-05, "loss": 0.1868, "step": 3713 }, { "epoch": 23.938955823293174, "grad_norm": 6.007972240447998, "learning_rate": 3.801935483870968e-05, "loss": 0.1628, "step": 3714 }, { "epoch": 23.945381526104416, "grad_norm": 11.650959968566895, "learning_rate": 3.801612903225807e-05, "loss": 0.3906, "step": 3715 }, { "epoch": 23.951807228915662, "grad_norm": 100.73197174072266, "learning_rate": 3.801290322580645e-05, "loss": 0.5471, "step": 3716 }, { "epoch": 23.958232931726908, "grad_norm": 7.651844024658203, "learning_rate": 3.800967741935484e-05, "loss": 0.325, "step": 3717 }, { "epoch": 23.964658634538154, "grad_norm": 8.737321853637695, "learning_rate": 3.8006451612903224e-05, "loss": 0.3141, "step": 3718 }, { "epoch": 23.971084337349396, "grad_norm": 19.55518913269043, "learning_rate": 3.8003225806451616e-05, "loss": 0.2611, "step": 3719 }, { "epoch": 23.97751004016064, "grad_norm": 4.501558303833008, "learning_rate": 3.8e-05, "loss": 0.2561, "step": 3720 }, { "epoch": 23.983935742971887, "grad_norm": 12.080550193786621, "learning_rate": 3.799677419354839e-05, "loss": 0.3705, "step": 3721 }, { "epoch": 23.990361445783133, "grad_norm": 23.894962310791016, "learning_rate": 3.799354838709678e-05, "loss": 0.2426, "step": 3722 }, { "epoch": 23.99678714859438, "grad_norm": 17.967885971069336, "learning_rate": 3.7990322580645166e-05, "loss": 0.5931, "step": 3723 }, { "epoch": 24.006425702811246, "grad_norm": 7.613768577575684, "learning_rate": 3.7987096774193545e-05, "loss": 0.4884, "step": 3724 }, { "epoch": 24.01285140562249, "grad_norm": 17.435705184936523, "learning_rate": 3.798387096774194e-05, "loss": 0.5446, "step": 3725 }, { "epoch": 24.019277108433734, "grad_norm": 7.789377212524414, "learning_rate": 3.798064516129032e-05, "loss": 0.4948, "step": 3726 }, { "epoch": 24.02570281124498, "grad_norm": 6.914918899536133, "learning_rate": 3.797741935483871e-05, "loss": 0.3253, "step": 3727 }, { "epoch": 24.032128514056225, "grad_norm": 8.129278182983398, "learning_rate": 3.79741935483871e-05, "loss": 0.401, "step": 3728 }, { "epoch": 24.03855421686747, "grad_norm": 7.718605995178223, "learning_rate": 3.797096774193549e-05, "loss": 0.4095, "step": 3729 }, { "epoch": 24.044979919678713, "grad_norm": 13.235170364379883, "learning_rate": 3.796774193548387e-05, "loss": 0.5822, "step": 3730 }, { "epoch": 24.05140562248996, "grad_norm": 12.603840827941895, "learning_rate": 3.7964516129032265e-05, "loss": 0.4101, "step": 3731 }, { "epoch": 24.057831325301205, "grad_norm": 10.590435028076172, "learning_rate": 3.7961290322580644e-05, "loss": 0.609, "step": 3732 }, { "epoch": 24.06425702811245, "grad_norm": 4.743679046630859, "learning_rate": 3.795806451612903e-05, "loss": 0.3129, "step": 3733 }, { "epoch": 24.070682730923696, "grad_norm": 11.503169059753418, "learning_rate": 3.795483870967742e-05, "loss": 0.3304, "step": 3734 }, { "epoch": 24.07710843373494, "grad_norm": 7.882920742034912, "learning_rate": 3.795161290322581e-05, "loss": 0.3395, "step": 3735 }, { "epoch": 24.083534136546184, "grad_norm": 14.431835174560547, "learning_rate": 3.7948387096774194e-05, "loss": 0.3107, "step": 3736 }, { "epoch": 24.08995983935743, "grad_norm": 6.4873504638671875, "learning_rate": 3.7945161290322586e-05, "loss": 0.4185, "step": 3737 }, { "epoch": 24.096385542168676, "grad_norm": 12.798707962036133, "learning_rate": 3.794193548387097e-05, "loss": 0.3895, "step": 3738 }, { "epoch": 24.102811244979918, "grad_norm": 5.158512115478516, "learning_rate": 3.793870967741935e-05, "loss": 0.2175, "step": 3739 }, { "epoch": 24.109236947791164, "grad_norm": 8.390878677368164, "learning_rate": 3.793548387096774e-05, "loss": 0.3216, "step": 3740 }, { "epoch": 24.11566265060241, "grad_norm": 5.771450519561768, "learning_rate": 3.793225806451613e-05, "loss": 0.4027, "step": 3741 }, { "epoch": 24.122088353413655, "grad_norm": 13.2645902633667, "learning_rate": 3.7929032258064515e-05, "loss": 0.4384, "step": 3742 }, { "epoch": 24.1285140562249, "grad_norm": 6.194428443908691, "learning_rate": 3.792580645161291e-05, "loss": 0.3019, "step": 3743 }, { "epoch": 24.134939759036143, "grad_norm": 7.279941558837891, "learning_rate": 3.792258064516129e-05, "loss": 0.3151, "step": 3744 }, { "epoch": 24.14136546184739, "grad_norm": 11.429244995117188, "learning_rate": 3.791935483870968e-05, "loss": 0.4651, "step": 3745 }, { "epoch": 24.147791164658635, "grad_norm": 13.069605827331543, "learning_rate": 3.791612903225807e-05, "loss": 0.6465, "step": 3746 }, { "epoch": 24.15421686746988, "grad_norm": 35.47580337524414, "learning_rate": 3.791290322580645e-05, "loss": 0.402, "step": 3747 }, { "epoch": 24.160642570281123, "grad_norm": 8.927448272705078, "learning_rate": 3.7909677419354836e-05, "loss": 0.4339, "step": 3748 }, { "epoch": 24.16706827309237, "grad_norm": 10.941190719604492, "learning_rate": 3.790645161290323e-05, "loss": 0.3009, "step": 3749 }, { "epoch": 24.173493975903614, "grad_norm": 23.952592849731445, "learning_rate": 3.7903225806451614e-05, "loss": 0.7558, "step": 3750 }, { "epoch": 24.17991967871486, "grad_norm": 13.455610275268555, "learning_rate": 3.79e-05, "loss": 0.2023, "step": 3751 }, { "epoch": 24.186345381526106, "grad_norm": 6.496456146240234, "learning_rate": 3.789677419354839e-05, "loss": 0.2206, "step": 3752 }, { "epoch": 24.19277108433735, "grad_norm": 24.507034301757812, "learning_rate": 3.789354838709678e-05, "loss": 0.648, "step": 3753 }, { "epoch": 24.199196787148594, "grad_norm": 28.49405860900879, "learning_rate": 3.7890322580645164e-05, "loss": 0.2938, "step": 3754 }, { "epoch": 24.20562248995984, "grad_norm": 4.402164936065674, "learning_rate": 3.788709677419355e-05, "loss": 0.3549, "step": 3755 }, { "epoch": 24.212048192771086, "grad_norm": 7.4764533042907715, "learning_rate": 3.7883870967741935e-05, "loss": 0.4506, "step": 3756 }, { "epoch": 24.218473895582328, "grad_norm": 13.422089576721191, "learning_rate": 3.788064516129032e-05, "loss": 0.5167, "step": 3757 }, { "epoch": 24.224899598393574, "grad_norm": 8.87651538848877, "learning_rate": 3.787741935483871e-05, "loss": 0.3118, "step": 3758 }, { "epoch": 24.23132530120482, "grad_norm": 7.710142135620117, "learning_rate": 3.78741935483871e-05, "loss": 0.2151, "step": 3759 }, { "epoch": 24.237751004016065, "grad_norm": 8.872915267944336, "learning_rate": 3.7870967741935485e-05, "loss": 0.5635, "step": 3760 }, { "epoch": 24.24417670682731, "grad_norm": 4.658843040466309, "learning_rate": 3.786774193548388e-05, "loss": 0.2797, "step": 3761 }, { "epoch": 24.250602409638553, "grad_norm": 7.29288387298584, "learning_rate": 3.786451612903226e-05, "loss": 0.3945, "step": 3762 }, { "epoch": 24.2570281124498, "grad_norm": 12.364317893981934, "learning_rate": 3.786129032258064e-05, "loss": 0.4307, "step": 3763 }, { "epoch": 24.263453815261045, "grad_norm": 13.242073059082031, "learning_rate": 3.7858064516129034e-05, "loss": 0.317, "step": 3764 }, { "epoch": 24.26987951807229, "grad_norm": 7.414330005645752, "learning_rate": 3.785483870967742e-05, "loss": 0.472, "step": 3765 }, { "epoch": 24.276305220883533, "grad_norm": 7.314948558807373, "learning_rate": 3.7851612903225806e-05, "loss": 0.1263, "step": 3766 }, { "epoch": 24.28273092369478, "grad_norm": 12.771259307861328, "learning_rate": 3.78483870967742e-05, "loss": 0.4163, "step": 3767 }, { "epoch": 24.289156626506024, "grad_norm": 14.20347785949707, "learning_rate": 3.7845161290322584e-05, "loss": 0.3266, "step": 3768 }, { "epoch": 24.29558232931727, "grad_norm": 11.11436653137207, "learning_rate": 3.784193548387097e-05, "loss": 0.6315, "step": 3769 }, { "epoch": 24.302008032128516, "grad_norm": 15.377031326293945, "learning_rate": 3.7838709677419355e-05, "loss": 0.6363, "step": 3770 }, { "epoch": 24.308433734939758, "grad_norm": 11.926899909973145, "learning_rate": 3.783548387096774e-05, "loss": 0.4461, "step": 3771 }, { "epoch": 24.314859437751004, "grad_norm": 9.450200080871582, "learning_rate": 3.7832258064516127e-05, "loss": 0.3889, "step": 3772 }, { "epoch": 24.32128514056225, "grad_norm": 9.116235733032227, "learning_rate": 3.782903225806452e-05, "loss": 0.3745, "step": 3773 }, { "epoch": 24.327710843373495, "grad_norm": 14.904829025268555, "learning_rate": 3.7825806451612905e-05, "loss": 0.4045, "step": 3774 }, { "epoch": 24.334136546184737, "grad_norm": 28.700475692749023, "learning_rate": 3.782258064516129e-05, "loss": 0.6177, "step": 3775 }, { "epoch": 24.340562248995983, "grad_norm": 24.251752853393555, "learning_rate": 3.781935483870968e-05, "loss": 0.2774, "step": 3776 }, { "epoch": 24.34698795180723, "grad_norm": 16.4367733001709, "learning_rate": 3.781612903225807e-05, "loss": 0.6644, "step": 3777 }, { "epoch": 24.353413654618475, "grad_norm": 7.760530471801758, "learning_rate": 3.781290322580645e-05, "loss": 0.3185, "step": 3778 }, { "epoch": 24.35983935742972, "grad_norm": 11.849804878234863, "learning_rate": 3.780967741935484e-05, "loss": 0.4713, "step": 3779 }, { "epoch": 24.366265060240963, "grad_norm": 8.21606159210205, "learning_rate": 3.7806451612903226e-05, "loss": 0.4134, "step": 3780 }, { "epoch": 24.37269076305221, "grad_norm": 4.84442663192749, "learning_rate": 3.780322580645161e-05, "loss": 0.1659, "step": 3781 }, { "epoch": 24.379116465863454, "grad_norm": 4.243147373199463, "learning_rate": 3.7800000000000004e-05, "loss": 0.3357, "step": 3782 }, { "epoch": 24.3855421686747, "grad_norm": 12.39113712310791, "learning_rate": 3.779677419354839e-05, "loss": 0.2336, "step": 3783 }, { "epoch": 24.391967871485942, "grad_norm": 18.611888885498047, "learning_rate": 3.7793548387096775e-05, "loss": 0.5232, "step": 3784 }, { "epoch": 24.398393574297188, "grad_norm": 14.981047630310059, "learning_rate": 3.779032258064517e-05, "loss": 0.3748, "step": 3785 }, { "epoch": 24.404819277108434, "grad_norm": 11.267483711242676, "learning_rate": 3.778709677419355e-05, "loss": 0.5133, "step": 3786 }, { "epoch": 24.41124497991968, "grad_norm": 8.136117935180664, "learning_rate": 3.778387096774193e-05, "loss": 0.3736, "step": 3787 }, { "epoch": 24.417670682730925, "grad_norm": 11.5302095413208, "learning_rate": 3.7780645161290325e-05, "loss": 0.4114, "step": 3788 }, { "epoch": 24.424096385542168, "grad_norm": 10.162724494934082, "learning_rate": 3.777741935483871e-05, "loss": 0.4153, "step": 3789 }, { "epoch": 24.430522088353413, "grad_norm": 12.919342994689941, "learning_rate": 3.7774193548387096e-05, "loss": 0.4693, "step": 3790 }, { "epoch": 24.43694779116466, "grad_norm": 9.84970760345459, "learning_rate": 3.777096774193549e-05, "loss": 0.4317, "step": 3791 }, { "epoch": 24.443373493975905, "grad_norm": 13.756994247436523, "learning_rate": 3.7767741935483875e-05, "loss": 0.2569, "step": 3792 }, { "epoch": 24.449799196787147, "grad_norm": 13.947928428649902, "learning_rate": 3.776451612903226e-05, "loss": 0.7614, "step": 3793 }, { "epoch": 24.456224899598393, "grad_norm": 10.057026863098145, "learning_rate": 3.7761290322580646e-05, "loss": 0.4255, "step": 3794 }, { "epoch": 24.46265060240964, "grad_norm": 10.968335151672363, "learning_rate": 3.775806451612903e-05, "loss": 0.2329, "step": 3795 }, { "epoch": 24.469076305220884, "grad_norm": 14.995638847351074, "learning_rate": 3.775483870967742e-05, "loss": 0.3127, "step": 3796 }, { "epoch": 24.47550200803213, "grad_norm": 4.555752754211426, "learning_rate": 3.775161290322581e-05, "loss": 0.2553, "step": 3797 }, { "epoch": 24.481927710843372, "grad_norm": 7.384092330932617, "learning_rate": 3.7748387096774196e-05, "loss": 0.3345, "step": 3798 }, { "epoch": 24.488353413654618, "grad_norm": 6.508727073669434, "learning_rate": 3.774516129032258e-05, "loss": 0.1711, "step": 3799 }, { "epoch": 24.494779116465864, "grad_norm": 11.817814826965332, "learning_rate": 3.7741935483870974e-05, "loss": 0.4883, "step": 3800 }, { "epoch": 24.50120481927711, "grad_norm": 17.43389892578125, "learning_rate": 3.773870967741936e-05, "loss": 0.3737, "step": 3801 }, { "epoch": 24.507630522088352, "grad_norm": 9.521485328674316, "learning_rate": 3.773548387096774e-05, "loss": 0.3388, "step": 3802 }, { "epoch": 24.514056224899598, "grad_norm": 10.025880813598633, "learning_rate": 3.773225806451613e-05, "loss": 0.3079, "step": 3803 }, { "epoch": 24.520481927710843, "grad_norm": 3.1796813011169434, "learning_rate": 3.772903225806452e-05, "loss": 0.1809, "step": 3804 }, { "epoch": 24.52690763052209, "grad_norm": 22.31972885131836, "learning_rate": 3.77258064516129e-05, "loss": 0.2985, "step": 3805 }, { "epoch": 24.533333333333335, "grad_norm": 7.648327350616455, "learning_rate": 3.7722580645161295e-05, "loss": 0.2385, "step": 3806 }, { "epoch": 24.539759036144577, "grad_norm": 13.380800247192383, "learning_rate": 3.771935483870968e-05, "loss": 0.5629, "step": 3807 }, { "epoch": 24.546184738955823, "grad_norm": 17.832002639770508, "learning_rate": 3.7716129032258066e-05, "loss": 0.3571, "step": 3808 }, { "epoch": 24.55261044176707, "grad_norm": 8.603607177734375, "learning_rate": 3.771290322580645e-05, "loss": 0.1698, "step": 3809 }, { "epoch": 24.559036144578315, "grad_norm": 7.535625457763672, "learning_rate": 3.770967741935484e-05, "loss": 0.4394, "step": 3810 }, { "epoch": 24.565461847389557, "grad_norm": 13.141098976135254, "learning_rate": 3.770645161290322e-05, "loss": 0.4086, "step": 3811 }, { "epoch": 24.571887550200803, "grad_norm": 10.692342758178711, "learning_rate": 3.7703225806451616e-05, "loss": 0.4569, "step": 3812 }, { "epoch": 24.57831325301205, "grad_norm": 10.804984092712402, "learning_rate": 3.77e-05, "loss": 0.4451, "step": 3813 }, { "epoch": 24.584738955823294, "grad_norm": 12.235997200012207, "learning_rate": 3.769677419354839e-05, "loss": 0.2411, "step": 3814 }, { "epoch": 24.59116465863454, "grad_norm": 23.159191131591797, "learning_rate": 3.769354838709678e-05, "loss": 0.6003, "step": 3815 }, { "epoch": 24.597590361445782, "grad_norm": 15.829947471618652, "learning_rate": 3.7690322580645165e-05, "loss": 0.562, "step": 3816 }, { "epoch": 24.604016064257028, "grad_norm": 7.475739479064941, "learning_rate": 3.7687096774193544e-05, "loss": 0.3374, "step": 3817 }, { "epoch": 24.610441767068274, "grad_norm": 12.706560134887695, "learning_rate": 3.768387096774194e-05, "loss": 0.4598, "step": 3818 }, { "epoch": 24.61686746987952, "grad_norm": 13.210616111755371, "learning_rate": 3.768064516129032e-05, "loss": 0.4233, "step": 3819 }, { "epoch": 24.62329317269076, "grad_norm": 14.952659606933594, "learning_rate": 3.767741935483871e-05, "loss": 0.4247, "step": 3820 }, { "epoch": 24.629718875502007, "grad_norm": 15.996359825134277, "learning_rate": 3.76741935483871e-05, "loss": 0.5378, "step": 3821 }, { "epoch": 24.636144578313253, "grad_norm": 7.584722995758057, "learning_rate": 3.7670967741935486e-05, "loss": 0.3957, "step": 3822 }, { "epoch": 24.6425702811245, "grad_norm": 8.814048767089844, "learning_rate": 3.766774193548387e-05, "loss": 0.5571, "step": 3823 }, { "epoch": 24.648995983935745, "grad_norm": 9.913712501525879, "learning_rate": 3.7664516129032265e-05, "loss": 0.3789, "step": 3824 }, { "epoch": 24.655421686746987, "grad_norm": 13.352310180664062, "learning_rate": 3.7661290322580644e-05, "loss": 0.4188, "step": 3825 }, { "epoch": 24.661847389558233, "grad_norm": 11.500670433044434, "learning_rate": 3.765806451612903e-05, "loss": 0.6083, "step": 3826 }, { "epoch": 24.66827309236948, "grad_norm": 13.059135437011719, "learning_rate": 3.765483870967742e-05, "loss": 0.3699, "step": 3827 }, { "epoch": 24.674698795180724, "grad_norm": 5.880025863647461, "learning_rate": 3.765161290322581e-05, "loss": 0.2697, "step": 3828 }, { "epoch": 24.681124497991966, "grad_norm": 5.068190097808838, "learning_rate": 3.764838709677419e-05, "loss": 0.2067, "step": 3829 }, { "epoch": 24.687550200803212, "grad_norm": 4.424836158752441, "learning_rate": 3.7645161290322586e-05, "loss": 0.1772, "step": 3830 }, { "epoch": 24.693975903614458, "grad_norm": 9.741162300109863, "learning_rate": 3.764193548387097e-05, "loss": 0.4867, "step": 3831 }, { "epoch": 24.700401606425704, "grad_norm": 6.616354465484619, "learning_rate": 3.763870967741936e-05, "loss": 0.2837, "step": 3832 }, { "epoch": 24.70682730923695, "grad_norm": 28.401580810546875, "learning_rate": 3.763548387096774e-05, "loss": 0.5059, "step": 3833 }, { "epoch": 24.71325301204819, "grad_norm": 5.901049613952637, "learning_rate": 3.763225806451613e-05, "loss": 0.2455, "step": 3834 }, { "epoch": 24.719678714859437, "grad_norm": 16.770845413208008, "learning_rate": 3.7629032258064514e-05, "loss": 0.3839, "step": 3835 }, { "epoch": 24.726104417670683, "grad_norm": 7.653078556060791, "learning_rate": 3.762580645161291e-05, "loss": 0.2563, "step": 3836 }, { "epoch": 24.73253012048193, "grad_norm": 15.274239540100098, "learning_rate": 3.762258064516129e-05, "loss": 0.5933, "step": 3837 }, { "epoch": 24.73895582329317, "grad_norm": 7.110038757324219, "learning_rate": 3.761935483870968e-05, "loss": 0.5618, "step": 3838 }, { "epoch": 24.745381526104417, "grad_norm": 5.1029276847839355, "learning_rate": 3.761612903225807e-05, "loss": 0.172, "step": 3839 }, { "epoch": 24.751807228915663, "grad_norm": 4.8744378089904785, "learning_rate": 3.761290322580645e-05, "loss": 0.1954, "step": 3840 }, { "epoch": 24.75823293172691, "grad_norm": 12.327849388122559, "learning_rate": 3.7609677419354835e-05, "loss": 0.2383, "step": 3841 }, { "epoch": 24.764658634538154, "grad_norm": 5.266777515411377, "learning_rate": 3.760645161290323e-05, "loss": 0.4851, "step": 3842 }, { "epoch": 24.771084337349397, "grad_norm": 13.18642520904541, "learning_rate": 3.7603225806451613e-05, "loss": 0.4304, "step": 3843 }, { "epoch": 24.777510040160642, "grad_norm": 8.777108192443848, "learning_rate": 3.76e-05, "loss": 0.3466, "step": 3844 }, { "epoch": 24.783935742971888, "grad_norm": 23.054344177246094, "learning_rate": 3.759677419354839e-05, "loss": 0.6136, "step": 3845 }, { "epoch": 24.790361445783134, "grad_norm": 5.684847831726074, "learning_rate": 3.759354838709678e-05, "loss": 0.1349, "step": 3846 }, { "epoch": 24.796787148594376, "grad_norm": 13.264238357543945, "learning_rate": 3.759032258064516e-05, "loss": 0.3619, "step": 3847 }, { "epoch": 24.803212851405622, "grad_norm": 9.369147300720215, "learning_rate": 3.758709677419355e-05, "loss": 0.4419, "step": 3848 }, { "epoch": 24.809638554216868, "grad_norm": 7.674674987792969, "learning_rate": 3.7583870967741934e-05, "loss": 0.3328, "step": 3849 }, { "epoch": 24.816064257028113, "grad_norm": 3.9471516609191895, "learning_rate": 3.758064516129032e-05, "loss": 0.1772, "step": 3850 }, { "epoch": 24.82248995983936, "grad_norm": 13.151703834533691, "learning_rate": 3.757741935483871e-05, "loss": 0.3374, "step": 3851 }, { "epoch": 24.8289156626506, "grad_norm": 8.539238929748535, "learning_rate": 3.75741935483871e-05, "loss": 0.4699, "step": 3852 }, { "epoch": 24.835341365461847, "grad_norm": 9.732882499694824, "learning_rate": 3.7570967741935484e-05, "loss": 0.3215, "step": 3853 }, { "epoch": 24.841767068273093, "grad_norm": 12.205499649047852, "learning_rate": 3.7567741935483877e-05, "loss": 0.3764, "step": 3854 }, { "epoch": 24.84819277108434, "grad_norm": 17.83991813659668, "learning_rate": 3.756451612903226e-05, "loss": 0.4979, "step": 3855 }, { "epoch": 24.85461847389558, "grad_norm": 17.69977569580078, "learning_rate": 3.756129032258064e-05, "loss": 0.4884, "step": 3856 }, { "epoch": 24.861044176706827, "grad_norm": 14.239144325256348, "learning_rate": 3.7558064516129034e-05, "loss": 0.3894, "step": 3857 }, { "epoch": 24.867469879518072, "grad_norm": 32.07857131958008, "learning_rate": 3.755483870967742e-05, "loss": 0.6338, "step": 3858 }, { "epoch": 24.873895582329318, "grad_norm": 6.63215446472168, "learning_rate": 3.7551612903225805e-05, "loss": 0.3401, "step": 3859 }, { "epoch": 24.880321285140564, "grad_norm": 21.489086151123047, "learning_rate": 3.75483870967742e-05, "loss": 0.4222, "step": 3860 }, { "epoch": 24.886746987951806, "grad_norm": 12.778510093688965, "learning_rate": 3.754516129032258e-05, "loss": 0.2868, "step": 3861 }, { "epoch": 24.893172690763052, "grad_norm": 9.432244300842285, "learning_rate": 3.754193548387097e-05, "loss": 0.3511, "step": 3862 }, { "epoch": 24.899598393574298, "grad_norm": 7.279962062835693, "learning_rate": 3.753870967741936e-05, "loss": 0.3855, "step": 3863 }, { "epoch": 24.906024096385543, "grad_norm": 25.64647102355957, "learning_rate": 3.753548387096774e-05, "loss": 0.4174, "step": 3864 }, { "epoch": 24.912449799196786, "grad_norm": 9.372386932373047, "learning_rate": 3.7532258064516126e-05, "loss": 0.2809, "step": 3865 }, { "epoch": 24.91887550200803, "grad_norm": 9.750895500183105, "learning_rate": 3.752903225806452e-05, "loss": 0.3515, "step": 3866 }, { "epoch": 24.925301204819277, "grad_norm": 18.888044357299805, "learning_rate": 3.7525806451612904e-05, "loss": 0.5293, "step": 3867 }, { "epoch": 24.931726907630523, "grad_norm": 18.566450119018555, "learning_rate": 3.752258064516129e-05, "loss": 0.5964, "step": 3868 }, { "epoch": 24.93815261044177, "grad_norm": 24.83380126953125, "learning_rate": 3.751935483870968e-05, "loss": 0.7114, "step": 3869 }, { "epoch": 24.94457831325301, "grad_norm": 10.00092887878418, "learning_rate": 3.751612903225807e-05, "loss": 0.3724, "step": 3870 }, { "epoch": 24.951004016064257, "grad_norm": 6.718848705291748, "learning_rate": 3.7512903225806454e-05, "loss": 0.4068, "step": 3871 }, { "epoch": 24.957429718875503, "grad_norm": 14.081612586975098, "learning_rate": 3.750967741935484e-05, "loss": 0.5048, "step": 3872 }, { "epoch": 24.96385542168675, "grad_norm": 12.210514068603516, "learning_rate": 3.7506451612903225e-05, "loss": 0.4424, "step": 3873 }, { "epoch": 24.97028112449799, "grad_norm": 10.527420997619629, "learning_rate": 3.750322580645161e-05, "loss": 0.6564, "step": 3874 }, { "epoch": 24.976706827309236, "grad_norm": 17.2098445892334, "learning_rate": 3.7500000000000003e-05, "loss": 0.5857, "step": 3875 }, { "epoch": 24.983132530120482, "grad_norm": 8.51539421081543, "learning_rate": 3.749677419354839e-05, "loss": 0.3752, "step": 3876 }, { "epoch": 24.989558232931728, "grad_norm": 12.4055757522583, "learning_rate": 3.7493548387096775e-05, "loss": 0.4439, "step": 3877 }, { "epoch": 24.99598393574297, "grad_norm": 4.093153476715088, "learning_rate": 3.749032258064517e-05, "loss": 0.1671, "step": 3878 }, { "epoch": 25.00562248995984, "grad_norm": 13.271063804626465, "learning_rate": 3.7487096774193546e-05, "loss": 0.349, "step": 3879 }, { "epoch": 25.012048192771083, "grad_norm": 18.726642608642578, "learning_rate": 3.748387096774193e-05, "loss": 0.3196, "step": 3880 }, { "epoch": 25.01847389558233, "grad_norm": 19.781835556030273, "learning_rate": 3.7480645161290325e-05, "loss": 0.5892, "step": 3881 }, { "epoch": 25.024899598393574, "grad_norm": 3.150259494781494, "learning_rate": 3.747741935483871e-05, "loss": 0.19, "step": 3882 }, { "epoch": 25.03132530120482, "grad_norm": 22.82227897644043, "learning_rate": 3.7474193548387096e-05, "loss": 0.4454, "step": 3883 }, { "epoch": 25.037751004016066, "grad_norm": 15.628226280212402, "learning_rate": 3.747096774193549e-05, "loss": 0.6034, "step": 3884 }, { "epoch": 25.044176706827308, "grad_norm": 5.12390661239624, "learning_rate": 3.7467741935483874e-05, "loss": 0.4032, "step": 3885 }, { "epoch": 25.050602409638554, "grad_norm": 8.09774112701416, "learning_rate": 3.746451612903226e-05, "loss": 0.4698, "step": 3886 }, { "epoch": 25.0570281124498, "grad_norm": 35.53449630737305, "learning_rate": 3.7461290322580646e-05, "loss": 0.5639, "step": 3887 }, { "epoch": 25.063453815261045, "grad_norm": 28.73720359802246, "learning_rate": 3.745806451612903e-05, "loss": 0.332, "step": 3888 }, { "epoch": 25.069879518072288, "grad_norm": 4.2680206298828125, "learning_rate": 3.745483870967742e-05, "loss": 0.2774, "step": 3889 }, { "epoch": 25.076305220883533, "grad_norm": 9.122130393981934, "learning_rate": 3.745161290322581e-05, "loss": 0.3166, "step": 3890 }, { "epoch": 25.08273092369478, "grad_norm": 23.750032424926758, "learning_rate": 3.7448387096774195e-05, "loss": 0.6688, "step": 3891 }, { "epoch": 25.089156626506025, "grad_norm": 6.954874515533447, "learning_rate": 3.744516129032258e-05, "loss": 0.2266, "step": 3892 }, { "epoch": 25.09558232931727, "grad_norm": 5.304136753082275, "learning_rate": 3.744193548387097e-05, "loss": 0.3198, "step": 3893 }, { "epoch": 25.102008032128513, "grad_norm": 15.160226821899414, "learning_rate": 3.743870967741936e-05, "loss": 0.3739, "step": 3894 }, { "epoch": 25.10843373493976, "grad_norm": 6.6342387199401855, "learning_rate": 3.743548387096774e-05, "loss": 0.3264, "step": 3895 }, { "epoch": 25.114859437751004, "grad_norm": 5.066497325897217, "learning_rate": 3.743225806451613e-05, "loss": 0.2623, "step": 3896 }, { "epoch": 25.12128514056225, "grad_norm": 8.057430267333984, "learning_rate": 3.7429032258064516e-05, "loss": 0.4991, "step": 3897 }, { "epoch": 25.127710843373492, "grad_norm": 11.057938575744629, "learning_rate": 3.74258064516129e-05, "loss": 0.313, "step": 3898 }, { "epoch": 25.134136546184738, "grad_norm": 5.435968399047852, "learning_rate": 3.7422580645161294e-05, "loss": 0.221, "step": 3899 }, { "epoch": 25.140562248995984, "grad_norm": 4.614466667175293, "learning_rate": 3.741935483870968e-05, "loss": 0.4118, "step": 3900 }, { "epoch": 25.14698795180723, "grad_norm": 4.516018867492676, "learning_rate": 3.7416129032258066e-05, "loss": 0.5135, "step": 3901 }, { "epoch": 25.153413654618475, "grad_norm": 65.19731903076172, "learning_rate": 3.741290322580646e-05, "loss": 0.4062, "step": 3902 }, { "epoch": 25.159839357429718, "grad_norm": 7.697643280029297, "learning_rate": 3.740967741935484e-05, "loss": 0.2318, "step": 3903 }, { "epoch": 25.166265060240963, "grad_norm": 8.097184181213379, "learning_rate": 3.740645161290322e-05, "loss": 0.3556, "step": 3904 }, { "epoch": 25.17269076305221, "grad_norm": 4.229158401489258, "learning_rate": 3.7403225806451615e-05, "loss": 0.4298, "step": 3905 }, { "epoch": 25.179116465863455, "grad_norm": 17.153854370117188, "learning_rate": 3.74e-05, "loss": 0.51, "step": 3906 }, { "epoch": 25.185542168674697, "grad_norm": 22.802358627319336, "learning_rate": 3.739677419354839e-05, "loss": 0.1908, "step": 3907 }, { "epoch": 25.191967871485943, "grad_norm": 14.880060195922852, "learning_rate": 3.739354838709678e-05, "loss": 0.3906, "step": 3908 }, { "epoch": 25.19839357429719, "grad_norm": 40.24662780761719, "learning_rate": 3.7390322580645165e-05, "loss": 0.2642, "step": 3909 }, { "epoch": 25.204819277108435, "grad_norm": 8.988543510437012, "learning_rate": 3.738709677419355e-05, "loss": 0.3246, "step": 3910 }, { "epoch": 25.21124497991968, "grad_norm": 25.356931686401367, "learning_rate": 3.7383870967741936e-05, "loss": 0.2562, "step": 3911 }, { "epoch": 25.217670682730922, "grad_norm": 23.760068893432617, "learning_rate": 3.738064516129032e-05, "loss": 0.4145, "step": 3912 }, { "epoch": 25.22409638554217, "grad_norm": 11.780166625976562, "learning_rate": 3.737741935483871e-05, "loss": 0.6944, "step": 3913 }, { "epoch": 25.230522088353414, "grad_norm": 19.654348373413086, "learning_rate": 3.73741935483871e-05, "loss": 0.366, "step": 3914 }, { "epoch": 25.23694779116466, "grad_norm": 10.712606430053711, "learning_rate": 3.7370967741935486e-05, "loss": 0.3606, "step": 3915 }, { "epoch": 25.243373493975902, "grad_norm": 7.062700271606445, "learning_rate": 3.736774193548387e-05, "loss": 0.3504, "step": 3916 }, { "epoch": 25.249799196787148, "grad_norm": 3.1352059841156006, "learning_rate": 3.7364516129032264e-05, "loss": 0.2679, "step": 3917 }, { "epoch": 25.256224899598394, "grad_norm": 5.32490348815918, "learning_rate": 3.736129032258064e-05, "loss": 0.1877, "step": 3918 }, { "epoch": 25.26265060240964, "grad_norm": 84.48810577392578, "learning_rate": 3.735806451612903e-05, "loss": 0.5433, "step": 3919 }, { "epoch": 25.269076305220885, "grad_norm": 27.047529220581055, "learning_rate": 3.735483870967742e-05, "loss": 0.6644, "step": 3920 }, { "epoch": 25.275502008032127, "grad_norm": 7.422890663146973, "learning_rate": 3.735161290322581e-05, "loss": 0.3414, "step": 3921 }, { "epoch": 25.281927710843373, "grad_norm": 11.859882354736328, "learning_rate": 3.734838709677419e-05, "loss": 0.2141, "step": 3922 }, { "epoch": 25.28835341365462, "grad_norm": 10.716955184936523, "learning_rate": 3.7345161290322585e-05, "loss": 0.6556, "step": 3923 }, { "epoch": 25.294779116465865, "grad_norm": 1.9097329378128052, "learning_rate": 3.734193548387097e-05, "loss": 0.6154, "step": 3924 }, { "epoch": 25.301204819277107, "grad_norm": 6.511972427368164, "learning_rate": 3.7338709677419357e-05, "loss": 0.4527, "step": 3925 }, { "epoch": 25.307630522088353, "grad_norm": 11.2675199508667, "learning_rate": 3.733548387096774e-05, "loss": 0.2728, "step": 3926 }, { "epoch": 25.3140562248996, "grad_norm": 18.679492950439453, "learning_rate": 3.733225806451613e-05, "loss": 0.169, "step": 3927 }, { "epoch": 25.320481927710844, "grad_norm": 9.220023155212402, "learning_rate": 3.7329032258064514e-05, "loss": 0.3061, "step": 3928 }, { "epoch": 25.32690763052209, "grad_norm": 11.823091506958008, "learning_rate": 3.7325806451612906e-05, "loss": 0.5532, "step": 3929 }, { "epoch": 25.333333333333332, "grad_norm": 4.1291422843933105, "learning_rate": 3.732258064516129e-05, "loss": 0.3038, "step": 3930 }, { "epoch": 25.339759036144578, "grad_norm": 13.388586044311523, "learning_rate": 3.731935483870968e-05, "loss": 0.3643, "step": 3931 }, { "epoch": 25.346184738955824, "grad_norm": 13.94215202331543, "learning_rate": 3.731612903225807e-05, "loss": 0.6883, "step": 3932 }, { "epoch": 25.35261044176707, "grad_norm": 12.447965621948242, "learning_rate": 3.7312903225806456e-05, "loss": 0.4111, "step": 3933 }, { "epoch": 25.35903614457831, "grad_norm": 13.264385223388672, "learning_rate": 3.730967741935484e-05, "loss": 0.5006, "step": 3934 }, { "epoch": 25.365461847389557, "grad_norm": 223.52528381347656, "learning_rate": 3.730645161290323e-05, "loss": 0.3452, "step": 3935 }, { "epoch": 25.371887550200803, "grad_norm": 8.120107650756836, "learning_rate": 3.730322580645161e-05, "loss": 0.4009, "step": 3936 }, { "epoch": 25.37831325301205, "grad_norm": 20.616453170776367, "learning_rate": 3.73e-05, "loss": 0.1317, "step": 3937 }, { "epoch": 25.384738955823295, "grad_norm": 2.8170838356018066, "learning_rate": 3.729677419354839e-05, "loss": 0.1164, "step": 3938 }, { "epoch": 25.391164658634537, "grad_norm": 11.438002586364746, "learning_rate": 3.729354838709678e-05, "loss": 0.5241, "step": 3939 }, { "epoch": 25.397590361445783, "grad_norm": 1.26485276222229, "learning_rate": 3.729032258064516e-05, "loss": 0.1498, "step": 3940 }, { "epoch": 25.40401606425703, "grad_norm": 24.95465087890625, "learning_rate": 3.7287096774193555e-05, "loss": 0.1509, "step": 3941 }, { "epoch": 25.410441767068274, "grad_norm": 3.208717107772827, "learning_rate": 3.7283870967741934e-05, "loss": 0.314, "step": 3942 }, { "epoch": 25.416867469879517, "grad_norm": 9.943354606628418, "learning_rate": 3.728064516129032e-05, "loss": 0.3935, "step": 3943 }, { "epoch": 25.423293172690762, "grad_norm": 1.8202235698699951, "learning_rate": 3.727741935483871e-05, "loss": 0.3933, "step": 3944 }, { "epoch": 25.429718875502008, "grad_norm": 5.949231147766113, "learning_rate": 3.72741935483871e-05, "loss": 0.1572, "step": 3945 }, { "epoch": 25.436144578313254, "grad_norm": 5.3781819343566895, "learning_rate": 3.7270967741935484e-05, "loss": 0.3454, "step": 3946 }, { "epoch": 25.4425702811245, "grad_norm": 45.80697250366211, "learning_rate": 3.7267741935483876e-05, "loss": 0.3899, "step": 3947 }, { "epoch": 25.448995983935742, "grad_norm": 23.391408920288086, "learning_rate": 3.726451612903226e-05, "loss": 0.5113, "step": 3948 }, { "epoch": 25.455421686746988, "grad_norm": 20.647647857666016, "learning_rate": 3.726129032258065e-05, "loss": 0.3611, "step": 3949 }, { "epoch": 25.461847389558233, "grad_norm": 14.814657211303711, "learning_rate": 3.725806451612903e-05, "loss": 0.4942, "step": 3950 }, { "epoch": 25.46827309236948, "grad_norm": 1.2421363592147827, "learning_rate": 3.725483870967742e-05, "loss": 0.318, "step": 3951 }, { "epoch": 25.47469879518072, "grad_norm": 14.254593849182129, "learning_rate": 3.7251612903225805e-05, "loss": 0.2837, "step": 3952 }, { "epoch": 25.481124497991967, "grad_norm": 10.189422607421875, "learning_rate": 3.72483870967742e-05, "loss": 0.4281, "step": 3953 }, { "epoch": 25.487550200803213, "grad_norm": 12.169082641601562, "learning_rate": 3.724516129032258e-05, "loss": 0.4418, "step": 3954 }, { "epoch": 25.49397590361446, "grad_norm": 21.705795288085938, "learning_rate": 3.724193548387097e-05, "loss": 0.3959, "step": 3955 }, { "epoch": 25.500401606425704, "grad_norm": 19.3573055267334, "learning_rate": 3.723870967741936e-05, "loss": 0.6411, "step": 3956 }, { "epoch": 25.506827309236947, "grad_norm": 4.710412502288818, "learning_rate": 3.723548387096774e-05, "loss": 0.4317, "step": 3957 }, { "epoch": 25.513253012048192, "grad_norm": 1.3653723001480103, "learning_rate": 3.723225806451613e-05, "loss": 0.1995, "step": 3958 }, { "epoch": 25.519678714859438, "grad_norm": 12.758195877075195, "learning_rate": 3.722903225806452e-05, "loss": 0.5589, "step": 3959 }, { "epoch": 25.526104417670684, "grad_norm": 6.351041316986084, "learning_rate": 3.7225806451612904e-05, "loss": 0.4958, "step": 3960 }, { "epoch": 25.532530120481926, "grad_norm": 8.93304443359375, "learning_rate": 3.722258064516129e-05, "loss": 0.3354, "step": 3961 }, { "epoch": 25.538955823293172, "grad_norm": 24.9128475189209, "learning_rate": 3.721935483870968e-05, "loss": 0.7949, "step": 3962 }, { "epoch": 25.545381526104418, "grad_norm": 11.835153579711914, "learning_rate": 3.721612903225807e-05, "loss": 0.6286, "step": 3963 }, { "epoch": 25.551807228915663, "grad_norm": 10.2212553024292, "learning_rate": 3.721290322580645e-05, "loss": 0.302, "step": 3964 }, { "epoch": 25.55823293172691, "grad_norm": 6.873178005218506, "learning_rate": 3.720967741935484e-05, "loss": 0.3681, "step": 3965 }, { "epoch": 25.56465863453815, "grad_norm": 5.241540908813477, "learning_rate": 3.7206451612903225e-05, "loss": 0.5307, "step": 3966 }, { "epoch": 25.571084337349397, "grad_norm": 2.7602367401123047, "learning_rate": 3.720322580645161e-05, "loss": 0.514, "step": 3967 }, { "epoch": 25.577510040160643, "grad_norm": 6.888190746307373, "learning_rate": 3.72e-05, "loss": 0.5718, "step": 3968 }, { "epoch": 25.58393574297189, "grad_norm": 9.872206687927246, "learning_rate": 3.719677419354839e-05, "loss": 0.1495, "step": 3969 }, { "epoch": 25.59036144578313, "grad_norm": 14.953062057495117, "learning_rate": 3.7193548387096774e-05, "loss": 0.6279, "step": 3970 }, { "epoch": 25.596787148594377, "grad_norm": 6.911475658416748, "learning_rate": 3.719032258064517e-05, "loss": 0.4967, "step": 3971 }, { "epoch": 25.603212851405623, "grad_norm": 6.398134231567383, "learning_rate": 3.718709677419355e-05, "loss": 0.3203, "step": 3972 }, { "epoch": 25.60963855421687, "grad_norm": 8.863290786743164, "learning_rate": 3.718387096774194e-05, "loss": 0.4075, "step": 3973 }, { "epoch": 25.616064257028114, "grad_norm": 4.139695167541504, "learning_rate": 3.7180645161290324e-05, "loss": 0.1847, "step": 3974 }, { "epoch": 25.622489959839356, "grad_norm": 10.078216552734375, "learning_rate": 3.717741935483871e-05, "loss": 0.2568, "step": 3975 }, { "epoch": 25.628915662650602, "grad_norm": 6.454030513763428, "learning_rate": 3.7174193548387095e-05, "loss": 0.1736, "step": 3976 }, { "epoch": 25.635341365461848, "grad_norm": 13.343243598937988, "learning_rate": 3.717096774193549e-05, "loss": 0.2631, "step": 3977 }, { "epoch": 25.641767068273094, "grad_norm": 8.88530158996582, "learning_rate": 3.7167741935483874e-05, "loss": 0.5215, "step": 3978 }, { "epoch": 25.648192771084336, "grad_norm": 5.829901218414307, "learning_rate": 3.716451612903226e-05, "loss": 0.3979, "step": 3979 }, { "epoch": 25.65461847389558, "grad_norm": 40.960391998291016, "learning_rate": 3.716129032258065e-05, "loss": 0.6614, "step": 3980 }, { "epoch": 25.661044176706827, "grad_norm": 32.168025970458984, "learning_rate": 3.715806451612903e-05, "loss": 0.6188, "step": 3981 }, { "epoch": 25.667469879518073, "grad_norm": 64.25288391113281, "learning_rate": 3.7154838709677416e-05, "loss": 0.3279, "step": 3982 }, { "epoch": 25.67389558232932, "grad_norm": 36.47962951660156, "learning_rate": 3.715161290322581e-05, "loss": 0.4498, "step": 3983 }, { "epoch": 25.68032128514056, "grad_norm": 7.405526638031006, "learning_rate": 3.7148387096774195e-05, "loss": 0.5733, "step": 3984 }, { "epoch": 25.686746987951807, "grad_norm": 17.19173812866211, "learning_rate": 3.714516129032258e-05, "loss": 0.4939, "step": 3985 }, { "epoch": 25.693172690763053, "grad_norm": 42.35985565185547, "learning_rate": 3.714193548387097e-05, "loss": 0.6622, "step": 3986 }, { "epoch": 25.6995983935743, "grad_norm": 15.556333541870117, "learning_rate": 3.713870967741936e-05, "loss": 0.2919, "step": 3987 }, { "epoch": 25.70602409638554, "grad_norm": 21.986055374145508, "learning_rate": 3.7135483870967744e-05, "loss": 0.2399, "step": 3988 }, { "epoch": 25.712449799196786, "grad_norm": 11.113306045532227, "learning_rate": 3.713225806451613e-05, "loss": 0.2889, "step": 3989 }, { "epoch": 25.718875502008032, "grad_norm": 25.253616333007812, "learning_rate": 3.7129032258064516e-05, "loss": 0.3098, "step": 3990 }, { "epoch": 25.725301204819278, "grad_norm": 3.5777363777160645, "learning_rate": 3.71258064516129e-05, "loss": 0.269, "step": 3991 }, { "epoch": 25.731726907630524, "grad_norm": 6.865363121032715, "learning_rate": 3.7122580645161294e-05, "loss": 0.2158, "step": 3992 }, { "epoch": 25.738152610441766, "grad_norm": 22.351234436035156, "learning_rate": 3.711935483870968e-05, "loss": 0.4974, "step": 3993 }, { "epoch": 25.74457831325301, "grad_norm": 13.1561279296875, "learning_rate": 3.7116129032258065e-05, "loss": 0.2486, "step": 3994 }, { "epoch": 25.751004016064257, "grad_norm": 11.688264846801758, "learning_rate": 3.711290322580646e-05, "loss": 0.1744, "step": 3995 }, { "epoch": 25.757429718875503, "grad_norm": 9.910943031311035, "learning_rate": 3.710967741935484e-05, "loss": 0.1824, "step": 3996 }, { "epoch": 25.763855421686745, "grad_norm": 18.520610809326172, "learning_rate": 3.710645161290323e-05, "loss": 0.2918, "step": 3997 }, { "epoch": 25.77028112449799, "grad_norm": 8.854121208190918, "learning_rate": 3.7103225806451615e-05, "loss": 0.5446, "step": 3998 }, { "epoch": 25.776706827309237, "grad_norm": 17.077762603759766, "learning_rate": 3.71e-05, "loss": 0.3727, "step": 3999 }, { "epoch": 25.783132530120483, "grad_norm": 15.138181686401367, "learning_rate": 3.7096774193548386e-05, "loss": 0.6447, "step": 4000 }, { "epoch": 25.78955823293173, "grad_norm": 7.037140369415283, "learning_rate": 3.709354838709678e-05, "loss": 0.3361, "step": 4001 }, { "epoch": 25.79598393574297, "grad_norm": 6.1297197341918945, "learning_rate": 3.7090322580645164e-05, "loss": 0.2824, "step": 4002 }, { "epoch": 25.802409638554217, "grad_norm": 9.210156440734863, "learning_rate": 3.708709677419355e-05, "loss": 0.5734, "step": 4003 }, { "epoch": 25.808835341365462, "grad_norm": 5.59216833114624, "learning_rate": 3.7083870967741936e-05, "loss": 0.363, "step": 4004 }, { "epoch": 25.815261044176708, "grad_norm": 21.953102111816406, "learning_rate": 3.708064516129032e-05, "loss": 0.3887, "step": 4005 }, { "epoch": 25.82168674698795, "grad_norm": 11.124159812927246, "learning_rate": 3.707741935483871e-05, "loss": 0.3952, "step": 4006 }, { "epoch": 25.828112449799196, "grad_norm": 8.67715835571289, "learning_rate": 3.70741935483871e-05, "loss": 0.3404, "step": 4007 }, { "epoch": 25.834538152610442, "grad_norm": 14.769949913024902, "learning_rate": 3.7070967741935485e-05, "loss": 0.2002, "step": 4008 }, { "epoch": 25.840963855421688, "grad_norm": 29.541336059570312, "learning_rate": 3.706774193548387e-05, "loss": 0.526, "step": 4009 }, { "epoch": 25.847389558232933, "grad_norm": 13.445619583129883, "learning_rate": 3.7064516129032264e-05, "loss": 0.4216, "step": 4010 }, { "epoch": 25.853815261044176, "grad_norm": 5.176745414733887, "learning_rate": 3.706129032258065e-05, "loss": 0.207, "step": 4011 }, { "epoch": 25.86024096385542, "grad_norm": 4.194936275482178, "learning_rate": 3.7058064516129035e-05, "loss": 0.2906, "step": 4012 }, { "epoch": 25.866666666666667, "grad_norm": 22.152660369873047, "learning_rate": 3.705483870967742e-05, "loss": 0.3686, "step": 4013 }, { "epoch": 25.873092369477913, "grad_norm": 4.610465049743652, "learning_rate": 3.7051612903225806e-05, "loss": 0.332, "step": 4014 }, { "epoch": 25.879518072289155, "grad_norm": 10.795495986938477, "learning_rate": 3.704838709677419e-05, "loss": 0.3757, "step": 4015 }, { "epoch": 25.8859437751004, "grad_norm": 8.739364624023438, "learning_rate": 3.7045161290322585e-05, "loss": 0.5811, "step": 4016 }, { "epoch": 25.892369477911647, "grad_norm": 10.986830711364746, "learning_rate": 3.704193548387097e-05, "loss": 0.2382, "step": 4017 }, { "epoch": 25.898795180722892, "grad_norm": 5.930974006652832, "learning_rate": 3.7038709677419356e-05, "loss": 0.2735, "step": 4018 }, { "epoch": 25.905220883534138, "grad_norm": 15.03441047668457, "learning_rate": 3.703548387096774e-05, "loss": 0.6127, "step": 4019 }, { "epoch": 25.91164658634538, "grad_norm": 15.744409561157227, "learning_rate": 3.703225806451613e-05, "loss": 0.6865, "step": 4020 }, { "epoch": 25.918072289156626, "grad_norm": 16.848773956298828, "learning_rate": 3.702903225806452e-05, "loss": 0.4531, "step": 4021 }, { "epoch": 25.924497991967872, "grad_norm": 13.034455299377441, "learning_rate": 3.7025806451612906e-05, "loss": 0.4602, "step": 4022 }, { "epoch": 25.930923694779118, "grad_norm": 17.852264404296875, "learning_rate": 3.702258064516129e-05, "loss": 0.66, "step": 4023 }, { "epoch": 25.93734939759036, "grad_norm": 3.7156879901885986, "learning_rate": 3.701935483870968e-05, "loss": 0.2178, "step": 4024 }, { "epoch": 25.943775100401606, "grad_norm": 45.897438049316406, "learning_rate": 3.701612903225807e-05, "loss": 0.4876, "step": 4025 }, { "epoch": 25.95020080321285, "grad_norm": 14.23676586151123, "learning_rate": 3.7012903225806455e-05, "loss": 0.3898, "step": 4026 }, { "epoch": 25.956626506024097, "grad_norm": 12.012144088745117, "learning_rate": 3.700967741935484e-05, "loss": 0.1736, "step": 4027 }, { "epoch": 25.963052208835343, "grad_norm": 14.982361793518066, "learning_rate": 3.700645161290323e-05, "loss": 0.7406, "step": 4028 }, { "epoch": 25.969477911646585, "grad_norm": 4.938498497009277, "learning_rate": 3.700322580645161e-05, "loss": 0.4006, "step": 4029 }, { "epoch": 25.97590361445783, "grad_norm": 10.112564086914062, "learning_rate": 3.7e-05, "loss": 0.5449, "step": 4030 }, { "epoch": 25.982329317269077, "grad_norm": 32.29616928100586, "learning_rate": 3.699677419354839e-05, "loss": 0.5553, "step": 4031 }, { "epoch": 25.988755020080323, "grad_norm": 39.634151458740234, "learning_rate": 3.6993548387096776e-05, "loss": 0.4689, "step": 4032 }, { "epoch": 25.995180722891565, "grad_norm": 16.482114791870117, "learning_rate": 3.699032258064516e-05, "loss": 0.241, "step": 4033 }, { "epoch": 26.004819277108435, "grad_norm": 6.734865188598633, "learning_rate": 3.6987096774193554e-05, "loss": 0.2549, "step": 4034 }, { "epoch": 26.011244979919677, "grad_norm": 12.434769630432129, "learning_rate": 3.6983870967741933e-05, "loss": 0.4543, "step": 4035 }, { "epoch": 26.017670682730923, "grad_norm": 4.157763481140137, "learning_rate": 3.6980645161290326e-05, "loss": 0.2233, "step": 4036 }, { "epoch": 26.02409638554217, "grad_norm": 14.214927673339844, "learning_rate": 3.697741935483871e-05, "loss": 0.3916, "step": 4037 }, { "epoch": 26.030522088353415, "grad_norm": 39.090614318847656, "learning_rate": 3.69741935483871e-05, "loss": 0.3315, "step": 4038 }, { "epoch": 26.036947791164657, "grad_norm": 12.129817962646484, "learning_rate": 3.697096774193548e-05, "loss": 0.5528, "step": 4039 }, { "epoch": 26.043373493975903, "grad_norm": 5.870711803436279, "learning_rate": 3.6967741935483876e-05, "loss": 0.4162, "step": 4040 }, { "epoch": 26.04979919678715, "grad_norm": 15.525108337402344, "learning_rate": 3.696451612903226e-05, "loss": 0.4572, "step": 4041 }, { "epoch": 26.056224899598394, "grad_norm": 11.84371566772461, "learning_rate": 3.696129032258065e-05, "loss": 0.2103, "step": 4042 }, { "epoch": 26.06265060240964, "grad_norm": 12.466302871704102, "learning_rate": 3.695806451612903e-05, "loss": 0.4114, "step": 4043 }, { "epoch": 26.069076305220882, "grad_norm": 10.68691349029541, "learning_rate": 3.695483870967742e-05, "loss": 0.2024, "step": 4044 }, { "epoch": 26.075502008032128, "grad_norm": 13.88906192779541, "learning_rate": 3.6951612903225804e-05, "loss": 0.4182, "step": 4045 }, { "epoch": 26.081927710843374, "grad_norm": 1.8483012914657593, "learning_rate": 3.6948387096774197e-05, "loss": 0.6169, "step": 4046 }, { "epoch": 26.08835341365462, "grad_norm": 19.672792434692383, "learning_rate": 3.694516129032258e-05, "loss": 0.435, "step": 4047 }, { "epoch": 26.094779116465862, "grad_norm": 9.747788429260254, "learning_rate": 3.694193548387097e-05, "loss": 0.5335, "step": 4048 }, { "epoch": 26.101204819277108, "grad_norm": 15.086952209472656, "learning_rate": 3.693870967741936e-05, "loss": 0.2702, "step": 4049 }, { "epoch": 26.107630522088353, "grad_norm": 5.964535713195801, "learning_rate": 3.6935483870967746e-05, "loss": 0.4406, "step": 4050 }, { "epoch": 26.1140562248996, "grad_norm": 7.829090595245361, "learning_rate": 3.693225806451613e-05, "loss": 0.4404, "step": 4051 }, { "epoch": 26.120481927710845, "grad_norm": 18.27556800842285, "learning_rate": 3.692903225806452e-05, "loss": 0.404, "step": 4052 }, { "epoch": 26.126907630522087, "grad_norm": 13.064620971679688, "learning_rate": 3.69258064516129e-05, "loss": 0.3576, "step": 4053 }, { "epoch": 26.133333333333333, "grad_norm": 4.329887866973877, "learning_rate": 3.692258064516129e-05, "loss": 0.4031, "step": 4054 }, { "epoch": 26.13975903614458, "grad_norm": 12.285958290100098, "learning_rate": 3.691935483870968e-05, "loss": 0.3583, "step": 4055 }, { "epoch": 26.146184738955824, "grad_norm": 23.143766403198242, "learning_rate": 3.691612903225807e-05, "loss": 0.5711, "step": 4056 }, { "epoch": 26.152610441767067, "grad_norm": 6.937382698059082, "learning_rate": 3.691290322580645e-05, "loss": 0.3185, "step": 4057 }, { "epoch": 26.159036144578312, "grad_norm": 8.956275939941406, "learning_rate": 3.690967741935484e-05, "loss": 0.4925, "step": 4058 }, { "epoch": 26.165461847389558, "grad_norm": 15.297065734863281, "learning_rate": 3.6906451612903224e-05, "loss": 0.3109, "step": 4059 }, { "epoch": 26.171887550200804, "grad_norm": 15.820121765136719, "learning_rate": 3.690322580645162e-05, "loss": 0.5234, "step": 4060 }, { "epoch": 26.17831325301205, "grad_norm": 11.2079496383667, "learning_rate": 3.69e-05, "loss": 0.5488, "step": 4061 }, { "epoch": 26.184738955823292, "grad_norm": 11.202858924865723, "learning_rate": 3.689677419354839e-05, "loss": 0.6219, "step": 4062 }, { "epoch": 26.191164658634538, "grad_norm": 19.305017471313477, "learning_rate": 3.6893548387096774e-05, "loss": 0.5995, "step": 4063 }, { "epoch": 26.197590361445783, "grad_norm": 6.136499404907227, "learning_rate": 3.6890322580645166e-05, "loss": 0.0997, "step": 4064 }, { "epoch": 26.20401606425703, "grad_norm": 6.979678630828857, "learning_rate": 3.688709677419355e-05, "loss": 0.2905, "step": 4065 }, { "epoch": 26.21044176706827, "grad_norm": 4.4715399742126465, "learning_rate": 3.688387096774194e-05, "loss": 0.2437, "step": 4066 }, { "epoch": 26.216867469879517, "grad_norm": 12.850160598754883, "learning_rate": 3.6880645161290323e-05, "loss": 0.2263, "step": 4067 }, { "epoch": 26.223293172690763, "grad_norm": 19.5216007232666, "learning_rate": 3.687741935483871e-05, "loss": 0.5349, "step": 4068 }, { "epoch": 26.22971887550201, "grad_norm": 11.945704460144043, "learning_rate": 3.6874193548387095e-05, "loss": 0.1771, "step": 4069 }, { "epoch": 26.236144578313255, "grad_norm": 2.2464280128479004, "learning_rate": 3.687096774193549e-05, "loss": 0.2796, "step": 4070 }, { "epoch": 26.242570281124497, "grad_norm": 4.710880756378174, "learning_rate": 3.686774193548387e-05, "loss": 0.3643, "step": 4071 }, { "epoch": 26.248995983935743, "grad_norm": 20.255962371826172, "learning_rate": 3.686451612903226e-05, "loss": 0.391, "step": 4072 }, { "epoch": 26.25542168674699, "grad_norm": 10.187041282653809, "learning_rate": 3.686129032258065e-05, "loss": 0.6603, "step": 4073 }, { "epoch": 26.261847389558234, "grad_norm": 21.417911529541016, "learning_rate": 3.685806451612903e-05, "loss": 0.5803, "step": 4074 }, { "epoch": 26.268273092369476, "grad_norm": 11.71616268157959, "learning_rate": 3.685483870967742e-05, "loss": 0.4132, "step": 4075 }, { "epoch": 26.274698795180722, "grad_norm": 6.141506195068359, "learning_rate": 3.685161290322581e-05, "loss": 0.4127, "step": 4076 }, { "epoch": 26.281124497991968, "grad_norm": 8.097460746765137, "learning_rate": 3.6848387096774194e-05, "loss": 0.3407, "step": 4077 }, { "epoch": 26.287550200803214, "grad_norm": 22.13886070251465, "learning_rate": 3.684516129032258e-05, "loss": 0.36, "step": 4078 }, { "epoch": 26.29397590361446, "grad_norm": 14.532417297363281, "learning_rate": 3.684193548387097e-05, "loss": 0.5631, "step": 4079 }, { "epoch": 26.3004016064257, "grad_norm": 14.402106285095215, "learning_rate": 3.683870967741936e-05, "loss": 0.6169, "step": 4080 }, { "epoch": 26.306827309236947, "grad_norm": 8.808382987976074, "learning_rate": 3.6835483870967744e-05, "loss": 0.4475, "step": 4081 }, { "epoch": 26.313253012048193, "grad_norm": 11.68379020690918, "learning_rate": 3.683225806451613e-05, "loss": 0.1615, "step": 4082 }, { "epoch": 26.31967871485944, "grad_norm": 9.685131072998047, "learning_rate": 3.6829032258064515e-05, "loss": 0.4612, "step": 4083 }, { "epoch": 26.32610441767068, "grad_norm": 3.6677086353302, "learning_rate": 3.682580645161291e-05, "loss": 0.2059, "step": 4084 }, { "epoch": 26.332530120481927, "grad_norm": 1.551535725593567, "learning_rate": 3.682258064516129e-05, "loss": 0.4483, "step": 4085 }, { "epoch": 26.338955823293173, "grad_norm": 15.927703857421875, "learning_rate": 3.681935483870968e-05, "loss": 0.2188, "step": 4086 }, { "epoch": 26.34538152610442, "grad_norm": 5.317329406738281, "learning_rate": 3.6816129032258065e-05, "loss": 0.1403, "step": 4087 }, { "epoch": 26.351807228915664, "grad_norm": 6.312209129333496, "learning_rate": 3.681290322580646e-05, "loss": 0.2585, "step": 4088 }, { "epoch": 26.358232931726906, "grad_norm": 13.135560989379883, "learning_rate": 3.6809677419354836e-05, "loss": 0.2254, "step": 4089 }, { "epoch": 26.364658634538152, "grad_norm": 3.94274640083313, "learning_rate": 3.680645161290323e-05, "loss": 0.3956, "step": 4090 }, { "epoch": 26.371084337349398, "grad_norm": 24.320985794067383, "learning_rate": 3.6803225806451614e-05, "loss": 0.7559, "step": 4091 }, { "epoch": 26.377510040160644, "grad_norm": 37.74164581298828, "learning_rate": 3.68e-05, "loss": 0.4924, "step": 4092 }, { "epoch": 26.383935742971886, "grad_norm": 8.344623565673828, "learning_rate": 3.6796774193548386e-05, "loss": 0.2938, "step": 4093 }, { "epoch": 26.39036144578313, "grad_norm": 25.99919319152832, "learning_rate": 3.679354838709678e-05, "loss": 0.6432, "step": 4094 }, { "epoch": 26.396787148594377, "grad_norm": 12.236382484436035, "learning_rate": 3.6790322580645164e-05, "loss": 0.4289, "step": 4095 }, { "epoch": 26.403212851405623, "grad_norm": 7.473093032836914, "learning_rate": 3.678709677419355e-05, "loss": 0.4424, "step": 4096 }, { "epoch": 26.40963855421687, "grad_norm": 18.1293888092041, "learning_rate": 3.6783870967741935e-05, "loss": 0.333, "step": 4097 }, { "epoch": 26.41606425702811, "grad_norm": 13.852118492126465, "learning_rate": 3.678064516129032e-05, "loss": 0.351, "step": 4098 }, { "epoch": 26.422489959839357, "grad_norm": 50.05815505981445, "learning_rate": 3.6777419354838714e-05, "loss": 0.2159, "step": 4099 }, { "epoch": 26.428915662650603, "grad_norm": 23.023826599121094, "learning_rate": 3.67741935483871e-05, "loss": 0.3855, "step": 4100 }, { "epoch": 26.43534136546185, "grad_norm": 18.22159767150879, "learning_rate": 3.6770967741935485e-05, "loss": 0.6342, "step": 4101 }, { "epoch": 26.44176706827309, "grad_norm": 11.289281845092773, "learning_rate": 3.676774193548387e-05, "loss": 0.2532, "step": 4102 }, { "epoch": 26.448192771084337, "grad_norm": 5.7099714279174805, "learning_rate": 3.676451612903226e-05, "loss": 0.2596, "step": 4103 }, { "epoch": 26.454618473895582, "grad_norm": 38.302268981933594, "learning_rate": 3.676129032258065e-05, "loss": 0.4966, "step": 4104 }, { "epoch": 26.461044176706828, "grad_norm": 5.014087677001953, "learning_rate": 3.6758064516129035e-05, "loss": 0.2048, "step": 4105 }, { "epoch": 26.467469879518074, "grad_norm": 6.826943874359131, "learning_rate": 3.675483870967742e-05, "loss": 0.2242, "step": 4106 }, { "epoch": 26.473895582329316, "grad_norm": 11.199810028076172, "learning_rate": 3.6751612903225806e-05, "loss": 0.4949, "step": 4107 }, { "epoch": 26.480321285140562, "grad_norm": 24.269317626953125, "learning_rate": 3.674838709677419e-05, "loss": 0.4432, "step": 4108 }, { "epoch": 26.486746987951808, "grad_norm": 18.493675231933594, "learning_rate": 3.6745161290322584e-05, "loss": 0.3726, "step": 4109 }, { "epoch": 26.493172690763053, "grad_norm": 31.52870750427246, "learning_rate": 3.674193548387097e-05, "loss": 0.3682, "step": 4110 }, { "epoch": 26.499598393574296, "grad_norm": 22.846542358398438, "learning_rate": 3.6738709677419356e-05, "loss": 0.495, "step": 4111 }, { "epoch": 26.50602409638554, "grad_norm": 18.490676879882812, "learning_rate": 3.673548387096775e-05, "loss": 0.5803, "step": 4112 }, { "epoch": 26.512449799196787, "grad_norm": 26.6693058013916, "learning_rate": 3.673225806451613e-05, "loss": 0.2869, "step": 4113 }, { "epoch": 26.518875502008033, "grad_norm": 12.527046203613281, "learning_rate": 3.672903225806452e-05, "loss": 0.3127, "step": 4114 }, { "epoch": 26.52530120481928, "grad_norm": 5.780581474304199, "learning_rate": 3.6725806451612905e-05, "loss": 0.2857, "step": 4115 }, { "epoch": 26.53172690763052, "grad_norm": 13.224693298339844, "learning_rate": 3.672258064516129e-05, "loss": 0.246, "step": 4116 }, { "epoch": 26.538152610441767, "grad_norm": 12.415292739868164, "learning_rate": 3.6719354838709677e-05, "loss": 0.4923, "step": 4117 }, { "epoch": 26.544578313253012, "grad_norm": 18.3674259185791, "learning_rate": 3.671612903225807e-05, "loss": 0.6207, "step": 4118 }, { "epoch": 26.551004016064258, "grad_norm": 8.771575927734375, "learning_rate": 3.6712903225806455e-05, "loss": 0.5926, "step": 4119 }, { "epoch": 26.5574297188755, "grad_norm": 9.403834342956543, "learning_rate": 3.670967741935484e-05, "loss": 0.1802, "step": 4120 }, { "epoch": 26.563855421686746, "grad_norm": 7.591729640960693, "learning_rate": 3.6706451612903226e-05, "loss": 0.4565, "step": 4121 }, { "epoch": 26.570281124497992, "grad_norm": 49.91875076293945, "learning_rate": 3.670322580645161e-05, "loss": 0.397, "step": 4122 }, { "epoch": 26.576706827309238, "grad_norm": 23.01400375366211, "learning_rate": 3.6700000000000004e-05, "loss": 0.4818, "step": 4123 }, { "epoch": 26.583132530120483, "grad_norm": 15.777756690979004, "learning_rate": 3.669677419354839e-05, "loss": 0.6331, "step": 4124 }, { "epoch": 26.589558232931726, "grad_norm": 10.970367431640625, "learning_rate": 3.6693548387096776e-05, "loss": 0.5726, "step": 4125 }, { "epoch": 26.59598393574297, "grad_norm": 6.700285911560059, "learning_rate": 3.669032258064516e-05, "loss": 0.3167, "step": 4126 }, { "epoch": 26.602409638554217, "grad_norm": 25.86379623413086, "learning_rate": 3.6687096774193554e-05, "loss": 0.5568, "step": 4127 }, { "epoch": 26.608835341365463, "grad_norm": 11.878816604614258, "learning_rate": 3.668387096774193e-05, "loss": 0.3831, "step": 4128 }, { "epoch": 26.615261044176705, "grad_norm": 15.033171653747559, "learning_rate": 3.6680645161290325e-05, "loss": 0.4668, "step": 4129 }, { "epoch": 26.62168674698795, "grad_norm": 5.569443702697754, "learning_rate": 3.667741935483871e-05, "loss": 0.3934, "step": 4130 }, { "epoch": 26.628112449799197, "grad_norm": 7.361006736755371, "learning_rate": 3.66741935483871e-05, "loss": 0.2159, "step": 4131 }, { "epoch": 26.634538152610443, "grad_norm": 12.217090606689453, "learning_rate": 3.667096774193548e-05, "loss": 0.2241, "step": 4132 }, { "epoch": 26.64096385542169, "grad_norm": 8.200261116027832, "learning_rate": 3.6667741935483875e-05, "loss": 0.1592, "step": 4133 }, { "epoch": 26.64738955823293, "grad_norm": 7.819206237792969, "learning_rate": 3.666451612903226e-05, "loss": 0.5207, "step": 4134 }, { "epoch": 26.653815261044176, "grad_norm": 9.05933952331543, "learning_rate": 3.6661290322580646e-05, "loss": 0.259, "step": 4135 }, { "epoch": 26.660240963855422, "grad_norm": 14.790998458862305, "learning_rate": 3.665806451612903e-05, "loss": 0.423, "step": 4136 }, { "epoch": 26.666666666666668, "grad_norm": 17.647274017333984, "learning_rate": 3.665483870967742e-05, "loss": 0.5067, "step": 4137 }, { "epoch": 26.67309236947791, "grad_norm": 26.77994728088379, "learning_rate": 3.665161290322581e-05, "loss": 0.1762, "step": 4138 }, { "epoch": 26.679518072289156, "grad_norm": 17.892826080322266, "learning_rate": 3.6648387096774196e-05, "loss": 0.4034, "step": 4139 }, { "epoch": 26.6859437751004, "grad_norm": 9.800741195678711, "learning_rate": 3.664516129032258e-05, "loss": 0.3673, "step": 4140 }, { "epoch": 26.692369477911647, "grad_norm": 14.978422164916992, "learning_rate": 3.664193548387097e-05, "loss": 0.5954, "step": 4141 }, { "epoch": 26.698795180722893, "grad_norm": 8.4230318069458, "learning_rate": 3.663870967741936e-05, "loss": 0.2439, "step": 4142 }, { "epoch": 26.705220883534135, "grad_norm": 3.7257070541381836, "learning_rate": 3.6635483870967746e-05, "loss": 0.4635, "step": 4143 }, { "epoch": 26.71164658634538, "grad_norm": 13.033432006835938, "learning_rate": 3.663225806451613e-05, "loss": 0.561, "step": 4144 }, { "epoch": 26.718072289156627, "grad_norm": 13.433618545532227, "learning_rate": 3.662903225806452e-05, "loss": 0.6744, "step": 4145 }, { "epoch": 26.724497991967873, "grad_norm": 33.33242416381836, "learning_rate": 3.66258064516129e-05, "loss": 0.5156, "step": 4146 }, { "epoch": 26.730923694779115, "grad_norm": 11.30380630493164, "learning_rate": 3.6622580645161295e-05, "loss": 0.4248, "step": 4147 }, { "epoch": 26.73734939759036, "grad_norm": 10.879453659057617, "learning_rate": 3.661935483870968e-05, "loss": 0.4093, "step": 4148 }, { "epoch": 26.743775100401606, "grad_norm": 12.208602905273438, "learning_rate": 3.661612903225807e-05, "loss": 0.6133, "step": 4149 }, { "epoch": 26.750200803212852, "grad_norm": 10.823945999145508, "learning_rate": 3.661290322580645e-05, "loss": 0.3148, "step": 4150 }, { "epoch": 26.756626506024098, "grad_norm": 13.647924423217773, "learning_rate": 3.6609677419354845e-05, "loss": 0.605, "step": 4151 }, { "epoch": 26.76305220883534, "grad_norm": 26.50774383544922, "learning_rate": 3.6606451612903224e-05, "loss": 0.5738, "step": 4152 }, { "epoch": 26.769477911646586, "grad_norm": 26.47618293762207, "learning_rate": 3.6603225806451616e-05, "loss": 0.3534, "step": 4153 }, { "epoch": 26.77590361445783, "grad_norm": 11.444644927978516, "learning_rate": 3.66e-05, "loss": 0.2068, "step": 4154 }, { "epoch": 26.782329317269078, "grad_norm": 9.312736511230469, "learning_rate": 3.659677419354839e-05, "loss": 0.1915, "step": 4155 }, { "epoch": 26.78875502008032, "grad_norm": 27.823551177978516, "learning_rate": 3.659354838709677e-05, "loss": 0.4936, "step": 4156 }, { "epoch": 26.795180722891565, "grad_norm": 23.66026496887207, "learning_rate": 3.6590322580645166e-05, "loss": 0.2243, "step": 4157 }, { "epoch": 26.80160642570281, "grad_norm": 15.176197052001953, "learning_rate": 3.658709677419355e-05, "loss": 0.272, "step": 4158 }, { "epoch": 26.808032128514057, "grad_norm": 14.177088737487793, "learning_rate": 3.658387096774194e-05, "loss": 0.2839, "step": 4159 }, { "epoch": 26.814457831325303, "grad_norm": 8.05355167388916, "learning_rate": 3.658064516129032e-05, "loss": 0.4411, "step": 4160 }, { "epoch": 26.820883534136545, "grad_norm": 17.31687355041504, "learning_rate": 3.657741935483871e-05, "loss": 0.492, "step": 4161 }, { "epoch": 26.82730923694779, "grad_norm": 8.907538414001465, "learning_rate": 3.65741935483871e-05, "loss": 0.4618, "step": 4162 }, { "epoch": 26.833734939759037, "grad_norm": 29.080583572387695, "learning_rate": 3.657096774193549e-05, "loss": 0.5253, "step": 4163 }, { "epoch": 26.840160642570282, "grad_norm": 3.589393377304077, "learning_rate": 3.656774193548387e-05, "loss": 0.4254, "step": 4164 }, { "epoch": 26.846586345381525, "grad_norm": 3.899044990539551, "learning_rate": 3.656451612903226e-05, "loss": 0.3519, "step": 4165 }, { "epoch": 26.85301204819277, "grad_norm": 6.502165794372559, "learning_rate": 3.656129032258065e-05, "loss": 0.2075, "step": 4166 }, { "epoch": 26.859437751004016, "grad_norm": 11.762040138244629, "learning_rate": 3.655806451612903e-05, "loss": 0.5108, "step": 4167 }, { "epoch": 26.865863453815262, "grad_norm": 8.363067626953125, "learning_rate": 3.655483870967742e-05, "loss": 0.313, "step": 4168 }, { "epoch": 26.872289156626508, "grad_norm": 23.653657913208008, "learning_rate": 3.655161290322581e-05, "loss": 0.3884, "step": 4169 }, { "epoch": 26.87871485943775, "grad_norm": 33.05450439453125, "learning_rate": 3.6548387096774194e-05, "loss": 0.3827, "step": 4170 }, { "epoch": 26.885140562248996, "grad_norm": 23.378171920776367, "learning_rate": 3.654516129032258e-05, "loss": 0.1331, "step": 4171 }, { "epoch": 26.89156626506024, "grad_norm": 3.934807777404785, "learning_rate": 3.654193548387097e-05, "loss": 0.3292, "step": 4172 }, { "epoch": 26.897991967871487, "grad_norm": 10.304234504699707, "learning_rate": 3.653870967741936e-05, "loss": 0.3116, "step": 4173 }, { "epoch": 26.90441767068273, "grad_norm": 4.033294200897217, "learning_rate": 3.653548387096774e-05, "loss": 0.178, "step": 4174 }, { "epoch": 26.910843373493975, "grad_norm": 38.633811950683594, "learning_rate": 3.653225806451613e-05, "loss": 0.3843, "step": 4175 }, { "epoch": 26.91726907630522, "grad_norm": 28.91898536682129, "learning_rate": 3.6529032258064515e-05, "loss": 0.4663, "step": 4176 }, { "epoch": 26.923694779116467, "grad_norm": 5.520235538482666, "learning_rate": 3.652580645161291e-05, "loss": 0.2932, "step": 4177 }, { "epoch": 26.930120481927712, "grad_norm": 7.3311262130737305, "learning_rate": 3.652258064516129e-05, "loss": 0.4938, "step": 4178 }, { "epoch": 26.936546184738955, "grad_norm": 19.662939071655273, "learning_rate": 3.651935483870968e-05, "loss": 0.6419, "step": 4179 }, { "epoch": 26.9429718875502, "grad_norm": 12.424399375915527, "learning_rate": 3.6516129032258064e-05, "loss": 0.3389, "step": 4180 }, { "epoch": 26.949397590361446, "grad_norm": 7.506296634674072, "learning_rate": 3.651290322580646e-05, "loss": 0.2266, "step": 4181 }, { "epoch": 26.955823293172692, "grad_norm": 7.772710800170898, "learning_rate": 3.650967741935484e-05, "loss": 0.3417, "step": 4182 }, { "epoch": 26.962248995983934, "grad_norm": 17.711257934570312, "learning_rate": 3.650645161290323e-05, "loss": 0.4435, "step": 4183 }, { "epoch": 26.96867469879518, "grad_norm": 15.408105850219727, "learning_rate": 3.6503225806451614e-05, "loss": 0.3832, "step": 4184 }, { "epoch": 26.975100401606426, "grad_norm": 18.22599983215332, "learning_rate": 3.65e-05, "loss": 0.4079, "step": 4185 }, { "epoch": 26.98152610441767, "grad_norm": 58.068416595458984, "learning_rate": 3.649677419354839e-05, "loss": 0.5016, "step": 4186 }, { "epoch": 26.987951807228917, "grad_norm": 7.384829044342041, "learning_rate": 3.649354838709678e-05, "loss": 0.2557, "step": 4187 }, { "epoch": 26.99437751004016, "grad_norm": 22.858884811401367, "learning_rate": 3.6490322580645163e-05, "loss": 0.2491, "step": 4188 }, { "epoch": 27.004016064257026, "grad_norm": 27.504348754882812, "learning_rate": 3.648709677419355e-05, "loss": 0.9517, "step": 4189 }, { "epoch": 27.010441767068272, "grad_norm": 13.137199401855469, "learning_rate": 3.648387096774194e-05, "loss": 0.4121, "step": 4190 }, { "epoch": 27.016867469879518, "grad_norm": 20.050344467163086, "learning_rate": 3.648064516129032e-05, "loss": 0.5115, "step": 4191 }, { "epoch": 27.023293172690764, "grad_norm": 11.354583740234375, "learning_rate": 3.647741935483871e-05, "loss": 0.491, "step": 4192 }, { "epoch": 27.02971887550201, "grad_norm": 8.213598251342773, "learning_rate": 3.64741935483871e-05, "loss": 0.2527, "step": 4193 }, { "epoch": 27.03614457831325, "grad_norm": 1.1976649761199951, "learning_rate": 3.6470967741935484e-05, "loss": 0.1138, "step": 4194 }, { "epoch": 27.042570281124497, "grad_norm": 21.51055145263672, "learning_rate": 3.646774193548387e-05, "loss": 0.196, "step": 4195 }, { "epoch": 27.048995983935743, "grad_norm": 24.403728485107422, "learning_rate": 3.646451612903226e-05, "loss": 0.2954, "step": 4196 }, { "epoch": 27.05542168674699, "grad_norm": 10.647163391113281, "learning_rate": 3.646129032258065e-05, "loss": 0.3686, "step": 4197 }, { "epoch": 27.06184738955823, "grad_norm": 7.548893451690674, "learning_rate": 3.6458064516129034e-05, "loss": 0.6108, "step": 4198 }, { "epoch": 27.068273092369477, "grad_norm": 12.78284740447998, "learning_rate": 3.645483870967742e-05, "loss": 0.3193, "step": 4199 }, { "epoch": 27.074698795180723, "grad_norm": 5.065481185913086, "learning_rate": 3.6451612903225805e-05, "loss": 0.2617, "step": 4200 }, { "epoch": 27.08112449799197, "grad_norm": 6.324392318725586, "learning_rate": 3.64483870967742e-05, "loss": 0.3891, "step": 4201 }, { "epoch": 27.087550200803214, "grad_norm": 39.798091888427734, "learning_rate": 3.6445161290322584e-05, "loss": 0.185, "step": 4202 }, { "epoch": 27.093975903614457, "grad_norm": 13.330360412597656, "learning_rate": 3.644193548387097e-05, "loss": 0.179, "step": 4203 }, { "epoch": 27.100401606425702, "grad_norm": 23.57560157775879, "learning_rate": 3.6438709677419355e-05, "loss": 0.3428, "step": 4204 }, { "epoch": 27.106827309236948, "grad_norm": 4.774203300476074, "learning_rate": 3.643548387096775e-05, "loss": 0.3351, "step": 4205 }, { "epoch": 27.113253012048194, "grad_norm": 9.83584976196289, "learning_rate": 3.6432258064516126e-05, "loss": 0.2034, "step": 4206 }, { "epoch": 27.119678714859436, "grad_norm": 6.521700382232666, "learning_rate": 3.642903225806452e-05, "loss": 0.4428, "step": 4207 }, { "epoch": 27.126104417670682, "grad_norm": 13.767409324645996, "learning_rate": 3.6425806451612905e-05, "loss": 0.6469, "step": 4208 }, { "epoch": 27.132530120481928, "grad_norm": 15.518377304077148, "learning_rate": 3.642258064516129e-05, "loss": 0.4428, "step": 4209 }, { "epoch": 27.138955823293173, "grad_norm": 13.661438941955566, "learning_rate": 3.641935483870968e-05, "loss": 0.404, "step": 4210 }, { "epoch": 27.14538152610442, "grad_norm": 12.477380752563477, "learning_rate": 3.641612903225807e-05, "loss": 0.1645, "step": 4211 }, { "epoch": 27.15180722891566, "grad_norm": 7.032232284545898, "learning_rate": 3.6412903225806454e-05, "loss": 0.2264, "step": 4212 }, { "epoch": 27.158232931726907, "grad_norm": 7.768043518066406, "learning_rate": 3.640967741935484e-05, "loss": 0.3648, "step": 4213 }, { "epoch": 27.164658634538153, "grad_norm": 15.558004379272461, "learning_rate": 3.6406451612903226e-05, "loss": 0.3911, "step": 4214 }, { "epoch": 27.1710843373494, "grad_norm": 18.29813575744629, "learning_rate": 3.640322580645161e-05, "loss": 0.4616, "step": 4215 }, { "epoch": 27.17751004016064, "grad_norm": 13.815125465393066, "learning_rate": 3.6400000000000004e-05, "loss": 0.1635, "step": 4216 }, { "epoch": 27.183935742971887, "grad_norm": 7.796751499176025, "learning_rate": 3.639677419354839e-05, "loss": 0.6381, "step": 4217 }, { "epoch": 27.190361445783132, "grad_norm": 14.241780281066895, "learning_rate": 3.6393548387096775e-05, "loss": 0.4345, "step": 4218 }, { "epoch": 27.196787148594378, "grad_norm": 13.495820999145508, "learning_rate": 3.639032258064516e-05, "loss": 0.2261, "step": 4219 }, { "epoch": 27.203212851405624, "grad_norm": 7.695445537567139, "learning_rate": 3.6387096774193553e-05, "loss": 0.2042, "step": 4220 }, { "epoch": 27.209638554216866, "grad_norm": 16.641725540161133, "learning_rate": 3.638387096774194e-05, "loss": 0.3738, "step": 4221 }, { "epoch": 27.216064257028112, "grad_norm": 15.776366233825684, "learning_rate": 3.6380645161290325e-05, "loss": 0.6748, "step": 4222 }, { "epoch": 27.222489959839358, "grad_norm": 25.045276641845703, "learning_rate": 3.637741935483871e-05, "loss": 0.717, "step": 4223 }, { "epoch": 27.228915662650603, "grad_norm": 4.632023811340332, "learning_rate": 3.6374193548387096e-05, "loss": 0.2181, "step": 4224 }, { "epoch": 27.235341365461846, "grad_norm": 11.070460319519043, "learning_rate": 3.637096774193549e-05, "loss": 0.2483, "step": 4225 }, { "epoch": 27.24176706827309, "grad_norm": 32.10017776489258, "learning_rate": 3.6367741935483874e-05, "loss": 0.5497, "step": 4226 }, { "epoch": 27.248192771084337, "grad_norm": 83.93248748779297, "learning_rate": 3.636451612903226e-05, "loss": 0.5839, "step": 4227 }, { "epoch": 27.254618473895583, "grad_norm": 5.37335729598999, "learning_rate": 3.6361290322580646e-05, "loss": 0.5037, "step": 4228 }, { "epoch": 27.26104417670683, "grad_norm": 6.760342121124268, "learning_rate": 3.635806451612904e-05, "loss": 0.4476, "step": 4229 }, { "epoch": 27.26746987951807, "grad_norm": 29.417898178100586, "learning_rate": 3.635483870967742e-05, "loss": 0.6961, "step": 4230 }, { "epoch": 27.273895582329317, "grad_norm": 5.987867832183838, "learning_rate": 3.635161290322581e-05, "loss": 0.3074, "step": 4231 }, { "epoch": 27.280321285140563, "grad_norm": 5.516406059265137, "learning_rate": 3.6348387096774195e-05, "loss": 0.2041, "step": 4232 }, { "epoch": 27.28674698795181, "grad_norm": 19.469554901123047, "learning_rate": 3.634516129032258e-05, "loss": 0.6247, "step": 4233 }, { "epoch": 27.29317269076305, "grad_norm": 3.552039384841919, "learning_rate": 3.634193548387097e-05, "loss": 0.2914, "step": 4234 }, { "epoch": 27.299598393574296, "grad_norm": 23.875335693359375, "learning_rate": 3.633870967741936e-05, "loss": 0.4512, "step": 4235 }, { "epoch": 27.306024096385542, "grad_norm": 4.76434326171875, "learning_rate": 3.6335483870967745e-05, "loss": 0.4471, "step": 4236 }, { "epoch": 27.312449799196788, "grad_norm": 35.473575592041016, "learning_rate": 3.633225806451613e-05, "loss": 0.351, "step": 4237 }, { "epoch": 27.318875502008034, "grad_norm": 6.143667221069336, "learning_rate": 3.6329032258064517e-05, "loss": 0.364, "step": 4238 }, { "epoch": 27.325301204819276, "grad_norm": 16.36777114868164, "learning_rate": 3.63258064516129e-05, "loss": 0.4786, "step": 4239 }, { "epoch": 27.33172690763052, "grad_norm": 2.903956651687622, "learning_rate": 3.6322580645161295e-05, "loss": 0.1977, "step": 4240 }, { "epoch": 27.338152610441767, "grad_norm": 7.947327613830566, "learning_rate": 3.631935483870968e-05, "loss": 0.5789, "step": 4241 }, { "epoch": 27.344578313253013, "grad_norm": 30.677623748779297, "learning_rate": 3.6316129032258066e-05, "loss": 0.1752, "step": 4242 }, { "epoch": 27.351004016064255, "grad_norm": 2.4974868297576904, "learning_rate": 3.631290322580645e-05, "loss": 0.0794, "step": 4243 }, { "epoch": 27.3574297188755, "grad_norm": 39.242095947265625, "learning_rate": 3.6309677419354844e-05, "loss": 0.3559, "step": 4244 }, { "epoch": 27.363855421686747, "grad_norm": 7.027824878692627, "learning_rate": 3.630645161290322e-05, "loss": 0.22, "step": 4245 }, { "epoch": 27.370281124497993, "grad_norm": 8.365204811096191, "learning_rate": 3.6303225806451616e-05, "loss": 0.295, "step": 4246 }, { "epoch": 27.37670682730924, "grad_norm": 10.571039199829102, "learning_rate": 3.63e-05, "loss": 0.7932, "step": 4247 }, { "epoch": 27.38313253012048, "grad_norm": 45.13370895385742, "learning_rate": 3.629677419354839e-05, "loss": 0.5799, "step": 4248 }, { "epoch": 27.389558232931726, "grad_norm": 15.486091613769531, "learning_rate": 3.629354838709678e-05, "loss": 0.3418, "step": 4249 }, { "epoch": 27.395983935742972, "grad_norm": 14.900108337402344, "learning_rate": 3.6290322580645165e-05, "loss": 0.5158, "step": 4250 }, { "epoch": 27.402409638554218, "grad_norm": 5.339759826660156, "learning_rate": 3.628709677419355e-05, "loss": 0.3138, "step": 4251 }, { "epoch": 27.40883534136546, "grad_norm": 26.73668670654297, "learning_rate": 3.628387096774194e-05, "loss": 0.4321, "step": 4252 }, { "epoch": 27.415261044176706, "grad_norm": 1.5949734449386597, "learning_rate": 3.628064516129032e-05, "loss": 0.5244, "step": 4253 }, { "epoch": 27.42168674698795, "grad_norm": 8.757027626037598, "learning_rate": 3.627741935483871e-05, "loss": 0.2639, "step": 4254 }, { "epoch": 27.428112449799197, "grad_norm": 20.91477394104004, "learning_rate": 3.62741935483871e-05, "loss": 0.7739, "step": 4255 }, { "epoch": 27.434538152610443, "grad_norm": 13.475547790527344, "learning_rate": 3.6270967741935486e-05, "loss": 0.5294, "step": 4256 }, { "epoch": 27.440963855421685, "grad_norm": 8.792112350463867, "learning_rate": 3.626774193548387e-05, "loss": 0.3224, "step": 4257 }, { "epoch": 27.44738955823293, "grad_norm": 18.92685890197754, "learning_rate": 3.626451612903226e-05, "loss": 0.2912, "step": 4258 }, { "epoch": 27.453815261044177, "grad_norm": 32.36457443237305, "learning_rate": 3.626129032258065e-05, "loss": 0.3925, "step": 4259 }, { "epoch": 27.460240963855423, "grad_norm": 16.610652923583984, "learning_rate": 3.6258064516129036e-05, "loss": 0.5223, "step": 4260 }, { "epoch": 27.466666666666665, "grad_norm": 6.309193134307861, "learning_rate": 3.625483870967742e-05, "loss": 0.3135, "step": 4261 }, { "epoch": 27.47309236947791, "grad_norm": 12.987339973449707, "learning_rate": 3.625161290322581e-05, "loss": 0.3729, "step": 4262 }, { "epoch": 27.479518072289157, "grad_norm": 14.611266136169434, "learning_rate": 3.624838709677419e-05, "loss": 0.4312, "step": 4263 }, { "epoch": 27.485943775100402, "grad_norm": 9.30414867401123, "learning_rate": 3.6245161290322586e-05, "loss": 0.2627, "step": 4264 }, { "epoch": 27.492369477911648, "grad_norm": 14.64678955078125, "learning_rate": 3.624193548387097e-05, "loss": 0.6301, "step": 4265 }, { "epoch": 27.49879518072289, "grad_norm": 25.288637161254883, "learning_rate": 3.623870967741936e-05, "loss": 0.3511, "step": 4266 }, { "epoch": 27.505220883534136, "grad_norm": 13.984898567199707, "learning_rate": 3.623548387096774e-05, "loss": 0.4006, "step": 4267 }, { "epoch": 27.511646586345382, "grad_norm": 11.492188453674316, "learning_rate": 3.623225806451613e-05, "loss": 0.4272, "step": 4268 }, { "epoch": 27.518072289156628, "grad_norm": 12.745936393737793, "learning_rate": 3.6229032258064514e-05, "loss": 0.4686, "step": 4269 }, { "epoch": 27.52449799196787, "grad_norm": 7.512050151824951, "learning_rate": 3.6225806451612907e-05, "loss": 0.327, "step": 4270 }, { "epoch": 27.530923694779116, "grad_norm": 12.419880867004395, "learning_rate": 3.622258064516129e-05, "loss": 0.3476, "step": 4271 }, { "epoch": 27.53734939759036, "grad_norm": 11.189887046813965, "learning_rate": 3.621935483870968e-05, "loss": 0.4341, "step": 4272 }, { "epoch": 27.543775100401607, "grad_norm": 21.083494186401367, "learning_rate": 3.621612903225807e-05, "loss": 0.4593, "step": 4273 }, { "epoch": 27.550200803212853, "grad_norm": 8.139703750610352, "learning_rate": 3.6212903225806456e-05, "loss": 0.4212, "step": 4274 }, { "epoch": 27.556626506024095, "grad_norm": 20.878355026245117, "learning_rate": 3.620967741935484e-05, "loss": 0.5372, "step": 4275 }, { "epoch": 27.56305220883534, "grad_norm": 6.255265712738037, "learning_rate": 3.620645161290323e-05, "loss": 0.2603, "step": 4276 }, { "epoch": 27.569477911646587, "grad_norm": 44.34809875488281, "learning_rate": 3.620322580645161e-05, "loss": 0.5075, "step": 4277 }, { "epoch": 27.575903614457832, "grad_norm": 18.32000160217285, "learning_rate": 3.62e-05, "loss": 0.6333, "step": 4278 }, { "epoch": 27.582329317269075, "grad_norm": 14.12602710723877, "learning_rate": 3.619677419354839e-05, "loss": 0.6106, "step": 4279 }, { "epoch": 27.58875502008032, "grad_norm": 9.476492881774902, "learning_rate": 3.619354838709678e-05, "loss": 0.3925, "step": 4280 }, { "epoch": 27.595180722891566, "grad_norm": 8.164690017700195, "learning_rate": 3.619032258064516e-05, "loss": 0.5957, "step": 4281 }, { "epoch": 27.601606425702812, "grad_norm": 10.530028343200684, "learning_rate": 3.618709677419355e-05, "loss": 0.4052, "step": 4282 }, { "epoch": 27.608032128514058, "grad_norm": 14.371582984924316, "learning_rate": 3.618387096774194e-05, "loss": 0.3383, "step": 4283 }, { "epoch": 27.6144578313253, "grad_norm": 14.989771842956543, "learning_rate": 3.618064516129032e-05, "loss": 0.5085, "step": 4284 }, { "epoch": 27.620883534136546, "grad_norm": 15.24472427368164, "learning_rate": 3.617741935483871e-05, "loss": 0.6995, "step": 4285 }, { "epoch": 27.62730923694779, "grad_norm": 5.476454734802246, "learning_rate": 3.61741935483871e-05, "loss": 0.2579, "step": 4286 }, { "epoch": 27.633734939759037, "grad_norm": 14.837824821472168, "learning_rate": 3.6170967741935484e-05, "loss": 0.3322, "step": 4287 }, { "epoch": 27.64016064257028, "grad_norm": 8.518501281738281, "learning_rate": 3.6167741935483876e-05, "loss": 0.2903, "step": 4288 }, { "epoch": 27.646586345381525, "grad_norm": 25.392595291137695, "learning_rate": 3.616451612903226e-05, "loss": 0.2916, "step": 4289 }, { "epoch": 27.65301204819277, "grad_norm": 14.082252502441406, "learning_rate": 3.616129032258065e-05, "loss": 0.5415, "step": 4290 }, { "epoch": 27.659437751004017, "grad_norm": 14.505256652832031, "learning_rate": 3.6158064516129034e-05, "loss": 0.4704, "step": 4291 }, { "epoch": 27.665863453815263, "grad_norm": 9.718148231506348, "learning_rate": 3.615483870967742e-05, "loss": 0.2676, "step": 4292 }, { "epoch": 27.672289156626505, "grad_norm": 9.553210258483887, "learning_rate": 3.6151612903225805e-05, "loss": 0.4382, "step": 4293 }, { "epoch": 27.67871485943775, "grad_norm": 4.367851257324219, "learning_rate": 3.61483870967742e-05, "loss": 0.3649, "step": 4294 }, { "epoch": 27.685140562248996, "grad_norm": 19.260889053344727, "learning_rate": 3.614516129032258e-05, "loss": 0.4571, "step": 4295 }, { "epoch": 27.691566265060242, "grad_norm": 6.617330551147461, "learning_rate": 3.614193548387097e-05, "loss": 0.6796, "step": 4296 }, { "epoch": 27.697991967871484, "grad_norm": 11.828193664550781, "learning_rate": 3.6138709677419355e-05, "loss": 0.5028, "step": 4297 }, { "epoch": 27.70441767068273, "grad_norm": 3.643627643585205, "learning_rate": 3.613548387096775e-05, "loss": 0.455, "step": 4298 }, { "epoch": 27.710843373493976, "grad_norm": 21.114778518676758, "learning_rate": 3.613225806451613e-05, "loss": 0.4533, "step": 4299 }, { "epoch": 27.71726907630522, "grad_norm": 4.696392059326172, "learning_rate": 3.612903225806452e-05, "loss": 0.444, "step": 4300 }, { "epoch": 27.723694779116467, "grad_norm": 21.568145751953125, "learning_rate": 3.6125806451612904e-05, "loss": 0.407, "step": 4301 }, { "epoch": 27.73012048192771, "grad_norm": 8.853513717651367, "learning_rate": 3.612258064516129e-05, "loss": 0.474, "step": 4302 }, { "epoch": 27.736546184738955, "grad_norm": 4.583315372467041, "learning_rate": 3.611935483870968e-05, "loss": 0.1964, "step": 4303 }, { "epoch": 27.7429718875502, "grad_norm": 17.80683135986328, "learning_rate": 3.611612903225807e-05, "loss": 0.355, "step": 4304 }, { "epoch": 27.749397590361447, "grad_norm": 4.549399375915527, "learning_rate": 3.6112903225806454e-05, "loss": 0.1827, "step": 4305 }, { "epoch": 27.75582329317269, "grad_norm": 13.927249908447266, "learning_rate": 3.610967741935484e-05, "loss": 0.5394, "step": 4306 }, { "epoch": 27.762248995983935, "grad_norm": 6.272693157196045, "learning_rate": 3.6106451612903225e-05, "loss": 0.5019, "step": 4307 }, { "epoch": 27.76867469879518, "grad_norm": 10.246725082397461, "learning_rate": 3.610322580645161e-05, "loss": 0.1843, "step": 4308 }, { "epoch": 27.775100401606426, "grad_norm": 12.054572105407715, "learning_rate": 3.61e-05, "loss": 0.4779, "step": 4309 }, { "epoch": 27.781526104417672, "grad_norm": 22.866901397705078, "learning_rate": 3.609677419354839e-05, "loss": 0.4473, "step": 4310 }, { "epoch": 27.787951807228914, "grad_norm": 18.864221572875977, "learning_rate": 3.6093548387096775e-05, "loss": 0.5814, "step": 4311 }, { "epoch": 27.79437751004016, "grad_norm": 11.862114906311035, "learning_rate": 3.609032258064517e-05, "loss": 0.3338, "step": 4312 }, { "epoch": 27.800803212851406, "grad_norm": 9.272457122802734, "learning_rate": 3.608709677419355e-05, "loss": 0.3421, "step": 4313 }, { "epoch": 27.80722891566265, "grad_norm": 8.52086067199707, "learning_rate": 3.608387096774194e-05, "loss": 0.3238, "step": 4314 }, { "epoch": 27.813654618473894, "grad_norm": 12.542362213134766, "learning_rate": 3.6080645161290324e-05, "loss": 0.3557, "step": 4315 }, { "epoch": 27.82008032128514, "grad_norm": 5.989299297332764, "learning_rate": 3.607741935483871e-05, "loss": 0.3957, "step": 4316 }, { "epoch": 27.826506024096386, "grad_norm": 5.863888263702393, "learning_rate": 3.6074193548387096e-05, "loss": 0.2624, "step": 4317 }, { "epoch": 27.83293172690763, "grad_norm": 6.3034281730651855, "learning_rate": 3.607096774193549e-05, "loss": 0.2367, "step": 4318 }, { "epoch": 27.839357429718877, "grad_norm": 11.321639060974121, "learning_rate": 3.6067741935483874e-05, "loss": 0.2358, "step": 4319 }, { "epoch": 27.84578313253012, "grad_norm": 18.24770164489746, "learning_rate": 3.606451612903226e-05, "loss": 0.3595, "step": 4320 }, { "epoch": 27.852208835341365, "grad_norm": 11.802704811096191, "learning_rate": 3.6061290322580645e-05, "loss": 0.3771, "step": 4321 }, { "epoch": 27.85863453815261, "grad_norm": 7.513056755065918, "learning_rate": 3.605806451612904e-05, "loss": 0.473, "step": 4322 }, { "epoch": 27.865060240963857, "grad_norm": 26.060497283935547, "learning_rate": 3.605483870967742e-05, "loss": 0.369, "step": 4323 }, { "epoch": 27.8714859437751, "grad_norm": 15.003191947937012, "learning_rate": 3.605161290322581e-05, "loss": 0.4099, "step": 4324 }, { "epoch": 27.877911646586345, "grad_norm": 10.269225120544434, "learning_rate": 3.6048387096774195e-05, "loss": 0.4728, "step": 4325 }, { "epoch": 27.88433734939759, "grad_norm": 9.339336395263672, "learning_rate": 3.604516129032258e-05, "loss": 0.3058, "step": 4326 }, { "epoch": 27.890763052208836, "grad_norm": 4.122340679168701, "learning_rate": 3.604193548387097e-05, "loss": 0.3491, "step": 4327 }, { "epoch": 27.897188755020082, "grad_norm": 4.3994269371032715, "learning_rate": 3.603870967741936e-05, "loss": 0.145, "step": 4328 }, { "epoch": 27.903614457831324, "grad_norm": 19.13685417175293, "learning_rate": 3.6035483870967745e-05, "loss": 0.5168, "step": 4329 }, { "epoch": 27.91004016064257, "grad_norm": 5.423758506774902, "learning_rate": 3.603225806451613e-05, "loss": 0.4897, "step": 4330 }, { "epoch": 27.916465863453816, "grad_norm": 10.137432098388672, "learning_rate": 3.6029032258064516e-05, "loss": 0.2844, "step": 4331 }, { "epoch": 27.92289156626506, "grad_norm": 6.293140411376953, "learning_rate": 3.60258064516129e-05, "loss": 0.45, "step": 4332 }, { "epoch": 27.929317269076304, "grad_norm": 12.626398086547852, "learning_rate": 3.6022580645161294e-05, "loss": 0.2824, "step": 4333 }, { "epoch": 27.93574297188755, "grad_norm": 6.51725435256958, "learning_rate": 3.601935483870968e-05, "loss": 0.5469, "step": 4334 }, { "epoch": 27.942168674698795, "grad_norm": 19.914541244506836, "learning_rate": 3.6016129032258066e-05, "loss": 0.8368, "step": 4335 }, { "epoch": 27.94859437751004, "grad_norm": 8.62860107421875, "learning_rate": 3.601290322580646e-05, "loss": 0.2182, "step": 4336 }, { "epoch": 27.955020080321287, "grad_norm": 24.898576736450195, "learning_rate": 3.6009677419354844e-05, "loss": 0.4604, "step": 4337 }, { "epoch": 27.96144578313253, "grad_norm": 19.965681076049805, "learning_rate": 3.600645161290322e-05, "loss": 0.45, "step": 4338 }, { "epoch": 27.967871485943775, "grad_norm": 26.633987426757812, "learning_rate": 3.6003225806451615e-05, "loss": 0.3543, "step": 4339 }, { "epoch": 27.97429718875502, "grad_norm": 17.327051162719727, "learning_rate": 3.6e-05, "loss": 0.4524, "step": 4340 }, { "epoch": 27.980722891566266, "grad_norm": 7.139040470123291, "learning_rate": 3.599677419354839e-05, "loss": 0.2788, "step": 4341 }, { "epoch": 27.98714859437751, "grad_norm": 13.684487342834473, "learning_rate": 3.599354838709678e-05, "loss": 0.5145, "step": 4342 }, { "epoch": 27.993574297188754, "grad_norm": 8.572149276733398, "learning_rate": 3.5990322580645165e-05, "loss": 0.3046, "step": 4343 }, { "epoch": 28.00321285140562, "grad_norm": 8.136265754699707, "learning_rate": 3.598709677419355e-05, "loss": 0.2298, "step": 4344 }, { "epoch": 28.009638554216867, "grad_norm": 17.288101196289062, "learning_rate": 3.5983870967741936e-05, "loss": 0.3058, "step": 4345 }, { "epoch": 28.016064257028113, "grad_norm": 33.5849494934082, "learning_rate": 3.598064516129032e-05, "loss": 0.4692, "step": 4346 }, { "epoch": 28.02248995983936, "grad_norm": 8.767496109008789, "learning_rate": 3.597741935483871e-05, "loss": 0.3198, "step": 4347 }, { "epoch": 28.0289156626506, "grad_norm": 16.897043228149414, "learning_rate": 3.59741935483871e-05, "loss": 0.6517, "step": 4348 }, { "epoch": 28.035341365461846, "grad_norm": 19.857393264770508, "learning_rate": 3.5970967741935486e-05, "loss": 0.4791, "step": 4349 }, { "epoch": 28.041767068273092, "grad_norm": 20.533300399780273, "learning_rate": 3.596774193548387e-05, "loss": 0.5274, "step": 4350 }, { "epoch": 28.048192771084338, "grad_norm": 12.247456550598145, "learning_rate": 3.5964516129032264e-05, "loss": 0.4692, "step": 4351 }, { "epoch": 28.054618473895584, "grad_norm": 15.57494831085205, "learning_rate": 3.596129032258065e-05, "loss": 0.4724, "step": 4352 }, { "epoch": 28.061044176706826, "grad_norm": 10.731034278869629, "learning_rate": 3.5958064516129035e-05, "loss": 0.3759, "step": 4353 }, { "epoch": 28.06746987951807, "grad_norm": 2.865828037261963, "learning_rate": 3.595483870967742e-05, "loss": 0.1266, "step": 4354 }, { "epoch": 28.073895582329317, "grad_norm": 17.785696029663086, "learning_rate": 3.595161290322581e-05, "loss": 0.7353, "step": 4355 }, { "epoch": 28.080321285140563, "grad_norm": 18.87442398071289, "learning_rate": 3.594838709677419e-05, "loss": 0.433, "step": 4356 }, { "epoch": 28.086746987951805, "grad_norm": 18.534725189208984, "learning_rate": 3.5945161290322585e-05, "loss": 0.5035, "step": 4357 }, { "epoch": 28.09317269076305, "grad_norm": 9.858053207397461, "learning_rate": 3.594193548387097e-05, "loss": 0.5987, "step": 4358 }, { "epoch": 28.099598393574297, "grad_norm": 11.376391410827637, "learning_rate": 3.5938709677419356e-05, "loss": 0.3019, "step": 4359 }, { "epoch": 28.106024096385543, "grad_norm": 5.48152494430542, "learning_rate": 3.593548387096774e-05, "loss": 0.2731, "step": 4360 }, { "epoch": 28.11244979919679, "grad_norm": 24.823904037475586, "learning_rate": 3.5932258064516135e-05, "loss": 0.4973, "step": 4361 }, { "epoch": 28.11887550200803, "grad_norm": 32.60622787475586, "learning_rate": 3.5929032258064514e-05, "loss": 0.5262, "step": 4362 }, { "epoch": 28.125301204819277, "grad_norm": 7.155497074127197, "learning_rate": 3.5925806451612906e-05, "loss": 0.1821, "step": 4363 }, { "epoch": 28.131726907630522, "grad_norm": 13.403438568115234, "learning_rate": 3.592258064516129e-05, "loss": 0.7627, "step": 4364 }, { "epoch": 28.138152610441768, "grad_norm": 14.66135025024414, "learning_rate": 3.591935483870968e-05, "loss": 0.2099, "step": 4365 }, { "epoch": 28.14457831325301, "grad_norm": 20.34657096862793, "learning_rate": 3.591612903225807e-05, "loss": 0.415, "step": 4366 }, { "epoch": 28.151004016064256, "grad_norm": 13.323319435119629, "learning_rate": 3.5912903225806456e-05, "loss": 0.3779, "step": 4367 }, { "epoch": 28.157429718875502, "grad_norm": 9.32179069519043, "learning_rate": 3.590967741935484e-05, "loss": 0.4911, "step": 4368 }, { "epoch": 28.163855421686748, "grad_norm": 11.446380615234375, "learning_rate": 3.590645161290323e-05, "loss": 0.6548, "step": 4369 }, { "epoch": 28.170281124497993, "grad_norm": 4.463781356811523, "learning_rate": 3.590322580645161e-05, "loss": 0.1976, "step": 4370 }, { "epoch": 28.176706827309236, "grad_norm": 12.145200729370117, "learning_rate": 3.59e-05, "loss": 0.1965, "step": 4371 }, { "epoch": 28.18313253012048, "grad_norm": 26.067176818847656, "learning_rate": 3.589677419354839e-05, "loss": 0.4004, "step": 4372 }, { "epoch": 28.189558232931727, "grad_norm": 17.703786849975586, "learning_rate": 3.589354838709678e-05, "loss": 0.4977, "step": 4373 }, { "epoch": 28.195983935742973, "grad_norm": 12.927376747131348, "learning_rate": 3.589032258064516e-05, "loss": 0.3546, "step": 4374 }, { "epoch": 28.202409638554215, "grad_norm": 6.384668350219727, "learning_rate": 3.5887096774193555e-05, "loss": 0.3727, "step": 4375 }, { "epoch": 28.20883534136546, "grad_norm": 17.349224090576172, "learning_rate": 3.588387096774194e-05, "loss": 0.3022, "step": 4376 }, { "epoch": 28.215261044176707, "grad_norm": 5.340559959411621, "learning_rate": 3.588064516129032e-05, "loss": 0.3567, "step": 4377 }, { "epoch": 28.221686746987952, "grad_norm": 6.483819484710693, "learning_rate": 3.587741935483871e-05, "loss": 0.3145, "step": 4378 }, { "epoch": 28.228112449799198, "grad_norm": 15.234386444091797, "learning_rate": 3.58741935483871e-05, "loss": 0.5992, "step": 4379 }, { "epoch": 28.23453815261044, "grad_norm": 5.069404602050781, "learning_rate": 3.5870967741935483e-05, "loss": 0.1868, "step": 4380 }, { "epoch": 28.240963855421686, "grad_norm": 2.8382959365844727, "learning_rate": 3.5867741935483876e-05, "loss": 0.1433, "step": 4381 }, { "epoch": 28.247389558232932, "grad_norm": 7.6022748947143555, "learning_rate": 3.586451612903226e-05, "loss": 0.2804, "step": 4382 }, { "epoch": 28.253815261044178, "grad_norm": 17.26947593688965, "learning_rate": 3.586129032258065e-05, "loss": 0.4091, "step": 4383 }, { "epoch": 28.26024096385542, "grad_norm": 5.356723308563232, "learning_rate": 3.585806451612903e-05, "loss": 0.1231, "step": 4384 }, { "epoch": 28.266666666666666, "grad_norm": 4.399486064910889, "learning_rate": 3.585483870967742e-05, "loss": 0.2654, "step": 4385 }, { "epoch": 28.27309236947791, "grad_norm": 12.982123374938965, "learning_rate": 3.5851612903225804e-05, "loss": 0.38, "step": 4386 }, { "epoch": 28.279518072289157, "grad_norm": 22.85491371154785, "learning_rate": 3.58483870967742e-05, "loss": 0.2578, "step": 4387 }, { "epoch": 28.285943775100403, "grad_norm": 19.38821029663086, "learning_rate": 3.584516129032258e-05, "loss": 0.5743, "step": 4388 }, { "epoch": 28.292369477911645, "grad_norm": 21.65048599243164, "learning_rate": 3.584193548387097e-05, "loss": 0.2905, "step": 4389 }, { "epoch": 28.29879518072289, "grad_norm": 10.187932014465332, "learning_rate": 3.583870967741936e-05, "loss": 0.5117, "step": 4390 }, { "epoch": 28.305220883534137, "grad_norm": 9.300326347351074, "learning_rate": 3.5835483870967746e-05, "loss": 0.3129, "step": 4391 }, { "epoch": 28.311646586345383, "grad_norm": 5.9116950035095215, "learning_rate": 3.583225806451613e-05, "loss": 0.2485, "step": 4392 }, { "epoch": 28.318072289156625, "grad_norm": 11.087654113769531, "learning_rate": 3.582903225806452e-05, "loss": 0.426, "step": 4393 }, { "epoch": 28.32449799196787, "grad_norm": 5.782369613647461, "learning_rate": 3.5825806451612904e-05, "loss": 0.5665, "step": 4394 }, { "epoch": 28.330923694779116, "grad_norm": 10.621213912963867, "learning_rate": 3.582258064516129e-05, "loss": 0.6023, "step": 4395 }, { "epoch": 28.337349397590362, "grad_norm": 21.951210021972656, "learning_rate": 3.581935483870968e-05, "loss": 0.3589, "step": 4396 }, { "epoch": 28.343775100401608, "grad_norm": 16.506956100463867, "learning_rate": 3.581612903225807e-05, "loss": 0.3738, "step": 4397 }, { "epoch": 28.35020080321285, "grad_norm": 7.499403476715088, "learning_rate": 3.581290322580645e-05, "loss": 0.2966, "step": 4398 }, { "epoch": 28.356626506024096, "grad_norm": 12.14697265625, "learning_rate": 3.5809677419354846e-05, "loss": 0.6689, "step": 4399 }, { "epoch": 28.36305220883534, "grad_norm": 8.156789779663086, "learning_rate": 3.580645161290323e-05, "loss": 0.2613, "step": 4400 }, { "epoch": 28.369477911646587, "grad_norm": 7.2217302322387695, "learning_rate": 3.580322580645161e-05, "loss": 0.4509, "step": 4401 }, { "epoch": 28.37590361445783, "grad_norm": 8.559857368469238, "learning_rate": 3.58e-05, "loss": 0.5132, "step": 4402 }, { "epoch": 28.382329317269075, "grad_norm": 15.109590530395508, "learning_rate": 3.579677419354839e-05, "loss": 0.2765, "step": 4403 }, { "epoch": 28.38875502008032, "grad_norm": 25.888263702392578, "learning_rate": 3.5793548387096774e-05, "loss": 0.5655, "step": 4404 }, { "epoch": 28.395180722891567, "grad_norm": 22.484193801879883, "learning_rate": 3.579032258064517e-05, "loss": 0.496, "step": 4405 }, { "epoch": 28.401606425702813, "grad_norm": 8.045501708984375, "learning_rate": 3.578709677419355e-05, "loss": 0.3954, "step": 4406 }, { "epoch": 28.408032128514055, "grad_norm": 7.9421067237854, "learning_rate": 3.578387096774194e-05, "loss": 0.3853, "step": 4407 }, { "epoch": 28.4144578313253, "grad_norm": 7.723948001861572, "learning_rate": 3.5780645161290324e-05, "loss": 0.3686, "step": 4408 }, { "epoch": 28.420883534136546, "grad_norm": 8.911657333374023, "learning_rate": 3.577741935483871e-05, "loss": 0.2306, "step": 4409 }, { "epoch": 28.427309236947792, "grad_norm": 16.35186767578125, "learning_rate": 3.5774193548387095e-05, "loss": 0.5203, "step": 4410 }, { "epoch": 28.433734939759034, "grad_norm": 5.800744533538818, "learning_rate": 3.577096774193549e-05, "loss": 0.4046, "step": 4411 }, { "epoch": 28.44016064257028, "grad_norm": 19.195270538330078, "learning_rate": 3.5767741935483873e-05, "loss": 0.2564, "step": 4412 }, { "epoch": 28.446586345381526, "grad_norm": 3.561566114425659, "learning_rate": 3.576451612903226e-05, "loss": 0.1543, "step": 4413 }, { "epoch": 28.45301204819277, "grad_norm": 11.428424835205078, "learning_rate": 3.576129032258065e-05, "loss": 0.3284, "step": 4414 }, { "epoch": 28.459437751004018, "grad_norm": 6.938541412353516, "learning_rate": 3.575806451612904e-05, "loss": 0.2622, "step": 4415 }, { "epoch": 28.46586345381526, "grad_norm": 5.017120838165283, "learning_rate": 3.5754838709677416e-05, "loss": 0.1457, "step": 4416 }, { "epoch": 28.472289156626506, "grad_norm": 8.32591724395752, "learning_rate": 3.575161290322581e-05, "loss": 0.5262, "step": 4417 }, { "epoch": 28.47871485943775, "grad_norm": 3.1145052909851074, "learning_rate": 3.5748387096774194e-05, "loss": 0.3214, "step": 4418 }, { "epoch": 28.485140562248997, "grad_norm": 8.822089195251465, "learning_rate": 3.574516129032258e-05, "loss": 0.4589, "step": 4419 }, { "epoch": 28.49156626506024, "grad_norm": 18.782556533813477, "learning_rate": 3.574193548387097e-05, "loss": 0.8776, "step": 4420 }, { "epoch": 28.497991967871485, "grad_norm": 9.430704116821289, "learning_rate": 3.573870967741936e-05, "loss": 0.3812, "step": 4421 }, { "epoch": 28.50441767068273, "grad_norm": 14.544929504394531, "learning_rate": 3.5735483870967744e-05, "loss": 0.3563, "step": 4422 }, { "epoch": 28.510843373493977, "grad_norm": 11.406376838684082, "learning_rate": 3.573225806451613e-05, "loss": 0.4876, "step": 4423 }, { "epoch": 28.517269076305222, "grad_norm": 14.434685707092285, "learning_rate": 3.5729032258064515e-05, "loss": 0.6569, "step": 4424 }, { "epoch": 28.523694779116465, "grad_norm": 5.93366813659668, "learning_rate": 3.57258064516129e-05, "loss": 0.1822, "step": 4425 }, { "epoch": 28.53012048192771, "grad_norm": 8.117956161499023, "learning_rate": 3.5722580645161294e-05, "loss": 0.2833, "step": 4426 }, { "epoch": 28.536546184738956, "grad_norm": 9.106802940368652, "learning_rate": 3.571935483870968e-05, "loss": 0.397, "step": 4427 }, { "epoch": 28.542971887550202, "grad_norm": 9.322734832763672, "learning_rate": 3.5716129032258065e-05, "loss": 0.3009, "step": 4428 }, { "epoch": 28.549397590361444, "grad_norm": 5.754122257232666, "learning_rate": 3.571290322580646e-05, "loss": 0.3205, "step": 4429 }, { "epoch": 28.55582329317269, "grad_norm": 19.95819664001465, "learning_rate": 3.570967741935484e-05, "loss": 0.4603, "step": 4430 }, { "epoch": 28.562248995983936, "grad_norm": 78.4732894897461, "learning_rate": 3.570645161290323e-05, "loss": 0.376, "step": 4431 }, { "epoch": 28.56867469879518, "grad_norm": 11.893593788146973, "learning_rate": 3.5703225806451615e-05, "loss": 0.2679, "step": 4432 }, { "epoch": 28.575100401606427, "grad_norm": 10.233795166015625, "learning_rate": 3.57e-05, "loss": 0.2588, "step": 4433 }, { "epoch": 28.58152610441767, "grad_norm": 28.665786743164062, "learning_rate": 3.5696774193548386e-05, "loss": 0.265, "step": 4434 }, { "epoch": 28.587951807228915, "grad_norm": 11.780948638916016, "learning_rate": 3.569354838709678e-05, "loss": 0.2193, "step": 4435 }, { "epoch": 28.59437751004016, "grad_norm": 10.881484985351562, "learning_rate": 3.5690322580645164e-05, "loss": 0.439, "step": 4436 }, { "epoch": 28.600803212851407, "grad_norm": 16.04619026184082, "learning_rate": 3.568709677419355e-05, "loss": 0.6332, "step": 4437 }, { "epoch": 28.60722891566265, "grad_norm": 11.193946838378906, "learning_rate": 3.568387096774194e-05, "loss": 0.3309, "step": 4438 }, { "epoch": 28.613654618473895, "grad_norm": 9.718118667602539, "learning_rate": 3.568064516129033e-05, "loss": 0.3555, "step": 4439 }, { "epoch": 28.62008032128514, "grad_norm": 4.051966667175293, "learning_rate": 3.567741935483871e-05, "loss": 0.2179, "step": 4440 }, { "epoch": 28.626506024096386, "grad_norm": 8.934828758239746, "learning_rate": 3.56741935483871e-05, "loss": 0.3519, "step": 4441 }, { "epoch": 28.632931726907632, "grad_norm": 6.5467610359191895, "learning_rate": 3.5670967741935485e-05, "loss": 0.375, "step": 4442 }, { "epoch": 28.639357429718874, "grad_norm": 14.989519119262695, "learning_rate": 3.566774193548387e-05, "loss": 0.5312, "step": 4443 }, { "epoch": 28.64578313253012, "grad_norm": 17.68325424194336, "learning_rate": 3.5664516129032264e-05, "loss": 0.3143, "step": 4444 }, { "epoch": 28.652208835341366, "grad_norm": 4.50241231918335, "learning_rate": 3.566129032258065e-05, "loss": 0.2644, "step": 4445 }, { "epoch": 28.65863453815261, "grad_norm": 6.1389851570129395, "learning_rate": 3.5658064516129035e-05, "loss": 0.3634, "step": 4446 }, { "epoch": 28.665060240963854, "grad_norm": 10.479798316955566, "learning_rate": 3.565483870967742e-05, "loss": 0.3624, "step": 4447 }, { "epoch": 28.6714859437751, "grad_norm": 7.154819488525391, "learning_rate": 3.5651612903225806e-05, "loss": 0.5821, "step": 4448 }, { "epoch": 28.677911646586345, "grad_norm": 6.765829086303711, "learning_rate": 3.564838709677419e-05, "loss": 0.4532, "step": 4449 }, { "epoch": 28.68433734939759, "grad_norm": 16.36671257019043, "learning_rate": 3.5645161290322585e-05, "loss": 0.3324, "step": 4450 }, { "epoch": 28.690763052208837, "grad_norm": 17.13541030883789, "learning_rate": 3.564193548387097e-05, "loss": 0.4908, "step": 4451 }, { "epoch": 28.69718875502008, "grad_norm": 15.335646629333496, "learning_rate": 3.5638709677419356e-05, "loss": 0.4051, "step": 4452 }, { "epoch": 28.703614457831325, "grad_norm": 11.217277526855469, "learning_rate": 3.563548387096775e-05, "loss": 0.3939, "step": 4453 }, { "epoch": 28.71004016064257, "grad_norm": 6.104141712188721, "learning_rate": 3.5632258064516134e-05, "loss": 0.371, "step": 4454 }, { "epoch": 28.716465863453816, "grad_norm": 6.57197904586792, "learning_rate": 3.562903225806451e-05, "loss": 0.454, "step": 4455 }, { "epoch": 28.72289156626506, "grad_norm": 29.40903663635254, "learning_rate": 3.5625806451612906e-05, "loss": 0.5299, "step": 4456 }, { "epoch": 28.729317269076304, "grad_norm": 18.876178741455078, "learning_rate": 3.562258064516129e-05, "loss": 0.2806, "step": 4457 }, { "epoch": 28.73574297188755, "grad_norm": 10.437382698059082, "learning_rate": 3.561935483870968e-05, "loss": 0.3611, "step": 4458 }, { "epoch": 28.742168674698796, "grad_norm": 4.793471336364746, "learning_rate": 3.561612903225807e-05, "loss": 0.2154, "step": 4459 }, { "epoch": 28.74859437751004, "grad_norm": 25.32998275756836, "learning_rate": 3.5612903225806455e-05, "loss": 0.39, "step": 4460 }, { "epoch": 28.755020080321284, "grad_norm": 13.85676383972168, "learning_rate": 3.560967741935484e-05, "loss": 0.3394, "step": 4461 }, { "epoch": 28.76144578313253, "grad_norm": 8.828316688537598, "learning_rate": 3.560645161290323e-05, "loss": 0.4005, "step": 4462 }, { "epoch": 28.767871485943775, "grad_norm": 7.191306114196777, "learning_rate": 3.560322580645161e-05, "loss": 0.3652, "step": 4463 }, { "epoch": 28.77429718875502, "grad_norm": 29.758790969848633, "learning_rate": 3.56e-05, "loss": 0.5418, "step": 4464 }, { "epoch": 28.780722891566263, "grad_norm": 24.762277603149414, "learning_rate": 3.559677419354839e-05, "loss": 0.7172, "step": 4465 }, { "epoch": 28.78714859437751, "grad_norm": 4.1125054359436035, "learning_rate": 3.5593548387096776e-05, "loss": 0.2734, "step": 4466 }, { "epoch": 28.793574297188755, "grad_norm": 7.17116641998291, "learning_rate": 3.559032258064516e-05, "loss": 0.3819, "step": 4467 }, { "epoch": 28.8, "grad_norm": 10.2844820022583, "learning_rate": 3.5587096774193554e-05, "loss": 0.4686, "step": 4468 }, { "epoch": 28.806425702811246, "grad_norm": 7.971284866333008, "learning_rate": 3.558387096774194e-05, "loss": 0.2192, "step": 4469 }, { "epoch": 28.81285140562249, "grad_norm": 14.984542846679688, "learning_rate": 3.5580645161290326e-05, "loss": 0.7716, "step": 4470 }, { "epoch": 28.819277108433734, "grad_norm": 6.99550724029541, "learning_rate": 3.557741935483871e-05, "loss": 0.246, "step": 4471 }, { "epoch": 28.82570281124498, "grad_norm": 9.829756736755371, "learning_rate": 3.55741935483871e-05, "loss": 0.6184, "step": 4472 }, { "epoch": 28.832128514056226, "grad_norm": 7.689605236053467, "learning_rate": 3.557096774193548e-05, "loss": 0.6293, "step": 4473 }, { "epoch": 28.83855421686747, "grad_norm": 14.706714630126953, "learning_rate": 3.5567741935483875e-05, "loss": 0.3821, "step": 4474 }, { "epoch": 28.844979919678714, "grad_norm": 3.801832914352417, "learning_rate": 3.556451612903226e-05, "loss": 0.3754, "step": 4475 }, { "epoch": 28.85140562248996, "grad_norm": 12.31419563293457, "learning_rate": 3.556129032258065e-05, "loss": 0.4129, "step": 4476 }, { "epoch": 28.857831325301206, "grad_norm": 15.753545761108398, "learning_rate": 3.555806451612904e-05, "loss": 0.3064, "step": 4477 }, { "epoch": 28.86425702811245, "grad_norm": 10.198423385620117, "learning_rate": 3.5554838709677425e-05, "loss": 0.3307, "step": 4478 }, { "epoch": 28.870682730923694, "grad_norm": 17.400312423706055, "learning_rate": 3.5551612903225804e-05, "loss": 0.231, "step": 4479 }, { "epoch": 28.87710843373494, "grad_norm": 13.591204643249512, "learning_rate": 3.5548387096774196e-05, "loss": 0.5273, "step": 4480 }, { "epoch": 28.883534136546185, "grad_norm": 17.68254852294922, "learning_rate": 3.554516129032258e-05, "loss": 0.4727, "step": 4481 }, { "epoch": 28.88995983935743, "grad_norm": 11.180360794067383, "learning_rate": 3.554193548387097e-05, "loss": 0.2779, "step": 4482 }, { "epoch": 28.896385542168673, "grad_norm": 17.195289611816406, "learning_rate": 3.553870967741936e-05, "loss": 0.3121, "step": 4483 }, { "epoch": 28.90281124497992, "grad_norm": 16.6837158203125, "learning_rate": 3.5535483870967746e-05, "loss": 0.5591, "step": 4484 }, { "epoch": 28.909236947791165, "grad_norm": 15.69422435760498, "learning_rate": 3.553225806451613e-05, "loss": 0.6055, "step": 4485 }, { "epoch": 28.91566265060241, "grad_norm": 6.993812084197998, "learning_rate": 3.552903225806452e-05, "loss": 0.2531, "step": 4486 }, { "epoch": 28.922088353413656, "grad_norm": 7.358951091766357, "learning_rate": 3.55258064516129e-05, "loss": 0.4634, "step": 4487 }, { "epoch": 28.9285140562249, "grad_norm": 4.75485372543335, "learning_rate": 3.552258064516129e-05, "loss": 0.2111, "step": 4488 }, { "epoch": 28.934939759036144, "grad_norm": 2.788290023803711, "learning_rate": 3.551935483870968e-05, "loss": 0.1576, "step": 4489 }, { "epoch": 28.94136546184739, "grad_norm": 11.018362045288086, "learning_rate": 3.551612903225807e-05, "loss": 0.1693, "step": 4490 }, { "epoch": 28.947791164658636, "grad_norm": 23.068559646606445, "learning_rate": 3.551290322580645e-05, "loss": 0.3545, "step": 4491 }, { "epoch": 28.954216867469878, "grad_norm": 40.243228912353516, "learning_rate": 3.5509677419354845e-05, "loss": 0.5419, "step": 4492 }, { "epoch": 28.960642570281124, "grad_norm": 39.74689483642578, "learning_rate": 3.550645161290323e-05, "loss": 0.4352, "step": 4493 }, { "epoch": 28.96706827309237, "grad_norm": 28.04096031188965, "learning_rate": 3.550322580645161e-05, "loss": 0.4819, "step": 4494 }, { "epoch": 28.973493975903615, "grad_norm": 31.000967025756836, "learning_rate": 3.55e-05, "loss": 0.4626, "step": 4495 }, { "epoch": 28.97991967871486, "grad_norm": 5.050440788269043, "learning_rate": 3.549677419354839e-05, "loss": 0.4426, "step": 4496 }, { "epoch": 28.986345381526103, "grad_norm": 4.866138935089111, "learning_rate": 3.5493548387096774e-05, "loss": 0.2577, "step": 4497 }, { "epoch": 28.99277108433735, "grad_norm": 4.1925249099731445, "learning_rate": 3.5490322580645166e-05, "loss": 0.1919, "step": 4498 }, { "epoch": 29.002409638554216, "grad_norm": 28.03898811340332, "learning_rate": 3.548709677419355e-05, "loss": 0.4095, "step": 4499 }, { "epoch": 29.00883534136546, "grad_norm": 13.71825122833252, "learning_rate": 3.548387096774194e-05, "loss": 0.3092, "step": 4500 }, { "epoch": 29.015261044176707, "grad_norm": 19.899288177490234, "learning_rate": 3.548064516129033e-05, "loss": 0.3418, "step": 4501 }, { "epoch": 29.021686746987953, "grad_norm": 5.539183139801025, "learning_rate": 3.547741935483871e-05, "loss": 0.3818, "step": 4502 }, { "epoch": 29.028112449799195, "grad_norm": 13.304910659790039, "learning_rate": 3.5474193548387095e-05, "loss": 0.287, "step": 4503 }, { "epoch": 29.03453815261044, "grad_norm": 6.960597038269043, "learning_rate": 3.547096774193549e-05, "loss": 0.4307, "step": 4504 }, { "epoch": 29.040963855421687, "grad_norm": 9.412762641906738, "learning_rate": 3.546774193548387e-05, "loss": 0.6728, "step": 4505 }, { "epoch": 29.047389558232933, "grad_norm": 18.86697006225586, "learning_rate": 3.546451612903226e-05, "loss": 0.4853, "step": 4506 }, { "epoch": 29.05381526104418, "grad_norm": 5.115467071533203, "learning_rate": 3.546129032258065e-05, "loss": 0.391, "step": 4507 }, { "epoch": 29.06024096385542, "grad_norm": 11.421082496643066, "learning_rate": 3.545806451612904e-05, "loss": 0.4451, "step": 4508 }, { "epoch": 29.066666666666666, "grad_norm": 9.356465339660645, "learning_rate": 3.545483870967742e-05, "loss": 0.2345, "step": 4509 }, { "epoch": 29.073092369477912, "grad_norm": 9.969861030578613, "learning_rate": 3.545161290322581e-05, "loss": 0.3463, "step": 4510 }, { "epoch": 29.079518072289158, "grad_norm": 13.507002830505371, "learning_rate": 3.5448387096774194e-05, "loss": 0.4577, "step": 4511 }, { "epoch": 29.0859437751004, "grad_norm": 7.505441188812256, "learning_rate": 3.544516129032258e-05, "loss": 0.5621, "step": 4512 }, { "epoch": 29.092369477911646, "grad_norm": 9.505091667175293, "learning_rate": 3.544193548387097e-05, "loss": 0.3909, "step": 4513 }, { "epoch": 29.09879518072289, "grad_norm": 12.414438247680664, "learning_rate": 3.543870967741936e-05, "loss": 0.3666, "step": 4514 }, { "epoch": 29.105220883534137, "grad_norm": 3.5067801475524902, "learning_rate": 3.5435483870967744e-05, "loss": 0.3689, "step": 4515 }, { "epoch": 29.111646586345383, "grad_norm": 8.617573738098145, "learning_rate": 3.5432258064516136e-05, "loss": 0.3589, "step": 4516 }, { "epoch": 29.118072289156625, "grad_norm": 8.806960105895996, "learning_rate": 3.5429032258064515e-05, "loss": 0.4843, "step": 4517 }, { "epoch": 29.12449799196787, "grad_norm": 1.782534122467041, "learning_rate": 3.54258064516129e-05, "loss": 0.1378, "step": 4518 }, { "epoch": 29.130923694779117, "grad_norm": 9.200276374816895, "learning_rate": 3.542258064516129e-05, "loss": 0.5521, "step": 4519 }, { "epoch": 29.137349397590363, "grad_norm": 15.692605018615723, "learning_rate": 3.541935483870968e-05, "loss": 0.1778, "step": 4520 }, { "epoch": 29.143775100401605, "grad_norm": 20.358016967773438, "learning_rate": 3.5416129032258065e-05, "loss": 0.5381, "step": 4521 }, { "epoch": 29.15020080321285, "grad_norm": 8.514213562011719, "learning_rate": 3.541290322580646e-05, "loss": 0.5519, "step": 4522 }, { "epoch": 29.156626506024097, "grad_norm": 15.070356369018555, "learning_rate": 3.540967741935484e-05, "loss": 0.483, "step": 4523 }, { "epoch": 29.163052208835342, "grad_norm": 5.357034683227539, "learning_rate": 3.540645161290323e-05, "loss": 0.2313, "step": 4524 }, { "epoch": 29.169477911646588, "grad_norm": 11.72697639465332, "learning_rate": 3.5403225806451614e-05, "loss": 0.3155, "step": 4525 }, { "epoch": 29.17590361445783, "grad_norm": 17.144865036010742, "learning_rate": 3.54e-05, "loss": 0.4888, "step": 4526 }, { "epoch": 29.182329317269076, "grad_norm": 14.252843856811523, "learning_rate": 3.5396774193548386e-05, "loss": 0.399, "step": 4527 }, { "epoch": 29.188755020080322, "grad_norm": 8.343183517456055, "learning_rate": 3.539354838709678e-05, "loss": 0.318, "step": 4528 }, { "epoch": 29.195180722891568, "grad_norm": 8.909079551696777, "learning_rate": 3.5390322580645164e-05, "loss": 0.5578, "step": 4529 }, { "epoch": 29.20160642570281, "grad_norm": 10.144410133361816, "learning_rate": 3.538709677419355e-05, "loss": 0.6876, "step": 4530 }, { "epoch": 29.208032128514056, "grad_norm": 21.87989044189453, "learning_rate": 3.538387096774194e-05, "loss": 0.5516, "step": 4531 }, { "epoch": 29.2144578313253, "grad_norm": 5.562446117401123, "learning_rate": 3.538064516129033e-05, "loss": 0.6707, "step": 4532 }, { "epoch": 29.220883534136547, "grad_norm": 3.644254446029663, "learning_rate": 3.5377419354838707e-05, "loss": 0.278, "step": 4533 }, { "epoch": 29.227309236947793, "grad_norm": 12.784631729125977, "learning_rate": 3.53741935483871e-05, "loss": 0.5939, "step": 4534 }, { "epoch": 29.233734939759035, "grad_norm": 16.557979583740234, "learning_rate": 3.5370967741935485e-05, "loss": 0.7224, "step": 4535 }, { "epoch": 29.24016064257028, "grad_norm": 12.598020553588867, "learning_rate": 3.536774193548387e-05, "loss": 0.3574, "step": 4536 }, { "epoch": 29.246586345381527, "grad_norm": 14.108193397521973, "learning_rate": 3.536451612903226e-05, "loss": 0.5609, "step": 4537 }, { "epoch": 29.253012048192772, "grad_norm": 39.9857292175293, "learning_rate": 3.536129032258065e-05, "loss": 0.2657, "step": 4538 }, { "epoch": 29.259437751004015, "grad_norm": 43.87123107910156, "learning_rate": 3.5358064516129034e-05, "loss": 0.3713, "step": 4539 }, { "epoch": 29.26586345381526, "grad_norm": 32.7863883972168, "learning_rate": 3.535483870967743e-05, "loss": 0.3902, "step": 4540 }, { "epoch": 29.272289156626506, "grad_norm": 14.50200080871582, "learning_rate": 3.5351612903225806e-05, "loss": 0.7033, "step": 4541 }, { "epoch": 29.278714859437752, "grad_norm": 10.087211608886719, "learning_rate": 3.534838709677419e-05, "loss": 0.5354, "step": 4542 }, { "epoch": 29.285140562248998, "grad_norm": 13.732751846313477, "learning_rate": 3.5345161290322584e-05, "loss": 0.6634, "step": 4543 }, { "epoch": 29.29156626506024, "grad_norm": 6.787538051605225, "learning_rate": 3.534193548387097e-05, "loss": 0.3947, "step": 4544 }, { "epoch": 29.297991967871486, "grad_norm": 14.589661598205566, "learning_rate": 3.5338709677419355e-05, "loss": 0.3658, "step": 4545 }, { "epoch": 29.30441767068273, "grad_norm": 7.337014675140381, "learning_rate": 3.533548387096775e-05, "loss": 0.363, "step": 4546 }, { "epoch": 29.310843373493977, "grad_norm": 7.740478515625, "learning_rate": 3.5332258064516134e-05, "loss": 0.202, "step": 4547 }, { "epoch": 29.31726907630522, "grad_norm": 3.924834728240967, "learning_rate": 3.532903225806452e-05, "loss": 0.3799, "step": 4548 }, { "epoch": 29.323694779116465, "grad_norm": 9.891263008117676, "learning_rate": 3.5325806451612905e-05, "loss": 0.3655, "step": 4549 }, { "epoch": 29.33012048192771, "grad_norm": 24.282121658325195, "learning_rate": 3.532258064516129e-05, "loss": 0.4722, "step": 4550 }, { "epoch": 29.336546184738957, "grad_norm": 18.959749221801758, "learning_rate": 3.5319354838709676e-05, "loss": 0.405, "step": 4551 }, { "epoch": 29.3429718875502, "grad_norm": 12.247556686401367, "learning_rate": 3.531612903225807e-05, "loss": 0.6677, "step": 4552 }, { "epoch": 29.349397590361445, "grad_norm": 30.760364532470703, "learning_rate": 3.5312903225806455e-05, "loss": 0.3147, "step": 4553 }, { "epoch": 29.35582329317269, "grad_norm": 8.296587944030762, "learning_rate": 3.530967741935484e-05, "loss": 0.4559, "step": 4554 }, { "epoch": 29.362248995983936, "grad_norm": 9.978439331054688, "learning_rate": 3.530645161290323e-05, "loss": 0.2715, "step": 4555 }, { "epoch": 29.368674698795182, "grad_norm": 6.160807132720947, "learning_rate": 3.530322580645161e-05, "loss": 0.1577, "step": 4556 }, { "epoch": 29.375100401606424, "grad_norm": 14.40156364440918, "learning_rate": 3.53e-05, "loss": 0.5031, "step": 4557 }, { "epoch": 29.38152610441767, "grad_norm": 34.31580352783203, "learning_rate": 3.529677419354839e-05, "loss": 0.3619, "step": 4558 }, { "epoch": 29.387951807228916, "grad_norm": 5.880702018737793, "learning_rate": 3.5293548387096776e-05, "loss": 0.2414, "step": 4559 }, { "epoch": 29.39437751004016, "grad_norm": 12.20317268371582, "learning_rate": 3.529032258064516e-05, "loss": 0.3535, "step": 4560 }, { "epoch": 29.400803212851407, "grad_norm": 13.670466423034668, "learning_rate": 3.5287096774193554e-05, "loss": 0.4191, "step": 4561 }, { "epoch": 29.40722891566265, "grad_norm": 23.741539001464844, "learning_rate": 3.528387096774194e-05, "loss": 0.5846, "step": 4562 }, { "epoch": 29.413654618473895, "grad_norm": 15.380813598632812, "learning_rate": 3.5280645161290325e-05, "loss": 0.3195, "step": 4563 }, { "epoch": 29.42008032128514, "grad_norm": 45.83262252807617, "learning_rate": 3.527741935483871e-05, "loss": 0.3089, "step": 4564 }, { "epoch": 29.426506024096387, "grad_norm": 60.46442794799805, "learning_rate": 3.52741935483871e-05, "loss": 0.4724, "step": 4565 }, { "epoch": 29.43293172690763, "grad_norm": 16.180992126464844, "learning_rate": 3.527096774193548e-05, "loss": 0.3476, "step": 4566 }, { "epoch": 29.439357429718875, "grad_norm": 7.576791763305664, "learning_rate": 3.5267741935483875e-05, "loss": 0.1252, "step": 4567 }, { "epoch": 29.44578313253012, "grad_norm": 13.034098625183105, "learning_rate": 3.526451612903226e-05, "loss": 0.4149, "step": 4568 }, { "epoch": 29.452208835341366, "grad_norm": 7.69444465637207, "learning_rate": 3.5261290322580646e-05, "loss": 0.2743, "step": 4569 }, { "epoch": 29.45863453815261, "grad_norm": 7.977758884429932, "learning_rate": 3.525806451612904e-05, "loss": 0.2486, "step": 4570 }, { "epoch": 29.465060240963854, "grad_norm": 10.0961275100708, "learning_rate": 3.5254838709677424e-05, "loss": 0.2852, "step": 4571 }, { "epoch": 29.4714859437751, "grad_norm": 22.03255844116211, "learning_rate": 3.52516129032258e-05, "loss": 0.4951, "step": 4572 }, { "epoch": 29.477911646586346, "grad_norm": 14.151461601257324, "learning_rate": 3.5248387096774196e-05, "loss": 0.5372, "step": 4573 }, { "epoch": 29.48433734939759, "grad_norm": 17.330150604248047, "learning_rate": 3.524516129032258e-05, "loss": 0.3685, "step": 4574 }, { "epoch": 29.490763052208834, "grad_norm": 15.267715454101562, "learning_rate": 3.524193548387097e-05, "loss": 0.2776, "step": 4575 }, { "epoch": 29.49718875502008, "grad_norm": 6.457451343536377, "learning_rate": 3.523870967741936e-05, "loss": 0.3247, "step": 4576 }, { "epoch": 29.503614457831326, "grad_norm": 5.20994758605957, "learning_rate": 3.5235483870967745e-05, "loss": 0.3512, "step": 4577 }, { "epoch": 29.51004016064257, "grad_norm": 9.175708770751953, "learning_rate": 3.523225806451613e-05, "loss": 0.2544, "step": 4578 }, { "epoch": 29.516465863453817, "grad_norm": 15.397836685180664, "learning_rate": 3.5229032258064524e-05, "loss": 0.5033, "step": 4579 }, { "epoch": 29.52289156626506, "grad_norm": 13.610023498535156, "learning_rate": 3.52258064516129e-05, "loss": 0.4384, "step": 4580 }, { "epoch": 29.529317269076305, "grad_norm": 3.9360158443450928, "learning_rate": 3.522258064516129e-05, "loss": 0.2923, "step": 4581 }, { "epoch": 29.53574297188755, "grad_norm": 10.335957527160645, "learning_rate": 3.521935483870968e-05, "loss": 0.529, "step": 4582 }, { "epoch": 29.542168674698797, "grad_norm": 11.944522857666016, "learning_rate": 3.5216129032258066e-05, "loss": 0.3633, "step": 4583 }, { "epoch": 29.54859437751004, "grad_norm": 5.521923542022705, "learning_rate": 3.521290322580645e-05, "loss": 0.3024, "step": 4584 }, { "epoch": 29.555020080321285, "grad_norm": 5.510916709899902, "learning_rate": 3.5209677419354845e-05, "loss": 0.2121, "step": 4585 }, { "epoch": 29.56144578313253, "grad_norm": 13.550652503967285, "learning_rate": 3.520645161290323e-05, "loss": 0.3595, "step": 4586 }, { "epoch": 29.567871485943776, "grad_norm": 9.476385116577148, "learning_rate": 3.520322580645161e-05, "loss": 0.4278, "step": 4587 }, { "epoch": 29.57429718875502, "grad_norm": 12.142529487609863, "learning_rate": 3.52e-05, "loss": 0.3203, "step": 4588 }, { "epoch": 29.580722891566264, "grad_norm": 2.877157688140869, "learning_rate": 3.519677419354839e-05, "loss": 0.198, "step": 4589 }, { "epoch": 29.58714859437751, "grad_norm": 14.60959529876709, "learning_rate": 3.519354838709677e-05, "loss": 0.6082, "step": 4590 }, { "epoch": 29.593574297188756, "grad_norm": 7.345745086669922, "learning_rate": 3.5190322580645166e-05, "loss": 0.4649, "step": 4591 }, { "epoch": 29.6, "grad_norm": 9.703039169311523, "learning_rate": 3.518709677419355e-05, "loss": 0.3462, "step": 4592 }, { "epoch": 29.606425702811244, "grad_norm": 15.606653213500977, "learning_rate": 3.518387096774194e-05, "loss": 0.3713, "step": 4593 }, { "epoch": 29.61285140562249, "grad_norm": 23.6650390625, "learning_rate": 3.518064516129033e-05, "loss": 0.7682, "step": 4594 }, { "epoch": 29.619277108433735, "grad_norm": 15.071745872497559, "learning_rate": 3.517741935483871e-05, "loss": 0.2669, "step": 4595 }, { "epoch": 29.62570281124498, "grad_norm": 26.182260513305664, "learning_rate": 3.5174193548387094e-05, "loss": 0.455, "step": 4596 }, { "epoch": 29.632128514056227, "grad_norm": 13.388492584228516, "learning_rate": 3.517096774193549e-05, "loss": 0.5083, "step": 4597 }, { "epoch": 29.63855421686747, "grad_norm": 5.469744682312012, "learning_rate": 3.516774193548387e-05, "loss": 0.3287, "step": 4598 }, { "epoch": 29.644979919678715, "grad_norm": 8.513886451721191, "learning_rate": 3.516451612903226e-05, "loss": 0.3744, "step": 4599 }, { "epoch": 29.65140562248996, "grad_norm": 23.28455352783203, "learning_rate": 3.516129032258065e-05, "loss": 0.386, "step": 4600 }, { "epoch": 29.657831325301206, "grad_norm": 28.933013916015625, "learning_rate": 3.5158064516129036e-05, "loss": 0.6475, "step": 4601 }, { "epoch": 29.66425702811245, "grad_norm": 11.272064208984375, "learning_rate": 3.515483870967742e-05, "loss": 0.5561, "step": 4602 }, { "epoch": 29.670682730923694, "grad_norm": 8.609187126159668, "learning_rate": 3.515161290322581e-05, "loss": 0.3558, "step": 4603 }, { "epoch": 29.67710843373494, "grad_norm": 13.954886436462402, "learning_rate": 3.5148387096774193e-05, "loss": 0.3736, "step": 4604 }, { "epoch": 29.683534136546186, "grad_norm": 14.688809394836426, "learning_rate": 3.514516129032258e-05, "loss": 0.407, "step": 4605 }, { "epoch": 29.689959839357428, "grad_norm": 17.147062301635742, "learning_rate": 3.514193548387097e-05, "loss": 0.3683, "step": 4606 }, { "epoch": 29.696385542168674, "grad_norm": 8.771656036376953, "learning_rate": 3.513870967741936e-05, "loss": 0.2324, "step": 4607 }, { "epoch": 29.70281124497992, "grad_norm": 6.221181392669678, "learning_rate": 3.513548387096774e-05, "loss": 0.2347, "step": 4608 }, { "epoch": 29.709236947791165, "grad_norm": 24.813615798950195, "learning_rate": 3.5132258064516136e-05, "loss": 0.4011, "step": 4609 }, { "epoch": 29.71566265060241, "grad_norm": 6.829839706420898, "learning_rate": 3.512903225806452e-05, "loss": 0.4136, "step": 4610 }, { "epoch": 29.722088353413653, "grad_norm": 16.4216365814209, "learning_rate": 3.51258064516129e-05, "loss": 0.3735, "step": 4611 }, { "epoch": 29.7285140562249, "grad_norm": 7.738377571105957, "learning_rate": 3.512258064516129e-05, "loss": 0.3226, "step": 4612 }, { "epoch": 29.734939759036145, "grad_norm": 30.254058837890625, "learning_rate": 3.511935483870968e-05, "loss": 0.4423, "step": 4613 }, { "epoch": 29.74136546184739, "grad_norm": 11.676312446594238, "learning_rate": 3.5116129032258064e-05, "loss": 0.1645, "step": 4614 }, { "epoch": 29.747791164658633, "grad_norm": 14.890725135803223, "learning_rate": 3.5112903225806457e-05, "loss": 0.3037, "step": 4615 }, { "epoch": 29.75421686746988, "grad_norm": 4.145571231842041, "learning_rate": 3.510967741935484e-05, "loss": 0.2695, "step": 4616 }, { "epoch": 29.760642570281124, "grad_norm": 27.460773468017578, "learning_rate": 3.510645161290323e-05, "loss": 0.6934, "step": 4617 }, { "epoch": 29.76706827309237, "grad_norm": 5.3283867835998535, "learning_rate": 3.510322580645162e-05, "loss": 0.2314, "step": 4618 }, { "epoch": 29.773493975903616, "grad_norm": 5.297540187835693, "learning_rate": 3.51e-05, "loss": 0.3461, "step": 4619 }, { "epoch": 29.779919678714858, "grad_norm": 14.54029369354248, "learning_rate": 3.5096774193548385e-05, "loss": 0.7181, "step": 4620 }, { "epoch": 29.786345381526104, "grad_norm": 7.189522743225098, "learning_rate": 3.509354838709678e-05, "loss": 0.3133, "step": 4621 }, { "epoch": 29.79277108433735, "grad_norm": 27.557415008544922, "learning_rate": 3.509032258064516e-05, "loss": 0.3314, "step": 4622 }, { "epoch": 29.799196787148595, "grad_norm": 13.034272193908691, "learning_rate": 3.508709677419355e-05, "loss": 0.6659, "step": 4623 }, { "epoch": 29.805622489959838, "grad_norm": 6.538400173187256, "learning_rate": 3.508387096774194e-05, "loss": 0.209, "step": 4624 }, { "epoch": 29.812048192771083, "grad_norm": 4.411250114440918, "learning_rate": 3.508064516129033e-05, "loss": 0.4553, "step": 4625 }, { "epoch": 29.81847389558233, "grad_norm": 6.928696632385254, "learning_rate": 3.5077419354838706e-05, "loss": 0.2192, "step": 4626 }, { "epoch": 29.824899598393575, "grad_norm": 7.371878147125244, "learning_rate": 3.50741935483871e-05, "loss": 0.5374, "step": 4627 }, { "epoch": 29.83132530120482, "grad_norm": 9.117181777954102, "learning_rate": 3.5070967741935484e-05, "loss": 0.1122, "step": 4628 }, { "epoch": 29.837751004016063, "grad_norm": 7.287140369415283, "learning_rate": 3.506774193548387e-05, "loss": 0.726, "step": 4629 }, { "epoch": 29.84417670682731, "grad_norm": 9.295029640197754, "learning_rate": 3.506451612903226e-05, "loss": 0.4038, "step": 4630 }, { "epoch": 29.850602409638554, "grad_norm": 7.117849826812744, "learning_rate": 3.506129032258065e-05, "loss": 0.3266, "step": 4631 }, { "epoch": 29.8570281124498, "grad_norm": 23.779891967773438, "learning_rate": 3.5058064516129034e-05, "loss": 0.5073, "step": 4632 }, { "epoch": 29.863453815261042, "grad_norm": 3.793759822845459, "learning_rate": 3.5054838709677426e-05, "loss": 0.262, "step": 4633 }, { "epoch": 29.86987951807229, "grad_norm": 29.234272003173828, "learning_rate": 3.5051612903225805e-05, "loss": 0.4211, "step": 4634 }, { "epoch": 29.876305220883534, "grad_norm": 15.941598892211914, "learning_rate": 3.504838709677419e-05, "loss": 0.4291, "step": 4635 }, { "epoch": 29.88273092369478, "grad_norm": 17.635541915893555, "learning_rate": 3.5045161290322583e-05, "loss": 0.6461, "step": 4636 }, { "epoch": 29.889156626506026, "grad_norm": 14.938704490661621, "learning_rate": 3.504193548387097e-05, "loss": 0.3422, "step": 4637 }, { "epoch": 29.895582329317268, "grad_norm": 17.356042861938477, "learning_rate": 3.5038709677419355e-05, "loss": 0.5081, "step": 4638 }, { "epoch": 29.902008032128514, "grad_norm": 19.977169036865234, "learning_rate": 3.503548387096775e-05, "loss": 0.3399, "step": 4639 }, { "epoch": 29.90843373493976, "grad_norm": 11.897782325744629, "learning_rate": 3.503225806451613e-05, "loss": 0.5001, "step": 4640 }, { "epoch": 29.914859437751005, "grad_norm": 47.11195373535156, "learning_rate": 3.502903225806452e-05, "loss": 0.631, "step": 4641 }, { "epoch": 29.921285140562247, "grad_norm": 9.253669738769531, "learning_rate": 3.5025806451612905e-05, "loss": 0.2765, "step": 4642 }, { "epoch": 29.927710843373493, "grad_norm": 11.400917053222656, "learning_rate": 3.502258064516129e-05, "loss": 0.2714, "step": 4643 }, { "epoch": 29.93413654618474, "grad_norm": 23.391292572021484, "learning_rate": 3.5019354838709676e-05, "loss": 0.3144, "step": 4644 }, { "epoch": 29.940562248995985, "grad_norm": 8.992465019226074, "learning_rate": 3.501612903225807e-05, "loss": 0.4456, "step": 4645 }, { "epoch": 29.94698795180723, "grad_norm": 3.2777185440063477, "learning_rate": 3.5012903225806454e-05, "loss": 0.0906, "step": 4646 }, { "epoch": 29.953413654618473, "grad_norm": 18.557254791259766, "learning_rate": 3.500967741935484e-05, "loss": 0.3902, "step": 4647 }, { "epoch": 29.95983935742972, "grad_norm": 19.950408935546875, "learning_rate": 3.500645161290323e-05, "loss": 0.6064, "step": 4648 }, { "epoch": 29.966265060240964, "grad_norm": 15.705765724182129, "learning_rate": 3.500322580645162e-05, "loss": 0.424, "step": 4649 }, { "epoch": 29.97269076305221, "grad_norm": 6.9212846755981445, "learning_rate": 3.5e-05, "loss": 0.2254, "step": 4650 }, { "epoch": 29.979116465863452, "grad_norm": 7.090415000915527, "learning_rate": 3.499677419354839e-05, "loss": 0.1738, "step": 4651 }, { "epoch": 29.985542168674698, "grad_norm": 12.372200012207031, "learning_rate": 3.4993548387096775e-05, "loss": 0.5801, "step": 4652 }, { "epoch": 29.991967871485944, "grad_norm": 21.470733642578125, "learning_rate": 3.499032258064516e-05, "loss": 0.2591, "step": 4653 }, { "epoch": 30.00160642570281, "grad_norm": 10.116034507751465, "learning_rate": 3.498709677419355e-05, "loss": 0.3537, "step": 4654 }, { "epoch": 30.008032128514056, "grad_norm": 64.2928695678711, "learning_rate": 3.498387096774194e-05, "loss": 0.625, "step": 4655 }, { "epoch": 30.014457831325302, "grad_norm": 10.380890846252441, "learning_rate": 3.4980645161290325e-05, "loss": 0.3396, "step": 4656 }, { "epoch": 30.020883534136548, "grad_norm": 21.00541877746582, "learning_rate": 3.497741935483872e-05, "loss": 0.4195, "step": 4657 }, { "epoch": 30.02730923694779, "grad_norm": 13.85729694366455, "learning_rate": 3.4974193548387096e-05, "loss": 0.4831, "step": 4658 }, { "epoch": 30.033734939759036, "grad_norm": 6.824368953704834, "learning_rate": 3.497096774193548e-05, "loss": 0.3935, "step": 4659 }, { "epoch": 30.04016064257028, "grad_norm": 10.120881080627441, "learning_rate": 3.4967741935483874e-05, "loss": 0.2568, "step": 4660 }, { "epoch": 30.046586345381527, "grad_norm": 26.54462432861328, "learning_rate": 3.496451612903226e-05, "loss": 0.4397, "step": 4661 }, { "epoch": 30.05301204819277, "grad_norm": 17.880359649658203, "learning_rate": 3.4961290322580646e-05, "loss": 0.2255, "step": 4662 }, { "epoch": 30.059437751004015, "grad_norm": 5.388489723205566, "learning_rate": 3.495806451612904e-05, "loss": 0.2146, "step": 4663 }, { "epoch": 30.06586345381526, "grad_norm": 8.721562385559082, "learning_rate": 3.4954838709677424e-05, "loss": 0.2509, "step": 4664 }, { "epoch": 30.072289156626507, "grad_norm": 11.676772117614746, "learning_rate": 3.49516129032258e-05, "loss": 0.3822, "step": 4665 }, { "epoch": 30.078714859437753, "grad_norm": 8.992287635803223, "learning_rate": 3.4948387096774195e-05, "loss": 0.3218, "step": 4666 }, { "epoch": 30.085140562248995, "grad_norm": 7.580782413482666, "learning_rate": 3.494516129032258e-05, "loss": 0.2938, "step": 4667 }, { "epoch": 30.09156626506024, "grad_norm": 20.80172348022461, "learning_rate": 3.494193548387097e-05, "loss": 0.2503, "step": 4668 }, { "epoch": 30.097991967871486, "grad_norm": 53.675872802734375, "learning_rate": 3.493870967741936e-05, "loss": 0.1582, "step": 4669 }, { "epoch": 30.104417670682732, "grad_norm": 32.95481872558594, "learning_rate": 3.4935483870967745e-05, "loss": 0.4591, "step": 4670 }, { "epoch": 30.110843373493974, "grad_norm": 8.642096519470215, "learning_rate": 3.493225806451613e-05, "loss": 0.7279, "step": 4671 }, { "epoch": 30.11726907630522, "grad_norm": 44.990665435791016, "learning_rate": 3.492903225806452e-05, "loss": 0.3928, "step": 4672 }, { "epoch": 30.123694779116466, "grad_norm": 4.273314952850342, "learning_rate": 3.49258064516129e-05, "loss": 0.266, "step": 4673 }, { "epoch": 30.13012048192771, "grad_norm": 5.82922887802124, "learning_rate": 3.492258064516129e-05, "loss": 0.326, "step": 4674 }, { "epoch": 30.136546184738958, "grad_norm": 10.731138229370117, "learning_rate": 3.491935483870968e-05, "loss": 0.4135, "step": 4675 }, { "epoch": 30.1429718875502, "grad_norm": 7.735910415649414, "learning_rate": 3.4916129032258066e-05, "loss": 0.2674, "step": 4676 }, { "epoch": 30.149397590361446, "grad_norm": 8.15026569366455, "learning_rate": 3.491290322580645e-05, "loss": 0.2413, "step": 4677 }, { "epoch": 30.15582329317269, "grad_norm": 8.778281211853027, "learning_rate": 3.4909677419354844e-05, "loss": 0.3868, "step": 4678 }, { "epoch": 30.162248995983937, "grad_norm": 4.467827320098877, "learning_rate": 3.490645161290323e-05, "loss": 0.2277, "step": 4679 }, { "epoch": 30.16867469879518, "grad_norm": 8.998677253723145, "learning_rate": 3.4903225806451616e-05, "loss": 0.2767, "step": 4680 }, { "epoch": 30.175100401606425, "grad_norm": 56.90595245361328, "learning_rate": 3.49e-05, "loss": 0.5007, "step": 4681 }, { "epoch": 30.18152610441767, "grad_norm": 3.848004102706909, "learning_rate": 3.489677419354839e-05, "loss": 0.2738, "step": 4682 }, { "epoch": 30.187951807228917, "grad_norm": 28.649700164794922, "learning_rate": 3.489354838709677e-05, "loss": 0.6195, "step": 4683 }, { "epoch": 30.194377510040162, "grad_norm": 7.581337928771973, "learning_rate": 3.4890322580645165e-05, "loss": 0.2407, "step": 4684 }, { "epoch": 30.200803212851405, "grad_norm": 9.950549125671387, "learning_rate": 3.488709677419355e-05, "loss": 0.5537, "step": 4685 }, { "epoch": 30.20722891566265, "grad_norm": 13.60373306274414, "learning_rate": 3.4883870967741937e-05, "loss": 0.4766, "step": 4686 }, { "epoch": 30.213654618473896, "grad_norm": 12.286860466003418, "learning_rate": 3.488064516129033e-05, "loss": 0.5908, "step": 4687 }, { "epoch": 30.220080321285142, "grad_norm": 9.434650421142578, "learning_rate": 3.4877419354838715e-05, "loss": 0.1089, "step": 4688 }, { "epoch": 30.226506024096384, "grad_norm": 7.490346431732178, "learning_rate": 3.4874193548387094e-05, "loss": 0.3972, "step": 4689 }, { "epoch": 30.23293172690763, "grad_norm": 9.068060874938965, "learning_rate": 3.4870967741935486e-05, "loss": 0.5381, "step": 4690 }, { "epoch": 30.239357429718876, "grad_norm": 22.214439392089844, "learning_rate": 3.486774193548387e-05, "loss": 0.8533, "step": 4691 }, { "epoch": 30.24578313253012, "grad_norm": 12.566352844238281, "learning_rate": 3.486451612903226e-05, "loss": 0.2995, "step": 4692 }, { "epoch": 30.252208835341367, "grad_norm": 7.526809215545654, "learning_rate": 3.486129032258065e-05, "loss": 0.3585, "step": 4693 }, { "epoch": 30.25863453815261, "grad_norm": 16.540752410888672, "learning_rate": 3.4858064516129036e-05, "loss": 0.4495, "step": 4694 }, { "epoch": 30.265060240963855, "grad_norm": 22.293062210083008, "learning_rate": 3.485483870967742e-05, "loss": 0.3975, "step": 4695 }, { "epoch": 30.2714859437751, "grad_norm": 5.961122512817383, "learning_rate": 3.485161290322581e-05, "loss": 0.2385, "step": 4696 }, { "epoch": 30.277911646586347, "grad_norm": 21.313703536987305, "learning_rate": 3.484838709677419e-05, "loss": 0.3843, "step": 4697 }, { "epoch": 30.28433734939759, "grad_norm": 13.86690902709961, "learning_rate": 3.484516129032258e-05, "loss": 0.4912, "step": 4698 }, { "epoch": 30.290763052208835, "grad_norm": 22.01365089416504, "learning_rate": 3.484193548387097e-05, "loss": 0.4225, "step": 4699 }, { "epoch": 30.29718875502008, "grad_norm": 16.77301025390625, "learning_rate": 3.483870967741936e-05, "loss": 0.5724, "step": 4700 }, { "epoch": 30.303614457831326, "grad_norm": 14.651970863342285, "learning_rate": 3.483548387096774e-05, "loss": 0.5684, "step": 4701 }, { "epoch": 30.310040160642572, "grad_norm": 6.224092960357666, "learning_rate": 3.4832258064516135e-05, "loss": 0.1741, "step": 4702 }, { "epoch": 30.316465863453814, "grad_norm": 16.684232711791992, "learning_rate": 3.482903225806452e-05, "loss": 0.3315, "step": 4703 }, { "epoch": 30.32289156626506, "grad_norm": 8.566338539123535, "learning_rate": 3.48258064516129e-05, "loss": 0.2686, "step": 4704 }, { "epoch": 30.329317269076306, "grad_norm": 16.13995361328125, "learning_rate": 3.482258064516129e-05, "loss": 0.813, "step": 4705 }, { "epoch": 30.33574297188755, "grad_norm": 23.622207641601562, "learning_rate": 3.481935483870968e-05, "loss": 0.5121, "step": 4706 }, { "epoch": 30.342168674698794, "grad_norm": 39.631107330322266, "learning_rate": 3.4816129032258064e-05, "loss": 0.3629, "step": 4707 }, { "epoch": 30.34859437751004, "grad_norm": 10.893561363220215, "learning_rate": 3.4812903225806456e-05, "loss": 0.3888, "step": 4708 }, { "epoch": 30.355020080321285, "grad_norm": 7.3543381690979, "learning_rate": 3.480967741935484e-05, "loss": 0.4268, "step": 4709 }, { "epoch": 30.36144578313253, "grad_norm": 2.788393974304199, "learning_rate": 3.480645161290323e-05, "loss": 0.1379, "step": 4710 }, { "epoch": 30.367871485943777, "grad_norm": 10.631136894226074, "learning_rate": 3.480322580645162e-05, "loss": 0.3542, "step": 4711 }, { "epoch": 30.37429718875502, "grad_norm": 13.893434524536133, "learning_rate": 3.48e-05, "loss": 0.4789, "step": 4712 }, { "epoch": 30.380722891566265, "grad_norm": 7.342748641967773, "learning_rate": 3.4796774193548385e-05, "loss": 0.3214, "step": 4713 }, { "epoch": 30.38714859437751, "grad_norm": 5.921910285949707, "learning_rate": 3.479354838709678e-05, "loss": 0.3064, "step": 4714 }, { "epoch": 30.393574297188756, "grad_norm": 10.756214141845703, "learning_rate": 3.479032258064516e-05, "loss": 0.3614, "step": 4715 }, { "epoch": 30.4, "grad_norm": 48.494110107421875, "learning_rate": 3.478709677419355e-05, "loss": 0.5227, "step": 4716 }, { "epoch": 30.406425702811244, "grad_norm": 10.223780632019043, "learning_rate": 3.478387096774194e-05, "loss": 0.5695, "step": 4717 }, { "epoch": 30.41285140562249, "grad_norm": 11.683802604675293, "learning_rate": 3.478064516129033e-05, "loss": 0.1904, "step": 4718 }, { "epoch": 30.419277108433736, "grad_norm": 3.8224050998687744, "learning_rate": 3.477741935483871e-05, "loss": 0.148, "step": 4719 }, { "epoch": 30.42570281124498, "grad_norm": 9.211217880249023, "learning_rate": 3.47741935483871e-05, "loss": 0.337, "step": 4720 }, { "epoch": 30.432128514056224, "grad_norm": 4.717530250549316, "learning_rate": 3.4770967741935484e-05, "loss": 0.1076, "step": 4721 }, { "epoch": 30.43855421686747, "grad_norm": 11.97302532196045, "learning_rate": 3.476774193548387e-05, "loss": 0.4131, "step": 4722 }, { "epoch": 30.444979919678715, "grad_norm": 9.433725357055664, "learning_rate": 3.476451612903226e-05, "loss": 0.3467, "step": 4723 }, { "epoch": 30.45140562248996, "grad_norm": 4.882419586181641, "learning_rate": 3.476129032258065e-05, "loss": 0.4912, "step": 4724 }, { "epoch": 30.457831325301203, "grad_norm": 9.374972343444824, "learning_rate": 3.475806451612903e-05, "loss": 0.4024, "step": 4725 }, { "epoch": 30.46425702811245, "grad_norm": 29.706863403320312, "learning_rate": 3.4754838709677426e-05, "loss": 0.441, "step": 4726 }, { "epoch": 30.470682730923695, "grad_norm": 22.66929054260254, "learning_rate": 3.475161290322581e-05, "loss": 0.6515, "step": 4727 }, { "epoch": 30.47710843373494, "grad_norm": 7.812443733215332, "learning_rate": 3.474838709677419e-05, "loss": 0.5112, "step": 4728 }, { "epoch": 30.483534136546183, "grad_norm": 12.08406925201416, "learning_rate": 3.474516129032258e-05, "loss": 0.3524, "step": 4729 }, { "epoch": 30.48995983935743, "grad_norm": 6.281970024108887, "learning_rate": 3.474193548387097e-05, "loss": 0.4149, "step": 4730 }, { "epoch": 30.496385542168674, "grad_norm": 140.68666076660156, "learning_rate": 3.4738709677419354e-05, "loss": 0.6603, "step": 4731 }, { "epoch": 30.50281124497992, "grad_norm": 4.880781173706055, "learning_rate": 3.473548387096775e-05, "loss": 0.2819, "step": 4732 }, { "epoch": 30.509236947791166, "grad_norm": 5.169513702392578, "learning_rate": 3.473225806451613e-05, "loss": 0.5655, "step": 4733 }, { "epoch": 30.51566265060241, "grad_norm": 10.540338516235352, "learning_rate": 3.472903225806452e-05, "loss": 0.61, "step": 4734 }, { "epoch": 30.522088353413654, "grad_norm": 6.493156433105469, "learning_rate": 3.4725806451612904e-05, "loss": 0.571, "step": 4735 }, { "epoch": 30.5285140562249, "grad_norm": 9.77694320678711, "learning_rate": 3.472258064516129e-05, "loss": 0.7046, "step": 4736 }, { "epoch": 30.534939759036146, "grad_norm": 11.16375732421875, "learning_rate": 3.4719354838709675e-05, "loss": 0.4748, "step": 4737 }, { "epoch": 30.54136546184739, "grad_norm": 10.89980697631836, "learning_rate": 3.471612903225807e-05, "loss": 0.3716, "step": 4738 }, { "epoch": 30.547791164658634, "grad_norm": 15.936318397521973, "learning_rate": 3.4712903225806454e-05, "loss": 0.4285, "step": 4739 }, { "epoch": 30.55421686746988, "grad_norm": 9.273109436035156, "learning_rate": 3.470967741935484e-05, "loss": 0.3466, "step": 4740 }, { "epoch": 30.560642570281125, "grad_norm": 18.03838539123535, "learning_rate": 3.470645161290323e-05, "loss": 0.5346, "step": 4741 }, { "epoch": 30.56706827309237, "grad_norm": 4.447118282318115, "learning_rate": 3.470322580645162e-05, "loss": 0.3432, "step": 4742 }, { "epoch": 30.573493975903613, "grad_norm": 13.52332878112793, "learning_rate": 3.4699999999999996e-05, "loss": 0.3432, "step": 4743 }, { "epoch": 30.57991967871486, "grad_norm": 12.325512886047363, "learning_rate": 3.469677419354839e-05, "loss": 0.633, "step": 4744 }, { "epoch": 30.586345381526105, "grad_norm": 17.381410598754883, "learning_rate": 3.4693548387096775e-05, "loss": 0.4355, "step": 4745 }, { "epoch": 30.59277108433735, "grad_norm": 3.691178321838379, "learning_rate": 3.469032258064516e-05, "loss": 0.3013, "step": 4746 }, { "epoch": 30.599196787148593, "grad_norm": 18.726837158203125, "learning_rate": 3.468709677419355e-05, "loss": 0.3892, "step": 4747 }, { "epoch": 30.60562248995984, "grad_norm": 10.567513465881348, "learning_rate": 3.468387096774194e-05, "loss": 0.414, "step": 4748 }, { "epoch": 30.612048192771084, "grad_norm": 2.870380401611328, "learning_rate": 3.4680645161290324e-05, "loss": 0.1027, "step": 4749 }, { "epoch": 30.61847389558233, "grad_norm": 20.106990814208984, "learning_rate": 3.467741935483872e-05, "loss": 0.9047, "step": 4750 }, { "epoch": 30.624899598393576, "grad_norm": 7.13020658493042, "learning_rate": 3.4674193548387096e-05, "loss": 0.4969, "step": 4751 }, { "epoch": 30.631325301204818, "grad_norm": 12.57993221282959, "learning_rate": 3.467096774193548e-05, "loss": 0.4056, "step": 4752 }, { "epoch": 30.637751004016064, "grad_norm": 6.647968292236328, "learning_rate": 3.4667741935483874e-05, "loss": 0.5739, "step": 4753 }, { "epoch": 30.64417670682731, "grad_norm": 6.530850887298584, "learning_rate": 3.466451612903226e-05, "loss": 0.517, "step": 4754 }, { "epoch": 30.650602409638555, "grad_norm": 7.211880683898926, "learning_rate": 3.4661290322580645e-05, "loss": 0.2687, "step": 4755 }, { "epoch": 30.6570281124498, "grad_norm": 6.9250407218933105, "learning_rate": 3.465806451612904e-05, "loss": 0.6246, "step": 4756 }, { "epoch": 30.663453815261043, "grad_norm": 14.664347648620605, "learning_rate": 3.4654838709677423e-05, "loss": 0.4085, "step": 4757 }, { "epoch": 30.66987951807229, "grad_norm": 13.064602851867676, "learning_rate": 3.465161290322581e-05, "loss": 0.6664, "step": 4758 }, { "epoch": 30.676305220883535, "grad_norm": 7.945522308349609, "learning_rate": 3.4648387096774195e-05, "loss": 0.4467, "step": 4759 }, { "epoch": 30.68273092369478, "grad_norm": 6.475647926330566, "learning_rate": 3.464516129032258e-05, "loss": 0.4126, "step": 4760 }, { "epoch": 30.689156626506023, "grad_norm": 5.80706262588501, "learning_rate": 3.4641935483870966e-05, "loss": 0.3144, "step": 4761 }, { "epoch": 30.69558232931727, "grad_norm": 5.307289123535156, "learning_rate": 3.463870967741936e-05, "loss": 0.371, "step": 4762 }, { "epoch": 30.702008032128514, "grad_norm": 8.874464988708496, "learning_rate": 3.4635483870967744e-05, "loss": 0.2514, "step": 4763 }, { "epoch": 30.70843373493976, "grad_norm": 73.15216064453125, "learning_rate": 3.463225806451613e-05, "loss": 0.2706, "step": 4764 }, { "epoch": 30.714859437751002, "grad_norm": 4.164714336395264, "learning_rate": 3.462903225806452e-05, "loss": 0.3714, "step": 4765 }, { "epoch": 30.721285140562248, "grad_norm": 3.7690813541412354, "learning_rate": 3.46258064516129e-05, "loss": 0.2249, "step": 4766 }, { "epoch": 30.727710843373494, "grad_norm": 14.356710433959961, "learning_rate": 3.462258064516129e-05, "loss": 0.7156, "step": 4767 }, { "epoch": 30.73413654618474, "grad_norm": 11.754584312438965, "learning_rate": 3.461935483870968e-05, "loss": 0.4539, "step": 4768 }, { "epoch": 30.740562248995985, "grad_norm": 13.782341957092285, "learning_rate": 3.4616129032258065e-05, "loss": 0.3949, "step": 4769 }, { "epoch": 30.746987951807228, "grad_norm": 13.075738906860352, "learning_rate": 3.461290322580645e-05, "loss": 0.3274, "step": 4770 }, { "epoch": 30.753413654618473, "grad_norm": 14.006796836853027, "learning_rate": 3.4609677419354844e-05, "loss": 0.7397, "step": 4771 }, { "epoch": 30.75983935742972, "grad_norm": 9.783364295959473, "learning_rate": 3.460645161290323e-05, "loss": 0.7371, "step": 4772 }, { "epoch": 30.766265060240965, "grad_norm": 7.334598541259766, "learning_rate": 3.4603225806451615e-05, "loss": 0.4455, "step": 4773 }, { "epoch": 30.77269076305221, "grad_norm": 14.776786804199219, "learning_rate": 3.46e-05, "loss": 0.5451, "step": 4774 }, { "epoch": 30.779116465863453, "grad_norm": 5.850310802459717, "learning_rate": 3.4596774193548386e-05, "loss": 0.299, "step": 4775 }, { "epoch": 30.7855421686747, "grad_norm": 4.828121185302734, "learning_rate": 3.459354838709677e-05, "loss": 0.3259, "step": 4776 }, { "epoch": 30.791967871485944, "grad_norm": 15.246344566345215, "learning_rate": 3.4590322580645165e-05, "loss": 0.688, "step": 4777 }, { "epoch": 30.79839357429719, "grad_norm": 5.969934940338135, "learning_rate": 3.458709677419355e-05, "loss": 0.1631, "step": 4778 }, { "epoch": 30.804819277108432, "grad_norm": 6.2513580322265625, "learning_rate": 3.4583870967741936e-05, "loss": 0.539, "step": 4779 }, { "epoch": 30.811244979919678, "grad_norm": 14.03940200805664, "learning_rate": 3.458064516129033e-05, "loss": 0.3506, "step": 4780 }, { "epoch": 30.817670682730924, "grad_norm": 21.918052673339844, "learning_rate": 3.4577419354838714e-05, "loss": 0.4367, "step": 4781 }, { "epoch": 30.82409638554217, "grad_norm": 19.069599151611328, "learning_rate": 3.457419354838709e-05, "loss": 0.3663, "step": 4782 }, { "epoch": 30.830522088353412, "grad_norm": 18.969409942626953, "learning_rate": 3.4570967741935486e-05, "loss": 0.5893, "step": 4783 }, { "epoch": 30.836947791164658, "grad_norm": 4.134420871734619, "learning_rate": 3.456774193548387e-05, "loss": 0.2581, "step": 4784 }, { "epoch": 30.843373493975903, "grad_norm": 12.034717559814453, "learning_rate": 3.456451612903226e-05, "loss": 0.5086, "step": 4785 }, { "epoch": 30.84979919678715, "grad_norm": 27.054550170898438, "learning_rate": 3.456129032258065e-05, "loss": 0.4761, "step": 4786 }, { "epoch": 30.856224899598395, "grad_norm": 37.371585845947266, "learning_rate": 3.4558064516129035e-05, "loss": 0.4446, "step": 4787 }, { "epoch": 30.862650602409637, "grad_norm": 12.680460929870605, "learning_rate": 3.455483870967742e-05, "loss": 0.3431, "step": 4788 }, { "epoch": 30.869076305220883, "grad_norm": 8.712678909301758, "learning_rate": 3.4551612903225813e-05, "loss": 0.3708, "step": 4789 }, { "epoch": 30.87550200803213, "grad_norm": 12.688152313232422, "learning_rate": 3.454838709677419e-05, "loss": 0.2318, "step": 4790 }, { "epoch": 30.881927710843375, "grad_norm": 37.96174240112305, "learning_rate": 3.454516129032258e-05, "loss": 0.5982, "step": 4791 }, { "epoch": 30.88835341365462, "grad_norm": 9.689079284667969, "learning_rate": 3.454193548387097e-05, "loss": 0.2406, "step": 4792 }, { "epoch": 30.894779116465863, "grad_norm": 8.3427152633667, "learning_rate": 3.4538709677419356e-05, "loss": 0.4513, "step": 4793 }, { "epoch": 30.90120481927711, "grad_norm": 13.315619468688965, "learning_rate": 3.453548387096774e-05, "loss": 0.5532, "step": 4794 }, { "epoch": 30.907630522088354, "grad_norm": 8.886188507080078, "learning_rate": 3.4532258064516134e-05, "loss": 0.494, "step": 4795 }, { "epoch": 30.9140562248996, "grad_norm": 16.436824798583984, "learning_rate": 3.452903225806452e-05, "loss": 0.3958, "step": 4796 }, { "epoch": 30.920481927710842, "grad_norm": 8.831319808959961, "learning_rate": 3.4525806451612906e-05, "loss": 0.392, "step": 4797 }, { "epoch": 30.926907630522088, "grad_norm": 12.565145492553711, "learning_rate": 3.452258064516129e-05, "loss": 0.5829, "step": 4798 }, { "epoch": 30.933333333333334, "grad_norm": 8.331588745117188, "learning_rate": 3.451935483870968e-05, "loss": 0.3461, "step": 4799 }, { "epoch": 30.93975903614458, "grad_norm": 5.085567951202393, "learning_rate": 3.451612903225806e-05, "loss": 0.3762, "step": 4800 }, { "epoch": 30.94618473895582, "grad_norm": 4.353314399719238, "learning_rate": 3.4512903225806456e-05, "loss": 0.2717, "step": 4801 }, { "epoch": 30.952610441767067, "grad_norm": 7.895242214202881, "learning_rate": 3.450967741935484e-05, "loss": 0.4087, "step": 4802 }, { "epoch": 30.959036144578313, "grad_norm": 5.583315849304199, "learning_rate": 3.450645161290323e-05, "loss": 0.284, "step": 4803 }, { "epoch": 30.96546184738956, "grad_norm": 12.36276626586914, "learning_rate": 3.450322580645162e-05, "loss": 0.7326, "step": 4804 }, { "epoch": 30.971887550200805, "grad_norm": 15.292177200317383, "learning_rate": 3.45e-05, "loss": 0.5425, "step": 4805 }, { "epoch": 30.978313253012047, "grad_norm": 20.838666915893555, "learning_rate": 3.4496774193548384e-05, "loss": 0.499, "step": 4806 }, { "epoch": 30.984738955823293, "grad_norm": 9.431572914123535, "learning_rate": 3.4493548387096777e-05, "loss": 0.2688, "step": 4807 }, { "epoch": 30.99116465863454, "grad_norm": 12.164697647094727, "learning_rate": 3.449032258064516e-05, "loss": 0.6601, "step": 4808 }, { "epoch": 31.000803212851405, "grad_norm": 12.518818855285645, "learning_rate": 3.448709677419355e-05, "loss": 0.3119, "step": 4809 }, { "epoch": 31.00722891566265, "grad_norm": 64.92481231689453, "learning_rate": 3.448387096774194e-05, "loss": 0.3993, "step": 4810 }, { "epoch": 31.013654618473897, "grad_norm": 3.275960922241211, "learning_rate": 3.4480645161290326e-05, "loss": 0.2224, "step": 4811 }, { "epoch": 31.02008032128514, "grad_norm": 6.1343183517456055, "learning_rate": 3.447741935483871e-05, "loss": 0.1944, "step": 4812 }, { "epoch": 31.026506024096385, "grad_norm": 4.799317836761475, "learning_rate": 3.44741935483871e-05, "loss": 0.3883, "step": 4813 }, { "epoch": 31.03293172690763, "grad_norm": 15.117255210876465, "learning_rate": 3.447096774193548e-05, "loss": 0.3616, "step": 4814 }, { "epoch": 31.039357429718876, "grad_norm": 9.995397567749023, "learning_rate": 3.446774193548387e-05, "loss": 0.3685, "step": 4815 }, { "epoch": 31.045783132530122, "grad_norm": 7.215076446533203, "learning_rate": 3.446451612903226e-05, "loss": 0.2329, "step": 4816 }, { "epoch": 31.052208835341364, "grad_norm": 6.821028709411621, "learning_rate": 3.446129032258065e-05, "loss": 0.5459, "step": 4817 }, { "epoch": 31.05863453815261, "grad_norm": 7.712496757507324, "learning_rate": 3.445806451612903e-05, "loss": 0.4402, "step": 4818 }, { "epoch": 31.065060240963856, "grad_norm": 8.21654224395752, "learning_rate": 3.4454838709677425e-05, "loss": 0.2209, "step": 4819 }, { "epoch": 31.0714859437751, "grad_norm": 11.595871925354004, "learning_rate": 3.445161290322581e-05, "loss": 0.3373, "step": 4820 }, { "epoch": 31.077911646586344, "grad_norm": 12.584671974182129, "learning_rate": 3.444838709677419e-05, "loss": 0.4795, "step": 4821 }, { "epoch": 31.08433734939759, "grad_norm": 12.353883743286133, "learning_rate": 3.444516129032258e-05, "loss": 0.5609, "step": 4822 }, { "epoch": 31.090763052208835, "grad_norm": 5.230612754821777, "learning_rate": 3.444193548387097e-05, "loss": 0.2013, "step": 4823 }, { "epoch": 31.09718875502008, "grad_norm": 38.467899322509766, "learning_rate": 3.4438709677419354e-05, "loss": 0.5071, "step": 4824 }, { "epoch": 31.103614457831327, "grad_norm": 5.14841890335083, "learning_rate": 3.4435483870967746e-05, "loss": 0.2425, "step": 4825 }, { "epoch": 31.11004016064257, "grad_norm": 13.691155433654785, "learning_rate": 3.443225806451613e-05, "loss": 0.3718, "step": 4826 }, { "epoch": 31.116465863453815, "grad_norm": 10.407744407653809, "learning_rate": 3.442903225806452e-05, "loss": 0.6807, "step": 4827 }, { "epoch": 31.12289156626506, "grad_norm": 4.169535160064697, "learning_rate": 3.442580645161291e-05, "loss": 0.2109, "step": 4828 }, { "epoch": 31.129317269076306, "grad_norm": 7.427701473236084, "learning_rate": 3.442258064516129e-05, "loss": 0.3527, "step": 4829 }, { "epoch": 31.13574297188755, "grad_norm": 9.187349319458008, "learning_rate": 3.4419354838709675e-05, "loss": 0.3748, "step": 4830 }, { "epoch": 31.142168674698794, "grad_norm": 13.336970329284668, "learning_rate": 3.441612903225807e-05, "loss": 0.4713, "step": 4831 }, { "epoch": 31.14859437751004, "grad_norm": 8.548060417175293, "learning_rate": 3.441290322580645e-05, "loss": 0.568, "step": 4832 }, { "epoch": 31.155020080321286, "grad_norm": 12.989392280578613, "learning_rate": 3.440967741935484e-05, "loss": 0.2738, "step": 4833 }, { "epoch": 31.16144578313253, "grad_norm": 8.400954246520996, "learning_rate": 3.440645161290323e-05, "loss": 0.5375, "step": 4834 }, { "epoch": 31.167871485943774, "grad_norm": 9.974100112915039, "learning_rate": 3.440322580645162e-05, "loss": 0.2433, "step": 4835 }, { "epoch": 31.17429718875502, "grad_norm": 11.234983444213867, "learning_rate": 3.4399999999999996e-05, "loss": 0.4147, "step": 4836 }, { "epoch": 31.180722891566266, "grad_norm": 9.991860389709473, "learning_rate": 3.439677419354839e-05, "loss": 0.4504, "step": 4837 }, { "epoch": 31.18714859437751, "grad_norm": 6.359915256500244, "learning_rate": 3.4393548387096774e-05, "loss": 0.3135, "step": 4838 }, { "epoch": 31.193574297188754, "grad_norm": 10.825556755065918, "learning_rate": 3.439032258064516e-05, "loss": 0.4346, "step": 4839 }, { "epoch": 31.2, "grad_norm": 16.867708206176758, "learning_rate": 3.438709677419355e-05, "loss": 0.5943, "step": 4840 }, { "epoch": 31.206425702811245, "grad_norm": 9.90170955657959, "learning_rate": 3.438387096774194e-05, "loss": 0.4512, "step": 4841 }, { "epoch": 31.21285140562249, "grad_norm": 5.727481365203857, "learning_rate": 3.4380645161290324e-05, "loss": 0.1232, "step": 4842 }, { "epoch": 31.219277108433737, "grad_norm": 9.173099517822266, "learning_rate": 3.4377419354838716e-05, "loss": 0.3941, "step": 4843 }, { "epoch": 31.22570281124498, "grad_norm": 12.251770973205566, "learning_rate": 3.4374193548387095e-05, "loss": 0.2779, "step": 4844 }, { "epoch": 31.232128514056225, "grad_norm": 10.684024810791016, "learning_rate": 3.437096774193548e-05, "loss": 0.6208, "step": 4845 }, { "epoch": 31.23855421686747, "grad_norm": 7.057666301727295, "learning_rate": 3.436774193548387e-05, "loss": 0.4796, "step": 4846 }, { "epoch": 31.244979919678716, "grad_norm": 19.62442398071289, "learning_rate": 3.436451612903226e-05, "loss": 0.5365, "step": 4847 }, { "epoch": 31.25140562248996, "grad_norm": 20.49799156188965, "learning_rate": 3.4361290322580645e-05, "loss": 0.228, "step": 4848 }, { "epoch": 31.257831325301204, "grad_norm": 20.987009048461914, "learning_rate": 3.435806451612904e-05, "loss": 0.5498, "step": 4849 }, { "epoch": 31.26425702811245, "grad_norm": 10.664656639099121, "learning_rate": 3.435483870967742e-05, "loss": 0.7214, "step": 4850 }, { "epoch": 31.270682730923696, "grad_norm": 12.154764175415039, "learning_rate": 3.435161290322581e-05, "loss": 0.4316, "step": 4851 }, { "epoch": 31.27710843373494, "grad_norm": 12.026331901550293, "learning_rate": 3.4348387096774194e-05, "loss": 0.3958, "step": 4852 }, { "epoch": 31.283534136546184, "grad_norm": 8.943201065063477, "learning_rate": 3.434516129032258e-05, "loss": 0.3897, "step": 4853 }, { "epoch": 31.28995983935743, "grad_norm": 12.711644172668457, "learning_rate": 3.4341935483870966e-05, "loss": 0.4427, "step": 4854 }, { "epoch": 31.296385542168675, "grad_norm": 14.993087768554688, "learning_rate": 3.433870967741936e-05, "loss": 0.683, "step": 4855 }, { "epoch": 31.30281124497992, "grad_norm": 14.060883522033691, "learning_rate": 3.4335483870967744e-05, "loss": 0.527, "step": 4856 }, { "epoch": 31.309236947791163, "grad_norm": 8.444601058959961, "learning_rate": 3.433225806451613e-05, "loss": 0.4681, "step": 4857 }, { "epoch": 31.31566265060241, "grad_norm": 5.825417518615723, "learning_rate": 3.432903225806452e-05, "loss": 0.4059, "step": 4858 }, { "epoch": 31.322088353413655, "grad_norm": 10.970907211303711, "learning_rate": 3.432580645161291e-05, "loss": 0.4677, "step": 4859 }, { "epoch": 31.3285140562249, "grad_norm": 11.442695617675781, "learning_rate": 3.432258064516129e-05, "loss": 0.3384, "step": 4860 }, { "epoch": 31.334939759036146, "grad_norm": 8.07578182220459, "learning_rate": 3.431935483870968e-05, "loss": 0.2963, "step": 4861 }, { "epoch": 31.34136546184739, "grad_norm": 12.65831470489502, "learning_rate": 3.4316129032258065e-05, "loss": 0.3944, "step": 4862 }, { "epoch": 31.347791164658634, "grad_norm": 6.7806315422058105, "learning_rate": 3.431290322580645e-05, "loss": 0.278, "step": 4863 }, { "epoch": 31.35421686746988, "grad_norm": 5.676780700683594, "learning_rate": 3.430967741935484e-05, "loss": 0.2607, "step": 4864 }, { "epoch": 31.360642570281126, "grad_norm": 10.435507774353027, "learning_rate": 3.430645161290323e-05, "loss": 0.749, "step": 4865 }, { "epoch": 31.367068273092368, "grad_norm": 9.081254005432129, "learning_rate": 3.4303225806451615e-05, "loss": 0.4989, "step": 4866 }, { "epoch": 31.373493975903614, "grad_norm": 11.456199645996094, "learning_rate": 3.430000000000001e-05, "loss": 0.4042, "step": 4867 }, { "epoch": 31.37991967871486, "grad_norm": 10.472579002380371, "learning_rate": 3.4296774193548386e-05, "loss": 0.4411, "step": 4868 }, { "epoch": 31.386345381526105, "grad_norm": 5.046639919281006, "learning_rate": 3.429354838709677e-05, "loss": 0.2135, "step": 4869 }, { "epoch": 31.39277108433735, "grad_norm": 9.137092590332031, "learning_rate": 3.4290322580645164e-05, "loss": 0.4184, "step": 4870 }, { "epoch": 31.399196787148593, "grad_norm": 7.991880893707275, "learning_rate": 3.428709677419355e-05, "loss": 0.5243, "step": 4871 }, { "epoch": 31.40562248995984, "grad_norm": 12.165648460388184, "learning_rate": 3.4283870967741936e-05, "loss": 0.5558, "step": 4872 }, { "epoch": 31.412048192771085, "grad_norm": 10.835708618164062, "learning_rate": 3.428064516129033e-05, "loss": 0.3405, "step": 4873 }, { "epoch": 31.41847389558233, "grad_norm": 8.606618881225586, "learning_rate": 3.4277419354838714e-05, "loss": 0.3788, "step": 4874 }, { "epoch": 31.424899598393573, "grad_norm": 10.743022918701172, "learning_rate": 3.427419354838709e-05, "loss": 0.3737, "step": 4875 }, { "epoch": 31.43132530120482, "grad_norm": 11.251443862915039, "learning_rate": 3.4270967741935485e-05, "loss": 0.2732, "step": 4876 }, { "epoch": 31.437751004016064, "grad_norm": 11.753552436828613, "learning_rate": 3.426774193548387e-05, "loss": 0.8349, "step": 4877 }, { "epoch": 31.44417670682731, "grad_norm": 11.250746726989746, "learning_rate": 3.4264516129032257e-05, "loss": 0.4557, "step": 4878 }, { "epoch": 31.450602409638556, "grad_norm": 12.481562614440918, "learning_rate": 3.426129032258065e-05, "loss": 0.469, "step": 4879 }, { "epoch": 31.457028112449798, "grad_norm": 21.34601593017578, "learning_rate": 3.4258064516129035e-05, "loss": 0.152, "step": 4880 }, { "epoch": 31.463453815261044, "grad_norm": 15.1026029586792, "learning_rate": 3.425483870967742e-05, "loss": 0.143, "step": 4881 }, { "epoch": 31.46987951807229, "grad_norm": 22.156206130981445, "learning_rate": 3.425161290322581e-05, "loss": 0.5579, "step": 4882 }, { "epoch": 31.476305220883535, "grad_norm": 17.434385299682617, "learning_rate": 3.424838709677419e-05, "loss": 0.4838, "step": 4883 }, { "epoch": 31.482730923694778, "grad_norm": 7.035708904266357, "learning_rate": 3.424516129032258e-05, "loss": 0.2849, "step": 4884 }, { "epoch": 31.489156626506023, "grad_norm": 18.921274185180664, "learning_rate": 3.424193548387097e-05, "loss": 0.4897, "step": 4885 }, { "epoch": 31.49558232931727, "grad_norm": 20.71222686767578, "learning_rate": 3.4238709677419356e-05, "loss": 0.3836, "step": 4886 }, { "epoch": 31.502008032128515, "grad_norm": 5.488966941833496, "learning_rate": 3.423548387096774e-05, "loss": 0.4013, "step": 4887 }, { "epoch": 31.50843373493976, "grad_norm": 3.5079455375671387, "learning_rate": 3.4232258064516134e-05, "loss": 0.1936, "step": 4888 }, { "epoch": 31.514859437751003, "grad_norm": 6.092197895050049, "learning_rate": 3.422903225806452e-05, "loss": 0.238, "step": 4889 }, { "epoch": 31.52128514056225, "grad_norm": 11.624839782714844, "learning_rate": 3.4225806451612905e-05, "loss": 0.3253, "step": 4890 }, { "epoch": 31.527710843373494, "grad_norm": 7.597599029541016, "learning_rate": 3.422258064516129e-05, "loss": 0.5836, "step": 4891 }, { "epoch": 31.53413654618474, "grad_norm": 8.467843055725098, "learning_rate": 3.421935483870968e-05, "loss": 0.4034, "step": 4892 }, { "epoch": 31.540562248995982, "grad_norm": 7.572290420532227, "learning_rate": 3.421612903225806e-05, "loss": 0.5992, "step": 4893 }, { "epoch": 31.54698795180723, "grad_norm": 5.157258987426758, "learning_rate": 3.4212903225806455e-05, "loss": 0.1911, "step": 4894 }, { "epoch": 31.553413654618474, "grad_norm": 11.708489418029785, "learning_rate": 3.420967741935484e-05, "loss": 0.2413, "step": 4895 }, { "epoch": 31.55983935742972, "grad_norm": 3.8122329711914062, "learning_rate": 3.4206451612903226e-05, "loss": 0.217, "step": 4896 }, { "epoch": 31.566265060240966, "grad_norm": 8.099039077758789, "learning_rate": 3.420322580645162e-05, "loss": 0.5016, "step": 4897 }, { "epoch": 31.572690763052208, "grad_norm": 6.667783260345459, "learning_rate": 3.4200000000000005e-05, "loss": 0.4512, "step": 4898 }, { "epoch": 31.579116465863454, "grad_norm": 8.844972610473633, "learning_rate": 3.4196774193548384e-05, "loss": 0.3114, "step": 4899 }, { "epoch": 31.5855421686747, "grad_norm": 33.420372009277344, "learning_rate": 3.4193548387096776e-05, "loss": 0.2798, "step": 4900 }, { "epoch": 31.591967871485945, "grad_norm": 5.944699287414551, "learning_rate": 3.419032258064516e-05, "loss": 0.3604, "step": 4901 }, { "epoch": 31.598393574297187, "grad_norm": 48.24216079711914, "learning_rate": 3.418709677419355e-05, "loss": 0.3918, "step": 4902 }, { "epoch": 31.604819277108433, "grad_norm": 9.531773567199707, "learning_rate": 3.418387096774194e-05, "loss": 0.3104, "step": 4903 }, { "epoch": 31.61124497991968, "grad_norm": 6.924391269683838, "learning_rate": 3.4180645161290326e-05, "loss": 0.5693, "step": 4904 }, { "epoch": 31.617670682730925, "grad_norm": 7.82374382019043, "learning_rate": 3.417741935483871e-05, "loss": 0.188, "step": 4905 }, { "epoch": 31.62409638554217, "grad_norm": 10.98991870880127, "learning_rate": 3.41741935483871e-05, "loss": 0.4859, "step": 4906 }, { "epoch": 31.630522088353413, "grad_norm": 5.740747928619385, "learning_rate": 3.417096774193548e-05, "loss": 0.4405, "step": 4907 }, { "epoch": 31.63694779116466, "grad_norm": 35.94227981567383, "learning_rate": 3.416774193548387e-05, "loss": 0.4601, "step": 4908 }, { "epoch": 31.643373493975904, "grad_norm": 16.861572265625, "learning_rate": 3.416451612903226e-05, "loss": 0.485, "step": 4909 }, { "epoch": 31.64979919678715, "grad_norm": 11.367107391357422, "learning_rate": 3.416129032258065e-05, "loss": 0.3257, "step": 4910 }, { "epoch": 31.656224899598392, "grad_norm": 10.35955810546875, "learning_rate": 3.415806451612903e-05, "loss": 0.7196, "step": 4911 }, { "epoch": 31.662650602409638, "grad_norm": 18.169652938842773, "learning_rate": 3.4154838709677425e-05, "loss": 0.4604, "step": 4912 }, { "epoch": 31.669076305220884, "grad_norm": 13.844120979309082, "learning_rate": 3.415161290322581e-05, "loss": 0.7599, "step": 4913 }, { "epoch": 31.67550200803213, "grad_norm": 10.647388458251953, "learning_rate": 3.414838709677419e-05, "loss": 0.6257, "step": 4914 }, { "epoch": 31.681927710843375, "grad_norm": 5.533727645874023, "learning_rate": 3.414516129032258e-05, "loss": 0.2133, "step": 4915 }, { "epoch": 31.688353413654617, "grad_norm": 9.717207908630371, "learning_rate": 3.414193548387097e-05, "loss": 0.4359, "step": 4916 }, { "epoch": 31.694779116465863, "grad_norm": 4.5330634117126465, "learning_rate": 3.413870967741935e-05, "loss": 0.188, "step": 4917 }, { "epoch": 31.70120481927711, "grad_norm": 6.919510841369629, "learning_rate": 3.4135483870967746e-05, "loss": 0.105, "step": 4918 }, { "epoch": 31.707630522088355, "grad_norm": 13.408164978027344, "learning_rate": 3.413225806451613e-05, "loss": 0.464, "step": 4919 }, { "epoch": 31.714056224899597, "grad_norm": 11.11752700805664, "learning_rate": 3.412903225806452e-05, "loss": 0.4269, "step": 4920 }, { "epoch": 31.720481927710843, "grad_norm": 3.006042957305908, "learning_rate": 3.412580645161291e-05, "loss": 0.2305, "step": 4921 }, { "epoch": 31.72690763052209, "grad_norm": 9.753578186035156, "learning_rate": 3.412258064516129e-05, "loss": 0.2261, "step": 4922 }, { "epoch": 31.733333333333334, "grad_norm": 9.861339569091797, "learning_rate": 3.4119354838709674e-05, "loss": 0.3553, "step": 4923 }, { "epoch": 31.739759036144576, "grad_norm": 8.409860610961914, "learning_rate": 3.411612903225807e-05, "loss": 0.4439, "step": 4924 }, { "epoch": 31.746184738955822, "grad_norm": 10.438196182250977, "learning_rate": 3.411290322580645e-05, "loss": 0.6153, "step": 4925 }, { "epoch": 31.752610441767068, "grad_norm": 8.519087791442871, "learning_rate": 3.410967741935484e-05, "loss": 0.3398, "step": 4926 }, { "epoch": 31.759036144578314, "grad_norm": 7.500062942504883, "learning_rate": 3.410645161290323e-05, "loss": 0.4701, "step": 4927 }, { "epoch": 31.76546184738956, "grad_norm": 20.48846435546875, "learning_rate": 3.4103225806451616e-05, "loss": 0.4858, "step": 4928 }, { "epoch": 31.771887550200802, "grad_norm": 6.9436516761779785, "learning_rate": 3.41e-05, "loss": 0.3056, "step": 4929 }, { "epoch": 31.778313253012048, "grad_norm": 12.100627899169922, "learning_rate": 3.409677419354839e-05, "loss": 0.52, "step": 4930 }, { "epoch": 31.784738955823293, "grad_norm": 12.258086204528809, "learning_rate": 3.4093548387096774e-05, "loss": 0.56, "step": 4931 }, { "epoch": 31.79116465863454, "grad_norm": 20.842622756958008, "learning_rate": 3.409032258064516e-05, "loss": 0.5915, "step": 4932 }, { "epoch": 31.797590361445785, "grad_norm": 8.346820831298828, "learning_rate": 3.408709677419355e-05, "loss": 0.3057, "step": 4933 }, { "epoch": 31.804016064257027, "grad_norm": 6.1718363761901855, "learning_rate": 3.408387096774194e-05, "loss": 0.262, "step": 4934 }, { "epoch": 31.810441767068273, "grad_norm": 7.345884799957275, "learning_rate": 3.408064516129032e-05, "loss": 0.4175, "step": 4935 }, { "epoch": 31.81686746987952, "grad_norm": 6.648449897766113, "learning_rate": 3.4077419354838716e-05, "loss": 0.3723, "step": 4936 }, { "epoch": 31.823293172690764, "grad_norm": 8.315817832946777, "learning_rate": 3.40741935483871e-05, "loss": 0.4482, "step": 4937 }, { "epoch": 31.829718875502007, "grad_norm": 5.174087047576904, "learning_rate": 3.407096774193548e-05, "loss": 0.4207, "step": 4938 }, { "epoch": 31.836144578313252, "grad_norm": 14.390337944030762, "learning_rate": 3.406774193548387e-05, "loss": 0.3391, "step": 4939 }, { "epoch": 31.842570281124498, "grad_norm": 9.118597030639648, "learning_rate": 3.406451612903226e-05, "loss": 0.3349, "step": 4940 }, { "epoch": 31.848995983935744, "grad_norm": 5.614077091217041, "learning_rate": 3.4061290322580644e-05, "loss": 0.2521, "step": 4941 }, { "epoch": 31.855421686746986, "grad_norm": 7.3776092529296875, "learning_rate": 3.405806451612904e-05, "loss": 0.6277, "step": 4942 }, { "epoch": 31.861847389558232, "grad_norm": 12.465639114379883, "learning_rate": 3.405483870967742e-05, "loss": 0.3928, "step": 4943 }, { "epoch": 31.868273092369478, "grad_norm": 6.221485137939453, "learning_rate": 3.405161290322581e-05, "loss": 0.1959, "step": 4944 }, { "epoch": 31.874698795180723, "grad_norm": 66.59747314453125, "learning_rate": 3.4048387096774194e-05, "loss": 0.5387, "step": 4945 }, { "epoch": 31.88112449799197, "grad_norm": 10.624136924743652, "learning_rate": 3.404516129032258e-05, "loss": 0.3529, "step": 4946 }, { "epoch": 31.88755020080321, "grad_norm": 7.587216377258301, "learning_rate": 3.4041935483870965e-05, "loss": 0.2636, "step": 4947 }, { "epoch": 31.893975903614457, "grad_norm": 12.660070419311523, "learning_rate": 3.403870967741936e-05, "loss": 0.5777, "step": 4948 }, { "epoch": 31.900401606425703, "grad_norm": 8.296857833862305, "learning_rate": 3.4035483870967743e-05, "loss": 0.3896, "step": 4949 }, { "epoch": 31.90682730923695, "grad_norm": 7.494345664978027, "learning_rate": 3.403225806451613e-05, "loss": 0.3402, "step": 4950 }, { "epoch": 31.913253012048195, "grad_norm": 7.603762626647949, "learning_rate": 3.402903225806452e-05, "loss": 0.2375, "step": 4951 }, { "epoch": 31.919678714859437, "grad_norm": 3.5509204864501953, "learning_rate": 3.402580645161291e-05, "loss": 0.1875, "step": 4952 }, { "epoch": 31.926104417670683, "grad_norm": 29.07172393798828, "learning_rate": 3.4022580645161286e-05, "loss": 0.4039, "step": 4953 }, { "epoch": 31.93253012048193, "grad_norm": 7.8196120262146, "learning_rate": 3.401935483870968e-05, "loss": 0.4654, "step": 4954 }, { "epoch": 31.938955823293174, "grad_norm": 12.666004180908203, "learning_rate": 3.4016129032258064e-05, "loss": 0.4107, "step": 4955 }, { "epoch": 31.945381526104416, "grad_norm": 5.361547470092773, "learning_rate": 3.401290322580645e-05, "loss": 0.3941, "step": 4956 }, { "epoch": 31.951807228915662, "grad_norm": 12.512109756469727, "learning_rate": 3.400967741935484e-05, "loss": 0.3579, "step": 4957 }, { "epoch": 31.958232931726908, "grad_norm": 8.250642776489258, "learning_rate": 3.400645161290323e-05, "loss": 0.3314, "step": 4958 }, { "epoch": 31.964658634538154, "grad_norm": 18.126174926757812, "learning_rate": 3.4003225806451614e-05, "loss": 0.3109, "step": 4959 }, { "epoch": 31.971084337349396, "grad_norm": 9.785690307617188, "learning_rate": 3.4000000000000007e-05, "loss": 0.2564, "step": 4960 }, { "epoch": 31.97751004016064, "grad_norm": 7.876813888549805, "learning_rate": 3.3996774193548385e-05, "loss": 0.3257, "step": 4961 }, { "epoch": 31.983935742971887, "grad_norm": 11.123963356018066, "learning_rate": 3.399354838709677e-05, "loss": 0.5465, "step": 4962 }, { "epoch": 31.990361445783133, "grad_norm": 8.461526870727539, "learning_rate": 3.3990322580645164e-05, "loss": 0.3548, "step": 4963 }, { "epoch": 31.99678714859438, "grad_norm": 9.515050888061523, "learning_rate": 3.398709677419355e-05, "loss": 0.2838, "step": 4964 }, { "epoch": 32.00642570281124, "grad_norm": 9.40609359741211, "learning_rate": 3.3983870967741935e-05, "loss": 0.5542, "step": 4965 }, { "epoch": 32.01285140562249, "grad_norm": 6.570440292358398, "learning_rate": 3.398064516129033e-05, "loss": 0.2014, "step": 4966 }, { "epoch": 32.019277108433734, "grad_norm": 11.935186386108398, "learning_rate": 3.397741935483871e-05, "loss": 0.3513, "step": 4967 }, { "epoch": 32.02570281124498, "grad_norm": 18.681978225708008, "learning_rate": 3.39741935483871e-05, "loss": 0.3235, "step": 4968 }, { "epoch": 32.032128514056225, "grad_norm": 18.745018005371094, "learning_rate": 3.3970967741935485e-05, "loss": 0.7955, "step": 4969 }, { "epoch": 32.03855421686747, "grad_norm": 8.381315231323242, "learning_rate": 3.396774193548387e-05, "loss": 0.3958, "step": 4970 }, { "epoch": 32.04497991967872, "grad_norm": 26.04526138305664, "learning_rate": 3.3964516129032256e-05, "loss": 0.5505, "step": 4971 }, { "epoch": 32.05140562248996, "grad_norm": 12.308353424072266, "learning_rate": 3.396129032258065e-05, "loss": 0.4636, "step": 4972 }, { "epoch": 32.0578313253012, "grad_norm": 6.398587226867676, "learning_rate": 3.3958064516129034e-05, "loss": 0.4118, "step": 4973 }, { "epoch": 32.06425702811245, "grad_norm": 7.293213367462158, "learning_rate": 3.395483870967742e-05, "loss": 0.2856, "step": 4974 }, { "epoch": 32.07068273092369, "grad_norm": 6.736018180847168, "learning_rate": 3.395161290322581e-05, "loss": 0.6653, "step": 4975 }, { "epoch": 32.07710843373494, "grad_norm": 20.791847229003906, "learning_rate": 3.39483870967742e-05, "loss": 0.4331, "step": 4976 }, { "epoch": 32.083534136546184, "grad_norm": 11.898348808288574, "learning_rate": 3.394516129032258e-05, "loss": 0.2554, "step": 4977 }, { "epoch": 32.08995983935743, "grad_norm": 10.132537841796875, "learning_rate": 3.394193548387097e-05, "loss": 0.3234, "step": 4978 }, { "epoch": 32.096385542168676, "grad_norm": 11.143256187438965, "learning_rate": 3.3938709677419355e-05, "loss": 0.4041, "step": 4979 }, { "epoch": 32.10281124497992, "grad_norm": 5.630221366882324, "learning_rate": 3.393548387096774e-05, "loss": 0.2489, "step": 4980 }, { "epoch": 32.10923694779117, "grad_norm": 6.730753421783447, "learning_rate": 3.3932258064516133e-05, "loss": 0.3009, "step": 4981 }, { "epoch": 32.11566265060241, "grad_norm": 10.053725242614746, "learning_rate": 3.392903225806452e-05, "loss": 0.4649, "step": 4982 }, { "epoch": 32.12208835341365, "grad_norm": 9.980286598205566, "learning_rate": 3.3925806451612905e-05, "loss": 0.5116, "step": 4983 }, { "epoch": 32.1285140562249, "grad_norm": 6.502622604370117, "learning_rate": 3.392258064516129e-05, "loss": 0.183, "step": 4984 }, { "epoch": 32.13493975903614, "grad_norm": 10.773932456970215, "learning_rate": 3.3919354838709676e-05, "loss": 0.2717, "step": 4985 }, { "epoch": 32.14136546184739, "grad_norm": 11.753673553466797, "learning_rate": 3.391612903225806e-05, "loss": 0.1731, "step": 4986 }, { "epoch": 32.147791164658635, "grad_norm": 10.650184631347656, "learning_rate": 3.3912903225806454e-05, "loss": 0.3909, "step": 4987 }, { "epoch": 32.15421686746988, "grad_norm": 6.143210411071777, "learning_rate": 3.390967741935484e-05, "loss": 0.344, "step": 4988 }, { "epoch": 32.16064257028113, "grad_norm": 8.448853492736816, "learning_rate": 3.3906451612903226e-05, "loss": 0.4042, "step": 4989 }, { "epoch": 32.16706827309237, "grad_norm": 11.272095680236816, "learning_rate": 3.390322580645162e-05, "loss": 0.5293, "step": 4990 }, { "epoch": 32.17349397590361, "grad_norm": 29.943960189819336, "learning_rate": 3.3900000000000004e-05, "loss": 0.3777, "step": 4991 }, { "epoch": 32.17991967871486, "grad_norm": 5.137420654296875, "learning_rate": 3.389677419354838e-05, "loss": 0.211, "step": 4992 }, { "epoch": 32.1863453815261, "grad_norm": 14.269719123840332, "learning_rate": 3.3893548387096775e-05, "loss": 0.5484, "step": 4993 }, { "epoch": 32.19277108433735, "grad_norm": 8.091329574584961, "learning_rate": 3.389032258064516e-05, "loss": 0.3273, "step": 4994 }, { "epoch": 32.199196787148594, "grad_norm": 9.742180824279785, "learning_rate": 3.388709677419355e-05, "loss": 0.3512, "step": 4995 }, { "epoch": 32.205622489959836, "grad_norm": 5.409653186798096, "learning_rate": 3.388387096774194e-05, "loss": 0.5078, "step": 4996 }, { "epoch": 32.212048192771086, "grad_norm": 10.68325424194336, "learning_rate": 3.3880645161290325e-05, "loss": 0.4024, "step": 4997 }, { "epoch": 32.21847389558233, "grad_norm": 13.66907024383545, "learning_rate": 3.387741935483871e-05, "loss": 0.3109, "step": 4998 }, { "epoch": 32.22489959839358, "grad_norm": 9.847329139709473, "learning_rate": 3.38741935483871e-05, "loss": 0.4206, "step": 4999 }, { "epoch": 32.23132530120482, "grad_norm": 9.731266975402832, "learning_rate": 3.387096774193548e-05, "loss": 0.3765, "step": 5000 }, { "epoch": 32.23775100401606, "grad_norm": 9.961601257324219, "learning_rate": 3.386774193548387e-05, "loss": 0.374, "step": 5001 }, { "epoch": 32.24417670682731, "grad_norm": 3.0801022052764893, "learning_rate": 3.386451612903226e-05, "loss": 0.1137, "step": 5002 }, { "epoch": 32.25060240963855, "grad_norm": 9.170217514038086, "learning_rate": 3.3861290322580646e-05, "loss": 0.2522, "step": 5003 }, { "epoch": 32.2570281124498, "grad_norm": 8.345402717590332, "learning_rate": 3.385806451612903e-05, "loss": 0.334, "step": 5004 }, { "epoch": 32.263453815261045, "grad_norm": 12.97867202758789, "learning_rate": 3.3854838709677424e-05, "loss": 0.4996, "step": 5005 }, { "epoch": 32.26987951807229, "grad_norm": 28.508296966552734, "learning_rate": 3.385161290322581e-05, "loss": 0.3811, "step": 5006 }, { "epoch": 32.276305220883536, "grad_norm": 3.2929646968841553, "learning_rate": 3.3848387096774196e-05, "loss": 0.1211, "step": 5007 }, { "epoch": 32.28273092369478, "grad_norm": 9.836917877197266, "learning_rate": 3.384516129032258e-05, "loss": 0.6521, "step": 5008 }, { "epoch": 32.28915662650602, "grad_norm": 10.2801513671875, "learning_rate": 3.384193548387097e-05, "loss": 0.6406, "step": 5009 }, { "epoch": 32.29558232931727, "grad_norm": 11.163098335266113, "learning_rate": 3.383870967741935e-05, "loss": 0.5535, "step": 5010 }, { "epoch": 32.30200803212851, "grad_norm": 3.582934856414795, "learning_rate": 3.3835483870967745e-05, "loss": 0.271, "step": 5011 }, { "epoch": 32.30843373493976, "grad_norm": 11.678404808044434, "learning_rate": 3.383225806451613e-05, "loss": 0.4606, "step": 5012 }, { "epoch": 32.314859437751004, "grad_norm": 5.172752857208252, "learning_rate": 3.382903225806452e-05, "loss": 0.2296, "step": 5013 }, { "epoch": 32.321285140562246, "grad_norm": 5.47462272644043, "learning_rate": 3.382580645161291e-05, "loss": 0.337, "step": 5014 }, { "epoch": 32.327710843373495, "grad_norm": 10.73227310180664, "learning_rate": 3.382258064516129e-05, "loss": 0.5193, "step": 5015 }, { "epoch": 32.33413654618474, "grad_norm": 14.887462615966797, "learning_rate": 3.3819354838709674e-05, "loss": 0.3752, "step": 5016 }, { "epoch": 32.34056224899599, "grad_norm": 4.866388320922852, "learning_rate": 3.3816129032258066e-05, "loss": 0.3142, "step": 5017 }, { "epoch": 32.34698795180723, "grad_norm": 10.946242332458496, "learning_rate": 3.381290322580645e-05, "loss": 0.3411, "step": 5018 }, { "epoch": 32.35341365461847, "grad_norm": 10.54703426361084, "learning_rate": 3.380967741935484e-05, "loss": 0.4346, "step": 5019 }, { "epoch": 32.35983935742972, "grad_norm": 30.525110244750977, "learning_rate": 3.380645161290323e-05, "loss": 0.4047, "step": 5020 }, { "epoch": 32.36626506024096, "grad_norm": 18.16117286682129, "learning_rate": 3.3803225806451616e-05, "loss": 0.4932, "step": 5021 }, { "epoch": 32.37269076305221, "grad_norm": 8.005058288574219, "learning_rate": 3.38e-05, "loss": 0.4437, "step": 5022 }, { "epoch": 32.379116465863454, "grad_norm": 15.118056297302246, "learning_rate": 3.379677419354839e-05, "loss": 0.4159, "step": 5023 }, { "epoch": 32.3855421686747, "grad_norm": 7.172489166259766, "learning_rate": 3.379354838709677e-05, "loss": 0.3173, "step": 5024 }, { "epoch": 32.391967871485946, "grad_norm": 7.840950012207031, "learning_rate": 3.379032258064516e-05, "loss": 0.543, "step": 5025 }, { "epoch": 32.39839357429719, "grad_norm": 15.173521041870117, "learning_rate": 3.378709677419355e-05, "loss": 0.2966, "step": 5026 }, { "epoch": 32.40481927710843, "grad_norm": 16.73192024230957, "learning_rate": 3.378387096774194e-05, "loss": 0.3818, "step": 5027 }, { "epoch": 32.41124497991968, "grad_norm": 13.168893814086914, "learning_rate": 3.378064516129032e-05, "loss": 0.5515, "step": 5028 }, { "epoch": 32.41767068273092, "grad_norm": 11.273982048034668, "learning_rate": 3.3777419354838715e-05, "loss": 0.5238, "step": 5029 }, { "epoch": 32.42409638554217, "grad_norm": 4.083312034606934, "learning_rate": 3.37741935483871e-05, "loss": 0.1643, "step": 5030 }, { "epoch": 32.43052208835341, "grad_norm": 8.299437522888184, "learning_rate": 3.377096774193548e-05, "loss": 0.6351, "step": 5031 }, { "epoch": 32.436947791164656, "grad_norm": 8.306997299194336, "learning_rate": 3.376774193548387e-05, "loss": 0.5573, "step": 5032 }, { "epoch": 32.443373493975905, "grad_norm": 7.657629489898682, "learning_rate": 3.376451612903226e-05, "loss": 0.2787, "step": 5033 }, { "epoch": 32.44979919678715, "grad_norm": 12.384552001953125, "learning_rate": 3.3761290322580644e-05, "loss": 0.6521, "step": 5034 }, { "epoch": 32.456224899598396, "grad_norm": 5.351294040679932, "learning_rate": 3.3758064516129036e-05, "loss": 0.3133, "step": 5035 }, { "epoch": 32.46265060240964, "grad_norm": 5.760582447052002, "learning_rate": 3.375483870967742e-05, "loss": 0.2264, "step": 5036 }, { "epoch": 32.46907630522088, "grad_norm": 4.657513618469238, "learning_rate": 3.375161290322581e-05, "loss": 0.1732, "step": 5037 }, { "epoch": 32.47550200803213, "grad_norm": 6.641335487365723, "learning_rate": 3.37483870967742e-05, "loss": 0.3037, "step": 5038 }, { "epoch": 32.48192771084337, "grad_norm": 12.90054702758789, "learning_rate": 3.374516129032258e-05, "loss": 0.3876, "step": 5039 }, { "epoch": 32.48835341365462, "grad_norm": 6.851719379425049, "learning_rate": 3.3741935483870965e-05, "loss": 0.2756, "step": 5040 }, { "epoch": 32.494779116465864, "grad_norm": 8.49681282043457, "learning_rate": 3.373870967741936e-05, "loss": 0.3361, "step": 5041 }, { "epoch": 32.501204819277106, "grad_norm": 9.063687324523926, "learning_rate": 3.373548387096774e-05, "loss": 0.3022, "step": 5042 }, { "epoch": 32.507630522088355, "grad_norm": 7.473578453063965, "learning_rate": 3.373225806451613e-05, "loss": 0.2468, "step": 5043 }, { "epoch": 32.5140562248996, "grad_norm": 6.44875431060791, "learning_rate": 3.372903225806452e-05, "loss": 0.3408, "step": 5044 }, { "epoch": 32.52048192771084, "grad_norm": 12.498738288879395, "learning_rate": 3.372580645161291e-05, "loss": 0.8577, "step": 5045 }, { "epoch": 32.52690763052209, "grad_norm": 14.054572105407715, "learning_rate": 3.372258064516129e-05, "loss": 0.6787, "step": 5046 }, { "epoch": 32.53333333333333, "grad_norm": 13.146361351013184, "learning_rate": 3.371935483870968e-05, "loss": 0.7099, "step": 5047 }, { "epoch": 32.53975903614458, "grad_norm": 16.298561096191406, "learning_rate": 3.3716129032258064e-05, "loss": 0.7604, "step": 5048 }, { "epoch": 32.54618473895582, "grad_norm": 7.883800983428955, "learning_rate": 3.371290322580645e-05, "loss": 0.4724, "step": 5049 }, { "epoch": 32.552610441767065, "grad_norm": 17.65770149230957, "learning_rate": 3.370967741935484e-05, "loss": 0.441, "step": 5050 }, { "epoch": 32.559036144578315, "grad_norm": 7.330566883087158, "learning_rate": 3.370645161290323e-05, "loss": 0.4088, "step": 5051 }, { "epoch": 32.56546184738956, "grad_norm": 14.614306449890137, "learning_rate": 3.3703225806451614e-05, "loss": 0.3403, "step": 5052 }, { "epoch": 32.571887550200806, "grad_norm": 5.1206488609313965, "learning_rate": 3.3700000000000006e-05, "loss": 0.2285, "step": 5053 }, { "epoch": 32.57831325301205, "grad_norm": 10.363106727600098, "learning_rate": 3.3696774193548385e-05, "loss": 0.4643, "step": 5054 }, { "epoch": 32.58473895582329, "grad_norm": 8.932682037353516, "learning_rate": 3.369354838709677e-05, "loss": 0.2082, "step": 5055 }, { "epoch": 32.59116465863454, "grad_norm": 8.560995101928711, "learning_rate": 3.369032258064516e-05, "loss": 0.3216, "step": 5056 }, { "epoch": 32.59759036144578, "grad_norm": 5.254962921142578, "learning_rate": 3.368709677419355e-05, "loss": 0.3375, "step": 5057 }, { "epoch": 32.60401606425703, "grad_norm": 7.156805992126465, "learning_rate": 3.3683870967741935e-05, "loss": 0.3386, "step": 5058 }, { "epoch": 32.610441767068274, "grad_norm": 7.371613502502441, "learning_rate": 3.368064516129033e-05, "loss": 0.3952, "step": 5059 }, { "epoch": 32.616867469879516, "grad_norm": 16.123058319091797, "learning_rate": 3.367741935483871e-05, "loss": 0.526, "step": 5060 }, { "epoch": 32.623293172690765, "grad_norm": 20.4727783203125, "learning_rate": 3.36741935483871e-05, "loss": 0.4466, "step": 5061 }, { "epoch": 32.62971887550201, "grad_norm": 10.780274391174316, "learning_rate": 3.3670967741935484e-05, "loss": 0.3602, "step": 5062 }, { "epoch": 32.63614457831325, "grad_norm": 7.496013164520264, "learning_rate": 3.366774193548387e-05, "loss": 0.3813, "step": 5063 }, { "epoch": 32.6425702811245, "grad_norm": 12.352389335632324, "learning_rate": 3.3664516129032256e-05, "loss": 0.2508, "step": 5064 }, { "epoch": 32.64899598393574, "grad_norm": 7.686119556427002, "learning_rate": 3.366129032258065e-05, "loss": 0.3699, "step": 5065 }, { "epoch": 32.65542168674699, "grad_norm": 9.85208797454834, "learning_rate": 3.3658064516129034e-05, "loss": 0.4138, "step": 5066 }, { "epoch": 32.66184738955823, "grad_norm": 9.99354362487793, "learning_rate": 3.365483870967742e-05, "loss": 0.6804, "step": 5067 }, { "epoch": 32.668273092369475, "grad_norm": 4.382440090179443, "learning_rate": 3.365161290322581e-05, "loss": 0.1762, "step": 5068 }, { "epoch": 32.674698795180724, "grad_norm": 11.280421257019043, "learning_rate": 3.36483870967742e-05, "loss": 0.2784, "step": 5069 }, { "epoch": 32.681124497991966, "grad_norm": 6.809791564941406, "learning_rate": 3.3645161290322577e-05, "loss": 0.3586, "step": 5070 }, { "epoch": 32.687550200803216, "grad_norm": 5.399776935577393, "learning_rate": 3.364193548387097e-05, "loss": 0.2842, "step": 5071 }, { "epoch": 32.69397590361446, "grad_norm": 21.313735961914062, "learning_rate": 3.3638709677419355e-05, "loss": 0.2816, "step": 5072 }, { "epoch": 32.7004016064257, "grad_norm": 10.894521713256836, "learning_rate": 3.363548387096774e-05, "loss": 0.3641, "step": 5073 }, { "epoch": 32.70682730923695, "grad_norm": 22.539648056030273, "learning_rate": 3.363225806451613e-05, "loss": 0.3668, "step": 5074 }, { "epoch": 32.71325301204819, "grad_norm": 5.421961784362793, "learning_rate": 3.362903225806452e-05, "loss": 0.2616, "step": 5075 }, { "epoch": 32.71967871485944, "grad_norm": 6.978095531463623, "learning_rate": 3.3625806451612904e-05, "loss": 0.2401, "step": 5076 }, { "epoch": 32.72610441767068, "grad_norm": 7.541128158569336, "learning_rate": 3.36225806451613e-05, "loss": 0.4851, "step": 5077 }, { "epoch": 32.732530120481925, "grad_norm": 9.797240257263184, "learning_rate": 3.3619354838709676e-05, "loss": 0.2752, "step": 5078 }, { "epoch": 32.738955823293175, "grad_norm": 12.50036907196045, "learning_rate": 3.361612903225806e-05, "loss": 0.4103, "step": 5079 }, { "epoch": 32.74538152610442, "grad_norm": 4.38849401473999, "learning_rate": 3.3612903225806454e-05, "loss": 0.1629, "step": 5080 }, { "epoch": 32.75180722891566, "grad_norm": 20.86431884765625, "learning_rate": 3.360967741935484e-05, "loss": 0.4998, "step": 5081 }, { "epoch": 32.75823293172691, "grad_norm": 8.00760555267334, "learning_rate": 3.3606451612903225e-05, "loss": 0.2272, "step": 5082 }, { "epoch": 32.76465863453815, "grad_norm": 7.303984642028809, "learning_rate": 3.360322580645162e-05, "loss": 0.5893, "step": 5083 }, { "epoch": 32.7710843373494, "grad_norm": 11.320795059204102, "learning_rate": 3.3600000000000004e-05, "loss": 0.5789, "step": 5084 }, { "epoch": 32.77751004016064, "grad_norm": 8.184514999389648, "learning_rate": 3.359677419354839e-05, "loss": 0.5404, "step": 5085 }, { "epoch": 32.783935742971885, "grad_norm": 4.144282817840576, "learning_rate": 3.3593548387096775e-05, "loss": 0.2532, "step": 5086 }, { "epoch": 32.790361445783134, "grad_norm": 12.754829406738281, "learning_rate": 3.359032258064516e-05, "loss": 0.3649, "step": 5087 }, { "epoch": 32.796787148594376, "grad_norm": 10.824511528015137, "learning_rate": 3.3587096774193546e-05, "loss": 0.2298, "step": 5088 }, { "epoch": 32.803212851405625, "grad_norm": 8.604979515075684, "learning_rate": 3.358387096774194e-05, "loss": 0.2136, "step": 5089 }, { "epoch": 32.80963855421687, "grad_norm": 8.839131355285645, "learning_rate": 3.3580645161290325e-05, "loss": 0.2212, "step": 5090 }, { "epoch": 32.81606425702811, "grad_norm": 9.920293807983398, "learning_rate": 3.357741935483871e-05, "loss": 0.3226, "step": 5091 }, { "epoch": 32.82248995983936, "grad_norm": 15.792895317077637, "learning_rate": 3.35741935483871e-05, "loss": 0.4799, "step": 5092 }, { "epoch": 32.8289156626506, "grad_norm": 15.934447288513184, "learning_rate": 3.357096774193548e-05, "loss": 0.5208, "step": 5093 }, { "epoch": 32.83534136546185, "grad_norm": 10.921751976013184, "learning_rate": 3.356774193548387e-05, "loss": 0.4832, "step": 5094 }, { "epoch": 32.84176706827309, "grad_norm": 7.769094944000244, "learning_rate": 3.356451612903226e-05, "loss": 0.2748, "step": 5095 }, { "epoch": 32.848192771084335, "grad_norm": 8.74075984954834, "learning_rate": 3.3561290322580646e-05, "loss": 0.3887, "step": 5096 }, { "epoch": 32.854618473895584, "grad_norm": 7.7968430519104, "learning_rate": 3.355806451612903e-05, "loss": 0.3988, "step": 5097 }, { "epoch": 32.86104417670683, "grad_norm": 12.011748313903809, "learning_rate": 3.3554838709677424e-05, "loss": 0.4293, "step": 5098 }, { "epoch": 32.86746987951807, "grad_norm": 7.903160095214844, "learning_rate": 3.355161290322581e-05, "loss": 0.2728, "step": 5099 }, { "epoch": 32.87389558232932, "grad_norm": 21.283348083496094, "learning_rate": 3.3548387096774195e-05, "loss": 0.4684, "step": 5100 }, { "epoch": 32.88032128514056, "grad_norm": 7.2879180908203125, "learning_rate": 3.354516129032258e-05, "loss": 0.301, "step": 5101 }, { "epoch": 32.88674698795181, "grad_norm": 18.778913497924805, "learning_rate": 3.354193548387097e-05, "loss": 0.2053, "step": 5102 }, { "epoch": 32.89317269076305, "grad_norm": 7.587935924530029, "learning_rate": 3.353870967741935e-05, "loss": 0.2761, "step": 5103 }, { "epoch": 32.899598393574294, "grad_norm": 13.120819091796875, "learning_rate": 3.3535483870967745e-05, "loss": 0.5347, "step": 5104 }, { "epoch": 32.90602409638554, "grad_norm": 6.814049243927002, "learning_rate": 3.353225806451613e-05, "loss": 0.305, "step": 5105 }, { "epoch": 32.912449799196786, "grad_norm": 5.33759069442749, "learning_rate": 3.3529032258064516e-05, "loss": 0.2543, "step": 5106 }, { "epoch": 32.918875502008035, "grad_norm": 17.6287899017334, "learning_rate": 3.352580645161291e-05, "loss": 0.513, "step": 5107 }, { "epoch": 32.92530120481928, "grad_norm": 9.067915916442871, "learning_rate": 3.3522580645161294e-05, "loss": 0.4111, "step": 5108 }, { "epoch": 32.93172690763052, "grad_norm": 11.230489730834961, "learning_rate": 3.351935483870968e-05, "loss": 0.6276, "step": 5109 }, { "epoch": 32.93815261044177, "grad_norm": 7.579833507537842, "learning_rate": 3.3516129032258066e-05, "loss": 0.6147, "step": 5110 }, { "epoch": 32.94457831325301, "grad_norm": 5.267874717712402, "learning_rate": 3.351290322580645e-05, "loss": 0.2332, "step": 5111 }, { "epoch": 32.95100401606426, "grad_norm": 10.671943664550781, "learning_rate": 3.350967741935484e-05, "loss": 0.2156, "step": 5112 }, { "epoch": 32.9574297188755, "grad_norm": 15.636852264404297, "learning_rate": 3.350645161290323e-05, "loss": 0.6915, "step": 5113 }, { "epoch": 32.963855421686745, "grad_norm": 14.445869445800781, "learning_rate": 3.3503225806451615e-05, "loss": 0.4656, "step": 5114 }, { "epoch": 32.970281124497994, "grad_norm": 4.994806289672852, "learning_rate": 3.35e-05, "loss": 0.2123, "step": 5115 }, { "epoch": 32.976706827309236, "grad_norm": 6.116507530212402, "learning_rate": 3.3496774193548394e-05, "loss": 0.2677, "step": 5116 }, { "epoch": 32.98313253012048, "grad_norm": 12.012433052062988, "learning_rate": 3.349354838709677e-05, "loss": 0.3897, "step": 5117 }, { "epoch": 32.98955823293173, "grad_norm": 6.6964569091796875, "learning_rate": 3.349032258064516e-05, "loss": 0.3613, "step": 5118 }, { "epoch": 32.99598393574297, "grad_norm": 6.253640651702881, "learning_rate": 3.348709677419355e-05, "loss": 0.3462, "step": 5119 }, { "epoch": 33.00562248995984, "grad_norm": 5.782875061035156, "learning_rate": 3.3483870967741936e-05, "loss": 0.445, "step": 5120 }, { "epoch": 33.01204819277108, "grad_norm": 11.054311752319336, "learning_rate": 3.348064516129032e-05, "loss": 0.436, "step": 5121 }, { "epoch": 33.01847389558233, "grad_norm": 5.52626895904541, "learning_rate": 3.3477419354838715e-05, "loss": 0.4787, "step": 5122 }, { "epoch": 33.024899598393574, "grad_norm": 9.582834243774414, "learning_rate": 3.34741935483871e-05, "loss": 0.5379, "step": 5123 }, { "epoch": 33.03132530120482, "grad_norm": 16.092056274414062, "learning_rate": 3.3470967741935486e-05, "loss": 0.5593, "step": 5124 }, { "epoch": 33.037751004016066, "grad_norm": 22.406442642211914, "learning_rate": 3.346774193548387e-05, "loss": 0.4229, "step": 5125 }, { "epoch": 33.04417670682731, "grad_norm": 9.648240089416504, "learning_rate": 3.346451612903226e-05, "loss": 0.3315, "step": 5126 }, { "epoch": 33.05060240963856, "grad_norm": 14.769278526306152, "learning_rate": 3.346129032258064e-05, "loss": 0.4905, "step": 5127 }, { "epoch": 33.0570281124498, "grad_norm": 9.054863929748535, "learning_rate": 3.3458064516129036e-05, "loss": 0.4047, "step": 5128 }, { "epoch": 33.06345381526104, "grad_norm": 2.351970672607422, "learning_rate": 3.345483870967742e-05, "loss": 0.1553, "step": 5129 }, { "epoch": 33.06987951807229, "grad_norm": 3.93961238861084, "learning_rate": 3.345161290322581e-05, "loss": 0.2461, "step": 5130 }, { "epoch": 33.07630522088353, "grad_norm": 14.294519424438477, "learning_rate": 3.34483870967742e-05, "loss": 0.5101, "step": 5131 }, { "epoch": 33.08273092369478, "grad_norm": 9.143071174621582, "learning_rate": 3.344516129032258e-05, "loss": 0.2199, "step": 5132 }, { "epoch": 33.089156626506025, "grad_norm": 26.57145881652832, "learning_rate": 3.3441935483870964e-05, "loss": 0.3045, "step": 5133 }, { "epoch": 33.09558232931727, "grad_norm": 31.5360164642334, "learning_rate": 3.343870967741936e-05, "loss": 0.3673, "step": 5134 }, { "epoch": 33.102008032128516, "grad_norm": 13.080411911010742, "learning_rate": 3.343548387096774e-05, "loss": 0.4004, "step": 5135 }, { "epoch": 33.10843373493976, "grad_norm": 22.433542251586914, "learning_rate": 3.343225806451613e-05, "loss": 0.3866, "step": 5136 }, { "epoch": 33.114859437751, "grad_norm": 3.3597686290740967, "learning_rate": 3.342903225806452e-05, "loss": 0.4394, "step": 5137 }, { "epoch": 33.12128514056225, "grad_norm": 5.710781574249268, "learning_rate": 3.3425806451612906e-05, "loss": 0.3802, "step": 5138 }, { "epoch": 33.12771084337349, "grad_norm": 11.925800323486328, "learning_rate": 3.342258064516129e-05, "loss": 0.2402, "step": 5139 }, { "epoch": 33.13413654618474, "grad_norm": 14.353483200073242, "learning_rate": 3.341935483870968e-05, "loss": 0.3086, "step": 5140 }, { "epoch": 33.140562248995984, "grad_norm": 16.931076049804688, "learning_rate": 3.3416129032258063e-05, "loss": 0.3786, "step": 5141 }, { "epoch": 33.146987951807226, "grad_norm": 10.32979965209961, "learning_rate": 3.341290322580645e-05, "loss": 0.6545, "step": 5142 }, { "epoch": 33.153413654618475, "grad_norm": 6.14680814743042, "learning_rate": 3.340967741935484e-05, "loss": 0.287, "step": 5143 }, { "epoch": 33.15983935742972, "grad_norm": 5.997348785400391, "learning_rate": 3.340645161290323e-05, "loss": 0.3276, "step": 5144 }, { "epoch": 33.16626506024097, "grad_norm": 11.437928199768066, "learning_rate": 3.340322580645161e-05, "loss": 0.4642, "step": 5145 }, { "epoch": 33.17269076305221, "grad_norm": 11.78586196899414, "learning_rate": 3.3400000000000005e-05, "loss": 0.302, "step": 5146 }, { "epoch": 33.17911646586345, "grad_norm": 16.44386100769043, "learning_rate": 3.339677419354839e-05, "loss": 0.3062, "step": 5147 }, { "epoch": 33.1855421686747, "grad_norm": 7.2427849769592285, "learning_rate": 3.339354838709678e-05, "loss": 0.123, "step": 5148 }, { "epoch": 33.19196787148594, "grad_norm": 15.364038467407227, "learning_rate": 3.339032258064516e-05, "loss": 0.3203, "step": 5149 }, { "epoch": 33.19839357429719, "grad_norm": 6.255305767059326, "learning_rate": 3.338709677419355e-05, "loss": 0.2164, "step": 5150 }, { "epoch": 33.204819277108435, "grad_norm": 3.9021265506744385, "learning_rate": 3.3383870967741934e-05, "loss": 0.18, "step": 5151 }, { "epoch": 33.21124497991968, "grad_norm": 15.852017402648926, "learning_rate": 3.3380645161290326e-05, "loss": 0.468, "step": 5152 }, { "epoch": 33.217670682730926, "grad_norm": 16.43008041381836, "learning_rate": 3.337741935483871e-05, "loss": 0.3048, "step": 5153 }, { "epoch": 33.22409638554217, "grad_norm": 4.836376667022705, "learning_rate": 3.33741935483871e-05, "loss": 0.2296, "step": 5154 }, { "epoch": 33.23052208835341, "grad_norm": 10.429946899414062, "learning_rate": 3.3370967741935484e-05, "loss": 0.5233, "step": 5155 }, { "epoch": 33.23694779116466, "grad_norm": 4.601631164550781, "learning_rate": 3.336774193548387e-05, "loss": 0.2313, "step": 5156 }, { "epoch": 33.2433734939759, "grad_norm": 14.774652481079102, "learning_rate": 3.3364516129032255e-05, "loss": 0.6357, "step": 5157 }, { "epoch": 33.24979919678715, "grad_norm": 12.073070526123047, "learning_rate": 3.336129032258065e-05, "loss": 0.145, "step": 5158 }, { "epoch": 33.256224899598394, "grad_norm": 13.161429405212402, "learning_rate": 3.335806451612903e-05, "loss": 0.1396, "step": 5159 }, { "epoch": 33.262650602409636, "grad_norm": 4.929690361022949, "learning_rate": 3.335483870967742e-05, "loss": 0.5564, "step": 5160 }, { "epoch": 33.269076305220885, "grad_norm": 10.061797142028809, "learning_rate": 3.335161290322581e-05, "loss": 0.4914, "step": 5161 }, { "epoch": 33.27550200803213, "grad_norm": 13.493315696716309, "learning_rate": 3.33483870967742e-05, "loss": 0.2794, "step": 5162 }, { "epoch": 33.28192771084338, "grad_norm": 21.888776779174805, "learning_rate": 3.334516129032258e-05, "loss": 0.4982, "step": 5163 }, { "epoch": 33.28835341365462, "grad_norm": 12.673678398132324, "learning_rate": 3.334193548387097e-05, "loss": 0.3833, "step": 5164 }, { "epoch": 33.29477911646586, "grad_norm": 4.462961673736572, "learning_rate": 3.3338709677419354e-05, "loss": 0.2415, "step": 5165 }, { "epoch": 33.30120481927711, "grad_norm": 15.551483154296875, "learning_rate": 3.333548387096774e-05, "loss": 0.2902, "step": 5166 }, { "epoch": 33.30763052208835, "grad_norm": 5.6094818115234375, "learning_rate": 3.333225806451613e-05, "loss": 0.3574, "step": 5167 }, { "epoch": 33.3140562248996, "grad_norm": 7.467251300811768, "learning_rate": 3.332903225806452e-05, "loss": 0.3868, "step": 5168 }, { "epoch": 33.320481927710844, "grad_norm": 4.16880464553833, "learning_rate": 3.3325806451612904e-05, "loss": 0.1553, "step": 5169 }, { "epoch": 33.326907630522086, "grad_norm": 27.510988235473633, "learning_rate": 3.3322580645161296e-05, "loss": 0.6247, "step": 5170 }, { "epoch": 33.333333333333336, "grad_norm": 18.113189697265625, "learning_rate": 3.3319354838709675e-05, "loss": 0.5188, "step": 5171 }, { "epoch": 33.33975903614458, "grad_norm": 10.194452285766602, "learning_rate": 3.331612903225807e-05, "loss": 0.3318, "step": 5172 }, { "epoch": 33.34618473895582, "grad_norm": 13.484223365783691, "learning_rate": 3.3312903225806453e-05, "loss": 0.5396, "step": 5173 }, { "epoch": 33.35261044176707, "grad_norm": 36.04429244995117, "learning_rate": 3.330967741935484e-05, "loss": 0.6185, "step": 5174 }, { "epoch": 33.35903614457831, "grad_norm": 14.09532642364502, "learning_rate": 3.3306451612903225e-05, "loss": 0.4011, "step": 5175 }, { "epoch": 33.36546184738956, "grad_norm": 10.85665512084961, "learning_rate": 3.330322580645162e-05, "loss": 0.2739, "step": 5176 }, { "epoch": 33.3718875502008, "grad_norm": 8.738136291503906, "learning_rate": 3.33e-05, "loss": 0.3038, "step": 5177 }, { "epoch": 33.378313253012045, "grad_norm": 16.27664566040039, "learning_rate": 3.329677419354839e-05, "loss": 0.5888, "step": 5178 }, { "epoch": 33.384738955823295, "grad_norm": 6.046577453613281, "learning_rate": 3.3293548387096774e-05, "loss": 0.2991, "step": 5179 }, { "epoch": 33.39116465863454, "grad_norm": 18.556617736816406, "learning_rate": 3.329032258064516e-05, "loss": 0.4812, "step": 5180 }, { "epoch": 33.397590361445786, "grad_norm": 21.26654815673828, "learning_rate": 3.3287096774193546e-05, "loss": 0.3433, "step": 5181 }, { "epoch": 33.40401606425703, "grad_norm": 20.215927124023438, "learning_rate": 3.328387096774194e-05, "loss": 0.7065, "step": 5182 }, { "epoch": 33.41044176706827, "grad_norm": 16.09368133544922, "learning_rate": 3.3280645161290324e-05, "loss": 0.3749, "step": 5183 }, { "epoch": 33.41686746987952, "grad_norm": 3.3133151531219482, "learning_rate": 3.327741935483871e-05, "loss": 0.5087, "step": 5184 }, { "epoch": 33.42329317269076, "grad_norm": 8.975677490234375, "learning_rate": 3.32741935483871e-05, "loss": 0.4504, "step": 5185 }, { "epoch": 33.429718875502004, "grad_norm": 6.498867988586426, "learning_rate": 3.327096774193549e-05, "loss": 0.3196, "step": 5186 }, { "epoch": 33.436144578313254, "grad_norm": 27.371559143066406, "learning_rate": 3.3267741935483874e-05, "loss": 0.5103, "step": 5187 }, { "epoch": 33.442570281124496, "grad_norm": 1.6298147439956665, "learning_rate": 3.326451612903226e-05, "loss": 0.4336, "step": 5188 }, { "epoch": 33.448995983935745, "grad_norm": 9.605629920959473, "learning_rate": 3.3261290322580645e-05, "loss": 0.3292, "step": 5189 }, { "epoch": 33.45542168674699, "grad_norm": 19.144725799560547, "learning_rate": 3.325806451612903e-05, "loss": 0.4815, "step": 5190 }, { "epoch": 33.46184738955823, "grad_norm": 6.563143253326416, "learning_rate": 3.325483870967742e-05, "loss": 0.4682, "step": 5191 }, { "epoch": 33.46827309236948, "grad_norm": 5.437958717346191, "learning_rate": 3.325161290322581e-05, "loss": 0.4588, "step": 5192 }, { "epoch": 33.47469879518072, "grad_norm": 9.897064208984375, "learning_rate": 3.3248387096774195e-05, "loss": 0.3216, "step": 5193 }, { "epoch": 33.48112449799197, "grad_norm": 11.157413482666016, "learning_rate": 3.324516129032258e-05, "loss": 0.2628, "step": 5194 }, { "epoch": 33.48755020080321, "grad_norm": 7.260941028594971, "learning_rate": 3.3241935483870966e-05, "loss": 0.4157, "step": 5195 }, { "epoch": 33.493975903614455, "grad_norm": 17.571386337280273, "learning_rate": 3.323870967741935e-05, "loss": 0.5621, "step": 5196 }, { "epoch": 33.500401606425704, "grad_norm": 20.285083770751953, "learning_rate": 3.3235483870967744e-05, "loss": 0.4285, "step": 5197 }, { "epoch": 33.50682730923695, "grad_norm": 15.624967575073242, "learning_rate": 3.323225806451613e-05, "loss": 0.422, "step": 5198 }, { "epoch": 33.513253012048196, "grad_norm": 20.77851676940918, "learning_rate": 3.3229032258064516e-05, "loss": 0.5638, "step": 5199 }, { "epoch": 33.51967871485944, "grad_norm": 5.494460105895996, "learning_rate": 3.322580645161291e-05, "loss": 0.3042, "step": 5200 }, { "epoch": 33.52610441767068, "grad_norm": 53.90639114379883, "learning_rate": 3.3222580645161294e-05, "loss": 0.3124, "step": 5201 }, { "epoch": 33.53253012048193, "grad_norm": 7.549752235412598, "learning_rate": 3.321935483870968e-05, "loss": 0.3373, "step": 5202 }, { "epoch": 33.53895582329317, "grad_norm": 24.990907669067383, "learning_rate": 3.3216129032258065e-05, "loss": 0.5457, "step": 5203 }, { "epoch": 33.54538152610442, "grad_norm": 3.716010570526123, "learning_rate": 3.321290322580645e-05, "loss": 0.298, "step": 5204 }, { "epoch": 33.55180722891566, "grad_norm": 27.4989070892334, "learning_rate": 3.320967741935484e-05, "loss": 0.3456, "step": 5205 }, { "epoch": 33.558232931726906, "grad_norm": 3.2892377376556396, "learning_rate": 3.320645161290323e-05, "loss": 0.2475, "step": 5206 }, { "epoch": 33.564658634538155, "grad_norm": 10.099933624267578, "learning_rate": 3.3203225806451615e-05, "loss": 0.3696, "step": 5207 }, { "epoch": 33.5710843373494, "grad_norm": 8.404860496520996, "learning_rate": 3.32e-05, "loss": 0.3815, "step": 5208 }, { "epoch": 33.57751004016064, "grad_norm": 13.524598121643066, "learning_rate": 3.319677419354839e-05, "loss": 0.4389, "step": 5209 }, { "epoch": 33.58393574297189, "grad_norm": 14.091160774230957, "learning_rate": 3.319354838709677e-05, "loss": 0.6585, "step": 5210 }, { "epoch": 33.59036144578313, "grad_norm": 16.263614654541016, "learning_rate": 3.3190322580645165e-05, "loss": 0.7596, "step": 5211 }, { "epoch": 33.59678714859438, "grad_norm": 16.71834945678711, "learning_rate": 3.318709677419355e-05, "loss": 0.6972, "step": 5212 }, { "epoch": 33.60321285140562, "grad_norm": 12.467730522155762, "learning_rate": 3.3183870967741936e-05, "loss": 0.2327, "step": 5213 }, { "epoch": 33.609638554216865, "grad_norm": 16.96485137939453, "learning_rate": 3.318064516129032e-05, "loss": 0.619, "step": 5214 }, { "epoch": 33.616064257028114, "grad_norm": 16.805627822875977, "learning_rate": 3.3177419354838714e-05, "loss": 0.3093, "step": 5215 }, { "epoch": 33.622489959839356, "grad_norm": 4.137359619140625, "learning_rate": 3.31741935483871e-05, "loss": 0.3833, "step": 5216 }, { "epoch": 33.628915662650606, "grad_norm": 22.23040199279785, "learning_rate": 3.3170967741935486e-05, "loss": 0.5984, "step": 5217 }, { "epoch": 33.63534136546185, "grad_norm": 4.48109769821167, "learning_rate": 3.316774193548387e-05, "loss": 0.1391, "step": 5218 }, { "epoch": 33.64176706827309, "grad_norm": 7.737450122833252, "learning_rate": 3.316451612903226e-05, "loss": 0.1837, "step": 5219 }, { "epoch": 33.64819277108434, "grad_norm": 4.544374465942383, "learning_rate": 3.316129032258064e-05, "loss": 0.2914, "step": 5220 }, { "epoch": 33.65461847389558, "grad_norm": 7.9277167320251465, "learning_rate": 3.3158064516129035e-05, "loss": 0.5955, "step": 5221 }, { "epoch": 33.661044176706824, "grad_norm": 37.01979446411133, "learning_rate": 3.315483870967742e-05, "loss": 0.7617, "step": 5222 }, { "epoch": 33.66746987951807, "grad_norm": 14.198821067810059, "learning_rate": 3.3151612903225807e-05, "loss": 0.4218, "step": 5223 }, { "epoch": 33.673895582329315, "grad_norm": 34.50285720825195, "learning_rate": 3.31483870967742e-05, "loss": 0.5673, "step": 5224 }, { "epoch": 33.680321285140565, "grad_norm": 8.058566093444824, "learning_rate": 3.3145161290322585e-05, "loss": 0.1504, "step": 5225 }, { "epoch": 33.68674698795181, "grad_norm": 26.646255493164062, "learning_rate": 3.314193548387097e-05, "loss": 0.4948, "step": 5226 }, { "epoch": 33.69317269076305, "grad_norm": 6.725916862487793, "learning_rate": 3.3138709677419356e-05, "loss": 0.4927, "step": 5227 }, { "epoch": 33.6995983935743, "grad_norm": 7.340423583984375, "learning_rate": 3.313548387096774e-05, "loss": 0.4574, "step": 5228 }, { "epoch": 33.70602409638554, "grad_norm": 6.43344259262085, "learning_rate": 3.313225806451613e-05, "loss": 0.2549, "step": 5229 }, { "epoch": 33.71244979919679, "grad_norm": 15.994149208068848, "learning_rate": 3.312903225806452e-05, "loss": 0.6433, "step": 5230 }, { "epoch": 33.71887550200803, "grad_norm": 19.309846878051758, "learning_rate": 3.3125806451612906e-05, "loss": 0.4421, "step": 5231 }, { "epoch": 33.725301204819274, "grad_norm": 14.497772216796875, "learning_rate": 3.312258064516129e-05, "loss": 0.2408, "step": 5232 }, { "epoch": 33.731726907630524, "grad_norm": 29.02393341064453, "learning_rate": 3.311935483870968e-05, "loss": 0.3641, "step": 5233 }, { "epoch": 33.738152610441766, "grad_norm": 20.079002380371094, "learning_rate": 3.311612903225806e-05, "loss": 0.3479, "step": 5234 }, { "epoch": 33.744578313253015, "grad_norm": 14.671927452087402, "learning_rate": 3.3112903225806455e-05, "loss": 0.2019, "step": 5235 }, { "epoch": 33.75100401606426, "grad_norm": 1.5020099878311157, "learning_rate": 3.310967741935484e-05, "loss": 0.6781, "step": 5236 }, { "epoch": 33.7574297188755, "grad_norm": 2.9166970252990723, "learning_rate": 3.310645161290323e-05, "loss": 0.343, "step": 5237 }, { "epoch": 33.76385542168675, "grad_norm": 45.5682258605957, "learning_rate": 3.310322580645161e-05, "loss": 0.4925, "step": 5238 }, { "epoch": 33.77028112449799, "grad_norm": 6.877400875091553, "learning_rate": 3.3100000000000005e-05, "loss": 0.5557, "step": 5239 }, { "epoch": 33.77670682730923, "grad_norm": 47.9395866394043, "learning_rate": 3.309677419354839e-05, "loss": 0.3847, "step": 5240 }, { "epoch": 33.78313253012048, "grad_norm": 2.7443602085113525, "learning_rate": 3.3093548387096776e-05, "loss": 0.181, "step": 5241 }, { "epoch": 33.789558232931725, "grad_norm": 23.059467315673828, "learning_rate": 3.309032258064516e-05, "loss": 0.6055, "step": 5242 }, { "epoch": 33.795983935742974, "grad_norm": 13.589239120483398, "learning_rate": 3.308709677419355e-05, "loss": 0.5884, "step": 5243 }, { "epoch": 33.80240963855422, "grad_norm": 7.641796112060547, "learning_rate": 3.3083870967741933e-05, "loss": 0.5091, "step": 5244 }, { "epoch": 33.80883534136546, "grad_norm": 10.45595932006836, "learning_rate": 3.3080645161290326e-05, "loss": 0.4616, "step": 5245 }, { "epoch": 33.81526104417671, "grad_norm": 20.89236068725586, "learning_rate": 3.307741935483871e-05, "loss": 0.5292, "step": 5246 }, { "epoch": 33.82168674698795, "grad_norm": 15.288496971130371, "learning_rate": 3.30741935483871e-05, "loss": 0.4796, "step": 5247 }, { "epoch": 33.8281124497992, "grad_norm": 8.652997016906738, "learning_rate": 3.307096774193549e-05, "loss": 0.2672, "step": 5248 }, { "epoch": 33.83453815261044, "grad_norm": 3.0159125328063965, "learning_rate": 3.306774193548387e-05, "loss": 0.2477, "step": 5249 }, { "epoch": 33.840963855421684, "grad_norm": 10.525106430053711, "learning_rate": 3.306451612903226e-05, "loss": 0.6596, "step": 5250 }, { "epoch": 33.84738955823293, "grad_norm": 7.179155349731445, "learning_rate": 3.306129032258065e-05, "loss": 0.4415, "step": 5251 }, { "epoch": 33.853815261044176, "grad_norm": 10.363061904907227, "learning_rate": 3.305806451612903e-05, "loss": 0.6188, "step": 5252 }, { "epoch": 33.860240963855425, "grad_norm": 5.430462837219238, "learning_rate": 3.305483870967742e-05, "loss": 0.1861, "step": 5253 }, { "epoch": 33.86666666666667, "grad_norm": 5.214340686798096, "learning_rate": 3.305161290322581e-05, "loss": 0.442, "step": 5254 }, { "epoch": 33.87309236947791, "grad_norm": 3.662494421005249, "learning_rate": 3.3048387096774197e-05, "loss": 0.3213, "step": 5255 }, { "epoch": 33.87951807228916, "grad_norm": 44.9022331237793, "learning_rate": 3.304516129032258e-05, "loss": 0.4602, "step": 5256 }, { "epoch": 33.8859437751004, "grad_norm": 14.559405326843262, "learning_rate": 3.304193548387097e-05, "loss": 0.2812, "step": 5257 }, { "epoch": 33.89236947791164, "grad_norm": 14.691364288330078, "learning_rate": 3.3038709677419354e-05, "loss": 0.4814, "step": 5258 }, { "epoch": 33.89879518072289, "grad_norm": 28.037694931030273, "learning_rate": 3.303548387096774e-05, "loss": 0.5307, "step": 5259 }, { "epoch": 33.905220883534135, "grad_norm": 29.852867126464844, "learning_rate": 3.303225806451613e-05, "loss": 0.4871, "step": 5260 }, { "epoch": 33.911646586345384, "grad_norm": 15.837162971496582, "learning_rate": 3.302903225806452e-05, "loss": 0.54, "step": 5261 }, { "epoch": 33.918072289156626, "grad_norm": 10.953420639038086, "learning_rate": 3.30258064516129e-05, "loss": 0.2409, "step": 5262 }, { "epoch": 33.92449799196787, "grad_norm": 7.706067085266113, "learning_rate": 3.3022580645161296e-05, "loss": 0.2522, "step": 5263 }, { "epoch": 33.93092369477912, "grad_norm": 21.0522403717041, "learning_rate": 3.3019354838709675e-05, "loss": 0.5268, "step": 5264 }, { "epoch": 33.93734939759036, "grad_norm": 10.163681983947754, "learning_rate": 3.301612903225807e-05, "loss": 0.2386, "step": 5265 }, { "epoch": 33.94377510040161, "grad_norm": 35.623146057128906, "learning_rate": 3.301290322580645e-05, "loss": 0.3778, "step": 5266 }, { "epoch": 33.95020080321285, "grad_norm": 9.207220077514648, "learning_rate": 3.300967741935484e-05, "loss": 0.6483, "step": 5267 }, { "epoch": 33.956626506024094, "grad_norm": 20.63681983947754, "learning_rate": 3.3006451612903224e-05, "loss": 0.3543, "step": 5268 }, { "epoch": 33.96305220883534, "grad_norm": 7.005767822265625, "learning_rate": 3.300322580645162e-05, "loss": 0.2062, "step": 5269 }, { "epoch": 33.969477911646585, "grad_norm": 6.338621139526367, "learning_rate": 3.3e-05, "loss": 0.2689, "step": 5270 }, { "epoch": 33.975903614457835, "grad_norm": 5.056144714355469, "learning_rate": 3.299677419354839e-05, "loss": 0.2638, "step": 5271 }, { "epoch": 33.98232931726908, "grad_norm": 3.4547719955444336, "learning_rate": 3.2993548387096774e-05, "loss": 0.3547, "step": 5272 }, { "epoch": 33.98875502008032, "grad_norm": 18.125883102416992, "learning_rate": 3.299032258064516e-05, "loss": 0.298, "step": 5273 }, { "epoch": 33.99518072289157, "grad_norm": 4.493209362030029, "learning_rate": 3.298709677419355e-05, "loss": 0.3632, "step": 5274 }, { "epoch": 34.00481927710843, "grad_norm": 11.311495780944824, "learning_rate": 3.298387096774194e-05, "loss": 0.5212, "step": 5275 }, { "epoch": 34.01124497991968, "grad_norm": 9.279659271240234, "learning_rate": 3.2980645161290324e-05, "loss": 0.4468, "step": 5276 }, { "epoch": 34.01767068273092, "grad_norm": 15.719539642333984, "learning_rate": 3.297741935483871e-05, "loss": 0.4909, "step": 5277 }, { "epoch": 34.024096385542165, "grad_norm": 64.32344055175781, "learning_rate": 3.29741935483871e-05, "loss": 0.5329, "step": 5278 }, { "epoch": 34.030522088353415, "grad_norm": 10.253863334655762, "learning_rate": 3.297096774193549e-05, "loss": 0.4787, "step": 5279 }, { "epoch": 34.03694779116466, "grad_norm": 56.48024368286133, "learning_rate": 3.296774193548387e-05, "loss": 0.4743, "step": 5280 }, { "epoch": 34.043373493975906, "grad_norm": 4.040931224822998, "learning_rate": 3.296451612903226e-05, "loss": 0.2935, "step": 5281 }, { "epoch": 34.04979919678715, "grad_norm": 8.746171951293945, "learning_rate": 3.2961290322580645e-05, "loss": 0.314, "step": 5282 }, { "epoch": 34.05622489959839, "grad_norm": 1.8355048894882202, "learning_rate": 3.295806451612903e-05, "loss": 0.244, "step": 5283 }, { "epoch": 34.06265060240964, "grad_norm": 15.8955078125, "learning_rate": 3.295483870967742e-05, "loss": 0.3434, "step": 5284 }, { "epoch": 34.06907630522088, "grad_norm": 20.625028610229492, "learning_rate": 3.295161290322581e-05, "loss": 0.3112, "step": 5285 }, { "epoch": 34.07550200803213, "grad_norm": 51.78456115722656, "learning_rate": 3.2948387096774194e-05, "loss": 0.3524, "step": 5286 }, { "epoch": 34.081927710843374, "grad_norm": 7.04843807220459, "learning_rate": 3.294516129032259e-05, "loss": 0.2888, "step": 5287 }, { "epoch": 34.088353413654616, "grad_norm": 14.448735237121582, "learning_rate": 3.2941935483870966e-05, "loss": 0.2909, "step": 5288 }, { "epoch": 34.094779116465865, "grad_norm": 12.346388816833496, "learning_rate": 3.293870967741936e-05, "loss": 0.2337, "step": 5289 }, { "epoch": 34.10120481927711, "grad_norm": 9.603854179382324, "learning_rate": 3.2935483870967744e-05, "loss": 0.1376, "step": 5290 }, { "epoch": 34.10763052208836, "grad_norm": 5.429609298706055, "learning_rate": 3.293225806451613e-05, "loss": 0.3868, "step": 5291 }, { "epoch": 34.1140562248996, "grad_norm": 12.963313102722168, "learning_rate": 3.2929032258064515e-05, "loss": 0.3662, "step": 5292 }, { "epoch": 34.12048192771084, "grad_norm": 12.592516899108887, "learning_rate": 3.292580645161291e-05, "loss": 0.3984, "step": 5293 }, { "epoch": 34.12690763052209, "grad_norm": 15.48464584350586, "learning_rate": 3.292258064516129e-05, "loss": 0.341, "step": 5294 }, { "epoch": 34.13333333333333, "grad_norm": 19.304603576660156, "learning_rate": 3.291935483870968e-05, "loss": 0.2691, "step": 5295 }, { "epoch": 34.139759036144575, "grad_norm": 25.74827766418457, "learning_rate": 3.2916129032258065e-05, "loss": 0.3494, "step": 5296 }, { "epoch": 34.146184738955824, "grad_norm": 5.286754608154297, "learning_rate": 3.291290322580645e-05, "loss": 0.1687, "step": 5297 }, { "epoch": 34.15261044176707, "grad_norm": 44.7780876159668, "learning_rate": 3.290967741935484e-05, "loss": 0.4872, "step": 5298 }, { "epoch": 34.159036144578316, "grad_norm": 12.101757049560547, "learning_rate": 3.290645161290323e-05, "loss": 0.3208, "step": 5299 }, { "epoch": 34.16546184738956, "grad_norm": 4.410048484802246, "learning_rate": 3.2903225806451614e-05, "loss": 0.1188, "step": 5300 }, { "epoch": 34.1718875502008, "grad_norm": 5.569417476654053, "learning_rate": 3.29e-05, "loss": 0.2541, "step": 5301 }, { "epoch": 34.17831325301205, "grad_norm": 15.643996238708496, "learning_rate": 3.289677419354839e-05, "loss": 0.3789, "step": 5302 }, { "epoch": 34.18473895582329, "grad_norm": 23.44184112548828, "learning_rate": 3.289354838709677e-05, "loss": 0.6205, "step": 5303 }, { "epoch": 34.19116465863454, "grad_norm": 5.782539367675781, "learning_rate": 3.2890322580645164e-05, "loss": 0.1906, "step": 5304 }, { "epoch": 34.19759036144578, "grad_norm": 8.144661903381348, "learning_rate": 3.288709677419355e-05, "loss": 0.4467, "step": 5305 }, { "epoch": 34.204016064257026, "grad_norm": 2.089284896850586, "learning_rate": 3.2883870967741935e-05, "loss": 0.1749, "step": 5306 }, { "epoch": 34.210441767068275, "grad_norm": 8.761879920959473, "learning_rate": 3.288064516129032e-05, "loss": 0.2152, "step": 5307 }, { "epoch": 34.21686746987952, "grad_norm": 27.037567138671875, "learning_rate": 3.2877419354838714e-05, "loss": 0.3284, "step": 5308 }, { "epoch": 34.22329317269077, "grad_norm": 13.301836967468262, "learning_rate": 3.28741935483871e-05, "loss": 0.3962, "step": 5309 }, { "epoch": 34.22971887550201, "grad_norm": 12.505327224731445, "learning_rate": 3.2870967741935485e-05, "loss": 0.4134, "step": 5310 }, { "epoch": 34.23614457831325, "grad_norm": 6.182512283325195, "learning_rate": 3.286774193548387e-05, "loss": 0.4469, "step": 5311 }, { "epoch": 34.2425702811245, "grad_norm": 6.626654148101807, "learning_rate": 3.2864516129032256e-05, "loss": 0.2522, "step": 5312 }, { "epoch": 34.24899598393574, "grad_norm": 11.183098793029785, "learning_rate": 3.286129032258065e-05, "loss": 0.2171, "step": 5313 }, { "epoch": 34.255421686746985, "grad_norm": 17.548267364501953, "learning_rate": 3.2858064516129035e-05, "loss": 0.2681, "step": 5314 }, { "epoch": 34.261847389558234, "grad_norm": 18.453786849975586, "learning_rate": 3.285483870967742e-05, "loss": 0.6791, "step": 5315 }, { "epoch": 34.268273092369476, "grad_norm": 8.084096908569336, "learning_rate": 3.2851612903225806e-05, "loss": 0.3145, "step": 5316 }, { "epoch": 34.274698795180726, "grad_norm": 7.7443342208862305, "learning_rate": 3.28483870967742e-05, "loss": 0.3661, "step": 5317 }, { "epoch": 34.28112449799197, "grad_norm": 13.685210227966309, "learning_rate": 3.2845161290322584e-05, "loss": 0.4631, "step": 5318 }, { "epoch": 34.28755020080321, "grad_norm": 15.570971488952637, "learning_rate": 3.284193548387097e-05, "loss": 0.5296, "step": 5319 }, { "epoch": 34.29397590361446, "grad_norm": 4.9748382568359375, "learning_rate": 3.2838709677419356e-05, "loss": 0.4872, "step": 5320 }, { "epoch": 34.3004016064257, "grad_norm": 6.558889865875244, "learning_rate": 3.283548387096774e-05, "loss": 0.1708, "step": 5321 }, { "epoch": 34.30682730923695, "grad_norm": 13.436909675598145, "learning_rate": 3.283225806451613e-05, "loss": 0.3547, "step": 5322 }, { "epoch": 34.31325301204819, "grad_norm": 8.820411682128906, "learning_rate": 3.282903225806452e-05, "loss": 0.1888, "step": 5323 }, { "epoch": 34.319678714859435, "grad_norm": 36.967041015625, "learning_rate": 3.2825806451612905e-05, "loss": 0.5259, "step": 5324 }, { "epoch": 34.326104417670685, "grad_norm": 8.038119316101074, "learning_rate": 3.282258064516129e-05, "loss": 0.3043, "step": 5325 }, { "epoch": 34.33253012048193, "grad_norm": 9.889986038208008, "learning_rate": 3.2819354838709683e-05, "loss": 0.3569, "step": 5326 }, { "epoch": 34.338955823293176, "grad_norm": 18.9923152923584, "learning_rate": 3.281612903225806e-05, "loss": 0.41, "step": 5327 }, { "epoch": 34.34538152610442, "grad_norm": 18.741920471191406, "learning_rate": 3.2812903225806455e-05, "loss": 0.448, "step": 5328 }, { "epoch": 34.35180722891566, "grad_norm": 16.23822784423828, "learning_rate": 3.280967741935484e-05, "loss": 0.2923, "step": 5329 }, { "epoch": 34.35823293172691, "grad_norm": 12.194328308105469, "learning_rate": 3.2806451612903226e-05, "loss": 0.214, "step": 5330 }, { "epoch": 34.36465863453815, "grad_norm": 14.581013679504395, "learning_rate": 3.280322580645161e-05, "loss": 0.6128, "step": 5331 }, { "epoch": 34.371084337349394, "grad_norm": 8.605831146240234, "learning_rate": 3.2800000000000004e-05, "loss": 0.5151, "step": 5332 }, { "epoch": 34.377510040160644, "grad_norm": 18.684127807617188, "learning_rate": 3.279677419354839e-05, "loss": 0.5015, "step": 5333 }, { "epoch": 34.383935742971886, "grad_norm": 37.84653854370117, "learning_rate": 3.2793548387096776e-05, "loss": 0.5257, "step": 5334 }, { "epoch": 34.390361445783135, "grad_norm": 23.066274642944336, "learning_rate": 3.279032258064516e-05, "loss": 0.4104, "step": 5335 }, { "epoch": 34.39678714859438, "grad_norm": 14.862425804138184, "learning_rate": 3.278709677419355e-05, "loss": 0.4095, "step": 5336 }, { "epoch": 34.40321285140562, "grad_norm": 11.27126693725586, "learning_rate": 3.278387096774194e-05, "loss": 0.3923, "step": 5337 }, { "epoch": 34.40963855421687, "grad_norm": 11.053305625915527, "learning_rate": 3.2780645161290325e-05, "loss": 0.4686, "step": 5338 }, { "epoch": 34.41606425702811, "grad_norm": 13.504538536071777, "learning_rate": 3.277741935483871e-05, "loss": 0.4933, "step": 5339 }, { "epoch": 34.42248995983936, "grad_norm": 4.427469730377197, "learning_rate": 3.27741935483871e-05, "loss": 0.6187, "step": 5340 }, { "epoch": 34.4289156626506, "grad_norm": 7.790173053741455, "learning_rate": 3.277096774193549e-05, "loss": 0.3792, "step": 5341 }, { "epoch": 34.435341365461845, "grad_norm": 22.81084632873535, "learning_rate": 3.276774193548387e-05, "loss": 0.4017, "step": 5342 }, { "epoch": 34.441767068273094, "grad_norm": 9.75485610961914, "learning_rate": 3.276451612903226e-05, "loss": 0.421, "step": 5343 }, { "epoch": 34.44819277108434, "grad_norm": 26.460710525512695, "learning_rate": 3.2761290322580646e-05, "loss": 0.3409, "step": 5344 }, { "epoch": 34.454618473895586, "grad_norm": 13.541646957397461, "learning_rate": 3.275806451612903e-05, "loss": 0.4671, "step": 5345 }, { "epoch": 34.46104417670683, "grad_norm": 17.221267700195312, "learning_rate": 3.275483870967742e-05, "loss": 0.3354, "step": 5346 }, { "epoch": 34.46746987951807, "grad_norm": 7.91865873336792, "learning_rate": 3.275161290322581e-05, "loss": 0.4245, "step": 5347 }, { "epoch": 34.47389558232932, "grad_norm": 22.164472579956055, "learning_rate": 3.2748387096774196e-05, "loss": 0.1472, "step": 5348 }, { "epoch": 34.48032128514056, "grad_norm": 4.371875286102295, "learning_rate": 3.274516129032258e-05, "loss": 0.2362, "step": 5349 }, { "epoch": 34.486746987951804, "grad_norm": 4.290115833282471, "learning_rate": 3.274193548387097e-05, "loss": 0.216, "step": 5350 }, { "epoch": 34.49317269076305, "grad_norm": 3.468684434890747, "learning_rate": 3.273870967741935e-05, "loss": 0.2003, "step": 5351 }, { "epoch": 34.499598393574296, "grad_norm": 15.373802185058594, "learning_rate": 3.2735483870967746e-05, "loss": 0.3756, "step": 5352 }, { "epoch": 34.506024096385545, "grad_norm": 63.4660530090332, "learning_rate": 3.273225806451613e-05, "loss": 0.621, "step": 5353 }, { "epoch": 34.51244979919679, "grad_norm": 8.791980743408203, "learning_rate": 3.272903225806452e-05, "loss": 0.239, "step": 5354 }, { "epoch": 34.51887550200803, "grad_norm": 12.719888687133789, "learning_rate": 3.27258064516129e-05, "loss": 0.2473, "step": 5355 }, { "epoch": 34.52530120481928, "grad_norm": 15.415653228759766, "learning_rate": 3.2722580645161295e-05, "loss": 0.1493, "step": 5356 }, { "epoch": 34.53172690763052, "grad_norm": 10.61021900177002, "learning_rate": 3.271935483870968e-05, "loss": 0.2941, "step": 5357 }, { "epoch": 34.53815261044177, "grad_norm": 13.241339683532715, "learning_rate": 3.271612903225807e-05, "loss": 0.3874, "step": 5358 }, { "epoch": 34.54457831325301, "grad_norm": 18.68042755126953, "learning_rate": 3.271290322580645e-05, "loss": 0.3877, "step": 5359 }, { "epoch": 34.551004016064255, "grad_norm": 16.452863693237305, "learning_rate": 3.270967741935484e-05, "loss": 0.3755, "step": 5360 }, { "epoch": 34.557429718875504, "grad_norm": 6.052667140960693, "learning_rate": 3.270645161290323e-05, "loss": 0.2664, "step": 5361 }, { "epoch": 34.563855421686746, "grad_norm": 4.287425994873047, "learning_rate": 3.2703225806451616e-05, "loss": 0.391, "step": 5362 }, { "epoch": 34.570281124497996, "grad_norm": 10.026632308959961, "learning_rate": 3.27e-05, "loss": 0.1897, "step": 5363 }, { "epoch": 34.57670682730924, "grad_norm": 10.006071090698242, "learning_rate": 3.269677419354839e-05, "loss": 0.5647, "step": 5364 }, { "epoch": 34.58313253012048, "grad_norm": 16.052688598632812, "learning_rate": 3.269354838709678e-05, "loss": 0.3837, "step": 5365 }, { "epoch": 34.58955823293173, "grad_norm": 16.487384796142578, "learning_rate": 3.269032258064516e-05, "loss": 0.3149, "step": 5366 }, { "epoch": 34.59598393574297, "grad_norm": 11.875029563903809, "learning_rate": 3.268709677419355e-05, "loss": 0.4691, "step": 5367 }, { "epoch": 34.602409638554214, "grad_norm": 6.996983528137207, "learning_rate": 3.268387096774194e-05, "loss": 0.3647, "step": 5368 }, { "epoch": 34.60883534136546, "grad_norm": 2.5628910064697266, "learning_rate": 3.268064516129032e-05, "loss": 0.1411, "step": 5369 }, { "epoch": 34.615261044176705, "grad_norm": 4.959655284881592, "learning_rate": 3.267741935483871e-05, "loss": 0.3205, "step": 5370 }, { "epoch": 34.621686746987955, "grad_norm": 5.449699878692627, "learning_rate": 3.26741935483871e-05, "loss": 0.109, "step": 5371 }, { "epoch": 34.6281124497992, "grad_norm": 2.7058780193328857, "learning_rate": 3.267096774193549e-05, "loss": 0.285, "step": 5372 }, { "epoch": 34.63453815261044, "grad_norm": 41.5511474609375, "learning_rate": 3.266774193548387e-05, "loss": 0.5741, "step": 5373 }, { "epoch": 34.64096385542169, "grad_norm": 12.07894515991211, "learning_rate": 3.266451612903226e-05, "loss": 0.4396, "step": 5374 }, { "epoch": 34.64738955823293, "grad_norm": 28.204702377319336, "learning_rate": 3.2661290322580644e-05, "loss": 0.618, "step": 5375 }, { "epoch": 34.65381526104418, "grad_norm": 21.339139938354492, "learning_rate": 3.2658064516129037e-05, "loss": 0.3937, "step": 5376 }, { "epoch": 34.66024096385542, "grad_norm": 10.631587982177734, "learning_rate": 3.265483870967742e-05, "loss": 0.5313, "step": 5377 }, { "epoch": 34.666666666666664, "grad_norm": 7.94950532913208, "learning_rate": 3.265161290322581e-05, "loss": 0.2108, "step": 5378 }, { "epoch": 34.673092369477914, "grad_norm": 5.094762802124023, "learning_rate": 3.2648387096774194e-05, "loss": 0.1534, "step": 5379 }, { "epoch": 34.679518072289156, "grad_norm": 13.843807220458984, "learning_rate": 3.2645161290322586e-05, "loss": 0.3109, "step": 5380 }, { "epoch": 34.6859437751004, "grad_norm": 4.300271987915039, "learning_rate": 3.2641935483870965e-05, "loss": 0.4502, "step": 5381 }, { "epoch": 34.69236947791165, "grad_norm": 3.5963335037231445, "learning_rate": 3.263870967741936e-05, "loss": 0.1441, "step": 5382 }, { "epoch": 34.69879518072289, "grad_norm": 6.4116926193237305, "learning_rate": 3.263548387096774e-05, "loss": 0.6294, "step": 5383 }, { "epoch": 34.70522088353414, "grad_norm": 19.482908248901367, "learning_rate": 3.263225806451613e-05, "loss": 0.3548, "step": 5384 }, { "epoch": 34.71164658634538, "grad_norm": 7.126611232757568, "learning_rate": 3.2629032258064515e-05, "loss": 0.4843, "step": 5385 }, { "epoch": 34.71807228915662, "grad_norm": 6.5585618019104, "learning_rate": 3.262580645161291e-05, "loss": 0.5844, "step": 5386 }, { "epoch": 34.72449799196787, "grad_norm": 21.5140380859375, "learning_rate": 3.262258064516129e-05, "loss": 0.1966, "step": 5387 }, { "epoch": 34.730923694779115, "grad_norm": 26.69683265686035, "learning_rate": 3.261935483870968e-05, "loss": 0.3146, "step": 5388 }, { "epoch": 34.737349397590364, "grad_norm": 30.40497398376465, "learning_rate": 3.2616129032258064e-05, "loss": 0.3234, "step": 5389 }, { "epoch": 34.743775100401606, "grad_norm": 19.65346336364746, "learning_rate": 3.261290322580645e-05, "loss": 0.6119, "step": 5390 }, { "epoch": 34.75020080321285, "grad_norm": 16.724151611328125, "learning_rate": 3.260967741935484e-05, "loss": 0.4038, "step": 5391 }, { "epoch": 34.7566265060241, "grad_norm": 12.496109008789062, "learning_rate": 3.260645161290323e-05, "loss": 0.1617, "step": 5392 }, { "epoch": 34.76305220883534, "grad_norm": 3.243312120437622, "learning_rate": 3.2603225806451614e-05, "loss": 0.4557, "step": 5393 }, { "epoch": 34.76947791164659, "grad_norm": 11.934861183166504, "learning_rate": 3.26e-05, "loss": 0.2605, "step": 5394 }, { "epoch": 34.77590361445783, "grad_norm": 16.367359161376953, "learning_rate": 3.259677419354839e-05, "loss": 0.7706, "step": 5395 }, { "epoch": 34.782329317269074, "grad_norm": 34.021339416503906, "learning_rate": 3.259354838709678e-05, "loss": 0.4513, "step": 5396 }, { "epoch": 34.78875502008032, "grad_norm": 6.6607666015625, "learning_rate": 3.2590322580645163e-05, "loss": 0.3412, "step": 5397 }, { "epoch": 34.795180722891565, "grad_norm": 5.475970268249512, "learning_rate": 3.258709677419355e-05, "loss": 0.2487, "step": 5398 }, { "epoch": 34.801606425702815, "grad_norm": 5.121541976928711, "learning_rate": 3.2583870967741935e-05, "loss": 0.3236, "step": 5399 }, { "epoch": 34.80803212851406, "grad_norm": 6.651093482971191, "learning_rate": 3.258064516129033e-05, "loss": 0.3199, "step": 5400 }, { "epoch": 34.8144578313253, "grad_norm": 30.93170738220215, "learning_rate": 3.257741935483871e-05, "loss": 0.3337, "step": 5401 }, { "epoch": 34.82088353413655, "grad_norm": 9.607050895690918, "learning_rate": 3.25741935483871e-05, "loss": 0.4099, "step": 5402 }, { "epoch": 34.82730923694779, "grad_norm": 6.646697521209717, "learning_rate": 3.2570967741935484e-05, "loss": 0.3336, "step": 5403 }, { "epoch": 34.83373493975903, "grad_norm": 3.1575300693511963, "learning_rate": 3.256774193548387e-05, "loss": 0.2615, "step": 5404 }, { "epoch": 34.84016064257028, "grad_norm": 10.019610404968262, "learning_rate": 3.2564516129032256e-05, "loss": 0.3026, "step": 5405 }, { "epoch": 34.846586345381525, "grad_norm": 6.552170753479004, "learning_rate": 3.256129032258065e-05, "loss": 0.3263, "step": 5406 }, { "epoch": 34.853012048192774, "grad_norm": 14.541943550109863, "learning_rate": 3.2558064516129034e-05, "loss": 0.4095, "step": 5407 }, { "epoch": 34.859437751004016, "grad_norm": 17.510112762451172, "learning_rate": 3.255483870967742e-05, "loss": 0.5686, "step": 5408 }, { "epoch": 34.86586345381526, "grad_norm": 6.658080101013184, "learning_rate": 3.2551612903225806e-05, "loss": 0.2104, "step": 5409 }, { "epoch": 34.87228915662651, "grad_norm": 5.936549186706543, "learning_rate": 3.25483870967742e-05, "loss": 0.0825, "step": 5410 }, { "epoch": 34.87871485943775, "grad_norm": 11.453723907470703, "learning_rate": 3.2545161290322584e-05, "loss": 0.2076, "step": 5411 }, { "epoch": 34.885140562249, "grad_norm": 4.345839977264404, "learning_rate": 3.254193548387097e-05, "loss": 0.4516, "step": 5412 }, { "epoch": 34.89156626506024, "grad_norm": 10.817652702331543, "learning_rate": 3.2538709677419355e-05, "loss": 0.3639, "step": 5413 }, { "epoch": 34.897991967871484, "grad_norm": 8.624922752380371, "learning_rate": 3.253548387096774e-05, "loss": 0.4372, "step": 5414 }, { "epoch": 34.90441767068273, "grad_norm": 14.36325454711914, "learning_rate": 3.253225806451613e-05, "loss": 0.5458, "step": 5415 }, { "epoch": 34.910843373493975, "grad_norm": 14.913798332214355, "learning_rate": 3.252903225806452e-05, "loss": 0.3204, "step": 5416 }, { "epoch": 34.91726907630522, "grad_norm": 6.740121841430664, "learning_rate": 3.2525806451612905e-05, "loss": 0.2218, "step": 5417 }, { "epoch": 34.92369477911647, "grad_norm": 3.721067428588867, "learning_rate": 3.252258064516129e-05, "loss": 0.3943, "step": 5418 }, { "epoch": 34.93012048192771, "grad_norm": 12.574819564819336, "learning_rate": 3.251935483870968e-05, "loss": 0.5995, "step": 5419 }, { "epoch": 34.93654618473896, "grad_norm": 9.942288398742676, "learning_rate": 3.251612903225806e-05, "loss": 0.5131, "step": 5420 }, { "epoch": 34.9429718875502, "grad_norm": 37.067649841308594, "learning_rate": 3.2512903225806454e-05, "loss": 0.5901, "step": 5421 }, { "epoch": 34.94939759036144, "grad_norm": 4.716465473175049, "learning_rate": 3.250967741935484e-05, "loss": 0.2483, "step": 5422 }, { "epoch": 34.95582329317269, "grad_norm": 27.83123207092285, "learning_rate": 3.2506451612903226e-05, "loss": 0.4237, "step": 5423 }, { "epoch": 34.962248995983934, "grad_norm": 15.447527885437012, "learning_rate": 3.250322580645162e-05, "loss": 0.3924, "step": 5424 }, { "epoch": 34.96867469879518, "grad_norm": 8.977574348449707, "learning_rate": 3.2500000000000004e-05, "loss": 0.5182, "step": 5425 }, { "epoch": 34.975100401606426, "grad_norm": 11.05328369140625, "learning_rate": 3.249677419354839e-05, "loss": 0.3793, "step": 5426 }, { "epoch": 34.98152610441767, "grad_norm": 9.055485725402832, "learning_rate": 3.2493548387096775e-05, "loss": 0.2006, "step": 5427 }, { "epoch": 34.98795180722892, "grad_norm": 25.973255157470703, "learning_rate": 3.249032258064516e-05, "loss": 0.4973, "step": 5428 }, { "epoch": 34.99437751004016, "grad_norm": 25.465959548950195, "learning_rate": 3.248709677419355e-05, "loss": 0.3757, "step": 5429 }, { "epoch": 35.00401606425703, "grad_norm": 8.004831314086914, "learning_rate": 3.248387096774194e-05, "loss": 0.3448, "step": 5430 }, { "epoch": 35.01044176706827, "grad_norm": 17.825056076049805, "learning_rate": 3.2480645161290325e-05, "loss": 0.4122, "step": 5431 }, { "epoch": 35.01686746987952, "grad_norm": 9.172942161560059, "learning_rate": 3.247741935483871e-05, "loss": 0.2281, "step": 5432 }, { "epoch": 35.023293172690764, "grad_norm": 12.538999557495117, "learning_rate": 3.2474193548387096e-05, "loss": 0.3606, "step": 5433 }, { "epoch": 35.029718875502006, "grad_norm": 17.627153396606445, "learning_rate": 3.247096774193549e-05, "loss": 0.3577, "step": 5434 }, { "epoch": 35.036144578313255, "grad_norm": 18.345481872558594, "learning_rate": 3.2467741935483875e-05, "loss": 0.4142, "step": 5435 }, { "epoch": 35.0425702811245, "grad_norm": 3.9042086601257324, "learning_rate": 3.246451612903226e-05, "loss": 0.4187, "step": 5436 }, { "epoch": 35.04899598393574, "grad_norm": 23.733837127685547, "learning_rate": 3.2461290322580646e-05, "loss": 0.1685, "step": 5437 }, { "epoch": 35.05542168674699, "grad_norm": 8.3946533203125, "learning_rate": 3.245806451612903e-05, "loss": 0.3532, "step": 5438 }, { "epoch": 35.06184738955823, "grad_norm": 27.50323486328125, "learning_rate": 3.2454838709677424e-05, "loss": 0.3644, "step": 5439 }, { "epoch": 35.06827309236948, "grad_norm": 13.26169490814209, "learning_rate": 3.245161290322581e-05, "loss": 0.545, "step": 5440 }, { "epoch": 35.07469879518072, "grad_norm": 8.135394096374512, "learning_rate": 3.2448387096774196e-05, "loss": 0.2788, "step": 5441 }, { "epoch": 35.081124497991965, "grad_norm": 2.763094425201416, "learning_rate": 3.244516129032258e-05, "loss": 0.2671, "step": 5442 }, { "epoch": 35.087550200803214, "grad_norm": 5.547766208648682, "learning_rate": 3.244193548387097e-05, "loss": 0.5142, "step": 5443 }, { "epoch": 35.09397590361446, "grad_norm": 15.316363334655762, "learning_rate": 3.243870967741935e-05, "loss": 0.2721, "step": 5444 }, { "epoch": 35.100401606425706, "grad_norm": 12.457051277160645, "learning_rate": 3.2435483870967745e-05, "loss": 0.3664, "step": 5445 }, { "epoch": 35.10682730923695, "grad_norm": 11.09540843963623, "learning_rate": 3.243225806451613e-05, "loss": 0.5103, "step": 5446 }, { "epoch": 35.11325301204819, "grad_norm": 14.836630821228027, "learning_rate": 3.2429032258064517e-05, "loss": 0.2301, "step": 5447 }, { "epoch": 35.11967871485944, "grad_norm": 14.438716888427734, "learning_rate": 3.24258064516129e-05, "loss": 0.4703, "step": 5448 }, { "epoch": 35.12610441767068, "grad_norm": 9.318597793579102, "learning_rate": 3.2422580645161295e-05, "loss": 0.4446, "step": 5449 }, { "epoch": 35.13253012048193, "grad_norm": 10.43554401397705, "learning_rate": 3.241935483870968e-05, "loss": 0.2603, "step": 5450 }, { "epoch": 35.13895582329317, "grad_norm": 7.793519496917725, "learning_rate": 3.2416129032258066e-05, "loss": 0.2962, "step": 5451 }, { "epoch": 35.145381526104416, "grad_norm": 25.615049362182617, "learning_rate": 3.241290322580645e-05, "loss": 0.3341, "step": 5452 }, { "epoch": 35.151807228915665, "grad_norm": 26.797048568725586, "learning_rate": 3.240967741935484e-05, "loss": 0.4963, "step": 5453 }, { "epoch": 35.15823293172691, "grad_norm": 10.25236701965332, "learning_rate": 3.240645161290323e-05, "loss": 0.3803, "step": 5454 }, { "epoch": 35.16465863453815, "grad_norm": 28.90460777282715, "learning_rate": 3.2403225806451616e-05, "loss": 0.4453, "step": 5455 }, { "epoch": 35.1710843373494, "grad_norm": 108.27642059326172, "learning_rate": 3.24e-05, "loss": 0.3912, "step": 5456 }, { "epoch": 35.17751004016064, "grad_norm": 4.043552398681641, "learning_rate": 3.239677419354839e-05, "loss": 0.3217, "step": 5457 }, { "epoch": 35.18393574297189, "grad_norm": 7.753203392028809, "learning_rate": 3.239354838709678e-05, "loss": 0.2534, "step": 5458 }, { "epoch": 35.19036144578313, "grad_norm": 17.845378875732422, "learning_rate": 3.239032258064516e-05, "loss": 0.2594, "step": 5459 }, { "epoch": 35.196787148594375, "grad_norm": 13.382160186767578, "learning_rate": 3.238709677419355e-05, "loss": 0.5896, "step": 5460 }, { "epoch": 35.203212851405624, "grad_norm": 13.345788955688477, "learning_rate": 3.238387096774194e-05, "loss": 0.5169, "step": 5461 }, { "epoch": 35.209638554216866, "grad_norm": 10.215737342834473, "learning_rate": 3.238064516129032e-05, "loss": 0.3336, "step": 5462 }, { "epoch": 35.216064257028115, "grad_norm": 7.021327495574951, "learning_rate": 3.2377419354838715e-05, "loss": 0.1152, "step": 5463 }, { "epoch": 35.22248995983936, "grad_norm": 11.165067672729492, "learning_rate": 3.23741935483871e-05, "loss": 0.448, "step": 5464 }, { "epoch": 35.2289156626506, "grad_norm": 6.533969402313232, "learning_rate": 3.2370967741935486e-05, "loss": 0.3716, "step": 5465 }, { "epoch": 35.23534136546185, "grad_norm": 4.578881740570068, "learning_rate": 3.236774193548387e-05, "loss": 0.4237, "step": 5466 }, { "epoch": 35.24176706827309, "grad_norm": 12.964484214782715, "learning_rate": 3.236451612903226e-05, "loss": 0.3693, "step": 5467 }, { "epoch": 35.24819277108434, "grad_norm": 30.053396224975586, "learning_rate": 3.2361290322580644e-05, "loss": 0.3955, "step": 5468 }, { "epoch": 35.25461847389558, "grad_norm": 15.601627349853516, "learning_rate": 3.2358064516129036e-05, "loss": 0.2841, "step": 5469 }, { "epoch": 35.261044176706825, "grad_norm": 14.437689781188965, "learning_rate": 3.235483870967742e-05, "loss": 0.5518, "step": 5470 }, { "epoch": 35.267469879518075, "grad_norm": 10.411277770996094, "learning_rate": 3.235161290322581e-05, "loss": 0.5127, "step": 5471 }, { "epoch": 35.27389558232932, "grad_norm": 18.3750057220459, "learning_rate": 3.234838709677419e-05, "loss": 0.7928, "step": 5472 }, { "epoch": 35.28032128514056, "grad_norm": 6.082028865814209, "learning_rate": 3.2345161290322586e-05, "loss": 0.5243, "step": 5473 }, { "epoch": 35.28674698795181, "grad_norm": 27.345378875732422, "learning_rate": 3.234193548387097e-05, "loss": 0.5386, "step": 5474 }, { "epoch": 35.29317269076305, "grad_norm": 20.332548141479492, "learning_rate": 3.233870967741936e-05, "loss": 0.2736, "step": 5475 }, { "epoch": 35.2995983935743, "grad_norm": 18.904752731323242, "learning_rate": 3.233548387096774e-05, "loss": 0.4089, "step": 5476 }, { "epoch": 35.30602409638554, "grad_norm": 24.92040252685547, "learning_rate": 3.233225806451613e-05, "loss": 0.2937, "step": 5477 }, { "epoch": 35.312449799196784, "grad_norm": 3.1968045234680176, "learning_rate": 3.232903225806452e-05, "loss": 0.2969, "step": 5478 }, { "epoch": 35.318875502008034, "grad_norm": 15.632599830627441, "learning_rate": 3.232580645161291e-05, "loss": 0.3465, "step": 5479 }, { "epoch": 35.325301204819276, "grad_norm": 7.5860514640808105, "learning_rate": 3.232258064516129e-05, "loss": 0.2266, "step": 5480 }, { "epoch": 35.331726907630525, "grad_norm": 35.181312561035156, "learning_rate": 3.231935483870968e-05, "loss": 0.4457, "step": 5481 }, { "epoch": 35.33815261044177, "grad_norm": 12.638949394226074, "learning_rate": 3.2316129032258064e-05, "loss": 0.5463, "step": 5482 }, { "epoch": 35.34457831325301, "grad_norm": 135.1776123046875, "learning_rate": 3.231290322580645e-05, "loss": 0.5742, "step": 5483 }, { "epoch": 35.35100401606426, "grad_norm": 4.746722221374512, "learning_rate": 3.230967741935484e-05, "loss": 0.5246, "step": 5484 }, { "epoch": 35.3574297188755, "grad_norm": 1.316657543182373, "learning_rate": 3.230645161290323e-05, "loss": 0.1699, "step": 5485 }, { "epoch": 35.36385542168675, "grad_norm": 17.933202743530273, "learning_rate": 3.230322580645161e-05, "loss": 0.2935, "step": 5486 }, { "epoch": 35.37028112449799, "grad_norm": 5.0334672927856445, "learning_rate": 3.2300000000000006e-05, "loss": 0.1406, "step": 5487 }, { "epoch": 35.376706827309235, "grad_norm": 10.195870399475098, "learning_rate": 3.229677419354839e-05, "loss": 0.1936, "step": 5488 }, { "epoch": 35.383132530120484, "grad_norm": 17.063074111938477, "learning_rate": 3.229354838709678e-05, "loss": 0.5509, "step": 5489 }, { "epoch": 35.389558232931726, "grad_norm": 15.252354621887207, "learning_rate": 3.229032258064516e-05, "loss": 0.7148, "step": 5490 }, { "epoch": 35.39598393574297, "grad_norm": 5.019874095916748, "learning_rate": 3.228709677419355e-05, "loss": 0.2609, "step": 5491 }, { "epoch": 35.40240963855422, "grad_norm": 20.922687530517578, "learning_rate": 3.2283870967741934e-05, "loss": 0.3649, "step": 5492 }, { "epoch": 35.40883534136546, "grad_norm": 11.698369026184082, "learning_rate": 3.228064516129033e-05, "loss": 0.7169, "step": 5493 }, { "epoch": 35.41526104417671, "grad_norm": 5.548719882965088, "learning_rate": 3.227741935483871e-05, "loss": 0.4117, "step": 5494 }, { "epoch": 35.42168674698795, "grad_norm": 4.585262775421143, "learning_rate": 3.22741935483871e-05, "loss": 0.1809, "step": 5495 }, { "epoch": 35.428112449799194, "grad_norm": 2.6978323459625244, "learning_rate": 3.2270967741935484e-05, "loss": 0.1045, "step": 5496 }, { "epoch": 35.43453815261044, "grad_norm": 2.884577751159668, "learning_rate": 3.2267741935483876e-05, "loss": 0.3722, "step": 5497 }, { "epoch": 35.440963855421685, "grad_norm": 9.078351020812988, "learning_rate": 3.2264516129032255e-05, "loss": 0.6063, "step": 5498 }, { "epoch": 35.447389558232935, "grad_norm": 17.608020782470703, "learning_rate": 3.226129032258065e-05, "loss": 0.401, "step": 5499 }, { "epoch": 35.45381526104418, "grad_norm": 7.209024429321289, "learning_rate": 3.2258064516129034e-05, "loss": 0.2498, "step": 5500 }, { "epoch": 35.46024096385542, "grad_norm": 8.795251846313477, "learning_rate": 3.225483870967742e-05, "loss": 0.5324, "step": 5501 }, { "epoch": 35.46666666666667, "grad_norm": 1.9356951713562012, "learning_rate": 3.225161290322581e-05, "loss": 0.202, "step": 5502 }, { "epoch": 35.47309236947791, "grad_norm": 17.593793869018555, "learning_rate": 3.22483870967742e-05, "loss": 0.6249, "step": 5503 }, { "epoch": 35.47951807228916, "grad_norm": 6.886632442474365, "learning_rate": 3.224516129032258e-05, "loss": 0.1446, "step": 5504 }, { "epoch": 35.4859437751004, "grad_norm": 14.775959968566895, "learning_rate": 3.224193548387097e-05, "loss": 0.1759, "step": 5505 }, { "epoch": 35.492369477911645, "grad_norm": 8.718326568603516, "learning_rate": 3.2238709677419355e-05, "loss": 0.7787, "step": 5506 }, { "epoch": 35.498795180722894, "grad_norm": 12.129290580749512, "learning_rate": 3.223548387096774e-05, "loss": 0.4003, "step": 5507 }, { "epoch": 35.505220883534136, "grad_norm": 17.090478897094727, "learning_rate": 3.223225806451613e-05, "loss": 0.4022, "step": 5508 }, { "epoch": 35.51164658634538, "grad_norm": 33.139495849609375, "learning_rate": 3.222903225806452e-05, "loss": 0.4018, "step": 5509 }, { "epoch": 35.51807228915663, "grad_norm": 8.571504592895508, "learning_rate": 3.2225806451612904e-05, "loss": 0.3901, "step": 5510 }, { "epoch": 35.52449799196787, "grad_norm": 12.466554641723633, "learning_rate": 3.222258064516129e-05, "loss": 0.5231, "step": 5511 }, { "epoch": 35.53092369477912, "grad_norm": 33.41510009765625, "learning_rate": 3.221935483870968e-05, "loss": 0.1434, "step": 5512 }, { "epoch": 35.53734939759036, "grad_norm": 26.295488357543945, "learning_rate": 3.221612903225806e-05, "loss": 0.5809, "step": 5513 }, { "epoch": 35.543775100401604, "grad_norm": 12.074775695800781, "learning_rate": 3.2212903225806454e-05, "loss": 0.2916, "step": 5514 }, { "epoch": 35.55020080321285, "grad_norm": 19.672019958496094, "learning_rate": 3.220967741935484e-05, "loss": 0.4719, "step": 5515 }, { "epoch": 35.556626506024095, "grad_norm": 7.670921802520752, "learning_rate": 3.2206451612903225e-05, "loss": 0.6865, "step": 5516 }, { "epoch": 35.563052208835344, "grad_norm": 8.147035598754883, "learning_rate": 3.220322580645162e-05, "loss": 0.3482, "step": 5517 }, { "epoch": 35.56947791164659, "grad_norm": 2.1091434955596924, "learning_rate": 3.2200000000000003e-05, "loss": 0.2687, "step": 5518 }, { "epoch": 35.57590361445783, "grad_norm": 3.9542384147644043, "learning_rate": 3.219677419354839e-05, "loss": 0.3653, "step": 5519 }, { "epoch": 35.58232931726908, "grad_norm": 8.210250854492188, "learning_rate": 3.2193548387096775e-05, "loss": 0.5122, "step": 5520 }, { "epoch": 35.58875502008032, "grad_norm": 26.43406867980957, "learning_rate": 3.219032258064516e-05, "loss": 0.6355, "step": 5521 }, { "epoch": 35.59518072289157, "grad_norm": 12.335522651672363, "learning_rate": 3.2187096774193546e-05, "loss": 0.5192, "step": 5522 }, { "epoch": 35.60160642570281, "grad_norm": 9.429933547973633, "learning_rate": 3.218387096774194e-05, "loss": 0.2545, "step": 5523 }, { "epoch": 35.608032128514054, "grad_norm": 11.858834266662598, "learning_rate": 3.2180645161290324e-05, "loss": 0.3705, "step": 5524 }, { "epoch": 35.6144578313253, "grad_norm": 6.946832656860352, "learning_rate": 3.217741935483871e-05, "loss": 0.2968, "step": 5525 }, { "epoch": 35.620883534136546, "grad_norm": 8.8915433883667, "learning_rate": 3.21741935483871e-05, "loss": 0.1705, "step": 5526 }, { "epoch": 35.62730923694779, "grad_norm": 14.909684181213379, "learning_rate": 3.217096774193549e-05, "loss": 0.4424, "step": 5527 }, { "epoch": 35.63373493975904, "grad_norm": 9.842211723327637, "learning_rate": 3.2167741935483874e-05, "loss": 0.2476, "step": 5528 }, { "epoch": 35.64016064257028, "grad_norm": 10.228586196899414, "learning_rate": 3.216451612903226e-05, "loss": 0.2413, "step": 5529 }, { "epoch": 35.64658634538153, "grad_norm": 37.013824462890625, "learning_rate": 3.2161290322580645e-05, "loss": 0.573, "step": 5530 }, { "epoch": 35.65301204819277, "grad_norm": 15.918956756591797, "learning_rate": 3.215806451612903e-05, "loss": 0.452, "step": 5531 }, { "epoch": 35.65943775100401, "grad_norm": 4.924895286560059, "learning_rate": 3.2154838709677424e-05, "loss": 0.2594, "step": 5532 }, { "epoch": 35.66586345381526, "grad_norm": 34.252281188964844, "learning_rate": 3.215161290322581e-05, "loss": 0.4515, "step": 5533 }, { "epoch": 35.672289156626505, "grad_norm": 19.007081985473633, "learning_rate": 3.2148387096774195e-05, "loss": 0.1646, "step": 5534 }, { "epoch": 35.678714859437754, "grad_norm": 11.484898567199707, "learning_rate": 3.214516129032258e-05, "loss": 0.2741, "step": 5535 }, { "epoch": 35.685140562248996, "grad_norm": 15.257747650146484, "learning_rate": 3.214193548387097e-05, "loss": 0.5662, "step": 5536 }, { "epoch": 35.69156626506024, "grad_norm": 10.410507202148438, "learning_rate": 3.213870967741935e-05, "loss": 0.4922, "step": 5537 }, { "epoch": 35.69799196787149, "grad_norm": 13.434195518493652, "learning_rate": 3.2135483870967745e-05, "loss": 0.375, "step": 5538 }, { "epoch": 35.70441767068273, "grad_norm": 36.51369857788086, "learning_rate": 3.213225806451613e-05, "loss": 0.5381, "step": 5539 }, { "epoch": 35.71084337349397, "grad_norm": 5.3395819664001465, "learning_rate": 3.2129032258064516e-05, "loss": 0.437, "step": 5540 }, { "epoch": 35.71726907630522, "grad_norm": 9.895275115966797, "learning_rate": 3.212580645161291e-05, "loss": 0.5265, "step": 5541 }, { "epoch": 35.723694779116464, "grad_norm": 1.8676414489746094, "learning_rate": 3.2122580645161294e-05, "loss": 0.3834, "step": 5542 }, { "epoch": 35.73012048192771, "grad_norm": 3.4245567321777344, "learning_rate": 3.211935483870968e-05, "loss": 0.234, "step": 5543 }, { "epoch": 35.736546184738955, "grad_norm": 11.433897018432617, "learning_rate": 3.2116129032258066e-05, "loss": 0.273, "step": 5544 }, { "epoch": 35.7429718875502, "grad_norm": 10.60086727142334, "learning_rate": 3.211290322580645e-05, "loss": 0.7603, "step": 5545 }, { "epoch": 35.74939759036145, "grad_norm": 12.38659954071045, "learning_rate": 3.210967741935484e-05, "loss": 0.4142, "step": 5546 }, { "epoch": 35.75582329317269, "grad_norm": 17.533754348754883, "learning_rate": 3.210645161290323e-05, "loss": 0.7119, "step": 5547 }, { "epoch": 35.76224899598394, "grad_norm": 24.551029205322266, "learning_rate": 3.2103225806451615e-05, "loss": 0.2908, "step": 5548 }, { "epoch": 35.76867469879518, "grad_norm": 4.9779486656188965, "learning_rate": 3.21e-05, "loss": 0.5862, "step": 5549 }, { "epoch": 35.77510040160642, "grad_norm": 10.850873947143555, "learning_rate": 3.2096774193548393e-05, "loss": 0.32, "step": 5550 }, { "epoch": 35.78152610441767, "grad_norm": 21.66697883605957, "learning_rate": 3.209354838709678e-05, "loss": 0.5529, "step": 5551 }, { "epoch": 35.787951807228914, "grad_norm": 18.095638275146484, "learning_rate": 3.209032258064516e-05, "loss": 0.2557, "step": 5552 }, { "epoch": 35.794377510040164, "grad_norm": 9.307257652282715, "learning_rate": 3.208709677419355e-05, "loss": 0.5323, "step": 5553 }, { "epoch": 35.800803212851406, "grad_norm": 7.768399715423584, "learning_rate": 3.2083870967741936e-05, "loss": 0.2598, "step": 5554 }, { "epoch": 35.80722891566265, "grad_norm": 8.253592491149902, "learning_rate": 3.208064516129032e-05, "loss": 0.3678, "step": 5555 }, { "epoch": 35.8136546184739, "grad_norm": 15.745942115783691, "learning_rate": 3.2077419354838714e-05, "loss": 0.3902, "step": 5556 }, { "epoch": 35.82008032128514, "grad_norm": 30.051679611206055, "learning_rate": 3.20741935483871e-05, "loss": 0.3725, "step": 5557 }, { "epoch": 35.82650602409639, "grad_norm": 3.7464630603790283, "learning_rate": 3.2070967741935486e-05, "loss": 0.3734, "step": 5558 }, { "epoch": 35.83293172690763, "grad_norm": 8.845803260803223, "learning_rate": 3.206774193548387e-05, "loss": 0.3561, "step": 5559 }, { "epoch": 35.83935742971887, "grad_norm": 10.094987869262695, "learning_rate": 3.206451612903226e-05, "loss": 0.6115, "step": 5560 }, { "epoch": 35.84578313253012, "grad_norm": 8.625713348388672, "learning_rate": 3.206129032258064e-05, "loss": 0.3357, "step": 5561 }, { "epoch": 35.852208835341365, "grad_norm": 9.704543113708496, "learning_rate": 3.2058064516129036e-05, "loss": 0.3299, "step": 5562 }, { "epoch": 35.85863453815261, "grad_norm": 24.989822387695312, "learning_rate": 3.205483870967742e-05, "loss": 0.4168, "step": 5563 }, { "epoch": 35.86506024096386, "grad_norm": 4.26202917098999, "learning_rate": 3.205161290322581e-05, "loss": 0.2852, "step": 5564 }, { "epoch": 35.8714859437751, "grad_norm": 5.122116565704346, "learning_rate": 3.20483870967742e-05, "loss": 0.3362, "step": 5565 }, { "epoch": 35.87791164658635, "grad_norm": 4.845158100128174, "learning_rate": 3.2045161290322585e-05, "loss": 0.1028, "step": 5566 }, { "epoch": 35.88433734939759, "grad_norm": 4.977379322052002, "learning_rate": 3.204193548387097e-05, "loss": 0.3272, "step": 5567 }, { "epoch": 35.89076305220883, "grad_norm": 19.28655242919922, "learning_rate": 3.2038709677419357e-05, "loss": 0.4535, "step": 5568 }, { "epoch": 35.89718875502008, "grad_norm": 5.527268409729004, "learning_rate": 3.203548387096774e-05, "loss": 0.3673, "step": 5569 }, { "epoch": 35.903614457831324, "grad_norm": 3.7805802822113037, "learning_rate": 3.203225806451613e-05, "loss": 0.4224, "step": 5570 }, { "epoch": 35.91004016064257, "grad_norm": 10.210881233215332, "learning_rate": 3.202903225806452e-05, "loss": 0.5039, "step": 5571 }, { "epoch": 35.916465863453816, "grad_norm": 7.625177383422852, "learning_rate": 3.2025806451612906e-05, "loss": 0.7841, "step": 5572 }, { "epoch": 35.92289156626506, "grad_norm": 4.410627365112305, "learning_rate": 3.202258064516129e-05, "loss": 0.2551, "step": 5573 }, { "epoch": 35.92931726907631, "grad_norm": 12.829569816589355, "learning_rate": 3.201935483870968e-05, "loss": 0.3406, "step": 5574 }, { "epoch": 35.93574297188755, "grad_norm": 11.289910316467285, "learning_rate": 3.201612903225807e-05, "loss": 0.4907, "step": 5575 }, { "epoch": 35.94216867469879, "grad_norm": 12.327706336975098, "learning_rate": 3.201290322580645e-05, "loss": 0.5047, "step": 5576 }, { "epoch": 35.94859437751004, "grad_norm": 11.80156135559082, "learning_rate": 3.200967741935484e-05, "loss": 0.4307, "step": 5577 }, { "epoch": 35.95502008032128, "grad_norm": 16.207761764526367, "learning_rate": 3.200645161290323e-05, "loss": 0.3051, "step": 5578 }, { "epoch": 35.96144578313253, "grad_norm": 6.912444591522217, "learning_rate": 3.200322580645161e-05, "loss": 0.6576, "step": 5579 }, { "epoch": 35.967871485943775, "grad_norm": 17.65351676940918, "learning_rate": 3.2000000000000005e-05, "loss": 0.5901, "step": 5580 }, { "epoch": 35.97429718875502, "grad_norm": 11.51978874206543, "learning_rate": 3.199677419354839e-05, "loss": 0.3035, "step": 5581 }, { "epoch": 35.980722891566266, "grad_norm": 28.39366912841797, "learning_rate": 3.199354838709678e-05, "loss": 0.5978, "step": 5582 }, { "epoch": 35.98714859437751, "grad_norm": 26.02519416809082, "learning_rate": 3.199032258064516e-05, "loss": 0.4749, "step": 5583 }, { "epoch": 35.99357429718876, "grad_norm": 10.318015098571777, "learning_rate": 3.198709677419355e-05, "loss": 0.3529, "step": 5584 }, { "epoch": 36.00321285140562, "grad_norm": 30.89751434326172, "learning_rate": 3.1983870967741934e-05, "loss": 0.3234, "step": 5585 }, { "epoch": 36.00963855421687, "grad_norm": 6.288119316101074, "learning_rate": 3.1980645161290326e-05, "loss": 0.3903, "step": 5586 }, { "epoch": 36.01606425702811, "grad_norm": 15.191824913024902, "learning_rate": 3.197741935483871e-05, "loss": 0.5086, "step": 5587 }, { "epoch": 36.022489959839355, "grad_norm": 9.135346412658691, "learning_rate": 3.19741935483871e-05, "loss": 0.3637, "step": 5588 }, { "epoch": 36.028915662650604, "grad_norm": 8.927167892456055, "learning_rate": 3.197096774193549e-05, "loss": 0.4104, "step": 5589 }, { "epoch": 36.035341365461846, "grad_norm": 12.020379066467285, "learning_rate": 3.1967741935483876e-05, "loss": 0.4422, "step": 5590 }, { "epoch": 36.041767068273096, "grad_norm": 5.27871561050415, "learning_rate": 3.1964516129032255e-05, "loss": 0.1246, "step": 5591 }, { "epoch": 36.04819277108434, "grad_norm": 11.78709602355957, "learning_rate": 3.196129032258065e-05, "loss": 0.2669, "step": 5592 }, { "epoch": 36.05461847389558, "grad_norm": 20.893604278564453, "learning_rate": 3.195806451612903e-05, "loss": 0.4316, "step": 5593 }, { "epoch": 36.06104417670683, "grad_norm": 5.936152458190918, "learning_rate": 3.195483870967742e-05, "loss": 0.2934, "step": 5594 }, { "epoch": 36.06746987951807, "grad_norm": 25.101951599121094, "learning_rate": 3.195161290322581e-05, "loss": 0.3952, "step": 5595 }, { "epoch": 36.073895582329314, "grad_norm": 4.486937046051025, "learning_rate": 3.19483870967742e-05, "loss": 0.3082, "step": 5596 }, { "epoch": 36.08032128514056, "grad_norm": 26.378997802734375, "learning_rate": 3.194516129032258e-05, "loss": 0.3733, "step": 5597 }, { "epoch": 36.086746987951805, "grad_norm": 7.339746475219727, "learning_rate": 3.194193548387097e-05, "loss": 0.2464, "step": 5598 }, { "epoch": 36.093172690763055, "grad_norm": 3.7711172103881836, "learning_rate": 3.1938709677419354e-05, "loss": 0.3643, "step": 5599 }, { "epoch": 36.0995983935743, "grad_norm": 29.57771873474121, "learning_rate": 3.193548387096774e-05, "loss": 0.6506, "step": 5600 }, { "epoch": 36.10602409638554, "grad_norm": 14.468658447265625, "learning_rate": 3.193225806451613e-05, "loss": 0.8176, "step": 5601 }, { "epoch": 36.11244979919679, "grad_norm": 14.935766220092773, "learning_rate": 3.192903225806452e-05, "loss": 0.5264, "step": 5602 }, { "epoch": 36.11887550200803, "grad_norm": 22.91436767578125, "learning_rate": 3.1925806451612904e-05, "loss": 0.4209, "step": 5603 }, { "epoch": 36.12530120481928, "grad_norm": 2.4386184215545654, "learning_rate": 3.1922580645161296e-05, "loss": 0.2017, "step": 5604 }, { "epoch": 36.13172690763052, "grad_norm": 3.7922441959381104, "learning_rate": 3.191935483870968e-05, "loss": 0.18, "step": 5605 }, { "epoch": 36.138152610441765, "grad_norm": 5.171318054199219, "learning_rate": 3.191612903225807e-05, "loss": 0.4023, "step": 5606 }, { "epoch": 36.144578313253014, "grad_norm": 10.681859970092773, "learning_rate": 3.191290322580645e-05, "loss": 0.3972, "step": 5607 }, { "epoch": 36.151004016064256, "grad_norm": 16.952959060668945, "learning_rate": 3.190967741935484e-05, "loss": 0.423, "step": 5608 }, { "epoch": 36.157429718875505, "grad_norm": 7.920362949371338, "learning_rate": 3.1906451612903225e-05, "loss": 0.6349, "step": 5609 }, { "epoch": 36.16385542168675, "grad_norm": 15.017330169677734, "learning_rate": 3.190322580645162e-05, "loss": 0.3214, "step": 5610 }, { "epoch": 36.17028112449799, "grad_norm": 11.486355781555176, "learning_rate": 3.19e-05, "loss": 0.201, "step": 5611 }, { "epoch": 36.17670682730924, "grad_norm": 9.5280122756958, "learning_rate": 3.189677419354839e-05, "loss": 0.5894, "step": 5612 }, { "epoch": 36.18313253012048, "grad_norm": 20.936124801635742, "learning_rate": 3.189354838709678e-05, "loss": 0.5586, "step": 5613 }, { "epoch": 36.189558232931724, "grad_norm": 45.51131820678711, "learning_rate": 3.189032258064517e-05, "loss": 0.575, "step": 5614 }, { "epoch": 36.19598393574297, "grad_norm": 2.4010677337646484, "learning_rate": 3.1887096774193546e-05, "loss": 0.0792, "step": 5615 }, { "epoch": 36.202409638554215, "grad_norm": 21.027301788330078, "learning_rate": 3.188387096774194e-05, "loss": 0.3981, "step": 5616 }, { "epoch": 36.208835341365464, "grad_norm": 28.407289505004883, "learning_rate": 3.1880645161290324e-05, "loss": 0.5864, "step": 5617 }, { "epoch": 36.21526104417671, "grad_norm": 6.470643997192383, "learning_rate": 3.187741935483871e-05, "loss": 0.1644, "step": 5618 }, { "epoch": 36.22168674698795, "grad_norm": 8.766337394714355, "learning_rate": 3.18741935483871e-05, "loss": 0.3502, "step": 5619 }, { "epoch": 36.2281124497992, "grad_norm": 7.370247840881348, "learning_rate": 3.187096774193549e-05, "loss": 0.203, "step": 5620 }, { "epoch": 36.23453815261044, "grad_norm": 5.8531174659729, "learning_rate": 3.1867741935483874e-05, "loss": 0.1848, "step": 5621 }, { "epoch": 36.24096385542169, "grad_norm": 11.177678108215332, "learning_rate": 3.186451612903226e-05, "loss": 0.4648, "step": 5622 }, { "epoch": 36.24738955823293, "grad_norm": 6.722128868103027, "learning_rate": 3.1861290322580645e-05, "loss": 0.606, "step": 5623 }, { "epoch": 36.253815261044174, "grad_norm": 5.854172706604004, "learning_rate": 3.185806451612903e-05, "loss": 0.4168, "step": 5624 }, { "epoch": 36.26024096385542, "grad_norm": 15.370866775512695, "learning_rate": 3.185483870967742e-05, "loss": 0.1948, "step": 5625 }, { "epoch": 36.266666666666666, "grad_norm": 27.207868576049805, "learning_rate": 3.185161290322581e-05, "loss": 0.463, "step": 5626 }, { "epoch": 36.273092369477915, "grad_norm": 7.793202877044678, "learning_rate": 3.1848387096774195e-05, "loss": 0.3063, "step": 5627 }, { "epoch": 36.27951807228916, "grad_norm": 6.8958611488342285, "learning_rate": 3.184516129032259e-05, "loss": 0.382, "step": 5628 }, { "epoch": 36.2859437751004, "grad_norm": 21.36222267150879, "learning_rate": 3.184193548387097e-05, "loss": 0.2285, "step": 5629 }, { "epoch": 36.29236947791165, "grad_norm": 5.637589454650879, "learning_rate": 3.183870967741935e-05, "loss": 0.4769, "step": 5630 }, { "epoch": 36.29879518072289, "grad_norm": 15.44418716430664, "learning_rate": 3.1835483870967744e-05, "loss": 0.3312, "step": 5631 }, { "epoch": 36.30522088353413, "grad_norm": 5.310580253601074, "learning_rate": 3.183225806451613e-05, "loss": 0.2151, "step": 5632 }, { "epoch": 36.31164658634538, "grad_norm": 18.822532653808594, "learning_rate": 3.1829032258064516e-05, "loss": 0.4439, "step": 5633 }, { "epoch": 36.318072289156625, "grad_norm": 14.933919906616211, "learning_rate": 3.182580645161291e-05, "loss": 0.3476, "step": 5634 }, { "epoch": 36.324497991967874, "grad_norm": 5.362615585327148, "learning_rate": 3.1822580645161294e-05, "loss": 0.3596, "step": 5635 }, { "epoch": 36.330923694779116, "grad_norm": 7.355966567993164, "learning_rate": 3.181935483870968e-05, "loss": 0.3774, "step": 5636 }, { "epoch": 36.33734939759036, "grad_norm": 4.104691982269287, "learning_rate": 3.1816129032258065e-05, "loss": 0.3025, "step": 5637 }, { "epoch": 36.34377510040161, "grad_norm": 7.912505149841309, "learning_rate": 3.181290322580645e-05, "loss": 0.3568, "step": 5638 }, { "epoch": 36.35020080321285, "grad_norm": 20.353099822998047, "learning_rate": 3.1809677419354837e-05, "loss": 0.3695, "step": 5639 }, { "epoch": 36.3566265060241, "grad_norm": 6.80876350402832, "learning_rate": 3.180645161290323e-05, "loss": 0.2118, "step": 5640 }, { "epoch": 36.36305220883534, "grad_norm": 21.47087860107422, "learning_rate": 3.1803225806451615e-05, "loss": 0.3597, "step": 5641 }, { "epoch": 36.369477911646584, "grad_norm": 12.653823852539062, "learning_rate": 3.18e-05, "loss": 0.196, "step": 5642 }, { "epoch": 36.37590361445783, "grad_norm": 5.548609733581543, "learning_rate": 3.179677419354839e-05, "loss": 0.1885, "step": 5643 }, { "epoch": 36.382329317269075, "grad_norm": 11.441323280334473, "learning_rate": 3.179354838709678e-05, "loss": 0.5419, "step": 5644 }, { "epoch": 36.388755020080325, "grad_norm": 23.821226119995117, "learning_rate": 3.1790322580645164e-05, "loss": 0.2031, "step": 5645 }, { "epoch": 36.39518072289157, "grad_norm": 19.433631896972656, "learning_rate": 3.178709677419355e-05, "loss": 0.3422, "step": 5646 }, { "epoch": 36.40160642570281, "grad_norm": 18.156978607177734, "learning_rate": 3.1783870967741936e-05, "loss": 0.5713, "step": 5647 }, { "epoch": 36.40803212851406, "grad_norm": 25.945247650146484, "learning_rate": 3.178064516129032e-05, "loss": 0.4364, "step": 5648 }, { "epoch": 36.4144578313253, "grad_norm": 12.387240409851074, "learning_rate": 3.1777419354838714e-05, "loss": 0.5234, "step": 5649 }, { "epoch": 36.42088353413654, "grad_norm": 8.442341804504395, "learning_rate": 3.17741935483871e-05, "loss": 0.3153, "step": 5650 }, { "epoch": 36.42730923694779, "grad_norm": 39.99456024169922, "learning_rate": 3.1770967741935485e-05, "loss": 0.2582, "step": 5651 }, { "epoch": 36.433734939759034, "grad_norm": 6.552863121032715, "learning_rate": 3.176774193548388e-05, "loss": 0.2231, "step": 5652 }, { "epoch": 36.440160642570284, "grad_norm": 12.137663841247559, "learning_rate": 3.176451612903226e-05, "loss": 0.3668, "step": 5653 }, { "epoch": 36.446586345381526, "grad_norm": 9.732499122619629, "learning_rate": 3.176129032258064e-05, "loss": 0.5847, "step": 5654 }, { "epoch": 36.45301204819277, "grad_norm": 2.0515594482421875, "learning_rate": 3.1758064516129035e-05, "loss": 0.1962, "step": 5655 }, { "epoch": 36.45943775100402, "grad_norm": 12.615513801574707, "learning_rate": 3.175483870967742e-05, "loss": 0.3273, "step": 5656 }, { "epoch": 36.46586345381526, "grad_norm": 11.941787719726562, "learning_rate": 3.1751612903225806e-05, "loss": 0.3062, "step": 5657 }, { "epoch": 36.47228915662651, "grad_norm": 96.43598937988281, "learning_rate": 3.17483870967742e-05, "loss": 0.5971, "step": 5658 }, { "epoch": 36.47871485943775, "grad_norm": 15.073031425476074, "learning_rate": 3.1745161290322585e-05, "loss": 0.8849, "step": 5659 }, { "epoch": 36.48514056224899, "grad_norm": 12.250102996826172, "learning_rate": 3.174193548387097e-05, "loss": 0.224, "step": 5660 }, { "epoch": 36.49156626506024, "grad_norm": 16.165542602539062, "learning_rate": 3.1738709677419356e-05, "loss": 0.481, "step": 5661 }, { "epoch": 36.497991967871485, "grad_norm": 5.1485514640808105, "learning_rate": 3.173548387096774e-05, "loss": 0.4428, "step": 5662 }, { "epoch": 36.504417670682734, "grad_norm": 12.305085182189941, "learning_rate": 3.173225806451613e-05, "loss": 0.3777, "step": 5663 }, { "epoch": 36.51084337349398, "grad_norm": 15.629007339477539, "learning_rate": 3.172903225806452e-05, "loss": 0.3801, "step": 5664 }, { "epoch": 36.51726907630522, "grad_norm": 4.881900310516357, "learning_rate": 3.1725806451612906e-05, "loss": 0.4364, "step": 5665 }, { "epoch": 36.52369477911647, "grad_norm": 6.519387245178223, "learning_rate": 3.172258064516129e-05, "loss": 0.2474, "step": 5666 }, { "epoch": 36.53012048192771, "grad_norm": 6.623727321624756, "learning_rate": 3.1719354838709684e-05, "loss": 0.5748, "step": 5667 }, { "epoch": 36.53654618473895, "grad_norm": 2.101713180541992, "learning_rate": 3.171612903225807e-05, "loss": 0.1724, "step": 5668 }, { "epoch": 36.5429718875502, "grad_norm": 8.153928756713867, "learning_rate": 3.171290322580645e-05, "loss": 0.3511, "step": 5669 }, { "epoch": 36.549397590361444, "grad_norm": 12.248947143554688, "learning_rate": 3.170967741935484e-05, "loss": 0.5024, "step": 5670 }, { "epoch": 36.55582329317269, "grad_norm": 20.86134147644043, "learning_rate": 3.170645161290323e-05, "loss": 0.4614, "step": 5671 }, { "epoch": 36.562248995983936, "grad_norm": 4.313559055328369, "learning_rate": 3.170322580645161e-05, "loss": 0.2026, "step": 5672 }, { "epoch": 36.56867469879518, "grad_norm": 20.622039794921875, "learning_rate": 3.1700000000000005e-05, "loss": 0.5763, "step": 5673 }, { "epoch": 36.57510040160643, "grad_norm": 47.373558044433594, "learning_rate": 3.169677419354839e-05, "loss": 0.6822, "step": 5674 }, { "epoch": 36.58152610441767, "grad_norm": 5.043015480041504, "learning_rate": 3.1693548387096776e-05, "loss": 0.2258, "step": 5675 }, { "epoch": 36.58795180722892, "grad_norm": 28.81036376953125, "learning_rate": 3.169032258064517e-05, "loss": 0.4724, "step": 5676 }, { "epoch": 36.59437751004016, "grad_norm": 5.469024658203125, "learning_rate": 3.168709677419355e-05, "loss": 0.5338, "step": 5677 }, { "epoch": 36.6008032128514, "grad_norm": 15.516695022583008, "learning_rate": 3.168387096774193e-05, "loss": 0.4347, "step": 5678 }, { "epoch": 36.60722891566265, "grad_norm": 9.022218704223633, "learning_rate": 3.1680645161290326e-05, "loss": 0.3909, "step": 5679 }, { "epoch": 36.613654618473895, "grad_norm": 23.481327056884766, "learning_rate": 3.167741935483871e-05, "loss": 0.3148, "step": 5680 }, { "epoch": 36.620080321285144, "grad_norm": 19.02046012878418, "learning_rate": 3.16741935483871e-05, "loss": 0.4885, "step": 5681 }, { "epoch": 36.626506024096386, "grad_norm": 8.212757110595703, "learning_rate": 3.167096774193549e-05, "loss": 0.2105, "step": 5682 }, { "epoch": 36.63293172690763, "grad_norm": 14.005293846130371, "learning_rate": 3.1667741935483875e-05, "loss": 0.5624, "step": 5683 }, { "epoch": 36.63935742971888, "grad_norm": 5.90186071395874, "learning_rate": 3.166451612903226e-05, "loss": 0.2991, "step": 5684 }, { "epoch": 36.64578313253012, "grad_norm": 5.6647562980651855, "learning_rate": 3.166129032258065e-05, "loss": 0.2142, "step": 5685 }, { "epoch": 36.65220883534136, "grad_norm": 5.271583557128906, "learning_rate": 3.165806451612903e-05, "loss": 0.4369, "step": 5686 }, { "epoch": 36.65863453815261, "grad_norm": 43.8637580871582, "learning_rate": 3.165483870967742e-05, "loss": 0.558, "step": 5687 }, { "epoch": 36.665060240963854, "grad_norm": 17.867956161499023, "learning_rate": 3.165161290322581e-05, "loss": 0.3103, "step": 5688 }, { "epoch": 36.6714859437751, "grad_norm": 7.599201202392578, "learning_rate": 3.1648387096774196e-05, "loss": 0.4165, "step": 5689 }, { "epoch": 36.677911646586345, "grad_norm": 30.82840347290039, "learning_rate": 3.164516129032258e-05, "loss": 0.4404, "step": 5690 }, { "epoch": 36.68433734939759, "grad_norm": 5.99545955657959, "learning_rate": 3.1641935483870975e-05, "loss": 0.3954, "step": 5691 }, { "epoch": 36.69076305220884, "grad_norm": 2.6357879638671875, "learning_rate": 3.1638709677419354e-05, "loss": 0.1813, "step": 5692 }, { "epoch": 36.69718875502008, "grad_norm": 4.6345930099487305, "learning_rate": 3.163548387096774e-05, "loss": 0.1976, "step": 5693 }, { "epoch": 36.70361445783133, "grad_norm": 8.675241470336914, "learning_rate": 3.163225806451613e-05, "loss": 0.4776, "step": 5694 }, { "epoch": 36.71004016064257, "grad_norm": 21.467803955078125, "learning_rate": 3.162903225806452e-05, "loss": 0.5604, "step": 5695 }, { "epoch": 36.71646586345381, "grad_norm": 6.144473552703857, "learning_rate": 3.16258064516129e-05, "loss": 0.4415, "step": 5696 }, { "epoch": 36.72289156626506, "grad_norm": 16.230268478393555, "learning_rate": 3.1622580645161296e-05, "loss": 0.6331, "step": 5697 }, { "epoch": 36.729317269076304, "grad_norm": 10.471996307373047, "learning_rate": 3.161935483870968e-05, "loss": 0.4971, "step": 5698 }, { "epoch": 36.735742971887554, "grad_norm": 11.536189079284668, "learning_rate": 3.161612903225807e-05, "loss": 0.4392, "step": 5699 }, { "epoch": 36.742168674698796, "grad_norm": 18.079795837402344, "learning_rate": 3.161290322580645e-05, "loss": 0.2977, "step": 5700 }, { "epoch": 36.74859437751004, "grad_norm": 23.70166778564453, "learning_rate": 3.160967741935484e-05, "loss": 0.5075, "step": 5701 }, { "epoch": 36.75502008032129, "grad_norm": 12.857414245605469, "learning_rate": 3.1606451612903224e-05, "loss": 0.3127, "step": 5702 }, { "epoch": 36.76144578313253, "grad_norm": 9.60131549835205, "learning_rate": 3.160322580645162e-05, "loss": 0.1895, "step": 5703 }, { "epoch": 36.76787148594377, "grad_norm": 31.889434814453125, "learning_rate": 3.16e-05, "loss": 0.4174, "step": 5704 }, { "epoch": 36.77429718875502, "grad_norm": 11.1685152053833, "learning_rate": 3.159677419354839e-05, "loss": 0.3957, "step": 5705 }, { "epoch": 36.78072289156626, "grad_norm": 8.477087020874023, "learning_rate": 3.159354838709678e-05, "loss": 0.49, "step": 5706 }, { "epoch": 36.78714859437751, "grad_norm": 9.56146240234375, "learning_rate": 3.1590322580645166e-05, "loss": 0.5723, "step": 5707 }, { "epoch": 36.793574297188755, "grad_norm": 21.058082580566406, "learning_rate": 3.1587096774193545e-05, "loss": 0.4338, "step": 5708 }, { "epoch": 36.8, "grad_norm": 31.182214736938477, "learning_rate": 3.158387096774194e-05, "loss": 0.4851, "step": 5709 }, { "epoch": 36.80642570281125, "grad_norm": 8.968403816223145, "learning_rate": 3.1580645161290323e-05, "loss": 0.3983, "step": 5710 }, { "epoch": 36.81285140562249, "grad_norm": 6.617511749267578, "learning_rate": 3.157741935483871e-05, "loss": 0.1582, "step": 5711 }, { "epoch": 36.81927710843374, "grad_norm": 6.8708391189575195, "learning_rate": 3.15741935483871e-05, "loss": 0.3943, "step": 5712 }, { "epoch": 36.82570281124498, "grad_norm": 9.843267440795898, "learning_rate": 3.157096774193549e-05, "loss": 0.3178, "step": 5713 }, { "epoch": 36.83212851405622, "grad_norm": 1.5683332681655884, "learning_rate": 3.156774193548387e-05, "loss": 0.2147, "step": 5714 }, { "epoch": 36.83855421686747, "grad_norm": 3.176229476928711, "learning_rate": 3.1564516129032265e-05, "loss": 0.1045, "step": 5715 }, { "epoch": 36.844979919678714, "grad_norm": 19.252534866333008, "learning_rate": 3.1561290322580644e-05, "loss": 0.4297, "step": 5716 }, { "epoch": 36.85140562248996, "grad_norm": 11.503835678100586, "learning_rate": 3.155806451612903e-05, "loss": 0.4954, "step": 5717 }, { "epoch": 36.857831325301206, "grad_norm": 2.9410290718078613, "learning_rate": 3.155483870967742e-05, "loss": 0.1787, "step": 5718 }, { "epoch": 36.86425702811245, "grad_norm": 6.322312831878662, "learning_rate": 3.155161290322581e-05, "loss": 0.4433, "step": 5719 }, { "epoch": 36.8706827309237, "grad_norm": 8.861842155456543, "learning_rate": 3.1548387096774194e-05, "loss": 0.4807, "step": 5720 }, { "epoch": 36.87710843373494, "grad_norm": 2.9272401332855225, "learning_rate": 3.1545161290322587e-05, "loss": 0.3067, "step": 5721 }, { "epoch": 36.88353413654618, "grad_norm": 16.161285400390625, "learning_rate": 3.154193548387097e-05, "loss": 0.3377, "step": 5722 }, { "epoch": 36.88995983935743, "grad_norm": 8.317582130432129, "learning_rate": 3.153870967741936e-05, "loss": 0.349, "step": 5723 }, { "epoch": 36.89638554216867, "grad_norm": 10.643819808959961, "learning_rate": 3.1535483870967744e-05, "loss": 0.1654, "step": 5724 }, { "epoch": 36.90281124497992, "grad_norm": 5.318249225616455, "learning_rate": 3.153225806451613e-05, "loss": 0.2449, "step": 5725 }, { "epoch": 36.909236947791165, "grad_norm": 12.61851978302002, "learning_rate": 3.1529032258064515e-05, "loss": 0.4002, "step": 5726 }, { "epoch": 36.91566265060241, "grad_norm": 4.739556312561035, "learning_rate": 3.152580645161291e-05, "loss": 0.2316, "step": 5727 }, { "epoch": 36.922088353413656, "grad_norm": 15.036227226257324, "learning_rate": 3.152258064516129e-05, "loss": 0.2393, "step": 5728 }, { "epoch": 36.9285140562249, "grad_norm": 23.610763549804688, "learning_rate": 3.151935483870968e-05, "loss": 0.3356, "step": 5729 }, { "epoch": 36.93493975903615, "grad_norm": 3.9302151203155518, "learning_rate": 3.151612903225807e-05, "loss": 0.4912, "step": 5730 }, { "epoch": 36.94136546184739, "grad_norm": 4.319417476654053, "learning_rate": 3.151290322580645e-05, "loss": 0.2851, "step": 5731 }, { "epoch": 36.94779116465863, "grad_norm": 7.0891008377075195, "learning_rate": 3.1509677419354836e-05, "loss": 0.3076, "step": 5732 }, { "epoch": 36.95421686746988, "grad_norm": 29.41510009765625, "learning_rate": 3.150645161290323e-05, "loss": 0.4099, "step": 5733 }, { "epoch": 36.960642570281124, "grad_norm": 11.885838508605957, "learning_rate": 3.1503225806451614e-05, "loss": 0.455, "step": 5734 }, { "epoch": 36.967068273092366, "grad_norm": 28.126789093017578, "learning_rate": 3.15e-05, "loss": 0.3082, "step": 5735 }, { "epoch": 36.973493975903615, "grad_norm": 3.8031365871429443, "learning_rate": 3.149677419354839e-05, "loss": 0.128, "step": 5736 }, { "epoch": 36.97991967871486, "grad_norm": 9.510915756225586, "learning_rate": 3.149354838709678e-05, "loss": 0.3862, "step": 5737 }, { "epoch": 36.98634538152611, "grad_norm": 14.088754653930664, "learning_rate": 3.1490322580645164e-05, "loss": 0.2654, "step": 5738 }, { "epoch": 36.99277108433735, "grad_norm": 8.670621871948242, "learning_rate": 3.148709677419355e-05, "loss": 0.4537, "step": 5739 }, { "epoch": 37.00240963855422, "grad_norm": 20.629701614379883, "learning_rate": 3.1483870967741935e-05, "loss": 0.4553, "step": 5740 }, { "epoch": 37.00883534136546, "grad_norm": 6.2381272315979, "learning_rate": 3.148064516129032e-05, "loss": 0.2952, "step": 5741 }, { "epoch": 37.015261044176704, "grad_norm": 11.07998275756836, "learning_rate": 3.1477419354838713e-05, "loss": 0.2014, "step": 5742 }, { "epoch": 37.02168674698795, "grad_norm": 14.502941131591797, "learning_rate": 3.14741935483871e-05, "loss": 0.6068, "step": 5743 }, { "epoch": 37.028112449799195, "grad_norm": 7.779422760009766, "learning_rate": 3.1470967741935485e-05, "loss": 0.3657, "step": 5744 }, { "epoch": 37.034538152610445, "grad_norm": 12.206379890441895, "learning_rate": 3.146774193548388e-05, "loss": 0.3509, "step": 5745 }, { "epoch": 37.04096385542169, "grad_norm": 37.425926208496094, "learning_rate": 3.146451612903226e-05, "loss": 0.3327, "step": 5746 }, { "epoch": 37.04738955823293, "grad_norm": 6.251177787780762, "learning_rate": 3.146129032258064e-05, "loss": 0.2529, "step": 5747 }, { "epoch": 37.05381526104418, "grad_norm": 5.436710357666016, "learning_rate": 3.1458064516129034e-05, "loss": 0.4576, "step": 5748 }, { "epoch": 37.06024096385542, "grad_norm": 3.6658122539520264, "learning_rate": 3.145483870967742e-05, "loss": 0.1771, "step": 5749 }, { "epoch": 37.06666666666667, "grad_norm": 10.995026588439941, "learning_rate": 3.1451612903225806e-05, "loss": 0.4588, "step": 5750 }, { "epoch": 37.07309236947791, "grad_norm": 7.3017144203186035, "learning_rate": 3.14483870967742e-05, "loss": 0.3142, "step": 5751 }, { "epoch": 37.079518072289154, "grad_norm": 5.449059009552002, "learning_rate": 3.1445161290322584e-05, "loss": 0.369, "step": 5752 }, { "epoch": 37.085943775100404, "grad_norm": 10.191740036010742, "learning_rate": 3.144193548387097e-05, "loss": 0.5555, "step": 5753 }, { "epoch": 37.092369477911646, "grad_norm": 12.187673568725586, "learning_rate": 3.143870967741936e-05, "loss": 0.4287, "step": 5754 }, { "epoch": 37.09879518072289, "grad_norm": 4.816283702850342, "learning_rate": 3.143548387096774e-05, "loss": 0.2596, "step": 5755 }, { "epoch": 37.10522088353414, "grad_norm": 5.797554016113281, "learning_rate": 3.143225806451613e-05, "loss": 0.4462, "step": 5756 }, { "epoch": 37.11164658634538, "grad_norm": 8.361015319824219, "learning_rate": 3.142903225806452e-05, "loss": 0.1667, "step": 5757 }, { "epoch": 37.11807228915663, "grad_norm": 12.410035133361816, "learning_rate": 3.1425806451612905e-05, "loss": 0.8445, "step": 5758 }, { "epoch": 37.12449799196787, "grad_norm": 11.729413032531738, "learning_rate": 3.142258064516129e-05, "loss": 0.4867, "step": 5759 }, { "epoch": 37.13092369477911, "grad_norm": 11.917558670043945, "learning_rate": 3.141935483870968e-05, "loss": 0.6573, "step": 5760 }, { "epoch": 37.13734939759036, "grad_norm": 15.020445823669434, "learning_rate": 3.141612903225807e-05, "loss": 0.334, "step": 5761 }, { "epoch": 37.143775100401605, "grad_norm": 9.140774726867676, "learning_rate": 3.141290322580645e-05, "loss": 0.1397, "step": 5762 }, { "epoch": 37.150200803212854, "grad_norm": 1.8136711120605469, "learning_rate": 3.140967741935484e-05, "loss": 0.3196, "step": 5763 }, { "epoch": 37.1566265060241, "grad_norm": 34.90462112426758, "learning_rate": 3.1406451612903226e-05, "loss": 0.192, "step": 5764 }, { "epoch": 37.16305220883534, "grad_norm": 16.888742446899414, "learning_rate": 3.140322580645161e-05, "loss": 0.3706, "step": 5765 }, { "epoch": 37.16947791164659, "grad_norm": 15.61706256866455, "learning_rate": 3.1400000000000004e-05, "loss": 0.5069, "step": 5766 }, { "epoch": 37.17590361445783, "grad_norm": 8.10888385772705, "learning_rate": 3.139677419354839e-05, "loss": 0.4149, "step": 5767 }, { "epoch": 37.18232931726908, "grad_norm": 20.86008071899414, "learning_rate": 3.1393548387096776e-05, "loss": 0.4137, "step": 5768 }, { "epoch": 37.18875502008032, "grad_norm": 19.472814559936523, "learning_rate": 3.139032258064517e-05, "loss": 0.2896, "step": 5769 }, { "epoch": 37.195180722891564, "grad_norm": 9.436918258666992, "learning_rate": 3.138709677419355e-05, "loss": 0.4759, "step": 5770 }, { "epoch": 37.20160642570281, "grad_norm": 25.17368507385254, "learning_rate": 3.138387096774193e-05, "loss": 0.55, "step": 5771 }, { "epoch": 37.208032128514056, "grad_norm": 11.49077033996582, "learning_rate": 3.1380645161290325e-05, "loss": 0.2421, "step": 5772 }, { "epoch": 37.2144578313253, "grad_norm": 14.978940963745117, "learning_rate": 3.137741935483871e-05, "loss": 0.4459, "step": 5773 }, { "epoch": 37.22088353413655, "grad_norm": 11.639519691467285, "learning_rate": 3.13741935483871e-05, "loss": 0.3219, "step": 5774 }, { "epoch": 37.22730923694779, "grad_norm": 11.84907054901123, "learning_rate": 3.137096774193549e-05, "loss": 0.5591, "step": 5775 }, { "epoch": 37.23373493975904, "grad_norm": 8.199860572814941, "learning_rate": 3.1367741935483875e-05, "loss": 0.1343, "step": 5776 }, { "epoch": 37.24016064257028, "grad_norm": 4.020561695098877, "learning_rate": 3.136451612903226e-05, "loss": 0.2526, "step": 5777 }, { "epoch": 37.24658634538152, "grad_norm": 11.831121444702148, "learning_rate": 3.1361290322580646e-05, "loss": 0.455, "step": 5778 }, { "epoch": 37.25301204819277, "grad_norm": 7.433454990386963, "learning_rate": 3.135806451612903e-05, "loss": 0.3751, "step": 5779 }, { "epoch": 37.259437751004015, "grad_norm": 5.382028102874756, "learning_rate": 3.135483870967742e-05, "loss": 0.2966, "step": 5780 }, { "epoch": 37.265863453815264, "grad_norm": 17.91098976135254, "learning_rate": 3.135161290322581e-05, "loss": 0.2965, "step": 5781 }, { "epoch": 37.272289156626506, "grad_norm": 4.283815860748291, "learning_rate": 3.1348387096774196e-05, "loss": 0.2464, "step": 5782 }, { "epoch": 37.27871485943775, "grad_norm": 3.1905717849731445, "learning_rate": 3.134516129032258e-05, "loss": 0.227, "step": 5783 }, { "epoch": 37.285140562249, "grad_norm": 10.196014404296875, "learning_rate": 3.1341935483870974e-05, "loss": 0.505, "step": 5784 }, { "epoch": 37.29156626506024, "grad_norm": 24.2209415435791, "learning_rate": 3.133870967741936e-05, "loss": 0.394, "step": 5785 }, { "epoch": 37.29799196787149, "grad_norm": 16.17068862915039, "learning_rate": 3.133548387096774e-05, "loss": 0.6786, "step": 5786 }, { "epoch": 37.30441767068273, "grad_norm": 12.783659934997559, "learning_rate": 3.133225806451613e-05, "loss": 0.3353, "step": 5787 }, { "epoch": 37.310843373493974, "grad_norm": 22.023649215698242, "learning_rate": 3.132903225806452e-05, "loss": 0.4617, "step": 5788 }, { "epoch": 37.31726907630522, "grad_norm": 8.083380699157715, "learning_rate": 3.13258064516129e-05, "loss": 0.6749, "step": 5789 }, { "epoch": 37.323694779116465, "grad_norm": 9.441837310791016, "learning_rate": 3.1322580645161295e-05, "loss": 0.2733, "step": 5790 }, { "epoch": 37.33012048192771, "grad_norm": 6.871176242828369, "learning_rate": 3.131935483870968e-05, "loss": 0.4594, "step": 5791 }, { "epoch": 37.33654618473896, "grad_norm": 11.893040657043457, "learning_rate": 3.1316129032258067e-05, "loss": 0.4642, "step": 5792 }, { "epoch": 37.3429718875502, "grad_norm": 8.694024085998535, "learning_rate": 3.131290322580646e-05, "loss": 0.2779, "step": 5793 }, { "epoch": 37.34939759036145, "grad_norm": 9.107980728149414, "learning_rate": 3.130967741935484e-05, "loss": 0.2875, "step": 5794 }, { "epoch": 37.35582329317269, "grad_norm": 9.644786834716797, "learning_rate": 3.1306451612903224e-05, "loss": 0.5894, "step": 5795 }, { "epoch": 37.36224899598393, "grad_norm": 4.8915863037109375, "learning_rate": 3.1303225806451616e-05, "loss": 0.3463, "step": 5796 }, { "epoch": 37.36867469879518, "grad_norm": 16.39217758178711, "learning_rate": 3.13e-05, "loss": 0.3112, "step": 5797 }, { "epoch": 37.375100401606424, "grad_norm": 15.237439155578613, "learning_rate": 3.129677419354839e-05, "loss": 0.532, "step": 5798 }, { "epoch": 37.381526104417674, "grad_norm": 4.54960823059082, "learning_rate": 3.129354838709678e-05, "loss": 0.2987, "step": 5799 }, { "epoch": 37.387951807228916, "grad_norm": 11.05309009552002, "learning_rate": 3.1290322580645166e-05, "loss": 0.2708, "step": 5800 }, { "epoch": 37.39437751004016, "grad_norm": 7.264349460601807, "learning_rate": 3.1287096774193545e-05, "loss": 0.3898, "step": 5801 }, { "epoch": 37.40080321285141, "grad_norm": 6.16087532043457, "learning_rate": 3.128387096774194e-05, "loss": 0.2761, "step": 5802 }, { "epoch": 37.40722891566265, "grad_norm": 8.944108009338379, "learning_rate": 3.128064516129032e-05, "loss": 0.3687, "step": 5803 }, { "epoch": 37.4136546184739, "grad_norm": 5.164125442504883, "learning_rate": 3.127741935483871e-05, "loss": 0.3702, "step": 5804 }, { "epoch": 37.42008032128514, "grad_norm": 5.1288580894470215, "learning_rate": 3.12741935483871e-05, "loss": 0.2277, "step": 5805 }, { "epoch": 37.42650602409638, "grad_norm": 23.163137435913086, "learning_rate": 3.127096774193549e-05, "loss": 0.4179, "step": 5806 }, { "epoch": 37.43293172690763, "grad_norm": 22.147802352905273, "learning_rate": 3.126774193548387e-05, "loss": 0.3998, "step": 5807 }, { "epoch": 37.439357429718875, "grad_norm": 19.483558654785156, "learning_rate": 3.1264516129032265e-05, "loss": 0.3165, "step": 5808 }, { "epoch": 37.44578313253012, "grad_norm": 2.9181015491485596, "learning_rate": 3.1261290322580644e-05, "loss": 0.4441, "step": 5809 }, { "epoch": 37.45220883534137, "grad_norm": 16.817424774169922, "learning_rate": 3.125806451612903e-05, "loss": 0.5186, "step": 5810 }, { "epoch": 37.45863453815261, "grad_norm": 8.699834823608398, "learning_rate": 3.125483870967742e-05, "loss": 0.3343, "step": 5811 }, { "epoch": 37.46506024096386, "grad_norm": 5.227504253387451, "learning_rate": 3.125161290322581e-05, "loss": 0.3339, "step": 5812 }, { "epoch": 37.4714859437751, "grad_norm": 10.956294059753418, "learning_rate": 3.1248387096774194e-05, "loss": 0.4476, "step": 5813 }, { "epoch": 37.47791164658634, "grad_norm": 8.89838695526123, "learning_rate": 3.1245161290322586e-05, "loss": 0.3362, "step": 5814 }, { "epoch": 37.48433734939759, "grad_norm": 22.05040740966797, "learning_rate": 3.124193548387097e-05, "loss": 0.4498, "step": 5815 }, { "epoch": 37.490763052208834, "grad_norm": 9.59993839263916, "learning_rate": 3.123870967741936e-05, "loss": 0.3566, "step": 5816 }, { "epoch": 37.49718875502008, "grad_norm": 6.337425708770752, "learning_rate": 3.123548387096774e-05, "loss": 0.3332, "step": 5817 }, { "epoch": 37.503614457831326, "grad_norm": 3.20990252494812, "learning_rate": 3.123225806451613e-05, "loss": 0.171, "step": 5818 }, { "epoch": 37.51004016064257, "grad_norm": 15.7053861618042, "learning_rate": 3.1229032258064515e-05, "loss": 0.5642, "step": 5819 }, { "epoch": 37.51646586345382, "grad_norm": 15.6851806640625, "learning_rate": 3.122580645161291e-05, "loss": 0.3206, "step": 5820 }, { "epoch": 37.52289156626506, "grad_norm": 47.57050323486328, "learning_rate": 3.122258064516129e-05, "loss": 0.2829, "step": 5821 }, { "epoch": 37.52931726907631, "grad_norm": 14.897741317749023, "learning_rate": 3.121935483870968e-05, "loss": 0.3821, "step": 5822 }, { "epoch": 37.53574297188755, "grad_norm": 5.2519965171813965, "learning_rate": 3.121612903225807e-05, "loss": 0.2527, "step": 5823 }, { "epoch": 37.54216867469879, "grad_norm": 4.785907745361328, "learning_rate": 3.1212903225806457e-05, "loss": 0.4167, "step": 5824 }, { "epoch": 37.54859437751004, "grad_norm": 2.3471410274505615, "learning_rate": 3.1209677419354836e-05, "loss": 0.1027, "step": 5825 }, { "epoch": 37.555020080321285, "grad_norm": 3.741581916809082, "learning_rate": 3.120645161290323e-05, "loss": 0.2362, "step": 5826 }, { "epoch": 37.56144578313253, "grad_norm": 17.509702682495117, "learning_rate": 3.1203225806451614e-05, "loss": 0.3417, "step": 5827 }, { "epoch": 37.567871485943776, "grad_norm": 7.825664520263672, "learning_rate": 3.12e-05, "loss": 0.3939, "step": 5828 }, { "epoch": 37.57429718875502, "grad_norm": 10.027355194091797, "learning_rate": 3.119677419354839e-05, "loss": 0.3092, "step": 5829 }, { "epoch": 37.58072289156627, "grad_norm": 9.868109703063965, "learning_rate": 3.119354838709678e-05, "loss": 0.4844, "step": 5830 }, { "epoch": 37.58714859437751, "grad_norm": 8.292624473571777, "learning_rate": 3.119032258064516e-05, "loss": 0.2782, "step": 5831 }, { "epoch": 37.59357429718875, "grad_norm": 8.098797798156738, "learning_rate": 3.118709677419355e-05, "loss": 0.2262, "step": 5832 }, { "epoch": 37.6, "grad_norm": 10.319704055786133, "learning_rate": 3.1183870967741935e-05, "loss": 0.4207, "step": 5833 }, { "epoch": 37.606425702811244, "grad_norm": 8.930638313293457, "learning_rate": 3.118064516129032e-05, "loss": 0.5004, "step": 5834 }, { "epoch": 37.61285140562249, "grad_norm": 16.701677322387695, "learning_rate": 3.117741935483871e-05, "loss": 0.4021, "step": 5835 }, { "epoch": 37.619277108433735, "grad_norm": 11.013955116271973, "learning_rate": 3.11741935483871e-05, "loss": 0.2559, "step": 5836 }, { "epoch": 37.62570281124498, "grad_norm": 8.443303108215332, "learning_rate": 3.1170967741935484e-05, "loss": 0.4251, "step": 5837 }, { "epoch": 37.63212851405623, "grad_norm": 11.102761268615723, "learning_rate": 3.116774193548388e-05, "loss": 0.5872, "step": 5838 }, { "epoch": 37.63855421686747, "grad_norm": 23.161226272583008, "learning_rate": 3.116451612903226e-05, "loss": 0.7214, "step": 5839 }, { "epoch": 37.64497991967872, "grad_norm": 28.699281692504883, "learning_rate": 3.116129032258064e-05, "loss": 0.5903, "step": 5840 }, { "epoch": 37.65140562248996, "grad_norm": 3.2727291584014893, "learning_rate": 3.1158064516129034e-05, "loss": 0.4142, "step": 5841 }, { "epoch": 37.6578313253012, "grad_norm": 8.630890846252441, "learning_rate": 3.115483870967742e-05, "loss": 0.5308, "step": 5842 }, { "epoch": 37.66425702811245, "grad_norm": 3.604700803756714, "learning_rate": 3.1151612903225805e-05, "loss": 0.3221, "step": 5843 }, { "epoch": 37.670682730923694, "grad_norm": 10.573225975036621, "learning_rate": 3.11483870967742e-05, "loss": 0.3009, "step": 5844 }, { "epoch": 37.67710843373494, "grad_norm": 20.87038230895996, "learning_rate": 3.1145161290322584e-05, "loss": 0.4517, "step": 5845 }, { "epoch": 37.683534136546186, "grad_norm": 14.945470809936523, "learning_rate": 3.114193548387097e-05, "loss": 0.4202, "step": 5846 }, { "epoch": 37.68995983935743, "grad_norm": 3.4107282161712646, "learning_rate": 3.113870967741936e-05, "loss": 0.542, "step": 5847 }, { "epoch": 37.69638554216868, "grad_norm": 10.834370613098145, "learning_rate": 3.113548387096774e-05, "loss": 0.3329, "step": 5848 }, { "epoch": 37.70281124497992, "grad_norm": 11.179727554321289, "learning_rate": 3.1132258064516126e-05, "loss": 0.4828, "step": 5849 }, { "epoch": 37.70923694779116, "grad_norm": 7.655165672302246, "learning_rate": 3.112903225806452e-05, "loss": 0.3511, "step": 5850 }, { "epoch": 37.71566265060241, "grad_norm": 12.636025428771973, "learning_rate": 3.1125806451612905e-05, "loss": 0.4783, "step": 5851 }, { "epoch": 37.72208835341365, "grad_norm": 8.397698402404785, "learning_rate": 3.112258064516129e-05, "loss": 0.1584, "step": 5852 }, { "epoch": 37.7285140562249, "grad_norm": 27.737714767456055, "learning_rate": 3.111935483870968e-05, "loss": 0.5741, "step": 5853 }, { "epoch": 37.734939759036145, "grad_norm": 8.396973609924316, "learning_rate": 3.111612903225807e-05, "loss": 0.2482, "step": 5854 }, { "epoch": 37.74136546184739, "grad_norm": 23.77396583557129, "learning_rate": 3.1112903225806454e-05, "loss": 0.3839, "step": 5855 }, { "epoch": 37.747791164658636, "grad_norm": 8.94832706451416, "learning_rate": 3.110967741935484e-05, "loss": 0.2381, "step": 5856 }, { "epoch": 37.75421686746988, "grad_norm": 11.563477516174316, "learning_rate": 3.1106451612903226e-05, "loss": 0.4718, "step": 5857 }, { "epoch": 37.76064257028113, "grad_norm": 19.208927154541016, "learning_rate": 3.110322580645161e-05, "loss": 0.2345, "step": 5858 }, { "epoch": 37.76706827309237, "grad_norm": 5.3177690505981445, "learning_rate": 3.1100000000000004e-05, "loss": 0.2849, "step": 5859 }, { "epoch": 37.77349397590361, "grad_norm": 13.551410675048828, "learning_rate": 3.109677419354839e-05, "loss": 0.3145, "step": 5860 }, { "epoch": 37.77991967871486, "grad_norm": 25.624698638916016, "learning_rate": 3.1093548387096775e-05, "loss": 0.4112, "step": 5861 }, { "epoch": 37.786345381526104, "grad_norm": 10.981632232666016, "learning_rate": 3.109032258064517e-05, "loss": 0.39, "step": 5862 }, { "epoch": 37.792771084337346, "grad_norm": 56.516395568847656, "learning_rate": 3.108709677419355e-05, "loss": 0.4045, "step": 5863 }, { "epoch": 37.799196787148595, "grad_norm": 19.873573303222656, "learning_rate": 3.108387096774193e-05, "loss": 0.2771, "step": 5864 }, { "epoch": 37.80562248995984, "grad_norm": 5.404311656951904, "learning_rate": 3.1080645161290325e-05, "loss": 0.2869, "step": 5865 }, { "epoch": 37.81204819277109, "grad_norm": 4.462423801422119, "learning_rate": 3.107741935483871e-05, "loss": 0.1861, "step": 5866 }, { "epoch": 37.81847389558233, "grad_norm": 3.439892053604126, "learning_rate": 3.1074193548387096e-05, "loss": 0.2996, "step": 5867 }, { "epoch": 37.82489959839357, "grad_norm": 40.08491516113281, "learning_rate": 3.107096774193549e-05, "loss": 0.6398, "step": 5868 }, { "epoch": 37.83132530120482, "grad_norm": 5.806554317474365, "learning_rate": 3.1067741935483874e-05, "loss": 0.4487, "step": 5869 }, { "epoch": 37.83775100401606, "grad_norm": 10.081775665283203, "learning_rate": 3.106451612903226e-05, "loss": 0.3018, "step": 5870 }, { "epoch": 37.84417670682731, "grad_norm": 12.209001541137695, "learning_rate": 3.1061290322580646e-05, "loss": 0.4671, "step": 5871 }, { "epoch": 37.850602409638554, "grad_norm": 14.28848648071289, "learning_rate": 3.105806451612903e-05, "loss": 0.3593, "step": 5872 }, { "epoch": 37.8570281124498, "grad_norm": 14.327451705932617, "learning_rate": 3.105483870967742e-05, "loss": 0.3601, "step": 5873 }, { "epoch": 37.863453815261046, "grad_norm": 11.91434383392334, "learning_rate": 3.105161290322581e-05, "loss": 0.2773, "step": 5874 }, { "epoch": 37.86987951807229, "grad_norm": 24.224279403686523, "learning_rate": 3.1048387096774195e-05, "loss": 0.4723, "step": 5875 }, { "epoch": 37.87630522088354, "grad_norm": 7.242455005645752, "learning_rate": 3.104516129032258e-05, "loss": 0.2813, "step": 5876 }, { "epoch": 37.88273092369478, "grad_norm": 6.245669841766357, "learning_rate": 3.1041935483870974e-05, "loss": 0.2577, "step": 5877 }, { "epoch": 37.88915662650602, "grad_norm": 13.317036628723145, "learning_rate": 3.103870967741936e-05, "loss": 0.4293, "step": 5878 }, { "epoch": 37.89558232931727, "grad_norm": 6.445785999298096, "learning_rate": 3.103548387096774e-05, "loss": 0.1905, "step": 5879 }, { "epoch": 37.90200803212851, "grad_norm": 4.108989238739014, "learning_rate": 3.103225806451613e-05, "loss": 0.3032, "step": 5880 }, { "epoch": 37.908433734939756, "grad_norm": 14.529451370239258, "learning_rate": 3.1029032258064516e-05, "loss": 0.3119, "step": 5881 }, { "epoch": 37.914859437751005, "grad_norm": 7.043300151824951, "learning_rate": 3.10258064516129e-05, "loss": 0.1944, "step": 5882 }, { "epoch": 37.92128514056225, "grad_norm": 16.886375427246094, "learning_rate": 3.1022580645161295e-05, "loss": 0.5712, "step": 5883 }, { "epoch": 37.9277108433735, "grad_norm": 5.221289157867432, "learning_rate": 3.101935483870968e-05, "loss": 0.1501, "step": 5884 }, { "epoch": 37.93413654618474, "grad_norm": 15.144210815429688, "learning_rate": 3.1016129032258066e-05, "loss": 0.4718, "step": 5885 }, { "epoch": 37.94056224899598, "grad_norm": 7.692596435546875, "learning_rate": 3.101290322580646e-05, "loss": 0.4663, "step": 5886 }, { "epoch": 37.94698795180723, "grad_norm": 38.775856018066406, "learning_rate": 3.100967741935484e-05, "loss": 0.5303, "step": 5887 }, { "epoch": 37.95341365461847, "grad_norm": 4.686007022857666, "learning_rate": 3.100645161290322e-05, "loss": 0.102, "step": 5888 }, { "epoch": 37.95983935742972, "grad_norm": 8.135615348815918, "learning_rate": 3.1003225806451616e-05, "loss": 0.5472, "step": 5889 }, { "epoch": 37.966265060240964, "grad_norm": 15.751120567321777, "learning_rate": 3.1e-05, "loss": 0.2791, "step": 5890 }, { "epoch": 37.972690763052206, "grad_norm": 8.364289283752441, "learning_rate": 3.099677419354839e-05, "loss": 0.485, "step": 5891 }, { "epoch": 37.979116465863456, "grad_norm": 18.580533981323242, "learning_rate": 3.099354838709678e-05, "loss": 0.1701, "step": 5892 }, { "epoch": 37.9855421686747, "grad_norm": 10.539810180664062, "learning_rate": 3.0990322580645165e-05, "loss": 0.4719, "step": 5893 }, { "epoch": 37.99196787148595, "grad_norm": 10.230181694030762, "learning_rate": 3.098709677419355e-05, "loss": 0.4651, "step": 5894 }, { "epoch": 38.00160642570281, "grad_norm": 7.165057182312012, "learning_rate": 3.098387096774194e-05, "loss": 0.3958, "step": 5895 }, { "epoch": 38.00803212851405, "grad_norm": 6.539088249206543, "learning_rate": 3.098064516129032e-05, "loss": 0.3907, "step": 5896 }, { "epoch": 38.0144578313253, "grad_norm": 32.25239562988281, "learning_rate": 3.097741935483871e-05, "loss": 0.4206, "step": 5897 }, { "epoch": 38.020883534136544, "grad_norm": 8.500343322753906, "learning_rate": 3.09741935483871e-05, "loss": 0.5267, "step": 5898 }, { "epoch": 38.027309236947794, "grad_norm": 13.166576385498047, "learning_rate": 3.0970967741935486e-05, "loss": 0.4064, "step": 5899 }, { "epoch": 38.033734939759036, "grad_norm": 8.414752960205078, "learning_rate": 3.096774193548387e-05, "loss": 0.8163, "step": 5900 }, { "epoch": 38.04016064257028, "grad_norm": 10.002311706542969, "learning_rate": 3.0964516129032264e-05, "loss": 0.5494, "step": 5901 }, { "epoch": 38.04658634538153, "grad_norm": 27.17428970336914, "learning_rate": 3.096129032258064e-05, "loss": 0.5693, "step": 5902 }, { "epoch": 38.05301204819277, "grad_norm": 4.208230495452881, "learning_rate": 3.095806451612903e-05, "loss": 0.1663, "step": 5903 }, { "epoch": 38.05943775100402, "grad_norm": 6.257015228271484, "learning_rate": 3.095483870967742e-05, "loss": 0.4476, "step": 5904 }, { "epoch": 38.06586345381526, "grad_norm": 6.740920066833496, "learning_rate": 3.095161290322581e-05, "loss": 0.5657, "step": 5905 }, { "epoch": 38.0722891566265, "grad_norm": 7.769279479980469, "learning_rate": 3.094838709677419e-05, "loss": 0.3909, "step": 5906 }, { "epoch": 38.07871485943775, "grad_norm": 20.742603302001953, "learning_rate": 3.0945161290322585e-05, "loss": 0.4974, "step": 5907 }, { "epoch": 38.085140562248995, "grad_norm": 8.708683013916016, "learning_rate": 3.094193548387097e-05, "loss": 0.3605, "step": 5908 }, { "epoch": 38.091566265060244, "grad_norm": 8.55104923248291, "learning_rate": 3.093870967741936e-05, "loss": 0.3276, "step": 5909 }, { "epoch": 38.09799196787149, "grad_norm": 5.262679100036621, "learning_rate": 3.093548387096774e-05, "loss": 0.3647, "step": 5910 }, { "epoch": 38.10441767068273, "grad_norm": 5.701866149902344, "learning_rate": 3.093225806451613e-05, "loss": 0.1672, "step": 5911 }, { "epoch": 38.11084337349398, "grad_norm": 8.248499870300293, "learning_rate": 3.0929032258064514e-05, "loss": 0.5205, "step": 5912 }, { "epoch": 38.11726907630522, "grad_norm": 7.8667216300964355, "learning_rate": 3.0925806451612906e-05, "loss": 0.1907, "step": 5913 }, { "epoch": 38.12369477911646, "grad_norm": 2.9063191413879395, "learning_rate": 3.092258064516129e-05, "loss": 0.2639, "step": 5914 }, { "epoch": 38.13012048192771, "grad_norm": 34.79776382446289, "learning_rate": 3.091935483870968e-05, "loss": 0.3917, "step": 5915 }, { "epoch": 38.136546184738954, "grad_norm": 11.730731964111328, "learning_rate": 3.091612903225807e-05, "loss": 0.5526, "step": 5916 }, { "epoch": 38.1429718875502, "grad_norm": 20.715055465698242, "learning_rate": 3.0912903225806456e-05, "loss": 0.2172, "step": 5917 }, { "epoch": 38.149397590361446, "grad_norm": 5.624989986419678, "learning_rate": 3.0909677419354835e-05, "loss": 0.4719, "step": 5918 }, { "epoch": 38.15582329317269, "grad_norm": 2.856282949447632, "learning_rate": 3.090645161290323e-05, "loss": 0.1905, "step": 5919 }, { "epoch": 38.16224899598394, "grad_norm": 5.736635208129883, "learning_rate": 3.090322580645161e-05, "loss": 0.4758, "step": 5920 }, { "epoch": 38.16867469879518, "grad_norm": 7.8090057373046875, "learning_rate": 3.09e-05, "loss": 0.5089, "step": 5921 }, { "epoch": 38.17510040160643, "grad_norm": 13.241496086120605, "learning_rate": 3.089677419354839e-05, "loss": 0.6758, "step": 5922 }, { "epoch": 38.18152610441767, "grad_norm": 5.2994465827941895, "learning_rate": 3.089354838709678e-05, "loss": 0.2743, "step": 5923 }, { "epoch": 38.18795180722891, "grad_norm": 5.023465156555176, "learning_rate": 3.089032258064516e-05, "loss": 0.3261, "step": 5924 }, { "epoch": 38.19437751004016, "grad_norm": 9.117459297180176, "learning_rate": 3.0887096774193555e-05, "loss": 0.4851, "step": 5925 }, { "epoch": 38.200803212851405, "grad_norm": 5.0066375732421875, "learning_rate": 3.0883870967741934e-05, "loss": 0.332, "step": 5926 }, { "epoch": 38.207228915662654, "grad_norm": 5.4498796463012695, "learning_rate": 3.088064516129032e-05, "loss": 0.3571, "step": 5927 }, { "epoch": 38.213654618473896, "grad_norm": 7.338947772979736, "learning_rate": 3.087741935483871e-05, "loss": 0.5144, "step": 5928 }, { "epoch": 38.22008032128514, "grad_norm": 8.074202537536621, "learning_rate": 3.08741935483871e-05, "loss": 0.599, "step": 5929 }, { "epoch": 38.22650602409639, "grad_norm": 9.023102760314941, "learning_rate": 3.0870967741935484e-05, "loss": 0.363, "step": 5930 }, { "epoch": 38.23293172690763, "grad_norm": 10.829524993896484, "learning_rate": 3.0867741935483876e-05, "loss": 0.6571, "step": 5931 }, { "epoch": 38.23935742971887, "grad_norm": 7.657089710235596, "learning_rate": 3.086451612903226e-05, "loss": 0.2573, "step": 5932 }, { "epoch": 38.24578313253012, "grad_norm": 5.328897953033447, "learning_rate": 3.086129032258065e-05, "loss": 0.2979, "step": 5933 }, { "epoch": 38.252208835341364, "grad_norm": 6.814082622528076, "learning_rate": 3.0858064516129033e-05, "loss": 0.335, "step": 5934 }, { "epoch": 38.25863453815261, "grad_norm": 6.098175525665283, "learning_rate": 3.085483870967742e-05, "loss": 0.1237, "step": 5935 }, { "epoch": 38.265060240963855, "grad_norm": 30.122983932495117, "learning_rate": 3.0851612903225805e-05, "loss": 0.4537, "step": 5936 }, { "epoch": 38.2714859437751, "grad_norm": 8.92806339263916, "learning_rate": 3.08483870967742e-05, "loss": 0.4621, "step": 5937 }, { "epoch": 38.27791164658635, "grad_norm": 3.46572208404541, "learning_rate": 3.084516129032258e-05, "loss": 0.1488, "step": 5938 }, { "epoch": 38.28433734939759, "grad_norm": 15.393075942993164, "learning_rate": 3.084193548387097e-05, "loss": 0.2582, "step": 5939 }, { "epoch": 38.29076305220884, "grad_norm": 6.290627956390381, "learning_rate": 3.083870967741936e-05, "loss": 0.3043, "step": 5940 }, { "epoch": 38.29718875502008, "grad_norm": 9.991666793823242, "learning_rate": 3.083548387096774e-05, "loss": 0.3409, "step": 5941 }, { "epoch": 38.30361445783132, "grad_norm": 5.6222968101501465, "learning_rate": 3.0832258064516126e-05, "loss": 0.2692, "step": 5942 }, { "epoch": 38.31004016064257, "grad_norm": 12.820935249328613, "learning_rate": 3.082903225806452e-05, "loss": 0.4969, "step": 5943 }, { "epoch": 38.316465863453814, "grad_norm": 5.533440589904785, "learning_rate": 3.0825806451612904e-05, "loss": 0.3573, "step": 5944 }, { "epoch": 38.32289156626506, "grad_norm": 11.105622291564941, "learning_rate": 3.082258064516129e-05, "loss": 0.3339, "step": 5945 }, { "epoch": 38.329317269076306, "grad_norm": 22.717121124267578, "learning_rate": 3.081935483870968e-05, "loss": 0.4682, "step": 5946 }, { "epoch": 38.33574297188755, "grad_norm": 8.801589012145996, "learning_rate": 3.081612903225807e-05, "loss": 0.6847, "step": 5947 }, { "epoch": 38.3421686746988, "grad_norm": 2.6915624141693115, "learning_rate": 3.0812903225806454e-05, "loss": 0.3602, "step": 5948 }, { "epoch": 38.34859437751004, "grad_norm": 11.668885231018066, "learning_rate": 3.080967741935484e-05, "loss": 0.2498, "step": 5949 }, { "epoch": 38.35502008032128, "grad_norm": 9.35225772857666, "learning_rate": 3.0806451612903225e-05, "loss": 0.3138, "step": 5950 }, { "epoch": 38.36144578313253, "grad_norm": 5.648863315582275, "learning_rate": 3.080322580645161e-05, "loss": 0.3569, "step": 5951 }, { "epoch": 38.36787148594377, "grad_norm": 5.983272552490234, "learning_rate": 3.08e-05, "loss": 0.2384, "step": 5952 }, { "epoch": 38.37429718875502, "grad_norm": 4.617569923400879, "learning_rate": 3.079677419354839e-05, "loss": 0.07, "step": 5953 }, { "epoch": 38.380722891566265, "grad_norm": 11.145262718200684, "learning_rate": 3.0793548387096775e-05, "loss": 0.5386, "step": 5954 }, { "epoch": 38.38714859437751, "grad_norm": 15.22333812713623, "learning_rate": 3.079032258064517e-05, "loss": 0.3015, "step": 5955 }, { "epoch": 38.393574297188756, "grad_norm": 5.786472320556641, "learning_rate": 3.078709677419355e-05, "loss": 0.4016, "step": 5956 }, { "epoch": 38.4, "grad_norm": 7.878530502319336, "learning_rate": 3.078387096774193e-05, "loss": 0.4568, "step": 5957 }, { "epoch": 38.40642570281125, "grad_norm": 8.465771675109863, "learning_rate": 3.0780645161290324e-05, "loss": 0.3466, "step": 5958 }, { "epoch": 38.41285140562249, "grad_norm": 4.608964920043945, "learning_rate": 3.077741935483871e-05, "loss": 0.1928, "step": 5959 }, { "epoch": 38.41927710843373, "grad_norm": 8.95073127746582, "learning_rate": 3.0774193548387096e-05, "loss": 0.4995, "step": 5960 }, { "epoch": 38.42570281124498, "grad_norm": 12.233649253845215, "learning_rate": 3.077096774193549e-05, "loss": 0.3137, "step": 5961 }, { "epoch": 38.432128514056224, "grad_norm": 7.77998161315918, "learning_rate": 3.0767741935483874e-05, "loss": 0.4359, "step": 5962 }, { "epoch": 38.43855421686747, "grad_norm": 11.95057201385498, "learning_rate": 3.076451612903226e-05, "loss": 0.3696, "step": 5963 }, { "epoch": 38.444979919678715, "grad_norm": 7.7411298751831055, "learning_rate": 3.076129032258065e-05, "loss": 0.4365, "step": 5964 }, { "epoch": 38.45140562248996, "grad_norm": 4.851945400238037, "learning_rate": 3.075806451612903e-05, "loss": 0.2002, "step": 5965 }, { "epoch": 38.45783132530121, "grad_norm": 13.119889259338379, "learning_rate": 3.075483870967742e-05, "loss": 0.3056, "step": 5966 }, { "epoch": 38.46425702811245, "grad_norm": 12.966410636901855, "learning_rate": 3.075161290322581e-05, "loss": 0.6913, "step": 5967 }, { "epoch": 38.47068273092369, "grad_norm": 8.225021362304688, "learning_rate": 3.0748387096774195e-05, "loss": 0.3732, "step": 5968 }, { "epoch": 38.47710843373494, "grad_norm": 4.566901206970215, "learning_rate": 3.074516129032258e-05, "loss": 0.2806, "step": 5969 }, { "epoch": 38.48353413654618, "grad_norm": 8.929239273071289, "learning_rate": 3.074193548387097e-05, "loss": 0.4138, "step": 5970 }, { "epoch": 38.48995983935743, "grad_norm": 31.875154495239258, "learning_rate": 3.073870967741936e-05, "loss": 0.3428, "step": 5971 }, { "epoch": 38.496385542168674, "grad_norm": 4.525157928466797, "learning_rate": 3.0735483870967745e-05, "loss": 0.3878, "step": 5972 }, { "epoch": 38.50281124497992, "grad_norm": 17.142343521118164, "learning_rate": 3.073225806451613e-05, "loss": 0.6511, "step": 5973 }, { "epoch": 38.509236947791166, "grad_norm": 5.362333297729492, "learning_rate": 3.0729032258064516e-05, "loss": 0.3938, "step": 5974 }, { "epoch": 38.51566265060241, "grad_norm": 6.177488327026367, "learning_rate": 3.07258064516129e-05, "loss": 0.4916, "step": 5975 }, { "epoch": 38.52208835341366, "grad_norm": 5.305450916290283, "learning_rate": 3.0722580645161294e-05, "loss": 0.4056, "step": 5976 }, { "epoch": 38.5285140562249, "grad_norm": 9.385419845581055, "learning_rate": 3.071935483870968e-05, "loss": 0.2991, "step": 5977 }, { "epoch": 38.53493975903614, "grad_norm": 23.751615524291992, "learning_rate": 3.0716129032258066e-05, "loss": 0.1959, "step": 5978 }, { "epoch": 38.54136546184739, "grad_norm": 6.777273178100586, "learning_rate": 3.071290322580646e-05, "loss": 0.2475, "step": 5979 }, { "epoch": 38.54779116465863, "grad_norm": 9.021295547485352, "learning_rate": 3.070967741935484e-05, "loss": 0.2196, "step": 5980 }, { "epoch": 38.55421686746988, "grad_norm": 3.6615238189697266, "learning_rate": 3.070645161290322e-05, "loss": 0.1668, "step": 5981 }, { "epoch": 38.560642570281125, "grad_norm": 7.978438854217529, "learning_rate": 3.0703225806451615e-05, "loss": 0.7117, "step": 5982 }, { "epoch": 38.56706827309237, "grad_norm": 5.438230991363525, "learning_rate": 3.07e-05, "loss": 0.3166, "step": 5983 }, { "epoch": 38.57349397590362, "grad_norm": 20.87507438659668, "learning_rate": 3.0696774193548387e-05, "loss": 0.4034, "step": 5984 }, { "epoch": 38.57991967871486, "grad_norm": 3.8371503353118896, "learning_rate": 3.069354838709678e-05, "loss": 0.1612, "step": 5985 }, { "epoch": 38.5863453815261, "grad_norm": 6.821069717407227, "learning_rate": 3.0690322580645165e-05, "loss": 0.6453, "step": 5986 }, { "epoch": 38.59277108433735, "grad_norm": 6.733366966247559, "learning_rate": 3.068709677419355e-05, "loss": 0.3419, "step": 5987 }, { "epoch": 38.59919678714859, "grad_norm": 10.263066291809082, "learning_rate": 3.0683870967741936e-05, "loss": 0.433, "step": 5988 }, { "epoch": 38.60562248995984, "grad_norm": 6.387729644775391, "learning_rate": 3.068064516129032e-05, "loss": 0.595, "step": 5989 }, { "epoch": 38.612048192771084, "grad_norm": 8.1942777633667, "learning_rate": 3.067741935483871e-05, "loss": 0.4157, "step": 5990 }, { "epoch": 38.618473895582326, "grad_norm": 7.409229755401611, "learning_rate": 3.06741935483871e-05, "loss": 0.1694, "step": 5991 }, { "epoch": 38.624899598393576, "grad_norm": 17.81806182861328, "learning_rate": 3.0670967741935486e-05, "loss": 0.657, "step": 5992 }, { "epoch": 38.63132530120482, "grad_norm": 9.689438819885254, "learning_rate": 3.066774193548387e-05, "loss": 0.3048, "step": 5993 }, { "epoch": 38.63775100401607, "grad_norm": 8.236137390136719, "learning_rate": 3.0664516129032264e-05, "loss": 0.5141, "step": 5994 }, { "epoch": 38.64417670682731, "grad_norm": 10.036909103393555, "learning_rate": 3.066129032258065e-05, "loss": 0.4397, "step": 5995 }, { "epoch": 38.65060240963855, "grad_norm": 40.55611801147461, "learning_rate": 3.065806451612903e-05, "loss": 0.2028, "step": 5996 }, { "epoch": 38.6570281124498, "grad_norm": 8.331366539001465, "learning_rate": 3.065483870967742e-05, "loss": 0.446, "step": 5997 }, { "epoch": 38.66345381526104, "grad_norm": 8.453489303588867, "learning_rate": 3.065161290322581e-05, "loss": 0.7433, "step": 5998 }, { "epoch": 38.66987951807229, "grad_norm": 7.3203654289245605, "learning_rate": 3.064838709677419e-05, "loss": 0.4449, "step": 5999 }, { "epoch": 38.676305220883535, "grad_norm": 7.8431220054626465, "learning_rate": 3.0645161290322585e-05, "loss": 0.1987, "step": 6000 }, { "epoch": 38.68273092369478, "grad_norm": 36.71531677246094, "learning_rate": 3.064193548387097e-05, "loss": 0.4747, "step": 6001 }, { "epoch": 38.689156626506026, "grad_norm": 10.441764831542969, "learning_rate": 3.0638709677419356e-05, "loss": 0.4414, "step": 6002 }, { "epoch": 38.69558232931727, "grad_norm": 8.324170112609863, "learning_rate": 3.063548387096775e-05, "loss": 0.342, "step": 6003 }, { "epoch": 38.70200803212851, "grad_norm": 7.772348403930664, "learning_rate": 3.063225806451613e-05, "loss": 0.3827, "step": 6004 }, { "epoch": 38.70843373493976, "grad_norm": 20.33013343811035, "learning_rate": 3.0629032258064513e-05, "loss": 0.519, "step": 6005 }, { "epoch": 38.714859437751, "grad_norm": 1.75920569896698, "learning_rate": 3.0625806451612906e-05, "loss": 0.1791, "step": 6006 }, { "epoch": 38.72128514056225, "grad_norm": 3.4262499809265137, "learning_rate": 3.062258064516129e-05, "loss": 0.2149, "step": 6007 }, { "epoch": 38.727710843373494, "grad_norm": 8.931417465209961, "learning_rate": 3.061935483870968e-05, "loss": 0.4949, "step": 6008 }, { "epoch": 38.734136546184736, "grad_norm": 19.907312393188477, "learning_rate": 3.061612903225807e-05, "loss": 0.1228, "step": 6009 }, { "epoch": 38.740562248995985, "grad_norm": 7.011288166046143, "learning_rate": 3.0612903225806456e-05, "loss": 0.3333, "step": 6010 }, { "epoch": 38.74698795180723, "grad_norm": 4.11329460144043, "learning_rate": 3.0609677419354835e-05, "loss": 0.193, "step": 6011 }, { "epoch": 38.75341365461848, "grad_norm": 5.579688549041748, "learning_rate": 3.060645161290323e-05, "loss": 0.1826, "step": 6012 }, { "epoch": 38.75983935742972, "grad_norm": 9.381750106811523, "learning_rate": 3.060322580645161e-05, "loss": 0.2674, "step": 6013 }, { "epoch": 38.76626506024096, "grad_norm": 6.117944717407227, "learning_rate": 3.06e-05, "loss": 0.316, "step": 6014 }, { "epoch": 38.77269076305221, "grad_norm": 8.318022727966309, "learning_rate": 3.059677419354839e-05, "loss": 0.3244, "step": 6015 }, { "epoch": 38.77911646586345, "grad_norm": 12.416167259216309, "learning_rate": 3.0593548387096777e-05, "loss": 0.3094, "step": 6016 }, { "epoch": 38.7855421686747, "grad_norm": 4.92892599105835, "learning_rate": 3.059032258064516e-05, "loss": 0.4695, "step": 6017 }, { "epoch": 38.791967871485944, "grad_norm": 13.840306282043457, "learning_rate": 3.0587096774193555e-05, "loss": 0.4197, "step": 6018 }, { "epoch": 38.79839357429719, "grad_norm": 5.367740631103516, "learning_rate": 3.0583870967741934e-05, "loss": 0.3411, "step": 6019 }, { "epoch": 38.804819277108436, "grad_norm": 15.250980377197266, "learning_rate": 3.058064516129032e-05, "loss": 0.4311, "step": 6020 }, { "epoch": 38.81124497991968, "grad_norm": 6.996432304382324, "learning_rate": 3.057741935483871e-05, "loss": 0.416, "step": 6021 }, { "epoch": 38.81767068273092, "grad_norm": 6.88331937789917, "learning_rate": 3.05741935483871e-05, "loss": 0.3045, "step": 6022 }, { "epoch": 38.82409638554217, "grad_norm": 8.824649810791016, "learning_rate": 3.057096774193548e-05, "loss": 0.3629, "step": 6023 }, { "epoch": 38.83052208835341, "grad_norm": 19.154268264770508, "learning_rate": 3.0567741935483876e-05, "loss": 0.4482, "step": 6024 }, { "epoch": 38.83694779116466, "grad_norm": 3.0454936027526855, "learning_rate": 3.056451612903226e-05, "loss": 0.1892, "step": 6025 }, { "epoch": 38.8433734939759, "grad_norm": 12.867469787597656, "learning_rate": 3.056129032258065e-05, "loss": 0.3474, "step": 6026 }, { "epoch": 38.849799196787146, "grad_norm": 9.004088401794434, "learning_rate": 3.055806451612903e-05, "loss": 0.379, "step": 6027 }, { "epoch": 38.856224899598395, "grad_norm": 10.409926414489746, "learning_rate": 3.055483870967742e-05, "loss": 0.2355, "step": 6028 }, { "epoch": 38.86265060240964, "grad_norm": 12.1473970413208, "learning_rate": 3.0551612903225804e-05, "loss": 0.3361, "step": 6029 }, { "epoch": 38.86907630522089, "grad_norm": 6.9965434074401855, "learning_rate": 3.05483870967742e-05, "loss": 0.6544, "step": 6030 }, { "epoch": 38.87550200803213, "grad_norm": 25.81473159790039, "learning_rate": 3.054516129032258e-05, "loss": 0.1597, "step": 6031 }, { "epoch": 38.88192771084337, "grad_norm": 47.36817932128906, "learning_rate": 3.054193548387097e-05, "loss": 0.8284, "step": 6032 }, { "epoch": 38.88835341365462, "grad_norm": 24.231552124023438, "learning_rate": 3.053870967741936e-05, "loss": 0.4392, "step": 6033 }, { "epoch": 38.89477911646586, "grad_norm": 27.292850494384766, "learning_rate": 3.0535483870967746e-05, "loss": 0.7931, "step": 6034 }, { "epoch": 38.90120481927711, "grad_norm": 9.90556812286377, "learning_rate": 3.0532258064516125e-05, "loss": 0.324, "step": 6035 }, { "epoch": 38.907630522088354, "grad_norm": 10.867264747619629, "learning_rate": 3.052903225806452e-05, "loss": 0.4362, "step": 6036 }, { "epoch": 38.914056224899596, "grad_norm": 8.526222229003906, "learning_rate": 3.0525806451612904e-05, "loss": 0.561, "step": 6037 }, { "epoch": 38.920481927710846, "grad_norm": 10.561077117919922, "learning_rate": 3.052258064516129e-05, "loss": 0.4042, "step": 6038 }, { "epoch": 38.92690763052209, "grad_norm": 8.768333435058594, "learning_rate": 3.051935483870968e-05, "loss": 0.7804, "step": 6039 }, { "epoch": 38.93333333333333, "grad_norm": 7.357405662536621, "learning_rate": 3.0516129032258067e-05, "loss": 0.4261, "step": 6040 }, { "epoch": 38.93975903614458, "grad_norm": 11.28842830657959, "learning_rate": 3.0512903225806457e-05, "loss": 0.3438, "step": 6041 }, { "epoch": 38.94618473895582, "grad_norm": 14.773565292358398, "learning_rate": 3.0509677419354842e-05, "loss": 0.8307, "step": 6042 }, { "epoch": 38.95261044176707, "grad_norm": 9.064266204833984, "learning_rate": 3.0506451612903225e-05, "loss": 0.5, "step": 6043 }, { "epoch": 38.95903614457831, "grad_norm": 6.41776180267334, "learning_rate": 3.0503225806451614e-05, "loss": 0.2534, "step": 6044 }, { "epoch": 38.965461847389555, "grad_norm": 16.090579986572266, "learning_rate": 3.05e-05, "loss": 0.3246, "step": 6045 }, { "epoch": 38.971887550200805, "grad_norm": 20.246036529541016, "learning_rate": 3.049677419354839e-05, "loss": 0.2759, "step": 6046 }, { "epoch": 38.97831325301205, "grad_norm": 7.846811294555664, "learning_rate": 3.0493548387096778e-05, "loss": 0.2726, "step": 6047 }, { "epoch": 38.984738955823296, "grad_norm": 13.555459976196289, "learning_rate": 3.0490322580645163e-05, "loss": 0.4964, "step": 6048 }, { "epoch": 38.99116465863454, "grad_norm": 11.057339668273926, "learning_rate": 3.0487096774193552e-05, "loss": 0.3097, "step": 6049 }, { "epoch": 39.00080321285141, "grad_norm": 33.917728424072266, "learning_rate": 3.0483870967741935e-05, "loss": 0.581, "step": 6050 }, { "epoch": 39.00722891566265, "grad_norm": 13.876530647277832, "learning_rate": 3.0480645161290324e-05, "loss": 0.4212, "step": 6051 }, { "epoch": 39.01365461847389, "grad_norm": 10.79062557220459, "learning_rate": 3.047741935483871e-05, "loss": 0.3187, "step": 6052 }, { "epoch": 39.02008032128514, "grad_norm": 6.798964023590088, "learning_rate": 3.04741935483871e-05, "loss": 0.2513, "step": 6053 }, { "epoch": 39.026506024096385, "grad_norm": 7.7326836585998535, "learning_rate": 3.0470967741935484e-05, "loss": 0.3052, "step": 6054 }, { "epoch": 39.03293172690763, "grad_norm": 11.515254020690918, "learning_rate": 3.0467741935483873e-05, "loss": 0.668, "step": 6055 }, { "epoch": 39.039357429718876, "grad_norm": 6.081240177154541, "learning_rate": 3.0464516129032262e-05, "loss": 0.3775, "step": 6056 }, { "epoch": 39.04578313253012, "grad_norm": 11.391200065612793, "learning_rate": 3.0461290322580648e-05, "loss": 0.3609, "step": 6057 }, { "epoch": 39.05220883534137, "grad_norm": 13.058281898498535, "learning_rate": 3.045806451612903e-05, "loss": 0.6265, "step": 6058 }, { "epoch": 39.05863453815261, "grad_norm": 10.587712287902832, "learning_rate": 3.045483870967742e-05, "loss": 0.457, "step": 6059 }, { "epoch": 39.06506024096385, "grad_norm": 8.399224281311035, "learning_rate": 3.0451612903225805e-05, "loss": 0.6035, "step": 6060 }, { "epoch": 39.0714859437751, "grad_norm": 4.91502571105957, "learning_rate": 3.0448387096774194e-05, "loss": 0.2351, "step": 6061 }, { "epoch": 39.077911646586344, "grad_norm": 10.726064682006836, "learning_rate": 3.0445161290322583e-05, "loss": 0.3618, "step": 6062 }, { "epoch": 39.08433734939759, "grad_norm": 11.127147674560547, "learning_rate": 3.044193548387097e-05, "loss": 0.2971, "step": 6063 }, { "epoch": 39.090763052208835, "grad_norm": 5.913999080657959, "learning_rate": 3.0438709677419358e-05, "loss": 0.3869, "step": 6064 }, { "epoch": 39.09718875502008, "grad_norm": 9.896709442138672, "learning_rate": 3.0435483870967747e-05, "loss": 0.6919, "step": 6065 }, { "epoch": 39.10361445783133, "grad_norm": 16.652156829833984, "learning_rate": 3.043225806451613e-05, "loss": 0.68, "step": 6066 }, { "epoch": 39.11004016064257, "grad_norm": 6.053792476654053, "learning_rate": 3.0429032258064515e-05, "loss": 0.2019, "step": 6067 }, { "epoch": 39.11646586345382, "grad_norm": 4.514317512512207, "learning_rate": 3.0425806451612904e-05, "loss": 0.3948, "step": 6068 }, { "epoch": 39.12289156626506, "grad_norm": 8.868717193603516, "learning_rate": 3.042258064516129e-05, "loss": 0.2082, "step": 6069 }, { "epoch": 39.1293172690763, "grad_norm": 8.467733383178711, "learning_rate": 3.041935483870968e-05, "loss": 0.6172, "step": 6070 }, { "epoch": 39.13574297188755, "grad_norm": 4.261107921600342, "learning_rate": 3.041612903225807e-05, "loss": 0.1841, "step": 6071 }, { "epoch": 39.142168674698794, "grad_norm": 7.55536413192749, "learning_rate": 3.0412903225806454e-05, "loss": 0.4239, "step": 6072 }, { "epoch": 39.14859437751004, "grad_norm": 7.639307975769043, "learning_rate": 3.0409677419354843e-05, "loss": 0.3586, "step": 6073 }, { "epoch": 39.155020080321286, "grad_norm": 5.348148822784424, "learning_rate": 3.0406451612903226e-05, "loss": 0.316, "step": 6074 }, { "epoch": 39.16144578313253, "grad_norm": 12.686758995056152, "learning_rate": 3.040322580645161e-05, "loss": 0.3002, "step": 6075 }, { "epoch": 39.16787148594378, "grad_norm": 3.6061036586761475, "learning_rate": 3.04e-05, "loss": 0.1905, "step": 6076 }, { "epoch": 39.17429718875502, "grad_norm": 5.347690105438232, "learning_rate": 3.039677419354839e-05, "loss": 0.4595, "step": 6077 }, { "epoch": 39.18072289156626, "grad_norm": 6.2441325187683105, "learning_rate": 3.0393548387096775e-05, "loss": 0.2136, "step": 6078 }, { "epoch": 39.18714859437751, "grad_norm": 7.431967735290527, "learning_rate": 3.0390322580645164e-05, "loss": 0.3344, "step": 6079 }, { "epoch": 39.19357429718875, "grad_norm": 11.056731224060059, "learning_rate": 3.0387096774193553e-05, "loss": 0.4924, "step": 6080 }, { "epoch": 39.2, "grad_norm": 7.162702560424805, "learning_rate": 3.0383870967741936e-05, "loss": 0.1716, "step": 6081 }, { "epoch": 39.206425702811245, "grad_norm": 6.790876388549805, "learning_rate": 3.038064516129032e-05, "loss": 0.3776, "step": 6082 }, { "epoch": 39.21285140562249, "grad_norm": 9.399481773376465, "learning_rate": 3.037741935483871e-05, "loss": 0.2764, "step": 6083 }, { "epoch": 39.21927710843374, "grad_norm": 8.350303649902344, "learning_rate": 3.0374193548387096e-05, "loss": 0.283, "step": 6084 }, { "epoch": 39.22570281124498, "grad_norm": 7.0762763023376465, "learning_rate": 3.0370967741935485e-05, "loss": 0.3239, "step": 6085 }, { "epoch": 39.23212851405623, "grad_norm": 6.275753021240234, "learning_rate": 3.0367741935483874e-05, "loss": 0.3109, "step": 6086 }, { "epoch": 39.23855421686747, "grad_norm": 7.432380199432373, "learning_rate": 3.036451612903226e-05, "loss": 0.4953, "step": 6087 }, { "epoch": 39.24497991967871, "grad_norm": 11.857890129089355, "learning_rate": 3.036129032258065e-05, "loss": 0.4306, "step": 6088 }, { "epoch": 39.25140562248996, "grad_norm": 11.780378341674805, "learning_rate": 3.035806451612903e-05, "loss": 0.4632, "step": 6089 }, { "epoch": 39.257831325301204, "grad_norm": 14.75621223449707, "learning_rate": 3.035483870967742e-05, "loss": 0.4544, "step": 6090 }, { "epoch": 39.264257028112446, "grad_norm": 11.045230865478516, "learning_rate": 3.0351612903225806e-05, "loss": 0.3254, "step": 6091 }, { "epoch": 39.270682730923696, "grad_norm": 4.590087413787842, "learning_rate": 3.0348387096774195e-05, "loss": 0.1291, "step": 6092 }, { "epoch": 39.27710843373494, "grad_norm": 7.897561550140381, "learning_rate": 3.034516129032258e-05, "loss": 0.3084, "step": 6093 }, { "epoch": 39.28353413654619, "grad_norm": 12.774271011352539, "learning_rate": 3.034193548387097e-05, "loss": 0.5007, "step": 6094 }, { "epoch": 39.28995983935743, "grad_norm": 9.120759010314941, "learning_rate": 3.033870967741936e-05, "loss": 0.3368, "step": 6095 }, { "epoch": 39.29638554216867, "grad_norm": 6.34688138961792, "learning_rate": 3.0335483870967745e-05, "loss": 0.2538, "step": 6096 }, { "epoch": 39.30281124497992, "grad_norm": 7.8273115158081055, "learning_rate": 3.0332258064516127e-05, "loss": 0.3051, "step": 6097 }, { "epoch": 39.30923694779116, "grad_norm": 10.977193832397461, "learning_rate": 3.0329032258064516e-05, "loss": 0.2741, "step": 6098 }, { "epoch": 39.31566265060241, "grad_norm": 6.785089492797852, "learning_rate": 3.0325806451612902e-05, "loss": 0.3339, "step": 6099 }, { "epoch": 39.322088353413655, "grad_norm": 7.351836204528809, "learning_rate": 3.032258064516129e-05, "loss": 0.1682, "step": 6100 }, { "epoch": 39.3285140562249, "grad_norm": 1.9578649997711182, "learning_rate": 3.031935483870968e-05, "loss": 0.3389, "step": 6101 }, { "epoch": 39.334939759036146, "grad_norm": 11.664630889892578, "learning_rate": 3.0316129032258066e-05, "loss": 0.3697, "step": 6102 }, { "epoch": 39.34136546184739, "grad_norm": 24.740617752075195, "learning_rate": 3.0312903225806455e-05, "loss": 0.6421, "step": 6103 }, { "epoch": 39.34779116465864, "grad_norm": 8.288354873657227, "learning_rate": 3.0309677419354844e-05, "loss": 0.29, "step": 6104 }, { "epoch": 39.35421686746988, "grad_norm": 8.916730880737305, "learning_rate": 3.0306451612903226e-05, "loss": 0.2546, "step": 6105 }, { "epoch": 39.36064257028112, "grad_norm": 9.186280250549316, "learning_rate": 3.0303225806451612e-05, "loss": 0.3023, "step": 6106 }, { "epoch": 39.36706827309237, "grad_norm": 7.919623374938965, "learning_rate": 3.03e-05, "loss": 0.537, "step": 6107 }, { "epoch": 39.373493975903614, "grad_norm": 13.130017280578613, "learning_rate": 3.0296774193548387e-05, "loss": 0.3702, "step": 6108 }, { "epoch": 39.379919678714856, "grad_norm": 6.732974052429199, "learning_rate": 3.0293548387096776e-05, "loss": 0.2993, "step": 6109 }, { "epoch": 39.386345381526105, "grad_norm": 13.253392219543457, "learning_rate": 3.0290322580645165e-05, "loss": 0.5838, "step": 6110 }, { "epoch": 39.39277108433735, "grad_norm": 11.33309268951416, "learning_rate": 3.028709677419355e-05, "loss": 0.6461, "step": 6111 }, { "epoch": 39.3991967871486, "grad_norm": 16.52901840209961, "learning_rate": 3.028387096774194e-05, "loss": 0.3199, "step": 6112 }, { "epoch": 39.40562248995984, "grad_norm": 6.578482151031494, "learning_rate": 3.0280645161290322e-05, "loss": 0.4275, "step": 6113 }, { "epoch": 39.41204819277108, "grad_norm": 8.36320972442627, "learning_rate": 3.027741935483871e-05, "loss": 0.268, "step": 6114 }, { "epoch": 39.41847389558233, "grad_norm": 12.021721839904785, "learning_rate": 3.0274193548387097e-05, "loss": 0.5533, "step": 6115 }, { "epoch": 39.42489959839357, "grad_norm": 16.47233009338379, "learning_rate": 3.0270967741935486e-05, "loss": 0.5364, "step": 6116 }, { "epoch": 39.43132530120482, "grad_norm": 2.6744885444641113, "learning_rate": 3.0267741935483872e-05, "loss": 0.1235, "step": 6117 }, { "epoch": 39.437751004016064, "grad_norm": 8.109593391418457, "learning_rate": 3.026451612903226e-05, "loss": 0.3156, "step": 6118 }, { "epoch": 39.44417670682731, "grad_norm": 4.069103717803955, "learning_rate": 3.026129032258065e-05, "loss": 0.2115, "step": 6119 }, { "epoch": 39.450602409638556, "grad_norm": 12.88331127166748, "learning_rate": 3.0258064516129032e-05, "loss": 0.4875, "step": 6120 }, { "epoch": 39.4570281124498, "grad_norm": 5.571795463562012, "learning_rate": 3.0254838709677418e-05, "loss": 0.241, "step": 6121 }, { "epoch": 39.46345381526105, "grad_norm": 10.224008560180664, "learning_rate": 3.0251612903225807e-05, "loss": 0.3117, "step": 6122 }, { "epoch": 39.46987951807229, "grad_norm": 6.231414794921875, "learning_rate": 3.0248387096774193e-05, "loss": 0.4389, "step": 6123 }, { "epoch": 39.47630522088353, "grad_norm": 11.460674285888672, "learning_rate": 3.0245161290322582e-05, "loss": 0.2393, "step": 6124 }, { "epoch": 39.48273092369478, "grad_norm": 12.937274932861328, "learning_rate": 3.024193548387097e-05, "loss": 0.2799, "step": 6125 }, { "epoch": 39.48915662650602, "grad_norm": 10.973774909973145, "learning_rate": 3.0238709677419357e-05, "loss": 0.4434, "step": 6126 }, { "epoch": 39.495582329317266, "grad_norm": 2.419341802597046, "learning_rate": 3.0235483870967746e-05, "loss": 0.1975, "step": 6127 }, { "epoch": 39.502008032128515, "grad_norm": 7.902740001678467, "learning_rate": 3.0232258064516128e-05, "loss": 0.3904, "step": 6128 }, { "epoch": 39.50843373493976, "grad_norm": 10.058815956115723, "learning_rate": 3.0229032258064517e-05, "loss": 0.6757, "step": 6129 }, { "epoch": 39.51485943775101, "grad_norm": 12.601351737976074, "learning_rate": 3.0225806451612903e-05, "loss": 0.7539, "step": 6130 }, { "epoch": 39.52128514056225, "grad_norm": 8.253425598144531, "learning_rate": 3.0222580645161292e-05, "loss": 0.3828, "step": 6131 }, { "epoch": 39.52771084337349, "grad_norm": 4.454915523529053, "learning_rate": 3.0219354838709678e-05, "loss": 0.2525, "step": 6132 }, { "epoch": 39.53413654618474, "grad_norm": 5.5158610343933105, "learning_rate": 3.0216129032258067e-05, "loss": 0.3917, "step": 6133 }, { "epoch": 39.54056224899598, "grad_norm": 10.649209022521973, "learning_rate": 3.0212903225806456e-05, "loss": 0.5176, "step": 6134 }, { "epoch": 39.54698795180723, "grad_norm": 3.7877163887023926, "learning_rate": 3.0209677419354842e-05, "loss": 0.1558, "step": 6135 }, { "epoch": 39.553413654618474, "grad_norm": 19.511322021484375, "learning_rate": 3.0206451612903224e-05, "loss": 0.2641, "step": 6136 }, { "epoch": 39.559839357429716, "grad_norm": 9.400845527648926, "learning_rate": 3.0203225806451613e-05, "loss": 0.3207, "step": 6137 }, { "epoch": 39.566265060240966, "grad_norm": 8.740376472473145, "learning_rate": 3.02e-05, "loss": 0.4364, "step": 6138 }, { "epoch": 39.57269076305221, "grad_norm": 10.781871795654297, "learning_rate": 3.0196774193548388e-05, "loss": 0.3647, "step": 6139 }, { "epoch": 39.57911646586346, "grad_norm": 10.401551246643066, "learning_rate": 3.0193548387096777e-05, "loss": 0.3356, "step": 6140 }, { "epoch": 39.5855421686747, "grad_norm": 7.88689661026001, "learning_rate": 3.0190322580645163e-05, "loss": 0.3783, "step": 6141 }, { "epoch": 39.59196787148594, "grad_norm": 11.246660232543945, "learning_rate": 3.0187096774193552e-05, "loss": 0.3141, "step": 6142 }, { "epoch": 39.59839357429719, "grad_norm": 7.904202938079834, "learning_rate": 3.018387096774194e-05, "loss": 0.3587, "step": 6143 }, { "epoch": 39.60481927710843, "grad_norm": 6.08286714553833, "learning_rate": 3.0180645161290323e-05, "loss": 0.2831, "step": 6144 }, { "epoch": 39.611244979919675, "grad_norm": 10.059313774108887, "learning_rate": 3.017741935483871e-05, "loss": 0.3896, "step": 6145 }, { "epoch": 39.617670682730925, "grad_norm": 10.92316722869873, "learning_rate": 3.0174193548387098e-05, "loss": 0.2738, "step": 6146 }, { "epoch": 39.62409638554217, "grad_norm": 14.938305854797363, "learning_rate": 3.0170967741935484e-05, "loss": 0.3135, "step": 6147 }, { "epoch": 39.630522088353416, "grad_norm": 4.147114276885986, "learning_rate": 3.0167741935483873e-05, "loss": 0.278, "step": 6148 }, { "epoch": 39.63694779116466, "grad_norm": 4.004580020904541, "learning_rate": 3.0164516129032262e-05, "loss": 0.267, "step": 6149 }, { "epoch": 39.6433734939759, "grad_norm": 5.345688819885254, "learning_rate": 3.0161290322580648e-05, "loss": 0.2059, "step": 6150 }, { "epoch": 39.64979919678715, "grad_norm": 11.313531875610352, "learning_rate": 3.015806451612903e-05, "loss": 0.3557, "step": 6151 }, { "epoch": 39.65622489959839, "grad_norm": 10.535335540771484, "learning_rate": 3.015483870967742e-05, "loss": 0.3686, "step": 6152 }, { "epoch": 39.66265060240964, "grad_norm": 9.011679649353027, "learning_rate": 3.0151612903225808e-05, "loss": 0.3519, "step": 6153 }, { "epoch": 39.669076305220884, "grad_norm": 13.829543113708496, "learning_rate": 3.0148387096774194e-05, "loss": 0.2754, "step": 6154 }, { "epoch": 39.675502008032126, "grad_norm": 7.738396167755127, "learning_rate": 3.0145161290322583e-05, "loss": 0.2869, "step": 6155 }, { "epoch": 39.681927710843375, "grad_norm": 6.867644309997559, "learning_rate": 3.014193548387097e-05, "loss": 0.1715, "step": 6156 }, { "epoch": 39.68835341365462, "grad_norm": 11.305328369140625, "learning_rate": 3.0138709677419358e-05, "loss": 0.4199, "step": 6157 }, { "epoch": 39.69477911646587, "grad_norm": 4.753394603729248, "learning_rate": 3.0135483870967747e-05, "loss": 0.3124, "step": 6158 }, { "epoch": 39.70120481927711, "grad_norm": 9.194375038146973, "learning_rate": 3.013225806451613e-05, "loss": 0.3892, "step": 6159 }, { "epoch": 39.70763052208835, "grad_norm": 6.598696708679199, "learning_rate": 3.0129032258064515e-05, "loss": 0.2185, "step": 6160 }, { "epoch": 39.7140562248996, "grad_norm": 3.3363215923309326, "learning_rate": 3.0125806451612904e-05, "loss": 0.1532, "step": 6161 }, { "epoch": 39.72048192771084, "grad_norm": 22.706880569458008, "learning_rate": 3.012258064516129e-05, "loss": 0.325, "step": 6162 }, { "epoch": 39.726907630522085, "grad_norm": 9.354252815246582, "learning_rate": 3.011935483870968e-05, "loss": 0.2775, "step": 6163 }, { "epoch": 39.733333333333334, "grad_norm": 7.122140407562256, "learning_rate": 3.0116129032258068e-05, "loss": 0.3207, "step": 6164 }, { "epoch": 39.73975903614458, "grad_norm": 7.257811069488525, "learning_rate": 3.0112903225806454e-05, "loss": 0.2578, "step": 6165 }, { "epoch": 39.746184738955826, "grad_norm": 19.124794006347656, "learning_rate": 3.0109677419354843e-05, "loss": 0.7878, "step": 6166 }, { "epoch": 39.75261044176707, "grad_norm": 10.387858390808105, "learning_rate": 3.0106451612903225e-05, "loss": 0.4372, "step": 6167 }, { "epoch": 39.75903614457831, "grad_norm": 9.952337265014648, "learning_rate": 3.0103225806451614e-05, "loss": 0.3327, "step": 6168 }, { "epoch": 39.76546184738956, "grad_norm": 9.576473236083984, "learning_rate": 3.01e-05, "loss": 0.3094, "step": 6169 }, { "epoch": 39.7718875502008, "grad_norm": 9.117764472961426, "learning_rate": 3.009677419354839e-05, "loss": 0.3021, "step": 6170 }, { "epoch": 39.77831325301205, "grad_norm": 31.763853073120117, "learning_rate": 3.0093548387096775e-05, "loss": 0.4629, "step": 6171 }, { "epoch": 39.78473895582329, "grad_norm": 5.19777250289917, "learning_rate": 3.0090322580645164e-05, "loss": 0.2777, "step": 6172 }, { "epoch": 39.791164658634536, "grad_norm": 19.59998321533203, "learning_rate": 3.0087096774193553e-05, "loss": 0.5446, "step": 6173 }, { "epoch": 39.797590361445785, "grad_norm": 8.651665687561035, "learning_rate": 3.008387096774194e-05, "loss": 0.4775, "step": 6174 }, { "epoch": 39.80401606425703, "grad_norm": 13.553994178771973, "learning_rate": 3.008064516129032e-05, "loss": 0.3569, "step": 6175 }, { "epoch": 39.810441767068276, "grad_norm": 6.027576446533203, "learning_rate": 3.007741935483871e-05, "loss": 0.3258, "step": 6176 }, { "epoch": 39.81686746987952, "grad_norm": 13.823768615722656, "learning_rate": 3.00741935483871e-05, "loss": 0.1186, "step": 6177 }, { "epoch": 39.82329317269076, "grad_norm": 16.861473083496094, "learning_rate": 3.0070967741935485e-05, "loss": 0.3221, "step": 6178 }, { "epoch": 39.82971887550201, "grad_norm": 9.246525764465332, "learning_rate": 3.0067741935483874e-05, "loss": 0.4084, "step": 6179 }, { "epoch": 39.83614457831325, "grad_norm": 4.405760765075684, "learning_rate": 3.006451612903226e-05, "loss": 0.1507, "step": 6180 }, { "epoch": 39.842570281124495, "grad_norm": 5.635703086853027, "learning_rate": 3.006129032258065e-05, "loss": 0.1315, "step": 6181 }, { "epoch": 39.848995983935744, "grad_norm": 8.94025993347168, "learning_rate": 3.0058064516129038e-05, "loss": 0.5043, "step": 6182 }, { "epoch": 39.855421686746986, "grad_norm": 7.676534652709961, "learning_rate": 3.005483870967742e-05, "loss": 0.3587, "step": 6183 }, { "epoch": 39.861847389558235, "grad_norm": 8.111066818237305, "learning_rate": 3.0051612903225806e-05, "loss": 0.26, "step": 6184 }, { "epoch": 39.86827309236948, "grad_norm": 12.999688148498535, "learning_rate": 3.0048387096774195e-05, "loss": 0.3756, "step": 6185 }, { "epoch": 39.87469879518072, "grad_norm": 17.190013885498047, "learning_rate": 3.004516129032258e-05, "loss": 0.7044, "step": 6186 }, { "epoch": 39.88112449799197, "grad_norm": 7.743106842041016, "learning_rate": 3.004193548387097e-05, "loss": 0.2958, "step": 6187 }, { "epoch": 39.88755020080321, "grad_norm": 9.227507591247559, "learning_rate": 3.003870967741936e-05, "loss": 0.3601, "step": 6188 }, { "epoch": 39.89397590361446, "grad_norm": 9.78299617767334, "learning_rate": 3.0035483870967744e-05, "loss": 0.4776, "step": 6189 }, { "epoch": 39.9004016064257, "grad_norm": 9.875475883483887, "learning_rate": 3.0032258064516127e-05, "loss": 0.3798, "step": 6190 }, { "epoch": 39.906827309236945, "grad_norm": 7.845644474029541, "learning_rate": 3.0029032258064516e-05, "loss": 0.3086, "step": 6191 }, { "epoch": 39.913253012048195, "grad_norm": 8.83162784576416, "learning_rate": 3.0025806451612905e-05, "loss": 0.3016, "step": 6192 }, { "epoch": 39.91967871485944, "grad_norm": 8.769092559814453, "learning_rate": 3.002258064516129e-05, "loss": 0.382, "step": 6193 }, { "epoch": 39.926104417670686, "grad_norm": 25.790464401245117, "learning_rate": 3.001935483870968e-05, "loss": 0.3192, "step": 6194 }, { "epoch": 39.93253012048193, "grad_norm": 11.367005348205566, "learning_rate": 3.0016129032258065e-05, "loss": 0.5316, "step": 6195 }, { "epoch": 39.93895582329317, "grad_norm": 12.066268920898438, "learning_rate": 3.0012903225806455e-05, "loss": 0.5832, "step": 6196 }, { "epoch": 39.94538152610442, "grad_norm": 8.02226448059082, "learning_rate": 3.0009677419354844e-05, "loss": 0.4204, "step": 6197 }, { "epoch": 39.95180722891566, "grad_norm": 7.105018615722656, "learning_rate": 3.0006451612903226e-05, "loss": 0.2565, "step": 6198 }, { "epoch": 39.958232931726904, "grad_norm": 6.651101589202881, "learning_rate": 3.000322580645161e-05, "loss": 0.3038, "step": 6199 }, { "epoch": 39.964658634538154, "grad_norm": 12.386326789855957, "learning_rate": 3e-05, "loss": 0.3884, "step": 6200 }, { "epoch": 39.971084337349396, "grad_norm": 5.556015491485596, "learning_rate": 2.9996774193548386e-05, "loss": 0.4125, "step": 6201 }, { "epoch": 39.977510040160645, "grad_norm": 18.035593032836914, "learning_rate": 2.9993548387096776e-05, "loss": 0.7221, "step": 6202 }, { "epoch": 39.98393574297189, "grad_norm": 5.499904632568359, "learning_rate": 2.9990322580645165e-05, "loss": 0.351, "step": 6203 }, { "epoch": 39.99036144578313, "grad_norm": 5.670801639556885, "learning_rate": 2.998709677419355e-05, "loss": 0.2567, "step": 6204 }, { "epoch": 39.99678714859438, "grad_norm": 6.987402439117432, "learning_rate": 2.998387096774194e-05, "loss": 0.376, "step": 6205 }, { "epoch": 40.00642570281124, "grad_norm": 5.582454204559326, "learning_rate": 2.9980645161290322e-05, "loss": 0.193, "step": 6206 }, { "epoch": 40.01285140562249, "grad_norm": 9.42165470123291, "learning_rate": 2.997741935483871e-05, "loss": 0.5689, "step": 6207 }, { "epoch": 40.019277108433734, "grad_norm": 10.062102317810059, "learning_rate": 2.9974193548387097e-05, "loss": 0.5846, "step": 6208 }, { "epoch": 40.02570281124498, "grad_norm": 17.2452392578125, "learning_rate": 2.9970967741935486e-05, "loss": 0.9049, "step": 6209 }, { "epoch": 40.032128514056225, "grad_norm": 11.584144592285156, "learning_rate": 2.996774193548387e-05, "loss": 0.2614, "step": 6210 }, { "epoch": 40.03855421686747, "grad_norm": 10.661826133728027, "learning_rate": 2.996451612903226e-05, "loss": 0.3696, "step": 6211 }, { "epoch": 40.04497991967872, "grad_norm": 4.485554218292236, "learning_rate": 2.996129032258065e-05, "loss": 0.2865, "step": 6212 }, { "epoch": 40.05140562248996, "grad_norm": 9.933930397033691, "learning_rate": 2.9958064516129035e-05, "loss": 0.1189, "step": 6213 }, { "epoch": 40.0578313253012, "grad_norm": 6.061844348907471, "learning_rate": 2.9954838709677418e-05, "loss": 0.274, "step": 6214 }, { "epoch": 40.06425702811245, "grad_norm": 10.114497184753418, "learning_rate": 2.9951612903225807e-05, "loss": 0.5406, "step": 6215 }, { "epoch": 40.07068273092369, "grad_norm": 8.409180641174316, "learning_rate": 2.9948387096774196e-05, "loss": 0.4201, "step": 6216 }, { "epoch": 40.07710843373494, "grad_norm": 11.682641983032227, "learning_rate": 2.994516129032258e-05, "loss": 0.458, "step": 6217 }, { "epoch": 40.083534136546184, "grad_norm": 12.655980110168457, "learning_rate": 2.994193548387097e-05, "loss": 0.3277, "step": 6218 }, { "epoch": 40.08995983935743, "grad_norm": 18.823911666870117, "learning_rate": 2.9938709677419356e-05, "loss": 0.4421, "step": 6219 }, { "epoch": 40.096385542168676, "grad_norm": 11.123300552368164, "learning_rate": 2.9935483870967745e-05, "loss": 0.3887, "step": 6220 }, { "epoch": 40.10281124497992, "grad_norm": 14.42996597290039, "learning_rate": 2.9932258064516134e-05, "loss": 0.3624, "step": 6221 }, { "epoch": 40.10923694779117, "grad_norm": 7.255551338195801, "learning_rate": 2.9929032258064517e-05, "loss": 0.2, "step": 6222 }, { "epoch": 40.11566265060241, "grad_norm": 8.748655319213867, "learning_rate": 2.9925806451612903e-05, "loss": 0.3911, "step": 6223 }, { "epoch": 40.12208835341365, "grad_norm": 12.77657413482666, "learning_rate": 2.992258064516129e-05, "loss": 0.338, "step": 6224 }, { "epoch": 40.1285140562249, "grad_norm": 5.438656330108643, "learning_rate": 2.9919354838709677e-05, "loss": 0.226, "step": 6225 }, { "epoch": 40.13493975903614, "grad_norm": 23.357824325561523, "learning_rate": 2.9916129032258066e-05, "loss": 0.3758, "step": 6226 }, { "epoch": 40.14136546184739, "grad_norm": 2.5881574153900146, "learning_rate": 2.9912903225806456e-05, "loss": 0.1121, "step": 6227 }, { "epoch": 40.147791164658635, "grad_norm": 15.082082748413086, "learning_rate": 2.990967741935484e-05, "loss": 0.5259, "step": 6228 }, { "epoch": 40.15421686746988, "grad_norm": 14.366121292114258, "learning_rate": 2.9906451612903224e-05, "loss": 0.4147, "step": 6229 }, { "epoch": 40.16064257028113, "grad_norm": 18.011606216430664, "learning_rate": 2.9903225806451613e-05, "loss": 0.5512, "step": 6230 }, { "epoch": 40.16706827309237, "grad_norm": 11.603331565856934, "learning_rate": 2.9900000000000002e-05, "loss": 0.1826, "step": 6231 }, { "epoch": 40.17349397590361, "grad_norm": 15.349370002746582, "learning_rate": 2.9896774193548387e-05, "loss": 0.5545, "step": 6232 }, { "epoch": 40.17991967871486, "grad_norm": 10.94444465637207, "learning_rate": 2.9893548387096777e-05, "loss": 0.3079, "step": 6233 }, { "epoch": 40.1863453815261, "grad_norm": 8.80532455444336, "learning_rate": 2.9890322580645162e-05, "loss": 0.4358, "step": 6234 }, { "epoch": 40.19277108433735, "grad_norm": 9.486279487609863, "learning_rate": 2.988709677419355e-05, "loss": 0.521, "step": 6235 }, { "epoch": 40.199196787148594, "grad_norm": 9.812552452087402, "learning_rate": 2.988387096774194e-05, "loss": 0.4289, "step": 6236 }, { "epoch": 40.205622489959836, "grad_norm": 20.112327575683594, "learning_rate": 2.9880645161290323e-05, "loss": 0.6103, "step": 6237 }, { "epoch": 40.212048192771086, "grad_norm": 5.478322982788086, "learning_rate": 2.987741935483871e-05, "loss": 0.3285, "step": 6238 }, { "epoch": 40.21847389558233, "grad_norm": 25.86673355102539, "learning_rate": 2.9874193548387098e-05, "loss": 0.5022, "step": 6239 }, { "epoch": 40.22489959839358, "grad_norm": 5.394374847412109, "learning_rate": 2.9870967741935487e-05, "loss": 0.2824, "step": 6240 }, { "epoch": 40.23132530120482, "grad_norm": 14.067180633544922, "learning_rate": 2.9867741935483872e-05, "loss": 0.3094, "step": 6241 }, { "epoch": 40.23775100401606, "grad_norm": 14.688261032104492, "learning_rate": 2.986451612903226e-05, "loss": 0.1494, "step": 6242 }, { "epoch": 40.24417670682731, "grad_norm": 11.919589042663574, "learning_rate": 2.9861290322580647e-05, "loss": 0.3637, "step": 6243 }, { "epoch": 40.25060240963855, "grad_norm": 6.438371658325195, "learning_rate": 2.9858064516129036e-05, "loss": 0.2786, "step": 6244 }, { "epoch": 40.2570281124498, "grad_norm": 5.723232269287109, "learning_rate": 2.985483870967742e-05, "loss": 0.218, "step": 6245 }, { "epoch": 40.263453815261045, "grad_norm": 5.948709487915039, "learning_rate": 2.9851612903225808e-05, "loss": 0.4018, "step": 6246 }, { "epoch": 40.26987951807229, "grad_norm": 8.577008247375488, "learning_rate": 2.9848387096774193e-05, "loss": 0.31, "step": 6247 }, { "epoch": 40.276305220883536, "grad_norm": 7.233649253845215, "learning_rate": 2.9845161290322582e-05, "loss": 0.3854, "step": 6248 }, { "epoch": 40.28273092369478, "grad_norm": 11.907279968261719, "learning_rate": 2.9841935483870968e-05, "loss": 0.2993, "step": 6249 }, { "epoch": 40.28915662650602, "grad_norm": 6.599133491516113, "learning_rate": 2.9838709677419357e-05, "loss": 0.5375, "step": 6250 }, { "epoch": 40.29558232931727, "grad_norm": 5.027306079864502, "learning_rate": 2.9835483870967746e-05, "loss": 0.4405, "step": 6251 }, { "epoch": 40.30200803212851, "grad_norm": 6.8314056396484375, "learning_rate": 2.9832258064516132e-05, "loss": 0.6821, "step": 6252 }, { "epoch": 40.30843373493976, "grad_norm": 7.957648754119873, "learning_rate": 2.9829032258064514e-05, "loss": 0.3467, "step": 6253 }, { "epoch": 40.314859437751004, "grad_norm": 14.509231567382812, "learning_rate": 2.9825806451612903e-05, "loss": 0.281, "step": 6254 }, { "epoch": 40.321285140562246, "grad_norm": 3.813133478164673, "learning_rate": 2.9822580645161293e-05, "loss": 0.2628, "step": 6255 }, { "epoch": 40.327710843373495, "grad_norm": 9.593283653259277, "learning_rate": 2.9819354838709678e-05, "loss": 0.4152, "step": 6256 }, { "epoch": 40.33413654618474, "grad_norm": 22.260778427124023, "learning_rate": 2.9816129032258067e-05, "loss": 0.4317, "step": 6257 }, { "epoch": 40.34056224899599, "grad_norm": 8.149809837341309, "learning_rate": 2.9812903225806453e-05, "loss": 0.2961, "step": 6258 }, { "epoch": 40.34698795180723, "grad_norm": 6.4103102684021, "learning_rate": 2.9809677419354842e-05, "loss": 0.1976, "step": 6259 }, { "epoch": 40.35341365461847, "grad_norm": 11.1505708694458, "learning_rate": 2.9806451612903224e-05, "loss": 0.5885, "step": 6260 }, { "epoch": 40.35983935742972, "grad_norm": 17.737964630126953, "learning_rate": 2.9803225806451614e-05, "loss": 0.2074, "step": 6261 }, { "epoch": 40.36626506024096, "grad_norm": 15.826858520507812, "learning_rate": 2.98e-05, "loss": 0.3744, "step": 6262 }, { "epoch": 40.37269076305221, "grad_norm": 9.097360610961914, "learning_rate": 2.979677419354839e-05, "loss": 0.4274, "step": 6263 }, { "epoch": 40.379116465863454, "grad_norm": 5.404153347015381, "learning_rate": 2.9793548387096774e-05, "loss": 0.2142, "step": 6264 }, { "epoch": 40.3855421686747, "grad_norm": 12.85130500793457, "learning_rate": 2.9790322580645163e-05, "loss": 0.3006, "step": 6265 }, { "epoch": 40.391967871485946, "grad_norm": 7.733003616333008, "learning_rate": 2.9787096774193552e-05, "loss": 0.5895, "step": 6266 }, { "epoch": 40.39839357429719, "grad_norm": 11.343696594238281, "learning_rate": 2.9783870967741938e-05, "loss": 0.1278, "step": 6267 }, { "epoch": 40.40481927710843, "grad_norm": 14.244894027709961, "learning_rate": 2.978064516129032e-05, "loss": 0.6294, "step": 6268 }, { "epoch": 40.41124497991968, "grad_norm": 13.454228401184082, "learning_rate": 2.977741935483871e-05, "loss": 0.4574, "step": 6269 }, { "epoch": 40.41767068273092, "grad_norm": 10.953049659729004, "learning_rate": 2.97741935483871e-05, "loss": 0.476, "step": 6270 }, { "epoch": 40.42409638554217, "grad_norm": 4.176214218139648, "learning_rate": 2.9770967741935484e-05, "loss": 0.3008, "step": 6271 }, { "epoch": 40.43052208835341, "grad_norm": 8.146389961242676, "learning_rate": 2.9767741935483873e-05, "loss": 0.2529, "step": 6272 }, { "epoch": 40.436947791164656, "grad_norm": 11.692792892456055, "learning_rate": 2.976451612903226e-05, "loss": 0.3829, "step": 6273 }, { "epoch": 40.443373493975905, "grad_norm": 5.175943374633789, "learning_rate": 2.9761290322580648e-05, "loss": 0.2203, "step": 6274 }, { "epoch": 40.44979919678715, "grad_norm": 11.560386657714844, "learning_rate": 2.9758064516129037e-05, "loss": 0.2863, "step": 6275 }, { "epoch": 40.456224899598396, "grad_norm": 19.583484649658203, "learning_rate": 2.975483870967742e-05, "loss": 0.3811, "step": 6276 }, { "epoch": 40.46265060240964, "grad_norm": 9.847454071044922, "learning_rate": 2.9751612903225805e-05, "loss": 0.3476, "step": 6277 }, { "epoch": 40.46907630522088, "grad_norm": 9.494978904724121, "learning_rate": 2.9748387096774194e-05, "loss": 0.2646, "step": 6278 }, { "epoch": 40.47550200803213, "grad_norm": 11.53348445892334, "learning_rate": 2.9745161290322583e-05, "loss": 0.2905, "step": 6279 }, { "epoch": 40.48192771084337, "grad_norm": 6.893826961517334, "learning_rate": 2.974193548387097e-05, "loss": 0.4325, "step": 6280 }, { "epoch": 40.48835341365462, "grad_norm": 14.420075416564941, "learning_rate": 2.9738709677419358e-05, "loss": 0.4208, "step": 6281 }, { "epoch": 40.494779116465864, "grad_norm": 8.938996315002441, "learning_rate": 2.9735483870967744e-05, "loss": 0.3269, "step": 6282 }, { "epoch": 40.501204819277106, "grad_norm": 12.786513328552246, "learning_rate": 2.9732258064516133e-05, "loss": 0.32, "step": 6283 }, { "epoch": 40.507630522088355, "grad_norm": 8.776115417480469, "learning_rate": 2.9729032258064515e-05, "loss": 0.3854, "step": 6284 }, { "epoch": 40.5140562248996, "grad_norm": 11.052535057067871, "learning_rate": 2.9725806451612904e-05, "loss": 0.3017, "step": 6285 }, { "epoch": 40.52048192771084, "grad_norm": 20.98642349243164, "learning_rate": 2.972258064516129e-05, "loss": 0.5774, "step": 6286 }, { "epoch": 40.52690763052209, "grad_norm": 5.441983699798584, "learning_rate": 2.971935483870968e-05, "loss": 0.22, "step": 6287 }, { "epoch": 40.53333333333333, "grad_norm": 15.294764518737793, "learning_rate": 2.9716129032258065e-05, "loss": 0.8829, "step": 6288 }, { "epoch": 40.53975903614458, "grad_norm": 14.808940887451172, "learning_rate": 2.9712903225806454e-05, "loss": 0.5859, "step": 6289 }, { "epoch": 40.54618473895582, "grad_norm": 9.007969856262207, "learning_rate": 2.9709677419354843e-05, "loss": 0.4438, "step": 6290 }, { "epoch": 40.552610441767065, "grad_norm": 11.875972747802734, "learning_rate": 2.970645161290323e-05, "loss": 0.4963, "step": 6291 }, { "epoch": 40.559036144578315, "grad_norm": 8.582267761230469, "learning_rate": 2.970322580645161e-05, "loss": 0.2964, "step": 6292 }, { "epoch": 40.56546184738956, "grad_norm": 5.066148281097412, "learning_rate": 2.97e-05, "loss": 0.19, "step": 6293 }, { "epoch": 40.571887550200806, "grad_norm": 11.28389835357666, "learning_rate": 2.969677419354839e-05, "loss": 0.4556, "step": 6294 }, { "epoch": 40.57831325301205, "grad_norm": 12.36076831817627, "learning_rate": 2.9693548387096775e-05, "loss": 0.3715, "step": 6295 }, { "epoch": 40.58473895582329, "grad_norm": 19.087217330932617, "learning_rate": 2.9690322580645164e-05, "loss": 0.4999, "step": 6296 }, { "epoch": 40.59116465863454, "grad_norm": 6.723677158355713, "learning_rate": 2.968709677419355e-05, "loss": 0.3897, "step": 6297 }, { "epoch": 40.59759036144578, "grad_norm": 7.645511627197266, "learning_rate": 2.968387096774194e-05, "loss": 0.1649, "step": 6298 }, { "epoch": 40.60401606425703, "grad_norm": 8.330875396728516, "learning_rate": 2.968064516129032e-05, "loss": 0.4922, "step": 6299 }, { "epoch": 40.610441767068274, "grad_norm": 6.16609525680542, "learning_rate": 2.967741935483871e-05, "loss": 0.224, "step": 6300 }, { "epoch": 40.616867469879516, "grad_norm": 24.139820098876953, "learning_rate": 2.9674193548387096e-05, "loss": 0.7061, "step": 6301 }, { "epoch": 40.623293172690765, "grad_norm": 9.286352157592773, "learning_rate": 2.9670967741935485e-05, "loss": 0.4704, "step": 6302 }, { "epoch": 40.62971887550201, "grad_norm": 9.594524383544922, "learning_rate": 2.9667741935483874e-05, "loss": 0.2349, "step": 6303 }, { "epoch": 40.63614457831325, "grad_norm": 16.041568756103516, "learning_rate": 2.966451612903226e-05, "loss": 0.5234, "step": 6304 }, { "epoch": 40.6425702811245, "grad_norm": 9.742194175720215, "learning_rate": 2.966129032258065e-05, "loss": 0.3838, "step": 6305 }, { "epoch": 40.64899598393574, "grad_norm": 4.969516277313232, "learning_rate": 2.9658064516129035e-05, "loss": 0.2213, "step": 6306 }, { "epoch": 40.65542168674699, "grad_norm": 7.873771667480469, "learning_rate": 2.9654838709677417e-05, "loss": 0.5295, "step": 6307 }, { "epoch": 40.66184738955823, "grad_norm": 6.861062526702881, "learning_rate": 2.9651612903225806e-05, "loss": 0.4669, "step": 6308 }, { "epoch": 40.668273092369475, "grad_norm": 6.735121726989746, "learning_rate": 2.9648387096774195e-05, "loss": 0.4326, "step": 6309 }, { "epoch": 40.674698795180724, "grad_norm": 13.442108154296875, "learning_rate": 2.964516129032258e-05, "loss": 0.278, "step": 6310 }, { "epoch": 40.681124497991966, "grad_norm": 9.6975736618042, "learning_rate": 2.964193548387097e-05, "loss": 0.5546, "step": 6311 }, { "epoch": 40.687550200803216, "grad_norm": 15.758618354797363, "learning_rate": 2.9638709677419356e-05, "loss": 0.539, "step": 6312 }, { "epoch": 40.69397590361446, "grad_norm": 16.775028228759766, "learning_rate": 2.9635483870967745e-05, "loss": 0.4863, "step": 6313 }, { "epoch": 40.7004016064257, "grad_norm": 7.583291053771973, "learning_rate": 2.9632258064516134e-05, "loss": 0.3468, "step": 6314 }, { "epoch": 40.70682730923695, "grad_norm": 9.965566635131836, "learning_rate": 2.9629032258064516e-05, "loss": 0.2397, "step": 6315 }, { "epoch": 40.71325301204819, "grad_norm": 5.792891979217529, "learning_rate": 2.9625806451612902e-05, "loss": 0.2815, "step": 6316 }, { "epoch": 40.71967871485944, "grad_norm": 5.82417106628418, "learning_rate": 2.962258064516129e-05, "loss": 0.5042, "step": 6317 }, { "epoch": 40.72610441767068, "grad_norm": 10.426259994506836, "learning_rate": 2.961935483870968e-05, "loss": 0.3984, "step": 6318 }, { "epoch": 40.732530120481925, "grad_norm": 17.898395538330078, "learning_rate": 2.9616129032258066e-05, "loss": 0.4279, "step": 6319 }, { "epoch": 40.738955823293175, "grad_norm": 9.621062278747559, "learning_rate": 2.9612903225806455e-05, "loss": 0.4105, "step": 6320 }, { "epoch": 40.74538152610442, "grad_norm": 14.492431640625, "learning_rate": 2.960967741935484e-05, "loss": 0.6618, "step": 6321 }, { "epoch": 40.75180722891566, "grad_norm": 23.66358757019043, "learning_rate": 2.960645161290323e-05, "loss": 0.6343, "step": 6322 }, { "epoch": 40.75823293172691, "grad_norm": 8.92880916595459, "learning_rate": 2.9603225806451612e-05, "loss": 0.2891, "step": 6323 }, { "epoch": 40.76465863453815, "grad_norm": 5.579379081726074, "learning_rate": 2.96e-05, "loss": 0.2144, "step": 6324 }, { "epoch": 40.7710843373494, "grad_norm": 6.535120487213135, "learning_rate": 2.9596774193548387e-05, "loss": 0.2009, "step": 6325 }, { "epoch": 40.77751004016064, "grad_norm": 14.664642333984375, "learning_rate": 2.9593548387096776e-05, "loss": 0.7092, "step": 6326 }, { "epoch": 40.783935742971885, "grad_norm": 5.466191291809082, "learning_rate": 2.9590322580645162e-05, "loss": 0.2007, "step": 6327 }, { "epoch": 40.790361445783134, "grad_norm": 11.601641654968262, "learning_rate": 2.958709677419355e-05, "loss": 0.2541, "step": 6328 }, { "epoch": 40.796787148594376, "grad_norm": 7.783644676208496, "learning_rate": 2.958387096774194e-05, "loss": 0.4687, "step": 6329 }, { "epoch": 40.803212851405625, "grad_norm": 10.886876106262207, "learning_rate": 2.9580645161290322e-05, "loss": 0.333, "step": 6330 }, { "epoch": 40.80963855421687, "grad_norm": 14.039788246154785, "learning_rate": 2.9577419354838708e-05, "loss": 0.5032, "step": 6331 }, { "epoch": 40.81606425702811, "grad_norm": 7.154376983642578, "learning_rate": 2.9574193548387097e-05, "loss": 0.1325, "step": 6332 }, { "epoch": 40.82248995983936, "grad_norm": 5.0315375328063965, "learning_rate": 2.9570967741935486e-05, "loss": 0.3581, "step": 6333 }, { "epoch": 40.8289156626506, "grad_norm": 11.620443344116211, "learning_rate": 2.9567741935483872e-05, "loss": 0.573, "step": 6334 }, { "epoch": 40.83534136546185, "grad_norm": 8.90910816192627, "learning_rate": 2.956451612903226e-05, "loss": 0.3816, "step": 6335 }, { "epoch": 40.84176706827309, "grad_norm": 15.178013801574707, "learning_rate": 2.9561290322580647e-05, "loss": 0.2738, "step": 6336 }, { "epoch": 40.848192771084335, "grad_norm": 19.588829040527344, "learning_rate": 2.9558064516129036e-05, "loss": 0.5759, "step": 6337 }, { "epoch": 40.854618473895584, "grad_norm": 9.050081253051758, "learning_rate": 2.9554838709677418e-05, "loss": 0.4801, "step": 6338 }, { "epoch": 40.86104417670683, "grad_norm": 8.779565811157227, "learning_rate": 2.9551612903225807e-05, "loss": 0.2961, "step": 6339 }, { "epoch": 40.86746987951807, "grad_norm": 26.200305938720703, "learning_rate": 2.9548387096774193e-05, "loss": 0.4495, "step": 6340 }, { "epoch": 40.87389558232932, "grad_norm": 8.522408485412598, "learning_rate": 2.9545161290322582e-05, "loss": 0.1442, "step": 6341 }, { "epoch": 40.88032128514056, "grad_norm": 11.6705322265625, "learning_rate": 2.954193548387097e-05, "loss": 0.4377, "step": 6342 }, { "epoch": 40.88674698795181, "grad_norm": 7.606959342956543, "learning_rate": 2.9538709677419357e-05, "loss": 0.288, "step": 6343 }, { "epoch": 40.89317269076305, "grad_norm": 13.959073066711426, "learning_rate": 2.9535483870967746e-05, "loss": 0.6022, "step": 6344 }, { "epoch": 40.899598393574294, "grad_norm": 16.486488342285156, "learning_rate": 2.953225806451613e-05, "loss": 0.5287, "step": 6345 }, { "epoch": 40.90602409638554, "grad_norm": 10.657832145690918, "learning_rate": 2.9529032258064514e-05, "loss": 0.2119, "step": 6346 }, { "epoch": 40.912449799196786, "grad_norm": 7.983388900756836, "learning_rate": 2.9525806451612903e-05, "loss": 0.2834, "step": 6347 }, { "epoch": 40.918875502008035, "grad_norm": 12.500000953674316, "learning_rate": 2.9522580645161292e-05, "loss": 0.4991, "step": 6348 }, { "epoch": 40.92530120481928, "grad_norm": 12.63341236114502, "learning_rate": 2.9519354838709678e-05, "loss": 0.3198, "step": 6349 }, { "epoch": 40.93172690763052, "grad_norm": 7.21807336807251, "learning_rate": 2.9516129032258067e-05, "loss": 0.2699, "step": 6350 }, { "epoch": 40.93815261044177, "grad_norm": 9.61251163482666, "learning_rate": 2.9512903225806453e-05, "loss": 0.3068, "step": 6351 }, { "epoch": 40.94457831325301, "grad_norm": 4.528140544891357, "learning_rate": 2.950967741935484e-05, "loss": 0.2471, "step": 6352 }, { "epoch": 40.95100401606426, "grad_norm": 4.159992694854736, "learning_rate": 2.950645161290323e-05, "loss": 0.1705, "step": 6353 }, { "epoch": 40.9574297188755, "grad_norm": 10.80868911743164, "learning_rate": 2.9503225806451613e-05, "loss": 0.3687, "step": 6354 }, { "epoch": 40.963855421686745, "grad_norm": 4.339812755584717, "learning_rate": 2.95e-05, "loss": 0.1466, "step": 6355 }, { "epoch": 40.970281124497994, "grad_norm": 8.861429214477539, "learning_rate": 2.9496774193548388e-05, "loss": 0.2025, "step": 6356 }, { "epoch": 40.976706827309236, "grad_norm": 10.997163772583008, "learning_rate": 2.9493548387096777e-05, "loss": 0.5043, "step": 6357 }, { "epoch": 40.98313253012048, "grad_norm": 6.496946334838867, "learning_rate": 2.9490322580645163e-05, "loss": 0.2398, "step": 6358 }, { "epoch": 40.98955823293173, "grad_norm": 9.07165813446045, "learning_rate": 2.9487096774193552e-05, "loss": 0.6483, "step": 6359 }, { "epoch": 40.99598393574297, "grad_norm": 10.39330005645752, "learning_rate": 2.9483870967741937e-05, "loss": 0.2731, "step": 6360 }, { "epoch": 41.00562248995984, "grad_norm": 10.955066680908203, "learning_rate": 2.9480645161290327e-05, "loss": 0.402, "step": 6361 }, { "epoch": 41.01204819277108, "grad_norm": 3.9042422771453857, "learning_rate": 2.947741935483871e-05, "loss": 0.161, "step": 6362 }, { "epoch": 41.01847389558233, "grad_norm": 46.7545051574707, "learning_rate": 2.9474193548387098e-05, "loss": 0.5952, "step": 6363 }, { "epoch": 41.024899598393574, "grad_norm": 8.091851234436035, "learning_rate": 2.9470967741935484e-05, "loss": 0.4207, "step": 6364 }, { "epoch": 41.03132530120482, "grad_norm": 7.219583988189697, "learning_rate": 2.9467741935483873e-05, "loss": 0.1879, "step": 6365 }, { "epoch": 41.037751004016066, "grad_norm": 27.137096405029297, "learning_rate": 2.9464516129032262e-05, "loss": 0.5465, "step": 6366 }, { "epoch": 41.04417670682731, "grad_norm": 7.2918925285339355, "learning_rate": 2.9461290322580648e-05, "loss": 0.4107, "step": 6367 }, { "epoch": 41.05060240963856, "grad_norm": 13.589943885803223, "learning_rate": 2.9458064516129037e-05, "loss": 0.3704, "step": 6368 }, { "epoch": 41.0570281124498, "grad_norm": 10.81155014038086, "learning_rate": 2.945483870967742e-05, "loss": 0.2769, "step": 6369 }, { "epoch": 41.06345381526104, "grad_norm": 9.804186820983887, "learning_rate": 2.9451612903225805e-05, "loss": 0.4692, "step": 6370 }, { "epoch": 41.06987951807229, "grad_norm": 4.391042709350586, "learning_rate": 2.9448387096774194e-05, "loss": 0.3581, "step": 6371 }, { "epoch": 41.07630522088353, "grad_norm": 12.604618072509766, "learning_rate": 2.9445161290322583e-05, "loss": 0.5416, "step": 6372 }, { "epoch": 41.08273092369478, "grad_norm": 11.404884338378906, "learning_rate": 2.944193548387097e-05, "loss": 0.1756, "step": 6373 }, { "epoch": 41.089156626506025, "grad_norm": 17.609394073486328, "learning_rate": 2.9438709677419358e-05, "loss": 0.3473, "step": 6374 }, { "epoch": 41.09558232931727, "grad_norm": 23.493101119995117, "learning_rate": 2.9435483870967743e-05, "loss": 0.2168, "step": 6375 }, { "epoch": 41.102008032128516, "grad_norm": 13.53542709350586, "learning_rate": 2.9432258064516132e-05, "loss": 0.5173, "step": 6376 }, { "epoch": 41.10843373493976, "grad_norm": 10.67378044128418, "learning_rate": 2.9429032258064515e-05, "loss": 0.5963, "step": 6377 }, { "epoch": 41.114859437751, "grad_norm": 5.354868412017822, "learning_rate": 2.9425806451612904e-05, "loss": 0.4114, "step": 6378 }, { "epoch": 41.12128514056225, "grad_norm": 18.472047805786133, "learning_rate": 2.942258064516129e-05, "loss": 0.2185, "step": 6379 }, { "epoch": 41.12771084337349, "grad_norm": 1.75070321559906, "learning_rate": 2.941935483870968e-05, "loss": 0.2615, "step": 6380 }, { "epoch": 41.13413654618474, "grad_norm": 7.621248722076416, "learning_rate": 2.9416129032258068e-05, "loss": 0.2727, "step": 6381 }, { "epoch": 41.140562248995984, "grad_norm": 13.547602653503418, "learning_rate": 2.9412903225806454e-05, "loss": 0.4495, "step": 6382 }, { "epoch": 41.146987951807226, "grad_norm": 18.681028366088867, "learning_rate": 2.9409677419354843e-05, "loss": 0.3891, "step": 6383 }, { "epoch": 41.153413654618475, "grad_norm": 11.977970123291016, "learning_rate": 2.940645161290323e-05, "loss": 0.557, "step": 6384 }, { "epoch": 41.15983935742972, "grad_norm": 19.894590377807617, "learning_rate": 2.940322580645161e-05, "loss": 0.3303, "step": 6385 }, { "epoch": 41.16626506024097, "grad_norm": 24.672922134399414, "learning_rate": 2.94e-05, "loss": 0.3864, "step": 6386 }, { "epoch": 41.17269076305221, "grad_norm": 7.34035587310791, "learning_rate": 2.939677419354839e-05, "loss": 0.3032, "step": 6387 }, { "epoch": 41.17911646586345, "grad_norm": 16.590166091918945, "learning_rate": 2.9393548387096775e-05, "loss": 0.2427, "step": 6388 }, { "epoch": 41.1855421686747, "grad_norm": 27.68614959716797, "learning_rate": 2.9390322580645164e-05, "loss": 0.2587, "step": 6389 }, { "epoch": 41.19196787148594, "grad_norm": 13.365257263183594, "learning_rate": 2.938709677419355e-05, "loss": 0.4573, "step": 6390 }, { "epoch": 41.19839357429719, "grad_norm": 3.6395912170410156, "learning_rate": 2.938387096774194e-05, "loss": 0.1778, "step": 6391 }, { "epoch": 41.204819277108435, "grad_norm": 8.318143844604492, "learning_rate": 2.9380645161290328e-05, "loss": 0.5419, "step": 6392 }, { "epoch": 41.21124497991968, "grad_norm": 27.23712158203125, "learning_rate": 2.937741935483871e-05, "loss": 0.4059, "step": 6393 }, { "epoch": 41.217670682730926, "grad_norm": 16.456846237182617, "learning_rate": 2.9374193548387096e-05, "loss": 0.3351, "step": 6394 }, { "epoch": 41.22409638554217, "grad_norm": 37.408138275146484, "learning_rate": 2.9370967741935485e-05, "loss": 0.4418, "step": 6395 }, { "epoch": 41.23052208835341, "grad_norm": 13.674593925476074, "learning_rate": 2.9367741935483874e-05, "loss": 0.5388, "step": 6396 }, { "epoch": 41.23694779116466, "grad_norm": 13.808338165283203, "learning_rate": 2.936451612903226e-05, "loss": 0.3212, "step": 6397 }, { "epoch": 41.2433734939759, "grad_norm": 14.423866271972656, "learning_rate": 2.936129032258065e-05, "loss": 0.227, "step": 6398 }, { "epoch": 41.24979919678715, "grad_norm": 19.06783103942871, "learning_rate": 2.9358064516129034e-05, "loss": 0.2916, "step": 6399 }, { "epoch": 41.256224899598394, "grad_norm": 5.961629867553711, "learning_rate": 2.9354838709677417e-05, "loss": 0.3564, "step": 6400 }, { "epoch": 41.262650602409636, "grad_norm": 4.940728664398193, "learning_rate": 2.9351612903225806e-05, "loss": 0.1861, "step": 6401 }, { "epoch": 41.269076305220885, "grad_norm": 6.431789875030518, "learning_rate": 2.9348387096774195e-05, "loss": 0.4051, "step": 6402 }, { "epoch": 41.27550200803213, "grad_norm": 3.5491909980773926, "learning_rate": 2.934516129032258e-05, "loss": 0.3314, "step": 6403 }, { "epoch": 41.28192771084338, "grad_norm": 27.5286808013916, "learning_rate": 2.934193548387097e-05, "loss": 0.6377, "step": 6404 }, { "epoch": 41.28835341365462, "grad_norm": 15.064513206481934, "learning_rate": 2.933870967741936e-05, "loss": 0.4382, "step": 6405 }, { "epoch": 41.29477911646586, "grad_norm": 4.266350746154785, "learning_rate": 2.9335483870967744e-05, "loss": 0.2384, "step": 6406 }, { "epoch": 41.30120481927711, "grad_norm": 7.980227947235107, "learning_rate": 2.9332258064516133e-05, "loss": 0.24, "step": 6407 }, { "epoch": 41.30763052208835, "grad_norm": 3.299412965774536, "learning_rate": 2.9329032258064516e-05, "loss": 0.4812, "step": 6408 }, { "epoch": 41.3140562248996, "grad_norm": 5.640388011932373, "learning_rate": 2.93258064516129e-05, "loss": 0.3519, "step": 6409 }, { "epoch": 41.320481927710844, "grad_norm": 24.115039825439453, "learning_rate": 2.932258064516129e-05, "loss": 1.0606, "step": 6410 }, { "epoch": 41.326907630522086, "grad_norm": 25.85173797607422, "learning_rate": 2.931935483870968e-05, "loss": 0.4397, "step": 6411 }, { "epoch": 41.333333333333336, "grad_norm": 29.616731643676758, "learning_rate": 2.9316129032258065e-05, "loss": 0.2738, "step": 6412 }, { "epoch": 41.33975903614458, "grad_norm": 19.92890167236328, "learning_rate": 2.9312903225806454e-05, "loss": 0.3173, "step": 6413 }, { "epoch": 41.34618473895582, "grad_norm": 4.123230934143066, "learning_rate": 2.930967741935484e-05, "loss": 0.1821, "step": 6414 }, { "epoch": 41.35261044176707, "grad_norm": 8.785836219787598, "learning_rate": 2.930645161290323e-05, "loss": 0.4272, "step": 6415 }, { "epoch": 41.35903614457831, "grad_norm": 4.527311325073242, "learning_rate": 2.930322580645161e-05, "loss": 0.2419, "step": 6416 }, { "epoch": 41.36546184738956, "grad_norm": 55.296783447265625, "learning_rate": 2.93e-05, "loss": 0.4831, "step": 6417 }, { "epoch": 41.3718875502008, "grad_norm": 27.641511917114258, "learning_rate": 2.9296774193548386e-05, "loss": 0.3465, "step": 6418 }, { "epoch": 41.378313253012045, "grad_norm": 23.457494735717773, "learning_rate": 2.9293548387096775e-05, "loss": 0.202, "step": 6419 }, { "epoch": 41.384738955823295, "grad_norm": 30.40280532836914, "learning_rate": 2.9290322580645165e-05, "loss": 0.2805, "step": 6420 }, { "epoch": 41.39116465863454, "grad_norm": 8.332939147949219, "learning_rate": 2.928709677419355e-05, "loss": 0.3492, "step": 6421 }, { "epoch": 41.397590361445786, "grad_norm": 6.887089252471924, "learning_rate": 2.928387096774194e-05, "loss": 0.2714, "step": 6422 }, { "epoch": 41.40401606425703, "grad_norm": 17.75130844116211, "learning_rate": 2.9280645161290325e-05, "loss": 0.6416, "step": 6423 }, { "epoch": 41.41044176706827, "grad_norm": 4.955935001373291, "learning_rate": 2.9277419354838707e-05, "loss": 0.3265, "step": 6424 }, { "epoch": 41.41686746987952, "grad_norm": 17.003820419311523, "learning_rate": 2.9274193548387097e-05, "loss": 0.3491, "step": 6425 }, { "epoch": 41.42329317269076, "grad_norm": 10.830525398254395, "learning_rate": 2.9270967741935486e-05, "loss": 0.4462, "step": 6426 }, { "epoch": 41.429718875502004, "grad_norm": 11.440634727478027, "learning_rate": 2.926774193548387e-05, "loss": 0.3072, "step": 6427 }, { "epoch": 41.436144578313254, "grad_norm": 6.113356113433838, "learning_rate": 2.926451612903226e-05, "loss": 0.4384, "step": 6428 }, { "epoch": 41.442570281124496, "grad_norm": 20.473146438598633, "learning_rate": 2.926129032258065e-05, "loss": 0.4991, "step": 6429 }, { "epoch": 41.448995983935745, "grad_norm": 4.031788349151611, "learning_rate": 2.9258064516129035e-05, "loss": 0.1383, "step": 6430 }, { "epoch": 41.45542168674699, "grad_norm": 2.1391494274139404, "learning_rate": 2.9254838709677424e-05, "loss": 0.2231, "step": 6431 }, { "epoch": 41.46184738955823, "grad_norm": 1.4914976358413696, "learning_rate": 2.9251612903225807e-05, "loss": 0.3179, "step": 6432 }, { "epoch": 41.46827309236948, "grad_norm": 46.45566940307617, "learning_rate": 2.9248387096774192e-05, "loss": 0.401, "step": 6433 }, { "epoch": 41.47469879518072, "grad_norm": 15.59701919555664, "learning_rate": 2.924516129032258e-05, "loss": 0.1661, "step": 6434 }, { "epoch": 41.48112449799197, "grad_norm": 34.32433319091797, "learning_rate": 2.924193548387097e-05, "loss": 0.5147, "step": 6435 }, { "epoch": 41.48755020080321, "grad_norm": 6.101129055023193, "learning_rate": 2.9238709677419356e-05, "loss": 0.4952, "step": 6436 }, { "epoch": 41.493975903614455, "grad_norm": 2.434997081756592, "learning_rate": 2.9235483870967745e-05, "loss": 0.2604, "step": 6437 }, { "epoch": 41.500401606425704, "grad_norm": 28.514781951904297, "learning_rate": 2.923225806451613e-05, "loss": 0.7032, "step": 6438 }, { "epoch": 41.50682730923695, "grad_norm": 18.661020278930664, "learning_rate": 2.9229032258064513e-05, "loss": 0.3814, "step": 6439 }, { "epoch": 41.513253012048196, "grad_norm": 15.32109546661377, "learning_rate": 2.9225806451612902e-05, "loss": 0.2496, "step": 6440 }, { "epoch": 41.51967871485944, "grad_norm": 13.20781135559082, "learning_rate": 2.922258064516129e-05, "loss": 0.3645, "step": 6441 }, { "epoch": 41.52610441767068, "grad_norm": 4.247008323669434, "learning_rate": 2.9219354838709677e-05, "loss": 0.2054, "step": 6442 }, { "epoch": 41.53253012048193, "grad_norm": 11.783411979675293, "learning_rate": 2.9216129032258066e-05, "loss": 0.3828, "step": 6443 }, { "epoch": 41.53895582329317, "grad_norm": 11.775041580200195, "learning_rate": 2.9212903225806455e-05, "loss": 0.4787, "step": 6444 }, { "epoch": 41.54538152610442, "grad_norm": 35.98434066772461, "learning_rate": 2.920967741935484e-05, "loss": 0.4345, "step": 6445 }, { "epoch": 41.55180722891566, "grad_norm": 13.112329483032227, "learning_rate": 2.920645161290323e-05, "loss": 0.6817, "step": 6446 }, { "epoch": 41.558232931726906, "grad_norm": 13.980307579040527, "learning_rate": 2.9203225806451613e-05, "loss": 0.3869, "step": 6447 }, { "epoch": 41.564658634538155, "grad_norm": 9.487811088562012, "learning_rate": 2.9199999999999998e-05, "loss": 0.3123, "step": 6448 }, { "epoch": 41.5710843373494, "grad_norm": 14.615299224853516, "learning_rate": 2.9196774193548387e-05, "loss": 0.3238, "step": 6449 }, { "epoch": 41.57751004016064, "grad_norm": 18.95184898376465, "learning_rate": 2.9193548387096776e-05, "loss": 0.6151, "step": 6450 }, { "epoch": 41.58393574297189, "grad_norm": 12.669065475463867, "learning_rate": 2.9190322580645162e-05, "loss": 0.3177, "step": 6451 }, { "epoch": 41.59036144578313, "grad_norm": 32.1550407409668, "learning_rate": 2.918709677419355e-05, "loss": 0.3783, "step": 6452 }, { "epoch": 41.59678714859438, "grad_norm": 17.019187927246094, "learning_rate": 2.9183870967741937e-05, "loss": 0.1775, "step": 6453 }, { "epoch": 41.60321285140562, "grad_norm": 45.064727783203125, "learning_rate": 2.9180645161290326e-05, "loss": 0.2633, "step": 6454 }, { "epoch": 41.609638554216865, "grad_norm": 5.583099842071533, "learning_rate": 2.917741935483871e-05, "loss": 0.4463, "step": 6455 }, { "epoch": 41.616064257028114, "grad_norm": 31.274343490600586, "learning_rate": 2.9174193548387097e-05, "loss": 0.6361, "step": 6456 }, { "epoch": 41.622489959839356, "grad_norm": 13.543919563293457, "learning_rate": 2.9170967741935483e-05, "loss": 0.4508, "step": 6457 }, { "epoch": 41.628915662650606, "grad_norm": 19.73433494567871, "learning_rate": 2.9167741935483872e-05, "loss": 0.2851, "step": 6458 }, { "epoch": 41.63534136546185, "grad_norm": 26.111902236938477, "learning_rate": 2.916451612903226e-05, "loss": 0.4708, "step": 6459 }, { "epoch": 41.64176706827309, "grad_norm": 5.593952655792236, "learning_rate": 2.9161290322580647e-05, "loss": 0.1307, "step": 6460 }, { "epoch": 41.64819277108434, "grad_norm": 14.517417907714844, "learning_rate": 2.9158064516129036e-05, "loss": 0.606, "step": 6461 }, { "epoch": 41.65461847389558, "grad_norm": 20.40607452392578, "learning_rate": 2.9154838709677422e-05, "loss": 0.3257, "step": 6462 }, { "epoch": 41.661044176706824, "grad_norm": 22.641263961791992, "learning_rate": 2.9151612903225804e-05, "loss": 0.4091, "step": 6463 }, { "epoch": 41.66746987951807, "grad_norm": 15.048809051513672, "learning_rate": 2.9148387096774193e-05, "loss": 0.4184, "step": 6464 }, { "epoch": 41.673895582329315, "grad_norm": 3.4175517559051514, "learning_rate": 2.9145161290322582e-05, "loss": 0.116, "step": 6465 }, { "epoch": 41.680321285140565, "grad_norm": 18.022714614868164, "learning_rate": 2.9141935483870968e-05, "loss": 0.3383, "step": 6466 }, { "epoch": 41.68674698795181, "grad_norm": 27.65518569946289, "learning_rate": 2.9138709677419357e-05, "loss": 0.4361, "step": 6467 }, { "epoch": 41.69317269076305, "grad_norm": 10.957051277160645, "learning_rate": 2.9135483870967746e-05, "loss": 0.2894, "step": 6468 }, { "epoch": 41.6995983935743, "grad_norm": 16.197490692138672, "learning_rate": 2.9132258064516132e-05, "loss": 0.5559, "step": 6469 }, { "epoch": 41.70602409638554, "grad_norm": 14.728711128234863, "learning_rate": 2.912903225806452e-05, "loss": 0.4478, "step": 6470 }, { "epoch": 41.71244979919679, "grad_norm": 44.906700134277344, "learning_rate": 2.9125806451612903e-05, "loss": 0.4328, "step": 6471 }, { "epoch": 41.71887550200803, "grad_norm": 12.671550750732422, "learning_rate": 2.912258064516129e-05, "loss": 0.4974, "step": 6472 }, { "epoch": 41.725301204819274, "grad_norm": 18.095272064208984, "learning_rate": 2.9119354838709678e-05, "loss": 0.3317, "step": 6473 }, { "epoch": 41.731726907630524, "grad_norm": 16.2546329498291, "learning_rate": 2.9116129032258067e-05, "loss": 0.4982, "step": 6474 }, { "epoch": 41.738152610441766, "grad_norm": 25.89173126220703, "learning_rate": 2.9112903225806453e-05, "loss": 0.2363, "step": 6475 }, { "epoch": 41.744578313253015, "grad_norm": 20.011075973510742, "learning_rate": 2.9109677419354842e-05, "loss": 0.612, "step": 6476 }, { "epoch": 41.75100401606426, "grad_norm": 28.92971420288086, "learning_rate": 2.9106451612903228e-05, "loss": 0.3588, "step": 6477 }, { "epoch": 41.7574297188755, "grad_norm": 2.7886245250701904, "learning_rate": 2.910322580645161e-05, "loss": 0.0974, "step": 6478 }, { "epoch": 41.76385542168675, "grad_norm": 14.718083381652832, "learning_rate": 2.91e-05, "loss": 0.3582, "step": 6479 }, { "epoch": 41.77028112449799, "grad_norm": 12.11688232421875, "learning_rate": 2.909677419354839e-05, "loss": 0.3974, "step": 6480 }, { "epoch": 41.77670682730923, "grad_norm": 10.664522171020508, "learning_rate": 2.9093548387096774e-05, "loss": 0.389, "step": 6481 }, { "epoch": 41.78313253012048, "grad_norm": 28.822641372680664, "learning_rate": 2.9090322580645163e-05, "loss": 0.273, "step": 6482 }, { "epoch": 41.789558232931725, "grad_norm": 17.76539421081543, "learning_rate": 2.9087096774193552e-05, "loss": 0.4666, "step": 6483 }, { "epoch": 41.795983935742974, "grad_norm": 13.047259330749512, "learning_rate": 2.9083870967741938e-05, "loss": 0.2168, "step": 6484 }, { "epoch": 41.80240963855422, "grad_norm": 1.2455103397369385, "learning_rate": 2.9080645161290327e-05, "loss": 0.3249, "step": 6485 }, { "epoch": 41.80883534136546, "grad_norm": 17.277000427246094, "learning_rate": 2.907741935483871e-05, "loss": 0.5042, "step": 6486 }, { "epoch": 41.81526104417671, "grad_norm": 26.382888793945312, "learning_rate": 2.9074193548387095e-05, "loss": 0.3903, "step": 6487 }, { "epoch": 41.82168674698795, "grad_norm": 37.09638595581055, "learning_rate": 2.9070967741935484e-05, "loss": 0.4239, "step": 6488 }, { "epoch": 41.8281124497992, "grad_norm": 15.411828994750977, "learning_rate": 2.9067741935483873e-05, "loss": 0.213, "step": 6489 }, { "epoch": 41.83453815261044, "grad_norm": 2.6462810039520264, "learning_rate": 2.906451612903226e-05, "loss": 0.505, "step": 6490 }, { "epoch": 41.840963855421684, "grad_norm": 27.357542037963867, "learning_rate": 2.9061290322580648e-05, "loss": 0.5308, "step": 6491 }, { "epoch": 41.84738955823293, "grad_norm": 12.871070861816406, "learning_rate": 2.9058064516129037e-05, "loss": 0.3455, "step": 6492 }, { "epoch": 41.853815261044176, "grad_norm": 14.102704048156738, "learning_rate": 2.9054838709677423e-05, "loss": 0.5642, "step": 6493 }, { "epoch": 41.860240963855425, "grad_norm": 3.8940041065216064, "learning_rate": 2.9051612903225805e-05, "loss": 0.2688, "step": 6494 }, { "epoch": 41.86666666666667, "grad_norm": 16.313518524169922, "learning_rate": 2.9048387096774194e-05, "loss": 0.6037, "step": 6495 }, { "epoch": 41.87309236947791, "grad_norm": 12.367256164550781, "learning_rate": 2.904516129032258e-05, "loss": 0.525, "step": 6496 }, { "epoch": 41.87951807228916, "grad_norm": 5.579915523529053, "learning_rate": 2.904193548387097e-05, "loss": 0.1067, "step": 6497 }, { "epoch": 41.8859437751004, "grad_norm": 12.912336349487305, "learning_rate": 2.9038709677419358e-05, "loss": 0.3033, "step": 6498 }, { "epoch": 41.89236947791164, "grad_norm": 12.584343910217285, "learning_rate": 2.9035483870967744e-05, "loss": 0.4731, "step": 6499 }, { "epoch": 41.89879518072289, "grad_norm": 39.62017822265625, "learning_rate": 2.9032258064516133e-05, "loss": 0.3506, "step": 6500 }, { "epoch": 41.905220883534135, "grad_norm": 37.86575698852539, "learning_rate": 2.902903225806452e-05, "loss": 0.1906, "step": 6501 }, { "epoch": 41.911646586345384, "grad_norm": 4.321551322937012, "learning_rate": 2.90258064516129e-05, "loss": 0.2251, "step": 6502 }, { "epoch": 41.918072289156626, "grad_norm": 20.599306106567383, "learning_rate": 2.902258064516129e-05, "loss": 0.5727, "step": 6503 }, { "epoch": 41.92449799196787, "grad_norm": 8.284601211547852, "learning_rate": 2.901935483870968e-05, "loss": 0.3208, "step": 6504 }, { "epoch": 41.93092369477912, "grad_norm": 27.536115646362305, "learning_rate": 2.9016129032258065e-05, "loss": 0.3425, "step": 6505 }, { "epoch": 41.93734939759036, "grad_norm": 12.899789810180664, "learning_rate": 2.9012903225806454e-05, "loss": 0.1652, "step": 6506 }, { "epoch": 41.94377510040161, "grad_norm": 29.432279586791992, "learning_rate": 2.9009677419354843e-05, "loss": 0.279, "step": 6507 }, { "epoch": 41.95020080321285, "grad_norm": 25.048315048217773, "learning_rate": 2.900645161290323e-05, "loss": 0.2825, "step": 6508 }, { "epoch": 41.956626506024094, "grad_norm": 38.12482833862305, "learning_rate": 2.900322580645161e-05, "loss": 0.5903, "step": 6509 }, { "epoch": 41.96305220883534, "grad_norm": 8.797537803649902, "learning_rate": 2.9e-05, "loss": 0.5009, "step": 6510 }, { "epoch": 41.969477911646585, "grad_norm": 8.04027271270752, "learning_rate": 2.8996774193548386e-05, "loss": 0.3116, "step": 6511 }, { "epoch": 41.975903614457835, "grad_norm": 4.137181758880615, "learning_rate": 2.8993548387096775e-05, "loss": 0.402, "step": 6512 }, { "epoch": 41.98232931726908, "grad_norm": 4.612218379974365, "learning_rate": 2.8990322580645164e-05, "loss": 0.329, "step": 6513 }, { "epoch": 41.98875502008032, "grad_norm": 11.837823867797852, "learning_rate": 2.898709677419355e-05, "loss": 0.6103, "step": 6514 }, { "epoch": 41.99518072289157, "grad_norm": 16.795251846313477, "learning_rate": 2.898387096774194e-05, "loss": 0.453, "step": 6515 }, { "epoch": 42.00481927710843, "grad_norm": 19.423694610595703, "learning_rate": 2.8980645161290325e-05, "loss": 0.3714, "step": 6516 }, { "epoch": 42.01124497991968, "grad_norm": 9.492229461669922, "learning_rate": 2.8977419354838707e-05, "loss": 0.294, "step": 6517 }, { "epoch": 42.01767068273092, "grad_norm": 3.647932291030884, "learning_rate": 2.8974193548387096e-05, "loss": 0.2929, "step": 6518 }, { "epoch": 42.024096385542165, "grad_norm": 13.90912914276123, "learning_rate": 2.8970967741935485e-05, "loss": 0.8038, "step": 6519 }, { "epoch": 42.030522088353415, "grad_norm": 13.989714622497559, "learning_rate": 2.896774193548387e-05, "loss": 0.2836, "step": 6520 }, { "epoch": 42.03694779116466, "grad_norm": 3.640889883041382, "learning_rate": 2.896451612903226e-05, "loss": 0.4456, "step": 6521 }, { "epoch": 42.043373493975906, "grad_norm": 2.7626760005950928, "learning_rate": 2.896129032258065e-05, "loss": 0.2346, "step": 6522 }, { "epoch": 42.04979919678715, "grad_norm": 25.356672286987305, "learning_rate": 2.8958064516129035e-05, "loss": 0.2201, "step": 6523 }, { "epoch": 42.05622489959839, "grad_norm": 24.749866485595703, "learning_rate": 2.8954838709677424e-05, "loss": 0.2817, "step": 6524 }, { "epoch": 42.06265060240964, "grad_norm": 14.178688049316406, "learning_rate": 2.8951612903225806e-05, "loss": 0.3063, "step": 6525 }, { "epoch": 42.06907630522088, "grad_norm": 8.941123008728027, "learning_rate": 2.8948387096774192e-05, "loss": 0.3612, "step": 6526 }, { "epoch": 42.07550200803213, "grad_norm": 6.763352394104004, "learning_rate": 2.894516129032258e-05, "loss": 0.2731, "step": 6527 }, { "epoch": 42.081927710843374, "grad_norm": 17.24951934814453, "learning_rate": 2.894193548387097e-05, "loss": 0.1412, "step": 6528 }, { "epoch": 42.088353413654616, "grad_norm": 10.621767044067383, "learning_rate": 2.8938709677419356e-05, "loss": 0.3396, "step": 6529 }, { "epoch": 42.094779116465865, "grad_norm": 7.4521098136901855, "learning_rate": 2.8935483870967745e-05, "loss": 0.2537, "step": 6530 }, { "epoch": 42.10120481927711, "grad_norm": 28.525728225708008, "learning_rate": 2.8932258064516134e-05, "loss": 0.5647, "step": 6531 }, { "epoch": 42.10763052208836, "grad_norm": 13.039239883422852, "learning_rate": 2.892903225806452e-05, "loss": 0.6092, "step": 6532 }, { "epoch": 42.1140562248996, "grad_norm": 5.438146591186523, "learning_rate": 2.8925806451612902e-05, "loss": 0.1866, "step": 6533 }, { "epoch": 42.12048192771084, "grad_norm": 16.92996597290039, "learning_rate": 2.892258064516129e-05, "loss": 0.1527, "step": 6534 }, { "epoch": 42.12690763052209, "grad_norm": 29.660860061645508, "learning_rate": 2.8919354838709677e-05, "loss": 0.4872, "step": 6535 }, { "epoch": 42.13333333333333, "grad_norm": 24.270360946655273, "learning_rate": 2.8916129032258066e-05, "loss": 0.2678, "step": 6536 }, { "epoch": 42.139759036144575, "grad_norm": 5.05695104598999, "learning_rate": 2.8912903225806455e-05, "loss": 0.5839, "step": 6537 }, { "epoch": 42.146184738955824, "grad_norm": 46.3470573425293, "learning_rate": 2.890967741935484e-05, "loss": 0.4857, "step": 6538 }, { "epoch": 42.15261044176707, "grad_norm": 27.042503356933594, "learning_rate": 2.890645161290323e-05, "loss": 0.3998, "step": 6539 }, { "epoch": 42.159036144578316, "grad_norm": 20.670459747314453, "learning_rate": 2.8903225806451615e-05, "loss": 0.4623, "step": 6540 }, { "epoch": 42.16546184738956, "grad_norm": 10.695233345031738, "learning_rate": 2.8899999999999998e-05, "loss": 0.3, "step": 6541 }, { "epoch": 42.1718875502008, "grad_norm": 7.559938430786133, "learning_rate": 2.8896774193548387e-05, "loss": 0.4516, "step": 6542 }, { "epoch": 42.17831325301205, "grad_norm": 49.97371292114258, "learning_rate": 2.8893548387096776e-05, "loss": 0.5807, "step": 6543 }, { "epoch": 42.18473895582329, "grad_norm": 12.84338092803955, "learning_rate": 2.889032258064516e-05, "loss": 0.4465, "step": 6544 }, { "epoch": 42.19116465863454, "grad_norm": 5.99408483505249, "learning_rate": 2.888709677419355e-05, "loss": 0.1534, "step": 6545 }, { "epoch": 42.19759036144578, "grad_norm": 3.9121546745300293, "learning_rate": 2.888387096774194e-05, "loss": 0.2868, "step": 6546 }, { "epoch": 42.204016064257026, "grad_norm": 7.14483118057251, "learning_rate": 2.8880645161290326e-05, "loss": 0.2493, "step": 6547 }, { "epoch": 42.210441767068275, "grad_norm": 126.54428100585938, "learning_rate": 2.8877419354838708e-05, "loss": 0.5644, "step": 6548 }, { "epoch": 42.21686746987952, "grad_norm": 4.993978500366211, "learning_rate": 2.8874193548387097e-05, "loss": 0.3273, "step": 6549 }, { "epoch": 42.22329317269077, "grad_norm": 22.117237091064453, "learning_rate": 2.8870967741935483e-05, "loss": 0.2615, "step": 6550 }, { "epoch": 42.22971887550201, "grad_norm": 10.396655082702637, "learning_rate": 2.8867741935483872e-05, "loss": 0.3633, "step": 6551 }, { "epoch": 42.23614457831325, "grad_norm": 12.159486770629883, "learning_rate": 2.886451612903226e-05, "loss": 0.5671, "step": 6552 }, { "epoch": 42.2425702811245, "grad_norm": 8.943056106567383, "learning_rate": 2.8861290322580647e-05, "loss": 0.3339, "step": 6553 }, { "epoch": 42.24899598393574, "grad_norm": 11.198452949523926, "learning_rate": 2.8858064516129036e-05, "loss": 0.4929, "step": 6554 }, { "epoch": 42.255421686746985, "grad_norm": 11.710260391235352, "learning_rate": 2.8854838709677425e-05, "loss": 0.4732, "step": 6555 }, { "epoch": 42.261847389558234, "grad_norm": 8.742714881896973, "learning_rate": 2.8851612903225804e-05, "loss": 0.4535, "step": 6556 }, { "epoch": 42.268273092369476, "grad_norm": 6.768779277801514, "learning_rate": 2.8848387096774193e-05, "loss": 0.4719, "step": 6557 }, { "epoch": 42.274698795180726, "grad_norm": 38.314552307128906, "learning_rate": 2.8845161290322582e-05, "loss": 0.5176, "step": 6558 }, { "epoch": 42.28112449799197, "grad_norm": 7.618805885314941, "learning_rate": 2.8841935483870968e-05, "loss": 0.6727, "step": 6559 }, { "epoch": 42.28755020080321, "grad_norm": 17.493375778198242, "learning_rate": 2.8838709677419357e-05, "loss": 0.3869, "step": 6560 }, { "epoch": 42.29397590361446, "grad_norm": 35.73940658569336, "learning_rate": 2.8835483870967746e-05, "loss": 0.6107, "step": 6561 }, { "epoch": 42.3004016064257, "grad_norm": 22.030193328857422, "learning_rate": 2.883225806451613e-05, "loss": 0.2901, "step": 6562 }, { "epoch": 42.30682730923695, "grad_norm": 20.12778663635254, "learning_rate": 2.882903225806452e-05, "loss": 0.5775, "step": 6563 }, { "epoch": 42.31325301204819, "grad_norm": 4.758580207824707, "learning_rate": 2.8825806451612903e-05, "loss": 0.2941, "step": 6564 }, { "epoch": 42.319678714859435, "grad_norm": 4.488368034362793, "learning_rate": 2.882258064516129e-05, "loss": 0.2335, "step": 6565 }, { "epoch": 42.326104417670685, "grad_norm": 21.24589729309082, "learning_rate": 2.8819354838709678e-05, "loss": 0.3435, "step": 6566 }, { "epoch": 42.33253012048193, "grad_norm": 3.197713851928711, "learning_rate": 2.8816129032258067e-05, "loss": 0.0899, "step": 6567 }, { "epoch": 42.338955823293176, "grad_norm": 14.932108879089355, "learning_rate": 2.8812903225806452e-05, "loss": 0.2832, "step": 6568 }, { "epoch": 42.34538152610442, "grad_norm": 52.966434478759766, "learning_rate": 2.880967741935484e-05, "loss": 0.7154, "step": 6569 }, { "epoch": 42.35180722891566, "grad_norm": 15.930668830871582, "learning_rate": 2.880645161290323e-05, "loss": 0.2447, "step": 6570 }, { "epoch": 42.35823293172691, "grad_norm": 25.298585891723633, "learning_rate": 2.8803225806451616e-05, "loss": 0.7442, "step": 6571 }, { "epoch": 42.36465863453815, "grad_norm": 13.741133689880371, "learning_rate": 2.88e-05, "loss": 0.3088, "step": 6572 }, { "epoch": 42.371084337349394, "grad_norm": 7.673280239105225, "learning_rate": 2.8796774193548388e-05, "loss": 0.1742, "step": 6573 }, { "epoch": 42.377510040160644, "grad_norm": 6.676046371459961, "learning_rate": 2.8793548387096773e-05, "loss": 0.534, "step": 6574 }, { "epoch": 42.383935742971886, "grad_norm": 21.3979434967041, "learning_rate": 2.8790322580645163e-05, "loss": 0.3777, "step": 6575 }, { "epoch": 42.390361445783135, "grad_norm": 5.632671356201172, "learning_rate": 2.878709677419355e-05, "loss": 0.4078, "step": 6576 }, { "epoch": 42.39678714859438, "grad_norm": 11.85025405883789, "learning_rate": 2.8783870967741937e-05, "loss": 0.5814, "step": 6577 }, { "epoch": 42.40321285140562, "grad_norm": 3.1556146144866943, "learning_rate": 2.8780645161290326e-05, "loss": 0.3839, "step": 6578 }, { "epoch": 42.40963855421687, "grad_norm": 57.29412078857422, "learning_rate": 2.877741935483871e-05, "loss": 0.6965, "step": 6579 }, { "epoch": 42.41606425702811, "grad_norm": 21.540342330932617, "learning_rate": 2.8774193548387095e-05, "loss": 0.2558, "step": 6580 }, { "epoch": 42.42248995983936, "grad_norm": 11.79300308227539, "learning_rate": 2.8770967741935484e-05, "loss": 0.3529, "step": 6581 }, { "epoch": 42.4289156626506, "grad_norm": 14.630714416503906, "learning_rate": 2.8767741935483873e-05, "loss": 0.5035, "step": 6582 }, { "epoch": 42.435341365461845, "grad_norm": 9.79832649230957, "learning_rate": 2.876451612903226e-05, "loss": 0.229, "step": 6583 }, { "epoch": 42.441767068273094, "grad_norm": 23.549869537353516, "learning_rate": 2.8761290322580648e-05, "loss": 0.3643, "step": 6584 }, { "epoch": 42.44819277108434, "grad_norm": 13.715536117553711, "learning_rate": 2.8758064516129037e-05, "loss": 0.5212, "step": 6585 }, { "epoch": 42.454618473895586, "grad_norm": 3.749884605407715, "learning_rate": 2.8754838709677422e-05, "loss": 0.2688, "step": 6586 }, { "epoch": 42.46104417670683, "grad_norm": 31.969261169433594, "learning_rate": 2.8751612903225805e-05, "loss": 0.2151, "step": 6587 }, { "epoch": 42.46746987951807, "grad_norm": 16.28118133544922, "learning_rate": 2.8748387096774194e-05, "loss": 0.4813, "step": 6588 }, { "epoch": 42.47389558232932, "grad_norm": 58.193092346191406, "learning_rate": 2.874516129032258e-05, "loss": 0.5579, "step": 6589 }, { "epoch": 42.48032128514056, "grad_norm": 20.35835838317871, "learning_rate": 2.874193548387097e-05, "loss": 0.313, "step": 6590 }, { "epoch": 42.486746987951804, "grad_norm": 22.954818725585938, "learning_rate": 2.8738709677419358e-05, "loss": 0.6557, "step": 6591 }, { "epoch": 42.49317269076305, "grad_norm": 15.608080863952637, "learning_rate": 2.8735483870967743e-05, "loss": 0.2087, "step": 6592 }, { "epoch": 42.499598393574296, "grad_norm": 5.526078701019287, "learning_rate": 2.8732258064516132e-05, "loss": 0.3361, "step": 6593 }, { "epoch": 42.506024096385545, "grad_norm": 14.25655746459961, "learning_rate": 2.872903225806452e-05, "loss": 0.1641, "step": 6594 }, { "epoch": 42.51244979919679, "grad_norm": 2.661907434463501, "learning_rate": 2.8725806451612904e-05, "loss": 0.2818, "step": 6595 }, { "epoch": 42.51887550200803, "grad_norm": 32.00822448730469, "learning_rate": 2.872258064516129e-05, "loss": 0.386, "step": 6596 }, { "epoch": 42.52530120481928, "grad_norm": 20.089019775390625, "learning_rate": 2.871935483870968e-05, "loss": 0.4198, "step": 6597 }, { "epoch": 42.53172690763052, "grad_norm": 15.728816986083984, "learning_rate": 2.8716129032258064e-05, "loss": 0.7435, "step": 6598 }, { "epoch": 42.53815261044177, "grad_norm": 20.86606788635254, "learning_rate": 2.8712903225806453e-05, "loss": 0.5141, "step": 6599 }, { "epoch": 42.54457831325301, "grad_norm": 9.863104820251465, "learning_rate": 2.8709677419354843e-05, "loss": 0.387, "step": 6600 }, { "epoch": 42.551004016064255, "grad_norm": 17.823741912841797, "learning_rate": 2.8706451612903228e-05, "loss": 0.2621, "step": 6601 }, { "epoch": 42.557429718875504, "grad_norm": 3.486635684967041, "learning_rate": 2.8703225806451617e-05, "loss": 0.1442, "step": 6602 }, { "epoch": 42.563855421686746, "grad_norm": 18.411460876464844, "learning_rate": 2.87e-05, "loss": 0.1615, "step": 6603 }, { "epoch": 42.570281124497996, "grad_norm": 22.654903411865234, "learning_rate": 2.8696774193548385e-05, "loss": 0.3527, "step": 6604 }, { "epoch": 42.57670682730924, "grad_norm": 11.981880187988281, "learning_rate": 2.8693548387096774e-05, "loss": 0.218, "step": 6605 }, { "epoch": 42.58313253012048, "grad_norm": 21.71693992614746, "learning_rate": 2.8690322580645164e-05, "loss": 0.7003, "step": 6606 }, { "epoch": 42.58955823293173, "grad_norm": 12.591666221618652, "learning_rate": 2.868709677419355e-05, "loss": 0.3669, "step": 6607 }, { "epoch": 42.59598393574297, "grad_norm": 19.23882484436035, "learning_rate": 2.868387096774194e-05, "loss": 0.6059, "step": 6608 }, { "epoch": 42.602409638554214, "grad_norm": 15.993809700012207, "learning_rate": 2.8680645161290327e-05, "loss": 0.3144, "step": 6609 }, { "epoch": 42.60883534136546, "grad_norm": 5.413948059082031, "learning_rate": 2.8677419354838713e-05, "loss": 0.2038, "step": 6610 }, { "epoch": 42.615261044176705, "grad_norm": 12.532094955444336, "learning_rate": 2.8674193548387095e-05, "loss": 0.5906, "step": 6611 }, { "epoch": 42.621686746987955, "grad_norm": 3.497258424758911, "learning_rate": 2.8670967741935485e-05, "loss": 0.2743, "step": 6612 }, { "epoch": 42.6281124497992, "grad_norm": 3.668253183364868, "learning_rate": 2.866774193548387e-05, "loss": 0.3167, "step": 6613 }, { "epoch": 42.63453815261044, "grad_norm": 32.78018569946289, "learning_rate": 2.866451612903226e-05, "loss": 0.3331, "step": 6614 }, { "epoch": 42.64096385542169, "grad_norm": 10.806419372558594, "learning_rate": 2.866129032258065e-05, "loss": 0.5142, "step": 6615 }, { "epoch": 42.64738955823293, "grad_norm": 11.463756561279297, "learning_rate": 2.8658064516129034e-05, "loss": 0.336, "step": 6616 }, { "epoch": 42.65381526104418, "grad_norm": 3.9553205966949463, "learning_rate": 2.8654838709677423e-05, "loss": 0.2766, "step": 6617 }, { "epoch": 42.66024096385542, "grad_norm": 12.904817581176758, "learning_rate": 2.8651612903225806e-05, "loss": 0.3738, "step": 6618 }, { "epoch": 42.666666666666664, "grad_norm": 28.580669403076172, "learning_rate": 2.864838709677419e-05, "loss": 0.4438, "step": 6619 }, { "epoch": 42.673092369477914, "grad_norm": 15.65837574005127, "learning_rate": 2.864516129032258e-05, "loss": 0.5993, "step": 6620 }, { "epoch": 42.679518072289156, "grad_norm": 15.734526634216309, "learning_rate": 2.864193548387097e-05, "loss": 0.4239, "step": 6621 }, { "epoch": 42.6859437751004, "grad_norm": 22.69672393798828, "learning_rate": 2.8638709677419355e-05, "loss": 0.3902, "step": 6622 }, { "epoch": 42.69236947791165, "grad_norm": 3.5264434814453125, "learning_rate": 2.8635483870967744e-05, "loss": 0.3996, "step": 6623 }, { "epoch": 42.69879518072289, "grad_norm": 25.34654426574707, "learning_rate": 2.8632258064516133e-05, "loss": 0.4029, "step": 6624 }, { "epoch": 42.70522088353414, "grad_norm": 9.282565116882324, "learning_rate": 2.862903225806452e-05, "loss": 0.3702, "step": 6625 }, { "epoch": 42.71164658634538, "grad_norm": 29.906553268432617, "learning_rate": 2.86258064516129e-05, "loss": 0.7075, "step": 6626 }, { "epoch": 42.71807228915662, "grad_norm": 10.461369514465332, "learning_rate": 2.862258064516129e-05, "loss": 0.4917, "step": 6627 }, { "epoch": 42.72449799196787, "grad_norm": 4.153209686279297, "learning_rate": 2.8619354838709676e-05, "loss": 0.2322, "step": 6628 }, { "epoch": 42.730923694779115, "grad_norm": 17.49810218811035, "learning_rate": 2.8616129032258065e-05, "loss": 0.582, "step": 6629 }, { "epoch": 42.737349397590364, "grad_norm": 7.352227210998535, "learning_rate": 2.8612903225806454e-05, "loss": 0.2683, "step": 6630 }, { "epoch": 42.743775100401606, "grad_norm": 9.718156814575195, "learning_rate": 2.860967741935484e-05, "loss": 0.3907, "step": 6631 }, { "epoch": 42.75020080321285, "grad_norm": 21.936248779296875, "learning_rate": 2.860645161290323e-05, "loss": 0.5127, "step": 6632 }, { "epoch": 42.7566265060241, "grad_norm": 20.134647369384766, "learning_rate": 2.8603225806451618e-05, "loss": 0.2973, "step": 6633 }, { "epoch": 42.76305220883534, "grad_norm": 42.430030822753906, "learning_rate": 2.86e-05, "loss": 0.6529, "step": 6634 }, { "epoch": 42.76947791164659, "grad_norm": 9.5061616897583, "learning_rate": 2.8596774193548386e-05, "loss": 0.3625, "step": 6635 }, { "epoch": 42.77590361445783, "grad_norm": 9.63257122039795, "learning_rate": 2.8593548387096775e-05, "loss": 0.1802, "step": 6636 }, { "epoch": 42.782329317269074, "grad_norm": 11.573139190673828, "learning_rate": 2.859032258064516e-05, "loss": 0.4291, "step": 6637 }, { "epoch": 42.78875502008032, "grad_norm": 8.03136157989502, "learning_rate": 2.858709677419355e-05, "loss": 0.4794, "step": 6638 }, { "epoch": 42.795180722891565, "grad_norm": 13.316296577453613, "learning_rate": 2.858387096774194e-05, "loss": 0.3648, "step": 6639 }, { "epoch": 42.801606425702815, "grad_norm": 6.12437105178833, "learning_rate": 2.8580645161290325e-05, "loss": 0.2912, "step": 6640 }, { "epoch": 42.80803212851406, "grad_norm": 8.676396369934082, "learning_rate": 2.8577419354838714e-05, "loss": 0.4833, "step": 6641 }, { "epoch": 42.8144578313253, "grad_norm": 9.081809997558594, "learning_rate": 2.8574193548387096e-05, "loss": 0.4054, "step": 6642 }, { "epoch": 42.82088353413655, "grad_norm": 6.590710639953613, "learning_rate": 2.8570967741935482e-05, "loss": 0.4321, "step": 6643 }, { "epoch": 42.82730923694779, "grad_norm": 3.0779361724853516, "learning_rate": 2.856774193548387e-05, "loss": 0.2284, "step": 6644 }, { "epoch": 42.83373493975903, "grad_norm": 4.258868217468262, "learning_rate": 2.856451612903226e-05, "loss": 0.1286, "step": 6645 }, { "epoch": 42.84016064257028, "grad_norm": 5.895387172698975, "learning_rate": 2.8561290322580646e-05, "loss": 0.2678, "step": 6646 }, { "epoch": 42.846586345381525, "grad_norm": 2.163433790206909, "learning_rate": 2.8558064516129035e-05, "loss": 0.2758, "step": 6647 }, { "epoch": 42.853012048192774, "grad_norm": 18.054523468017578, "learning_rate": 2.8554838709677424e-05, "loss": 0.4267, "step": 6648 }, { "epoch": 42.859437751004016, "grad_norm": 19.89259147644043, "learning_rate": 2.8551612903225807e-05, "loss": 0.4594, "step": 6649 }, { "epoch": 42.86586345381526, "grad_norm": 17.33949851989746, "learning_rate": 2.8548387096774192e-05, "loss": 0.3435, "step": 6650 }, { "epoch": 42.87228915662651, "grad_norm": 23.780107498168945, "learning_rate": 2.854516129032258e-05, "loss": 0.4056, "step": 6651 }, { "epoch": 42.87871485943775, "grad_norm": 10.372529983520508, "learning_rate": 2.8541935483870967e-05, "loss": 0.2239, "step": 6652 }, { "epoch": 42.885140562249, "grad_norm": 4.318125247955322, "learning_rate": 2.8538709677419356e-05, "loss": 0.3175, "step": 6653 }, { "epoch": 42.89156626506024, "grad_norm": 13.448472023010254, "learning_rate": 2.8535483870967745e-05, "loss": 0.3122, "step": 6654 }, { "epoch": 42.897991967871484, "grad_norm": 38.07373046875, "learning_rate": 2.853225806451613e-05, "loss": 0.2741, "step": 6655 }, { "epoch": 42.90441767068273, "grad_norm": 9.852231979370117, "learning_rate": 2.852903225806452e-05, "loss": 0.2354, "step": 6656 }, { "epoch": 42.910843373493975, "grad_norm": 24.564279556274414, "learning_rate": 2.8525806451612902e-05, "loss": 0.3465, "step": 6657 }, { "epoch": 42.91726907630522, "grad_norm": 7.254668235778809, "learning_rate": 2.852258064516129e-05, "loss": 0.2273, "step": 6658 }, { "epoch": 42.92369477911647, "grad_norm": 9.012991905212402, "learning_rate": 2.8519354838709677e-05, "loss": 0.2883, "step": 6659 }, { "epoch": 42.93012048192771, "grad_norm": 4.570892333984375, "learning_rate": 2.8516129032258066e-05, "loss": 0.2372, "step": 6660 }, { "epoch": 42.93654618473896, "grad_norm": 3.266794204711914, "learning_rate": 2.8512903225806452e-05, "loss": 0.3045, "step": 6661 }, { "epoch": 42.9429718875502, "grad_norm": 8.601818084716797, "learning_rate": 2.850967741935484e-05, "loss": 0.4291, "step": 6662 }, { "epoch": 42.94939759036144, "grad_norm": 28.675874710083008, "learning_rate": 2.850645161290323e-05, "loss": 0.3389, "step": 6663 }, { "epoch": 42.95582329317269, "grad_norm": 6.831585884094238, "learning_rate": 2.8503225806451616e-05, "loss": 0.2434, "step": 6664 }, { "epoch": 42.962248995983934, "grad_norm": 4.693180084228516, "learning_rate": 2.8499999999999998e-05, "loss": 0.3191, "step": 6665 }, { "epoch": 42.96867469879518, "grad_norm": 14.722373962402344, "learning_rate": 2.8496774193548387e-05, "loss": 0.4943, "step": 6666 }, { "epoch": 42.975100401606426, "grad_norm": 12.714482307434082, "learning_rate": 2.8493548387096773e-05, "loss": 0.5414, "step": 6667 }, { "epoch": 42.98152610441767, "grad_norm": 3.1973443031311035, "learning_rate": 2.8490322580645162e-05, "loss": 0.1836, "step": 6668 }, { "epoch": 42.98795180722892, "grad_norm": 12.558316230773926, "learning_rate": 2.848709677419355e-05, "loss": 0.2773, "step": 6669 }, { "epoch": 42.99437751004016, "grad_norm": 6.685543537139893, "learning_rate": 2.8483870967741937e-05, "loss": 0.3668, "step": 6670 }, { "epoch": 43.00401606425703, "grad_norm": 18.2491455078125, "learning_rate": 2.8480645161290326e-05, "loss": 0.6725, "step": 6671 }, { "epoch": 43.01044176706827, "grad_norm": 9.413459777832031, "learning_rate": 2.8477419354838715e-05, "loss": 0.4961, "step": 6672 }, { "epoch": 43.01686746987952, "grad_norm": 5.445452690124512, "learning_rate": 2.8474193548387097e-05, "loss": 0.3344, "step": 6673 }, { "epoch": 43.023293172690764, "grad_norm": 16.14012336730957, "learning_rate": 2.8470967741935483e-05, "loss": 0.3047, "step": 6674 }, { "epoch": 43.029718875502006, "grad_norm": 23.61553192138672, "learning_rate": 2.8467741935483872e-05, "loss": 0.3597, "step": 6675 }, { "epoch": 43.036144578313255, "grad_norm": 8.387807846069336, "learning_rate": 2.8464516129032258e-05, "loss": 0.5431, "step": 6676 }, { "epoch": 43.0425702811245, "grad_norm": 13.433372497558594, "learning_rate": 2.8461290322580647e-05, "loss": 0.3047, "step": 6677 }, { "epoch": 43.04899598393574, "grad_norm": 9.207900047302246, "learning_rate": 2.8458064516129036e-05, "loss": 0.4129, "step": 6678 }, { "epoch": 43.05542168674699, "grad_norm": 15.730074882507324, "learning_rate": 2.8454838709677422e-05, "loss": 0.4095, "step": 6679 }, { "epoch": 43.06184738955823, "grad_norm": 16.317651748657227, "learning_rate": 2.845161290322581e-05, "loss": 0.5563, "step": 6680 }, { "epoch": 43.06827309236948, "grad_norm": 14.295950889587402, "learning_rate": 2.8448387096774193e-05, "loss": 0.2825, "step": 6681 }, { "epoch": 43.07469879518072, "grad_norm": 26.057340621948242, "learning_rate": 2.844516129032258e-05, "loss": 0.4753, "step": 6682 }, { "epoch": 43.081124497991965, "grad_norm": 5.296412944793701, "learning_rate": 2.8441935483870968e-05, "loss": 0.289, "step": 6683 }, { "epoch": 43.087550200803214, "grad_norm": 2.319585084915161, "learning_rate": 2.8438709677419357e-05, "loss": 0.2048, "step": 6684 }, { "epoch": 43.09397590361446, "grad_norm": 4.2651214599609375, "learning_rate": 2.8435483870967743e-05, "loss": 0.2031, "step": 6685 }, { "epoch": 43.100401606425706, "grad_norm": 3.668553113937378, "learning_rate": 2.8432258064516132e-05, "loss": 0.1822, "step": 6686 }, { "epoch": 43.10682730923695, "grad_norm": 41.677520751953125, "learning_rate": 2.842903225806452e-05, "loss": 0.532, "step": 6687 }, { "epoch": 43.11325301204819, "grad_norm": 8.05997371673584, "learning_rate": 2.8425806451612903e-05, "loss": 0.3141, "step": 6688 }, { "epoch": 43.11967871485944, "grad_norm": 8.000168800354004, "learning_rate": 2.842258064516129e-05, "loss": 0.368, "step": 6689 }, { "epoch": 43.12610441767068, "grad_norm": 14.222890853881836, "learning_rate": 2.8419354838709678e-05, "loss": 0.3414, "step": 6690 }, { "epoch": 43.13253012048193, "grad_norm": 8.885315895080566, "learning_rate": 2.8416129032258064e-05, "loss": 0.3185, "step": 6691 }, { "epoch": 43.13895582329317, "grad_norm": 12.043113708496094, "learning_rate": 2.8412903225806453e-05, "loss": 0.5638, "step": 6692 }, { "epoch": 43.145381526104416, "grad_norm": 10.388110160827637, "learning_rate": 2.8409677419354842e-05, "loss": 0.3335, "step": 6693 }, { "epoch": 43.151807228915665, "grad_norm": 5.933961391448975, "learning_rate": 2.8406451612903228e-05, "loss": 0.4252, "step": 6694 }, { "epoch": 43.15823293172691, "grad_norm": 14.904802322387695, "learning_rate": 2.8403225806451617e-05, "loss": 0.6857, "step": 6695 }, { "epoch": 43.16465863453815, "grad_norm": 16.01428985595703, "learning_rate": 2.84e-05, "loss": 0.4258, "step": 6696 }, { "epoch": 43.1710843373494, "grad_norm": 15.524032592773438, "learning_rate": 2.8396774193548388e-05, "loss": 0.6646, "step": 6697 }, { "epoch": 43.17751004016064, "grad_norm": 11.796770095825195, "learning_rate": 2.8393548387096774e-05, "loss": 0.4222, "step": 6698 }, { "epoch": 43.18393574297189, "grad_norm": 23.041919708251953, "learning_rate": 2.8390322580645163e-05, "loss": 0.6412, "step": 6699 }, { "epoch": 43.19036144578313, "grad_norm": 14.940279960632324, "learning_rate": 2.838709677419355e-05, "loss": 0.4133, "step": 6700 }, { "epoch": 43.196787148594375, "grad_norm": 4.074092388153076, "learning_rate": 2.8383870967741938e-05, "loss": 0.2652, "step": 6701 }, { "epoch": 43.203212851405624, "grad_norm": 10.873602867126465, "learning_rate": 2.8380645161290327e-05, "loss": 0.296, "step": 6702 }, { "epoch": 43.209638554216866, "grad_norm": 21.940494537353516, "learning_rate": 2.8377419354838713e-05, "loss": 0.4665, "step": 6703 }, { "epoch": 43.216064257028115, "grad_norm": 3.5396106243133545, "learning_rate": 2.8374193548387095e-05, "loss": 0.3641, "step": 6704 }, { "epoch": 43.22248995983936, "grad_norm": 9.491764068603516, "learning_rate": 2.8370967741935484e-05, "loss": 0.338, "step": 6705 }, { "epoch": 43.2289156626506, "grad_norm": 9.054986000061035, "learning_rate": 2.836774193548387e-05, "loss": 0.2483, "step": 6706 }, { "epoch": 43.23534136546185, "grad_norm": 18.397733688354492, "learning_rate": 2.836451612903226e-05, "loss": 0.3552, "step": 6707 }, { "epoch": 43.24176706827309, "grad_norm": 26.184112548828125, "learning_rate": 2.8361290322580648e-05, "loss": 0.3907, "step": 6708 }, { "epoch": 43.24819277108434, "grad_norm": 17.460962295532227, "learning_rate": 2.8358064516129034e-05, "loss": 0.3126, "step": 6709 }, { "epoch": 43.25461847389558, "grad_norm": 13.50922679901123, "learning_rate": 2.8354838709677423e-05, "loss": 0.2834, "step": 6710 }, { "epoch": 43.261044176706825, "grad_norm": 23.02869987487793, "learning_rate": 2.8351612903225812e-05, "loss": 0.7164, "step": 6711 }, { "epoch": 43.267469879518075, "grad_norm": 20.13482093811035, "learning_rate": 2.8348387096774194e-05, "loss": 0.4084, "step": 6712 }, { "epoch": 43.27389558232932, "grad_norm": 11.480851173400879, "learning_rate": 2.834516129032258e-05, "loss": 0.236, "step": 6713 }, { "epoch": 43.28032128514056, "grad_norm": 2.0088582038879395, "learning_rate": 2.834193548387097e-05, "loss": 0.1697, "step": 6714 }, { "epoch": 43.28674698795181, "grad_norm": 4.0671234130859375, "learning_rate": 2.8338709677419355e-05, "loss": 0.1984, "step": 6715 }, { "epoch": 43.29317269076305, "grad_norm": 8.5120210647583, "learning_rate": 2.8335483870967744e-05, "loss": 0.319, "step": 6716 }, { "epoch": 43.2995983935743, "grad_norm": 10.04362964630127, "learning_rate": 2.8332258064516133e-05, "loss": 0.4216, "step": 6717 }, { "epoch": 43.30602409638554, "grad_norm": 18.794187545776367, "learning_rate": 2.832903225806452e-05, "loss": 0.2895, "step": 6718 }, { "epoch": 43.312449799196784, "grad_norm": 30.463552474975586, "learning_rate": 2.8325806451612908e-05, "loss": 0.4392, "step": 6719 }, { "epoch": 43.318875502008034, "grad_norm": 3.475888252258301, "learning_rate": 2.832258064516129e-05, "loss": 0.3297, "step": 6720 }, { "epoch": 43.325301204819276, "grad_norm": 4.274607181549072, "learning_rate": 2.831935483870968e-05, "loss": 0.1842, "step": 6721 }, { "epoch": 43.331726907630525, "grad_norm": 12.863090515136719, "learning_rate": 2.8316129032258065e-05, "loss": 0.5476, "step": 6722 }, { "epoch": 43.33815261044177, "grad_norm": 29.999561309814453, "learning_rate": 2.8312903225806454e-05, "loss": 0.4945, "step": 6723 }, { "epoch": 43.34457831325301, "grad_norm": 9.853028297424316, "learning_rate": 2.830967741935484e-05, "loss": 0.2157, "step": 6724 }, { "epoch": 43.35100401606426, "grad_norm": 5.254396915435791, "learning_rate": 2.830645161290323e-05, "loss": 0.2493, "step": 6725 }, { "epoch": 43.3574297188755, "grad_norm": 4.600759506225586, "learning_rate": 2.8303225806451618e-05, "loss": 0.2067, "step": 6726 }, { "epoch": 43.36385542168675, "grad_norm": 39.382083892822266, "learning_rate": 2.83e-05, "loss": 0.4498, "step": 6727 }, { "epoch": 43.37028112449799, "grad_norm": 5.9874396324157715, "learning_rate": 2.8296774193548386e-05, "loss": 0.5603, "step": 6728 }, { "epoch": 43.376706827309235, "grad_norm": 27.753080368041992, "learning_rate": 2.8293548387096775e-05, "loss": 0.3967, "step": 6729 }, { "epoch": 43.383132530120484, "grad_norm": 33.88929748535156, "learning_rate": 2.829032258064516e-05, "loss": 0.5982, "step": 6730 }, { "epoch": 43.389558232931726, "grad_norm": 6.377817630767822, "learning_rate": 2.828709677419355e-05, "loss": 0.72, "step": 6731 }, { "epoch": 43.39598393574297, "grad_norm": 25.425432205200195, "learning_rate": 2.828387096774194e-05, "loss": 0.6998, "step": 6732 }, { "epoch": 43.40240963855422, "grad_norm": 14.6969633102417, "learning_rate": 2.8280645161290325e-05, "loss": 0.2498, "step": 6733 }, { "epoch": 43.40883534136546, "grad_norm": 15.658797264099121, "learning_rate": 2.8277419354838714e-05, "loss": 0.4189, "step": 6734 }, { "epoch": 43.41526104417671, "grad_norm": 12.207951545715332, "learning_rate": 2.8274193548387096e-05, "loss": 0.3085, "step": 6735 }, { "epoch": 43.42168674698795, "grad_norm": 14.466145515441895, "learning_rate": 2.8270967741935485e-05, "loss": 0.2746, "step": 6736 }, { "epoch": 43.428112449799194, "grad_norm": 4.593943119049072, "learning_rate": 2.826774193548387e-05, "loss": 0.2333, "step": 6737 }, { "epoch": 43.43453815261044, "grad_norm": 11.59003734588623, "learning_rate": 2.826451612903226e-05, "loss": 0.6601, "step": 6738 }, { "epoch": 43.440963855421685, "grad_norm": 15.316638946533203, "learning_rate": 2.8261290322580646e-05, "loss": 0.6111, "step": 6739 }, { "epoch": 43.447389558232935, "grad_norm": 13.631606101989746, "learning_rate": 2.8258064516129035e-05, "loss": 0.4192, "step": 6740 }, { "epoch": 43.45381526104418, "grad_norm": 13.417119979858398, "learning_rate": 2.8254838709677424e-05, "loss": 0.3091, "step": 6741 }, { "epoch": 43.46024096385542, "grad_norm": 4.775461196899414, "learning_rate": 2.825161290322581e-05, "loss": 0.3039, "step": 6742 }, { "epoch": 43.46666666666667, "grad_norm": 8.746034622192383, "learning_rate": 2.8248387096774192e-05, "loss": 0.3685, "step": 6743 }, { "epoch": 43.47309236947791, "grad_norm": 6.623194217681885, "learning_rate": 2.824516129032258e-05, "loss": 0.497, "step": 6744 }, { "epoch": 43.47951807228916, "grad_norm": 20.85746192932129, "learning_rate": 2.8241935483870967e-05, "loss": 0.4081, "step": 6745 }, { "epoch": 43.4859437751004, "grad_norm": 22.43221664428711, "learning_rate": 2.8238709677419356e-05, "loss": 0.1406, "step": 6746 }, { "epoch": 43.492369477911645, "grad_norm": 20.673870086669922, "learning_rate": 2.8235483870967745e-05, "loss": 0.4375, "step": 6747 }, { "epoch": 43.498795180722894, "grad_norm": 9.968413352966309, "learning_rate": 2.823225806451613e-05, "loss": 0.465, "step": 6748 }, { "epoch": 43.505220883534136, "grad_norm": 15.134805679321289, "learning_rate": 2.822903225806452e-05, "loss": 0.3162, "step": 6749 }, { "epoch": 43.51164658634538, "grad_norm": 8.793292045593262, "learning_rate": 2.822580645161291e-05, "loss": 0.3331, "step": 6750 }, { "epoch": 43.51807228915663, "grad_norm": 9.473102569580078, "learning_rate": 2.822258064516129e-05, "loss": 0.3846, "step": 6751 }, { "epoch": 43.52449799196787, "grad_norm": 31.31662940979004, "learning_rate": 2.8219354838709677e-05, "loss": 0.6424, "step": 6752 }, { "epoch": 43.53092369477912, "grad_norm": 4.94683313369751, "learning_rate": 2.8216129032258066e-05, "loss": 0.1472, "step": 6753 }, { "epoch": 43.53734939759036, "grad_norm": 15.710400581359863, "learning_rate": 2.821290322580645e-05, "loss": 0.1079, "step": 6754 }, { "epoch": 43.543775100401604, "grad_norm": 4.964228630065918, "learning_rate": 2.820967741935484e-05, "loss": 0.3242, "step": 6755 }, { "epoch": 43.55020080321285, "grad_norm": 36.75010299682617, "learning_rate": 2.820645161290323e-05, "loss": 0.3004, "step": 6756 }, { "epoch": 43.556626506024095, "grad_norm": 6.553594589233398, "learning_rate": 2.8203225806451615e-05, "loss": 0.3385, "step": 6757 }, { "epoch": 43.563052208835344, "grad_norm": 4.5034637451171875, "learning_rate": 2.8199999999999998e-05, "loss": 0.3014, "step": 6758 }, { "epoch": 43.56947791164659, "grad_norm": 15.362098693847656, "learning_rate": 2.8196774193548387e-05, "loss": 0.4095, "step": 6759 }, { "epoch": 43.57590361445783, "grad_norm": 21.467561721801758, "learning_rate": 2.8193548387096776e-05, "loss": 0.5842, "step": 6760 }, { "epoch": 43.58232931726908, "grad_norm": 14.817920684814453, "learning_rate": 2.819032258064516e-05, "loss": 0.5157, "step": 6761 }, { "epoch": 43.58875502008032, "grad_norm": 3.483119487762451, "learning_rate": 2.818709677419355e-05, "loss": 0.1283, "step": 6762 }, { "epoch": 43.59518072289157, "grad_norm": 12.376260757446289, "learning_rate": 2.8183870967741936e-05, "loss": 0.3808, "step": 6763 }, { "epoch": 43.60160642570281, "grad_norm": 3.3037521839141846, "learning_rate": 2.8180645161290325e-05, "loss": 0.2832, "step": 6764 }, { "epoch": 43.608032128514054, "grad_norm": 3.2405002117156982, "learning_rate": 2.8177419354838715e-05, "loss": 0.416, "step": 6765 }, { "epoch": 43.6144578313253, "grad_norm": 12.076570510864258, "learning_rate": 2.8174193548387097e-05, "loss": 0.5537, "step": 6766 }, { "epoch": 43.620883534136546, "grad_norm": 28.19305992126465, "learning_rate": 2.8170967741935483e-05, "loss": 0.4467, "step": 6767 }, { "epoch": 43.62730923694779, "grad_norm": 14.162671089172363, "learning_rate": 2.816774193548387e-05, "loss": 0.3503, "step": 6768 }, { "epoch": 43.63373493975904, "grad_norm": 5.748995304107666, "learning_rate": 2.8164516129032257e-05, "loss": 0.088, "step": 6769 }, { "epoch": 43.64016064257028, "grad_norm": 5.971461296081543, "learning_rate": 2.8161290322580646e-05, "loss": 0.1858, "step": 6770 }, { "epoch": 43.64658634538153, "grad_norm": 10.451157569885254, "learning_rate": 2.8158064516129036e-05, "loss": 0.307, "step": 6771 }, { "epoch": 43.65301204819277, "grad_norm": 14.860811233520508, "learning_rate": 2.815483870967742e-05, "loss": 0.1933, "step": 6772 }, { "epoch": 43.65943775100401, "grad_norm": 22.72530746459961, "learning_rate": 2.815161290322581e-05, "loss": 0.2687, "step": 6773 }, { "epoch": 43.66586345381526, "grad_norm": 8.792468070983887, "learning_rate": 2.8148387096774193e-05, "loss": 0.4274, "step": 6774 }, { "epoch": 43.672289156626505, "grad_norm": 6.260234832763672, "learning_rate": 2.8145161290322582e-05, "loss": 0.2984, "step": 6775 }, { "epoch": 43.678714859437754, "grad_norm": 11.928750991821289, "learning_rate": 2.8141935483870967e-05, "loss": 0.3446, "step": 6776 }, { "epoch": 43.685140562248996, "grad_norm": 3.787475347518921, "learning_rate": 2.8138709677419357e-05, "loss": 0.4552, "step": 6777 }, { "epoch": 43.69156626506024, "grad_norm": 12.032944679260254, "learning_rate": 2.8135483870967742e-05, "loss": 0.3839, "step": 6778 }, { "epoch": 43.69799196787149, "grad_norm": 7.6623640060424805, "learning_rate": 2.813225806451613e-05, "loss": 0.2755, "step": 6779 }, { "epoch": 43.70441767068273, "grad_norm": 6.245262622833252, "learning_rate": 2.812903225806452e-05, "loss": 0.271, "step": 6780 }, { "epoch": 43.71084337349397, "grad_norm": 8.632566452026367, "learning_rate": 2.8125806451612906e-05, "loss": 0.3211, "step": 6781 }, { "epoch": 43.71726907630522, "grad_norm": 35.69687271118164, "learning_rate": 2.812258064516129e-05, "loss": 0.4545, "step": 6782 }, { "epoch": 43.723694779116464, "grad_norm": 6.8690996170043945, "learning_rate": 2.8119354838709678e-05, "loss": 0.3358, "step": 6783 }, { "epoch": 43.73012048192771, "grad_norm": 13.997332572937012, "learning_rate": 2.8116129032258067e-05, "loss": 0.3756, "step": 6784 }, { "epoch": 43.736546184738955, "grad_norm": 14.500986099243164, "learning_rate": 2.8112903225806452e-05, "loss": 0.4343, "step": 6785 }, { "epoch": 43.7429718875502, "grad_norm": 6.411348342895508, "learning_rate": 2.810967741935484e-05, "loss": 0.2533, "step": 6786 }, { "epoch": 43.74939759036145, "grad_norm": 20.794523239135742, "learning_rate": 2.8106451612903227e-05, "loss": 0.6036, "step": 6787 }, { "epoch": 43.75582329317269, "grad_norm": 7.699295520782471, "learning_rate": 2.8103225806451616e-05, "loss": 0.3155, "step": 6788 }, { "epoch": 43.76224899598394, "grad_norm": 5.100988388061523, "learning_rate": 2.8100000000000005e-05, "loss": 0.1083, "step": 6789 }, { "epoch": 43.76867469879518, "grad_norm": 5.684778690338135, "learning_rate": 2.8096774193548388e-05, "loss": 0.2996, "step": 6790 }, { "epoch": 43.77510040160642, "grad_norm": 13.526923179626465, "learning_rate": 2.8093548387096773e-05, "loss": 0.2279, "step": 6791 }, { "epoch": 43.78152610441767, "grad_norm": 22.5965576171875, "learning_rate": 2.8090322580645163e-05, "loss": 0.4843, "step": 6792 }, { "epoch": 43.787951807228914, "grad_norm": 3.7002511024475098, "learning_rate": 2.8087096774193548e-05, "loss": 0.2993, "step": 6793 }, { "epoch": 43.794377510040164, "grad_norm": 4.362335681915283, "learning_rate": 2.8083870967741937e-05, "loss": 0.2453, "step": 6794 }, { "epoch": 43.800803212851406, "grad_norm": 55.75126647949219, "learning_rate": 2.8080645161290326e-05, "loss": 0.4895, "step": 6795 }, { "epoch": 43.80722891566265, "grad_norm": 23.267148971557617, "learning_rate": 2.8077419354838712e-05, "loss": 0.2785, "step": 6796 }, { "epoch": 43.8136546184739, "grad_norm": 10.303601264953613, "learning_rate": 2.8074193548387094e-05, "loss": 0.3599, "step": 6797 }, { "epoch": 43.82008032128514, "grad_norm": 6.044642925262451, "learning_rate": 2.8070967741935484e-05, "loss": 0.1854, "step": 6798 }, { "epoch": 43.82650602409639, "grad_norm": 8.53931713104248, "learning_rate": 2.8067741935483873e-05, "loss": 0.3445, "step": 6799 }, { "epoch": 43.83293172690763, "grad_norm": 13.233942985534668, "learning_rate": 2.806451612903226e-05, "loss": 0.4787, "step": 6800 }, { "epoch": 43.83935742971887, "grad_norm": 2.9064509868621826, "learning_rate": 2.8061290322580647e-05, "loss": 0.4762, "step": 6801 }, { "epoch": 43.84578313253012, "grad_norm": 13.290702819824219, "learning_rate": 2.8058064516129033e-05, "loss": 0.2637, "step": 6802 }, { "epoch": 43.852208835341365, "grad_norm": 24.513887405395508, "learning_rate": 2.8054838709677422e-05, "loss": 0.5552, "step": 6803 }, { "epoch": 43.85863453815261, "grad_norm": 6.9784464836120605, "learning_rate": 2.805161290322581e-05, "loss": 0.2604, "step": 6804 }, { "epoch": 43.86506024096386, "grad_norm": 17.452672958374023, "learning_rate": 2.8048387096774194e-05, "loss": 0.6822, "step": 6805 }, { "epoch": 43.8714859437751, "grad_norm": 3.0743229389190674, "learning_rate": 2.804516129032258e-05, "loss": 0.1848, "step": 6806 }, { "epoch": 43.87791164658635, "grad_norm": 12.680451393127441, "learning_rate": 2.804193548387097e-05, "loss": 0.2127, "step": 6807 }, { "epoch": 43.88433734939759, "grad_norm": 15.48709487915039, "learning_rate": 2.8038709677419354e-05, "loss": 0.2836, "step": 6808 }, { "epoch": 43.89076305220883, "grad_norm": 17.883853912353516, "learning_rate": 2.8035483870967743e-05, "loss": 0.6809, "step": 6809 }, { "epoch": 43.89718875502008, "grad_norm": 21.227394104003906, "learning_rate": 2.8032258064516132e-05, "loss": 0.4711, "step": 6810 }, { "epoch": 43.903614457831324, "grad_norm": 5.94002628326416, "learning_rate": 2.8029032258064518e-05, "loss": 0.3401, "step": 6811 }, { "epoch": 43.91004016064257, "grad_norm": 6.315562725067139, "learning_rate": 2.8025806451612907e-05, "loss": 0.3062, "step": 6812 }, { "epoch": 43.916465863453816, "grad_norm": 14.617342948913574, "learning_rate": 2.802258064516129e-05, "loss": 0.3545, "step": 6813 }, { "epoch": 43.92289156626506, "grad_norm": 8.281838417053223, "learning_rate": 2.801935483870968e-05, "loss": 0.2394, "step": 6814 }, { "epoch": 43.92931726907631, "grad_norm": 13.658869743347168, "learning_rate": 2.8016129032258064e-05, "loss": 0.4259, "step": 6815 }, { "epoch": 43.93574297188755, "grad_norm": 32.967681884765625, "learning_rate": 2.8012903225806453e-05, "loss": 0.283, "step": 6816 }, { "epoch": 43.94216867469879, "grad_norm": 24.796314239501953, "learning_rate": 2.800967741935484e-05, "loss": 0.6273, "step": 6817 }, { "epoch": 43.94859437751004, "grad_norm": 27.25303840637207, "learning_rate": 2.8006451612903228e-05, "loss": 0.3233, "step": 6818 }, { "epoch": 43.95502008032128, "grad_norm": 5.304405212402344, "learning_rate": 2.8003225806451617e-05, "loss": 0.2223, "step": 6819 }, { "epoch": 43.96144578313253, "grad_norm": 3.2969284057617188, "learning_rate": 2.8000000000000003e-05, "loss": 0.2823, "step": 6820 }, { "epoch": 43.967871485943775, "grad_norm": 13.653912544250488, "learning_rate": 2.7996774193548385e-05, "loss": 0.5284, "step": 6821 }, { "epoch": 43.97429718875502, "grad_norm": 7.784904479980469, "learning_rate": 2.7993548387096774e-05, "loss": 0.5421, "step": 6822 }, { "epoch": 43.980722891566266, "grad_norm": 44.00801467895508, "learning_rate": 2.7990322580645163e-05, "loss": 0.4896, "step": 6823 }, { "epoch": 43.98714859437751, "grad_norm": 1.921966791152954, "learning_rate": 2.798709677419355e-05, "loss": 0.3199, "step": 6824 }, { "epoch": 43.99357429718876, "grad_norm": 2.68086838722229, "learning_rate": 2.7983870967741938e-05, "loss": 0.3593, "step": 6825 }, { "epoch": 44.00321285140562, "grad_norm": 13.781465530395508, "learning_rate": 2.7980645161290324e-05, "loss": 0.3904, "step": 6826 }, { "epoch": 44.00963855421687, "grad_norm": 4.967447757720947, "learning_rate": 2.7977419354838713e-05, "loss": 0.4696, "step": 6827 }, { "epoch": 44.01606425702811, "grad_norm": 25.121315002441406, "learning_rate": 2.7974193548387095e-05, "loss": 0.4914, "step": 6828 }, { "epoch": 44.022489959839355, "grad_norm": 16.892288208007812, "learning_rate": 2.7970967741935484e-05, "loss": 0.2182, "step": 6829 }, { "epoch": 44.028915662650604, "grad_norm": 12.387653350830078, "learning_rate": 2.796774193548387e-05, "loss": 0.2502, "step": 6830 }, { "epoch": 44.035341365461846, "grad_norm": 22.114953994750977, "learning_rate": 2.796451612903226e-05, "loss": 0.3734, "step": 6831 }, { "epoch": 44.041767068273096, "grad_norm": 11.741511344909668, "learning_rate": 2.7961290322580645e-05, "loss": 0.3374, "step": 6832 }, { "epoch": 44.04819277108434, "grad_norm": 10.65000057220459, "learning_rate": 2.7958064516129034e-05, "loss": 0.3339, "step": 6833 }, { "epoch": 44.05461847389558, "grad_norm": 18.841829299926758, "learning_rate": 2.7954838709677423e-05, "loss": 0.5361, "step": 6834 }, { "epoch": 44.06104417670683, "grad_norm": 8.790011405944824, "learning_rate": 2.795161290322581e-05, "loss": 0.4829, "step": 6835 }, { "epoch": 44.06746987951807, "grad_norm": 12.730033874511719, "learning_rate": 2.794838709677419e-05, "loss": 0.352, "step": 6836 }, { "epoch": 44.073895582329314, "grad_norm": 9.295753479003906, "learning_rate": 2.794516129032258e-05, "loss": 0.2833, "step": 6837 }, { "epoch": 44.08032128514056, "grad_norm": 15.86755084991455, "learning_rate": 2.794193548387097e-05, "loss": 0.5615, "step": 6838 }, { "epoch": 44.086746987951805, "grad_norm": 4.967320919036865, "learning_rate": 2.7938709677419355e-05, "loss": 0.2008, "step": 6839 }, { "epoch": 44.093172690763055, "grad_norm": 10.217352867126465, "learning_rate": 2.7935483870967744e-05, "loss": 0.3707, "step": 6840 }, { "epoch": 44.0995983935743, "grad_norm": 8.465452194213867, "learning_rate": 2.793225806451613e-05, "loss": 0.1939, "step": 6841 }, { "epoch": 44.10602409638554, "grad_norm": 11.681164741516113, "learning_rate": 2.792903225806452e-05, "loss": 0.4426, "step": 6842 }, { "epoch": 44.11244979919679, "grad_norm": 4.8117289543151855, "learning_rate": 2.7925806451612908e-05, "loss": 0.1109, "step": 6843 }, { "epoch": 44.11887550200803, "grad_norm": 3.3044464588165283, "learning_rate": 2.792258064516129e-05, "loss": 0.2298, "step": 6844 }, { "epoch": 44.12530120481928, "grad_norm": 17.49867820739746, "learning_rate": 2.7919354838709676e-05, "loss": 0.7885, "step": 6845 }, { "epoch": 44.13172690763052, "grad_norm": 13.99487018585205, "learning_rate": 2.7916129032258065e-05, "loss": 0.3025, "step": 6846 }, { "epoch": 44.138152610441765, "grad_norm": 8.633150100708008, "learning_rate": 2.7912903225806454e-05, "loss": 0.4117, "step": 6847 }, { "epoch": 44.144578313253014, "grad_norm": 23.775375366210938, "learning_rate": 2.790967741935484e-05, "loss": 0.3583, "step": 6848 }, { "epoch": 44.151004016064256, "grad_norm": 11.426551818847656, "learning_rate": 2.790645161290323e-05, "loss": 0.2657, "step": 6849 }, { "epoch": 44.157429718875505, "grad_norm": 7.076809406280518, "learning_rate": 2.7903225806451615e-05, "loss": 0.3851, "step": 6850 }, { "epoch": 44.16385542168675, "grad_norm": 19.182126998901367, "learning_rate": 2.7900000000000004e-05, "loss": 0.4886, "step": 6851 }, { "epoch": 44.17028112449799, "grad_norm": 17.561138153076172, "learning_rate": 2.7896774193548386e-05, "loss": 0.3367, "step": 6852 }, { "epoch": 44.17670682730924, "grad_norm": 21.28853416442871, "learning_rate": 2.7893548387096775e-05, "loss": 0.2852, "step": 6853 }, { "epoch": 44.18313253012048, "grad_norm": 7.009654521942139, "learning_rate": 2.789032258064516e-05, "loss": 0.3669, "step": 6854 }, { "epoch": 44.189558232931724, "grad_norm": 8.598198890686035, "learning_rate": 2.788709677419355e-05, "loss": 0.4536, "step": 6855 }, { "epoch": 44.19598393574297, "grad_norm": 8.854788780212402, "learning_rate": 2.7883870967741936e-05, "loss": 0.2109, "step": 6856 }, { "epoch": 44.202409638554215, "grad_norm": 16.505802154541016, "learning_rate": 2.7880645161290325e-05, "loss": 0.5645, "step": 6857 }, { "epoch": 44.208835341365464, "grad_norm": 25.098060607910156, "learning_rate": 2.7877419354838714e-05, "loss": 0.3542, "step": 6858 }, { "epoch": 44.21526104417671, "grad_norm": 3.7883615493774414, "learning_rate": 2.78741935483871e-05, "loss": 0.2621, "step": 6859 }, { "epoch": 44.22168674698795, "grad_norm": 7.6027140617370605, "learning_rate": 2.7870967741935482e-05, "loss": 0.3174, "step": 6860 }, { "epoch": 44.2281124497992, "grad_norm": 13.390052795410156, "learning_rate": 2.786774193548387e-05, "loss": 0.2523, "step": 6861 }, { "epoch": 44.23453815261044, "grad_norm": 13.84585189819336, "learning_rate": 2.786451612903226e-05, "loss": 0.326, "step": 6862 }, { "epoch": 44.24096385542169, "grad_norm": 31.29087257385254, "learning_rate": 2.7861290322580646e-05, "loss": 0.6442, "step": 6863 }, { "epoch": 44.24738955823293, "grad_norm": 13.927441596984863, "learning_rate": 2.7858064516129035e-05, "loss": 0.5507, "step": 6864 }, { "epoch": 44.253815261044174, "grad_norm": 5.651756763458252, "learning_rate": 2.785483870967742e-05, "loss": 0.2651, "step": 6865 }, { "epoch": 44.26024096385542, "grad_norm": 10.20797348022461, "learning_rate": 2.785161290322581e-05, "loss": 0.3519, "step": 6866 }, { "epoch": 44.266666666666666, "grad_norm": 5.713867664337158, "learning_rate": 2.7848387096774192e-05, "loss": 0.4034, "step": 6867 }, { "epoch": 44.273092369477915, "grad_norm": 7.181176662445068, "learning_rate": 2.784516129032258e-05, "loss": 0.2654, "step": 6868 }, { "epoch": 44.27951807228916, "grad_norm": 6.2992682456970215, "learning_rate": 2.7841935483870967e-05, "loss": 0.3464, "step": 6869 }, { "epoch": 44.2859437751004, "grad_norm": 9.360143661499023, "learning_rate": 2.7838709677419356e-05, "loss": 0.3, "step": 6870 }, { "epoch": 44.29236947791165, "grad_norm": 31.08097267150879, "learning_rate": 2.7835483870967742e-05, "loss": 0.2994, "step": 6871 }, { "epoch": 44.29879518072289, "grad_norm": 9.359074592590332, "learning_rate": 2.783225806451613e-05, "loss": 0.3136, "step": 6872 }, { "epoch": 44.30522088353413, "grad_norm": 3.30641770362854, "learning_rate": 2.782903225806452e-05, "loss": 0.1017, "step": 6873 }, { "epoch": 44.31164658634538, "grad_norm": 8.923386573791504, "learning_rate": 2.7825806451612906e-05, "loss": 0.324, "step": 6874 }, { "epoch": 44.318072289156625, "grad_norm": 3.442554473876953, "learning_rate": 2.7822580645161288e-05, "loss": 0.2654, "step": 6875 }, { "epoch": 44.324497991967874, "grad_norm": 4.057717800140381, "learning_rate": 2.7819354838709677e-05, "loss": 0.3923, "step": 6876 }, { "epoch": 44.330923694779116, "grad_norm": 28.114049911499023, "learning_rate": 2.7816129032258066e-05, "loss": 0.4425, "step": 6877 }, { "epoch": 44.33734939759036, "grad_norm": 8.620261192321777, "learning_rate": 2.7812903225806452e-05, "loss": 0.5297, "step": 6878 }, { "epoch": 44.34377510040161, "grad_norm": 8.21008586883545, "learning_rate": 2.780967741935484e-05, "loss": 0.2584, "step": 6879 }, { "epoch": 44.35020080321285, "grad_norm": 9.203254699707031, "learning_rate": 2.7806451612903227e-05, "loss": 0.4331, "step": 6880 }, { "epoch": 44.3566265060241, "grad_norm": 4.646229267120361, "learning_rate": 2.7803225806451616e-05, "loss": 0.1372, "step": 6881 }, { "epoch": 44.36305220883534, "grad_norm": 21.98133087158203, "learning_rate": 2.7800000000000005e-05, "loss": 0.4472, "step": 6882 }, { "epoch": 44.369477911646584, "grad_norm": 9.01758098602295, "learning_rate": 2.7796774193548387e-05, "loss": 0.279, "step": 6883 }, { "epoch": 44.37590361445783, "grad_norm": 13.577691078186035, "learning_rate": 2.7793548387096773e-05, "loss": 0.5702, "step": 6884 }, { "epoch": 44.382329317269075, "grad_norm": 17.965213775634766, "learning_rate": 2.7790322580645162e-05, "loss": 0.3742, "step": 6885 }, { "epoch": 44.388755020080325, "grad_norm": 6.1531805992126465, "learning_rate": 2.778709677419355e-05, "loss": 0.2489, "step": 6886 }, { "epoch": 44.39518072289157, "grad_norm": 8.711943626403809, "learning_rate": 2.7783870967741937e-05, "loss": 0.3715, "step": 6887 }, { "epoch": 44.40160642570281, "grad_norm": 7.721774578094482, "learning_rate": 2.7780645161290326e-05, "loss": 0.4928, "step": 6888 }, { "epoch": 44.40803212851406, "grad_norm": 13.884897232055664, "learning_rate": 2.777741935483871e-05, "loss": 0.3663, "step": 6889 }, { "epoch": 44.4144578313253, "grad_norm": 2.4375722408294678, "learning_rate": 2.77741935483871e-05, "loss": 0.0901, "step": 6890 }, { "epoch": 44.42088353413654, "grad_norm": 4.913372039794922, "learning_rate": 2.7770967741935483e-05, "loss": 0.3197, "step": 6891 }, { "epoch": 44.42730923694779, "grad_norm": 7.561152458190918, "learning_rate": 2.7767741935483872e-05, "loss": 0.0987, "step": 6892 }, { "epoch": 44.433734939759034, "grad_norm": 12.266944885253906, "learning_rate": 2.7764516129032258e-05, "loss": 0.5354, "step": 6893 }, { "epoch": 44.440160642570284, "grad_norm": 64.62538146972656, "learning_rate": 2.7761290322580647e-05, "loss": 0.6413, "step": 6894 }, { "epoch": 44.446586345381526, "grad_norm": 10.379334449768066, "learning_rate": 2.7758064516129033e-05, "loss": 0.2744, "step": 6895 }, { "epoch": 44.45301204819277, "grad_norm": 9.378013610839844, "learning_rate": 2.7754838709677422e-05, "loss": 0.1986, "step": 6896 }, { "epoch": 44.45943775100402, "grad_norm": 8.778912544250488, "learning_rate": 2.775161290322581e-05, "loss": 0.5781, "step": 6897 }, { "epoch": 44.46586345381526, "grad_norm": 15.69161605834961, "learning_rate": 2.7748387096774193e-05, "loss": 0.3799, "step": 6898 }, { "epoch": 44.47228915662651, "grad_norm": 3.6642401218414307, "learning_rate": 2.774516129032258e-05, "loss": 0.2257, "step": 6899 }, { "epoch": 44.47871485943775, "grad_norm": 10.678464889526367, "learning_rate": 2.7741935483870968e-05, "loss": 0.5172, "step": 6900 }, { "epoch": 44.48514056224899, "grad_norm": 8.227027893066406, "learning_rate": 2.7738709677419357e-05, "loss": 0.3963, "step": 6901 }, { "epoch": 44.49156626506024, "grad_norm": 8.743165016174316, "learning_rate": 2.7735483870967743e-05, "loss": 0.3841, "step": 6902 }, { "epoch": 44.497991967871485, "grad_norm": 4.1107988357543945, "learning_rate": 2.7732258064516132e-05, "loss": 0.257, "step": 6903 }, { "epoch": 44.504417670682734, "grad_norm": 20.394306182861328, "learning_rate": 2.7729032258064518e-05, "loss": 0.3556, "step": 6904 }, { "epoch": 44.51084337349398, "grad_norm": 8.683865547180176, "learning_rate": 2.7725806451612907e-05, "loss": 0.6032, "step": 6905 }, { "epoch": 44.51726907630522, "grad_norm": 16.039892196655273, "learning_rate": 2.772258064516129e-05, "loss": 0.382, "step": 6906 }, { "epoch": 44.52369477911647, "grad_norm": 9.259767532348633, "learning_rate": 2.7719354838709678e-05, "loss": 0.5282, "step": 6907 }, { "epoch": 44.53012048192771, "grad_norm": 12.762166976928711, "learning_rate": 2.7716129032258064e-05, "loss": 0.2086, "step": 6908 }, { "epoch": 44.53654618473895, "grad_norm": 18.991640090942383, "learning_rate": 2.7712903225806453e-05, "loss": 0.2565, "step": 6909 }, { "epoch": 44.5429718875502, "grad_norm": 3.554856538772583, "learning_rate": 2.7709677419354842e-05, "loss": 0.2762, "step": 6910 }, { "epoch": 44.549397590361444, "grad_norm": 41.357845306396484, "learning_rate": 2.7706451612903228e-05, "loss": 0.585, "step": 6911 }, { "epoch": 44.55582329317269, "grad_norm": 8.762624740600586, "learning_rate": 2.7703225806451617e-05, "loss": 0.2028, "step": 6912 }, { "epoch": 44.562248995983936, "grad_norm": 15.254287719726562, "learning_rate": 2.7700000000000002e-05, "loss": 0.2523, "step": 6913 }, { "epoch": 44.56867469879518, "grad_norm": 8.73322582244873, "learning_rate": 2.7696774193548385e-05, "loss": 0.3568, "step": 6914 }, { "epoch": 44.57510040160643, "grad_norm": 10.7490234375, "learning_rate": 2.7693548387096774e-05, "loss": 0.3507, "step": 6915 }, { "epoch": 44.58152610441767, "grad_norm": 38.842010498046875, "learning_rate": 2.7690322580645163e-05, "loss": 0.2609, "step": 6916 }, { "epoch": 44.58795180722892, "grad_norm": 9.496431350708008, "learning_rate": 2.768709677419355e-05, "loss": 0.3116, "step": 6917 }, { "epoch": 44.59437751004016, "grad_norm": 11.148895263671875, "learning_rate": 2.7683870967741938e-05, "loss": 0.2755, "step": 6918 }, { "epoch": 44.6008032128514, "grad_norm": 4.362517833709717, "learning_rate": 2.7680645161290323e-05, "loss": 0.3415, "step": 6919 }, { "epoch": 44.60722891566265, "grad_norm": 7.6442131996154785, "learning_rate": 2.7677419354838713e-05, "loss": 0.2816, "step": 6920 }, { "epoch": 44.613654618473895, "grad_norm": 6.909961700439453, "learning_rate": 2.76741935483871e-05, "loss": 0.3509, "step": 6921 }, { "epoch": 44.620080321285144, "grad_norm": 11.767563819885254, "learning_rate": 2.7670967741935484e-05, "loss": 0.2322, "step": 6922 }, { "epoch": 44.626506024096386, "grad_norm": 14.8084135055542, "learning_rate": 2.766774193548387e-05, "loss": 0.2117, "step": 6923 }, { "epoch": 44.63293172690763, "grad_norm": 8.15322208404541, "learning_rate": 2.766451612903226e-05, "loss": 0.2885, "step": 6924 }, { "epoch": 44.63935742971888, "grad_norm": 5.973489761352539, "learning_rate": 2.7661290322580648e-05, "loss": 0.3968, "step": 6925 }, { "epoch": 44.64578313253012, "grad_norm": 3.967755079269409, "learning_rate": 2.7658064516129034e-05, "loss": 0.3011, "step": 6926 }, { "epoch": 44.65220883534136, "grad_norm": 3.722294807434082, "learning_rate": 2.7654838709677423e-05, "loss": 0.3434, "step": 6927 }, { "epoch": 44.65863453815261, "grad_norm": 3.8664894104003906, "learning_rate": 2.765161290322581e-05, "loss": 0.3503, "step": 6928 }, { "epoch": 44.665060240963854, "grad_norm": 5.753806114196777, "learning_rate": 2.7648387096774197e-05, "loss": 0.3958, "step": 6929 }, { "epoch": 44.6714859437751, "grad_norm": 7.068022727966309, "learning_rate": 2.764516129032258e-05, "loss": 0.3049, "step": 6930 }, { "epoch": 44.677911646586345, "grad_norm": 2.7280662059783936, "learning_rate": 2.764193548387097e-05, "loss": 0.2642, "step": 6931 }, { "epoch": 44.68433734939759, "grad_norm": 9.796874046325684, "learning_rate": 2.7638709677419355e-05, "loss": 0.4006, "step": 6932 }, { "epoch": 44.69076305220884, "grad_norm": 7.615346908569336, "learning_rate": 2.7635483870967744e-05, "loss": 0.2927, "step": 6933 }, { "epoch": 44.69718875502008, "grad_norm": 18.2475643157959, "learning_rate": 2.763225806451613e-05, "loss": 0.5445, "step": 6934 }, { "epoch": 44.70361445783133, "grad_norm": 20.09774398803711, "learning_rate": 2.762903225806452e-05, "loss": 0.2908, "step": 6935 }, { "epoch": 44.71004016064257, "grad_norm": 5.8736042976379395, "learning_rate": 2.7625806451612908e-05, "loss": 0.4792, "step": 6936 }, { "epoch": 44.71646586345381, "grad_norm": 12.511962890625, "learning_rate": 2.762258064516129e-05, "loss": 0.3092, "step": 6937 }, { "epoch": 44.72289156626506, "grad_norm": 4.009990215301514, "learning_rate": 2.7619354838709676e-05, "loss": 0.2881, "step": 6938 }, { "epoch": 44.729317269076304, "grad_norm": 18.873798370361328, "learning_rate": 2.7616129032258065e-05, "loss": 0.5296, "step": 6939 }, { "epoch": 44.735742971887554, "grad_norm": 10.255088806152344, "learning_rate": 2.7612903225806454e-05, "loss": 0.5785, "step": 6940 }, { "epoch": 44.742168674698796, "grad_norm": 9.561141967773438, "learning_rate": 2.760967741935484e-05, "loss": 0.4497, "step": 6941 }, { "epoch": 44.74859437751004, "grad_norm": 10.926531791687012, "learning_rate": 2.760645161290323e-05, "loss": 0.4333, "step": 6942 }, { "epoch": 44.75502008032129, "grad_norm": 6.936130046844482, "learning_rate": 2.7603225806451614e-05, "loss": 0.2235, "step": 6943 }, { "epoch": 44.76144578313253, "grad_norm": 14.271961212158203, "learning_rate": 2.7600000000000003e-05, "loss": 0.2948, "step": 6944 }, { "epoch": 44.76787148594377, "grad_norm": 18.94078826904297, "learning_rate": 2.7596774193548386e-05, "loss": 0.6231, "step": 6945 }, { "epoch": 44.77429718875502, "grad_norm": 6.972793102264404, "learning_rate": 2.7593548387096775e-05, "loss": 0.2694, "step": 6946 }, { "epoch": 44.78072289156626, "grad_norm": 18.604516983032227, "learning_rate": 2.759032258064516e-05, "loss": 0.4709, "step": 6947 }, { "epoch": 44.78714859437751, "grad_norm": 6.3241496086120605, "learning_rate": 2.758709677419355e-05, "loss": 0.1282, "step": 6948 }, { "epoch": 44.793574297188755, "grad_norm": 7.5470733642578125, "learning_rate": 2.758387096774194e-05, "loss": 0.2758, "step": 6949 }, { "epoch": 44.8, "grad_norm": 20.774259567260742, "learning_rate": 2.7580645161290324e-05, "loss": 0.6413, "step": 6950 }, { "epoch": 44.80642570281125, "grad_norm": 14.131318092346191, "learning_rate": 2.7577419354838714e-05, "loss": 0.3279, "step": 6951 }, { "epoch": 44.81285140562249, "grad_norm": 21.61092185974121, "learning_rate": 2.75741935483871e-05, "loss": 0.4117, "step": 6952 }, { "epoch": 44.81927710843374, "grad_norm": 10.120919227600098, "learning_rate": 2.757096774193548e-05, "loss": 0.3954, "step": 6953 }, { "epoch": 44.82570281124498, "grad_norm": 12.90715503692627, "learning_rate": 2.756774193548387e-05, "loss": 0.4147, "step": 6954 }, { "epoch": 44.83212851405622, "grad_norm": 3.7653543949127197, "learning_rate": 2.756451612903226e-05, "loss": 0.2713, "step": 6955 }, { "epoch": 44.83855421686747, "grad_norm": 3.0917582511901855, "learning_rate": 2.7561290322580645e-05, "loss": 0.3238, "step": 6956 }, { "epoch": 44.844979919678714, "grad_norm": 13.555360794067383, "learning_rate": 2.7558064516129035e-05, "loss": 0.2575, "step": 6957 }, { "epoch": 44.85140562248996, "grad_norm": 21.488788604736328, "learning_rate": 2.755483870967742e-05, "loss": 0.1317, "step": 6958 }, { "epoch": 44.857831325301206, "grad_norm": 13.099515914916992, "learning_rate": 2.755161290322581e-05, "loss": 0.2707, "step": 6959 }, { "epoch": 44.86425702811245, "grad_norm": 6.055099010467529, "learning_rate": 2.75483870967742e-05, "loss": 0.1145, "step": 6960 }, { "epoch": 44.8706827309237, "grad_norm": 17.687658309936523, "learning_rate": 2.754516129032258e-05, "loss": 0.2469, "step": 6961 }, { "epoch": 44.87710843373494, "grad_norm": 14.245097160339355, "learning_rate": 2.7541935483870966e-05, "loss": 0.5239, "step": 6962 }, { "epoch": 44.88353413654618, "grad_norm": 21.04055404663086, "learning_rate": 2.7538709677419356e-05, "loss": 0.3081, "step": 6963 }, { "epoch": 44.88995983935743, "grad_norm": 11.812444686889648, "learning_rate": 2.7535483870967745e-05, "loss": 0.249, "step": 6964 }, { "epoch": 44.89638554216867, "grad_norm": 3.729907989501953, "learning_rate": 2.753225806451613e-05, "loss": 0.3252, "step": 6965 }, { "epoch": 44.90281124497992, "grad_norm": 16.830406188964844, "learning_rate": 2.752903225806452e-05, "loss": 0.6387, "step": 6966 }, { "epoch": 44.909236947791165, "grad_norm": 4.8757524490356445, "learning_rate": 2.7525806451612905e-05, "loss": 0.3161, "step": 6967 }, { "epoch": 44.91566265060241, "grad_norm": 22.162996292114258, "learning_rate": 2.7522580645161287e-05, "loss": 0.3198, "step": 6968 }, { "epoch": 44.922088353413656, "grad_norm": 14.672492980957031, "learning_rate": 2.7519354838709677e-05, "loss": 0.3377, "step": 6969 }, { "epoch": 44.9285140562249, "grad_norm": 17.673242568969727, "learning_rate": 2.7516129032258066e-05, "loss": 0.29, "step": 6970 }, { "epoch": 44.93493975903615, "grad_norm": 21.853038787841797, "learning_rate": 2.751290322580645e-05, "loss": 0.5022, "step": 6971 }, { "epoch": 44.94136546184739, "grad_norm": 8.240283012390137, "learning_rate": 2.750967741935484e-05, "loss": 0.3795, "step": 6972 }, { "epoch": 44.94779116465863, "grad_norm": 3.2563037872314453, "learning_rate": 2.750645161290323e-05, "loss": 0.0761, "step": 6973 }, { "epoch": 44.95421686746988, "grad_norm": 6.325511455535889, "learning_rate": 2.7503225806451615e-05, "loss": 0.3089, "step": 6974 }, { "epoch": 44.960642570281124, "grad_norm": 17.892465591430664, "learning_rate": 2.7500000000000004e-05, "loss": 0.4504, "step": 6975 }, { "epoch": 44.967068273092366, "grad_norm": 25.034475326538086, "learning_rate": 2.7496774193548387e-05, "loss": 0.2194, "step": 6976 }, { "epoch": 44.973493975903615, "grad_norm": 10.796599388122559, "learning_rate": 2.7493548387096772e-05, "loss": 0.2749, "step": 6977 }, { "epoch": 44.97991967871486, "grad_norm": 2.895017623901367, "learning_rate": 2.749032258064516e-05, "loss": 0.1352, "step": 6978 }, { "epoch": 44.98634538152611, "grad_norm": 20.882680892944336, "learning_rate": 2.748709677419355e-05, "loss": 0.4309, "step": 6979 }, { "epoch": 44.99277108433735, "grad_norm": 7.473291873931885, "learning_rate": 2.7483870967741936e-05, "loss": 0.1673, "step": 6980 }, { "epoch": 45.00240963855422, "grad_norm": 4.953526020050049, "learning_rate": 2.7480645161290325e-05, "loss": 0.2421, "step": 6981 }, { "epoch": 45.00883534136546, "grad_norm": 16.390207290649414, "learning_rate": 2.747741935483871e-05, "loss": 0.3902, "step": 6982 }, { "epoch": 45.015261044176704, "grad_norm": 15.741551399230957, "learning_rate": 2.74741935483871e-05, "loss": 0.621, "step": 6983 }, { "epoch": 45.02168674698795, "grad_norm": 3.974667549133301, "learning_rate": 2.7470967741935483e-05, "loss": 0.2514, "step": 6984 }, { "epoch": 45.028112449799195, "grad_norm": 12.478424072265625, "learning_rate": 2.746774193548387e-05, "loss": 0.3901, "step": 6985 }, { "epoch": 45.034538152610445, "grad_norm": 7.067190647125244, "learning_rate": 2.7464516129032257e-05, "loss": 0.2104, "step": 6986 }, { "epoch": 45.04096385542169, "grad_norm": 4.973394870758057, "learning_rate": 2.7461290322580646e-05, "loss": 0.3209, "step": 6987 }, { "epoch": 45.04738955823293, "grad_norm": 13.868691444396973, "learning_rate": 2.7458064516129035e-05, "loss": 0.3833, "step": 6988 }, { "epoch": 45.05381526104418, "grad_norm": 6.328377723693848, "learning_rate": 2.745483870967742e-05, "loss": 0.404, "step": 6989 }, { "epoch": 45.06024096385542, "grad_norm": 11.86628246307373, "learning_rate": 2.745161290322581e-05, "loss": 0.5527, "step": 6990 }, { "epoch": 45.06666666666667, "grad_norm": 8.408854484558105, "learning_rate": 2.7448387096774196e-05, "loss": 0.4742, "step": 6991 }, { "epoch": 45.07309236947791, "grad_norm": 3.1634626388549805, "learning_rate": 2.744516129032258e-05, "loss": 0.262, "step": 6992 }, { "epoch": 45.079518072289154, "grad_norm": 21.521987915039062, "learning_rate": 2.7441935483870967e-05, "loss": 0.4755, "step": 6993 }, { "epoch": 45.085943775100404, "grad_norm": 9.697423934936523, "learning_rate": 2.7438709677419357e-05, "loss": 0.2425, "step": 6994 }, { "epoch": 45.092369477911646, "grad_norm": 12.333966255187988, "learning_rate": 2.7435483870967742e-05, "loss": 0.2961, "step": 6995 }, { "epoch": 45.09879518072289, "grad_norm": 23.582733154296875, "learning_rate": 2.743225806451613e-05, "loss": 0.6169, "step": 6996 }, { "epoch": 45.10522088353414, "grad_norm": 12.291025161743164, "learning_rate": 2.7429032258064517e-05, "loss": 0.2601, "step": 6997 }, { "epoch": 45.11164658634538, "grad_norm": 7.050225734710693, "learning_rate": 2.7425806451612906e-05, "loss": 0.443, "step": 6998 }, { "epoch": 45.11807228915663, "grad_norm": 11.503120422363281, "learning_rate": 2.7422580645161295e-05, "loss": 0.2582, "step": 6999 }, { "epoch": 45.12449799196787, "grad_norm": 4.971824645996094, "learning_rate": 2.7419354838709678e-05, "loss": 0.1714, "step": 7000 }, { "epoch": 45.13092369477911, "grad_norm": 6.2998809814453125, "learning_rate": 2.7416129032258063e-05, "loss": 0.4874, "step": 7001 }, { "epoch": 45.13734939759036, "grad_norm": 11.31480884552002, "learning_rate": 2.7412903225806452e-05, "loss": 0.1597, "step": 7002 }, { "epoch": 45.143775100401605, "grad_norm": 2.9157633781433105, "learning_rate": 2.740967741935484e-05, "loss": 0.0843, "step": 7003 }, { "epoch": 45.150200803212854, "grad_norm": 4.757986545562744, "learning_rate": 2.7406451612903227e-05, "loss": 0.4096, "step": 7004 }, { "epoch": 45.1566265060241, "grad_norm": 9.172698020935059, "learning_rate": 2.7403225806451616e-05, "loss": 0.5362, "step": 7005 }, { "epoch": 45.16305220883534, "grad_norm": 41.897159576416016, "learning_rate": 2.7400000000000002e-05, "loss": 0.5798, "step": 7006 }, { "epoch": 45.16947791164659, "grad_norm": 28.34930992126465, "learning_rate": 2.7396774193548384e-05, "loss": 0.5634, "step": 7007 }, { "epoch": 45.17590361445783, "grad_norm": 16.106287002563477, "learning_rate": 2.7393548387096773e-05, "loss": 0.4939, "step": 7008 }, { "epoch": 45.18232931726908, "grad_norm": 9.254327774047852, "learning_rate": 2.7390322580645162e-05, "loss": 0.3457, "step": 7009 }, { "epoch": 45.18875502008032, "grad_norm": 4.67654275894165, "learning_rate": 2.7387096774193548e-05, "loss": 0.334, "step": 7010 }, { "epoch": 45.195180722891564, "grad_norm": 4.334888458251953, "learning_rate": 2.7383870967741937e-05, "loss": 0.2377, "step": 7011 }, { "epoch": 45.20160642570281, "grad_norm": 11.53090763092041, "learning_rate": 2.7380645161290326e-05, "loss": 0.3484, "step": 7012 }, { "epoch": 45.208032128514056, "grad_norm": 20.35521125793457, "learning_rate": 2.7377419354838712e-05, "loss": 0.189, "step": 7013 }, { "epoch": 45.2144578313253, "grad_norm": 9.268329620361328, "learning_rate": 2.73741935483871e-05, "loss": 0.2079, "step": 7014 }, { "epoch": 45.22088353413655, "grad_norm": 41.26242446899414, "learning_rate": 2.7370967741935483e-05, "loss": 0.3478, "step": 7015 }, { "epoch": 45.22730923694779, "grad_norm": 6.987249851226807, "learning_rate": 2.736774193548387e-05, "loss": 0.3791, "step": 7016 }, { "epoch": 45.23373493975904, "grad_norm": 2.9701550006866455, "learning_rate": 2.7364516129032258e-05, "loss": 0.1213, "step": 7017 }, { "epoch": 45.24016064257028, "grad_norm": 5.494978427886963, "learning_rate": 2.7361290322580647e-05, "loss": 0.4848, "step": 7018 }, { "epoch": 45.24658634538152, "grad_norm": 2.749946117401123, "learning_rate": 2.7358064516129033e-05, "loss": 0.1091, "step": 7019 }, { "epoch": 45.25301204819277, "grad_norm": 44.61805725097656, "learning_rate": 2.7354838709677422e-05, "loss": 0.3458, "step": 7020 }, { "epoch": 45.259437751004015, "grad_norm": 30.593229293823242, "learning_rate": 2.7351612903225808e-05, "loss": 0.3021, "step": 7021 }, { "epoch": 45.265863453815264, "grad_norm": 10.4358549118042, "learning_rate": 2.7348387096774197e-05, "loss": 0.4084, "step": 7022 }, { "epoch": 45.272289156626506, "grad_norm": 50.26017379760742, "learning_rate": 2.734516129032258e-05, "loss": 0.5564, "step": 7023 }, { "epoch": 45.27871485943775, "grad_norm": 9.532809257507324, "learning_rate": 2.734193548387097e-05, "loss": 0.1537, "step": 7024 }, { "epoch": 45.285140562249, "grad_norm": 6.01202392578125, "learning_rate": 2.7338709677419354e-05, "loss": 0.3939, "step": 7025 }, { "epoch": 45.29156626506024, "grad_norm": 15.853163719177246, "learning_rate": 2.7335483870967743e-05, "loss": 0.6945, "step": 7026 }, { "epoch": 45.29799196787149, "grad_norm": 15.747126579284668, "learning_rate": 2.7332258064516132e-05, "loss": 0.4912, "step": 7027 }, { "epoch": 45.30441767068273, "grad_norm": 15.148466110229492, "learning_rate": 2.7329032258064518e-05, "loss": 0.4801, "step": 7028 }, { "epoch": 45.310843373493974, "grad_norm": 13.332650184631348, "learning_rate": 2.7325806451612907e-05, "loss": 0.3971, "step": 7029 }, { "epoch": 45.31726907630522, "grad_norm": 11.451433181762695, "learning_rate": 2.7322580645161293e-05, "loss": 0.311, "step": 7030 }, { "epoch": 45.323694779116465, "grad_norm": 6.843425750732422, "learning_rate": 2.7319354838709675e-05, "loss": 0.5084, "step": 7031 }, { "epoch": 45.33012048192771, "grad_norm": 14.802313804626465, "learning_rate": 2.7316129032258064e-05, "loss": 0.3101, "step": 7032 }, { "epoch": 45.33654618473896, "grad_norm": 7.238044261932373, "learning_rate": 2.7312903225806453e-05, "loss": 0.1879, "step": 7033 }, { "epoch": 45.3429718875502, "grad_norm": 26.3364315032959, "learning_rate": 2.730967741935484e-05, "loss": 0.2349, "step": 7034 }, { "epoch": 45.34939759036145, "grad_norm": 9.192367553710938, "learning_rate": 2.7306451612903228e-05, "loss": 0.4511, "step": 7035 }, { "epoch": 45.35582329317269, "grad_norm": 7.694331645965576, "learning_rate": 2.7303225806451617e-05, "loss": 0.1959, "step": 7036 }, { "epoch": 45.36224899598393, "grad_norm": 11.606061935424805, "learning_rate": 2.7300000000000003e-05, "loss": 0.1496, "step": 7037 }, { "epoch": 45.36867469879518, "grad_norm": 7.0536322593688965, "learning_rate": 2.7296774193548392e-05, "loss": 0.1406, "step": 7038 }, { "epoch": 45.375100401606424, "grad_norm": 11.57717227935791, "learning_rate": 2.7293548387096774e-05, "loss": 0.4443, "step": 7039 }, { "epoch": 45.381526104417674, "grad_norm": 4.647211074829102, "learning_rate": 2.729032258064516e-05, "loss": 0.1547, "step": 7040 }, { "epoch": 45.387951807228916, "grad_norm": 3.981508255004883, "learning_rate": 2.728709677419355e-05, "loss": 0.2217, "step": 7041 }, { "epoch": 45.39437751004016, "grad_norm": 8.15679931640625, "learning_rate": 2.7283870967741938e-05, "loss": 0.2747, "step": 7042 }, { "epoch": 45.40080321285141, "grad_norm": 15.342512130737305, "learning_rate": 2.7280645161290324e-05, "loss": 0.355, "step": 7043 }, { "epoch": 45.40722891566265, "grad_norm": 8.487688064575195, "learning_rate": 2.7277419354838713e-05, "loss": 0.2575, "step": 7044 }, { "epoch": 45.4136546184739, "grad_norm": 6.8262939453125, "learning_rate": 2.72741935483871e-05, "loss": 0.3199, "step": 7045 }, { "epoch": 45.42008032128514, "grad_norm": 7.680093288421631, "learning_rate": 2.727096774193548e-05, "loss": 0.4068, "step": 7046 }, { "epoch": 45.42650602409638, "grad_norm": 12.171651840209961, "learning_rate": 2.726774193548387e-05, "loss": 0.4664, "step": 7047 }, { "epoch": 45.43293172690763, "grad_norm": 3.3113505840301514, "learning_rate": 2.726451612903226e-05, "loss": 0.1115, "step": 7048 }, { "epoch": 45.439357429718875, "grad_norm": 8.699409484863281, "learning_rate": 2.7261290322580645e-05, "loss": 0.3461, "step": 7049 }, { "epoch": 45.44578313253012, "grad_norm": 7.033812046051025, "learning_rate": 2.7258064516129034e-05, "loss": 0.5398, "step": 7050 }, { "epoch": 45.45220883534137, "grad_norm": 23.014801025390625, "learning_rate": 2.7254838709677423e-05, "loss": 0.2993, "step": 7051 }, { "epoch": 45.45863453815261, "grad_norm": 14.371550559997559, "learning_rate": 2.725161290322581e-05, "loss": 0.4693, "step": 7052 }, { "epoch": 45.46506024096386, "grad_norm": 18.63987922668457, "learning_rate": 2.7248387096774198e-05, "loss": 0.3321, "step": 7053 }, { "epoch": 45.4714859437751, "grad_norm": 8.730029106140137, "learning_rate": 2.724516129032258e-05, "loss": 0.4647, "step": 7054 }, { "epoch": 45.47791164658634, "grad_norm": 4.668944835662842, "learning_rate": 2.7241935483870966e-05, "loss": 0.3174, "step": 7055 }, { "epoch": 45.48433734939759, "grad_norm": 7.551980972290039, "learning_rate": 2.7238709677419355e-05, "loss": 0.2952, "step": 7056 }, { "epoch": 45.490763052208834, "grad_norm": 13.217572212219238, "learning_rate": 2.7235483870967744e-05, "loss": 0.3932, "step": 7057 }, { "epoch": 45.49718875502008, "grad_norm": 4.339266300201416, "learning_rate": 2.723225806451613e-05, "loss": 0.1503, "step": 7058 }, { "epoch": 45.503614457831326, "grad_norm": 48.8013916015625, "learning_rate": 2.722903225806452e-05, "loss": 0.4836, "step": 7059 }, { "epoch": 45.51004016064257, "grad_norm": 10.434020042419434, "learning_rate": 2.7225806451612905e-05, "loss": 0.1941, "step": 7060 }, { "epoch": 45.51646586345382, "grad_norm": 5.4785261154174805, "learning_rate": 2.7222580645161294e-05, "loss": 0.2044, "step": 7061 }, { "epoch": 45.52289156626506, "grad_norm": 3.8729588985443115, "learning_rate": 2.7219354838709676e-05, "loss": 0.1777, "step": 7062 }, { "epoch": 45.52931726907631, "grad_norm": 11.190072059631348, "learning_rate": 2.7216129032258065e-05, "loss": 0.4519, "step": 7063 }, { "epoch": 45.53574297188755, "grad_norm": 8.425976753234863, "learning_rate": 2.721290322580645e-05, "loss": 0.5576, "step": 7064 }, { "epoch": 45.54216867469879, "grad_norm": 5.497267246246338, "learning_rate": 2.720967741935484e-05, "loss": 0.3691, "step": 7065 }, { "epoch": 45.54859437751004, "grad_norm": 9.134222984313965, "learning_rate": 2.720645161290323e-05, "loss": 0.5217, "step": 7066 }, { "epoch": 45.555020080321285, "grad_norm": 15.075708389282227, "learning_rate": 2.7203225806451615e-05, "loss": 0.2711, "step": 7067 }, { "epoch": 45.56144578313253, "grad_norm": 7.794920444488525, "learning_rate": 2.7200000000000004e-05, "loss": 0.3074, "step": 7068 }, { "epoch": 45.567871485943776, "grad_norm": 18.269620895385742, "learning_rate": 2.719677419354839e-05, "loss": 0.4908, "step": 7069 }, { "epoch": 45.57429718875502, "grad_norm": 11.50418758392334, "learning_rate": 2.7193548387096772e-05, "loss": 0.4462, "step": 7070 }, { "epoch": 45.58072289156627, "grad_norm": 19.06467628479004, "learning_rate": 2.719032258064516e-05, "loss": 0.5923, "step": 7071 }, { "epoch": 45.58714859437751, "grad_norm": 17.572175979614258, "learning_rate": 2.718709677419355e-05, "loss": 0.3826, "step": 7072 }, { "epoch": 45.59357429718875, "grad_norm": 16.95147705078125, "learning_rate": 2.7183870967741936e-05, "loss": 0.3809, "step": 7073 }, { "epoch": 45.6, "grad_norm": 17.974960327148438, "learning_rate": 2.7180645161290325e-05, "loss": 0.3299, "step": 7074 }, { "epoch": 45.606425702811244, "grad_norm": 7.986781120300293, "learning_rate": 2.7177419354838714e-05, "loss": 0.2059, "step": 7075 }, { "epoch": 45.61285140562249, "grad_norm": 18.82197380065918, "learning_rate": 2.71741935483871e-05, "loss": 0.3579, "step": 7076 }, { "epoch": 45.619277108433735, "grad_norm": 12.29666805267334, "learning_rate": 2.7170967741935482e-05, "loss": 0.4514, "step": 7077 }, { "epoch": 45.62570281124498, "grad_norm": 6.2474141120910645, "learning_rate": 2.716774193548387e-05, "loss": 0.4023, "step": 7078 }, { "epoch": 45.63212851405623, "grad_norm": 8.986905097961426, "learning_rate": 2.7164516129032257e-05, "loss": 0.3793, "step": 7079 }, { "epoch": 45.63855421686747, "grad_norm": 25.711942672729492, "learning_rate": 2.7161290322580646e-05, "loss": 0.5694, "step": 7080 }, { "epoch": 45.64497991967872, "grad_norm": 15.683199882507324, "learning_rate": 2.7158064516129035e-05, "loss": 0.262, "step": 7081 }, { "epoch": 45.65140562248996, "grad_norm": 4.228419303894043, "learning_rate": 2.715483870967742e-05, "loss": 0.1466, "step": 7082 }, { "epoch": 45.6578313253012, "grad_norm": 11.49387264251709, "learning_rate": 2.715161290322581e-05, "loss": 0.4445, "step": 7083 }, { "epoch": 45.66425702811245, "grad_norm": 8.715044975280762, "learning_rate": 2.7148387096774195e-05, "loss": 0.263, "step": 7084 }, { "epoch": 45.670682730923694, "grad_norm": 5.543072700500488, "learning_rate": 2.7145161290322578e-05, "loss": 0.1514, "step": 7085 }, { "epoch": 45.67710843373494, "grad_norm": 8.305353164672852, "learning_rate": 2.7141935483870967e-05, "loss": 0.3972, "step": 7086 }, { "epoch": 45.683534136546186, "grad_norm": 12.693012237548828, "learning_rate": 2.7138709677419356e-05, "loss": 0.3597, "step": 7087 }, { "epoch": 45.68995983935743, "grad_norm": 63.93213653564453, "learning_rate": 2.7135483870967742e-05, "loss": 0.4551, "step": 7088 }, { "epoch": 45.69638554216868, "grad_norm": 21.871427536010742, "learning_rate": 2.713225806451613e-05, "loss": 0.8596, "step": 7089 }, { "epoch": 45.70281124497992, "grad_norm": 17.79832649230957, "learning_rate": 2.712903225806452e-05, "loss": 0.3031, "step": 7090 }, { "epoch": 45.70923694779116, "grad_norm": 5.4100141525268555, "learning_rate": 2.7125806451612906e-05, "loss": 0.3472, "step": 7091 }, { "epoch": 45.71566265060241, "grad_norm": 8.123132705688477, "learning_rate": 2.7122580645161295e-05, "loss": 0.4081, "step": 7092 }, { "epoch": 45.72208835341365, "grad_norm": 3.6061999797821045, "learning_rate": 2.7119354838709677e-05, "loss": 0.134, "step": 7093 }, { "epoch": 45.7285140562249, "grad_norm": 8.485245704650879, "learning_rate": 2.7116129032258063e-05, "loss": 0.6341, "step": 7094 }, { "epoch": 45.734939759036145, "grad_norm": 6.815056324005127, "learning_rate": 2.7112903225806452e-05, "loss": 0.2401, "step": 7095 }, { "epoch": 45.74136546184739, "grad_norm": 8.649275779724121, "learning_rate": 2.710967741935484e-05, "loss": 0.406, "step": 7096 }, { "epoch": 45.747791164658636, "grad_norm": 9.177726745605469, "learning_rate": 2.7106451612903227e-05, "loss": 0.3829, "step": 7097 }, { "epoch": 45.75421686746988, "grad_norm": 4.999598503112793, "learning_rate": 2.7103225806451616e-05, "loss": 0.2884, "step": 7098 }, { "epoch": 45.76064257028113, "grad_norm": 4.77438497543335, "learning_rate": 2.7100000000000005e-05, "loss": 0.1518, "step": 7099 }, { "epoch": 45.76706827309237, "grad_norm": 5.652735233306885, "learning_rate": 2.709677419354839e-05, "loss": 0.3851, "step": 7100 }, { "epoch": 45.77349397590361, "grad_norm": 12.919740676879883, "learning_rate": 2.7093548387096773e-05, "loss": 0.3265, "step": 7101 }, { "epoch": 45.77991967871486, "grad_norm": 17.5384521484375, "learning_rate": 2.7090322580645162e-05, "loss": 0.1766, "step": 7102 }, { "epoch": 45.786345381526104, "grad_norm": 28.26291275024414, "learning_rate": 2.7087096774193548e-05, "loss": 0.4099, "step": 7103 }, { "epoch": 45.792771084337346, "grad_norm": 9.184724807739258, "learning_rate": 2.7083870967741937e-05, "loss": 0.2384, "step": 7104 }, { "epoch": 45.799196787148595, "grad_norm": 13.117751121520996, "learning_rate": 2.7080645161290326e-05, "loss": 0.5372, "step": 7105 }, { "epoch": 45.80562248995984, "grad_norm": 6.63515567779541, "learning_rate": 2.707741935483871e-05, "loss": 0.2871, "step": 7106 }, { "epoch": 45.81204819277109, "grad_norm": 8.41846752166748, "learning_rate": 2.70741935483871e-05, "loss": 0.2829, "step": 7107 }, { "epoch": 45.81847389558233, "grad_norm": 29.03434181213379, "learning_rate": 2.7070967741935486e-05, "loss": 0.2638, "step": 7108 }, { "epoch": 45.82489959839357, "grad_norm": 10.832329750061035, "learning_rate": 2.706774193548387e-05, "loss": 0.37, "step": 7109 }, { "epoch": 45.83132530120482, "grad_norm": 6.399376392364502, "learning_rate": 2.7064516129032258e-05, "loss": 0.4601, "step": 7110 }, { "epoch": 45.83775100401606, "grad_norm": 18.04751205444336, "learning_rate": 2.7061290322580647e-05, "loss": 0.4963, "step": 7111 }, { "epoch": 45.84417670682731, "grad_norm": 6.756120681762695, "learning_rate": 2.7058064516129033e-05, "loss": 0.2397, "step": 7112 }, { "epoch": 45.850602409638554, "grad_norm": 3.7867448329925537, "learning_rate": 2.705483870967742e-05, "loss": 0.2026, "step": 7113 }, { "epoch": 45.8570281124498, "grad_norm": 8.091314315795898, "learning_rate": 2.705161290322581e-05, "loss": 0.3137, "step": 7114 }, { "epoch": 45.863453815261046, "grad_norm": 10.77336311340332, "learning_rate": 2.7048387096774196e-05, "loss": 0.3519, "step": 7115 }, { "epoch": 45.86987951807229, "grad_norm": 4.726949214935303, "learning_rate": 2.704516129032258e-05, "loss": 0.3659, "step": 7116 }, { "epoch": 45.87630522088354, "grad_norm": 12.863045692443848, "learning_rate": 2.7041935483870968e-05, "loss": 0.4061, "step": 7117 }, { "epoch": 45.88273092369478, "grad_norm": 4.354026794433594, "learning_rate": 2.7038709677419354e-05, "loss": 0.1581, "step": 7118 }, { "epoch": 45.88915662650602, "grad_norm": 14.14599895477295, "learning_rate": 2.7035483870967743e-05, "loss": 0.4125, "step": 7119 }, { "epoch": 45.89558232931727, "grad_norm": 11.542196273803711, "learning_rate": 2.7032258064516132e-05, "loss": 0.3523, "step": 7120 }, { "epoch": 45.90200803212851, "grad_norm": 4.841372966766357, "learning_rate": 2.7029032258064517e-05, "loss": 0.2457, "step": 7121 }, { "epoch": 45.908433734939756, "grad_norm": 10.962514877319336, "learning_rate": 2.7025806451612907e-05, "loss": 0.3111, "step": 7122 }, { "epoch": 45.914859437751005, "grad_norm": 17.7298526763916, "learning_rate": 2.7022580645161292e-05, "loss": 0.6367, "step": 7123 }, { "epoch": 45.92128514056225, "grad_norm": 5.713824272155762, "learning_rate": 2.7019354838709675e-05, "loss": 0.1861, "step": 7124 }, { "epoch": 45.9277108433735, "grad_norm": 18.34485626220703, "learning_rate": 2.7016129032258064e-05, "loss": 0.3081, "step": 7125 }, { "epoch": 45.93413654618474, "grad_norm": 19.108673095703125, "learning_rate": 2.7012903225806453e-05, "loss": 0.4799, "step": 7126 }, { "epoch": 45.94056224899598, "grad_norm": 6.695702075958252, "learning_rate": 2.700967741935484e-05, "loss": 0.1686, "step": 7127 }, { "epoch": 45.94698795180723, "grad_norm": 12.093729019165039, "learning_rate": 2.7006451612903228e-05, "loss": 0.4933, "step": 7128 }, { "epoch": 45.95341365461847, "grad_norm": 10.242436408996582, "learning_rate": 2.7003225806451617e-05, "loss": 0.3228, "step": 7129 }, { "epoch": 45.95983935742972, "grad_norm": 6.4422526359558105, "learning_rate": 2.7000000000000002e-05, "loss": 0.2797, "step": 7130 }, { "epoch": 45.966265060240964, "grad_norm": 8.544782638549805, "learning_rate": 2.699677419354839e-05, "loss": 0.3584, "step": 7131 }, { "epoch": 45.972690763052206, "grad_norm": 10.161892890930176, "learning_rate": 2.6993548387096774e-05, "loss": 0.1152, "step": 7132 }, { "epoch": 45.979116465863456, "grad_norm": 2.201974630355835, "learning_rate": 2.699032258064516e-05, "loss": 0.1515, "step": 7133 }, { "epoch": 45.9855421686747, "grad_norm": 7.9581804275512695, "learning_rate": 2.698709677419355e-05, "loss": 0.6336, "step": 7134 }, { "epoch": 45.99196787148595, "grad_norm": 7.809915542602539, "learning_rate": 2.6983870967741938e-05, "loss": 0.4042, "step": 7135 }, { "epoch": 46.00160642570281, "grad_norm": 8.604445457458496, "learning_rate": 2.6980645161290323e-05, "loss": 0.3809, "step": 7136 }, { "epoch": 46.00803212851405, "grad_norm": 12.202855110168457, "learning_rate": 2.6977419354838712e-05, "loss": 0.6438, "step": 7137 }, { "epoch": 46.0144578313253, "grad_norm": 15.134500503540039, "learning_rate": 2.69741935483871e-05, "loss": 0.5023, "step": 7138 }, { "epoch": 46.020883534136544, "grad_norm": 6.3560919761657715, "learning_rate": 2.6970967741935487e-05, "loss": 0.2597, "step": 7139 }, { "epoch": 46.027309236947794, "grad_norm": 9.29275131225586, "learning_rate": 2.696774193548387e-05, "loss": 0.3097, "step": 7140 }, { "epoch": 46.033734939759036, "grad_norm": 11.41480827331543, "learning_rate": 2.696451612903226e-05, "loss": 0.3943, "step": 7141 }, { "epoch": 46.04016064257028, "grad_norm": 15.14299201965332, "learning_rate": 2.6961290322580644e-05, "loss": 0.4718, "step": 7142 }, { "epoch": 46.04658634538153, "grad_norm": 12.160162925720215, "learning_rate": 2.6958064516129034e-05, "loss": 0.466, "step": 7143 }, { "epoch": 46.05301204819277, "grad_norm": 5.529515743255615, "learning_rate": 2.6954838709677423e-05, "loss": 0.3542, "step": 7144 }, { "epoch": 46.05943775100402, "grad_norm": 16.489351272583008, "learning_rate": 2.695161290322581e-05, "loss": 0.4471, "step": 7145 }, { "epoch": 46.06586345381526, "grad_norm": 42.0651741027832, "learning_rate": 2.6948387096774197e-05, "loss": 0.4773, "step": 7146 }, { "epoch": 46.0722891566265, "grad_norm": 9.29786491394043, "learning_rate": 2.694516129032258e-05, "loss": 0.3473, "step": 7147 }, { "epoch": 46.07871485943775, "grad_norm": 24.962848663330078, "learning_rate": 2.6941935483870965e-05, "loss": 0.4306, "step": 7148 }, { "epoch": 46.085140562248995, "grad_norm": 29.43343162536621, "learning_rate": 2.6938709677419355e-05, "loss": 0.6086, "step": 7149 }, { "epoch": 46.091566265060244, "grad_norm": 6.407468318939209, "learning_rate": 2.6935483870967744e-05, "loss": 0.3582, "step": 7150 }, { "epoch": 46.09799196787149, "grad_norm": 18.998441696166992, "learning_rate": 2.693225806451613e-05, "loss": 0.4037, "step": 7151 }, { "epoch": 46.10441767068273, "grad_norm": 6.336942195892334, "learning_rate": 2.692903225806452e-05, "loss": 0.3617, "step": 7152 }, { "epoch": 46.11084337349398, "grad_norm": 12.114083290100098, "learning_rate": 2.6925806451612908e-05, "loss": 0.2695, "step": 7153 }, { "epoch": 46.11726907630522, "grad_norm": 12.334098815917969, "learning_rate": 2.6922580645161293e-05, "loss": 0.3527, "step": 7154 }, { "epoch": 46.12369477911646, "grad_norm": 10.633148193359375, "learning_rate": 2.6919354838709676e-05, "loss": 0.0984, "step": 7155 }, { "epoch": 46.13012048192771, "grad_norm": 13.504015922546387, "learning_rate": 2.6916129032258065e-05, "loss": 0.4265, "step": 7156 }, { "epoch": 46.136546184738954, "grad_norm": 15.06752872467041, "learning_rate": 2.691290322580645e-05, "loss": 0.3469, "step": 7157 }, { "epoch": 46.1429718875502, "grad_norm": 9.2946138381958, "learning_rate": 2.690967741935484e-05, "loss": 0.4944, "step": 7158 }, { "epoch": 46.149397590361446, "grad_norm": 8.700982093811035, "learning_rate": 2.690645161290323e-05, "loss": 0.2355, "step": 7159 }, { "epoch": 46.15582329317269, "grad_norm": 9.374290466308594, "learning_rate": 2.6903225806451614e-05, "loss": 0.2506, "step": 7160 }, { "epoch": 46.16224899598394, "grad_norm": 3.9652962684631348, "learning_rate": 2.6900000000000003e-05, "loss": 0.3625, "step": 7161 }, { "epoch": 46.16867469879518, "grad_norm": 7.104926586151123, "learning_rate": 2.6896774193548392e-05, "loss": 0.3141, "step": 7162 }, { "epoch": 46.17510040160643, "grad_norm": 22.54165267944336, "learning_rate": 2.689354838709677e-05, "loss": 0.4229, "step": 7163 }, { "epoch": 46.18152610441767, "grad_norm": 8.237348556518555, "learning_rate": 2.689032258064516e-05, "loss": 0.3264, "step": 7164 }, { "epoch": 46.18795180722891, "grad_norm": 6.234988212585449, "learning_rate": 2.688709677419355e-05, "loss": 0.4064, "step": 7165 }, { "epoch": 46.19437751004016, "grad_norm": 14.445403099060059, "learning_rate": 2.6883870967741935e-05, "loss": 0.4292, "step": 7166 }, { "epoch": 46.200803212851405, "grad_norm": 10.208695411682129, "learning_rate": 2.6880645161290324e-05, "loss": 0.6038, "step": 7167 }, { "epoch": 46.207228915662654, "grad_norm": 6.355615139007568, "learning_rate": 2.6877419354838713e-05, "loss": 0.2193, "step": 7168 }, { "epoch": 46.213654618473896, "grad_norm": 7.528269290924072, "learning_rate": 2.68741935483871e-05, "loss": 0.3973, "step": 7169 }, { "epoch": 46.22008032128514, "grad_norm": 8.044272422790527, "learning_rate": 2.6870967741935488e-05, "loss": 0.4957, "step": 7170 }, { "epoch": 46.22650602409639, "grad_norm": 12.977578163146973, "learning_rate": 2.686774193548387e-05, "loss": 0.4722, "step": 7171 }, { "epoch": 46.23293172690763, "grad_norm": 5.7378339767456055, "learning_rate": 2.6864516129032256e-05, "loss": 0.2294, "step": 7172 }, { "epoch": 46.23935742971887, "grad_norm": 41.1268196105957, "learning_rate": 2.6861290322580645e-05, "loss": 0.4895, "step": 7173 }, { "epoch": 46.24578313253012, "grad_norm": 36.98874282836914, "learning_rate": 2.6858064516129034e-05, "loss": 0.7428, "step": 7174 }, { "epoch": 46.252208835341364, "grad_norm": 9.553424835205078, "learning_rate": 2.685483870967742e-05, "loss": 0.4593, "step": 7175 }, { "epoch": 46.25863453815261, "grad_norm": 10.204558372497559, "learning_rate": 2.685161290322581e-05, "loss": 0.3906, "step": 7176 }, { "epoch": 46.265060240963855, "grad_norm": 10.45246410369873, "learning_rate": 2.68483870967742e-05, "loss": 0.3707, "step": 7177 }, { "epoch": 46.2714859437751, "grad_norm": 15.480936050415039, "learning_rate": 2.6845161290322584e-05, "loss": 0.6136, "step": 7178 }, { "epoch": 46.27791164658635, "grad_norm": 11.834671020507812, "learning_rate": 2.6841935483870966e-05, "loss": 0.306, "step": 7179 }, { "epoch": 46.28433734939759, "grad_norm": 15.976999282836914, "learning_rate": 2.6838709677419355e-05, "loss": 0.3464, "step": 7180 }, { "epoch": 46.29076305220884, "grad_norm": 7.373652458190918, "learning_rate": 2.683548387096774e-05, "loss": 0.7012, "step": 7181 }, { "epoch": 46.29718875502008, "grad_norm": 12.3583984375, "learning_rate": 2.683225806451613e-05, "loss": 0.4657, "step": 7182 }, { "epoch": 46.30361445783132, "grad_norm": 44.270877838134766, "learning_rate": 2.682903225806452e-05, "loss": 0.3368, "step": 7183 }, { "epoch": 46.31004016064257, "grad_norm": 15.512558937072754, "learning_rate": 2.6825806451612905e-05, "loss": 0.3726, "step": 7184 }, { "epoch": 46.316465863453814, "grad_norm": 4.616358757019043, "learning_rate": 2.6822580645161294e-05, "loss": 0.3347, "step": 7185 }, { "epoch": 46.32289156626506, "grad_norm": 11.112526893615723, "learning_rate": 2.6819354838709676e-05, "loss": 0.2993, "step": 7186 }, { "epoch": 46.329317269076306, "grad_norm": 12.45744800567627, "learning_rate": 2.6816129032258062e-05, "loss": 0.6624, "step": 7187 }, { "epoch": 46.33574297188755, "grad_norm": 11.872723579406738, "learning_rate": 2.681290322580645e-05, "loss": 0.2209, "step": 7188 }, { "epoch": 46.3421686746988, "grad_norm": 7.217973709106445, "learning_rate": 2.680967741935484e-05, "loss": 0.1621, "step": 7189 }, { "epoch": 46.34859437751004, "grad_norm": 7.5334672927856445, "learning_rate": 2.6806451612903226e-05, "loss": 0.292, "step": 7190 }, { "epoch": 46.35502008032128, "grad_norm": 7.652576923370361, "learning_rate": 2.6803225806451615e-05, "loss": 0.4126, "step": 7191 }, { "epoch": 46.36144578313253, "grad_norm": 9.172607421875, "learning_rate": 2.6800000000000004e-05, "loss": 0.4724, "step": 7192 }, { "epoch": 46.36787148594377, "grad_norm": 13.341243743896484, "learning_rate": 2.679677419354839e-05, "loss": 0.5024, "step": 7193 }, { "epoch": 46.37429718875502, "grad_norm": 5.773445129394531, "learning_rate": 2.6793548387096772e-05, "loss": 0.2725, "step": 7194 }, { "epoch": 46.380722891566265, "grad_norm": 10.269645690917969, "learning_rate": 2.679032258064516e-05, "loss": 0.5031, "step": 7195 }, { "epoch": 46.38714859437751, "grad_norm": 6.449611186981201, "learning_rate": 2.6787096774193547e-05, "loss": 0.1897, "step": 7196 }, { "epoch": 46.393574297188756, "grad_norm": 8.311967849731445, "learning_rate": 2.6783870967741936e-05, "loss": 0.2273, "step": 7197 }, { "epoch": 46.4, "grad_norm": 20.120512008666992, "learning_rate": 2.6780645161290325e-05, "loss": 0.3988, "step": 7198 }, { "epoch": 46.40642570281125, "grad_norm": 8.852877616882324, "learning_rate": 2.677741935483871e-05, "loss": 0.282, "step": 7199 }, { "epoch": 46.41285140562249, "grad_norm": 7.935997486114502, "learning_rate": 2.67741935483871e-05, "loss": 0.2696, "step": 7200 }, { "epoch": 46.41927710843373, "grad_norm": 7.958041667938232, "learning_rate": 2.677096774193549e-05, "loss": 0.3305, "step": 7201 }, { "epoch": 46.42570281124498, "grad_norm": 15.39387321472168, "learning_rate": 2.676774193548387e-05, "loss": 0.5107, "step": 7202 }, { "epoch": 46.432128514056224, "grad_norm": 29.810483932495117, "learning_rate": 2.6764516129032257e-05, "loss": 0.3749, "step": 7203 }, { "epoch": 46.43855421686747, "grad_norm": 18.234947204589844, "learning_rate": 2.6761290322580646e-05, "loss": 0.4672, "step": 7204 }, { "epoch": 46.444979919678715, "grad_norm": 14.385478973388672, "learning_rate": 2.6758064516129032e-05, "loss": 0.4476, "step": 7205 }, { "epoch": 46.45140562248996, "grad_norm": 4.2725958824157715, "learning_rate": 2.675483870967742e-05, "loss": 0.3275, "step": 7206 }, { "epoch": 46.45783132530121, "grad_norm": 8.81848430633545, "learning_rate": 2.675161290322581e-05, "loss": 0.311, "step": 7207 }, { "epoch": 46.46425702811245, "grad_norm": 10.859129905700684, "learning_rate": 2.6748387096774196e-05, "loss": 0.5038, "step": 7208 }, { "epoch": 46.47068273092369, "grad_norm": 10.292496681213379, "learning_rate": 2.6745161290322585e-05, "loss": 0.6015, "step": 7209 }, { "epoch": 46.47710843373494, "grad_norm": 9.055238723754883, "learning_rate": 2.6741935483870967e-05, "loss": 0.2523, "step": 7210 }, { "epoch": 46.48353413654618, "grad_norm": 9.930639266967773, "learning_rate": 2.6738709677419353e-05, "loss": 0.3834, "step": 7211 }, { "epoch": 46.48995983935743, "grad_norm": 19.387767791748047, "learning_rate": 2.6735483870967742e-05, "loss": 0.504, "step": 7212 }, { "epoch": 46.496385542168674, "grad_norm": 3.6721794605255127, "learning_rate": 2.673225806451613e-05, "loss": 0.1744, "step": 7213 }, { "epoch": 46.50281124497992, "grad_norm": 27.692277908325195, "learning_rate": 2.6729032258064517e-05, "loss": 0.2808, "step": 7214 }, { "epoch": 46.509236947791166, "grad_norm": 6.922491550445557, "learning_rate": 2.6725806451612906e-05, "loss": 0.3134, "step": 7215 }, { "epoch": 46.51566265060241, "grad_norm": 14.335888862609863, "learning_rate": 2.6722580645161295e-05, "loss": 0.5529, "step": 7216 }, { "epoch": 46.52208835341366, "grad_norm": 6.639595031738281, "learning_rate": 2.6719354838709677e-05, "loss": 0.2518, "step": 7217 }, { "epoch": 46.5285140562249, "grad_norm": 10.980874061584473, "learning_rate": 2.6716129032258063e-05, "loss": 0.4673, "step": 7218 }, { "epoch": 46.53493975903614, "grad_norm": 6.9065022468566895, "learning_rate": 2.6712903225806452e-05, "loss": 0.4081, "step": 7219 }, { "epoch": 46.54136546184739, "grad_norm": 13.673219680786133, "learning_rate": 2.6709677419354838e-05, "loss": 0.4864, "step": 7220 }, { "epoch": 46.54779116465863, "grad_norm": 13.78917121887207, "learning_rate": 2.6706451612903227e-05, "loss": 0.7241, "step": 7221 }, { "epoch": 46.55421686746988, "grad_norm": 5.415931701660156, "learning_rate": 2.6703225806451616e-05, "loss": 0.2377, "step": 7222 }, { "epoch": 46.560642570281125, "grad_norm": 35.20075607299805, "learning_rate": 2.6700000000000002e-05, "loss": 0.5837, "step": 7223 }, { "epoch": 46.56706827309237, "grad_norm": 13.854165077209473, "learning_rate": 2.669677419354839e-05, "loss": 0.6756, "step": 7224 }, { "epoch": 46.57349397590362, "grad_norm": 9.141483306884766, "learning_rate": 2.6693548387096773e-05, "loss": 0.1371, "step": 7225 }, { "epoch": 46.57991967871486, "grad_norm": 6.428073406219482, "learning_rate": 2.669032258064516e-05, "loss": 0.2379, "step": 7226 }, { "epoch": 46.5863453815261, "grad_norm": 14.487081527709961, "learning_rate": 2.6687096774193548e-05, "loss": 0.1709, "step": 7227 }, { "epoch": 46.59277108433735, "grad_norm": 12.93925666809082, "learning_rate": 2.6683870967741937e-05, "loss": 0.2966, "step": 7228 }, { "epoch": 46.59919678714859, "grad_norm": 4.203712463378906, "learning_rate": 2.6680645161290323e-05, "loss": 0.4276, "step": 7229 }, { "epoch": 46.60562248995984, "grad_norm": 12.739593505859375, "learning_rate": 2.6677419354838712e-05, "loss": 0.2645, "step": 7230 }, { "epoch": 46.612048192771084, "grad_norm": 19.38926124572754, "learning_rate": 2.66741935483871e-05, "loss": 0.626, "step": 7231 }, { "epoch": 46.618473895582326, "grad_norm": 13.6903715133667, "learning_rate": 2.6670967741935487e-05, "loss": 0.4484, "step": 7232 }, { "epoch": 46.624899598393576, "grad_norm": 11.740065574645996, "learning_rate": 2.666774193548387e-05, "loss": 0.3799, "step": 7233 }, { "epoch": 46.63132530120482, "grad_norm": 12.595711708068848, "learning_rate": 2.6664516129032258e-05, "loss": 0.3813, "step": 7234 }, { "epoch": 46.63775100401607, "grad_norm": 12.0125732421875, "learning_rate": 2.6661290322580644e-05, "loss": 0.3179, "step": 7235 }, { "epoch": 46.64417670682731, "grad_norm": 10.742772102355957, "learning_rate": 2.6658064516129033e-05, "loss": 0.1643, "step": 7236 }, { "epoch": 46.65060240963855, "grad_norm": 14.545305252075195, "learning_rate": 2.6654838709677422e-05, "loss": 0.3094, "step": 7237 }, { "epoch": 46.6570281124498, "grad_norm": 6.945914268493652, "learning_rate": 2.6651612903225808e-05, "loss": 0.3154, "step": 7238 }, { "epoch": 46.66345381526104, "grad_norm": 15.65876293182373, "learning_rate": 2.6648387096774197e-05, "loss": 0.1851, "step": 7239 }, { "epoch": 46.66987951807229, "grad_norm": 3.967341423034668, "learning_rate": 2.6645161290322586e-05, "loss": 0.2484, "step": 7240 }, { "epoch": 46.676305220883535, "grad_norm": 10.874924659729004, "learning_rate": 2.6641935483870968e-05, "loss": 0.2068, "step": 7241 }, { "epoch": 46.68273092369478, "grad_norm": 11.025912284851074, "learning_rate": 2.6638709677419354e-05, "loss": 0.681, "step": 7242 }, { "epoch": 46.689156626506026, "grad_norm": 10.571778297424316, "learning_rate": 2.6635483870967743e-05, "loss": 0.1805, "step": 7243 }, { "epoch": 46.69558232931727, "grad_norm": 17.87015151977539, "learning_rate": 2.663225806451613e-05, "loss": 0.3544, "step": 7244 }, { "epoch": 46.70200803212851, "grad_norm": 20.815881729125977, "learning_rate": 2.6629032258064518e-05, "loss": 0.3987, "step": 7245 }, { "epoch": 46.70843373493976, "grad_norm": 11.724069595336914, "learning_rate": 2.6625806451612907e-05, "loss": 0.2491, "step": 7246 }, { "epoch": 46.714859437751, "grad_norm": 16.261798858642578, "learning_rate": 2.6622580645161293e-05, "loss": 0.3291, "step": 7247 }, { "epoch": 46.72128514056225, "grad_norm": 12.788185119628906, "learning_rate": 2.6619354838709682e-05, "loss": 0.4198, "step": 7248 }, { "epoch": 46.727710843373494, "grad_norm": 6.038384914398193, "learning_rate": 2.6616129032258064e-05, "loss": 0.3621, "step": 7249 }, { "epoch": 46.734136546184736, "grad_norm": 6.111212730407715, "learning_rate": 2.661290322580645e-05, "loss": 0.343, "step": 7250 }, { "epoch": 46.740562248995985, "grad_norm": 13.837345123291016, "learning_rate": 2.660967741935484e-05, "loss": 0.4622, "step": 7251 }, { "epoch": 46.74698795180723, "grad_norm": 25.075969696044922, "learning_rate": 2.6606451612903228e-05, "loss": 0.411, "step": 7252 }, { "epoch": 46.75341365461848, "grad_norm": 5.429897308349609, "learning_rate": 2.6603225806451614e-05, "loss": 0.301, "step": 7253 }, { "epoch": 46.75983935742972, "grad_norm": 9.607427597045898, "learning_rate": 2.6600000000000003e-05, "loss": 0.4502, "step": 7254 }, { "epoch": 46.76626506024096, "grad_norm": 20.408924102783203, "learning_rate": 2.6596774193548392e-05, "loss": 0.2675, "step": 7255 }, { "epoch": 46.77269076305221, "grad_norm": 7.440409183502197, "learning_rate": 2.6593548387096774e-05, "loss": 0.3384, "step": 7256 }, { "epoch": 46.77911646586345, "grad_norm": 9.499515533447266, "learning_rate": 2.659032258064516e-05, "loss": 0.255, "step": 7257 }, { "epoch": 46.7855421686747, "grad_norm": 12.534220695495605, "learning_rate": 2.658709677419355e-05, "loss": 0.2402, "step": 7258 }, { "epoch": 46.791967871485944, "grad_norm": 8.169532775878906, "learning_rate": 2.6583870967741935e-05, "loss": 0.2741, "step": 7259 }, { "epoch": 46.79839357429719, "grad_norm": 10.07476806640625, "learning_rate": 2.6580645161290324e-05, "loss": 0.3474, "step": 7260 }, { "epoch": 46.804819277108436, "grad_norm": 7.838259220123291, "learning_rate": 2.6577419354838713e-05, "loss": 0.1536, "step": 7261 }, { "epoch": 46.81124497991968, "grad_norm": 6.4369683265686035, "learning_rate": 2.65741935483871e-05, "loss": 0.2962, "step": 7262 }, { "epoch": 46.81767068273092, "grad_norm": 20.496774673461914, "learning_rate": 2.6570967741935488e-05, "loss": 0.4136, "step": 7263 }, { "epoch": 46.82409638554217, "grad_norm": 3.7203965187072754, "learning_rate": 2.656774193548387e-05, "loss": 0.2189, "step": 7264 }, { "epoch": 46.83052208835341, "grad_norm": 13.347314834594727, "learning_rate": 2.656451612903226e-05, "loss": 0.4985, "step": 7265 }, { "epoch": 46.83694779116466, "grad_norm": 9.057223320007324, "learning_rate": 2.6561290322580645e-05, "loss": 0.2784, "step": 7266 }, { "epoch": 46.8433734939759, "grad_norm": 12.840435028076172, "learning_rate": 2.6558064516129034e-05, "loss": 0.4724, "step": 7267 }, { "epoch": 46.849799196787146, "grad_norm": 8.541585922241211, "learning_rate": 2.655483870967742e-05, "loss": 0.2986, "step": 7268 }, { "epoch": 46.856224899598395, "grad_norm": 8.478144645690918, "learning_rate": 2.655161290322581e-05, "loss": 0.2512, "step": 7269 }, { "epoch": 46.86265060240964, "grad_norm": 19.176023483276367, "learning_rate": 2.6548387096774198e-05, "loss": 0.5744, "step": 7270 }, { "epoch": 46.86907630522089, "grad_norm": 14.181173324584961, "learning_rate": 2.6545161290322584e-05, "loss": 0.6045, "step": 7271 }, { "epoch": 46.87550200803213, "grad_norm": 7.799170017242432, "learning_rate": 2.6541935483870966e-05, "loss": 0.2764, "step": 7272 }, { "epoch": 46.88192771084337, "grad_norm": 14.989847183227539, "learning_rate": 2.6538709677419355e-05, "loss": 0.1442, "step": 7273 }, { "epoch": 46.88835341365462, "grad_norm": 7.352709770202637, "learning_rate": 2.653548387096774e-05, "loss": 0.4375, "step": 7274 }, { "epoch": 46.89477911646586, "grad_norm": 12.735052108764648, "learning_rate": 2.653225806451613e-05, "loss": 0.5133, "step": 7275 }, { "epoch": 46.90120481927711, "grad_norm": 10.59900951385498, "learning_rate": 2.652903225806452e-05, "loss": 0.4199, "step": 7276 }, { "epoch": 46.907630522088354, "grad_norm": 15.412851333618164, "learning_rate": 2.6525806451612905e-05, "loss": 0.1168, "step": 7277 }, { "epoch": 46.914056224899596, "grad_norm": 8.667595863342285, "learning_rate": 2.6522580645161294e-05, "loss": 0.3008, "step": 7278 }, { "epoch": 46.920481927710846, "grad_norm": 13.81971263885498, "learning_rate": 2.6519354838709683e-05, "loss": 0.6258, "step": 7279 }, { "epoch": 46.92690763052209, "grad_norm": 12.666378021240234, "learning_rate": 2.6516129032258065e-05, "loss": 0.5614, "step": 7280 }, { "epoch": 46.93333333333333, "grad_norm": 14.676054954528809, "learning_rate": 2.651290322580645e-05, "loss": 0.4309, "step": 7281 }, { "epoch": 46.93975903614458, "grad_norm": 7.0286993980407715, "learning_rate": 2.650967741935484e-05, "loss": 0.2915, "step": 7282 }, { "epoch": 46.94618473895582, "grad_norm": 10.832810401916504, "learning_rate": 2.6506451612903226e-05, "loss": 0.3795, "step": 7283 }, { "epoch": 46.95261044176707, "grad_norm": 14.61186695098877, "learning_rate": 2.6503225806451615e-05, "loss": 0.3443, "step": 7284 }, { "epoch": 46.95903614457831, "grad_norm": 7.4846320152282715, "learning_rate": 2.6500000000000004e-05, "loss": 0.1827, "step": 7285 }, { "epoch": 46.965461847389555, "grad_norm": 13.394513130187988, "learning_rate": 2.649677419354839e-05, "loss": 0.5611, "step": 7286 }, { "epoch": 46.971887550200805, "grad_norm": 12.430180549621582, "learning_rate": 2.649354838709678e-05, "loss": 0.5253, "step": 7287 }, { "epoch": 46.97831325301205, "grad_norm": 18.45766258239746, "learning_rate": 2.649032258064516e-05, "loss": 0.27, "step": 7288 }, { "epoch": 46.984738955823296, "grad_norm": 18.832233428955078, "learning_rate": 2.6487096774193547e-05, "loss": 0.3465, "step": 7289 }, { "epoch": 46.99116465863454, "grad_norm": 10.187567710876465, "learning_rate": 2.6483870967741936e-05, "loss": 0.2024, "step": 7290 }, { "epoch": 47.00080321285141, "grad_norm": 20.410381317138672, "learning_rate": 2.6480645161290325e-05, "loss": 0.3418, "step": 7291 }, { "epoch": 47.00722891566265, "grad_norm": 7.537530899047852, "learning_rate": 2.647741935483871e-05, "loss": 0.3986, "step": 7292 }, { "epoch": 47.01365461847389, "grad_norm": 11.456846237182617, "learning_rate": 2.64741935483871e-05, "loss": 0.6728, "step": 7293 }, { "epoch": 47.02008032128514, "grad_norm": 20.24042320251465, "learning_rate": 2.647096774193549e-05, "loss": 0.6505, "step": 7294 }, { "epoch": 47.026506024096385, "grad_norm": 5.3581318855285645, "learning_rate": 2.646774193548387e-05, "loss": 0.1351, "step": 7295 }, { "epoch": 47.03293172690763, "grad_norm": 9.344194412231445, "learning_rate": 2.6464516129032257e-05, "loss": 0.3508, "step": 7296 }, { "epoch": 47.039357429718876, "grad_norm": 17.222421646118164, "learning_rate": 2.6461290322580646e-05, "loss": 0.2782, "step": 7297 }, { "epoch": 47.04578313253012, "grad_norm": 4.819244861602783, "learning_rate": 2.645806451612903e-05, "loss": 0.3763, "step": 7298 }, { "epoch": 47.05220883534137, "grad_norm": 8.584444999694824, "learning_rate": 2.645483870967742e-05, "loss": 0.4484, "step": 7299 }, { "epoch": 47.05863453815261, "grad_norm": 8.692585945129395, "learning_rate": 2.645161290322581e-05, "loss": 0.5802, "step": 7300 }, { "epoch": 47.06506024096385, "grad_norm": 7.743775367736816, "learning_rate": 2.6448387096774195e-05, "loss": 0.2479, "step": 7301 }, { "epoch": 47.0714859437751, "grad_norm": 34.18688201904297, "learning_rate": 2.6445161290322585e-05, "loss": 0.6241, "step": 7302 }, { "epoch": 47.077911646586344, "grad_norm": 5.154871940612793, "learning_rate": 2.6441935483870967e-05, "loss": 0.2457, "step": 7303 }, { "epoch": 47.08433734939759, "grad_norm": 8.160508155822754, "learning_rate": 2.6438709677419356e-05, "loss": 0.343, "step": 7304 }, { "epoch": 47.090763052208835, "grad_norm": 26.78085708618164, "learning_rate": 2.643548387096774e-05, "loss": 0.2526, "step": 7305 }, { "epoch": 47.09718875502008, "grad_norm": 10.492220878601074, "learning_rate": 2.643225806451613e-05, "loss": 0.3198, "step": 7306 }, { "epoch": 47.10361445783133, "grad_norm": 16.645021438598633, "learning_rate": 2.6429032258064516e-05, "loss": 0.4639, "step": 7307 }, { "epoch": 47.11004016064257, "grad_norm": 11.388580322265625, "learning_rate": 2.6425806451612906e-05, "loss": 0.5201, "step": 7308 }, { "epoch": 47.11646586345382, "grad_norm": 7.791767120361328, "learning_rate": 2.6422580645161295e-05, "loss": 0.2196, "step": 7309 }, { "epoch": 47.12289156626506, "grad_norm": 17.93011474609375, "learning_rate": 2.641935483870968e-05, "loss": 0.4765, "step": 7310 }, { "epoch": 47.1293172690763, "grad_norm": 5.527345180511475, "learning_rate": 2.6416129032258063e-05, "loss": 0.1365, "step": 7311 }, { "epoch": 47.13574297188755, "grad_norm": 6.090267658233643, "learning_rate": 2.6412903225806452e-05, "loss": 0.237, "step": 7312 }, { "epoch": 47.142168674698794, "grad_norm": 7.704984664916992, "learning_rate": 2.6409677419354837e-05, "loss": 0.2429, "step": 7313 }, { "epoch": 47.14859437751004, "grad_norm": 6.177865505218506, "learning_rate": 2.6406451612903227e-05, "loss": 0.3969, "step": 7314 }, { "epoch": 47.155020080321286, "grad_norm": 9.047676086425781, "learning_rate": 2.6403225806451616e-05, "loss": 0.299, "step": 7315 }, { "epoch": 47.16144578313253, "grad_norm": 7.459741115570068, "learning_rate": 2.64e-05, "loss": 0.361, "step": 7316 }, { "epoch": 47.16787148594378, "grad_norm": 9.237130165100098, "learning_rate": 2.639677419354839e-05, "loss": 0.5029, "step": 7317 }, { "epoch": 47.17429718875502, "grad_norm": 9.10013484954834, "learning_rate": 2.639354838709678e-05, "loss": 0.428, "step": 7318 }, { "epoch": 47.18072289156626, "grad_norm": 11.908211708068848, "learning_rate": 2.6390322580645162e-05, "loss": 0.4207, "step": 7319 }, { "epoch": 47.18714859437751, "grad_norm": 8.715445518493652, "learning_rate": 2.6387096774193548e-05, "loss": 0.2803, "step": 7320 }, { "epoch": 47.19357429718875, "grad_norm": 4.716795444488525, "learning_rate": 2.6383870967741937e-05, "loss": 0.1883, "step": 7321 }, { "epoch": 47.2, "grad_norm": 6.706210613250732, "learning_rate": 2.6380645161290322e-05, "loss": 0.1935, "step": 7322 }, { "epoch": 47.206425702811245, "grad_norm": 9.332176208496094, "learning_rate": 2.637741935483871e-05, "loss": 0.1578, "step": 7323 }, { "epoch": 47.21285140562249, "grad_norm": 14.829580307006836, "learning_rate": 2.63741935483871e-05, "loss": 0.3149, "step": 7324 }, { "epoch": 47.21927710843374, "grad_norm": 12.741120338439941, "learning_rate": 2.6370967741935486e-05, "loss": 0.7361, "step": 7325 }, { "epoch": 47.22570281124498, "grad_norm": 9.008341789245605, "learning_rate": 2.636774193548387e-05, "loss": 0.4736, "step": 7326 }, { "epoch": 47.23212851405623, "grad_norm": 8.72082233428955, "learning_rate": 2.6364516129032258e-05, "loss": 0.3045, "step": 7327 }, { "epoch": 47.23855421686747, "grad_norm": 11.110326766967773, "learning_rate": 2.6361290322580647e-05, "loss": 0.4881, "step": 7328 }, { "epoch": 47.24497991967871, "grad_norm": 9.515909194946289, "learning_rate": 2.6358064516129032e-05, "loss": 0.3427, "step": 7329 }, { "epoch": 47.25140562248996, "grad_norm": 21.83661460876465, "learning_rate": 2.635483870967742e-05, "loss": 0.4362, "step": 7330 }, { "epoch": 47.257831325301204, "grad_norm": 6.880281448364258, "learning_rate": 2.6351612903225807e-05, "loss": 0.3282, "step": 7331 }, { "epoch": 47.264257028112446, "grad_norm": 17.847017288208008, "learning_rate": 2.6348387096774196e-05, "loss": 0.4137, "step": 7332 }, { "epoch": 47.270682730923696, "grad_norm": 13.730484962463379, "learning_rate": 2.6345161290322585e-05, "loss": 0.5183, "step": 7333 }, { "epoch": 47.27710843373494, "grad_norm": 16.830459594726562, "learning_rate": 2.6341935483870968e-05, "loss": 0.3141, "step": 7334 }, { "epoch": 47.28353413654619, "grad_norm": 13.774622917175293, "learning_rate": 2.6338709677419353e-05, "loss": 0.4008, "step": 7335 }, { "epoch": 47.28995983935743, "grad_norm": 8.078412055969238, "learning_rate": 2.6335483870967743e-05, "loss": 0.2425, "step": 7336 }, { "epoch": 47.29638554216867, "grad_norm": 7.775115013122559, "learning_rate": 2.6332258064516128e-05, "loss": 0.4107, "step": 7337 }, { "epoch": 47.30281124497992, "grad_norm": 8.916316032409668, "learning_rate": 2.6329032258064517e-05, "loss": 0.3471, "step": 7338 }, { "epoch": 47.30923694779116, "grad_norm": 8.002837181091309, "learning_rate": 2.6325806451612906e-05, "loss": 0.3616, "step": 7339 }, { "epoch": 47.31566265060241, "grad_norm": 12.208673477172852, "learning_rate": 2.6322580645161292e-05, "loss": 0.486, "step": 7340 }, { "epoch": 47.322088353413655, "grad_norm": 11.955066680908203, "learning_rate": 2.631935483870968e-05, "loss": 0.4159, "step": 7341 }, { "epoch": 47.3285140562249, "grad_norm": 3.943674087524414, "learning_rate": 2.6316129032258064e-05, "loss": 0.1368, "step": 7342 }, { "epoch": 47.334939759036146, "grad_norm": 8.586464881896973, "learning_rate": 2.6312903225806453e-05, "loss": 0.3657, "step": 7343 }, { "epoch": 47.34136546184739, "grad_norm": 12.047308921813965, "learning_rate": 2.630967741935484e-05, "loss": 0.3333, "step": 7344 }, { "epoch": 47.34779116465864, "grad_norm": 17.013080596923828, "learning_rate": 2.6306451612903228e-05, "loss": 0.9154, "step": 7345 }, { "epoch": 47.35421686746988, "grad_norm": 11.520630836486816, "learning_rate": 2.6303225806451613e-05, "loss": 0.277, "step": 7346 }, { "epoch": 47.36064257028112, "grad_norm": 9.722981452941895, "learning_rate": 2.6300000000000002e-05, "loss": 0.2216, "step": 7347 }, { "epoch": 47.36706827309237, "grad_norm": 6.760676383972168, "learning_rate": 2.629677419354839e-05, "loss": 0.293, "step": 7348 }, { "epoch": 47.373493975903614, "grad_norm": 2.6471195220947266, "learning_rate": 2.6293548387096777e-05, "loss": 0.0782, "step": 7349 }, { "epoch": 47.379919678714856, "grad_norm": 7.006134986877441, "learning_rate": 2.629032258064516e-05, "loss": 0.527, "step": 7350 }, { "epoch": 47.386345381526105, "grad_norm": 6.4443559646606445, "learning_rate": 2.628709677419355e-05, "loss": 0.4061, "step": 7351 }, { "epoch": 47.39277108433735, "grad_norm": 19.383590698242188, "learning_rate": 2.6283870967741934e-05, "loss": 0.859, "step": 7352 }, { "epoch": 47.3991967871486, "grad_norm": 11.178043365478516, "learning_rate": 2.6280645161290323e-05, "loss": 0.2731, "step": 7353 }, { "epoch": 47.40562248995984, "grad_norm": 8.473057746887207, "learning_rate": 2.6277419354838712e-05, "loss": 0.4296, "step": 7354 }, { "epoch": 47.41204819277108, "grad_norm": 4.489865303039551, "learning_rate": 2.6274193548387098e-05, "loss": 0.2131, "step": 7355 }, { "epoch": 47.41847389558233, "grad_norm": 2.1776366233825684, "learning_rate": 2.6270967741935487e-05, "loss": 0.0938, "step": 7356 }, { "epoch": 47.42489959839357, "grad_norm": 8.398616790771484, "learning_rate": 2.6267741935483876e-05, "loss": 0.3568, "step": 7357 }, { "epoch": 47.43132530120482, "grad_norm": 4.290478706359863, "learning_rate": 2.626451612903226e-05, "loss": 0.2996, "step": 7358 }, { "epoch": 47.437751004016064, "grad_norm": 4.138827323913574, "learning_rate": 2.6261290322580644e-05, "loss": 0.2929, "step": 7359 }, { "epoch": 47.44417670682731, "grad_norm": 15.624200820922852, "learning_rate": 2.6258064516129033e-05, "loss": 0.4768, "step": 7360 }, { "epoch": 47.450602409638556, "grad_norm": 10.877457618713379, "learning_rate": 2.625483870967742e-05, "loss": 0.2763, "step": 7361 }, { "epoch": 47.4570281124498, "grad_norm": 5.883092403411865, "learning_rate": 2.6251612903225808e-05, "loss": 0.4536, "step": 7362 }, { "epoch": 47.46345381526105, "grad_norm": 11.218177795410156, "learning_rate": 2.6248387096774197e-05, "loss": 0.5398, "step": 7363 }, { "epoch": 47.46987951807229, "grad_norm": 9.613465309143066, "learning_rate": 2.6245161290322583e-05, "loss": 0.3692, "step": 7364 }, { "epoch": 47.47630522088353, "grad_norm": 8.035354614257812, "learning_rate": 2.6241935483870965e-05, "loss": 0.2324, "step": 7365 }, { "epoch": 47.48273092369478, "grad_norm": 8.753141403198242, "learning_rate": 2.6238709677419354e-05, "loss": 0.4222, "step": 7366 }, { "epoch": 47.48915662650602, "grad_norm": 33.723384857177734, "learning_rate": 2.6235483870967744e-05, "loss": 0.2571, "step": 7367 }, { "epoch": 47.495582329317266, "grad_norm": 7.033966541290283, "learning_rate": 2.623225806451613e-05, "loss": 0.3555, "step": 7368 }, { "epoch": 47.502008032128515, "grad_norm": 7.8734354972839355, "learning_rate": 2.622903225806452e-05, "loss": 0.108, "step": 7369 }, { "epoch": 47.50843373493976, "grad_norm": 9.28868579864502, "learning_rate": 2.6225806451612904e-05, "loss": 0.3316, "step": 7370 }, { "epoch": 47.51485943775101, "grad_norm": 7.382019519805908, "learning_rate": 2.6222580645161293e-05, "loss": 0.2475, "step": 7371 }, { "epoch": 47.52128514056225, "grad_norm": 14.945975303649902, "learning_rate": 2.6219354838709682e-05, "loss": 0.4274, "step": 7372 }, { "epoch": 47.52771084337349, "grad_norm": 19.218955993652344, "learning_rate": 2.6216129032258065e-05, "loss": 0.2726, "step": 7373 }, { "epoch": 47.53413654618474, "grad_norm": 7.774152755737305, "learning_rate": 2.621290322580645e-05, "loss": 0.2714, "step": 7374 }, { "epoch": 47.54056224899598, "grad_norm": 10.862861633300781, "learning_rate": 2.620967741935484e-05, "loss": 0.2654, "step": 7375 }, { "epoch": 47.54698795180723, "grad_norm": 19.114744186401367, "learning_rate": 2.6206451612903225e-05, "loss": 0.6569, "step": 7376 }, { "epoch": 47.553413654618474, "grad_norm": 7.064778804779053, "learning_rate": 2.6203225806451614e-05, "loss": 0.2607, "step": 7377 }, { "epoch": 47.559839357429716, "grad_norm": 4.810049057006836, "learning_rate": 2.6200000000000003e-05, "loss": 0.2657, "step": 7378 }, { "epoch": 47.566265060240966, "grad_norm": 6.739880561828613, "learning_rate": 2.619677419354839e-05, "loss": 0.4459, "step": 7379 }, { "epoch": 47.57269076305221, "grad_norm": 15.264415740966797, "learning_rate": 2.6193548387096778e-05, "loss": 0.521, "step": 7380 }, { "epoch": 47.57911646586346, "grad_norm": 7.088973045349121, "learning_rate": 2.619032258064516e-05, "loss": 0.3875, "step": 7381 }, { "epoch": 47.5855421686747, "grad_norm": 17.041973114013672, "learning_rate": 2.618709677419355e-05, "loss": 0.237, "step": 7382 }, { "epoch": 47.59196787148594, "grad_norm": 11.564109802246094, "learning_rate": 2.6183870967741935e-05, "loss": 0.2965, "step": 7383 }, { "epoch": 47.59839357429719, "grad_norm": 35.53783416748047, "learning_rate": 2.6180645161290324e-05, "loss": 0.4106, "step": 7384 }, { "epoch": 47.60481927710843, "grad_norm": 7.371618270874023, "learning_rate": 2.617741935483871e-05, "loss": 0.3066, "step": 7385 }, { "epoch": 47.611244979919675, "grad_norm": 3.1006076335906982, "learning_rate": 2.61741935483871e-05, "loss": 0.105, "step": 7386 }, { "epoch": 47.617670682730925, "grad_norm": 7.96307897567749, "learning_rate": 2.6170967741935488e-05, "loss": 0.354, "step": 7387 }, { "epoch": 47.62409638554217, "grad_norm": 7.29679012298584, "learning_rate": 2.6167741935483874e-05, "loss": 0.4511, "step": 7388 }, { "epoch": 47.630522088353416, "grad_norm": 8.883305549621582, "learning_rate": 2.6164516129032256e-05, "loss": 0.3614, "step": 7389 }, { "epoch": 47.63694779116466, "grad_norm": 14.209417343139648, "learning_rate": 2.6161290322580645e-05, "loss": 0.3319, "step": 7390 }, { "epoch": 47.6433734939759, "grad_norm": 7.48521614074707, "learning_rate": 2.6158064516129034e-05, "loss": 0.4111, "step": 7391 }, { "epoch": 47.64979919678715, "grad_norm": 9.679743766784668, "learning_rate": 2.615483870967742e-05, "loss": 0.2857, "step": 7392 }, { "epoch": 47.65622489959839, "grad_norm": 13.434447288513184, "learning_rate": 2.615161290322581e-05, "loss": 0.494, "step": 7393 }, { "epoch": 47.66265060240964, "grad_norm": 12.346441268920898, "learning_rate": 2.6148387096774195e-05, "loss": 0.4745, "step": 7394 }, { "epoch": 47.669076305220884, "grad_norm": 8.285652160644531, "learning_rate": 2.6145161290322584e-05, "loss": 0.4013, "step": 7395 }, { "epoch": 47.675502008032126, "grad_norm": 7.769961357116699, "learning_rate": 2.6141935483870966e-05, "loss": 0.2817, "step": 7396 }, { "epoch": 47.681927710843375, "grad_norm": 10.071239471435547, "learning_rate": 2.6138709677419355e-05, "loss": 0.2449, "step": 7397 }, { "epoch": 47.68835341365462, "grad_norm": 12.655542373657227, "learning_rate": 2.613548387096774e-05, "loss": 0.4522, "step": 7398 }, { "epoch": 47.69477911646587, "grad_norm": 24.04180145263672, "learning_rate": 2.613225806451613e-05, "loss": 0.4314, "step": 7399 }, { "epoch": 47.70120481927711, "grad_norm": 3.734616279602051, "learning_rate": 2.6129032258064516e-05, "loss": 0.1539, "step": 7400 }, { "epoch": 47.70763052208835, "grad_norm": 6.310758590698242, "learning_rate": 2.6125806451612905e-05, "loss": 0.2033, "step": 7401 }, { "epoch": 47.7140562248996, "grad_norm": 9.145163536071777, "learning_rate": 2.6122580645161294e-05, "loss": 0.4516, "step": 7402 }, { "epoch": 47.72048192771084, "grad_norm": 23.51615333557129, "learning_rate": 2.611935483870968e-05, "loss": 0.7049, "step": 7403 }, { "epoch": 47.726907630522085, "grad_norm": 7.188912868499756, "learning_rate": 2.6116129032258062e-05, "loss": 0.4128, "step": 7404 }, { "epoch": 47.733333333333334, "grad_norm": 9.33350944519043, "learning_rate": 2.611290322580645e-05, "loss": 0.2436, "step": 7405 }, { "epoch": 47.73975903614458, "grad_norm": 4.01378870010376, "learning_rate": 2.610967741935484e-05, "loss": 0.13, "step": 7406 }, { "epoch": 47.746184738955826, "grad_norm": 6.315446376800537, "learning_rate": 2.6106451612903226e-05, "loss": 0.4524, "step": 7407 }, { "epoch": 47.75261044176707, "grad_norm": 10.369728088378906, "learning_rate": 2.6103225806451615e-05, "loss": 0.1441, "step": 7408 }, { "epoch": 47.75903614457831, "grad_norm": 11.843390464782715, "learning_rate": 2.61e-05, "loss": 0.4291, "step": 7409 }, { "epoch": 47.76546184738956, "grad_norm": 8.756425857543945, "learning_rate": 2.609677419354839e-05, "loss": 0.2552, "step": 7410 }, { "epoch": 47.7718875502008, "grad_norm": 12.534231185913086, "learning_rate": 2.609354838709678e-05, "loss": 0.4176, "step": 7411 }, { "epoch": 47.77831325301205, "grad_norm": 8.994430541992188, "learning_rate": 2.609032258064516e-05, "loss": 0.4346, "step": 7412 }, { "epoch": 47.78473895582329, "grad_norm": 6.750273704528809, "learning_rate": 2.6087096774193547e-05, "loss": 0.2434, "step": 7413 }, { "epoch": 47.791164658634536, "grad_norm": 4.953756332397461, "learning_rate": 2.6083870967741936e-05, "loss": 0.2435, "step": 7414 }, { "epoch": 47.797590361445785, "grad_norm": 7.137415409088135, "learning_rate": 2.6080645161290322e-05, "loss": 0.4264, "step": 7415 }, { "epoch": 47.80401606425703, "grad_norm": 7.228492259979248, "learning_rate": 2.607741935483871e-05, "loss": 0.3252, "step": 7416 }, { "epoch": 47.810441767068276, "grad_norm": 7.928820610046387, "learning_rate": 2.60741935483871e-05, "loss": 0.3687, "step": 7417 }, { "epoch": 47.81686746987952, "grad_norm": 9.987020492553711, "learning_rate": 2.6070967741935486e-05, "loss": 0.3506, "step": 7418 }, { "epoch": 47.82329317269076, "grad_norm": 13.459898948669434, "learning_rate": 2.6067741935483875e-05, "loss": 0.2351, "step": 7419 }, { "epoch": 47.82971887550201, "grad_norm": 5.3769097328186035, "learning_rate": 2.6064516129032257e-05, "loss": 0.1839, "step": 7420 }, { "epoch": 47.83614457831325, "grad_norm": 7.626886367797852, "learning_rate": 2.6061290322580646e-05, "loss": 0.2553, "step": 7421 }, { "epoch": 47.842570281124495, "grad_norm": 7.728386878967285, "learning_rate": 2.6058064516129032e-05, "loss": 0.4954, "step": 7422 }, { "epoch": 47.848995983935744, "grad_norm": 23.231168746948242, "learning_rate": 2.605483870967742e-05, "loss": 0.3006, "step": 7423 }, { "epoch": 47.855421686746986, "grad_norm": 9.40967082977295, "learning_rate": 2.6051612903225807e-05, "loss": 0.2446, "step": 7424 }, { "epoch": 47.861847389558235, "grad_norm": 9.694540977478027, "learning_rate": 2.6048387096774196e-05, "loss": 0.5539, "step": 7425 }, { "epoch": 47.86827309236948, "grad_norm": 16.387805938720703, "learning_rate": 2.6045161290322585e-05, "loss": 0.3348, "step": 7426 }, { "epoch": 47.87469879518072, "grad_norm": 12.39390754699707, "learning_rate": 2.604193548387097e-05, "loss": 0.2534, "step": 7427 }, { "epoch": 47.88112449799197, "grad_norm": 14.936233520507812, "learning_rate": 2.6038709677419353e-05, "loss": 0.3441, "step": 7428 }, { "epoch": 47.88755020080321, "grad_norm": 10.211873054504395, "learning_rate": 2.6035483870967742e-05, "loss": 0.4763, "step": 7429 }, { "epoch": 47.89397590361446, "grad_norm": 18.60927963256836, "learning_rate": 2.603225806451613e-05, "loss": 0.5546, "step": 7430 }, { "epoch": 47.9004016064257, "grad_norm": 9.190190315246582, "learning_rate": 2.6029032258064517e-05, "loss": 0.3585, "step": 7431 }, { "epoch": 47.906827309236945, "grad_norm": 15.060502052307129, "learning_rate": 2.6025806451612906e-05, "loss": 0.1743, "step": 7432 }, { "epoch": 47.913253012048195, "grad_norm": 8.675372123718262, "learning_rate": 2.602258064516129e-05, "loss": 0.5859, "step": 7433 }, { "epoch": 47.91967871485944, "grad_norm": 8.857006072998047, "learning_rate": 2.601935483870968e-05, "loss": 0.4177, "step": 7434 }, { "epoch": 47.926104417670686, "grad_norm": 6.70772123336792, "learning_rate": 2.6016129032258063e-05, "loss": 0.3366, "step": 7435 }, { "epoch": 47.93253012048193, "grad_norm": 5.904670715332031, "learning_rate": 2.6012903225806452e-05, "loss": 0.2122, "step": 7436 }, { "epoch": 47.93895582329317, "grad_norm": 5.836674213409424, "learning_rate": 2.6009677419354838e-05, "loss": 0.3155, "step": 7437 }, { "epoch": 47.94538152610442, "grad_norm": 10.916560173034668, "learning_rate": 2.6006451612903227e-05, "loss": 0.5105, "step": 7438 }, { "epoch": 47.95180722891566, "grad_norm": 9.968976974487305, "learning_rate": 2.6003225806451613e-05, "loss": 0.2759, "step": 7439 }, { "epoch": 47.958232931726904, "grad_norm": 14.638289451599121, "learning_rate": 2.6000000000000002e-05, "loss": 0.4972, "step": 7440 }, { "epoch": 47.964658634538154, "grad_norm": 8.458620071411133, "learning_rate": 2.599677419354839e-05, "loss": 0.2235, "step": 7441 }, { "epoch": 47.971084337349396, "grad_norm": 14.342974662780762, "learning_rate": 2.5993548387096777e-05, "loss": 0.3176, "step": 7442 }, { "epoch": 47.977510040160645, "grad_norm": 14.906553268432617, "learning_rate": 2.599032258064516e-05, "loss": 0.5548, "step": 7443 }, { "epoch": 47.98393574297189, "grad_norm": 5.038039207458496, "learning_rate": 2.5987096774193548e-05, "loss": 0.395, "step": 7444 }, { "epoch": 47.99036144578313, "grad_norm": 16.817899703979492, "learning_rate": 2.5983870967741937e-05, "loss": 0.3293, "step": 7445 }, { "epoch": 47.99678714859438, "grad_norm": 10.273216247558594, "learning_rate": 2.5980645161290323e-05, "loss": 0.4012, "step": 7446 }, { "epoch": 48.00642570281124, "grad_norm": 12.171708106994629, "learning_rate": 2.5977419354838712e-05, "loss": 0.359, "step": 7447 }, { "epoch": 48.01285140562249, "grad_norm": 15.000752449035645, "learning_rate": 2.5974193548387098e-05, "loss": 0.5549, "step": 7448 }, { "epoch": 48.019277108433734, "grad_norm": 8.745424270629883, "learning_rate": 2.5970967741935487e-05, "loss": 0.245, "step": 7449 }, { "epoch": 48.02570281124498, "grad_norm": 6.592627048492432, "learning_rate": 2.5967741935483876e-05, "loss": 0.3178, "step": 7450 }, { "epoch": 48.032128514056225, "grad_norm": 8.83784294128418, "learning_rate": 2.5964516129032258e-05, "loss": 0.4148, "step": 7451 }, { "epoch": 48.03855421686747, "grad_norm": 11.262797355651855, "learning_rate": 2.5961290322580644e-05, "loss": 0.4927, "step": 7452 }, { "epoch": 48.04497991967872, "grad_norm": 14.8204984664917, "learning_rate": 2.5958064516129033e-05, "loss": 0.3159, "step": 7453 }, { "epoch": 48.05140562248996, "grad_norm": 16.764467239379883, "learning_rate": 2.5954838709677422e-05, "loss": 0.8506, "step": 7454 }, { "epoch": 48.0578313253012, "grad_norm": 14.480372428894043, "learning_rate": 2.5951612903225808e-05, "loss": 0.5622, "step": 7455 }, { "epoch": 48.06425702811245, "grad_norm": 9.822854995727539, "learning_rate": 2.5948387096774197e-05, "loss": 0.4708, "step": 7456 }, { "epoch": 48.07068273092369, "grad_norm": 14.187322616577148, "learning_rate": 2.5945161290322583e-05, "loss": 0.4599, "step": 7457 }, { "epoch": 48.07710843373494, "grad_norm": 5.118282794952393, "learning_rate": 2.594193548387097e-05, "loss": 0.1634, "step": 7458 }, { "epoch": 48.083534136546184, "grad_norm": 9.521262168884277, "learning_rate": 2.5938709677419354e-05, "loss": 0.3296, "step": 7459 }, { "epoch": 48.08995983935743, "grad_norm": 10.682927131652832, "learning_rate": 2.5935483870967743e-05, "loss": 0.1965, "step": 7460 }, { "epoch": 48.096385542168676, "grad_norm": 11.155173301696777, "learning_rate": 2.593225806451613e-05, "loss": 0.3866, "step": 7461 }, { "epoch": 48.10281124497992, "grad_norm": 22.453344345092773, "learning_rate": 2.5929032258064518e-05, "loss": 0.505, "step": 7462 }, { "epoch": 48.10923694779117, "grad_norm": 14.916485786437988, "learning_rate": 2.5925806451612904e-05, "loss": 0.4131, "step": 7463 }, { "epoch": 48.11566265060241, "grad_norm": 6.890715599060059, "learning_rate": 2.5922580645161293e-05, "loss": 0.2385, "step": 7464 }, { "epoch": 48.12208835341365, "grad_norm": 22.19963836669922, "learning_rate": 2.5919354838709682e-05, "loss": 0.4006, "step": 7465 }, { "epoch": 48.1285140562249, "grad_norm": 7.502270698547363, "learning_rate": 2.5916129032258064e-05, "loss": 0.2873, "step": 7466 }, { "epoch": 48.13493975903614, "grad_norm": 9.167383193969727, "learning_rate": 2.591290322580645e-05, "loss": 0.4824, "step": 7467 }, { "epoch": 48.14136546184739, "grad_norm": 13.00903034210205, "learning_rate": 2.590967741935484e-05, "loss": 0.6574, "step": 7468 }, { "epoch": 48.147791164658635, "grad_norm": 24.03598403930664, "learning_rate": 2.5906451612903228e-05, "loss": 0.5537, "step": 7469 }, { "epoch": 48.15421686746988, "grad_norm": 7.7911601066589355, "learning_rate": 2.5903225806451614e-05, "loss": 0.1503, "step": 7470 }, { "epoch": 48.16064257028113, "grad_norm": 8.100223541259766, "learning_rate": 2.5900000000000003e-05, "loss": 0.2188, "step": 7471 }, { "epoch": 48.16706827309237, "grad_norm": 7.419651031494141, "learning_rate": 2.589677419354839e-05, "loss": 0.3233, "step": 7472 }, { "epoch": 48.17349397590361, "grad_norm": 23.9660701751709, "learning_rate": 2.5893548387096778e-05, "loss": 0.3903, "step": 7473 }, { "epoch": 48.17991967871486, "grad_norm": 52.683738708496094, "learning_rate": 2.589032258064516e-05, "loss": 0.248, "step": 7474 }, { "epoch": 48.1863453815261, "grad_norm": 9.070000648498535, "learning_rate": 2.588709677419355e-05, "loss": 0.2584, "step": 7475 }, { "epoch": 48.19277108433735, "grad_norm": 10.758220672607422, "learning_rate": 2.5883870967741935e-05, "loss": 0.5202, "step": 7476 }, { "epoch": 48.199196787148594, "grad_norm": 14.597804069519043, "learning_rate": 2.5880645161290324e-05, "loss": 0.414, "step": 7477 }, { "epoch": 48.205622489959836, "grad_norm": 19.783496856689453, "learning_rate": 2.587741935483871e-05, "loss": 0.4264, "step": 7478 }, { "epoch": 48.212048192771086, "grad_norm": 5.999805927276611, "learning_rate": 2.58741935483871e-05, "loss": 0.308, "step": 7479 }, { "epoch": 48.21847389558233, "grad_norm": 26.079755783081055, "learning_rate": 2.5870967741935488e-05, "loss": 0.5538, "step": 7480 }, { "epoch": 48.22489959839358, "grad_norm": 17.771982192993164, "learning_rate": 2.5867741935483873e-05, "loss": 0.5084, "step": 7481 }, { "epoch": 48.23132530120482, "grad_norm": 14.638922691345215, "learning_rate": 2.5864516129032256e-05, "loss": 0.3465, "step": 7482 }, { "epoch": 48.23775100401606, "grad_norm": 12.316288948059082, "learning_rate": 2.5861290322580645e-05, "loss": 0.3134, "step": 7483 }, { "epoch": 48.24417670682731, "grad_norm": 6.255400657653809, "learning_rate": 2.5858064516129034e-05, "loss": 0.2037, "step": 7484 }, { "epoch": 48.25060240963855, "grad_norm": 5.530472755432129, "learning_rate": 2.585483870967742e-05, "loss": 0.1906, "step": 7485 }, { "epoch": 48.2570281124498, "grad_norm": 9.3776273727417, "learning_rate": 2.585161290322581e-05, "loss": 0.3607, "step": 7486 }, { "epoch": 48.263453815261045, "grad_norm": 8.340258598327637, "learning_rate": 2.5848387096774194e-05, "loss": 0.4929, "step": 7487 }, { "epoch": 48.26987951807229, "grad_norm": 8.598422050476074, "learning_rate": 2.5845161290322583e-05, "loss": 0.2149, "step": 7488 }, { "epoch": 48.276305220883536, "grad_norm": 12.88918399810791, "learning_rate": 2.5841935483870973e-05, "loss": 0.4312, "step": 7489 }, { "epoch": 48.28273092369478, "grad_norm": 12.507916450500488, "learning_rate": 2.5838709677419355e-05, "loss": 0.4113, "step": 7490 }, { "epoch": 48.28915662650602, "grad_norm": 11.388280868530273, "learning_rate": 2.583548387096774e-05, "loss": 0.5119, "step": 7491 }, { "epoch": 48.29558232931727, "grad_norm": 7.734696388244629, "learning_rate": 2.583225806451613e-05, "loss": 0.1979, "step": 7492 }, { "epoch": 48.30200803212851, "grad_norm": 14.1597261428833, "learning_rate": 2.582903225806452e-05, "loss": 0.5638, "step": 7493 }, { "epoch": 48.30843373493976, "grad_norm": 8.127717971801758, "learning_rate": 2.5825806451612904e-05, "loss": 0.3133, "step": 7494 }, { "epoch": 48.314859437751004, "grad_norm": 8.835047721862793, "learning_rate": 2.5822580645161294e-05, "loss": 0.189, "step": 7495 }, { "epoch": 48.321285140562246, "grad_norm": 10.303837776184082, "learning_rate": 2.581935483870968e-05, "loss": 0.3677, "step": 7496 }, { "epoch": 48.327710843373495, "grad_norm": 9.111800193786621, "learning_rate": 2.581612903225807e-05, "loss": 0.2038, "step": 7497 }, { "epoch": 48.33413654618474, "grad_norm": 8.253888130187988, "learning_rate": 2.581290322580645e-05, "loss": 0.2758, "step": 7498 }, { "epoch": 48.34056224899599, "grad_norm": 9.18570613861084, "learning_rate": 2.580967741935484e-05, "loss": 0.2697, "step": 7499 }, { "epoch": 48.34698795180723, "grad_norm": 11.45446491241455, "learning_rate": 2.5806451612903226e-05, "loss": 0.1991, "step": 7500 }, { "epoch": 48.35341365461847, "grad_norm": 15.723467826843262, "learning_rate": 2.5803225806451615e-05, "loss": 0.2285, "step": 7501 }, { "epoch": 48.35983935742972, "grad_norm": 15.756359100341797, "learning_rate": 2.58e-05, "loss": 0.5417, "step": 7502 }, { "epoch": 48.36626506024096, "grad_norm": 65.93222045898438, "learning_rate": 2.579677419354839e-05, "loss": 0.5307, "step": 7503 }, { "epoch": 48.37269076305221, "grad_norm": 7.640913486480713, "learning_rate": 2.579354838709678e-05, "loss": 0.3455, "step": 7504 }, { "epoch": 48.379116465863454, "grad_norm": 5.256098747253418, "learning_rate": 2.579032258064516e-05, "loss": 0.1024, "step": 7505 }, { "epoch": 48.3855421686747, "grad_norm": 10.693883895874023, "learning_rate": 2.5787096774193547e-05, "loss": 0.7921, "step": 7506 }, { "epoch": 48.391967871485946, "grad_norm": 13.3115816116333, "learning_rate": 2.5783870967741936e-05, "loss": 0.4074, "step": 7507 }, { "epoch": 48.39839357429719, "grad_norm": 10.864363670349121, "learning_rate": 2.5780645161290325e-05, "loss": 0.4087, "step": 7508 }, { "epoch": 48.40481927710843, "grad_norm": 7.853160381317139, "learning_rate": 2.577741935483871e-05, "loss": 0.4714, "step": 7509 }, { "epoch": 48.41124497991968, "grad_norm": 6.1512227058410645, "learning_rate": 2.57741935483871e-05, "loss": 0.3669, "step": 7510 }, { "epoch": 48.41767068273092, "grad_norm": 14.644233703613281, "learning_rate": 2.5770967741935485e-05, "loss": 0.5149, "step": 7511 }, { "epoch": 48.42409638554217, "grad_norm": 11.841924667358398, "learning_rate": 2.5767741935483874e-05, "loss": 0.2435, "step": 7512 }, { "epoch": 48.43052208835341, "grad_norm": 8.716687202453613, "learning_rate": 2.5764516129032257e-05, "loss": 0.2523, "step": 7513 }, { "epoch": 48.436947791164656, "grad_norm": 8.493523597717285, "learning_rate": 2.5761290322580646e-05, "loss": 0.2525, "step": 7514 }, { "epoch": 48.443373493975905, "grad_norm": 25.162965774536133, "learning_rate": 2.575806451612903e-05, "loss": 0.4809, "step": 7515 }, { "epoch": 48.44979919678715, "grad_norm": 14.450448036193848, "learning_rate": 2.575483870967742e-05, "loss": 0.4001, "step": 7516 }, { "epoch": 48.456224899598396, "grad_norm": 11.017749786376953, "learning_rate": 2.575161290322581e-05, "loss": 0.4511, "step": 7517 }, { "epoch": 48.46265060240964, "grad_norm": 7.726383686065674, "learning_rate": 2.5748387096774195e-05, "loss": 0.5101, "step": 7518 }, { "epoch": 48.46907630522088, "grad_norm": 15.656213760375977, "learning_rate": 2.5745161290322584e-05, "loss": 0.2695, "step": 7519 }, { "epoch": 48.47550200803213, "grad_norm": 17.534475326538086, "learning_rate": 2.574193548387097e-05, "loss": 0.2916, "step": 7520 }, { "epoch": 48.48192771084337, "grad_norm": 8.310469627380371, "learning_rate": 2.5738709677419352e-05, "loss": 0.2943, "step": 7521 }, { "epoch": 48.48835341365462, "grad_norm": 8.537556648254395, "learning_rate": 2.573548387096774e-05, "loss": 0.2848, "step": 7522 }, { "epoch": 48.494779116465864, "grad_norm": 23.97414207458496, "learning_rate": 2.573225806451613e-05, "loss": 0.435, "step": 7523 }, { "epoch": 48.501204819277106, "grad_norm": 9.10914134979248, "learning_rate": 2.5729032258064516e-05, "loss": 0.3868, "step": 7524 }, { "epoch": 48.507630522088355, "grad_norm": 10.071598052978516, "learning_rate": 2.5725806451612905e-05, "loss": 0.3119, "step": 7525 }, { "epoch": 48.5140562248996, "grad_norm": 13.566871643066406, "learning_rate": 2.572258064516129e-05, "loss": 0.2887, "step": 7526 }, { "epoch": 48.52048192771084, "grad_norm": 10.234539031982422, "learning_rate": 2.571935483870968e-05, "loss": 0.3784, "step": 7527 }, { "epoch": 48.52690763052209, "grad_norm": 5.931802272796631, "learning_rate": 2.571612903225807e-05, "loss": 0.2315, "step": 7528 }, { "epoch": 48.53333333333333, "grad_norm": 18.701194763183594, "learning_rate": 2.571290322580645e-05, "loss": 0.5559, "step": 7529 }, { "epoch": 48.53975903614458, "grad_norm": 4.452053070068359, "learning_rate": 2.5709677419354837e-05, "loss": 0.0937, "step": 7530 }, { "epoch": 48.54618473895582, "grad_norm": 14.434877395629883, "learning_rate": 2.5706451612903226e-05, "loss": 0.5117, "step": 7531 }, { "epoch": 48.552610441767065, "grad_norm": 12.552257537841797, "learning_rate": 2.5703225806451616e-05, "loss": 0.4522, "step": 7532 }, { "epoch": 48.559036144578315, "grad_norm": 5.643826484680176, "learning_rate": 2.57e-05, "loss": 0.1234, "step": 7533 }, { "epoch": 48.56546184738956, "grad_norm": 23.589406967163086, "learning_rate": 2.569677419354839e-05, "loss": 0.4092, "step": 7534 }, { "epoch": 48.571887550200806, "grad_norm": 15.036336898803711, "learning_rate": 2.5693548387096776e-05, "loss": 0.3412, "step": 7535 }, { "epoch": 48.57831325301205, "grad_norm": 7.675848007202148, "learning_rate": 2.5690322580645165e-05, "loss": 0.2922, "step": 7536 }, { "epoch": 48.58473895582329, "grad_norm": 15.543149948120117, "learning_rate": 2.5687096774193547e-05, "loss": 0.3064, "step": 7537 }, { "epoch": 48.59116465863454, "grad_norm": 5.805973529815674, "learning_rate": 2.5683870967741937e-05, "loss": 0.2827, "step": 7538 }, { "epoch": 48.59759036144578, "grad_norm": 10.230324745178223, "learning_rate": 2.5680645161290322e-05, "loss": 0.5144, "step": 7539 }, { "epoch": 48.60401606425703, "grad_norm": 10.143692970275879, "learning_rate": 2.567741935483871e-05, "loss": 0.3909, "step": 7540 }, { "epoch": 48.610441767068274, "grad_norm": 6.096628189086914, "learning_rate": 2.5674193548387097e-05, "loss": 0.2899, "step": 7541 }, { "epoch": 48.616867469879516, "grad_norm": 10.395247459411621, "learning_rate": 2.5670967741935486e-05, "loss": 0.3871, "step": 7542 }, { "epoch": 48.623293172690765, "grad_norm": 4.172732353210449, "learning_rate": 2.5667741935483875e-05, "loss": 0.1913, "step": 7543 }, { "epoch": 48.62971887550201, "grad_norm": 9.419051170349121, "learning_rate": 2.5664516129032258e-05, "loss": 0.6411, "step": 7544 }, { "epoch": 48.63614457831325, "grad_norm": 5.029129505157471, "learning_rate": 2.5661290322580643e-05, "loss": 0.2088, "step": 7545 }, { "epoch": 48.6425702811245, "grad_norm": 8.29389476776123, "learning_rate": 2.5658064516129032e-05, "loss": 0.2754, "step": 7546 }, { "epoch": 48.64899598393574, "grad_norm": 10.348937034606934, "learning_rate": 2.565483870967742e-05, "loss": 0.3488, "step": 7547 }, { "epoch": 48.65542168674699, "grad_norm": 10.009827613830566, "learning_rate": 2.5651612903225807e-05, "loss": 0.2846, "step": 7548 }, { "epoch": 48.66184738955823, "grad_norm": 5.043690204620361, "learning_rate": 2.5648387096774196e-05, "loss": 0.2536, "step": 7549 }, { "epoch": 48.668273092369475, "grad_norm": 14.59740924835205, "learning_rate": 2.5645161290322582e-05, "loss": 0.3329, "step": 7550 }, { "epoch": 48.674698795180724, "grad_norm": 5.227171897888184, "learning_rate": 2.564193548387097e-05, "loss": 0.2362, "step": 7551 }, { "epoch": 48.681124497991966, "grad_norm": 8.430366516113281, "learning_rate": 2.5638709677419353e-05, "loss": 0.2151, "step": 7552 }, { "epoch": 48.687550200803216, "grad_norm": 8.662446022033691, "learning_rate": 2.5635483870967743e-05, "loss": 0.2942, "step": 7553 }, { "epoch": 48.69397590361446, "grad_norm": 10.417023658752441, "learning_rate": 2.5632258064516128e-05, "loss": 0.6037, "step": 7554 }, { "epoch": 48.7004016064257, "grad_norm": 10.235279083251953, "learning_rate": 2.5629032258064517e-05, "loss": 0.3301, "step": 7555 }, { "epoch": 48.70682730923695, "grad_norm": 13.685818672180176, "learning_rate": 2.5625806451612906e-05, "loss": 0.2535, "step": 7556 }, { "epoch": 48.71325301204819, "grad_norm": 22.248779296875, "learning_rate": 2.5622580645161292e-05, "loss": 0.6793, "step": 7557 }, { "epoch": 48.71967871485944, "grad_norm": 10.01766586303711, "learning_rate": 2.561935483870968e-05, "loss": 0.2688, "step": 7558 }, { "epoch": 48.72610441767068, "grad_norm": 14.821676254272461, "learning_rate": 2.5616129032258067e-05, "loss": 0.3094, "step": 7559 }, { "epoch": 48.732530120481925, "grad_norm": 13.70749568939209, "learning_rate": 2.561290322580645e-05, "loss": 0.6555, "step": 7560 }, { "epoch": 48.738955823293175, "grad_norm": 9.983782768249512, "learning_rate": 2.560967741935484e-05, "loss": 0.2916, "step": 7561 }, { "epoch": 48.74538152610442, "grad_norm": 10.303790092468262, "learning_rate": 2.5606451612903227e-05, "loss": 0.3188, "step": 7562 }, { "epoch": 48.75180722891566, "grad_norm": 6.68446683883667, "learning_rate": 2.5603225806451613e-05, "loss": 0.4501, "step": 7563 }, { "epoch": 48.75823293172691, "grad_norm": 14.191903114318848, "learning_rate": 2.5600000000000002e-05, "loss": 0.4097, "step": 7564 }, { "epoch": 48.76465863453815, "grad_norm": 11.864990234375, "learning_rate": 2.5596774193548388e-05, "loss": 0.3938, "step": 7565 }, { "epoch": 48.7710843373494, "grad_norm": 7.944164276123047, "learning_rate": 2.5593548387096777e-05, "loss": 0.3406, "step": 7566 }, { "epoch": 48.77751004016064, "grad_norm": 6.351147651672363, "learning_rate": 2.5590322580645166e-05, "loss": 0.271, "step": 7567 }, { "epoch": 48.783935742971885, "grad_norm": 10.896486282348633, "learning_rate": 2.558709677419355e-05, "loss": 0.2968, "step": 7568 }, { "epoch": 48.790361445783134, "grad_norm": 24.812089920043945, "learning_rate": 2.5583870967741934e-05, "loss": 0.3177, "step": 7569 }, { "epoch": 48.796787148594376, "grad_norm": 11.386155128479004, "learning_rate": 2.5580645161290323e-05, "loss": 0.3565, "step": 7570 }, { "epoch": 48.803212851405625, "grad_norm": 8.908379554748535, "learning_rate": 2.5577419354838712e-05, "loss": 0.4052, "step": 7571 }, { "epoch": 48.80963855421687, "grad_norm": 10.968917846679688, "learning_rate": 2.5574193548387098e-05, "loss": 0.3164, "step": 7572 }, { "epoch": 48.81606425702811, "grad_norm": 11.091605186462402, "learning_rate": 2.5570967741935487e-05, "loss": 0.5443, "step": 7573 }, { "epoch": 48.82248995983936, "grad_norm": 13.662432670593262, "learning_rate": 2.5567741935483873e-05, "loss": 0.4549, "step": 7574 }, { "epoch": 48.8289156626506, "grad_norm": 9.536479949951172, "learning_rate": 2.5564516129032255e-05, "loss": 0.4645, "step": 7575 }, { "epoch": 48.83534136546185, "grad_norm": 7.510072708129883, "learning_rate": 2.5561290322580644e-05, "loss": 0.2558, "step": 7576 }, { "epoch": 48.84176706827309, "grad_norm": 12.2244291305542, "learning_rate": 2.5558064516129033e-05, "loss": 0.6137, "step": 7577 }, { "epoch": 48.848192771084335, "grad_norm": 9.642499923706055, "learning_rate": 2.555483870967742e-05, "loss": 0.55, "step": 7578 }, { "epoch": 48.854618473895584, "grad_norm": 19.21794319152832, "learning_rate": 2.5551612903225808e-05, "loss": 0.3471, "step": 7579 }, { "epoch": 48.86104417670683, "grad_norm": 14.097293853759766, "learning_rate": 2.5548387096774197e-05, "loss": 0.5445, "step": 7580 }, { "epoch": 48.86746987951807, "grad_norm": 8.055538177490234, "learning_rate": 2.5545161290322583e-05, "loss": 0.3327, "step": 7581 }, { "epoch": 48.87389558232932, "grad_norm": 14.528288841247559, "learning_rate": 2.5541935483870972e-05, "loss": 0.7102, "step": 7582 }, { "epoch": 48.88032128514056, "grad_norm": 3.971771717071533, "learning_rate": 2.5538709677419354e-05, "loss": 0.3157, "step": 7583 }, { "epoch": 48.88674698795181, "grad_norm": 7.807028770446777, "learning_rate": 2.553548387096774e-05, "loss": 0.346, "step": 7584 }, { "epoch": 48.89317269076305, "grad_norm": 12.07834529876709, "learning_rate": 2.553225806451613e-05, "loss": 0.2952, "step": 7585 }, { "epoch": 48.899598393574294, "grad_norm": 13.986035346984863, "learning_rate": 2.5529032258064518e-05, "loss": 0.3164, "step": 7586 }, { "epoch": 48.90602409638554, "grad_norm": 7.318881034851074, "learning_rate": 2.5525806451612904e-05, "loss": 0.2909, "step": 7587 }, { "epoch": 48.912449799196786, "grad_norm": 7.019761562347412, "learning_rate": 2.5522580645161293e-05, "loss": 0.3118, "step": 7588 }, { "epoch": 48.918875502008035, "grad_norm": 16.205825805664062, "learning_rate": 2.551935483870968e-05, "loss": 0.521, "step": 7589 }, { "epoch": 48.92530120481928, "grad_norm": 6.116755485534668, "learning_rate": 2.5516129032258068e-05, "loss": 0.2949, "step": 7590 }, { "epoch": 48.93172690763052, "grad_norm": 11.2490816116333, "learning_rate": 2.551290322580645e-05, "loss": 0.4218, "step": 7591 }, { "epoch": 48.93815261044177, "grad_norm": 21.145475387573242, "learning_rate": 2.550967741935484e-05, "loss": 0.5768, "step": 7592 }, { "epoch": 48.94457831325301, "grad_norm": 9.178004264831543, "learning_rate": 2.5506451612903225e-05, "loss": 0.2609, "step": 7593 }, { "epoch": 48.95100401606426, "grad_norm": 12.10524845123291, "learning_rate": 2.5503225806451614e-05, "loss": 0.5101, "step": 7594 }, { "epoch": 48.9574297188755, "grad_norm": 17.83115577697754, "learning_rate": 2.5500000000000003e-05, "loss": 0.6416, "step": 7595 }, { "epoch": 48.963855421686745, "grad_norm": 4.071291923522949, "learning_rate": 2.549677419354839e-05, "loss": 0.1569, "step": 7596 }, { "epoch": 48.970281124497994, "grad_norm": 15.604609489440918, "learning_rate": 2.5493548387096778e-05, "loss": 0.24, "step": 7597 }, { "epoch": 48.976706827309236, "grad_norm": 19.095062255859375, "learning_rate": 2.5490322580645164e-05, "loss": 0.4578, "step": 7598 }, { "epoch": 48.98313253012048, "grad_norm": 3.925823926925659, "learning_rate": 2.5487096774193546e-05, "loss": 0.163, "step": 7599 }, { "epoch": 48.98955823293173, "grad_norm": 9.436901092529297, "learning_rate": 2.5483870967741935e-05, "loss": 0.305, "step": 7600 }, { "epoch": 48.99598393574297, "grad_norm": 8.594244956970215, "learning_rate": 2.5480645161290324e-05, "loss": 0.2947, "step": 7601 }, { "epoch": 49.00562248995984, "grad_norm": 23.83285140991211, "learning_rate": 2.547741935483871e-05, "loss": 0.221, "step": 7602 }, { "epoch": 49.01204819277108, "grad_norm": 2.445651054382324, "learning_rate": 2.54741935483871e-05, "loss": 0.2426, "step": 7603 }, { "epoch": 49.01847389558233, "grad_norm": 10.815884590148926, "learning_rate": 2.5470967741935485e-05, "loss": 0.4952, "step": 7604 }, { "epoch": 49.024899598393574, "grad_norm": 10.6751708984375, "learning_rate": 2.5467741935483874e-05, "loss": 0.3367, "step": 7605 }, { "epoch": 49.03132530120482, "grad_norm": 20.34043312072754, "learning_rate": 2.5464516129032263e-05, "loss": 0.4869, "step": 7606 }, { "epoch": 49.037751004016066, "grad_norm": 15.5840482711792, "learning_rate": 2.5461290322580645e-05, "loss": 0.1948, "step": 7607 }, { "epoch": 49.04417670682731, "grad_norm": 15.750227928161621, "learning_rate": 2.545806451612903e-05, "loss": 0.5343, "step": 7608 }, { "epoch": 49.05060240963856, "grad_norm": 22.083396911621094, "learning_rate": 2.545483870967742e-05, "loss": 0.2937, "step": 7609 }, { "epoch": 49.0570281124498, "grad_norm": 10.513264656066895, "learning_rate": 2.545161290322581e-05, "loss": 0.5155, "step": 7610 }, { "epoch": 49.06345381526104, "grad_norm": 19.27533531188965, "learning_rate": 2.5448387096774195e-05, "loss": 0.3316, "step": 7611 }, { "epoch": 49.06987951807229, "grad_norm": 9.07114028930664, "learning_rate": 2.5445161290322584e-05, "loss": 0.2195, "step": 7612 }, { "epoch": 49.07630522088353, "grad_norm": 16.76089859008789, "learning_rate": 2.544193548387097e-05, "loss": 0.3338, "step": 7613 }, { "epoch": 49.08273092369478, "grad_norm": 17.410972595214844, "learning_rate": 2.5438709677419352e-05, "loss": 0.5045, "step": 7614 }, { "epoch": 49.089156626506025, "grad_norm": 15.121792793273926, "learning_rate": 2.543548387096774e-05, "loss": 0.5146, "step": 7615 }, { "epoch": 49.09558232931727, "grad_norm": 3.893282890319824, "learning_rate": 2.543225806451613e-05, "loss": 0.2135, "step": 7616 }, { "epoch": 49.102008032128516, "grad_norm": 5.419770240783691, "learning_rate": 2.5429032258064516e-05, "loss": 0.3104, "step": 7617 }, { "epoch": 49.10843373493976, "grad_norm": 2.880505084991455, "learning_rate": 2.5425806451612905e-05, "loss": 0.1602, "step": 7618 }, { "epoch": 49.114859437751, "grad_norm": 7.712810039520264, "learning_rate": 2.5422580645161294e-05, "loss": 0.392, "step": 7619 }, { "epoch": 49.12128514056225, "grad_norm": 27.92974853515625, "learning_rate": 2.541935483870968e-05, "loss": 0.3189, "step": 7620 }, { "epoch": 49.12771084337349, "grad_norm": 15.083423614501953, "learning_rate": 2.541612903225807e-05, "loss": 0.3281, "step": 7621 }, { "epoch": 49.13413654618474, "grad_norm": 21.015953063964844, "learning_rate": 2.541290322580645e-05, "loss": 0.6078, "step": 7622 }, { "epoch": 49.140562248995984, "grad_norm": 6.7753143310546875, "learning_rate": 2.5409677419354837e-05, "loss": 0.3391, "step": 7623 }, { "epoch": 49.146987951807226, "grad_norm": 24.01690101623535, "learning_rate": 2.5406451612903226e-05, "loss": 0.4399, "step": 7624 }, { "epoch": 49.153413654618475, "grad_norm": 9.137487411499023, "learning_rate": 2.5403225806451615e-05, "loss": 0.4035, "step": 7625 }, { "epoch": 49.15983935742972, "grad_norm": 43.34367370605469, "learning_rate": 2.54e-05, "loss": 0.4665, "step": 7626 }, { "epoch": 49.16626506024097, "grad_norm": 13.31959056854248, "learning_rate": 2.539677419354839e-05, "loss": 0.7722, "step": 7627 }, { "epoch": 49.17269076305221, "grad_norm": 2.8568050861358643, "learning_rate": 2.5393548387096776e-05, "loss": 0.1491, "step": 7628 }, { "epoch": 49.17911646586345, "grad_norm": 12.000348091125488, "learning_rate": 2.5390322580645165e-05, "loss": 0.288, "step": 7629 }, { "epoch": 49.1855421686747, "grad_norm": 13.878386497497559, "learning_rate": 2.5387096774193547e-05, "loss": 0.5455, "step": 7630 }, { "epoch": 49.19196787148594, "grad_norm": 21.48633575439453, "learning_rate": 2.5383870967741936e-05, "loss": 0.2497, "step": 7631 }, { "epoch": 49.19839357429719, "grad_norm": 24.037124633789062, "learning_rate": 2.5380645161290322e-05, "loss": 0.5409, "step": 7632 }, { "epoch": 49.204819277108435, "grad_norm": 25.118059158325195, "learning_rate": 2.537741935483871e-05, "loss": 0.475, "step": 7633 }, { "epoch": 49.21124497991968, "grad_norm": 14.426032066345215, "learning_rate": 2.53741935483871e-05, "loss": 0.3872, "step": 7634 }, { "epoch": 49.217670682730926, "grad_norm": 19.089134216308594, "learning_rate": 2.5370967741935486e-05, "loss": 0.3623, "step": 7635 }, { "epoch": 49.22409638554217, "grad_norm": 8.479887008666992, "learning_rate": 2.5367741935483875e-05, "loss": 0.4481, "step": 7636 }, { "epoch": 49.23052208835341, "grad_norm": 7.5994133949279785, "learning_rate": 2.536451612903226e-05, "loss": 0.2979, "step": 7637 }, { "epoch": 49.23694779116466, "grad_norm": 12.518744468688965, "learning_rate": 2.5361290322580643e-05, "loss": 0.2847, "step": 7638 }, { "epoch": 49.2433734939759, "grad_norm": 11.880411148071289, "learning_rate": 2.5358064516129032e-05, "loss": 0.4362, "step": 7639 }, { "epoch": 49.24979919678715, "grad_norm": 10.349522590637207, "learning_rate": 2.535483870967742e-05, "loss": 0.1985, "step": 7640 }, { "epoch": 49.256224899598394, "grad_norm": 5.099886894226074, "learning_rate": 2.5351612903225807e-05, "loss": 0.575, "step": 7641 }, { "epoch": 49.262650602409636, "grad_norm": 13.337523460388184, "learning_rate": 2.5348387096774196e-05, "loss": 0.4591, "step": 7642 }, { "epoch": 49.269076305220885, "grad_norm": 3.480051040649414, "learning_rate": 2.5345161290322585e-05, "loss": 0.2308, "step": 7643 }, { "epoch": 49.27550200803213, "grad_norm": 12.660725593566895, "learning_rate": 2.534193548387097e-05, "loss": 0.1891, "step": 7644 }, { "epoch": 49.28192771084338, "grad_norm": 24.13661003112793, "learning_rate": 2.5338709677419353e-05, "loss": 0.3361, "step": 7645 }, { "epoch": 49.28835341365462, "grad_norm": 36.14182662963867, "learning_rate": 2.5335483870967742e-05, "loss": 0.725, "step": 7646 }, { "epoch": 49.29477911646586, "grad_norm": 5.6312055587768555, "learning_rate": 2.5332258064516128e-05, "loss": 0.3277, "step": 7647 }, { "epoch": 49.30120481927711, "grad_norm": 10.050627708435059, "learning_rate": 2.5329032258064517e-05, "loss": 0.246, "step": 7648 }, { "epoch": 49.30763052208835, "grad_norm": 19.20332145690918, "learning_rate": 2.5325806451612906e-05, "loss": 0.1768, "step": 7649 }, { "epoch": 49.3140562248996, "grad_norm": 27.738140106201172, "learning_rate": 2.532258064516129e-05, "loss": 0.2855, "step": 7650 }, { "epoch": 49.320481927710844, "grad_norm": 3.008701801300049, "learning_rate": 2.531935483870968e-05, "loss": 0.1819, "step": 7651 }, { "epoch": 49.326907630522086, "grad_norm": 12.252419471740723, "learning_rate": 2.5316129032258066e-05, "loss": 0.5306, "step": 7652 }, { "epoch": 49.333333333333336, "grad_norm": 6.8027424812316895, "learning_rate": 2.531290322580645e-05, "loss": 0.4289, "step": 7653 }, { "epoch": 49.33975903614458, "grad_norm": 7.841494560241699, "learning_rate": 2.5309677419354838e-05, "loss": 0.3308, "step": 7654 }, { "epoch": 49.34618473895582, "grad_norm": 30.18126106262207, "learning_rate": 2.5306451612903227e-05, "loss": 0.412, "step": 7655 }, { "epoch": 49.35261044176707, "grad_norm": 8.266347885131836, "learning_rate": 2.5303225806451613e-05, "loss": 0.1881, "step": 7656 }, { "epoch": 49.35903614457831, "grad_norm": 11.841397285461426, "learning_rate": 2.5300000000000002e-05, "loss": 0.5173, "step": 7657 }, { "epoch": 49.36546184738956, "grad_norm": 9.60648250579834, "learning_rate": 2.529677419354839e-05, "loss": 0.4166, "step": 7658 }, { "epoch": 49.3718875502008, "grad_norm": 28.42780303955078, "learning_rate": 2.5293548387096777e-05, "loss": 0.373, "step": 7659 }, { "epoch": 49.378313253012045, "grad_norm": 13.826963424682617, "learning_rate": 2.5290322580645166e-05, "loss": 0.2849, "step": 7660 }, { "epoch": 49.384738955823295, "grad_norm": 6.789860248565674, "learning_rate": 2.5287096774193548e-05, "loss": 0.3432, "step": 7661 }, { "epoch": 49.39116465863454, "grad_norm": 23.681339263916016, "learning_rate": 2.5283870967741934e-05, "loss": 0.4049, "step": 7662 }, { "epoch": 49.397590361445786, "grad_norm": 18.392866134643555, "learning_rate": 2.5280645161290323e-05, "loss": 0.299, "step": 7663 }, { "epoch": 49.40401606425703, "grad_norm": 7.797584056854248, "learning_rate": 2.5277419354838712e-05, "loss": 0.1612, "step": 7664 }, { "epoch": 49.41044176706827, "grad_norm": 12.985987663269043, "learning_rate": 2.5274193548387098e-05, "loss": 0.3247, "step": 7665 }, { "epoch": 49.41686746987952, "grad_norm": 28.017818450927734, "learning_rate": 2.5270967741935487e-05, "loss": 0.5232, "step": 7666 }, { "epoch": 49.42329317269076, "grad_norm": 24.287857055664062, "learning_rate": 2.5267741935483872e-05, "loss": 0.1851, "step": 7667 }, { "epoch": 49.429718875502004, "grad_norm": 4.182886123657227, "learning_rate": 2.526451612903226e-05, "loss": 0.2917, "step": 7668 }, { "epoch": 49.436144578313254, "grad_norm": 3.568812370300293, "learning_rate": 2.5261290322580644e-05, "loss": 0.3332, "step": 7669 }, { "epoch": 49.442570281124496, "grad_norm": 6.321065902709961, "learning_rate": 2.5258064516129033e-05, "loss": 0.3475, "step": 7670 }, { "epoch": 49.448995983935745, "grad_norm": 10.587515830993652, "learning_rate": 2.525483870967742e-05, "loss": 0.2918, "step": 7671 }, { "epoch": 49.45542168674699, "grad_norm": 3.4091734886169434, "learning_rate": 2.5251612903225808e-05, "loss": 0.39, "step": 7672 }, { "epoch": 49.46184738955823, "grad_norm": 9.624696731567383, "learning_rate": 2.5248387096774197e-05, "loss": 0.4924, "step": 7673 }, { "epoch": 49.46827309236948, "grad_norm": 15.236140251159668, "learning_rate": 2.5245161290322582e-05, "loss": 0.2639, "step": 7674 }, { "epoch": 49.47469879518072, "grad_norm": 23.112852096557617, "learning_rate": 2.524193548387097e-05, "loss": 0.3807, "step": 7675 }, { "epoch": 49.48112449799197, "grad_norm": 4.315866470336914, "learning_rate": 2.5238709677419357e-05, "loss": 0.1919, "step": 7676 }, { "epoch": 49.48755020080321, "grad_norm": 9.826911926269531, "learning_rate": 2.523548387096774e-05, "loss": 0.3276, "step": 7677 }, { "epoch": 49.493975903614455, "grad_norm": 5.546341896057129, "learning_rate": 2.523225806451613e-05, "loss": 0.3597, "step": 7678 }, { "epoch": 49.500401606425704, "grad_norm": 34.8401985168457, "learning_rate": 2.5229032258064518e-05, "loss": 0.4701, "step": 7679 }, { "epoch": 49.50682730923695, "grad_norm": 9.16073226928711, "learning_rate": 2.5225806451612903e-05, "loss": 0.2222, "step": 7680 }, { "epoch": 49.513253012048196, "grad_norm": 22.314193725585938, "learning_rate": 2.5222580645161293e-05, "loss": 0.6181, "step": 7681 }, { "epoch": 49.51967871485944, "grad_norm": 17.764490127563477, "learning_rate": 2.521935483870968e-05, "loss": 0.255, "step": 7682 }, { "epoch": 49.52610441767068, "grad_norm": 15.271461486816406, "learning_rate": 2.5216129032258067e-05, "loss": 0.2033, "step": 7683 }, { "epoch": 49.53253012048193, "grad_norm": 4.263397216796875, "learning_rate": 2.521290322580645e-05, "loss": 0.5331, "step": 7684 }, { "epoch": 49.53895582329317, "grad_norm": 12.870203971862793, "learning_rate": 2.520967741935484e-05, "loss": 0.4561, "step": 7685 }, { "epoch": 49.54538152610442, "grad_norm": 16.30982208251953, "learning_rate": 2.5206451612903224e-05, "loss": 0.2788, "step": 7686 }, { "epoch": 49.55180722891566, "grad_norm": 15.480277061462402, "learning_rate": 2.5203225806451614e-05, "loss": 0.3127, "step": 7687 }, { "epoch": 49.558232931726906, "grad_norm": 23.504053115844727, "learning_rate": 2.5200000000000003e-05, "loss": 0.3645, "step": 7688 }, { "epoch": 49.564658634538155, "grad_norm": 5.765514373779297, "learning_rate": 2.519677419354839e-05, "loss": 0.1985, "step": 7689 }, { "epoch": 49.5710843373494, "grad_norm": 16.487415313720703, "learning_rate": 2.5193548387096777e-05, "loss": 0.3302, "step": 7690 }, { "epoch": 49.57751004016064, "grad_norm": 12.604153633117676, "learning_rate": 2.5190322580645163e-05, "loss": 0.4542, "step": 7691 }, { "epoch": 49.58393574297189, "grad_norm": 33.562129974365234, "learning_rate": 2.5187096774193545e-05, "loss": 0.4208, "step": 7692 }, { "epoch": 49.59036144578313, "grad_norm": 9.971574783325195, "learning_rate": 2.5183870967741935e-05, "loss": 0.4409, "step": 7693 }, { "epoch": 49.59678714859438, "grad_norm": 8.265867233276367, "learning_rate": 2.5180645161290324e-05, "loss": 0.2153, "step": 7694 }, { "epoch": 49.60321285140562, "grad_norm": 9.165416717529297, "learning_rate": 2.517741935483871e-05, "loss": 0.5177, "step": 7695 }, { "epoch": 49.609638554216865, "grad_norm": 6.912754058837891, "learning_rate": 2.51741935483871e-05, "loss": 0.3885, "step": 7696 }, { "epoch": 49.616064257028114, "grad_norm": 12.908419609069824, "learning_rate": 2.5170967741935488e-05, "loss": 0.514, "step": 7697 }, { "epoch": 49.622489959839356, "grad_norm": 15.795821189880371, "learning_rate": 2.5167741935483873e-05, "loss": 0.3854, "step": 7698 }, { "epoch": 49.628915662650606, "grad_norm": 9.789693832397461, "learning_rate": 2.5164516129032262e-05, "loss": 0.1485, "step": 7699 }, { "epoch": 49.63534136546185, "grad_norm": 7.784651756286621, "learning_rate": 2.5161290322580645e-05, "loss": 0.305, "step": 7700 }, { "epoch": 49.64176706827309, "grad_norm": 26.687732696533203, "learning_rate": 2.515806451612903e-05, "loss": 0.3918, "step": 7701 }, { "epoch": 49.64819277108434, "grad_norm": 30.07540512084961, "learning_rate": 2.515483870967742e-05, "loss": 0.2935, "step": 7702 }, { "epoch": 49.65461847389558, "grad_norm": 3.889354705810547, "learning_rate": 2.515161290322581e-05, "loss": 0.1385, "step": 7703 }, { "epoch": 49.661044176706824, "grad_norm": 41.81861114501953, "learning_rate": 2.5148387096774194e-05, "loss": 0.7025, "step": 7704 }, { "epoch": 49.66746987951807, "grad_norm": 28.982210159301758, "learning_rate": 2.5145161290322583e-05, "loss": 0.3136, "step": 7705 }, { "epoch": 49.673895582329315, "grad_norm": 17.495967864990234, "learning_rate": 2.5141935483870972e-05, "loss": 0.3702, "step": 7706 }, { "epoch": 49.680321285140565, "grad_norm": 6.550161838531494, "learning_rate": 2.5138709677419358e-05, "loss": 0.2055, "step": 7707 }, { "epoch": 49.68674698795181, "grad_norm": 4.174894332885742, "learning_rate": 2.513548387096774e-05, "loss": 0.2811, "step": 7708 }, { "epoch": 49.69317269076305, "grad_norm": 9.416892051696777, "learning_rate": 2.513225806451613e-05, "loss": 0.4234, "step": 7709 }, { "epoch": 49.6995983935743, "grad_norm": 19.705209732055664, "learning_rate": 2.5129032258064515e-05, "loss": 0.2581, "step": 7710 }, { "epoch": 49.70602409638554, "grad_norm": 27.68037223815918, "learning_rate": 2.5125806451612904e-05, "loss": 0.4862, "step": 7711 }, { "epoch": 49.71244979919679, "grad_norm": 9.180672645568848, "learning_rate": 2.5122580645161294e-05, "loss": 0.3779, "step": 7712 }, { "epoch": 49.71887550200803, "grad_norm": 39.80828094482422, "learning_rate": 2.511935483870968e-05, "loss": 0.3707, "step": 7713 }, { "epoch": 49.725301204819274, "grad_norm": 17.967239379882812, "learning_rate": 2.511612903225807e-05, "loss": 0.8435, "step": 7714 }, { "epoch": 49.731726907630524, "grad_norm": 23.417409896850586, "learning_rate": 2.511290322580645e-05, "loss": 0.2822, "step": 7715 }, { "epoch": 49.738152610441766, "grad_norm": 3.485077142715454, "learning_rate": 2.5109677419354836e-05, "loss": 0.2535, "step": 7716 }, { "epoch": 49.744578313253015, "grad_norm": 5.279077053070068, "learning_rate": 2.5106451612903225e-05, "loss": 0.4378, "step": 7717 }, { "epoch": 49.75100401606426, "grad_norm": 5.495669841766357, "learning_rate": 2.5103225806451615e-05, "loss": 0.4132, "step": 7718 }, { "epoch": 49.7574297188755, "grad_norm": 26.621482849121094, "learning_rate": 2.51e-05, "loss": 0.4862, "step": 7719 }, { "epoch": 49.76385542168675, "grad_norm": 13.55643367767334, "learning_rate": 2.509677419354839e-05, "loss": 0.2367, "step": 7720 }, { "epoch": 49.77028112449799, "grad_norm": 16.232969284057617, "learning_rate": 2.509354838709678e-05, "loss": 0.3513, "step": 7721 }, { "epoch": 49.77670682730923, "grad_norm": 7.686426639556885, "learning_rate": 2.5090322580645164e-05, "loss": 0.2927, "step": 7722 }, { "epoch": 49.78313253012048, "grad_norm": 4.0712127685546875, "learning_rate": 2.5087096774193546e-05, "loss": 0.1928, "step": 7723 }, { "epoch": 49.789558232931725, "grad_norm": 8.723182678222656, "learning_rate": 2.5083870967741936e-05, "loss": 0.36, "step": 7724 }, { "epoch": 49.795983935742974, "grad_norm": 8.421792030334473, "learning_rate": 2.508064516129032e-05, "loss": 0.2531, "step": 7725 }, { "epoch": 49.80240963855422, "grad_norm": 11.453817367553711, "learning_rate": 2.507741935483871e-05, "loss": 0.5077, "step": 7726 }, { "epoch": 49.80883534136546, "grad_norm": 9.25241756439209, "learning_rate": 2.50741935483871e-05, "loss": 0.4972, "step": 7727 }, { "epoch": 49.81526104417671, "grad_norm": 16.431671142578125, "learning_rate": 2.5070967741935485e-05, "loss": 0.4517, "step": 7728 }, { "epoch": 49.82168674698795, "grad_norm": 15.004380226135254, "learning_rate": 2.5067741935483874e-05, "loss": 0.3203, "step": 7729 }, { "epoch": 49.8281124497992, "grad_norm": 21.64318084716797, "learning_rate": 2.5064516129032263e-05, "loss": 0.3877, "step": 7730 }, { "epoch": 49.83453815261044, "grad_norm": 8.871678352355957, "learning_rate": 2.5061290322580642e-05, "loss": 0.426, "step": 7731 }, { "epoch": 49.840963855421684, "grad_norm": 19.95966148376465, "learning_rate": 2.505806451612903e-05, "loss": 0.2942, "step": 7732 }, { "epoch": 49.84738955823293, "grad_norm": 14.53647518157959, "learning_rate": 2.505483870967742e-05, "loss": 0.3056, "step": 7733 }, { "epoch": 49.853815261044176, "grad_norm": 20.69304656982422, "learning_rate": 2.5051612903225806e-05, "loss": 0.3557, "step": 7734 }, { "epoch": 49.860240963855425, "grad_norm": 11.834735870361328, "learning_rate": 2.5048387096774195e-05, "loss": 0.3514, "step": 7735 }, { "epoch": 49.86666666666667, "grad_norm": 14.096933364868164, "learning_rate": 2.5045161290322584e-05, "loss": 0.3675, "step": 7736 }, { "epoch": 49.87309236947791, "grad_norm": 8.124133110046387, "learning_rate": 2.504193548387097e-05, "loss": 0.4418, "step": 7737 }, { "epoch": 49.87951807228916, "grad_norm": 2.522624969482422, "learning_rate": 2.503870967741936e-05, "loss": 0.6761, "step": 7738 }, { "epoch": 49.8859437751004, "grad_norm": 3.54211163520813, "learning_rate": 2.503548387096774e-05, "loss": 0.6599, "step": 7739 }, { "epoch": 49.89236947791164, "grad_norm": 21.243162155151367, "learning_rate": 2.5032258064516127e-05, "loss": 0.5464, "step": 7740 }, { "epoch": 49.89879518072289, "grad_norm": 12.032764434814453, "learning_rate": 2.5029032258064516e-05, "loss": 0.279, "step": 7741 }, { "epoch": 49.905220883534135, "grad_norm": 8.496574401855469, "learning_rate": 2.5025806451612905e-05, "loss": 0.6284, "step": 7742 }, { "epoch": 49.911646586345384, "grad_norm": 38.81671142578125, "learning_rate": 2.502258064516129e-05, "loss": 0.5215, "step": 7743 }, { "epoch": 49.918072289156626, "grad_norm": 8.88745403289795, "learning_rate": 2.501935483870968e-05, "loss": 0.2298, "step": 7744 }, { "epoch": 49.92449799196787, "grad_norm": 17.426877975463867, "learning_rate": 2.501612903225807e-05, "loss": 0.2468, "step": 7745 }, { "epoch": 49.93092369477912, "grad_norm": 12.882323265075684, "learning_rate": 2.5012903225806455e-05, "loss": 0.6876, "step": 7746 }, { "epoch": 49.93734939759036, "grad_norm": 3.0084011554718018, "learning_rate": 2.5009677419354837e-05, "loss": 0.1019, "step": 7747 }, { "epoch": 49.94377510040161, "grad_norm": 16.404069900512695, "learning_rate": 2.5006451612903226e-05, "loss": 0.5195, "step": 7748 }, { "epoch": 49.95020080321285, "grad_norm": 29.739299774169922, "learning_rate": 2.5003225806451612e-05, "loss": 0.4234, "step": 7749 }, { "epoch": 49.956626506024094, "grad_norm": 10.592819213867188, "learning_rate": 2.5e-05, "loss": 0.4381, "step": 7750 }, { "epoch": 49.96305220883534, "grad_norm": 17.669448852539062, "learning_rate": 2.499677419354839e-05, "loss": 0.1763, "step": 7751 }, { "epoch": 49.969477911646585, "grad_norm": 9.096053123474121, "learning_rate": 2.4993548387096773e-05, "loss": 0.4378, "step": 7752 }, { "epoch": 49.975903614457835, "grad_norm": 9.515458106994629, "learning_rate": 2.4990322580645162e-05, "loss": 0.3382, "step": 7753 }, { "epoch": 49.98232931726908, "grad_norm": 3.8511054515838623, "learning_rate": 2.498709677419355e-05, "loss": 0.1861, "step": 7754 }, { "epoch": 49.98875502008032, "grad_norm": 26.332658767700195, "learning_rate": 2.4983870967741937e-05, "loss": 0.4574, "step": 7755 }, { "epoch": 49.99518072289157, "grad_norm": 18.813478469848633, "learning_rate": 2.4980645161290322e-05, "loss": 0.6093, "step": 7756 }, { "epoch": 50.00481927710843, "grad_norm": 10.327225685119629, "learning_rate": 2.497741935483871e-05, "loss": 0.5326, "step": 7757 }, { "epoch": 50.01124497991968, "grad_norm": 5.4410014152526855, "learning_rate": 2.4974193548387097e-05, "loss": 0.1471, "step": 7758 }, { "epoch": 50.01767068273092, "grad_norm": 14.124382019042969, "learning_rate": 2.4970967741935486e-05, "loss": 0.3781, "step": 7759 }, { "epoch": 50.024096385542165, "grad_norm": 20.2247371673584, "learning_rate": 2.4967741935483872e-05, "loss": 0.2546, "step": 7760 }, { "epoch": 50.030522088353415, "grad_norm": 3.5908029079437256, "learning_rate": 2.4964516129032258e-05, "loss": 0.1592, "step": 7761 }, { "epoch": 50.03694779116466, "grad_norm": 14.085161209106445, "learning_rate": 2.4961290322580647e-05, "loss": 0.3925, "step": 7762 }, { "epoch": 50.043373493975906, "grad_norm": 21.729328155517578, "learning_rate": 2.4958064516129036e-05, "loss": 0.2315, "step": 7763 }, { "epoch": 50.04979919678715, "grad_norm": 5.588596820831299, "learning_rate": 2.4954838709677418e-05, "loss": 0.3164, "step": 7764 }, { "epoch": 50.05622489959839, "grad_norm": 11.348555564880371, "learning_rate": 2.4951612903225807e-05, "loss": 0.2741, "step": 7765 }, { "epoch": 50.06265060240964, "grad_norm": 11.441289901733398, "learning_rate": 2.4948387096774196e-05, "loss": 0.3781, "step": 7766 }, { "epoch": 50.06907630522088, "grad_norm": 12.591486930847168, "learning_rate": 2.4945161290322582e-05, "loss": 0.3806, "step": 7767 }, { "epoch": 50.07550200803213, "grad_norm": 8.595272064208984, "learning_rate": 2.4941935483870968e-05, "loss": 0.6667, "step": 7768 }, { "epoch": 50.081927710843374, "grad_norm": 11.78748893737793, "learning_rate": 2.4938709677419357e-05, "loss": 0.3656, "step": 7769 }, { "epoch": 50.088353413654616, "grad_norm": 23.81410789489746, "learning_rate": 2.4935483870967742e-05, "loss": 0.298, "step": 7770 }, { "epoch": 50.094779116465865, "grad_norm": 24.007802963256836, "learning_rate": 2.493225806451613e-05, "loss": 0.4294, "step": 7771 }, { "epoch": 50.10120481927711, "grad_norm": 5.40529727935791, "learning_rate": 2.4929032258064517e-05, "loss": 0.3197, "step": 7772 }, { "epoch": 50.10763052208836, "grad_norm": 5.496341228485107, "learning_rate": 2.4925806451612903e-05, "loss": 0.2052, "step": 7773 }, { "epoch": 50.1140562248996, "grad_norm": 7.7460036277771, "learning_rate": 2.4922580645161292e-05, "loss": 0.2899, "step": 7774 }, { "epoch": 50.12048192771084, "grad_norm": 19.393953323364258, "learning_rate": 2.491935483870968e-05, "loss": 0.3566, "step": 7775 }, { "epoch": 50.12690763052209, "grad_norm": 16.686927795410156, "learning_rate": 2.4916129032258063e-05, "loss": 0.4157, "step": 7776 }, { "epoch": 50.13333333333333, "grad_norm": 13.15408706665039, "learning_rate": 2.4912903225806453e-05, "loss": 0.2833, "step": 7777 }, { "epoch": 50.139759036144575, "grad_norm": 15.261802673339844, "learning_rate": 2.490967741935484e-05, "loss": 0.3525, "step": 7778 }, { "epoch": 50.146184738955824, "grad_norm": 8.186882019042969, "learning_rate": 2.4906451612903227e-05, "loss": 0.1443, "step": 7779 }, { "epoch": 50.15261044176707, "grad_norm": 18.068401336669922, "learning_rate": 2.4903225806451613e-05, "loss": 0.295, "step": 7780 }, { "epoch": 50.159036144578316, "grad_norm": 23.788930892944336, "learning_rate": 2.4900000000000002e-05, "loss": 0.5342, "step": 7781 }, { "epoch": 50.16546184738956, "grad_norm": 7.242359638214111, "learning_rate": 2.4896774193548388e-05, "loss": 0.2274, "step": 7782 }, { "epoch": 50.1718875502008, "grad_norm": 38.67057418823242, "learning_rate": 2.4893548387096777e-05, "loss": 0.3325, "step": 7783 }, { "epoch": 50.17831325301205, "grad_norm": 22.881227493286133, "learning_rate": 2.4890322580645163e-05, "loss": 0.3517, "step": 7784 }, { "epoch": 50.18473895582329, "grad_norm": 8.892716407775879, "learning_rate": 2.488709677419355e-05, "loss": 0.5471, "step": 7785 }, { "epoch": 50.19116465863454, "grad_norm": 9.693360328674316, "learning_rate": 2.4883870967741937e-05, "loss": 0.2072, "step": 7786 }, { "epoch": 50.19759036144578, "grad_norm": 3.8569388389587402, "learning_rate": 2.4880645161290323e-05, "loss": 0.5463, "step": 7787 }, { "epoch": 50.204016064257026, "grad_norm": 9.976887702941895, "learning_rate": 2.487741935483871e-05, "loss": 0.5333, "step": 7788 }, { "epoch": 50.210441767068275, "grad_norm": 5.310797691345215, "learning_rate": 2.4874193548387098e-05, "loss": 0.4974, "step": 7789 }, { "epoch": 50.21686746987952, "grad_norm": 37.02044677734375, "learning_rate": 2.4870967741935487e-05, "loss": 0.2735, "step": 7790 }, { "epoch": 50.22329317269077, "grad_norm": 29.371646881103516, "learning_rate": 2.486774193548387e-05, "loss": 0.303, "step": 7791 }, { "epoch": 50.22971887550201, "grad_norm": 8.202502250671387, "learning_rate": 2.486451612903226e-05, "loss": 0.3101, "step": 7792 }, { "epoch": 50.23614457831325, "grad_norm": 8.387465476989746, "learning_rate": 2.4861290322580648e-05, "loss": 0.4882, "step": 7793 }, { "epoch": 50.2425702811245, "grad_norm": 14.096904754638672, "learning_rate": 2.4858064516129033e-05, "loss": 0.3244, "step": 7794 }, { "epoch": 50.24899598393574, "grad_norm": 19.479860305786133, "learning_rate": 2.485483870967742e-05, "loss": 0.2013, "step": 7795 }, { "epoch": 50.255421686746985, "grad_norm": 8.73294448852539, "learning_rate": 2.4851612903225808e-05, "loss": 0.167, "step": 7796 }, { "epoch": 50.261847389558234, "grad_norm": 11.10960865020752, "learning_rate": 2.4848387096774194e-05, "loss": 0.3938, "step": 7797 }, { "epoch": 50.268273092369476, "grad_norm": 4.9773664474487305, "learning_rate": 2.4845161290322583e-05, "loss": 0.287, "step": 7798 }, { "epoch": 50.274698795180726, "grad_norm": 21.52992057800293, "learning_rate": 2.484193548387097e-05, "loss": 0.6338, "step": 7799 }, { "epoch": 50.28112449799197, "grad_norm": 11.26776123046875, "learning_rate": 2.4838709677419354e-05, "loss": 0.2352, "step": 7800 }, { "epoch": 50.28755020080321, "grad_norm": 31.8348445892334, "learning_rate": 2.4835483870967743e-05, "loss": 0.1465, "step": 7801 }, { "epoch": 50.29397590361446, "grad_norm": 4.086488246917725, "learning_rate": 2.4832258064516132e-05, "loss": 0.1009, "step": 7802 }, { "epoch": 50.3004016064257, "grad_norm": 16.736846923828125, "learning_rate": 2.4829032258064515e-05, "loss": 0.4327, "step": 7803 }, { "epoch": 50.30682730923695, "grad_norm": 37.2752799987793, "learning_rate": 2.4825806451612904e-05, "loss": 0.4868, "step": 7804 }, { "epoch": 50.31325301204819, "grad_norm": 9.65968132019043, "learning_rate": 2.4822580645161293e-05, "loss": 0.3479, "step": 7805 }, { "epoch": 50.319678714859435, "grad_norm": 7.5970540046691895, "learning_rate": 2.481935483870968e-05, "loss": 0.4625, "step": 7806 }, { "epoch": 50.326104417670685, "grad_norm": 30.136985778808594, "learning_rate": 2.4816129032258064e-05, "loss": 0.3668, "step": 7807 }, { "epoch": 50.33253012048193, "grad_norm": 4.839444160461426, "learning_rate": 2.4812903225806454e-05, "loss": 0.4216, "step": 7808 }, { "epoch": 50.338955823293176, "grad_norm": 7.5022687911987305, "learning_rate": 2.480967741935484e-05, "loss": 0.4809, "step": 7809 }, { "epoch": 50.34538152610442, "grad_norm": 11.398527145385742, "learning_rate": 2.480645161290323e-05, "loss": 0.4852, "step": 7810 }, { "epoch": 50.35180722891566, "grad_norm": 24.938743591308594, "learning_rate": 2.4803225806451614e-05, "loss": 0.3238, "step": 7811 }, { "epoch": 50.35823293172691, "grad_norm": 23.33205795288086, "learning_rate": 2.48e-05, "loss": 0.496, "step": 7812 }, { "epoch": 50.36465863453815, "grad_norm": 20.512386322021484, "learning_rate": 2.479677419354839e-05, "loss": 0.3008, "step": 7813 }, { "epoch": 50.371084337349394, "grad_norm": 4.491701126098633, "learning_rate": 2.4793548387096778e-05, "loss": 0.1783, "step": 7814 }, { "epoch": 50.377510040160644, "grad_norm": 33.778255462646484, "learning_rate": 2.479032258064516e-05, "loss": 0.3431, "step": 7815 }, { "epoch": 50.383935742971886, "grad_norm": 3.497976303100586, "learning_rate": 2.478709677419355e-05, "loss": 0.1653, "step": 7816 }, { "epoch": 50.390361445783135, "grad_norm": 16.71702766418457, "learning_rate": 2.478387096774194e-05, "loss": 0.4439, "step": 7817 }, { "epoch": 50.39678714859438, "grad_norm": 18.49319076538086, "learning_rate": 2.4780645161290324e-05, "loss": 0.2322, "step": 7818 }, { "epoch": 50.40321285140562, "grad_norm": 4.251752853393555, "learning_rate": 2.477741935483871e-05, "loss": 0.1895, "step": 7819 }, { "epoch": 50.40963855421687, "grad_norm": 15.102534294128418, "learning_rate": 2.47741935483871e-05, "loss": 0.4017, "step": 7820 }, { "epoch": 50.41606425702811, "grad_norm": 3.8668735027313232, "learning_rate": 2.4770967741935485e-05, "loss": 0.1383, "step": 7821 }, { "epoch": 50.42248995983936, "grad_norm": 7.937746524810791, "learning_rate": 2.476774193548387e-05, "loss": 0.2966, "step": 7822 }, { "epoch": 50.4289156626506, "grad_norm": 14.065214157104492, "learning_rate": 2.476451612903226e-05, "loss": 0.3851, "step": 7823 }, { "epoch": 50.435341365461845, "grad_norm": 18.456520080566406, "learning_rate": 2.4761290322580645e-05, "loss": 0.6978, "step": 7824 }, { "epoch": 50.441767068273094, "grad_norm": 7.908552646636963, "learning_rate": 2.4758064516129034e-05, "loss": 0.3695, "step": 7825 }, { "epoch": 50.44819277108434, "grad_norm": 43.91160583496094, "learning_rate": 2.475483870967742e-05, "loss": 0.2835, "step": 7826 }, { "epoch": 50.454618473895586, "grad_norm": 21.103286743164062, "learning_rate": 2.4751612903225806e-05, "loss": 0.5109, "step": 7827 }, { "epoch": 50.46104417670683, "grad_norm": 13.172975540161133, "learning_rate": 2.4748387096774195e-05, "loss": 0.168, "step": 7828 }, { "epoch": 50.46746987951807, "grad_norm": 5.4694414138793945, "learning_rate": 2.4745161290322584e-05, "loss": 0.1325, "step": 7829 }, { "epoch": 50.47389558232932, "grad_norm": 28.56239891052246, "learning_rate": 2.4741935483870966e-05, "loss": 0.341, "step": 7830 }, { "epoch": 50.48032128514056, "grad_norm": 25.879531860351562, "learning_rate": 2.4738709677419355e-05, "loss": 0.6538, "step": 7831 }, { "epoch": 50.486746987951804, "grad_norm": 19.022659301757812, "learning_rate": 2.4735483870967744e-05, "loss": 0.3337, "step": 7832 }, { "epoch": 50.49317269076305, "grad_norm": 17.558395385742188, "learning_rate": 2.473225806451613e-05, "loss": 0.6202, "step": 7833 }, { "epoch": 50.499598393574296, "grad_norm": 43.645355224609375, "learning_rate": 2.4729032258064516e-05, "loss": 0.3481, "step": 7834 }, { "epoch": 50.506024096385545, "grad_norm": 16.42315101623535, "learning_rate": 2.4725806451612905e-05, "loss": 0.3558, "step": 7835 }, { "epoch": 50.51244979919679, "grad_norm": 20.981382369995117, "learning_rate": 2.472258064516129e-05, "loss": 0.5825, "step": 7836 }, { "epoch": 50.51887550200803, "grad_norm": 8.873213768005371, "learning_rate": 2.471935483870968e-05, "loss": 0.1309, "step": 7837 }, { "epoch": 50.52530120481928, "grad_norm": 7.9421257972717285, "learning_rate": 2.4716129032258065e-05, "loss": 0.26, "step": 7838 }, { "epoch": 50.53172690763052, "grad_norm": 4.573850154876709, "learning_rate": 2.471290322580645e-05, "loss": 0.2072, "step": 7839 }, { "epoch": 50.53815261044177, "grad_norm": 50.874000549316406, "learning_rate": 2.470967741935484e-05, "loss": 0.5966, "step": 7840 }, { "epoch": 50.54457831325301, "grad_norm": 6.029675006866455, "learning_rate": 2.470645161290323e-05, "loss": 0.3612, "step": 7841 }, { "epoch": 50.551004016064255, "grad_norm": 20.898456573486328, "learning_rate": 2.470322580645161e-05, "loss": 0.3987, "step": 7842 }, { "epoch": 50.557429718875504, "grad_norm": 7.480789661407471, "learning_rate": 2.47e-05, "loss": 0.436, "step": 7843 }, { "epoch": 50.563855421686746, "grad_norm": 8.485367774963379, "learning_rate": 2.469677419354839e-05, "loss": 0.1817, "step": 7844 }, { "epoch": 50.570281124497996, "grad_norm": 12.173864364624023, "learning_rate": 2.4693548387096775e-05, "loss": 0.4442, "step": 7845 }, { "epoch": 50.57670682730924, "grad_norm": 20.57891273498535, "learning_rate": 2.469032258064516e-05, "loss": 0.2414, "step": 7846 }, { "epoch": 50.58313253012048, "grad_norm": 6.7440290451049805, "learning_rate": 2.468709677419355e-05, "loss": 0.3338, "step": 7847 }, { "epoch": 50.58955823293173, "grad_norm": 14.237239837646484, "learning_rate": 2.4683870967741936e-05, "loss": 0.405, "step": 7848 }, { "epoch": 50.59598393574297, "grad_norm": 7.581586837768555, "learning_rate": 2.4680645161290325e-05, "loss": 0.2514, "step": 7849 }, { "epoch": 50.602409638554214, "grad_norm": 22.663238525390625, "learning_rate": 2.467741935483871e-05, "loss": 0.2788, "step": 7850 }, { "epoch": 50.60883534136546, "grad_norm": 11.40705680847168, "learning_rate": 2.4674193548387097e-05, "loss": 0.4029, "step": 7851 }, { "epoch": 50.615261044176705, "grad_norm": 3.5533502101898193, "learning_rate": 2.4670967741935486e-05, "loss": 0.2995, "step": 7852 }, { "epoch": 50.621686746987955, "grad_norm": 11.857319831848145, "learning_rate": 2.4667741935483875e-05, "loss": 0.5712, "step": 7853 }, { "epoch": 50.6281124497992, "grad_norm": 10.829794883728027, "learning_rate": 2.4664516129032257e-05, "loss": 0.2036, "step": 7854 }, { "epoch": 50.63453815261044, "grad_norm": 3.156787157058716, "learning_rate": 2.4661290322580646e-05, "loss": 0.2658, "step": 7855 }, { "epoch": 50.64096385542169, "grad_norm": 23.98505401611328, "learning_rate": 2.4658064516129035e-05, "loss": 0.5997, "step": 7856 }, { "epoch": 50.64738955823293, "grad_norm": 12.106834411621094, "learning_rate": 2.4654838709677418e-05, "loss": 0.4189, "step": 7857 }, { "epoch": 50.65381526104418, "grad_norm": 16.467947006225586, "learning_rate": 2.4651612903225807e-05, "loss": 0.3902, "step": 7858 }, { "epoch": 50.66024096385542, "grad_norm": 17.002504348754883, "learning_rate": 2.4648387096774196e-05, "loss": 0.3117, "step": 7859 }, { "epoch": 50.666666666666664, "grad_norm": 8.162485122680664, "learning_rate": 2.464516129032258e-05, "loss": 0.704, "step": 7860 }, { "epoch": 50.673092369477914, "grad_norm": 6.961740493774414, "learning_rate": 2.4641935483870967e-05, "loss": 0.1949, "step": 7861 }, { "epoch": 50.679518072289156, "grad_norm": 16.809755325317383, "learning_rate": 2.4638709677419356e-05, "loss": 0.3787, "step": 7862 }, { "epoch": 50.6859437751004, "grad_norm": 37.494712829589844, "learning_rate": 2.4635483870967742e-05, "loss": 0.1381, "step": 7863 }, { "epoch": 50.69236947791165, "grad_norm": 30.66067123413086, "learning_rate": 2.463225806451613e-05, "loss": 0.45, "step": 7864 }, { "epoch": 50.69879518072289, "grad_norm": 18.820323944091797, "learning_rate": 2.4629032258064517e-05, "loss": 0.5649, "step": 7865 }, { "epoch": 50.70522088353414, "grad_norm": 10.316113471984863, "learning_rate": 2.4625806451612902e-05, "loss": 0.4828, "step": 7866 }, { "epoch": 50.71164658634538, "grad_norm": 2.5599329471588135, "learning_rate": 2.462258064516129e-05, "loss": 0.3115, "step": 7867 }, { "epoch": 50.71807228915662, "grad_norm": 30.232643127441406, "learning_rate": 2.461935483870968e-05, "loss": 0.3998, "step": 7868 }, { "epoch": 50.72449799196787, "grad_norm": 17.545969009399414, "learning_rate": 2.4616129032258063e-05, "loss": 0.3968, "step": 7869 }, { "epoch": 50.730923694779115, "grad_norm": 27.089401245117188, "learning_rate": 2.4612903225806452e-05, "loss": 0.3968, "step": 7870 }, { "epoch": 50.737349397590364, "grad_norm": 4.025360107421875, "learning_rate": 2.460967741935484e-05, "loss": 0.2968, "step": 7871 }, { "epoch": 50.743775100401606, "grad_norm": 9.197416305541992, "learning_rate": 2.4606451612903227e-05, "loss": 0.2352, "step": 7872 }, { "epoch": 50.75020080321285, "grad_norm": 35.44845199584961, "learning_rate": 2.4603225806451613e-05, "loss": 0.5679, "step": 7873 }, { "epoch": 50.7566265060241, "grad_norm": 49.75244140625, "learning_rate": 2.46e-05, "loss": 0.564, "step": 7874 }, { "epoch": 50.76305220883534, "grad_norm": 20.223724365234375, "learning_rate": 2.4596774193548387e-05, "loss": 0.6771, "step": 7875 }, { "epoch": 50.76947791164659, "grad_norm": 3.6950225830078125, "learning_rate": 2.4593548387096776e-05, "loss": 0.2528, "step": 7876 }, { "epoch": 50.77590361445783, "grad_norm": 12.592577934265137, "learning_rate": 2.4590322580645162e-05, "loss": 0.4935, "step": 7877 }, { "epoch": 50.782329317269074, "grad_norm": 4.7222514152526855, "learning_rate": 2.4587096774193548e-05, "loss": 0.3565, "step": 7878 }, { "epoch": 50.78875502008032, "grad_norm": 39.9859619140625, "learning_rate": 2.4583870967741937e-05, "loss": 0.2743, "step": 7879 }, { "epoch": 50.795180722891565, "grad_norm": 33.608116149902344, "learning_rate": 2.4580645161290326e-05, "loss": 0.3843, "step": 7880 }, { "epoch": 50.801606425702815, "grad_norm": 27.388479232788086, "learning_rate": 2.457741935483871e-05, "loss": 0.4346, "step": 7881 }, { "epoch": 50.80803212851406, "grad_norm": 8.486562728881836, "learning_rate": 2.4574193548387097e-05, "loss": 0.2263, "step": 7882 }, { "epoch": 50.8144578313253, "grad_norm": 12.616827964782715, "learning_rate": 2.4570967741935487e-05, "loss": 0.3879, "step": 7883 }, { "epoch": 50.82088353413655, "grad_norm": 21.050033569335938, "learning_rate": 2.4567741935483872e-05, "loss": 0.4482, "step": 7884 }, { "epoch": 50.82730923694779, "grad_norm": 18.41457176208496, "learning_rate": 2.4564516129032258e-05, "loss": 0.3314, "step": 7885 }, { "epoch": 50.83373493975903, "grad_norm": 17.34960174560547, "learning_rate": 2.4561290322580647e-05, "loss": 0.365, "step": 7886 }, { "epoch": 50.84016064257028, "grad_norm": 27.314470291137695, "learning_rate": 2.4558064516129033e-05, "loss": 0.3203, "step": 7887 }, { "epoch": 50.846586345381525, "grad_norm": 16.89081382751465, "learning_rate": 2.4554838709677422e-05, "loss": 0.4925, "step": 7888 }, { "epoch": 50.853012048192774, "grad_norm": 42.42847442626953, "learning_rate": 2.4551612903225808e-05, "loss": 0.3106, "step": 7889 }, { "epoch": 50.859437751004016, "grad_norm": 8.10696792602539, "learning_rate": 2.4548387096774193e-05, "loss": 0.3249, "step": 7890 }, { "epoch": 50.86586345381526, "grad_norm": 7.206893444061279, "learning_rate": 2.4545161290322582e-05, "loss": 0.3403, "step": 7891 }, { "epoch": 50.87228915662651, "grad_norm": 180.69232177734375, "learning_rate": 2.454193548387097e-05, "loss": 0.6175, "step": 7892 }, { "epoch": 50.87871485943775, "grad_norm": 34.81171798706055, "learning_rate": 2.4538709677419354e-05, "loss": 0.7732, "step": 7893 }, { "epoch": 50.885140562249, "grad_norm": 30.119707107543945, "learning_rate": 2.4535483870967743e-05, "loss": 0.4406, "step": 7894 }, { "epoch": 50.89156626506024, "grad_norm": 18.501375198364258, "learning_rate": 2.4532258064516132e-05, "loss": 0.2004, "step": 7895 }, { "epoch": 50.897991967871484, "grad_norm": 16.2963924407959, "learning_rate": 2.4529032258064518e-05, "loss": 0.7725, "step": 7896 }, { "epoch": 50.90441767068273, "grad_norm": 7.059051990509033, "learning_rate": 2.4525806451612903e-05, "loss": 0.415, "step": 7897 }, { "epoch": 50.910843373493975, "grad_norm": 11.05044937133789, "learning_rate": 2.4522580645161292e-05, "loss": 0.2042, "step": 7898 }, { "epoch": 50.91726907630522, "grad_norm": 10.031455039978027, "learning_rate": 2.4519354838709678e-05, "loss": 0.4616, "step": 7899 }, { "epoch": 50.92369477911647, "grad_norm": 9.972018241882324, "learning_rate": 2.4516129032258064e-05, "loss": 0.2133, "step": 7900 }, { "epoch": 50.93012048192771, "grad_norm": 4.165627479553223, "learning_rate": 2.4512903225806453e-05, "loss": 0.3323, "step": 7901 }, { "epoch": 50.93654618473896, "grad_norm": 64.24018859863281, "learning_rate": 2.450967741935484e-05, "loss": 0.3492, "step": 7902 }, { "epoch": 50.9429718875502, "grad_norm": 23.38990020751953, "learning_rate": 2.4506451612903228e-05, "loss": 0.4554, "step": 7903 }, { "epoch": 50.94939759036144, "grad_norm": 3.1118757724761963, "learning_rate": 2.4503225806451614e-05, "loss": 0.2911, "step": 7904 }, { "epoch": 50.95582329317269, "grad_norm": 15.509034156799316, "learning_rate": 2.45e-05, "loss": 0.5191, "step": 7905 }, { "epoch": 50.962248995983934, "grad_norm": 27.183837890625, "learning_rate": 2.4496774193548388e-05, "loss": 0.4929, "step": 7906 }, { "epoch": 50.96867469879518, "grad_norm": 68.3900375366211, "learning_rate": 2.4493548387096777e-05, "loss": 0.379, "step": 7907 }, { "epoch": 50.975100401606426, "grad_norm": 9.06173324584961, "learning_rate": 2.449032258064516e-05, "loss": 0.4785, "step": 7908 }, { "epoch": 50.98152610441767, "grad_norm": 8.531025886535645, "learning_rate": 2.448709677419355e-05, "loss": 0.1798, "step": 7909 }, { "epoch": 50.98795180722892, "grad_norm": 16.389379501342773, "learning_rate": 2.4483870967741938e-05, "loss": 0.2436, "step": 7910 }, { "epoch": 50.99437751004016, "grad_norm": 32.17607879638672, "learning_rate": 2.4480645161290324e-05, "loss": 0.4466, "step": 7911 }, { "epoch": 51.00401606425703, "grad_norm": 8.790007591247559, "learning_rate": 2.447741935483871e-05, "loss": 0.2472, "step": 7912 }, { "epoch": 51.01044176706827, "grad_norm": 27.13780975341797, "learning_rate": 2.44741935483871e-05, "loss": 0.3552, "step": 7913 }, { "epoch": 51.01686746987952, "grad_norm": 7.7093071937561035, "learning_rate": 2.4470967741935484e-05, "loss": 0.2197, "step": 7914 }, { "epoch": 51.023293172690764, "grad_norm": 7.645054817199707, "learning_rate": 2.4467741935483873e-05, "loss": 0.6927, "step": 7915 }, { "epoch": 51.029718875502006, "grad_norm": 9.283169746398926, "learning_rate": 2.446451612903226e-05, "loss": 0.2558, "step": 7916 }, { "epoch": 51.036144578313255, "grad_norm": 11.194573402404785, "learning_rate": 2.4461290322580645e-05, "loss": 0.2744, "step": 7917 }, { "epoch": 51.0425702811245, "grad_norm": 7.675858974456787, "learning_rate": 2.4458064516129034e-05, "loss": 0.2294, "step": 7918 }, { "epoch": 51.04899598393574, "grad_norm": 5.739564895629883, "learning_rate": 2.4454838709677423e-05, "loss": 0.1607, "step": 7919 }, { "epoch": 51.05542168674699, "grad_norm": 8.649431228637695, "learning_rate": 2.4451612903225805e-05, "loss": 0.277, "step": 7920 }, { "epoch": 51.06184738955823, "grad_norm": 26.89166831970215, "learning_rate": 2.4448387096774194e-05, "loss": 0.3421, "step": 7921 }, { "epoch": 51.06827309236948, "grad_norm": 13.743285179138184, "learning_rate": 2.4445161290322583e-05, "loss": 0.3571, "step": 7922 }, { "epoch": 51.07469879518072, "grad_norm": 12.528355598449707, "learning_rate": 2.444193548387097e-05, "loss": 0.397, "step": 7923 }, { "epoch": 51.081124497991965, "grad_norm": 14.641589164733887, "learning_rate": 2.4438709677419355e-05, "loss": 0.4014, "step": 7924 }, { "epoch": 51.087550200803214, "grad_norm": 20.590463638305664, "learning_rate": 2.4435483870967744e-05, "loss": 0.3751, "step": 7925 }, { "epoch": 51.09397590361446, "grad_norm": 24.94268035888672, "learning_rate": 2.443225806451613e-05, "loss": 0.3843, "step": 7926 }, { "epoch": 51.100401606425706, "grad_norm": 38.1427001953125, "learning_rate": 2.442903225806452e-05, "loss": 0.2542, "step": 7927 }, { "epoch": 51.10682730923695, "grad_norm": 5.416262149810791, "learning_rate": 2.4425806451612904e-05, "loss": 0.1268, "step": 7928 }, { "epoch": 51.11325301204819, "grad_norm": 44.850791931152344, "learning_rate": 2.442258064516129e-05, "loss": 0.2601, "step": 7929 }, { "epoch": 51.11967871485944, "grad_norm": 3.8422813415527344, "learning_rate": 2.441935483870968e-05, "loss": 0.1817, "step": 7930 }, { "epoch": 51.12610441767068, "grad_norm": 11.233837127685547, "learning_rate": 2.4416129032258065e-05, "loss": 0.3408, "step": 7931 }, { "epoch": 51.13253012048193, "grad_norm": 21.68353843688965, "learning_rate": 2.441290322580645e-05, "loss": 0.5813, "step": 7932 }, { "epoch": 51.13895582329317, "grad_norm": 11.82320785522461, "learning_rate": 2.440967741935484e-05, "loss": 0.2681, "step": 7933 }, { "epoch": 51.145381526104416, "grad_norm": 14.328413009643555, "learning_rate": 2.440645161290323e-05, "loss": 0.3976, "step": 7934 }, { "epoch": 51.151807228915665, "grad_norm": 16.711706161499023, "learning_rate": 2.4403225806451614e-05, "loss": 0.3528, "step": 7935 }, { "epoch": 51.15823293172691, "grad_norm": 21.782272338867188, "learning_rate": 2.44e-05, "loss": 0.4702, "step": 7936 }, { "epoch": 51.16465863453815, "grad_norm": 5.621126651763916, "learning_rate": 2.439677419354839e-05, "loss": 0.262, "step": 7937 }, { "epoch": 51.1710843373494, "grad_norm": 3.846756935119629, "learning_rate": 2.4393548387096775e-05, "loss": 0.3866, "step": 7938 }, { "epoch": 51.17751004016064, "grad_norm": 11.22616195678711, "learning_rate": 2.439032258064516e-05, "loss": 0.2806, "step": 7939 }, { "epoch": 51.18393574297189, "grad_norm": 20.1240234375, "learning_rate": 2.438709677419355e-05, "loss": 0.5411, "step": 7940 }, { "epoch": 51.19036144578313, "grad_norm": 47.26263427734375, "learning_rate": 2.4383870967741935e-05, "loss": 0.4764, "step": 7941 }, { "epoch": 51.196787148594375, "grad_norm": 2.1134257316589355, "learning_rate": 2.4380645161290325e-05, "loss": 0.1561, "step": 7942 }, { "epoch": 51.203212851405624, "grad_norm": 25.763051986694336, "learning_rate": 2.437741935483871e-05, "loss": 0.4546, "step": 7943 }, { "epoch": 51.209638554216866, "grad_norm": 9.428690910339355, "learning_rate": 2.4374193548387096e-05, "loss": 0.2127, "step": 7944 }, { "epoch": 51.216064257028115, "grad_norm": 11.282415390014648, "learning_rate": 2.4370967741935485e-05, "loss": 0.4514, "step": 7945 }, { "epoch": 51.22248995983936, "grad_norm": 16.36469268798828, "learning_rate": 2.4367741935483874e-05, "loss": 0.3658, "step": 7946 }, { "epoch": 51.2289156626506, "grad_norm": 3.9918582439422607, "learning_rate": 2.436451612903226e-05, "loss": 0.2602, "step": 7947 }, { "epoch": 51.23534136546185, "grad_norm": 14.90851879119873, "learning_rate": 2.4361290322580646e-05, "loss": 0.3255, "step": 7948 }, { "epoch": 51.24176706827309, "grad_norm": 40.65071105957031, "learning_rate": 2.4358064516129035e-05, "loss": 0.4962, "step": 7949 }, { "epoch": 51.24819277108434, "grad_norm": 21.313310623168945, "learning_rate": 2.435483870967742e-05, "loss": 0.4965, "step": 7950 }, { "epoch": 51.25461847389558, "grad_norm": 8.620077133178711, "learning_rate": 2.4351612903225806e-05, "loss": 0.3243, "step": 7951 }, { "epoch": 51.261044176706825, "grad_norm": 6.609145164489746, "learning_rate": 2.4348387096774195e-05, "loss": 0.4284, "step": 7952 }, { "epoch": 51.267469879518075, "grad_norm": 11.732816696166992, "learning_rate": 2.434516129032258e-05, "loss": 0.3809, "step": 7953 }, { "epoch": 51.27389558232932, "grad_norm": 89.32655334472656, "learning_rate": 2.434193548387097e-05, "loss": 0.7073, "step": 7954 }, { "epoch": 51.28032128514056, "grad_norm": 18.160627365112305, "learning_rate": 2.4338709677419356e-05, "loss": 0.3491, "step": 7955 }, { "epoch": 51.28674698795181, "grad_norm": 5.830833435058594, "learning_rate": 2.433548387096774e-05, "loss": 0.2027, "step": 7956 }, { "epoch": 51.29317269076305, "grad_norm": 11.250349044799805, "learning_rate": 2.433225806451613e-05, "loss": 0.6381, "step": 7957 }, { "epoch": 51.2995983935743, "grad_norm": 25.50080108642578, "learning_rate": 2.432903225806452e-05, "loss": 0.4575, "step": 7958 }, { "epoch": 51.30602409638554, "grad_norm": 17.619102478027344, "learning_rate": 2.4325806451612905e-05, "loss": 0.5365, "step": 7959 }, { "epoch": 51.312449799196784, "grad_norm": 26.801677703857422, "learning_rate": 2.432258064516129e-05, "loss": 0.5289, "step": 7960 }, { "epoch": 51.318875502008034, "grad_norm": 22.622299194335938, "learning_rate": 2.431935483870968e-05, "loss": 0.417, "step": 7961 }, { "epoch": 51.325301204819276, "grad_norm": 17.942440032958984, "learning_rate": 2.4316129032258066e-05, "loss": 0.4314, "step": 7962 }, { "epoch": 51.331726907630525, "grad_norm": 43.323204040527344, "learning_rate": 2.431290322580645e-05, "loss": 0.4967, "step": 7963 }, { "epoch": 51.33815261044177, "grad_norm": 2.964613199234009, "learning_rate": 2.430967741935484e-05, "loss": 0.335, "step": 7964 }, { "epoch": 51.34457831325301, "grad_norm": 16.270296096801758, "learning_rate": 2.4306451612903226e-05, "loss": 0.397, "step": 7965 }, { "epoch": 51.35100401606426, "grad_norm": 17.898948669433594, "learning_rate": 2.4303225806451612e-05, "loss": 0.2519, "step": 7966 }, { "epoch": 51.3574297188755, "grad_norm": 9.048871994018555, "learning_rate": 2.43e-05, "loss": 0.3196, "step": 7967 }, { "epoch": 51.36385542168675, "grad_norm": 9.635146141052246, "learning_rate": 2.4296774193548387e-05, "loss": 0.4209, "step": 7968 }, { "epoch": 51.37028112449799, "grad_norm": 10.960365295410156, "learning_rate": 2.4293548387096776e-05, "loss": 0.4258, "step": 7969 }, { "epoch": 51.376706827309235, "grad_norm": 11.035155296325684, "learning_rate": 2.429032258064516e-05, "loss": 0.5778, "step": 7970 }, { "epoch": 51.383132530120484, "grad_norm": 28.865737915039062, "learning_rate": 2.4287096774193547e-05, "loss": 0.4883, "step": 7971 }, { "epoch": 51.389558232931726, "grad_norm": 8.050721168518066, "learning_rate": 2.4283870967741936e-05, "loss": 0.3393, "step": 7972 }, { "epoch": 51.39598393574297, "grad_norm": 7.149768352508545, "learning_rate": 2.4280645161290326e-05, "loss": 0.2182, "step": 7973 }, { "epoch": 51.40240963855422, "grad_norm": 13.811445236206055, "learning_rate": 2.427741935483871e-05, "loss": 0.228, "step": 7974 }, { "epoch": 51.40883534136546, "grad_norm": 7.882261753082275, "learning_rate": 2.4274193548387097e-05, "loss": 0.2326, "step": 7975 }, { "epoch": 51.41526104417671, "grad_norm": 10.02605152130127, "learning_rate": 2.4270967741935486e-05, "loss": 0.4318, "step": 7976 }, { "epoch": 51.42168674698795, "grad_norm": 19.623472213745117, "learning_rate": 2.4267741935483872e-05, "loss": 0.423, "step": 7977 }, { "epoch": 51.428112449799194, "grad_norm": 20.10013198852539, "learning_rate": 2.4264516129032257e-05, "loss": 0.6021, "step": 7978 }, { "epoch": 51.43453815261044, "grad_norm": 11.140539169311523, "learning_rate": 2.4261290322580647e-05, "loss": 0.3807, "step": 7979 }, { "epoch": 51.440963855421685, "grad_norm": 3.387063503265381, "learning_rate": 2.4258064516129032e-05, "loss": 0.3098, "step": 7980 }, { "epoch": 51.447389558232935, "grad_norm": 17.409130096435547, "learning_rate": 2.425483870967742e-05, "loss": 0.2543, "step": 7981 }, { "epoch": 51.45381526104418, "grad_norm": 12.743817329406738, "learning_rate": 2.4251612903225807e-05, "loss": 0.3844, "step": 7982 }, { "epoch": 51.46024096385542, "grad_norm": 10.746816635131836, "learning_rate": 2.4248387096774193e-05, "loss": 0.3955, "step": 7983 }, { "epoch": 51.46666666666667, "grad_norm": 11.387640953063965, "learning_rate": 2.4245161290322582e-05, "loss": 0.3994, "step": 7984 }, { "epoch": 51.47309236947791, "grad_norm": 8.937162399291992, "learning_rate": 2.424193548387097e-05, "loss": 0.0921, "step": 7985 }, { "epoch": 51.47951807228916, "grad_norm": 4.577577114105225, "learning_rate": 2.4238709677419357e-05, "loss": 0.2755, "step": 7986 }, { "epoch": 51.4859437751004, "grad_norm": 8.885736465454102, "learning_rate": 2.4235483870967742e-05, "loss": 0.2764, "step": 7987 }, { "epoch": 51.492369477911645, "grad_norm": 30.68227195739746, "learning_rate": 2.423225806451613e-05, "loss": 0.4723, "step": 7988 }, { "epoch": 51.498795180722894, "grad_norm": 12.687495231628418, "learning_rate": 2.4229032258064517e-05, "loss": 0.5045, "step": 7989 }, { "epoch": 51.505220883534136, "grad_norm": 13.093008041381836, "learning_rate": 2.4225806451612903e-05, "loss": 0.713, "step": 7990 }, { "epoch": 51.51164658634538, "grad_norm": 15.059205055236816, "learning_rate": 2.4222580645161292e-05, "loss": 0.2448, "step": 7991 }, { "epoch": 51.51807228915663, "grad_norm": 14.680448532104492, "learning_rate": 2.4219354838709678e-05, "loss": 0.3987, "step": 7992 }, { "epoch": 51.52449799196787, "grad_norm": 10.157474517822266, "learning_rate": 2.4216129032258067e-05, "loss": 0.4368, "step": 7993 }, { "epoch": 51.53092369477912, "grad_norm": 12.830140113830566, "learning_rate": 2.4212903225806452e-05, "loss": 0.3151, "step": 7994 }, { "epoch": 51.53734939759036, "grad_norm": 16.276796340942383, "learning_rate": 2.4209677419354838e-05, "loss": 0.3313, "step": 7995 }, { "epoch": 51.543775100401604, "grad_norm": 5.43107795715332, "learning_rate": 2.4206451612903227e-05, "loss": 0.2543, "step": 7996 }, { "epoch": 51.55020080321285, "grad_norm": 18.192853927612305, "learning_rate": 2.4203225806451616e-05, "loss": 0.1367, "step": 7997 }, { "epoch": 51.556626506024095, "grad_norm": 17.853139877319336, "learning_rate": 2.4200000000000002e-05, "loss": 0.3585, "step": 7998 }, { "epoch": 51.563052208835344, "grad_norm": 6.708072662353516, "learning_rate": 2.4196774193548388e-05, "loss": 0.236, "step": 7999 }, { "epoch": 51.56947791164659, "grad_norm": 9.600107192993164, "learning_rate": 2.4193548387096777e-05, "loss": 0.2428, "step": 8000 }, { "epoch": 51.57590361445783, "grad_norm": 10.898974418640137, "learning_rate": 2.4190322580645163e-05, "loss": 0.3755, "step": 8001 }, { "epoch": 51.58232931726908, "grad_norm": 31.758817672729492, "learning_rate": 2.4187096774193548e-05, "loss": 0.3739, "step": 8002 }, { "epoch": 51.58875502008032, "grad_norm": 25.48935890197754, "learning_rate": 2.4183870967741937e-05, "loss": 0.3121, "step": 8003 }, { "epoch": 51.59518072289157, "grad_norm": 17.02418327331543, "learning_rate": 2.4180645161290323e-05, "loss": 0.5697, "step": 8004 }, { "epoch": 51.60160642570281, "grad_norm": 32.68305969238281, "learning_rate": 2.417741935483871e-05, "loss": 0.449, "step": 8005 }, { "epoch": 51.608032128514054, "grad_norm": 10.676518440246582, "learning_rate": 2.4174193548387098e-05, "loss": 0.5167, "step": 8006 }, { "epoch": 51.6144578313253, "grad_norm": 5.8127241134643555, "learning_rate": 2.4170967741935484e-05, "loss": 0.2557, "step": 8007 }, { "epoch": 51.620883534136546, "grad_norm": 17.662410736083984, "learning_rate": 2.4167741935483873e-05, "loss": 0.6294, "step": 8008 }, { "epoch": 51.62730923694779, "grad_norm": 21.423376083374023, "learning_rate": 2.416451612903226e-05, "loss": 0.5446, "step": 8009 }, { "epoch": 51.63373493975904, "grad_norm": 56.42667007446289, "learning_rate": 2.4161290322580648e-05, "loss": 0.6137, "step": 8010 }, { "epoch": 51.64016064257028, "grad_norm": 25.29070281982422, "learning_rate": 2.4158064516129033e-05, "loss": 0.5964, "step": 8011 }, { "epoch": 51.64658634538153, "grad_norm": 13.428545951843262, "learning_rate": 2.4154838709677422e-05, "loss": 0.5252, "step": 8012 }, { "epoch": 51.65301204819277, "grad_norm": 11.24696159362793, "learning_rate": 2.4151612903225808e-05, "loss": 0.1648, "step": 8013 }, { "epoch": 51.65943775100401, "grad_norm": 24.121034622192383, "learning_rate": 2.4148387096774194e-05, "loss": 0.3653, "step": 8014 }, { "epoch": 51.66586345381526, "grad_norm": 17.003238677978516, "learning_rate": 2.4145161290322583e-05, "loss": 0.5045, "step": 8015 }, { "epoch": 51.672289156626505, "grad_norm": 15.34350872039795, "learning_rate": 2.414193548387097e-05, "loss": 0.2991, "step": 8016 }, { "epoch": 51.678714859437754, "grad_norm": 38.94831848144531, "learning_rate": 2.4138709677419354e-05, "loss": 0.4777, "step": 8017 }, { "epoch": 51.685140562248996, "grad_norm": 6.121136665344238, "learning_rate": 2.4135483870967743e-05, "loss": 0.3682, "step": 8018 }, { "epoch": 51.69156626506024, "grad_norm": 18.59524917602539, "learning_rate": 2.413225806451613e-05, "loss": 0.3007, "step": 8019 }, { "epoch": 51.69799196787149, "grad_norm": 12.583378791809082, "learning_rate": 2.4129032258064518e-05, "loss": 0.3911, "step": 8020 }, { "epoch": 51.70441767068273, "grad_norm": 8.083178520202637, "learning_rate": 2.4125806451612904e-05, "loss": 0.3157, "step": 8021 }, { "epoch": 51.71084337349397, "grad_norm": 13.245803833007812, "learning_rate": 2.4122580645161293e-05, "loss": 0.156, "step": 8022 }, { "epoch": 51.71726907630522, "grad_norm": 4.684439659118652, "learning_rate": 2.411935483870968e-05, "loss": 0.1501, "step": 8023 }, { "epoch": 51.723694779116464, "grad_norm": 19.760347366333008, "learning_rate": 2.4116129032258068e-05, "loss": 0.2818, "step": 8024 }, { "epoch": 51.73012048192771, "grad_norm": 14.422463417053223, "learning_rate": 2.4112903225806453e-05, "loss": 0.3241, "step": 8025 }, { "epoch": 51.736546184738955, "grad_norm": 38.69111633300781, "learning_rate": 2.410967741935484e-05, "loss": 0.2581, "step": 8026 }, { "epoch": 51.7429718875502, "grad_norm": 27.08409881591797, "learning_rate": 2.4106451612903228e-05, "loss": 0.2455, "step": 8027 }, { "epoch": 51.74939759036145, "grad_norm": 14.63601016998291, "learning_rate": 2.4103225806451614e-05, "loss": 0.5275, "step": 8028 }, { "epoch": 51.75582329317269, "grad_norm": 1.7915555238723755, "learning_rate": 2.41e-05, "loss": 0.2632, "step": 8029 }, { "epoch": 51.76224899598394, "grad_norm": 24.30817222595215, "learning_rate": 2.409677419354839e-05, "loss": 0.489, "step": 8030 }, { "epoch": 51.76867469879518, "grad_norm": 6.5229973793029785, "learning_rate": 2.4093548387096774e-05, "loss": 0.32, "step": 8031 }, { "epoch": 51.77510040160642, "grad_norm": 14.194367408752441, "learning_rate": 2.4090322580645164e-05, "loss": 0.5742, "step": 8032 }, { "epoch": 51.78152610441767, "grad_norm": 8.758734703063965, "learning_rate": 2.408709677419355e-05, "loss": 0.3111, "step": 8033 }, { "epoch": 51.787951807228914, "grad_norm": 26.297895431518555, "learning_rate": 2.4083870967741935e-05, "loss": 0.4666, "step": 8034 }, { "epoch": 51.794377510040164, "grad_norm": 25.84332847595215, "learning_rate": 2.4080645161290324e-05, "loss": 0.4972, "step": 8035 }, { "epoch": 51.800803212851406, "grad_norm": 3.611940622329712, "learning_rate": 2.407741935483871e-05, "loss": 0.2925, "step": 8036 }, { "epoch": 51.80722891566265, "grad_norm": 15.861573219299316, "learning_rate": 2.40741935483871e-05, "loss": 0.602, "step": 8037 }, { "epoch": 51.8136546184739, "grad_norm": 7.611725330352783, "learning_rate": 2.4070967741935485e-05, "loss": 0.1726, "step": 8038 }, { "epoch": 51.82008032128514, "grad_norm": 8.403093338012695, "learning_rate": 2.4067741935483874e-05, "loss": 0.3764, "step": 8039 }, { "epoch": 51.82650602409639, "grad_norm": 35.663360595703125, "learning_rate": 2.406451612903226e-05, "loss": 0.2569, "step": 8040 }, { "epoch": 51.83293172690763, "grad_norm": 5.697908878326416, "learning_rate": 2.4061290322580645e-05, "loss": 0.2325, "step": 8041 }, { "epoch": 51.83935742971887, "grad_norm": 12.787976264953613, "learning_rate": 2.4058064516129034e-05, "loss": 0.4676, "step": 8042 }, { "epoch": 51.84578313253012, "grad_norm": 2.3859143257141113, "learning_rate": 2.405483870967742e-05, "loss": 0.1576, "step": 8043 }, { "epoch": 51.852208835341365, "grad_norm": 10.443259239196777, "learning_rate": 2.4051612903225806e-05, "loss": 0.2775, "step": 8044 }, { "epoch": 51.85863453815261, "grad_norm": 18.091411590576172, "learning_rate": 2.4048387096774195e-05, "loss": 0.4728, "step": 8045 }, { "epoch": 51.86506024096386, "grad_norm": 20.59971809387207, "learning_rate": 2.404516129032258e-05, "loss": 0.6636, "step": 8046 }, { "epoch": 51.8714859437751, "grad_norm": 2.902738571166992, "learning_rate": 2.404193548387097e-05, "loss": 0.2027, "step": 8047 }, { "epoch": 51.87791164658635, "grad_norm": 25.35239601135254, "learning_rate": 2.4038709677419355e-05, "loss": 0.2789, "step": 8048 }, { "epoch": 51.88433734939759, "grad_norm": 3.8270299434661865, "learning_rate": 2.4035483870967744e-05, "loss": 0.1628, "step": 8049 }, { "epoch": 51.89076305220883, "grad_norm": 1.6309503316879272, "learning_rate": 2.403225806451613e-05, "loss": 0.1599, "step": 8050 }, { "epoch": 51.89718875502008, "grad_norm": 23.256610870361328, "learning_rate": 2.402903225806452e-05, "loss": 0.2995, "step": 8051 }, { "epoch": 51.903614457831324, "grad_norm": 19.56418228149414, "learning_rate": 2.4025806451612905e-05, "loss": 0.3426, "step": 8052 }, { "epoch": 51.91004016064257, "grad_norm": 8.846311569213867, "learning_rate": 2.402258064516129e-05, "loss": 0.2643, "step": 8053 }, { "epoch": 51.916465863453816, "grad_norm": 2.6177990436553955, "learning_rate": 2.401935483870968e-05, "loss": 0.1164, "step": 8054 }, { "epoch": 51.92289156626506, "grad_norm": 11.830642700195312, "learning_rate": 2.4016129032258065e-05, "loss": 0.4337, "step": 8055 }, { "epoch": 51.92931726907631, "grad_norm": 4.561389923095703, "learning_rate": 2.401290322580645e-05, "loss": 0.283, "step": 8056 }, { "epoch": 51.93574297188755, "grad_norm": 5.829712867736816, "learning_rate": 2.400967741935484e-05, "loss": 0.5333, "step": 8057 }, { "epoch": 51.94216867469879, "grad_norm": 38.50145721435547, "learning_rate": 2.4006451612903226e-05, "loss": 0.2867, "step": 8058 }, { "epoch": 51.94859437751004, "grad_norm": 9.366156578063965, "learning_rate": 2.4003225806451615e-05, "loss": 0.3655, "step": 8059 }, { "epoch": 51.95502008032128, "grad_norm": 7.1955718994140625, "learning_rate": 2.4e-05, "loss": 0.6113, "step": 8060 }, { "epoch": 51.96144578313253, "grad_norm": 9.481236457824707, "learning_rate": 2.399677419354839e-05, "loss": 0.2836, "step": 8061 }, { "epoch": 51.967871485943775, "grad_norm": 15.627847671508789, "learning_rate": 2.3993548387096775e-05, "loss": 0.4015, "step": 8062 }, { "epoch": 51.97429718875502, "grad_norm": 22.57843780517578, "learning_rate": 2.3990322580645165e-05, "loss": 0.501, "step": 8063 }, { "epoch": 51.980722891566266, "grad_norm": 13.117259979248047, "learning_rate": 2.398709677419355e-05, "loss": 0.3421, "step": 8064 }, { "epoch": 51.98714859437751, "grad_norm": 9.737043380737305, "learning_rate": 2.3983870967741936e-05, "loss": 0.4491, "step": 8065 }, { "epoch": 51.99357429718876, "grad_norm": 17.02623176574707, "learning_rate": 2.3980645161290325e-05, "loss": 0.2992, "step": 8066 }, { "epoch": 52.00321285140562, "grad_norm": 10.747393608093262, "learning_rate": 2.397741935483871e-05, "loss": 0.3807, "step": 8067 }, { "epoch": 52.00963855421687, "grad_norm": 9.585186004638672, "learning_rate": 2.3974193548387096e-05, "loss": 0.5367, "step": 8068 }, { "epoch": 52.01606425702811, "grad_norm": 18.412141799926758, "learning_rate": 2.3970967741935486e-05, "loss": 0.4417, "step": 8069 }, { "epoch": 52.022489959839355, "grad_norm": 4.656810760498047, "learning_rate": 2.396774193548387e-05, "loss": 0.3505, "step": 8070 }, { "epoch": 52.028915662650604, "grad_norm": 11.704150199890137, "learning_rate": 2.3964516129032257e-05, "loss": 0.4869, "step": 8071 }, { "epoch": 52.035341365461846, "grad_norm": 6.191903114318848, "learning_rate": 2.3961290322580646e-05, "loss": 0.2169, "step": 8072 }, { "epoch": 52.041767068273096, "grad_norm": 17.066646575927734, "learning_rate": 2.3958064516129035e-05, "loss": 0.3549, "step": 8073 }, { "epoch": 52.04819277108434, "grad_norm": 11.043472290039062, "learning_rate": 2.395483870967742e-05, "loss": 0.1994, "step": 8074 }, { "epoch": 52.05461847389558, "grad_norm": 34.02171325683594, "learning_rate": 2.3951612903225807e-05, "loss": 0.3143, "step": 8075 }, { "epoch": 52.06104417670683, "grad_norm": 13.354324340820312, "learning_rate": 2.3948387096774196e-05, "loss": 0.2753, "step": 8076 }, { "epoch": 52.06746987951807, "grad_norm": 20.37383460998535, "learning_rate": 2.394516129032258e-05, "loss": 0.4087, "step": 8077 }, { "epoch": 52.073895582329314, "grad_norm": 17.095458984375, "learning_rate": 2.394193548387097e-05, "loss": 0.6865, "step": 8078 }, { "epoch": 52.08032128514056, "grad_norm": 38.01460266113281, "learning_rate": 2.3938709677419356e-05, "loss": 0.2647, "step": 8079 }, { "epoch": 52.086746987951805, "grad_norm": 4.579261779785156, "learning_rate": 2.3935483870967742e-05, "loss": 0.2479, "step": 8080 }, { "epoch": 52.093172690763055, "grad_norm": 16.111370086669922, "learning_rate": 2.393225806451613e-05, "loss": 0.4104, "step": 8081 }, { "epoch": 52.0995983935743, "grad_norm": 3.0662641525268555, "learning_rate": 2.3929032258064517e-05, "loss": 0.3539, "step": 8082 }, { "epoch": 52.10602409638554, "grad_norm": 4.475181579589844, "learning_rate": 2.3925806451612902e-05, "loss": 0.224, "step": 8083 }, { "epoch": 52.11244979919679, "grad_norm": 18.113697052001953, "learning_rate": 2.392258064516129e-05, "loss": 0.2735, "step": 8084 }, { "epoch": 52.11887550200803, "grad_norm": 30.275104522705078, "learning_rate": 2.391935483870968e-05, "loss": 0.4484, "step": 8085 }, { "epoch": 52.12530120481928, "grad_norm": 3.248425245285034, "learning_rate": 2.3916129032258066e-05, "loss": 0.2337, "step": 8086 }, { "epoch": 52.13172690763052, "grad_norm": 8.943284034729004, "learning_rate": 2.3912903225806452e-05, "loss": 0.5197, "step": 8087 }, { "epoch": 52.138152610441765, "grad_norm": 37.08747863769531, "learning_rate": 2.390967741935484e-05, "loss": 0.4138, "step": 8088 }, { "epoch": 52.144578313253014, "grad_norm": 17.066055297851562, "learning_rate": 2.3906451612903227e-05, "loss": 0.603, "step": 8089 }, { "epoch": 52.151004016064256, "grad_norm": 24.13787841796875, "learning_rate": 2.3903225806451616e-05, "loss": 0.6181, "step": 8090 }, { "epoch": 52.157429718875505, "grad_norm": 4.416585445404053, "learning_rate": 2.39e-05, "loss": 0.2887, "step": 8091 }, { "epoch": 52.16385542168675, "grad_norm": 15.44590950012207, "learning_rate": 2.3896774193548387e-05, "loss": 0.2466, "step": 8092 }, { "epoch": 52.17028112449799, "grad_norm": 17.599407196044922, "learning_rate": 2.3893548387096776e-05, "loss": 0.3833, "step": 8093 }, { "epoch": 52.17670682730924, "grad_norm": 11.302499771118164, "learning_rate": 2.3890322580645162e-05, "loss": 0.3959, "step": 8094 }, { "epoch": 52.18313253012048, "grad_norm": 35.227359771728516, "learning_rate": 2.3887096774193548e-05, "loss": 0.2273, "step": 8095 }, { "epoch": 52.189558232931724, "grad_norm": 11.181041717529297, "learning_rate": 2.3883870967741937e-05, "loss": 0.2693, "step": 8096 }, { "epoch": 52.19598393574297, "grad_norm": 10.890218734741211, "learning_rate": 2.3880645161290323e-05, "loss": 0.3091, "step": 8097 }, { "epoch": 52.202409638554215, "grad_norm": 12.897887229919434, "learning_rate": 2.387741935483871e-05, "loss": 0.5035, "step": 8098 }, { "epoch": 52.208835341365464, "grad_norm": 14.554824829101562, "learning_rate": 2.3874193548387097e-05, "loss": 0.4345, "step": 8099 }, { "epoch": 52.21526104417671, "grad_norm": 12.428888320922852, "learning_rate": 2.3870967741935486e-05, "loss": 0.3184, "step": 8100 }, { "epoch": 52.22168674698795, "grad_norm": 4.598308563232422, "learning_rate": 2.3867741935483872e-05, "loss": 0.3414, "step": 8101 }, { "epoch": 52.2281124497992, "grad_norm": 22.748010635375977, "learning_rate": 2.386451612903226e-05, "loss": 0.3142, "step": 8102 }, { "epoch": 52.23453815261044, "grad_norm": 6.288761138916016, "learning_rate": 2.3861290322580647e-05, "loss": 0.2342, "step": 8103 }, { "epoch": 52.24096385542169, "grad_norm": 9.051536560058594, "learning_rate": 2.3858064516129033e-05, "loss": 0.3297, "step": 8104 }, { "epoch": 52.24738955823293, "grad_norm": 16.70203971862793, "learning_rate": 2.3854838709677422e-05, "loss": 0.3269, "step": 8105 }, { "epoch": 52.253815261044174, "grad_norm": 7.593077182769775, "learning_rate": 2.3851612903225807e-05, "loss": 0.5274, "step": 8106 }, { "epoch": 52.26024096385542, "grad_norm": 24.681657791137695, "learning_rate": 2.3848387096774193e-05, "loss": 0.566, "step": 8107 }, { "epoch": 52.266666666666666, "grad_norm": 7.1664958000183105, "learning_rate": 2.3845161290322582e-05, "loss": 0.2286, "step": 8108 }, { "epoch": 52.273092369477915, "grad_norm": 8.753214836120605, "learning_rate": 2.3841935483870968e-05, "loss": 0.4958, "step": 8109 }, { "epoch": 52.27951807228916, "grad_norm": 8.255121231079102, "learning_rate": 2.3838709677419354e-05, "loss": 0.3493, "step": 8110 }, { "epoch": 52.2859437751004, "grad_norm": 3.498838186264038, "learning_rate": 2.3835483870967743e-05, "loss": 0.1284, "step": 8111 }, { "epoch": 52.29236947791165, "grad_norm": 17.349441528320312, "learning_rate": 2.3832258064516132e-05, "loss": 0.4293, "step": 8112 }, { "epoch": 52.29879518072289, "grad_norm": 9.098011016845703, "learning_rate": 2.3829032258064518e-05, "loss": 0.5564, "step": 8113 }, { "epoch": 52.30522088353413, "grad_norm": 7.172802925109863, "learning_rate": 2.3825806451612903e-05, "loss": 0.1803, "step": 8114 }, { "epoch": 52.31164658634538, "grad_norm": 31.557104110717773, "learning_rate": 2.3822580645161292e-05, "loss": 0.3436, "step": 8115 }, { "epoch": 52.318072289156625, "grad_norm": 13.723363876342773, "learning_rate": 2.3819354838709678e-05, "loss": 0.2265, "step": 8116 }, { "epoch": 52.324497991967874, "grad_norm": 31.19764518737793, "learning_rate": 2.3816129032258067e-05, "loss": 0.7183, "step": 8117 }, { "epoch": 52.330923694779116, "grad_norm": 32.9656982421875, "learning_rate": 2.3812903225806453e-05, "loss": 0.5815, "step": 8118 }, { "epoch": 52.33734939759036, "grad_norm": 11.222369194030762, "learning_rate": 2.380967741935484e-05, "loss": 0.419, "step": 8119 }, { "epoch": 52.34377510040161, "grad_norm": 12.610513687133789, "learning_rate": 2.3806451612903228e-05, "loss": 0.2826, "step": 8120 }, { "epoch": 52.35020080321285, "grad_norm": 24.402748107910156, "learning_rate": 2.3803225806451613e-05, "loss": 0.2345, "step": 8121 }, { "epoch": 52.3566265060241, "grad_norm": 10.545089721679688, "learning_rate": 2.38e-05, "loss": 0.379, "step": 8122 }, { "epoch": 52.36305220883534, "grad_norm": 18.98798179626465, "learning_rate": 2.3796774193548388e-05, "loss": 0.4619, "step": 8123 }, { "epoch": 52.369477911646584, "grad_norm": 12.2634916305542, "learning_rate": 2.3793548387096777e-05, "loss": 0.2213, "step": 8124 }, { "epoch": 52.37590361445783, "grad_norm": 6.6680908203125, "learning_rate": 2.3790322580645163e-05, "loss": 0.2736, "step": 8125 }, { "epoch": 52.382329317269075, "grad_norm": 8.572311401367188, "learning_rate": 2.378709677419355e-05, "loss": 0.3748, "step": 8126 }, { "epoch": 52.388755020080325, "grad_norm": 5.896097660064697, "learning_rate": 2.3783870967741938e-05, "loss": 0.3627, "step": 8127 }, { "epoch": 52.39518072289157, "grad_norm": 25.102691650390625, "learning_rate": 2.3780645161290324e-05, "loss": 0.169, "step": 8128 }, { "epoch": 52.40160642570281, "grad_norm": 15.214530944824219, "learning_rate": 2.3777419354838713e-05, "loss": 0.385, "step": 8129 }, { "epoch": 52.40803212851406, "grad_norm": 33.84724807739258, "learning_rate": 2.37741935483871e-05, "loss": 0.3434, "step": 8130 }, { "epoch": 52.4144578313253, "grad_norm": 19.71137046813965, "learning_rate": 2.3770967741935484e-05, "loss": 0.4918, "step": 8131 }, { "epoch": 52.42088353413654, "grad_norm": 7.752753734588623, "learning_rate": 2.3767741935483873e-05, "loss": 0.412, "step": 8132 }, { "epoch": 52.42730923694779, "grad_norm": 8.716812133789062, "learning_rate": 2.376451612903226e-05, "loss": 0.4591, "step": 8133 }, { "epoch": 52.433734939759034, "grad_norm": 10.061591148376465, "learning_rate": 2.3761290322580645e-05, "loss": 0.52, "step": 8134 }, { "epoch": 52.440160642570284, "grad_norm": 7.996492385864258, "learning_rate": 2.3758064516129034e-05, "loss": 0.4106, "step": 8135 }, { "epoch": 52.446586345381526, "grad_norm": 12.492325782775879, "learning_rate": 2.3754838709677423e-05, "loss": 0.5658, "step": 8136 }, { "epoch": 52.45301204819277, "grad_norm": 5.3040995597839355, "learning_rate": 2.375161290322581e-05, "loss": 0.2628, "step": 8137 }, { "epoch": 52.45943775100402, "grad_norm": 3.332998752593994, "learning_rate": 2.3748387096774194e-05, "loss": 0.3883, "step": 8138 }, { "epoch": 52.46586345381526, "grad_norm": 18.356298446655273, "learning_rate": 2.3745161290322583e-05, "loss": 0.3152, "step": 8139 }, { "epoch": 52.47228915662651, "grad_norm": 30.78119468688965, "learning_rate": 2.374193548387097e-05, "loss": 0.6408, "step": 8140 }, { "epoch": 52.47871485943775, "grad_norm": 6.657002925872803, "learning_rate": 2.3738709677419358e-05, "loss": 0.3908, "step": 8141 }, { "epoch": 52.48514056224899, "grad_norm": 34.41271209716797, "learning_rate": 2.3735483870967744e-05, "loss": 0.3347, "step": 8142 }, { "epoch": 52.49156626506024, "grad_norm": 34.50281524658203, "learning_rate": 2.373225806451613e-05, "loss": 0.3983, "step": 8143 }, { "epoch": 52.497991967871485, "grad_norm": 9.197553634643555, "learning_rate": 2.372903225806452e-05, "loss": 0.3471, "step": 8144 }, { "epoch": 52.504417670682734, "grad_norm": 8.924246788024902, "learning_rate": 2.3725806451612904e-05, "loss": 0.7213, "step": 8145 }, { "epoch": 52.51084337349398, "grad_norm": 5.407959938049316, "learning_rate": 2.372258064516129e-05, "loss": 0.4396, "step": 8146 }, { "epoch": 52.51726907630522, "grad_norm": 5.791236400604248, "learning_rate": 2.371935483870968e-05, "loss": 0.4704, "step": 8147 }, { "epoch": 52.52369477911647, "grad_norm": 17.43137550354004, "learning_rate": 2.3716129032258068e-05, "loss": 0.4014, "step": 8148 }, { "epoch": 52.53012048192771, "grad_norm": 22.848552703857422, "learning_rate": 2.371290322580645e-05, "loss": 0.234, "step": 8149 }, { "epoch": 52.53654618473895, "grad_norm": 12.931133270263672, "learning_rate": 2.370967741935484e-05, "loss": 0.2031, "step": 8150 }, { "epoch": 52.5429718875502, "grad_norm": 9.372171401977539, "learning_rate": 2.370645161290323e-05, "loss": 0.3079, "step": 8151 }, { "epoch": 52.549397590361444, "grad_norm": 7.085123538970947, "learning_rate": 2.3703225806451614e-05, "loss": 0.3068, "step": 8152 }, { "epoch": 52.55582329317269, "grad_norm": 10.450413703918457, "learning_rate": 2.37e-05, "loss": 0.4625, "step": 8153 }, { "epoch": 52.562248995983936, "grad_norm": 18.879087448120117, "learning_rate": 2.369677419354839e-05, "loss": 0.5237, "step": 8154 }, { "epoch": 52.56867469879518, "grad_norm": 4.8610687255859375, "learning_rate": 2.3693548387096775e-05, "loss": 0.2769, "step": 8155 }, { "epoch": 52.57510040160643, "grad_norm": 15.067874908447266, "learning_rate": 2.3690322580645164e-05, "loss": 0.3628, "step": 8156 }, { "epoch": 52.58152610441767, "grad_norm": 23.624408721923828, "learning_rate": 2.368709677419355e-05, "loss": 0.4927, "step": 8157 }, { "epoch": 52.58795180722892, "grad_norm": 8.74249267578125, "learning_rate": 2.3683870967741935e-05, "loss": 0.2474, "step": 8158 }, { "epoch": 52.59437751004016, "grad_norm": 10.204761505126953, "learning_rate": 2.3680645161290324e-05, "loss": 0.4097, "step": 8159 }, { "epoch": 52.6008032128514, "grad_norm": 30.295042037963867, "learning_rate": 2.367741935483871e-05, "loss": 0.4261, "step": 8160 }, { "epoch": 52.60722891566265, "grad_norm": 7.265615463256836, "learning_rate": 2.3674193548387096e-05, "loss": 0.4759, "step": 8161 }, { "epoch": 52.613654618473895, "grad_norm": 12.68472957611084, "learning_rate": 2.3670967741935485e-05, "loss": 0.2506, "step": 8162 }, { "epoch": 52.620080321285144, "grad_norm": 12.311478614807129, "learning_rate": 2.3667741935483874e-05, "loss": 0.3994, "step": 8163 }, { "epoch": 52.626506024096386, "grad_norm": 14.358757972717285, "learning_rate": 2.366451612903226e-05, "loss": 0.2961, "step": 8164 }, { "epoch": 52.63293172690763, "grad_norm": 26.97091293334961, "learning_rate": 2.3661290322580646e-05, "loss": 0.4422, "step": 8165 }, { "epoch": 52.63935742971888, "grad_norm": 17.1400089263916, "learning_rate": 2.3658064516129035e-05, "loss": 0.367, "step": 8166 }, { "epoch": 52.64578313253012, "grad_norm": 16.38471794128418, "learning_rate": 2.365483870967742e-05, "loss": 0.4956, "step": 8167 }, { "epoch": 52.65220883534136, "grad_norm": 2.284325361251831, "learning_rate": 2.365161290322581e-05, "loss": 0.2515, "step": 8168 }, { "epoch": 52.65863453815261, "grad_norm": 15.250675201416016, "learning_rate": 2.3648387096774195e-05, "loss": 0.2258, "step": 8169 }, { "epoch": 52.665060240963854, "grad_norm": 30.84919548034668, "learning_rate": 2.364516129032258e-05, "loss": 0.2336, "step": 8170 }, { "epoch": 52.6714859437751, "grad_norm": 14.392685890197754, "learning_rate": 2.364193548387097e-05, "loss": 0.3536, "step": 8171 }, { "epoch": 52.677911646586345, "grad_norm": 32.1840705871582, "learning_rate": 2.3638709677419356e-05, "loss": 0.327, "step": 8172 }, { "epoch": 52.68433734939759, "grad_norm": 7.6351165771484375, "learning_rate": 2.363548387096774e-05, "loss": 0.4738, "step": 8173 }, { "epoch": 52.69076305220884, "grad_norm": 39.03428649902344, "learning_rate": 2.363225806451613e-05, "loss": 0.3609, "step": 8174 }, { "epoch": 52.69718875502008, "grad_norm": 11.571566581726074, "learning_rate": 2.362903225806452e-05, "loss": 0.2729, "step": 8175 }, { "epoch": 52.70361445783133, "grad_norm": 6.223926544189453, "learning_rate": 2.3625806451612905e-05, "loss": 0.2794, "step": 8176 }, { "epoch": 52.71004016064257, "grad_norm": 8.157527923583984, "learning_rate": 2.362258064516129e-05, "loss": 0.3023, "step": 8177 }, { "epoch": 52.71646586345381, "grad_norm": 11.346107482910156, "learning_rate": 2.361935483870968e-05, "loss": 0.329, "step": 8178 }, { "epoch": 52.72289156626506, "grad_norm": 12.262207984924316, "learning_rate": 2.3616129032258066e-05, "loss": 0.4172, "step": 8179 }, { "epoch": 52.729317269076304, "grad_norm": 8.661438941955566, "learning_rate": 2.361290322580645e-05, "loss": 0.2948, "step": 8180 }, { "epoch": 52.735742971887554, "grad_norm": 14.647418975830078, "learning_rate": 2.360967741935484e-05, "loss": 0.3699, "step": 8181 }, { "epoch": 52.742168674698796, "grad_norm": 10.23572063446045, "learning_rate": 2.3606451612903226e-05, "loss": 0.6544, "step": 8182 }, { "epoch": 52.74859437751004, "grad_norm": 11.017683982849121, "learning_rate": 2.3603225806451615e-05, "loss": 0.5544, "step": 8183 }, { "epoch": 52.75502008032129, "grad_norm": 3.084645986557007, "learning_rate": 2.36e-05, "loss": 0.2864, "step": 8184 }, { "epoch": 52.76144578313253, "grad_norm": 9.378654479980469, "learning_rate": 2.3596774193548387e-05, "loss": 0.4152, "step": 8185 }, { "epoch": 52.76787148594377, "grad_norm": 7.328895568847656, "learning_rate": 2.3593548387096776e-05, "loss": 0.298, "step": 8186 }, { "epoch": 52.77429718875502, "grad_norm": 6.148715972900391, "learning_rate": 2.3590322580645165e-05, "loss": 0.3436, "step": 8187 }, { "epoch": 52.78072289156626, "grad_norm": 12.245357513427734, "learning_rate": 2.3587096774193547e-05, "loss": 0.4581, "step": 8188 }, { "epoch": 52.78714859437751, "grad_norm": 7.7537384033203125, "learning_rate": 2.3583870967741936e-05, "loss": 0.3928, "step": 8189 }, { "epoch": 52.793574297188755, "grad_norm": 9.093396186828613, "learning_rate": 2.3580645161290325e-05, "loss": 0.3304, "step": 8190 }, { "epoch": 52.8, "grad_norm": 44.94588851928711, "learning_rate": 2.357741935483871e-05, "loss": 0.7973, "step": 8191 }, { "epoch": 52.80642570281125, "grad_norm": 5.131511688232422, "learning_rate": 2.3574193548387097e-05, "loss": 0.147, "step": 8192 }, { "epoch": 52.81285140562249, "grad_norm": 9.563542366027832, "learning_rate": 2.3570967741935486e-05, "loss": 0.2853, "step": 8193 }, { "epoch": 52.81927710843374, "grad_norm": 10.275428771972656, "learning_rate": 2.356774193548387e-05, "loss": 0.4042, "step": 8194 }, { "epoch": 52.82570281124498, "grad_norm": 28.26894187927246, "learning_rate": 2.356451612903226e-05, "loss": 0.3508, "step": 8195 }, { "epoch": 52.83212851405622, "grad_norm": 21.31001853942871, "learning_rate": 2.3561290322580646e-05, "loss": 0.5548, "step": 8196 }, { "epoch": 52.83855421686747, "grad_norm": 15.793660163879395, "learning_rate": 2.3558064516129032e-05, "loss": 0.1508, "step": 8197 }, { "epoch": 52.844979919678714, "grad_norm": 12.68259334564209, "learning_rate": 2.355483870967742e-05, "loss": 0.5297, "step": 8198 }, { "epoch": 52.85140562248996, "grad_norm": 11.19294261932373, "learning_rate": 2.355161290322581e-05, "loss": 0.4634, "step": 8199 }, { "epoch": 52.857831325301206, "grad_norm": 15.848125457763672, "learning_rate": 2.3548387096774193e-05, "loss": 0.2198, "step": 8200 }, { "epoch": 52.86425702811245, "grad_norm": 5.8256940841674805, "learning_rate": 2.3545161290322582e-05, "loss": 0.1852, "step": 8201 }, { "epoch": 52.8706827309237, "grad_norm": 2.3815712928771973, "learning_rate": 2.354193548387097e-05, "loss": 0.0918, "step": 8202 }, { "epoch": 52.87710843373494, "grad_norm": 11.205114364624023, "learning_rate": 2.3538709677419357e-05, "loss": 0.2701, "step": 8203 }, { "epoch": 52.88353413654618, "grad_norm": 5.144824981689453, "learning_rate": 2.3535483870967742e-05, "loss": 0.2058, "step": 8204 }, { "epoch": 52.88995983935743, "grad_norm": 12.035234451293945, "learning_rate": 2.353225806451613e-05, "loss": 0.2484, "step": 8205 }, { "epoch": 52.89638554216867, "grad_norm": 4.575941562652588, "learning_rate": 2.3529032258064517e-05, "loss": 0.4211, "step": 8206 }, { "epoch": 52.90281124497992, "grad_norm": 5.683244228363037, "learning_rate": 2.3525806451612906e-05, "loss": 0.1479, "step": 8207 }, { "epoch": 52.909236947791165, "grad_norm": 7.498319625854492, "learning_rate": 2.3522580645161292e-05, "loss": 0.2869, "step": 8208 }, { "epoch": 52.91566265060241, "grad_norm": 7.516854286193848, "learning_rate": 2.3519354838709678e-05, "loss": 0.2932, "step": 8209 }, { "epoch": 52.922088353413656, "grad_norm": 11.625347137451172, "learning_rate": 2.3516129032258067e-05, "loss": 0.4942, "step": 8210 }, { "epoch": 52.9285140562249, "grad_norm": 15.427197456359863, "learning_rate": 2.3512903225806456e-05, "loss": 0.2677, "step": 8211 }, { "epoch": 52.93493975903615, "grad_norm": 12.825974464416504, "learning_rate": 2.3509677419354838e-05, "loss": 0.2707, "step": 8212 }, { "epoch": 52.94136546184739, "grad_norm": 4.500626087188721, "learning_rate": 2.3506451612903227e-05, "loss": 0.1586, "step": 8213 }, { "epoch": 52.94779116465863, "grad_norm": 12.574734687805176, "learning_rate": 2.3503225806451616e-05, "loss": 0.5164, "step": 8214 }, { "epoch": 52.95421686746988, "grad_norm": 19.33399200439453, "learning_rate": 2.35e-05, "loss": 0.2725, "step": 8215 }, { "epoch": 52.960642570281124, "grad_norm": 4.709247589111328, "learning_rate": 2.3496774193548388e-05, "loss": 0.2349, "step": 8216 }, { "epoch": 52.967068273092366, "grad_norm": 20.856355667114258, "learning_rate": 2.3493548387096777e-05, "loss": 0.3951, "step": 8217 }, { "epoch": 52.973493975903615, "grad_norm": 22.589303970336914, "learning_rate": 2.3490322580645163e-05, "loss": 0.5879, "step": 8218 }, { "epoch": 52.97991967871486, "grad_norm": 25.500757217407227, "learning_rate": 2.3487096774193548e-05, "loss": 0.47, "step": 8219 }, { "epoch": 52.98634538152611, "grad_norm": 5.7067952156066895, "learning_rate": 2.3483870967741937e-05, "loss": 0.2288, "step": 8220 }, { "epoch": 52.99277108433735, "grad_norm": 6.415181636810303, "learning_rate": 2.3480645161290323e-05, "loss": 0.2996, "step": 8221 }, { "epoch": 53.00240963855422, "grad_norm": 8.701748847961426, "learning_rate": 2.3477419354838712e-05, "loss": 0.2768, "step": 8222 }, { "epoch": 53.00883534136546, "grad_norm": 21.65620231628418, "learning_rate": 2.3474193548387098e-05, "loss": 0.4861, "step": 8223 }, { "epoch": 53.015261044176704, "grad_norm": 6.926136016845703, "learning_rate": 2.3470967741935484e-05, "loss": 0.2962, "step": 8224 }, { "epoch": 53.02168674698795, "grad_norm": 21.783540725708008, "learning_rate": 2.3467741935483873e-05, "loss": 0.3511, "step": 8225 }, { "epoch": 53.028112449799195, "grad_norm": 10.001421928405762, "learning_rate": 2.3464516129032262e-05, "loss": 0.3674, "step": 8226 }, { "epoch": 53.034538152610445, "grad_norm": 8.99609661102295, "learning_rate": 2.3461290322580644e-05, "loss": 0.3033, "step": 8227 }, { "epoch": 53.04096385542169, "grad_norm": 10.718971252441406, "learning_rate": 2.3458064516129033e-05, "loss": 0.3766, "step": 8228 }, { "epoch": 53.04738955823293, "grad_norm": 11.687832832336426, "learning_rate": 2.3454838709677422e-05, "loss": 0.2885, "step": 8229 }, { "epoch": 53.05381526104418, "grad_norm": 6.156092166900635, "learning_rate": 2.3451612903225808e-05, "loss": 0.1934, "step": 8230 }, { "epoch": 53.06024096385542, "grad_norm": 9.334355354309082, "learning_rate": 2.3448387096774194e-05, "loss": 0.3456, "step": 8231 }, { "epoch": 53.06666666666667, "grad_norm": 15.367499351501465, "learning_rate": 2.3445161290322583e-05, "loss": 0.4402, "step": 8232 }, { "epoch": 53.07309236947791, "grad_norm": 5.297476291656494, "learning_rate": 2.344193548387097e-05, "loss": 0.2478, "step": 8233 }, { "epoch": 53.079518072289154, "grad_norm": 15.547532081604004, "learning_rate": 2.3438709677419358e-05, "loss": 0.4224, "step": 8234 }, { "epoch": 53.085943775100404, "grad_norm": 4.526370525360107, "learning_rate": 2.3435483870967743e-05, "loss": 0.2261, "step": 8235 }, { "epoch": 53.092369477911646, "grad_norm": 9.905190467834473, "learning_rate": 2.343225806451613e-05, "loss": 0.4036, "step": 8236 }, { "epoch": 53.09879518072289, "grad_norm": 11.190105438232422, "learning_rate": 2.3429032258064518e-05, "loss": 0.401, "step": 8237 }, { "epoch": 53.10522088353414, "grad_norm": 9.352090835571289, "learning_rate": 2.3425806451612907e-05, "loss": 0.2911, "step": 8238 }, { "epoch": 53.11164658634538, "grad_norm": 13.679755210876465, "learning_rate": 2.342258064516129e-05, "loss": 0.3681, "step": 8239 }, { "epoch": 53.11807228915663, "grad_norm": 6.102626323699951, "learning_rate": 2.341935483870968e-05, "loss": 0.2689, "step": 8240 }, { "epoch": 53.12449799196787, "grad_norm": 6.76973819732666, "learning_rate": 2.3416129032258068e-05, "loss": 0.5491, "step": 8241 }, { "epoch": 53.13092369477911, "grad_norm": 6.648318767547607, "learning_rate": 2.3412903225806453e-05, "loss": 0.2539, "step": 8242 }, { "epoch": 53.13734939759036, "grad_norm": 5.897964954376221, "learning_rate": 2.340967741935484e-05, "loss": 0.2546, "step": 8243 }, { "epoch": 53.143775100401605, "grad_norm": 9.212940216064453, "learning_rate": 2.3406451612903228e-05, "loss": 0.3863, "step": 8244 }, { "epoch": 53.150200803212854, "grad_norm": 9.867010116577148, "learning_rate": 2.3403225806451614e-05, "loss": 0.2428, "step": 8245 }, { "epoch": 53.1566265060241, "grad_norm": 8.644875526428223, "learning_rate": 2.3400000000000003e-05, "loss": 0.5559, "step": 8246 }, { "epoch": 53.16305220883534, "grad_norm": 8.848932266235352, "learning_rate": 2.339677419354839e-05, "loss": 0.2197, "step": 8247 }, { "epoch": 53.16947791164659, "grad_norm": 5.316514015197754, "learning_rate": 2.3393548387096774e-05, "loss": 0.2031, "step": 8248 }, { "epoch": 53.17590361445783, "grad_norm": 8.20737361907959, "learning_rate": 2.3390322580645163e-05, "loss": 0.6644, "step": 8249 }, { "epoch": 53.18232931726908, "grad_norm": 30.55251121520996, "learning_rate": 2.338709677419355e-05, "loss": 0.2492, "step": 8250 }, { "epoch": 53.18875502008032, "grad_norm": 12.162216186523438, "learning_rate": 2.3383870967741935e-05, "loss": 0.5683, "step": 8251 }, { "epoch": 53.195180722891564, "grad_norm": 4.718019962310791, "learning_rate": 2.3380645161290324e-05, "loss": 0.4399, "step": 8252 }, { "epoch": 53.20160642570281, "grad_norm": 6.3041486740112305, "learning_rate": 2.3377419354838713e-05, "loss": 0.3822, "step": 8253 }, { "epoch": 53.208032128514056, "grad_norm": 19.497589111328125, "learning_rate": 2.3374193548387095e-05, "loss": 0.4233, "step": 8254 }, { "epoch": 53.2144578313253, "grad_norm": 8.054879188537598, "learning_rate": 2.3370967741935484e-05, "loss": 0.3112, "step": 8255 }, { "epoch": 53.22088353413655, "grad_norm": 18.4410400390625, "learning_rate": 2.3367741935483874e-05, "loss": 0.3243, "step": 8256 }, { "epoch": 53.22730923694779, "grad_norm": 24.496326446533203, "learning_rate": 2.336451612903226e-05, "loss": 0.4034, "step": 8257 }, { "epoch": 53.23373493975904, "grad_norm": 10.712265968322754, "learning_rate": 2.3361290322580645e-05, "loss": 0.2252, "step": 8258 }, { "epoch": 53.24016064257028, "grad_norm": 11.203986167907715, "learning_rate": 2.3358064516129034e-05, "loss": 0.3404, "step": 8259 }, { "epoch": 53.24658634538152, "grad_norm": 11.955147743225098, "learning_rate": 2.335483870967742e-05, "loss": 0.1982, "step": 8260 }, { "epoch": 53.25301204819277, "grad_norm": 22.370769500732422, "learning_rate": 2.335161290322581e-05, "loss": 0.3923, "step": 8261 }, { "epoch": 53.259437751004015, "grad_norm": 69.5159912109375, "learning_rate": 2.3348387096774195e-05, "loss": 0.4774, "step": 8262 }, { "epoch": 53.265863453815264, "grad_norm": 41.823455810546875, "learning_rate": 2.334516129032258e-05, "loss": 0.4773, "step": 8263 }, { "epoch": 53.272289156626506, "grad_norm": 22.556560516357422, "learning_rate": 2.334193548387097e-05, "loss": 0.6652, "step": 8264 }, { "epoch": 53.27871485943775, "grad_norm": 18.4066162109375, "learning_rate": 2.333870967741936e-05, "loss": 0.3076, "step": 8265 }, { "epoch": 53.285140562249, "grad_norm": 4.966412544250488, "learning_rate": 2.333548387096774e-05, "loss": 0.2308, "step": 8266 }, { "epoch": 53.29156626506024, "grad_norm": 12.54556941986084, "learning_rate": 2.333225806451613e-05, "loss": 0.3352, "step": 8267 }, { "epoch": 53.29799196787149, "grad_norm": 4.066286563873291, "learning_rate": 2.332903225806452e-05, "loss": 0.0797, "step": 8268 }, { "epoch": 53.30441767068273, "grad_norm": 11.763428688049316, "learning_rate": 2.3325806451612905e-05, "loss": 0.276, "step": 8269 }, { "epoch": 53.310843373493974, "grad_norm": 5.341954708099365, "learning_rate": 2.332258064516129e-05, "loss": 0.3142, "step": 8270 }, { "epoch": 53.31726907630522, "grad_norm": 8.416061401367188, "learning_rate": 2.331935483870968e-05, "loss": 0.3072, "step": 8271 }, { "epoch": 53.323694779116465, "grad_norm": 4.44813346862793, "learning_rate": 2.3316129032258065e-05, "loss": 0.3128, "step": 8272 }, { "epoch": 53.33012048192771, "grad_norm": 31.923831939697266, "learning_rate": 2.3312903225806454e-05, "loss": 0.4148, "step": 8273 }, { "epoch": 53.33654618473896, "grad_norm": 19.055574417114258, "learning_rate": 2.330967741935484e-05, "loss": 0.3211, "step": 8274 }, { "epoch": 53.3429718875502, "grad_norm": 9.28176498413086, "learning_rate": 2.3306451612903226e-05, "loss": 0.3816, "step": 8275 }, { "epoch": 53.34939759036145, "grad_norm": 8.076173782348633, "learning_rate": 2.3303225806451615e-05, "loss": 0.1602, "step": 8276 }, { "epoch": 53.35582329317269, "grad_norm": 6.182373523712158, "learning_rate": 2.3300000000000004e-05, "loss": 0.2814, "step": 8277 }, { "epoch": 53.36224899598393, "grad_norm": 6.746546268463135, "learning_rate": 2.3296774193548386e-05, "loss": 0.5494, "step": 8278 }, { "epoch": 53.36867469879518, "grad_norm": 23.564970016479492, "learning_rate": 2.3293548387096775e-05, "loss": 0.4398, "step": 8279 }, { "epoch": 53.375100401606424, "grad_norm": 32.98740005493164, "learning_rate": 2.3290322580645164e-05, "loss": 0.3091, "step": 8280 }, { "epoch": 53.381526104417674, "grad_norm": 13.36709976196289, "learning_rate": 2.328709677419355e-05, "loss": 0.328, "step": 8281 }, { "epoch": 53.387951807228916, "grad_norm": 10.17408275604248, "learning_rate": 2.3283870967741936e-05, "loss": 0.3987, "step": 8282 }, { "epoch": 53.39437751004016, "grad_norm": 12.74755573272705, "learning_rate": 2.3280645161290325e-05, "loss": 0.2384, "step": 8283 }, { "epoch": 53.40080321285141, "grad_norm": 27.887941360473633, "learning_rate": 2.327741935483871e-05, "loss": 0.5737, "step": 8284 }, { "epoch": 53.40722891566265, "grad_norm": 7.900670528411865, "learning_rate": 2.3274193548387096e-05, "loss": 0.4428, "step": 8285 }, { "epoch": 53.4136546184739, "grad_norm": 23.622058868408203, "learning_rate": 2.3270967741935485e-05, "loss": 0.702, "step": 8286 }, { "epoch": 53.42008032128514, "grad_norm": 7.479599952697754, "learning_rate": 2.326774193548387e-05, "loss": 0.3937, "step": 8287 }, { "epoch": 53.42650602409638, "grad_norm": 7.873880386352539, "learning_rate": 2.326451612903226e-05, "loss": 0.2511, "step": 8288 }, { "epoch": 53.43293172690763, "grad_norm": 16.077035903930664, "learning_rate": 2.3261290322580646e-05, "loss": 0.2034, "step": 8289 }, { "epoch": 53.439357429718875, "grad_norm": 11.047595024108887, "learning_rate": 2.325806451612903e-05, "loss": 0.4855, "step": 8290 }, { "epoch": 53.44578313253012, "grad_norm": 14.157708168029785, "learning_rate": 2.325483870967742e-05, "loss": 0.1996, "step": 8291 }, { "epoch": 53.45220883534137, "grad_norm": 24.572710037231445, "learning_rate": 2.325161290322581e-05, "loss": 0.5108, "step": 8292 }, { "epoch": 53.45863453815261, "grad_norm": 9.51484489440918, "learning_rate": 2.3248387096774192e-05, "loss": 0.2238, "step": 8293 }, { "epoch": 53.46506024096386, "grad_norm": 12.352067947387695, "learning_rate": 2.324516129032258e-05, "loss": 0.5004, "step": 8294 }, { "epoch": 53.4714859437751, "grad_norm": 10.702413558959961, "learning_rate": 2.324193548387097e-05, "loss": 0.346, "step": 8295 }, { "epoch": 53.47791164658634, "grad_norm": 8.898735046386719, "learning_rate": 2.3238709677419356e-05, "loss": 0.4768, "step": 8296 }, { "epoch": 53.48433734939759, "grad_norm": 16.125944137573242, "learning_rate": 2.3235483870967742e-05, "loss": 0.2747, "step": 8297 }, { "epoch": 53.490763052208834, "grad_norm": 18.798227310180664, "learning_rate": 2.323225806451613e-05, "loss": 0.2947, "step": 8298 }, { "epoch": 53.49718875502008, "grad_norm": 11.303023338317871, "learning_rate": 2.3229032258064517e-05, "loss": 0.3995, "step": 8299 }, { "epoch": 53.503614457831326, "grad_norm": 4.502843856811523, "learning_rate": 2.3225806451612906e-05, "loss": 0.2131, "step": 8300 }, { "epoch": 53.51004016064257, "grad_norm": 120.75431823730469, "learning_rate": 2.322258064516129e-05, "loss": 0.1751, "step": 8301 }, { "epoch": 53.51646586345382, "grad_norm": 5.556122779846191, "learning_rate": 2.3219354838709677e-05, "loss": 0.3798, "step": 8302 }, { "epoch": 53.52289156626506, "grad_norm": 7.369353771209717, "learning_rate": 2.3216129032258066e-05, "loss": 0.2101, "step": 8303 }, { "epoch": 53.52931726907631, "grad_norm": 19.702688217163086, "learning_rate": 2.3212903225806455e-05, "loss": 0.2155, "step": 8304 }, { "epoch": 53.53574297188755, "grad_norm": 20.115610122680664, "learning_rate": 2.3209677419354838e-05, "loss": 0.7926, "step": 8305 }, { "epoch": 53.54216867469879, "grad_norm": 21.149585723876953, "learning_rate": 2.3206451612903227e-05, "loss": 0.4006, "step": 8306 }, { "epoch": 53.54859437751004, "grad_norm": 10.79874324798584, "learning_rate": 2.3203225806451616e-05, "loss": 0.2469, "step": 8307 }, { "epoch": 53.555020080321285, "grad_norm": 3.876410722732544, "learning_rate": 2.32e-05, "loss": 0.1942, "step": 8308 }, { "epoch": 53.56144578313253, "grad_norm": 138.236328125, "learning_rate": 2.3196774193548387e-05, "loss": 0.4059, "step": 8309 }, { "epoch": 53.567871485943776, "grad_norm": 8.328344345092773, "learning_rate": 2.3193548387096776e-05, "loss": 0.4306, "step": 8310 }, { "epoch": 53.57429718875502, "grad_norm": 22.117929458618164, "learning_rate": 2.3190322580645162e-05, "loss": 0.2319, "step": 8311 }, { "epoch": 53.58072289156627, "grad_norm": 14.114813804626465, "learning_rate": 2.318709677419355e-05, "loss": 0.2542, "step": 8312 }, { "epoch": 53.58714859437751, "grad_norm": 15.354537963867188, "learning_rate": 2.3183870967741937e-05, "loss": 0.3231, "step": 8313 }, { "epoch": 53.59357429718875, "grad_norm": 16.20968246459961, "learning_rate": 2.3180645161290323e-05, "loss": 0.3517, "step": 8314 }, { "epoch": 53.6, "grad_norm": 11.874384880065918, "learning_rate": 2.317741935483871e-05, "loss": 0.3668, "step": 8315 }, { "epoch": 53.606425702811244, "grad_norm": 5.504416465759277, "learning_rate": 2.31741935483871e-05, "loss": 0.3002, "step": 8316 }, { "epoch": 53.61285140562249, "grad_norm": 6.981999397277832, "learning_rate": 2.3170967741935483e-05, "loss": 0.4585, "step": 8317 }, { "epoch": 53.619277108433735, "grad_norm": 13.765620231628418, "learning_rate": 2.3167741935483872e-05, "loss": 0.1122, "step": 8318 }, { "epoch": 53.62570281124498, "grad_norm": 34.78927993774414, "learning_rate": 2.316451612903226e-05, "loss": 0.3797, "step": 8319 }, { "epoch": 53.63212851405623, "grad_norm": 5.076805114746094, "learning_rate": 2.3161290322580644e-05, "loss": 0.2638, "step": 8320 }, { "epoch": 53.63855421686747, "grad_norm": 6.946958065032959, "learning_rate": 2.3158064516129033e-05, "loss": 0.4098, "step": 8321 }, { "epoch": 53.64497991967872, "grad_norm": 8.196404457092285, "learning_rate": 2.3154838709677422e-05, "loss": 0.1747, "step": 8322 }, { "epoch": 53.65140562248996, "grad_norm": 10.79149341583252, "learning_rate": 2.3151612903225807e-05, "loss": 0.4131, "step": 8323 }, { "epoch": 53.6578313253012, "grad_norm": 14.560623168945312, "learning_rate": 2.3148387096774193e-05, "loss": 0.4258, "step": 8324 }, { "epoch": 53.66425702811245, "grad_norm": 8.49012279510498, "learning_rate": 2.3145161290322582e-05, "loss": 0.324, "step": 8325 }, { "epoch": 53.670682730923694, "grad_norm": 16.416461944580078, "learning_rate": 2.3141935483870968e-05, "loss": 0.3957, "step": 8326 }, { "epoch": 53.67710843373494, "grad_norm": 21.223831176757812, "learning_rate": 2.3138709677419357e-05, "loss": 0.8334, "step": 8327 }, { "epoch": 53.683534136546186, "grad_norm": 7.540114879608154, "learning_rate": 2.3135483870967743e-05, "loss": 0.307, "step": 8328 }, { "epoch": 53.68995983935743, "grad_norm": 8.149510383605957, "learning_rate": 2.313225806451613e-05, "loss": 0.2482, "step": 8329 }, { "epoch": 53.69638554216868, "grad_norm": 10.45870304107666, "learning_rate": 2.3129032258064518e-05, "loss": 0.1764, "step": 8330 }, { "epoch": 53.70281124497992, "grad_norm": 10.581807136535645, "learning_rate": 2.3125806451612907e-05, "loss": 0.2446, "step": 8331 }, { "epoch": 53.70923694779116, "grad_norm": 6.281113147735596, "learning_rate": 2.312258064516129e-05, "loss": 0.2405, "step": 8332 }, { "epoch": 53.71566265060241, "grad_norm": 5.621829986572266, "learning_rate": 2.3119354838709678e-05, "loss": 0.2548, "step": 8333 }, { "epoch": 53.72208835341365, "grad_norm": 30.775007247924805, "learning_rate": 2.3116129032258067e-05, "loss": 0.495, "step": 8334 }, { "epoch": 53.7285140562249, "grad_norm": 11.327947616577148, "learning_rate": 2.3112903225806453e-05, "loss": 0.4149, "step": 8335 }, { "epoch": 53.734939759036145, "grad_norm": 16.101247787475586, "learning_rate": 2.310967741935484e-05, "loss": 0.426, "step": 8336 }, { "epoch": 53.74136546184739, "grad_norm": 11.41106128692627, "learning_rate": 2.3106451612903228e-05, "loss": 0.4838, "step": 8337 }, { "epoch": 53.747791164658636, "grad_norm": 18.39268684387207, "learning_rate": 2.3103225806451613e-05, "loss": 0.5373, "step": 8338 }, { "epoch": 53.75421686746988, "grad_norm": 3.2083046436309814, "learning_rate": 2.3100000000000002e-05, "loss": 0.3609, "step": 8339 }, { "epoch": 53.76064257028113, "grad_norm": 5.519745826721191, "learning_rate": 2.3096774193548388e-05, "loss": 0.2982, "step": 8340 }, { "epoch": 53.76706827309237, "grad_norm": 2.290193557739258, "learning_rate": 2.3093548387096774e-05, "loss": 0.0827, "step": 8341 }, { "epoch": 53.77349397590361, "grad_norm": 7.434123516082764, "learning_rate": 2.3090322580645163e-05, "loss": 0.2238, "step": 8342 }, { "epoch": 53.77991967871486, "grad_norm": 14.786657333374023, "learning_rate": 2.3087096774193552e-05, "loss": 0.4986, "step": 8343 }, { "epoch": 53.786345381526104, "grad_norm": 23.127622604370117, "learning_rate": 2.3083870967741934e-05, "loss": 0.7047, "step": 8344 }, { "epoch": 53.792771084337346, "grad_norm": 7.939465522766113, "learning_rate": 2.3080645161290323e-05, "loss": 0.2002, "step": 8345 }, { "epoch": 53.799196787148595, "grad_norm": 17.955379486083984, "learning_rate": 2.3077419354838713e-05, "loss": 0.317, "step": 8346 }, { "epoch": 53.80562248995984, "grad_norm": 25.543127059936523, "learning_rate": 2.3074193548387098e-05, "loss": 0.3845, "step": 8347 }, { "epoch": 53.81204819277109, "grad_norm": 3.479508399963379, "learning_rate": 2.3070967741935484e-05, "loss": 0.2321, "step": 8348 }, { "epoch": 53.81847389558233, "grad_norm": 9.420294761657715, "learning_rate": 2.3067741935483873e-05, "loss": 0.3785, "step": 8349 }, { "epoch": 53.82489959839357, "grad_norm": 21.328758239746094, "learning_rate": 2.306451612903226e-05, "loss": 0.5227, "step": 8350 }, { "epoch": 53.83132530120482, "grad_norm": 9.286895751953125, "learning_rate": 2.3061290322580648e-05, "loss": 0.2875, "step": 8351 }, { "epoch": 53.83775100401606, "grad_norm": 7.846019268035889, "learning_rate": 2.3058064516129034e-05, "loss": 0.3466, "step": 8352 }, { "epoch": 53.84417670682731, "grad_norm": 7.937396049499512, "learning_rate": 2.305483870967742e-05, "loss": 0.2654, "step": 8353 }, { "epoch": 53.850602409638554, "grad_norm": 6.336033344268799, "learning_rate": 2.305161290322581e-05, "loss": 0.3662, "step": 8354 }, { "epoch": 53.8570281124498, "grad_norm": 14.893156051635742, "learning_rate": 2.3048387096774194e-05, "loss": 0.2041, "step": 8355 }, { "epoch": 53.863453815261046, "grad_norm": 18.370309829711914, "learning_rate": 2.304516129032258e-05, "loss": 0.1974, "step": 8356 }, { "epoch": 53.86987951807229, "grad_norm": 9.14780330657959, "learning_rate": 2.304193548387097e-05, "loss": 0.5408, "step": 8357 }, { "epoch": 53.87630522088354, "grad_norm": 19.900146484375, "learning_rate": 2.3038709677419358e-05, "loss": 0.401, "step": 8358 }, { "epoch": 53.88273092369478, "grad_norm": 18.146625518798828, "learning_rate": 2.303548387096774e-05, "loss": 0.3172, "step": 8359 }, { "epoch": 53.88915662650602, "grad_norm": 7.422125339508057, "learning_rate": 2.303225806451613e-05, "loss": 0.2436, "step": 8360 }, { "epoch": 53.89558232931727, "grad_norm": 7.922678470611572, "learning_rate": 2.302903225806452e-05, "loss": 0.3584, "step": 8361 }, { "epoch": 53.90200803212851, "grad_norm": 30.633106231689453, "learning_rate": 2.3025806451612904e-05, "loss": 0.1878, "step": 8362 }, { "epoch": 53.908433734939756, "grad_norm": 16.327789306640625, "learning_rate": 2.302258064516129e-05, "loss": 0.4898, "step": 8363 }, { "epoch": 53.914859437751005, "grad_norm": 13.100521087646484, "learning_rate": 2.301935483870968e-05, "loss": 0.1955, "step": 8364 }, { "epoch": 53.92128514056225, "grad_norm": 40.81662368774414, "learning_rate": 2.3016129032258065e-05, "loss": 0.5192, "step": 8365 }, { "epoch": 53.9277108433735, "grad_norm": 8.070446014404297, "learning_rate": 2.3012903225806454e-05, "loss": 0.1874, "step": 8366 }, { "epoch": 53.93413654618474, "grad_norm": 20.501379013061523, "learning_rate": 2.300967741935484e-05, "loss": 0.6058, "step": 8367 }, { "epoch": 53.94056224899598, "grad_norm": 17.012359619140625, "learning_rate": 2.3006451612903225e-05, "loss": 0.5161, "step": 8368 }, { "epoch": 53.94698795180723, "grad_norm": 6.825001239776611, "learning_rate": 2.3003225806451614e-05, "loss": 0.2893, "step": 8369 }, { "epoch": 53.95341365461847, "grad_norm": 11.69084358215332, "learning_rate": 2.3000000000000003e-05, "loss": 0.4011, "step": 8370 }, { "epoch": 53.95983935742972, "grad_norm": 10.629890441894531, "learning_rate": 2.2996774193548386e-05, "loss": 0.5205, "step": 8371 }, { "epoch": 53.966265060240964, "grad_norm": 18.049488067626953, "learning_rate": 2.2993548387096775e-05, "loss": 0.5525, "step": 8372 }, { "epoch": 53.972690763052206, "grad_norm": 6.248203277587891, "learning_rate": 2.2990322580645164e-05, "loss": 0.3927, "step": 8373 }, { "epoch": 53.979116465863456, "grad_norm": 3.812784194946289, "learning_rate": 2.298709677419355e-05, "loss": 0.201, "step": 8374 }, { "epoch": 53.9855421686747, "grad_norm": 4.748183250427246, "learning_rate": 2.2983870967741935e-05, "loss": 0.3054, "step": 8375 }, { "epoch": 53.99196787148595, "grad_norm": 17.712604522705078, "learning_rate": 2.2980645161290324e-05, "loss": 0.3009, "step": 8376 }, { "epoch": 54.00160642570281, "grad_norm": 13.749829292297363, "learning_rate": 2.297741935483871e-05, "loss": 0.3062, "step": 8377 }, { "epoch": 54.00803212851405, "grad_norm": 15.6046724319458, "learning_rate": 2.29741935483871e-05, "loss": 0.399, "step": 8378 }, { "epoch": 54.0144578313253, "grad_norm": 6.2341461181640625, "learning_rate": 2.2970967741935485e-05, "loss": 0.3346, "step": 8379 }, { "epoch": 54.020883534136544, "grad_norm": 10.791793823242188, "learning_rate": 2.296774193548387e-05, "loss": 0.345, "step": 8380 }, { "epoch": 54.027309236947794, "grad_norm": 13.776445388793945, "learning_rate": 2.296451612903226e-05, "loss": 0.4772, "step": 8381 }, { "epoch": 54.033734939759036, "grad_norm": 11.872284889221191, "learning_rate": 2.296129032258065e-05, "loss": 0.6125, "step": 8382 }, { "epoch": 54.04016064257028, "grad_norm": 12.807524681091309, "learning_rate": 2.295806451612903e-05, "loss": 0.398, "step": 8383 }, { "epoch": 54.04658634538153, "grad_norm": 7.532231330871582, "learning_rate": 2.295483870967742e-05, "loss": 0.3279, "step": 8384 }, { "epoch": 54.05301204819277, "grad_norm": 6.155656814575195, "learning_rate": 2.295161290322581e-05, "loss": 0.274, "step": 8385 }, { "epoch": 54.05943775100402, "grad_norm": 17.68522834777832, "learning_rate": 2.2948387096774195e-05, "loss": 0.3053, "step": 8386 }, { "epoch": 54.06586345381526, "grad_norm": 9.57635498046875, "learning_rate": 2.294516129032258e-05, "loss": 0.4926, "step": 8387 }, { "epoch": 54.0722891566265, "grad_norm": 17.79399871826172, "learning_rate": 2.294193548387097e-05, "loss": 0.3546, "step": 8388 }, { "epoch": 54.07871485943775, "grad_norm": 11.403151512145996, "learning_rate": 2.2938709677419356e-05, "loss": 0.3882, "step": 8389 }, { "epoch": 54.085140562248995, "grad_norm": 21.70346450805664, "learning_rate": 2.293548387096774e-05, "loss": 0.3876, "step": 8390 }, { "epoch": 54.091566265060244, "grad_norm": 11.989421844482422, "learning_rate": 2.293225806451613e-05, "loss": 0.2117, "step": 8391 }, { "epoch": 54.09799196787149, "grad_norm": 31.049123764038086, "learning_rate": 2.2929032258064516e-05, "loss": 0.3176, "step": 8392 }, { "epoch": 54.10441767068273, "grad_norm": 7.346314907073975, "learning_rate": 2.2925806451612905e-05, "loss": 0.1921, "step": 8393 }, { "epoch": 54.11084337349398, "grad_norm": 8.167415618896484, "learning_rate": 2.292258064516129e-05, "loss": 0.429, "step": 8394 }, { "epoch": 54.11726907630522, "grad_norm": 5.863715171813965, "learning_rate": 2.2919354838709677e-05, "loss": 0.1999, "step": 8395 }, { "epoch": 54.12369477911646, "grad_norm": 43.099708557128906, "learning_rate": 2.2916129032258066e-05, "loss": 0.4493, "step": 8396 }, { "epoch": 54.13012048192771, "grad_norm": 23.057842254638672, "learning_rate": 2.2912903225806455e-05, "loss": 0.4246, "step": 8397 }, { "epoch": 54.136546184738954, "grad_norm": 6.822643280029297, "learning_rate": 2.2909677419354837e-05, "loss": 0.3338, "step": 8398 }, { "epoch": 54.1429718875502, "grad_norm": 27.255983352661133, "learning_rate": 2.2906451612903226e-05, "loss": 0.3511, "step": 8399 }, { "epoch": 54.149397590361446, "grad_norm": 25.616750717163086, "learning_rate": 2.2903225806451615e-05, "loss": 0.4165, "step": 8400 }, { "epoch": 54.15582329317269, "grad_norm": 8.794234275817871, "learning_rate": 2.29e-05, "loss": 0.5835, "step": 8401 }, { "epoch": 54.16224899598394, "grad_norm": 12.0003023147583, "learning_rate": 2.2896774193548387e-05, "loss": 0.5206, "step": 8402 }, { "epoch": 54.16867469879518, "grad_norm": 19.58059310913086, "learning_rate": 2.2893548387096776e-05, "loss": 0.3215, "step": 8403 }, { "epoch": 54.17510040160643, "grad_norm": 12.256162643432617, "learning_rate": 2.289032258064516e-05, "loss": 0.3107, "step": 8404 }, { "epoch": 54.18152610441767, "grad_norm": 8.344029426574707, "learning_rate": 2.288709677419355e-05, "loss": 0.4487, "step": 8405 }, { "epoch": 54.18795180722891, "grad_norm": 13.046971321105957, "learning_rate": 2.2883870967741936e-05, "loss": 0.2868, "step": 8406 }, { "epoch": 54.19437751004016, "grad_norm": 8.71540641784668, "learning_rate": 2.2880645161290322e-05, "loss": 0.3882, "step": 8407 }, { "epoch": 54.200803212851405, "grad_norm": 20.569599151611328, "learning_rate": 2.287741935483871e-05, "loss": 0.4769, "step": 8408 }, { "epoch": 54.207228915662654, "grad_norm": 5.630842685699463, "learning_rate": 2.28741935483871e-05, "loss": 0.171, "step": 8409 }, { "epoch": 54.213654618473896, "grad_norm": 6.288757801055908, "learning_rate": 2.2870967741935483e-05, "loss": 0.2788, "step": 8410 }, { "epoch": 54.22008032128514, "grad_norm": 14.417759895324707, "learning_rate": 2.286774193548387e-05, "loss": 0.2032, "step": 8411 }, { "epoch": 54.22650602409639, "grad_norm": 3.220332145690918, "learning_rate": 2.286451612903226e-05, "loss": 0.1012, "step": 8412 }, { "epoch": 54.23293172690763, "grad_norm": 4.627926349639893, "learning_rate": 2.2861290322580646e-05, "loss": 0.2163, "step": 8413 }, { "epoch": 54.23935742971887, "grad_norm": 16.823896408081055, "learning_rate": 2.2858064516129032e-05, "loss": 0.4158, "step": 8414 }, { "epoch": 54.24578313253012, "grad_norm": 22.844118118286133, "learning_rate": 2.285483870967742e-05, "loss": 0.4529, "step": 8415 }, { "epoch": 54.252208835341364, "grad_norm": 6.229879856109619, "learning_rate": 2.2851612903225807e-05, "loss": 0.2603, "step": 8416 }, { "epoch": 54.25863453815261, "grad_norm": 15.625415802001953, "learning_rate": 2.2848387096774196e-05, "loss": 0.2916, "step": 8417 }, { "epoch": 54.265060240963855, "grad_norm": 16.774845123291016, "learning_rate": 2.2845161290322582e-05, "loss": 0.4592, "step": 8418 }, { "epoch": 54.2714859437751, "grad_norm": 9.841038703918457, "learning_rate": 2.2841935483870967e-05, "loss": 0.3685, "step": 8419 }, { "epoch": 54.27791164658635, "grad_norm": 10.515037536621094, "learning_rate": 2.2838709677419357e-05, "loss": 0.1953, "step": 8420 }, { "epoch": 54.28433734939759, "grad_norm": 6.79695463180542, "learning_rate": 2.2835483870967746e-05, "loss": 0.2174, "step": 8421 }, { "epoch": 54.29076305220884, "grad_norm": 12.98112678527832, "learning_rate": 2.2832258064516128e-05, "loss": 0.4299, "step": 8422 }, { "epoch": 54.29718875502008, "grad_norm": 7.760638236999512, "learning_rate": 2.2829032258064517e-05, "loss": 0.2868, "step": 8423 }, { "epoch": 54.30361445783132, "grad_norm": 12.599329948425293, "learning_rate": 2.2825806451612906e-05, "loss": 0.1765, "step": 8424 }, { "epoch": 54.31004016064257, "grad_norm": 23.419090270996094, "learning_rate": 2.2822580645161292e-05, "loss": 0.4412, "step": 8425 }, { "epoch": 54.316465863453814, "grad_norm": 8.903014183044434, "learning_rate": 2.2819354838709678e-05, "loss": 0.2736, "step": 8426 }, { "epoch": 54.32289156626506, "grad_norm": 14.48536491394043, "learning_rate": 2.2816129032258067e-05, "loss": 0.4822, "step": 8427 }, { "epoch": 54.329317269076306, "grad_norm": 5.25405216217041, "learning_rate": 2.2812903225806452e-05, "loss": 0.1755, "step": 8428 }, { "epoch": 54.33574297188755, "grad_norm": 14.245885848999023, "learning_rate": 2.2809677419354838e-05, "loss": 0.3308, "step": 8429 }, { "epoch": 54.3421686746988, "grad_norm": 10.2273530960083, "learning_rate": 2.2806451612903227e-05, "loss": 0.308, "step": 8430 }, { "epoch": 54.34859437751004, "grad_norm": 21.637840270996094, "learning_rate": 2.2803225806451613e-05, "loss": 0.8864, "step": 8431 }, { "epoch": 54.35502008032128, "grad_norm": 11.223241806030273, "learning_rate": 2.2800000000000002e-05, "loss": 0.3382, "step": 8432 }, { "epoch": 54.36144578313253, "grad_norm": 10.660079002380371, "learning_rate": 2.2796774193548388e-05, "loss": 0.4869, "step": 8433 }, { "epoch": 54.36787148594377, "grad_norm": 10.986165046691895, "learning_rate": 2.2793548387096773e-05, "loss": 0.1343, "step": 8434 }, { "epoch": 54.37429718875502, "grad_norm": 7.854736328125, "learning_rate": 2.2790322580645162e-05, "loss": 0.3769, "step": 8435 }, { "epoch": 54.380722891566265, "grad_norm": 13.818857192993164, "learning_rate": 2.278709677419355e-05, "loss": 0.373, "step": 8436 }, { "epoch": 54.38714859437751, "grad_norm": 18.59088897705078, "learning_rate": 2.2783870967741934e-05, "loss": 0.3554, "step": 8437 }, { "epoch": 54.393574297188756, "grad_norm": 10.841196060180664, "learning_rate": 2.2780645161290323e-05, "loss": 0.2456, "step": 8438 }, { "epoch": 54.4, "grad_norm": 14.500370979309082, "learning_rate": 2.2777419354838712e-05, "loss": 0.7044, "step": 8439 }, { "epoch": 54.40642570281125, "grad_norm": 9.290318489074707, "learning_rate": 2.2774193548387098e-05, "loss": 0.1771, "step": 8440 }, { "epoch": 54.41285140562249, "grad_norm": 6.213766098022461, "learning_rate": 2.2770967741935483e-05, "loss": 0.3519, "step": 8441 }, { "epoch": 54.41927710843373, "grad_norm": 12.652292251586914, "learning_rate": 2.2767741935483873e-05, "loss": 0.2998, "step": 8442 }, { "epoch": 54.42570281124498, "grad_norm": 12.128206253051758, "learning_rate": 2.2764516129032258e-05, "loss": 0.2428, "step": 8443 }, { "epoch": 54.432128514056224, "grad_norm": 11.67878246307373, "learning_rate": 2.2761290322580647e-05, "loss": 0.3452, "step": 8444 }, { "epoch": 54.43855421686747, "grad_norm": 6.545104503631592, "learning_rate": 2.2758064516129033e-05, "loss": 0.1215, "step": 8445 }, { "epoch": 54.444979919678715, "grad_norm": 7.045259952545166, "learning_rate": 2.275483870967742e-05, "loss": 0.3289, "step": 8446 }, { "epoch": 54.45140562248996, "grad_norm": 10.646787643432617, "learning_rate": 2.2751612903225808e-05, "loss": 0.4421, "step": 8447 }, { "epoch": 54.45783132530121, "grad_norm": 5.892731666564941, "learning_rate": 2.2748387096774197e-05, "loss": 0.1766, "step": 8448 }, { "epoch": 54.46425702811245, "grad_norm": 28.08163070678711, "learning_rate": 2.274516129032258e-05, "loss": 0.555, "step": 8449 }, { "epoch": 54.47068273092369, "grad_norm": 11.691849708557129, "learning_rate": 2.274193548387097e-05, "loss": 0.5127, "step": 8450 }, { "epoch": 54.47710843373494, "grad_norm": 11.321060180664062, "learning_rate": 2.2738709677419357e-05, "loss": 0.4491, "step": 8451 }, { "epoch": 54.48353413654618, "grad_norm": 18.37396812438965, "learning_rate": 2.2735483870967743e-05, "loss": 0.5451, "step": 8452 }, { "epoch": 54.48995983935743, "grad_norm": 12.87100887298584, "learning_rate": 2.273225806451613e-05, "loss": 0.3576, "step": 8453 }, { "epoch": 54.496385542168674, "grad_norm": 14.43994140625, "learning_rate": 2.2729032258064518e-05, "loss": 0.2515, "step": 8454 }, { "epoch": 54.50281124497992, "grad_norm": 8.156157493591309, "learning_rate": 2.2725806451612904e-05, "loss": 0.3508, "step": 8455 }, { "epoch": 54.509236947791166, "grad_norm": 8.141383171081543, "learning_rate": 2.2722580645161293e-05, "loss": 0.2898, "step": 8456 }, { "epoch": 54.51566265060241, "grad_norm": 6.67735481262207, "learning_rate": 2.271935483870968e-05, "loss": 0.2724, "step": 8457 }, { "epoch": 54.52208835341366, "grad_norm": 22.766267776489258, "learning_rate": 2.2716129032258064e-05, "loss": 0.2188, "step": 8458 }, { "epoch": 54.5285140562249, "grad_norm": 8.806476593017578, "learning_rate": 2.2712903225806453e-05, "loss": 0.3195, "step": 8459 }, { "epoch": 54.53493975903614, "grad_norm": 7.251330852508545, "learning_rate": 2.2709677419354842e-05, "loss": 0.5384, "step": 8460 }, { "epoch": 54.54136546184739, "grad_norm": 5.065263748168945, "learning_rate": 2.2706451612903225e-05, "loss": 0.2803, "step": 8461 }, { "epoch": 54.54779116465863, "grad_norm": 9.343049049377441, "learning_rate": 2.2703225806451614e-05, "loss": 0.451, "step": 8462 }, { "epoch": 54.55421686746988, "grad_norm": 17.51178741455078, "learning_rate": 2.2700000000000003e-05, "loss": 0.3858, "step": 8463 }, { "epoch": 54.560642570281125, "grad_norm": 16.26546287536621, "learning_rate": 2.2696774193548385e-05, "loss": 0.3496, "step": 8464 }, { "epoch": 54.56706827309237, "grad_norm": 17.10704803466797, "learning_rate": 2.2693548387096774e-05, "loss": 0.4368, "step": 8465 }, { "epoch": 54.57349397590362, "grad_norm": 12.622209548950195, "learning_rate": 2.2690322580645163e-05, "loss": 0.4572, "step": 8466 }, { "epoch": 54.57991967871486, "grad_norm": 19.86355972290039, "learning_rate": 2.268709677419355e-05, "loss": 0.3753, "step": 8467 }, { "epoch": 54.5863453815261, "grad_norm": 11.994933128356934, "learning_rate": 2.2683870967741935e-05, "loss": 0.4428, "step": 8468 }, { "epoch": 54.59277108433735, "grad_norm": 10.118098258972168, "learning_rate": 2.2680645161290324e-05, "loss": 0.3342, "step": 8469 }, { "epoch": 54.59919678714859, "grad_norm": 13.859382629394531, "learning_rate": 2.267741935483871e-05, "loss": 0.4736, "step": 8470 }, { "epoch": 54.60562248995984, "grad_norm": 8.856517791748047, "learning_rate": 2.26741935483871e-05, "loss": 0.3362, "step": 8471 }, { "epoch": 54.612048192771084, "grad_norm": 21.210254669189453, "learning_rate": 2.2670967741935484e-05, "loss": 0.4124, "step": 8472 }, { "epoch": 54.618473895582326, "grad_norm": 8.24007511138916, "learning_rate": 2.266774193548387e-05, "loss": 0.2457, "step": 8473 }, { "epoch": 54.624899598393576, "grad_norm": 4.133711338043213, "learning_rate": 2.266451612903226e-05, "loss": 0.2939, "step": 8474 }, { "epoch": 54.63132530120482, "grad_norm": 8.7335844039917, "learning_rate": 2.266129032258065e-05, "loss": 0.5945, "step": 8475 }, { "epoch": 54.63775100401607, "grad_norm": 44.19302749633789, "learning_rate": 2.265806451612903e-05, "loss": 0.62, "step": 8476 }, { "epoch": 54.64417670682731, "grad_norm": 18.173147201538086, "learning_rate": 2.265483870967742e-05, "loss": 0.2149, "step": 8477 }, { "epoch": 54.65060240963855, "grad_norm": 5.337989330291748, "learning_rate": 2.265161290322581e-05, "loss": 0.2154, "step": 8478 }, { "epoch": 54.6570281124498, "grad_norm": 28.23044776916504, "learning_rate": 2.2648387096774195e-05, "loss": 0.3515, "step": 8479 }, { "epoch": 54.66345381526104, "grad_norm": 5.26331090927124, "learning_rate": 2.264516129032258e-05, "loss": 0.1719, "step": 8480 }, { "epoch": 54.66987951807229, "grad_norm": 11.440985679626465, "learning_rate": 2.264193548387097e-05, "loss": 0.5752, "step": 8481 }, { "epoch": 54.676305220883535, "grad_norm": 27.5925235748291, "learning_rate": 2.2638709677419355e-05, "loss": 0.3038, "step": 8482 }, { "epoch": 54.68273092369478, "grad_norm": 12.992650985717773, "learning_rate": 2.2635483870967744e-05, "loss": 0.3246, "step": 8483 }, { "epoch": 54.689156626506026, "grad_norm": 12.047537803649902, "learning_rate": 2.263225806451613e-05, "loss": 0.2144, "step": 8484 }, { "epoch": 54.69558232931727, "grad_norm": 11.760979652404785, "learning_rate": 2.2629032258064516e-05, "loss": 0.4073, "step": 8485 }, { "epoch": 54.70200803212851, "grad_norm": 15.953472137451172, "learning_rate": 2.2625806451612905e-05, "loss": 0.2723, "step": 8486 }, { "epoch": 54.70843373493976, "grad_norm": 48.28562545776367, "learning_rate": 2.2622580645161294e-05, "loss": 0.3774, "step": 8487 }, { "epoch": 54.714859437751, "grad_norm": 11.319221496582031, "learning_rate": 2.2619354838709676e-05, "loss": 0.394, "step": 8488 }, { "epoch": 54.72128514056225, "grad_norm": 25.354339599609375, "learning_rate": 2.2616129032258065e-05, "loss": 0.398, "step": 8489 }, { "epoch": 54.727710843373494, "grad_norm": 10.26396369934082, "learning_rate": 2.2612903225806454e-05, "loss": 0.5299, "step": 8490 }, { "epoch": 54.734136546184736, "grad_norm": 5.180341720581055, "learning_rate": 2.260967741935484e-05, "loss": 0.3374, "step": 8491 }, { "epoch": 54.740562248995985, "grad_norm": 20.890583038330078, "learning_rate": 2.2606451612903226e-05, "loss": 0.3253, "step": 8492 }, { "epoch": 54.74698795180723, "grad_norm": 11.784181594848633, "learning_rate": 2.2603225806451615e-05, "loss": 0.4972, "step": 8493 }, { "epoch": 54.75341365461848, "grad_norm": 11.176065444946289, "learning_rate": 2.26e-05, "loss": 0.4442, "step": 8494 }, { "epoch": 54.75983935742972, "grad_norm": 15.535170555114746, "learning_rate": 2.259677419354839e-05, "loss": 0.5232, "step": 8495 }, { "epoch": 54.76626506024096, "grad_norm": 11.982329368591309, "learning_rate": 2.2593548387096775e-05, "loss": 0.4337, "step": 8496 }, { "epoch": 54.77269076305221, "grad_norm": 15.98304557800293, "learning_rate": 2.259032258064516e-05, "loss": 0.6144, "step": 8497 }, { "epoch": 54.77911646586345, "grad_norm": 11.610346794128418, "learning_rate": 2.258709677419355e-05, "loss": 0.6706, "step": 8498 }, { "epoch": 54.7855421686747, "grad_norm": 19.181673049926758, "learning_rate": 2.2583870967741936e-05, "loss": 0.7958, "step": 8499 }, { "epoch": 54.791967871485944, "grad_norm": 11.167648315429688, "learning_rate": 2.258064516129032e-05, "loss": 0.5, "step": 8500 }, { "epoch": 54.79839357429719, "grad_norm": 7.595747470855713, "learning_rate": 2.257741935483871e-05, "loss": 0.2712, "step": 8501 }, { "epoch": 54.804819277108436, "grad_norm": 7.268211364746094, "learning_rate": 2.25741935483871e-05, "loss": 0.2649, "step": 8502 }, { "epoch": 54.81124497991968, "grad_norm": 11.189077377319336, "learning_rate": 2.2570967741935485e-05, "loss": 0.2139, "step": 8503 }, { "epoch": 54.81767068273092, "grad_norm": 16.902263641357422, "learning_rate": 2.256774193548387e-05, "loss": 0.7524, "step": 8504 }, { "epoch": 54.82409638554217, "grad_norm": 7.132918357849121, "learning_rate": 2.256451612903226e-05, "loss": 0.3721, "step": 8505 }, { "epoch": 54.83052208835341, "grad_norm": 10.559561729431152, "learning_rate": 2.2561290322580646e-05, "loss": 0.4323, "step": 8506 }, { "epoch": 54.83694779116466, "grad_norm": 27.611228942871094, "learning_rate": 2.255806451612903e-05, "loss": 0.4647, "step": 8507 }, { "epoch": 54.8433734939759, "grad_norm": 27.151979446411133, "learning_rate": 2.255483870967742e-05, "loss": 0.3163, "step": 8508 }, { "epoch": 54.849799196787146, "grad_norm": 14.139945030212402, "learning_rate": 2.2551612903225806e-05, "loss": 0.3505, "step": 8509 }, { "epoch": 54.856224899598395, "grad_norm": 7.544675827026367, "learning_rate": 2.2548387096774195e-05, "loss": 0.4659, "step": 8510 }, { "epoch": 54.86265060240964, "grad_norm": 18.284303665161133, "learning_rate": 2.254516129032258e-05, "loss": 0.1759, "step": 8511 }, { "epoch": 54.86907630522089, "grad_norm": 11.007237434387207, "learning_rate": 2.2541935483870967e-05, "loss": 0.3673, "step": 8512 }, { "epoch": 54.87550200803213, "grad_norm": 14.772024154663086, "learning_rate": 2.2538709677419356e-05, "loss": 0.4108, "step": 8513 }, { "epoch": 54.88192771084337, "grad_norm": 17.722105026245117, "learning_rate": 2.2535483870967745e-05, "loss": 0.4973, "step": 8514 }, { "epoch": 54.88835341365462, "grad_norm": 9.889018058776855, "learning_rate": 2.2532258064516127e-05, "loss": 0.298, "step": 8515 }, { "epoch": 54.89477911646586, "grad_norm": 27.08592987060547, "learning_rate": 2.2529032258064517e-05, "loss": 0.6055, "step": 8516 }, { "epoch": 54.90120481927711, "grad_norm": 4.843697547912598, "learning_rate": 2.2525806451612906e-05, "loss": 0.3375, "step": 8517 }, { "epoch": 54.907630522088354, "grad_norm": 26.64925765991211, "learning_rate": 2.252258064516129e-05, "loss": 0.4536, "step": 8518 }, { "epoch": 54.914056224899596, "grad_norm": 21.97613525390625, "learning_rate": 2.2519354838709677e-05, "loss": 0.5756, "step": 8519 }, { "epoch": 54.920481927710846, "grad_norm": 9.045520782470703, "learning_rate": 2.2516129032258066e-05, "loss": 0.3218, "step": 8520 }, { "epoch": 54.92690763052209, "grad_norm": 10.410262107849121, "learning_rate": 2.2512903225806452e-05, "loss": 0.5427, "step": 8521 }, { "epoch": 54.93333333333333, "grad_norm": 16.729581832885742, "learning_rate": 2.250967741935484e-05, "loss": 0.6556, "step": 8522 }, { "epoch": 54.93975903614458, "grad_norm": 11.595735549926758, "learning_rate": 2.2506451612903227e-05, "loss": 0.4958, "step": 8523 }, { "epoch": 54.94618473895582, "grad_norm": 21.757770538330078, "learning_rate": 2.2503225806451612e-05, "loss": 0.4464, "step": 8524 }, { "epoch": 54.95261044176707, "grad_norm": 11.143238067626953, "learning_rate": 2.25e-05, "loss": 0.4316, "step": 8525 }, { "epoch": 54.95903614457831, "grad_norm": 4.23774528503418, "learning_rate": 2.249677419354839e-05, "loss": 0.1953, "step": 8526 }, { "epoch": 54.965461847389555, "grad_norm": 11.753876686096191, "learning_rate": 2.2493548387096773e-05, "loss": 0.5141, "step": 8527 }, { "epoch": 54.971887550200805, "grad_norm": 9.055468559265137, "learning_rate": 2.2490322580645162e-05, "loss": 0.2674, "step": 8528 }, { "epoch": 54.97831325301205, "grad_norm": 10.938447952270508, "learning_rate": 2.248709677419355e-05, "loss": 0.2212, "step": 8529 }, { "epoch": 54.984738955823296, "grad_norm": 6.932473659515381, "learning_rate": 2.2483870967741937e-05, "loss": 0.3795, "step": 8530 }, { "epoch": 54.99116465863454, "grad_norm": 7.431621074676514, "learning_rate": 2.2480645161290322e-05, "loss": 0.1868, "step": 8531 }, { "epoch": 55.00080321285141, "grad_norm": 5.932088375091553, "learning_rate": 2.247741935483871e-05, "loss": 0.2497, "step": 8532 }, { "epoch": 55.00722891566265, "grad_norm": 5.837884902954102, "learning_rate": 2.2474193548387097e-05, "loss": 0.4429, "step": 8533 }, { "epoch": 55.01365461847389, "grad_norm": 10.562032699584961, "learning_rate": 2.2470967741935483e-05, "loss": 0.3016, "step": 8534 }, { "epoch": 55.02008032128514, "grad_norm": 7.105339527130127, "learning_rate": 2.2467741935483872e-05, "loss": 0.3579, "step": 8535 }, { "epoch": 55.026506024096385, "grad_norm": 13.342361450195312, "learning_rate": 2.2464516129032258e-05, "loss": 0.6718, "step": 8536 }, { "epoch": 55.03293172690763, "grad_norm": 15.84011459350586, "learning_rate": 2.2461290322580647e-05, "loss": 0.2733, "step": 8537 }, { "epoch": 55.039357429718876, "grad_norm": 9.15815544128418, "learning_rate": 2.2458064516129033e-05, "loss": 0.2488, "step": 8538 }, { "epoch": 55.04578313253012, "grad_norm": 9.707568168640137, "learning_rate": 2.2454838709677418e-05, "loss": 0.3251, "step": 8539 }, { "epoch": 55.05220883534137, "grad_norm": 10.708809852600098, "learning_rate": 2.2451612903225807e-05, "loss": 0.2329, "step": 8540 }, { "epoch": 55.05863453815261, "grad_norm": 14.532754898071289, "learning_rate": 2.2448387096774196e-05, "loss": 0.3992, "step": 8541 }, { "epoch": 55.06506024096385, "grad_norm": 63.81163024902344, "learning_rate": 2.2445161290322582e-05, "loss": 0.3332, "step": 8542 }, { "epoch": 55.0714859437751, "grad_norm": 4.509560585021973, "learning_rate": 2.2441935483870968e-05, "loss": 0.2569, "step": 8543 }, { "epoch": 55.077911646586344, "grad_norm": 15.501545906066895, "learning_rate": 2.2438709677419357e-05, "loss": 0.3614, "step": 8544 }, { "epoch": 55.08433734939759, "grad_norm": 13.603327751159668, "learning_rate": 2.2435483870967743e-05, "loss": 0.3095, "step": 8545 }, { "epoch": 55.090763052208835, "grad_norm": 12.705198287963867, "learning_rate": 2.243225806451613e-05, "loss": 0.5176, "step": 8546 }, { "epoch": 55.09718875502008, "grad_norm": 14.662434577941895, "learning_rate": 2.2429032258064517e-05, "loss": 0.1377, "step": 8547 }, { "epoch": 55.10361445783133, "grad_norm": 9.61970329284668, "learning_rate": 2.2425806451612903e-05, "loss": 0.4234, "step": 8548 }, { "epoch": 55.11004016064257, "grad_norm": 13.737009048461914, "learning_rate": 2.2422580645161292e-05, "loss": 0.2586, "step": 8549 }, { "epoch": 55.11646586345382, "grad_norm": 11.436576843261719, "learning_rate": 2.2419354838709678e-05, "loss": 0.2833, "step": 8550 }, { "epoch": 55.12289156626506, "grad_norm": 16.325359344482422, "learning_rate": 2.2416129032258064e-05, "loss": 0.3617, "step": 8551 }, { "epoch": 55.1293172690763, "grad_norm": 25.398967742919922, "learning_rate": 2.2412903225806453e-05, "loss": 0.7892, "step": 8552 }, { "epoch": 55.13574297188755, "grad_norm": 15.046781539916992, "learning_rate": 2.2409677419354842e-05, "loss": 0.409, "step": 8553 }, { "epoch": 55.142168674698794, "grad_norm": 5.889947414398193, "learning_rate": 2.2406451612903228e-05, "loss": 0.2549, "step": 8554 }, { "epoch": 55.14859437751004, "grad_norm": 5.153244972229004, "learning_rate": 2.2403225806451613e-05, "loss": 0.3781, "step": 8555 }, { "epoch": 55.155020080321286, "grad_norm": 3.470503568649292, "learning_rate": 2.2400000000000002e-05, "loss": 0.0891, "step": 8556 }, { "epoch": 55.16144578313253, "grad_norm": 14.361204147338867, "learning_rate": 2.2396774193548388e-05, "loss": 0.5326, "step": 8557 }, { "epoch": 55.16787148594378, "grad_norm": 7.615780353546143, "learning_rate": 2.2393548387096774e-05, "loss": 0.2178, "step": 8558 }, { "epoch": 55.17429718875502, "grad_norm": 14.312039375305176, "learning_rate": 2.2390322580645163e-05, "loss": 0.5492, "step": 8559 }, { "epoch": 55.18072289156626, "grad_norm": 12.28155517578125, "learning_rate": 2.238709677419355e-05, "loss": 0.3199, "step": 8560 }, { "epoch": 55.18714859437751, "grad_norm": 14.964739799499512, "learning_rate": 2.2383870967741938e-05, "loss": 0.2764, "step": 8561 }, { "epoch": 55.19357429718875, "grad_norm": 11.691545486450195, "learning_rate": 2.2380645161290323e-05, "loss": 0.6281, "step": 8562 }, { "epoch": 55.2, "grad_norm": 15.367417335510254, "learning_rate": 2.237741935483871e-05, "loss": 0.2162, "step": 8563 }, { "epoch": 55.206425702811245, "grad_norm": 10.817863464355469, "learning_rate": 2.2374193548387098e-05, "loss": 0.5776, "step": 8564 }, { "epoch": 55.21285140562249, "grad_norm": 4.893016338348389, "learning_rate": 2.2370967741935487e-05, "loss": 0.1515, "step": 8565 }, { "epoch": 55.21927710843374, "grad_norm": 76.17369079589844, "learning_rate": 2.2367741935483873e-05, "loss": 0.5265, "step": 8566 }, { "epoch": 55.22570281124498, "grad_norm": 10.022130966186523, "learning_rate": 2.236451612903226e-05, "loss": 0.3503, "step": 8567 }, { "epoch": 55.23212851405623, "grad_norm": 4.5736918449401855, "learning_rate": 2.2361290322580648e-05, "loss": 0.2042, "step": 8568 }, { "epoch": 55.23855421686747, "grad_norm": 5.41405725479126, "learning_rate": 2.2358064516129034e-05, "loss": 0.257, "step": 8569 }, { "epoch": 55.24497991967871, "grad_norm": 14.09325122833252, "learning_rate": 2.235483870967742e-05, "loss": 0.4657, "step": 8570 }, { "epoch": 55.25140562248996, "grad_norm": 6.147930145263672, "learning_rate": 2.2351612903225808e-05, "loss": 0.301, "step": 8571 }, { "epoch": 55.257831325301204, "grad_norm": 10.405645370483398, "learning_rate": 2.2348387096774194e-05, "loss": 0.3648, "step": 8572 }, { "epoch": 55.264257028112446, "grad_norm": 16.811588287353516, "learning_rate": 2.234516129032258e-05, "loss": 0.1474, "step": 8573 }, { "epoch": 55.270682730923696, "grad_norm": 12.5346097946167, "learning_rate": 2.234193548387097e-05, "loss": 0.2272, "step": 8574 }, { "epoch": 55.27710843373494, "grad_norm": 11.720685005187988, "learning_rate": 2.2338709677419355e-05, "loss": 0.4281, "step": 8575 }, { "epoch": 55.28353413654619, "grad_norm": 7.665368556976318, "learning_rate": 2.2335483870967744e-05, "loss": 0.4909, "step": 8576 }, { "epoch": 55.28995983935743, "grad_norm": 9.19697380065918, "learning_rate": 2.233225806451613e-05, "loss": 0.2732, "step": 8577 }, { "epoch": 55.29638554216867, "grad_norm": 5.48228120803833, "learning_rate": 2.2329032258064515e-05, "loss": 0.0861, "step": 8578 }, { "epoch": 55.30281124497992, "grad_norm": 7.856325626373291, "learning_rate": 2.2325806451612904e-05, "loss": 0.2534, "step": 8579 }, { "epoch": 55.30923694779116, "grad_norm": 6.7892889976501465, "learning_rate": 2.2322580645161293e-05, "loss": 0.2566, "step": 8580 }, { "epoch": 55.31566265060241, "grad_norm": 14.19792652130127, "learning_rate": 2.231935483870968e-05, "loss": 0.2682, "step": 8581 }, { "epoch": 55.322088353413655, "grad_norm": 9.52595043182373, "learning_rate": 2.2316129032258065e-05, "loss": 0.145, "step": 8582 }, { "epoch": 55.3285140562249, "grad_norm": 10.982439994812012, "learning_rate": 2.2312903225806454e-05, "loss": 0.3915, "step": 8583 }, { "epoch": 55.334939759036146, "grad_norm": 11.477272987365723, "learning_rate": 2.230967741935484e-05, "loss": 0.2278, "step": 8584 }, { "epoch": 55.34136546184739, "grad_norm": 10.797319412231445, "learning_rate": 2.2306451612903225e-05, "loss": 0.64, "step": 8585 }, { "epoch": 55.34779116465864, "grad_norm": 7.242995262145996, "learning_rate": 2.2303225806451614e-05, "loss": 0.3682, "step": 8586 }, { "epoch": 55.35421686746988, "grad_norm": 10.93015193939209, "learning_rate": 2.23e-05, "loss": 0.4536, "step": 8587 }, { "epoch": 55.36064257028112, "grad_norm": 7.892663478851318, "learning_rate": 2.229677419354839e-05, "loss": 0.3562, "step": 8588 }, { "epoch": 55.36706827309237, "grad_norm": 4.391317844390869, "learning_rate": 2.2293548387096775e-05, "loss": 0.2335, "step": 8589 }, { "epoch": 55.373493975903614, "grad_norm": 5.018495082855225, "learning_rate": 2.229032258064516e-05, "loss": 0.2743, "step": 8590 }, { "epoch": 55.379919678714856, "grad_norm": 23.228118896484375, "learning_rate": 2.228709677419355e-05, "loss": 0.3471, "step": 8591 }, { "epoch": 55.386345381526105, "grad_norm": 12.47262954711914, "learning_rate": 2.228387096774194e-05, "loss": 0.5308, "step": 8592 }, { "epoch": 55.39277108433735, "grad_norm": 13.958786010742188, "learning_rate": 2.2280645161290324e-05, "loss": 0.4448, "step": 8593 }, { "epoch": 55.3991967871486, "grad_norm": 26.531251907348633, "learning_rate": 2.227741935483871e-05, "loss": 0.3076, "step": 8594 }, { "epoch": 55.40562248995984, "grad_norm": 12.500147819519043, "learning_rate": 2.22741935483871e-05, "loss": 0.3647, "step": 8595 }, { "epoch": 55.41204819277108, "grad_norm": 9.438719749450684, "learning_rate": 2.2270967741935485e-05, "loss": 0.4949, "step": 8596 }, { "epoch": 55.41847389558233, "grad_norm": 8.78107738494873, "learning_rate": 2.226774193548387e-05, "loss": 0.4027, "step": 8597 }, { "epoch": 55.42489959839357, "grad_norm": 13.401511192321777, "learning_rate": 2.226451612903226e-05, "loss": 0.3322, "step": 8598 }, { "epoch": 55.43132530120482, "grad_norm": 7.910576820373535, "learning_rate": 2.2261290322580645e-05, "loss": 0.2871, "step": 8599 }, { "epoch": 55.437751004016064, "grad_norm": 16.672958374023438, "learning_rate": 2.2258064516129034e-05, "loss": 0.1428, "step": 8600 }, { "epoch": 55.44417670682731, "grad_norm": 12.26131820678711, "learning_rate": 2.225483870967742e-05, "loss": 0.3522, "step": 8601 }, { "epoch": 55.450602409638556, "grad_norm": 4.030019283294678, "learning_rate": 2.2251612903225806e-05, "loss": 0.1764, "step": 8602 }, { "epoch": 55.4570281124498, "grad_norm": 7.910061359405518, "learning_rate": 2.2248387096774195e-05, "loss": 0.3847, "step": 8603 }, { "epoch": 55.46345381526105, "grad_norm": 10.645905494689941, "learning_rate": 2.224516129032258e-05, "loss": 0.436, "step": 8604 }, { "epoch": 55.46987951807229, "grad_norm": 14.487853050231934, "learning_rate": 2.224193548387097e-05, "loss": 0.3871, "step": 8605 }, { "epoch": 55.47630522088353, "grad_norm": 7.780862808227539, "learning_rate": 2.2238709677419355e-05, "loss": 0.3007, "step": 8606 }, { "epoch": 55.48273092369478, "grad_norm": 23.91556167602539, "learning_rate": 2.2235483870967745e-05, "loss": 0.3843, "step": 8607 }, { "epoch": 55.48915662650602, "grad_norm": 8.151308059692383, "learning_rate": 2.223225806451613e-05, "loss": 0.3094, "step": 8608 }, { "epoch": 55.495582329317266, "grad_norm": 8.402750968933105, "learning_rate": 2.2229032258064516e-05, "loss": 0.5662, "step": 8609 }, { "epoch": 55.502008032128515, "grad_norm": 19.720529556274414, "learning_rate": 2.2225806451612905e-05, "loss": 0.4755, "step": 8610 }, { "epoch": 55.50843373493976, "grad_norm": 8.896076202392578, "learning_rate": 2.222258064516129e-05, "loss": 0.3642, "step": 8611 }, { "epoch": 55.51485943775101, "grad_norm": 39.95230484008789, "learning_rate": 2.2219354838709676e-05, "loss": 0.6581, "step": 8612 }, { "epoch": 55.52128514056225, "grad_norm": 10.662940979003906, "learning_rate": 2.2216129032258066e-05, "loss": 0.508, "step": 8613 }, { "epoch": 55.52771084337349, "grad_norm": 12.849797248840332, "learning_rate": 2.221290322580645e-05, "loss": 0.5543, "step": 8614 }, { "epoch": 55.53413654618474, "grad_norm": 7.7869439125061035, "learning_rate": 2.220967741935484e-05, "loss": 0.3322, "step": 8615 }, { "epoch": 55.54056224899598, "grad_norm": 8.519017219543457, "learning_rate": 2.2206451612903226e-05, "loss": 0.3016, "step": 8616 }, { "epoch": 55.54698795180723, "grad_norm": 4.385989665985107, "learning_rate": 2.2203225806451615e-05, "loss": 0.3375, "step": 8617 }, { "epoch": 55.553413654618474, "grad_norm": 11.488164901733398, "learning_rate": 2.22e-05, "loss": 0.6654, "step": 8618 }, { "epoch": 55.559839357429716, "grad_norm": 7.156603813171387, "learning_rate": 2.219677419354839e-05, "loss": 0.4083, "step": 8619 }, { "epoch": 55.566265060240966, "grad_norm": 8.37711238861084, "learning_rate": 2.2193548387096776e-05, "loss": 0.2772, "step": 8620 }, { "epoch": 55.57269076305221, "grad_norm": 9.572151184082031, "learning_rate": 2.219032258064516e-05, "loss": 0.5409, "step": 8621 }, { "epoch": 55.57911646586346, "grad_norm": 11.088610649108887, "learning_rate": 2.218709677419355e-05, "loss": 0.366, "step": 8622 }, { "epoch": 55.5855421686747, "grad_norm": 7.1232194900512695, "learning_rate": 2.2183870967741936e-05, "loss": 0.1458, "step": 8623 }, { "epoch": 55.59196787148594, "grad_norm": 11.26762866973877, "learning_rate": 2.2180645161290322e-05, "loss": 0.4042, "step": 8624 }, { "epoch": 55.59839357429719, "grad_norm": 13.158461570739746, "learning_rate": 2.217741935483871e-05, "loss": 0.4026, "step": 8625 }, { "epoch": 55.60481927710843, "grad_norm": 9.620585441589355, "learning_rate": 2.2174193548387097e-05, "loss": 0.3708, "step": 8626 }, { "epoch": 55.611244979919675, "grad_norm": 12.72431755065918, "learning_rate": 2.2170967741935486e-05, "loss": 0.3786, "step": 8627 }, { "epoch": 55.617670682730925, "grad_norm": 15.565566062927246, "learning_rate": 2.216774193548387e-05, "loss": 0.3666, "step": 8628 }, { "epoch": 55.62409638554217, "grad_norm": 6.551695823669434, "learning_rate": 2.216451612903226e-05, "loss": 0.3616, "step": 8629 }, { "epoch": 55.630522088353416, "grad_norm": 14.437466621398926, "learning_rate": 2.2161290322580646e-05, "loss": 0.4566, "step": 8630 }, { "epoch": 55.63694779116466, "grad_norm": 5.045342922210693, "learning_rate": 2.2158064516129035e-05, "loss": 0.3316, "step": 8631 }, { "epoch": 55.6433734939759, "grad_norm": 11.923730850219727, "learning_rate": 2.215483870967742e-05, "loss": 0.3413, "step": 8632 }, { "epoch": 55.64979919678715, "grad_norm": 10.376564025878906, "learning_rate": 2.2151612903225807e-05, "loss": 0.5697, "step": 8633 }, { "epoch": 55.65622489959839, "grad_norm": 6.664711952209473, "learning_rate": 2.2148387096774196e-05, "loss": 0.2893, "step": 8634 }, { "epoch": 55.66265060240964, "grad_norm": 9.557512283325195, "learning_rate": 2.214516129032258e-05, "loss": 0.2536, "step": 8635 }, { "epoch": 55.669076305220884, "grad_norm": 34.185176849365234, "learning_rate": 2.2141935483870967e-05, "loss": 0.4323, "step": 8636 }, { "epoch": 55.675502008032126, "grad_norm": 11.210018157958984, "learning_rate": 2.2138709677419356e-05, "loss": 0.4843, "step": 8637 }, { "epoch": 55.681927710843375, "grad_norm": 6.084740161895752, "learning_rate": 2.2135483870967742e-05, "loss": 0.2804, "step": 8638 }, { "epoch": 55.68835341365462, "grad_norm": 14.629530906677246, "learning_rate": 2.2132258064516128e-05, "loss": 0.2894, "step": 8639 }, { "epoch": 55.69477911646587, "grad_norm": 10.512938499450684, "learning_rate": 2.2129032258064517e-05, "loss": 0.4535, "step": 8640 }, { "epoch": 55.70120481927711, "grad_norm": 4.921356678009033, "learning_rate": 2.2125806451612903e-05, "loss": 0.1727, "step": 8641 }, { "epoch": 55.70763052208835, "grad_norm": 14.173575401306152, "learning_rate": 2.2122580645161292e-05, "loss": 0.5986, "step": 8642 }, { "epoch": 55.7140562248996, "grad_norm": 12.359382629394531, "learning_rate": 2.2119354838709677e-05, "loss": 0.5175, "step": 8643 }, { "epoch": 55.72048192771084, "grad_norm": 47.55646514892578, "learning_rate": 2.2116129032258067e-05, "loss": 0.1858, "step": 8644 }, { "epoch": 55.726907630522085, "grad_norm": 8.07618236541748, "learning_rate": 2.2112903225806452e-05, "loss": 0.2372, "step": 8645 }, { "epoch": 55.733333333333334, "grad_norm": 9.390458106994629, "learning_rate": 2.210967741935484e-05, "loss": 0.2099, "step": 8646 }, { "epoch": 55.73975903614458, "grad_norm": 7.443897247314453, "learning_rate": 2.2106451612903227e-05, "loss": 0.2595, "step": 8647 }, { "epoch": 55.746184738955826, "grad_norm": 44.52476119995117, "learning_rate": 2.2103225806451613e-05, "loss": 0.1599, "step": 8648 }, { "epoch": 55.75261044176707, "grad_norm": 4.062342166900635, "learning_rate": 2.2100000000000002e-05, "loss": 0.1235, "step": 8649 }, { "epoch": 55.75903614457831, "grad_norm": 12.857741355895996, "learning_rate": 2.2096774193548388e-05, "loss": 0.438, "step": 8650 }, { "epoch": 55.76546184738956, "grad_norm": 17.171611785888672, "learning_rate": 2.2093548387096773e-05, "loss": 0.2877, "step": 8651 }, { "epoch": 55.7718875502008, "grad_norm": 5.952408790588379, "learning_rate": 2.2090322580645162e-05, "loss": 0.208, "step": 8652 }, { "epoch": 55.77831325301205, "grad_norm": 4.203178405761719, "learning_rate": 2.2087096774193548e-05, "loss": 0.344, "step": 8653 }, { "epoch": 55.78473895582329, "grad_norm": 7.157889366149902, "learning_rate": 2.2083870967741937e-05, "loss": 0.2085, "step": 8654 }, { "epoch": 55.791164658634536, "grad_norm": 3.7217538356781006, "learning_rate": 2.2080645161290323e-05, "loss": 0.2866, "step": 8655 }, { "epoch": 55.797590361445785, "grad_norm": 12.63580322265625, "learning_rate": 2.2077419354838712e-05, "loss": 0.4789, "step": 8656 }, { "epoch": 55.80401606425703, "grad_norm": 9.700967788696289, "learning_rate": 2.2074193548387098e-05, "loss": 0.3257, "step": 8657 }, { "epoch": 55.810441767068276, "grad_norm": 7.882602214813232, "learning_rate": 2.2070967741935487e-05, "loss": 0.2899, "step": 8658 }, { "epoch": 55.81686746987952, "grad_norm": 5.732178211212158, "learning_rate": 2.2067741935483872e-05, "loss": 0.2265, "step": 8659 }, { "epoch": 55.82329317269076, "grad_norm": 20.332868576049805, "learning_rate": 2.2064516129032258e-05, "loss": 0.2371, "step": 8660 }, { "epoch": 55.82971887550201, "grad_norm": 22.15435028076172, "learning_rate": 2.2061290322580647e-05, "loss": 0.3942, "step": 8661 }, { "epoch": 55.83614457831325, "grad_norm": 5.40201997756958, "learning_rate": 2.2058064516129033e-05, "loss": 0.3517, "step": 8662 }, { "epoch": 55.842570281124495, "grad_norm": 5.9558868408203125, "learning_rate": 2.205483870967742e-05, "loss": 0.1712, "step": 8663 }, { "epoch": 55.848995983935744, "grad_norm": 18.968584060668945, "learning_rate": 2.2051612903225808e-05, "loss": 0.423, "step": 8664 }, { "epoch": 55.855421686746986, "grad_norm": 13.731377601623535, "learning_rate": 2.2048387096774193e-05, "loss": 0.4228, "step": 8665 }, { "epoch": 55.861847389558235, "grad_norm": 6.2737202644348145, "learning_rate": 2.2045161290322583e-05, "loss": 0.2851, "step": 8666 }, { "epoch": 55.86827309236948, "grad_norm": 7.209270000457764, "learning_rate": 2.2041935483870968e-05, "loss": 0.3049, "step": 8667 }, { "epoch": 55.87469879518072, "grad_norm": 13.226442337036133, "learning_rate": 2.2038709677419357e-05, "loss": 0.3548, "step": 8668 }, { "epoch": 55.88112449799197, "grad_norm": 8.084733963012695, "learning_rate": 2.2035483870967743e-05, "loss": 0.5073, "step": 8669 }, { "epoch": 55.88755020080321, "grad_norm": 3.8346149921417236, "learning_rate": 2.2032258064516132e-05, "loss": 0.0913, "step": 8670 }, { "epoch": 55.89397590361446, "grad_norm": 9.224181175231934, "learning_rate": 2.2029032258064518e-05, "loss": 0.1511, "step": 8671 }, { "epoch": 55.9004016064257, "grad_norm": 7.488786697387695, "learning_rate": 2.2025806451612904e-05, "loss": 0.3973, "step": 8672 }, { "epoch": 55.906827309236945, "grad_norm": 11.812317848205566, "learning_rate": 2.2022580645161293e-05, "loss": 0.4855, "step": 8673 }, { "epoch": 55.913253012048195, "grad_norm": 13.790559768676758, "learning_rate": 2.201935483870968e-05, "loss": 0.3173, "step": 8674 }, { "epoch": 55.91967871485944, "grad_norm": 17.67720603942871, "learning_rate": 2.2016129032258064e-05, "loss": 0.4412, "step": 8675 }, { "epoch": 55.926104417670686, "grad_norm": 15.28653621673584, "learning_rate": 2.2012903225806453e-05, "loss": 0.1868, "step": 8676 }, { "epoch": 55.93253012048193, "grad_norm": 37.73622131347656, "learning_rate": 2.200967741935484e-05, "loss": 0.381, "step": 8677 }, { "epoch": 55.93895582329317, "grad_norm": 10.915465354919434, "learning_rate": 2.2006451612903225e-05, "loss": 0.4217, "step": 8678 }, { "epoch": 55.94538152610442, "grad_norm": 12.561117172241211, "learning_rate": 2.2003225806451614e-05, "loss": 0.6582, "step": 8679 }, { "epoch": 55.95180722891566, "grad_norm": 10.849084854125977, "learning_rate": 2.2000000000000003e-05, "loss": 0.1655, "step": 8680 }, { "epoch": 55.958232931726904, "grad_norm": 7.689713478088379, "learning_rate": 2.199677419354839e-05, "loss": 0.3625, "step": 8681 }, { "epoch": 55.964658634538154, "grad_norm": 17.688087463378906, "learning_rate": 2.1993548387096774e-05, "loss": 0.5306, "step": 8682 }, { "epoch": 55.971084337349396, "grad_norm": 13.462970733642578, "learning_rate": 2.1990322580645163e-05, "loss": 0.4577, "step": 8683 }, { "epoch": 55.977510040160645, "grad_norm": 7.935078144073486, "learning_rate": 2.198709677419355e-05, "loss": 0.245, "step": 8684 }, { "epoch": 55.98393574297189, "grad_norm": 13.536547660827637, "learning_rate": 2.1983870967741938e-05, "loss": 0.3644, "step": 8685 }, { "epoch": 55.99036144578313, "grad_norm": 12.135068893432617, "learning_rate": 2.1980645161290324e-05, "loss": 0.2474, "step": 8686 }, { "epoch": 55.99678714859438, "grad_norm": 6.948619365692139, "learning_rate": 2.197741935483871e-05, "loss": 0.3404, "step": 8687 }, { "epoch": 56.00642570281124, "grad_norm": 11.424154281616211, "learning_rate": 2.19741935483871e-05, "loss": 0.8036, "step": 8688 }, { "epoch": 56.01285140562249, "grad_norm": 9.528037071228027, "learning_rate": 2.1970967741935484e-05, "loss": 0.747, "step": 8689 }, { "epoch": 56.019277108433734, "grad_norm": 16.633663177490234, "learning_rate": 2.196774193548387e-05, "loss": 0.4554, "step": 8690 }, { "epoch": 56.02570281124498, "grad_norm": 4.000925064086914, "learning_rate": 2.196451612903226e-05, "loss": 0.1013, "step": 8691 }, { "epoch": 56.032128514056225, "grad_norm": 8.875505447387695, "learning_rate": 2.1961290322580648e-05, "loss": 0.3843, "step": 8692 }, { "epoch": 56.03855421686747, "grad_norm": 6.119199752807617, "learning_rate": 2.1958064516129034e-05, "loss": 0.208, "step": 8693 }, { "epoch": 56.04497991967872, "grad_norm": 5.5667219161987305, "learning_rate": 2.195483870967742e-05, "loss": 0.485, "step": 8694 }, { "epoch": 56.05140562248996, "grad_norm": 5.949770927429199, "learning_rate": 2.195161290322581e-05, "loss": 0.2, "step": 8695 }, { "epoch": 56.0578313253012, "grad_norm": 5.440288543701172, "learning_rate": 2.1948387096774194e-05, "loss": 0.3691, "step": 8696 }, { "epoch": 56.06425702811245, "grad_norm": 11.838462829589844, "learning_rate": 2.1945161290322584e-05, "loss": 0.4935, "step": 8697 }, { "epoch": 56.07068273092369, "grad_norm": 8.925100326538086, "learning_rate": 2.194193548387097e-05, "loss": 0.3506, "step": 8698 }, { "epoch": 56.07710843373494, "grad_norm": 7.168172359466553, "learning_rate": 2.1938709677419355e-05, "loss": 0.3311, "step": 8699 }, { "epoch": 56.083534136546184, "grad_norm": 5.269953727722168, "learning_rate": 2.1935483870967744e-05, "loss": 0.2537, "step": 8700 }, { "epoch": 56.08995983935743, "grad_norm": 9.519928932189941, "learning_rate": 2.193225806451613e-05, "loss": 0.2534, "step": 8701 }, { "epoch": 56.096385542168676, "grad_norm": 6.486670017242432, "learning_rate": 2.1929032258064515e-05, "loss": 0.3979, "step": 8702 }, { "epoch": 56.10281124497992, "grad_norm": 8.861202239990234, "learning_rate": 2.1925806451612905e-05, "loss": 0.3244, "step": 8703 }, { "epoch": 56.10923694779117, "grad_norm": 10.010149955749512, "learning_rate": 2.192258064516129e-05, "loss": 0.2271, "step": 8704 }, { "epoch": 56.11566265060241, "grad_norm": 8.255145072937012, "learning_rate": 2.191935483870968e-05, "loss": 0.2819, "step": 8705 }, { "epoch": 56.12208835341365, "grad_norm": 7.834157943725586, "learning_rate": 2.1916129032258065e-05, "loss": 0.2203, "step": 8706 }, { "epoch": 56.1285140562249, "grad_norm": 12.677055358886719, "learning_rate": 2.1912903225806454e-05, "loss": 0.467, "step": 8707 }, { "epoch": 56.13493975903614, "grad_norm": 8.586658477783203, "learning_rate": 2.190967741935484e-05, "loss": 0.178, "step": 8708 }, { "epoch": 56.14136546184739, "grad_norm": 3.924391508102417, "learning_rate": 2.190645161290323e-05, "loss": 0.1479, "step": 8709 }, { "epoch": 56.147791164658635, "grad_norm": 12.814491271972656, "learning_rate": 2.1903225806451615e-05, "loss": 0.3194, "step": 8710 }, { "epoch": 56.15421686746988, "grad_norm": 9.131176948547363, "learning_rate": 2.19e-05, "loss": 0.3526, "step": 8711 }, { "epoch": 56.16064257028113, "grad_norm": 15.14548110961914, "learning_rate": 2.189677419354839e-05, "loss": 0.4451, "step": 8712 }, { "epoch": 56.16706827309237, "grad_norm": 5.8254714012146, "learning_rate": 2.1893548387096775e-05, "loss": 0.1871, "step": 8713 }, { "epoch": 56.17349397590361, "grad_norm": 9.74003791809082, "learning_rate": 2.189032258064516e-05, "loss": 0.5304, "step": 8714 }, { "epoch": 56.17991967871486, "grad_norm": 7.779250621795654, "learning_rate": 2.188709677419355e-05, "loss": 0.3348, "step": 8715 }, { "epoch": 56.1863453815261, "grad_norm": 17.825660705566406, "learning_rate": 2.1883870967741936e-05, "loss": 0.4811, "step": 8716 }, { "epoch": 56.19277108433735, "grad_norm": 4.359278202056885, "learning_rate": 2.188064516129032e-05, "loss": 0.1135, "step": 8717 }, { "epoch": 56.199196787148594, "grad_norm": 5.257340431213379, "learning_rate": 2.187741935483871e-05, "loss": 0.4108, "step": 8718 }, { "epoch": 56.205622489959836, "grad_norm": 7.054483413696289, "learning_rate": 2.18741935483871e-05, "loss": 0.4883, "step": 8719 }, { "epoch": 56.212048192771086, "grad_norm": 5.385734558105469, "learning_rate": 2.1870967741935485e-05, "loss": 0.2597, "step": 8720 }, { "epoch": 56.21847389558233, "grad_norm": 15.322784423828125, "learning_rate": 2.186774193548387e-05, "loss": 0.4646, "step": 8721 }, { "epoch": 56.22489959839358, "grad_norm": 16.316513061523438, "learning_rate": 2.186451612903226e-05, "loss": 0.4363, "step": 8722 }, { "epoch": 56.23132530120482, "grad_norm": 13.28602409362793, "learning_rate": 2.1861290322580646e-05, "loss": 0.4949, "step": 8723 }, { "epoch": 56.23775100401606, "grad_norm": 6.062453269958496, "learning_rate": 2.1858064516129035e-05, "loss": 0.3074, "step": 8724 }, { "epoch": 56.24417670682731, "grad_norm": 6.416532039642334, "learning_rate": 2.185483870967742e-05, "loss": 0.4699, "step": 8725 }, { "epoch": 56.25060240963855, "grad_norm": 4.163134574890137, "learning_rate": 2.1851612903225806e-05, "loss": 0.3867, "step": 8726 }, { "epoch": 56.2570281124498, "grad_norm": 7.284613609313965, "learning_rate": 2.1848387096774195e-05, "loss": 0.2992, "step": 8727 }, { "epoch": 56.263453815261045, "grad_norm": 7.560714244842529, "learning_rate": 2.184516129032258e-05, "loss": 0.2676, "step": 8728 }, { "epoch": 56.26987951807229, "grad_norm": 8.745549201965332, "learning_rate": 2.1841935483870967e-05, "loss": 0.2702, "step": 8729 }, { "epoch": 56.276305220883536, "grad_norm": 10.052202224731445, "learning_rate": 2.1838709677419356e-05, "loss": 0.3456, "step": 8730 }, { "epoch": 56.28273092369478, "grad_norm": 9.54332160949707, "learning_rate": 2.1835483870967745e-05, "loss": 0.4571, "step": 8731 }, { "epoch": 56.28915662650602, "grad_norm": 7.400074481964111, "learning_rate": 2.183225806451613e-05, "loss": 0.2972, "step": 8732 }, { "epoch": 56.29558232931727, "grad_norm": 16.43437957763672, "learning_rate": 2.1829032258064516e-05, "loss": 0.332, "step": 8733 }, { "epoch": 56.30200803212851, "grad_norm": 8.802884101867676, "learning_rate": 2.1825806451612906e-05, "loss": 0.3282, "step": 8734 }, { "epoch": 56.30843373493976, "grad_norm": 5.075444221496582, "learning_rate": 2.182258064516129e-05, "loss": 0.3269, "step": 8735 }, { "epoch": 56.314859437751004, "grad_norm": 5.7245612144470215, "learning_rate": 2.181935483870968e-05, "loss": 0.2473, "step": 8736 }, { "epoch": 56.321285140562246, "grad_norm": 11.065630912780762, "learning_rate": 2.1816129032258066e-05, "loss": 0.4469, "step": 8737 }, { "epoch": 56.327710843373495, "grad_norm": 6.706663131713867, "learning_rate": 2.1812903225806452e-05, "loss": 0.2949, "step": 8738 }, { "epoch": 56.33413654618474, "grad_norm": 6.4748663902282715, "learning_rate": 2.180967741935484e-05, "loss": 0.239, "step": 8739 }, { "epoch": 56.34056224899599, "grad_norm": 5.249592304229736, "learning_rate": 2.1806451612903227e-05, "loss": 0.298, "step": 8740 }, { "epoch": 56.34698795180723, "grad_norm": 11.112825393676758, "learning_rate": 2.1803225806451612e-05, "loss": 0.2156, "step": 8741 }, { "epoch": 56.35341365461847, "grad_norm": 3.1304922103881836, "learning_rate": 2.18e-05, "loss": 0.1402, "step": 8742 }, { "epoch": 56.35983935742972, "grad_norm": 8.764928817749023, "learning_rate": 2.179677419354839e-05, "loss": 0.3241, "step": 8743 }, { "epoch": 56.36626506024096, "grad_norm": 7.818082332611084, "learning_rate": 2.1793548387096776e-05, "loss": 0.474, "step": 8744 }, { "epoch": 56.37269076305221, "grad_norm": 8.735101699829102, "learning_rate": 2.1790322580645162e-05, "loss": 0.3375, "step": 8745 }, { "epoch": 56.379116465863454, "grad_norm": 68.05123901367188, "learning_rate": 2.178709677419355e-05, "loss": 0.1667, "step": 8746 }, { "epoch": 56.3855421686747, "grad_norm": 10.836974143981934, "learning_rate": 2.1783870967741937e-05, "loss": 0.2639, "step": 8747 }, { "epoch": 56.391967871485946, "grad_norm": 26.01140022277832, "learning_rate": 2.1780645161290322e-05, "loss": 0.3888, "step": 8748 }, { "epoch": 56.39839357429719, "grad_norm": 14.343374252319336, "learning_rate": 2.177741935483871e-05, "loss": 0.4707, "step": 8749 }, { "epoch": 56.40481927710843, "grad_norm": 13.487356185913086, "learning_rate": 2.1774193548387097e-05, "loss": 0.2867, "step": 8750 }, { "epoch": 56.41124497991968, "grad_norm": 9.889737129211426, "learning_rate": 2.1770967741935486e-05, "loss": 0.3563, "step": 8751 }, { "epoch": 56.41767068273092, "grad_norm": 11.052881240844727, "learning_rate": 2.1767741935483872e-05, "loss": 0.373, "step": 8752 }, { "epoch": 56.42409638554217, "grad_norm": 7.310066223144531, "learning_rate": 2.1764516129032258e-05, "loss": 0.4895, "step": 8753 }, { "epoch": 56.43052208835341, "grad_norm": 22.759078979492188, "learning_rate": 2.1761290322580647e-05, "loss": 0.3085, "step": 8754 }, { "epoch": 56.436947791164656, "grad_norm": 10.78786849975586, "learning_rate": 2.1758064516129036e-05, "loss": 0.4039, "step": 8755 }, { "epoch": 56.443373493975905, "grad_norm": 9.45719051361084, "learning_rate": 2.1754838709677418e-05, "loss": 0.2825, "step": 8756 }, { "epoch": 56.44979919678715, "grad_norm": 11.615312576293945, "learning_rate": 2.1751612903225807e-05, "loss": 0.6164, "step": 8757 }, { "epoch": 56.456224899598396, "grad_norm": 10.019189834594727, "learning_rate": 2.1748387096774196e-05, "loss": 0.3085, "step": 8758 }, { "epoch": 56.46265060240964, "grad_norm": 6.6183319091796875, "learning_rate": 2.1745161290322582e-05, "loss": 0.2739, "step": 8759 }, { "epoch": 56.46907630522088, "grad_norm": 17.82005500793457, "learning_rate": 2.1741935483870968e-05, "loss": 0.5573, "step": 8760 }, { "epoch": 56.47550200803213, "grad_norm": 10.386798858642578, "learning_rate": 2.1738709677419357e-05, "loss": 0.3633, "step": 8761 }, { "epoch": 56.48192771084337, "grad_norm": 8.648918151855469, "learning_rate": 2.1735483870967743e-05, "loss": 0.5237, "step": 8762 }, { "epoch": 56.48835341365462, "grad_norm": 13.551389694213867, "learning_rate": 2.173225806451613e-05, "loss": 0.3468, "step": 8763 }, { "epoch": 56.494779116465864, "grad_norm": 5.992164611816406, "learning_rate": 2.1729032258064517e-05, "loss": 0.3584, "step": 8764 }, { "epoch": 56.501204819277106, "grad_norm": 4.908015251159668, "learning_rate": 2.1725806451612903e-05, "loss": 0.2305, "step": 8765 }, { "epoch": 56.507630522088355, "grad_norm": 14.309395790100098, "learning_rate": 2.1722580645161292e-05, "loss": 0.2828, "step": 8766 }, { "epoch": 56.5140562248996, "grad_norm": 8.650580406188965, "learning_rate": 2.1719354838709678e-05, "loss": 0.3606, "step": 8767 }, { "epoch": 56.52048192771084, "grad_norm": 6.656639099121094, "learning_rate": 2.1716129032258064e-05, "loss": 0.2192, "step": 8768 }, { "epoch": 56.52690763052209, "grad_norm": 10.542424201965332, "learning_rate": 2.1712903225806453e-05, "loss": 0.3483, "step": 8769 }, { "epoch": 56.53333333333333, "grad_norm": 8.81373405456543, "learning_rate": 2.1709677419354842e-05, "loss": 0.3913, "step": 8770 }, { "epoch": 56.53975903614458, "grad_norm": 9.540003776550293, "learning_rate": 2.1706451612903227e-05, "loss": 0.3272, "step": 8771 }, { "epoch": 56.54618473895582, "grad_norm": 33.406192779541016, "learning_rate": 2.1703225806451613e-05, "loss": 0.6621, "step": 8772 }, { "epoch": 56.552610441767065, "grad_norm": 12.687828063964844, "learning_rate": 2.1700000000000002e-05, "loss": 0.3788, "step": 8773 }, { "epoch": 56.559036144578315, "grad_norm": 14.543844223022461, "learning_rate": 2.1696774193548388e-05, "loss": 0.5409, "step": 8774 }, { "epoch": 56.56546184738956, "grad_norm": 11.019344329833984, "learning_rate": 2.1693548387096777e-05, "loss": 0.2272, "step": 8775 }, { "epoch": 56.571887550200806, "grad_norm": 6.344858169555664, "learning_rate": 2.1690322580645163e-05, "loss": 0.15, "step": 8776 }, { "epoch": 56.57831325301205, "grad_norm": 8.182905197143555, "learning_rate": 2.168709677419355e-05, "loss": 0.1942, "step": 8777 }, { "epoch": 56.58473895582329, "grad_norm": 12.414752006530762, "learning_rate": 2.1683870967741938e-05, "loss": 0.259, "step": 8778 }, { "epoch": 56.59116465863454, "grad_norm": 7.532511234283447, "learning_rate": 2.1680645161290323e-05, "loss": 0.6781, "step": 8779 }, { "epoch": 56.59759036144578, "grad_norm": 13.333368301391602, "learning_rate": 2.167741935483871e-05, "loss": 0.628, "step": 8780 }, { "epoch": 56.60401606425703, "grad_norm": 6.300346374511719, "learning_rate": 2.1674193548387098e-05, "loss": 0.2775, "step": 8781 }, { "epoch": 56.610441767068274, "grad_norm": 7.075736045837402, "learning_rate": 2.1670967741935487e-05, "loss": 0.3087, "step": 8782 }, { "epoch": 56.616867469879516, "grad_norm": 6.176878929138184, "learning_rate": 2.166774193548387e-05, "loss": 0.1784, "step": 8783 }, { "epoch": 56.623293172690765, "grad_norm": 14.503384590148926, "learning_rate": 2.166451612903226e-05, "loss": 0.7506, "step": 8784 }, { "epoch": 56.62971887550201, "grad_norm": 6.9588303565979, "learning_rate": 2.1661290322580648e-05, "loss": 0.5273, "step": 8785 }, { "epoch": 56.63614457831325, "grad_norm": 14.446167945861816, "learning_rate": 2.1658064516129033e-05, "loss": 0.4074, "step": 8786 }, { "epoch": 56.6425702811245, "grad_norm": 9.930401802062988, "learning_rate": 2.165483870967742e-05, "loss": 0.5163, "step": 8787 }, { "epoch": 56.64899598393574, "grad_norm": 17.62592124938965, "learning_rate": 2.1651612903225808e-05, "loss": 0.337, "step": 8788 }, { "epoch": 56.65542168674699, "grad_norm": 10.84517765045166, "learning_rate": 2.1648387096774194e-05, "loss": 0.4533, "step": 8789 }, { "epoch": 56.66184738955823, "grad_norm": 19.2664794921875, "learning_rate": 2.1645161290322583e-05, "loss": 0.3333, "step": 8790 }, { "epoch": 56.668273092369475, "grad_norm": 12.833342552185059, "learning_rate": 2.164193548387097e-05, "loss": 0.5747, "step": 8791 }, { "epoch": 56.674698795180724, "grad_norm": 16.56853485107422, "learning_rate": 2.1638709677419354e-05, "loss": 0.3858, "step": 8792 }, { "epoch": 56.681124497991966, "grad_norm": 2.7916438579559326, "learning_rate": 2.1635483870967744e-05, "loss": 0.0577, "step": 8793 }, { "epoch": 56.687550200803216, "grad_norm": 17.794113159179688, "learning_rate": 2.1632258064516133e-05, "loss": 0.3608, "step": 8794 }, { "epoch": 56.69397590361446, "grad_norm": 7.771078109741211, "learning_rate": 2.1629032258064515e-05, "loss": 0.3363, "step": 8795 }, { "epoch": 56.7004016064257, "grad_norm": 9.335284233093262, "learning_rate": 2.1625806451612904e-05, "loss": 0.2402, "step": 8796 }, { "epoch": 56.70682730923695, "grad_norm": 11.78723430633545, "learning_rate": 2.1622580645161293e-05, "loss": 0.3344, "step": 8797 }, { "epoch": 56.71325301204819, "grad_norm": 4.338823318481445, "learning_rate": 2.161935483870968e-05, "loss": 0.2999, "step": 8798 }, { "epoch": 56.71967871485944, "grad_norm": 22.69545555114746, "learning_rate": 2.1616129032258065e-05, "loss": 0.3738, "step": 8799 }, { "epoch": 56.72610441767068, "grad_norm": 11.205507278442383, "learning_rate": 2.1612903225806454e-05, "loss": 0.5246, "step": 8800 }, { "epoch": 56.732530120481925, "grad_norm": 9.014156341552734, "learning_rate": 2.160967741935484e-05, "loss": 0.4231, "step": 8801 }, { "epoch": 56.738955823293175, "grad_norm": 9.005908012390137, "learning_rate": 2.160645161290323e-05, "loss": 0.3498, "step": 8802 }, { "epoch": 56.74538152610442, "grad_norm": 6.755700588226318, "learning_rate": 2.1603225806451614e-05, "loss": 0.3605, "step": 8803 }, { "epoch": 56.75180722891566, "grad_norm": 10.323579788208008, "learning_rate": 2.16e-05, "loss": 0.2011, "step": 8804 }, { "epoch": 56.75823293172691, "grad_norm": 8.431421279907227, "learning_rate": 2.159677419354839e-05, "loss": 0.4687, "step": 8805 }, { "epoch": 56.76465863453815, "grad_norm": 8.393315315246582, "learning_rate": 2.1593548387096778e-05, "loss": 0.3966, "step": 8806 }, { "epoch": 56.7710843373494, "grad_norm": 9.872955322265625, "learning_rate": 2.159032258064516e-05, "loss": 0.4304, "step": 8807 }, { "epoch": 56.77751004016064, "grad_norm": 9.706896781921387, "learning_rate": 2.158709677419355e-05, "loss": 0.3573, "step": 8808 }, { "epoch": 56.783935742971885, "grad_norm": 54.175071716308594, "learning_rate": 2.158387096774194e-05, "loss": 0.3169, "step": 8809 }, { "epoch": 56.790361445783134, "grad_norm": 10.058055877685547, "learning_rate": 2.1580645161290324e-05, "loss": 0.463, "step": 8810 }, { "epoch": 56.796787148594376, "grad_norm": 5.063609600067139, "learning_rate": 2.157741935483871e-05, "loss": 0.2601, "step": 8811 }, { "epoch": 56.803212851405625, "grad_norm": 29.27873992919922, "learning_rate": 2.15741935483871e-05, "loss": 0.2555, "step": 8812 }, { "epoch": 56.80963855421687, "grad_norm": 8.35562515258789, "learning_rate": 2.1570967741935485e-05, "loss": 0.2681, "step": 8813 }, { "epoch": 56.81606425702811, "grad_norm": 18.948848724365234, "learning_rate": 2.1567741935483874e-05, "loss": 0.619, "step": 8814 }, { "epoch": 56.82248995983936, "grad_norm": 19.424821853637695, "learning_rate": 2.156451612903226e-05, "loss": 0.4347, "step": 8815 }, { "epoch": 56.8289156626506, "grad_norm": 11.166234016418457, "learning_rate": 2.1561290322580645e-05, "loss": 0.2159, "step": 8816 }, { "epoch": 56.83534136546185, "grad_norm": 5.139346122741699, "learning_rate": 2.1558064516129034e-05, "loss": 0.2819, "step": 8817 }, { "epoch": 56.84176706827309, "grad_norm": 8.886659622192383, "learning_rate": 2.155483870967742e-05, "loss": 0.3346, "step": 8818 }, { "epoch": 56.848192771084335, "grad_norm": 6.665624618530273, "learning_rate": 2.1551612903225806e-05, "loss": 0.2753, "step": 8819 }, { "epoch": 56.854618473895584, "grad_norm": 7.807168483734131, "learning_rate": 2.1548387096774195e-05, "loss": 0.3061, "step": 8820 }, { "epoch": 56.86104417670683, "grad_norm": 15.260660171508789, "learning_rate": 2.1545161290322584e-05, "loss": 0.4657, "step": 8821 }, { "epoch": 56.86746987951807, "grad_norm": 7.755295753479004, "learning_rate": 2.1541935483870966e-05, "loss": 0.2857, "step": 8822 }, { "epoch": 56.87389558232932, "grad_norm": 5.231827735900879, "learning_rate": 2.1538709677419355e-05, "loss": 0.3654, "step": 8823 }, { "epoch": 56.88032128514056, "grad_norm": 9.5991792678833, "learning_rate": 2.1535483870967744e-05, "loss": 0.3311, "step": 8824 }, { "epoch": 56.88674698795181, "grad_norm": 7.4894022941589355, "learning_rate": 2.153225806451613e-05, "loss": 0.3143, "step": 8825 }, { "epoch": 56.89317269076305, "grad_norm": 16.396608352661133, "learning_rate": 2.1529032258064516e-05, "loss": 0.2163, "step": 8826 }, { "epoch": 56.899598393574294, "grad_norm": 11.663924217224121, "learning_rate": 2.1525806451612905e-05, "loss": 0.3809, "step": 8827 }, { "epoch": 56.90602409638554, "grad_norm": 7.647030830383301, "learning_rate": 2.152258064516129e-05, "loss": 0.421, "step": 8828 }, { "epoch": 56.912449799196786, "grad_norm": 12.443014144897461, "learning_rate": 2.151935483870968e-05, "loss": 0.3846, "step": 8829 }, { "epoch": 56.918875502008035, "grad_norm": 6.3491997718811035, "learning_rate": 2.1516129032258066e-05, "loss": 0.1604, "step": 8830 }, { "epoch": 56.92530120481928, "grad_norm": 27.74869155883789, "learning_rate": 2.151290322580645e-05, "loss": 0.4338, "step": 8831 }, { "epoch": 56.93172690763052, "grad_norm": 7.429074287414551, "learning_rate": 2.150967741935484e-05, "loss": 0.144, "step": 8832 }, { "epoch": 56.93815261044177, "grad_norm": 4.405149459838867, "learning_rate": 2.150645161290323e-05, "loss": 0.1445, "step": 8833 }, { "epoch": 56.94457831325301, "grad_norm": 9.08588695526123, "learning_rate": 2.1503225806451612e-05, "loss": 0.2441, "step": 8834 }, { "epoch": 56.95100401606426, "grad_norm": 4.721147537231445, "learning_rate": 2.15e-05, "loss": 0.2247, "step": 8835 }, { "epoch": 56.9574297188755, "grad_norm": 10.584759712219238, "learning_rate": 2.149677419354839e-05, "loss": 0.5143, "step": 8836 }, { "epoch": 56.963855421686745, "grad_norm": 9.20551872253418, "learning_rate": 2.1493548387096776e-05, "loss": 0.305, "step": 8837 }, { "epoch": 56.970281124497994, "grad_norm": 14.668428421020508, "learning_rate": 2.149032258064516e-05, "loss": 0.6224, "step": 8838 }, { "epoch": 56.976706827309236, "grad_norm": 5.38633394241333, "learning_rate": 2.148709677419355e-05, "loss": 0.4261, "step": 8839 }, { "epoch": 56.98313253012048, "grad_norm": 11.668970108032227, "learning_rate": 2.1483870967741936e-05, "loss": 0.4571, "step": 8840 }, { "epoch": 56.98955823293173, "grad_norm": 6.386092662811279, "learning_rate": 2.1480645161290325e-05, "loss": 0.4041, "step": 8841 }, { "epoch": 56.99598393574297, "grad_norm": 8.362720489501953, "learning_rate": 2.147741935483871e-05, "loss": 0.3916, "step": 8842 }, { "epoch": 57.00562248995984, "grad_norm": 41.13007736206055, "learning_rate": 2.1474193548387097e-05, "loss": 0.4097, "step": 8843 }, { "epoch": 57.01204819277108, "grad_norm": 38.23492431640625, "learning_rate": 2.1470967741935486e-05, "loss": 0.3937, "step": 8844 }, { "epoch": 57.01847389558233, "grad_norm": 4.9647440910339355, "learning_rate": 2.1467741935483875e-05, "loss": 0.3085, "step": 8845 }, { "epoch": 57.024899598393574, "grad_norm": 11.283077239990234, "learning_rate": 2.1464516129032257e-05, "loss": 0.4047, "step": 8846 }, { "epoch": 57.03132530120482, "grad_norm": 11.520797729492188, "learning_rate": 2.1461290322580646e-05, "loss": 0.4022, "step": 8847 }, { "epoch": 57.037751004016066, "grad_norm": 21.587053298950195, "learning_rate": 2.1458064516129035e-05, "loss": 0.5079, "step": 8848 }, { "epoch": 57.04417670682731, "grad_norm": 13.062283515930176, "learning_rate": 2.145483870967742e-05, "loss": 0.5331, "step": 8849 }, { "epoch": 57.05060240963856, "grad_norm": 22.35247802734375, "learning_rate": 2.1451612903225807e-05, "loss": 0.6297, "step": 8850 }, { "epoch": 57.0570281124498, "grad_norm": 24.352432250976562, "learning_rate": 2.1448387096774196e-05, "loss": 0.3727, "step": 8851 }, { "epoch": 57.06345381526104, "grad_norm": 11.04910659790039, "learning_rate": 2.144516129032258e-05, "loss": 0.4281, "step": 8852 }, { "epoch": 57.06987951807229, "grad_norm": 6.241481304168701, "learning_rate": 2.1441935483870967e-05, "loss": 0.227, "step": 8853 }, { "epoch": 57.07630522088353, "grad_norm": 9.898970603942871, "learning_rate": 2.1438709677419356e-05, "loss": 0.4566, "step": 8854 }, { "epoch": 57.08273092369478, "grad_norm": 12.564654350280762, "learning_rate": 2.1435483870967742e-05, "loss": 0.4417, "step": 8855 }, { "epoch": 57.089156626506025, "grad_norm": 3.269186496734619, "learning_rate": 2.143225806451613e-05, "loss": 0.263, "step": 8856 }, { "epoch": 57.09558232931727, "grad_norm": 1.3358097076416016, "learning_rate": 2.1429032258064517e-05, "loss": 0.2496, "step": 8857 }, { "epoch": 57.102008032128516, "grad_norm": 56.16328048706055, "learning_rate": 2.1425806451612903e-05, "loss": 0.4676, "step": 8858 }, { "epoch": 57.10843373493976, "grad_norm": 3.805755138397217, "learning_rate": 2.142258064516129e-05, "loss": 0.1475, "step": 8859 }, { "epoch": 57.114859437751, "grad_norm": 15.333080291748047, "learning_rate": 2.141935483870968e-05, "loss": 0.1878, "step": 8860 }, { "epoch": 57.12128514056225, "grad_norm": 13.972169876098633, "learning_rate": 2.1416129032258063e-05, "loss": 0.5682, "step": 8861 }, { "epoch": 57.12771084337349, "grad_norm": 19.931886672973633, "learning_rate": 2.1412903225806452e-05, "loss": 0.6524, "step": 8862 }, { "epoch": 57.13413654618474, "grad_norm": 8.36837387084961, "learning_rate": 2.140967741935484e-05, "loss": 0.2788, "step": 8863 }, { "epoch": 57.140562248995984, "grad_norm": 23.038318634033203, "learning_rate": 2.1406451612903227e-05, "loss": 0.2724, "step": 8864 }, { "epoch": 57.146987951807226, "grad_norm": 39.56023406982422, "learning_rate": 2.1403225806451613e-05, "loss": 0.472, "step": 8865 }, { "epoch": 57.153413654618475, "grad_norm": 33.63346862792969, "learning_rate": 2.1400000000000002e-05, "loss": 0.4385, "step": 8866 }, { "epoch": 57.15983935742972, "grad_norm": 11.86148738861084, "learning_rate": 2.1396774193548387e-05, "loss": 0.4636, "step": 8867 }, { "epoch": 57.16626506024097, "grad_norm": 2.1453261375427246, "learning_rate": 2.1393548387096777e-05, "loss": 0.3152, "step": 8868 }, { "epoch": 57.17269076305221, "grad_norm": 37.2838020324707, "learning_rate": 2.1390322580645162e-05, "loss": 0.4771, "step": 8869 }, { "epoch": 57.17911646586345, "grad_norm": 7.1012959480285645, "learning_rate": 2.1387096774193548e-05, "loss": 0.1819, "step": 8870 }, { "epoch": 57.1855421686747, "grad_norm": 3.8950352668762207, "learning_rate": 2.1383870967741937e-05, "loss": 0.274, "step": 8871 }, { "epoch": 57.19196787148594, "grad_norm": 3.221719980239868, "learning_rate": 2.1380645161290326e-05, "loss": 0.5154, "step": 8872 }, { "epoch": 57.19839357429719, "grad_norm": 11.48817253112793, "learning_rate": 2.137741935483871e-05, "loss": 0.3369, "step": 8873 }, { "epoch": 57.204819277108435, "grad_norm": 35.423213958740234, "learning_rate": 2.1374193548387098e-05, "loss": 0.6501, "step": 8874 }, { "epoch": 57.21124497991968, "grad_norm": 15.543707847595215, "learning_rate": 2.1370967741935487e-05, "loss": 0.2858, "step": 8875 }, { "epoch": 57.217670682730926, "grad_norm": 7.04957914352417, "learning_rate": 2.1367741935483872e-05, "loss": 0.1752, "step": 8876 }, { "epoch": 57.22409638554217, "grad_norm": 11.577259063720703, "learning_rate": 2.1364516129032258e-05, "loss": 0.3613, "step": 8877 }, { "epoch": 57.23052208835341, "grad_norm": 6.930624485015869, "learning_rate": 2.1361290322580647e-05, "loss": 0.284, "step": 8878 }, { "epoch": 57.23694779116466, "grad_norm": 21.70224380493164, "learning_rate": 2.1358064516129033e-05, "loss": 0.3994, "step": 8879 }, { "epoch": 57.2433734939759, "grad_norm": 3.264004945755005, "learning_rate": 2.1354838709677422e-05, "loss": 0.3144, "step": 8880 }, { "epoch": 57.24979919678715, "grad_norm": 8.883803367614746, "learning_rate": 2.1351612903225808e-05, "loss": 0.4279, "step": 8881 }, { "epoch": 57.256224899598394, "grad_norm": 18.18349266052246, "learning_rate": 2.1348387096774193e-05, "loss": 0.2611, "step": 8882 }, { "epoch": 57.262650602409636, "grad_norm": 9.592351913452148, "learning_rate": 2.1345161290322583e-05, "loss": 0.4518, "step": 8883 }, { "epoch": 57.269076305220885, "grad_norm": 10.02164363861084, "learning_rate": 2.134193548387097e-05, "loss": 0.3874, "step": 8884 }, { "epoch": 57.27550200803213, "grad_norm": 25.81232452392578, "learning_rate": 2.1338709677419354e-05, "loss": 0.6614, "step": 8885 }, { "epoch": 57.28192771084338, "grad_norm": 4.312160491943359, "learning_rate": 2.1335483870967743e-05, "loss": 0.1658, "step": 8886 }, { "epoch": 57.28835341365462, "grad_norm": 10.155618667602539, "learning_rate": 2.1332258064516132e-05, "loss": 0.6134, "step": 8887 }, { "epoch": 57.29477911646586, "grad_norm": 12.104937553405762, "learning_rate": 2.1329032258064514e-05, "loss": 0.3442, "step": 8888 }, { "epoch": 57.30120481927711, "grad_norm": 2.5294737815856934, "learning_rate": 2.1325806451612904e-05, "loss": 0.2439, "step": 8889 }, { "epoch": 57.30763052208835, "grad_norm": 13.501090049743652, "learning_rate": 2.1322580645161293e-05, "loss": 0.3949, "step": 8890 }, { "epoch": 57.3140562248996, "grad_norm": 5.663064479827881, "learning_rate": 2.131935483870968e-05, "loss": 0.2216, "step": 8891 }, { "epoch": 57.320481927710844, "grad_norm": 22.899948120117188, "learning_rate": 2.1316129032258064e-05, "loss": 0.4177, "step": 8892 }, { "epoch": 57.326907630522086, "grad_norm": 1.7142165899276733, "learning_rate": 2.1312903225806453e-05, "loss": 0.0586, "step": 8893 }, { "epoch": 57.333333333333336, "grad_norm": 10.371576309204102, "learning_rate": 2.130967741935484e-05, "loss": 0.3098, "step": 8894 }, { "epoch": 57.33975903614458, "grad_norm": 3.729853630065918, "learning_rate": 2.1306451612903228e-05, "loss": 0.3536, "step": 8895 }, { "epoch": 57.34618473895582, "grad_norm": 46.67816925048828, "learning_rate": 2.1303225806451614e-05, "loss": 0.4181, "step": 8896 }, { "epoch": 57.35261044176707, "grad_norm": 27.09041404724121, "learning_rate": 2.13e-05, "loss": 0.3114, "step": 8897 }, { "epoch": 57.35903614457831, "grad_norm": 20.768775939941406, "learning_rate": 2.129677419354839e-05, "loss": 0.3252, "step": 8898 }, { "epoch": 57.36546184738956, "grad_norm": 21.001028060913086, "learning_rate": 2.1293548387096778e-05, "loss": 0.3564, "step": 8899 }, { "epoch": 57.3718875502008, "grad_norm": 3.0573160648345947, "learning_rate": 2.129032258064516e-05, "loss": 0.2465, "step": 8900 }, { "epoch": 57.378313253012045, "grad_norm": 10.572975158691406, "learning_rate": 2.128709677419355e-05, "loss": 0.126, "step": 8901 }, { "epoch": 57.384738955823295, "grad_norm": 21.53617286682129, "learning_rate": 2.1283870967741938e-05, "loss": 0.4179, "step": 8902 }, { "epoch": 57.39116465863454, "grad_norm": 6.849263668060303, "learning_rate": 2.1280645161290324e-05, "loss": 0.211, "step": 8903 }, { "epoch": 57.397590361445786, "grad_norm": 22.288541793823242, "learning_rate": 2.127741935483871e-05, "loss": 0.6479, "step": 8904 }, { "epoch": 57.40401606425703, "grad_norm": 32.790611267089844, "learning_rate": 2.12741935483871e-05, "loss": 0.3542, "step": 8905 }, { "epoch": 57.41044176706827, "grad_norm": 24.320270538330078, "learning_rate": 2.1270967741935484e-05, "loss": 0.3372, "step": 8906 }, { "epoch": 57.41686746987952, "grad_norm": 5.4632368087768555, "learning_rate": 2.1267741935483873e-05, "loss": 0.4196, "step": 8907 }, { "epoch": 57.42329317269076, "grad_norm": 2.029021978378296, "learning_rate": 2.126451612903226e-05, "loss": 0.2366, "step": 8908 }, { "epoch": 57.429718875502004, "grad_norm": 6.615001201629639, "learning_rate": 2.1261290322580645e-05, "loss": 0.542, "step": 8909 }, { "epoch": 57.436144578313254, "grad_norm": 23.90488052368164, "learning_rate": 2.1258064516129034e-05, "loss": 0.3652, "step": 8910 }, { "epoch": 57.442570281124496, "grad_norm": 11.174256324768066, "learning_rate": 2.1254838709677423e-05, "loss": 0.4141, "step": 8911 }, { "epoch": 57.448995983935745, "grad_norm": 30.302106857299805, "learning_rate": 2.1251612903225805e-05, "loss": 0.7001, "step": 8912 }, { "epoch": 57.45542168674699, "grad_norm": 7.130460739135742, "learning_rate": 2.1248387096774194e-05, "loss": 0.2773, "step": 8913 }, { "epoch": 57.46184738955823, "grad_norm": 3.3757712841033936, "learning_rate": 2.1245161290322583e-05, "loss": 0.2758, "step": 8914 }, { "epoch": 57.46827309236948, "grad_norm": 34.09059143066406, "learning_rate": 2.124193548387097e-05, "loss": 0.4418, "step": 8915 }, { "epoch": 57.47469879518072, "grad_norm": 9.991086959838867, "learning_rate": 2.1238709677419355e-05, "loss": 0.3414, "step": 8916 }, { "epoch": 57.48112449799197, "grad_norm": 4.146393775939941, "learning_rate": 2.1235483870967744e-05, "loss": 0.3213, "step": 8917 }, { "epoch": 57.48755020080321, "grad_norm": 4.262486934661865, "learning_rate": 2.123225806451613e-05, "loss": 0.2065, "step": 8918 }, { "epoch": 57.493975903614455, "grad_norm": 27.092464447021484, "learning_rate": 2.122903225806452e-05, "loss": 0.3297, "step": 8919 }, { "epoch": 57.500401606425704, "grad_norm": 9.889774322509766, "learning_rate": 2.1225806451612904e-05, "loss": 0.3463, "step": 8920 }, { "epoch": 57.50682730923695, "grad_norm": 4.925448417663574, "learning_rate": 2.122258064516129e-05, "loss": 0.2864, "step": 8921 }, { "epoch": 57.513253012048196, "grad_norm": 10.106683731079102, "learning_rate": 2.121935483870968e-05, "loss": 0.3854, "step": 8922 }, { "epoch": 57.51967871485944, "grad_norm": 5.753169059753418, "learning_rate": 2.121612903225807e-05, "loss": 0.2525, "step": 8923 }, { "epoch": 57.52610441767068, "grad_norm": 2.655975818634033, "learning_rate": 2.121290322580645e-05, "loss": 0.4834, "step": 8924 }, { "epoch": 57.53253012048193, "grad_norm": 2.5549917221069336, "learning_rate": 2.120967741935484e-05, "loss": 0.4451, "step": 8925 }, { "epoch": 57.53895582329317, "grad_norm": 12.825581550598145, "learning_rate": 2.120645161290323e-05, "loss": 0.4805, "step": 8926 }, { "epoch": 57.54538152610442, "grad_norm": 12.083800315856934, "learning_rate": 2.120322580645161e-05, "loss": 0.1683, "step": 8927 }, { "epoch": 57.55180722891566, "grad_norm": 8.607394218444824, "learning_rate": 2.12e-05, "loss": 0.1895, "step": 8928 }, { "epoch": 57.558232931726906, "grad_norm": 13.065925598144531, "learning_rate": 2.119677419354839e-05, "loss": 0.2015, "step": 8929 }, { "epoch": 57.564658634538155, "grad_norm": 30.59544563293457, "learning_rate": 2.1193548387096775e-05, "loss": 0.344, "step": 8930 }, { "epoch": 57.5710843373494, "grad_norm": 17.501293182373047, "learning_rate": 2.119032258064516e-05, "loss": 0.4128, "step": 8931 }, { "epoch": 57.57751004016064, "grad_norm": 18.695676803588867, "learning_rate": 2.118709677419355e-05, "loss": 0.2818, "step": 8932 }, { "epoch": 57.58393574297189, "grad_norm": 17.981130599975586, "learning_rate": 2.1183870967741936e-05, "loss": 0.2977, "step": 8933 }, { "epoch": 57.59036144578313, "grad_norm": 48.40337371826172, "learning_rate": 2.1180645161290325e-05, "loss": 0.4372, "step": 8934 }, { "epoch": 57.59678714859438, "grad_norm": 8.642415046691895, "learning_rate": 2.117741935483871e-05, "loss": 0.3158, "step": 8935 }, { "epoch": 57.60321285140562, "grad_norm": 8.160150527954102, "learning_rate": 2.1174193548387096e-05, "loss": 0.3047, "step": 8936 }, { "epoch": 57.609638554216865, "grad_norm": 70.14936828613281, "learning_rate": 2.1170967741935485e-05, "loss": 0.4604, "step": 8937 }, { "epoch": 57.616064257028114, "grad_norm": 16.266855239868164, "learning_rate": 2.1167741935483874e-05, "loss": 0.2556, "step": 8938 }, { "epoch": 57.622489959839356, "grad_norm": 4.297211170196533, "learning_rate": 2.1164516129032257e-05, "loss": 0.099, "step": 8939 }, { "epoch": 57.628915662650606, "grad_norm": 2.578967809677124, "learning_rate": 2.1161290322580646e-05, "loss": 0.1592, "step": 8940 }, { "epoch": 57.63534136546185, "grad_norm": 10.939672470092773, "learning_rate": 2.1158064516129035e-05, "loss": 0.2288, "step": 8941 }, { "epoch": 57.64176706827309, "grad_norm": 17.799726486206055, "learning_rate": 2.115483870967742e-05, "loss": 0.4355, "step": 8942 }, { "epoch": 57.64819277108434, "grad_norm": 28.915359497070312, "learning_rate": 2.1151612903225806e-05, "loss": 0.1927, "step": 8943 }, { "epoch": 57.65461847389558, "grad_norm": 27.95438575744629, "learning_rate": 2.1148387096774195e-05, "loss": 0.2658, "step": 8944 }, { "epoch": 57.661044176706824, "grad_norm": 16.93316650390625, "learning_rate": 2.114516129032258e-05, "loss": 0.4134, "step": 8945 }, { "epoch": 57.66746987951807, "grad_norm": 11.945950508117676, "learning_rate": 2.114193548387097e-05, "loss": 0.3038, "step": 8946 }, { "epoch": 57.673895582329315, "grad_norm": 6.363576412200928, "learning_rate": 2.1138709677419356e-05, "loss": 0.2382, "step": 8947 }, { "epoch": 57.680321285140565, "grad_norm": 23.207223892211914, "learning_rate": 2.113548387096774e-05, "loss": 0.4614, "step": 8948 }, { "epoch": 57.68674698795181, "grad_norm": 5.331193447113037, "learning_rate": 2.113225806451613e-05, "loss": 0.2986, "step": 8949 }, { "epoch": 57.69317269076305, "grad_norm": 17.237531661987305, "learning_rate": 2.112903225806452e-05, "loss": 0.3699, "step": 8950 }, { "epoch": 57.6995983935743, "grad_norm": 4.623847007751465, "learning_rate": 2.1125806451612902e-05, "loss": 0.2359, "step": 8951 }, { "epoch": 57.70602409638554, "grad_norm": 12.216511726379395, "learning_rate": 2.112258064516129e-05, "loss": 0.3061, "step": 8952 }, { "epoch": 57.71244979919679, "grad_norm": 13.159395217895508, "learning_rate": 2.111935483870968e-05, "loss": 0.392, "step": 8953 }, { "epoch": 57.71887550200803, "grad_norm": 4.364858627319336, "learning_rate": 2.1116129032258066e-05, "loss": 0.2256, "step": 8954 }, { "epoch": 57.725301204819274, "grad_norm": 11.092702865600586, "learning_rate": 2.111290322580645e-05, "loss": 0.4015, "step": 8955 }, { "epoch": 57.731726907630524, "grad_norm": 6.621644973754883, "learning_rate": 2.110967741935484e-05, "loss": 0.4231, "step": 8956 }, { "epoch": 57.738152610441766, "grad_norm": 38.80104446411133, "learning_rate": 2.1106451612903226e-05, "loss": 0.669, "step": 8957 }, { "epoch": 57.744578313253015, "grad_norm": 26.21578598022461, "learning_rate": 2.1103225806451616e-05, "loss": 0.6119, "step": 8958 }, { "epoch": 57.75100401606426, "grad_norm": 13.392189979553223, "learning_rate": 2.11e-05, "loss": 0.5705, "step": 8959 }, { "epoch": 57.7574297188755, "grad_norm": 40.255126953125, "learning_rate": 2.1096774193548387e-05, "loss": 0.3913, "step": 8960 }, { "epoch": 57.76385542168675, "grad_norm": 43.38566589355469, "learning_rate": 2.1093548387096776e-05, "loss": 0.5543, "step": 8961 }, { "epoch": 57.77028112449799, "grad_norm": 35.84101867675781, "learning_rate": 2.1090322580645162e-05, "loss": 0.4397, "step": 8962 }, { "epoch": 57.77670682730923, "grad_norm": 6.20531702041626, "learning_rate": 2.1087096774193547e-05, "loss": 0.2225, "step": 8963 }, { "epoch": 57.78313253012048, "grad_norm": 7.39669132232666, "learning_rate": 2.1083870967741937e-05, "loss": 0.5617, "step": 8964 }, { "epoch": 57.789558232931725, "grad_norm": 10.397893905639648, "learning_rate": 2.1080645161290326e-05, "loss": 0.2044, "step": 8965 }, { "epoch": 57.795983935742974, "grad_norm": 10.236897468566895, "learning_rate": 2.1077419354838708e-05, "loss": 0.3184, "step": 8966 }, { "epoch": 57.80240963855422, "grad_norm": 5.873462677001953, "learning_rate": 2.1074193548387097e-05, "loss": 0.3147, "step": 8967 }, { "epoch": 57.80883534136546, "grad_norm": 17.290508270263672, "learning_rate": 2.1070967741935486e-05, "loss": 0.3668, "step": 8968 }, { "epoch": 57.81526104417671, "grad_norm": 30.3098087310791, "learning_rate": 2.1067741935483872e-05, "loss": 0.3827, "step": 8969 }, { "epoch": 57.82168674698795, "grad_norm": 45.81795883178711, "learning_rate": 2.1064516129032258e-05, "loss": 0.4199, "step": 8970 }, { "epoch": 57.8281124497992, "grad_norm": 25.994211196899414, "learning_rate": 2.1061290322580647e-05, "loss": 0.4242, "step": 8971 }, { "epoch": 57.83453815261044, "grad_norm": 5.615973949432373, "learning_rate": 2.1058064516129032e-05, "loss": 0.3336, "step": 8972 }, { "epoch": 57.840963855421684, "grad_norm": 43.247718811035156, "learning_rate": 2.105483870967742e-05, "loss": 0.311, "step": 8973 }, { "epoch": 57.84738955823293, "grad_norm": 10.844341278076172, "learning_rate": 2.1051612903225807e-05, "loss": 0.4523, "step": 8974 }, { "epoch": 57.853815261044176, "grad_norm": 16.253812789916992, "learning_rate": 2.1048387096774193e-05, "loss": 0.5746, "step": 8975 }, { "epoch": 57.860240963855425, "grad_norm": 19.828901290893555, "learning_rate": 2.1045161290322582e-05, "loss": 0.2561, "step": 8976 }, { "epoch": 57.86666666666667, "grad_norm": 37.440101623535156, "learning_rate": 2.104193548387097e-05, "loss": 0.444, "step": 8977 }, { "epoch": 57.87309236947791, "grad_norm": 2.940559148788452, "learning_rate": 2.1038709677419353e-05, "loss": 0.2594, "step": 8978 }, { "epoch": 57.87951807228916, "grad_norm": 11.638213157653809, "learning_rate": 2.1035483870967743e-05, "loss": 0.3734, "step": 8979 }, { "epoch": 57.8859437751004, "grad_norm": 3.7992658615112305, "learning_rate": 2.103225806451613e-05, "loss": 0.3329, "step": 8980 }, { "epoch": 57.89236947791164, "grad_norm": 7.331028938293457, "learning_rate": 2.1029032258064517e-05, "loss": 0.5475, "step": 8981 }, { "epoch": 57.89879518072289, "grad_norm": 12.07387924194336, "learning_rate": 2.1025806451612903e-05, "loss": 0.3299, "step": 8982 }, { "epoch": 57.905220883534135, "grad_norm": 57.29989242553711, "learning_rate": 2.1022580645161292e-05, "loss": 0.3593, "step": 8983 }, { "epoch": 57.911646586345384, "grad_norm": 10.215331077575684, "learning_rate": 2.1019354838709678e-05, "loss": 0.5582, "step": 8984 }, { "epoch": 57.918072289156626, "grad_norm": 14.922645568847656, "learning_rate": 2.1016129032258067e-05, "loss": 0.3021, "step": 8985 }, { "epoch": 57.92449799196787, "grad_norm": 15.902966499328613, "learning_rate": 2.1012903225806453e-05, "loss": 0.6132, "step": 8986 }, { "epoch": 57.93092369477912, "grad_norm": 22.968215942382812, "learning_rate": 2.100967741935484e-05, "loss": 0.3475, "step": 8987 }, { "epoch": 57.93734939759036, "grad_norm": 3.7389888763427734, "learning_rate": 2.1006451612903227e-05, "loss": 0.1642, "step": 8988 }, { "epoch": 57.94377510040161, "grad_norm": 26.050573348999023, "learning_rate": 2.1003225806451617e-05, "loss": 0.3776, "step": 8989 }, { "epoch": 57.95020080321285, "grad_norm": 14.682300567626953, "learning_rate": 2.1e-05, "loss": 0.2069, "step": 8990 }, { "epoch": 57.956626506024094, "grad_norm": 23.31763458251953, "learning_rate": 2.0996774193548388e-05, "loss": 0.2895, "step": 8991 }, { "epoch": 57.96305220883534, "grad_norm": 10.67624282836914, "learning_rate": 2.0993548387096777e-05, "loss": 0.3296, "step": 8992 }, { "epoch": 57.969477911646585, "grad_norm": 12.774197578430176, "learning_rate": 2.0990322580645163e-05, "loss": 0.1668, "step": 8993 }, { "epoch": 57.975903614457835, "grad_norm": 1.6005241870880127, "learning_rate": 2.098709677419355e-05, "loss": 0.1642, "step": 8994 }, { "epoch": 57.98232931726908, "grad_norm": 25.485586166381836, "learning_rate": 2.0983870967741938e-05, "loss": 0.4681, "step": 8995 }, { "epoch": 57.98875502008032, "grad_norm": 15.839212417602539, "learning_rate": 2.0980645161290323e-05, "loss": 0.4667, "step": 8996 }, { "epoch": 57.99518072289157, "grad_norm": 21.574026107788086, "learning_rate": 2.097741935483871e-05, "loss": 0.4516, "step": 8997 }, { "epoch": 58.00481927710843, "grad_norm": 6.355813503265381, "learning_rate": 2.0974193548387098e-05, "loss": 0.283, "step": 8998 }, { "epoch": 58.01124497991968, "grad_norm": 12.271167755126953, "learning_rate": 2.0970967741935484e-05, "loss": 0.2639, "step": 8999 }, { "epoch": 58.01767068273092, "grad_norm": 9.334641456604004, "learning_rate": 2.0967741935483873e-05, "loss": 0.5279, "step": 9000 }, { "epoch": 58.024096385542165, "grad_norm": 14.684054374694824, "learning_rate": 2.096451612903226e-05, "loss": 0.3071, "step": 9001 }, { "epoch": 58.030522088353415, "grad_norm": 23.3890380859375, "learning_rate": 2.0961290322580644e-05, "loss": 0.4475, "step": 9002 }, { "epoch": 58.03694779116466, "grad_norm": 15.797439575195312, "learning_rate": 2.0958064516129033e-05, "loss": 0.2738, "step": 9003 }, { "epoch": 58.043373493975906, "grad_norm": 10.502391815185547, "learning_rate": 2.0954838709677422e-05, "loss": 0.3576, "step": 9004 }, { "epoch": 58.04979919678715, "grad_norm": 5.75930643081665, "learning_rate": 2.0951612903225805e-05, "loss": 0.2114, "step": 9005 }, { "epoch": 58.05622489959839, "grad_norm": 16.41192054748535, "learning_rate": 2.0948387096774194e-05, "loss": 0.2152, "step": 9006 }, { "epoch": 58.06265060240964, "grad_norm": 7.640543460845947, "learning_rate": 2.0945161290322583e-05, "loss": 0.4764, "step": 9007 }, { "epoch": 58.06907630522088, "grad_norm": 8.690766334533691, "learning_rate": 2.094193548387097e-05, "loss": 0.3006, "step": 9008 }, { "epoch": 58.07550200803213, "grad_norm": 9.586685180664062, "learning_rate": 2.0938709677419354e-05, "loss": 0.5461, "step": 9009 }, { "epoch": 58.081927710843374, "grad_norm": 12.367435455322266, "learning_rate": 2.0935483870967743e-05, "loss": 0.2107, "step": 9010 }, { "epoch": 58.088353413654616, "grad_norm": 8.395047187805176, "learning_rate": 2.093225806451613e-05, "loss": 0.6964, "step": 9011 }, { "epoch": 58.094779116465865, "grad_norm": 20.398012161254883, "learning_rate": 2.0929032258064518e-05, "loss": 0.3282, "step": 9012 }, { "epoch": 58.10120481927711, "grad_norm": 20.571441650390625, "learning_rate": 2.0925806451612904e-05, "loss": 0.623, "step": 9013 }, { "epoch": 58.10763052208836, "grad_norm": 19.022672653198242, "learning_rate": 2.092258064516129e-05, "loss": 0.2892, "step": 9014 }, { "epoch": 58.1140562248996, "grad_norm": 6.569518089294434, "learning_rate": 2.091935483870968e-05, "loss": 0.3159, "step": 9015 }, { "epoch": 58.12048192771084, "grad_norm": 25.804847717285156, "learning_rate": 2.0916129032258068e-05, "loss": 0.3049, "step": 9016 }, { "epoch": 58.12690763052209, "grad_norm": 2.585737943649292, "learning_rate": 2.091290322580645e-05, "loss": 0.3498, "step": 9017 }, { "epoch": 58.13333333333333, "grad_norm": 24.966625213623047, "learning_rate": 2.090967741935484e-05, "loss": 0.5447, "step": 9018 }, { "epoch": 58.139759036144575, "grad_norm": 16.226186752319336, "learning_rate": 2.090645161290323e-05, "loss": 0.288, "step": 9019 }, { "epoch": 58.146184738955824, "grad_norm": 12.70874309539795, "learning_rate": 2.0903225806451614e-05, "loss": 0.3677, "step": 9020 }, { "epoch": 58.15261044176707, "grad_norm": 9.580146789550781, "learning_rate": 2.09e-05, "loss": 0.2669, "step": 9021 }, { "epoch": 58.159036144578316, "grad_norm": 7.174164295196533, "learning_rate": 2.089677419354839e-05, "loss": 0.3552, "step": 9022 }, { "epoch": 58.16546184738956, "grad_norm": 9.463091850280762, "learning_rate": 2.0893548387096775e-05, "loss": 0.2249, "step": 9023 }, { "epoch": 58.1718875502008, "grad_norm": 6.390652656555176, "learning_rate": 2.0890322580645164e-05, "loss": 0.606, "step": 9024 }, { "epoch": 58.17831325301205, "grad_norm": 56.88545608520508, "learning_rate": 2.088709677419355e-05, "loss": 0.1105, "step": 9025 }, { "epoch": 58.18473895582329, "grad_norm": 3.734297513961792, "learning_rate": 2.0883870967741935e-05, "loss": 0.319, "step": 9026 }, { "epoch": 58.19116465863454, "grad_norm": 6.537109375, "learning_rate": 2.0880645161290324e-05, "loss": 0.3482, "step": 9027 }, { "epoch": 58.19759036144578, "grad_norm": 17.711395263671875, "learning_rate": 2.0877419354838713e-05, "loss": 0.2868, "step": 9028 }, { "epoch": 58.204016064257026, "grad_norm": 2.7024073600769043, "learning_rate": 2.0874193548387096e-05, "loss": 0.2462, "step": 9029 }, { "epoch": 58.210441767068275, "grad_norm": 9.136384963989258, "learning_rate": 2.0870967741935485e-05, "loss": 0.3469, "step": 9030 }, { "epoch": 58.21686746987952, "grad_norm": 11.415303230285645, "learning_rate": 2.0867741935483874e-05, "loss": 0.3704, "step": 9031 }, { "epoch": 58.22329317269077, "grad_norm": 14.084053039550781, "learning_rate": 2.0864516129032256e-05, "loss": 0.2726, "step": 9032 }, { "epoch": 58.22971887550201, "grad_norm": 101.10780334472656, "learning_rate": 2.0861290322580645e-05, "loss": 0.6164, "step": 9033 }, { "epoch": 58.23614457831325, "grad_norm": 28.189495086669922, "learning_rate": 2.0858064516129034e-05, "loss": 0.2655, "step": 9034 }, { "epoch": 58.2425702811245, "grad_norm": 37.41965866088867, "learning_rate": 2.085483870967742e-05, "loss": 0.3872, "step": 9035 }, { "epoch": 58.24899598393574, "grad_norm": 19.557361602783203, "learning_rate": 2.0851612903225806e-05, "loss": 0.1899, "step": 9036 }, { "epoch": 58.255421686746985, "grad_norm": 23.382835388183594, "learning_rate": 2.0848387096774195e-05, "loss": 0.5186, "step": 9037 }, { "epoch": 58.261847389558234, "grad_norm": 13.943532943725586, "learning_rate": 2.084516129032258e-05, "loss": 0.2829, "step": 9038 }, { "epoch": 58.268273092369476, "grad_norm": 20.929784774780273, "learning_rate": 2.084193548387097e-05, "loss": 0.3037, "step": 9039 }, { "epoch": 58.274698795180726, "grad_norm": 16.079931259155273, "learning_rate": 2.0838709677419355e-05, "loss": 0.6522, "step": 9040 }, { "epoch": 58.28112449799197, "grad_norm": 8.813535690307617, "learning_rate": 2.083548387096774e-05, "loss": 0.1887, "step": 9041 }, { "epoch": 58.28755020080321, "grad_norm": 3.5318267345428467, "learning_rate": 2.083225806451613e-05, "loss": 0.2968, "step": 9042 }, { "epoch": 58.29397590361446, "grad_norm": 14.79654312133789, "learning_rate": 2.082903225806452e-05, "loss": 0.4774, "step": 9043 }, { "epoch": 58.3004016064257, "grad_norm": 24.100587844848633, "learning_rate": 2.08258064516129e-05, "loss": 0.5219, "step": 9044 }, { "epoch": 58.30682730923695, "grad_norm": 31.739179611206055, "learning_rate": 2.082258064516129e-05, "loss": 0.1761, "step": 9045 }, { "epoch": 58.31325301204819, "grad_norm": 26.073009490966797, "learning_rate": 2.081935483870968e-05, "loss": 0.4782, "step": 9046 }, { "epoch": 58.319678714859435, "grad_norm": 16.983394622802734, "learning_rate": 2.0816129032258065e-05, "loss": 0.2359, "step": 9047 }, { "epoch": 58.326104417670685, "grad_norm": 12.534893035888672, "learning_rate": 2.081290322580645e-05, "loss": 0.3831, "step": 9048 }, { "epoch": 58.33253012048193, "grad_norm": 36.905311584472656, "learning_rate": 2.080967741935484e-05, "loss": 0.5326, "step": 9049 }, { "epoch": 58.338955823293176, "grad_norm": 4.6371355056762695, "learning_rate": 2.0806451612903226e-05, "loss": 0.3584, "step": 9050 }, { "epoch": 58.34538152610442, "grad_norm": 11.814962387084961, "learning_rate": 2.0803225806451615e-05, "loss": 0.6045, "step": 9051 }, { "epoch": 58.35180722891566, "grad_norm": 9.344985961914062, "learning_rate": 2.08e-05, "loss": 0.0869, "step": 9052 }, { "epoch": 58.35823293172691, "grad_norm": 13.64812183380127, "learning_rate": 2.0796774193548386e-05, "loss": 0.4166, "step": 9053 }, { "epoch": 58.36465863453815, "grad_norm": 36.66450500488281, "learning_rate": 2.0793548387096776e-05, "loss": 0.3395, "step": 9054 }, { "epoch": 58.371084337349394, "grad_norm": 5.628081321716309, "learning_rate": 2.0790322580645165e-05, "loss": 0.3125, "step": 9055 }, { "epoch": 58.377510040160644, "grad_norm": 24.24956703186035, "learning_rate": 2.0787096774193547e-05, "loss": 0.1698, "step": 9056 }, { "epoch": 58.383935742971886, "grad_norm": 53.931617736816406, "learning_rate": 2.0783870967741936e-05, "loss": 0.6704, "step": 9057 }, { "epoch": 58.390361445783135, "grad_norm": 83.31592559814453, "learning_rate": 2.0780645161290325e-05, "loss": 0.4623, "step": 9058 }, { "epoch": 58.39678714859438, "grad_norm": 15.073403358459473, "learning_rate": 2.077741935483871e-05, "loss": 0.5932, "step": 9059 }, { "epoch": 58.40321285140562, "grad_norm": 12.441214561462402, "learning_rate": 2.0774193548387097e-05, "loss": 0.4374, "step": 9060 }, { "epoch": 58.40963855421687, "grad_norm": 50.35538864135742, "learning_rate": 2.0770967741935486e-05, "loss": 0.2617, "step": 9061 }, { "epoch": 58.41606425702811, "grad_norm": 127.88896179199219, "learning_rate": 2.076774193548387e-05, "loss": 0.4924, "step": 9062 }, { "epoch": 58.42248995983936, "grad_norm": 3.602724075317383, "learning_rate": 2.076451612903226e-05, "loss": 0.3676, "step": 9063 }, { "epoch": 58.4289156626506, "grad_norm": 18.067615509033203, "learning_rate": 2.0761290322580646e-05, "loss": 0.425, "step": 9064 }, { "epoch": 58.435341365461845, "grad_norm": 9.142279624938965, "learning_rate": 2.0758064516129032e-05, "loss": 0.183, "step": 9065 }, { "epoch": 58.441767068273094, "grad_norm": 19.075071334838867, "learning_rate": 2.075483870967742e-05, "loss": 0.3347, "step": 9066 }, { "epoch": 58.44819277108434, "grad_norm": 21.90576171875, "learning_rate": 2.0751612903225807e-05, "loss": 0.4127, "step": 9067 }, { "epoch": 58.454618473895586, "grad_norm": 20.86139488220215, "learning_rate": 2.0748387096774192e-05, "loss": 0.2389, "step": 9068 }, { "epoch": 58.46104417670683, "grad_norm": 17.031923294067383, "learning_rate": 2.074516129032258e-05, "loss": 0.3086, "step": 9069 }, { "epoch": 58.46746987951807, "grad_norm": 20.86531639099121, "learning_rate": 2.074193548387097e-05, "loss": 0.5671, "step": 9070 }, { "epoch": 58.47389558232932, "grad_norm": 14.9439697265625, "learning_rate": 2.0738709677419353e-05, "loss": 0.2387, "step": 9071 }, { "epoch": 58.48032128514056, "grad_norm": 5.960389614105225, "learning_rate": 2.0735483870967742e-05, "loss": 0.3953, "step": 9072 }, { "epoch": 58.486746987951804, "grad_norm": 36.69559097290039, "learning_rate": 2.073225806451613e-05, "loss": 0.6929, "step": 9073 }, { "epoch": 58.49317269076305, "grad_norm": 8.265134811401367, "learning_rate": 2.0729032258064517e-05, "loss": 0.2506, "step": 9074 }, { "epoch": 58.499598393574296, "grad_norm": 32.24312973022461, "learning_rate": 2.0725806451612903e-05, "loss": 0.3566, "step": 9075 }, { "epoch": 58.506024096385545, "grad_norm": 1.365784764289856, "learning_rate": 2.072258064516129e-05, "loss": 0.2865, "step": 9076 }, { "epoch": 58.51244979919679, "grad_norm": 6.910459518432617, "learning_rate": 2.0719354838709677e-05, "loss": 0.3684, "step": 9077 }, { "epoch": 58.51887550200803, "grad_norm": 5.0871901512146, "learning_rate": 2.0716129032258066e-05, "loss": 0.5015, "step": 9078 }, { "epoch": 58.52530120481928, "grad_norm": 20.97478485107422, "learning_rate": 2.0712903225806452e-05, "loss": 0.4884, "step": 9079 }, { "epoch": 58.53172690763052, "grad_norm": 9.652345657348633, "learning_rate": 2.0709677419354838e-05, "loss": 0.3367, "step": 9080 }, { "epoch": 58.53815261044177, "grad_norm": 20.927658081054688, "learning_rate": 2.0706451612903227e-05, "loss": 0.2846, "step": 9081 }, { "epoch": 58.54457831325301, "grad_norm": 4.662885665893555, "learning_rate": 2.0703225806451616e-05, "loss": 0.2957, "step": 9082 }, { "epoch": 58.551004016064255, "grad_norm": 17.635658264160156, "learning_rate": 2.07e-05, "loss": 0.4691, "step": 9083 }, { "epoch": 58.557429718875504, "grad_norm": 46.83458709716797, "learning_rate": 2.0696774193548387e-05, "loss": 0.5128, "step": 9084 }, { "epoch": 58.563855421686746, "grad_norm": 22.198671340942383, "learning_rate": 2.0693548387096777e-05, "loss": 0.3795, "step": 9085 }, { "epoch": 58.570281124497996, "grad_norm": 4.8026556968688965, "learning_rate": 2.0690322580645162e-05, "loss": 0.5535, "step": 9086 }, { "epoch": 58.57670682730924, "grad_norm": 18.91691017150879, "learning_rate": 2.0687096774193548e-05, "loss": 0.2102, "step": 9087 }, { "epoch": 58.58313253012048, "grad_norm": 10.94459056854248, "learning_rate": 2.0683870967741937e-05, "loss": 0.4945, "step": 9088 }, { "epoch": 58.58955823293173, "grad_norm": 90.03770446777344, "learning_rate": 2.0680645161290323e-05, "loss": 0.3952, "step": 9089 }, { "epoch": 58.59598393574297, "grad_norm": 21.50893783569336, "learning_rate": 2.0677419354838712e-05, "loss": 0.3368, "step": 9090 }, { "epoch": 58.602409638554214, "grad_norm": 109.1624755859375, "learning_rate": 2.0674193548387098e-05, "loss": 0.5266, "step": 9091 }, { "epoch": 58.60883534136546, "grad_norm": 20.683197021484375, "learning_rate": 2.0670967741935483e-05, "loss": 0.4256, "step": 9092 }, { "epoch": 58.615261044176705, "grad_norm": 37.26167678833008, "learning_rate": 2.0667741935483872e-05, "loss": 0.6136, "step": 9093 }, { "epoch": 58.621686746987955, "grad_norm": 19.088563919067383, "learning_rate": 2.066451612903226e-05, "loss": 0.5883, "step": 9094 }, { "epoch": 58.6281124497992, "grad_norm": 17.627483367919922, "learning_rate": 2.0661290322580644e-05, "loss": 0.2369, "step": 9095 }, { "epoch": 58.63453815261044, "grad_norm": 11.21904182434082, "learning_rate": 2.0658064516129033e-05, "loss": 0.2504, "step": 9096 }, { "epoch": 58.64096385542169, "grad_norm": 4.280948162078857, "learning_rate": 2.0654838709677422e-05, "loss": 0.2547, "step": 9097 }, { "epoch": 58.64738955823293, "grad_norm": 11.046573638916016, "learning_rate": 2.0651612903225808e-05, "loss": 0.2744, "step": 9098 }, { "epoch": 58.65381526104418, "grad_norm": 29.542797088623047, "learning_rate": 2.0648387096774193e-05, "loss": 0.5232, "step": 9099 }, { "epoch": 58.66024096385542, "grad_norm": 3.5500810146331787, "learning_rate": 2.0645161290322582e-05, "loss": 0.3692, "step": 9100 }, { "epoch": 58.666666666666664, "grad_norm": 7.775793075561523, "learning_rate": 2.0641935483870968e-05, "loss": 0.3427, "step": 9101 }, { "epoch": 58.673092369477914, "grad_norm": 21.1373348236084, "learning_rate": 2.0638709677419354e-05, "loss": 0.3102, "step": 9102 }, { "epoch": 58.679518072289156, "grad_norm": 7.074936866760254, "learning_rate": 2.0635483870967743e-05, "loss": 0.2812, "step": 9103 }, { "epoch": 58.6859437751004, "grad_norm": 23.22747230529785, "learning_rate": 2.063225806451613e-05, "loss": 0.4681, "step": 9104 }, { "epoch": 58.69236947791165, "grad_norm": 22.966455459594727, "learning_rate": 2.0629032258064518e-05, "loss": 0.4809, "step": 9105 }, { "epoch": 58.69879518072289, "grad_norm": 17.001924514770508, "learning_rate": 2.0625806451612903e-05, "loss": 0.3057, "step": 9106 }, { "epoch": 58.70522088353414, "grad_norm": 15.174927711486816, "learning_rate": 2.062258064516129e-05, "loss": 0.192, "step": 9107 }, { "epoch": 58.71164658634538, "grad_norm": 10.509733200073242, "learning_rate": 2.0619354838709678e-05, "loss": 0.2436, "step": 9108 }, { "epoch": 58.71807228915662, "grad_norm": 4.360930919647217, "learning_rate": 2.0616129032258067e-05, "loss": 0.2228, "step": 9109 }, { "epoch": 58.72449799196787, "grad_norm": 3.629084348678589, "learning_rate": 2.0612903225806453e-05, "loss": 0.2293, "step": 9110 }, { "epoch": 58.730923694779115, "grad_norm": 27.28446388244629, "learning_rate": 2.060967741935484e-05, "loss": 0.3415, "step": 9111 }, { "epoch": 58.737349397590364, "grad_norm": 2.484060525894165, "learning_rate": 2.0606451612903228e-05, "loss": 0.2059, "step": 9112 }, { "epoch": 58.743775100401606, "grad_norm": 6.284982681274414, "learning_rate": 2.0603225806451614e-05, "loss": 0.3417, "step": 9113 }, { "epoch": 58.75020080321285, "grad_norm": 36.72643280029297, "learning_rate": 2.06e-05, "loss": 0.456, "step": 9114 }, { "epoch": 58.7566265060241, "grad_norm": 43.21123504638672, "learning_rate": 2.059677419354839e-05, "loss": 0.3595, "step": 9115 }, { "epoch": 58.76305220883534, "grad_norm": 19.441051483154297, "learning_rate": 2.0593548387096774e-05, "loss": 0.2591, "step": 9116 }, { "epoch": 58.76947791164659, "grad_norm": 12.940001487731934, "learning_rate": 2.0590322580645163e-05, "loss": 0.3027, "step": 9117 }, { "epoch": 58.77590361445783, "grad_norm": 32.98129653930664, "learning_rate": 2.058709677419355e-05, "loss": 0.6896, "step": 9118 }, { "epoch": 58.782329317269074, "grad_norm": 47.37898635864258, "learning_rate": 2.0583870967741935e-05, "loss": 0.3045, "step": 9119 }, { "epoch": 58.78875502008032, "grad_norm": 4.632575511932373, "learning_rate": 2.0580645161290324e-05, "loss": 0.1671, "step": 9120 }, { "epoch": 58.795180722891565, "grad_norm": 3.631967067718506, "learning_rate": 2.0577419354838713e-05, "loss": 0.2411, "step": 9121 }, { "epoch": 58.801606425702815, "grad_norm": 14.60024356842041, "learning_rate": 2.0574193548387095e-05, "loss": 0.2221, "step": 9122 }, { "epoch": 58.80803212851406, "grad_norm": 20.486724853515625, "learning_rate": 2.0570967741935484e-05, "loss": 0.33, "step": 9123 }, { "epoch": 58.8144578313253, "grad_norm": 7.692007064819336, "learning_rate": 2.0567741935483873e-05, "loss": 0.3742, "step": 9124 }, { "epoch": 58.82088353413655, "grad_norm": 36.43500900268555, "learning_rate": 2.056451612903226e-05, "loss": 0.2989, "step": 9125 }, { "epoch": 58.82730923694779, "grad_norm": 5.355484485626221, "learning_rate": 2.0561290322580645e-05, "loss": 0.4262, "step": 9126 }, { "epoch": 58.83373493975903, "grad_norm": 37.499244689941406, "learning_rate": 2.0558064516129034e-05, "loss": 0.3992, "step": 9127 }, { "epoch": 58.84016064257028, "grad_norm": 30.56252670288086, "learning_rate": 2.055483870967742e-05, "loss": 0.7595, "step": 9128 }, { "epoch": 58.846586345381525, "grad_norm": 15.795065879821777, "learning_rate": 2.055161290322581e-05, "loss": 0.2758, "step": 9129 }, { "epoch": 58.853012048192774, "grad_norm": 18.938302993774414, "learning_rate": 2.0548387096774194e-05, "loss": 0.2787, "step": 9130 }, { "epoch": 58.859437751004016, "grad_norm": 71.398681640625, "learning_rate": 2.054516129032258e-05, "loss": 0.4797, "step": 9131 }, { "epoch": 58.86586345381526, "grad_norm": 17.164358139038086, "learning_rate": 2.054193548387097e-05, "loss": 0.2863, "step": 9132 }, { "epoch": 58.87228915662651, "grad_norm": 5.177850246429443, "learning_rate": 2.0538709677419358e-05, "loss": 0.368, "step": 9133 }, { "epoch": 58.87871485943775, "grad_norm": 36.271766662597656, "learning_rate": 2.053548387096774e-05, "loss": 0.6979, "step": 9134 }, { "epoch": 58.885140562249, "grad_norm": 57.040550231933594, "learning_rate": 2.053225806451613e-05, "loss": 0.3205, "step": 9135 }, { "epoch": 58.89156626506024, "grad_norm": 9.817535400390625, "learning_rate": 2.052903225806452e-05, "loss": 0.4354, "step": 9136 }, { "epoch": 58.897991967871484, "grad_norm": 4.816279888153076, "learning_rate": 2.0525806451612904e-05, "loss": 0.3228, "step": 9137 }, { "epoch": 58.90441767068273, "grad_norm": 9.161266326904297, "learning_rate": 2.052258064516129e-05, "loss": 0.3497, "step": 9138 }, { "epoch": 58.910843373493975, "grad_norm": 9.228181838989258, "learning_rate": 2.051935483870968e-05, "loss": 0.3052, "step": 9139 }, { "epoch": 58.91726907630522, "grad_norm": 8.775249481201172, "learning_rate": 2.0516129032258065e-05, "loss": 0.2751, "step": 9140 }, { "epoch": 58.92369477911647, "grad_norm": 30.777456283569336, "learning_rate": 2.051290322580645e-05, "loss": 0.2811, "step": 9141 }, { "epoch": 58.93012048192771, "grad_norm": 5.078250408172607, "learning_rate": 2.050967741935484e-05, "loss": 0.3931, "step": 9142 }, { "epoch": 58.93654618473896, "grad_norm": 4.560163974761963, "learning_rate": 2.0506451612903225e-05, "loss": 0.1516, "step": 9143 }, { "epoch": 58.9429718875502, "grad_norm": 11.004239082336426, "learning_rate": 2.0503225806451615e-05, "loss": 0.1852, "step": 9144 }, { "epoch": 58.94939759036144, "grad_norm": 8.768962860107422, "learning_rate": 2.05e-05, "loss": 0.3188, "step": 9145 }, { "epoch": 58.95582329317269, "grad_norm": 13.65294075012207, "learning_rate": 2.0496774193548386e-05, "loss": 0.3793, "step": 9146 }, { "epoch": 58.962248995983934, "grad_norm": 30.412059783935547, "learning_rate": 2.0493548387096775e-05, "loss": 0.2997, "step": 9147 }, { "epoch": 58.96867469879518, "grad_norm": 4.691768169403076, "learning_rate": 2.0490322580645164e-05, "loss": 0.1784, "step": 9148 }, { "epoch": 58.975100401606426, "grad_norm": 5.138479709625244, "learning_rate": 2.048709677419355e-05, "loss": 0.1869, "step": 9149 }, { "epoch": 58.98152610441767, "grad_norm": 25.88495635986328, "learning_rate": 2.0483870967741936e-05, "loss": 0.262, "step": 9150 }, { "epoch": 58.98795180722892, "grad_norm": 4.677053928375244, "learning_rate": 2.0480645161290325e-05, "loss": 0.1978, "step": 9151 }, { "epoch": 58.99437751004016, "grad_norm": 15.915729522705078, "learning_rate": 2.047741935483871e-05, "loss": 0.2991, "step": 9152 }, { "epoch": 59.00401606425703, "grad_norm": 32.058265686035156, "learning_rate": 2.0474193548387096e-05, "loss": 0.527, "step": 9153 }, { "epoch": 59.01044176706827, "grad_norm": 4.630568981170654, "learning_rate": 2.0470967741935485e-05, "loss": 0.342, "step": 9154 }, { "epoch": 59.01686746987952, "grad_norm": 4.16118049621582, "learning_rate": 2.046774193548387e-05, "loss": 0.2512, "step": 9155 }, { "epoch": 59.023293172690764, "grad_norm": 17.208776473999023, "learning_rate": 2.046451612903226e-05, "loss": 0.6951, "step": 9156 }, { "epoch": 59.029718875502006, "grad_norm": 18.815900802612305, "learning_rate": 2.0461290322580646e-05, "loss": 0.2941, "step": 9157 }, { "epoch": 59.036144578313255, "grad_norm": 8.632052421569824, "learning_rate": 2.045806451612903e-05, "loss": 0.2446, "step": 9158 }, { "epoch": 59.0425702811245, "grad_norm": 6.119904518127441, "learning_rate": 2.045483870967742e-05, "loss": 0.1882, "step": 9159 }, { "epoch": 59.04899598393574, "grad_norm": 37.24540710449219, "learning_rate": 2.045161290322581e-05, "loss": 0.3816, "step": 9160 }, { "epoch": 59.05542168674699, "grad_norm": 10.688478469848633, "learning_rate": 2.0448387096774195e-05, "loss": 0.2776, "step": 9161 }, { "epoch": 59.06184738955823, "grad_norm": 6.652769088745117, "learning_rate": 2.044516129032258e-05, "loss": 0.3776, "step": 9162 }, { "epoch": 59.06827309236948, "grad_norm": 7.740002155303955, "learning_rate": 2.044193548387097e-05, "loss": 0.3383, "step": 9163 }, { "epoch": 59.07469879518072, "grad_norm": 15.469229698181152, "learning_rate": 2.0438709677419356e-05, "loss": 0.2709, "step": 9164 }, { "epoch": 59.081124497991965, "grad_norm": 9.187816619873047, "learning_rate": 2.043548387096774e-05, "loss": 0.3976, "step": 9165 }, { "epoch": 59.087550200803214, "grad_norm": 58.321475982666016, "learning_rate": 2.043225806451613e-05, "loss": 0.3889, "step": 9166 }, { "epoch": 59.09397590361446, "grad_norm": 15.321309089660645, "learning_rate": 2.0429032258064516e-05, "loss": 0.6316, "step": 9167 }, { "epoch": 59.100401606425706, "grad_norm": 5.067667484283447, "learning_rate": 2.0425806451612905e-05, "loss": 0.234, "step": 9168 }, { "epoch": 59.10682730923695, "grad_norm": 12.189191818237305, "learning_rate": 2.042258064516129e-05, "loss": 0.27, "step": 9169 }, { "epoch": 59.11325301204819, "grad_norm": 7.784904479980469, "learning_rate": 2.0419354838709677e-05, "loss": 0.3473, "step": 9170 }, { "epoch": 59.11967871485944, "grad_norm": 8.33542537689209, "learning_rate": 2.0416129032258066e-05, "loss": 0.5803, "step": 9171 }, { "epoch": 59.12610441767068, "grad_norm": 13.087316513061523, "learning_rate": 2.0412903225806455e-05, "loss": 0.1739, "step": 9172 }, { "epoch": 59.13253012048193, "grad_norm": 4.892680644989014, "learning_rate": 2.040967741935484e-05, "loss": 0.4842, "step": 9173 }, { "epoch": 59.13895582329317, "grad_norm": 76.5669174194336, "learning_rate": 2.0406451612903226e-05, "loss": 0.3856, "step": 9174 }, { "epoch": 59.145381526104416, "grad_norm": 20.01005744934082, "learning_rate": 2.0403225806451615e-05, "loss": 0.3628, "step": 9175 }, { "epoch": 59.151807228915665, "grad_norm": 8.392603874206543, "learning_rate": 2.04e-05, "loss": 0.2218, "step": 9176 }, { "epoch": 59.15823293172691, "grad_norm": 6.091121673583984, "learning_rate": 2.0396774193548387e-05, "loss": 0.3341, "step": 9177 }, { "epoch": 59.16465863453815, "grad_norm": 6.692813873291016, "learning_rate": 2.0393548387096776e-05, "loss": 0.5881, "step": 9178 }, { "epoch": 59.1710843373494, "grad_norm": 25.243732452392578, "learning_rate": 2.0390322580645162e-05, "loss": 0.3659, "step": 9179 }, { "epoch": 59.17751004016064, "grad_norm": 9.658066749572754, "learning_rate": 2.0387096774193547e-05, "loss": 0.3526, "step": 9180 }, { "epoch": 59.18393574297189, "grad_norm": 13.961220741271973, "learning_rate": 2.0383870967741937e-05, "loss": 0.461, "step": 9181 }, { "epoch": 59.19036144578313, "grad_norm": 94.07540893554688, "learning_rate": 2.0380645161290322e-05, "loss": 0.4136, "step": 9182 }, { "epoch": 59.196787148594375, "grad_norm": 12.138782501220703, "learning_rate": 2.037741935483871e-05, "loss": 0.1374, "step": 9183 }, { "epoch": 59.203212851405624, "grad_norm": 8.165399551391602, "learning_rate": 2.0374193548387097e-05, "loss": 0.3151, "step": 9184 }, { "epoch": 59.209638554216866, "grad_norm": 10.426656723022461, "learning_rate": 2.0370967741935483e-05, "loss": 0.5055, "step": 9185 }, { "epoch": 59.216064257028115, "grad_norm": 3.102954387664795, "learning_rate": 2.0367741935483872e-05, "loss": 0.3463, "step": 9186 }, { "epoch": 59.22248995983936, "grad_norm": 5.744695663452148, "learning_rate": 2.036451612903226e-05, "loss": 0.3373, "step": 9187 }, { "epoch": 59.2289156626506, "grad_norm": 10.432522773742676, "learning_rate": 2.0361290322580647e-05, "loss": 0.4732, "step": 9188 }, { "epoch": 59.23534136546185, "grad_norm": 19.9555721282959, "learning_rate": 2.0358064516129032e-05, "loss": 0.5699, "step": 9189 }, { "epoch": 59.24176706827309, "grad_norm": 44.09833908081055, "learning_rate": 2.035483870967742e-05, "loss": 0.3137, "step": 9190 }, { "epoch": 59.24819277108434, "grad_norm": 125.0845718383789, "learning_rate": 2.0351612903225807e-05, "loss": 0.2186, "step": 9191 }, { "epoch": 59.25461847389558, "grad_norm": 3.1696367263793945, "learning_rate": 2.0348387096774193e-05, "loss": 0.1496, "step": 9192 }, { "epoch": 59.261044176706825, "grad_norm": 15.366622924804688, "learning_rate": 2.0345161290322582e-05, "loss": 0.2383, "step": 9193 }, { "epoch": 59.267469879518075, "grad_norm": 8.954635620117188, "learning_rate": 2.0341935483870968e-05, "loss": 0.3718, "step": 9194 }, { "epoch": 59.27389558232932, "grad_norm": 28.590639114379883, "learning_rate": 2.0338709677419357e-05, "loss": 0.2697, "step": 9195 }, { "epoch": 59.28032128514056, "grad_norm": 18.365028381347656, "learning_rate": 2.0335483870967742e-05, "loss": 0.2413, "step": 9196 }, { "epoch": 59.28674698795181, "grad_norm": 12.51225471496582, "learning_rate": 2.0332258064516128e-05, "loss": 0.3614, "step": 9197 }, { "epoch": 59.29317269076305, "grad_norm": 34.09514236450195, "learning_rate": 2.0329032258064517e-05, "loss": 0.3312, "step": 9198 }, { "epoch": 59.2995983935743, "grad_norm": 16.773223876953125, "learning_rate": 2.0325806451612906e-05, "loss": 0.3738, "step": 9199 }, { "epoch": 59.30602409638554, "grad_norm": 21.301376342773438, "learning_rate": 2.0322580645161292e-05, "loss": 0.4462, "step": 9200 }, { "epoch": 59.312449799196784, "grad_norm": 26.745691299438477, "learning_rate": 2.0319354838709678e-05, "loss": 0.2153, "step": 9201 }, { "epoch": 59.318875502008034, "grad_norm": 10.933475494384766, "learning_rate": 2.0316129032258067e-05, "loss": 0.4181, "step": 9202 }, { "epoch": 59.325301204819276, "grad_norm": 7.6714935302734375, "learning_rate": 2.0312903225806453e-05, "loss": 0.4806, "step": 9203 }, { "epoch": 59.331726907630525, "grad_norm": 13.920511245727539, "learning_rate": 2.0309677419354838e-05, "loss": 0.361, "step": 9204 }, { "epoch": 59.33815261044177, "grad_norm": 14.864977836608887, "learning_rate": 2.0306451612903227e-05, "loss": 0.8068, "step": 9205 }, { "epoch": 59.34457831325301, "grad_norm": 6.159852981567383, "learning_rate": 2.0303225806451613e-05, "loss": 0.2011, "step": 9206 }, { "epoch": 59.35100401606426, "grad_norm": 18.138126373291016, "learning_rate": 2.0300000000000002e-05, "loss": 0.3285, "step": 9207 }, { "epoch": 59.3574297188755, "grad_norm": 6.654239654541016, "learning_rate": 2.0296774193548388e-05, "loss": 0.6516, "step": 9208 }, { "epoch": 59.36385542168675, "grad_norm": 7.654695510864258, "learning_rate": 2.0293548387096774e-05, "loss": 0.4331, "step": 9209 }, { "epoch": 59.37028112449799, "grad_norm": 6.340070724487305, "learning_rate": 2.0290322580645163e-05, "loss": 0.3331, "step": 9210 }, { "epoch": 59.376706827309235, "grad_norm": 13.71240234375, "learning_rate": 2.028709677419355e-05, "loss": 0.4537, "step": 9211 }, { "epoch": 59.383132530120484, "grad_norm": 54.93801498413086, "learning_rate": 2.0283870967741937e-05, "loss": 0.4698, "step": 9212 }, { "epoch": 59.389558232931726, "grad_norm": 10.04604721069336, "learning_rate": 2.0280645161290323e-05, "loss": 0.3379, "step": 9213 }, { "epoch": 59.39598393574297, "grad_norm": 10.334918022155762, "learning_rate": 2.0277419354838712e-05, "loss": 0.2755, "step": 9214 }, { "epoch": 59.40240963855422, "grad_norm": 8.16031265258789, "learning_rate": 2.0274193548387098e-05, "loss": 0.2891, "step": 9215 }, { "epoch": 59.40883534136546, "grad_norm": 23.21932029724121, "learning_rate": 2.0270967741935484e-05, "loss": 0.161, "step": 9216 }, { "epoch": 59.41526104417671, "grad_norm": 4.536672115325928, "learning_rate": 2.0267741935483873e-05, "loss": 0.1722, "step": 9217 }, { "epoch": 59.42168674698795, "grad_norm": 10.988138198852539, "learning_rate": 2.026451612903226e-05, "loss": 0.2914, "step": 9218 }, { "epoch": 59.428112449799194, "grad_norm": 13.659801483154297, "learning_rate": 2.0261290322580644e-05, "loss": 0.2631, "step": 9219 }, { "epoch": 59.43453815261044, "grad_norm": 21.380977630615234, "learning_rate": 2.0258064516129033e-05, "loss": 0.3259, "step": 9220 }, { "epoch": 59.440963855421685, "grad_norm": 4.5861639976501465, "learning_rate": 2.025483870967742e-05, "loss": 0.3957, "step": 9221 }, { "epoch": 59.447389558232935, "grad_norm": 6.142553806304932, "learning_rate": 2.0251612903225808e-05, "loss": 0.7274, "step": 9222 }, { "epoch": 59.45381526104418, "grad_norm": 25.6002140045166, "learning_rate": 2.0248387096774194e-05, "loss": 0.2716, "step": 9223 }, { "epoch": 59.46024096385542, "grad_norm": 10.356910705566406, "learning_rate": 2.0245161290322583e-05, "loss": 0.2971, "step": 9224 }, { "epoch": 59.46666666666667, "grad_norm": 9.56591510772705, "learning_rate": 2.024193548387097e-05, "loss": 0.352, "step": 9225 }, { "epoch": 59.47309236947791, "grad_norm": 9.764209747314453, "learning_rate": 2.0238709677419358e-05, "loss": 0.135, "step": 9226 }, { "epoch": 59.47951807228916, "grad_norm": 10.520560264587402, "learning_rate": 2.0235483870967743e-05, "loss": 0.1991, "step": 9227 }, { "epoch": 59.4859437751004, "grad_norm": 10.000173568725586, "learning_rate": 2.023225806451613e-05, "loss": 0.3636, "step": 9228 }, { "epoch": 59.492369477911645, "grad_norm": 17.833608627319336, "learning_rate": 2.0229032258064518e-05, "loss": 0.4178, "step": 9229 }, { "epoch": 59.498795180722894, "grad_norm": 46.96345138549805, "learning_rate": 2.0225806451612904e-05, "loss": 0.3611, "step": 9230 }, { "epoch": 59.505220883534136, "grad_norm": 19.657012939453125, "learning_rate": 2.022258064516129e-05, "loss": 0.4584, "step": 9231 }, { "epoch": 59.51164658634538, "grad_norm": 19.507356643676758, "learning_rate": 2.021935483870968e-05, "loss": 0.3371, "step": 9232 }, { "epoch": 59.51807228915663, "grad_norm": 51.15565872192383, "learning_rate": 2.0216129032258064e-05, "loss": 0.7348, "step": 9233 }, { "epoch": 59.52449799196787, "grad_norm": 13.263320922851562, "learning_rate": 2.0212903225806454e-05, "loss": 0.436, "step": 9234 }, { "epoch": 59.53092369477912, "grad_norm": 30.70184326171875, "learning_rate": 2.020967741935484e-05, "loss": 0.5636, "step": 9235 }, { "epoch": 59.53734939759036, "grad_norm": 18.529268264770508, "learning_rate": 2.020645161290323e-05, "loss": 0.3429, "step": 9236 }, { "epoch": 59.543775100401604, "grad_norm": 41.163238525390625, "learning_rate": 2.0203225806451614e-05, "loss": 0.376, "step": 9237 }, { "epoch": 59.55020080321285, "grad_norm": 12.305435180664062, "learning_rate": 2.0200000000000003e-05, "loss": 0.2476, "step": 9238 }, { "epoch": 59.556626506024095, "grad_norm": 6.893301486968994, "learning_rate": 2.019677419354839e-05, "loss": 0.4593, "step": 9239 }, { "epoch": 59.563052208835344, "grad_norm": 19.061065673828125, "learning_rate": 2.0193548387096775e-05, "loss": 0.2235, "step": 9240 }, { "epoch": 59.56947791164659, "grad_norm": 8.193607330322266, "learning_rate": 2.0190322580645164e-05, "loss": 0.3083, "step": 9241 }, { "epoch": 59.57590361445783, "grad_norm": 9.685818672180176, "learning_rate": 2.018709677419355e-05, "loss": 0.2541, "step": 9242 }, { "epoch": 59.58232931726908, "grad_norm": 11.364033699035645, "learning_rate": 2.0183870967741935e-05, "loss": 0.0895, "step": 9243 }, { "epoch": 59.58875502008032, "grad_norm": 23.400043487548828, "learning_rate": 2.0180645161290324e-05, "loss": 0.4836, "step": 9244 }, { "epoch": 59.59518072289157, "grad_norm": 12.60353946685791, "learning_rate": 2.017741935483871e-05, "loss": 0.4098, "step": 9245 }, { "epoch": 59.60160642570281, "grad_norm": 4.720575332641602, "learning_rate": 2.0174193548387096e-05, "loss": 0.3315, "step": 9246 }, { "epoch": 59.608032128514054, "grad_norm": 5.993205547332764, "learning_rate": 2.0170967741935485e-05, "loss": 0.2367, "step": 9247 }, { "epoch": 59.6144578313253, "grad_norm": 6.8719964027404785, "learning_rate": 2.016774193548387e-05, "loss": 0.2712, "step": 9248 }, { "epoch": 59.620883534136546, "grad_norm": 20.597522735595703, "learning_rate": 2.016451612903226e-05, "loss": 0.2087, "step": 9249 }, { "epoch": 59.62730923694779, "grad_norm": 6.489467620849609, "learning_rate": 2.0161290322580645e-05, "loss": 0.2273, "step": 9250 }, { "epoch": 59.63373493975904, "grad_norm": 2.1704649925231934, "learning_rate": 2.0158064516129034e-05, "loss": 0.3575, "step": 9251 }, { "epoch": 59.64016064257028, "grad_norm": 11.05473804473877, "learning_rate": 2.015483870967742e-05, "loss": 0.3566, "step": 9252 }, { "epoch": 59.64658634538153, "grad_norm": 25.671863555908203, "learning_rate": 2.015161290322581e-05, "loss": 0.2865, "step": 9253 }, { "epoch": 59.65301204819277, "grad_norm": 3.341527223587036, "learning_rate": 2.0148387096774195e-05, "loss": 0.3156, "step": 9254 }, { "epoch": 59.65943775100401, "grad_norm": 14.57736873626709, "learning_rate": 2.014516129032258e-05, "loss": 0.3585, "step": 9255 }, { "epoch": 59.66586345381526, "grad_norm": 10.28858757019043, "learning_rate": 2.014193548387097e-05, "loss": 0.2505, "step": 9256 }, { "epoch": 59.672289156626505, "grad_norm": 17.76874542236328, "learning_rate": 2.0138709677419355e-05, "loss": 0.4328, "step": 9257 }, { "epoch": 59.678714859437754, "grad_norm": 11.477797508239746, "learning_rate": 2.013548387096774e-05, "loss": 0.3393, "step": 9258 }, { "epoch": 59.685140562248996, "grad_norm": 12.378521919250488, "learning_rate": 2.013225806451613e-05, "loss": 0.2109, "step": 9259 }, { "epoch": 59.69156626506024, "grad_norm": 13.138596534729004, "learning_rate": 2.0129032258064516e-05, "loss": 0.4352, "step": 9260 }, { "epoch": 59.69799196787149, "grad_norm": 19.40541648864746, "learning_rate": 2.0125806451612905e-05, "loss": 0.3089, "step": 9261 }, { "epoch": 59.70441767068273, "grad_norm": 7.590924263000488, "learning_rate": 2.012258064516129e-05, "loss": 0.3289, "step": 9262 }, { "epoch": 59.71084337349397, "grad_norm": 9.144556045532227, "learning_rate": 2.011935483870968e-05, "loss": 0.3028, "step": 9263 }, { "epoch": 59.71726907630522, "grad_norm": 4.428866386413574, "learning_rate": 2.0116129032258065e-05, "loss": 0.3239, "step": 9264 }, { "epoch": 59.723694779116464, "grad_norm": 23.803325653076172, "learning_rate": 2.0112903225806454e-05, "loss": 0.2971, "step": 9265 }, { "epoch": 59.73012048192771, "grad_norm": 7.384566783905029, "learning_rate": 2.010967741935484e-05, "loss": 0.1683, "step": 9266 }, { "epoch": 59.736546184738955, "grad_norm": 12.385125160217285, "learning_rate": 2.0106451612903226e-05, "loss": 0.4217, "step": 9267 }, { "epoch": 59.7429718875502, "grad_norm": 12.828734397888184, "learning_rate": 2.0103225806451615e-05, "loss": 0.1955, "step": 9268 }, { "epoch": 59.74939759036145, "grad_norm": 9.780828475952148, "learning_rate": 2.01e-05, "loss": 0.4291, "step": 9269 }, { "epoch": 59.75582329317269, "grad_norm": 6.246034622192383, "learning_rate": 2.0096774193548386e-05, "loss": 0.2728, "step": 9270 }, { "epoch": 59.76224899598394, "grad_norm": 28.157848358154297, "learning_rate": 2.0093548387096775e-05, "loss": 0.577, "step": 9271 }, { "epoch": 59.76867469879518, "grad_norm": 13.088212013244629, "learning_rate": 2.009032258064516e-05, "loss": 0.4332, "step": 9272 }, { "epoch": 59.77510040160642, "grad_norm": 3.1321074962615967, "learning_rate": 2.008709677419355e-05, "loss": 0.2144, "step": 9273 }, { "epoch": 59.78152610441767, "grad_norm": 1.8144793510437012, "learning_rate": 2.0083870967741936e-05, "loss": 0.1966, "step": 9274 }, { "epoch": 59.787951807228914, "grad_norm": 6.29011344909668, "learning_rate": 2.0080645161290325e-05, "loss": 0.5706, "step": 9275 }, { "epoch": 59.794377510040164, "grad_norm": 16.1497859954834, "learning_rate": 2.007741935483871e-05, "loss": 0.4211, "step": 9276 }, { "epoch": 59.800803212851406, "grad_norm": 11.51296615600586, "learning_rate": 2.00741935483871e-05, "loss": 0.6079, "step": 9277 }, { "epoch": 59.80722891566265, "grad_norm": 3.6844375133514404, "learning_rate": 2.0070967741935486e-05, "loss": 0.2111, "step": 9278 }, { "epoch": 59.8136546184739, "grad_norm": 4.0074872970581055, "learning_rate": 2.006774193548387e-05, "loss": 0.0932, "step": 9279 }, { "epoch": 59.82008032128514, "grad_norm": 4.107634544372559, "learning_rate": 2.006451612903226e-05, "loss": 0.1026, "step": 9280 }, { "epoch": 59.82650602409639, "grad_norm": 5.682419776916504, "learning_rate": 2.0061290322580646e-05, "loss": 0.3503, "step": 9281 }, { "epoch": 59.83293172690763, "grad_norm": 8.482921600341797, "learning_rate": 2.0058064516129032e-05, "loss": 0.3933, "step": 9282 }, { "epoch": 59.83935742971887, "grad_norm": 10.187932014465332, "learning_rate": 2.005483870967742e-05, "loss": 0.4039, "step": 9283 }, { "epoch": 59.84578313253012, "grad_norm": 15.187365531921387, "learning_rate": 2.0051612903225807e-05, "loss": 0.3342, "step": 9284 }, { "epoch": 59.852208835341365, "grad_norm": 4.51331901550293, "learning_rate": 2.0048387096774192e-05, "loss": 0.3036, "step": 9285 }, { "epoch": 59.85863453815261, "grad_norm": 4.661794662475586, "learning_rate": 2.004516129032258e-05, "loss": 0.3546, "step": 9286 }, { "epoch": 59.86506024096386, "grad_norm": 9.888216972351074, "learning_rate": 2.004193548387097e-05, "loss": 0.4052, "step": 9287 }, { "epoch": 59.8714859437751, "grad_norm": 9.704700469970703, "learning_rate": 2.0038709677419356e-05, "loss": 0.1687, "step": 9288 }, { "epoch": 59.87791164658635, "grad_norm": 11.151368141174316, "learning_rate": 2.0035483870967742e-05, "loss": 0.2161, "step": 9289 }, { "epoch": 59.88433734939759, "grad_norm": 21.456745147705078, "learning_rate": 2.003225806451613e-05, "loss": 0.3703, "step": 9290 }, { "epoch": 59.89076305220883, "grad_norm": 10.577315330505371, "learning_rate": 2.0029032258064517e-05, "loss": 0.7826, "step": 9291 }, { "epoch": 59.89718875502008, "grad_norm": 2.5193750858306885, "learning_rate": 2.0025806451612906e-05, "loss": 0.2451, "step": 9292 }, { "epoch": 59.903614457831324, "grad_norm": 16.4716854095459, "learning_rate": 2.002258064516129e-05, "loss": 0.4223, "step": 9293 }, { "epoch": 59.91004016064257, "grad_norm": 7.930906772613525, "learning_rate": 2.0019354838709677e-05, "loss": 0.3699, "step": 9294 }, { "epoch": 59.916465863453816, "grad_norm": 21.169593811035156, "learning_rate": 2.0016129032258066e-05, "loss": 0.5019, "step": 9295 }, { "epoch": 59.92289156626506, "grad_norm": 36.64851760864258, "learning_rate": 2.0012903225806452e-05, "loss": 0.3843, "step": 9296 }, { "epoch": 59.92931726907631, "grad_norm": 11.05106258392334, "learning_rate": 2.0009677419354838e-05, "loss": 0.3023, "step": 9297 }, { "epoch": 59.93574297188755, "grad_norm": 24.166126251220703, "learning_rate": 2.0006451612903227e-05, "loss": 0.3805, "step": 9298 }, { "epoch": 59.94216867469879, "grad_norm": 19.40261459350586, "learning_rate": 2.0003225806451616e-05, "loss": 0.3827, "step": 9299 }, { "epoch": 59.94859437751004, "grad_norm": 6.838233947753906, "learning_rate": 2e-05, "loss": 0.5694, "step": 9300 }, { "epoch": 59.95502008032128, "grad_norm": 9.272538185119629, "learning_rate": 1.9996774193548387e-05, "loss": 0.3051, "step": 9301 }, { "epoch": 59.96144578313253, "grad_norm": 4.955071926116943, "learning_rate": 1.9993548387096776e-05, "loss": 0.2046, "step": 9302 }, { "epoch": 59.967871485943775, "grad_norm": 13.013553619384766, "learning_rate": 1.9990322580645162e-05, "loss": 0.3814, "step": 9303 }, { "epoch": 59.97429718875502, "grad_norm": 16.01833724975586, "learning_rate": 1.998709677419355e-05, "loss": 0.3157, "step": 9304 }, { "epoch": 59.980722891566266, "grad_norm": 9.672798156738281, "learning_rate": 1.9983870967741937e-05, "loss": 0.157, "step": 9305 }, { "epoch": 59.98714859437751, "grad_norm": 14.965426445007324, "learning_rate": 1.9980645161290323e-05, "loss": 0.4102, "step": 9306 }, { "epoch": 59.99357429718876, "grad_norm": 17.4371395111084, "learning_rate": 1.9977419354838712e-05, "loss": 0.2637, "step": 9307 }, { "epoch": 60.00321285140562, "grad_norm": 13.365357398986816, "learning_rate": 1.9974193548387097e-05, "loss": 0.3287, "step": 9308 }, { "epoch": 60.00963855421687, "grad_norm": 15.162031173706055, "learning_rate": 1.9970967741935483e-05, "loss": 0.4021, "step": 9309 }, { "epoch": 60.01606425702811, "grad_norm": 4.42145299911499, "learning_rate": 1.9967741935483872e-05, "loss": 0.1488, "step": 9310 }, { "epoch": 60.022489959839355, "grad_norm": 10.98394775390625, "learning_rate": 1.9964516129032258e-05, "loss": 0.5063, "step": 9311 }, { "epoch": 60.028915662650604, "grad_norm": 18.501277923583984, "learning_rate": 1.9961290322580647e-05, "loss": 0.2824, "step": 9312 }, { "epoch": 60.035341365461846, "grad_norm": 3.588265895843506, "learning_rate": 1.9958064516129033e-05, "loss": 0.2624, "step": 9313 }, { "epoch": 60.041767068273096, "grad_norm": 14.163512229919434, "learning_rate": 1.9954838709677422e-05, "loss": 0.2819, "step": 9314 }, { "epoch": 60.04819277108434, "grad_norm": 19.995464324951172, "learning_rate": 1.9951612903225808e-05, "loss": 0.5975, "step": 9315 }, { "epoch": 60.05461847389558, "grad_norm": 11.591009140014648, "learning_rate": 1.9948387096774193e-05, "loss": 0.1454, "step": 9316 }, { "epoch": 60.06104417670683, "grad_norm": 9.983513832092285, "learning_rate": 1.9945161290322582e-05, "loss": 0.0958, "step": 9317 }, { "epoch": 60.06746987951807, "grad_norm": 20.428842544555664, "learning_rate": 1.9941935483870968e-05, "loss": 0.4025, "step": 9318 }, { "epoch": 60.073895582329314, "grad_norm": 7.049335956573486, "learning_rate": 1.9938709677419357e-05, "loss": 0.3437, "step": 9319 }, { "epoch": 60.08032128514056, "grad_norm": 5.66385555267334, "learning_rate": 1.9935483870967743e-05, "loss": 0.0858, "step": 9320 }, { "epoch": 60.086746987951805, "grad_norm": 35.450782775878906, "learning_rate": 1.993225806451613e-05, "loss": 0.3804, "step": 9321 }, { "epoch": 60.093172690763055, "grad_norm": 14.838239669799805, "learning_rate": 1.9929032258064518e-05, "loss": 0.3041, "step": 9322 }, { "epoch": 60.0995983935743, "grad_norm": 18.232765197753906, "learning_rate": 1.9925806451612903e-05, "loss": 0.3222, "step": 9323 }, { "epoch": 60.10602409638554, "grad_norm": 4.33119535446167, "learning_rate": 1.992258064516129e-05, "loss": 0.2647, "step": 9324 }, { "epoch": 60.11244979919679, "grad_norm": 8.910563468933105, "learning_rate": 1.9919354838709678e-05, "loss": 0.5906, "step": 9325 }, { "epoch": 60.11887550200803, "grad_norm": 6.362738609313965, "learning_rate": 1.9916129032258067e-05, "loss": 0.2111, "step": 9326 }, { "epoch": 60.12530120481928, "grad_norm": 6.24000358581543, "learning_rate": 1.9912903225806453e-05, "loss": 0.2776, "step": 9327 }, { "epoch": 60.13172690763052, "grad_norm": 16.57542610168457, "learning_rate": 1.990967741935484e-05, "loss": 0.3155, "step": 9328 }, { "epoch": 60.138152610441765, "grad_norm": 26.572423934936523, "learning_rate": 1.9906451612903228e-05, "loss": 0.3613, "step": 9329 }, { "epoch": 60.144578313253014, "grad_norm": 8.437740325927734, "learning_rate": 1.9903225806451613e-05, "loss": 0.3619, "step": 9330 }, { "epoch": 60.151004016064256, "grad_norm": 8.93577766418457, "learning_rate": 1.9900000000000003e-05, "loss": 0.2259, "step": 9331 }, { "epoch": 60.157429718875505, "grad_norm": 14.028823852539062, "learning_rate": 1.9896774193548388e-05, "loss": 0.187, "step": 9332 }, { "epoch": 60.16385542168675, "grad_norm": 8.623002052307129, "learning_rate": 1.9893548387096774e-05, "loss": 0.1905, "step": 9333 }, { "epoch": 60.17028112449799, "grad_norm": 7.695211887359619, "learning_rate": 1.9890322580645163e-05, "loss": 0.1821, "step": 9334 }, { "epoch": 60.17670682730924, "grad_norm": 12.807120323181152, "learning_rate": 1.988709677419355e-05, "loss": 0.5289, "step": 9335 }, { "epoch": 60.18313253012048, "grad_norm": 1.9192783832550049, "learning_rate": 1.9883870967741935e-05, "loss": 0.1989, "step": 9336 }, { "epoch": 60.189558232931724, "grad_norm": 7.782166004180908, "learning_rate": 1.9880645161290324e-05, "loss": 0.4129, "step": 9337 }, { "epoch": 60.19598393574297, "grad_norm": 15.92729377746582, "learning_rate": 1.9877419354838713e-05, "loss": 0.405, "step": 9338 }, { "epoch": 60.202409638554215, "grad_norm": 10.257293701171875, "learning_rate": 1.98741935483871e-05, "loss": 0.4805, "step": 9339 }, { "epoch": 60.208835341365464, "grad_norm": 12.483553886413574, "learning_rate": 1.9870967741935484e-05, "loss": 0.4112, "step": 9340 }, { "epoch": 60.21526104417671, "grad_norm": 20.318397521972656, "learning_rate": 1.9867741935483873e-05, "loss": 0.3908, "step": 9341 }, { "epoch": 60.22168674698795, "grad_norm": 18.056350708007812, "learning_rate": 1.986451612903226e-05, "loss": 0.4139, "step": 9342 }, { "epoch": 60.2281124497992, "grad_norm": 17.556861877441406, "learning_rate": 1.9861290322580648e-05, "loss": 0.3479, "step": 9343 }, { "epoch": 60.23453815261044, "grad_norm": 34.50197219848633, "learning_rate": 1.9858064516129034e-05, "loss": 0.2878, "step": 9344 }, { "epoch": 60.24096385542169, "grad_norm": 5.38253116607666, "learning_rate": 1.985483870967742e-05, "loss": 0.3593, "step": 9345 }, { "epoch": 60.24738955823293, "grad_norm": 4.190635681152344, "learning_rate": 1.985161290322581e-05, "loss": 0.1603, "step": 9346 }, { "epoch": 60.253815261044174, "grad_norm": 11.066614151000977, "learning_rate": 1.9848387096774194e-05, "loss": 0.3783, "step": 9347 }, { "epoch": 60.26024096385542, "grad_norm": 8.19511604309082, "learning_rate": 1.984516129032258e-05, "loss": 0.3397, "step": 9348 }, { "epoch": 60.266666666666666, "grad_norm": 12.144669532775879, "learning_rate": 1.984193548387097e-05, "loss": 0.3616, "step": 9349 }, { "epoch": 60.273092369477915, "grad_norm": 31.826047897338867, "learning_rate": 1.9838709677419358e-05, "loss": 0.1955, "step": 9350 }, { "epoch": 60.27951807228916, "grad_norm": 3.9482359886169434, "learning_rate": 1.983548387096774e-05, "loss": 0.2069, "step": 9351 }, { "epoch": 60.2859437751004, "grad_norm": 17.920564651489258, "learning_rate": 1.983225806451613e-05, "loss": 0.3491, "step": 9352 }, { "epoch": 60.29236947791165, "grad_norm": 20.23564338684082, "learning_rate": 1.982903225806452e-05, "loss": 0.41, "step": 9353 }, { "epoch": 60.29879518072289, "grad_norm": 11.578191757202148, "learning_rate": 1.9825806451612904e-05, "loss": 0.2416, "step": 9354 }, { "epoch": 60.30522088353413, "grad_norm": 11.47775650024414, "learning_rate": 1.982258064516129e-05, "loss": 0.3227, "step": 9355 }, { "epoch": 60.31164658634538, "grad_norm": 19.78510093688965, "learning_rate": 1.981935483870968e-05, "loss": 0.5851, "step": 9356 }, { "epoch": 60.318072289156625, "grad_norm": 7.5485124588012695, "learning_rate": 1.9816129032258065e-05, "loss": 0.3399, "step": 9357 }, { "epoch": 60.324497991967874, "grad_norm": 7.536389350891113, "learning_rate": 1.9812903225806454e-05, "loss": 0.568, "step": 9358 }, { "epoch": 60.330923694779116, "grad_norm": 20.729055404663086, "learning_rate": 1.980967741935484e-05, "loss": 0.339, "step": 9359 }, { "epoch": 60.33734939759036, "grad_norm": 15.462143898010254, "learning_rate": 1.9806451612903225e-05, "loss": 0.6116, "step": 9360 }, { "epoch": 60.34377510040161, "grad_norm": 18.558147430419922, "learning_rate": 1.9803225806451614e-05, "loss": 0.3744, "step": 9361 }, { "epoch": 60.35020080321285, "grad_norm": 76.81745910644531, "learning_rate": 1.9800000000000004e-05, "loss": 0.3237, "step": 9362 }, { "epoch": 60.3566265060241, "grad_norm": 15.331165313720703, "learning_rate": 1.9796774193548386e-05, "loss": 0.6061, "step": 9363 }, { "epoch": 60.36305220883534, "grad_norm": 6.535224914550781, "learning_rate": 1.9793548387096775e-05, "loss": 0.3566, "step": 9364 }, { "epoch": 60.369477911646584, "grad_norm": 18.09334945678711, "learning_rate": 1.9790322580645164e-05, "loss": 0.5265, "step": 9365 }, { "epoch": 60.37590361445783, "grad_norm": 14.047391891479492, "learning_rate": 1.978709677419355e-05, "loss": 0.4094, "step": 9366 }, { "epoch": 60.382329317269075, "grad_norm": 14.468518257141113, "learning_rate": 1.9783870967741935e-05, "loss": 0.1305, "step": 9367 }, { "epoch": 60.388755020080325, "grad_norm": 14.920417785644531, "learning_rate": 1.9780645161290325e-05, "loss": 0.3372, "step": 9368 }, { "epoch": 60.39518072289157, "grad_norm": 9.277619361877441, "learning_rate": 1.977741935483871e-05, "loss": 0.3221, "step": 9369 }, { "epoch": 60.40160642570281, "grad_norm": 10.75175666809082, "learning_rate": 1.97741935483871e-05, "loss": 0.3315, "step": 9370 }, { "epoch": 60.40803212851406, "grad_norm": 11.533734321594238, "learning_rate": 1.9770967741935485e-05, "loss": 0.3719, "step": 9371 }, { "epoch": 60.4144578313253, "grad_norm": 5.460799694061279, "learning_rate": 1.976774193548387e-05, "loss": 0.1184, "step": 9372 }, { "epoch": 60.42088353413654, "grad_norm": 5.037088871002197, "learning_rate": 1.976451612903226e-05, "loss": 0.2566, "step": 9373 }, { "epoch": 60.42730923694779, "grad_norm": 23.759326934814453, "learning_rate": 1.9761290322580646e-05, "loss": 0.2469, "step": 9374 }, { "epoch": 60.433734939759034, "grad_norm": 6.2601637840271, "learning_rate": 1.975806451612903e-05, "loss": 0.2436, "step": 9375 }, { "epoch": 60.440160642570284, "grad_norm": 16.92035675048828, "learning_rate": 1.975483870967742e-05, "loss": 0.584, "step": 9376 }, { "epoch": 60.446586345381526, "grad_norm": 5.768692493438721, "learning_rate": 1.975161290322581e-05, "loss": 0.4617, "step": 9377 }, { "epoch": 60.45301204819277, "grad_norm": 41.99406051635742, "learning_rate": 1.9748387096774195e-05, "loss": 0.3216, "step": 9378 }, { "epoch": 60.45943775100402, "grad_norm": 9.35854721069336, "learning_rate": 1.974516129032258e-05, "loss": 0.2841, "step": 9379 }, { "epoch": 60.46586345381526, "grad_norm": 9.281303405761719, "learning_rate": 1.974193548387097e-05, "loss": 0.3963, "step": 9380 }, { "epoch": 60.47228915662651, "grad_norm": 2.704362154006958, "learning_rate": 1.9738709677419356e-05, "loss": 0.1198, "step": 9381 }, { "epoch": 60.47871485943775, "grad_norm": 5.618377685546875, "learning_rate": 1.9735483870967745e-05, "loss": 0.3112, "step": 9382 }, { "epoch": 60.48514056224899, "grad_norm": 8.564105033874512, "learning_rate": 1.973225806451613e-05, "loss": 0.4218, "step": 9383 }, { "epoch": 60.49156626506024, "grad_norm": 2.6416707038879395, "learning_rate": 1.9729032258064516e-05, "loss": 0.2417, "step": 9384 }, { "epoch": 60.497991967871485, "grad_norm": 23.868412017822266, "learning_rate": 1.9725806451612905e-05, "loss": 0.2051, "step": 9385 }, { "epoch": 60.504417670682734, "grad_norm": 3.3882405757904053, "learning_rate": 1.972258064516129e-05, "loss": 0.2334, "step": 9386 }, { "epoch": 60.51084337349398, "grad_norm": 15.30805492401123, "learning_rate": 1.9719354838709677e-05, "loss": 0.1729, "step": 9387 }, { "epoch": 60.51726907630522, "grad_norm": 12.87706470489502, "learning_rate": 1.9716129032258066e-05, "loss": 0.4501, "step": 9388 }, { "epoch": 60.52369477911647, "grad_norm": 8.88161849975586, "learning_rate": 1.9712903225806455e-05, "loss": 0.3979, "step": 9389 }, { "epoch": 60.53012048192771, "grad_norm": 8.245670318603516, "learning_rate": 1.9709677419354837e-05, "loss": 0.3618, "step": 9390 }, { "epoch": 60.53654618473895, "grad_norm": 7.632021903991699, "learning_rate": 1.9706451612903226e-05, "loss": 0.2217, "step": 9391 }, { "epoch": 60.5429718875502, "grad_norm": 9.017716407775879, "learning_rate": 1.9703225806451615e-05, "loss": 0.2344, "step": 9392 }, { "epoch": 60.549397590361444, "grad_norm": 14.28629207611084, "learning_rate": 1.97e-05, "loss": 0.4032, "step": 9393 }, { "epoch": 60.55582329317269, "grad_norm": 28.120223999023438, "learning_rate": 1.9696774193548387e-05, "loss": 0.4107, "step": 9394 }, { "epoch": 60.562248995983936, "grad_norm": 11.623565673828125, "learning_rate": 1.9693548387096776e-05, "loss": 0.2654, "step": 9395 }, { "epoch": 60.56867469879518, "grad_norm": 9.6266508102417, "learning_rate": 1.969032258064516e-05, "loss": 0.367, "step": 9396 }, { "epoch": 60.57510040160643, "grad_norm": 28.703737258911133, "learning_rate": 1.968709677419355e-05, "loss": 0.5614, "step": 9397 }, { "epoch": 60.58152610441767, "grad_norm": 12.736942291259766, "learning_rate": 1.9683870967741936e-05, "loss": 0.4889, "step": 9398 }, { "epoch": 60.58795180722892, "grad_norm": 1.9867396354675293, "learning_rate": 1.9680645161290322e-05, "loss": 0.1748, "step": 9399 }, { "epoch": 60.59437751004016, "grad_norm": 5.394249439239502, "learning_rate": 1.967741935483871e-05, "loss": 0.3337, "step": 9400 }, { "epoch": 60.6008032128514, "grad_norm": 22.235496520996094, "learning_rate": 1.96741935483871e-05, "loss": 0.4973, "step": 9401 }, { "epoch": 60.60722891566265, "grad_norm": 25.78319549560547, "learning_rate": 1.9670967741935483e-05, "loss": 0.5021, "step": 9402 }, { "epoch": 60.613654618473895, "grad_norm": 3.7703003883361816, "learning_rate": 1.9667741935483872e-05, "loss": 0.37, "step": 9403 }, { "epoch": 60.620080321285144, "grad_norm": 15.169534683227539, "learning_rate": 1.966451612903226e-05, "loss": 0.3757, "step": 9404 }, { "epoch": 60.626506024096386, "grad_norm": 8.472518920898438, "learning_rate": 1.9661290322580647e-05, "loss": 0.2725, "step": 9405 }, { "epoch": 60.63293172690763, "grad_norm": 12.222375869750977, "learning_rate": 1.9658064516129032e-05, "loss": 0.5031, "step": 9406 }, { "epoch": 60.63935742971888, "grad_norm": 6.5167412757873535, "learning_rate": 1.965483870967742e-05, "loss": 0.3457, "step": 9407 }, { "epoch": 60.64578313253012, "grad_norm": 17.629520416259766, "learning_rate": 1.9651612903225807e-05, "loss": 0.2177, "step": 9408 }, { "epoch": 60.65220883534136, "grad_norm": 39.50954055786133, "learning_rate": 1.9648387096774196e-05, "loss": 0.5744, "step": 9409 }, { "epoch": 60.65863453815261, "grad_norm": 4.0484619140625, "learning_rate": 1.9645161290322582e-05, "loss": 0.2436, "step": 9410 }, { "epoch": 60.665060240963854, "grad_norm": 6.473830699920654, "learning_rate": 1.9641935483870968e-05, "loss": 0.5005, "step": 9411 }, { "epoch": 60.6714859437751, "grad_norm": 6.9217658042907715, "learning_rate": 1.9638709677419357e-05, "loss": 0.3176, "step": 9412 }, { "epoch": 60.677911646586345, "grad_norm": 44.11043167114258, "learning_rate": 1.9635483870967746e-05, "loss": 0.503, "step": 9413 }, { "epoch": 60.68433734939759, "grad_norm": 7.895200252532959, "learning_rate": 1.9632258064516128e-05, "loss": 0.1875, "step": 9414 }, { "epoch": 60.69076305220884, "grad_norm": 12.242607116699219, "learning_rate": 1.9629032258064517e-05, "loss": 0.2236, "step": 9415 }, { "epoch": 60.69718875502008, "grad_norm": 18.12982940673828, "learning_rate": 1.9625806451612906e-05, "loss": 0.3243, "step": 9416 }, { "epoch": 60.70361445783133, "grad_norm": 21.7982234954834, "learning_rate": 1.9622580645161292e-05, "loss": 0.4143, "step": 9417 }, { "epoch": 60.71004016064257, "grad_norm": 5.568426132202148, "learning_rate": 1.9619354838709678e-05, "loss": 0.1753, "step": 9418 }, { "epoch": 60.71646586345381, "grad_norm": 7.257507801055908, "learning_rate": 1.9616129032258067e-05, "loss": 0.4524, "step": 9419 }, { "epoch": 60.72289156626506, "grad_norm": 12.766108512878418, "learning_rate": 1.9612903225806452e-05, "loss": 0.3261, "step": 9420 }, { "epoch": 60.729317269076304, "grad_norm": 21.869285583496094, "learning_rate": 1.9609677419354838e-05, "loss": 0.3699, "step": 9421 }, { "epoch": 60.735742971887554, "grad_norm": 13.876508712768555, "learning_rate": 1.9606451612903227e-05, "loss": 0.218, "step": 9422 }, { "epoch": 60.742168674698796, "grad_norm": 7.639199256896973, "learning_rate": 1.9603225806451613e-05, "loss": 0.3614, "step": 9423 }, { "epoch": 60.74859437751004, "grad_norm": 8.89919662475586, "learning_rate": 1.9600000000000002e-05, "loss": 0.3385, "step": 9424 }, { "epoch": 60.75502008032129, "grad_norm": 23.711544036865234, "learning_rate": 1.9596774193548388e-05, "loss": 0.5725, "step": 9425 }, { "epoch": 60.76144578313253, "grad_norm": 6.041403770446777, "learning_rate": 1.9593548387096773e-05, "loss": 0.1985, "step": 9426 }, { "epoch": 60.76787148594377, "grad_norm": 16.424665451049805, "learning_rate": 1.9590322580645163e-05, "loss": 0.3654, "step": 9427 }, { "epoch": 60.77429718875502, "grad_norm": 11.633688926696777, "learning_rate": 1.958709677419355e-05, "loss": 0.6126, "step": 9428 }, { "epoch": 60.78072289156626, "grad_norm": 18.818525314331055, "learning_rate": 1.9583870967741934e-05, "loss": 0.5344, "step": 9429 }, { "epoch": 60.78714859437751, "grad_norm": 11.252857208251953, "learning_rate": 1.9580645161290323e-05, "loss": 0.2431, "step": 9430 }, { "epoch": 60.793574297188755, "grad_norm": 7.627513885498047, "learning_rate": 1.9577419354838712e-05, "loss": 0.4756, "step": 9431 }, { "epoch": 60.8, "grad_norm": 7.292386531829834, "learning_rate": 1.9574193548387098e-05, "loss": 0.3698, "step": 9432 }, { "epoch": 60.80642570281125, "grad_norm": 12.094573020935059, "learning_rate": 1.9570967741935484e-05, "loss": 0.1719, "step": 9433 }, { "epoch": 60.81285140562249, "grad_norm": 13.472612380981445, "learning_rate": 1.9567741935483873e-05, "loss": 0.2635, "step": 9434 }, { "epoch": 60.81927710843374, "grad_norm": 8.319987297058105, "learning_rate": 1.956451612903226e-05, "loss": 0.3595, "step": 9435 }, { "epoch": 60.82570281124498, "grad_norm": 7.624417304992676, "learning_rate": 1.9561290322580648e-05, "loss": 0.6168, "step": 9436 }, { "epoch": 60.83212851405622, "grad_norm": 4.2403340339660645, "learning_rate": 1.9558064516129033e-05, "loss": 0.2718, "step": 9437 }, { "epoch": 60.83855421686747, "grad_norm": 18.256906509399414, "learning_rate": 1.955483870967742e-05, "loss": 0.4609, "step": 9438 }, { "epoch": 60.844979919678714, "grad_norm": 4.0248260498046875, "learning_rate": 1.9551612903225808e-05, "loss": 0.1781, "step": 9439 }, { "epoch": 60.85140562248996, "grad_norm": 11.672380447387695, "learning_rate": 1.9548387096774197e-05, "loss": 0.3496, "step": 9440 }, { "epoch": 60.857831325301206, "grad_norm": 10.179880142211914, "learning_rate": 1.954516129032258e-05, "loss": 0.2294, "step": 9441 }, { "epoch": 60.86425702811245, "grad_norm": 17.92652130126953, "learning_rate": 1.954193548387097e-05, "loss": 0.3971, "step": 9442 }, { "epoch": 60.8706827309237, "grad_norm": 8.745756149291992, "learning_rate": 1.9538709677419358e-05, "loss": 0.383, "step": 9443 }, { "epoch": 60.87710843373494, "grad_norm": 25.452007293701172, "learning_rate": 1.9535483870967743e-05, "loss": 0.7699, "step": 9444 }, { "epoch": 60.88353413654618, "grad_norm": 20.630525588989258, "learning_rate": 1.953225806451613e-05, "loss": 0.2463, "step": 9445 }, { "epoch": 60.88995983935743, "grad_norm": 8.950231552124023, "learning_rate": 1.9529032258064518e-05, "loss": 0.4841, "step": 9446 }, { "epoch": 60.89638554216867, "grad_norm": 26.68023681640625, "learning_rate": 1.9525806451612904e-05, "loss": 0.5713, "step": 9447 }, { "epoch": 60.90281124497992, "grad_norm": 7.446045398712158, "learning_rate": 1.9522580645161293e-05, "loss": 0.6533, "step": 9448 }, { "epoch": 60.909236947791165, "grad_norm": 5.440520763397217, "learning_rate": 1.951935483870968e-05, "loss": 0.2954, "step": 9449 }, { "epoch": 60.91566265060241, "grad_norm": 19.99081802368164, "learning_rate": 1.9516129032258064e-05, "loss": 0.5171, "step": 9450 }, { "epoch": 60.922088353413656, "grad_norm": 19.26583480834961, "learning_rate": 1.9512903225806453e-05, "loss": 0.6286, "step": 9451 }, { "epoch": 60.9285140562249, "grad_norm": 58.678741455078125, "learning_rate": 1.9509677419354843e-05, "loss": 0.4074, "step": 9452 }, { "epoch": 60.93493975903615, "grad_norm": 15.339599609375, "learning_rate": 1.9506451612903225e-05, "loss": 0.3695, "step": 9453 }, { "epoch": 60.94136546184739, "grad_norm": 11.52511215209961, "learning_rate": 1.9503225806451614e-05, "loss": 0.3686, "step": 9454 }, { "epoch": 60.94779116465863, "grad_norm": 32.18100357055664, "learning_rate": 1.9500000000000003e-05, "loss": 0.6175, "step": 9455 }, { "epoch": 60.95421686746988, "grad_norm": 7.146939754486084, "learning_rate": 1.949677419354839e-05, "loss": 0.2127, "step": 9456 }, { "epoch": 60.960642570281124, "grad_norm": 3.694227933883667, "learning_rate": 1.9493548387096774e-05, "loss": 0.2837, "step": 9457 }, { "epoch": 60.967068273092366, "grad_norm": 7.610121726989746, "learning_rate": 1.9490322580645164e-05, "loss": 0.2411, "step": 9458 }, { "epoch": 60.973493975903615, "grad_norm": 20.67191505432129, "learning_rate": 1.948709677419355e-05, "loss": 0.4607, "step": 9459 }, { "epoch": 60.97991967871486, "grad_norm": 20.784345626831055, "learning_rate": 1.9483870967741935e-05, "loss": 0.3585, "step": 9460 }, { "epoch": 60.98634538152611, "grad_norm": 4.99907112121582, "learning_rate": 1.9480645161290324e-05, "loss": 0.3936, "step": 9461 }, { "epoch": 60.99277108433735, "grad_norm": 6.467520713806152, "learning_rate": 1.947741935483871e-05, "loss": 0.2754, "step": 9462 }, { "epoch": 61.00240963855422, "grad_norm": 7.21141242980957, "learning_rate": 1.94741935483871e-05, "loss": 0.2084, "step": 9463 }, { "epoch": 61.00883534136546, "grad_norm": 15.530294418334961, "learning_rate": 1.9470967741935485e-05, "loss": 0.3924, "step": 9464 }, { "epoch": 61.015261044176704, "grad_norm": 12.550902366638184, "learning_rate": 1.946774193548387e-05, "loss": 0.2714, "step": 9465 }, { "epoch": 61.02168674698795, "grad_norm": 11.072308540344238, "learning_rate": 1.946451612903226e-05, "loss": 0.6821, "step": 9466 }, { "epoch": 61.028112449799195, "grad_norm": 19.398576736450195, "learning_rate": 1.946129032258065e-05, "loss": 0.3318, "step": 9467 }, { "epoch": 61.034538152610445, "grad_norm": 28.596553802490234, "learning_rate": 1.945806451612903e-05, "loss": 0.8695, "step": 9468 }, { "epoch": 61.04096385542169, "grad_norm": 13.130526542663574, "learning_rate": 1.945483870967742e-05, "loss": 0.3675, "step": 9469 }, { "epoch": 61.04738955823293, "grad_norm": 7.137133598327637, "learning_rate": 1.945161290322581e-05, "loss": 0.2253, "step": 9470 }, { "epoch": 61.05381526104418, "grad_norm": 21.438814163208008, "learning_rate": 1.9448387096774195e-05, "loss": 0.257, "step": 9471 }, { "epoch": 61.06024096385542, "grad_norm": 14.198332786560059, "learning_rate": 1.944516129032258e-05, "loss": 0.2962, "step": 9472 }, { "epoch": 61.06666666666667, "grad_norm": 3.4983136653900146, "learning_rate": 1.944193548387097e-05, "loss": 0.2417, "step": 9473 }, { "epoch": 61.07309236947791, "grad_norm": 11.65748405456543, "learning_rate": 1.9438709677419355e-05, "loss": 0.4761, "step": 9474 }, { "epoch": 61.079518072289154, "grad_norm": 7.695066928863525, "learning_rate": 1.9435483870967744e-05, "loss": 0.4617, "step": 9475 }, { "epoch": 61.085943775100404, "grad_norm": 14.416770935058594, "learning_rate": 1.943225806451613e-05, "loss": 0.4971, "step": 9476 }, { "epoch": 61.092369477911646, "grad_norm": 25.26616668701172, "learning_rate": 1.9429032258064516e-05, "loss": 0.5252, "step": 9477 }, { "epoch": 61.09879518072289, "grad_norm": 4.284099102020264, "learning_rate": 1.9425806451612905e-05, "loss": 0.1624, "step": 9478 }, { "epoch": 61.10522088353414, "grad_norm": 5.643074989318848, "learning_rate": 1.9422580645161294e-05, "loss": 0.1378, "step": 9479 }, { "epoch": 61.11164658634538, "grad_norm": 8.418787956237793, "learning_rate": 1.9419354838709676e-05, "loss": 0.8039, "step": 9480 }, { "epoch": 61.11807228915663, "grad_norm": 6.278980255126953, "learning_rate": 1.9416129032258065e-05, "loss": 0.1671, "step": 9481 }, { "epoch": 61.12449799196787, "grad_norm": 3.2343714237213135, "learning_rate": 1.9412903225806454e-05, "loss": 0.5145, "step": 9482 }, { "epoch": 61.13092369477911, "grad_norm": 17.759994506835938, "learning_rate": 1.940967741935484e-05, "loss": 0.3327, "step": 9483 }, { "epoch": 61.13734939759036, "grad_norm": 6.744078159332275, "learning_rate": 1.9406451612903226e-05, "loss": 0.3358, "step": 9484 }, { "epoch": 61.143775100401605, "grad_norm": 11.365160942077637, "learning_rate": 1.9403225806451615e-05, "loss": 0.3316, "step": 9485 }, { "epoch": 61.150200803212854, "grad_norm": 39.07899856567383, "learning_rate": 1.94e-05, "loss": 0.2712, "step": 9486 }, { "epoch": 61.1566265060241, "grad_norm": 11.237308502197266, "learning_rate": 1.939677419354839e-05, "loss": 0.3013, "step": 9487 }, { "epoch": 61.16305220883534, "grad_norm": 7.658923149108887, "learning_rate": 1.9393548387096775e-05, "loss": 0.2435, "step": 9488 }, { "epoch": 61.16947791164659, "grad_norm": 6.5088887214660645, "learning_rate": 1.939032258064516e-05, "loss": 0.436, "step": 9489 }, { "epoch": 61.17590361445783, "grad_norm": 11.75362491607666, "learning_rate": 1.938709677419355e-05, "loss": 0.3215, "step": 9490 }, { "epoch": 61.18232931726908, "grad_norm": 9.754264831542969, "learning_rate": 1.938387096774194e-05, "loss": 0.2208, "step": 9491 }, { "epoch": 61.18875502008032, "grad_norm": 6.404338359832764, "learning_rate": 1.938064516129032e-05, "loss": 0.3904, "step": 9492 }, { "epoch": 61.195180722891564, "grad_norm": 11.367752075195312, "learning_rate": 1.937741935483871e-05, "loss": 0.4816, "step": 9493 }, { "epoch": 61.20160642570281, "grad_norm": 5.692886829376221, "learning_rate": 1.93741935483871e-05, "loss": 0.4908, "step": 9494 }, { "epoch": 61.208032128514056, "grad_norm": 25.708751678466797, "learning_rate": 1.9370967741935482e-05, "loss": 0.2884, "step": 9495 }, { "epoch": 61.2144578313253, "grad_norm": 8.295884132385254, "learning_rate": 1.936774193548387e-05, "loss": 0.2691, "step": 9496 }, { "epoch": 61.22088353413655, "grad_norm": 6.284309387207031, "learning_rate": 1.936451612903226e-05, "loss": 0.662, "step": 9497 }, { "epoch": 61.22730923694779, "grad_norm": 6.995025157928467, "learning_rate": 1.9361290322580646e-05, "loss": 0.4916, "step": 9498 }, { "epoch": 61.23373493975904, "grad_norm": 17.39888572692871, "learning_rate": 1.9358064516129032e-05, "loss": 0.5167, "step": 9499 }, { "epoch": 61.24016064257028, "grad_norm": 15.65650463104248, "learning_rate": 1.935483870967742e-05, "loss": 0.2934, "step": 9500 }, { "epoch": 61.24658634538152, "grad_norm": 7.527843475341797, "learning_rate": 1.9351612903225807e-05, "loss": 0.3647, "step": 9501 }, { "epoch": 61.25301204819277, "grad_norm": 12.518664360046387, "learning_rate": 1.9348387096774196e-05, "loss": 0.4948, "step": 9502 }, { "epoch": 61.259437751004015, "grad_norm": 11.538191795349121, "learning_rate": 1.934516129032258e-05, "loss": 0.4665, "step": 9503 }, { "epoch": 61.265863453815264, "grad_norm": 11.52187728881836, "learning_rate": 1.9341935483870967e-05, "loss": 0.522, "step": 9504 }, { "epoch": 61.272289156626506, "grad_norm": 9.006144523620605, "learning_rate": 1.9338709677419356e-05, "loss": 0.238, "step": 9505 }, { "epoch": 61.27871485943775, "grad_norm": 2.991748571395874, "learning_rate": 1.9335483870967745e-05, "loss": 0.112, "step": 9506 }, { "epoch": 61.285140562249, "grad_norm": 6.447067737579346, "learning_rate": 1.9332258064516128e-05, "loss": 0.0971, "step": 9507 }, { "epoch": 61.29156626506024, "grad_norm": 7.8167243003845215, "learning_rate": 1.9329032258064517e-05, "loss": 0.228, "step": 9508 }, { "epoch": 61.29799196787149, "grad_norm": 5.219592571258545, "learning_rate": 1.9325806451612906e-05, "loss": 0.4196, "step": 9509 }, { "epoch": 61.30441767068273, "grad_norm": 18.89302635192871, "learning_rate": 1.932258064516129e-05, "loss": 0.5988, "step": 9510 }, { "epoch": 61.310843373493974, "grad_norm": 13.291326522827148, "learning_rate": 1.9319354838709677e-05, "loss": 0.5646, "step": 9511 }, { "epoch": 61.31726907630522, "grad_norm": 12.36815071105957, "learning_rate": 1.9316129032258066e-05, "loss": 0.3289, "step": 9512 }, { "epoch": 61.323694779116465, "grad_norm": 6.500981330871582, "learning_rate": 1.9312903225806452e-05, "loss": 0.2803, "step": 9513 }, { "epoch": 61.33012048192771, "grad_norm": 4.8310933113098145, "learning_rate": 1.930967741935484e-05, "loss": 0.3667, "step": 9514 }, { "epoch": 61.33654618473896, "grad_norm": 21.408260345458984, "learning_rate": 1.9306451612903227e-05, "loss": 0.4774, "step": 9515 }, { "epoch": 61.3429718875502, "grad_norm": 13.298478126525879, "learning_rate": 1.9303225806451612e-05, "loss": 0.4366, "step": 9516 }, { "epoch": 61.34939759036145, "grad_norm": 10.904467582702637, "learning_rate": 1.93e-05, "loss": 0.319, "step": 9517 }, { "epoch": 61.35582329317269, "grad_norm": 6.848258018493652, "learning_rate": 1.929677419354839e-05, "loss": 0.2532, "step": 9518 }, { "epoch": 61.36224899598393, "grad_norm": 15.374462127685547, "learning_rate": 1.9293548387096773e-05, "loss": 0.4647, "step": 9519 }, { "epoch": 61.36867469879518, "grad_norm": 12.149232864379883, "learning_rate": 1.9290322580645162e-05, "loss": 0.251, "step": 9520 }, { "epoch": 61.375100401606424, "grad_norm": 10.891679763793945, "learning_rate": 1.928709677419355e-05, "loss": 0.263, "step": 9521 }, { "epoch": 61.381526104417674, "grad_norm": 17.43327522277832, "learning_rate": 1.9283870967741937e-05, "loss": 0.5077, "step": 9522 }, { "epoch": 61.387951807228916, "grad_norm": 26.601158142089844, "learning_rate": 1.9280645161290323e-05, "loss": 0.4488, "step": 9523 }, { "epoch": 61.39437751004016, "grad_norm": 26.194307327270508, "learning_rate": 1.927741935483871e-05, "loss": 0.3291, "step": 9524 }, { "epoch": 61.40080321285141, "grad_norm": 9.560626029968262, "learning_rate": 1.9274193548387097e-05, "loss": 0.7248, "step": 9525 }, { "epoch": 61.40722891566265, "grad_norm": 16.739116668701172, "learning_rate": 1.9270967741935486e-05, "loss": 0.2527, "step": 9526 }, { "epoch": 61.4136546184739, "grad_norm": 9.241950988769531, "learning_rate": 1.9267741935483872e-05, "loss": 0.2113, "step": 9527 }, { "epoch": 61.42008032128514, "grad_norm": 41.295650482177734, "learning_rate": 1.9264516129032258e-05, "loss": 0.4377, "step": 9528 }, { "epoch": 61.42650602409638, "grad_norm": 18.407686233520508, "learning_rate": 1.9261290322580647e-05, "loss": 0.7104, "step": 9529 }, { "epoch": 61.43293172690763, "grad_norm": 17.42197036743164, "learning_rate": 1.9258064516129033e-05, "loss": 0.1698, "step": 9530 }, { "epoch": 61.439357429718875, "grad_norm": 5.024514675140381, "learning_rate": 1.925483870967742e-05, "loss": 0.178, "step": 9531 }, { "epoch": 61.44578313253012, "grad_norm": 5.508563041687012, "learning_rate": 1.9251612903225807e-05, "loss": 0.2852, "step": 9532 }, { "epoch": 61.45220883534137, "grad_norm": 21.54368782043457, "learning_rate": 1.9248387096774197e-05, "loss": 0.3036, "step": 9533 }, { "epoch": 61.45863453815261, "grad_norm": 26.56414794921875, "learning_rate": 1.924516129032258e-05, "loss": 0.2629, "step": 9534 }, { "epoch": 61.46506024096386, "grad_norm": 31.558603286743164, "learning_rate": 1.9241935483870968e-05, "loss": 0.4547, "step": 9535 }, { "epoch": 61.4714859437751, "grad_norm": 14.988080978393555, "learning_rate": 1.9238709677419357e-05, "loss": 0.3115, "step": 9536 }, { "epoch": 61.47791164658634, "grad_norm": 13.038777351379395, "learning_rate": 1.9235483870967743e-05, "loss": 0.5211, "step": 9537 }, { "epoch": 61.48433734939759, "grad_norm": 10.480180740356445, "learning_rate": 1.923225806451613e-05, "loss": 0.327, "step": 9538 }, { "epoch": 61.490763052208834, "grad_norm": 12.662388801574707, "learning_rate": 1.9229032258064518e-05, "loss": 0.5676, "step": 9539 }, { "epoch": 61.49718875502008, "grad_norm": 4.1775336265563965, "learning_rate": 1.9225806451612903e-05, "loss": 0.1343, "step": 9540 }, { "epoch": 61.503614457831326, "grad_norm": 2.7760393619537354, "learning_rate": 1.9222580645161292e-05, "loss": 0.1828, "step": 9541 }, { "epoch": 61.51004016064257, "grad_norm": 130.70083618164062, "learning_rate": 1.9219354838709678e-05, "loss": 0.4521, "step": 9542 }, { "epoch": 61.51646586345382, "grad_norm": 8.472541809082031, "learning_rate": 1.9216129032258064e-05, "loss": 0.4612, "step": 9543 }, { "epoch": 61.52289156626506, "grad_norm": 13.911072731018066, "learning_rate": 1.9212903225806453e-05, "loss": 0.5107, "step": 9544 }, { "epoch": 61.52931726907631, "grad_norm": 8.122194290161133, "learning_rate": 1.9209677419354842e-05, "loss": 0.2592, "step": 9545 }, { "epoch": 61.53574297188755, "grad_norm": 38.17155075073242, "learning_rate": 1.9206451612903224e-05, "loss": 0.4724, "step": 9546 }, { "epoch": 61.54216867469879, "grad_norm": 7.693421840667725, "learning_rate": 1.9203225806451613e-05, "loss": 0.4182, "step": 9547 }, { "epoch": 61.54859437751004, "grad_norm": 8.661027908325195, "learning_rate": 1.9200000000000003e-05, "loss": 0.4307, "step": 9548 }, { "epoch": 61.555020080321285, "grad_norm": 6.620832443237305, "learning_rate": 1.9196774193548388e-05, "loss": 0.3062, "step": 9549 }, { "epoch": 61.56144578313253, "grad_norm": 16.646377563476562, "learning_rate": 1.9193548387096774e-05, "loss": 0.4645, "step": 9550 }, { "epoch": 61.567871485943776, "grad_norm": 23.28106689453125, "learning_rate": 1.9190322580645163e-05, "loss": 0.3142, "step": 9551 }, { "epoch": 61.57429718875502, "grad_norm": 7.381101608276367, "learning_rate": 1.918709677419355e-05, "loss": 0.3461, "step": 9552 }, { "epoch": 61.58072289156627, "grad_norm": 47.9754524230957, "learning_rate": 1.9183870967741938e-05, "loss": 0.4299, "step": 9553 }, { "epoch": 61.58714859437751, "grad_norm": 11.07827377319336, "learning_rate": 1.9180645161290324e-05, "loss": 0.3426, "step": 9554 }, { "epoch": 61.59357429718875, "grad_norm": 8.019710540771484, "learning_rate": 1.917741935483871e-05, "loss": 0.2302, "step": 9555 }, { "epoch": 61.6, "grad_norm": 28.872867584228516, "learning_rate": 1.91741935483871e-05, "loss": 0.2863, "step": 9556 }, { "epoch": 61.606425702811244, "grad_norm": 6.486356735229492, "learning_rate": 1.9170967741935487e-05, "loss": 0.4414, "step": 9557 }, { "epoch": 61.61285140562249, "grad_norm": 16.612823486328125, "learning_rate": 1.916774193548387e-05, "loss": 0.5527, "step": 9558 }, { "epoch": 61.619277108433735, "grad_norm": 5.898576736450195, "learning_rate": 1.916451612903226e-05, "loss": 0.1337, "step": 9559 }, { "epoch": 61.62570281124498, "grad_norm": 3.8020310401916504, "learning_rate": 1.9161290322580648e-05, "loss": 0.1566, "step": 9560 }, { "epoch": 61.63212851405623, "grad_norm": 12.342406272888184, "learning_rate": 1.9158064516129034e-05, "loss": 0.3717, "step": 9561 }, { "epoch": 61.63855421686747, "grad_norm": 10.145712852478027, "learning_rate": 1.915483870967742e-05, "loss": 0.2856, "step": 9562 }, { "epoch": 61.64497991967872, "grad_norm": 6.796682357788086, "learning_rate": 1.915161290322581e-05, "loss": 0.1087, "step": 9563 }, { "epoch": 61.65140562248996, "grad_norm": 7.937827110290527, "learning_rate": 1.9148387096774194e-05, "loss": 0.4599, "step": 9564 }, { "epoch": 61.6578313253012, "grad_norm": 12.184178352355957, "learning_rate": 1.914516129032258e-05, "loss": 0.3123, "step": 9565 }, { "epoch": 61.66425702811245, "grad_norm": 13.46915340423584, "learning_rate": 1.914193548387097e-05, "loss": 0.5155, "step": 9566 }, { "epoch": 61.670682730923694, "grad_norm": 9.856911659240723, "learning_rate": 1.9138709677419355e-05, "loss": 0.2721, "step": 9567 }, { "epoch": 61.67710843373494, "grad_norm": 11.262436866760254, "learning_rate": 1.9135483870967744e-05, "loss": 0.4364, "step": 9568 }, { "epoch": 61.683534136546186, "grad_norm": 6.429572582244873, "learning_rate": 1.913225806451613e-05, "loss": 0.2539, "step": 9569 }, { "epoch": 61.68995983935743, "grad_norm": 6.093557357788086, "learning_rate": 1.9129032258064515e-05, "loss": 0.3907, "step": 9570 }, { "epoch": 61.69638554216868, "grad_norm": 10.328697204589844, "learning_rate": 1.9125806451612904e-05, "loss": 0.2653, "step": 9571 }, { "epoch": 61.70281124497992, "grad_norm": 17.99159049987793, "learning_rate": 1.9122580645161293e-05, "loss": 0.3737, "step": 9572 }, { "epoch": 61.70923694779116, "grad_norm": 10.65272045135498, "learning_rate": 1.9119354838709676e-05, "loss": 0.5232, "step": 9573 }, { "epoch": 61.71566265060241, "grad_norm": 8.691675186157227, "learning_rate": 1.9116129032258065e-05, "loss": 0.3365, "step": 9574 }, { "epoch": 61.72208835341365, "grad_norm": 6.479521751403809, "learning_rate": 1.9112903225806454e-05, "loss": 0.2964, "step": 9575 }, { "epoch": 61.7285140562249, "grad_norm": 3.4989683628082275, "learning_rate": 1.910967741935484e-05, "loss": 0.1285, "step": 9576 }, { "epoch": 61.734939759036145, "grad_norm": 9.696212768554688, "learning_rate": 1.9106451612903225e-05, "loss": 0.4897, "step": 9577 }, { "epoch": 61.74136546184739, "grad_norm": 7.472391605377197, "learning_rate": 1.9103225806451614e-05, "loss": 0.3554, "step": 9578 }, { "epoch": 61.747791164658636, "grad_norm": 6.294517517089844, "learning_rate": 1.91e-05, "loss": 0.3187, "step": 9579 }, { "epoch": 61.75421686746988, "grad_norm": 5.058650493621826, "learning_rate": 1.909677419354839e-05, "loss": 0.1545, "step": 9580 }, { "epoch": 61.76064257028113, "grad_norm": 7.868680000305176, "learning_rate": 1.9093548387096775e-05, "loss": 0.3489, "step": 9581 }, { "epoch": 61.76706827309237, "grad_norm": 32.94089889526367, "learning_rate": 1.909032258064516e-05, "loss": 0.3448, "step": 9582 }, { "epoch": 61.77349397590361, "grad_norm": 14.297347068786621, "learning_rate": 1.908709677419355e-05, "loss": 0.4996, "step": 9583 }, { "epoch": 61.77991967871486, "grad_norm": 5.024224281311035, "learning_rate": 1.908387096774194e-05, "loss": 0.3137, "step": 9584 }, { "epoch": 61.786345381526104, "grad_norm": 7.859261512756348, "learning_rate": 1.908064516129032e-05, "loss": 0.4877, "step": 9585 }, { "epoch": 61.792771084337346, "grad_norm": 10.859615325927734, "learning_rate": 1.907741935483871e-05, "loss": 0.549, "step": 9586 }, { "epoch": 61.799196787148595, "grad_norm": 5.212913513183594, "learning_rate": 1.90741935483871e-05, "loss": 0.3133, "step": 9587 }, { "epoch": 61.80562248995984, "grad_norm": 8.831284523010254, "learning_rate": 1.9070967741935485e-05, "loss": 0.201, "step": 9588 }, { "epoch": 61.81204819277109, "grad_norm": 8.6402006149292, "learning_rate": 1.906774193548387e-05, "loss": 0.3954, "step": 9589 }, { "epoch": 61.81847389558233, "grad_norm": 5.604293346405029, "learning_rate": 1.906451612903226e-05, "loss": 0.2906, "step": 9590 }, { "epoch": 61.82489959839357, "grad_norm": 19.40337371826172, "learning_rate": 1.9061290322580646e-05, "loss": 0.4035, "step": 9591 }, { "epoch": 61.83132530120482, "grad_norm": 7.3694376945495605, "learning_rate": 1.9058064516129035e-05, "loss": 0.4213, "step": 9592 }, { "epoch": 61.83775100401606, "grad_norm": 14.637478828430176, "learning_rate": 1.905483870967742e-05, "loss": 0.644, "step": 9593 }, { "epoch": 61.84417670682731, "grad_norm": 3.8890459537506104, "learning_rate": 1.9051612903225806e-05, "loss": 0.1514, "step": 9594 }, { "epoch": 61.850602409638554, "grad_norm": 9.868946075439453, "learning_rate": 1.9048387096774195e-05, "loss": 0.2621, "step": 9595 }, { "epoch": 61.8570281124498, "grad_norm": 11.05150318145752, "learning_rate": 1.9045161290322584e-05, "loss": 0.1814, "step": 9596 }, { "epoch": 61.863453815261046, "grad_norm": 9.76126766204834, "learning_rate": 1.9041935483870967e-05, "loss": 0.5206, "step": 9597 }, { "epoch": 61.86987951807229, "grad_norm": 5.689285755157471, "learning_rate": 1.9038709677419356e-05, "loss": 0.4125, "step": 9598 }, { "epoch": 61.87630522088354, "grad_norm": 3.6605052947998047, "learning_rate": 1.9035483870967745e-05, "loss": 0.2075, "step": 9599 }, { "epoch": 61.88273092369478, "grad_norm": 7.307962417602539, "learning_rate": 1.9032258064516127e-05, "loss": 0.4287, "step": 9600 }, { "epoch": 61.88915662650602, "grad_norm": 3.9787702560424805, "learning_rate": 1.9029032258064516e-05, "loss": 0.2173, "step": 9601 }, { "epoch": 61.89558232931727, "grad_norm": 11.726875305175781, "learning_rate": 1.9025806451612905e-05, "loss": 0.2942, "step": 9602 }, { "epoch": 61.90200803212851, "grad_norm": 12.592400550842285, "learning_rate": 1.902258064516129e-05, "loss": 0.3532, "step": 9603 }, { "epoch": 61.908433734939756, "grad_norm": 21.862417221069336, "learning_rate": 1.9019354838709677e-05, "loss": 0.4835, "step": 9604 }, { "epoch": 61.914859437751005, "grad_norm": 12.806248664855957, "learning_rate": 1.9016129032258066e-05, "loss": 0.406, "step": 9605 }, { "epoch": 61.92128514056225, "grad_norm": 29.582958221435547, "learning_rate": 1.901290322580645e-05, "loss": 0.2612, "step": 9606 }, { "epoch": 61.9277108433735, "grad_norm": 20.902490615844727, "learning_rate": 1.900967741935484e-05, "loss": 0.2486, "step": 9607 }, { "epoch": 61.93413654618474, "grad_norm": 10.843552589416504, "learning_rate": 1.9006451612903226e-05, "loss": 0.2064, "step": 9608 }, { "epoch": 61.94056224899598, "grad_norm": 1.4859529733657837, "learning_rate": 1.9003225806451612e-05, "loss": 0.0797, "step": 9609 }, { "epoch": 61.94698795180723, "grad_norm": 5.628155708312988, "learning_rate": 1.9e-05, "loss": 0.3158, "step": 9610 }, { "epoch": 61.95341365461847, "grad_norm": 12.720137596130371, "learning_rate": 1.899677419354839e-05, "loss": 0.3264, "step": 9611 }, { "epoch": 61.95983935742972, "grad_norm": 20.766376495361328, "learning_rate": 1.8993548387096772e-05, "loss": 0.3767, "step": 9612 }, { "epoch": 61.966265060240964, "grad_norm": 8.344478607177734, "learning_rate": 1.899032258064516e-05, "loss": 0.3171, "step": 9613 }, { "epoch": 61.972690763052206, "grad_norm": 7.552705764770508, "learning_rate": 1.898709677419355e-05, "loss": 0.4655, "step": 9614 }, { "epoch": 61.979116465863456, "grad_norm": 6.634243488311768, "learning_rate": 1.8983870967741936e-05, "loss": 0.5101, "step": 9615 }, { "epoch": 61.9855421686747, "grad_norm": 6.709218978881836, "learning_rate": 1.8980645161290322e-05, "loss": 0.3274, "step": 9616 }, { "epoch": 61.99196787148595, "grad_norm": 7.903200626373291, "learning_rate": 1.897741935483871e-05, "loss": 0.2239, "step": 9617 }, { "epoch": 62.00160642570281, "grad_norm": 10.77597427368164, "learning_rate": 1.8974193548387097e-05, "loss": 0.3425, "step": 9618 }, { "epoch": 62.00803212851405, "grad_norm": 4.277984142303467, "learning_rate": 1.8970967741935486e-05, "loss": 0.3459, "step": 9619 }, { "epoch": 62.0144578313253, "grad_norm": 15.131847381591797, "learning_rate": 1.896774193548387e-05, "loss": 0.4189, "step": 9620 }, { "epoch": 62.020883534136544, "grad_norm": 11.356260299682617, "learning_rate": 1.8964516129032257e-05, "loss": 0.3584, "step": 9621 }, { "epoch": 62.027309236947794, "grad_norm": 8.59472942352295, "learning_rate": 1.8961290322580646e-05, "loss": 0.3492, "step": 9622 }, { "epoch": 62.033734939759036, "grad_norm": 7.473719596862793, "learning_rate": 1.8958064516129036e-05, "loss": 0.2198, "step": 9623 }, { "epoch": 62.04016064257028, "grad_norm": 7.29683780670166, "learning_rate": 1.8954838709677418e-05, "loss": 0.5551, "step": 9624 }, { "epoch": 62.04658634538153, "grad_norm": 22.236051559448242, "learning_rate": 1.8951612903225807e-05, "loss": 0.3207, "step": 9625 }, { "epoch": 62.05301204819277, "grad_norm": 14.062541961669922, "learning_rate": 1.8948387096774196e-05, "loss": 0.4198, "step": 9626 }, { "epoch": 62.05943775100402, "grad_norm": 6.249337673187256, "learning_rate": 1.8945161290322582e-05, "loss": 0.1804, "step": 9627 }, { "epoch": 62.06586345381526, "grad_norm": 11.686328887939453, "learning_rate": 1.8941935483870967e-05, "loss": 0.3431, "step": 9628 }, { "epoch": 62.0722891566265, "grad_norm": 6.1454572677612305, "learning_rate": 1.8938709677419357e-05, "loss": 0.1536, "step": 9629 }, { "epoch": 62.07871485943775, "grad_norm": 5.169063568115234, "learning_rate": 1.8935483870967742e-05, "loss": 0.3318, "step": 9630 }, { "epoch": 62.085140562248995, "grad_norm": 13.883646965026855, "learning_rate": 1.893225806451613e-05, "loss": 0.3389, "step": 9631 }, { "epoch": 62.091566265060244, "grad_norm": 13.979205131530762, "learning_rate": 1.8929032258064517e-05, "loss": 0.3923, "step": 9632 }, { "epoch": 62.09799196787149, "grad_norm": 12.958463668823242, "learning_rate": 1.8925806451612903e-05, "loss": 0.3075, "step": 9633 }, { "epoch": 62.10441767068273, "grad_norm": 27.442472457885742, "learning_rate": 1.8922580645161292e-05, "loss": 0.2829, "step": 9634 }, { "epoch": 62.11084337349398, "grad_norm": 8.348760604858398, "learning_rate": 1.8919354838709678e-05, "loss": 0.435, "step": 9635 }, { "epoch": 62.11726907630522, "grad_norm": 15.384303092956543, "learning_rate": 1.8916129032258063e-05, "loss": 0.401, "step": 9636 }, { "epoch": 62.12369477911646, "grad_norm": 28.026790618896484, "learning_rate": 1.8912903225806452e-05, "loss": 0.343, "step": 9637 }, { "epoch": 62.13012048192771, "grad_norm": 24.78782844543457, "learning_rate": 1.890967741935484e-05, "loss": 0.2481, "step": 9638 }, { "epoch": 62.136546184738954, "grad_norm": 12.840600967407227, "learning_rate": 1.8906451612903224e-05, "loss": 0.6537, "step": 9639 }, { "epoch": 62.1429718875502, "grad_norm": 6.220372200012207, "learning_rate": 1.8903225806451613e-05, "loss": 0.2886, "step": 9640 }, { "epoch": 62.149397590361446, "grad_norm": 7.786555290222168, "learning_rate": 1.8900000000000002e-05, "loss": 0.3611, "step": 9641 }, { "epoch": 62.15582329317269, "grad_norm": 31.780391693115234, "learning_rate": 1.8896774193548388e-05, "loss": 0.355, "step": 9642 }, { "epoch": 62.16224899598394, "grad_norm": 46.37759017944336, "learning_rate": 1.8893548387096773e-05, "loss": 0.2125, "step": 9643 }, { "epoch": 62.16867469879518, "grad_norm": 1.8406972885131836, "learning_rate": 1.8890322580645163e-05, "loss": 0.0582, "step": 9644 }, { "epoch": 62.17510040160643, "grad_norm": 5.719559192657471, "learning_rate": 1.8887096774193548e-05, "loss": 0.2102, "step": 9645 }, { "epoch": 62.18152610441767, "grad_norm": 23.340410232543945, "learning_rate": 1.8883870967741937e-05, "loss": 0.443, "step": 9646 }, { "epoch": 62.18795180722891, "grad_norm": 6.775402069091797, "learning_rate": 1.8880645161290323e-05, "loss": 0.2135, "step": 9647 }, { "epoch": 62.19437751004016, "grad_norm": 3.2821128368377686, "learning_rate": 1.887741935483871e-05, "loss": 0.4962, "step": 9648 }, { "epoch": 62.200803212851405, "grad_norm": 13.616622924804688, "learning_rate": 1.8874193548387098e-05, "loss": 0.2548, "step": 9649 }, { "epoch": 62.207228915662654, "grad_norm": 4.5289225578308105, "learning_rate": 1.8870967741935487e-05, "loss": 0.1221, "step": 9650 }, { "epoch": 62.213654618473896, "grad_norm": 21.19673728942871, "learning_rate": 1.886774193548387e-05, "loss": 0.196, "step": 9651 }, { "epoch": 62.22008032128514, "grad_norm": 10.627843856811523, "learning_rate": 1.886451612903226e-05, "loss": 0.3621, "step": 9652 }, { "epoch": 62.22650602409639, "grad_norm": 8.584006309509277, "learning_rate": 1.8861290322580647e-05, "loss": 0.4083, "step": 9653 }, { "epoch": 62.23293172690763, "grad_norm": 9.78405475616455, "learning_rate": 1.8858064516129033e-05, "loss": 0.2702, "step": 9654 }, { "epoch": 62.23935742971887, "grad_norm": 46.52910232543945, "learning_rate": 1.885483870967742e-05, "loss": 0.4815, "step": 9655 }, { "epoch": 62.24578313253012, "grad_norm": 11.77009105682373, "learning_rate": 1.8851612903225808e-05, "loss": 0.3789, "step": 9656 }, { "epoch": 62.252208835341364, "grad_norm": 7.7988762855529785, "learning_rate": 1.8848387096774194e-05, "loss": 0.1645, "step": 9657 }, { "epoch": 62.25863453815261, "grad_norm": 75.49591827392578, "learning_rate": 1.8845161290322583e-05, "loss": 0.5794, "step": 9658 }, { "epoch": 62.265060240963855, "grad_norm": 10.495624542236328, "learning_rate": 1.884193548387097e-05, "loss": 0.6683, "step": 9659 }, { "epoch": 62.2714859437751, "grad_norm": 19.97315788269043, "learning_rate": 1.8838709677419354e-05, "loss": 0.5533, "step": 9660 }, { "epoch": 62.27791164658635, "grad_norm": 7.178805828094482, "learning_rate": 1.8835483870967743e-05, "loss": 0.2529, "step": 9661 }, { "epoch": 62.28433734939759, "grad_norm": 11.54531192779541, "learning_rate": 1.8832258064516132e-05, "loss": 0.3115, "step": 9662 }, { "epoch": 62.29076305220884, "grad_norm": 14.092905044555664, "learning_rate": 1.8829032258064515e-05, "loss": 0.4832, "step": 9663 }, { "epoch": 62.29718875502008, "grad_norm": 11.232972145080566, "learning_rate": 1.8825806451612904e-05, "loss": 0.2922, "step": 9664 }, { "epoch": 62.30361445783132, "grad_norm": 17.975797653198242, "learning_rate": 1.8822580645161293e-05, "loss": 0.3254, "step": 9665 }, { "epoch": 62.31004016064257, "grad_norm": 6.4371418952941895, "learning_rate": 1.881935483870968e-05, "loss": 0.2211, "step": 9666 }, { "epoch": 62.316465863453814, "grad_norm": 9.581252098083496, "learning_rate": 1.8816129032258064e-05, "loss": 0.4782, "step": 9667 }, { "epoch": 62.32289156626506, "grad_norm": 10.07673454284668, "learning_rate": 1.8812903225806453e-05, "loss": 0.1187, "step": 9668 }, { "epoch": 62.329317269076306, "grad_norm": 4.540850639343262, "learning_rate": 1.880967741935484e-05, "loss": 0.3447, "step": 9669 }, { "epoch": 62.33574297188755, "grad_norm": 13.745126724243164, "learning_rate": 1.8806451612903225e-05, "loss": 0.274, "step": 9670 }, { "epoch": 62.3421686746988, "grad_norm": 52.507015228271484, "learning_rate": 1.8803225806451614e-05, "loss": 0.4374, "step": 9671 }, { "epoch": 62.34859437751004, "grad_norm": 9.654534339904785, "learning_rate": 1.88e-05, "loss": 0.4066, "step": 9672 }, { "epoch": 62.35502008032128, "grad_norm": 8.01167106628418, "learning_rate": 1.879677419354839e-05, "loss": 0.2701, "step": 9673 }, { "epoch": 62.36144578313253, "grad_norm": 4.575211524963379, "learning_rate": 1.8793548387096774e-05, "loss": 0.3591, "step": 9674 }, { "epoch": 62.36787148594377, "grad_norm": 11.232206344604492, "learning_rate": 1.879032258064516e-05, "loss": 0.136, "step": 9675 }, { "epoch": 62.37429718875502, "grad_norm": 4.817242622375488, "learning_rate": 1.878709677419355e-05, "loss": 0.2748, "step": 9676 }, { "epoch": 62.380722891566265, "grad_norm": 6.369555950164795, "learning_rate": 1.8783870967741938e-05, "loss": 0.1726, "step": 9677 }, { "epoch": 62.38714859437751, "grad_norm": 17.61492156982422, "learning_rate": 1.878064516129032e-05, "loss": 0.1664, "step": 9678 }, { "epoch": 62.393574297188756, "grad_norm": 8.077603340148926, "learning_rate": 1.877741935483871e-05, "loss": 0.1911, "step": 9679 }, { "epoch": 62.4, "grad_norm": 20.880260467529297, "learning_rate": 1.87741935483871e-05, "loss": 0.4595, "step": 9680 }, { "epoch": 62.40642570281125, "grad_norm": 10.486918449401855, "learning_rate": 1.8770967741935484e-05, "loss": 0.3646, "step": 9681 }, { "epoch": 62.41285140562249, "grad_norm": 4.134535312652588, "learning_rate": 1.876774193548387e-05, "loss": 0.1772, "step": 9682 }, { "epoch": 62.41927710843373, "grad_norm": 16.91661834716797, "learning_rate": 1.876451612903226e-05, "loss": 0.6181, "step": 9683 }, { "epoch": 62.42570281124498, "grad_norm": 6.6716790199279785, "learning_rate": 1.8761290322580645e-05, "loss": 0.4687, "step": 9684 }, { "epoch": 62.432128514056224, "grad_norm": 13.060693740844727, "learning_rate": 1.8758064516129034e-05, "loss": 0.2872, "step": 9685 }, { "epoch": 62.43855421686747, "grad_norm": 17.18409538269043, "learning_rate": 1.875483870967742e-05, "loss": 0.3202, "step": 9686 }, { "epoch": 62.444979919678715, "grad_norm": 13.160019874572754, "learning_rate": 1.8751612903225806e-05, "loss": 0.2579, "step": 9687 }, { "epoch": 62.45140562248996, "grad_norm": 3.8252170085906982, "learning_rate": 1.8748387096774195e-05, "loss": 0.2788, "step": 9688 }, { "epoch": 62.45783132530121, "grad_norm": 18.099130630493164, "learning_rate": 1.8745161290322584e-05, "loss": 0.523, "step": 9689 }, { "epoch": 62.46425702811245, "grad_norm": 28.236635208129883, "learning_rate": 1.8741935483870966e-05, "loss": 0.2499, "step": 9690 }, { "epoch": 62.47068273092369, "grad_norm": 5.916816234588623, "learning_rate": 1.8738709677419355e-05, "loss": 0.4461, "step": 9691 }, { "epoch": 62.47710843373494, "grad_norm": 12.29343318939209, "learning_rate": 1.8735483870967744e-05, "loss": 0.3505, "step": 9692 }, { "epoch": 62.48353413654618, "grad_norm": 7.651863098144531, "learning_rate": 1.873225806451613e-05, "loss": 0.453, "step": 9693 }, { "epoch": 62.48995983935743, "grad_norm": 10.476935386657715, "learning_rate": 1.8729032258064516e-05, "loss": 0.3547, "step": 9694 }, { "epoch": 62.496385542168674, "grad_norm": 9.309611320495605, "learning_rate": 1.8725806451612905e-05, "loss": 0.2419, "step": 9695 }, { "epoch": 62.50281124497992, "grad_norm": 20.272077560424805, "learning_rate": 1.872258064516129e-05, "loss": 0.1872, "step": 9696 }, { "epoch": 62.509236947791166, "grad_norm": 7.254884719848633, "learning_rate": 1.871935483870968e-05, "loss": 0.2428, "step": 9697 }, { "epoch": 62.51566265060241, "grad_norm": 77.98880767822266, "learning_rate": 1.8716129032258065e-05, "loss": 0.562, "step": 9698 }, { "epoch": 62.52208835341366, "grad_norm": 11.199357032775879, "learning_rate": 1.871290322580645e-05, "loss": 0.5032, "step": 9699 }, { "epoch": 62.5285140562249, "grad_norm": 3.0330114364624023, "learning_rate": 1.870967741935484e-05, "loss": 0.1887, "step": 9700 }, { "epoch": 62.53493975903614, "grad_norm": 13.039134979248047, "learning_rate": 1.870645161290323e-05, "loss": 0.3403, "step": 9701 }, { "epoch": 62.54136546184739, "grad_norm": 11.4985990524292, "learning_rate": 1.870322580645161e-05, "loss": 0.3769, "step": 9702 }, { "epoch": 62.54779116465863, "grad_norm": 18.24168586730957, "learning_rate": 1.87e-05, "loss": 0.3247, "step": 9703 }, { "epoch": 62.55421686746988, "grad_norm": 4.625901699066162, "learning_rate": 1.869677419354839e-05, "loss": 0.2061, "step": 9704 }, { "epoch": 62.560642570281125, "grad_norm": 16.656164169311523, "learning_rate": 1.8693548387096775e-05, "loss": 0.4306, "step": 9705 }, { "epoch": 62.56706827309237, "grad_norm": 6.759951591491699, "learning_rate": 1.869032258064516e-05, "loss": 0.4275, "step": 9706 }, { "epoch": 62.57349397590362, "grad_norm": 21.598039627075195, "learning_rate": 1.868709677419355e-05, "loss": 0.4515, "step": 9707 }, { "epoch": 62.57991967871486, "grad_norm": 11.443148612976074, "learning_rate": 1.8683870967741936e-05, "loss": 0.2571, "step": 9708 }, { "epoch": 62.5863453815261, "grad_norm": 9.249371528625488, "learning_rate": 1.868064516129032e-05, "loss": 0.3321, "step": 9709 }, { "epoch": 62.59277108433735, "grad_norm": 5.192340850830078, "learning_rate": 1.867741935483871e-05, "loss": 0.3547, "step": 9710 }, { "epoch": 62.59919678714859, "grad_norm": 16.143138885498047, "learning_rate": 1.8674193548387096e-05, "loss": 0.4463, "step": 9711 }, { "epoch": 62.60562248995984, "grad_norm": 10.024649620056152, "learning_rate": 1.8670967741935485e-05, "loss": 0.2909, "step": 9712 }, { "epoch": 62.612048192771084, "grad_norm": 87.45343780517578, "learning_rate": 1.866774193548387e-05, "loss": 0.4338, "step": 9713 }, { "epoch": 62.618473895582326, "grad_norm": 11.787094116210938, "learning_rate": 1.8664516129032257e-05, "loss": 0.3814, "step": 9714 }, { "epoch": 62.624899598393576, "grad_norm": 6.262829780578613, "learning_rate": 1.8661290322580646e-05, "loss": 0.372, "step": 9715 }, { "epoch": 62.63132530120482, "grad_norm": 17.3543701171875, "learning_rate": 1.8658064516129035e-05, "loss": 0.3295, "step": 9716 }, { "epoch": 62.63775100401607, "grad_norm": 4.072933197021484, "learning_rate": 1.865483870967742e-05, "loss": 0.1813, "step": 9717 }, { "epoch": 62.64417670682731, "grad_norm": 5.904683589935303, "learning_rate": 1.8651612903225806e-05, "loss": 0.2884, "step": 9718 }, { "epoch": 62.65060240963855, "grad_norm": 11.815750122070312, "learning_rate": 1.8648387096774196e-05, "loss": 0.3833, "step": 9719 }, { "epoch": 62.6570281124498, "grad_norm": 105.71170043945312, "learning_rate": 1.864516129032258e-05, "loss": 0.6994, "step": 9720 }, { "epoch": 62.66345381526104, "grad_norm": 8.813092231750488, "learning_rate": 1.8641935483870967e-05, "loss": 0.4691, "step": 9721 }, { "epoch": 62.66987951807229, "grad_norm": 8.863479614257812, "learning_rate": 1.8638709677419356e-05, "loss": 0.4973, "step": 9722 }, { "epoch": 62.676305220883535, "grad_norm": 18.416099548339844, "learning_rate": 1.8635483870967742e-05, "loss": 0.4095, "step": 9723 }, { "epoch": 62.68273092369478, "grad_norm": 11.533865928649902, "learning_rate": 1.863225806451613e-05, "loss": 0.5877, "step": 9724 }, { "epoch": 62.689156626506026, "grad_norm": 29.161422729492188, "learning_rate": 1.8629032258064517e-05, "loss": 0.6444, "step": 9725 }, { "epoch": 62.69558232931727, "grad_norm": 80.7207260131836, "learning_rate": 1.8625806451612902e-05, "loss": 0.4828, "step": 9726 }, { "epoch": 62.70200803212851, "grad_norm": 50.23865509033203, "learning_rate": 1.862258064516129e-05, "loss": 0.2966, "step": 9727 }, { "epoch": 62.70843373493976, "grad_norm": 12.693290710449219, "learning_rate": 1.861935483870968e-05, "loss": 0.3243, "step": 9728 }, { "epoch": 62.714859437751, "grad_norm": 8.013206481933594, "learning_rate": 1.8616129032258066e-05, "loss": 0.3625, "step": 9729 }, { "epoch": 62.72128514056225, "grad_norm": 12.976834297180176, "learning_rate": 1.8612903225806452e-05, "loss": 0.6393, "step": 9730 }, { "epoch": 62.727710843373494, "grad_norm": 12.51193904876709, "learning_rate": 1.860967741935484e-05, "loss": 0.24, "step": 9731 }, { "epoch": 62.734136546184736, "grad_norm": 9.0816650390625, "learning_rate": 1.8606451612903227e-05, "loss": 0.3157, "step": 9732 }, { "epoch": 62.740562248995985, "grad_norm": 14.028429985046387, "learning_rate": 1.8603225806451612e-05, "loss": 0.4407, "step": 9733 }, { "epoch": 62.74698795180723, "grad_norm": 9.341959953308105, "learning_rate": 1.86e-05, "loss": 0.5226, "step": 9734 }, { "epoch": 62.75341365461848, "grad_norm": 8.009300231933594, "learning_rate": 1.8596774193548387e-05, "loss": 0.3962, "step": 9735 }, { "epoch": 62.75983935742972, "grad_norm": 8.177054405212402, "learning_rate": 1.8593548387096776e-05, "loss": 0.2967, "step": 9736 }, { "epoch": 62.76626506024096, "grad_norm": 6.9784255027771, "learning_rate": 1.8590322580645162e-05, "loss": 0.2672, "step": 9737 }, { "epoch": 62.77269076305221, "grad_norm": 11.331185340881348, "learning_rate": 1.8587096774193548e-05, "loss": 0.2982, "step": 9738 }, { "epoch": 62.77911646586345, "grad_norm": 7.931328296661377, "learning_rate": 1.8583870967741937e-05, "loss": 0.4397, "step": 9739 }, { "epoch": 62.7855421686747, "grad_norm": 14.728872299194336, "learning_rate": 1.8580645161290326e-05, "loss": 0.5373, "step": 9740 }, { "epoch": 62.791967871485944, "grad_norm": 9.965849876403809, "learning_rate": 1.8577419354838708e-05, "loss": 0.4921, "step": 9741 }, { "epoch": 62.79839357429719, "grad_norm": 7.985283851623535, "learning_rate": 1.8574193548387097e-05, "loss": 0.4064, "step": 9742 }, { "epoch": 62.804819277108436, "grad_norm": 9.83627700805664, "learning_rate": 1.8570967741935486e-05, "loss": 0.1439, "step": 9743 }, { "epoch": 62.81124497991968, "grad_norm": 17.54963493347168, "learning_rate": 1.8567741935483872e-05, "loss": 0.4302, "step": 9744 }, { "epoch": 62.81767068273092, "grad_norm": 32.442665100097656, "learning_rate": 1.8564516129032258e-05, "loss": 0.3573, "step": 9745 }, { "epoch": 62.82409638554217, "grad_norm": 8.191428184509277, "learning_rate": 1.8561290322580647e-05, "loss": 0.2666, "step": 9746 }, { "epoch": 62.83052208835341, "grad_norm": 7.087864875793457, "learning_rate": 1.8558064516129033e-05, "loss": 0.3541, "step": 9747 }, { "epoch": 62.83694779116466, "grad_norm": 7.227071285247803, "learning_rate": 1.855483870967742e-05, "loss": 0.3096, "step": 9748 }, { "epoch": 62.8433734939759, "grad_norm": 6.592286109924316, "learning_rate": 1.8551612903225807e-05, "loss": 0.3517, "step": 9749 }, { "epoch": 62.849799196787146, "grad_norm": 11.852201461791992, "learning_rate": 1.8548387096774193e-05, "loss": 0.432, "step": 9750 }, { "epoch": 62.856224899598395, "grad_norm": 4.767253875732422, "learning_rate": 1.8545161290322582e-05, "loss": 0.3927, "step": 9751 }, { "epoch": 62.86265060240964, "grad_norm": 5.952033519744873, "learning_rate": 1.8541935483870968e-05, "loss": 0.1974, "step": 9752 }, { "epoch": 62.86907630522089, "grad_norm": 34.88759994506836, "learning_rate": 1.8538709677419354e-05, "loss": 0.5331, "step": 9753 }, { "epoch": 62.87550200803213, "grad_norm": 7.861747741699219, "learning_rate": 1.8535483870967743e-05, "loss": 0.3376, "step": 9754 }, { "epoch": 62.88192771084337, "grad_norm": 20.451377868652344, "learning_rate": 1.8532258064516132e-05, "loss": 0.3407, "step": 9755 }, { "epoch": 62.88835341365462, "grad_norm": 6.09533166885376, "learning_rate": 1.8529032258064518e-05, "loss": 0.2811, "step": 9756 }, { "epoch": 62.89477911646586, "grad_norm": 6.255895614624023, "learning_rate": 1.8525806451612903e-05, "loss": 0.1843, "step": 9757 }, { "epoch": 62.90120481927711, "grad_norm": 7.785248756408691, "learning_rate": 1.8522580645161292e-05, "loss": 0.3834, "step": 9758 }, { "epoch": 62.907630522088354, "grad_norm": 8.665393829345703, "learning_rate": 1.8519354838709678e-05, "loss": 0.2122, "step": 9759 }, { "epoch": 62.914056224899596, "grad_norm": 4.308812618255615, "learning_rate": 1.8516129032258064e-05, "loss": 0.1369, "step": 9760 }, { "epoch": 62.920481927710846, "grad_norm": 7.673308372497559, "learning_rate": 1.8512903225806453e-05, "loss": 0.2095, "step": 9761 }, { "epoch": 62.92690763052209, "grad_norm": 4.612572193145752, "learning_rate": 1.850967741935484e-05, "loss": 0.337, "step": 9762 }, { "epoch": 62.93333333333333, "grad_norm": 7.7105326652526855, "learning_rate": 1.8506451612903228e-05, "loss": 0.2447, "step": 9763 }, { "epoch": 62.93975903614458, "grad_norm": 19.88982391357422, "learning_rate": 1.8503225806451613e-05, "loss": 0.6019, "step": 9764 }, { "epoch": 62.94618473895582, "grad_norm": 3.9709079265594482, "learning_rate": 1.85e-05, "loss": 0.215, "step": 9765 }, { "epoch": 62.95261044176707, "grad_norm": 5.355020523071289, "learning_rate": 1.8496774193548388e-05, "loss": 0.3123, "step": 9766 }, { "epoch": 62.95903614457831, "grad_norm": 17.647573471069336, "learning_rate": 1.8493548387096777e-05, "loss": 0.3197, "step": 9767 }, { "epoch": 62.965461847389555, "grad_norm": 5.736093997955322, "learning_rate": 1.8490322580645163e-05, "loss": 0.4513, "step": 9768 }, { "epoch": 62.971887550200805, "grad_norm": 8.651398658752441, "learning_rate": 1.848709677419355e-05, "loss": 0.3575, "step": 9769 }, { "epoch": 62.97831325301205, "grad_norm": 7.770142078399658, "learning_rate": 1.8483870967741938e-05, "loss": 0.2065, "step": 9770 }, { "epoch": 62.984738955823296, "grad_norm": 10.868807792663574, "learning_rate": 1.8480645161290323e-05, "loss": 0.5083, "step": 9771 }, { "epoch": 62.99116465863454, "grad_norm": 13.765331268310547, "learning_rate": 1.847741935483871e-05, "loss": 0.2819, "step": 9772 }, { "epoch": 63.00080321285141, "grad_norm": 4.232555389404297, "learning_rate": 1.8474193548387098e-05, "loss": 0.1835, "step": 9773 }, { "epoch": 63.00722891566265, "grad_norm": 7.283231735229492, "learning_rate": 1.8470967741935484e-05, "loss": 0.294, "step": 9774 }, { "epoch": 63.01365461847389, "grad_norm": 9.76679801940918, "learning_rate": 1.8467741935483873e-05, "loss": 0.4458, "step": 9775 }, { "epoch": 63.02008032128514, "grad_norm": 5.43214750289917, "learning_rate": 1.846451612903226e-05, "loss": 0.1559, "step": 9776 }, { "epoch": 63.026506024096385, "grad_norm": 20.14143943786621, "learning_rate": 1.8461290322580644e-05, "loss": 0.2121, "step": 9777 }, { "epoch": 63.03293172690763, "grad_norm": 25.20077133178711, "learning_rate": 1.8458064516129034e-05, "loss": 0.4078, "step": 9778 }, { "epoch": 63.039357429718876, "grad_norm": 4.114442348480225, "learning_rate": 1.845483870967742e-05, "loss": 0.1006, "step": 9779 }, { "epoch": 63.04578313253012, "grad_norm": 15.807417869567871, "learning_rate": 1.845161290322581e-05, "loss": 0.3194, "step": 9780 }, { "epoch": 63.05220883534137, "grad_norm": 5.105607509613037, "learning_rate": 1.8448387096774194e-05, "loss": 0.289, "step": 9781 }, { "epoch": 63.05863453815261, "grad_norm": 6.8404974937438965, "learning_rate": 1.8445161290322583e-05, "loss": 0.2991, "step": 9782 }, { "epoch": 63.06506024096385, "grad_norm": 12.635913848876953, "learning_rate": 1.844193548387097e-05, "loss": 0.2775, "step": 9783 }, { "epoch": 63.0714859437751, "grad_norm": 7.646561622619629, "learning_rate": 1.8438709677419355e-05, "loss": 0.2883, "step": 9784 }, { "epoch": 63.077911646586344, "grad_norm": 12.723731994628906, "learning_rate": 1.8435483870967744e-05, "loss": 0.1433, "step": 9785 }, { "epoch": 63.08433734939759, "grad_norm": 14.251596450805664, "learning_rate": 1.843225806451613e-05, "loss": 0.2916, "step": 9786 }, { "epoch": 63.090763052208835, "grad_norm": 6.453538417816162, "learning_rate": 1.8429032258064515e-05, "loss": 0.3101, "step": 9787 }, { "epoch": 63.09718875502008, "grad_norm": 11.861954689025879, "learning_rate": 1.8425806451612904e-05, "loss": 0.3958, "step": 9788 }, { "epoch": 63.10361445783133, "grad_norm": 8.531064987182617, "learning_rate": 1.842258064516129e-05, "loss": 0.5842, "step": 9789 }, { "epoch": 63.11004016064257, "grad_norm": 6.653960227966309, "learning_rate": 1.841935483870968e-05, "loss": 0.3596, "step": 9790 }, { "epoch": 63.11646586345382, "grad_norm": 19.01388168334961, "learning_rate": 1.8416129032258065e-05, "loss": 0.3419, "step": 9791 }, { "epoch": 63.12289156626506, "grad_norm": 9.143494606018066, "learning_rate": 1.8412903225806454e-05, "loss": 0.3369, "step": 9792 }, { "epoch": 63.1293172690763, "grad_norm": 41.32926940917969, "learning_rate": 1.840967741935484e-05, "loss": 0.4022, "step": 9793 }, { "epoch": 63.13574297188755, "grad_norm": 13.609683990478516, "learning_rate": 1.840645161290323e-05, "loss": 0.428, "step": 9794 }, { "epoch": 63.142168674698794, "grad_norm": 6.522578239440918, "learning_rate": 1.8403225806451614e-05, "loss": 0.3193, "step": 9795 }, { "epoch": 63.14859437751004, "grad_norm": 4.820715427398682, "learning_rate": 1.84e-05, "loss": 0.1942, "step": 9796 }, { "epoch": 63.155020080321286, "grad_norm": 19.860010147094727, "learning_rate": 1.839677419354839e-05, "loss": 0.381, "step": 9797 }, { "epoch": 63.16144578313253, "grad_norm": 3.4498186111450195, "learning_rate": 1.8393548387096775e-05, "loss": 0.1728, "step": 9798 }, { "epoch": 63.16787148594378, "grad_norm": 4.222604751586914, "learning_rate": 1.839032258064516e-05, "loss": 0.3432, "step": 9799 }, { "epoch": 63.17429718875502, "grad_norm": 12.404746055603027, "learning_rate": 1.838709677419355e-05, "loss": 0.32, "step": 9800 }, { "epoch": 63.18072289156626, "grad_norm": 9.703675270080566, "learning_rate": 1.8383870967741935e-05, "loss": 0.4009, "step": 9801 }, { "epoch": 63.18714859437751, "grad_norm": 9.559374809265137, "learning_rate": 1.8380645161290324e-05, "loss": 0.2373, "step": 9802 }, { "epoch": 63.19357429718875, "grad_norm": 7.693971157073975, "learning_rate": 1.837741935483871e-05, "loss": 0.2411, "step": 9803 }, { "epoch": 63.2, "grad_norm": 8.48255729675293, "learning_rate": 1.8374193548387096e-05, "loss": 0.4534, "step": 9804 }, { "epoch": 63.206425702811245, "grad_norm": 28.69174575805664, "learning_rate": 1.8370967741935485e-05, "loss": 0.4548, "step": 9805 }, { "epoch": 63.21285140562249, "grad_norm": 16.862607955932617, "learning_rate": 1.8367741935483874e-05, "loss": 0.526, "step": 9806 }, { "epoch": 63.21927710843374, "grad_norm": 6.327004432678223, "learning_rate": 1.836451612903226e-05, "loss": 0.1797, "step": 9807 }, { "epoch": 63.22570281124498, "grad_norm": 7.459498405456543, "learning_rate": 1.8361290322580645e-05, "loss": 0.2296, "step": 9808 }, { "epoch": 63.23212851405623, "grad_norm": 4.46823263168335, "learning_rate": 1.8358064516129035e-05, "loss": 0.1676, "step": 9809 }, { "epoch": 63.23855421686747, "grad_norm": 10.535564422607422, "learning_rate": 1.835483870967742e-05, "loss": 0.7857, "step": 9810 }, { "epoch": 63.24497991967871, "grad_norm": 11.483525276184082, "learning_rate": 1.8351612903225806e-05, "loss": 0.4036, "step": 9811 }, { "epoch": 63.25140562248996, "grad_norm": 10.023435592651367, "learning_rate": 1.8348387096774195e-05, "loss": 0.1695, "step": 9812 }, { "epoch": 63.257831325301204, "grad_norm": 8.284754753112793, "learning_rate": 1.834516129032258e-05, "loss": 0.4405, "step": 9813 }, { "epoch": 63.264257028112446, "grad_norm": 8.93739128112793, "learning_rate": 1.8341935483870966e-05, "loss": 0.3139, "step": 9814 }, { "epoch": 63.270682730923696, "grad_norm": 8.189189910888672, "learning_rate": 1.8338709677419356e-05, "loss": 0.2183, "step": 9815 }, { "epoch": 63.27710843373494, "grad_norm": 15.045328140258789, "learning_rate": 1.833548387096774e-05, "loss": 0.6126, "step": 9816 }, { "epoch": 63.28353413654619, "grad_norm": 5.770256519317627, "learning_rate": 1.833225806451613e-05, "loss": 0.338, "step": 9817 }, { "epoch": 63.28995983935743, "grad_norm": 10.165118217468262, "learning_rate": 1.8329032258064516e-05, "loss": 0.5432, "step": 9818 }, { "epoch": 63.29638554216867, "grad_norm": 8.450175285339355, "learning_rate": 1.8325806451612905e-05, "loss": 0.4286, "step": 9819 }, { "epoch": 63.30281124497992, "grad_norm": 10.1035795211792, "learning_rate": 1.832258064516129e-05, "loss": 0.1867, "step": 9820 }, { "epoch": 63.30923694779116, "grad_norm": 6.512495040893555, "learning_rate": 1.831935483870968e-05, "loss": 0.1272, "step": 9821 }, { "epoch": 63.31566265060241, "grad_norm": 11.429304122924805, "learning_rate": 1.8316129032258066e-05, "loss": 0.5088, "step": 9822 }, { "epoch": 63.322088353413655, "grad_norm": 17.501144409179688, "learning_rate": 1.831290322580645e-05, "loss": 0.2802, "step": 9823 }, { "epoch": 63.3285140562249, "grad_norm": 12.954021453857422, "learning_rate": 1.830967741935484e-05, "loss": 0.5017, "step": 9824 }, { "epoch": 63.334939759036146, "grad_norm": 16.6405029296875, "learning_rate": 1.8306451612903226e-05, "loss": 0.4163, "step": 9825 }, { "epoch": 63.34136546184739, "grad_norm": 6.4837327003479, "learning_rate": 1.8303225806451612e-05, "loss": 0.2994, "step": 9826 }, { "epoch": 63.34779116465864, "grad_norm": 7.364033222198486, "learning_rate": 1.83e-05, "loss": 0.4396, "step": 9827 }, { "epoch": 63.35421686746988, "grad_norm": 10.325826644897461, "learning_rate": 1.8296774193548387e-05, "loss": 0.4357, "step": 9828 }, { "epoch": 63.36064257028112, "grad_norm": 15.678816795349121, "learning_rate": 1.8293548387096776e-05, "loss": 0.5848, "step": 9829 }, { "epoch": 63.36706827309237, "grad_norm": 5.543935775756836, "learning_rate": 1.829032258064516e-05, "loss": 0.1658, "step": 9830 }, { "epoch": 63.373493975903614, "grad_norm": 12.550654411315918, "learning_rate": 1.828709677419355e-05, "loss": 0.5102, "step": 9831 }, { "epoch": 63.379919678714856, "grad_norm": 30.08832359313965, "learning_rate": 1.8283870967741936e-05, "loss": 0.3567, "step": 9832 }, { "epoch": 63.386345381526105, "grad_norm": 9.930652618408203, "learning_rate": 1.8280645161290325e-05, "loss": 0.2938, "step": 9833 }, { "epoch": 63.39277108433735, "grad_norm": 4.425585746765137, "learning_rate": 1.827741935483871e-05, "loss": 0.154, "step": 9834 }, { "epoch": 63.3991967871486, "grad_norm": 11.143013000488281, "learning_rate": 1.8274193548387097e-05, "loss": 0.4288, "step": 9835 }, { "epoch": 63.40562248995984, "grad_norm": 8.154888153076172, "learning_rate": 1.8270967741935486e-05, "loss": 0.1686, "step": 9836 }, { "epoch": 63.41204819277108, "grad_norm": 8.072259902954102, "learning_rate": 1.826774193548387e-05, "loss": 0.2049, "step": 9837 }, { "epoch": 63.41847389558233, "grad_norm": 11.192670822143555, "learning_rate": 1.8264516129032257e-05, "loss": 0.394, "step": 9838 }, { "epoch": 63.42489959839357, "grad_norm": 6.213763236999512, "learning_rate": 1.8261290322580646e-05, "loss": 0.2623, "step": 9839 }, { "epoch": 63.43132530120482, "grad_norm": 4.889918327331543, "learning_rate": 1.8258064516129032e-05, "loss": 0.2571, "step": 9840 }, { "epoch": 63.437751004016064, "grad_norm": 8.322469711303711, "learning_rate": 1.825483870967742e-05, "loss": 0.3069, "step": 9841 }, { "epoch": 63.44417670682731, "grad_norm": 9.122991561889648, "learning_rate": 1.8251612903225807e-05, "loss": 0.3532, "step": 9842 }, { "epoch": 63.450602409638556, "grad_norm": 8.564327239990234, "learning_rate": 1.8248387096774196e-05, "loss": 0.1854, "step": 9843 }, { "epoch": 63.4570281124498, "grad_norm": 6.274616241455078, "learning_rate": 1.8245161290322582e-05, "loss": 0.3158, "step": 9844 }, { "epoch": 63.46345381526105, "grad_norm": 6.283868312835693, "learning_rate": 1.824193548387097e-05, "loss": 0.1672, "step": 9845 }, { "epoch": 63.46987951807229, "grad_norm": 7.734771251678467, "learning_rate": 1.8238709677419357e-05, "loss": 0.5073, "step": 9846 }, { "epoch": 63.47630522088353, "grad_norm": 8.222757339477539, "learning_rate": 1.8235483870967742e-05, "loss": 0.4454, "step": 9847 }, { "epoch": 63.48273092369478, "grad_norm": 13.7427978515625, "learning_rate": 1.823225806451613e-05, "loss": 0.281, "step": 9848 }, { "epoch": 63.48915662650602, "grad_norm": 8.173940658569336, "learning_rate": 1.8229032258064517e-05, "loss": 0.1999, "step": 9849 }, { "epoch": 63.495582329317266, "grad_norm": 11.495317459106445, "learning_rate": 1.8225806451612903e-05, "loss": 0.6146, "step": 9850 }, { "epoch": 63.502008032128515, "grad_norm": 10.930644989013672, "learning_rate": 1.8222580645161292e-05, "loss": 0.7796, "step": 9851 }, { "epoch": 63.50843373493976, "grad_norm": 13.571673393249512, "learning_rate": 1.8219354838709678e-05, "loss": 0.4938, "step": 9852 }, { "epoch": 63.51485943775101, "grad_norm": 6.868824005126953, "learning_rate": 1.8216129032258063e-05, "loss": 0.3316, "step": 9853 }, { "epoch": 63.52128514056225, "grad_norm": 6.820518970489502, "learning_rate": 1.8212903225806452e-05, "loss": 0.3479, "step": 9854 }, { "epoch": 63.52771084337349, "grad_norm": 8.622833251953125, "learning_rate": 1.820967741935484e-05, "loss": 0.1527, "step": 9855 }, { "epoch": 63.53413654618474, "grad_norm": 32.86064529418945, "learning_rate": 1.8206451612903227e-05, "loss": 0.4072, "step": 9856 }, { "epoch": 63.54056224899598, "grad_norm": 14.367424011230469, "learning_rate": 1.8203225806451613e-05, "loss": 0.5724, "step": 9857 }, { "epoch": 63.54698795180723, "grad_norm": 14.740036964416504, "learning_rate": 1.8200000000000002e-05, "loss": 0.3006, "step": 9858 }, { "epoch": 63.553413654618474, "grad_norm": 4.9104108810424805, "learning_rate": 1.8196774193548388e-05, "loss": 0.1473, "step": 9859 }, { "epoch": 63.559839357429716, "grad_norm": 3.4905405044555664, "learning_rate": 1.8193548387096777e-05, "loss": 0.1659, "step": 9860 }, { "epoch": 63.566265060240966, "grad_norm": 10.957624435424805, "learning_rate": 1.8190322580645162e-05, "loss": 0.3486, "step": 9861 }, { "epoch": 63.57269076305221, "grad_norm": 11.494583129882812, "learning_rate": 1.8187096774193548e-05, "loss": 0.3487, "step": 9862 }, { "epoch": 63.57911646586346, "grad_norm": 6.247692108154297, "learning_rate": 1.8183870967741937e-05, "loss": 0.2546, "step": 9863 }, { "epoch": 63.5855421686747, "grad_norm": 10.209617614746094, "learning_rate": 1.8180645161290323e-05, "loss": 0.4431, "step": 9864 }, { "epoch": 63.59196787148594, "grad_norm": 18.24652862548828, "learning_rate": 1.817741935483871e-05, "loss": 0.5215, "step": 9865 }, { "epoch": 63.59839357429719, "grad_norm": 6.182807922363281, "learning_rate": 1.8174193548387098e-05, "loss": 0.3348, "step": 9866 }, { "epoch": 63.60481927710843, "grad_norm": 10.805475234985352, "learning_rate": 1.8170967741935483e-05, "loss": 0.5965, "step": 9867 }, { "epoch": 63.611244979919675, "grad_norm": 8.220215797424316, "learning_rate": 1.8167741935483873e-05, "loss": 0.2569, "step": 9868 }, { "epoch": 63.617670682730925, "grad_norm": 11.306218147277832, "learning_rate": 1.8164516129032258e-05, "loss": 0.3072, "step": 9869 }, { "epoch": 63.62409638554217, "grad_norm": 9.904862403869629, "learning_rate": 1.8161290322580647e-05, "loss": 0.4976, "step": 9870 }, { "epoch": 63.630522088353416, "grad_norm": 5.998709678649902, "learning_rate": 1.8158064516129033e-05, "loss": 0.143, "step": 9871 }, { "epoch": 63.63694779116466, "grad_norm": 10.673972129821777, "learning_rate": 1.8154838709677422e-05, "loss": 0.302, "step": 9872 }, { "epoch": 63.6433734939759, "grad_norm": 12.213934898376465, "learning_rate": 1.8151612903225808e-05, "loss": 0.3635, "step": 9873 }, { "epoch": 63.64979919678715, "grad_norm": 5.373374938964844, "learning_rate": 1.8148387096774194e-05, "loss": 0.1838, "step": 9874 }, { "epoch": 63.65622489959839, "grad_norm": 15.851325988769531, "learning_rate": 1.8145161290322583e-05, "loss": 0.3699, "step": 9875 }, { "epoch": 63.66265060240964, "grad_norm": 15.392342567443848, "learning_rate": 1.814193548387097e-05, "loss": 0.4946, "step": 9876 }, { "epoch": 63.669076305220884, "grad_norm": 16.96819496154785, "learning_rate": 1.8138709677419354e-05, "loss": 0.4025, "step": 9877 }, { "epoch": 63.675502008032126, "grad_norm": 6.149435520172119, "learning_rate": 1.8135483870967743e-05, "loss": 0.2228, "step": 9878 }, { "epoch": 63.681927710843375, "grad_norm": 7.550358295440674, "learning_rate": 1.813225806451613e-05, "loss": 0.2496, "step": 9879 }, { "epoch": 63.68835341365462, "grad_norm": 16.047534942626953, "learning_rate": 1.8129032258064518e-05, "loss": 0.3991, "step": 9880 }, { "epoch": 63.69477911646587, "grad_norm": 3.726963996887207, "learning_rate": 1.8125806451612904e-05, "loss": 0.1356, "step": 9881 }, { "epoch": 63.70120481927711, "grad_norm": 11.931534767150879, "learning_rate": 1.8122580645161293e-05, "loss": 0.3456, "step": 9882 }, { "epoch": 63.70763052208835, "grad_norm": 16.397377014160156, "learning_rate": 1.811935483870968e-05, "loss": 0.5073, "step": 9883 }, { "epoch": 63.7140562248996, "grad_norm": 22.51012420654297, "learning_rate": 1.8116129032258064e-05, "loss": 0.4702, "step": 9884 }, { "epoch": 63.72048192771084, "grad_norm": 7.681759834289551, "learning_rate": 1.8112903225806453e-05, "loss": 0.2371, "step": 9885 }, { "epoch": 63.726907630522085, "grad_norm": 4.63995885848999, "learning_rate": 1.810967741935484e-05, "loss": 0.2216, "step": 9886 }, { "epoch": 63.733333333333334, "grad_norm": 7.22000789642334, "learning_rate": 1.8106451612903228e-05, "loss": 0.4482, "step": 9887 }, { "epoch": 63.73975903614458, "grad_norm": 19.591249465942383, "learning_rate": 1.8103225806451614e-05, "loss": 0.6374, "step": 9888 }, { "epoch": 63.746184738955826, "grad_norm": 14.38123607635498, "learning_rate": 1.81e-05, "loss": 0.3575, "step": 9889 }, { "epoch": 63.75261044176707, "grad_norm": 5.382565498352051, "learning_rate": 1.809677419354839e-05, "loss": 0.3033, "step": 9890 }, { "epoch": 63.75903614457831, "grad_norm": 12.392477989196777, "learning_rate": 1.8093548387096774e-05, "loss": 0.4472, "step": 9891 }, { "epoch": 63.76546184738956, "grad_norm": 7.222362518310547, "learning_rate": 1.809032258064516e-05, "loss": 0.4718, "step": 9892 }, { "epoch": 63.7718875502008, "grad_norm": 12.84952449798584, "learning_rate": 1.808709677419355e-05, "loss": 0.6511, "step": 9893 }, { "epoch": 63.77831325301205, "grad_norm": 15.154152870178223, "learning_rate": 1.8083870967741938e-05, "loss": 0.3345, "step": 9894 }, { "epoch": 63.78473895582329, "grad_norm": 4.76397705078125, "learning_rate": 1.8080645161290324e-05, "loss": 0.1322, "step": 9895 }, { "epoch": 63.791164658634536, "grad_norm": 31.656139373779297, "learning_rate": 1.807741935483871e-05, "loss": 0.6394, "step": 9896 }, { "epoch": 63.797590361445785, "grad_norm": 4.927604675292969, "learning_rate": 1.80741935483871e-05, "loss": 0.1702, "step": 9897 }, { "epoch": 63.80401606425703, "grad_norm": 25.0007381439209, "learning_rate": 1.8070967741935484e-05, "loss": 0.3641, "step": 9898 }, { "epoch": 63.810441767068276, "grad_norm": 7.797725677490234, "learning_rate": 1.8067741935483874e-05, "loss": 0.3998, "step": 9899 }, { "epoch": 63.81686746987952, "grad_norm": 11.001249313354492, "learning_rate": 1.806451612903226e-05, "loss": 0.3832, "step": 9900 }, { "epoch": 63.82329317269076, "grad_norm": 7.700016975402832, "learning_rate": 1.8061290322580645e-05, "loss": 0.2139, "step": 9901 }, { "epoch": 63.82971887550201, "grad_norm": 32.945003509521484, "learning_rate": 1.8058064516129034e-05, "loss": 0.2648, "step": 9902 }, { "epoch": 63.83614457831325, "grad_norm": 10.473057746887207, "learning_rate": 1.805483870967742e-05, "loss": 0.3443, "step": 9903 }, { "epoch": 63.842570281124495, "grad_norm": 15.285216331481934, "learning_rate": 1.8051612903225805e-05, "loss": 0.5567, "step": 9904 }, { "epoch": 63.848995983935744, "grad_norm": 6.817166805267334, "learning_rate": 1.8048387096774195e-05, "loss": 0.2471, "step": 9905 }, { "epoch": 63.855421686746986, "grad_norm": 8.822260856628418, "learning_rate": 1.8045161290322584e-05, "loss": 0.3498, "step": 9906 }, { "epoch": 63.861847389558235, "grad_norm": 12.521121978759766, "learning_rate": 1.804193548387097e-05, "loss": 0.4293, "step": 9907 }, { "epoch": 63.86827309236948, "grad_norm": 19.839759826660156, "learning_rate": 1.8038709677419355e-05, "loss": 0.2633, "step": 9908 }, { "epoch": 63.87469879518072, "grad_norm": 10.083704948425293, "learning_rate": 1.8035483870967744e-05, "loss": 0.1633, "step": 9909 }, { "epoch": 63.88112449799197, "grad_norm": 6.260706424713135, "learning_rate": 1.803225806451613e-05, "loss": 0.3046, "step": 9910 }, { "epoch": 63.88755020080321, "grad_norm": 6.090437412261963, "learning_rate": 1.802903225806452e-05, "loss": 0.2146, "step": 9911 }, { "epoch": 63.89397590361446, "grad_norm": 15.127165794372559, "learning_rate": 1.8025806451612905e-05, "loss": 0.4418, "step": 9912 }, { "epoch": 63.9004016064257, "grad_norm": 8.010884284973145, "learning_rate": 1.802258064516129e-05, "loss": 0.4564, "step": 9913 }, { "epoch": 63.906827309236945, "grad_norm": 5.629749774932861, "learning_rate": 1.801935483870968e-05, "loss": 0.249, "step": 9914 }, { "epoch": 63.913253012048195, "grad_norm": 10.510514259338379, "learning_rate": 1.8016129032258065e-05, "loss": 0.2861, "step": 9915 }, { "epoch": 63.91967871485944, "grad_norm": 5.205867767333984, "learning_rate": 1.801290322580645e-05, "loss": 0.4444, "step": 9916 }, { "epoch": 63.926104417670686, "grad_norm": 7.900332927703857, "learning_rate": 1.800967741935484e-05, "loss": 0.3751, "step": 9917 }, { "epoch": 63.93253012048193, "grad_norm": 9.048799514770508, "learning_rate": 1.800645161290323e-05, "loss": 0.3282, "step": 9918 }, { "epoch": 63.93895582329317, "grad_norm": 13.756075859069824, "learning_rate": 1.800322580645161e-05, "loss": 0.5046, "step": 9919 }, { "epoch": 63.94538152610442, "grad_norm": 28.294862747192383, "learning_rate": 1.8e-05, "loss": 0.286, "step": 9920 }, { "epoch": 63.95180722891566, "grad_norm": 6.18265438079834, "learning_rate": 1.799677419354839e-05, "loss": 0.323, "step": 9921 }, { "epoch": 63.958232931726904, "grad_norm": 4.732179164886475, "learning_rate": 1.7993548387096775e-05, "loss": 0.0687, "step": 9922 }, { "epoch": 63.964658634538154, "grad_norm": 5.230437755584717, "learning_rate": 1.799032258064516e-05, "loss": 0.131, "step": 9923 }, { "epoch": 63.971084337349396, "grad_norm": 8.395163536071777, "learning_rate": 1.798709677419355e-05, "loss": 0.2947, "step": 9924 }, { "epoch": 63.977510040160645, "grad_norm": 11.542778968811035, "learning_rate": 1.7983870967741936e-05, "loss": 0.3948, "step": 9925 }, { "epoch": 63.98393574297189, "grad_norm": 9.166508674621582, "learning_rate": 1.7980645161290325e-05, "loss": 0.3548, "step": 9926 }, { "epoch": 63.99036144578313, "grad_norm": 11.849881172180176, "learning_rate": 1.797741935483871e-05, "loss": 0.7158, "step": 9927 }, { "epoch": 63.99678714859438, "grad_norm": 7.260274410247803, "learning_rate": 1.7974193548387096e-05, "loss": 0.477, "step": 9928 }, { "epoch": 64.00642570281124, "grad_norm": 18.527118682861328, "learning_rate": 1.7970967741935485e-05, "loss": 0.4076, "step": 9929 }, { "epoch": 64.01285140562248, "grad_norm": 12.370447158813477, "learning_rate": 1.796774193548387e-05, "loss": 0.4092, "step": 9930 }, { "epoch": 64.01927710843374, "grad_norm": 11.663895606994629, "learning_rate": 1.7964516129032257e-05, "loss": 0.3822, "step": 9931 }, { "epoch": 64.02570281124498, "grad_norm": 19.097816467285156, "learning_rate": 1.7961290322580646e-05, "loss": 0.3065, "step": 9932 }, { "epoch": 64.03212851405623, "grad_norm": 23.343910217285156, "learning_rate": 1.7958064516129035e-05, "loss": 0.4252, "step": 9933 }, { "epoch": 64.03855421686747, "grad_norm": 9.888547897338867, "learning_rate": 1.795483870967742e-05, "loss": 0.3497, "step": 9934 }, { "epoch": 64.04497991967871, "grad_norm": 8.186392784118652, "learning_rate": 1.7951612903225806e-05, "loss": 0.2475, "step": 9935 }, { "epoch": 64.05140562248997, "grad_norm": 8.300703048706055, "learning_rate": 1.7948387096774195e-05, "loss": 0.2818, "step": 9936 }, { "epoch": 64.05783132530121, "grad_norm": 11.63050651550293, "learning_rate": 1.794516129032258e-05, "loss": 0.4704, "step": 9937 }, { "epoch": 64.06425702811245, "grad_norm": 10.21927261352539, "learning_rate": 1.794193548387097e-05, "loss": 0.4679, "step": 9938 }, { "epoch": 64.07068273092369, "grad_norm": 15.657363891601562, "learning_rate": 1.7938709677419356e-05, "loss": 0.4611, "step": 9939 }, { "epoch": 64.07710843373494, "grad_norm": 14.994342803955078, "learning_rate": 1.7935483870967742e-05, "loss": 0.548, "step": 9940 }, { "epoch": 64.08353413654619, "grad_norm": 14.29724407196045, "learning_rate": 1.793225806451613e-05, "loss": 0.32, "step": 9941 }, { "epoch": 64.08995983935743, "grad_norm": 11.267457008361816, "learning_rate": 1.7929032258064516e-05, "loss": 0.4227, "step": 9942 }, { "epoch": 64.09638554216868, "grad_norm": 4.4873762130737305, "learning_rate": 1.7925806451612902e-05, "loss": 0.2063, "step": 9943 }, { "epoch": 64.10281124497992, "grad_norm": 4.108654499053955, "learning_rate": 1.792258064516129e-05, "loss": 0.1632, "step": 9944 }, { "epoch": 64.10923694779116, "grad_norm": 11.064501762390137, "learning_rate": 1.791935483870968e-05, "loss": 0.4995, "step": 9945 }, { "epoch": 64.1156626506024, "grad_norm": 8.859625816345215, "learning_rate": 1.7916129032258066e-05, "loss": 0.398, "step": 9946 }, { "epoch": 64.12208835341366, "grad_norm": 31.8221378326416, "learning_rate": 1.7912903225806452e-05, "loss": 0.3668, "step": 9947 }, { "epoch": 64.1285140562249, "grad_norm": 17.027812957763672, "learning_rate": 1.790967741935484e-05, "loss": 0.2034, "step": 9948 }, { "epoch": 64.13493975903614, "grad_norm": 7.153292179107666, "learning_rate": 1.7906451612903227e-05, "loss": 0.531, "step": 9949 }, { "epoch": 64.14136546184739, "grad_norm": 11.82421875, "learning_rate": 1.7903225806451616e-05, "loss": 0.5347, "step": 9950 }, { "epoch": 64.14779116465863, "grad_norm": 8.82397747039795, "learning_rate": 1.79e-05, "loss": 0.4971, "step": 9951 }, { "epoch": 64.15421686746988, "grad_norm": 10.209368705749512, "learning_rate": 1.7896774193548387e-05, "loss": 0.2627, "step": 9952 }, { "epoch": 64.16064257028113, "grad_norm": 7.967401027679443, "learning_rate": 1.7893548387096776e-05, "loss": 0.3777, "step": 9953 }, { "epoch": 64.16706827309237, "grad_norm": 7.969147682189941, "learning_rate": 1.7890322580645162e-05, "loss": 0.3616, "step": 9954 }, { "epoch": 64.17349397590361, "grad_norm": 6.659584999084473, "learning_rate": 1.7887096774193548e-05, "loss": 0.2636, "step": 9955 }, { "epoch": 64.17991967871485, "grad_norm": 16.089195251464844, "learning_rate": 1.7883870967741937e-05, "loss": 0.3265, "step": 9956 }, { "epoch": 64.18634538152611, "grad_norm": 12.692270278930664, "learning_rate": 1.7880645161290326e-05, "loss": 0.5512, "step": 9957 }, { "epoch": 64.19277108433735, "grad_norm": 16.737730026245117, "learning_rate": 1.7877419354838708e-05, "loss": 0.2965, "step": 9958 }, { "epoch": 64.1991967871486, "grad_norm": 7.713048458099365, "learning_rate": 1.7874193548387097e-05, "loss": 0.328, "step": 9959 }, { "epoch": 64.20562248995984, "grad_norm": 8.054083824157715, "learning_rate": 1.7870967741935486e-05, "loss": 0.4172, "step": 9960 }, { "epoch": 64.21204819277108, "grad_norm": 9.589738845825195, "learning_rate": 1.7867741935483872e-05, "loss": 0.4586, "step": 9961 }, { "epoch": 64.21847389558233, "grad_norm": 15.36292839050293, "learning_rate": 1.7864516129032258e-05, "loss": 0.3153, "step": 9962 }, { "epoch": 64.22489959839358, "grad_norm": 7.528014183044434, "learning_rate": 1.7861290322580647e-05, "loss": 0.2826, "step": 9963 }, { "epoch": 64.23132530120482, "grad_norm": 5.950640678405762, "learning_rate": 1.7858064516129033e-05, "loss": 0.2194, "step": 9964 }, { "epoch": 64.23775100401606, "grad_norm": 5.550477981567383, "learning_rate": 1.785483870967742e-05, "loss": 0.2741, "step": 9965 }, { "epoch": 64.2441767068273, "grad_norm": 14.78658676147461, "learning_rate": 1.7851612903225807e-05, "loss": 0.32, "step": 9966 }, { "epoch": 64.25060240963856, "grad_norm": 10.84672737121582, "learning_rate": 1.7848387096774193e-05, "loss": 0.4167, "step": 9967 }, { "epoch": 64.2570281124498, "grad_norm": 10.470044136047363, "learning_rate": 1.7845161290322582e-05, "loss": 0.315, "step": 9968 }, { "epoch": 64.26345381526104, "grad_norm": 16.243392944335938, "learning_rate": 1.784193548387097e-05, "loss": 0.4165, "step": 9969 }, { "epoch": 64.26987951807229, "grad_norm": 5.954764366149902, "learning_rate": 1.7838709677419354e-05, "loss": 0.1793, "step": 9970 }, { "epoch": 64.27630522088353, "grad_norm": 5.139837265014648, "learning_rate": 1.7835483870967743e-05, "loss": 0.1911, "step": 9971 }, { "epoch": 64.28273092369479, "grad_norm": 16.09231948852539, "learning_rate": 1.7832258064516132e-05, "loss": 0.3334, "step": 9972 }, { "epoch": 64.28915662650603, "grad_norm": 4.804646968841553, "learning_rate": 1.7829032258064517e-05, "loss": 0.1485, "step": 9973 }, { "epoch": 64.29558232931727, "grad_norm": 5.336272716522217, "learning_rate": 1.7825806451612903e-05, "loss": 0.2391, "step": 9974 }, { "epoch": 64.30200803212851, "grad_norm": 7.523683071136475, "learning_rate": 1.7822580645161292e-05, "loss": 0.3142, "step": 9975 }, { "epoch": 64.30843373493975, "grad_norm": 16.572860717773438, "learning_rate": 1.7819354838709678e-05, "loss": 0.1819, "step": 9976 }, { "epoch": 64.31485943775101, "grad_norm": 7.329529285430908, "learning_rate": 1.7816129032258067e-05, "loss": 0.2615, "step": 9977 }, { "epoch": 64.32128514056225, "grad_norm": 6.400355339050293, "learning_rate": 1.7812903225806453e-05, "loss": 0.1134, "step": 9978 }, { "epoch": 64.3277108433735, "grad_norm": 9.244804382324219, "learning_rate": 1.780967741935484e-05, "loss": 0.3334, "step": 9979 }, { "epoch": 64.33413654618474, "grad_norm": 8.177058219909668, "learning_rate": 1.7806451612903228e-05, "loss": 0.4191, "step": 9980 }, { "epoch": 64.34056224899598, "grad_norm": 6.261157989501953, "learning_rate": 1.7803225806451617e-05, "loss": 0.1579, "step": 9981 }, { "epoch": 64.34698795180722, "grad_norm": 5.050208568572998, "learning_rate": 1.78e-05, "loss": 0.1633, "step": 9982 }, { "epoch": 64.35341365461848, "grad_norm": 18.386980056762695, "learning_rate": 1.7796774193548388e-05, "loss": 0.2362, "step": 9983 }, { "epoch": 64.35983935742972, "grad_norm": 4.865265846252441, "learning_rate": 1.7793548387096777e-05, "loss": 0.192, "step": 9984 }, { "epoch": 64.36626506024096, "grad_norm": 10.274489402770996, "learning_rate": 1.7790322580645163e-05, "loss": 0.1903, "step": 9985 }, { "epoch": 64.3726907630522, "grad_norm": 6.090758323669434, "learning_rate": 1.778709677419355e-05, "loss": 0.3337, "step": 9986 }, { "epoch": 64.37911646586345, "grad_norm": 7.927453517913818, "learning_rate": 1.7783870967741938e-05, "loss": 0.3285, "step": 9987 }, { "epoch": 64.3855421686747, "grad_norm": 10.928866386413574, "learning_rate": 1.7780645161290323e-05, "loss": 0.2181, "step": 9988 }, { "epoch": 64.39196787148595, "grad_norm": 5.565443515777588, "learning_rate": 1.7777419354838712e-05, "loss": 0.2437, "step": 9989 }, { "epoch": 64.39839357429719, "grad_norm": 8.386938095092773, "learning_rate": 1.7774193548387098e-05, "loss": 0.3556, "step": 9990 }, { "epoch": 64.40481927710843, "grad_norm": 10.016650199890137, "learning_rate": 1.7770967741935484e-05, "loss": 0.4288, "step": 9991 }, { "epoch": 64.41124497991967, "grad_norm": 15.904959678649902, "learning_rate": 1.7767741935483873e-05, "loss": 0.2464, "step": 9992 }, { "epoch": 64.41767068273093, "grad_norm": 19.118061065673828, "learning_rate": 1.776451612903226e-05, "loss": 0.4177, "step": 9993 }, { "epoch": 64.42409638554217, "grad_norm": 5.091905117034912, "learning_rate": 1.7761290322580644e-05, "loss": 0.2378, "step": 9994 }, { "epoch": 64.43052208835341, "grad_norm": 8.881449699401855, "learning_rate": 1.7758064516129034e-05, "loss": 0.208, "step": 9995 }, { "epoch": 64.43694779116466, "grad_norm": 7.9268035888671875, "learning_rate": 1.7754838709677423e-05, "loss": 0.4137, "step": 9996 }, { "epoch": 64.4433734939759, "grad_norm": 9.965892791748047, "learning_rate": 1.7751612903225805e-05, "loss": 0.2725, "step": 9997 }, { "epoch": 64.44979919678715, "grad_norm": 17.162939071655273, "learning_rate": 1.7748387096774194e-05, "loss": 0.3656, "step": 9998 }, { "epoch": 64.4562248995984, "grad_norm": 4.1097517013549805, "learning_rate": 1.7745161290322583e-05, "loss": 0.2606, "step": 9999 }, { "epoch": 64.46265060240964, "grad_norm": 4.620773792266846, "learning_rate": 1.774193548387097e-05, "loss": 0.2208, "step": 10000 }, { "epoch": 64.46907630522088, "grad_norm": 11.279522895812988, "learning_rate": 1.7738709677419355e-05, "loss": 0.206, "step": 10001 }, { "epoch": 64.47550200803212, "grad_norm": 7.549008369445801, "learning_rate": 1.7735483870967744e-05, "loss": 0.169, "step": 10002 }, { "epoch": 64.48192771084338, "grad_norm": 6.304736137390137, "learning_rate": 1.773225806451613e-05, "loss": 0.4321, "step": 10003 }, { "epoch": 64.48835341365462, "grad_norm": 9.773940086364746, "learning_rate": 1.772903225806452e-05, "loss": 0.31, "step": 10004 }, { "epoch": 64.49477911646586, "grad_norm": 8.12804126739502, "learning_rate": 1.7725806451612904e-05, "loss": 0.3727, "step": 10005 }, { "epoch": 64.5012048192771, "grad_norm": 16.185523986816406, "learning_rate": 1.772258064516129e-05, "loss": 0.4804, "step": 10006 }, { "epoch": 64.50763052208835, "grad_norm": 8.706141471862793, "learning_rate": 1.771935483870968e-05, "loss": 0.4925, "step": 10007 }, { "epoch": 64.5140562248996, "grad_norm": 7.269900798797607, "learning_rate": 1.7716129032258068e-05, "loss": 0.2279, "step": 10008 }, { "epoch": 64.52048192771085, "grad_norm": 4.939067363739014, "learning_rate": 1.771290322580645e-05, "loss": 0.1894, "step": 10009 }, { "epoch": 64.52690763052209, "grad_norm": 4.155062675476074, "learning_rate": 1.770967741935484e-05, "loss": 0.1506, "step": 10010 }, { "epoch": 64.53333333333333, "grad_norm": 5.78749418258667, "learning_rate": 1.770645161290323e-05, "loss": 0.3325, "step": 10011 }, { "epoch": 64.53975903614457, "grad_norm": 17.813386917114258, "learning_rate": 1.7703225806451614e-05, "loss": 0.3632, "step": 10012 }, { "epoch": 64.54618473895583, "grad_norm": 10.380252838134766, "learning_rate": 1.77e-05, "loss": 0.4781, "step": 10013 }, { "epoch": 64.55261044176707, "grad_norm": 11.268101692199707, "learning_rate": 1.769677419354839e-05, "loss": 0.4637, "step": 10014 }, { "epoch": 64.55903614457831, "grad_norm": 13.469372749328613, "learning_rate": 1.7693548387096775e-05, "loss": 0.3314, "step": 10015 }, { "epoch": 64.56546184738956, "grad_norm": 12.748895645141602, "learning_rate": 1.7690322580645164e-05, "loss": 0.749, "step": 10016 }, { "epoch": 64.5718875502008, "grad_norm": 10.367271423339844, "learning_rate": 1.768709677419355e-05, "loss": 0.2094, "step": 10017 }, { "epoch": 64.57831325301204, "grad_norm": 16.59768295288086, "learning_rate": 1.7683870967741935e-05, "loss": 0.2921, "step": 10018 }, { "epoch": 64.5847389558233, "grad_norm": 21.90956687927246, "learning_rate": 1.7680645161290324e-05, "loss": 0.4351, "step": 10019 }, { "epoch": 64.59116465863454, "grad_norm": 15.705693244934082, "learning_rate": 1.7677419354838713e-05, "loss": 0.2939, "step": 10020 }, { "epoch": 64.59759036144578, "grad_norm": 13.245079040527344, "learning_rate": 1.7674193548387096e-05, "loss": 0.4006, "step": 10021 }, { "epoch": 64.60401606425702, "grad_norm": 11.905587196350098, "learning_rate": 1.7670967741935485e-05, "loss": 0.3861, "step": 10022 }, { "epoch": 64.61044176706827, "grad_norm": 6.731557846069336, "learning_rate": 1.7667741935483874e-05, "loss": 0.221, "step": 10023 }, { "epoch": 64.61686746987952, "grad_norm": 23.731460571289062, "learning_rate": 1.766451612903226e-05, "loss": 0.2883, "step": 10024 }, { "epoch": 64.62329317269077, "grad_norm": 18.010162353515625, "learning_rate": 1.7661290322580645e-05, "loss": 0.4197, "step": 10025 }, { "epoch": 64.62971887550201, "grad_norm": 6.745675086975098, "learning_rate": 1.7658064516129034e-05, "loss": 0.3383, "step": 10026 }, { "epoch": 64.63614457831325, "grad_norm": 6.9675140380859375, "learning_rate": 1.765483870967742e-05, "loss": 0.3917, "step": 10027 }, { "epoch": 64.64257028112449, "grad_norm": 30.533714294433594, "learning_rate": 1.7651612903225806e-05, "loss": 0.4292, "step": 10028 }, { "epoch": 64.64899598393575, "grad_norm": 27.542476654052734, "learning_rate": 1.7648387096774195e-05, "loss": 0.4614, "step": 10029 }, { "epoch": 64.65542168674699, "grad_norm": 11.92068862915039, "learning_rate": 1.764516129032258e-05, "loss": 0.4981, "step": 10030 }, { "epoch": 64.66184738955823, "grad_norm": 9.525392532348633, "learning_rate": 1.764193548387097e-05, "loss": 0.5983, "step": 10031 }, { "epoch": 64.66827309236947, "grad_norm": 13.908788681030273, "learning_rate": 1.7638709677419355e-05, "loss": 0.2395, "step": 10032 }, { "epoch": 64.67469879518072, "grad_norm": 8.119650840759277, "learning_rate": 1.763548387096774e-05, "loss": 0.4556, "step": 10033 }, { "epoch": 64.68112449799197, "grad_norm": 14.492985725402832, "learning_rate": 1.763225806451613e-05, "loss": 0.5003, "step": 10034 }, { "epoch": 64.68755020080322, "grad_norm": 8.481849670410156, "learning_rate": 1.762903225806452e-05, "loss": 0.3278, "step": 10035 }, { "epoch": 64.69397590361446, "grad_norm": 12.663870811462402, "learning_rate": 1.76258064516129e-05, "loss": 0.3199, "step": 10036 }, { "epoch": 64.7004016064257, "grad_norm": 16.337980270385742, "learning_rate": 1.762258064516129e-05, "loss": 0.3455, "step": 10037 }, { "epoch": 64.70682730923694, "grad_norm": 6.645645618438721, "learning_rate": 1.761935483870968e-05, "loss": 0.3207, "step": 10038 }, { "epoch": 64.7132530120482, "grad_norm": 7.15838623046875, "learning_rate": 1.7616129032258066e-05, "loss": 0.2675, "step": 10039 }, { "epoch": 64.71967871485944, "grad_norm": 9.976278305053711, "learning_rate": 1.761290322580645e-05, "loss": 0.4306, "step": 10040 }, { "epoch": 64.72610441767068, "grad_norm": 13.084236145019531, "learning_rate": 1.760967741935484e-05, "loss": 0.3015, "step": 10041 }, { "epoch": 64.73253012048193, "grad_norm": 7.27763032913208, "learning_rate": 1.7606451612903226e-05, "loss": 0.2099, "step": 10042 }, { "epoch": 64.73895582329317, "grad_norm": 6.973329067230225, "learning_rate": 1.7603225806451615e-05, "loss": 0.4628, "step": 10043 }, { "epoch": 64.74538152610442, "grad_norm": 10.32335376739502, "learning_rate": 1.76e-05, "loss": 0.5662, "step": 10044 }, { "epoch": 64.75180722891567, "grad_norm": 6.529433727264404, "learning_rate": 1.7596774193548387e-05, "loss": 0.1481, "step": 10045 }, { "epoch": 64.75823293172691, "grad_norm": 26.05270767211914, "learning_rate": 1.7593548387096776e-05, "loss": 0.3966, "step": 10046 }, { "epoch": 64.76465863453815, "grad_norm": 13.256425857543945, "learning_rate": 1.7590322580645165e-05, "loss": 0.5956, "step": 10047 }, { "epoch": 64.7710843373494, "grad_norm": 3.6206772327423096, "learning_rate": 1.7587096774193547e-05, "loss": 0.1892, "step": 10048 }, { "epoch": 64.77751004016065, "grad_norm": 17.8087158203125, "learning_rate": 1.7583870967741936e-05, "loss": 0.3103, "step": 10049 }, { "epoch": 64.78393574297189, "grad_norm": 5.036778926849365, "learning_rate": 1.7580645161290325e-05, "loss": 0.171, "step": 10050 }, { "epoch": 64.79036144578313, "grad_norm": 8.955246925354004, "learning_rate": 1.757741935483871e-05, "loss": 0.2857, "step": 10051 }, { "epoch": 64.79678714859438, "grad_norm": 8.175972938537598, "learning_rate": 1.7574193548387097e-05, "loss": 0.5615, "step": 10052 }, { "epoch": 64.80321285140562, "grad_norm": 9.33311939239502, "learning_rate": 1.7570967741935486e-05, "loss": 0.2717, "step": 10053 }, { "epoch": 64.80963855421686, "grad_norm": 4.937364101409912, "learning_rate": 1.756774193548387e-05, "loss": 0.1569, "step": 10054 }, { "epoch": 64.81606425702812, "grad_norm": 10.148744583129883, "learning_rate": 1.756451612903226e-05, "loss": 0.2933, "step": 10055 }, { "epoch": 64.82248995983936, "grad_norm": 13.261999130249023, "learning_rate": 1.7561290322580646e-05, "loss": 0.7515, "step": 10056 }, { "epoch": 64.8289156626506, "grad_norm": 9.244780540466309, "learning_rate": 1.7558064516129032e-05, "loss": 0.4054, "step": 10057 }, { "epoch": 64.83534136546184, "grad_norm": 9.398358345031738, "learning_rate": 1.755483870967742e-05, "loss": 0.4222, "step": 10058 }, { "epoch": 64.84176706827309, "grad_norm": 16.2795352935791, "learning_rate": 1.755161290322581e-05, "loss": 0.6061, "step": 10059 }, { "epoch": 64.84819277108434, "grad_norm": 12.500030517578125, "learning_rate": 1.7548387096774193e-05, "loss": 0.5718, "step": 10060 }, { "epoch": 64.85461847389558, "grad_norm": 7.16551399230957, "learning_rate": 1.754516129032258e-05, "loss": 0.3031, "step": 10061 }, { "epoch": 64.86104417670683, "grad_norm": 8.04210376739502, "learning_rate": 1.754193548387097e-05, "loss": 0.2436, "step": 10062 }, { "epoch": 64.86746987951807, "grad_norm": 7.099430084228516, "learning_rate": 1.7538709677419353e-05, "loss": 0.2607, "step": 10063 }, { "epoch": 64.87389558232931, "grad_norm": 8.888876914978027, "learning_rate": 1.7535483870967742e-05, "loss": 0.3941, "step": 10064 }, { "epoch": 64.88032128514057, "grad_norm": 5.115845203399658, "learning_rate": 1.753225806451613e-05, "loss": 0.2612, "step": 10065 }, { "epoch": 64.88674698795181, "grad_norm": 6.78074312210083, "learning_rate": 1.7529032258064517e-05, "loss": 0.3027, "step": 10066 }, { "epoch": 64.89317269076305, "grad_norm": 9.259164810180664, "learning_rate": 1.7525806451612903e-05, "loss": 0.3433, "step": 10067 }, { "epoch": 64.8995983935743, "grad_norm": 16.067644119262695, "learning_rate": 1.7522580645161292e-05, "loss": 0.2296, "step": 10068 }, { "epoch": 64.90602409638554, "grad_norm": 7.371352672576904, "learning_rate": 1.7519354838709677e-05, "loss": 0.2328, "step": 10069 }, { "epoch": 64.91244979919679, "grad_norm": 8.183927536010742, "learning_rate": 1.7516129032258067e-05, "loss": 0.2714, "step": 10070 }, { "epoch": 64.91887550200804, "grad_norm": 6.367239475250244, "learning_rate": 1.7512903225806452e-05, "loss": 0.3738, "step": 10071 }, { "epoch": 64.92530120481928, "grad_norm": 9.575989723205566, "learning_rate": 1.7509677419354838e-05, "loss": 0.3652, "step": 10072 }, { "epoch": 64.93172690763052, "grad_norm": 10.017301559448242, "learning_rate": 1.7506451612903227e-05, "loss": 0.4196, "step": 10073 }, { "epoch": 64.93815261044176, "grad_norm": 13.609647750854492, "learning_rate": 1.7503225806451616e-05, "loss": 0.4932, "step": 10074 }, { "epoch": 64.94457831325302, "grad_norm": 6.404902458190918, "learning_rate": 1.75e-05, "loss": 0.3951, "step": 10075 }, { "epoch": 64.95100401606426, "grad_norm": 11.248318672180176, "learning_rate": 1.7496774193548388e-05, "loss": 0.5114, "step": 10076 }, { "epoch": 64.9574297188755, "grad_norm": 8.378954887390137, "learning_rate": 1.7493548387096777e-05, "loss": 0.3531, "step": 10077 }, { "epoch": 64.96385542168674, "grad_norm": 10.642579078674316, "learning_rate": 1.7490322580645162e-05, "loss": 0.2436, "step": 10078 }, { "epoch": 64.97028112449799, "grad_norm": 6.523393154144287, "learning_rate": 1.7487096774193548e-05, "loss": 0.1881, "step": 10079 }, { "epoch": 64.97670682730924, "grad_norm": 6.16199254989624, "learning_rate": 1.7483870967741937e-05, "loss": 0.2897, "step": 10080 }, { "epoch": 64.98313253012049, "grad_norm": 4.203366279602051, "learning_rate": 1.7480645161290323e-05, "loss": 0.1554, "step": 10081 }, { "epoch": 64.98955823293173, "grad_norm": 6.520161151885986, "learning_rate": 1.7477419354838712e-05, "loss": 0.2987, "step": 10082 }, { "epoch": 64.99598393574297, "grad_norm": 16.249130249023438, "learning_rate": 1.7474193548387098e-05, "loss": 0.2384, "step": 10083 }, { "epoch": 65.00562248995983, "grad_norm": 24.976037979125977, "learning_rate": 1.7470967741935483e-05, "loss": 0.365, "step": 10084 }, { "epoch": 65.01204819277109, "grad_norm": 72.96903228759766, "learning_rate": 1.7467741935483872e-05, "loss": 0.3385, "step": 10085 }, { "epoch": 65.01847389558233, "grad_norm": 20.085847854614258, "learning_rate": 1.746451612903226e-05, "loss": 0.12, "step": 10086 }, { "epoch": 65.02489959839357, "grad_norm": 11.788900375366211, "learning_rate": 1.7461290322580644e-05, "loss": 0.1948, "step": 10087 }, { "epoch": 65.03132530120482, "grad_norm": 28.783374786376953, "learning_rate": 1.7458064516129033e-05, "loss": 0.6806, "step": 10088 }, { "epoch": 65.03775100401606, "grad_norm": 18.872350692749023, "learning_rate": 1.7454838709677422e-05, "loss": 0.5097, "step": 10089 }, { "epoch": 65.04417670682732, "grad_norm": 6.048608779907227, "learning_rate": 1.7451612903225808e-05, "loss": 0.2168, "step": 10090 }, { "epoch": 65.05060240963856, "grad_norm": 3.9220712184906006, "learning_rate": 1.7448387096774193e-05, "loss": 0.2469, "step": 10091 }, { "epoch": 65.0570281124498, "grad_norm": 4.4743852615356445, "learning_rate": 1.7445161290322583e-05, "loss": 0.3293, "step": 10092 }, { "epoch": 65.06345381526104, "grad_norm": 25.630615234375, "learning_rate": 1.7441935483870968e-05, "loss": 0.3556, "step": 10093 }, { "epoch": 65.06987951807228, "grad_norm": 16.669395446777344, "learning_rate": 1.7438709677419357e-05, "loss": 0.4097, "step": 10094 }, { "epoch": 65.07630522088354, "grad_norm": 14.726630210876465, "learning_rate": 1.7435483870967743e-05, "loss": 0.4296, "step": 10095 }, { "epoch": 65.08273092369478, "grad_norm": 36.570045471191406, "learning_rate": 1.743225806451613e-05, "loss": 0.6174, "step": 10096 }, { "epoch": 65.08915662650602, "grad_norm": 7.95161771774292, "learning_rate": 1.7429032258064518e-05, "loss": 0.0518, "step": 10097 }, { "epoch": 65.09558232931727, "grad_norm": 15.68359661102295, "learning_rate": 1.7425806451612904e-05, "loss": 0.33, "step": 10098 }, { "epoch": 65.10200803212851, "grad_norm": 16.19686508178711, "learning_rate": 1.742258064516129e-05, "loss": 0.386, "step": 10099 }, { "epoch": 65.10843373493977, "grad_norm": 12.856230735778809, "learning_rate": 1.741935483870968e-05, "loss": 0.3223, "step": 10100 }, { "epoch": 65.11485943775101, "grad_norm": 3.355464458465576, "learning_rate": 1.7416129032258068e-05, "loss": 0.2649, "step": 10101 }, { "epoch": 65.12128514056225, "grad_norm": 20.452178955078125, "learning_rate": 1.741290322580645e-05, "loss": 0.3655, "step": 10102 }, { "epoch": 65.12771084337349, "grad_norm": 69.774658203125, "learning_rate": 1.740967741935484e-05, "loss": 0.1663, "step": 10103 }, { "epoch": 65.13413654618473, "grad_norm": 5.630690574645996, "learning_rate": 1.7406451612903228e-05, "loss": 0.3414, "step": 10104 }, { "epoch": 65.14056224899599, "grad_norm": 19.186704635620117, "learning_rate": 1.7403225806451614e-05, "loss": 0.3691, "step": 10105 }, { "epoch": 65.14698795180723, "grad_norm": 39.19611358642578, "learning_rate": 1.74e-05, "loss": 0.2698, "step": 10106 }, { "epoch": 65.15341365461848, "grad_norm": 9.164335250854492, "learning_rate": 1.739677419354839e-05, "loss": 0.2808, "step": 10107 }, { "epoch": 65.15983935742972, "grad_norm": 10.050246238708496, "learning_rate": 1.7393548387096774e-05, "loss": 0.1772, "step": 10108 }, { "epoch": 65.16626506024096, "grad_norm": 10.16596508026123, "learning_rate": 1.7390322580645163e-05, "loss": 0.6774, "step": 10109 }, { "epoch": 65.1726907630522, "grad_norm": 22.29489517211914, "learning_rate": 1.738709677419355e-05, "loss": 0.3978, "step": 10110 }, { "epoch": 65.17911646586346, "grad_norm": 19.17732048034668, "learning_rate": 1.7383870967741935e-05, "loss": 0.4983, "step": 10111 }, { "epoch": 65.1855421686747, "grad_norm": 263.54302978515625, "learning_rate": 1.7380645161290324e-05, "loss": 0.3572, "step": 10112 }, { "epoch": 65.19196787148594, "grad_norm": 31.946731567382812, "learning_rate": 1.7377419354838713e-05, "loss": 0.5395, "step": 10113 }, { "epoch": 65.19839357429719, "grad_norm": 37.8673210144043, "learning_rate": 1.7374193548387095e-05, "loss": 0.4046, "step": 10114 }, { "epoch": 65.20481927710843, "grad_norm": 5.366594314575195, "learning_rate": 1.7370967741935484e-05, "loss": 0.1917, "step": 10115 }, { "epoch": 65.21124497991968, "grad_norm": 23.015079498291016, "learning_rate": 1.7367741935483873e-05, "loss": 0.4401, "step": 10116 }, { "epoch": 65.21767068273093, "grad_norm": 9.898681640625, "learning_rate": 1.736451612903226e-05, "loss": 0.1889, "step": 10117 }, { "epoch": 65.22409638554217, "grad_norm": 14.608948707580566, "learning_rate": 1.7361290322580645e-05, "loss": 0.2088, "step": 10118 }, { "epoch": 65.23052208835341, "grad_norm": 34.54685974121094, "learning_rate": 1.7358064516129034e-05, "loss": 0.3576, "step": 10119 }, { "epoch": 65.23694779116465, "grad_norm": 14.18813419342041, "learning_rate": 1.735483870967742e-05, "loss": 0.4812, "step": 10120 }, { "epoch": 65.24337349397591, "grad_norm": 9.647610664367676, "learning_rate": 1.735161290322581e-05, "loss": 0.1978, "step": 10121 }, { "epoch": 65.24979919678715, "grad_norm": 13.070293426513672, "learning_rate": 1.7348387096774194e-05, "loss": 0.4628, "step": 10122 }, { "epoch": 65.2562248995984, "grad_norm": 20.830289840698242, "learning_rate": 1.734516129032258e-05, "loss": 0.1523, "step": 10123 }, { "epoch": 65.26265060240964, "grad_norm": 6.429520606994629, "learning_rate": 1.734193548387097e-05, "loss": 0.3774, "step": 10124 }, { "epoch": 65.26907630522088, "grad_norm": 2.416311025619507, "learning_rate": 1.733870967741936e-05, "loss": 0.1235, "step": 10125 }, { "epoch": 65.27550200803213, "grad_norm": 23.76166343688965, "learning_rate": 1.733548387096774e-05, "loss": 0.535, "step": 10126 }, { "epoch": 65.28192771084338, "grad_norm": 21.751108169555664, "learning_rate": 1.733225806451613e-05, "loss": 0.476, "step": 10127 }, { "epoch": 65.28835341365462, "grad_norm": 8.325493812561035, "learning_rate": 1.732903225806452e-05, "loss": 0.4017, "step": 10128 }, { "epoch": 65.29477911646586, "grad_norm": 27.706327438354492, "learning_rate": 1.7325806451612905e-05, "loss": 0.6634, "step": 10129 }, { "epoch": 65.3012048192771, "grad_norm": 5.990884304046631, "learning_rate": 1.732258064516129e-05, "loss": 0.4528, "step": 10130 }, { "epoch": 65.30763052208836, "grad_norm": 12.45839786529541, "learning_rate": 1.731935483870968e-05, "loss": 0.2372, "step": 10131 }, { "epoch": 65.3140562248996, "grad_norm": 20.02515411376953, "learning_rate": 1.7316129032258065e-05, "loss": 0.6005, "step": 10132 }, { "epoch": 65.32048192771084, "grad_norm": 33.6424560546875, "learning_rate": 1.731290322580645e-05, "loss": 0.5125, "step": 10133 }, { "epoch": 65.32690763052209, "grad_norm": 51.45632553100586, "learning_rate": 1.730967741935484e-05, "loss": 0.6645, "step": 10134 }, { "epoch": 65.33333333333333, "grad_norm": 20.849992752075195, "learning_rate": 1.7306451612903226e-05, "loss": 0.6236, "step": 10135 }, { "epoch": 65.33975903614459, "grad_norm": 10.777569770812988, "learning_rate": 1.7303225806451615e-05, "loss": 0.2567, "step": 10136 }, { "epoch": 65.34618473895583, "grad_norm": 13.656564712524414, "learning_rate": 1.73e-05, "loss": 0.391, "step": 10137 }, { "epoch": 65.35261044176707, "grad_norm": 26.373607635498047, "learning_rate": 1.7296774193548386e-05, "loss": 0.2607, "step": 10138 }, { "epoch": 65.35903614457831, "grad_norm": 22.02272605895996, "learning_rate": 1.7293548387096775e-05, "loss": 0.4663, "step": 10139 }, { "epoch": 65.36546184738955, "grad_norm": 7.808395862579346, "learning_rate": 1.7290322580645164e-05, "loss": 0.481, "step": 10140 }, { "epoch": 65.3718875502008, "grad_norm": 1.3973948955535889, "learning_rate": 1.7287096774193547e-05, "loss": 0.2855, "step": 10141 }, { "epoch": 65.37831325301205, "grad_norm": 26.777027130126953, "learning_rate": 1.7283870967741936e-05, "loss": 0.295, "step": 10142 }, { "epoch": 65.3847389558233, "grad_norm": 5.609462261199951, "learning_rate": 1.7280645161290325e-05, "loss": 0.1665, "step": 10143 }, { "epoch": 65.39116465863454, "grad_norm": 3.952463150024414, "learning_rate": 1.727741935483871e-05, "loss": 0.3663, "step": 10144 }, { "epoch": 65.39759036144578, "grad_norm": 13.097654342651367, "learning_rate": 1.7274193548387096e-05, "loss": 0.405, "step": 10145 }, { "epoch": 65.40401606425702, "grad_norm": 16.805809020996094, "learning_rate": 1.7270967741935485e-05, "loss": 0.405, "step": 10146 }, { "epoch": 65.41044176706828, "grad_norm": 27.52912139892578, "learning_rate": 1.726774193548387e-05, "loss": 0.3975, "step": 10147 }, { "epoch": 65.41686746987952, "grad_norm": 16.823646545410156, "learning_rate": 1.726451612903226e-05, "loss": 0.2472, "step": 10148 }, { "epoch": 65.42329317269076, "grad_norm": 25.13182258605957, "learning_rate": 1.7261290322580646e-05, "loss": 0.3484, "step": 10149 }, { "epoch": 65.429718875502, "grad_norm": 4.677748680114746, "learning_rate": 1.725806451612903e-05, "loss": 0.064, "step": 10150 }, { "epoch": 65.43614457831325, "grad_norm": 24.092857360839844, "learning_rate": 1.725483870967742e-05, "loss": 0.4722, "step": 10151 }, { "epoch": 65.4425702811245, "grad_norm": 3.8807711601257324, "learning_rate": 1.725161290322581e-05, "loss": 0.0738, "step": 10152 }, { "epoch": 65.44899598393575, "grad_norm": 19.195890426635742, "learning_rate": 1.7248387096774192e-05, "loss": 0.2516, "step": 10153 }, { "epoch": 65.45542168674699, "grad_norm": 3.586853265762329, "learning_rate": 1.724516129032258e-05, "loss": 0.3304, "step": 10154 }, { "epoch": 65.46184738955823, "grad_norm": 9.031302452087402, "learning_rate": 1.724193548387097e-05, "loss": 0.2873, "step": 10155 }, { "epoch": 65.46827309236947, "grad_norm": 6.637688159942627, "learning_rate": 1.7238709677419356e-05, "loss": 0.2381, "step": 10156 }, { "epoch": 65.47469879518073, "grad_norm": 10.846821784973145, "learning_rate": 1.723548387096774e-05, "loss": 0.1792, "step": 10157 }, { "epoch": 65.48112449799197, "grad_norm": 6.318206310272217, "learning_rate": 1.723225806451613e-05, "loss": 0.0911, "step": 10158 }, { "epoch": 65.48755020080321, "grad_norm": 3.721013069152832, "learning_rate": 1.7229032258064516e-05, "loss": 0.2971, "step": 10159 }, { "epoch": 65.49397590361446, "grad_norm": 20.46941566467285, "learning_rate": 1.7225806451612906e-05, "loss": 0.4569, "step": 10160 }, { "epoch": 65.5004016064257, "grad_norm": 15.976509094238281, "learning_rate": 1.722258064516129e-05, "loss": 0.339, "step": 10161 }, { "epoch": 65.50682730923695, "grad_norm": 13.596953392028809, "learning_rate": 1.7219354838709677e-05, "loss": 0.3537, "step": 10162 }, { "epoch": 65.5132530120482, "grad_norm": 31.18654441833496, "learning_rate": 1.7216129032258066e-05, "loss": 0.4503, "step": 10163 }, { "epoch": 65.51967871485944, "grad_norm": 17.201208114624023, "learning_rate": 1.7212903225806455e-05, "loss": 0.7192, "step": 10164 }, { "epoch": 65.52610441767068, "grad_norm": 6.8100266456604, "learning_rate": 1.7209677419354837e-05, "loss": 0.5703, "step": 10165 }, { "epoch": 65.53253012048192, "grad_norm": 22.980632781982422, "learning_rate": 1.7206451612903227e-05, "loss": 0.288, "step": 10166 }, { "epoch": 65.53895582329318, "grad_norm": 14.049704551696777, "learning_rate": 1.7203225806451616e-05, "loss": 0.2033, "step": 10167 }, { "epoch": 65.54538152610442, "grad_norm": 2.9493134021759033, "learning_rate": 1.7199999999999998e-05, "loss": 0.3421, "step": 10168 }, { "epoch": 65.55180722891566, "grad_norm": 5.119200706481934, "learning_rate": 1.7196774193548387e-05, "loss": 0.3408, "step": 10169 }, { "epoch": 65.5582329317269, "grad_norm": 4.744615077972412, "learning_rate": 1.7193548387096776e-05, "loss": 0.3369, "step": 10170 }, { "epoch": 65.56465863453815, "grad_norm": 24.767290115356445, "learning_rate": 1.7190322580645162e-05, "loss": 0.2774, "step": 10171 }, { "epoch": 65.5710843373494, "grad_norm": 14.749650955200195, "learning_rate": 1.7187096774193548e-05, "loss": 0.3577, "step": 10172 }, { "epoch": 65.57751004016065, "grad_norm": 10.490717887878418, "learning_rate": 1.7183870967741937e-05, "loss": 0.4517, "step": 10173 }, { "epoch": 65.58393574297189, "grad_norm": 6.103773593902588, "learning_rate": 1.7180645161290322e-05, "loss": 0.2505, "step": 10174 }, { "epoch": 65.59036144578313, "grad_norm": 8.709396362304688, "learning_rate": 1.717741935483871e-05, "loss": 0.3716, "step": 10175 }, { "epoch": 65.59678714859437, "grad_norm": 19.045108795166016, "learning_rate": 1.7174193548387097e-05, "loss": 0.459, "step": 10176 }, { "epoch": 65.60321285140563, "grad_norm": 16.28843116760254, "learning_rate": 1.7170967741935483e-05, "loss": 0.2063, "step": 10177 }, { "epoch": 65.60963855421687, "grad_norm": 8.9552640914917, "learning_rate": 1.7167741935483872e-05, "loss": 0.3558, "step": 10178 }, { "epoch": 65.61606425702811, "grad_norm": 6.538630962371826, "learning_rate": 1.716451612903226e-05, "loss": 0.1802, "step": 10179 }, { "epoch": 65.62248995983936, "grad_norm": 2.2719807624816895, "learning_rate": 1.7161290322580643e-05, "loss": 0.3687, "step": 10180 }, { "epoch": 65.6289156626506, "grad_norm": 8.742609977722168, "learning_rate": 1.7158064516129032e-05, "loss": 0.3744, "step": 10181 }, { "epoch": 65.63534136546184, "grad_norm": 15.617135047912598, "learning_rate": 1.715483870967742e-05, "loss": 0.6115, "step": 10182 }, { "epoch": 65.6417670682731, "grad_norm": 8.013773918151855, "learning_rate": 1.7151612903225807e-05, "loss": 0.2744, "step": 10183 }, { "epoch": 65.64819277108434, "grad_norm": 12.639318466186523, "learning_rate": 1.7148387096774193e-05, "loss": 0.206, "step": 10184 }, { "epoch": 65.65461847389558, "grad_norm": 23.663047790527344, "learning_rate": 1.7145161290322582e-05, "loss": 0.3683, "step": 10185 }, { "epoch": 65.66104417670682, "grad_norm": 14.414694786071777, "learning_rate": 1.7141935483870968e-05, "loss": 0.416, "step": 10186 }, { "epoch": 65.66746987951807, "grad_norm": 15.948834419250488, "learning_rate": 1.7138709677419357e-05, "loss": 0.4622, "step": 10187 }, { "epoch": 65.67389558232932, "grad_norm": 5.143244743347168, "learning_rate": 1.7135483870967743e-05, "loss": 0.2543, "step": 10188 }, { "epoch": 65.68032128514056, "grad_norm": 15.468269348144531, "learning_rate": 1.7132258064516128e-05, "loss": 0.2284, "step": 10189 }, { "epoch": 65.6867469879518, "grad_norm": 32.6530876159668, "learning_rate": 1.7129032258064517e-05, "loss": 0.1957, "step": 10190 }, { "epoch": 65.69317269076305, "grad_norm": 29.200088500976562, "learning_rate": 1.7125806451612906e-05, "loss": 0.2565, "step": 10191 }, { "epoch": 65.69959839357429, "grad_norm": 19.17777442932129, "learning_rate": 1.712258064516129e-05, "loss": 0.2251, "step": 10192 }, { "epoch": 65.70602409638555, "grad_norm": 33.254371643066406, "learning_rate": 1.7119354838709678e-05, "loss": 0.38, "step": 10193 }, { "epoch": 65.71244979919679, "grad_norm": 11.348870277404785, "learning_rate": 1.7116129032258067e-05, "loss": 0.8052, "step": 10194 }, { "epoch": 65.71887550200803, "grad_norm": 19.012712478637695, "learning_rate": 1.7112903225806453e-05, "loss": 0.2309, "step": 10195 }, { "epoch": 65.72530120481927, "grad_norm": 12.464212417602539, "learning_rate": 1.710967741935484e-05, "loss": 0.4612, "step": 10196 }, { "epoch": 65.73172690763052, "grad_norm": 37.68912887573242, "learning_rate": 1.7106451612903227e-05, "loss": 0.4514, "step": 10197 }, { "epoch": 65.73815261044177, "grad_norm": 16.072551727294922, "learning_rate": 1.7103225806451613e-05, "loss": 0.3737, "step": 10198 }, { "epoch": 65.74457831325302, "grad_norm": 22.91742706298828, "learning_rate": 1.7100000000000002e-05, "loss": 0.291, "step": 10199 }, { "epoch": 65.75100401606426, "grad_norm": 19.020116806030273, "learning_rate": 1.7096774193548388e-05, "loss": 0.3167, "step": 10200 }, { "epoch": 65.7574297188755, "grad_norm": 8.6586332321167, "learning_rate": 1.7093548387096774e-05, "loss": 0.2518, "step": 10201 }, { "epoch": 65.76385542168674, "grad_norm": 7.950282096862793, "learning_rate": 1.7090322580645163e-05, "loss": 0.3439, "step": 10202 }, { "epoch": 65.770281124498, "grad_norm": 11.061857223510742, "learning_rate": 1.708709677419355e-05, "loss": 0.3292, "step": 10203 }, { "epoch": 65.77670682730924, "grad_norm": 16.504924774169922, "learning_rate": 1.7083870967741934e-05, "loss": 0.5399, "step": 10204 }, { "epoch": 65.78313253012048, "grad_norm": 4.376203536987305, "learning_rate": 1.7080645161290323e-05, "loss": 0.5281, "step": 10205 }, { "epoch": 65.78955823293173, "grad_norm": 4.496923923492432, "learning_rate": 1.7077419354838712e-05, "loss": 0.4342, "step": 10206 }, { "epoch": 65.79598393574297, "grad_norm": 39.51161575317383, "learning_rate": 1.7074193548387095e-05, "loss": 0.5851, "step": 10207 }, { "epoch": 65.80240963855422, "grad_norm": 9.51400089263916, "learning_rate": 1.7070967741935484e-05, "loss": 0.353, "step": 10208 }, { "epoch": 65.80883534136547, "grad_norm": 2.7644736766815186, "learning_rate": 1.7067741935483873e-05, "loss": 0.2974, "step": 10209 }, { "epoch": 65.81526104417671, "grad_norm": 16.73200225830078, "learning_rate": 1.706451612903226e-05, "loss": 0.4025, "step": 10210 }, { "epoch": 65.82168674698795, "grad_norm": 23.827062606811523, "learning_rate": 1.7061290322580644e-05, "loss": 0.2523, "step": 10211 }, { "epoch": 65.82811244979919, "grad_norm": 29.053634643554688, "learning_rate": 1.7058064516129033e-05, "loss": 0.4084, "step": 10212 }, { "epoch": 65.83453815261043, "grad_norm": 9.288148880004883, "learning_rate": 1.705483870967742e-05, "loss": 0.275, "step": 10213 }, { "epoch": 65.84096385542169, "grad_norm": 13.181745529174805, "learning_rate": 1.7051612903225808e-05, "loss": 0.3811, "step": 10214 }, { "epoch": 65.84738955823293, "grad_norm": 96.73228454589844, "learning_rate": 1.7048387096774194e-05, "loss": 0.588, "step": 10215 }, { "epoch": 65.85381526104418, "grad_norm": 7.410951614379883, "learning_rate": 1.704516129032258e-05, "loss": 0.2149, "step": 10216 }, { "epoch": 65.86024096385542, "grad_norm": 17.249204635620117, "learning_rate": 1.704193548387097e-05, "loss": 0.5934, "step": 10217 }, { "epoch": 65.86666666666666, "grad_norm": 22.445327758789062, "learning_rate": 1.7038709677419358e-05, "loss": 0.435, "step": 10218 }, { "epoch": 65.87309236947792, "grad_norm": 12.7310209274292, "learning_rate": 1.703548387096774e-05, "loss": 0.4846, "step": 10219 }, { "epoch": 65.87951807228916, "grad_norm": 18.16791534423828, "learning_rate": 1.703225806451613e-05, "loss": 0.5502, "step": 10220 }, { "epoch": 65.8859437751004, "grad_norm": 15.285391807556152, "learning_rate": 1.702903225806452e-05, "loss": 0.3263, "step": 10221 }, { "epoch": 65.89236947791164, "grad_norm": 25.587600708007812, "learning_rate": 1.7025806451612904e-05, "loss": 0.5953, "step": 10222 }, { "epoch": 65.89879518072289, "grad_norm": 10.022032737731934, "learning_rate": 1.702258064516129e-05, "loss": 0.4525, "step": 10223 }, { "epoch": 65.90522088353414, "grad_norm": 28.229843139648438, "learning_rate": 1.701935483870968e-05, "loss": 0.5679, "step": 10224 }, { "epoch": 65.91164658634538, "grad_norm": 8.888211250305176, "learning_rate": 1.7016129032258065e-05, "loss": 0.2594, "step": 10225 }, { "epoch": 65.91807228915663, "grad_norm": 11.572032928466797, "learning_rate": 1.7012903225806454e-05, "loss": 0.3464, "step": 10226 }, { "epoch": 65.92449799196787, "grad_norm": 3.6982429027557373, "learning_rate": 1.700967741935484e-05, "loss": 0.3423, "step": 10227 }, { "epoch": 65.93092369477911, "grad_norm": 10.921606063842773, "learning_rate": 1.7006451612903225e-05, "loss": 0.3915, "step": 10228 }, { "epoch": 65.93734939759037, "grad_norm": 12.65984058380127, "learning_rate": 1.7003225806451614e-05, "loss": 0.1923, "step": 10229 }, { "epoch": 65.94377510040161, "grad_norm": 34.16386795043945, "learning_rate": 1.7000000000000003e-05, "loss": 0.3143, "step": 10230 }, { "epoch": 65.95020080321285, "grad_norm": 9.376321792602539, "learning_rate": 1.6996774193548386e-05, "loss": 0.2932, "step": 10231 }, { "epoch": 65.9566265060241, "grad_norm": 13.886521339416504, "learning_rate": 1.6993548387096775e-05, "loss": 0.2877, "step": 10232 }, { "epoch": 65.96305220883534, "grad_norm": 22.125818252563477, "learning_rate": 1.6990322580645164e-05, "loss": 0.7489, "step": 10233 }, { "epoch": 65.96947791164659, "grad_norm": 2.498638153076172, "learning_rate": 1.698709677419355e-05, "loss": 0.1932, "step": 10234 }, { "epoch": 65.97590361445783, "grad_norm": 29.615745544433594, "learning_rate": 1.6983870967741935e-05, "loss": 0.5716, "step": 10235 }, { "epoch": 65.98232931726908, "grad_norm": 19.13153076171875, "learning_rate": 1.6980645161290324e-05, "loss": 0.5858, "step": 10236 }, { "epoch": 65.98875502008032, "grad_norm": 14.91498851776123, "learning_rate": 1.697741935483871e-05, "loss": 0.2559, "step": 10237 }, { "epoch": 65.99518072289156, "grad_norm": 22.39655113220215, "learning_rate": 1.69741935483871e-05, "loss": 0.3508, "step": 10238 }, { "epoch": 66.00481927710844, "grad_norm": 15.170304298400879, "learning_rate": 1.6970967741935485e-05, "loss": 0.61, "step": 10239 }, { "epoch": 66.01124497991968, "grad_norm": 5.10971212387085, "learning_rate": 1.696774193548387e-05, "loss": 0.1964, "step": 10240 }, { "epoch": 66.01767068273092, "grad_norm": 5.491192817687988, "learning_rate": 1.696451612903226e-05, "loss": 0.2151, "step": 10241 }, { "epoch": 66.02409638554217, "grad_norm": 18.150604248046875, "learning_rate": 1.6961290322580645e-05, "loss": 0.2029, "step": 10242 }, { "epoch": 66.03052208835341, "grad_norm": 7.986854553222656, "learning_rate": 1.695806451612903e-05, "loss": 0.2796, "step": 10243 }, { "epoch": 66.03694779116466, "grad_norm": 14.945721626281738, "learning_rate": 1.695483870967742e-05, "loss": 0.5202, "step": 10244 }, { "epoch": 66.0433734939759, "grad_norm": 13.018848419189453, "learning_rate": 1.695161290322581e-05, "loss": 0.3086, "step": 10245 }, { "epoch": 66.04979919678715, "grad_norm": 36.54106140136719, "learning_rate": 1.694838709677419e-05, "loss": 0.5158, "step": 10246 }, { "epoch": 66.05622489959839, "grad_norm": 4.811461925506592, "learning_rate": 1.694516129032258e-05, "loss": 0.3301, "step": 10247 }, { "epoch": 66.06265060240963, "grad_norm": 7.336611747741699, "learning_rate": 1.694193548387097e-05, "loss": 0.1539, "step": 10248 }, { "epoch": 66.06907630522089, "grad_norm": 13.797613143920898, "learning_rate": 1.6938709677419355e-05, "loss": 0.2441, "step": 10249 }, { "epoch": 66.07550200803213, "grad_norm": 14.07485294342041, "learning_rate": 1.693548387096774e-05, "loss": 0.3718, "step": 10250 }, { "epoch": 66.08192771084337, "grad_norm": 18.459951400756836, "learning_rate": 1.693225806451613e-05, "loss": 0.3097, "step": 10251 }, { "epoch": 66.08835341365462, "grad_norm": 20.24520492553711, "learning_rate": 1.6929032258064516e-05, "loss": 0.2907, "step": 10252 }, { "epoch": 66.09477911646586, "grad_norm": 21.637802124023438, "learning_rate": 1.6925806451612905e-05, "loss": 0.5276, "step": 10253 }, { "epoch": 66.10120481927711, "grad_norm": 3.401071310043335, "learning_rate": 1.692258064516129e-05, "loss": 0.4129, "step": 10254 }, { "epoch": 66.10763052208836, "grad_norm": 18.707550048828125, "learning_rate": 1.6919354838709676e-05, "loss": 0.3717, "step": 10255 }, { "epoch": 66.1140562248996, "grad_norm": 11.408756256103516, "learning_rate": 1.6916129032258066e-05, "loss": 0.4306, "step": 10256 }, { "epoch": 66.12048192771084, "grad_norm": 15.259897232055664, "learning_rate": 1.6912903225806455e-05, "loss": 0.3564, "step": 10257 }, { "epoch": 66.12690763052208, "grad_norm": 6.181063175201416, "learning_rate": 1.6909677419354837e-05, "loss": 0.313, "step": 10258 }, { "epoch": 66.13333333333334, "grad_norm": 9.26055908203125, "learning_rate": 1.6906451612903226e-05, "loss": 0.7424, "step": 10259 }, { "epoch": 66.13975903614458, "grad_norm": 8.03606128692627, "learning_rate": 1.6903225806451615e-05, "loss": 0.3784, "step": 10260 }, { "epoch": 66.14618473895582, "grad_norm": 9.534388542175293, "learning_rate": 1.69e-05, "loss": 0.3413, "step": 10261 }, { "epoch": 66.15261044176707, "grad_norm": 18.04337501525879, "learning_rate": 1.6896774193548387e-05, "loss": 0.4722, "step": 10262 }, { "epoch": 66.15903614457831, "grad_norm": 10.901326179504395, "learning_rate": 1.6893548387096776e-05, "loss": 0.4893, "step": 10263 }, { "epoch": 66.16546184738957, "grad_norm": 13.777867317199707, "learning_rate": 1.689032258064516e-05, "loss": 0.1916, "step": 10264 }, { "epoch": 66.17188755020081, "grad_norm": 28.929351806640625, "learning_rate": 1.688709677419355e-05, "loss": 0.3016, "step": 10265 }, { "epoch": 66.17831325301205, "grad_norm": 15.545781135559082, "learning_rate": 1.6883870967741936e-05, "loss": 0.1758, "step": 10266 }, { "epoch": 66.18473895582329, "grad_norm": 29.004329681396484, "learning_rate": 1.6880645161290322e-05, "loss": 0.6704, "step": 10267 }, { "epoch": 66.19116465863453, "grad_norm": 11.71139144897461, "learning_rate": 1.687741935483871e-05, "loss": 0.438, "step": 10268 }, { "epoch": 66.19759036144578, "grad_norm": 6.30149507522583, "learning_rate": 1.68741935483871e-05, "loss": 0.2803, "step": 10269 }, { "epoch": 66.20401606425703, "grad_norm": 15.298186302185059, "learning_rate": 1.6870967741935482e-05, "loss": 0.3677, "step": 10270 }, { "epoch": 66.21044176706827, "grad_norm": 5.12421989440918, "learning_rate": 1.686774193548387e-05, "loss": 0.314, "step": 10271 }, { "epoch": 66.21686746987952, "grad_norm": 8.358097076416016, "learning_rate": 1.686451612903226e-05, "loss": 0.3375, "step": 10272 }, { "epoch": 66.22329317269076, "grad_norm": 9.74433708190918, "learning_rate": 1.6861290322580646e-05, "loss": 0.3031, "step": 10273 }, { "epoch": 66.229718875502, "grad_norm": 13.62829303741455, "learning_rate": 1.6858064516129032e-05, "loss": 0.536, "step": 10274 }, { "epoch": 66.23614457831326, "grad_norm": 19.7477970123291, "learning_rate": 1.685483870967742e-05, "loss": 0.4731, "step": 10275 }, { "epoch": 66.2425702811245, "grad_norm": 3.0300204753875732, "learning_rate": 1.6851612903225807e-05, "loss": 0.157, "step": 10276 }, { "epoch": 66.24899598393574, "grad_norm": 30.163169860839844, "learning_rate": 1.6848387096774192e-05, "loss": 0.6384, "step": 10277 }, { "epoch": 66.25542168674698, "grad_norm": 17.64474868774414, "learning_rate": 1.684516129032258e-05, "loss": 0.4033, "step": 10278 }, { "epoch": 66.26184738955823, "grad_norm": 25.424936294555664, "learning_rate": 1.6841935483870967e-05, "loss": 0.3938, "step": 10279 }, { "epoch": 66.26827309236948, "grad_norm": 4.011849403381348, "learning_rate": 1.6838709677419356e-05, "loss": 0.3289, "step": 10280 }, { "epoch": 66.27469879518073, "grad_norm": 2.5604684352874756, "learning_rate": 1.6835483870967742e-05, "loss": 0.2255, "step": 10281 }, { "epoch": 66.28112449799197, "grad_norm": 2.3429932594299316, "learning_rate": 1.6832258064516128e-05, "loss": 0.3046, "step": 10282 }, { "epoch": 66.28755020080321, "grad_norm": 4.471564292907715, "learning_rate": 1.6829032258064517e-05, "loss": 0.5754, "step": 10283 }, { "epoch": 66.29397590361445, "grad_norm": 6.678257465362549, "learning_rate": 1.6825806451612906e-05, "loss": 0.3793, "step": 10284 }, { "epoch": 66.30040160642571, "grad_norm": 5.4217963218688965, "learning_rate": 1.6822580645161288e-05, "loss": 0.3532, "step": 10285 }, { "epoch": 66.30682730923695, "grad_norm": 20.130252838134766, "learning_rate": 1.6819354838709677e-05, "loss": 0.2929, "step": 10286 }, { "epoch": 66.3132530120482, "grad_norm": 24.565345764160156, "learning_rate": 1.6816129032258066e-05, "loss": 0.6761, "step": 10287 }, { "epoch": 66.31967871485944, "grad_norm": 9.235897064208984, "learning_rate": 1.6812903225806452e-05, "loss": 0.4994, "step": 10288 }, { "epoch": 66.32610441767068, "grad_norm": 11.860610961914062, "learning_rate": 1.6809677419354838e-05, "loss": 0.4031, "step": 10289 }, { "epoch": 66.33253012048193, "grad_norm": 38.7297477722168, "learning_rate": 1.6806451612903227e-05, "loss": 0.2621, "step": 10290 }, { "epoch": 66.33895582329318, "grad_norm": 13.613311767578125, "learning_rate": 1.6803225806451613e-05, "loss": 0.5383, "step": 10291 }, { "epoch": 66.34538152610442, "grad_norm": 7.676721096038818, "learning_rate": 1.6800000000000002e-05, "loss": 0.2954, "step": 10292 }, { "epoch": 66.35180722891566, "grad_norm": 11.364206314086914, "learning_rate": 1.6796774193548387e-05, "loss": 0.2481, "step": 10293 }, { "epoch": 66.3582329317269, "grad_norm": 7.718087673187256, "learning_rate": 1.6793548387096773e-05, "loss": 0.2434, "step": 10294 }, { "epoch": 66.36465863453816, "grad_norm": 1.596604585647583, "learning_rate": 1.6790322580645162e-05, "loss": 0.3229, "step": 10295 }, { "epoch": 66.3710843373494, "grad_norm": 4.053183078765869, "learning_rate": 1.678709677419355e-05, "loss": 0.21, "step": 10296 }, { "epoch": 66.37751004016064, "grad_norm": 10.14251708984375, "learning_rate": 1.6783870967741934e-05, "loss": 0.356, "step": 10297 }, { "epoch": 66.38393574297189, "grad_norm": 27.010231018066406, "learning_rate": 1.6780645161290323e-05, "loss": 0.4127, "step": 10298 }, { "epoch": 66.39036144578313, "grad_norm": 2.7019851207733154, "learning_rate": 1.6777419354838712e-05, "loss": 0.2886, "step": 10299 }, { "epoch": 66.39678714859438, "grad_norm": 5.333657741546631, "learning_rate": 1.6774193548387098e-05, "loss": 0.1859, "step": 10300 }, { "epoch": 66.40321285140563, "grad_norm": 33.975345611572266, "learning_rate": 1.6770967741935483e-05, "loss": 0.2747, "step": 10301 }, { "epoch": 66.40963855421687, "grad_norm": 27.263540267944336, "learning_rate": 1.6767741935483872e-05, "loss": 0.5416, "step": 10302 }, { "epoch": 66.41606425702811, "grad_norm": 6.375085830688477, "learning_rate": 1.6764516129032258e-05, "loss": 0.5264, "step": 10303 }, { "epoch": 66.42248995983935, "grad_norm": 6.062243461608887, "learning_rate": 1.6761290322580647e-05, "loss": 0.3329, "step": 10304 }, { "epoch": 66.4289156626506, "grad_norm": 5.074785232543945, "learning_rate": 1.6758064516129033e-05, "loss": 0.3515, "step": 10305 }, { "epoch": 66.43534136546185, "grad_norm": 15.960755348205566, "learning_rate": 1.675483870967742e-05, "loss": 0.1237, "step": 10306 }, { "epoch": 66.4417670682731, "grad_norm": 12.802389144897461, "learning_rate": 1.6751612903225808e-05, "loss": 0.3832, "step": 10307 }, { "epoch": 66.44819277108434, "grad_norm": 20.98552894592285, "learning_rate": 1.6748387096774197e-05, "loss": 0.1899, "step": 10308 }, { "epoch": 66.45461847389558, "grad_norm": 9.727273941040039, "learning_rate": 1.674516129032258e-05, "loss": 0.2035, "step": 10309 }, { "epoch": 66.46104417670682, "grad_norm": 20.910497665405273, "learning_rate": 1.6741935483870968e-05, "loss": 0.418, "step": 10310 }, { "epoch": 66.46746987951808, "grad_norm": 5.042041301727295, "learning_rate": 1.6738709677419357e-05, "loss": 0.2867, "step": 10311 }, { "epoch": 66.47389558232932, "grad_norm": 22.5440731048584, "learning_rate": 1.6735483870967743e-05, "loss": 0.5107, "step": 10312 }, { "epoch": 66.48032128514056, "grad_norm": 16.212013244628906, "learning_rate": 1.673225806451613e-05, "loss": 0.1484, "step": 10313 }, { "epoch": 66.4867469879518, "grad_norm": 46.3642463684082, "learning_rate": 1.6729032258064518e-05, "loss": 0.4314, "step": 10314 }, { "epoch": 66.49317269076305, "grad_norm": 4.145239353179932, "learning_rate": 1.6725806451612904e-05, "loss": 0.1763, "step": 10315 }, { "epoch": 66.4995983935743, "grad_norm": 9.736322402954102, "learning_rate": 1.672258064516129e-05, "loss": 0.317, "step": 10316 }, { "epoch": 66.50602409638554, "grad_norm": 21.191492080688477, "learning_rate": 1.671935483870968e-05, "loss": 0.5234, "step": 10317 }, { "epoch": 66.51244979919679, "grad_norm": 3.255852222442627, "learning_rate": 1.6716129032258064e-05, "loss": 0.2586, "step": 10318 }, { "epoch": 66.51887550200803, "grad_norm": 32.85953903198242, "learning_rate": 1.6712903225806453e-05, "loss": 0.4981, "step": 10319 }, { "epoch": 66.52530120481927, "grad_norm": 6.015186309814453, "learning_rate": 1.670967741935484e-05, "loss": 0.6064, "step": 10320 }, { "epoch": 66.53172690763053, "grad_norm": 17.3913631439209, "learning_rate": 1.6706451612903225e-05, "loss": 0.4737, "step": 10321 }, { "epoch": 66.53815261044177, "grad_norm": 12.264531135559082, "learning_rate": 1.6703225806451614e-05, "loss": 0.2781, "step": 10322 }, { "epoch": 66.54457831325301, "grad_norm": 34.31031799316406, "learning_rate": 1.6700000000000003e-05, "loss": 0.184, "step": 10323 }, { "epoch": 66.55100401606425, "grad_norm": 14.997517585754395, "learning_rate": 1.669677419354839e-05, "loss": 0.3137, "step": 10324 }, { "epoch": 66.5574297188755, "grad_norm": 6.347816467285156, "learning_rate": 1.6693548387096774e-05, "loss": 0.2481, "step": 10325 }, { "epoch": 66.56385542168675, "grad_norm": 7.798308849334717, "learning_rate": 1.6690322580645163e-05, "loss": 0.2423, "step": 10326 }, { "epoch": 66.570281124498, "grad_norm": 24.44102668762207, "learning_rate": 1.668709677419355e-05, "loss": 0.3897, "step": 10327 }, { "epoch": 66.57670682730924, "grad_norm": 4.789078235626221, "learning_rate": 1.6683870967741935e-05, "loss": 0.23, "step": 10328 }, { "epoch": 66.58313253012048, "grad_norm": 23.030183792114258, "learning_rate": 1.6680645161290324e-05, "loss": 0.3562, "step": 10329 }, { "epoch": 66.58955823293172, "grad_norm": 19.258499145507812, "learning_rate": 1.667741935483871e-05, "loss": 0.2827, "step": 10330 }, { "epoch": 66.59598393574298, "grad_norm": 5.8562517166137695, "learning_rate": 1.66741935483871e-05, "loss": 0.2771, "step": 10331 }, { "epoch": 66.60240963855422, "grad_norm": 14.985047340393066, "learning_rate": 1.6670967741935484e-05, "loss": 0.7143, "step": 10332 }, { "epoch": 66.60883534136546, "grad_norm": 7.591320037841797, "learning_rate": 1.666774193548387e-05, "loss": 0.2717, "step": 10333 }, { "epoch": 66.6152610441767, "grad_norm": 22.030858993530273, "learning_rate": 1.666451612903226e-05, "loss": 0.2882, "step": 10334 }, { "epoch": 66.62168674698795, "grad_norm": 11.313961029052734, "learning_rate": 1.6661290322580648e-05, "loss": 0.3227, "step": 10335 }, { "epoch": 66.6281124497992, "grad_norm": 3.4275567531585693, "learning_rate": 1.6658064516129034e-05, "loss": 0.4217, "step": 10336 }, { "epoch": 66.63453815261045, "grad_norm": 20.189987182617188, "learning_rate": 1.665483870967742e-05, "loss": 0.1894, "step": 10337 }, { "epoch": 66.64096385542169, "grad_norm": 5.579227924346924, "learning_rate": 1.665161290322581e-05, "loss": 0.1833, "step": 10338 }, { "epoch": 66.64738955823293, "grad_norm": 8.409218788146973, "learning_rate": 1.6648387096774194e-05, "loss": 0.227, "step": 10339 }, { "epoch": 66.65381526104417, "grad_norm": 19.727367401123047, "learning_rate": 1.664516129032258e-05, "loss": 0.2753, "step": 10340 }, { "epoch": 66.66024096385541, "grad_norm": 13.385237693786621, "learning_rate": 1.664193548387097e-05, "loss": 0.3329, "step": 10341 }, { "epoch": 66.66666666666667, "grad_norm": 14.256393432617188, "learning_rate": 1.6638709677419355e-05, "loss": 0.3481, "step": 10342 }, { "epoch": 66.67309236947791, "grad_norm": 12.374268531799316, "learning_rate": 1.6635483870967744e-05, "loss": 0.2933, "step": 10343 }, { "epoch": 66.67951807228916, "grad_norm": 20.896474838256836, "learning_rate": 1.663225806451613e-05, "loss": 0.3985, "step": 10344 }, { "epoch": 66.6859437751004, "grad_norm": 15.683452606201172, "learning_rate": 1.6629032258064515e-05, "loss": 0.3439, "step": 10345 }, { "epoch": 66.69236947791164, "grad_norm": 18.202085494995117, "learning_rate": 1.6625806451612904e-05, "loss": 0.5784, "step": 10346 }, { "epoch": 66.6987951807229, "grad_norm": 8.837230682373047, "learning_rate": 1.662258064516129e-05, "loss": 0.2409, "step": 10347 }, { "epoch": 66.70522088353414, "grad_norm": 17.348541259765625, "learning_rate": 1.6619354838709676e-05, "loss": 0.5247, "step": 10348 }, { "epoch": 66.71164658634538, "grad_norm": 4.214949131011963, "learning_rate": 1.6616129032258065e-05, "loss": 0.2293, "step": 10349 }, { "epoch": 66.71807228915662, "grad_norm": 43.998416900634766, "learning_rate": 1.6612903225806454e-05, "loss": 0.1186, "step": 10350 }, { "epoch": 66.72449799196787, "grad_norm": 15.885435104370117, "learning_rate": 1.660967741935484e-05, "loss": 0.3608, "step": 10351 }, { "epoch": 66.73092369477912, "grad_norm": 2.1057190895080566, "learning_rate": 1.6606451612903226e-05, "loss": 0.4422, "step": 10352 }, { "epoch": 66.73734939759036, "grad_norm": 11.877395629882812, "learning_rate": 1.6603225806451615e-05, "loss": 0.3136, "step": 10353 }, { "epoch": 66.7437751004016, "grad_norm": 12.885238647460938, "learning_rate": 1.66e-05, "loss": 0.4282, "step": 10354 }, { "epoch": 66.75020080321285, "grad_norm": 5.046059608459473, "learning_rate": 1.6596774193548386e-05, "loss": 0.4009, "step": 10355 }, { "epoch": 66.75662650602409, "grad_norm": 11.897374153137207, "learning_rate": 1.6593548387096775e-05, "loss": 0.2707, "step": 10356 }, { "epoch": 66.76305220883535, "grad_norm": 19.282445907592773, "learning_rate": 1.659032258064516e-05, "loss": 0.3166, "step": 10357 }, { "epoch": 66.76947791164659, "grad_norm": 13.348416328430176, "learning_rate": 1.658709677419355e-05, "loss": 0.2626, "step": 10358 }, { "epoch": 66.77590361445783, "grad_norm": 9.939349174499512, "learning_rate": 1.6583870967741936e-05, "loss": 0.3645, "step": 10359 }, { "epoch": 66.78232931726907, "grad_norm": 13.703206062316895, "learning_rate": 1.658064516129032e-05, "loss": 0.4108, "step": 10360 }, { "epoch": 66.78875502008032, "grad_norm": 1.1732847690582275, "learning_rate": 1.657741935483871e-05, "loss": 0.2493, "step": 10361 }, { "epoch": 66.79518072289157, "grad_norm": 10.375535011291504, "learning_rate": 1.65741935483871e-05, "loss": 0.2577, "step": 10362 }, { "epoch": 66.80160642570281, "grad_norm": 5.823345184326172, "learning_rate": 1.6570967741935485e-05, "loss": 0.5141, "step": 10363 }, { "epoch": 66.80803212851406, "grad_norm": 3.597538709640503, "learning_rate": 1.656774193548387e-05, "loss": 0.2042, "step": 10364 }, { "epoch": 66.8144578313253, "grad_norm": 8.260321617126465, "learning_rate": 1.656451612903226e-05, "loss": 0.2465, "step": 10365 }, { "epoch": 66.82088353413654, "grad_norm": 18.515342712402344, "learning_rate": 1.6561290322580646e-05, "loss": 0.4163, "step": 10366 }, { "epoch": 66.8273092369478, "grad_norm": 4.487364768981934, "learning_rate": 1.655806451612903e-05, "loss": 0.4265, "step": 10367 }, { "epoch": 66.83373493975904, "grad_norm": 31.469131469726562, "learning_rate": 1.655483870967742e-05, "loss": 0.4099, "step": 10368 }, { "epoch": 66.84016064257028, "grad_norm": 39.172393798828125, "learning_rate": 1.6551612903225806e-05, "loss": 0.2637, "step": 10369 }, { "epoch": 66.84658634538152, "grad_norm": 22.23819923400879, "learning_rate": 1.6548387096774195e-05, "loss": 0.188, "step": 10370 }, { "epoch": 66.85301204819277, "grad_norm": 5.611649513244629, "learning_rate": 1.654516129032258e-05, "loss": 0.2989, "step": 10371 }, { "epoch": 66.85943775100401, "grad_norm": 6.018378257751465, "learning_rate": 1.6541935483870967e-05, "loss": 0.3938, "step": 10372 }, { "epoch": 66.86586345381527, "grad_norm": 13.470643043518066, "learning_rate": 1.6538709677419356e-05, "loss": 0.2026, "step": 10373 }, { "epoch": 66.87228915662651, "grad_norm": 41.15129089355469, "learning_rate": 1.6535483870967745e-05, "loss": 0.453, "step": 10374 }, { "epoch": 66.87871485943775, "grad_norm": 3.349915027618408, "learning_rate": 1.653225806451613e-05, "loss": 0.2316, "step": 10375 }, { "epoch": 66.88514056224899, "grad_norm": 16.58843421936035, "learning_rate": 1.6529032258064516e-05, "loss": 0.2798, "step": 10376 }, { "epoch": 66.89156626506023, "grad_norm": 16.788288116455078, "learning_rate": 1.6525806451612905e-05, "loss": 0.0914, "step": 10377 }, { "epoch": 66.89799196787149, "grad_norm": 3.607210874557495, "learning_rate": 1.652258064516129e-05, "loss": 0.3581, "step": 10378 }, { "epoch": 66.90441767068273, "grad_norm": 56.15657424926758, "learning_rate": 1.6519354838709677e-05, "loss": 0.2193, "step": 10379 }, { "epoch": 66.91084337349398, "grad_norm": 13.851141929626465, "learning_rate": 1.6516129032258066e-05, "loss": 0.6256, "step": 10380 }, { "epoch": 66.91726907630522, "grad_norm": 7.20695161819458, "learning_rate": 1.651290322580645e-05, "loss": 0.4557, "step": 10381 }, { "epoch": 66.92369477911646, "grad_norm": 20.54633903503418, "learning_rate": 1.6509677419354837e-05, "loss": 0.3149, "step": 10382 }, { "epoch": 66.93012048192772, "grad_norm": 39.321556091308594, "learning_rate": 1.6506451612903226e-05, "loss": 0.7747, "step": 10383 }, { "epoch": 66.93654618473896, "grad_norm": 2.7747480869293213, "learning_rate": 1.6503225806451612e-05, "loss": 0.2191, "step": 10384 }, { "epoch": 66.9429718875502, "grad_norm": 7.181278228759766, "learning_rate": 1.65e-05, "loss": 0.4666, "step": 10385 }, { "epoch": 66.94939759036144, "grad_norm": 7.521988868713379, "learning_rate": 1.6496774193548387e-05, "loss": 0.2095, "step": 10386 }, { "epoch": 66.95582329317268, "grad_norm": 8.504829406738281, "learning_rate": 1.6493548387096776e-05, "loss": 0.2671, "step": 10387 }, { "epoch": 66.96224899598394, "grad_norm": 4.976464748382568, "learning_rate": 1.6490322580645162e-05, "loss": 0.1639, "step": 10388 }, { "epoch": 66.96867469879518, "grad_norm": 32.716007232666016, "learning_rate": 1.648709677419355e-05, "loss": 0.4262, "step": 10389 }, { "epoch": 66.97510040160643, "grad_norm": 3.234747886657715, "learning_rate": 1.6483870967741937e-05, "loss": 0.1594, "step": 10390 }, { "epoch": 66.98152610441767, "grad_norm": 57.15389633178711, "learning_rate": 1.6480645161290322e-05, "loss": 0.6043, "step": 10391 }, { "epoch": 66.98795180722891, "grad_norm": 9.779963493347168, "learning_rate": 1.647741935483871e-05, "loss": 0.424, "step": 10392 }, { "epoch": 66.99437751004017, "grad_norm": 22.146482467651367, "learning_rate": 1.6474193548387097e-05, "loss": 0.3826, "step": 10393 }, { "epoch": 67.00401606425703, "grad_norm": 5.07423210144043, "learning_rate": 1.6470967741935483e-05, "loss": 0.3918, "step": 10394 }, { "epoch": 67.01044176706827, "grad_norm": 3.7540626525878906, "learning_rate": 1.6467741935483872e-05, "loss": 0.3196, "step": 10395 }, { "epoch": 67.01686746987951, "grad_norm": 19.08336639404297, "learning_rate": 1.6464516129032258e-05, "loss": 0.4015, "step": 10396 }, { "epoch": 67.02329317269076, "grad_norm": 11.694981575012207, "learning_rate": 1.6461290322580647e-05, "loss": 0.406, "step": 10397 }, { "epoch": 67.02971887550201, "grad_norm": 4.070638179779053, "learning_rate": 1.6458064516129032e-05, "loss": 0.1846, "step": 10398 }, { "epoch": 67.03614457831326, "grad_norm": 18.79389190673828, "learning_rate": 1.645483870967742e-05, "loss": 0.4768, "step": 10399 }, { "epoch": 67.0425702811245, "grad_norm": 26.638351440429688, "learning_rate": 1.6451612903225807e-05, "loss": 0.7115, "step": 10400 }, { "epoch": 67.04899598393574, "grad_norm": 10.422534942626953, "learning_rate": 1.6448387096774196e-05, "loss": 0.2009, "step": 10401 }, { "epoch": 67.05542168674698, "grad_norm": 12.82841968536377, "learning_rate": 1.6445161290322582e-05, "loss": 0.3202, "step": 10402 }, { "epoch": 67.06184738955824, "grad_norm": 10.437336921691895, "learning_rate": 1.6441935483870968e-05, "loss": 0.4236, "step": 10403 }, { "epoch": 67.06827309236948, "grad_norm": 13.856006622314453, "learning_rate": 1.6438709677419357e-05, "loss": 0.1474, "step": 10404 }, { "epoch": 67.07469879518072, "grad_norm": 3.131486177444458, "learning_rate": 1.6435483870967743e-05, "loss": 0.1686, "step": 10405 }, { "epoch": 67.08112449799196, "grad_norm": 7.973912715911865, "learning_rate": 1.6432258064516128e-05, "loss": 0.3121, "step": 10406 }, { "epoch": 67.08755020080321, "grad_norm": 28.929975509643555, "learning_rate": 1.6429032258064517e-05, "loss": 0.397, "step": 10407 }, { "epoch": 67.09397590361446, "grad_norm": 13.439531326293945, "learning_rate": 1.6425806451612903e-05, "loss": 0.5968, "step": 10408 }, { "epoch": 67.1004016064257, "grad_norm": 8.71683406829834, "learning_rate": 1.6422580645161292e-05, "loss": 0.1438, "step": 10409 }, { "epoch": 67.10682730923695, "grad_norm": 3.2346811294555664, "learning_rate": 1.6419354838709678e-05, "loss": 0.2583, "step": 10410 }, { "epoch": 67.11325301204819, "grad_norm": 19.572542190551758, "learning_rate": 1.6416129032258064e-05, "loss": 0.2777, "step": 10411 }, { "epoch": 67.11967871485943, "grad_norm": 11.798100471496582, "learning_rate": 1.6412903225806453e-05, "loss": 0.2453, "step": 10412 }, { "epoch": 67.12610441767069, "grad_norm": 18.212800979614258, "learning_rate": 1.6409677419354842e-05, "loss": 0.2352, "step": 10413 }, { "epoch": 67.13253012048193, "grad_norm": 7.626398086547852, "learning_rate": 1.6406451612903227e-05, "loss": 0.5574, "step": 10414 }, { "epoch": 67.13895582329317, "grad_norm": 20.09790802001953, "learning_rate": 1.6403225806451613e-05, "loss": 0.3799, "step": 10415 }, { "epoch": 67.14538152610442, "grad_norm": 12.650323867797852, "learning_rate": 1.6400000000000002e-05, "loss": 0.3099, "step": 10416 }, { "epoch": 67.15180722891566, "grad_norm": 18.751859664916992, "learning_rate": 1.6396774193548388e-05, "loss": 0.4028, "step": 10417 }, { "epoch": 67.15823293172691, "grad_norm": 6.3935980796813965, "learning_rate": 1.6393548387096774e-05, "loss": 0.2405, "step": 10418 }, { "epoch": 67.16465863453816, "grad_norm": 3.943296432495117, "learning_rate": 1.6390322580645163e-05, "loss": 0.3861, "step": 10419 }, { "epoch": 67.1710843373494, "grad_norm": 18.201675415039062, "learning_rate": 1.638709677419355e-05, "loss": 0.2426, "step": 10420 }, { "epoch": 67.17751004016064, "grad_norm": 7.11331033706665, "learning_rate": 1.6383870967741934e-05, "loss": 0.154, "step": 10421 }, { "epoch": 67.18393574297188, "grad_norm": 19.281673431396484, "learning_rate": 1.6380645161290323e-05, "loss": 0.6394, "step": 10422 }, { "epoch": 67.19036144578314, "grad_norm": 2.6747233867645264, "learning_rate": 1.637741935483871e-05, "loss": 0.1023, "step": 10423 }, { "epoch": 67.19678714859438, "grad_norm": 19.04358673095703, "learning_rate": 1.6374193548387098e-05, "loss": 0.4928, "step": 10424 }, { "epoch": 67.20321285140562, "grad_norm": 5.273942470550537, "learning_rate": 1.6370967741935484e-05, "loss": 0.229, "step": 10425 }, { "epoch": 67.20963855421687, "grad_norm": 6.498650550842285, "learning_rate": 1.6367741935483873e-05, "loss": 0.3292, "step": 10426 }, { "epoch": 67.21606425702811, "grad_norm": 17.82896614074707, "learning_rate": 1.636451612903226e-05, "loss": 0.2041, "step": 10427 }, { "epoch": 67.22248995983935, "grad_norm": 17.021713256835938, "learning_rate": 1.6361290322580648e-05, "loss": 0.4212, "step": 10428 }, { "epoch": 67.2289156626506, "grad_norm": 5.214338302612305, "learning_rate": 1.6358064516129033e-05, "loss": 0.2028, "step": 10429 }, { "epoch": 67.23534136546185, "grad_norm": 28.415287017822266, "learning_rate": 1.635483870967742e-05, "loss": 0.4669, "step": 10430 }, { "epoch": 67.24176706827309, "grad_norm": 14.64244270324707, "learning_rate": 1.6351612903225808e-05, "loss": 0.1109, "step": 10431 }, { "epoch": 67.24819277108433, "grad_norm": 10.134042739868164, "learning_rate": 1.6348387096774194e-05, "loss": 0.2302, "step": 10432 }, { "epoch": 67.25461847389558, "grad_norm": 9.647066116333008, "learning_rate": 1.634516129032258e-05, "loss": 0.2806, "step": 10433 }, { "epoch": 67.26104417670683, "grad_norm": 9.949697494506836, "learning_rate": 1.634193548387097e-05, "loss": 0.2784, "step": 10434 }, { "epoch": 67.26746987951807, "grad_norm": 16.092082977294922, "learning_rate": 1.6338709677419354e-05, "loss": 0.2922, "step": 10435 }, { "epoch": 67.27389558232932, "grad_norm": 13.816109657287598, "learning_rate": 1.6335483870967743e-05, "loss": 0.3632, "step": 10436 }, { "epoch": 67.28032128514056, "grad_norm": 15.85423755645752, "learning_rate": 1.633225806451613e-05, "loss": 0.3094, "step": 10437 }, { "epoch": 67.2867469879518, "grad_norm": 8.836875915527344, "learning_rate": 1.6329032258064518e-05, "loss": 0.3069, "step": 10438 }, { "epoch": 67.29317269076306, "grad_norm": 55.80133819580078, "learning_rate": 1.6325806451612904e-05, "loss": 0.4427, "step": 10439 }, { "epoch": 67.2995983935743, "grad_norm": 11.255841255187988, "learning_rate": 1.6322580645161293e-05, "loss": 0.3185, "step": 10440 }, { "epoch": 67.30602409638554, "grad_norm": 5.779454231262207, "learning_rate": 1.631935483870968e-05, "loss": 0.2504, "step": 10441 }, { "epoch": 67.31244979919678, "grad_norm": 2.8867239952087402, "learning_rate": 1.6316129032258064e-05, "loss": 0.2667, "step": 10442 }, { "epoch": 67.31887550200803, "grad_norm": 5.329693794250488, "learning_rate": 1.6312903225806454e-05, "loss": 0.1618, "step": 10443 }, { "epoch": 67.32530120481928, "grad_norm": 21.797224044799805, "learning_rate": 1.630967741935484e-05, "loss": 0.3841, "step": 10444 }, { "epoch": 67.33172690763053, "grad_norm": 53.35844039916992, "learning_rate": 1.6306451612903225e-05, "loss": 0.504, "step": 10445 }, { "epoch": 67.33815261044177, "grad_norm": 44.24126434326172, "learning_rate": 1.6303225806451614e-05, "loss": 0.536, "step": 10446 }, { "epoch": 67.34457831325301, "grad_norm": 25.510278701782227, "learning_rate": 1.63e-05, "loss": 0.3581, "step": 10447 }, { "epoch": 67.35100401606425, "grad_norm": 83.83444213867188, "learning_rate": 1.629677419354839e-05, "loss": 0.2339, "step": 10448 }, { "epoch": 67.35742971887551, "grad_norm": 9.0806303024292, "learning_rate": 1.6293548387096775e-05, "loss": 0.5907, "step": 10449 }, { "epoch": 67.36385542168675, "grad_norm": 8.064671516418457, "learning_rate": 1.6290322580645164e-05, "loss": 0.5137, "step": 10450 }, { "epoch": 67.37028112449799, "grad_norm": 20.661821365356445, "learning_rate": 1.628709677419355e-05, "loss": 0.588, "step": 10451 }, { "epoch": 67.37670682730923, "grad_norm": 35.204742431640625, "learning_rate": 1.6283870967741935e-05, "loss": 0.2724, "step": 10452 }, { "epoch": 67.38313253012048, "grad_norm": 6.895562648773193, "learning_rate": 1.6280645161290324e-05, "loss": 0.4577, "step": 10453 }, { "epoch": 67.38955823293173, "grad_norm": 13.976546287536621, "learning_rate": 1.627741935483871e-05, "loss": 0.6187, "step": 10454 }, { "epoch": 67.39598393574298, "grad_norm": 16.65161895751953, "learning_rate": 1.62741935483871e-05, "loss": 0.4756, "step": 10455 }, { "epoch": 67.40240963855422, "grad_norm": 19.763765335083008, "learning_rate": 1.6270967741935485e-05, "loss": 0.365, "step": 10456 }, { "epoch": 67.40883534136546, "grad_norm": 11.763806343078613, "learning_rate": 1.626774193548387e-05, "loss": 0.2494, "step": 10457 }, { "epoch": 67.4152610441767, "grad_norm": 15.436226844787598, "learning_rate": 1.626451612903226e-05, "loss": 0.3674, "step": 10458 }, { "epoch": 67.42168674698796, "grad_norm": 29.9111270904541, "learning_rate": 1.6261290322580645e-05, "loss": 0.34, "step": 10459 }, { "epoch": 67.4281124497992, "grad_norm": 22.401865005493164, "learning_rate": 1.625806451612903e-05, "loss": 0.4346, "step": 10460 }, { "epoch": 67.43453815261044, "grad_norm": 5.080616474151611, "learning_rate": 1.625483870967742e-05, "loss": 0.2737, "step": 10461 }, { "epoch": 67.44096385542169, "grad_norm": 2.792421817779541, "learning_rate": 1.625161290322581e-05, "loss": 0.2888, "step": 10462 }, { "epoch": 67.44738955823293, "grad_norm": 17.218284606933594, "learning_rate": 1.6248387096774195e-05, "loss": 0.4155, "step": 10463 }, { "epoch": 67.45381526104417, "grad_norm": 14.737699508666992, "learning_rate": 1.624516129032258e-05, "loss": 0.409, "step": 10464 }, { "epoch": 67.46024096385543, "grad_norm": 9.341961860656738, "learning_rate": 1.624193548387097e-05, "loss": 0.3149, "step": 10465 }, { "epoch": 67.46666666666667, "grad_norm": 16.167211532592773, "learning_rate": 1.6238709677419355e-05, "loss": 0.3935, "step": 10466 }, { "epoch": 67.47309236947791, "grad_norm": 7.787516117095947, "learning_rate": 1.6235483870967744e-05, "loss": 0.4964, "step": 10467 }, { "epoch": 67.47951807228915, "grad_norm": 10.511207580566406, "learning_rate": 1.623225806451613e-05, "loss": 0.3051, "step": 10468 }, { "epoch": 67.4859437751004, "grad_norm": 6.3461785316467285, "learning_rate": 1.6229032258064516e-05, "loss": 0.3519, "step": 10469 }, { "epoch": 67.49236947791165, "grad_norm": 7.269052982330322, "learning_rate": 1.6225806451612905e-05, "loss": 0.324, "step": 10470 }, { "epoch": 67.4987951807229, "grad_norm": 8.222823143005371, "learning_rate": 1.622258064516129e-05, "loss": 0.336, "step": 10471 }, { "epoch": 67.50522088353414, "grad_norm": 11.708711624145508, "learning_rate": 1.6219354838709676e-05, "loss": 0.092, "step": 10472 }, { "epoch": 67.51164658634538, "grad_norm": 22.449600219726562, "learning_rate": 1.6216129032258065e-05, "loss": 0.5407, "step": 10473 }, { "epoch": 67.51807228915662, "grad_norm": 9.591660499572754, "learning_rate": 1.621290322580645e-05, "loss": 0.7395, "step": 10474 }, { "epoch": 67.52449799196788, "grad_norm": 5.6640448570251465, "learning_rate": 1.620967741935484e-05, "loss": 0.1425, "step": 10475 }, { "epoch": 67.53092369477912, "grad_norm": 46.52628707885742, "learning_rate": 1.6206451612903226e-05, "loss": 0.3655, "step": 10476 }, { "epoch": 67.53734939759036, "grad_norm": 23.155895233154297, "learning_rate": 1.6203225806451615e-05, "loss": 0.5128, "step": 10477 }, { "epoch": 67.5437751004016, "grad_norm": 18.353649139404297, "learning_rate": 1.62e-05, "loss": 0.3169, "step": 10478 }, { "epoch": 67.55020080321285, "grad_norm": 10.067021369934082, "learning_rate": 1.619677419354839e-05, "loss": 0.3315, "step": 10479 }, { "epoch": 67.5566265060241, "grad_norm": 7.366595268249512, "learning_rate": 1.6193548387096776e-05, "loss": 0.4038, "step": 10480 }, { "epoch": 67.56305220883534, "grad_norm": 13.127676010131836, "learning_rate": 1.619032258064516e-05, "loss": 0.4239, "step": 10481 }, { "epoch": 67.56947791164659, "grad_norm": 16.672344207763672, "learning_rate": 1.618709677419355e-05, "loss": 0.3963, "step": 10482 }, { "epoch": 67.57590361445783, "grad_norm": 5.212394714355469, "learning_rate": 1.6183870967741936e-05, "loss": 0.3844, "step": 10483 }, { "epoch": 67.58232931726907, "grad_norm": 12.69940185546875, "learning_rate": 1.6180645161290322e-05, "loss": 0.2157, "step": 10484 }, { "epoch": 67.58875502008033, "grad_norm": 10.817279815673828, "learning_rate": 1.617741935483871e-05, "loss": 0.41, "step": 10485 }, { "epoch": 67.59518072289157, "grad_norm": 28.64080238342285, "learning_rate": 1.6174193548387097e-05, "loss": 0.2755, "step": 10486 }, { "epoch": 67.60160642570281, "grad_norm": 24.166425704956055, "learning_rate": 1.6170967741935486e-05, "loss": 0.4513, "step": 10487 }, { "epoch": 67.60803212851405, "grad_norm": 3.3340110778808594, "learning_rate": 1.616774193548387e-05, "loss": 0.1511, "step": 10488 }, { "epoch": 67.6144578313253, "grad_norm": 5.743995666503906, "learning_rate": 1.616451612903226e-05, "loss": 0.5167, "step": 10489 }, { "epoch": 67.62088353413655, "grad_norm": 6.488898754119873, "learning_rate": 1.6161290322580646e-05, "loss": 0.1925, "step": 10490 }, { "epoch": 67.6273092369478, "grad_norm": 20.732561111450195, "learning_rate": 1.6158064516129032e-05, "loss": 0.4248, "step": 10491 }, { "epoch": 67.63373493975904, "grad_norm": 36.26223373413086, "learning_rate": 1.615483870967742e-05, "loss": 0.6021, "step": 10492 }, { "epoch": 67.64016064257028, "grad_norm": 3.576169729232788, "learning_rate": 1.6151612903225807e-05, "loss": 0.2186, "step": 10493 }, { "epoch": 67.64658634538152, "grad_norm": 10.065464973449707, "learning_rate": 1.6148387096774196e-05, "loss": 0.2496, "step": 10494 }, { "epoch": 67.65301204819278, "grad_norm": 13.22855281829834, "learning_rate": 1.614516129032258e-05, "loss": 0.4369, "step": 10495 }, { "epoch": 67.65943775100402, "grad_norm": 11.273487091064453, "learning_rate": 1.6141935483870967e-05, "loss": 0.4865, "step": 10496 }, { "epoch": 67.66586345381526, "grad_norm": 21.363515853881836, "learning_rate": 1.6138709677419356e-05, "loss": 0.3471, "step": 10497 }, { "epoch": 67.6722891566265, "grad_norm": 25.826826095581055, "learning_rate": 1.6135483870967742e-05, "loss": 0.4336, "step": 10498 }, { "epoch": 67.67871485943775, "grad_norm": 17.529109954833984, "learning_rate": 1.6132258064516128e-05, "loss": 0.5214, "step": 10499 }, { "epoch": 67.68514056224899, "grad_norm": 7.427643775939941, "learning_rate": 1.6129032258064517e-05, "loss": 0.1752, "step": 10500 }, { "epoch": 67.69156626506025, "grad_norm": 9.915264129638672, "learning_rate": 1.6125806451612906e-05, "loss": 0.2224, "step": 10501 }, { "epoch": 67.69799196787149, "grad_norm": 12.097508430480957, "learning_rate": 1.612258064516129e-05, "loss": 0.5537, "step": 10502 }, { "epoch": 67.70441767068273, "grad_norm": 5.678272247314453, "learning_rate": 1.6119354838709677e-05, "loss": 0.2502, "step": 10503 }, { "epoch": 67.71084337349397, "grad_norm": 19.948827743530273, "learning_rate": 1.6116129032258066e-05, "loss": 0.4973, "step": 10504 }, { "epoch": 67.71726907630521, "grad_norm": 28.72099494934082, "learning_rate": 1.6112903225806452e-05, "loss": 0.2866, "step": 10505 }, { "epoch": 67.72369477911647, "grad_norm": 8.29601764678955, "learning_rate": 1.610967741935484e-05, "loss": 0.5377, "step": 10506 }, { "epoch": 67.73012048192771, "grad_norm": 4.599223613739014, "learning_rate": 1.6106451612903227e-05, "loss": 0.2886, "step": 10507 }, { "epoch": 67.73654618473896, "grad_norm": 6.961331367492676, "learning_rate": 1.6103225806451613e-05, "loss": 0.6634, "step": 10508 }, { "epoch": 67.7429718875502, "grad_norm": 2.176243543624878, "learning_rate": 1.6100000000000002e-05, "loss": 0.0497, "step": 10509 }, { "epoch": 67.74939759036144, "grad_norm": 77.77631378173828, "learning_rate": 1.6096774193548387e-05, "loss": 0.6709, "step": 10510 }, { "epoch": 67.7558232931727, "grad_norm": 21.987144470214844, "learning_rate": 1.6093548387096773e-05, "loss": 0.2789, "step": 10511 }, { "epoch": 67.76224899598394, "grad_norm": 3.4024133682250977, "learning_rate": 1.6090322580645162e-05, "loss": 0.118, "step": 10512 }, { "epoch": 67.76867469879518, "grad_norm": 13.011076927185059, "learning_rate": 1.608709677419355e-05, "loss": 0.6067, "step": 10513 }, { "epoch": 67.77510040160642, "grad_norm": 15.455446243286133, "learning_rate": 1.6083870967741937e-05, "loss": 0.2354, "step": 10514 }, { "epoch": 67.78152610441767, "grad_norm": 8.4077787399292, "learning_rate": 1.6080645161290323e-05, "loss": 0.3605, "step": 10515 }, { "epoch": 67.78795180722892, "grad_norm": 14.789347648620605, "learning_rate": 1.6077419354838712e-05, "loss": 0.0973, "step": 10516 }, { "epoch": 67.79437751004016, "grad_norm": 8.536847114562988, "learning_rate": 1.6074193548387098e-05, "loss": 0.5246, "step": 10517 }, { "epoch": 67.8008032128514, "grad_norm": 12.28898811340332, "learning_rate": 1.6070967741935487e-05, "loss": 0.3394, "step": 10518 }, { "epoch": 67.80722891566265, "grad_norm": 22.38331413269043, "learning_rate": 1.6067741935483872e-05, "loss": 0.4047, "step": 10519 }, { "epoch": 67.81365461847389, "grad_norm": 9.023180961608887, "learning_rate": 1.6064516129032258e-05, "loss": 0.4978, "step": 10520 }, { "epoch": 67.82008032128515, "grad_norm": 17.153871536254883, "learning_rate": 1.6061290322580647e-05, "loss": 0.4987, "step": 10521 }, { "epoch": 67.82650602409639, "grad_norm": 8.869195938110352, "learning_rate": 1.6058064516129033e-05, "loss": 0.2829, "step": 10522 }, { "epoch": 67.83293172690763, "grad_norm": 8.634016036987305, "learning_rate": 1.605483870967742e-05, "loss": 0.4432, "step": 10523 }, { "epoch": 67.83935742971887, "grad_norm": 10.4974946975708, "learning_rate": 1.6051612903225808e-05, "loss": 0.5919, "step": 10524 }, { "epoch": 67.84578313253012, "grad_norm": 4.435631275177002, "learning_rate": 1.6048387096774197e-05, "loss": 0.266, "step": 10525 }, { "epoch": 67.85220883534137, "grad_norm": 7.522472858428955, "learning_rate": 1.604516129032258e-05, "loss": 0.3335, "step": 10526 }, { "epoch": 67.85863453815261, "grad_norm": 8.68961238861084, "learning_rate": 1.6041935483870968e-05, "loss": 0.4232, "step": 10527 }, { "epoch": 67.86506024096386, "grad_norm": 7.769514083862305, "learning_rate": 1.6038709677419357e-05, "loss": 0.5641, "step": 10528 }, { "epoch": 67.8714859437751, "grad_norm": 18.348865509033203, "learning_rate": 1.6035483870967743e-05, "loss": 0.4239, "step": 10529 }, { "epoch": 67.87791164658634, "grad_norm": 9.189146995544434, "learning_rate": 1.603225806451613e-05, "loss": 0.2811, "step": 10530 }, { "epoch": 67.88433734939758, "grad_norm": 4.932330131530762, "learning_rate": 1.6029032258064518e-05, "loss": 0.0629, "step": 10531 }, { "epoch": 67.89076305220884, "grad_norm": 15.745492935180664, "learning_rate": 1.6025806451612903e-05, "loss": 0.3722, "step": 10532 }, { "epoch": 67.89718875502008, "grad_norm": 17.052345275878906, "learning_rate": 1.6022580645161293e-05, "loss": 0.1314, "step": 10533 }, { "epoch": 67.90361445783132, "grad_norm": 11.753120422363281, "learning_rate": 1.6019354838709678e-05, "loss": 0.4135, "step": 10534 }, { "epoch": 67.91004016064257, "grad_norm": 13.31556224822998, "learning_rate": 1.6016129032258064e-05, "loss": 0.3596, "step": 10535 }, { "epoch": 67.91646586345381, "grad_norm": 28.429603576660156, "learning_rate": 1.6012903225806453e-05, "loss": 0.4864, "step": 10536 }, { "epoch": 67.92289156626506, "grad_norm": 15.455270767211914, "learning_rate": 1.600967741935484e-05, "loss": 0.4976, "step": 10537 }, { "epoch": 67.92931726907631, "grad_norm": 31.077653884887695, "learning_rate": 1.6006451612903224e-05, "loss": 0.6214, "step": 10538 }, { "epoch": 67.93574297188755, "grad_norm": 21.357421875, "learning_rate": 1.6003225806451614e-05, "loss": 0.385, "step": 10539 }, { "epoch": 67.94216867469879, "grad_norm": 10.81625747680664, "learning_rate": 1.6000000000000003e-05, "loss": 0.2929, "step": 10540 }, { "epoch": 67.94859437751003, "grad_norm": 17.633106231689453, "learning_rate": 1.599677419354839e-05, "loss": 0.2845, "step": 10541 }, { "epoch": 67.95502008032129, "grad_norm": 11.988134384155273, "learning_rate": 1.5993548387096774e-05, "loss": 0.3834, "step": 10542 }, { "epoch": 67.96144578313253, "grad_norm": 21.660245895385742, "learning_rate": 1.5990322580645163e-05, "loss": 0.5105, "step": 10543 }, { "epoch": 67.96787148594377, "grad_norm": 6.314919948577881, "learning_rate": 1.598709677419355e-05, "loss": 0.3936, "step": 10544 }, { "epoch": 67.97429718875502, "grad_norm": 6.728000164031982, "learning_rate": 1.5983870967741938e-05, "loss": 0.232, "step": 10545 }, { "epoch": 67.98072289156626, "grad_norm": 16.11785125732422, "learning_rate": 1.5980645161290324e-05, "loss": 0.2022, "step": 10546 }, { "epoch": 67.98714859437752, "grad_norm": 12.49441909790039, "learning_rate": 1.597741935483871e-05, "loss": 0.3449, "step": 10547 }, { "epoch": 67.99357429718876, "grad_norm": 5.72181510925293, "learning_rate": 1.59741935483871e-05, "loss": 0.2554, "step": 10548 }, { "epoch": 68.00321285140562, "grad_norm": 24.734596252441406, "learning_rate": 1.5970967741935484e-05, "loss": 0.4455, "step": 10549 }, { "epoch": 68.00963855421686, "grad_norm": 2.7591545581817627, "learning_rate": 1.596774193548387e-05, "loss": 0.1686, "step": 10550 }, { "epoch": 68.0160642570281, "grad_norm": 12.209945678710938, "learning_rate": 1.596451612903226e-05, "loss": 0.4762, "step": 10551 }, { "epoch": 68.02248995983936, "grad_norm": 20.200101852416992, "learning_rate": 1.5961290322580648e-05, "loss": 0.4594, "step": 10552 }, { "epoch": 68.0289156626506, "grad_norm": 10.111146926879883, "learning_rate": 1.5958064516129034e-05, "loss": 0.5527, "step": 10553 }, { "epoch": 68.03534136546185, "grad_norm": 19.14664077758789, "learning_rate": 1.595483870967742e-05, "loss": 0.2896, "step": 10554 }, { "epoch": 68.04176706827309, "grad_norm": 11.022343635559082, "learning_rate": 1.595161290322581e-05, "loss": 0.3354, "step": 10555 }, { "epoch": 68.04819277108433, "grad_norm": 8.74827766418457, "learning_rate": 1.5948387096774194e-05, "loss": 0.573, "step": 10556 }, { "epoch": 68.05461847389559, "grad_norm": 57.12520217895508, "learning_rate": 1.5945161290322583e-05, "loss": 0.3057, "step": 10557 }, { "epoch": 68.06104417670683, "grad_norm": 6.227731227874756, "learning_rate": 1.594193548387097e-05, "loss": 0.5069, "step": 10558 }, { "epoch": 68.06746987951807, "grad_norm": 5.064833641052246, "learning_rate": 1.5938709677419355e-05, "loss": 0.1155, "step": 10559 }, { "epoch": 68.07389558232931, "grad_norm": 17.669361114501953, "learning_rate": 1.5935483870967744e-05, "loss": 0.3915, "step": 10560 }, { "epoch": 68.08032128514056, "grad_norm": 8.929681777954102, "learning_rate": 1.593225806451613e-05, "loss": 0.6907, "step": 10561 }, { "epoch": 68.08674698795181, "grad_norm": 30.026554107666016, "learning_rate": 1.5929032258064515e-05, "loss": 0.6314, "step": 10562 }, { "epoch": 68.09317269076305, "grad_norm": 7.581883430480957, "learning_rate": 1.5925806451612904e-05, "loss": 0.4771, "step": 10563 }, { "epoch": 68.0995983935743, "grad_norm": 38.70206832885742, "learning_rate": 1.5922580645161294e-05, "loss": 0.4113, "step": 10564 }, { "epoch": 68.10602409638554, "grad_norm": 7.927651405334473, "learning_rate": 1.5919354838709676e-05, "loss": 0.368, "step": 10565 }, { "epoch": 68.11244979919678, "grad_norm": 2.4700963497161865, "learning_rate": 1.5916129032258065e-05, "loss": 0.2183, "step": 10566 }, { "epoch": 68.11887550200804, "grad_norm": 5.345782279968262, "learning_rate": 1.5912903225806454e-05, "loss": 0.2173, "step": 10567 }, { "epoch": 68.12530120481928, "grad_norm": 18.56327247619629, "learning_rate": 1.590967741935484e-05, "loss": 0.4111, "step": 10568 }, { "epoch": 68.13172690763052, "grad_norm": 8.787520408630371, "learning_rate": 1.5906451612903225e-05, "loss": 0.2968, "step": 10569 }, { "epoch": 68.13815261044176, "grad_norm": 17.994733810424805, "learning_rate": 1.5903225806451615e-05, "loss": 0.5317, "step": 10570 }, { "epoch": 68.144578313253, "grad_norm": 11.171520233154297, "learning_rate": 1.59e-05, "loss": 0.2002, "step": 10571 }, { "epoch": 68.15100401606426, "grad_norm": 16.0449275970459, "learning_rate": 1.589677419354839e-05, "loss": 0.393, "step": 10572 }, { "epoch": 68.1574297188755, "grad_norm": 12.703531265258789, "learning_rate": 1.5893548387096775e-05, "loss": 0.4736, "step": 10573 }, { "epoch": 68.16385542168675, "grad_norm": 42.4390869140625, "learning_rate": 1.589032258064516e-05, "loss": 0.625, "step": 10574 }, { "epoch": 68.17028112449799, "grad_norm": 8.919379234313965, "learning_rate": 1.588709677419355e-05, "loss": 0.3541, "step": 10575 }, { "epoch": 68.17670682730923, "grad_norm": 6.968826770782471, "learning_rate": 1.588387096774194e-05, "loss": 0.303, "step": 10576 }, { "epoch": 68.18313253012049, "grad_norm": 18.882709503173828, "learning_rate": 1.588064516129032e-05, "loss": 0.5748, "step": 10577 }, { "epoch": 68.18955823293173, "grad_norm": 4.714296340942383, "learning_rate": 1.587741935483871e-05, "loss": 0.1354, "step": 10578 }, { "epoch": 68.19598393574297, "grad_norm": 24.40467643737793, "learning_rate": 1.58741935483871e-05, "loss": 0.3962, "step": 10579 }, { "epoch": 68.20240963855422, "grad_norm": 24.71921730041504, "learning_rate": 1.5870967741935485e-05, "loss": 0.3697, "step": 10580 }, { "epoch": 68.20883534136546, "grad_norm": 6.215675354003906, "learning_rate": 1.586774193548387e-05, "loss": 0.3098, "step": 10581 }, { "epoch": 68.21526104417671, "grad_norm": 15.345829010009766, "learning_rate": 1.586451612903226e-05, "loss": 0.419, "step": 10582 }, { "epoch": 68.22168674698796, "grad_norm": 14.947734832763672, "learning_rate": 1.5861290322580646e-05, "loss": 0.6814, "step": 10583 }, { "epoch": 68.2281124497992, "grad_norm": 9.719700813293457, "learning_rate": 1.5858064516129035e-05, "loss": 0.2178, "step": 10584 }, { "epoch": 68.23453815261044, "grad_norm": 11.153547286987305, "learning_rate": 1.585483870967742e-05, "loss": 0.4634, "step": 10585 }, { "epoch": 68.24096385542168, "grad_norm": 3.7166178226470947, "learning_rate": 1.5851612903225806e-05, "loss": 0.0963, "step": 10586 }, { "epoch": 68.24738955823292, "grad_norm": 9.809256553649902, "learning_rate": 1.5848387096774195e-05, "loss": 0.2604, "step": 10587 }, { "epoch": 68.25381526104418, "grad_norm": 24.23883056640625, "learning_rate": 1.5845161290322584e-05, "loss": 0.5452, "step": 10588 }, { "epoch": 68.26024096385542, "grad_norm": 11.645397186279297, "learning_rate": 1.5841935483870967e-05, "loss": 0.9169, "step": 10589 }, { "epoch": 68.26666666666667, "grad_norm": 8.498936653137207, "learning_rate": 1.5838709677419356e-05, "loss": 0.2161, "step": 10590 }, { "epoch": 68.27309236947791, "grad_norm": 13.475993156433105, "learning_rate": 1.5835483870967745e-05, "loss": 0.3451, "step": 10591 }, { "epoch": 68.27951807228915, "grad_norm": 8.772191047668457, "learning_rate": 1.583225806451613e-05, "loss": 0.3551, "step": 10592 }, { "epoch": 68.2859437751004, "grad_norm": 8.398369789123535, "learning_rate": 1.5829032258064516e-05, "loss": 0.3614, "step": 10593 }, { "epoch": 68.29236947791165, "grad_norm": 12.57928466796875, "learning_rate": 1.5825806451612905e-05, "loss": 0.2726, "step": 10594 }, { "epoch": 68.29879518072289, "grad_norm": 4.487682819366455, "learning_rate": 1.582258064516129e-05, "loss": 0.0936, "step": 10595 }, { "epoch": 68.30522088353413, "grad_norm": 4.943413734436035, "learning_rate": 1.5819354838709677e-05, "loss": 0.1606, "step": 10596 }, { "epoch": 68.31164658634538, "grad_norm": 21.275279998779297, "learning_rate": 1.5816129032258066e-05, "loss": 0.3958, "step": 10597 }, { "epoch": 68.31807228915663, "grad_norm": 3.5227088928222656, "learning_rate": 1.581290322580645e-05, "loss": 0.2305, "step": 10598 }, { "epoch": 68.32449799196787, "grad_norm": 1.431037425994873, "learning_rate": 1.580967741935484e-05, "loss": 0.1617, "step": 10599 }, { "epoch": 68.33092369477912, "grad_norm": 10.993537902832031, "learning_rate": 1.5806451612903226e-05, "loss": 0.6315, "step": 10600 }, { "epoch": 68.33734939759036, "grad_norm": 8.846015930175781, "learning_rate": 1.5803225806451612e-05, "loss": 0.3051, "step": 10601 }, { "epoch": 68.3437751004016, "grad_norm": 5.014407634735107, "learning_rate": 1.58e-05, "loss": 0.2252, "step": 10602 }, { "epoch": 68.35020080321286, "grad_norm": 9.478387832641602, "learning_rate": 1.579677419354839e-05, "loss": 0.2391, "step": 10603 }, { "epoch": 68.3566265060241, "grad_norm": 8.287192344665527, "learning_rate": 1.5793548387096773e-05, "loss": 0.4191, "step": 10604 }, { "epoch": 68.36305220883534, "grad_norm": 12.497725486755371, "learning_rate": 1.5790322580645162e-05, "loss": 0.3064, "step": 10605 }, { "epoch": 68.36947791164658, "grad_norm": 7.9083147048950195, "learning_rate": 1.578709677419355e-05, "loss": 0.2741, "step": 10606 }, { "epoch": 68.37590361445783, "grad_norm": 18.19590187072754, "learning_rate": 1.5783870967741937e-05, "loss": 0.1586, "step": 10607 }, { "epoch": 68.38232931726908, "grad_norm": 4.363445281982422, "learning_rate": 1.5780645161290322e-05, "loss": 0.1613, "step": 10608 }, { "epoch": 68.38875502008032, "grad_norm": 2.299743413925171, "learning_rate": 1.577741935483871e-05, "loss": 0.1496, "step": 10609 }, { "epoch": 68.39518072289157, "grad_norm": 28.455020904541016, "learning_rate": 1.5774193548387097e-05, "loss": 0.1774, "step": 10610 }, { "epoch": 68.40160642570281, "grad_norm": 7.812076091766357, "learning_rate": 1.5770967741935486e-05, "loss": 0.3586, "step": 10611 }, { "epoch": 68.40803212851405, "grad_norm": 34.76642990112305, "learning_rate": 1.5767741935483872e-05, "loss": 0.2975, "step": 10612 }, { "epoch": 68.41445783132531, "grad_norm": 37.86546325683594, "learning_rate": 1.5764516129032258e-05, "loss": 0.2997, "step": 10613 }, { "epoch": 68.42088353413655, "grad_norm": 9.23888111114502, "learning_rate": 1.5761290322580647e-05, "loss": 0.6753, "step": 10614 }, { "epoch": 68.42730923694779, "grad_norm": 15.524300575256348, "learning_rate": 1.5758064516129036e-05, "loss": 0.2209, "step": 10615 }, { "epoch": 68.43373493975903, "grad_norm": 54.20780944824219, "learning_rate": 1.5754838709677418e-05, "loss": 0.5714, "step": 10616 }, { "epoch": 68.44016064257028, "grad_norm": 15.141263961791992, "learning_rate": 1.5751612903225807e-05, "loss": 0.5441, "step": 10617 }, { "epoch": 68.44658634538153, "grad_norm": 11.120865821838379, "learning_rate": 1.5748387096774196e-05, "loss": 0.2482, "step": 10618 }, { "epoch": 68.45301204819278, "grad_norm": 19.416318893432617, "learning_rate": 1.5745161290322582e-05, "loss": 0.5325, "step": 10619 }, { "epoch": 68.45943775100402, "grad_norm": 11.38959789276123, "learning_rate": 1.5741935483870968e-05, "loss": 0.3425, "step": 10620 }, { "epoch": 68.46586345381526, "grad_norm": 23.208166122436523, "learning_rate": 1.5738709677419357e-05, "loss": 0.3031, "step": 10621 }, { "epoch": 68.4722891566265, "grad_norm": 13.307455062866211, "learning_rate": 1.5735483870967742e-05, "loss": 0.1952, "step": 10622 }, { "epoch": 68.47871485943774, "grad_norm": 6.957086563110352, "learning_rate": 1.573225806451613e-05, "loss": 0.3382, "step": 10623 }, { "epoch": 68.485140562249, "grad_norm": 15.385689735412598, "learning_rate": 1.5729032258064517e-05, "loss": 0.3965, "step": 10624 }, { "epoch": 68.49156626506024, "grad_norm": 11.57314682006836, "learning_rate": 1.5725806451612903e-05, "loss": 0.4898, "step": 10625 }, { "epoch": 68.49799196787149, "grad_norm": 129.80404663085938, "learning_rate": 1.5722580645161292e-05, "loss": 0.4132, "step": 10626 }, { "epoch": 68.50441767068273, "grad_norm": 11.410440444946289, "learning_rate": 1.571935483870968e-05, "loss": 0.3478, "step": 10627 }, { "epoch": 68.51084337349397, "grad_norm": 23.39995765686035, "learning_rate": 1.5716129032258063e-05, "loss": 0.2588, "step": 10628 }, { "epoch": 68.51726907630523, "grad_norm": 11.083110809326172, "learning_rate": 1.5712903225806453e-05, "loss": 0.2938, "step": 10629 }, { "epoch": 68.52369477911647, "grad_norm": 7.668240070343018, "learning_rate": 1.570967741935484e-05, "loss": 0.2775, "step": 10630 }, { "epoch": 68.53012048192771, "grad_norm": 4.08010196685791, "learning_rate": 1.5706451612903224e-05, "loss": 0.1874, "step": 10631 }, { "epoch": 68.53654618473895, "grad_norm": 9.409632682800293, "learning_rate": 1.5703225806451613e-05, "loss": 0.1511, "step": 10632 }, { "epoch": 68.5429718875502, "grad_norm": 8.636675834655762, "learning_rate": 1.5700000000000002e-05, "loss": 0.2949, "step": 10633 }, { "epoch": 68.54939759036145, "grad_norm": 8.66132926940918, "learning_rate": 1.5696774193548388e-05, "loss": 0.405, "step": 10634 }, { "epoch": 68.5558232931727, "grad_norm": 9.490458488464355, "learning_rate": 1.5693548387096774e-05, "loss": 0.3429, "step": 10635 }, { "epoch": 68.56224899598394, "grad_norm": 10.333097457885742, "learning_rate": 1.5690322580645163e-05, "loss": 0.1986, "step": 10636 }, { "epoch": 68.56867469879518, "grad_norm": 28.51957893371582, "learning_rate": 1.568709677419355e-05, "loss": 0.4804, "step": 10637 }, { "epoch": 68.57510040160642, "grad_norm": 6.883852958679199, "learning_rate": 1.5683870967741937e-05, "loss": 0.4645, "step": 10638 }, { "epoch": 68.58152610441768, "grad_norm": 7.5570597648620605, "learning_rate": 1.5680645161290323e-05, "loss": 0.2374, "step": 10639 }, { "epoch": 68.58795180722892, "grad_norm": 20.404571533203125, "learning_rate": 1.567741935483871e-05, "loss": 0.6071, "step": 10640 }, { "epoch": 68.59437751004016, "grad_norm": 17.892255783081055, "learning_rate": 1.5674193548387098e-05, "loss": 0.4377, "step": 10641 }, { "epoch": 68.6008032128514, "grad_norm": 5.641824722290039, "learning_rate": 1.5670967741935487e-05, "loss": 0.4935, "step": 10642 }, { "epoch": 68.60722891566265, "grad_norm": 13.76407527923584, "learning_rate": 1.566774193548387e-05, "loss": 0.1292, "step": 10643 }, { "epoch": 68.6136546184739, "grad_norm": 15.194442749023438, "learning_rate": 1.566451612903226e-05, "loss": 0.1811, "step": 10644 }, { "epoch": 68.62008032128514, "grad_norm": 7.283559322357178, "learning_rate": 1.5661290322580648e-05, "loss": 0.7149, "step": 10645 }, { "epoch": 68.62650602409639, "grad_norm": 11.925901412963867, "learning_rate": 1.5658064516129033e-05, "loss": 0.1245, "step": 10646 }, { "epoch": 68.63293172690763, "grad_norm": 13.336392402648926, "learning_rate": 1.565483870967742e-05, "loss": 0.2681, "step": 10647 }, { "epoch": 68.63935742971887, "grad_norm": 10.098137855529785, "learning_rate": 1.5651612903225808e-05, "loss": 0.2532, "step": 10648 }, { "epoch": 68.64578313253013, "grad_norm": 9.31742000579834, "learning_rate": 1.5648387096774194e-05, "loss": 0.218, "step": 10649 }, { "epoch": 68.65220883534137, "grad_norm": 28.44939422607422, "learning_rate": 1.5645161290322583e-05, "loss": 0.5674, "step": 10650 }, { "epoch": 68.65863453815261, "grad_norm": 12.485673904418945, "learning_rate": 1.564193548387097e-05, "loss": 0.358, "step": 10651 }, { "epoch": 68.66506024096385, "grad_norm": 6.554621696472168, "learning_rate": 1.5638709677419354e-05, "loss": 0.3247, "step": 10652 }, { "epoch": 68.6714859437751, "grad_norm": 9.147716522216797, "learning_rate": 1.5635483870967743e-05, "loss": 0.2509, "step": 10653 }, { "epoch": 68.67791164658635, "grad_norm": 8.394251823425293, "learning_rate": 1.5632258064516132e-05, "loss": 0.2473, "step": 10654 }, { "epoch": 68.6843373493976, "grad_norm": 7.654251575469971, "learning_rate": 1.5629032258064515e-05, "loss": 0.337, "step": 10655 }, { "epoch": 68.69076305220884, "grad_norm": 3.0053579807281494, "learning_rate": 1.5625806451612904e-05, "loss": 0.3808, "step": 10656 }, { "epoch": 68.69718875502008, "grad_norm": 31.9644718170166, "learning_rate": 1.5622580645161293e-05, "loss": 0.4805, "step": 10657 }, { "epoch": 68.70361445783132, "grad_norm": 14.329480171203613, "learning_rate": 1.561935483870968e-05, "loss": 0.4955, "step": 10658 }, { "epoch": 68.71004016064256, "grad_norm": 12.232187271118164, "learning_rate": 1.5616129032258064e-05, "loss": 0.4105, "step": 10659 }, { "epoch": 68.71646586345382, "grad_norm": 71.07048797607422, "learning_rate": 1.5612903225806454e-05, "loss": 0.5477, "step": 10660 }, { "epoch": 68.72289156626506, "grad_norm": 9.75273609161377, "learning_rate": 1.560967741935484e-05, "loss": 0.529, "step": 10661 }, { "epoch": 68.7293172690763, "grad_norm": 5.479813098907471, "learning_rate": 1.5606451612903228e-05, "loss": 0.2833, "step": 10662 }, { "epoch": 68.73574297188755, "grad_norm": 3.4431142807006836, "learning_rate": 1.5603225806451614e-05, "loss": 0.1281, "step": 10663 }, { "epoch": 68.74216867469879, "grad_norm": 17.376041412353516, "learning_rate": 1.56e-05, "loss": 0.6231, "step": 10664 }, { "epoch": 68.74859437751005, "grad_norm": 13.3538818359375, "learning_rate": 1.559677419354839e-05, "loss": 0.34, "step": 10665 }, { "epoch": 68.75502008032129, "grad_norm": 9.469456672668457, "learning_rate": 1.5593548387096775e-05, "loss": 0.5415, "step": 10666 }, { "epoch": 68.76144578313253, "grad_norm": 18.128324508666992, "learning_rate": 1.559032258064516e-05, "loss": 0.3616, "step": 10667 }, { "epoch": 68.76787148594377, "grad_norm": 27.257583618164062, "learning_rate": 1.558709677419355e-05, "loss": 0.573, "step": 10668 }, { "epoch": 68.77429718875501, "grad_norm": 26.792673110961914, "learning_rate": 1.558387096774194e-05, "loss": 0.5296, "step": 10669 }, { "epoch": 68.78072289156627, "grad_norm": 10.15999698638916, "learning_rate": 1.558064516129032e-05, "loss": 0.4446, "step": 10670 }, { "epoch": 68.78714859437751, "grad_norm": 28.002161026000977, "learning_rate": 1.557741935483871e-05, "loss": 0.3417, "step": 10671 }, { "epoch": 68.79357429718875, "grad_norm": 2.712322235107422, "learning_rate": 1.55741935483871e-05, "loss": 0.2094, "step": 10672 }, { "epoch": 68.8, "grad_norm": 8.98261547088623, "learning_rate": 1.5570967741935485e-05, "loss": 0.2895, "step": 10673 }, { "epoch": 68.80642570281124, "grad_norm": 13.525171279907227, "learning_rate": 1.556774193548387e-05, "loss": 0.2693, "step": 10674 }, { "epoch": 68.8128514056225, "grad_norm": 12.369752883911133, "learning_rate": 1.556451612903226e-05, "loss": 0.3893, "step": 10675 }, { "epoch": 68.81927710843374, "grad_norm": 20.013729095458984, "learning_rate": 1.5561290322580645e-05, "loss": 0.2478, "step": 10676 }, { "epoch": 68.82570281124498, "grad_norm": 9.96346664428711, "learning_rate": 1.5558064516129034e-05, "loss": 0.4416, "step": 10677 }, { "epoch": 68.83212851405622, "grad_norm": 6.157644271850586, "learning_rate": 1.555483870967742e-05, "loss": 0.2184, "step": 10678 }, { "epoch": 68.83855421686746, "grad_norm": 9.519196510314941, "learning_rate": 1.5551612903225806e-05, "loss": 0.0985, "step": 10679 }, { "epoch": 68.84497991967872, "grad_norm": 66.89995574951172, "learning_rate": 1.5548387096774195e-05, "loss": 0.3935, "step": 10680 }, { "epoch": 68.85140562248996, "grad_norm": 5.623855113983154, "learning_rate": 1.5545161290322584e-05, "loss": 0.2131, "step": 10681 }, { "epoch": 68.8578313253012, "grad_norm": 31.671951293945312, "learning_rate": 1.5541935483870966e-05, "loss": 0.3775, "step": 10682 }, { "epoch": 68.86425702811245, "grad_norm": 10.987950325012207, "learning_rate": 1.5538709677419355e-05, "loss": 0.1918, "step": 10683 }, { "epoch": 68.87068273092369, "grad_norm": 35.44477844238281, "learning_rate": 1.5535483870967744e-05, "loss": 0.3129, "step": 10684 }, { "epoch": 68.87710843373495, "grad_norm": 25.091596603393555, "learning_rate": 1.553225806451613e-05, "loss": 0.1904, "step": 10685 }, { "epoch": 68.88353413654619, "grad_norm": 11.10919189453125, "learning_rate": 1.5529032258064516e-05, "loss": 0.2487, "step": 10686 }, { "epoch": 68.88995983935743, "grad_norm": 7.686821460723877, "learning_rate": 1.5525806451612905e-05, "loss": 0.5183, "step": 10687 }, { "epoch": 68.89638554216867, "grad_norm": 13.661816596984863, "learning_rate": 1.552258064516129e-05, "loss": 0.2635, "step": 10688 }, { "epoch": 68.90281124497992, "grad_norm": 4.656649112701416, "learning_rate": 1.551935483870968e-05, "loss": 0.1452, "step": 10689 }, { "epoch": 68.90923694779117, "grad_norm": 5.498107433319092, "learning_rate": 1.5516129032258065e-05, "loss": 0.2009, "step": 10690 }, { "epoch": 68.91566265060241, "grad_norm": 14.897470474243164, "learning_rate": 1.551290322580645e-05, "loss": 0.3178, "step": 10691 }, { "epoch": 68.92208835341366, "grad_norm": 7.4747419357299805, "learning_rate": 1.550967741935484e-05, "loss": 0.4626, "step": 10692 }, { "epoch": 68.9285140562249, "grad_norm": 8.382161140441895, "learning_rate": 1.550645161290323e-05, "loss": 0.1363, "step": 10693 }, { "epoch": 68.93493975903614, "grad_norm": 5.492730140686035, "learning_rate": 1.550322580645161e-05, "loss": 0.2928, "step": 10694 }, { "epoch": 68.94136546184738, "grad_norm": 16.34708595275879, "learning_rate": 1.55e-05, "loss": 0.2295, "step": 10695 }, { "epoch": 68.94779116465864, "grad_norm": 17.8239803314209, "learning_rate": 1.549677419354839e-05, "loss": 0.3414, "step": 10696 }, { "epoch": 68.95421686746988, "grad_norm": 7.830305576324463, "learning_rate": 1.5493548387096775e-05, "loss": 0.369, "step": 10697 }, { "epoch": 68.96064257028112, "grad_norm": 9.11733627319336, "learning_rate": 1.549032258064516e-05, "loss": 0.2896, "step": 10698 }, { "epoch": 68.96706827309237, "grad_norm": 6.85371732711792, "learning_rate": 1.548709677419355e-05, "loss": 0.2443, "step": 10699 }, { "epoch": 68.97349397590361, "grad_norm": 12.3607759475708, "learning_rate": 1.5483870967741936e-05, "loss": 0.2507, "step": 10700 }, { "epoch": 68.97991967871486, "grad_norm": 6.951343536376953, "learning_rate": 1.548064516129032e-05, "loss": 0.3215, "step": 10701 }, { "epoch": 68.9863453815261, "grad_norm": 7.611929893493652, "learning_rate": 1.547741935483871e-05, "loss": 0.3169, "step": 10702 }, { "epoch": 68.99277108433735, "grad_norm": 7.385714530944824, "learning_rate": 1.5474193548387096e-05, "loss": 0.1379, "step": 10703 }, { "epoch": 69.00240963855421, "grad_norm": 42.576576232910156, "learning_rate": 1.5470967741935486e-05, "loss": 0.2132, "step": 10704 }, { "epoch": 69.00883534136547, "grad_norm": 7.549677848815918, "learning_rate": 1.546774193548387e-05, "loss": 0.3813, "step": 10705 }, { "epoch": 69.01526104417671, "grad_norm": 18.839834213256836, "learning_rate": 1.5464516129032257e-05, "loss": 0.4377, "step": 10706 }, { "epoch": 69.02168674698795, "grad_norm": 4.506725311279297, "learning_rate": 1.5461290322580646e-05, "loss": 0.2013, "step": 10707 }, { "epoch": 69.0281124497992, "grad_norm": 8.271048545837402, "learning_rate": 1.5458064516129035e-05, "loss": 0.2948, "step": 10708 }, { "epoch": 69.03453815261044, "grad_norm": 12.028121948242188, "learning_rate": 1.5454838709677418e-05, "loss": 0.4815, "step": 10709 }, { "epoch": 69.04096385542168, "grad_norm": 5.504416465759277, "learning_rate": 1.5451612903225807e-05, "loss": 0.3594, "step": 10710 }, { "epoch": 69.04738955823294, "grad_norm": 10.487569808959961, "learning_rate": 1.5448387096774196e-05, "loss": 0.3971, "step": 10711 }, { "epoch": 69.05381526104418, "grad_norm": 30.185327529907227, "learning_rate": 1.544516129032258e-05, "loss": 0.7363, "step": 10712 }, { "epoch": 69.06024096385542, "grad_norm": 11.675618171691895, "learning_rate": 1.5441935483870967e-05, "loss": 0.1557, "step": 10713 }, { "epoch": 69.06666666666666, "grad_norm": 4.59666109085083, "learning_rate": 1.5438709677419356e-05, "loss": 0.1881, "step": 10714 }, { "epoch": 69.0730923694779, "grad_norm": 12.655189514160156, "learning_rate": 1.5435483870967742e-05, "loss": 0.4323, "step": 10715 }, { "epoch": 69.07951807228916, "grad_norm": 11.61057186126709, "learning_rate": 1.543225806451613e-05, "loss": 0.4766, "step": 10716 }, { "epoch": 69.0859437751004, "grad_norm": 17.51173210144043, "learning_rate": 1.5429032258064517e-05, "loss": 0.139, "step": 10717 }, { "epoch": 69.09236947791165, "grad_norm": 8.19989013671875, "learning_rate": 1.5425806451612902e-05, "loss": 0.1766, "step": 10718 }, { "epoch": 69.09879518072289, "grad_norm": 7.38560676574707, "learning_rate": 1.542258064516129e-05, "loss": 0.3389, "step": 10719 }, { "epoch": 69.10522088353413, "grad_norm": 18.595016479492188, "learning_rate": 1.541935483870968e-05, "loss": 0.5381, "step": 10720 }, { "epoch": 69.11164658634539, "grad_norm": 18.02616310119629, "learning_rate": 1.5416129032258063e-05, "loss": 0.5467, "step": 10721 }, { "epoch": 69.11807228915663, "grad_norm": 38.896141052246094, "learning_rate": 1.5412903225806452e-05, "loss": 0.6419, "step": 10722 }, { "epoch": 69.12449799196787, "grad_norm": 7.751284122467041, "learning_rate": 1.540967741935484e-05, "loss": 0.3291, "step": 10723 }, { "epoch": 69.13092369477911, "grad_norm": 71.83285522460938, "learning_rate": 1.5406451612903227e-05, "loss": 0.5027, "step": 10724 }, { "epoch": 69.13734939759036, "grad_norm": 8.610201835632324, "learning_rate": 1.5403225806451613e-05, "loss": 0.6825, "step": 10725 }, { "epoch": 69.14377510040161, "grad_norm": 4.555269241333008, "learning_rate": 1.54e-05, "loss": 0.176, "step": 10726 }, { "epoch": 69.15020080321285, "grad_norm": 10.480067253112793, "learning_rate": 1.5396774193548387e-05, "loss": 0.222, "step": 10727 }, { "epoch": 69.1566265060241, "grad_norm": 5.283851623535156, "learning_rate": 1.5393548387096776e-05, "loss": 0.4709, "step": 10728 }, { "epoch": 69.16305220883534, "grad_norm": 26.305604934692383, "learning_rate": 1.5390322580645162e-05, "loss": 0.4012, "step": 10729 }, { "epoch": 69.16947791164658, "grad_norm": 6.296440124511719, "learning_rate": 1.5387096774193548e-05, "loss": 0.1874, "step": 10730 }, { "epoch": 69.17590361445784, "grad_norm": 25.483911514282227, "learning_rate": 1.5383870967741937e-05, "loss": 0.3625, "step": 10731 }, { "epoch": 69.18232931726908, "grad_norm": 10.191533088684082, "learning_rate": 1.5380645161290326e-05, "loss": 0.2745, "step": 10732 }, { "epoch": 69.18875502008032, "grad_norm": 35.87990951538086, "learning_rate": 1.537741935483871e-05, "loss": 0.5288, "step": 10733 }, { "epoch": 69.19518072289156, "grad_norm": 5.195704936981201, "learning_rate": 1.5374193548387097e-05, "loss": 0.2796, "step": 10734 }, { "epoch": 69.2016064257028, "grad_norm": 17.526575088500977, "learning_rate": 1.5370967741935487e-05, "loss": 0.2956, "step": 10735 }, { "epoch": 69.20803212851406, "grad_norm": 25.284900665283203, "learning_rate": 1.5367741935483872e-05, "loss": 0.2586, "step": 10736 }, { "epoch": 69.2144578313253, "grad_norm": 12.069411277770996, "learning_rate": 1.5364516129032258e-05, "loss": 0.3707, "step": 10737 }, { "epoch": 69.22088353413655, "grad_norm": 8.034399032592773, "learning_rate": 1.5361290322580647e-05, "loss": 0.2628, "step": 10738 }, { "epoch": 69.22730923694779, "grad_norm": 46.28340148925781, "learning_rate": 1.5358064516129033e-05, "loss": 0.2507, "step": 10739 }, { "epoch": 69.23373493975903, "grad_norm": 5.162405967712402, "learning_rate": 1.535483870967742e-05, "loss": 0.2282, "step": 10740 }, { "epoch": 69.24016064257029, "grad_norm": 15.272724151611328, "learning_rate": 1.5351612903225808e-05, "loss": 0.5109, "step": 10741 }, { "epoch": 69.24658634538153, "grad_norm": 16.453596115112305, "learning_rate": 1.5348387096774193e-05, "loss": 0.586, "step": 10742 }, { "epoch": 69.25301204819277, "grad_norm": 29.18878746032715, "learning_rate": 1.5345161290322582e-05, "loss": 0.4911, "step": 10743 }, { "epoch": 69.25943775100401, "grad_norm": 5.554053783416748, "learning_rate": 1.5341935483870968e-05, "loss": 0.1786, "step": 10744 }, { "epoch": 69.26586345381526, "grad_norm": 10.119221687316895, "learning_rate": 1.5338709677419354e-05, "loss": 0.2792, "step": 10745 }, { "epoch": 69.2722891566265, "grad_norm": 8.037721633911133, "learning_rate": 1.5335483870967743e-05, "loss": 0.4538, "step": 10746 }, { "epoch": 69.27871485943776, "grad_norm": 12.10954475402832, "learning_rate": 1.5332258064516132e-05, "loss": 0.3257, "step": 10747 }, { "epoch": 69.285140562249, "grad_norm": 11.581870079040527, "learning_rate": 1.5329032258064514e-05, "loss": 0.569, "step": 10748 }, { "epoch": 69.29156626506024, "grad_norm": 6.98714542388916, "learning_rate": 1.5325806451612903e-05, "loss": 0.3619, "step": 10749 }, { "epoch": 69.29799196787148, "grad_norm": 16.5802059173584, "learning_rate": 1.5322580645161292e-05, "loss": 0.3051, "step": 10750 }, { "epoch": 69.30441767068272, "grad_norm": 17.24944496154785, "learning_rate": 1.5319354838709678e-05, "loss": 0.9158, "step": 10751 }, { "epoch": 69.31084337349398, "grad_norm": 12.367106437683105, "learning_rate": 1.5316129032258064e-05, "loss": 0.2628, "step": 10752 }, { "epoch": 69.31726907630522, "grad_norm": 7.160524368286133, "learning_rate": 1.5312903225806453e-05, "loss": 0.2876, "step": 10753 }, { "epoch": 69.32369477911647, "grad_norm": 16.02552604675293, "learning_rate": 1.530967741935484e-05, "loss": 0.3728, "step": 10754 }, { "epoch": 69.33012048192771, "grad_norm": 6.259268283843994, "learning_rate": 1.5306451612903228e-05, "loss": 0.4137, "step": 10755 }, { "epoch": 69.33654618473895, "grad_norm": 8.29456901550293, "learning_rate": 1.5303225806451613e-05, "loss": 0.3752, "step": 10756 }, { "epoch": 69.3429718875502, "grad_norm": 27.60688591003418, "learning_rate": 1.53e-05, "loss": 0.62, "step": 10757 }, { "epoch": 69.34939759036145, "grad_norm": 6.336475849151611, "learning_rate": 1.5296774193548388e-05, "loss": 0.1595, "step": 10758 }, { "epoch": 69.35582329317269, "grad_norm": 7.946515083312988, "learning_rate": 1.5293548387096777e-05, "loss": 0.2576, "step": 10759 }, { "epoch": 69.36224899598393, "grad_norm": 13.38344955444336, "learning_rate": 1.529032258064516e-05, "loss": 0.196, "step": 10760 }, { "epoch": 69.36867469879518, "grad_norm": 4.481596946716309, "learning_rate": 1.528709677419355e-05, "loss": 0.3184, "step": 10761 }, { "epoch": 69.37510040160643, "grad_norm": 16.099658966064453, "learning_rate": 1.5283870967741938e-05, "loss": 0.5706, "step": 10762 }, { "epoch": 69.38152610441767, "grad_norm": 8.547609329223633, "learning_rate": 1.5280645161290324e-05, "loss": 0.7765, "step": 10763 }, { "epoch": 69.38795180722892, "grad_norm": 9.588772773742676, "learning_rate": 1.527741935483871e-05, "loss": 0.4257, "step": 10764 }, { "epoch": 69.39437751004016, "grad_norm": 9.032954216003418, "learning_rate": 1.52741935483871e-05, "loss": 0.1318, "step": 10765 }, { "epoch": 69.4008032128514, "grad_norm": 19.494413375854492, "learning_rate": 1.5270967741935484e-05, "loss": 0.2698, "step": 10766 }, { "epoch": 69.40722891566266, "grad_norm": 13.050572395324707, "learning_rate": 1.5267741935483873e-05, "loss": 0.5899, "step": 10767 }, { "epoch": 69.4136546184739, "grad_norm": 11.77628231048584, "learning_rate": 1.526451612903226e-05, "loss": 0.3395, "step": 10768 }, { "epoch": 69.42008032128514, "grad_norm": 9.771559715270996, "learning_rate": 1.5261290322580645e-05, "loss": 0.4922, "step": 10769 }, { "epoch": 69.42650602409638, "grad_norm": 10.989027976989746, "learning_rate": 1.5258064516129034e-05, "loss": 0.5557, "step": 10770 }, { "epoch": 69.43293172690763, "grad_norm": 30.06463623046875, "learning_rate": 1.5254838709677421e-05, "loss": 0.4673, "step": 10771 }, { "epoch": 69.43935742971888, "grad_norm": 3.1940338611602783, "learning_rate": 1.5251612903225807e-05, "loss": 0.138, "step": 10772 }, { "epoch": 69.44578313253012, "grad_norm": 18.974138259887695, "learning_rate": 1.5248387096774194e-05, "loss": 0.2922, "step": 10773 }, { "epoch": 69.45220883534137, "grad_norm": 7.279081344604492, "learning_rate": 1.5245161290322582e-05, "loss": 0.343, "step": 10774 }, { "epoch": 69.45863453815261, "grad_norm": 17.735591888427734, "learning_rate": 1.5241935483870967e-05, "loss": 0.2517, "step": 10775 }, { "epoch": 69.46506024096385, "grad_norm": 8.263099670410156, "learning_rate": 1.5238709677419355e-05, "loss": 0.5585, "step": 10776 }, { "epoch": 69.47148594377511, "grad_norm": 39.34904479980469, "learning_rate": 1.5235483870967742e-05, "loss": 0.3462, "step": 10777 }, { "epoch": 69.47791164658635, "grad_norm": 6.3162007331848145, "learning_rate": 1.5232258064516131e-05, "loss": 0.4576, "step": 10778 }, { "epoch": 69.48433734939759, "grad_norm": 19.703153610229492, "learning_rate": 1.5229032258064515e-05, "loss": 0.3612, "step": 10779 }, { "epoch": 69.49076305220883, "grad_norm": 13.74382209777832, "learning_rate": 1.5225806451612903e-05, "loss": 0.2931, "step": 10780 }, { "epoch": 69.49718875502008, "grad_norm": 5.455018997192383, "learning_rate": 1.5222580645161292e-05, "loss": 0.2067, "step": 10781 }, { "epoch": 69.50361445783132, "grad_norm": 12.672283172607422, "learning_rate": 1.5219354838709679e-05, "loss": 0.267, "step": 10782 }, { "epoch": 69.51004016064257, "grad_norm": 38.2683219909668, "learning_rate": 1.5216129032258065e-05, "loss": 0.4735, "step": 10783 }, { "epoch": 69.51646586345382, "grad_norm": 6.171727180480957, "learning_rate": 1.5212903225806452e-05, "loss": 0.5244, "step": 10784 }, { "epoch": 69.52289156626506, "grad_norm": 22.848926544189453, "learning_rate": 1.520967741935484e-05, "loss": 0.3337, "step": 10785 }, { "epoch": 69.5293172690763, "grad_norm": 13.409778594970703, "learning_rate": 1.5206451612903227e-05, "loss": 0.5734, "step": 10786 }, { "epoch": 69.53574297188754, "grad_norm": 11.818140983581543, "learning_rate": 1.5203225806451613e-05, "loss": 0.2638, "step": 10787 }, { "epoch": 69.5421686746988, "grad_norm": 5.768368721008301, "learning_rate": 1.52e-05, "loss": 0.1727, "step": 10788 }, { "epoch": 69.54859437751004, "grad_norm": 18.735301971435547, "learning_rate": 1.5196774193548388e-05, "loss": 0.3657, "step": 10789 }, { "epoch": 69.55502008032128, "grad_norm": 12.718650817871094, "learning_rate": 1.5193548387096777e-05, "loss": 0.2154, "step": 10790 }, { "epoch": 69.56144578313253, "grad_norm": 10.436798095703125, "learning_rate": 1.519032258064516e-05, "loss": 0.2383, "step": 10791 }, { "epoch": 69.56787148594377, "grad_norm": 22.863449096679688, "learning_rate": 1.5187096774193548e-05, "loss": 0.5867, "step": 10792 }, { "epoch": 69.57429718875503, "grad_norm": 7.708557605743408, "learning_rate": 1.5183870967741937e-05, "loss": 0.1994, "step": 10793 }, { "epoch": 69.58072289156627, "grad_norm": 9.95473861694336, "learning_rate": 1.5180645161290325e-05, "loss": 0.4248, "step": 10794 }, { "epoch": 69.58714859437751, "grad_norm": 5.747898101806641, "learning_rate": 1.517741935483871e-05, "loss": 0.3794, "step": 10795 }, { "epoch": 69.59357429718875, "grad_norm": 5.563554763793945, "learning_rate": 1.5174193548387098e-05, "loss": 0.3243, "step": 10796 }, { "epoch": 69.6, "grad_norm": 8.184593200683594, "learning_rate": 1.5170967741935485e-05, "loss": 0.5526, "step": 10797 }, { "epoch": 69.60642570281125, "grad_norm": 15.511151313781738, "learning_rate": 1.5167741935483872e-05, "loss": 0.2888, "step": 10798 }, { "epoch": 69.6128514056225, "grad_norm": 8.65613842010498, "learning_rate": 1.5164516129032258e-05, "loss": 0.3134, "step": 10799 }, { "epoch": 69.61927710843374, "grad_norm": 6.611145496368408, "learning_rate": 1.5161290322580646e-05, "loss": 0.3222, "step": 10800 }, { "epoch": 69.62570281124498, "grad_norm": 4.621960639953613, "learning_rate": 1.5158064516129033e-05, "loss": 0.2003, "step": 10801 }, { "epoch": 69.63212851405622, "grad_norm": 16.777677536010742, "learning_rate": 1.5154838709677422e-05, "loss": 0.4029, "step": 10802 }, { "epoch": 69.63855421686748, "grad_norm": 12.173711776733398, "learning_rate": 1.5151612903225806e-05, "loss": 0.6705, "step": 10803 }, { "epoch": 69.64497991967872, "grad_norm": 17.381874084472656, "learning_rate": 1.5148387096774193e-05, "loss": 0.4889, "step": 10804 }, { "epoch": 69.65140562248996, "grad_norm": 9.944095611572266, "learning_rate": 1.5145161290322583e-05, "loss": 0.2902, "step": 10805 }, { "epoch": 69.6578313253012, "grad_norm": 8.977365493774414, "learning_rate": 1.514193548387097e-05, "loss": 0.2686, "step": 10806 }, { "epoch": 69.66425702811244, "grad_norm": 11.38762092590332, "learning_rate": 1.5138709677419356e-05, "loss": 0.5076, "step": 10807 }, { "epoch": 69.6706827309237, "grad_norm": 12.880350112915039, "learning_rate": 1.5135483870967743e-05, "loss": 0.107, "step": 10808 }, { "epoch": 69.67710843373494, "grad_norm": 10.214906692504883, "learning_rate": 1.513225806451613e-05, "loss": 0.2337, "step": 10809 }, { "epoch": 69.68353413654619, "grad_norm": 14.761795997619629, "learning_rate": 1.5129032258064516e-05, "loss": 0.4631, "step": 10810 }, { "epoch": 69.68995983935743, "grad_norm": 41.9510383605957, "learning_rate": 1.5125806451612904e-05, "loss": 0.4507, "step": 10811 }, { "epoch": 69.69638554216867, "grad_norm": 9.436812400817871, "learning_rate": 1.5122580645161291e-05, "loss": 0.3856, "step": 10812 }, { "epoch": 69.70281124497993, "grad_norm": 5.072901248931885, "learning_rate": 1.5119354838709678e-05, "loss": 0.1085, "step": 10813 }, { "epoch": 69.70923694779117, "grad_norm": 8.023699760437012, "learning_rate": 1.5116129032258064e-05, "loss": 0.1967, "step": 10814 }, { "epoch": 69.71566265060241, "grad_norm": 4.569772243499756, "learning_rate": 1.5112903225806452e-05, "loss": 0.2356, "step": 10815 }, { "epoch": 69.72208835341365, "grad_norm": 4.107898235321045, "learning_rate": 1.5109677419354839e-05, "loss": 0.1251, "step": 10816 }, { "epoch": 69.7285140562249, "grad_norm": 6.00607967376709, "learning_rate": 1.5106451612903228e-05, "loss": 0.2885, "step": 10817 }, { "epoch": 69.73493975903614, "grad_norm": 19.690916061401367, "learning_rate": 1.5103225806451612e-05, "loss": 0.429, "step": 10818 }, { "epoch": 69.7413654618474, "grad_norm": 7.404109954833984, "learning_rate": 1.51e-05, "loss": 0.5145, "step": 10819 }, { "epoch": 69.74779116465864, "grad_norm": 6.46242094039917, "learning_rate": 1.5096774193548389e-05, "loss": 0.5243, "step": 10820 }, { "epoch": 69.75421686746988, "grad_norm": 22.149742126464844, "learning_rate": 1.5093548387096776e-05, "loss": 0.1155, "step": 10821 }, { "epoch": 69.76064257028112, "grad_norm": 9.365462303161621, "learning_rate": 1.5090322580645162e-05, "loss": 0.4134, "step": 10822 }, { "epoch": 69.76706827309236, "grad_norm": 17.367456436157227, "learning_rate": 1.5087096774193549e-05, "loss": 0.2168, "step": 10823 }, { "epoch": 69.77349397590362, "grad_norm": 4.716618061065674, "learning_rate": 1.5083870967741936e-05, "loss": 0.269, "step": 10824 }, { "epoch": 69.77991967871486, "grad_norm": 7.3708062171936035, "learning_rate": 1.5080645161290324e-05, "loss": 0.6133, "step": 10825 }, { "epoch": 69.7863453815261, "grad_norm": 3.1852023601531982, "learning_rate": 1.507741935483871e-05, "loss": 0.1635, "step": 10826 }, { "epoch": 69.79277108433735, "grad_norm": 10.707415580749512, "learning_rate": 1.5074193548387097e-05, "loss": 0.2473, "step": 10827 }, { "epoch": 69.79919678714859, "grad_norm": 8.847635269165039, "learning_rate": 1.5070967741935484e-05, "loss": 0.3197, "step": 10828 }, { "epoch": 69.80562248995984, "grad_norm": 31.851327896118164, "learning_rate": 1.5067741935483873e-05, "loss": 0.3783, "step": 10829 }, { "epoch": 69.81204819277109, "grad_norm": 7.208048343658447, "learning_rate": 1.5064516129032257e-05, "loss": 0.2703, "step": 10830 }, { "epoch": 69.81847389558233, "grad_norm": 5.609486103057861, "learning_rate": 1.5061290322580645e-05, "loss": 0.3326, "step": 10831 }, { "epoch": 69.82489959839357, "grad_norm": 8.501166343688965, "learning_rate": 1.5058064516129034e-05, "loss": 0.1157, "step": 10832 }, { "epoch": 69.83132530120481, "grad_norm": 15.745312690734863, "learning_rate": 1.5054838709677421e-05, "loss": 0.4578, "step": 10833 }, { "epoch": 69.83775100401607, "grad_norm": 12.753639221191406, "learning_rate": 1.5051612903225807e-05, "loss": 0.5886, "step": 10834 }, { "epoch": 69.84417670682731, "grad_norm": 20.174501419067383, "learning_rate": 1.5048387096774194e-05, "loss": 0.4535, "step": 10835 }, { "epoch": 69.85060240963855, "grad_norm": 12.002389907836914, "learning_rate": 1.5045161290322582e-05, "loss": 0.4003, "step": 10836 }, { "epoch": 69.8570281124498, "grad_norm": 13.916284561157227, "learning_rate": 1.504193548387097e-05, "loss": 0.2284, "step": 10837 }, { "epoch": 69.86345381526104, "grad_norm": 9.86619758605957, "learning_rate": 1.5038709677419355e-05, "loss": 0.4448, "step": 10838 }, { "epoch": 69.8698795180723, "grad_norm": 7.631714344024658, "learning_rate": 1.5035483870967742e-05, "loss": 0.2892, "step": 10839 }, { "epoch": 69.87630522088354, "grad_norm": 11.616436004638672, "learning_rate": 1.503225806451613e-05, "loss": 0.275, "step": 10840 }, { "epoch": 69.88273092369478, "grad_norm": 25.031055450439453, "learning_rate": 1.5029032258064519e-05, "loss": 0.2925, "step": 10841 }, { "epoch": 69.88915662650602, "grad_norm": 15.989583969116211, "learning_rate": 1.5025806451612903e-05, "loss": 0.4076, "step": 10842 }, { "epoch": 69.89558232931726, "grad_norm": 9.128373146057129, "learning_rate": 1.502258064516129e-05, "loss": 0.6231, "step": 10843 }, { "epoch": 69.90200803212852, "grad_norm": 27.643688201904297, "learning_rate": 1.501935483870968e-05, "loss": 0.5587, "step": 10844 }, { "epoch": 69.90843373493976, "grad_norm": 7.164669990539551, "learning_rate": 1.5016129032258063e-05, "loss": 0.2898, "step": 10845 }, { "epoch": 69.914859437751, "grad_norm": 44.524662017822266, "learning_rate": 1.5012903225806452e-05, "loss": 0.4673, "step": 10846 }, { "epoch": 69.92128514056225, "grad_norm": 24.55091094970703, "learning_rate": 1.500967741935484e-05, "loss": 0.4647, "step": 10847 }, { "epoch": 69.92771084337349, "grad_norm": 6.6307525634765625, "learning_rate": 1.5006451612903227e-05, "loss": 0.2538, "step": 10848 }, { "epoch": 69.93413654618475, "grad_norm": 10.111104011535645, "learning_rate": 1.5003225806451613e-05, "loss": 0.5777, "step": 10849 }, { "epoch": 69.94056224899599, "grad_norm": 19.2015438079834, "learning_rate": 1.5e-05, "loss": 0.4224, "step": 10850 }, { "epoch": 69.94698795180723, "grad_norm": 15.468301773071289, "learning_rate": 1.4996774193548388e-05, "loss": 0.2125, "step": 10851 }, { "epoch": 69.95341365461847, "grad_norm": 10.710100173950195, "learning_rate": 1.4993548387096775e-05, "loss": 0.34, "step": 10852 }, { "epoch": 69.95983935742971, "grad_norm": 31.31772804260254, "learning_rate": 1.4990322580645161e-05, "loss": 0.4338, "step": 10853 }, { "epoch": 69.96626506024096, "grad_norm": 17.466182708740234, "learning_rate": 1.4987096774193548e-05, "loss": 0.5302, "step": 10854 }, { "epoch": 69.97269076305221, "grad_norm": 11.221983909606934, "learning_rate": 1.4983870967741936e-05, "loss": 0.1821, "step": 10855 }, { "epoch": 69.97911646586346, "grad_norm": 26.767623901367188, "learning_rate": 1.4980645161290325e-05, "loss": 0.6397, "step": 10856 }, { "epoch": 69.9855421686747, "grad_norm": 8.19560718536377, "learning_rate": 1.4977419354838709e-05, "loss": 0.3892, "step": 10857 }, { "epoch": 69.99196787148594, "grad_norm": 15.124629974365234, "learning_rate": 1.4974193548387098e-05, "loss": 0.2116, "step": 10858 }, { "epoch": 70.00160642570282, "grad_norm": 10.69879150390625, "learning_rate": 1.4970967741935485e-05, "loss": 0.3416, "step": 10859 }, { "epoch": 70.00803212851406, "grad_norm": 18.537559509277344, "learning_rate": 1.4967741935483873e-05, "loss": 0.4307, "step": 10860 }, { "epoch": 70.0144578313253, "grad_norm": 11.169251441955566, "learning_rate": 1.4964516129032258e-05, "loss": 0.2353, "step": 10861 }, { "epoch": 70.02088353413654, "grad_norm": 10.190794944763184, "learning_rate": 1.4961290322580646e-05, "loss": 0.4957, "step": 10862 }, { "epoch": 70.02730923694779, "grad_norm": 17.237293243408203, "learning_rate": 1.4958064516129033e-05, "loss": 0.3042, "step": 10863 }, { "epoch": 70.03373493975904, "grad_norm": 7.312419891357422, "learning_rate": 1.495483870967742e-05, "loss": 0.3613, "step": 10864 }, { "epoch": 70.04016064257029, "grad_norm": 14.317428588867188, "learning_rate": 1.4951612903225806e-05, "loss": 0.2212, "step": 10865 }, { "epoch": 70.04658634538153, "grad_norm": 33.10057830810547, "learning_rate": 1.4948387096774194e-05, "loss": 0.3828, "step": 10866 }, { "epoch": 70.05301204819277, "grad_norm": 9.200424194335938, "learning_rate": 1.4945161290322581e-05, "loss": 0.4871, "step": 10867 }, { "epoch": 70.05943775100401, "grad_norm": 11.2037353515625, "learning_rate": 1.494193548387097e-05, "loss": 0.2066, "step": 10868 }, { "epoch": 70.06586345381525, "grad_norm": 4.981110095977783, "learning_rate": 1.4938709677419354e-05, "loss": 0.473, "step": 10869 }, { "epoch": 70.07228915662651, "grad_norm": 9.915499687194824, "learning_rate": 1.4935483870967743e-05, "loss": 0.5214, "step": 10870 }, { "epoch": 70.07871485943775, "grad_norm": 13.782299995422363, "learning_rate": 1.493225806451613e-05, "loss": 0.2955, "step": 10871 }, { "epoch": 70.085140562249, "grad_norm": 18.023958206176758, "learning_rate": 1.4929032258064518e-05, "loss": 0.5221, "step": 10872 }, { "epoch": 70.09156626506024, "grad_norm": 4.836298942565918, "learning_rate": 1.4925806451612904e-05, "loss": 0.2065, "step": 10873 }, { "epoch": 70.09799196787148, "grad_norm": 8.734611511230469, "learning_rate": 1.4922580645161291e-05, "loss": 0.3498, "step": 10874 }, { "epoch": 70.10441767068274, "grad_norm": 17.223365783691406, "learning_rate": 1.4919354838709679e-05, "loss": 0.5199, "step": 10875 }, { "epoch": 70.11084337349398, "grad_norm": 11.100360870361328, "learning_rate": 1.4916129032258066e-05, "loss": 0.3699, "step": 10876 }, { "epoch": 70.11726907630522, "grad_norm": 14.61180591583252, "learning_rate": 1.4912903225806452e-05, "loss": 0.4816, "step": 10877 }, { "epoch": 70.12369477911646, "grad_norm": 18.75144386291504, "learning_rate": 1.4909677419354839e-05, "loss": 0.4923, "step": 10878 }, { "epoch": 70.1301204819277, "grad_norm": 23.213726043701172, "learning_rate": 1.4906451612903227e-05, "loss": 0.722, "step": 10879 }, { "epoch": 70.13654618473896, "grad_norm": 7.890746593475342, "learning_rate": 1.4903225806451612e-05, "loss": 0.3049, "step": 10880 }, { "epoch": 70.1429718875502, "grad_norm": 6.336200714111328, "learning_rate": 1.49e-05, "loss": 0.2785, "step": 10881 }, { "epoch": 70.14939759036145, "grad_norm": 6.452744960784912, "learning_rate": 1.4896774193548387e-05, "loss": 0.157, "step": 10882 }, { "epoch": 70.15582329317269, "grad_norm": 3.8633086681365967, "learning_rate": 1.4893548387096776e-05, "loss": 0.2155, "step": 10883 }, { "epoch": 70.16224899598393, "grad_norm": 8.247538566589355, "learning_rate": 1.489032258064516e-05, "loss": 0.3068, "step": 10884 }, { "epoch": 70.16867469879519, "grad_norm": 9.775053977966309, "learning_rate": 1.488709677419355e-05, "loss": 0.4473, "step": 10885 }, { "epoch": 70.17510040160643, "grad_norm": 16.631072998046875, "learning_rate": 1.4883870967741937e-05, "loss": 0.5309, "step": 10886 }, { "epoch": 70.18152610441767, "grad_norm": 8.09756088256836, "learning_rate": 1.4880645161290324e-05, "loss": 0.2835, "step": 10887 }, { "epoch": 70.18795180722891, "grad_norm": 6.906799793243408, "learning_rate": 1.487741935483871e-05, "loss": 0.1718, "step": 10888 }, { "epoch": 70.19437751004016, "grad_norm": 16.254623413085938, "learning_rate": 1.4874193548387097e-05, "loss": 0.2978, "step": 10889 }, { "epoch": 70.20080321285141, "grad_norm": 21.51219367980957, "learning_rate": 1.4870967741935485e-05, "loss": 0.417, "step": 10890 }, { "epoch": 70.20722891566265, "grad_norm": 12.325072288513184, "learning_rate": 1.4867741935483872e-05, "loss": 0.5739, "step": 10891 }, { "epoch": 70.2136546184739, "grad_norm": 7.419034957885742, "learning_rate": 1.4864516129032258e-05, "loss": 0.4109, "step": 10892 }, { "epoch": 70.22008032128514, "grad_norm": 13.04545783996582, "learning_rate": 1.4861290322580645e-05, "loss": 0.4595, "step": 10893 }, { "epoch": 70.22650602409638, "grad_norm": 8.254886627197266, "learning_rate": 1.4858064516129032e-05, "loss": 0.2129, "step": 10894 }, { "epoch": 70.23293172690764, "grad_norm": 7.214944839477539, "learning_rate": 1.4854838709677422e-05, "loss": 0.1882, "step": 10895 }, { "epoch": 70.23935742971888, "grad_norm": 8.895373344421387, "learning_rate": 1.4851612903225806e-05, "loss": 0.4338, "step": 10896 }, { "epoch": 70.24578313253012, "grad_norm": 7.346629619598389, "learning_rate": 1.4848387096774195e-05, "loss": 0.3392, "step": 10897 }, { "epoch": 70.25220883534136, "grad_norm": 9.5201997756958, "learning_rate": 1.4845161290322582e-05, "loss": 0.2141, "step": 10898 }, { "epoch": 70.2586345381526, "grad_norm": 13.904963493347168, "learning_rate": 1.484193548387097e-05, "loss": 0.571, "step": 10899 }, { "epoch": 70.26506024096386, "grad_norm": 15.208247184753418, "learning_rate": 1.4838709677419355e-05, "loss": 0.5933, "step": 10900 }, { "epoch": 70.2714859437751, "grad_norm": 15.873103141784668, "learning_rate": 1.4835483870967743e-05, "loss": 0.6, "step": 10901 }, { "epoch": 70.27791164658635, "grad_norm": 6.19387149810791, "learning_rate": 1.483225806451613e-05, "loss": 0.1572, "step": 10902 }, { "epoch": 70.28433734939759, "grad_norm": 19.402254104614258, "learning_rate": 1.4829032258064517e-05, "loss": 0.3054, "step": 10903 }, { "epoch": 70.29076305220883, "grad_norm": 6.442883014678955, "learning_rate": 1.4825806451612903e-05, "loss": 0.2738, "step": 10904 }, { "epoch": 70.29718875502007, "grad_norm": 22.63572883605957, "learning_rate": 1.482258064516129e-05, "loss": 0.4472, "step": 10905 }, { "epoch": 70.30361445783133, "grad_norm": 6.2055745124816895, "learning_rate": 1.4819354838709678e-05, "loss": 0.2907, "step": 10906 }, { "epoch": 70.31004016064257, "grad_norm": 29.26108169555664, "learning_rate": 1.4816129032258067e-05, "loss": 0.4802, "step": 10907 }, { "epoch": 70.31646586345381, "grad_norm": 5.224173545837402, "learning_rate": 1.4812903225806451e-05, "loss": 0.3987, "step": 10908 }, { "epoch": 70.32289156626506, "grad_norm": 4.991490364074707, "learning_rate": 1.480967741935484e-05, "loss": 0.2861, "step": 10909 }, { "epoch": 70.3293172690763, "grad_norm": 10.535131454467773, "learning_rate": 1.4806451612903227e-05, "loss": 0.4995, "step": 10910 }, { "epoch": 70.33574297188756, "grad_norm": 7.562111854553223, "learning_rate": 1.4803225806451615e-05, "loss": 0.3716, "step": 10911 }, { "epoch": 70.3421686746988, "grad_norm": 4.184596061706543, "learning_rate": 1.48e-05, "loss": 0.1808, "step": 10912 }, { "epoch": 70.34859437751004, "grad_norm": 8.111595153808594, "learning_rate": 1.4796774193548388e-05, "loss": 0.3777, "step": 10913 }, { "epoch": 70.35502008032128, "grad_norm": 6.250491142272949, "learning_rate": 1.4793548387096775e-05, "loss": 0.3713, "step": 10914 }, { "epoch": 70.36144578313252, "grad_norm": 7.074757099151611, "learning_rate": 1.4790322580645161e-05, "loss": 0.2833, "step": 10915 }, { "epoch": 70.36787148594378, "grad_norm": 10.245893478393555, "learning_rate": 1.4787096774193549e-05, "loss": 0.3268, "step": 10916 }, { "epoch": 70.37429718875502, "grad_norm": 5.50026273727417, "learning_rate": 1.4783870967741936e-05, "loss": 0.2348, "step": 10917 }, { "epoch": 70.38072289156626, "grad_norm": 5.011965274810791, "learning_rate": 1.4780645161290323e-05, "loss": 0.2299, "step": 10918 }, { "epoch": 70.3871485943775, "grad_norm": 11.91970443725586, "learning_rate": 1.4777419354838709e-05, "loss": 0.6042, "step": 10919 }, { "epoch": 70.39357429718875, "grad_norm": 8.520805358886719, "learning_rate": 1.4774193548387096e-05, "loss": 0.2843, "step": 10920 }, { "epoch": 70.4, "grad_norm": 3.376079797744751, "learning_rate": 1.4770967741935486e-05, "loss": 0.1045, "step": 10921 }, { "epoch": 70.40642570281125, "grad_norm": 11.856110572814941, "learning_rate": 1.4767741935483873e-05, "loss": 0.2305, "step": 10922 }, { "epoch": 70.41285140562249, "grad_norm": 7.104299545288086, "learning_rate": 1.4764516129032257e-05, "loss": 0.3432, "step": 10923 }, { "epoch": 70.41927710843373, "grad_norm": 8.879840850830078, "learning_rate": 1.4761290322580646e-05, "loss": 0.2106, "step": 10924 }, { "epoch": 70.42570281124497, "grad_norm": 14.910285949707031, "learning_rate": 1.4758064516129033e-05, "loss": 0.1293, "step": 10925 }, { "epoch": 70.43212851405623, "grad_norm": 7.790024757385254, "learning_rate": 1.475483870967742e-05, "loss": 0.3269, "step": 10926 }, { "epoch": 70.43855421686747, "grad_norm": 8.04168701171875, "learning_rate": 1.4751612903225807e-05, "loss": 0.2776, "step": 10927 }, { "epoch": 70.44497991967872, "grad_norm": 10.154553413391113, "learning_rate": 1.4748387096774194e-05, "loss": 0.2771, "step": 10928 }, { "epoch": 70.45140562248996, "grad_norm": 5.641875267028809, "learning_rate": 1.4745161290322581e-05, "loss": 0.2838, "step": 10929 }, { "epoch": 70.4578313253012, "grad_norm": 9.055374145507812, "learning_rate": 1.4741935483870969e-05, "loss": 0.5009, "step": 10930 }, { "epoch": 70.46425702811246, "grad_norm": 6.172160625457764, "learning_rate": 1.4738709677419354e-05, "loss": 0.3088, "step": 10931 }, { "epoch": 70.4706827309237, "grad_norm": 6.901937484741211, "learning_rate": 1.4735483870967742e-05, "loss": 0.4795, "step": 10932 }, { "epoch": 70.47710843373494, "grad_norm": 9.789116859436035, "learning_rate": 1.4732258064516131e-05, "loss": 0.1966, "step": 10933 }, { "epoch": 70.48353413654618, "grad_norm": 11.529409408569336, "learning_rate": 1.4729032258064518e-05, "loss": 0.244, "step": 10934 }, { "epoch": 70.48995983935743, "grad_norm": 19.730709075927734, "learning_rate": 1.4725806451612902e-05, "loss": 0.4737, "step": 10935 }, { "epoch": 70.49638554216868, "grad_norm": 17.457412719726562, "learning_rate": 1.4722580645161291e-05, "loss": 0.4268, "step": 10936 }, { "epoch": 70.50281124497992, "grad_norm": 16.662309646606445, "learning_rate": 1.4719354838709679e-05, "loss": 0.2708, "step": 10937 }, { "epoch": 70.50923694779117, "grad_norm": 6.666028022766113, "learning_rate": 1.4716129032258066e-05, "loss": 0.214, "step": 10938 }, { "epoch": 70.51566265060241, "grad_norm": 4.797404766082764, "learning_rate": 1.4712903225806452e-05, "loss": 0.2064, "step": 10939 }, { "epoch": 70.52208835341365, "grad_norm": 7.071338176727295, "learning_rate": 1.470967741935484e-05, "loss": 0.2656, "step": 10940 }, { "epoch": 70.52851405622489, "grad_norm": 12.279313087463379, "learning_rate": 1.4706451612903227e-05, "loss": 0.2592, "step": 10941 }, { "epoch": 70.53493975903615, "grad_norm": 14.592275619506836, "learning_rate": 1.4703225806451614e-05, "loss": 0.2775, "step": 10942 }, { "epoch": 70.54136546184739, "grad_norm": 14.299819946289062, "learning_rate": 1.47e-05, "loss": 0.4313, "step": 10943 }, { "epoch": 70.54779116465863, "grad_norm": 17.474538803100586, "learning_rate": 1.4696774193548387e-05, "loss": 0.305, "step": 10944 }, { "epoch": 70.55421686746988, "grad_norm": 5.811059474945068, "learning_rate": 1.4693548387096775e-05, "loss": 0.1884, "step": 10945 }, { "epoch": 70.56064257028112, "grad_norm": 22.059219360351562, "learning_rate": 1.4690322580645164e-05, "loss": 0.2154, "step": 10946 }, { "epoch": 70.56706827309237, "grad_norm": 3.159494161605835, "learning_rate": 1.4687096774193548e-05, "loss": 0.0539, "step": 10947 }, { "epoch": 70.57349397590362, "grad_norm": 3.5564522743225098, "learning_rate": 1.4683870967741937e-05, "loss": 0.0913, "step": 10948 }, { "epoch": 70.57991967871486, "grad_norm": 13.46561050415039, "learning_rate": 1.4680645161290324e-05, "loss": 0.571, "step": 10949 }, { "epoch": 70.5863453815261, "grad_norm": 8.463027954101562, "learning_rate": 1.4677419354838708e-05, "loss": 0.2978, "step": 10950 }, { "epoch": 70.59277108433734, "grad_norm": 17.826030731201172, "learning_rate": 1.4674193548387097e-05, "loss": 0.4626, "step": 10951 }, { "epoch": 70.5991967871486, "grad_norm": 22.761615753173828, "learning_rate": 1.4670967741935485e-05, "loss": 0.4892, "step": 10952 }, { "epoch": 70.60562248995984, "grad_norm": 8.105454444885254, "learning_rate": 1.4667741935483872e-05, "loss": 0.45, "step": 10953 }, { "epoch": 70.61204819277108, "grad_norm": 39.44690704345703, "learning_rate": 1.4664516129032258e-05, "loss": 0.1742, "step": 10954 }, { "epoch": 70.61847389558233, "grad_norm": 10.17225170135498, "learning_rate": 1.4661290322580645e-05, "loss": 0.3993, "step": 10955 }, { "epoch": 70.62489959839357, "grad_norm": 13.636817932128906, "learning_rate": 1.4658064516129033e-05, "loss": 0.5978, "step": 10956 }, { "epoch": 70.63132530120482, "grad_norm": 8.38656234741211, "learning_rate": 1.465483870967742e-05, "loss": 0.2386, "step": 10957 }, { "epoch": 70.63775100401607, "grad_norm": 7.679324626922607, "learning_rate": 1.4651612903225806e-05, "loss": 0.1315, "step": 10958 }, { "epoch": 70.64417670682731, "grad_norm": 6.663963317871094, "learning_rate": 1.4648387096774193e-05, "loss": 0.2345, "step": 10959 }, { "epoch": 70.65060240963855, "grad_norm": 22.231826782226562, "learning_rate": 1.4645161290322582e-05, "loss": 0.4123, "step": 10960 }, { "epoch": 70.6570281124498, "grad_norm": 11.233016967773438, "learning_rate": 1.464193548387097e-05, "loss": 0.3532, "step": 10961 }, { "epoch": 70.66345381526105, "grad_norm": 9.81601333618164, "learning_rate": 1.4638709677419354e-05, "loss": 0.0753, "step": 10962 }, { "epoch": 70.66987951807229, "grad_norm": 11.310175895690918, "learning_rate": 1.4635483870967743e-05, "loss": 0.4494, "step": 10963 }, { "epoch": 70.67630522088353, "grad_norm": 6.469303607940674, "learning_rate": 1.463225806451613e-05, "loss": 0.3153, "step": 10964 }, { "epoch": 70.68273092369478, "grad_norm": 23.348318099975586, "learning_rate": 1.4629032258064518e-05, "loss": 0.3226, "step": 10965 }, { "epoch": 70.68915662650602, "grad_norm": 11.995959281921387, "learning_rate": 1.4625806451612903e-05, "loss": 0.3247, "step": 10966 }, { "epoch": 70.69558232931728, "grad_norm": 9.243979454040527, "learning_rate": 1.462258064516129e-05, "loss": 0.5891, "step": 10967 }, { "epoch": 70.70200803212852, "grad_norm": 3.784895181655884, "learning_rate": 1.4619354838709678e-05, "loss": 0.1357, "step": 10968 }, { "epoch": 70.70843373493976, "grad_norm": 9.341614723205566, "learning_rate": 1.4616129032258066e-05, "loss": 0.2491, "step": 10969 }, { "epoch": 70.714859437751, "grad_norm": 11.213520050048828, "learning_rate": 1.4612903225806451e-05, "loss": 0.5273, "step": 10970 }, { "epoch": 70.72128514056224, "grad_norm": 4.6353983879089355, "learning_rate": 1.4609677419354839e-05, "loss": 0.1284, "step": 10971 }, { "epoch": 70.7277108433735, "grad_norm": 11.682771682739258, "learning_rate": 1.4606451612903228e-05, "loss": 0.3798, "step": 10972 }, { "epoch": 70.73413654618474, "grad_norm": 8.571578025817871, "learning_rate": 1.4603225806451615e-05, "loss": 0.3731, "step": 10973 }, { "epoch": 70.74056224899599, "grad_norm": 9.112040519714355, "learning_rate": 1.4599999999999999e-05, "loss": 0.4525, "step": 10974 }, { "epoch": 70.74698795180723, "grad_norm": 18.148237228393555, "learning_rate": 1.4596774193548388e-05, "loss": 0.5416, "step": 10975 }, { "epoch": 70.75341365461847, "grad_norm": 7.300929546356201, "learning_rate": 1.4593548387096776e-05, "loss": 0.3557, "step": 10976 }, { "epoch": 70.75983935742971, "grad_norm": 7.367361068725586, "learning_rate": 1.4590322580645163e-05, "loss": 0.2267, "step": 10977 }, { "epoch": 70.76626506024097, "grad_norm": 10.914345741271973, "learning_rate": 1.4587096774193549e-05, "loss": 0.5737, "step": 10978 }, { "epoch": 70.77269076305221, "grad_norm": 4.2912116050720215, "learning_rate": 1.4583870967741936e-05, "loss": 0.206, "step": 10979 }, { "epoch": 70.77911646586345, "grad_norm": 15.976140022277832, "learning_rate": 1.4580645161290324e-05, "loss": 0.3036, "step": 10980 }, { "epoch": 70.7855421686747, "grad_norm": 15.232151985168457, "learning_rate": 1.4577419354838711e-05, "loss": 0.3783, "step": 10981 }, { "epoch": 70.79196787148594, "grad_norm": 7.733218669891357, "learning_rate": 1.4574193548387097e-05, "loss": 0.4765, "step": 10982 }, { "epoch": 70.7983935742972, "grad_norm": 6.720606327056885, "learning_rate": 1.4570967741935484e-05, "loss": 0.2115, "step": 10983 }, { "epoch": 70.80481927710844, "grad_norm": 7.682621955871582, "learning_rate": 1.4567741935483873e-05, "loss": 0.2726, "step": 10984 }, { "epoch": 70.81124497991968, "grad_norm": 6.01506233215332, "learning_rate": 1.456451612903226e-05, "loss": 0.5284, "step": 10985 }, { "epoch": 70.81767068273092, "grad_norm": 6.366855621337891, "learning_rate": 1.4561290322580645e-05, "loss": 0.1336, "step": 10986 }, { "epoch": 70.82409638554216, "grad_norm": 36.59817886352539, "learning_rate": 1.4558064516129034e-05, "loss": 0.617, "step": 10987 }, { "epoch": 70.83052208835342, "grad_norm": 11.08757495880127, "learning_rate": 1.4554838709677421e-05, "loss": 0.3087, "step": 10988 }, { "epoch": 70.83694779116466, "grad_norm": 6.314168453216553, "learning_rate": 1.4551612903225805e-05, "loss": 0.1709, "step": 10989 }, { "epoch": 70.8433734939759, "grad_norm": 17.861787796020508, "learning_rate": 1.4548387096774194e-05, "loss": 0.2295, "step": 10990 }, { "epoch": 70.84979919678715, "grad_norm": 10.108424186706543, "learning_rate": 1.4545161290322582e-05, "loss": 0.3212, "step": 10991 }, { "epoch": 70.85622489959839, "grad_norm": 8.029388427734375, "learning_rate": 1.4541935483870969e-05, "loss": 0.3055, "step": 10992 }, { "epoch": 70.86265060240964, "grad_norm": 12.843503952026367, "learning_rate": 1.4538709677419355e-05, "loss": 0.4012, "step": 10993 }, { "epoch": 70.86907630522089, "grad_norm": 7.545074462890625, "learning_rate": 1.4535483870967742e-05, "loss": 0.351, "step": 10994 }, { "epoch": 70.87550200803213, "grad_norm": 13.22414493560791, "learning_rate": 1.453225806451613e-05, "loss": 0.6632, "step": 10995 }, { "epoch": 70.88192771084337, "grad_norm": 18.354400634765625, "learning_rate": 1.4529032258064519e-05, "loss": 0.5419, "step": 10996 }, { "epoch": 70.88835341365461, "grad_norm": 5.759202480316162, "learning_rate": 1.4525806451612903e-05, "loss": 0.2624, "step": 10997 }, { "epoch": 70.89477911646587, "grad_norm": 9.954826354980469, "learning_rate": 1.452258064516129e-05, "loss": 0.3957, "step": 10998 }, { "epoch": 70.90120481927711, "grad_norm": 11.832571983337402, "learning_rate": 1.4519354838709679e-05, "loss": 0.452, "step": 10999 }, { "epoch": 70.90763052208835, "grad_norm": 8.954046249389648, "learning_rate": 1.4516129032258066e-05, "loss": 0.3272, "step": 11000 }, { "epoch": 70.9140562248996, "grad_norm": 17.78217124938965, "learning_rate": 1.451290322580645e-05, "loss": 0.4588, "step": 11001 }, { "epoch": 70.92048192771084, "grad_norm": 7.798130512237549, "learning_rate": 1.450967741935484e-05, "loss": 0.3272, "step": 11002 }, { "epoch": 70.9269076305221, "grad_norm": 40.548072814941406, "learning_rate": 1.4506451612903227e-05, "loss": 0.1744, "step": 11003 }, { "epoch": 70.93333333333334, "grad_norm": 29.54244613647461, "learning_rate": 1.4503225806451614e-05, "loss": 0.2823, "step": 11004 }, { "epoch": 70.93975903614458, "grad_norm": 3.5287673473358154, "learning_rate": 1.45e-05, "loss": 0.1389, "step": 11005 }, { "epoch": 70.94618473895582, "grad_norm": 8.160722732543945, "learning_rate": 1.4496774193548387e-05, "loss": 0.42, "step": 11006 }, { "epoch": 70.95261044176706, "grad_norm": 29.435495376586914, "learning_rate": 1.4493548387096775e-05, "loss": 0.3915, "step": 11007 }, { "epoch": 70.95903614457832, "grad_norm": 13.414846420288086, "learning_rate": 1.4490322580645162e-05, "loss": 0.2546, "step": 11008 }, { "epoch": 70.96546184738956, "grad_norm": 9.70048999786377, "learning_rate": 1.4487096774193548e-05, "loss": 0.474, "step": 11009 }, { "epoch": 70.9718875502008, "grad_norm": 9.392085075378418, "learning_rate": 1.4483870967741935e-05, "loss": 0.3125, "step": 11010 }, { "epoch": 70.97831325301205, "grad_norm": 8.425556182861328, "learning_rate": 1.4480645161290324e-05, "loss": 0.1912, "step": 11011 }, { "epoch": 70.98473895582329, "grad_norm": 6.183394432067871, "learning_rate": 1.4477419354838712e-05, "loss": 0.2991, "step": 11012 }, { "epoch": 70.99116465863453, "grad_norm": 10.173599243164062, "learning_rate": 1.4474193548387096e-05, "loss": 0.3432, "step": 11013 }, { "epoch": 71.00080321285141, "grad_norm": 4.484566688537598, "learning_rate": 1.4470967741935485e-05, "loss": 0.1833, "step": 11014 }, { "epoch": 71.00722891566265, "grad_norm": 4.231504440307617, "learning_rate": 1.4467741935483872e-05, "loss": 0.1667, "step": 11015 }, { "epoch": 71.0136546184739, "grad_norm": 7.684971332550049, "learning_rate": 1.446451612903226e-05, "loss": 0.3855, "step": 11016 }, { "epoch": 71.02008032128514, "grad_norm": 10.918825149536133, "learning_rate": 1.4461290322580646e-05, "loss": 0.4468, "step": 11017 }, { "epoch": 71.02650602409639, "grad_norm": 56.55377197265625, "learning_rate": 1.4458064516129033e-05, "loss": 0.2637, "step": 11018 }, { "epoch": 71.03293172690763, "grad_norm": 8.748237609863281, "learning_rate": 1.445483870967742e-05, "loss": 0.4503, "step": 11019 }, { "epoch": 71.03935742971888, "grad_norm": 23.17003631591797, "learning_rate": 1.4451612903225808e-05, "loss": 0.2962, "step": 11020 }, { "epoch": 71.04578313253012, "grad_norm": 10.085155487060547, "learning_rate": 1.4448387096774193e-05, "loss": 0.3794, "step": 11021 }, { "epoch": 71.05220883534136, "grad_norm": 6.946505546569824, "learning_rate": 1.444516129032258e-05, "loss": 0.5378, "step": 11022 }, { "epoch": 71.05863453815262, "grad_norm": 3.2092995643615723, "learning_rate": 1.444193548387097e-05, "loss": 0.114, "step": 11023 }, { "epoch": 71.06506024096386, "grad_norm": 5.292834281921387, "learning_rate": 1.4438709677419354e-05, "loss": 0.3174, "step": 11024 }, { "epoch": 71.0714859437751, "grad_norm": 2.099088668823242, "learning_rate": 1.4435483870967741e-05, "loss": 0.0559, "step": 11025 }, { "epoch": 71.07791164658634, "grad_norm": 6.477895259857178, "learning_rate": 1.443225806451613e-05, "loss": 0.1567, "step": 11026 }, { "epoch": 71.08433734939759, "grad_norm": 16.844438552856445, "learning_rate": 1.4429032258064518e-05, "loss": 0.45, "step": 11027 }, { "epoch": 71.09076305220883, "grad_norm": 12.006876945495605, "learning_rate": 1.4425806451612902e-05, "loss": 0.3345, "step": 11028 }, { "epoch": 71.09718875502008, "grad_norm": 13.380739212036133, "learning_rate": 1.4422580645161291e-05, "loss": 0.2197, "step": 11029 }, { "epoch": 71.10361445783133, "grad_norm": 6.854885101318359, "learning_rate": 1.4419354838709678e-05, "loss": 0.2077, "step": 11030 }, { "epoch": 71.11004016064257, "grad_norm": 3.775723934173584, "learning_rate": 1.4416129032258066e-05, "loss": 0.1584, "step": 11031 }, { "epoch": 71.11646586345381, "grad_norm": 5.445743083953857, "learning_rate": 1.4412903225806451e-05, "loss": 0.2379, "step": 11032 }, { "epoch": 71.12289156626505, "grad_norm": 5.19208288192749, "learning_rate": 1.4409677419354839e-05, "loss": 0.1426, "step": 11033 }, { "epoch": 71.12931726907631, "grad_norm": 9.816346168518066, "learning_rate": 1.4406451612903226e-05, "loss": 0.358, "step": 11034 }, { "epoch": 71.13574297188755, "grad_norm": 3.9942524433135986, "learning_rate": 1.4403225806451615e-05, "loss": 0.2386, "step": 11035 }, { "epoch": 71.1421686746988, "grad_norm": 4.763792991638184, "learning_rate": 1.44e-05, "loss": 0.2803, "step": 11036 }, { "epoch": 71.14859437751004, "grad_norm": 9.398032188415527, "learning_rate": 1.4396774193548387e-05, "loss": 0.248, "step": 11037 }, { "epoch": 71.15502008032128, "grad_norm": 20.57535171508789, "learning_rate": 1.4393548387096776e-05, "loss": 0.4553, "step": 11038 }, { "epoch": 71.16144578313254, "grad_norm": 11.801006317138672, "learning_rate": 1.4390322580645163e-05, "loss": 0.2532, "step": 11039 }, { "epoch": 71.16787148594378, "grad_norm": 8.678600311279297, "learning_rate": 1.4387096774193547e-05, "loss": 0.3046, "step": 11040 }, { "epoch": 71.17429718875502, "grad_norm": 6.536129474639893, "learning_rate": 1.4383870967741936e-05, "loss": 0.1823, "step": 11041 }, { "epoch": 71.18072289156626, "grad_norm": 6.288715362548828, "learning_rate": 1.4380645161290324e-05, "loss": 0.2955, "step": 11042 }, { "epoch": 71.1871485943775, "grad_norm": 18.19637680053711, "learning_rate": 1.4377419354838711e-05, "loss": 0.257, "step": 11043 }, { "epoch": 71.19357429718876, "grad_norm": 5.6013994216918945, "learning_rate": 1.4374193548387097e-05, "loss": 0.3465, "step": 11044 }, { "epoch": 71.2, "grad_norm": 16.820369720458984, "learning_rate": 1.4370967741935484e-05, "loss": 0.3965, "step": 11045 }, { "epoch": 71.20642570281125, "grad_norm": 7.922639846801758, "learning_rate": 1.4367741935483872e-05, "loss": 0.2649, "step": 11046 }, { "epoch": 71.21285140562249, "grad_norm": 8.660283088684082, "learning_rate": 1.436451612903226e-05, "loss": 0.4147, "step": 11047 }, { "epoch": 71.21927710843373, "grad_norm": 10.495837211608887, "learning_rate": 1.4361290322580645e-05, "loss": 0.2714, "step": 11048 }, { "epoch": 71.22570281124499, "grad_norm": 7.213881015777588, "learning_rate": 1.4358064516129032e-05, "loss": 0.2126, "step": 11049 }, { "epoch": 71.23212851405623, "grad_norm": 11.03390884399414, "learning_rate": 1.4354838709677421e-05, "loss": 0.2964, "step": 11050 }, { "epoch": 71.23855421686747, "grad_norm": 3.638113021850586, "learning_rate": 1.4351612903225809e-05, "loss": 0.2643, "step": 11051 }, { "epoch": 71.24497991967871, "grad_norm": 22.19644546508789, "learning_rate": 1.4348387096774193e-05, "loss": 0.3173, "step": 11052 }, { "epoch": 71.25140562248995, "grad_norm": 11.081832885742188, "learning_rate": 1.4345161290322582e-05, "loss": 0.4651, "step": 11053 }, { "epoch": 71.25783132530121, "grad_norm": 6.449596881866455, "learning_rate": 1.434193548387097e-05, "loss": 0.2897, "step": 11054 }, { "epoch": 71.26425702811245, "grad_norm": 6.449328899383545, "learning_rate": 1.4338709677419357e-05, "loss": 0.2456, "step": 11055 }, { "epoch": 71.2706827309237, "grad_norm": 7.912364959716797, "learning_rate": 1.4335483870967742e-05, "loss": 0.4917, "step": 11056 }, { "epoch": 71.27710843373494, "grad_norm": 48.65501022338867, "learning_rate": 1.433225806451613e-05, "loss": 0.5039, "step": 11057 }, { "epoch": 71.28353413654618, "grad_norm": 30.2677059173584, "learning_rate": 1.4329032258064517e-05, "loss": 0.4875, "step": 11058 }, { "epoch": 71.28995983935744, "grad_norm": 27.696531295776367, "learning_rate": 1.4325806451612903e-05, "loss": 0.3785, "step": 11059 }, { "epoch": 71.29638554216868, "grad_norm": 7.09128999710083, "learning_rate": 1.432258064516129e-05, "loss": 0.3878, "step": 11060 }, { "epoch": 71.30281124497992, "grad_norm": 10.276581764221191, "learning_rate": 1.4319354838709678e-05, "loss": 0.5321, "step": 11061 }, { "epoch": 71.30923694779116, "grad_norm": 11.588987350463867, "learning_rate": 1.4316129032258067e-05, "loss": 0.3804, "step": 11062 }, { "epoch": 71.3156626506024, "grad_norm": 9.957548141479492, "learning_rate": 1.431290322580645e-05, "loss": 0.4672, "step": 11063 }, { "epoch": 71.32208835341365, "grad_norm": 10.235323905944824, "learning_rate": 1.4309677419354838e-05, "loss": 0.392, "step": 11064 }, { "epoch": 71.3285140562249, "grad_norm": 6.029904365539551, "learning_rate": 1.4306451612903227e-05, "loss": 0.2044, "step": 11065 }, { "epoch": 71.33493975903615, "grad_norm": 11.860097885131836, "learning_rate": 1.4303225806451615e-05, "loss": 0.1284, "step": 11066 }, { "epoch": 71.34136546184739, "grad_norm": 9.715838432312012, "learning_rate": 1.43e-05, "loss": 0.4113, "step": 11067 }, { "epoch": 71.34779116465863, "grad_norm": 23.721115112304688, "learning_rate": 1.4296774193548388e-05, "loss": 0.541, "step": 11068 }, { "epoch": 71.35421686746987, "grad_norm": 22.30900001525879, "learning_rate": 1.4293548387096775e-05, "loss": 0.6629, "step": 11069 }, { "epoch": 71.36064257028113, "grad_norm": 3.9875950813293457, "learning_rate": 1.4290322580645163e-05, "loss": 0.1619, "step": 11070 }, { "epoch": 71.36706827309237, "grad_norm": 8.287738800048828, "learning_rate": 1.4287096774193548e-05, "loss": 0.3254, "step": 11071 }, { "epoch": 71.37349397590361, "grad_norm": 20.454259872436523, "learning_rate": 1.4283870967741936e-05, "loss": 0.7085, "step": 11072 }, { "epoch": 71.37991967871486, "grad_norm": 10.19217300415039, "learning_rate": 1.4280645161290323e-05, "loss": 0.496, "step": 11073 }, { "epoch": 71.3863453815261, "grad_norm": 6.727989196777344, "learning_rate": 1.4277419354838712e-05, "loss": 0.3616, "step": 11074 }, { "epoch": 71.39277108433735, "grad_norm": 5.007495880126953, "learning_rate": 1.4274193548387096e-05, "loss": 0.1903, "step": 11075 }, { "epoch": 71.3991967871486, "grad_norm": 7.91377592086792, "learning_rate": 1.4270967741935484e-05, "loss": 0.386, "step": 11076 }, { "epoch": 71.40562248995984, "grad_norm": 14.098525047302246, "learning_rate": 1.4267741935483873e-05, "loss": 0.2671, "step": 11077 }, { "epoch": 71.41204819277108, "grad_norm": 8.179404258728027, "learning_rate": 1.426451612903226e-05, "loss": 0.4411, "step": 11078 }, { "epoch": 71.41847389558232, "grad_norm": 24.944496154785156, "learning_rate": 1.4261290322580646e-05, "loss": 0.342, "step": 11079 }, { "epoch": 71.42489959839358, "grad_norm": 5.6230974197387695, "learning_rate": 1.4258064516129033e-05, "loss": 0.2807, "step": 11080 }, { "epoch": 71.43132530120482, "grad_norm": 9.053861618041992, "learning_rate": 1.425483870967742e-05, "loss": 0.2938, "step": 11081 }, { "epoch": 71.43775100401606, "grad_norm": 12.357953071594238, "learning_rate": 1.4251612903225808e-05, "loss": 0.5166, "step": 11082 }, { "epoch": 71.4441767068273, "grad_norm": 8.977677345275879, "learning_rate": 1.4248387096774194e-05, "loss": 0.2942, "step": 11083 }, { "epoch": 71.45060240963855, "grad_norm": 7.596463680267334, "learning_rate": 1.4245161290322581e-05, "loss": 0.3635, "step": 11084 }, { "epoch": 71.4570281124498, "grad_norm": 11.390877723693848, "learning_rate": 1.4241935483870968e-05, "loss": 0.3602, "step": 11085 }, { "epoch": 71.46345381526105, "grad_norm": 5.406828880310059, "learning_rate": 1.4238709677419358e-05, "loss": 0.3418, "step": 11086 }, { "epoch": 71.46987951807229, "grad_norm": 7.114943027496338, "learning_rate": 1.4235483870967742e-05, "loss": 0.4162, "step": 11087 }, { "epoch": 71.47630522088353, "grad_norm": 19.29792022705078, "learning_rate": 1.4232258064516129e-05, "loss": 0.1961, "step": 11088 }, { "epoch": 71.48273092369477, "grad_norm": 6.969693183898926, "learning_rate": 1.4229032258064518e-05, "loss": 0.3381, "step": 11089 }, { "epoch": 71.48915662650603, "grad_norm": 8.042298316955566, "learning_rate": 1.4225806451612905e-05, "loss": 0.2679, "step": 11090 }, { "epoch": 71.49558232931727, "grad_norm": 21.305143356323242, "learning_rate": 1.422258064516129e-05, "loss": 0.2495, "step": 11091 }, { "epoch": 71.50200803212851, "grad_norm": 10.388924598693848, "learning_rate": 1.4219354838709679e-05, "loss": 0.1422, "step": 11092 }, { "epoch": 71.50843373493976, "grad_norm": 7.5247273445129395, "learning_rate": 1.4216129032258066e-05, "loss": 0.1148, "step": 11093 }, { "epoch": 71.514859437751, "grad_norm": 12.502607345581055, "learning_rate": 1.4212903225806452e-05, "loss": 0.5467, "step": 11094 }, { "epoch": 71.52128514056226, "grad_norm": 15.9971284866333, "learning_rate": 1.4209677419354839e-05, "loss": 0.4016, "step": 11095 }, { "epoch": 71.5277108433735, "grad_norm": 13.063100814819336, "learning_rate": 1.4206451612903226e-05, "loss": 0.3484, "step": 11096 }, { "epoch": 71.53413654618474, "grad_norm": 6.270208835601807, "learning_rate": 1.4203225806451614e-05, "loss": 0.271, "step": 11097 }, { "epoch": 71.54056224899598, "grad_norm": 11.219042778015137, "learning_rate": 1.42e-05, "loss": 0.3865, "step": 11098 }, { "epoch": 71.54698795180722, "grad_norm": 9.480997085571289, "learning_rate": 1.4196774193548387e-05, "loss": 0.2243, "step": 11099 }, { "epoch": 71.55341365461847, "grad_norm": 6.4439826011657715, "learning_rate": 1.4193548387096774e-05, "loss": 0.2225, "step": 11100 }, { "epoch": 71.55983935742972, "grad_norm": 14.571901321411133, "learning_rate": 1.4190322580645163e-05, "loss": 0.7471, "step": 11101 }, { "epoch": 71.56626506024097, "grad_norm": 27.068073272705078, "learning_rate": 1.4187096774193547e-05, "loss": 0.3174, "step": 11102 }, { "epoch": 71.57269076305221, "grad_norm": 28.571081161499023, "learning_rate": 1.4183870967741935e-05, "loss": 0.3041, "step": 11103 }, { "epoch": 71.57911646586345, "grad_norm": 11.372547149658203, "learning_rate": 1.4180645161290324e-05, "loss": 0.375, "step": 11104 }, { "epoch": 71.58554216867469, "grad_norm": 6.084019184112549, "learning_rate": 1.4177419354838711e-05, "loss": 0.3196, "step": 11105 }, { "epoch": 71.59196787148595, "grad_norm": 10.452221870422363, "learning_rate": 1.4174193548387097e-05, "loss": 0.2688, "step": 11106 }, { "epoch": 71.59839357429719, "grad_norm": 3.961237668991089, "learning_rate": 1.4170967741935484e-05, "loss": 0.1956, "step": 11107 }, { "epoch": 71.60481927710843, "grad_norm": 8.090140342712402, "learning_rate": 1.4167741935483872e-05, "loss": 0.4232, "step": 11108 }, { "epoch": 71.61124497991968, "grad_norm": 11.39536190032959, "learning_rate": 1.416451612903226e-05, "loss": 0.5409, "step": 11109 }, { "epoch": 71.61767068273092, "grad_norm": 7.730048656463623, "learning_rate": 1.4161290322580645e-05, "loss": 0.1426, "step": 11110 }, { "epoch": 71.62409638554217, "grad_norm": 11.48516845703125, "learning_rate": 1.4158064516129032e-05, "loss": 0.6373, "step": 11111 }, { "epoch": 71.63052208835342, "grad_norm": 19.413928985595703, "learning_rate": 1.415483870967742e-05, "loss": 0.3163, "step": 11112 }, { "epoch": 71.63694779116466, "grad_norm": 8.461963653564453, "learning_rate": 1.4151612903225809e-05, "loss": 0.5697, "step": 11113 }, { "epoch": 71.6433734939759, "grad_norm": 11.096430778503418, "learning_rate": 1.4148387096774193e-05, "loss": 0.2775, "step": 11114 }, { "epoch": 71.64979919678714, "grad_norm": 13.956893920898438, "learning_rate": 1.414516129032258e-05, "loss": 0.4194, "step": 11115 }, { "epoch": 71.6562248995984, "grad_norm": 15.063150405883789, "learning_rate": 1.414193548387097e-05, "loss": 0.4037, "step": 11116 }, { "epoch": 71.66265060240964, "grad_norm": 11.942245483398438, "learning_rate": 1.4138709677419357e-05, "loss": 0.2824, "step": 11117 }, { "epoch": 71.66907630522088, "grad_norm": 9.555983543395996, "learning_rate": 1.4135483870967743e-05, "loss": 0.2659, "step": 11118 }, { "epoch": 71.67550200803213, "grad_norm": 9.804084777832031, "learning_rate": 1.413225806451613e-05, "loss": 0.2444, "step": 11119 }, { "epoch": 71.68192771084337, "grad_norm": 4.715213298797607, "learning_rate": 1.4129032258064517e-05, "loss": 0.1054, "step": 11120 }, { "epoch": 71.68835341365462, "grad_norm": 12.288783073425293, "learning_rate": 1.4125806451612905e-05, "loss": 0.3218, "step": 11121 }, { "epoch": 71.69477911646587, "grad_norm": 8.42742919921875, "learning_rate": 1.412258064516129e-05, "loss": 0.1942, "step": 11122 }, { "epoch": 71.70120481927711, "grad_norm": 9.080729484558105, "learning_rate": 1.4119354838709678e-05, "loss": 0.2493, "step": 11123 }, { "epoch": 71.70763052208835, "grad_norm": 23.267807006835938, "learning_rate": 1.4116129032258065e-05, "loss": 0.4036, "step": 11124 }, { "epoch": 71.7140562248996, "grad_norm": 18.21878433227539, "learning_rate": 1.4112903225806454e-05, "loss": 0.5172, "step": 11125 }, { "epoch": 71.72048192771085, "grad_norm": 7.248779773712158, "learning_rate": 1.4109677419354838e-05, "loss": 0.2999, "step": 11126 }, { "epoch": 71.72690763052209, "grad_norm": 10.509697914123535, "learning_rate": 1.4106451612903226e-05, "loss": 0.3501, "step": 11127 }, { "epoch": 71.73333333333333, "grad_norm": 12.285347938537598, "learning_rate": 1.4103225806451615e-05, "loss": 0.4676, "step": 11128 }, { "epoch": 71.73975903614458, "grad_norm": 31.475086212158203, "learning_rate": 1.4099999999999999e-05, "loss": 0.4493, "step": 11129 }, { "epoch": 71.74618473895582, "grad_norm": 12.138216972351074, "learning_rate": 1.4096774193548388e-05, "loss": 0.3574, "step": 11130 }, { "epoch": 71.75261044176708, "grad_norm": 8.093310356140137, "learning_rate": 1.4093548387096775e-05, "loss": 0.2169, "step": 11131 }, { "epoch": 71.75903614457832, "grad_norm": 9.699089050292969, "learning_rate": 1.4090322580645163e-05, "loss": 0.2948, "step": 11132 }, { "epoch": 71.76546184738956, "grad_norm": 15.824109077453613, "learning_rate": 1.4087096774193548e-05, "loss": 0.7029, "step": 11133 }, { "epoch": 71.7718875502008, "grad_norm": 50.35569381713867, "learning_rate": 1.4083870967741936e-05, "loss": 0.7055, "step": 11134 }, { "epoch": 71.77831325301204, "grad_norm": 29.93772315979004, "learning_rate": 1.4080645161290323e-05, "loss": 0.4236, "step": 11135 }, { "epoch": 71.78473895582329, "grad_norm": 10.698310852050781, "learning_rate": 1.407741935483871e-05, "loss": 0.3462, "step": 11136 }, { "epoch": 71.79116465863454, "grad_norm": 34.09800720214844, "learning_rate": 1.4074193548387096e-05, "loss": 0.3181, "step": 11137 }, { "epoch": 71.79759036144578, "grad_norm": 11.738534927368164, "learning_rate": 1.4070967741935484e-05, "loss": 0.3263, "step": 11138 }, { "epoch": 71.80401606425703, "grad_norm": 10.462334632873535, "learning_rate": 1.4067741935483871e-05, "loss": 0.2669, "step": 11139 }, { "epoch": 71.81044176706827, "grad_norm": 11.501304626464844, "learning_rate": 1.406451612903226e-05, "loss": 0.4629, "step": 11140 }, { "epoch": 71.81686746987951, "grad_norm": 4.759537696838379, "learning_rate": 1.4061290322580644e-05, "loss": 0.1498, "step": 11141 }, { "epoch": 71.82329317269077, "grad_norm": 6.231377124786377, "learning_rate": 1.4058064516129033e-05, "loss": 0.2475, "step": 11142 }, { "epoch": 71.82971887550201, "grad_norm": 16.09848403930664, "learning_rate": 1.405483870967742e-05, "loss": 0.6558, "step": 11143 }, { "epoch": 71.83614457831325, "grad_norm": 4.8170013427734375, "learning_rate": 1.4051612903225808e-05, "loss": 0.2365, "step": 11144 }, { "epoch": 71.8425702811245, "grad_norm": 21.64299964904785, "learning_rate": 1.4048387096774194e-05, "loss": 0.5635, "step": 11145 }, { "epoch": 71.84899598393574, "grad_norm": 13.353261947631836, "learning_rate": 1.4045161290322581e-05, "loss": 0.5347, "step": 11146 }, { "epoch": 71.855421686747, "grad_norm": 6.550199031829834, "learning_rate": 1.4041935483870969e-05, "loss": 0.1546, "step": 11147 }, { "epoch": 71.86184738955824, "grad_norm": 16.34953498840332, "learning_rate": 1.4038709677419356e-05, "loss": 0.2491, "step": 11148 }, { "epoch": 71.86827309236948, "grad_norm": 10.823531150817871, "learning_rate": 1.4035483870967742e-05, "loss": 0.2307, "step": 11149 }, { "epoch": 71.87469879518072, "grad_norm": 11.412637710571289, "learning_rate": 1.403225806451613e-05, "loss": 0.375, "step": 11150 }, { "epoch": 71.88112449799196, "grad_norm": 36.121124267578125, "learning_rate": 1.4029032258064517e-05, "loss": 0.5822, "step": 11151 }, { "epoch": 71.88755020080322, "grad_norm": 14.634746551513672, "learning_rate": 1.4025806451612906e-05, "loss": 0.276, "step": 11152 }, { "epoch": 71.89397590361446, "grad_norm": 13.818081855773926, "learning_rate": 1.402258064516129e-05, "loss": 0.6157, "step": 11153 }, { "epoch": 71.9004016064257, "grad_norm": 6.903292655944824, "learning_rate": 1.4019354838709677e-05, "loss": 0.319, "step": 11154 }, { "epoch": 71.90682730923695, "grad_norm": 12.119915008544922, "learning_rate": 1.4016129032258066e-05, "loss": 0.2422, "step": 11155 }, { "epoch": 71.91325301204819, "grad_norm": 7.388041019439697, "learning_rate": 1.4012903225806454e-05, "loss": 0.2766, "step": 11156 }, { "epoch": 71.91967871485944, "grad_norm": 14.759033203125, "learning_rate": 1.400967741935484e-05, "loss": 0.3262, "step": 11157 }, { "epoch": 71.92610441767069, "grad_norm": 4.841273307800293, "learning_rate": 1.4006451612903227e-05, "loss": 0.3134, "step": 11158 }, { "epoch": 71.93253012048193, "grad_norm": 9.870925903320312, "learning_rate": 1.4003225806451614e-05, "loss": 0.4344, "step": 11159 }, { "epoch": 71.93895582329317, "grad_norm": 6.340672492980957, "learning_rate": 1.4000000000000001e-05, "loss": 0.233, "step": 11160 }, { "epoch": 71.94538152610441, "grad_norm": 3.626347541809082, "learning_rate": 1.3996774193548387e-05, "loss": 0.2584, "step": 11161 }, { "epoch": 71.95180722891567, "grad_norm": 12.438763618469238, "learning_rate": 1.3993548387096775e-05, "loss": 0.4569, "step": 11162 }, { "epoch": 71.95823293172691, "grad_norm": 5.387928485870361, "learning_rate": 1.3990322580645162e-05, "loss": 0.2059, "step": 11163 }, { "epoch": 71.96465863453815, "grad_norm": 7.875983238220215, "learning_rate": 1.3987096774193548e-05, "loss": 0.2983, "step": 11164 }, { "epoch": 71.9710843373494, "grad_norm": 7.9993438720703125, "learning_rate": 1.3983870967741935e-05, "loss": 0.3901, "step": 11165 }, { "epoch": 71.97751004016064, "grad_norm": 4.353142261505127, "learning_rate": 1.3980645161290323e-05, "loss": 0.1165, "step": 11166 }, { "epoch": 71.9839357429719, "grad_norm": 4.792303562164307, "learning_rate": 1.3977419354838712e-05, "loss": 0.1658, "step": 11167 }, { "epoch": 71.99036144578314, "grad_norm": 12.697009086608887, "learning_rate": 1.3974193548387096e-05, "loss": 0.3039, "step": 11168 }, { "epoch": 71.99678714859438, "grad_norm": 13.454633712768555, "learning_rate": 1.3970967741935485e-05, "loss": 0.276, "step": 11169 }, { "epoch": 72.00642570281124, "grad_norm": 6.940415382385254, "learning_rate": 1.3967741935483872e-05, "loss": 0.2373, "step": 11170 }, { "epoch": 72.01285140562248, "grad_norm": 8.547089576721191, "learning_rate": 1.396451612903226e-05, "loss": 0.4776, "step": 11171 }, { "epoch": 72.01927710843374, "grad_norm": 13.186166763305664, "learning_rate": 1.3961290322580645e-05, "loss": 0.2427, "step": 11172 }, { "epoch": 72.02570281124498, "grad_norm": 7.110153675079346, "learning_rate": 1.3958064516129033e-05, "loss": 0.2515, "step": 11173 }, { "epoch": 72.03212851405623, "grad_norm": 10.521536827087402, "learning_rate": 1.395483870967742e-05, "loss": 0.2605, "step": 11174 }, { "epoch": 72.03855421686747, "grad_norm": 5.014297008514404, "learning_rate": 1.3951612903225807e-05, "loss": 0.2286, "step": 11175 }, { "epoch": 72.04497991967871, "grad_norm": 12.138169288635254, "learning_rate": 1.3948387096774193e-05, "loss": 0.5448, "step": 11176 }, { "epoch": 72.05140562248997, "grad_norm": 8.7147855758667, "learning_rate": 1.394516129032258e-05, "loss": 0.2929, "step": 11177 }, { "epoch": 72.05783132530121, "grad_norm": 15.553010940551758, "learning_rate": 1.3941935483870968e-05, "loss": 0.292, "step": 11178 }, { "epoch": 72.06425702811245, "grad_norm": 8.154196739196777, "learning_rate": 1.3938709677419357e-05, "loss": 0.2721, "step": 11179 }, { "epoch": 72.07068273092369, "grad_norm": 10.274209022521973, "learning_rate": 1.3935483870967741e-05, "loss": 0.4492, "step": 11180 }, { "epoch": 72.07710843373494, "grad_norm": 8.874336242675781, "learning_rate": 1.393225806451613e-05, "loss": 0.4266, "step": 11181 }, { "epoch": 72.08353413654619, "grad_norm": 15.80324649810791, "learning_rate": 1.3929032258064518e-05, "loss": 0.4558, "step": 11182 }, { "epoch": 72.08995983935743, "grad_norm": 6.691143989562988, "learning_rate": 1.3925806451612905e-05, "loss": 0.319, "step": 11183 }, { "epoch": 72.09638554216868, "grad_norm": 12.170472145080566, "learning_rate": 1.392258064516129e-05, "loss": 0.3284, "step": 11184 }, { "epoch": 72.10281124497992, "grad_norm": 15.187764167785645, "learning_rate": 1.3919354838709678e-05, "loss": 0.4195, "step": 11185 }, { "epoch": 72.10923694779116, "grad_norm": 10.276333808898926, "learning_rate": 1.3916129032258065e-05, "loss": 0.2819, "step": 11186 }, { "epoch": 72.1156626506024, "grad_norm": 8.022777557373047, "learning_rate": 1.3912903225806453e-05, "loss": 0.1965, "step": 11187 }, { "epoch": 72.12208835341366, "grad_norm": 7.535183906555176, "learning_rate": 1.3909677419354839e-05, "loss": 0.2054, "step": 11188 }, { "epoch": 72.1285140562249, "grad_norm": 12.973431587219238, "learning_rate": 1.3906451612903226e-05, "loss": 0.3394, "step": 11189 }, { "epoch": 72.13493975903614, "grad_norm": 54.991817474365234, "learning_rate": 1.3903225806451613e-05, "loss": 0.4004, "step": 11190 }, { "epoch": 72.14136546184739, "grad_norm": 12.851848602294922, "learning_rate": 1.3900000000000002e-05, "loss": 0.5612, "step": 11191 }, { "epoch": 72.14779116465863, "grad_norm": 12.535515785217285, "learning_rate": 1.3896774193548386e-05, "loss": 0.4028, "step": 11192 }, { "epoch": 72.15421686746988, "grad_norm": 23.472654342651367, "learning_rate": 1.3893548387096776e-05, "loss": 0.3897, "step": 11193 }, { "epoch": 72.16064257028113, "grad_norm": 9.458663940429688, "learning_rate": 1.3890322580645163e-05, "loss": 0.2971, "step": 11194 }, { "epoch": 72.16706827309237, "grad_norm": 9.697534561157227, "learning_rate": 1.388709677419355e-05, "loss": 0.3566, "step": 11195 }, { "epoch": 72.17349397590361, "grad_norm": 33.77050018310547, "learning_rate": 1.3883870967741936e-05, "loss": 0.3754, "step": 11196 }, { "epoch": 72.17991967871485, "grad_norm": 13.533379554748535, "learning_rate": 1.3880645161290323e-05, "loss": 0.4482, "step": 11197 }, { "epoch": 72.18634538152611, "grad_norm": 19.63783073425293, "learning_rate": 1.3877419354838711e-05, "loss": 0.5791, "step": 11198 }, { "epoch": 72.19277108433735, "grad_norm": 10.703657150268555, "learning_rate": 1.3874193548387097e-05, "loss": 0.5228, "step": 11199 }, { "epoch": 72.1991967871486, "grad_norm": 16.37843132019043, "learning_rate": 1.3870967741935484e-05, "loss": 0.5376, "step": 11200 }, { "epoch": 72.20562248995984, "grad_norm": 19.6377010345459, "learning_rate": 1.3867741935483871e-05, "loss": 0.386, "step": 11201 }, { "epoch": 72.21204819277108, "grad_norm": 6.377631187438965, "learning_rate": 1.3864516129032259e-05, "loss": 0.1548, "step": 11202 }, { "epoch": 72.21847389558233, "grad_norm": 8.075305938720703, "learning_rate": 1.3861290322580644e-05, "loss": 0.2211, "step": 11203 }, { "epoch": 72.22489959839358, "grad_norm": 17.848186492919922, "learning_rate": 1.3858064516129032e-05, "loss": 0.4478, "step": 11204 }, { "epoch": 72.23132530120482, "grad_norm": 8.584609985351562, "learning_rate": 1.3854838709677421e-05, "loss": 0.3782, "step": 11205 }, { "epoch": 72.23775100401606, "grad_norm": 8.76755142211914, "learning_rate": 1.3851612903225808e-05, "loss": 0.2978, "step": 11206 }, { "epoch": 72.2441767068273, "grad_norm": 6.5244460105896, "learning_rate": 1.3848387096774192e-05, "loss": 0.4938, "step": 11207 }, { "epoch": 72.25060240963856, "grad_norm": 11.55421257019043, "learning_rate": 1.3845161290322581e-05, "loss": 0.3792, "step": 11208 }, { "epoch": 72.2570281124498, "grad_norm": 10.735637664794922, "learning_rate": 1.3841935483870969e-05, "loss": 0.384, "step": 11209 }, { "epoch": 72.26345381526104, "grad_norm": 4.4119086265563965, "learning_rate": 1.3838709677419356e-05, "loss": 0.21, "step": 11210 }, { "epoch": 72.26987951807229, "grad_norm": 21.207189559936523, "learning_rate": 1.3835483870967742e-05, "loss": 0.2824, "step": 11211 }, { "epoch": 72.27630522088353, "grad_norm": 6.932804584503174, "learning_rate": 1.383225806451613e-05, "loss": 0.2705, "step": 11212 }, { "epoch": 72.28273092369479, "grad_norm": 7.8481950759887695, "learning_rate": 1.3829032258064517e-05, "loss": 0.3365, "step": 11213 }, { "epoch": 72.28915662650603, "grad_norm": 11.413186073303223, "learning_rate": 1.3825806451612904e-05, "loss": 0.3673, "step": 11214 }, { "epoch": 72.29558232931727, "grad_norm": 15.507223129272461, "learning_rate": 1.382258064516129e-05, "loss": 0.5738, "step": 11215 }, { "epoch": 72.30200803212851, "grad_norm": 11.86655044555664, "learning_rate": 1.3819354838709677e-05, "loss": 0.1688, "step": 11216 }, { "epoch": 72.30843373493975, "grad_norm": 15.217381477355957, "learning_rate": 1.3816129032258065e-05, "loss": 0.4824, "step": 11217 }, { "epoch": 72.31485943775101, "grad_norm": 7.584263801574707, "learning_rate": 1.3812903225806454e-05, "loss": 0.1674, "step": 11218 }, { "epoch": 72.32128514056225, "grad_norm": 9.276656150817871, "learning_rate": 1.3809677419354838e-05, "loss": 0.3629, "step": 11219 }, { "epoch": 72.3277108433735, "grad_norm": 6.387947082519531, "learning_rate": 1.3806451612903227e-05, "loss": 0.3384, "step": 11220 }, { "epoch": 72.33413654618474, "grad_norm": 14.062156677246094, "learning_rate": 1.3803225806451614e-05, "loss": 0.2079, "step": 11221 }, { "epoch": 72.34056224899598, "grad_norm": 6.071409702301025, "learning_rate": 1.3800000000000002e-05, "loss": 0.1664, "step": 11222 }, { "epoch": 72.34698795180722, "grad_norm": 10.61702823638916, "learning_rate": 1.3796774193548387e-05, "loss": 0.398, "step": 11223 }, { "epoch": 72.35341365461848, "grad_norm": 9.32119083404541, "learning_rate": 1.3793548387096775e-05, "loss": 0.5346, "step": 11224 }, { "epoch": 72.35983935742972, "grad_norm": 13.296235084533691, "learning_rate": 1.3790322580645162e-05, "loss": 0.4321, "step": 11225 }, { "epoch": 72.36626506024096, "grad_norm": 8.84039306640625, "learning_rate": 1.378709677419355e-05, "loss": 0.587, "step": 11226 }, { "epoch": 72.3726907630522, "grad_norm": 7.301549911499023, "learning_rate": 1.3783870967741935e-05, "loss": 0.2148, "step": 11227 }, { "epoch": 72.37911646586345, "grad_norm": 10.338971138000488, "learning_rate": 1.3780645161290323e-05, "loss": 0.7361, "step": 11228 }, { "epoch": 72.3855421686747, "grad_norm": 10.943202018737793, "learning_rate": 1.377741935483871e-05, "loss": 0.3707, "step": 11229 }, { "epoch": 72.39196787148595, "grad_norm": 8.698347091674805, "learning_rate": 1.37741935483871e-05, "loss": 0.3408, "step": 11230 }, { "epoch": 72.39839357429719, "grad_norm": 15.035500526428223, "learning_rate": 1.3770967741935483e-05, "loss": 0.2091, "step": 11231 }, { "epoch": 72.40481927710843, "grad_norm": 19.142484664916992, "learning_rate": 1.3767741935483872e-05, "loss": 0.5335, "step": 11232 }, { "epoch": 72.41124497991967, "grad_norm": 7.206905364990234, "learning_rate": 1.376451612903226e-05, "loss": 0.4591, "step": 11233 }, { "epoch": 72.41767068273093, "grad_norm": 12.990989685058594, "learning_rate": 1.3761290322580644e-05, "loss": 0.3402, "step": 11234 }, { "epoch": 72.42409638554217, "grad_norm": 6.314487934112549, "learning_rate": 1.3758064516129033e-05, "loss": 0.3414, "step": 11235 }, { "epoch": 72.43052208835341, "grad_norm": 22.816909790039062, "learning_rate": 1.375483870967742e-05, "loss": 0.4961, "step": 11236 }, { "epoch": 72.43694779116466, "grad_norm": 21.189525604248047, "learning_rate": 1.3751612903225808e-05, "loss": 0.3958, "step": 11237 }, { "epoch": 72.4433734939759, "grad_norm": 3.90771746635437, "learning_rate": 1.3748387096774193e-05, "loss": 0.163, "step": 11238 }, { "epoch": 72.44979919678715, "grad_norm": 9.591474533081055, "learning_rate": 1.374516129032258e-05, "loss": 0.6245, "step": 11239 }, { "epoch": 72.4562248995984, "grad_norm": 9.701788902282715, "learning_rate": 1.3741935483870968e-05, "loss": 0.5528, "step": 11240 }, { "epoch": 72.46265060240964, "grad_norm": 14.7894287109375, "learning_rate": 1.3738709677419356e-05, "loss": 0.4054, "step": 11241 }, { "epoch": 72.46907630522088, "grad_norm": 9.72680377960205, "learning_rate": 1.3735483870967741e-05, "loss": 0.473, "step": 11242 }, { "epoch": 72.47550200803212, "grad_norm": 36.373592376708984, "learning_rate": 1.3732258064516129e-05, "loss": 0.8183, "step": 11243 }, { "epoch": 72.48192771084338, "grad_norm": 7.587238311767578, "learning_rate": 1.3729032258064518e-05, "loss": 0.4473, "step": 11244 }, { "epoch": 72.48835341365462, "grad_norm": 8.469470024108887, "learning_rate": 1.3725806451612905e-05, "loss": 0.2138, "step": 11245 }, { "epoch": 72.49477911646586, "grad_norm": 8.67419147491455, "learning_rate": 1.372258064516129e-05, "loss": 0.5415, "step": 11246 }, { "epoch": 72.5012048192771, "grad_norm": 11.152780532836914, "learning_rate": 1.3719354838709678e-05, "loss": 0.3757, "step": 11247 }, { "epoch": 72.50763052208835, "grad_norm": 7.690812587738037, "learning_rate": 1.3716129032258066e-05, "loss": 0.3434, "step": 11248 }, { "epoch": 72.5140562248996, "grad_norm": 26.807493209838867, "learning_rate": 1.3712903225806453e-05, "loss": 0.6349, "step": 11249 }, { "epoch": 72.52048192771085, "grad_norm": 15.630422592163086, "learning_rate": 1.3709677419354839e-05, "loss": 0.6065, "step": 11250 }, { "epoch": 72.52690763052209, "grad_norm": 10.228142738342285, "learning_rate": 1.3706451612903226e-05, "loss": 0.2508, "step": 11251 }, { "epoch": 72.53333333333333, "grad_norm": 5.772899627685547, "learning_rate": 1.3703225806451614e-05, "loss": 0.335, "step": 11252 }, { "epoch": 72.53975903614457, "grad_norm": 4.648911952972412, "learning_rate": 1.3700000000000001e-05, "loss": 0.1281, "step": 11253 }, { "epoch": 72.54618473895583, "grad_norm": 5.675931930541992, "learning_rate": 1.3696774193548387e-05, "loss": 0.1961, "step": 11254 }, { "epoch": 72.55261044176707, "grad_norm": 20.473026275634766, "learning_rate": 1.3693548387096774e-05, "loss": 0.5055, "step": 11255 }, { "epoch": 72.55903614457831, "grad_norm": 17.72543716430664, "learning_rate": 1.3690322580645163e-05, "loss": 0.7758, "step": 11256 }, { "epoch": 72.56546184738956, "grad_norm": 18.539348602294922, "learning_rate": 1.368709677419355e-05, "loss": 0.4777, "step": 11257 }, { "epoch": 72.5718875502008, "grad_norm": 5.7665510177612305, "learning_rate": 1.3683870967741935e-05, "loss": 0.3298, "step": 11258 }, { "epoch": 72.57831325301204, "grad_norm": 10.681336402893066, "learning_rate": 1.3680645161290324e-05, "loss": 0.3656, "step": 11259 }, { "epoch": 72.5847389558233, "grad_norm": 10.139114379882812, "learning_rate": 1.3677419354838711e-05, "loss": 0.4175, "step": 11260 }, { "epoch": 72.59116465863454, "grad_norm": 9.057086944580078, "learning_rate": 1.3674193548387098e-05, "loss": 0.3594, "step": 11261 }, { "epoch": 72.59759036144578, "grad_norm": 10.828425407409668, "learning_rate": 1.3670967741935484e-05, "loss": 0.3947, "step": 11262 }, { "epoch": 72.60401606425702, "grad_norm": 7.429467678070068, "learning_rate": 1.3667741935483872e-05, "loss": 0.345, "step": 11263 }, { "epoch": 72.61044176706827, "grad_norm": 6.688225746154785, "learning_rate": 1.3664516129032259e-05, "loss": 0.286, "step": 11264 }, { "epoch": 72.61686746987952, "grad_norm": 12.17105770111084, "learning_rate": 1.3661290322580646e-05, "loss": 0.2342, "step": 11265 }, { "epoch": 72.62329317269077, "grad_norm": 5.983339786529541, "learning_rate": 1.3658064516129032e-05, "loss": 0.3035, "step": 11266 }, { "epoch": 72.62971887550201, "grad_norm": 8.910189628601074, "learning_rate": 1.365483870967742e-05, "loss": 0.2434, "step": 11267 }, { "epoch": 72.63614457831325, "grad_norm": 11.031219482421875, "learning_rate": 1.3651612903225809e-05, "loss": 0.2326, "step": 11268 }, { "epoch": 72.64257028112449, "grad_norm": 15.016214370727539, "learning_rate": 1.3648387096774196e-05, "loss": 0.3446, "step": 11269 }, { "epoch": 72.64899598393575, "grad_norm": 7.74052095413208, "learning_rate": 1.364516129032258e-05, "loss": 0.2055, "step": 11270 }, { "epoch": 72.65542168674699, "grad_norm": 16.702167510986328, "learning_rate": 1.3641935483870969e-05, "loss": 0.4795, "step": 11271 }, { "epoch": 72.66184738955823, "grad_norm": 7.754491329193115, "learning_rate": 1.3638709677419357e-05, "loss": 0.4775, "step": 11272 }, { "epoch": 72.66827309236947, "grad_norm": 14.239830017089844, "learning_rate": 1.363548387096774e-05, "loss": 0.4173, "step": 11273 }, { "epoch": 72.67469879518072, "grad_norm": 17.913631439208984, "learning_rate": 1.363225806451613e-05, "loss": 0.4222, "step": 11274 }, { "epoch": 72.68112449799197, "grad_norm": 15.782352447509766, "learning_rate": 1.3629032258064517e-05, "loss": 0.6142, "step": 11275 }, { "epoch": 72.68755020080322, "grad_norm": 8.524056434631348, "learning_rate": 1.3625806451612904e-05, "loss": 0.3111, "step": 11276 }, { "epoch": 72.69397590361446, "grad_norm": 7.9599480628967285, "learning_rate": 1.362258064516129e-05, "loss": 0.2898, "step": 11277 }, { "epoch": 72.7004016064257, "grad_norm": 8.128677368164062, "learning_rate": 1.3619354838709678e-05, "loss": 0.3874, "step": 11278 }, { "epoch": 72.70682730923694, "grad_norm": 9.306640625, "learning_rate": 1.3616129032258065e-05, "loss": 0.383, "step": 11279 }, { "epoch": 72.7132530120482, "grad_norm": 4.5481767654418945, "learning_rate": 1.3612903225806452e-05, "loss": 0.1533, "step": 11280 }, { "epoch": 72.71967871485944, "grad_norm": 6.652541637420654, "learning_rate": 1.3609677419354838e-05, "loss": 0.1589, "step": 11281 }, { "epoch": 72.72610441767068, "grad_norm": 8.683876037597656, "learning_rate": 1.3606451612903225e-05, "loss": 0.2198, "step": 11282 }, { "epoch": 72.73253012048193, "grad_norm": 8.301579475402832, "learning_rate": 1.3603225806451615e-05, "loss": 0.2984, "step": 11283 }, { "epoch": 72.73895582329317, "grad_norm": 10.399547576904297, "learning_rate": 1.3600000000000002e-05, "loss": 0.2733, "step": 11284 }, { "epoch": 72.74538152610442, "grad_norm": 18.263988494873047, "learning_rate": 1.3596774193548386e-05, "loss": 0.265, "step": 11285 }, { "epoch": 72.75180722891567, "grad_norm": 7.712699890136719, "learning_rate": 1.3593548387096775e-05, "loss": 0.1995, "step": 11286 }, { "epoch": 72.75823293172691, "grad_norm": 8.027753829956055, "learning_rate": 1.3590322580645162e-05, "loss": 0.2242, "step": 11287 }, { "epoch": 72.76465863453815, "grad_norm": 7.339461803436279, "learning_rate": 1.358709677419355e-05, "loss": 0.2511, "step": 11288 }, { "epoch": 72.7710843373494, "grad_norm": 7.613354206085205, "learning_rate": 1.3583870967741936e-05, "loss": 0.2903, "step": 11289 }, { "epoch": 72.77751004016065, "grad_norm": 5.058289527893066, "learning_rate": 1.3580645161290323e-05, "loss": 0.2667, "step": 11290 }, { "epoch": 72.78393574297189, "grad_norm": 10.691937446594238, "learning_rate": 1.357741935483871e-05, "loss": 0.3603, "step": 11291 }, { "epoch": 72.79036144578313, "grad_norm": 12.799712181091309, "learning_rate": 1.3574193548387098e-05, "loss": 0.448, "step": 11292 }, { "epoch": 72.79678714859438, "grad_norm": 3.5879759788513184, "learning_rate": 1.3570967741935483e-05, "loss": 0.0993, "step": 11293 }, { "epoch": 72.80321285140562, "grad_norm": 11.347413063049316, "learning_rate": 1.3567741935483871e-05, "loss": 0.2583, "step": 11294 }, { "epoch": 72.80963855421686, "grad_norm": 9.471620559692383, "learning_rate": 1.356451612903226e-05, "loss": 0.2926, "step": 11295 }, { "epoch": 72.81606425702812, "grad_norm": 6.924047470092773, "learning_rate": 1.3561290322580647e-05, "loss": 0.2304, "step": 11296 }, { "epoch": 72.82248995983936, "grad_norm": 10.106683731079102, "learning_rate": 1.3558064516129031e-05, "loss": 0.2002, "step": 11297 }, { "epoch": 72.8289156626506, "grad_norm": 9.3718900680542, "learning_rate": 1.355483870967742e-05, "loss": 0.2072, "step": 11298 }, { "epoch": 72.83534136546184, "grad_norm": 5.6394877433776855, "learning_rate": 1.3551612903225808e-05, "loss": 0.1869, "step": 11299 }, { "epoch": 72.84176706827309, "grad_norm": 10.23232364654541, "learning_rate": 1.3548387096774195e-05, "loss": 0.2144, "step": 11300 }, { "epoch": 72.84819277108434, "grad_norm": 12.545649528503418, "learning_rate": 1.3545161290322581e-05, "loss": 0.2498, "step": 11301 }, { "epoch": 72.85461847389558, "grad_norm": 6.456254959106445, "learning_rate": 1.3541935483870968e-05, "loss": 0.4659, "step": 11302 }, { "epoch": 72.86104417670683, "grad_norm": 6.0113325119018555, "learning_rate": 1.3538709677419356e-05, "loss": 0.2448, "step": 11303 }, { "epoch": 72.86746987951807, "grad_norm": 7.351822376251221, "learning_rate": 1.3535483870967743e-05, "loss": 0.2345, "step": 11304 }, { "epoch": 72.87389558232931, "grad_norm": 9.770203590393066, "learning_rate": 1.3532258064516129e-05, "loss": 0.5253, "step": 11305 }, { "epoch": 72.88032128514057, "grad_norm": 14.109454154968262, "learning_rate": 1.3529032258064516e-05, "loss": 0.2539, "step": 11306 }, { "epoch": 72.88674698795181, "grad_norm": 9.201281547546387, "learning_rate": 1.3525806451612905e-05, "loss": 0.1731, "step": 11307 }, { "epoch": 72.89317269076305, "grad_norm": 6.4746880531311035, "learning_rate": 1.352258064516129e-05, "loss": 0.2627, "step": 11308 }, { "epoch": 72.8995983935743, "grad_norm": 15.915838241577148, "learning_rate": 1.3519354838709677e-05, "loss": 0.4418, "step": 11309 }, { "epoch": 72.90602409638554, "grad_norm": 10.150455474853516, "learning_rate": 1.3516129032258066e-05, "loss": 0.5447, "step": 11310 }, { "epoch": 72.91244979919679, "grad_norm": 7.2579193115234375, "learning_rate": 1.3512903225806453e-05, "loss": 0.188, "step": 11311 }, { "epoch": 72.91887550200804, "grad_norm": 13.76992416381836, "learning_rate": 1.3509677419354837e-05, "loss": 0.4232, "step": 11312 }, { "epoch": 72.92530120481928, "grad_norm": 8.448809623718262, "learning_rate": 1.3506451612903226e-05, "loss": 0.1795, "step": 11313 }, { "epoch": 72.93172690763052, "grad_norm": 9.63921070098877, "learning_rate": 1.3503225806451614e-05, "loss": 0.4233, "step": 11314 }, { "epoch": 72.93815261044176, "grad_norm": 6.409614086151123, "learning_rate": 1.3500000000000001e-05, "loss": 0.251, "step": 11315 }, { "epoch": 72.94457831325302, "grad_norm": 5.144510269165039, "learning_rate": 1.3496774193548387e-05, "loss": 0.1476, "step": 11316 }, { "epoch": 72.95100401606426, "grad_norm": 15.039368629455566, "learning_rate": 1.3493548387096774e-05, "loss": 0.4954, "step": 11317 }, { "epoch": 72.9574297188755, "grad_norm": 6.088171005249023, "learning_rate": 1.3490322580645162e-05, "loss": 0.2793, "step": 11318 }, { "epoch": 72.96385542168674, "grad_norm": 7.075700759887695, "learning_rate": 1.348709677419355e-05, "loss": 0.1944, "step": 11319 }, { "epoch": 72.97028112449799, "grad_norm": 11.347112655639648, "learning_rate": 1.3483870967741935e-05, "loss": 0.4888, "step": 11320 }, { "epoch": 72.97670682730924, "grad_norm": 13.520894050598145, "learning_rate": 1.3480645161290322e-05, "loss": 0.6034, "step": 11321 }, { "epoch": 72.98313253012049, "grad_norm": 10.706438064575195, "learning_rate": 1.3477419354838711e-05, "loss": 0.3002, "step": 11322 }, { "epoch": 72.98955823293173, "grad_norm": 7.518767356872559, "learning_rate": 1.3474193548387099e-05, "loss": 0.3357, "step": 11323 }, { "epoch": 72.99598393574297, "grad_norm": 6.3719353675842285, "learning_rate": 1.3470967741935483e-05, "loss": 0.2422, "step": 11324 }, { "epoch": 73.00562248995983, "grad_norm": 20.133941650390625, "learning_rate": 1.3467741935483872e-05, "loss": 0.3327, "step": 11325 }, { "epoch": 73.01204819277109, "grad_norm": 23.653345108032227, "learning_rate": 1.346451612903226e-05, "loss": 0.5903, "step": 11326 }, { "epoch": 73.01847389558233, "grad_norm": 16.571714401245117, "learning_rate": 1.3461290322580647e-05, "loss": 0.3657, "step": 11327 }, { "epoch": 73.02489959839357, "grad_norm": 5.037583351135254, "learning_rate": 1.3458064516129032e-05, "loss": 0.2443, "step": 11328 }, { "epoch": 73.03132530120482, "grad_norm": 27.4832706451416, "learning_rate": 1.345483870967742e-05, "loss": 0.7218, "step": 11329 }, { "epoch": 73.03775100401606, "grad_norm": 22.486772537231445, "learning_rate": 1.3451612903225807e-05, "loss": 0.4238, "step": 11330 }, { "epoch": 73.04417670682732, "grad_norm": 34.99040985107422, "learning_rate": 1.3448387096774196e-05, "loss": 0.5058, "step": 11331 }, { "epoch": 73.05060240963856, "grad_norm": 20.762529373168945, "learning_rate": 1.344516129032258e-05, "loss": 0.2158, "step": 11332 }, { "epoch": 73.0570281124498, "grad_norm": 24.859113693237305, "learning_rate": 1.3441935483870968e-05, "loss": 0.5054, "step": 11333 }, { "epoch": 73.06345381526104, "grad_norm": 11.430506706237793, "learning_rate": 1.3438709677419357e-05, "loss": 0.514, "step": 11334 }, { "epoch": 73.06987951807228, "grad_norm": 17.0384464263916, "learning_rate": 1.3435483870967744e-05, "loss": 0.3991, "step": 11335 }, { "epoch": 73.07630522088354, "grad_norm": 11.041083335876465, "learning_rate": 1.3432258064516128e-05, "loss": 0.369, "step": 11336 }, { "epoch": 73.08273092369478, "grad_norm": 5.469184398651123, "learning_rate": 1.3429032258064517e-05, "loss": 0.301, "step": 11337 }, { "epoch": 73.08915662650602, "grad_norm": 2.9531028270721436, "learning_rate": 1.3425806451612905e-05, "loss": 0.285, "step": 11338 }, { "epoch": 73.09558232931727, "grad_norm": 7.462621688842773, "learning_rate": 1.3422580645161292e-05, "loss": 0.3895, "step": 11339 }, { "epoch": 73.10200803212851, "grad_norm": 4.59721040725708, "learning_rate": 1.3419354838709678e-05, "loss": 0.133, "step": 11340 }, { "epoch": 73.10843373493977, "grad_norm": 13.543598175048828, "learning_rate": 1.3416129032258065e-05, "loss": 0.4282, "step": 11341 }, { "epoch": 73.11485943775101, "grad_norm": 42.603538513183594, "learning_rate": 1.3412903225806453e-05, "loss": 0.2774, "step": 11342 }, { "epoch": 73.12128514056225, "grad_norm": 3.2884609699249268, "learning_rate": 1.3409677419354838e-05, "loss": 0.4768, "step": 11343 }, { "epoch": 73.12771084337349, "grad_norm": 12.16231632232666, "learning_rate": 1.3406451612903226e-05, "loss": 0.3846, "step": 11344 }, { "epoch": 73.13413654618473, "grad_norm": 11.561821937561035, "learning_rate": 1.3403225806451613e-05, "loss": 0.3342, "step": 11345 }, { "epoch": 73.14056224899599, "grad_norm": 7.045403003692627, "learning_rate": 1.3400000000000002e-05, "loss": 0.0809, "step": 11346 }, { "epoch": 73.14698795180723, "grad_norm": 40.15856170654297, "learning_rate": 1.3396774193548386e-05, "loss": 0.6215, "step": 11347 }, { "epoch": 73.15341365461848, "grad_norm": 4.883693218231201, "learning_rate": 1.3393548387096774e-05, "loss": 0.2967, "step": 11348 }, { "epoch": 73.15983935742972, "grad_norm": 7.805540084838867, "learning_rate": 1.3390322580645163e-05, "loss": 0.2342, "step": 11349 }, { "epoch": 73.16626506024096, "grad_norm": 24.251911163330078, "learning_rate": 1.338709677419355e-05, "loss": 0.3518, "step": 11350 }, { "epoch": 73.1726907630522, "grad_norm": 16.301361083984375, "learning_rate": 1.3383870967741936e-05, "loss": 0.4973, "step": 11351 }, { "epoch": 73.17911646586346, "grad_norm": 15.407856941223145, "learning_rate": 1.3380645161290323e-05, "loss": 0.4956, "step": 11352 }, { "epoch": 73.1855421686747, "grad_norm": 10.63922119140625, "learning_rate": 1.337741935483871e-05, "loss": 0.4926, "step": 11353 }, { "epoch": 73.19196787148594, "grad_norm": 34.35561752319336, "learning_rate": 1.3374193548387098e-05, "loss": 0.4372, "step": 11354 }, { "epoch": 73.19839357429719, "grad_norm": 12.555814743041992, "learning_rate": 1.3370967741935484e-05, "loss": 0.3142, "step": 11355 }, { "epoch": 73.20481927710843, "grad_norm": 68.1099853515625, "learning_rate": 1.3367741935483871e-05, "loss": 0.4533, "step": 11356 }, { "epoch": 73.21124497991968, "grad_norm": 16.434480667114258, "learning_rate": 1.3364516129032258e-05, "loss": 0.4186, "step": 11357 }, { "epoch": 73.21767068273093, "grad_norm": 11.950297355651855, "learning_rate": 1.3361290322580648e-05, "loss": 0.371, "step": 11358 }, { "epoch": 73.22409638554217, "grad_norm": 11.668167114257812, "learning_rate": 1.3358064516129032e-05, "loss": 0.4727, "step": 11359 }, { "epoch": 73.23052208835341, "grad_norm": 20.645845413208008, "learning_rate": 1.3354838709677419e-05, "loss": 0.2693, "step": 11360 }, { "epoch": 73.23694779116465, "grad_norm": 7.1647629737854, "learning_rate": 1.3351612903225808e-05, "loss": 0.3108, "step": 11361 }, { "epoch": 73.24337349397591, "grad_norm": 23.46388816833496, "learning_rate": 1.3348387096774195e-05, "loss": 0.3553, "step": 11362 }, { "epoch": 73.24979919678715, "grad_norm": 6.832275867462158, "learning_rate": 1.334516129032258e-05, "loss": 0.1692, "step": 11363 }, { "epoch": 73.2562248995984, "grad_norm": 16.15122413635254, "learning_rate": 1.3341935483870969e-05, "loss": 0.3475, "step": 11364 }, { "epoch": 73.26265060240964, "grad_norm": 15.952591896057129, "learning_rate": 1.3338709677419356e-05, "loss": 0.421, "step": 11365 }, { "epoch": 73.26907630522088, "grad_norm": 32.5468864440918, "learning_rate": 1.3335483870967743e-05, "loss": 0.2322, "step": 11366 }, { "epoch": 73.27550200803213, "grad_norm": 23.181135177612305, "learning_rate": 1.3332258064516129e-05, "loss": 0.2287, "step": 11367 }, { "epoch": 73.28192771084338, "grad_norm": 22.045196533203125, "learning_rate": 1.3329032258064516e-05, "loss": 0.3509, "step": 11368 }, { "epoch": 73.28835341365462, "grad_norm": 21.170307159423828, "learning_rate": 1.3325806451612904e-05, "loss": 0.3623, "step": 11369 }, { "epoch": 73.29477911646586, "grad_norm": 14.699597358703613, "learning_rate": 1.3322580645161293e-05, "loss": 0.4141, "step": 11370 }, { "epoch": 73.3012048192771, "grad_norm": 20.321313858032227, "learning_rate": 1.3319354838709677e-05, "loss": 0.4516, "step": 11371 }, { "epoch": 73.30763052208836, "grad_norm": 20.6833438873291, "learning_rate": 1.3316129032258064e-05, "loss": 0.2872, "step": 11372 }, { "epoch": 73.3140562248996, "grad_norm": 27.274188995361328, "learning_rate": 1.3312903225806454e-05, "loss": 0.3778, "step": 11373 }, { "epoch": 73.32048192771084, "grad_norm": 22.74163055419922, "learning_rate": 1.3309677419354841e-05, "loss": 0.3315, "step": 11374 }, { "epoch": 73.32690763052209, "grad_norm": 6.300030708312988, "learning_rate": 1.3306451612903225e-05, "loss": 0.2499, "step": 11375 }, { "epoch": 73.33333333333333, "grad_norm": 33.02091598510742, "learning_rate": 1.3303225806451614e-05, "loss": 0.1544, "step": 11376 }, { "epoch": 73.33975903614459, "grad_norm": 19.975051879882812, "learning_rate": 1.3300000000000001e-05, "loss": 0.5953, "step": 11377 }, { "epoch": 73.34618473895583, "grad_norm": 17.59523582458496, "learning_rate": 1.3296774193548387e-05, "loss": 0.4499, "step": 11378 }, { "epoch": 73.35261044176707, "grad_norm": 8.947306632995605, "learning_rate": 1.3293548387096775e-05, "loss": 0.3288, "step": 11379 }, { "epoch": 73.35903614457831, "grad_norm": 13.899598121643066, "learning_rate": 1.3290322580645162e-05, "loss": 0.302, "step": 11380 }, { "epoch": 73.36546184738955, "grad_norm": 17.035247802734375, "learning_rate": 1.328709677419355e-05, "loss": 0.3158, "step": 11381 }, { "epoch": 73.3718875502008, "grad_norm": 61.80740737915039, "learning_rate": 1.3283870967741935e-05, "loss": 0.3176, "step": 11382 }, { "epoch": 73.37831325301205, "grad_norm": 18.655790328979492, "learning_rate": 1.3280645161290322e-05, "loss": 0.1951, "step": 11383 }, { "epoch": 73.3847389558233, "grad_norm": 35.517242431640625, "learning_rate": 1.327741935483871e-05, "loss": 0.3455, "step": 11384 }, { "epoch": 73.39116465863454, "grad_norm": 13.444392204284668, "learning_rate": 1.3274193548387099e-05, "loss": 0.1898, "step": 11385 }, { "epoch": 73.39759036144578, "grad_norm": 18.308887481689453, "learning_rate": 1.3270967741935483e-05, "loss": 0.349, "step": 11386 }, { "epoch": 73.40401606425702, "grad_norm": 10.407633781433105, "learning_rate": 1.326774193548387e-05, "loss": 0.2834, "step": 11387 }, { "epoch": 73.41044176706828, "grad_norm": 19.04804801940918, "learning_rate": 1.326451612903226e-05, "loss": 0.3219, "step": 11388 }, { "epoch": 73.41686746987952, "grad_norm": 15.314685821533203, "learning_rate": 1.3261290322580647e-05, "loss": 0.325, "step": 11389 }, { "epoch": 73.42329317269076, "grad_norm": 18.950908660888672, "learning_rate": 1.3258064516129033e-05, "loss": 0.4327, "step": 11390 }, { "epoch": 73.429718875502, "grad_norm": 6.734734535217285, "learning_rate": 1.325483870967742e-05, "loss": 0.4559, "step": 11391 }, { "epoch": 73.43614457831325, "grad_norm": 27.55721092224121, "learning_rate": 1.3251612903225807e-05, "loss": 0.268, "step": 11392 }, { "epoch": 73.4425702811245, "grad_norm": 7.41220760345459, "learning_rate": 1.3248387096774195e-05, "loss": 0.1429, "step": 11393 }, { "epoch": 73.44899598393575, "grad_norm": 26.387157440185547, "learning_rate": 1.324516129032258e-05, "loss": 0.2935, "step": 11394 }, { "epoch": 73.45542168674699, "grad_norm": 10.004230499267578, "learning_rate": 1.3241935483870968e-05, "loss": 0.4066, "step": 11395 }, { "epoch": 73.46184738955823, "grad_norm": 16.13243293762207, "learning_rate": 1.3238709677419355e-05, "loss": 0.4428, "step": 11396 }, { "epoch": 73.46827309236947, "grad_norm": 17.343515396118164, "learning_rate": 1.3235483870967744e-05, "loss": 0.4457, "step": 11397 }, { "epoch": 73.47469879518073, "grad_norm": 22.527324676513672, "learning_rate": 1.3232258064516128e-05, "loss": 0.3681, "step": 11398 }, { "epoch": 73.48112449799197, "grad_norm": 13.818653106689453, "learning_rate": 1.3229032258064516e-05, "loss": 0.2819, "step": 11399 }, { "epoch": 73.48755020080321, "grad_norm": 9.833773612976074, "learning_rate": 1.3225806451612905e-05, "loss": 0.1495, "step": 11400 }, { "epoch": 73.49397590361446, "grad_norm": 14.88431167602539, "learning_rate": 1.3222580645161292e-05, "loss": 0.2641, "step": 11401 }, { "epoch": 73.5004016064257, "grad_norm": 24.643842697143555, "learning_rate": 1.3219354838709678e-05, "loss": 0.7049, "step": 11402 }, { "epoch": 73.50682730923695, "grad_norm": 5.9392595291137695, "learning_rate": 1.3216129032258065e-05, "loss": 0.3406, "step": 11403 }, { "epoch": 73.5132530120482, "grad_norm": 8.14792537689209, "learning_rate": 1.3212903225806453e-05, "loss": 0.3755, "step": 11404 }, { "epoch": 73.51967871485944, "grad_norm": 21.068090438842773, "learning_rate": 1.320967741935484e-05, "loss": 0.4833, "step": 11405 }, { "epoch": 73.52610441767068, "grad_norm": 23.86457061767578, "learning_rate": 1.3206451612903226e-05, "loss": 0.203, "step": 11406 }, { "epoch": 73.53253012048192, "grad_norm": 9.917435646057129, "learning_rate": 1.3203225806451613e-05, "loss": 0.4489, "step": 11407 }, { "epoch": 73.53895582329318, "grad_norm": 5.638948917388916, "learning_rate": 1.32e-05, "loss": 0.223, "step": 11408 }, { "epoch": 73.54538152610442, "grad_norm": 23.224477767944336, "learning_rate": 1.319677419354839e-05, "loss": 0.3007, "step": 11409 }, { "epoch": 73.55180722891566, "grad_norm": 21.368724822998047, "learning_rate": 1.3193548387096774e-05, "loss": 0.5293, "step": 11410 }, { "epoch": 73.5582329317269, "grad_norm": 1.855591058731079, "learning_rate": 1.3190322580645161e-05, "loss": 0.1117, "step": 11411 }, { "epoch": 73.56465863453815, "grad_norm": 23.21368980407715, "learning_rate": 1.318709677419355e-05, "loss": 0.3015, "step": 11412 }, { "epoch": 73.5710843373494, "grad_norm": 2.647475004196167, "learning_rate": 1.3183870967741934e-05, "loss": 0.2272, "step": 11413 }, { "epoch": 73.57751004016065, "grad_norm": 8.46017074584961, "learning_rate": 1.3180645161290323e-05, "loss": 0.5201, "step": 11414 }, { "epoch": 73.58393574297189, "grad_norm": 10.57668685913086, "learning_rate": 1.317741935483871e-05, "loss": 0.5365, "step": 11415 }, { "epoch": 73.59036144578313, "grad_norm": 4.608310699462891, "learning_rate": 1.3174193548387098e-05, "loss": 0.0723, "step": 11416 }, { "epoch": 73.59678714859437, "grad_norm": 24.22916030883789, "learning_rate": 1.3170967741935484e-05, "loss": 0.2876, "step": 11417 }, { "epoch": 73.60321285140563, "grad_norm": 38.63987731933594, "learning_rate": 1.3167741935483871e-05, "loss": 0.3971, "step": 11418 }, { "epoch": 73.60963855421687, "grad_norm": 8.42976188659668, "learning_rate": 1.3164516129032259e-05, "loss": 0.4181, "step": 11419 }, { "epoch": 73.61606425702811, "grad_norm": 24.329917907714844, "learning_rate": 1.3161290322580646e-05, "loss": 0.5136, "step": 11420 }, { "epoch": 73.62248995983936, "grad_norm": 9.688454627990723, "learning_rate": 1.3158064516129032e-05, "loss": 0.4168, "step": 11421 }, { "epoch": 73.6289156626506, "grad_norm": 6.224981784820557, "learning_rate": 1.315483870967742e-05, "loss": 0.1688, "step": 11422 }, { "epoch": 73.63534136546184, "grad_norm": 16.723407745361328, "learning_rate": 1.3151612903225807e-05, "loss": 0.3766, "step": 11423 }, { "epoch": 73.6417670682731, "grad_norm": 132.2700653076172, "learning_rate": 1.3148387096774196e-05, "loss": 0.2605, "step": 11424 }, { "epoch": 73.64819277108434, "grad_norm": 14.502820014953613, "learning_rate": 1.314516129032258e-05, "loss": 0.3135, "step": 11425 }, { "epoch": 73.65461847389558, "grad_norm": 21.04939079284668, "learning_rate": 1.3141935483870967e-05, "loss": 0.491, "step": 11426 }, { "epoch": 73.66104417670682, "grad_norm": 12.942314147949219, "learning_rate": 1.3138709677419356e-05, "loss": 0.1806, "step": 11427 }, { "epoch": 73.66746987951807, "grad_norm": 22.56622314453125, "learning_rate": 1.3135483870967744e-05, "loss": 0.338, "step": 11428 }, { "epoch": 73.67389558232932, "grad_norm": 14.207125663757324, "learning_rate": 1.313225806451613e-05, "loss": 0.4905, "step": 11429 }, { "epoch": 73.68032128514056, "grad_norm": 29.734668731689453, "learning_rate": 1.3129032258064517e-05, "loss": 0.3598, "step": 11430 }, { "epoch": 73.6867469879518, "grad_norm": 13.876202583312988, "learning_rate": 1.3125806451612904e-05, "loss": 0.3095, "step": 11431 }, { "epoch": 73.69317269076305, "grad_norm": 15.448945999145508, "learning_rate": 1.3122580645161292e-05, "loss": 0.368, "step": 11432 }, { "epoch": 73.69959839357429, "grad_norm": 20.71588134765625, "learning_rate": 1.3119354838709677e-05, "loss": 0.2859, "step": 11433 }, { "epoch": 73.70602409638555, "grad_norm": 108.07635498046875, "learning_rate": 1.3116129032258065e-05, "loss": 0.3062, "step": 11434 }, { "epoch": 73.71244979919679, "grad_norm": 5.58262300491333, "learning_rate": 1.3112903225806452e-05, "loss": 0.3708, "step": 11435 }, { "epoch": 73.71887550200803, "grad_norm": 49.25411605834961, "learning_rate": 1.3109677419354841e-05, "loss": 0.3891, "step": 11436 }, { "epoch": 73.72530120481927, "grad_norm": 33.074066162109375, "learning_rate": 1.3106451612903225e-05, "loss": 0.4098, "step": 11437 }, { "epoch": 73.73172690763052, "grad_norm": 5.908755302429199, "learning_rate": 1.3103225806451613e-05, "loss": 0.1862, "step": 11438 }, { "epoch": 73.73815261044177, "grad_norm": 13.883624076843262, "learning_rate": 1.3100000000000002e-05, "loss": 0.3344, "step": 11439 }, { "epoch": 73.74457831325302, "grad_norm": 4.332909107208252, "learning_rate": 1.3096774193548389e-05, "loss": 0.3255, "step": 11440 }, { "epoch": 73.75100401606426, "grad_norm": 12.390989303588867, "learning_rate": 1.3093548387096775e-05, "loss": 0.4138, "step": 11441 }, { "epoch": 73.7574297188755, "grad_norm": 36.22265625, "learning_rate": 1.3090322580645162e-05, "loss": 0.3197, "step": 11442 }, { "epoch": 73.76385542168674, "grad_norm": 6.056005477905273, "learning_rate": 1.308709677419355e-05, "loss": 0.1621, "step": 11443 }, { "epoch": 73.770281124498, "grad_norm": 10.374762535095215, "learning_rate": 1.3083870967741937e-05, "loss": 0.3738, "step": 11444 }, { "epoch": 73.77670682730924, "grad_norm": 4.472620010375977, "learning_rate": 1.3080645161290323e-05, "loss": 0.3652, "step": 11445 }, { "epoch": 73.78313253012048, "grad_norm": 0.9809226989746094, "learning_rate": 1.307741935483871e-05, "loss": 0.1199, "step": 11446 }, { "epoch": 73.78955823293173, "grad_norm": 4.397385120391846, "learning_rate": 1.3074193548387097e-05, "loss": 0.2616, "step": 11447 }, { "epoch": 73.79598393574297, "grad_norm": 35.77999496459961, "learning_rate": 1.3070967741935483e-05, "loss": 0.4075, "step": 11448 }, { "epoch": 73.80240963855422, "grad_norm": 23.842607498168945, "learning_rate": 1.306774193548387e-05, "loss": 0.6694, "step": 11449 }, { "epoch": 73.80883534136547, "grad_norm": 12.526016235351562, "learning_rate": 1.3064516129032258e-05, "loss": 0.3631, "step": 11450 }, { "epoch": 73.81526104417671, "grad_norm": 41.495540618896484, "learning_rate": 1.3061290322580647e-05, "loss": 0.7689, "step": 11451 }, { "epoch": 73.82168674698795, "grad_norm": 22.897762298583984, "learning_rate": 1.3058064516129031e-05, "loss": 0.4853, "step": 11452 }, { "epoch": 73.82811244979919, "grad_norm": 15.246039390563965, "learning_rate": 1.305483870967742e-05, "loss": 0.3568, "step": 11453 }, { "epoch": 73.83453815261043, "grad_norm": 9.143372535705566, "learning_rate": 1.3051612903225808e-05, "loss": 0.2354, "step": 11454 }, { "epoch": 73.84096385542169, "grad_norm": 7.542817115783691, "learning_rate": 1.3048387096774195e-05, "loss": 0.4684, "step": 11455 }, { "epoch": 73.84738955823293, "grad_norm": 10.477184295654297, "learning_rate": 1.304516129032258e-05, "loss": 0.4532, "step": 11456 }, { "epoch": 73.85381526104418, "grad_norm": 6.7349700927734375, "learning_rate": 1.3041935483870968e-05, "loss": 0.31, "step": 11457 }, { "epoch": 73.86024096385542, "grad_norm": 27.04928207397461, "learning_rate": 1.3038709677419355e-05, "loss": 0.2873, "step": 11458 }, { "epoch": 73.86666666666666, "grad_norm": 7.7953596115112305, "learning_rate": 1.3035483870967743e-05, "loss": 0.3558, "step": 11459 }, { "epoch": 73.87309236947792, "grad_norm": 22.05460548400879, "learning_rate": 1.3032258064516129e-05, "loss": 0.4448, "step": 11460 }, { "epoch": 73.87951807228916, "grad_norm": 20.014366149902344, "learning_rate": 1.3029032258064516e-05, "loss": 0.4622, "step": 11461 }, { "epoch": 73.8859437751004, "grad_norm": 7.643558025360107, "learning_rate": 1.3025806451612903e-05, "loss": 0.2408, "step": 11462 }, { "epoch": 73.89236947791164, "grad_norm": 19.0252742767334, "learning_rate": 1.3022580645161292e-05, "loss": 0.3445, "step": 11463 }, { "epoch": 73.89879518072289, "grad_norm": 15.224724769592285, "learning_rate": 1.3019354838709676e-05, "loss": 0.3758, "step": 11464 }, { "epoch": 73.90522088353414, "grad_norm": 20.22138214111328, "learning_rate": 1.3016129032258066e-05, "loss": 0.6391, "step": 11465 }, { "epoch": 73.91164658634538, "grad_norm": 28.281105041503906, "learning_rate": 1.3012903225806453e-05, "loss": 0.3544, "step": 11466 }, { "epoch": 73.91807228915663, "grad_norm": 9.11500358581543, "learning_rate": 1.300967741935484e-05, "loss": 0.1831, "step": 11467 }, { "epoch": 73.92449799196787, "grad_norm": 16.54926109313965, "learning_rate": 1.3006451612903226e-05, "loss": 0.536, "step": 11468 }, { "epoch": 73.93092369477911, "grad_norm": 3.237851619720459, "learning_rate": 1.3003225806451613e-05, "loss": 0.2397, "step": 11469 }, { "epoch": 73.93734939759037, "grad_norm": 16.74538803100586, "learning_rate": 1.3000000000000001e-05, "loss": 0.3935, "step": 11470 }, { "epoch": 73.94377510040161, "grad_norm": 9.199413299560547, "learning_rate": 1.2996774193548388e-05, "loss": 0.3386, "step": 11471 }, { "epoch": 73.95020080321285, "grad_norm": 3.883836030960083, "learning_rate": 1.2993548387096774e-05, "loss": 0.3082, "step": 11472 }, { "epoch": 73.9566265060241, "grad_norm": 35.86561965942383, "learning_rate": 1.2990322580645161e-05, "loss": 0.3128, "step": 11473 }, { "epoch": 73.96305220883534, "grad_norm": 5.549123764038086, "learning_rate": 1.2987096774193549e-05, "loss": 0.0975, "step": 11474 }, { "epoch": 73.96947791164659, "grad_norm": 20.126209259033203, "learning_rate": 1.2983870967741938e-05, "loss": 0.3506, "step": 11475 }, { "epoch": 73.97590361445783, "grad_norm": 19.646194458007812, "learning_rate": 1.2980645161290322e-05, "loss": 0.4333, "step": 11476 }, { "epoch": 73.98232931726908, "grad_norm": 11.483283042907715, "learning_rate": 1.2977419354838711e-05, "loss": 0.4455, "step": 11477 }, { "epoch": 73.98875502008032, "grad_norm": 33.22245788574219, "learning_rate": 1.2974193548387098e-05, "loss": 0.1606, "step": 11478 }, { "epoch": 73.99518072289156, "grad_norm": 14.398019790649414, "learning_rate": 1.2970967741935486e-05, "loss": 0.2523, "step": 11479 }, { "epoch": 74.00481927710844, "grad_norm": 7.689138889312744, "learning_rate": 1.2967741935483872e-05, "loss": 0.1383, "step": 11480 }, { "epoch": 74.01124497991968, "grad_norm": 15.177087783813477, "learning_rate": 1.2964516129032259e-05, "loss": 0.2696, "step": 11481 }, { "epoch": 74.01767068273092, "grad_norm": 3.130850315093994, "learning_rate": 1.2961290322580646e-05, "loss": 0.1509, "step": 11482 }, { "epoch": 74.02409638554217, "grad_norm": 13.265777587890625, "learning_rate": 1.2958064516129032e-05, "loss": 0.445, "step": 11483 }, { "epoch": 74.03052208835341, "grad_norm": 32.22446060180664, "learning_rate": 1.295483870967742e-05, "loss": 0.3517, "step": 11484 }, { "epoch": 74.03694779116466, "grad_norm": 19.116994857788086, "learning_rate": 1.2951612903225807e-05, "loss": 0.3098, "step": 11485 }, { "epoch": 74.0433734939759, "grad_norm": 4.207432270050049, "learning_rate": 1.2948387096774194e-05, "loss": 0.1014, "step": 11486 }, { "epoch": 74.04979919678715, "grad_norm": 15.709590911865234, "learning_rate": 1.294516129032258e-05, "loss": 0.2252, "step": 11487 }, { "epoch": 74.05622489959839, "grad_norm": 45.097896575927734, "learning_rate": 1.2941935483870967e-05, "loss": 0.7388, "step": 11488 }, { "epoch": 74.06265060240963, "grad_norm": 26.667057037353516, "learning_rate": 1.2938709677419355e-05, "loss": 0.1915, "step": 11489 }, { "epoch": 74.06907630522089, "grad_norm": 14.748495101928711, "learning_rate": 1.2935483870967744e-05, "loss": 0.4572, "step": 11490 }, { "epoch": 74.07550200803213, "grad_norm": 7.374574184417725, "learning_rate": 1.2932258064516128e-05, "loss": 0.1771, "step": 11491 }, { "epoch": 74.08192771084337, "grad_norm": 17.126035690307617, "learning_rate": 1.2929032258064517e-05, "loss": 0.1548, "step": 11492 }, { "epoch": 74.08835341365462, "grad_norm": 3.4330151081085205, "learning_rate": 1.2925806451612904e-05, "loss": 0.2765, "step": 11493 }, { "epoch": 74.09477911646586, "grad_norm": 65.1695327758789, "learning_rate": 1.2922580645161292e-05, "loss": 0.6198, "step": 11494 }, { "epoch": 74.10120481927711, "grad_norm": 28.111162185668945, "learning_rate": 1.2919354838709677e-05, "loss": 0.2403, "step": 11495 }, { "epoch": 74.10763052208836, "grad_norm": 7.203749179840088, "learning_rate": 1.2916129032258065e-05, "loss": 0.3789, "step": 11496 }, { "epoch": 74.1140562248996, "grad_norm": 16.82021141052246, "learning_rate": 1.2912903225806452e-05, "loss": 0.4857, "step": 11497 }, { "epoch": 74.12048192771084, "grad_norm": 14.9382963180542, "learning_rate": 1.290967741935484e-05, "loss": 0.4787, "step": 11498 }, { "epoch": 74.12690763052208, "grad_norm": 20.407835006713867, "learning_rate": 1.2906451612903225e-05, "loss": 0.2704, "step": 11499 }, { "epoch": 74.13333333333334, "grad_norm": 4.354835510253906, "learning_rate": 1.2903225806451613e-05, "loss": 0.3328, "step": 11500 }, { "epoch": 74.13975903614458, "grad_norm": 18.795564651489258, "learning_rate": 1.29e-05, "loss": 0.216, "step": 11501 }, { "epoch": 74.14618473895582, "grad_norm": 21.180450439453125, "learning_rate": 1.289677419354839e-05, "loss": 0.363, "step": 11502 }, { "epoch": 74.15261044176707, "grad_norm": 3.337341547012329, "learning_rate": 1.2893548387096773e-05, "loss": 0.1999, "step": 11503 }, { "epoch": 74.15903614457831, "grad_norm": 23.486146926879883, "learning_rate": 1.2890322580645162e-05, "loss": 0.2286, "step": 11504 }, { "epoch": 74.16546184738957, "grad_norm": 14.341392517089844, "learning_rate": 1.288709677419355e-05, "loss": 0.2569, "step": 11505 }, { "epoch": 74.17188755020081, "grad_norm": 14.283839225769043, "learning_rate": 1.2883870967741937e-05, "loss": 0.4518, "step": 11506 }, { "epoch": 74.17831325301205, "grad_norm": 15.706862449645996, "learning_rate": 1.2880645161290323e-05, "loss": 0.6132, "step": 11507 }, { "epoch": 74.18473895582329, "grad_norm": 16.268569946289062, "learning_rate": 1.287741935483871e-05, "loss": 0.3586, "step": 11508 }, { "epoch": 74.19116465863453, "grad_norm": 6.701987266540527, "learning_rate": 1.2874193548387098e-05, "loss": 0.4842, "step": 11509 }, { "epoch": 74.19759036144578, "grad_norm": 5.620816707611084, "learning_rate": 1.2870967741935485e-05, "loss": 0.3911, "step": 11510 }, { "epoch": 74.20401606425703, "grad_norm": 8.548803329467773, "learning_rate": 1.286774193548387e-05, "loss": 0.2366, "step": 11511 }, { "epoch": 74.21044176706827, "grad_norm": 32.66473388671875, "learning_rate": 1.2864516129032258e-05, "loss": 0.4682, "step": 11512 }, { "epoch": 74.21686746987952, "grad_norm": 14.706523895263672, "learning_rate": 1.2861290322580646e-05, "loss": 0.3602, "step": 11513 }, { "epoch": 74.22329317269076, "grad_norm": 12.545736312866211, "learning_rate": 1.2858064516129035e-05, "loss": 0.3347, "step": 11514 }, { "epoch": 74.229718875502, "grad_norm": 41.482295989990234, "learning_rate": 1.2854838709677419e-05, "loss": 0.3284, "step": 11515 }, { "epoch": 74.23614457831326, "grad_norm": 10.141473770141602, "learning_rate": 1.2851612903225808e-05, "loss": 0.4288, "step": 11516 }, { "epoch": 74.2425702811245, "grad_norm": 39.23409652709961, "learning_rate": 1.2848387096774195e-05, "loss": 0.2535, "step": 11517 }, { "epoch": 74.24899598393574, "grad_norm": 23.78781509399414, "learning_rate": 1.2845161290322583e-05, "loss": 0.426, "step": 11518 }, { "epoch": 74.25542168674698, "grad_norm": 17.80862808227539, "learning_rate": 1.2841935483870968e-05, "loss": 0.246, "step": 11519 }, { "epoch": 74.26184738955823, "grad_norm": 6.141529560089111, "learning_rate": 1.2838709677419356e-05, "loss": 0.1966, "step": 11520 }, { "epoch": 74.26827309236948, "grad_norm": 4.923411846160889, "learning_rate": 1.2835483870967743e-05, "loss": 0.1887, "step": 11521 }, { "epoch": 74.27469879518073, "grad_norm": 11.076740264892578, "learning_rate": 1.2832258064516129e-05, "loss": 0.4225, "step": 11522 }, { "epoch": 74.28112449799197, "grad_norm": 8.31588077545166, "learning_rate": 1.2829032258064516e-05, "loss": 0.3115, "step": 11523 }, { "epoch": 74.28755020080321, "grad_norm": 18.905582427978516, "learning_rate": 1.2825806451612904e-05, "loss": 0.5305, "step": 11524 }, { "epoch": 74.29397590361445, "grad_norm": 17.958166122436523, "learning_rate": 1.2822580645161291e-05, "loss": 0.2191, "step": 11525 }, { "epoch": 74.30040160642571, "grad_norm": 5.075779914855957, "learning_rate": 1.2819354838709677e-05, "loss": 0.158, "step": 11526 }, { "epoch": 74.30682730923695, "grad_norm": 31.135601043701172, "learning_rate": 1.2816129032258064e-05, "loss": 0.3967, "step": 11527 }, { "epoch": 74.3132530120482, "grad_norm": 13.954828262329102, "learning_rate": 1.2812903225806453e-05, "loss": 0.2888, "step": 11528 }, { "epoch": 74.31967871485944, "grad_norm": 21.02936363220215, "learning_rate": 1.280967741935484e-05, "loss": 0.4815, "step": 11529 }, { "epoch": 74.32610441767068, "grad_norm": 13.324379920959473, "learning_rate": 1.2806451612903225e-05, "loss": 0.3773, "step": 11530 }, { "epoch": 74.33253012048193, "grad_norm": 18.31964111328125, "learning_rate": 1.2803225806451614e-05, "loss": 0.7145, "step": 11531 }, { "epoch": 74.33895582329318, "grad_norm": 37.68881607055664, "learning_rate": 1.2800000000000001e-05, "loss": 0.4556, "step": 11532 }, { "epoch": 74.34538152610442, "grad_norm": 3.193983316421509, "learning_rate": 1.2796774193548389e-05, "loss": 0.2311, "step": 11533 }, { "epoch": 74.35180722891566, "grad_norm": 22.996749877929688, "learning_rate": 1.2793548387096774e-05, "loss": 0.6013, "step": 11534 }, { "epoch": 74.3582329317269, "grad_norm": 10.228848457336426, "learning_rate": 1.2790322580645162e-05, "loss": 0.2465, "step": 11535 }, { "epoch": 74.36465863453816, "grad_norm": 4.621026515960693, "learning_rate": 1.2787096774193549e-05, "loss": 0.4566, "step": 11536 }, { "epoch": 74.3710843373494, "grad_norm": 45.011192321777344, "learning_rate": 1.2783870967741936e-05, "loss": 0.3542, "step": 11537 }, { "epoch": 74.37751004016064, "grad_norm": 3.1198489665985107, "learning_rate": 1.2780645161290322e-05, "loss": 0.2138, "step": 11538 }, { "epoch": 74.38393574297189, "grad_norm": 25.565616607666016, "learning_rate": 1.277741935483871e-05, "loss": 0.4766, "step": 11539 }, { "epoch": 74.39036144578313, "grad_norm": 10.316567420959473, "learning_rate": 1.2774193548387099e-05, "loss": 0.297, "step": 11540 }, { "epoch": 74.39678714859438, "grad_norm": 9.894896507263184, "learning_rate": 1.2770967741935486e-05, "loss": 0.3044, "step": 11541 }, { "epoch": 74.40321285140563, "grad_norm": 5.6963324546813965, "learning_rate": 1.276774193548387e-05, "loss": 0.4426, "step": 11542 }, { "epoch": 74.40963855421687, "grad_norm": 3.553941249847412, "learning_rate": 1.2764516129032259e-05, "loss": 0.1738, "step": 11543 }, { "epoch": 74.41606425702811, "grad_norm": 105.27574157714844, "learning_rate": 1.2761290322580647e-05, "loss": 0.3921, "step": 11544 }, { "epoch": 74.42248995983935, "grad_norm": 24.75446891784668, "learning_rate": 1.2758064516129034e-05, "loss": 0.4401, "step": 11545 }, { "epoch": 74.4289156626506, "grad_norm": 8.18326473236084, "learning_rate": 1.275483870967742e-05, "loss": 0.5807, "step": 11546 }, { "epoch": 74.43534136546185, "grad_norm": 10.866838455200195, "learning_rate": 1.2751612903225807e-05, "loss": 0.1588, "step": 11547 }, { "epoch": 74.4417670682731, "grad_norm": 34.161075592041016, "learning_rate": 1.2748387096774194e-05, "loss": 0.5973, "step": 11548 }, { "epoch": 74.44819277108434, "grad_norm": 9.107887268066406, "learning_rate": 1.2745161290322582e-05, "loss": 0.2231, "step": 11549 }, { "epoch": 74.45461847389558, "grad_norm": 4.749131202697754, "learning_rate": 1.2741935483870968e-05, "loss": 0.1127, "step": 11550 }, { "epoch": 74.46104417670682, "grad_norm": 15.354138374328613, "learning_rate": 1.2738709677419355e-05, "loss": 0.2482, "step": 11551 }, { "epoch": 74.46746987951808, "grad_norm": 29.42238426208496, "learning_rate": 1.2735483870967742e-05, "loss": 0.3353, "step": 11552 }, { "epoch": 74.47389558232932, "grad_norm": 19.567943572998047, "learning_rate": 1.2732258064516131e-05, "loss": 0.402, "step": 11553 }, { "epoch": 74.48032128514056, "grad_norm": 7.382462501525879, "learning_rate": 1.2729032258064515e-05, "loss": 0.3757, "step": 11554 }, { "epoch": 74.4867469879518, "grad_norm": 27.378185272216797, "learning_rate": 1.2725806451612905e-05, "loss": 0.4194, "step": 11555 }, { "epoch": 74.49317269076305, "grad_norm": 4.1370134353637695, "learning_rate": 1.2722580645161292e-05, "loss": 0.1881, "step": 11556 }, { "epoch": 74.4995983935743, "grad_norm": 19.874311447143555, "learning_rate": 1.2719354838709676e-05, "loss": 0.359, "step": 11557 }, { "epoch": 74.50602409638554, "grad_norm": 17.785734176635742, "learning_rate": 1.2716129032258065e-05, "loss": 0.2556, "step": 11558 }, { "epoch": 74.51244979919679, "grad_norm": 36.63900375366211, "learning_rate": 1.2712903225806452e-05, "loss": 0.3669, "step": 11559 }, { "epoch": 74.51887550200803, "grad_norm": 8.51177978515625, "learning_rate": 1.270967741935484e-05, "loss": 0.377, "step": 11560 }, { "epoch": 74.52530120481927, "grad_norm": 5.85182523727417, "learning_rate": 1.2706451612903226e-05, "loss": 0.2205, "step": 11561 }, { "epoch": 74.53172690763053, "grad_norm": 22.327878952026367, "learning_rate": 1.2703225806451613e-05, "loss": 0.182, "step": 11562 }, { "epoch": 74.53815261044177, "grad_norm": 8.685479164123535, "learning_rate": 1.27e-05, "loss": 0.1643, "step": 11563 }, { "epoch": 74.54457831325301, "grad_norm": 20.916067123413086, "learning_rate": 1.2696774193548388e-05, "loss": 0.4278, "step": 11564 }, { "epoch": 74.55100401606425, "grad_norm": 26.50739860534668, "learning_rate": 1.2693548387096773e-05, "loss": 0.4411, "step": 11565 }, { "epoch": 74.5574297188755, "grad_norm": 37.11213684082031, "learning_rate": 1.2690322580645161e-05, "loss": 0.363, "step": 11566 }, { "epoch": 74.56385542168675, "grad_norm": 67.53971862792969, "learning_rate": 1.268709677419355e-05, "loss": 0.3282, "step": 11567 }, { "epoch": 74.570281124498, "grad_norm": 20.683387756347656, "learning_rate": 1.2683870967741937e-05, "loss": 0.3982, "step": 11568 }, { "epoch": 74.57670682730924, "grad_norm": 12.622617721557617, "learning_rate": 1.2680645161290321e-05, "loss": 0.3406, "step": 11569 }, { "epoch": 74.58313253012048, "grad_norm": 23.250822067260742, "learning_rate": 1.267741935483871e-05, "loss": 0.524, "step": 11570 }, { "epoch": 74.58955823293172, "grad_norm": 18.241727828979492, "learning_rate": 1.2674193548387098e-05, "loss": 0.5097, "step": 11571 }, { "epoch": 74.59598393574298, "grad_norm": 8.816597938537598, "learning_rate": 1.2670967741935485e-05, "loss": 0.2905, "step": 11572 }, { "epoch": 74.60240963855422, "grad_norm": 13.000894546508789, "learning_rate": 1.2667741935483871e-05, "loss": 0.3552, "step": 11573 }, { "epoch": 74.60883534136546, "grad_norm": 14.983821868896484, "learning_rate": 1.2664516129032258e-05, "loss": 0.2925, "step": 11574 }, { "epoch": 74.6152610441767, "grad_norm": 3.2266650199890137, "learning_rate": 1.2661290322580646e-05, "loss": 0.2953, "step": 11575 }, { "epoch": 74.62168674698795, "grad_norm": 29.084224700927734, "learning_rate": 1.2658064516129033e-05, "loss": 0.568, "step": 11576 }, { "epoch": 74.6281124497992, "grad_norm": 10.312226295471191, "learning_rate": 1.2654838709677419e-05, "loss": 0.3073, "step": 11577 }, { "epoch": 74.63453815261045, "grad_norm": 24.255714416503906, "learning_rate": 1.2651612903225806e-05, "loss": 0.4222, "step": 11578 }, { "epoch": 74.64096385542169, "grad_norm": 4.58524227142334, "learning_rate": 1.2648387096774195e-05, "loss": 0.2817, "step": 11579 }, { "epoch": 74.64738955823293, "grad_norm": 9.884917259216309, "learning_rate": 1.2645161290322583e-05, "loss": 0.352, "step": 11580 }, { "epoch": 74.65381526104417, "grad_norm": 7.6994452476501465, "learning_rate": 1.2641935483870967e-05, "loss": 0.1612, "step": 11581 }, { "epoch": 74.66024096385541, "grad_norm": 17.983144760131836, "learning_rate": 1.2638709677419356e-05, "loss": 0.3167, "step": 11582 }, { "epoch": 74.66666666666667, "grad_norm": 5.981217384338379, "learning_rate": 1.2635483870967743e-05, "loss": 0.264, "step": 11583 }, { "epoch": 74.67309236947791, "grad_norm": 20.57732391357422, "learning_rate": 1.263225806451613e-05, "loss": 0.6789, "step": 11584 }, { "epoch": 74.67951807228916, "grad_norm": 54.10771179199219, "learning_rate": 1.2629032258064516e-05, "loss": 0.4692, "step": 11585 }, { "epoch": 74.6859437751004, "grad_norm": 17.4908447265625, "learning_rate": 1.2625806451612904e-05, "loss": 0.258, "step": 11586 }, { "epoch": 74.69236947791164, "grad_norm": 9.008246421813965, "learning_rate": 1.2622580645161291e-05, "loss": 0.1668, "step": 11587 }, { "epoch": 74.6987951807229, "grad_norm": 6.239560127258301, "learning_rate": 1.2619354838709679e-05, "loss": 0.3704, "step": 11588 }, { "epoch": 74.70522088353414, "grad_norm": 12.039253234863281, "learning_rate": 1.2616129032258064e-05, "loss": 0.3958, "step": 11589 }, { "epoch": 74.71164658634538, "grad_norm": 19.814380645751953, "learning_rate": 1.2612903225806452e-05, "loss": 0.3697, "step": 11590 }, { "epoch": 74.71807228915662, "grad_norm": 19.244949340820312, "learning_rate": 1.260967741935484e-05, "loss": 0.5607, "step": 11591 }, { "epoch": 74.72449799196787, "grad_norm": 32.420528411865234, "learning_rate": 1.2606451612903225e-05, "loss": 0.2488, "step": 11592 }, { "epoch": 74.73092369477912, "grad_norm": 20.876970291137695, "learning_rate": 1.2603225806451612e-05, "loss": 0.3627, "step": 11593 }, { "epoch": 74.73734939759036, "grad_norm": 2.2945144176483154, "learning_rate": 1.2600000000000001e-05, "loss": 0.3083, "step": 11594 }, { "epoch": 74.7437751004016, "grad_norm": 13.93698501586914, "learning_rate": 1.2596774193548389e-05, "loss": 0.5335, "step": 11595 }, { "epoch": 74.75020080321285, "grad_norm": 16.612266540527344, "learning_rate": 1.2593548387096773e-05, "loss": 0.1716, "step": 11596 }, { "epoch": 74.75662650602409, "grad_norm": 7.832287788391113, "learning_rate": 1.2590322580645162e-05, "loss": 0.2944, "step": 11597 }, { "epoch": 74.76305220883535, "grad_norm": 42.96680450439453, "learning_rate": 1.258709677419355e-05, "loss": 0.2254, "step": 11598 }, { "epoch": 74.76947791164659, "grad_norm": 24.25086784362793, "learning_rate": 1.2583870967741937e-05, "loss": 0.2514, "step": 11599 }, { "epoch": 74.77590361445783, "grad_norm": 10.481918334960938, "learning_rate": 1.2580645161290322e-05, "loss": 0.1376, "step": 11600 }, { "epoch": 74.78232931726907, "grad_norm": 30.323976516723633, "learning_rate": 1.257741935483871e-05, "loss": 0.3495, "step": 11601 }, { "epoch": 74.78875502008032, "grad_norm": 30.318248748779297, "learning_rate": 1.2574193548387097e-05, "loss": 0.4602, "step": 11602 }, { "epoch": 74.79518072289157, "grad_norm": 13.461414337158203, "learning_rate": 1.2570967741935486e-05, "loss": 0.435, "step": 11603 }, { "epoch": 74.80160642570281, "grad_norm": 6.42478609085083, "learning_rate": 1.256774193548387e-05, "loss": 0.2801, "step": 11604 }, { "epoch": 74.80803212851406, "grad_norm": 19.274505615234375, "learning_rate": 1.2564516129032258e-05, "loss": 0.3158, "step": 11605 }, { "epoch": 74.8144578313253, "grad_norm": 4.808783054351807, "learning_rate": 1.2561290322580647e-05, "loss": 0.1886, "step": 11606 }, { "epoch": 74.82088353413654, "grad_norm": 10.531542778015137, "learning_rate": 1.2558064516129034e-05, "loss": 0.6846, "step": 11607 }, { "epoch": 74.8273092369478, "grad_norm": 5.553062438964844, "learning_rate": 1.2554838709677418e-05, "loss": 0.3361, "step": 11608 }, { "epoch": 74.83373493975904, "grad_norm": 13.981711387634277, "learning_rate": 1.2551612903225807e-05, "loss": 0.5526, "step": 11609 }, { "epoch": 74.84016064257028, "grad_norm": 12.519765853881836, "learning_rate": 1.2548387096774195e-05, "loss": 0.5959, "step": 11610 }, { "epoch": 74.84658634538152, "grad_norm": 17.942235946655273, "learning_rate": 1.2545161290322582e-05, "loss": 0.3522, "step": 11611 }, { "epoch": 74.85301204819277, "grad_norm": 39.38404083251953, "learning_rate": 1.2541935483870968e-05, "loss": 0.4327, "step": 11612 }, { "epoch": 74.85943775100401, "grad_norm": 8.125992774963379, "learning_rate": 1.2538709677419355e-05, "loss": 0.1731, "step": 11613 }, { "epoch": 74.86586345381527, "grad_norm": 5.960559844970703, "learning_rate": 1.2535483870967743e-05, "loss": 0.1533, "step": 11614 }, { "epoch": 74.87228915662651, "grad_norm": 3.571913719177246, "learning_rate": 1.2532258064516132e-05, "loss": 0.1902, "step": 11615 }, { "epoch": 74.87871485943775, "grad_norm": 5.769121170043945, "learning_rate": 1.2529032258064516e-05, "loss": 0.2333, "step": 11616 }, { "epoch": 74.88514056224899, "grad_norm": 9.812116622924805, "learning_rate": 1.2525806451612903e-05, "loss": 0.3381, "step": 11617 }, { "epoch": 74.89156626506023, "grad_norm": 8.640517234802246, "learning_rate": 1.2522580645161292e-05, "loss": 0.3444, "step": 11618 }, { "epoch": 74.89799196787149, "grad_norm": 59.20040512084961, "learning_rate": 1.251935483870968e-05, "loss": 0.4532, "step": 11619 }, { "epoch": 74.90441767068273, "grad_norm": 12.75652027130127, "learning_rate": 1.2516129032258064e-05, "loss": 0.4157, "step": 11620 }, { "epoch": 74.91084337349398, "grad_norm": 13.089523315429688, "learning_rate": 1.2512903225806453e-05, "loss": 0.3452, "step": 11621 }, { "epoch": 74.91726907630522, "grad_norm": 6.318881988525391, "learning_rate": 1.250967741935484e-05, "loss": 0.4168, "step": 11622 }, { "epoch": 74.92369477911646, "grad_norm": 27.531593322753906, "learning_rate": 1.2506451612903227e-05, "loss": 0.2597, "step": 11623 }, { "epoch": 74.93012048192772, "grad_norm": 15.044590950012207, "learning_rate": 1.2503225806451613e-05, "loss": 0.3539, "step": 11624 }, { "epoch": 74.93654618473896, "grad_norm": 15.170028686523438, "learning_rate": 1.25e-05, "loss": 0.268, "step": 11625 }, { "epoch": 74.9429718875502, "grad_norm": 2.0349037647247314, "learning_rate": 1.2496774193548386e-05, "loss": 0.1248, "step": 11626 }, { "epoch": 74.94939759036144, "grad_norm": 3.1076388359069824, "learning_rate": 1.2493548387096775e-05, "loss": 0.1551, "step": 11627 }, { "epoch": 74.95582329317268, "grad_norm": 61.90915298461914, "learning_rate": 1.2490322580645161e-05, "loss": 0.3837, "step": 11628 }, { "epoch": 74.96224899598394, "grad_norm": 37.26755142211914, "learning_rate": 1.2487096774193549e-05, "loss": 0.4065, "step": 11629 }, { "epoch": 74.96867469879518, "grad_norm": 11.897700309753418, "learning_rate": 1.2483870967741936e-05, "loss": 0.2748, "step": 11630 }, { "epoch": 74.97510040160643, "grad_norm": 9.75853157043457, "learning_rate": 1.2480645161290323e-05, "loss": 0.1564, "step": 11631 }, { "epoch": 74.98152610441767, "grad_norm": 3.340036392211914, "learning_rate": 1.2477419354838709e-05, "loss": 0.5051, "step": 11632 }, { "epoch": 74.98795180722891, "grad_norm": 8.553160667419434, "learning_rate": 1.2474193548387098e-05, "loss": 0.1363, "step": 11633 }, { "epoch": 74.99437751004017, "grad_norm": 9.00985050201416, "learning_rate": 1.2470967741935484e-05, "loss": 0.371, "step": 11634 }, { "epoch": 75.00401606425703, "grad_norm": 16.631912231445312, "learning_rate": 1.2467741935483871e-05, "loss": 0.4821, "step": 11635 }, { "epoch": 75.01044176706827, "grad_norm": 6.395348072052002, "learning_rate": 1.2464516129032259e-05, "loss": 0.2004, "step": 11636 }, { "epoch": 75.01686746987951, "grad_norm": 9.57071590423584, "learning_rate": 1.2461290322580646e-05, "loss": 0.2818, "step": 11637 }, { "epoch": 75.02329317269076, "grad_norm": 19.26232147216797, "learning_rate": 1.2458064516129032e-05, "loss": 0.4435, "step": 11638 }, { "epoch": 75.02971887550201, "grad_norm": 7.9332661628723145, "learning_rate": 1.245483870967742e-05, "loss": 0.2882, "step": 11639 }, { "epoch": 75.03614457831326, "grad_norm": 16.95728302001953, "learning_rate": 1.2451612903225807e-05, "loss": 0.3048, "step": 11640 }, { "epoch": 75.0425702811245, "grad_norm": 10.997352600097656, "learning_rate": 1.2448387096774194e-05, "loss": 0.4797, "step": 11641 }, { "epoch": 75.04899598393574, "grad_norm": 25.212556838989258, "learning_rate": 1.2445161290322581e-05, "loss": 0.5261, "step": 11642 }, { "epoch": 75.05542168674698, "grad_norm": 24.807016372680664, "learning_rate": 1.2441935483870969e-05, "loss": 0.4822, "step": 11643 }, { "epoch": 75.06184738955824, "grad_norm": 17.699176788330078, "learning_rate": 1.2438709677419354e-05, "loss": 0.3208, "step": 11644 }, { "epoch": 75.06827309236948, "grad_norm": 8.06608772277832, "learning_rate": 1.2435483870967744e-05, "loss": 0.4336, "step": 11645 }, { "epoch": 75.07469879518072, "grad_norm": 6.864851474761963, "learning_rate": 1.243225806451613e-05, "loss": 0.2393, "step": 11646 }, { "epoch": 75.08112449799196, "grad_norm": 12.627239227294922, "learning_rate": 1.2429032258064517e-05, "loss": 0.2507, "step": 11647 }, { "epoch": 75.08755020080321, "grad_norm": 19.452266693115234, "learning_rate": 1.2425806451612904e-05, "loss": 0.3266, "step": 11648 }, { "epoch": 75.09397590361446, "grad_norm": 11.314794540405273, "learning_rate": 1.2422580645161291e-05, "loss": 0.5095, "step": 11649 }, { "epoch": 75.1004016064257, "grad_norm": 24.759471893310547, "learning_rate": 1.2419354838709677e-05, "loss": 0.2661, "step": 11650 }, { "epoch": 75.10682730923695, "grad_norm": 2.2317545413970947, "learning_rate": 1.2416129032258066e-05, "loss": 0.2124, "step": 11651 }, { "epoch": 75.11325301204819, "grad_norm": 14.7765531539917, "learning_rate": 1.2412903225806452e-05, "loss": 0.2574, "step": 11652 }, { "epoch": 75.11967871485943, "grad_norm": 14.010074615478516, "learning_rate": 1.240967741935484e-05, "loss": 0.3255, "step": 11653 }, { "epoch": 75.12610441767069, "grad_norm": 6.932831764221191, "learning_rate": 1.2406451612903227e-05, "loss": 0.2364, "step": 11654 }, { "epoch": 75.13253012048193, "grad_norm": 12.672649383544922, "learning_rate": 1.2403225806451614e-05, "loss": 0.3798, "step": 11655 }, { "epoch": 75.13895582329317, "grad_norm": 5.854065895080566, "learning_rate": 1.24e-05, "loss": 0.2509, "step": 11656 }, { "epoch": 75.14538152610442, "grad_norm": 5.648767471313477, "learning_rate": 1.2396774193548389e-05, "loss": 0.1229, "step": 11657 }, { "epoch": 75.15180722891566, "grad_norm": 5.313577175140381, "learning_rate": 1.2393548387096775e-05, "loss": 0.4128, "step": 11658 }, { "epoch": 75.15823293172691, "grad_norm": 20.338109970092773, "learning_rate": 1.2390322580645162e-05, "loss": 0.547, "step": 11659 }, { "epoch": 75.16465863453816, "grad_norm": 2.9373791217803955, "learning_rate": 1.238709677419355e-05, "loss": 0.2485, "step": 11660 }, { "epoch": 75.1710843373494, "grad_norm": 17.269302368164062, "learning_rate": 1.2383870967741935e-05, "loss": 0.3654, "step": 11661 }, { "epoch": 75.17751004016064, "grad_norm": 6.466867923736572, "learning_rate": 1.2380645161290323e-05, "loss": 0.2549, "step": 11662 }, { "epoch": 75.18393574297188, "grad_norm": 12.036404609680176, "learning_rate": 1.237741935483871e-05, "loss": 0.5174, "step": 11663 }, { "epoch": 75.19036144578314, "grad_norm": 50.22600173950195, "learning_rate": 1.2374193548387097e-05, "loss": 0.2246, "step": 11664 }, { "epoch": 75.19678714859438, "grad_norm": 19.81513023376465, "learning_rate": 1.2370967741935483e-05, "loss": 0.3252, "step": 11665 }, { "epoch": 75.20321285140562, "grad_norm": 16.90605926513672, "learning_rate": 1.2367741935483872e-05, "loss": 0.2404, "step": 11666 }, { "epoch": 75.20963855421687, "grad_norm": 17.238080978393555, "learning_rate": 1.2364516129032258e-05, "loss": 0.309, "step": 11667 }, { "epoch": 75.21606425702811, "grad_norm": 13.936511039733887, "learning_rate": 1.2361290322580645e-05, "loss": 0.2835, "step": 11668 }, { "epoch": 75.22248995983935, "grad_norm": 40.23421859741211, "learning_rate": 1.2358064516129033e-05, "loss": 0.4247, "step": 11669 }, { "epoch": 75.2289156626506, "grad_norm": 3.551815986633301, "learning_rate": 1.235483870967742e-05, "loss": 0.3358, "step": 11670 }, { "epoch": 75.23534136546185, "grad_norm": 12.293880462646484, "learning_rate": 1.2351612903225806e-05, "loss": 0.4873, "step": 11671 }, { "epoch": 75.24176706827309, "grad_norm": 65.65019226074219, "learning_rate": 1.2348387096774195e-05, "loss": 0.5813, "step": 11672 }, { "epoch": 75.24819277108433, "grad_norm": 4.1264543533325195, "learning_rate": 1.234516129032258e-05, "loss": 0.3333, "step": 11673 }, { "epoch": 75.25461847389558, "grad_norm": 10.062124252319336, "learning_rate": 1.2341935483870968e-05, "loss": 0.1595, "step": 11674 }, { "epoch": 75.26104417670683, "grad_norm": 20.437091827392578, "learning_rate": 1.2338709677419355e-05, "loss": 0.3972, "step": 11675 }, { "epoch": 75.26746987951807, "grad_norm": 11.178284645080566, "learning_rate": 1.2335483870967743e-05, "loss": 0.2619, "step": 11676 }, { "epoch": 75.27389558232932, "grad_norm": 22.889095306396484, "learning_rate": 1.2332258064516129e-05, "loss": 0.3147, "step": 11677 }, { "epoch": 75.28032128514056, "grad_norm": 27.301664352416992, "learning_rate": 1.2329032258064518e-05, "loss": 0.5763, "step": 11678 }, { "epoch": 75.2867469879518, "grad_norm": 12.12076473236084, "learning_rate": 1.2325806451612903e-05, "loss": 0.2778, "step": 11679 }, { "epoch": 75.29317269076306, "grad_norm": 4.181569576263428, "learning_rate": 1.232258064516129e-05, "loss": 0.2512, "step": 11680 }, { "epoch": 75.2995983935743, "grad_norm": 16.91254234313965, "learning_rate": 1.2319354838709678e-05, "loss": 0.1995, "step": 11681 }, { "epoch": 75.30602409638554, "grad_norm": 24.67270851135254, "learning_rate": 1.2316129032258066e-05, "loss": 0.5748, "step": 11682 }, { "epoch": 75.31244979919678, "grad_norm": 7.685676574707031, "learning_rate": 1.2312903225806451e-05, "loss": 0.3501, "step": 11683 }, { "epoch": 75.31887550200803, "grad_norm": 14.99482536315918, "learning_rate": 1.230967741935484e-05, "loss": 0.4965, "step": 11684 }, { "epoch": 75.32530120481928, "grad_norm": 29.66274642944336, "learning_rate": 1.2306451612903226e-05, "loss": 0.4238, "step": 11685 }, { "epoch": 75.33172690763053, "grad_norm": 11.875168800354004, "learning_rate": 1.2303225806451613e-05, "loss": 0.6036, "step": 11686 }, { "epoch": 75.33815261044177, "grad_norm": 23.324684143066406, "learning_rate": 1.23e-05, "loss": 0.5565, "step": 11687 }, { "epoch": 75.34457831325301, "grad_norm": 4.951862335205078, "learning_rate": 1.2296774193548388e-05, "loss": 0.3245, "step": 11688 }, { "epoch": 75.35100401606425, "grad_norm": 6.122068405151367, "learning_rate": 1.2293548387096774e-05, "loss": 0.2679, "step": 11689 }, { "epoch": 75.35742971887551, "grad_norm": 6.2981438636779785, "learning_rate": 1.2290322580645163e-05, "loss": 0.4101, "step": 11690 }, { "epoch": 75.36385542168675, "grad_norm": 32.11992645263672, "learning_rate": 1.2287096774193549e-05, "loss": 0.4956, "step": 11691 }, { "epoch": 75.37028112449799, "grad_norm": 8.094075202941895, "learning_rate": 1.2283870967741936e-05, "loss": 0.2812, "step": 11692 }, { "epoch": 75.37670682730923, "grad_norm": 6.152370452880859, "learning_rate": 1.2280645161290324e-05, "loss": 0.3231, "step": 11693 }, { "epoch": 75.38313253012048, "grad_norm": 54.42789840698242, "learning_rate": 1.2277419354838711e-05, "loss": 0.657, "step": 11694 }, { "epoch": 75.38955823293173, "grad_norm": 9.790698051452637, "learning_rate": 1.2274193548387097e-05, "loss": 0.489, "step": 11695 }, { "epoch": 75.39598393574298, "grad_norm": 19.267168045043945, "learning_rate": 1.2270967741935486e-05, "loss": 0.363, "step": 11696 }, { "epoch": 75.40240963855422, "grad_norm": 13.054150581359863, "learning_rate": 1.2267741935483871e-05, "loss": 0.3063, "step": 11697 }, { "epoch": 75.40883534136546, "grad_norm": 9.165116310119629, "learning_rate": 1.2264516129032259e-05, "loss": 0.6387, "step": 11698 }, { "epoch": 75.4152610441767, "grad_norm": 8.038104057312012, "learning_rate": 1.2261290322580646e-05, "loss": 0.2876, "step": 11699 }, { "epoch": 75.42168674698796, "grad_norm": 6.478921413421631, "learning_rate": 1.2258064516129032e-05, "loss": 0.4217, "step": 11700 }, { "epoch": 75.4281124497992, "grad_norm": 22.03668975830078, "learning_rate": 1.225483870967742e-05, "loss": 0.4123, "step": 11701 }, { "epoch": 75.43453815261044, "grad_norm": 5.594086170196533, "learning_rate": 1.2251612903225807e-05, "loss": 0.3104, "step": 11702 }, { "epoch": 75.44096385542169, "grad_norm": 6.233170509338379, "learning_rate": 1.2248387096774194e-05, "loss": 0.5797, "step": 11703 }, { "epoch": 75.44738955823293, "grad_norm": 10.306283950805664, "learning_rate": 1.224516129032258e-05, "loss": 0.4855, "step": 11704 }, { "epoch": 75.45381526104417, "grad_norm": 21.698606491088867, "learning_rate": 1.2241935483870969e-05, "loss": 0.3204, "step": 11705 }, { "epoch": 75.46024096385543, "grad_norm": 16.12855339050293, "learning_rate": 1.2238709677419355e-05, "loss": 0.58, "step": 11706 }, { "epoch": 75.46666666666667, "grad_norm": 28.990896224975586, "learning_rate": 1.2235483870967742e-05, "loss": 0.3594, "step": 11707 }, { "epoch": 75.47309236947791, "grad_norm": 22.155208587646484, "learning_rate": 1.223225806451613e-05, "loss": 0.3203, "step": 11708 }, { "epoch": 75.47951807228915, "grad_norm": 7.0428786277771, "learning_rate": 1.2229032258064517e-05, "loss": 0.3306, "step": 11709 }, { "epoch": 75.4859437751004, "grad_norm": 9.512901306152344, "learning_rate": 1.2225806451612903e-05, "loss": 0.2478, "step": 11710 }, { "epoch": 75.49236947791165, "grad_norm": 8.14437484741211, "learning_rate": 1.2222580645161292e-05, "loss": 0.5353, "step": 11711 }, { "epoch": 75.4987951807229, "grad_norm": 15.947444915771484, "learning_rate": 1.2219354838709677e-05, "loss": 0.4804, "step": 11712 }, { "epoch": 75.50522088353414, "grad_norm": 10.366927146911621, "learning_rate": 1.2216129032258065e-05, "loss": 0.508, "step": 11713 }, { "epoch": 75.51164658634538, "grad_norm": 3.6590704917907715, "learning_rate": 1.2212903225806452e-05, "loss": 0.4633, "step": 11714 }, { "epoch": 75.51807228915662, "grad_norm": 6.167420864105225, "learning_rate": 1.220967741935484e-05, "loss": 0.2981, "step": 11715 }, { "epoch": 75.52449799196788, "grad_norm": 11.79883861541748, "learning_rate": 1.2206451612903225e-05, "loss": 0.3849, "step": 11716 }, { "epoch": 75.53092369477912, "grad_norm": 2.9745211601257324, "learning_rate": 1.2203225806451614e-05, "loss": 0.1298, "step": 11717 }, { "epoch": 75.53734939759036, "grad_norm": 3.9464035034179688, "learning_rate": 1.22e-05, "loss": 0.3029, "step": 11718 }, { "epoch": 75.5437751004016, "grad_norm": 25.072816848754883, "learning_rate": 1.2196774193548387e-05, "loss": 0.1941, "step": 11719 }, { "epoch": 75.55020080321285, "grad_norm": 6.230973720550537, "learning_rate": 1.2193548387096775e-05, "loss": 0.3689, "step": 11720 }, { "epoch": 75.5566265060241, "grad_norm": 10.924762725830078, "learning_rate": 1.2190322580645162e-05, "loss": 0.4391, "step": 11721 }, { "epoch": 75.56305220883534, "grad_norm": 4.322564125061035, "learning_rate": 1.2187096774193548e-05, "loss": 0.1502, "step": 11722 }, { "epoch": 75.56947791164659, "grad_norm": 8.416579246520996, "learning_rate": 1.2183870967741937e-05, "loss": 0.2739, "step": 11723 }, { "epoch": 75.57590361445783, "grad_norm": 27.226802825927734, "learning_rate": 1.2180645161290323e-05, "loss": 0.2477, "step": 11724 }, { "epoch": 75.58232931726907, "grad_norm": 11.474567413330078, "learning_rate": 1.217741935483871e-05, "loss": 0.3597, "step": 11725 }, { "epoch": 75.58875502008033, "grad_norm": 6.605042457580566, "learning_rate": 1.2174193548387098e-05, "loss": 0.3093, "step": 11726 }, { "epoch": 75.59518072289157, "grad_norm": 22.390117645263672, "learning_rate": 1.2170967741935485e-05, "loss": 0.4054, "step": 11727 }, { "epoch": 75.60160642570281, "grad_norm": 14.055712699890137, "learning_rate": 1.216774193548387e-05, "loss": 0.4626, "step": 11728 }, { "epoch": 75.60803212851405, "grad_norm": 16.849376678466797, "learning_rate": 1.216451612903226e-05, "loss": 0.6217, "step": 11729 }, { "epoch": 75.6144578313253, "grad_norm": 15.167166709899902, "learning_rate": 1.2161290322580646e-05, "loss": 0.3434, "step": 11730 }, { "epoch": 75.62088353413655, "grad_norm": 3.293659210205078, "learning_rate": 1.2158064516129033e-05, "loss": 0.218, "step": 11731 }, { "epoch": 75.6273092369478, "grad_norm": 13.984874725341797, "learning_rate": 1.215483870967742e-05, "loss": 0.4617, "step": 11732 }, { "epoch": 75.63373493975904, "grad_norm": 3.4665112495422363, "learning_rate": 1.2151612903225806e-05, "loss": 0.1671, "step": 11733 }, { "epoch": 75.64016064257028, "grad_norm": 52.074806213378906, "learning_rate": 1.2148387096774193e-05, "loss": 0.3327, "step": 11734 }, { "epoch": 75.64658634538152, "grad_norm": 5.482022285461426, "learning_rate": 1.214516129032258e-05, "loss": 0.2221, "step": 11735 }, { "epoch": 75.65301204819278, "grad_norm": 52.08237075805664, "learning_rate": 1.2141935483870968e-05, "loss": 0.4875, "step": 11736 }, { "epoch": 75.65943775100402, "grad_norm": 38.10751724243164, "learning_rate": 1.2138709677419356e-05, "loss": 0.5696, "step": 11737 }, { "epoch": 75.66586345381526, "grad_norm": 11.488741874694824, "learning_rate": 1.2135483870967743e-05, "loss": 0.2526, "step": 11738 }, { "epoch": 75.6722891566265, "grad_norm": 6.284101486206055, "learning_rate": 1.2132258064516129e-05, "loss": 0.4524, "step": 11739 }, { "epoch": 75.67871485943775, "grad_norm": 14.67668342590332, "learning_rate": 1.2129032258064516e-05, "loss": 0.3108, "step": 11740 }, { "epoch": 75.68514056224899, "grad_norm": 19.985605239868164, "learning_rate": 1.2125806451612904e-05, "loss": 0.465, "step": 11741 }, { "epoch": 75.69156626506025, "grad_norm": 14.785624504089355, "learning_rate": 1.2122580645161291e-05, "loss": 0.3761, "step": 11742 }, { "epoch": 75.69799196787149, "grad_norm": 8.866311073303223, "learning_rate": 1.2119354838709678e-05, "loss": 0.3137, "step": 11743 }, { "epoch": 75.70441767068273, "grad_norm": 18.68756866455078, "learning_rate": 1.2116129032258066e-05, "loss": 0.4166, "step": 11744 }, { "epoch": 75.71084337349397, "grad_norm": 22.9813289642334, "learning_rate": 1.2112903225806451e-05, "loss": 0.3107, "step": 11745 }, { "epoch": 75.71726907630521, "grad_norm": 26.200153350830078, "learning_rate": 1.2109677419354839e-05, "loss": 0.4566, "step": 11746 }, { "epoch": 75.72369477911647, "grad_norm": 22.65890884399414, "learning_rate": 1.2106451612903226e-05, "loss": 0.2906, "step": 11747 }, { "epoch": 75.73012048192771, "grad_norm": 6.196310997009277, "learning_rate": 1.2103225806451614e-05, "loss": 0.3413, "step": 11748 }, { "epoch": 75.73654618473896, "grad_norm": 23.91090965270996, "learning_rate": 1.2100000000000001e-05, "loss": 0.6094, "step": 11749 }, { "epoch": 75.7429718875502, "grad_norm": 3.247143507003784, "learning_rate": 1.2096774193548388e-05, "loss": 0.1625, "step": 11750 }, { "epoch": 75.74939759036144, "grad_norm": 8.519095420837402, "learning_rate": 1.2093548387096774e-05, "loss": 0.5039, "step": 11751 }, { "epoch": 75.7558232931727, "grad_norm": 5.639012813568115, "learning_rate": 1.2090322580645162e-05, "loss": 0.0989, "step": 11752 }, { "epoch": 75.76224899598394, "grad_norm": 9.47154426574707, "learning_rate": 1.2087096774193549e-05, "loss": 0.202, "step": 11753 }, { "epoch": 75.76867469879518, "grad_norm": 50.40522003173828, "learning_rate": 1.2083870967741936e-05, "loss": 0.4247, "step": 11754 }, { "epoch": 75.77510040160642, "grad_norm": 64.52552795410156, "learning_rate": 1.2080645161290324e-05, "loss": 0.4638, "step": 11755 }, { "epoch": 75.78152610441767, "grad_norm": 8.470540046691895, "learning_rate": 1.2077419354838711e-05, "loss": 0.4151, "step": 11756 }, { "epoch": 75.78795180722892, "grad_norm": 4.574779510498047, "learning_rate": 1.2074193548387097e-05, "loss": 0.1981, "step": 11757 }, { "epoch": 75.79437751004016, "grad_norm": 28.754119873046875, "learning_rate": 1.2070967741935484e-05, "loss": 0.2585, "step": 11758 }, { "epoch": 75.8008032128514, "grad_norm": 71.98648071289062, "learning_rate": 1.2067741935483872e-05, "loss": 0.4412, "step": 11759 }, { "epoch": 75.80722891566265, "grad_norm": 14.67486572265625, "learning_rate": 1.2064516129032259e-05, "loss": 0.3662, "step": 11760 }, { "epoch": 75.81365461847389, "grad_norm": 41.33049774169922, "learning_rate": 1.2061290322580646e-05, "loss": 0.4456, "step": 11761 }, { "epoch": 75.82008032128515, "grad_norm": 26.411710739135742, "learning_rate": 1.2058064516129034e-05, "loss": 0.1947, "step": 11762 }, { "epoch": 75.82650602409639, "grad_norm": 7.407994747161865, "learning_rate": 1.205483870967742e-05, "loss": 0.3003, "step": 11763 }, { "epoch": 75.83293172690763, "grad_norm": 13.477943420410156, "learning_rate": 1.2051612903225807e-05, "loss": 0.5213, "step": 11764 }, { "epoch": 75.83935742971887, "grad_norm": 4.377828598022461, "learning_rate": 1.2048387096774194e-05, "loss": 0.1984, "step": 11765 }, { "epoch": 75.84578313253012, "grad_norm": 7.144232273101807, "learning_rate": 1.2045161290322582e-05, "loss": 0.3402, "step": 11766 }, { "epoch": 75.85220883534137, "grad_norm": 6.337428092956543, "learning_rate": 1.2041935483870967e-05, "loss": 0.1324, "step": 11767 }, { "epoch": 75.85863453815261, "grad_norm": 2.9135582447052, "learning_rate": 1.2038709677419355e-05, "loss": 0.2843, "step": 11768 }, { "epoch": 75.86506024096386, "grad_norm": 49.29400634765625, "learning_rate": 1.2035483870967742e-05, "loss": 0.1336, "step": 11769 }, { "epoch": 75.8714859437751, "grad_norm": 33.618953704833984, "learning_rate": 1.203225806451613e-05, "loss": 0.3369, "step": 11770 }, { "epoch": 75.87791164658634, "grad_norm": 20.15214729309082, "learning_rate": 1.2029032258064517e-05, "loss": 0.5504, "step": 11771 }, { "epoch": 75.88433734939758, "grad_norm": 10.516805648803711, "learning_rate": 1.2025806451612903e-05, "loss": 0.324, "step": 11772 }, { "epoch": 75.89076305220884, "grad_norm": 7.391190052032471, "learning_rate": 1.202258064516129e-05, "loss": 0.4871, "step": 11773 }, { "epoch": 75.89718875502008, "grad_norm": 4.597283363342285, "learning_rate": 1.2019354838709678e-05, "loss": 0.3309, "step": 11774 }, { "epoch": 75.90361445783132, "grad_norm": 35.95001220703125, "learning_rate": 1.2016129032258065e-05, "loss": 0.3089, "step": 11775 }, { "epoch": 75.91004016064257, "grad_norm": 2.858374834060669, "learning_rate": 1.2012903225806452e-05, "loss": 0.1515, "step": 11776 }, { "epoch": 75.91646586345381, "grad_norm": 18.9863338470459, "learning_rate": 1.200967741935484e-05, "loss": 0.534, "step": 11777 }, { "epoch": 75.92289156626506, "grad_norm": 13.574174880981445, "learning_rate": 1.2006451612903226e-05, "loss": 0.2167, "step": 11778 }, { "epoch": 75.92931726907631, "grad_norm": 33.444278717041016, "learning_rate": 1.2003225806451613e-05, "loss": 0.553, "step": 11779 }, { "epoch": 75.93574297188755, "grad_norm": 10.29061508178711, "learning_rate": 1.2e-05, "loss": 0.6449, "step": 11780 }, { "epoch": 75.94216867469879, "grad_norm": 15.575132369995117, "learning_rate": 1.1996774193548388e-05, "loss": 0.2271, "step": 11781 }, { "epoch": 75.94859437751003, "grad_norm": 25.181636810302734, "learning_rate": 1.1993548387096775e-05, "loss": 0.3043, "step": 11782 }, { "epoch": 75.95502008032129, "grad_norm": 16.899242401123047, "learning_rate": 1.1990322580645163e-05, "loss": 0.322, "step": 11783 }, { "epoch": 75.96144578313253, "grad_norm": 10.054367065429688, "learning_rate": 1.1987096774193548e-05, "loss": 0.2788, "step": 11784 }, { "epoch": 75.96787148594377, "grad_norm": 2.370502471923828, "learning_rate": 1.1983870967741936e-05, "loss": 0.153, "step": 11785 }, { "epoch": 75.97429718875502, "grad_norm": 17.594459533691406, "learning_rate": 1.1980645161290323e-05, "loss": 0.3373, "step": 11786 }, { "epoch": 75.98072289156626, "grad_norm": 5.892359733581543, "learning_rate": 1.197741935483871e-05, "loss": 0.3853, "step": 11787 }, { "epoch": 75.98714859437752, "grad_norm": 8.935941696166992, "learning_rate": 1.1974193548387098e-05, "loss": 0.2767, "step": 11788 }, { "epoch": 75.99357429718876, "grad_norm": 33.17708206176758, "learning_rate": 1.1970967741935485e-05, "loss": 0.1728, "step": 11789 }, { "epoch": 76.00321285140562, "grad_norm": 6.490870475769043, "learning_rate": 1.1967741935483871e-05, "loss": 0.3013, "step": 11790 }, { "epoch": 76.00963855421686, "grad_norm": 4.3857197761535645, "learning_rate": 1.1964516129032258e-05, "loss": 0.3842, "step": 11791 }, { "epoch": 76.0160642570281, "grad_norm": 11.252339363098145, "learning_rate": 1.1961290322580646e-05, "loss": 0.1335, "step": 11792 }, { "epoch": 76.02248995983936, "grad_norm": 8.842610359191895, "learning_rate": 1.1958064516129033e-05, "loss": 0.3227, "step": 11793 }, { "epoch": 76.0289156626506, "grad_norm": 55.57204055786133, "learning_rate": 1.195483870967742e-05, "loss": 0.3051, "step": 11794 }, { "epoch": 76.03534136546185, "grad_norm": 12.476533889770508, "learning_rate": 1.1951612903225808e-05, "loss": 0.6817, "step": 11795 }, { "epoch": 76.04176706827309, "grad_norm": 26.57796859741211, "learning_rate": 1.1948387096774194e-05, "loss": 0.5332, "step": 11796 }, { "epoch": 76.04819277108433, "grad_norm": 19.101600646972656, "learning_rate": 1.1945161290322581e-05, "loss": 0.4654, "step": 11797 }, { "epoch": 76.05461847389559, "grad_norm": 8.386314392089844, "learning_rate": 1.1941935483870968e-05, "loss": 0.2581, "step": 11798 }, { "epoch": 76.06104417670683, "grad_norm": 13.546136856079102, "learning_rate": 1.1938709677419356e-05, "loss": 0.356, "step": 11799 }, { "epoch": 76.06746987951807, "grad_norm": 32.65681076049805, "learning_rate": 1.1935483870967743e-05, "loss": 0.4978, "step": 11800 }, { "epoch": 76.07389558232931, "grad_norm": 11.591947555541992, "learning_rate": 1.193225806451613e-05, "loss": 0.1919, "step": 11801 }, { "epoch": 76.08032128514056, "grad_norm": 20.036544799804688, "learning_rate": 1.1929032258064516e-05, "loss": 0.4695, "step": 11802 }, { "epoch": 76.08674698795181, "grad_norm": 14.955492973327637, "learning_rate": 1.1925806451612904e-05, "loss": 0.5416, "step": 11803 }, { "epoch": 76.09317269076305, "grad_norm": 14.924878120422363, "learning_rate": 1.1922580645161291e-05, "loss": 0.3118, "step": 11804 }, { "epoch": 76.0995983935743, "grad_norm": 11.551287651062012, "learning_rate": 1.1919354838709677e-05, "loss": 0.6713, "step": 11805 }, { "epoch": 76.10602409638554, "grad_norm": 2.894089698791504, "learning_rate": 1.1916129032258066e-05, "loss": 0.219, "step": 11806 }, { "epoch": 76.11244979919678, "grad_norm": 26.072643280029297, "learning_rate": 1.1912903225806452e-05, "loss": 0.5239, "step": 11807 }, { "epoch": 76.11887550200804, "grad_norm": 5.514174938201904, "learning_rate": 1.1909677419354839e-05, "loss": 0.361, "step": 11808 }, { "epoch": 76.12530120481928, "grad_norm": 21.433074951171875, "learning_rate": 1.1906451612903226e-05, "loss": 0.462, "step": 11809 }, { "epoch": 76.13172690763052, "grad_norm": 5.963779449462891, "learning_rate": 1.1903225806451614e-05, "loss": 0.139, "step": 11810 }, { "epoch": 76.13815261044176, "grad_norm": 7.9572038650512695, "learning_rate": 1.19e-05, "loss": 0.3087, "step": 11811 }, { "epoch": 76.144578313253, "grad_norm": 8.192914962768555, "learning_rate": 1.1896774193548389e-05, "loss": 0.6668, "step": 11812 }, { "epoch": 76.15100401606426, "grad_norm": 14.279645919799805, "learning_rate": 1.1893548387096774e-05, "loss": 0.1814, "step": 11813 }, { "epoch": 76.1574297188755, "grad_norm": 7.694892406463623, "learning_rate": 1.1890322580645162e-05, "loss": 0.1644, "step": 11814 }, { "epoch": 76.16385542168675, "grad_norm": 20.04719352722168, "learning_rate": 1.188709677419355e-05, "loss": 0.5869, "step": 11815 }, { "epoch": 76.17028112449799, "grad_norm": 14.133217811584473, "learning_rate": 1.1883870967741937e-05, "loss": 0.4376, "step": 11816 }, { "epoch": 76.17670682730923, "grad_norm": 10.079792976379395, "learning_rate": 1.1880645161290322e-05, "loss": 0.3304, "step": 11817 }, { "epoch": 76.18313253012049, "grad_norm": 12.095648765563965, "learning_rate": 1.1877419354838711e-05, "loss": 0.3744, "step": 11818 }, { "epoch": 76.18955823293173, "grad_norm": 24.447813034057617, "learning_rate": 1.1874193548387097e-05, "loss": 0.2616, "step": 11819 }, { "epoch": 76.19598393574297, "grad_norm": 3.688683032989502, "learning_rate": 1.1870967741935484e-05, "loss": 0.0956, "step": 11820 }, { "epoch": 76.20240963855422, "grad_norm": 4.057191848754883, "learning_rate": 1.1867741935483872e-05, "loss": 0.3369, "step": 11821 }, { "epoch": 76.20883534136546, "grad_norm": 12.014185905456543, "learning_rate": 1.186451612903226e-05, "loss": 0.1164, "step": 11822 }, { "epoch": 76.21526104417671, "grad_norm": 14.485177040100098, "learning_rate": 1.1861290322580645e-05, "loss": 0.23, "step": 11823 }, { "epoch": 76.22168674698796, "grad_norm": 8.908011436462402, "learning_rate": 1.1858064516129034e-05, "loss": 0.3125, "step": 11824 }, { "epoch": 76.2281124497992, "grad_norm": 36.93851089477539, "learning_rate": 1.185483870967742e-05, "loss": 0.2539, "step": 11825 }, { "epoch": 76.23453815261044, "grad_norm": 9.55447006225586, "learning_rate": 1.1851612903225807e-05, "loss": 0.3728, "step": 11826 }, { "epoch": 76.24096385542168, "grad_norm": 5.859768390655518, "learning_rate": 1.1848387096774195e-05, "loss": 0.315, "step": 11827 }, { "epoch": 76.24738955823292, "grad_norm": 9.002429008483887, "learning_rate": 1.1845161290322582e-05, "loss": 0.0831, "step": 11828 }, { "epoch": 76.25381526104418, "grad_norm": 7.034671783447266, "learning_rate": 1.1841935483870968e-05, "loss": 0.2431, "step": 11829 }, { "epoch": 76.26024096385542, "grad_norm": 26.213726043701172, "learning_rate": 1.1838709677419355e-05, "loss": 0.4508, "step": 11830 }, { "epoch": 76.26666666666667, "grad_norm": 9.145035743713379, "learning_rate": 1.1835483870967743e-05, "loss": 0.3944, "step": 11831 }, { "epoch": 76.27309236947791, "grad_norm": 25.603025436401367, "learning_rate": 1.183225806451613e-05, "loss": 0.3769, "step": 11832 }, { "epoch": 76.27951807228915, "grad_norm": 35.986568450927734, "learning_rate": 1.1829032258064517e-05, "loss": 0.4934, "step": 11833 }, { "epoch": 76.2859437751004, "grad_norm": 17.130666732788086, "learning_rate": 1.1825806451612905e-05, "loss": 0.083, "step": 11834 }, { "epoch": 76.29236947791165, "grad_norm": 13.945481300354004, "learning_rate": 1.182258064516129e-05, "loss": 0.3857, "step": 11835 }, { "epoch": 76.29879518072289, "grad_norm": 10.46998405456543, "learning_rate": 1.1819354838709678e-05, "loss": 0.4162, "step": 11836 }, { "epoch": 76.30522088353413, "grad_norm": 15.919310569763184, "learning_rate": 1.1816129032258065e-05, "loss": 0.354, "step": 11837 }, { "epoch": 76.31164658634538, "grad_norm": 9.257962226867676, "learning_rate": 1.1812903225806453e-05, "loss": 0.4643, "step": 11838 }, { "epoch": 76.31807228915663, "grad_norm": 15.385805130004883, "learning_rate": 1.180967741935484e-05, "loss": 0.3632, "step": 11839 }, { "epoch": 76.32449799196787, "grad_norm": 4.760012149810791, "learning_rate": 1.1806451612903226e-05, "loss": 0.267, "step": 11840 }, { "epoch": 76.33092369477912, "grad_norm": 6.099640369415283, "learning_rate": 1.1803225806451613e-05, "loss": 0.1265, "step": 11841 }, { "epoch": 76.33734939759036, "grad_norm": 12.260600090026855, "learning_rate": 1.18e-05, "loss": 0.3179, "step": 11842 }, { "epoch": 76.3437751004016, "grad_norm": 4.140135288238525, "learning_rate": 1.1796774193548388e-05, "loss": 0.1425, "step": 11843 }, { "epoch": 76.35020080321286, "grad_norm": 15.92506217956543, "learning_rate": 1.1793548387096774e-05, "loss": 0.3991, "step": 11844 }, { "epoch": 76.3566265060241, "grad_norm": 6.707594871520996, "learning_rate": 1.1790322580645163e-05, "loss": 0.1723, "step": 11845 }, { "epoch": 76.36305220883534, "grad_norm": 18.906339645385742, "learning_rate": 1.1787096774193548e-05, "loss": 0.4108, "step": 11846 }, { "epoch": 76.36947791164658, "grad_norm": 3.984469413757324, "learning_rate": 1.1783870967741936e-05, "loss": 0.2486, "step": 11847 }, { "epoch": 76.37590361445783, "grad_norm": 14.8101224899292, "learning_rate": 1.1780645161290323e-05, "loss": 0.2866, "step": 11848 }, { "epoch": 76.38232931726908, "grad_norm": 25.59548568725586, "learning_rate": 1.177741935483871e-05, "loss": 0.4922, "step": 11849 }, { "epoch": 76.38875502008032, "grad_norm": 11.650177955627441, "learning_rate": 1.1774193548387096e-05, "loss": 0.4285, "step": 11850 }, { "epoch": 76.39518072289157, "grad_norm": 4.82424783706665, "learning_rate": 1.1770967741935485e-05, "loss": 0.2678, "step": 11851 }, { "epoch": 76.40160642570281, "grad_norm": 17.15419578552246, "learning_rate": 1.1767741935483871e-05, "loss": 0.4268, "step": 11852 }, { "epoch": 76.40803212851405, "grad_norm": 23.65662384033203, "learning_rate": 1.1764516129032259e-05, "loss": 0.3252, "step": 11853 }, { "epoch": 76.41445783132531, "grad_norm": 19.961166381835938, "learning_rate": 1.1761290322580646e-05, "loss": 0.1813, "step": 11854 }, { "epoch": 76.42088353413655, "grad_norm": 5.3306756019592285, "learning_rate": 1.1758064516129033e-05, "loss": 0.4233, "step": 11855 }, { "epoch": 76.42730923694779, "grad_norm": 6.795158863067627, "learning_rate": 1.1754838709677419e-05, "loss": 0.3795, "step": 11856 }, { "epoch": 76.43373493975903, "grad_norm": 9.088838577270508, "learning_rate": 1.1751612903225808e-05, "loss": 0.3178, "step": 11857 }, { "epoch": 76.44016064257028, "grad_norm": 20.663724899291992, "learning_rate": 1.1748387096774194e-05, "loss": 0.5844, "step": 11858 }, { "epoch": 76.44658634538153, "grad_norm": 31.67595672607422, "learning_rate": 1.1745161290322581e-05, "loss": 0.256, "step": 11859 }, { "epoch": 76.45301204819278, "grad_norm": 12.689080238342285, "learning_rate": 1.1741935483870969e-05, "loss": 0.4112, "step": 11860 }, { "epoch": 76.45943775100402, "grad_norm": 10.577312469482422, "learning_rate": 1.1738709677419356e-05, "loss": 0.3657, "step": 11861 }, { "epoch": 76.46586345381526, "grad_norm": 11.231584548950195, "learning_rate": 1.1735483870967742e-05, "loss": 0.2967, "step": 11862 }, { "epoch": 76.4722891566265, "grad_norm": 57.66011428833008, "learning_rate": 1.1732258064516131e-05, "loss": 0.3644, "step": 11863 }, { "epoch": 76.47871485943774, "grad_norm": 22.89069938659668, "learning_rate": 1.1729032258064517e-05, "loss": 0.4909, "step": 11864 }, { "epoch": 76.485140562249, "grad_norm": 12.818914413452148, "learning_rate": 1.1725806451612904e-05, "loss": 0.2711, "step": 11865 }, { "epoch": 76.49156626506024, "grad_norm": 49.43728256225586, "learning_rate": 1.1722580645161291e-05, "loss": 0.3878, "step": 11866 }, { "epoch": 76.49799196787149, "grad_norm": 9.24327564239502, "learning_rate": 1.1719354838709679e-05, "loss": 0.248, "step": 11867 }, { "epoch": 76.50441767068273, "grad_norm": 17.43622589111328, "learning_rate": 1.1716129032258064e-05, "loss": 0.4071, "step": 11868 }, { "epoch": 76.51084337349397, "grad_norm": 9.44344711303711, "learning_rate": 1.1712903225806454e-05, "loss": 0.4971, "step": 11869 }, { "epoch": 76.51726907630523, "grad_norm": 10.632987976074219, "learning_rate": 1.170967741935484e-05, "loss": 0.6283, "step": 11870 }, { "epoch": 76.52369477911647, "grad_norm": 36.728389739990234, "learning_rate": 1.1706451612903227e-05, "loss": 0.5189, "step": 11871 }, { "epoch": 76.53012048192771, "grad_norm": 197.85064697265625, "learning_rate": 1.1703225806451614e-05, "loss": 0.4161, "step": 11872 }, { "epoch": 76.53654618473895, "grad_norm": 6.306184768676758, "learning_rate": 1.1700000000000001e-05, "loss": 0.1422, "step": 11873 }, { "epoch": 76.5429718875502, "grad_norm": 17.08009910583496, "learning_rate": 1.1696774193548387e-05, "loss": 0.2258, "step": 11874 }, { "epoch": 76.54939759036145, "grad_norm": 14.22372055053711, "learning_rate": 1.1693548387096775e-05, "loss": 0.294, "step": 11875 }, { "epoch": 76.5558232931727, "grad_norm": 15.182585716247559, "learning_rate": 1.1690322580645162e-05, "loss": 0.3105, "step": 11876 }, { "epoch": 76.56224899598394, "grad_norm": 6.786001205444336, "learning_rate": 1.1687096774193548e-05, "loss": 0.3346, "step": 11877 }, { "epoch": 76.56867469879518, "grad_norm": 11.752196311950684, "learning_rate": 1.1683870967741937e-05, "loss": 0.2967, "step": 11878 }, { "epoch": 76.57510040160642, "grad_norm": 10.791647911071777, "learning_rate": 1.1680645161290322e-05, "loss": 0.299, "step": 11879 }, { "epoch": 76.58152610441768, "grad_norm": 5.934676170349121, "learning_rate": 1.167741935483871e-05, "loss": 0.565, "step": 11880 }, { "epoch": 76.58795180722892, "grad_norm": 5.584198474884033, "learning_rate": 1.1674193548387097e-05, "loss": 0.2659, "step": 11881 }, { "epoch": 76.59437751004016, "grad_norm": 14.97862434387207, "learning_rate": 1.1670967741935485e-05, "loss": 0.4374, "step": 11882 }, { "epoch": 76.6008032128514, "grad_norm": 4.679862976074219, "learning_rate": 1.166774193548387e-05, "loss": 0.3784, "step": 11883 }, { "epoch": 76.60722891566265, "grad_norm": 40.79656219482422, "learning_rate": 1.166451612903226e-05, "loss": 0.4363, "step": 11884 }, { "epoch": 76.6136546184739, "grad_norm": 23.113895416259766, "learning_rate": 1.1661290322580645e-05, "loss": 0.3194, "step": 11885 }, { "epoch": 76.62008032128514, "grad_norm": 11.006299018859863, "learning_rate": 1.1658064516129033e-05, "loss": 0.4906, "step": 11886 }, { "epoch": 76.62650602409639, "grad_norm": 31.979244232177734, "learning_rate": 1.165483870967742e-05, "loss": 0.2286, "step": 11887 }, { "epoch": 76.63293172690763, "grad_norm": 8.819523811340332, "learning_rate": 1.1651612903225807e-05, "loss": 0.4148, "step": 11888 }, { "epoch": 76.63935742971887, "grad_norm": 13.11689567565918, "learning_rate": 1.1648387096774193e-05, "loss": 0.3418, "step": 11889 }, { "epoch": 76.64578313253013, "grad_norm": 18.73499298095703, "learning_rate": 1.1645161290322582e-05, "loss": 0.4574, "step": 11890 }, { "epoch": 76.65220883534137, "grad_norm": 5.848644256591797, "learning_rate": 1.1641935483870968e-05, "loss": 0.3367, "step": 11891 }, { "epoch": 76.65863453815261, "grad_norm": 23.09615707397461, "learning_rate": 1.1638709677419355e-05, "loss": 0.4571, "step": 11892 }, { "epoch": 76.66506024096385, "grad_norm": 46.068214416503906, "learning_rate": 1.1635483870967743e-05, "loss": 0.5823, "step": 11893 }, { "epoch": 76.6714859437751, "grad_norm": 11.6139554977417, "learning_rate": 1.163225806451613e-05, "loss": 0.4936, "step": 11894 }, { "epoch": 76.67791164658635, "grad_norm": 319.3323669433594, "learning_rate": 1.1629032258064516e-05, "loss": 0.2346, "step": 11895 }, { "epoch": 76.6843373493976, "grad_norm": 9.941960334777832, "learning_rate": 1.1625806451612905e-05, "loss": 0.3343, "step": 11896 }, { "epoch": 76.69076305220884, "grad_norm": 14.008718490600586, "learning_rate": 1.162258064516129e-05, "loss": 0.2364, "step": 11897 }, { "epoch": 76.69718875502008, "grad_norm": 4.942590713500977, "learning_rate": 1.1619354838709678e-05, "loss": 0.1723, "step": 11898 }, { "epoch": 76.70361445783132, "grad_norm": 23.919240951538086, "learning_rate": 1.1616129032258065e-05, "loss": 0.4917, "step": 11899 }, { "epoch": 76.71004016064256, "grad_norm": 6.358809471130371, "learning_rate": 1.1612903225806453e-05, "loss": 0.2525, "step": 11900 }, { "epoch": 76.71646586345382, "grad_norm": 10.650776863098145, "learning_rate": 1.1609677419354839e-05, "loss": 0.3271, "step": 11901 }, { "epoch": 76.72289156626506, "grad_norm": 2.005423069000244, "learning_rate": 1.1606451612903228e-05, "loss": 0.1894, "step": 11902 }, { "epoch": 76.7293172690763, "grad_norm": 14.723344802856445, "learning_rate": 1.1603225806451613e-05, "loss": 0.513, "step": 11903 }, { "epoch": 76.73574297188755, "grad_norm": 22.23870086669922, "learning_rate": 1.16e-05, "loss": 0.2945, "step": 11904 }, { "epoch": 76.74216867469879, "grad_norm": 12.880244255065918, "learning_rate": 1.1596774193548388e-05, "loss": 0.5068, "step": 11905 }, { "epoch": 76.74859437751005, "grad_norm": 8.91006851196289, "learning_rate": 1.1593548387096776e-05, "loss": 0.3968, "step": 11906 }, { "epoch": 76.75502008032129, "grad_norm": 49.911407470703125, "learning_rate": 1.1590322580645161e-05, "loss": 0.4496, "step": 11907 }, { "epoch": 76.76144578313253, "grad_norm": 32.776084899902344, "learning_rate": 1.158709677419355e-05, "loss": 0.2152, "step": 11908 }, { "epoch": 76.76787148594377, "grad_norm": 8.270224571228027, "learning_rate": 1.1583870967741936e-05, "loss": 0.4652, "step": 11909 }, { "epoch": 76.77429718875501, "grad_norm": 4.970918655395508, "learning_rate": 1.1580645161290322e-05, "loss": 0.3501, "step": 11910 }, { "epoch": 76.78072289156627, "grad_norm": 19.54104995727539, "learning_rate": 1.1577419354838711e-05, "loss": 0.5843, "step": 11911 }, { "epoch": 76.78714859437751, "grad_norm": 12.159276962280273, "learning_rate": 1.1574193548387097e-05, "loss": 0.2274, "step": 11912 }, { "epoch": 76.79357429718875, "grad_norm": 15.227327346801758, "learning_rate": 1.1570967741935484e-05, "loss": 0.3973, "step": 11913 }, { "epoch": 76.8, "grad_norm": 5.316003322601318, "learning_rate": 1.1567741935483871e-05, "loss": 0.355, "step": 11914 }, { "epoch": 76.80642570281124, "grad_norm": 27.954151153564453, "learning_rate": 1.1564516129032259e-05, "loss": 0.2975, "step": 11915 }, { "epoch": 76.8128514056225, "grad_norm": 16.50263023376465, "learning_rate": 1.1561290322580644e-05, "loss": 0.3349, "step": 11916 }, { "epoch": 76.81927710843374, "grad_norm": 22.996488571166992, "learning_rate": 1.1558064516129034e-05, "loss": 0.5418, "step": 11917 }, { "epoch": 76.82570281124498, "grad_norm": 2.011739730834961, "learning_rate": 1.155483870967742e-05, "loss": 0.1019, "step": 11918 }, { "epoch": 76.83212851405622, "grad_norm": 21.87101173400879, "learning_rate": 1.1551612903225807e-05, "loss": 0.2752, "step": 11919 }, { "epoch": 76.83855421686746, "grad_norm": 5.161100387573242, "learning_rate": 1.1548387096774194e-05, "loss": 0.2778, "step": 11920 }, { "epoch": 76.84497991967872, "grad_norm": 11.340121269226074, "learning_rate": 1.1545161290322581e-05, "loss": 0.2987, "step": 11921 }, { "epoch": 76.85140562248996, "grad_norm": 10.205262184143066, "learning_rate": 1.1541935483870967e-05, "loss": 0.3265, "step": 11922 }, { "epoch": 76.8578313253012, "grad_norm": 4.1588287353515625, "learning_rate": 1.1538709677419356e-05, "loss": 0.2247, "step": 11923 }, { "epoch": 76.86425702811245, "grad_norm": 29.66635513305664, "learning_rate": 1.1535483870967742e-05, "loss": 0.6065, "step": 11924 }, { "epoch": 76.87068273092369, "grad_norm": 16.027795791625977, "learning_rate": 1.153225806451613e-05, "loss": 0.5173, "step": 11925 }, { "epoch": 76.87710843373495, "grad_norm": 9.213505744934082, "learning_rate": 1.1529032258064517e-05, "loss": 0.4603, "step": 11926 }, { "epoch": 76.88353413654619, "grad_norm": 2.676900863647461, "learning_rate": 1.1525806451612904e-05, "loss": 0.1344, "step": 11927 }, { "epoch": 76.88995983935743, "grad_norm": 25.207868576049805, "learning_rate": 1.152258064516129e-05, "loss": 0.4709, "step": 11928 }, { "epoch": 76.89638554216867, "grad_norm": 4.8578691482543945, "learning_rate": 1.1519354838709679e-05, "loss": 0.0986, "step": 11929 }, { "epoch": 76.90281124497992, "grad_norm": 9.212676048278809, "learning_rate": 1.1516129032258065e-05, "loss": 0.19, "step": 11930 }, { "epoch": 76.90923694779117, "grad_norm": 32.529151916503906, "learning_rate": 1.1512903225806452e-05, "loss": 0.4928, "step": 11931 }, { "epoch": 76.91566265060241, "grad_norm": 10.587340354919434, "learning_rate": 1.150967741935484e-05, "loss": 0.3408, "step": 11932 }, { "epoch": 76.92208835341366, "grad_norm": 4.164780139923096, "learning_rate": 1.1506451612903227e-05, "loss": 0.1722, "step": 11933 }, { "epoch": 76.9285140562249, "grad_norm": 33.08290100097656, "learning_rate": 1.1503225806451613e-05, "loss": 0.1667, "step": 11934 }, { "epoch": 76.93493975903614, "grad_norm": 23.211299896240234, "learning_rate": 1.1500000000000002e-05, "loss": 0.4411, "step": 11935 }, { "epoch": 76.94136546184738, "grad_norm": 6.224764347076416, "learning_rate": 1.1496774193548387e-05, "loss": 0.3363, "step": 11936 }, { "epoch": 76.94779116465864, "grad_norm": 6.467221736907959, "learning_rate": 1.1493548387096775e-05, "loss": 0.1582, "step": 11937 }, { "epoch": 76.95421686746988, "grad_norm": 5.449309349060059, "learning_rate": 1.1490322580645162e-05, "loss": 0.3369, "step": 11938 }, { "epoch": 76.96064257028112, "grad_norm": 14.75442123413086, "learning_rate": 1.148709677419355e-05, "loss": 0.4208, "step": 11939 }, { "epoch": 76.96706827309237, "grad_norm": 8.505314826965332, "learning_rate": 1.1483870967741935e-05, "loss": 0.3818, "step": 11940 }, { "epoch": 76.97349397590361, "grad_norm": 28.85272216796875, "learning_rate": 1.1480645161290324e-05, "loss": 0.6969, "step": 11941 }, { "epoch": 76.97991967871486, "grad_norm": 16.850570678710938, "learning_rate": 1.147741935483871e-05, "loss": 0.2394, "step": 11942 }, { "epoch": 76.9863453815261, "grad_norm": 21.611698150634766, "learning_rate": 1.1474193548387098e-05, "loss": 0.4138, "step": 11943 }, { "epoch": 76.99277108433735, "grad_norm": 17.928937911987305, "learning_rate": 1.1470967741935485e-05, "loss": 0.2863, "step": 11944 }, { "epoch": 77.00240963855421, "grad_norm": 16.166284561157227, "learning_rate": 1.146774193548387e-05, "loss": 0.4307, "step": 11945 }, { "epoch": 77.00883534136547, "grad_norm": 10.044533729553223, "learning_rate": 1.1464516129032258e-05, "loss": 0.3108, "step": 11946 }, { "epoch": 77.01526104417671, "grad_norm": 10.196000099182129, "learning_rate": 1.1461290322580645e-05, "loss": 0.2593, "step": 11947 }, { "epoch": 77.02168674698795, "grad_norm": 9.119309425354004, "learning_rate": 1.1458064516129033e-05, "loss": 0.1992, "step": 11948 }, { "epoch": 77.0281124497992, "grad_norm": 6.91757345199585, "learning_rate": 1.1454838709677419e-05, "loss": 0.4048, "step": 11949 }, { "epoch": 77.03453815261044, "grad_norm": 8.111795425415039, "learning_rate": 1.1451612903225808e-05, "loss": 0.2686, "step": 11950 }, { "epoch": 77.04096385542168, "grad_norm": 15.155050277709961, "learning_rate": 1.1448387096774193e-05, "loss": 0.213, "step": 11951 }, { "epoch": 77.04738955823294, "grad_norm": 12.686714172363281, "learning_rate": 1.144516129032258e-05, "loss": 0.4078, "step": 11952 }, { "epoch": 77.05381526104418, "grad_norm": 15.226890563964844, "learning_rate": 1.1441935483870968e-05, "loss": 0.3404, "step": 11953 }, { "epoch": 77.06024096385542, "grad_norm": 13.757600784301758, "learning_rate": 1.1438709677419356e-05, "loss": 0.1926, "step": 11954 }, { "epoch": 77.06666666666666, "grad_norm": 11.624473571777344, "learning_rate": 1.1435483870967741e-05, "loss": 0.4115, "step": 11955 }, { "epoch": 77.0730923694779, "grad_norm": 9.293159484863281, "learning_rate": 1.143225806451613e-05, "loss": 0.1682, "step": 11956 }, { "epoch": 77.07951807228916, "grad_norm": 8.988417625427246, "learning_rate": 1.1429032258064516e-05, "loss": 0.2104, "step": 11957 }, { "epoch": 77.0859437751004, "grad_norm": 23.15863800048828, "learning_rate": 1.1425806451612903e-05, "loss": 0.4795, "step": 11958 }, { "epoch": 77.09236947791165, "grad_norm": 12.085850715637207, "learning_rate": 1.1422580645161291e-05, "loss": 0.5135, "step": 11959 }, { "epoch": 77.09879518072289, "grad_norm": 22.91847038269043, "learning_rate": 1.1419354838709678e-05, "loss": 0.2724, "step": 11960 }, { "epoch": 77.10522088353413, "grad_norm": 6.604137897491455, "learning_rate": 1.1416129032258064e-05, "loss": 0.3635, "step": 11961 }, { "epoch": 77.11164658634539, "grad_norm": 12.417350769042969, "learning_rate": 1.1412903225806453e-05, "loss": 0.4341, "step": 11962 }, { "epoch": 77.11807228915663, "grad_norm": 11.877209663391113, "learning_rate": 1.1409677419354839e-05, "loss": 0.3229, "step": 11963 }, { "epoch": 77.12449799196787, "grad_norm": 6.2246551513671875, "learning_rate": 1.1406451612903226e-05, "loss": 0.1605, "step": 11964 }, { "epoch": 77.13092369477911, "grad_norm": 9.795791625976562, "learning_rate": 1.1403225806451614e-05, "loss": 0.3554, "step": 11965 }, { "epoch": 77.13734939759036, "grad_norm": 9.264416694641113, "learning_rate": 1.1400000000000001e-05, "loss": 0.3075, "step": 11966 }, { "epoch": 77.14377510040161, "grad_norm": 16.58245086669922, "learning_rate": 1.1396774193548387e-05, "loss": 0.244, "step": 11967 }, { "epoch": 77.15020080321285, "grad_norm": 6.181987762451172, "learning_rate": 1.1393548387096776e-05, "loss": 0.1432, "step": 11968 }, { "epoch": 77.1566265060241, "grad_norm": 20.329164505004883, "learning_rate": 1.1390322580645161e-05, "loss": 0.4244, "step": 11969 }, { "epoch": 77.16305220883534, "grad_norm": 27.623727798461914, "learning_rate": 1.1387096774193549e-05, "loss": 0.7257, "step": 11970 }, { "epoch": 77.16947791164658, "grad_norm": 11.51387882232666, "learning_rate": 1.1383870967741936e-05, "loss": 0.3775, "step": 11971 }, { "epoch": 77.17590361445784, "grad_norm": 3.488539934158325, "learning_rate": 1.1380645161290324e-05, "loss": 0.2062, "step": 11972 }, { "epoch": 77.18232931726908, "grad_norm": 2.8541066646575928, "learning_rate": 1.137741935483871e-05, "loss": 0.1029, "step": 11973 }, { "epoch": 77.18875502008032, "grad_norm": 8.161179542541504, "learning_rate": 1.1374193548387098e-05, "loss": 0.2157, "step": 11974 }, { "epoch": 77.19518072289156, "grad_norm": 8.632699966430664, "learning_rate": 1.1370967741935484e-05, "loss": 0.2535, "step": 11975 }, { "epoch": 77.2016064257028, "grad_norm": 2.874300241470337, "learning_rate": 1.1367741935483872e-05, "loss": 0.1183, "step": 11976 }, { "epoch": 77.20803212851406, "grad_norm": 3.8254640102386475, "learning_rate": 1.1364516129032259e-05, "loss": 0.2913, "step": 11977 }, { "epoch": 77.2144578313253, "grad_norm": 10.424912452697754, "learning_rate": 1.1361290322580646e-05, "loss": 0.1882, "step": 11978 }, { "epoch": 77.22088353413655, "grad_norm": 6.285780906677246, "learning_rate": 1.1358064516129032e-05, "loss": 0.2975, "step": 11979 }, { "epoch": 77.22730923694779, "grad_norm": 6.121952056884766, "learning_rate": 1.1354838709677421e-05, "loss": 0.2984, "step": 11980 }, { "epoch": 77.23373493975903, "grad_norm": 18.763668060302734, "learning_rate": 1.1351612903225807e-05, "loss": 0.198, "step": 11981 }, { "epoch": 77.24016064257029, "grad_norm": 5.354515552520752, "learning_rate": 1.1348387096774193e-05, "loss": 0.3762, "step": 11982 }, { "epoch": 77.24658634538153, "grad_norm": 21.846832275390625, "learning_rate": 1.1345161290322582e-05, "loss": 0.142, "step": 11983 }, { "epoch": 77.25301204819277, "grad_norm": 9.845346450805664, "learning_rate": 1.1341935483870967e-05, "loss": 0.3046, "step": 11984 }, { "epoch": 77.25943775100401, "grad_norm": 53.0323486328125, "learning_rate": 1.1338709677419355e-05, "loss": 0.5342, "step": 11985 }, { "epoch": 77.26586345381526, "grad_norm": 32.62080383300781, "learning_rate": 1.1335483870967742e-05, "loss": 0.28, "step": 11986 }, { "epoch": 77.2722891566265, "grad_norm": 58.24993133544922, "learning_rate": 1.133225806451613e-05, "loss": 0.3357, "step": 11987 }, { "epoch": 77.27871485943776, "grad_norm": 11.122698783874512, "learning_rate": 1.1329032258064515e-05, "loss": 0.6439, "step": 11988 }, { "epoch": 77.285140562249, "grad_norm": 10.467425346374512, "learning_rate": 1.1325806451612904e-05, "loss": 0.2226, "step": 11989 }, { "epoch": 77.29156626506024, "grad_norm": 11.753886222839355, "learning_rate": 1.132258064516129e-05, "loss": 0.2247, "step": 11990 }, { "epoch": 77.29799196787148, "grad_norm": 14.443418502807617, "learning_rate": 1.1319354838709678e-05, "loss": 0.2923, "step": 11991 }, { "epoch": 77.30441767068272, "grad_norm": 10.097341537475586, "learning_rate": 1.1316129032258065e-05, "loss": 0.4329, "step": 11992 }, { "epoch": 77.31084337349398, "grad_norm": 9.086316108703613, "learning_rate": 1.1312903225806452e-05, "loss": 0.2796, "step": 11993 }, { "epoch": 77.31726907630522, "grad_norm": 5.759650707244873, "learning_rate": 1.1309677419354838e-05, "loss": 0.2486, "step": 11994 }, { "epoch": 77.32369477911647, "grad_norm": 10.374469757080078, "learning_rate": 1.1306451612903227e-05, "loss": 0.4953, "step": 11995 }, { "epoch": 77.33012048192771, "grad_norm": 6.705284118652344, "learning_rate": 1.1303225806451613e-05, "loss": 0.1501, "step": 11996 }, { "epoch": 77.33654618473895, "grad_norm": 46.194942474365234, "learning_rate": 1.13e-05, "loss": 0.2155, "step": 11997 }, { "epoch": 77.3429718875502, "grad_norm": 16.392351150512695, "learning_rate": 1.1296774193548388e-05, "loss": 0.5478, "step": 11998 }, { "epoch": 77.34939759036145, "grad_norm": 10.040154457092285, "learning_rate": 1.1293548387096775e-05, "loss": 0.2485, "step": 11999 }, { "epoch": 77.35582329317269, "grad_norm": 7.689499378204346, "learning_rate": 1.129032258064516e-05, "loss": 0.2808, "step": 12000 }, { "epoch": 77.36224899598393, "grad_norm": 13.541933059692383, "learning_rate": 1.128709677419355e-05, "loss": 0.3208, "step": 12001 }, { "epoch": 77.36867469879518, "grad_norm": 5.054934024810791, "learning_rate": 1.1283870967741936e-05, "loss": 0.1656, "step": 12002 }, { "epoch": 77.37510040160643, "grad_norm": 3.6415863037109375, "learning_rate": 1.1280645161290323e-05, "loss": 0.2033, "step": 12003 }, { "epoch": 77.38152610441767, "grad_norm": 9.336079597473145, "learning_rate": 1.127741935483871e-05, "loss": 0.3877, "step": 12004 }, { "epoch": 77.38795180722892, "grad_norm": 11.847923278808594, "learning_rate": 1.1274193548387098e-05, "loss": 0.2858, "step": 12005 }, { "epoch": 77.39437751004016, "grad_norm": 20.84319496154785, "learning_rate": 1.1270967741935483e-05, "loss": 0.2213, "step": 12006 }, { "epoch": 77.4008032128514, "grad_norm": 5.866176605224609, "learning_rate": 1.1267741935483873e-05, "loss": 0.2006, "step": 12007 }, { "epoch": 77.40722891566266, "grad_norm": 10.968789100646973, "learning_rate": 1.1264516129032258e-05, "loss": 0.2169, "step": 12008 }, { "epoch": 77.4136546184739, "grad_norm": 12.077693939208984, "learning_rate": 1.1261290322580646e-05, "loss": 0.4064, "step": 12009 }, { "epoch": 77.42008032128514, "grad_norm": 15.532217025756836, "learning_rate": 1.1258064516129033e-05, "loss": 0.2296, "step": 12010 }, { "epoch": 77.42650602409638, "grad_norm": 16.174894332885742, "learning_rate": 1.125483870967742e-05, "loss": 0.5349, "step": 12011 }, { "epoch": 77.43293172690763, "grad_norm": 6.888463020324707, "learning_rate": 1.1251612903225806e-05, "loss": 0.5536, "step": 12012 }, { "epoch": 77.43935742971888, "grad_norm": 11.569600105285645, "learning_rate": 1.1248387096774195e-05, "loss": 0.2733, "step": 12013 }, { "epoch": 77.44578313253012, "grad_norm": 15.94694709777832, "learning_rate": 1.1245161290322581e-05, "loss": 0.2995, "step": 12014 }, { "epoch": 77.45220883534137, "grad_norm": 15.15534782409668, "learning_rate": 1.1241935483870968e-05, "loss": 0.7016, "step": 12015 }, { "epoch": 77.45863453815261, "grad_norm": 2.4328010082244873, "learning_rate": 1.1238709677419356e-05, "loss": 0.1666, "step": 12016 }, { "epoch": 77.46506024096385, "grad_norm": 6.339604377746582, "learning_rate": 1.1235483870967741e-05, "loss": 0.4314, "step": 12017 }, { "epoch": 77.47148594377511, "grad_norm": 40.58120346069336, "learning_rate": 1.1232258064516129e-05, "loss": 0.6944, "step": 12018 }, { "epoch": 77.47791164658635, "grad_norm": 6.0121870040893555, "learning_rate": 1.1229032258064516e-05, "loss": 0.2098, "step": 12019 }, { "epoch": 77.48433734939759, "grad_norm": 12.788175582885742, "learning_rate": 1.1225806451612904e-05, "loss": 0.2597, "step": 12020 }, { "epoch": 77.49076305220883, "grad_norm": 5.798655033111572, "learning_rate": 1.1222580645161291e-05, "loss": 0.2814, "step": 12021 }, { "epoch": 77.49718875502008, "grad_norm": 12.529336929321289, "learning_rate": 1.1219354838709678e-05, "loss": 0.3999, "step": 12022 }, { "epoch": 77.50361445783132, "grad_norm": 21.46907615661621, "learning_rate": 1.1216129032258064e-05, "loss": 0.56, "step": 12023 }, { "epoch": 77.51004016064257, "grad_norm": 4.553239345550537, "learning_rate": 1.1212903225806452e-05, "loss": 0.1802, "step": 12024 }, { "epoch": 77.51646586345382, "grad_norm": 19.027090072631836, "learning_rate": 1.1209677419354839e-05, "loss": 0.5831, "step": 12025 }, { "epoch": 77.52289156626506, "grad_norm": 15.28028678894043, "learning_rate": 1.1206451612903226e-05, "loss": 0.4331, "step": 12026 }, { "epoch": 77.5293172690763, "grad_norm": 5.189234733581543, "learning_rate": 1.1203225806451614e-05, "loss": 0.3185, "step": 12027 }, { "epoch": 77.53574297188754, "grad_norm": 7.206272125244141, "learning_rate": 1.1200000000000001e-05, "loss": 0.2897, "step": 12028 }, { "epoch": 77.5421686746988, "grad_norm": 23.66325569152832, "learning_rate": 1.1196774193548387e-05, "loss": 0.3351, "step": 12029 }, { "epoch": 77.54859437751004, "grad_norm": 18.59787368774414, "learning_rate": 1.1193548387096774e-05, "loss": 0.5721, "step": 12030 }, { "epoch": 77.55502008032128, "grad_norm": 8.323797225952148, "learning_rate": 1.1190322580645162e-05, "loss": 0.2025, "step": 12031 }, { "epoch": 77.56144578313253, "grad_norm": 9.393218994140625, "learning_rate": 1.1187096774193549e-05, "loss": 0.4015, "step": 12032 }, { "epoch": 77.56787148594377, "grad_norm": 19.548625946044922, "learning_rate": 1.1183870967741936e-05, "loss": 0.7573, "step": 12033 }, { "epoch": 77.57429718875503, "grad_norm": 27.24327850341797, "learning_rate": 1.1180645161290324e-05, "loss": 0.5788, "step": 12034 }, { "epoch": 77.58072289156627, "grad_norm": 8.198678970336914, "learning_rate": 1.117741935483871e-05, "loss": 0.5984, "step": 12035 }, { "epoch": 77.58714859437751, "grad_norm": 12.52865982055664, "learning_rate": 1.1174193548387097e-05, "loss": 0.6964, "step": 12036 }, { "epoch": 77.59357429718875, "grad_norm": 7.559333801269531, "learning_rate": 1.1170967741935484e-05, "loss": 0.2755, "step": 12037 }, { "epoch": 77.6, "grad_norm": 81.16634368896484, "learning_rate": 1.1167741935483872e-05, "loss": 0.3115, "step": 12038 }, { "epoch": 77.60642570281125, "grad_norm": 8.837461471557617, "learning_rate": 1.1164516129032258e-05, "loss": 0.2813, "step": 12039 }, { "epoch": 77.6128514056225, "grad_norm": 9.105329513549805, "learning_rate": 1.1161290322580647e-05, "loss": 0.425, "step": 12040 }, { "epoch": 77.61927710843374, "grad_norm": 14.416062355041504, "learning_rate": 1.1158064516129032e-05, "loss": 0.3141, "step": 12041 }, { "epoch": 77.62570281124498, "grad_norm": 8.66292953491211, "learning_rate": 1.115483870967742e-05, "loss": 0.3143, "step": 12042 }, { "epoch": 77.63212851405622, "grad_norm": 5.308629989624023, "learning_rate": 1.1151612903225807e-05, "loss": 0.4193, "step": 12043 }, { "epoch": 77.63855421686748, "grad_norm": 17.267282485961914, "learning_rate": 1.1148387096774195e-05, "loss": 0.389, "step": 12044 }, { "epoch": 77.64497991967872, "grad_norm": 25.03281593322754, "learning_rate": 1.114516129032258e-05, "loss": 0.3392, "step": 12045 }, { "epoch": 77.65140562248996, "grad_norm": 16.34506607055664, "learning_rate": 1.114193548387097e-05, "loss": 0.3911, "step": 12046 }, { "epoch": 77.6578313253012, "grad_norm": 6.430923938751221, "learning_rate": 1.1138709677419355e-05, "loss": 0.2048, "step": 12047 }, { "epoch": 77.66425702811244, "grad_norm": 8.459012985229492, "learning_rate": 1.1135483870967742e-05, "loss": 0.217, "step": 12048 }, { "epoch": 77.6706827309237, "grad_norm": 28.872507095336914, "learning_rate": 1.113225806451613e-05, "loss": 0.3124, "step": 12049 }, { "epoch": 77.67710843373494, "grad_norm": 11.1145601272583, "learning_rate": 1.1129032258064517e-05, "loss": 0.3644, "step": 12050 }, { "epoch": 77.68353413654619, "grad_norm": 7.445207595825195, "learning_rate": 1.1125806451612903e-05, "loss": 0.2559, "step": 12051 }, { "epoch": 77.68995983935743, "grad_norm": 18.790029525756836, "learning_rate": 1.112258064516129e-05, "loss": 0.3395, "step": 12052 }, { "epoch": 77.69638554216867, "grad_norm": 15.613697052001953, "learning_rate": 1.1119354838709678e-05, "loss": 0.4891, "step": 12053 }, { "epoch": 77.70281124497993, "grad_norm": 36.757572174072266, "learning_rate": 1.1116129032258065e-05, "loss": 0.4932, "step": 12054 }, { "epoch": 77.70923694779117, "grad_norm": 3.459099531173706, "learning_rate": 1.1112903225806453e-05, "loss": 0.0701, "step": 12055 }, { "epoch": 77.71566265060241, "grad_norm": 13.40372371673584, "learning_rate": 1.1109677419354838e-05, "loss": 0.4848, "step": 12056 }, { "epoch": 77.72208835341365, "grad_norm": 4.927857875823975, "learning_rate": 1.1106451612903226e-05, "loss": 0.1827, "step": 12057 }, { "epoch": 77.7285140562249, "grad_norm": 8.491528511047363, "learning_rate": 1.1103225806451613e-05, "loss": 0.2805, "step": 12058 }, { "epoch": 77.73493975903614, "grad_norm": 4.274961948394775, "learning_rate": 1.11e-05, "loss": 0.1167, "step": 12059 }, { "epoch": 77.7413654618474, "grad_norm": 5.6774444580078125, "learning_rate": 1.1096774193548388e-05, "loss": 0.1944, "step": 12060 }, { "epoch": 77.74779116465864, "grad_norm": 10.195443153381348, "learning_rate": 1.1093548387096775e-05, "loss": 0.3046, "step": 12061 }, { "epoch": 77.75421686746988, "grad_norm": 16.88578224182129, "learning_rate": 1.1090322580645161e-05, "loss": 0.4084, "step": 12062 }, { "epoch": 77.76064257028112, "grad_norm": 12.532694816589355, "learning_rate": 1.1087096774193548e-05, "loss": 0.3536, "step": 12063 }, { "epoch": 77.76706827309236, "grad_norm": 8.693267822265625, "learning_rate": 1.1083870967741936e-05, "loss": 0.3791, "step": 12064 }, { "epoch": 77.77349397590362, "grad_norm": 11.249631881713867, "learning_rate": 1.1080645161290323e-05, "loss": 0.4389, "step": 12065 }, { "epoch": 77.77991967871486, "grad_norm": 8.752057075500488, "learning_rate": 1.107741935483871e-05, "loss": 0.1887, "step": 12066 }, { "epoch": 77.7863453815261, "grad_norm": 11.632431983947754, "learning_rate": 1.1074193548387098e-05, "loss": 0.5621, "step": 12067 }, { "epoch": 77.79277108433735, "grad_norm": 8.055760383605957, "learning_rate": 1.1070967741935484e-05, "loss": 0.3042, "step": 12068 }, { "epoch": 77.79919678714859, "grad_norm": 3.888066053390503, "learning_rate": 1.1067741935483871e-05, "loss": 0.1645, "step": 12069 }, { "epoch": 77.80562248995984, "grad_norm": 4.099750518798828, "learning_rate": 1.1064516129032258e-05, "loss": 0.1984, "step": 12070 }, { "epoch": 77.81204819277109, "grad_norm": 10.477432250976562, "learning_rate": 1.1061290322580646e-05, "loss": 0.4585, "step": 12071 }, { "epoch": 77.81847389558233, "grad_norm": 18.55389976501465, "learning_rate": 1.1058064516129033e-05, "loss": 0.3803, "step": 12072 }, { "epoch": 77.82489959839357, "grad_norm": 8.573723793029785, "learning_rate": 1.105483870967742e-05, "loss": 0.3289, "step": 12073 }, { "epoch": 77.83132530120481, "grad_norm": 25.249130249023438, "learning_rate": 1.1051612903225806e-05, "loss": 0.482, "step": 12074 }, { "epoch": 77.83775100401607, "grad_norm": 10.63747501373291, "learning_rate": 1.1048387096774194e-05, "loss": 0.2899, "step": 12075 }, { "epoch": 77.84417670682731, "grad_norm": 5.531499862670898, "learning_rate": 1.1045161290322581e-05, "loss": 0.2208, "step": 12076 }, { "epoch": 77.85060240963855, "grad_norm": 8.522496223449707, "learning_rate": 1.1041935483870969e-05, "loss": 0.3951, "step": 12077 }, { "epoch": 77.8570281124498, "grad_norm": 70.65514373779297, "learning_rate": 1.1038709677419356e-05, "loss": 0.4804, "step": 12078 }, { "epoch": 77.86345381526104, "grad_norm": 29.04999542236328, "learning_rate": 1.1035483870967743e-05, "loss": 0.4405, "step": 12079 }, { "epoch": 77.8698795180723, "grad_norm": 19.73027229309082, "learning_rate": 1.1032258064516129e-05, "loss": 0.2662, "step": 12080 }, { "epoch": 77.87630522088354, "grad_norm": 15.918228149414062, "learning_rate": 1.1029032258064516e-05, "loss": 0.3981, "step": 12081 }, { "epoch": 77.88273092369478, "grad_norm": 12.819405555725098, "learning_rate": 1.1025806451612904e-05, "loss": 0.387, "step": 12082 }, { "epoch": 77.88915662650602, "grad_norm": 5.710006237030029, "learning_rate": 1.1022580645161291e-05, "loss": 0.1997, "step": 12083 }, { "epoch": 77.89558232931726, "grad_norm": 4.959819793701172, "learning_rate": 1.1019354838709679e-05, "loss": 0.2453, "step": 12084 }, { "epoch": 77.90200803212852, "grad_norm": 25.15949821472168, "learning_rate": 1.1016129032258066e-05, "loss": 0.2232, "step": 12085 }, { "epoch": 77.90843373493976, "grad_norm": 23.044010162353516, "learning_rate": 1.1012903225806452e-05, "loss": 0.595, "step": 12086 }, { "epoch": 77.914859437751, "grad_norm": 11.33812427520752, "learning_rate": 1.100967741935484e-05, "loss": 0.5319, "step": 12087 }, { "epoch": 77.92128514056225, "grad_norm": 12.750066757202148, "learning_rate": 1.1006451612903227e-05, "loss": 0.3665, "step": 12088 }, { "epoch": 77.92771084337349, "grad_norm": 18.582107543945312, "learning_rate": 1.1003225806451612e-05, "loss": 0.2711, "step": 12089 }, { "epoch": 77.93413654618475, "grad_norm": 13.041115760803223, "learning_rate": 1.1000000000000001e-05, "loss": 0.4489, "step": 12090 }, { "epoch": 77.94056224899599, "grad_norm": 5.9212260246276855, "learning_rate": 1.0996774193548387e-05, "loss": 0.2077, "step": 12091 }, { "epoch": 77.94698795180723, "grad_norm": 16.329776763916016, "learning_rate": 1.0993548387096775e-05, "loss": 0.3439, "step": 12092 }, { "epoch": 77.95341365461847, "grad_norm": 18.48341941833496, "learning_rate": 1.0990322580645162e-05, "loss": 0.4401, "step": 12093 }, { "epoch": 77.95983935742971, "grad_norm": 21.259078979492188, "learning_rate": 1.098709677419355e-05, "loss": 0.3677, "step": 12094 }, { "epoch": 77.96626506024096, "grad_norm": 6.8638916015625, "learning_rate": 1.0983870967741935e-05, "loss": 0.3529, "step": 12095 }, { "epoch": 77.97269076305221, "grad_norm": 4.1659345626831055, "learning_rate": 1.0980645161290324e-05, "loss": 0.1365, "step": 12096 }, { "epoch": 77.97911646586346, "grad_norm": 8.383546829223633, "learning_rate": 1.097741935483871e-05, "loss": 0.4167, "step": 12097 }, { "epoch": 77.9855421686747, "grad_norm": 17.116329193115234, "learning_rate": 1.0974193548387097e-05, "loss": 0.1992, "step": 12098 }, { "epoch": 77.99196787148594, "grad_norm": 6.15261697769165, "learning_rate": 1.0970967741935485e-05, "loss": 0.1078, "step": 12099 }, { "epoch": 78.00160642570282, "grad_norm": 7.721583843231201, "learning_rate": 1.0967741935483872e-05, "loss": 0.4065, "step": 12100 }, { "epoch": 78.00803212851406, "grad_norm": 13.152572631835938, "learning_rate": 1.0964516129032258e-05, "loss": 0.2812, "step": 12101 }, { "epoch": 78.0144578313253, "grad_norm": 10.833571434020996, "learning_rate": 1.0961290322580645e-05, "loss": 0.1741, "step": 12102 }, { "epoch": 78.02088353413654, "grad_norm": 18.407161712646484, "learning_rate": 1.0958064516129033e-05, "loss": 0.6969, "step": 12103 }, { "epoch": 78.02730923694779, "grad_norm": 15.966415405273438, "learning_rate": 1.095483870967742e-05, "loss": 0.4141, "step": 12104 }, { "epoch": 78.03373493975904, "grad_norm": 10.240878105163574, "learning_rate": 1.0951612903225807e-05, "loss": 0.3722, "step": 12105 }, { "epoch": 78.04016064257029, "grad_norm": 44.94757080078125, "learning_rate": 1.0948387096774195e-05, "loss": 0.5595, "step": 12106 }, { "epoch": 78.04658634538153, "grad_norm": 41.999237060546875, "learning_rate": 1.094516129032258e-05, "loss": 0.507, "step": 12107 }, { "epoch": 78.05301204819277, "grad_norm": 7.464653491973877, "learning_rate": 1.0941935483870968e-05, "loss": 0.3188, "step": 12108 }, { "epoch": 78.05943775100401, "grad_norm": 39.628353118896484, "learning_rate": 1.0938709677419355e-05, "loss": 0.2385, "step": 12109 }, { "epoch": 78.06586345381525, "grad_norm": 24.718690872192383, "learning_rate": 1.0935483870967743e-05, "loss": 0.3804, "step": 12110 }, { "epoch": 78.07228915662651, "grad_norm": 25.996662139892578, "learning_rate": 1.093225806451613e-05, "loss": 0.2227, "step": 12111 }, { "epoch": 78.07871485943775, "grad_norm": 9.736248016357422, "learning_rate": 1.0929032258064517e-05, "loss": 0.27, "step": 12112 }, { "epoch": 78.085140562249, "grad_norm": 12.746957778930664, "learning_rate": 1.0925806451612903e-05, "loss": 0.3624, "step": 12113 }, { "epoch": 78.09156626506024, "grad_norm": 11.252799034118652, "learning_rate": 1.092258064516129e-05, "loss": 0.2339, "step": 12114 }, { "epoch": 78.09799196787148, "grad_norm": 6.1728835105896, "learning_rate": 1.0919354838709678e-05, "loss": 0.4727, "step": 12115 }, { "epoch": 78.10441767068274, "grad_norm": 6.1424336433410645, "learning_rate": 1.0916129032258065e-05, "loss": 0.1274, "step": 12116 }, { "epoch": 78.11084337349398, "grad_norm": 10.644445419311523, "learning_rate": 1.0912903225806453e-05, "loss": 0.313, "step": 12117 }, { "epoch": 78.11726907630522, "grad_norm": 9.315019607543945, "learning_rate": 1.090967741935484e-05, "loss": 0.3614, "step": 12118 }, { "epoch": 78.12369477911646, "grad_norm": 9.522073745727539, "learning_rate": 1.0906451612903226e-05, "loss": 0.3191, "step": 12119 }, { "epoch": 78.1301204819277, "grad_norm": 12.39132308959961, "learning_rate": 1.0903225806451613e-05, "loss": 0.2041, "step": 12120 }, { "epoch": 78.13654618473896, "grad_norm": 6.9930267333984375, "learning_rate": 1.09e-05, "loss": 0.2961, "step": 12121 }, { "epoch": 78.1429718875502, "grad_norm": 3.4540975093841553, "learning_rate": 1.0896774193548388e-05, "loss": 0.1455, "step": 12122 }, { "epoch": 78.14939759036145, "grad_norm": 5.592596054077148, "learning_rate": 1.0893548387096775e-05, "loss": 0.104, "step": 12123 }, { "epoch": 78.15582329317269, "grad_norm": 10.436592102050781, "learning_rate": 1.0890322580645161e-05, "loss": 0.3636, "step": 12124 }, { "epoch": 78.16224899598393, "grad_norm": 12.247865676879883, "learning_rate": 1.0887096774193549e-05, "loss": 0.2697, "step": 12125 }, { "epoch": 78.16867469879519, "grad_norm": 15.919793128967285, "learning_rate": 1.0883870967741936e-05, "loss": 0.3872, "step": 12126 }, { "epoch": 78.17510040160643, "grad_norm": 7.694927215576172, "learning_rate": 1.0880645161290323e-05, "loss": 0.5258, "step": 12127 }, { "epoch": 78.18152610441767, "grad_norm": 10.122292518615723, "learning_rate": 1.0877419354838709e-05, "loss": 0.4701, "step": 12128 }, { "epoch": 78.18795180722891, "grad_norm": 9.04406452178955, "learning_rate": 1.0874193548387098e-05, "loss": 0.3773, "step": 12129 }, { "epoch": 78.19437751004016, "grad_norm": 16.697263717651367, "learning_rate": 1.0870967741935484e-05, "loss": 0.4657, "step": 12130 }, { "epoch": 78.20080321285141, "grad_norm": 75.33509063720703, "learning_rate": 1.0867741935483871e-05, "loss": 0.3566, "step": 12131 }, { "epoch": 78.20722891566265, "grad_norm": 23.193010330200195, "learning_rate": 1.0864516129032259e-05, "loss": 0.5646, "step": 12132 }, { "epoch": 78.2136546184739, "grad_norm": 10.317876815795898, "learning_rate": 1.0861290322580646e-05, "loss": 0.3679, "step": 12133 }, { "epoch": 78.22008032128514, "grad_norm": 7.5440521240234375, "learning_rate": 1.0858064516129032e-05, "loss": 0.2877, "step": 12134 }, { "epoch": 78.22650602409638, "grad_norm": 22.310016632080078, "learning_rate": 1.0854838709677421e-05, "loss": 0.6264, "step": 12135 }, { "epoch": 78.23293172690764, "grad_norm": 9.627022743225098, "learning_rate": 1.0851612903225807e-05, "loss": 0.481, "step": 12136 }, { "epoch": 78.23935742971888, "grad_norm": 5.983254432678223, "learning_rate": 1.0848387096774194e-05, "loss": 0.2453, "step": 12137 }, { "epoch": 78.24578313253012, "grad_norm": 8.371882438659668, "learning_rate": 1.0845161290322581e-05, "loss": 0.4289, "step": 12138 }, { "epoch": 78.25220883534136, "grad_norm": 6.756310939788818, "learning_rate": 1.0841935483870969e-05, "loss": 0.1946, "step": 12139 }, { "epoch": 78.2586345381526, "grad_norm": 12.398489952087402, "learning_rate": 1.0838709677419355e-05, "loss": 0.3679, "step": 12140 }, { "epoch": 78.26506024096386, "grad_norm": 11.352669715881348, "learning_rate": 1.0835483870967744e-05, "loss": 0.2515, "step": 12141 }, { "epoch": 78.2714859437751, "grad_norm": 19.967348098754883, "learning_rate": 1.083225806451613e-05, "loss": 0.4398, "step": 12142 }, { "epoch": 78.27791164658635, "grad_norm": 12.256321907043457, "learning_rate": 1.0829032258064517e-05, "loss": 0.4366, "step": 12143 }, { "epoch": 78.28433734939759, "grad_norm": 24.119333267211914, "learning_rate": 1.0825806451612904e-05, "loss": 0.3572, "step": 12144 }, { "epoch": 78.29076305220883, "grad_norm": 7.799607276916504, "learning_rate": 1.0822580645161292e-05, "loss": 0.4133, "step": 12145 }, { "epoch": 78.29718875502007, "grad_norm": 12.659296035766602, "learning_rate": 1.0819354838709677e-05, "loss": 0.209, "step": 12146 }, { "epoch": 78.30361445783133, "grad_norm": 13.936117172241211, "learning_rate": 1.0816129032258066e-05, "loss": 0.3623, "step": 12147 }, { "epoch": 78.31004016064257, "grad_norm": 38.26675796508789, "learning_rate": 1.0812903225806452e-05, "loss": 0.4798, "step": 12148 }, { "epoch": 78.31646586345381, "grad_norm": 7.736074447631836, "learning_rate": 1.080967741935484e-05, "loss": 0.1466, "step": 12149 }, { "epoch": 78.32289156626506, "grad_norm": 6.290682315826416, "learning_rate": 1.0806451612903227e-05, "loss": 0.297, "step": 12150 }, { "epoch": 78.3293172690763, "grad_norm": 4.584226131439209, "learning_rate": 1.0803225806451614e-05, "loss": 0.1214, "step": 12151 }, { "epoch": 78.33574297188756, "grad_norm": 8.083534240722656, "learning_rate": 1.08e-05, "loss": 0.2691, "step": 12152 }, { "epoch": 78.3421686746988, "grad_norm": 8.293120384216309, "learning_rate": 1.0796774193548389e-05, "loss": 0.4712, "step": 12153 }, { "epoch": 78.34859437751004, "grad_norm": 18.551044464111328, "learning_rate": 1.0793548387096775e-05, "loss": 0.4504, "step": 12154 }, { "epoch": 78.35502008032128, "grad_norm": 20.902507781982422, "learning_rate": 1.0790322580645162e-05, "loss": 0.3696, "step": 12155 }, { "epoch": 78.36144578313252, "grad_norm": 10.116701126098633, "learning_rate": 1.078709677419355e-05, "loss": 0.382, "step": 12156 }, { "epoch": 78.36787148594378, "grad_norm": 16.27604103088379, "learning_rate": 1.0783870967741937e-05, "loss": 0.44, "step": 12157 }, { "epoch": 78.37429718875502, "grad_norm": 6.101781845092773, "learning_rate": 1.0780645161290323e-05, "loss": 0.2806, "step": 12158 }, { "epoch": 78.38072289156626, "grad_norm": 7.814598560333252, "learning_rate": 1.077741935483871e-05, "loss": 0.3025, "step": 12159 }, { "epoch": 78.3871485943775, "grad_norm": 17.442459106445312, "learning_rate": 1.0774193548387097e-05, "loss": 0.2988, "step": 12160 }, { "epoch": 78.39357429718875, "grad_norm": 7.382140159606934, "learning_rate": 1.0770967741935483e-05, "loss": 0.2705, "step": 12161 }, { "epoch": 78.4, "grad_norm": 8.753013610839844, "learning_rate": 1.0767741935483872e-05, "loss": 0.2951, "step": 12162 }, { "epoch": 78.40642570281125, "grad_norm": 8.378053665161133, "learning_rate": 1.0764516129032258e-05, "loss": 0.4397, "step": 12163 }, { "epoch": 78.41285140562249, "grad_norm": 3.662109136581421, "learning_rate": 1.0761290322580645e-05, "loss": 0.1303, "step": 12164 }, { "epoch": 78.41927710843373, "grad_norm": 11.387707710266113, "learning_rate": 1.0758064516129033e-05, "loss": 0.4043, "step": 12165 }, { "epoch": 78.42570281124497, "grad_norm": 7.610923767089844, "learning_rate": 1.075483870967742e-05, "loss": 0.3914, "step": 12166 }, { "epoch": 78.43212851405623, "grad_norm": 10.601775169372559, "learning_rate": 1.0751612903225806e-05, "loss": 0.3922, "step": 12167 }, { "epoch": 78.43855421686747, "grad_norm": 12.943984985351562, "learning_rate": 1.0748387096774195e-05, "loss": 0.3742, "step": 12168 }, { "epoch": 78.44497991967872, "grad_norm": 31.106306076049805, "learning_rate": 1.074516129032258e-05, "loss": 0.385, "step": 12169 }, { "epoch": 78.45140562248996, "grad_norm": 9.692773818969727, "learning_rate": 1.0741935483870968e-05, "loss": 0.1932, "step": 12170 }, { "epoch": 78.4578313253012, "grad_norm": 7.006452560424805, "learning_rate": 1.0738709677419355e-05, "loss": 0.3136, "step": 12171 }, { "epoch": 78.46425702811246, "grad_norm": 7.908455848693848, "learning_rate": 1.0735483870967743e-05, "loss": 0.2678, "step": 12172 }, { "epoch": 78.4706827309237, "grad_norm": 30.935226440429688, "learning_rate": 1.0732258064516129e-05, "loss": 0.3793, "step": 12173 }, { "epoch": 78.47710843373494, "grad_norm": 15.217925071716309, "learning_rate": 1.0729032258064518e-05, "loss": 0.4243, "step": 12174 }, { "epoch": 78.48353413654618, "grad_norm": 2.6734187602996826, "learning_rate": 1.0725806451612903e-05, "loss": 0.1594, "step": 12175 }, { "epoch": 78.48995983935743, "grad_norm": 7.395491123199463, "learning_rate": 1.072258064516129e-05, "loss": 0.3062, "step": 12176 }, { "epoch": 78.49638554216868, "grad_norm": 18.61543846130371, "learning_rate": 1.0719354838709678e-05, "loss": 0.2824, "step": 12177 }, { "epoch": 78.50281124497992, "grad_norm": 2.8066134452819824, "learning_rate": 1.0716129032258066e-05, "loss": 0.1707, "step": 12178 }, { "epoch": 78.50923694779117, "grad_norm": 9.835318565368652, "learning_rate": 1.0712903225806451e-05, "loss": 0.52, "step": 12179 }, { "epoch": 78.51566265060241, "grad_norm": 9.944955825805664, "learning_rate": 1.070967741935484e-05, "loss": 0.1539, "step": 12180 }, { "epoch": 78.52208835341365, "grad_norm": 8.038522720336914, "learning_rate": 1.0706451612903226e-05, "loss": 0.6084, "step": 12181 }, { "epoch": 78.52851405622489, "grad_norm": 17.03618621826172, "learning_rate": 1.0703225806451613e-05, "loss": 0.5463, "step": 12182 }, { "epoch": 78.53493975903615, "grad_norm": 6.998373031616211, "learning_rate": 1.0700000000000001e-05, "loss": 0.3942, "step": 12183 }, { "epoch": 78.54136546184739, "grad_norm": 4.849183082580566, "learning_rate": 1.0696774193548388e-05, "loss": 0.112, "step": 12184 }, { "epoch": 78.54779116465863, "grad_norm": 3.361543655395508, "learning_rate": 1.0693548387096774e-05, "loss": 0.1784, "step": 12185 }, { "epoch": 78.55421686746988, "grad_norm": 5.912510871887207, "learning_rate": 1.0690322580645163e-05, "loss": 0.2761, "step": 12186 }, { "epoch": 78.56064257028112, "grad_norm": 11.530496597290039, "learning_rate": 1.0687096774193549e-05, "loss": 0.4251, "step": 12187 }, { "epoch": 78.56706827309237, "grad_norm": 21.319334030151367, "learning_rate": 1.0683870967741936e-05, "loss": 0.3421, "step": 12188 }, { "epoch": 78.57349397590362, "grad_norm": 6.388710975646973, "learning_rate": 1.0680645161290324e-05, "loss": 0.1691, "step": 12189 }, { "epoch": 78.57991967871486, "grad_norm": 7.354771614074707, "learning_rate": 1.0677419354838711e-05, "loss": 0.1657, "step": 12190 }, { "epoch": 78.5863453815261, "grad_norm": 19.010578155517578, "learning_rate": 1.0674193548387097e-05, "loss": 0.5155, "step": 12191 }, { "epoch": 78.59277108433734, "grad_norm": 24.106260299682617, "learning_rate": 1.0670967741935486e-05, "loss": 0.6252, "step": 12192 }, { "epoch": 78.5991967871486, "grad_norm": 10.451512336730957, "learning_rate": 1.0667741935483872e-05, "loss": 0.1961, "step": 12193 }, { "epoch": 78.60562248995984, "grad_norm": 11.318965911865234, "learning_rate": 1.0664516129032257e-05, "loss": 0.3078, "step": 12194 }, { "epoch": 78.61204819277108, "grad_norm": 7.038882732391357, "learning_rate": 1.0661290322580646e-05, "loss": 0.2871, "step": 12195 }, { "epoch": 78.61847389558233, "grad_norm": 31.22345542907715, "learning_rate": 1.0658064516129032e-05, "loss": 0.4248, "step": 12196 }, { "epoch": 78.62489959839357, "grad_norm": 30.207319259643555, "learning_rate": 1.065483870967742e-05, "loss": 0.4176, "step": 12197 }, { "epoch": 78.63132530120482, "grad_norm": 40.25847625732422, "learning_rate": 1.0651612903225807e-05, "loss": 0.3877, "step": 12198 }, { "epoch": 78.63775100401607, "grad_norm": 12.491936683654785, "learning_rate": 1.0648387096774194e-05, "loss": 0.3439, "step": 12199 }, { "epoch": 78.64417670682731, "grad_norm": 17.553552627563477, "learning_rate": 1.064516129032258e-05, "loss": 0.4757, "step": 12200 }, { "epoch": 78.65060240963855, "grad_norm": 4.782384395599365, "learning_rate": 1.0641935483870969e-05, "loss": 0.2806, "step": 12201 }, { "epoch": 78.6570281124498, "grad_norm": 12.995503425598145, "learning_rate": 1.0638709677419355e-05, "loss": 0.2236, "step": 12202 }, { "epoch": 78.66345381526105, "grad_norm": 10.114654541015625, "learning_rate": 1.0635483870967742e-05, "loss": 0.4816, "step": 12203 }, { "epoch": 78.66987951807229, "grad_norm": 21.228361129760742, "learning_rate": 1.063225806451613e-05, "loss": 0.366, "step": 12204 }, { "epoch": 78.67630522088353, "grad_norm": 12.233612060546875, "learning_rate": 1.0629032258064517e-05, "loss": 0.8476, "step": 12205 }, { "epoch": 78.68273092369478, "grad_norm": 4.901845932006836, "learning_rate": 1.0625806451612903e-05, "loss": 0.1628, "step": 12206 }, { "epoch": 78.68915662650602, "grad_norm": 9.862273216247559, "learning_rate": 1.0622580645161292e-05, "loss": 0.5275, "step": 12207 }, { "epoch": 78.69558232931728, "grad_norm": 17.10154914855957, "learning_rate": 1.0619354838709677e-05, "loss": 0.4647, "step": 12208 }, { "epoch": 78.70200803212852, "grad_norm": 7.6584625244140625, "learning_rate": 1.0616129032258065e-05, "loss": 0.4604, "step": 12209 }, { "epoch": 78.70843373493976, "grad_norm": 5.609926223754883, "learning_rate": 1.0612903225806452e-05, "loss": 0.1989, "step": 12210 }, { "epoch": 78.714859437751, "grad_norm": 6.3372626304626465, "learning_rate": 1.060967741935484e-05, "loss": 0.2472, "step": 12211 }, { "epoch": 78.72128514056224, "grad_norm": 10.882563591003418, "learning_rate": 1.0606451612903225e-05, "loss": 0.457, "step": 12212 }, { "epoch": 78.7277108433735, "grad_norm": 24.319347381591797, "learning_rate": 1.0603225806451614e-05, "loss": 0.3348, "step": 12213 }, { "epoch": 78.73413654618474, "grad_norm": 13.279581069946289, "learning_rate": 1.06e-05, "loss": 0.2354, "step": 12214 }, { "epoch": 78.74056224899599, "grad_norm": 10.497946739196777, "learning_rate": 1.0596774193548388e-05, "loss": 0.208, "step": 12215 }, { "epoch": 78.74698795180723, "grad_norm": 6.473867893218994, "learning_rate": 1.0593548387096775e-05, "loss": 0.3803, "step": 12216 }, { "epoch": 78.75341365461847, "grad_norm": 6.590905666351318, "learning_rate": 1.0590322580645162e-05, "loss": 0.2589, "step": 12217 }, { "epoch": 78.75983935742971, "grad_norm": 14.202570915222168, "learning_rate": 1.0587096774193548e-05, "loss": 0.4623, "step": 12218 }, { "epoch": 78.76626506024097, "grad_norm": 4.935111045837402, "learning_rate": 1.0583870967741937e-05, "loss": 0.175, "step": 12219 }, { "epoch": 78.77269076305221, "grad_norm": 5.719627857208252, "learning_rate": 1.0580645161290323e-05, "loss": 0.4758, "step": 12220 }, { "epoch": 78.77911646586345, "grad_norm": 5.274233341217041, "learning_rate": 1.057741935483871e-05, "loss": 0.296, "step": 12221 }, { "epoch": 78.7855421686747, "grad_norm": 5.157745838165283, "learning_rate": 1.0574193548387098e-05, "loss": 0.3286, "step": 12222 }, { "epoch": 78.79196787148594, "grad_norm": 7.130980491638184, "learning_rate": 1.0570967741935485e-05, "loss": 0.3202, "step": 12223 }, { "epoch": 78.7983935742972, "grad_norm": 29.147319793701172, "learning_rate": 1.056774193548387e-05, "loss": 0.2964, "step": 12224 }, { "epoch": 78.80481927710844, "grad_norm": 4.016019821166992, "learning_rate": 1.056451612903226e-05, "loss": 0.2664, "step": 12225 }, { "epoch": 78.81124497991968, "grad_norm": 6.885075569152832, "learning_rate": 1.0561290322580646e-05, "loss": 0.2379, "step": 12226 }, { "epoch": 78.81767068273092, "grad_norm": 7.986532688140869, "learning_rate": 1.0558064516129033e-05, "loss": 0.2567, "step": 12227 }, { "epoch": 78.82409638554216, "grad_norm": 6.225419998168945, "learning_rate": 1.055483870967742e-05, "loss": 0.277, "step": 12228 }, { "epoch": 78.83052208835342, "grad_norm": 6.752612590789795, "learning_rate": 1.0551612903225808e-05, "loss": 0.3666, "step": 12229 }, { "epoch": 78.83694779116466, "grad_norm": 18.75626564025879, "learning_rate": 1.0548387096774193e-05, "loss": 0.5476, "step": 12230 }, { "epoch": 78.8433734939759, "grad_norm": 13.217841148376465, "learning_rate": 1.0545161290322581e-05, "loss": 0.4748, "step": 12231 }, { "epoch": 78.84979919678715, "grad_norm": 17.0091609954834, "learning_rate": 1.0541935483870968e-05, "loss": 0.3229, "step": 12232 }, { "epoch": 78.85622489959839, "grad_norm": 7.849721431732178, "learning_rate": 1.0538709677419354e-05, "loss": 0.3513, "step": 12233 }, { "epoch": 78.86265060240964, "grad_norm": 4.987218379974365, "learning_rate": 1.0535483870967743e-05, "loss": 0.2498, "step": 12234 }, { "epoch": 78.86907630522089, "grad_norm": 34.59664535522461, "learning_rate": 1.0532258064516129e-05, "loss": 0.2858, "step": 12235 }, { "epoch": 78.87550200803213, "grad_norm": 12.813202857971191, "learning_rate": 1.0529032258064516e-05, "loss": 0.378, "step": 12236 }, { "epoch": 78.88192771084337, "grad_norm": 16.542709350585938, "learning_rate": 1.0525806451612904e-05, "loss": 0.6197, "step": 12237 }, { "epoch": 78.88835341365461, "grad_norm": 11.610065460205078, "learning_rate": 1.0522580645161291e-05, "loss": 0.2862, "step": 12238 }, { "epoch": 78.89477911646587, "grad_norm": 8.624369621276855, "learning_rate": 1.0519354838709677e-05, "loss": 0.2842, "step": 12239 }, { "epoch": 78.90120481927711, "grad_norm": 14.855216026306152, "learning_rate": 1.0516129032258066e-05, "loss": 0.5573, "step": 12240 }, { "epoch": 78.90763052208835, "grad_norm": 9.521042823791504, "learning_rate": 1.0512903225806452e-05, "loss": 0.576, "step": 12241 }, { "epoch": 78.9140562248996, "grad_norm": 21.43421173095703, "learning_rate": 1.0509677419354839e-05, "loss": 0.4589, "step": 12242 }, { "epoch": 78.92048192771084, "grad_norm": 18.326366424560547, "learning_rate": 1.0506451612903226e-05, "loss": 0.1428, "step": 12243 }, { "epoch": 78.9269076305221, "grad_norm": 21.709300994873047, "learning_rate": 1.0503225806451614e-05, "loss": 0.436, "step": 12244 }, { "epoch": 78.93333333333334, "grad_norm": 4.706326007843018, "learning_rate": 1.05e-05, "loss": 0.1573, "step": 12245 }, { "epoch": 78.93975903614458, "grad_norm": 21.682069778442383, "learning_rate": 1.0496774193548389e-05, "loss": 0.5516, "step": 12246 }, { "epoch": 78.94618473895582, "grad_norm": 12.470328330993652, "learning_rate": 1.0493548387096774e-05, "loss": 0.5263, "step": 12247 }, { "epoch": 78.95261044176706, "grad_norm": 13.142778396606445, "learning_rate": 1.0490322580645162e-05, "loss": 0.5687, "step": 12248 }, { "epoch": 78.95903614457832, "grad_norm": 12.954768180847168, "learning_rate": 1.0487096774193549e-05, "loss": 0.5369, "step": 12249 }, { "epoch": 78.96546184738956, "grad_norm": 8.51728630065918, "learning_rate": 1.0483870967741936e-05, "loss": 0.3354, "step": 12250 }, { "epoch": 78.9718875502008, "grad_norm": 21.30521583557129, "learning_rate": 1.0480645161290322e-05, "loss": 0.3691, "step": 12251 }, { "epoch": 78.97831325301205, "grad_norm": 11.072531700134277, "learning_rate": 1.0477419354838711e-05, "loss": 0.4447, "step": 12252 }, { "epoch": 78.98473895582329, "grad_norm": 22.876605987548828, "learning_rate": 1.0474193548387097e-05, "loss": 0.2757, "step": 12253 }, { "epoch": 78.99116465863453, "grad_norm": 15.740100860595703, "learning_rate": 1.0470967741935484e-05, "loss": 0.485, "step": 12254 }, { "epoch": 79.00080321285141, "grad_norm": 6.529194355010986, "learning_rate": 1.0467741935483872e-05, "loss": 0.3948, "step": 12255 }, { "epoch": 79.00722891566265, "grad_norm": 24.706811904907227, "learning_rate": 1.0464516129032259e-05, "loss": 0.1515, "step": 12256 }, { "epoch": 79.0136546184739, "grad_norm": 45.90966796875, "learning_rate": 1.0461290322580645e-05, "loss": 0.5125, "step": 12257 }, { "epoch": 79.02008032128514, "grad_norm": 21.354097366333008, "learning_rate": 1.0458064516129034e-05, "loss": 0.2596, "step": 12258 }, { "epoch": 79.02650602409639, "grad_norm": 4.2303147315979, "learning_rate": 1.045483870967742e-05, "loss": 0.2556, "step": 12259 }, { "epoch": 79.03293172690763, "grad_norm": 4.556866645812988, "learning_rate": 1.0451612903225807e-05, "loss": 0.1723, "step": 12260 }, { "epoch": 79.03935742971888, "grad_norm": 12.477094650268555, "learning_rate": 1.0448387096774194e-05, "loss": 0.2457, "step": 12261 }, { "epoch": 79.04578313253012, "grad_norm": 8.706198692321777, "learning_rate": 1.0445161290322582e-05, "loss": 0.2142, "step": 12262 }, { "epoch": 79.05220883534136, "grad_norm": 16.002830505371094, "learning_rate": 1.0441935483870968e-05, "loss": 0.6457, "step": 12263 }, { "epoch": 79.05863453815262, "grad_norm": 6.2293219566345215, "learning_rate": 1.0438709677419357e-05, "loss": 0.2094, "step": 12264 }, { "epoch": 79.06506024096386, "grad_norm": 14.630133628845215, "learning_rate": 1.0435483870967742e-05, "loss": 0.4324, "step": 12265 }, { "epoch": 79.0714859437751, "grad_norm": 16.614316940307617, "learning_rate": 1.0432258064516128e-05, "loss": 0.4592, "step": 12266 }, { "epoch": 79.07791164658634, "grad_norm": 11.812570571899414, "learning_rate": 1.0429032258064517e-05, "loss": 0.4087, "step": 12267 }, { "epoch": 79.08433734939759, "grad_norm": 17.801715850830078, "learning_rate": 1.0425806451612903e-05, "loss": 0.36, "step": 12268 }, { "epoch": 79.09076305220883, "grad_norm": 11.371745109558105, "learning_rate": 1.042258064516129e-05, "loss": 0.4857, "step": 12269 }, { "epoch": 79.09718875502008, "grad_norm": 8.167532920837402, "learning_rate": 1.0419354838709678e-05, "loss": 0.4477, "step": 12270 }, { "epoch": 79.10361445783133, "grad_norm": 13.014463424682617, "learning_rate": 1.0416129032258065e-05, "loss": 0.2366, "step": 12271 }, { "epoch": 79.11004016064257, "grad_norm": 6.01093053817749, "learning_rate": 1.041290322580645e-05, "loss": 0.272, "step": 12272 }, { "epoch": 79.11646586345381, "grad_norm": 28.812732696533203, "learning_rate": 1.040967741935484e-05, "loss": 0.5291, "step": 12273 }, { "epoch": 79.12289156626505, "grad_norm": 10.895368576049805, "learning_rate": 1.0406451612903226e-05, "loss": 0.2695, "step": 12274 }, { "epoch": 79.12931726907631, "grad_norm": 18.50235366821289, "learning_rate": 1.0403225806451613e-05, "loss": 0.3399, "step": 12275 }, { "epoch": 79.13574297188755, "grad_norm": 14.822324752807617, "learning_rate": 1.04e-05, "loss": 0.2396, "step": 12276 }, { "epoch": 79.1421686746988, "grad_norm": 17.335968017578125, "learning_rate": 1.0396774193548388e-05, "loss": 0.5105, "step": 12277 }, { "epoch": 79.14859437751004, "grad_norm": 8.717119216918945, "learning_rate": 1.0393548387096773e-05, "loss": 0.3209, "step": 12278 }, { "epoch": 79.15502008032128, "grad_norm": 10.427535057067871, "learning_rate": 1.0390322580645163e-05, "loss": 0.6286, "step": 12279 }, { "epoch": 79.16144578313254, "grad_norm": 5.058247089385986, "learning_rate": 1.0387096774193548e-05, "loss": 0.357, "step": 12280 }, { "epoch": 79.16787148594378, "grad_norm": 15.604729652404785, "learning_rate": 1.0383870967741936e-05, "loss": 0.2354, "step": 12281 }, { "epoch": 79.17429718875502, "grad_norm": 10.927157402038574, "learning_rate": 1.0380645161290323e-05, "loss": 0.2522, "step": 12282 }, { "epoch": 79.18072289156626, "grad_norm": 6.15898323059082, "learning_rate": 1.037741935483871e-05, "loss": 0.2373, "step": 12283 }, { "epoch": 79.1871485943775, "grad_norm": 1.9036158323287964, "learning_rate": 1.0374193548387096e-05, "loss": 0.0759, "step": 12284 }, { "epoch": 79.19357429718876, "grad_norm": 11.010747909545898, "learning_rate": 1.0370967741935485e-05, "loss": 0.4611, "step": 12285 }, { "epoch": 79.2, "grad_norm": 18.518455505371094, "learning_rate": 1.0367741935483871e-05, "loss": 0.2713, "step": 12286 }, { "epoch": 79.20642570281125, "grad_norm": 11.06638240814209, "learning_rate": 1.0364516129032258e-05, "loss": 0.3702, "step": 12287 }, { "epoch": 79.21285140562249, "grad_norm": 7.065963268280029, "learning_rate": 1.0361290322580646e-05, "loss": 0.3322, "step": 12288 }, { "epoch": 79.21927710843373, "grad_norm": 11.81475830078125, "learning_rate": 1.0358064516129033e-05, "loss": 0.2263, "step": 12289 }, { "epoch": 79.22570281124499, "grad_norm": 6.5353312492370605, "learning_rate": 1.0354838709677419e-05, "loss": 0.3224, "step": 12290 }, { "epoch": 79.23212851405623, "grad_norm": 13.228715896606445, "learning_rate": 1.0351612903225808e-05, "loss": 0.5948, "step": 12291 }, { "epoch": 79.23855421686747, "grad_norm": 8.920228004455566, "learning_rate": 1.0348387096774194e-05, "loss": 0.5754, "step": 12292 }, { "epoch": 79.24497991967871, "grad_norm": 8.100141525268555, "learning_rate": 1.0345161290322581e-05, "loss": 0.3438, "step": 12293 }, { "epoch": 79.25140562248995, "grad_norm": 5.53030252456665, "learning_rate": 1.0341935483870969e-05, "loss": 0.132, "step": 12294 }, { "epoch": 79.25783132530121, "grad_norm": 21.87738609313965, "learning_rate": 1.0338709677419356e-05, "loss": 0.3847, "step": 12295 }, { "epoch": 79.26425702811245, "grad_norm": 11.802020072937012, "learning_rate": 1.0335483870967742e-05, "loss": 0.3115, "step": 12296 }, { "epoch": 79.2706827309237, "grad_norm": 6.394933700561523, "learning_rate": 1.033225806451613e-05, "loss": 0.123, "step": 12297 }, { "epoch": 79.27710843373494, "grad_norm": 28.999454498291016, "learning_rate": 1.0329032258064516e-05, "loss": 0.253, "step": 12298 }, { "epoch": 79.28353413654618, "grad_norm": 17.681602478027344, "learning_rate": 1.0325806451612904e-05, "loss": 0.2402, "step": 12299 }, { "epoch": 79.28995983935744, "grad_norm": 10.693987846374512, "learning_rate": 1.0322580645161291e-05, "loss": 0.3912, "step": 12300 }, { "epoch": 79.29638554216868, "grad_norm": 8.58504581451416, "learning_rate": 1.0319354838709677e-05, "loss": 0.5394, "step": 12301 }, { "epoch": 79.30281124497992, "grad_norm": 5.595921039581299, "learning_rate": 1.0316129032258064e-05, "loss": 0.2203, "step": 12302 }, { "epoch": 79.30923694779116, "grad_norm": 14.945353507995605, "learning_rate": 1.0312903225806452e-05, "loss": 0.3645, "step": 12303 }, { "epoch": 79.3156626506024, "grad_norm": 18.76036262512207, "learning_rate": 1.0309677419354839e-05, "loss": 0.3692, "step": 12304 }, { "epoch": 79.32208835341365, "grad_norm": 12.113131523132324, "learning_rate": 1.0306451612903227e-05, "loss": 0.2562, "step": 12305 }, { "epoch": 79.3285140562249, "grad_norm": 12.176321029663086, "learning_rate": 1.0303225806451614e-05, "loss": 0.2165, "step": 12306 }, { "epoch": 79.33493975903615, "grad_norm": 9.519885063171387, "learning_rate": 1.03e-05, "loss": 0.4444, "step": 12307 }, { "epoch": 79.34136546184739, "grad_norm": 10.251164436340332, "learning_rate": 1.0296774193548387e-05, "loss": 0.3425, "step": 12308 }, { "epoch": 79.34779116465863, "grad_norm": 12.530195236206055, "learning_rate": 1.0293548387096774e-05, "loss": 0.3196, "step": 12309 }, { "epoch": 79.35421686746987, "grad_norm": 22.936853408813477, "learning_rate": 1.0290322580645162e-05, "loss": 0.4012, "step": 12310 }, { "epoch": 79.36064257028113, "grad_norm": 5.872519016265869, "learning_rate": 1.0287096774193548e-05, "loss": 0.2539, "step": 12311 }, { "epoch": 79.36706827309237, "grad_norm": 8.432003021240234, "learning_rate": 1.0283870967741937e-05, "loss": 0.2933, "step": 12312 }, { "epoch": 79.37349397590361, "grad_norm": 10.554841041564941, "learning_rate": 1.0280645161290322e-05, "loss": 0.7836, "step": 12313 }, { "epoch": 79.37991967871486, "grad_norm": 7.3376688957214355, "learning_rate": 1.027741935483871e-05, "loss": 0.1906, "step": 12314 }, { "epoch": 79.3863453815261, "grad_norm": 13.50447940826416, "learning_rate": 1.0274193548387097e-05, "loss": 0.2319, "step": 12315 }, { "epoch": 79.39277108433735, "grad_norm": 8.7621488571167, "learning_rate": 1.0270967741935485e-05, "loss": 0.3715, "step": 12316 }, { "epoch": 79.3991967871486, "grad_norm": 17.046371459960938, "learning_rate": 1.026774193548387e-05, "loss": 0.5212, "step": 12317 }, { "epoch": 79.40562248995984, "grad_norm": 14.606476783752441, "learning_rate": 1.026451612903226e-05, "loss": 0.4295, "step": 12318 }, { "epoch": 79.41204819277108, "grad_norm": 9.807774543762207, "learning_rate": 1.0261290322580645e-05, "loss": 0.2512, "step": 12319 }, { "epoch": 79.41847389558232, "grad_norm": 10.48374080657959, "learning_rate": 1.0258064516129032e-05, "loss": 0.5039, "step": 12320 }, { "epoch": 79.42489959839358, "grad_norm": 10.383699417114258, "learning_rate": 1.025483870967742e-05, "loss": 0.4272, "step": 12321 }, { "epoch": 79.43132530120482, "grad_norm": 13.304255485534668, "learning_rate": 1.0251612903225807e-05, "loss": 0.3835, "step": 12322 }, { "epoch": 79.43775100401606, "grad_norm": 18.228487014770508, "learning_rate": 1.0248387096774193e-05, "loss": 0.2246, "step": 12323 }, { "epoch": 79.4441767068273, "grad_norm": 11.014293670654297, "learning_rate": 1.0245161290322582e-05, "loss": 0.4167, "step": 12324 }, { "epoch": 79.45060240963855, "grad_norm": 8.304905891418457, "learning_rate": 1.0241935483870968e-05, "loss": 0.2586, "step": 12325 }, { "epoch": 79.4570281124498, "grad_norm": 6.001856327056885, "learning_rate": 1.0238709677419355e-05, "loss": 0.2874, "step": 12326 }, { "epoch": 79.46345381526105, "grad_norm": 6.47868537902832, "learning_rate": 1.0235483870967743e-05, "loss": 0.2787, "step": 12327 }, { "epoch": 79.46987951807229, "grad_norm": 6.417140007019043, "learning_rate": 1.023225806451613e-05, "loss": 0.4401, "step": 12328 }, { "epoch": 79.47630522088353, "grad_norm": 10.048418998718262, "learning_rate": 1.0229032258064516e-05, "loss": 0.575, "step": 12329 }, { "epoch": 79.48273092369477, "grad_norm": 9.320355415344238, "learning_rate": 1.0225806451612905e-05, "loss": 0.3278, "step": 12330 }, { "epoch": 79.48915662650603, "grad_norm": 6.770679950714111, "learning_rate": 1.022258064516129e-05, "loss": 0.453, "step": 12331 }, { "epoch": 79.49558232931727, "grad_norm": 6.652153015136719, "learning_rate": 1.0219354838709678e-05, "loss": 0.1553, "step": 12332 }, { "epoch": 79.50200803212851, "grad_norm": 9.603610038757324, "learning_rate": 1.0216129032258065e-05, "loss": 0.2198, "step": 12333 }, { "epoch": 79.50843373493976, "grad_norm": 15.951197624206543, "learning_rate": 1.0212903225806453e-05, "loss": 0.474, "step": 12334 }, { "epoch": 79.514859437751, "grad_norm": 13.255653381347656, "learning_rate": 1.0209677419354838e-05, "loss": 0.5585, "step": 12335 }, { "epoch": 79.52128514056226, "grad_norm": 14.626747131347656, "learning_rate": 1.0206451612903227e-05, "loss": 0.5115, "step": 12336 }, { "epoch": 79.5277108433735, "grad_norm": 4.7896342277526855, "learning_rate": 1.0203225806451613e-05, "loss": 0.1128, "step": 12337 }, { "epoch": 79.53413654618474, "grad_norm": 7.218379974365234, "learning_rate": 1.02e-05, "loss": 0.142, "step": 12338 }, { "epoch": 79.54056224899598, "grad_norm": 16.441303253173828, "learning_rate": 1.0196774193548388e-05, "loss": 0.3509, "step": 12339 }, { "epoch": 79.54698795180722, "grad_norm": 5.948690891265869, "learning_rate": 1.0193548387096774e-05, "loss": 0.1685, "step": 12340 }, { "epoch": 79.55341365461847, "grad_norm": 7.103861331939697, "learning_rate": 1.0190322580645161e-05, "loss": 0.344, "step": 12341 }, { "epoch": 79.55983935742972, "grad_norm": 7.421385765075684, "learning_rate": 1.0187096774193549e-05, "loss": 0.2155, "step": 12342 }, { "epoch": 79.56626506024097, "grad_norm": 4.095909595489502, "learning_rate": 1.0183870967741936e-05, "loss": 0.4148, "step": 12343 }, { "epoch": 79.57269076305221, "grad_norm": 13.631425857543945, "learning_rate": 1.0180645161290323e-05, "loss": 0.1675, "step": 12344 }, { "epoch": 79.57911646586345, "grad_norm": 22.31906509399414, "learning_rate": 1.017741935483871e-05, "loss": 0.5563, "step": 12345 }, { "epoch": 79.58554216867469, "grad_norm": 10.000852584838867, "learning_rate": 1.0174193548387096e-05, "loss": 0.2704, "step": 12346 }, { "epoch": 79.59196787148595, "grad_norm": 7.21949577331543, "learning_rate": 1.0170967741935484e-05, "loss": 0.2116, "step": 12347 }, { "epoch": 79.59839357429719, "grad_norm": 8.698410987854004, "learning_rate": 1.0167741935483871e-05, "loss": 0.2531, "step": 12348 }, { "epoch": 79.60481927710843, "grad_norm": 4.487123012542725, "learning_rate": 1.0164516129032259e-05, "loss": 0.3486, "step": 12349 }, { "epoch": 79.61124497991968, "grad_norm": 13.72161865234375, "learning_rate": 1.0161290322580646e-05, "loss": 0.3028, "step": 12350 }, { "epoch": 79.61767068273092, "grad_norm": 6.679986953735352, "learning_rate": 1.0158064516129033e-05, "loss": 0.2815, "step": 12351 }, { "epoch": 79.62409638554217, "grad_norm": 51.62168502807617, "learning_rate": 1.0154838709677419e-05, "loss": 0.3385, "step": 12352 }, { "epoch": 79.63052208835342, "grad_norm": 9.892707824707031, "learning_rate": 1.0151612903225807e-05, "loss": 0.2275, "step": 12353 }, { "epoch": 79.63694779116466, "grad_norm": 14.400070190429688, "learning_rate": 1.0148387096774194e-05, "loss": 0.3655, "step": 12354 }, { "epoch": 79.6433734939759, "grad_norm": 21.56974220275879, "learning_rate": 1.0145161290322581e-05, "loss": 0.521, "step": 12355 }, { "epoch": 79.64979919678714, "grad_norm": 5.210620403289795, "learning_rate": 1.0141935483870969e-05, "loss": 0.2711, "step": 12356 }, { "epoch": 79.6562248995984, "grad_norm": 8.817724227905273, "learning_rate": 1.0138709677419356e-05, "loss": 0.2486, "step": 12357 }, { "epoch": 79.66265060240964, "grad_norm": 7.599261283874512, "learning_rate": 1.0135483870967742e-05, "loss": 0.4093, "step": 12358 }, { "epoch": 79.66907630522088, "grad_norm": 16.035737991333008, "learning_rate": 1.013225806451613e-05, "loss": 0.3993, "step": 12359 }, { "epoch": 79.67550200803213, "grad_norm": 8.73398494720459, "learning_rate": 1.0129032258064517e-05, "loss": 0.313, "step": 12360 }, { "epoch": 79.68192771084337, "grad_norm": 6.795543193817139, "learning_rate": 1.0125806451612904e-05, "loss": 0.2686, "step": 12361 }, { "epoch": 79.68835341365462, "grad_norm": 8.09697151184082, "learning_rate": 1.0122580645161291e-05, "loss": 0.209, "step": 12362 }, { "epoch": 79.69477911646587, "grad_norm": 20.838695526123047, "learning_rate": 1.0119354838709679e-05, "loss": 0.3658, "step": 12363 }, { "epoch": 79.70120481927711, "grad_norm": 17.671724319458008, "learning_rate": 1.0116129032258065e-05, "loss": 0.4745, "step": 12364 }, { "epoch": 79.70763052208835, "grad_norm": 6.3593010902404785, "learning_rate": 1.0112903225806452e-05, "loss": 0.2404, "step": 12365 }, { "epoch": 79.7140562248996, "grad_norm": 11.601086616516113, "learning_rate": 1.010967741935484e-05, "loss": 0.1547, "step": 12366 }, { "epoch": 79.72048192771085, "grad_norm": 7.617368221282959, "learning_rate": 1.0106451612903227e-05, "loss": 0.2277, "step": 12367 }, { "epoch": 79.72690763052209, "grad_norm": 4.76111364364624, "learning_rate": 1.0103225806451614e-05, "loss": 0.1515, "step": 12368 }, { "epoch": 79.73333333333333, "grad_norm": 11.589303970336914, "learning_rate": 1.0100000000000002e-05, "loss": 0.4195, "step": 12369 }, { "epoch": 79.73975903614458, "grad_norm": 10.199393272399902, "learning_rate": 1.0096774193548387e-05, "loss": 0.3132, "step": 12370 }, { "epoch": 79.74618473895582, "grad_norm": 12.927929878234863, "learning_rate": 1.0093548387096775e-05, "loss": 0.2749, "step": 12371 }, { "epoch": 79.75261044176708, "grad_norm": 10.10861873626709, "learning_rate": 1.0090322580645162e-05, "loss": 0.5384, "step": 12372 }, { "epoch": 79.75903614457832, "grad_norm": 11.128927230834961, "learning_rate": 1.0087096774193548e-05, "loss": 0.3592, "step": 12373 }, { "epoch": 79.76546184738956, "grad_norm": 5.965092658996582, "learning_rate": 1.0083870967741935e-05, "loss": 0.1905, "step": 12374 }, { "epoch": 79.7718875502008, "grad_norm": 8.373833656311035, "learning_rate": 1.0080645161290323e-05, "loss": 0.4089, "step": 12375 }, { "epoch": 79.77831325301204, "grad_norm": 2.931881904602051, "learning_rate": 1.007741935483871e-05, "loss": 0.1033, "step": 12376 }, { "epoch": 79.78473895582329, "grad_norm": 13.307466506958008, "learning_rate": 1.0074193548387097e-05, "loss": 0.2869, "step": 12377 }, { "epoch": 79.79116465863454, "grad_norm": 9.861356735229492, "learning_rate": 1.0070967741935485e-05, "loss": 0.3144, "step": 12378 }, { "epoch": 79.79759036144578, "grad_norm": 7.33074951171875, "learning_rate": 1.006774193548387e-05, "loss": 0.2987, "step": 12379 }, { "epoch": 79.80401606425703, "grad_norm": 6.87556791305542, "learning_rate": 1.0064516129032258e-05, "loss": 0.4343, "step": 12380 }, { "epoch": 79.81044176706827, "grad_norm": 9.448430061340332, "learning_rate": 1.0061290322580645e-05, "loss": 0.2256, "step": 12381 }, { "epoch": 79.81686746987951, "grad_norm": 4.622623920440674, "learning_rate": 1.0058064516129033e-05, "loss": 0.1217, "step": 12382 }, { "epoch": 79.82329317269077, "grad_norm": 6.816863059997559, "learning_rate": 1.005483870967742e-05, "loss": 0.3676, "step": 12383 }, { "epoch": 79.82971887550201, "grad_norm": 3.47499418258667, "learning_rate": 1.0051612903225807e-05, "loss": 0.0666, "step": 12384 }, { "epoch": 79.83614457831325, "grad_norm": 12.951236724853516, "learning_rate": 1.0048387096774193e-05, "loss": 0.3968, "step": 12385 }, { "epoch": 79.8425702811245, "grad_norm": 7.287737846374512, "learning_rate": 1.004516129032258e-05, "loss": 0.2767, "step": 12386 }, { "epoch": 79.84899598393574, "grad_norm": 13.986059188842773, "learning_rate": 1.0041935483870968e-05, "loss": 0.5887, "step": 12387 }, { "epoch": 79.855421686747, "grad_norm": 15.717768669128418, "learning_rate": 1.0038709677419355e-05, "loss": 0.3813, "step": 12388 }, { "epoch": 79.86184738955824, "grad_norm": 12.56949234008789, "learning_rate": 1.0035483870967743e-05, "loss": 0.2214, "step": 12389 }, { "epoch": 79.86827309236948, "grad_norm": 10.237641334533691, "learning_rate": 1.003225806451613e-05, "loss": 0.5022, "step": 12390 }, { "epoch": 79.87469879518072, "grad_norm": 13.44831657409668, "learning_rate": 1.0029032258064516e-05, "loss": 0.4092, "step": 12391 }, { "epoch": 79.88112449799196, "grad_norm": 5.586889743804932, "learning_rate": 1.0025806451612903e-05, "loss": 0.4179, "step": 12392 }, { "epoch": 79.88755020080322, "grad_norm": 30.69989013671875, "learning_rate": 1.002258064516129e-05, "loss": 0.4213, "step": 12393 }, { "epoch": 79.89397590361446, "grad_norm": 7.789387226104736, "learning_rate": 1.0019354838709678e-05, "loss": 0.3776, "step": 12394 }, { "epoch": 79.9004016064257, "grad_norm": 7.938144683837891, "learning_rate": 1.0016129032258066e-05, "loss": 0.4333, "step": 12395 }, { "epoch": 79.90682730923695, "grad_norm": 8.365467071533203, "learning_rate": 1.0012903225806453e-05, "loss": 0.5478, "step": 12396 }, { "epoch": 79.91325301204819, "grad_norm": 18.459001541137695, "learning_rate": 1.0009677419354839e-05, "loss": 0.2894, "step": 12397 }, { "epoch": 79.91967871485944, "grad_norm": 15.56529426574707, "learning_rate": 1.0006451612903226e-05, "loss": 0.5389, "step": 12398 }, { "epoch": 79.92610441767069, "grad_norm": 9.508841514587402, "learning_rate": 1.0003225806451613e-05, "loss": 0.3504, "step": 12399 }, { "epoch": 79.93253012048193, "grad_norm": 9.607430458068848, "learning_rate": 1e-05, "loss": 0.3968, "step": 12400 }, { "epoch": 79.93895582329317, "grad_norm": 8.110116004943848, "learning_rate": 9.996774193548388e-06, "loss": 0.2064, "step": 12401 }, { "epoch": 79.94538152610441, "grad_norm": 8.781006813049316, "learning_rate": 9.993548387096776e-06, "loss": 0.3602, "step": 12402 }, { "epoch": 79.95180722891567, "grad_norm": 13.049358367919922, "learning_rate": 9.990322580645161e-06, "loss": 0.5509, "step": 12403 }, { "epoch": 79.95823293172691, "grad_norm": 14.580856323242188, "learning_rate": 9.987096774193549e-06, "loss": 0.5048, "step": 12404 }, { "epoch": 79.96465863453815, "grad_norm": 6.324873924255371, "learning_rate": 9.983870967741936e-06, "loss": 0.2441, "step": 12405 }, { "epoch": 79.9710843373494, "grad_norm": 8.587529182434082, "learning_rate": 9.980645161290324e-06, "loss": 0.3366, "step": 12406 }, { "epoch": 79.97751004016064, "grad_norm": 5.549382209777832, "learning_rate": 9.977419354838711e-06, "loss": 0.3789, "step": 12407 }, { "epoch": 79.9839357429719, "grad_norm": 6.9449262619018555, "learning_rate": 9.974193548387097e-06, "loss": 0.2168, "step": 12408 }, { "epoch": 79.99036144578314, "grad_norm": 9.919812202453613, "learning_rate": 9.970967741935484e-06, "loss": 0.2871, "step": 12409 }, { "epoch": 79.99678714859438, "grad_norm": 13.198134422302246, "learning_rate": 9.967741935483871e-06, "loss": 0.3919, "step": 12410 }, { "epoch": 80.00642570281124, "grad_norm": 8.080090522766113, "learning_rate": 9.964516129032259e-06, "loss": 0.3414, "step": 12411 }, { "epoch": 80.01285140562248, "grad_norm": 6.511715412139893, "learning_rate": 9.961290322580645e-06, "loss": 0.0829, "step": 12412 }, { "epoch": 80.01927710843374, "grad_norm": 10.260003089904785, "learning_rate": 9.958064516129034e-06, "loss": 0.3611, "step": 12413 }, { "epoch": 80.02570281124498, "grad_norm": 6.82750940322876, "learning_rate": 9.95483870967742e-06, "loss": 0.2806, "step": 12414 }, { "epoch": 80.03212851405623, "grad_norm": 22.277482986450195, "learning_rate": 9.951612903225807e-06, "loss": 0.3359, "step": 12415 }, { "epoch": 80.03855421686747, "grad_norm": 6.447524070739746, "learning_rate": 9.948387096774194e-06, "loss": 0.3248, "step": 12416 }, { "epoch": 80.04497991967871, "grad_norm": 8.480523109436035, "learning_rate": 9.945161290322582e-06, "loss": 0.3359, "step": 12417 }, { "epoch": 80.05140562248997, "grad_norm": 4.0681681632995605, "learning_rate": 9.941935483870967e-06, "loss": 0.253, "step": 12418 }, { "epoch": 80.05783132530121, "grad_norm": 7.291290760040283, "learning_rate": 9.938709677419356e-06, "loss": 0.3776, "step": 12419 }, { "epoch": 80.06425702811245, "grad_norm": 7.234158992767334, "learning_rate": 9.935483870967742e-06, "loss": 0.1628, "step": 12420 }, { "epoch": 80.07068273092369, "grad_norm": 10.641410827636719, "learning_rate": 9.93225806451613e-06, "loss": 0.5225, "step": 12421 }, { "epoch": 80.07710843373494, "grad_norm": 7.510659694671631, "learning_rate": 9.929032258064517e-06, "loss": 0.3901, "step": 12422 }, { "epoch": 80.08353413654619, "grad_norm": 7.665186405181885, "learning_rate": 9.925806451612904e-06, "loss": 0.1772, "step": 12423 }, { "epoch": 80.08995983935743, "grad_norm": 11.379207611083984, "learning_rate": 9.92258064516129e-06, "loss": 0.4109, "step": 12424 }, { "epoch": 80.09638554216868, "grad_norm": 11.880132675170898, "learning_rate": 9.919354838709679e-06, "loss": 0.2726, "step": 12425 }, { "epoch": 80.10281124497992, "grad_norm": 12.094263076782227, "learning_rate": 9.916129032258065e-06, "loss": 0.3408, "step": 12426 }, { "epoch": 80.10923694779116, "grad_norm": 12.987117767333984, "learning_rate": 9.912903225806452e-06, "loss": 0.482, "step": 12427 }, { "epoch": 80.1156626506024, "grad_norm": 18.40212059020996, "learning_rate": 9.90967741935484e-06, "loss": 0.3399, "step": 12428 }, { "epoch": 80.12208835341366, "grad_norm": 8.123409271240234, "learning_rate": 9.906451612903227e-06, "loss": 0.2956, "step": 12429 }, { "epoch": 80.1285140562249, "grad_norm": 20.138906478881836, "learning_rate": 9.903225806451613e-06, "loss": 0.5312, "step": 12430 }, { "epoch": 80.13493975903614, "grad_norm": 4.302764415740967, "learning_rate": 9.900000000000002e-06, "loss": 0.0962, "step": 12431 }, { "epoch": 80.14136546184739, "grad_norm": 6.982243537902832, "learning_rate": 9.896774193548387e-06, "loss": 0.2924, "step": 12432 }, { "epoch": 80.14779116465863, "grad_norm": 6.534780502319336, "learning_rate": 9.893548387096775e-06, "loss": 0.2883, "step": 12433 }, { "epoch": 80.15421686746988, "grad_norm": 9.324395179748535, "learning_rate": 9.890322580645162e-06, "loss": 0.4444, "step": 12434 }, { "epoch": 80.16064257028113, "grad_norm": 7.88824987411499, "learning_rate": 9.88709677419355e-06, "loss": 0.265, "step": 12435 }, { "epoch": 80.16706827309237, "grad_norm": 6.275410175323486, "learning_rate": 9.883870967741935e-06, "loss": 0.144, "step": 12436 }, { "epoch": 80.17349397590361, "grad_norm": 17.709157943725586, "learning_rate": 9.880645161290323e-06, "loss": 0.2526, "step": 12437 }, { "epoch": 80.17991967871485, "grad_norm": 9.730137825012207, "learning_rate": 9.87741935483871e-06, "loss": 0.3169, "step": 12438 }, { "epoch": 80.18634538152611, "grad_norm": 7.565020561218262, "learning_rate": 9.874193548387098e-06, "loss": 0.3691, "step": 12439 }, { "epoch": 80.19277108433735, "grad_norm": 5.209322929382324, "learning_rate": 9.870967741935485e-06, "loss": 0.1402, "step": 12440 }, { "epoch": 80.1991967871486, "grad_norm": 11.632399559020996, "learning_rate": 9.867741935483872e-06, "loss": 0.3861, "step": 12441 }, { "epoch": 80.20562248995984, "grad_norm": 12.060176849365234, "learning_rate": 9.864516129032258e-06, "loss": 0.4819, "step": 12442 }, { "epoch": 80.21204819277108, "grad_norm": 12.728686332702637, "learning_rate": 9.861290322580646e-06, "loss": 0.276, "step": 12443 }, { "epoch": 80.21847389558233, "grad_norm": 5.640647888183594, "learning_rate": 9.858064516129033e-06, "loss": 0.2565, "step": 12444 }, { "epoch": 80.22489959839358, "grad_norm": 24.000337600708008, "learning_rate": 9.854838709677419e-06, "loss": 0.6529, "step": 12445 }, { "epoch": 80.23132530120482, "grad_norm": 11.41061019897461, "learning_rate": 9.851612903225808e-06, "loss": 0.2594, "step": 12446 }, { "epoch": 80.23775100401606, "grad_norm": 4.969854831695557, "learning_rate": 9.848387096774193e-06, "loss": 0.1057, "step": 12447 }, { "epoch": 80.2441767068273, "grad_norm": 11.697362899780273, "learning_rate": 9.84516129032258e-06, "loss": 0.4213, "step": 12448 }, { "epoch": 80.25060240963856, "grad_norm": 11.480378150939941, "learning_rate": 9.841935483870968e-06, "loss": 0.2595, "step": 12449 }, { "epoch": 80.2570281124498, "grad_norm": 8.530617713928223, "learning_rate": 9.838709677419356e-06, "loss": 0.4027, "step": 12450 }, { "epoch": 80.26345381526104, "grad_norm": 9.553887367248535, "learning_rate": 9.835483870967741e-06, "loss": 0.3585, "step": 12451 }, { "epoch": 80.26987951807229, "grad_norm": 3.5006113052368164, "learning_rate": 9.83225806451613e-06, "loss": 0.0899, "step": 12452 }, { "epoch": 80.27630522088353, "grad_norm": 10.478350639343262, "learning_rate": 9.829032258064516e-06, "loss": 0.2988, "step": 12453 }, { "epoch": 80.28273092369479, "grad_norm": 16.863012313842773, "learning_rate": 9.825806451612904e-06, "loss": 0.3261, "step": 12454 }, { "epoch": 80.28915662650603, "grad_norm": 14.011431694030762, "learning_rate": 9.822580645161291e-06, "loss": 0.2816, "step": 12455 }, { "epoch": 80.29558232931727, "grad_norm": 8.797995567321777, "learning_rate": 9.819354838709678e-06, "loss": 0.2469, "step": 12456 }, { "epoch": 80.30200803212851, "grad_norm": 7.767399311065674, "learning_rate": 9.816129032258064e-06, "loss": 0.1853, "step": 12457 }, { "epoch": 80.30843373493975, "grad_norm": 15.385284423828125, "learning_rate": 9.812903225806453e-06, "loss": 0.4075, "step": 12458 }, { "epoch": 80.31485943775101, "grad_norm": 12.46799087524414, "learning_rate": 9.809677419354839e-06, "loss": 0.4051, "step": 12459 }, { "epoch": 80.32128514056225, "grad_norm": 5.815365314483643, "learning_rate": 9.806451612903226e-06, "loss": 0.2243, "step": 12460 }, { "epoch": 80.3277108433735, "grad_norm": 13.627558708190918, "learning_rate": 9.803225806451614e-06, "loss": 0.38, "step": 12461 }, { "epoch": 80.33413654618474, "grad_norm": 10.821681022644043, "learning_rate": 9.800000000000001e-06, "loss": 0.5381, "step": 12462 }, { "epoch": 80.34056224899598, "grad_norm": 6.546141147613525, "learning_rate": 9.796774193548387e-06, "loss": 0.2596, "step": 12463 }, { "epoch": 80.34698795180722, "grad_norm": 8.834405899047852, "learning_rate": 9.793548387096776e-06, "loss": 0.4732, "step": 12464 }, { "epoch": 80.35341365461848, "grad_norm": 9.239470481872559, "learning_rate": 9.790322580645162e-06, "loss": 0.5463, "step": 12465 }, { "epoch": 80.35983935742972, "grad_norm": 10.05118465423584, "learning_rate": 9.787096774193549e-06, "loss": 0.3873, "step": 12466 }, { "epoch": 80.36626506024096, "grad_norm": 8.727070808410645, "learning_rate": 9.783870967741936e-06, "loss": 0.2796, "step": 12467 }, { "epoch": 80.3726907630522, "grad_norm": 9.80595874786377, "learning_rate": 9.780645161290324e-06, "loss": 0.3631, "step": 12468 }, { "epoch": 80.37911646586345, "grad_norm": 7.761882305145264, "learning_rate": 9.77741935483871e-06, "loss": 0.4784, "step": 12469 }, { "epoch": 80.3855421686747, "grad_norm": 5.727697372436523, "learning_rate": 9.774193548387099e-06, "loss": 0.3291, "step": 12470 }, { "epoch": 80.39196787148595, "grad_norm": 10.550575256347656, "learning_rate": 9.770967741935484e-06, "loss": 0.4366, "step": 12471 }, { "epoch": 80.39839357429719, "grad_norm": 9.114117622375488, "learning_rate": 9.767741935483872e-06, "loss": 0.4044, "step": 12472 }, { "epoch": 80.40481927710843, "grad_norm": 4.457929611206055, "learning_rate": 9.764516129032259e-06, "loss": 0.1306, "step": 12473 }, { "epoch": 80.41124497991967, "grad_norm": 9.086007118225098, "learning_rate": 9.761290322580646e-06, "loss": 0.2217, "step": 12474 }, { "epoch": 80.41767068273093, "grad_norm": 6.479095935821533, "learning_rate": 9.758064516129032e-06, "loss": 0.2844, "step": 12475 }, { "epoch": 80.42409638554217, "grad_norm": 10.131092071533203, "learning_rate": 9.754838709677421e-06, "loss": 0.3783, "step": 12476 }, { "epoch": 80.43052208835341, "grad_norm": 30.403076171875, "learning_rate": 9.751612903225807e-06, "loss": 0.4833, "step": 12477 }, { "epoch": 80.43694779116466, "grad_norm": 16.141056060791016, "learning_rate": 9.748387096774194e-06, "loss": 0.3928, "step": 12478 }, { "epoch": 80.4433734939759, "grad_norm": 39.18837356567383, "learning_rate": 9.745161290322582e-06, "loss": 0.4231, "step": 12479 }, { "epoch": 80.44979919678715, "grad_norm": 5.021612167358398, "learning_rate": 9.741935483870967e-06, "loss": 0.3005, "step": 12480 }, { "epoch": 80.4562248995984, "grad_norm": 13.852899551391602, "learning_rate": 9.738709677419355e-06, "loss": 0.4184, "step": 12481 }, { "epoch": 80.46265060240964, "grad_norm": 9.713211059570312, "learning_rate": 9.735483870967742e-06, "loss": 0.37, "step": 12482 }, { "epoch": 80.46907630522088, "grad_norm": 5.278839588165283, "learning_rate": 9.73225806451613e-06, "loss": 0.1683, "step": 12483 }, { "epoch": 80.47550200803212, "grad_norm": 7.237608909606934, "learning_rate": 9.729032258064515e-06, "loss": 0.3918, "step": 12484 }, { "epoch": 80.48192771084338, "grad_norm": 22.545310974121094, "learning_rate": 9.725806451612904e-06, "loss": 0.2984, "step": 12485 }, { "epoch": 80.48835341365462, "grad_norm": 7.651182174682617, "learning_rate": 9.72258064516129e-06, "loss": 0.2916, "step": 12486 }, { "epoch": 80.49477911646586, "grad_norm": 67.4669189453125, "learning_rate": 9.719354838709678e-06, "loss": 0.3245, "step": 12487 }, { "epoch": 80.5012048192771, "grad_norm": 8.661114692687988, "learning_rate": 9.716129032258065e-06, "loss": 0.3456, "step": 12488 }, { "epoch": 80.50763052208835, "grad_norm": 3.2411348819732666, "learning_rate": 9.712903225806452e-06, "loss": 0.0945, "step": 12489 }, { "epoch": 80.5140562248996, "grad_norm": 26.24892234802246, "learning_rate": 9.709677419354838e-06, "loss": 0.3621, "step": 12490 }, { "epoch": 80.52048192771085, "grad_norm": 16.62833595275879, "learning_rate": 9.706451612903227e-06, "loss": 0.3188, "step": 12491 }, { "epoch": 80.52690763052209, "grad_norm": 9.842142105102539, "learning_rate": 9.703225806451613e-06, "loss": 0.4676, "step": 12492 }, { "epoch": 80.53333333333333, "grad_norm": 8.243788719177246, "learning_rate": 9.7e-06, "loss": 0.4, "step": 12493 }, { "epoch": 80.53975903614457, "grad_norm": 11.204283714294434, "learning_rate": 9.696774193548388e-06, "loss": 0.586, "step": 12494 }, { "epoch": 80.54618473895583, "grad_norm": 7.886009216308594, "learning_rate": 9.693548387096775e-06, "loss": 0.4121, "step": 12495 }, { "epoch": 80.55261044176707, "grad_norm": 9.496927261352539, "learning_rate": 9.69032258064516e-06, "loss": 0.5316, "step": 12496 }, { "epoch": 80.55903614457831, "grad_norm": 12.529535293579102, "learning_rate": 9.68709677419355e-06, "loss": 0.421, "step": 12497 }, { "epoch": 80.56546184738956, "grad_norm": 9.243574142456055, "learning_rate": 9.683870967741936e-06, "loss": 0.2523, "step": 12498 }, { "epoch": 80.5718875502008, "grad_norm": 3.9664723873138428, "learning_rate": 9.680645161290323e-06, "loss": 0.1911, "step": 12499 }, { "epoch": 80.57831325301204, "grad_norm": 6.6744585037231445, "learning_rate": 9.67741935483871e-06, "loss": 0.2444, "step": 12500 }, { "epoch": 80.5847389558233, "grad_norm": 17.444164276123047, "learning_rate": 9.674193548387098e-06, "loss": 0.4835, "step": 12501 }, { "epoch": 80.59116465863454, "grad_norm": 6.457406520843506, "learning_rate": 9.670967741935484e-06, "loss": 0.2426, "step": 12502 }, { "epoch": 80.59759036144578, "grad_norm": 23.755239486694336, "learning_rate": 9.667741935483873e-06, "loss": 0.3287, "step": 12503 }, { "epoch": 80.60401606425702, "grad_norm": 12.146163940429688, "learning_rate": 9.664516129032258e-06, "loss": 0.8601, "step": 12504 }, { "epoch": 80.61044176706827, "grad_norm": 8.044696807861328, "learning_rate": 9.661290322580646e-06, "loss": 0.228, "step": 12505 }, { "epoch": 80.61686746987952, "grad_norm": 8.812767028808594, "learning_rate": 9.658064516129033e-06, "loss": 0.1957, "step": 12506 }, { "epoch": 80.62329317269077, "grad_norm": 7.8762969970703125, "learning_rate": 9.65483870967742e-06, "loss": 0.1402, "step": 12507 }, { "epoch": 80.62971887550201, "grad_norm": 8.52807331085205, "learning_rate": 9.651612903225806e-06, "loss": 0.3696, "step": 12508 }, { "epoch": 80.63614457831325, "grad_norm": 5.0191826820373535, "learning_rate": 9.648387096774195e-06, "loss": 0.1186, "step": 12509 }, { "epoch": 80.64257028112449, "grad_norm": 6.587648391723633, "learning_rate": 9.645161290322581e-06, "loss": 0.1504, "step": 12510 }, { "epoch": 80.64899598393575, "grad_norm": 6.378141403198242, "learning_rate": 9.641935483870968e-06, "loss": 0.2213, "step": 12511 }, { "epoch": 80.65542168674699, "grad_norm": 14.755125999450684, "learning_rate": 9.638709677419356e-06, "loss": 0.5452, "step": 12512 }, { "epoch": 80.66184738955823, "grad_norm": 13.643917083740234, "learning_rate": 9.635483870967743e-06, "loss": 0.4837, "step": 12513 }, { "epoch": 80.66827309236947, "grad_norm": 15.65616226196289, "learning_rate": 9.632258064516129e-06, "loss": 0.5295, "step": 12514 }, { "epoch": 80.67469879518072, "grad_norm": 6.982420444488525, "learning_rate": 9.629032258064516e-06, "loss": 0.4054, "step": 12515 }, { "epoch": 80.68112449799197, "grad_norm": 9.566182136535645, "learning_rate": 9.625806451612904e-06, "loss": 0.6435, "step": 12516 }, { "epoch": 80.68755020080322, "grad_norm": 11.258644104003906, "learning_rate": 9.62258064516129e-06, "loss": 0.4374, "step": 12517 }, { "epoch": 80.69397590361446, "grad_norm": 4.5430097579956055, "learning_rate": 9.619354838709679e-06, "loss": 0.2319, "step": 12518 }, { "epoch": 80.7004016064257, "grad_norm": 14.106711387634277, "learning_rate": 9.616129032258064e-06, "loss": 0.2292, "step": 12519 }, { "epoch": 80.70682730923694, "grad_norm": 8.712564468383789, "learning_rate": 9.612903225806452e-06, "loss": 0.3139, "step": 12520 }, { "epoch": 80.7132530120482, "grad_norm": 9.21323013305664, "learning_rate": 9.609677419354839e-06, "loss": 0.1854, "step": 12521 }, { "epoch": 80.71967871485944, "grad_norm": 13.25653076171875, "learning_rate": 9.606451612903226e-06, "loss": 0.5786, "step": 12522 }, { "epoch": 80.72610441767068, "grad_norm": 28.11932945251465, "learning_rate": 9.603225806451612e-06, "loss": 0.668, "step": 12523 }, { "epoch": 80.73253012048193, "grad_norm": 12.695836067199707, "learning_rate": 9.600000000000001e-06, "loss": 0.3375, "step": 12524 }, { "epoch": 80.73895582329317, "grad_norm": 9.967041015625, "learning_rate": 9.596774193548387e-06, "loss": 0.1895, "step": 12525 }, { "epoch": 80.74538152610442, "grad_norm": 6.308195114135742, "learning_rate": 9.593548387096774e-06, "loss": 0.3453, "step": 12526 }, { "epoch": 80.75180722891567, "grad_norm": 11.478520393371582, "learning_rate": 9.590322580645162e-06, "loss": 0.3665, "step": 12527 }, { "epoch": 80.75823293172691, "grad_norm": 6.559749603271484, "learning_rate": 9.58709677419355e-06, "loss": 0.2532, "step": 12528 }, { "epoch": 80.76465863453815, "grad_norm": 6.96519136428833, "learning_rate": 9.583870967741935e-06, "loss": 0.2524, "step": 12529 }, { "epoch": 80.7710843373494, "grad_norm": 10.441873550415039, "learning_rate": 9.580645161290324e-06, "loss": 0.4292, "step": 12530 }, { "epoch": 80.77751004016065, "grad_norm": 3.8156888484954834, "learning_rate": 9.57741935483871e-06, "loss": 0.1574, "step": 12531 }, { "epoch": 80.78393574297189, "grad_norm": 9.740674018859863, "learning_rate": 9.574193548387097e-06, "loss": 0.3978, "step": 12532 }, { "epoch": 80.79036144578313, "grad_norm": 13.517998695373535, "learning_rate": 9.570967741935484e-06, "loss": 0.3663, "step": 12533 }, { "epoch": 80.79678714859438, "grad_norm": 9.621321678161621, "learning_rate": 9.567741935483872e-06, "loss": 0.2246, "step": 12534 }, { "epoch": 80.80321285140562, "grad_norm": 6.398690700531006, "learning_rate": 9.564516129032258e-06, "loss": 0.1966, "step": 12535 }, { "epoch": 80.80963855421686, "grad_norm": 7.881494522094727, "learning_rate": 9.561290322580647e-06, "loss": 0.2037, "step": 12536 }, { "epoch": 80.81606425702812, "grad_norm": 7.581315040588379, "learning_rate": 9.558064516129032e-06, "loss": 0.2492, "step": 12537 }, { "epoch": 80.82248995983936, "grad_norm": 8.862713813781738, "learning_rate": 9.55483870967742e-06, "loss": 0.3107, "step": 12538 }, { "epoch": 80.8289156626506, "grad_norm": 13.148063659667969, "learning_rate": 9.551612903225807e-06, "loss": 0.4687, "step": 12539 }, { "epoch": 80.83534136546184, "grad_norm": 10.889517784118652, "learning_rate": 9.548387096774195e-06, "loss": 0.2493, "step": 12540 }, { "epoch": 80.84176706827309, "grad_norm": 8.773075103759766, "learning_rate": 9.54516129032258e-06, "loss": 0.3919, "step": 12541 }, { "epoch": 80.84819277108434, "grad_norm": 11.43134593963623, "learning_rate": 9.54193548387097e-06, "loss": 0.3394, "step": 12542 }, { "epoch": 80.85461847389558, "grad_norm": 8.510411262512207, "learning_rate": 9.538709677419355e-06, "loss": 0.4714, "step": 12543 }, { "epoch": 80.86104417670683, "grad_norm": 9.317519187927246, "learning_rate": 9.535483870967743e-06, "loss": 0.6961, "step": 12544 }, { "epoch": 80.86746987951807, "grad_norm": 15.861013412475586, "learning_rate": 9.53225806451613e-06, "loss": 0.3973, "step": 12545 }, { "epoch": 80.87389558232931, "grad_norm": 9.617669105529785, "learning_rate": 9.529032258064517e-06, "loss": 0.3949, "step": 12546 }, { "epoch": 80.88032128514057, "grad_norm": 61.64981460571289, "learning_rate": 9.525806451612903e-06, "loss": 0.4491, "step": 12547 }, { "epoch": 80.88674698795181, "grad_norm": 4.247437953948975, "learning_rate": 9.522580645161292e-06, "loss": 0.1927, "step": 12548 }, { "epoch": 80.89317269076305, "grad_norm": 10.93249225616455, "learning_rate": 9.519354838709678e-06, "loss": 0.3384, "step": 12549 }, { "epoch": 80.8995983935743, "grad_norm": 8.3690767288208, "learning_rate": 9.516129032258064e-06, "loss": 0.2818, "step": 12550 }, { "epoch": 80.90602409638554, "grad_norm": 11.574996948242188, "learning_rate": 9.512903225806453e-06, "loss": 0.3376, "step": 12551 }, { "epoch": 80.91244979919679, "grad_norm": 10.277587890625, "learning_rate": 9.509677419354838e-06, "loss": 0.2699, "step": 12552 }, { "epoch": 80.91887550200804, "grad_norm": 2.552764892578125, "learning_rate": 9.506451612903226e-06, "loss": 0.1167, "step": 12553 }, { "epoch": 80.92530120481928, "grad_norm": 8.565356254577637, "learning_rate": 9.503225806451613e-06, "loss": 0.2252, "step": 12554 }, { "epoch": 80.93172690763052, "grad_norm": 16.587947845458984, "learning_rate": 9.5e-06, "loss": 0.481, "step": 12555 }, { "epoch": 80.93815261044176, "grad_norm": 10.874274253845215, "learning_rate": 9.496774193548386e-06, "loss": 0.3783, "step": 12556 }, { "epoch": 80.94457831325302, "grad_norm": 5.175418376922607, "learning_rate": 9.493548387096775e-06, "loss": 0.172, "step": 12557 }, { "epoch": 80.95100401606426, "grad_norm": 9.94865894317627, "learning_rate": 9.490322580645161e-06, "loss": 0.2612, "step": 12558 }, { "epoch": 80.9574297188755, "grad_norm": 7.057974815368652, "learning_rate": 9.487096774193548e-06, "loss": 0.2168, "step": 12559 }, { "epoch": 80.96385542168674, "grad_norm": 7.610111236572266, "learning_rate": 9.483870967741936e-06, "loss": 0.3548, "step": 12560 }, { "epoch": 80.97028112449799, "grad_norm": 12.026534080505371, "learning_rate": 9.480645161290323e-06, "loss": 0.3356, "step": 12561 }, { "epoch": 80.97670682730924, "grad_norm": 9.77940845489502, "learning_rate": 9.477419354838709e-06, "loss": 0.3263, "step": 12562 }, { "epoch": 80.98313253012049, "grad_norm": 13.10011100769043, "learning_rate": 9.474193548387098e-06, "loss": 0.4145, "step": 12563 }, { "epoch": 80.98955823293173, "grad_norm": 24.479475021362305, "learning_rate": 9.470967741935484e-06, "loss": 0.5178, "step": 12564 }, { "epoch": 80.99598393574297, "grad_norm": 7.193163871765137, "learning_rate": 9.467741935483871e-06, "loss": 0.2446, "step": 12565 }, { "epoch": 81.00562248995983, "grad_norm": 6.245288848876953, "learning_rate": 9.464516129032259e-06, "loss": 0.316, "step": 12566 }, { "epoch": 81.01204819277109, "grad_norm": 5.353343963623047, "learning_rate": 9.461290322580646e-06, "loss": 0.2277, "step": 12567 }, { "epoch": 81.01847389558233, "grad_norm": 4.116716384887695, "learning_rate": 9.458064516129032e-06, "loss": 0.1467, "step": 12568 }, { "epoch": 81.02489959839357, "grad_norm": 21.235153198242188, "learning_rate": 9.45483870967742e-06, "loss": 0.3261, "step": 12569 }, { "epoch": 81.03132530120482, "grad_norm": 9.457457542419434, "learning_rate": 9.451612903225806e-06, "loss": 0.2981, "step": 12570 }, { "epoch": 81.03775100401606, "grad_norm": 16.590728759765625, "learning_rate": 9.448387096774194e-06, "loss": 0.3237, "step": 12571 }, { "epoch": 81.04417670682732, "grad_norm": 19.525291442871094, "learning_rate": 9.445161290322581e-06, "loss": 0.2892, "step": 12572 }, { "epoch": 81.05060240963856, "grad_norm": 17.02222442626953, "learning_rate": 9.441935483870969e-06, "loss": 0.2706, "step": 12573 }, { "epoch": 81.0570281124498, "grad_norm": 11.087482452392578, "learning_rate": 9.438709677419354e-06, "loss": 0.2696, "step": 12574 }, { "epoch": 81.06345381526104, "grad_norm": 8.350217819213867, "learning_rate": 9.435483870967743e-06, "loss": 0.4827, "step": 12575 }, { "epoch": 81.06987951807228, "grad_norm": 14.044392585754395, "learning_rate": 9.43225806451613e-06, "loss": 0.2213, "step": 12576 }, { "epoch": 81.07630522088354, "grad_norm": 14.558335304260254, "learning_rate": 9.429032258064517e-06, "loss": 0.5788, "step": 12577 }, { "epoch": 81.08273092369478, "grad_norm": 20.247751235961914, "learning_rate": 9.425806451612904e-06, "loss": 0.2867, "step": 12578 }, { "epoch": 81.08915662650602, "grad_norm": 12.873541831970215, "learning_rate": 9.422580645161291e-06, "loss": 0.2974, "step": 12579 }, { "epoch": 81.09558232931727, "grad_norm": 4.659008502960205, "learning_rate": 9.419354838709677e-06, "loss": 0.1298, "step": 12580 }, { "epoch": 81.10200803212851, "grad_norm": 16.43574333190918, "learning_rate": 9.416129032258066e-06, "loss": 0.6435, "step": 12581 }, { "epoch": 81.10843373493977, "grad_norm": 14.953484535217285, "learning_rate": 9.412903225806452e-06, "loss": 0.2858, "step": 12582 }, { "epoch": 81.11485943775101, "grad_norm": 16.896080017089844, "learning_rate": 9.40967741935484e-06, "loss": 0.5713, "step": 12583 }, { "epoch": 81.12128514056225, "grad_norm": 15.119153022766113, "learning_rate": 9.406451612903227e-06, "loss": 0.2433, "step": 12584 }, { "epoch": 81.12771084337349, "grad_norm": 29.69403076171875, "learning_rate": 9.403225806451612e-06, "loss": 0.2412, "step": 12585 }, { "epoch": 81.13413654618473, "grad_norm": 30.00211524963379, "learning_rate": 9.4e-06, "loss": 0.3645, "step": 12586 }, { "epoch": 81.14056224899599, "grad_norm": 5.0769476890563965, "learning_rate": 9.396774193548387e-06, "loss": 0.6132, "step": 12587 }, { "epoch": 81.14698795180723, "grad_norm": 10.023415565490723, "learning_rate": 9.393548387096775e-06, "loss": 0.332, "step": 12588 }, { "epoch": 81.15341365461848, "grad_norm": 1.8538016080856323, "learning_rate": 9.39032258064516e-06, "loss": 0.2379, "step": 12589 }, { "epoch": 81.15983935742972, "grad_norm": 23.740467071533203, "learning_rate": 9.38709677419355e-06, "loss": 0.5032, "step": 12590 }, { "epoch": 81.16626506024096, "grad_norm": 30.523902893066406, "learning_rate": 9.383870967741935e-06, "loss": 0.6224, "step": 12591 }, { "epoch": 81.1726907630522, "grad_norm": 9.388754844665527, "learning_rate": 9.380645161290322e-06, "loss": 0.3992, "step": 12592 }, { "epoch": 81.17911646586346, "grad_norm": 1.582260012626648, "learning_rate": 9.37741935483871e-06, "loss": 0.2997, "step": 12593 }, { "epoch": 81.1855421686747, "grad_norm": 17.20458984375, "learning_rate": 9.374193548387097e-06, "loss": 0.2541, "step": 12594 }, { "epoch": 81.19196787148594, "grad_norm": 16.928556442260742, "learning_rate": 9.370967741935483e-06, "loss": 0.1706, "step": 12595 }, { "epoch": 81.19839357429719, "grad_norm": 19.850196838378906, "learning_rate": 9.367741935483872e-06, "loss": 0.4522, "step": 12596 }, { "epoch": 81.20481927710843, "grad_norm": 23.00525665283203, "learning_rate": 9.364516129032258e-06, "loss": 0.3359, "step": 12597 }, { "epoch": 81.21124497991968, "grad_norm": 1.3249250650405884, "learning_rate": 9.361290322580645e-06, "loss": 0.2371, "step": 12598 }, { "epoch": 81.21767068273093, "grad_norm": 16.15267562866211, "learning_rate": 9.358064516129033e-06, "loss": 0.3041, "step": 12599 }, { "epoch": 81.22409638554217, "grad_norm": 18.355831146240234, "learning_rate": 9.35483870967742e-06, "loss": 0.4662, "step": 12600 }, { "epoch": 81.23052208835341, "grad_norm": 10.319449424743652, "learning_rate": 9.351612903225806e-06, "loss": 0.2975, "step": 12601 }, { "epoch": 81.23694779116465, "grad_norm": 7.187723636627197, "learning_rate": 9.348387096774195e-06, "loss": 0.2446, "step": 12602 }, { "epoch": 81.24337349397591, "grad_norm": 16.08838653564453, "learning_rate": 9.34516129032258e-06, "loss": 0.68, "step": 12603 }, { "epoch": 81.24979919678715, "grad_norm": 5.286441802978516, "learning_rate": 9.341935483870968e-06, "loss": 0.2092, "step": 12604 }, { "epoch": 81.2562248995984, "grad_norm": 8.615823745727539, "learning_rate": 9.338709677419355e-06, "loss": 0.1983, "step": 12605 }, { "epoch": 81.26265060240964, "grad_norm": 6.848452091217041, "learning_rate": 9.335483870967743e-06, "loss": 0.3331, "step": 12606 }, { "epoch": 81.26907630522088, "grad_norm": 20.173290252685547, "learning_rate": 9.332258064516128e-06, "loss": 0.6545, "step": 12607 }, { "epoch": 81.27550200803213, "grad_norm": 25.228246688842773, "learning_rate": 9.329032258064518e-06, "loss": 0.2608, "step": 12608 }, { "epoch": 81.28192771084338, "grad_norm": 8.46483039855957, "learning_rate": 9.325806451612903e-06, "loss": 0.2349, "step": 12609 }, { "epoch": 81.28835341365462, "grad_norm": 5.357273578643799, "learning_rate": 9.32258064516129e-06, "loss": 0.5878, "step": 12610 }, { "epoch": 81.29477911646586, "grad_norm": 4.3581976890563965, "learning_rate": 9.319354838709678e-06, "loss": 0.3695, "step": 12611 }, { "epoch": 81.3012048192771, "grad_norm": 17.461034774780273, "learning_rate": 9.316129032258065e-06, "loss": 0.3404, "step": 12612 }, { "epoch": 81.30763052208836, "grad_norm": 5.008364677429199, "learning_rate": 9.312903225806451e-06, "loss": 0.1846, "step": 12613 }, { "epoch": 81.3140562248996, "grad_norm": 40.223838806152344, "learning_rate": 9.30967741935484e-06, "loss": 0.3695, "step": 12614 }, { "epoch": 81.32048192771084, "grad_norm": 9.706708908081055, "learning_rate": 9.306451612903226e-06, "loss": 0.3048, "step": 12615 }, { "epoch": 81.32690763052209, "grad_norm": 10.79458236694336, "learning_rate": 9.303225806451613e-06, "loss": 0.2811, "step": 12616 }, { "epoch": 81.33333333333333, "grad_norm": 9.677567481994629, "learning_rate": 9.3e-06, "loss": 0.5881, "step": 12617 }, { "epoch": 81.33975903614459, "grad_norm": 4.626848220825195, "learning_rate": 9.296774193548388e-06, "loss": 0.2188, "step": 12618 }, { "epoch": 81.34618473895583, "grad_norm": 22.76961326599121, "learning_rate": 9.293548387096774e-06, "loss": 0.8537, "step": 12619 }, { "epoch": 81.35261044176707, "grad_norm": 23.85955238342285, "learning_rate": 9.290322580645163e-06, "loss": 0.2372, "step": 12620 }, { "epoch": 81.35903614457831, "grad_norm": 5.253103733062744, "learning_rate": 9.287096774193549e-06, "loss": 0.2493, "step": 12621 }, { "epoch": 81.36546184738955, "grad_norm": 43.91737365722656, "learning_rate": 9.283870967741936e-06, "loss": 0.4623, "step": 12622 }, { "epoch": 81.3718875502008, "grad_norm": 12.894742965698242, "learning_rate": 9.280645161290323e-06, "loss": 0.2372, "step": 12623 }, { "epoch": 81.37831325301205, "grad_norm": 8.401308059692383, "learning_rate": 9.27741935483871e-06, "loss": 0.2332, "step": 12624 }, { "epoch": 81.3847389558233, "grad_norm": 19.084487915039062, "learning_rate": 9.274193548387097e-06, "loss": 0.346, "step": 12625 }, { "epoch": 81.39116465863454, "grad_norm": 8.979364395141602, "learning_rate": 9.270967741935484e-06, "loss": 0.4355, "step": 12626 }, { "epoch": 81.39759036144578, "grad_norm": 40.76905059814453, "learning_rate": 9.267741935483871e-06, "loss": 0.328, "step": 12627 }, { "epoch": 81.40401606425702, "grad_norm": 12.734367370605469, "learning_rate": 9.264516129032259e-06, "loss": 0.3993, "step": 12628 }, { "epoch": 81.41044176706828, "grad_norm": 26.20062255859375, "learning_rate": 9.261290322580646e-06, "loss": 0.2557, "step": 12629 }, { "epoch": 81.41686746987952, "grad_norm": 14.371213912963867, "learning_rate": 9.258064516129032e-06, "loss": 0.3065, "step": 12630 }, { "epoch": 81.42329317269076, "grad_norm": 47.68510055541992, "learning_rate": 9.25483870967742e-06, "loss": 0.5874, "step": 12631 }, { "epoch": 81.429718875502, "grad_norm": 21.566518783569336, "learning_rate": 9.251612903225807e-06, "loss": 0.39, "step": 12632 }, { "epoch": 81.43614457831325, "grad_norm": 30.99933624267578, "learning_rate": 9.248387096774194e-06, "loss": 0.3168, "step": 12633 }, { "epoch": 81.4425702811245, "grad_norm": 8.806756973266602, "learning_rate": 9.245161290322581e-06, "loss": 0.2629, "step": 12634 }, { "epoch": 81.44899598393575, "grad_norm": 10.453888893127441, "learning_rate": 9.241935483870969e-06, "loss": 0.3018, "step": 12635 }, { "epoch": 81.45542168674699, "grad_norm": 14.790909767150879, "learning_rate": 9.238709677419355e-06, "loss": 0.3327, "step": 12636 }, { "epoch": 81.46184738955823, "grad_norm": 6.687849521636963, "learning_rate": 9.235483870967742e-06, "loss": 0.477, "step": 12637 }, { "epoch": 81.46827309236947, "grad_norm": 20.45868682861328, "learning_rate": 9.23225806451613e-06, "loss": 0.2809, "step": 12638 }, { "epoch": 81.47469879518073, "grad_norm": 11.378606796264648, "learning_rate": 9.229032258064517e-06, "loss": 0.1795, "step": 12639 }, { "epoch": 81.48112449799197, "grad_norm": 18.382747650146484, "learning_rate": 9.225806451612904e-06, "loss": 0.3335, "step": 12640 }, { "epoch": 81.48755020080321, "grad_norm": 15.470297813415527, "learning_rate": 9.222580645161292e-06, "loss": 0.3488, "step": 12641 }, { "epoch": 81.49397590361446, "grad_norm": 29.464086532592773, "learning_rate": 9.219354838709677e-06, "loss": 0.328, "step": 12642 }, { "epoch": 81.5004016064257, "grad_norm": 14.698683738708496, "learning_rate": 9.216129032258065e-06, "loss": 0.2807, "step": 12643 }, { "epoch": 81.50682730923695, "grad_norm": 23.718780517578125, "learning_rate": 9.212903225806452e-06, "loss": 0.439, "step": 12644 }, { "epoch": 81.5132530120482, "grad_norm": 17.945730209350586, "learning_rate": 9.20967741935484e-06, "loss": 0.5132, "step": 12645 }, { "epoch": 81.51967871485944, "grad_norm": 3.8525376319885254, "learning_rate": 9.206451612903227e-06, "loss": 0.2834, "step": 12646 }, { "epoch": 81.52610441767068, "grad_norm": 47.928646087646484, "learning_rate": 9.203225806451614e-06, "loss": 0.403, "step": 12647 }, { "epoch": 81.53253012048192, "grad_norm": 45.776123046875, "learning_rate": 9.2e-06, "loss": 0.6497, "step": 12648 }, { "epoch": 81.53895582329318, "grad_norm": 14.323925971984863, "learning_rate": 9.196774193548387e-06, "loss": 0.3958, "step": 12649 }, { "epoch": 81.54538152610442, "grad_norm": 25.990781784057617, "learning_rate": 9.193548387096775e-06, "loss": 0.0822, "step": 12650 }, { "epoch": 81.55180722891566, "grad_norm": 10.829333305358887, "learning_rate": 9.190322580645162e-06, "loss": 0.3232, "step": 12651 }, { "epoch": 81.5582329317269, "grad_norm": 22.348575592041016, "learning_rate": 9.187096774193548e-06, "loss": 0.1959, "step": 12652 }, { "epoch": 81.56465863453815, "grad_norm": 8.704065322875977, "learning_rate": 9.183870967741937e-06, "loss": 0.3046, "step": 12653 }, { "epoch": 81.5710843373494, "grad_norm": 11.668932914733887, "learning_rate": 9.180645161290323e-06, "loss": 0.418, "step": 12654 }, { "epoch": 81.57751004016065, "grad_norm": 2.5315749645233154, "learning_rate": 9.17741935483871e-06, "loss": 0.259, "step": 12655 }, { "epoch": 81.58393574297189, "grad_norm": 5.179762840270996, "learning_rate": 9.174193548387098e-06, "loss": 0.2538, "step": 12656 }, { "epoch": 81.59036144578313, "grad_norm": 9.399178504943848, "learning_rate": 9.170967741935483e-06, "loss": 0.6671, "step": 12657 }, { "epoch": 81.59678714859437, "grad_norm": 16.269899368286133, "learning_rate": 9.16774193548387e-06, "loss": 0.43, "step": 12658 }, { "epoch": 81.60321285140563, "grad_norm": 47.67700958251953, "learning_rate": 9.164516129032258e-06, "loss": 0.7526, "step": 12659 }, { "epoch": 81.60963855421687, "grad_norm": 3.001101493835449, "learning_rate": 9.161290322580645e-06, "loss": 0.1047, "step": 12660 }, { "epoch": 81.61606425702811, "grad_norm": 9.961862564086914, "learning_rate": 9.158064516129033e-06, "loss": 0.4324, "step": 12661 }, { "epoch": 81.62248995983936, "grad_norm": 11.588841438293457, "learning_rate": 9.15483870967742e-06, "loss": 0.2336, "step": 12662 }, { "epoch": 81.6289156626506, "grad_norm": 5.895252227783203, "learning_rate": 9.151612903225806e-06, "loss": 0.2609, "step": 12663 }, { "epoch": 81.63534136546184, "grad_norm": 19.91753387451172, "learning_rate": 9.148387096774193e-06, "loss": 0.4276, "step": 12664 }, { "epoch": 81.6417670682731, "grad_norm": 3.513989210128784, "learning_rate": 9.14516129032258e-06, "loss": 0.2833, "step": 12665 }, { "epoch": 81.64819277108434, "grad_norm": 31.13001823425293, "learning_rate": 9.141935483870968e-06, "loss": 0.1812, "step": 12666 }, { "epoch": 81.65461847389558, "grad_norm": 7.279386043548584, "learning_rate": 9.138709677419356e-06, "loss": 0.4659, "step": 12667 }, { "epoch": 81.66104417670682, "grad_norm": 4.713456630706787, "learning_rate": 9.135483870967743e-06, "loss": 0.2672, "step": 12668 }, { "epoch": 81.66746987951807, "grad_norm": 7.2171244621276855, "learning_rate": 9.132258064516129e-06, "loss": 0.2475, "step": 12669 }, { "epoch": 81.67389558232932, "grad_norm": 6.376832008361816, "learning_rate": 9.129032258064516e-06, "loss": 0.2256, "step": 12670 }, { "epoch": 81.68032128514056, "grad_norm": 7.388886451721191, "learning_rate": 9.125806451612903e-06, "loss": 0.2037, "step": 12671 }, { "epoch": 81.6867469879518, "grad_norm": 4.627335071563721, "learning_rate": 9.122580645161291e-06, "loss": 0.2409, "step": 12672 }, { "epoch": 81.69317269076305, "grad_norm": 18.426881790161133, "learning_rate": 9.119354838709678e-06, "loss": 0.421, "step": 12673 }, { "epoch": 81.69959839357429, "grad_norm": 6.684193134307861, "learning_rate": 9.116129032258066e-06, "loss": 0.312, "step": 12674 }, { "epoch": 81.70602409638555, "grad_norm": 11.307368278503418, "learning_rate": 9.112903225806451e-06, "loss": 0.3473, "step": 12675 }, { "epoch": 81.71244979919679, "grad_norm": 22.313526153564453, "learning_rate": 9.109677419354839e-06, "loss": 0.6725, "step": 12676 }, { "epoch": 81.71887550200803, "grad_norm": 1.4874019622802734, "learning_rate": 9.106451612903226e-06, "loss": 0.2943, "step": 12677 }, { "epoch": 81.72530120481927, "grad_norm": 16.112430572509766, "learning_rate": 9.103225806451614e-06, "loss": 0.5065, "step": 12678 }, { "epoch": 81.73172690763052, "grad_norm": 8.241067886352539, "learning_rate": 9.100000000000001e-06, "loss": 0.4074, "step": 12679 }, { "epoch": 81.73815261044177, "grad_norm": 19.547603607177734, "learning_rate": 9.096774193548388e-06, "loss": 0.4195, "step": 12680 }, { "epoch": 81.74457831325302, "grad_norm": 25.08710479736328, "learning_rate": 9.093548387096774e-06, "loss": 0.3536, "step": 12681 }, { "epoch": 81.75100401606426, "grad_norm": 14.618867874145508, "learning_rate": 9.090322580645161e-06, "loss": 0.5065, "step": 12682 }, { "epoch": 81.7574297188755, "grad_norm": 1.8051973581314087, "learning_rate": 9.087096774193549e-06, "loss": 0.0701, "step": 12683 }, { "epoch": 81.76385542168674, "grad_norm": 10.57521915435791, "learning_rate": 9.083870967741936e-06, "loss": 0.4207, "step": 12684 }, { "epoch": 81.770281124498, "grad_norm": 2.1673829555511475, "learning_rate": 9.080645161290324e-06, "loss": 0.3269, "step": 12685 }, { "epoch": 81.77670682730924, "grad_norm": 11.337553024291992, "learning_rate": 9.077419354838711e-06, "loss": 0.3067, "step": 12686 }, { "epoch": 81.78313253012048, "grad_norm": 10.905994415283203, "learning_rate": 9.074193548387097e-06, "loss": 0.4114, "step": 12687 }, { "epoch": 81.78955823293173, "grad_norm": 25.94481658935547, "learning_rate": 9.070967741935484e-06, "loss": 0.5043, "step": 12688 }, { "epoch": 81.79598393574297, "grad_norm": 4.022121429443359, "learning_rate": 9.067741935483872e-06, "loss": 0.2327, "step": 12689 }, { "epoch": 81.80240963855422, "grad_norm": 9.776264190673828, "learning_rate": 9.064516129032259e-06, "loss": 0.308, "step": 12690 }, { "epoch": 81.80883534136547, "grad_norm": 49.088661193847656, "learning_rate": 9.061290322580646e-06, "loss": 0.5485, "step": 12691 }, { "epoch": 81.81526104417671, "grad_norm": 33.21071243286133, "learning_rate": 9.058064516129032e-06, "loss": 0.6968, "step": 12692 }, { "epoch": 81.82168674698795, "grad_norm": 23.745162963867188, "learning_rate": 9.05483870967742e-06, "loss": 0.4234, "step": 12693 }, { "epoch": 81.82811244979919, "grad_norm": 15.611031532287598, "learning_rate": 9.051612903225807e-06, "loss": 0.1919, "step": 12694 }, { "epoch": 81.83453815261043, "grad_norm": 19.802886962890625, "learning_rate": 9.048387096774194e-06, "loss": 0.4443, "step": 12695 }, { "epoch": 81.84096385542169, "grad_norm": 15.02939510345459, "learning_rate": 9.04516129032258e-06, "loss": 0.2349, "step": 12696 }, { "epoch": 81.84738955823293, "grad_norm": 21.07808494567871, "learning_rate": 9.041935483870969e-06, "loss": 0.4466, "step": 12697 }, { "epoch": 81.85381526104418, "grad_norm": 4.524120330810547, "learning_rate": 9.038709677419355e-06, "loss": 0.3068, "step": 12698 }, { "epoch": 81.86024096385542, "grad_norm": 9.44431209564209, "learning_rate": 9.035483870967742e-06, "loss": 0.2025, "step": 12699 }, { "epoch": 81.86666666666666, "grad_norm": 16.310260772705078, "learning_rate": 9.03225806451613e-06, "loss": 0.3462, "step": 12700 }, { "epoch": 81.87309236947792, "grad_norm": 27.29207992553711, "learning_rate": 9.029032258064517e-06, "loss": 0.2054, "step": 12701 }, { "epoch": 81.87951807228916, "grad_norm": 15.995261192321777, "learning_rate": 9.025806451612903e-06, "loss": 0.2099, "step": 12702 }, { "epoch": 81.8859437751004, "grad_norm": 20.431812286376953, "learning_rate": 9.022580645161292e-06, "loss": 0.4216, "step": 12703 }, { "epoch": 81.89236947791164, "grad_norm": 6.596088409423828, "learning_rate": 9.019354838709678e-06, "loss": 0.4835, "step": 12704 }, { "epoch": 81.89879518072289, "grad_norm": 14.270121574401855, "learning_rate": 9.016129032258065e-06, "loss": 0.4176, "step": 12705 }, { "epoch": 81.90522088353414, "grad_norm": 15.130910873413086, "learning_rate": 9.012903225806452e-06, "loss": 0.3391, "step": 12706 }, { "epoch": 81.91164658634538, "grad_norm": 7.0091471672058105, "learning_rate": 9.00967741935484e-06, "loss": 0.3388, "step": 12707 }, { "epoch": 81.91807228915663, "grad_norm": 15.669628143310547, "learning_rate": 9.006451612903225e-06, "loss": 0.4588, "step": 12708 }, { "epoch": 81.92449799196787, "grad_norm": 5.733871936798096, "learning_rate": 9.003225806451615e-06, "loss": 0.1368, "step": 12709 }, { "epoch": 81.93092369477911, "grad_norm": 13.75333023071289, "learning_rate": 9e-06, "loss": 0.3335, "step": 12710 }, { "epoch": 81.93734939759037, "grad_norm": 6.767966270446777, "learning_rate": 8.996774193548388e-06, "loss": 0.1443, "step": 12711 }, { "epoch": 81.94377510040161, "grad_norm": 13.57185173034668, "learning_rate": 8.993548387096775e-06, "loss": 0.1646, "step": 12712 }, { "epoch": 81.95020080321285, "grad_norm": 4.778683662414551, "learning_rate": 8.990322580645162e-06, "loss": 0.0939, "step": 12713 }, { "epoch": 81.9566265060241, "grad_norm": 19.055740356445312, "learning_rate": 8.987096774193548e-06, "loss": 0.4468, "step": 12714 }, { "epoch": 81.96305220883534, "grad_norm": 6.977706432342529, "learning_rate": 8.983870967741936e-06, "loss": 0.3638, "step": 12715 }, { "epoch": 81.96947791164659, "grad_norm": 46.64641571044922, "learning_rate": 8.980645161290323e-06, "loss": 0.41, "step": 12716 }, { "epoch": 81.97590361445783, "grad_norm": 7.401476860046387, "learning_rate": 8.97741935483871e-06, "loss": 0.4576, "step": 12717 }, { "epoch": 81.98232931726908, "grad_norm": 19.504119873046875, "learning_rate": 8.974193548387098e-06, "loss": 0.5245, "step": 12718 }, { "epoch": 81.98875502008032, "grad_norm": 5.376507759094238, "learning_rate": 8.970967741935485e-06, "loss": 0.1144, "step": 12719 }, { "epoch": 81.99518072289156, "grad_norm": 32.07115173339844, "learning_rate": 8.967741935483871e-06, "loss": 0.2496, "step": 12720 }, { "epoch": 82.00481927710844, "grad_norm": 3.0058224201202393, "learning_rate": 8.964516129032258e-06, "loss": 0.2785, "step": 12721 }, { "epoch": 82.01124497991968, "grad_norm": 21.75843048095703, "learning_rate": 8.961290322580646e-06, "loss": 0.4467, "step": 12722 }, { "epoch": 82.01767068273092, "grad_norm": 20.225000381469727, "learning_rate": 8.958064516129033e-06, "loss": 0.3014, "step": 12723 }, { "epoch": 82.02409638554217, "grad_norm": 3.5653789043426514, "learning_rate": 8.95483870967742e-06, "loss": 0.1002, "step": 12724 }, { "epoch": 82.03052208835341, "grad_norm": 25.922651290893555, "learning_rate": 8.951612903225808e-06, "loss": 0.3054, "step": 12725 }, { "epoch": 82.03694779116466, "grad_norm": 25.341886520385742, "learning_rate": 8.948387096774194e-06, "loss": 0.3316, "step": 12726 }, { "epoch": 82.0433734939759, "grad_norm": 31.645822525024414, "learning_rate": 8.945161290322581e-06, "loss": 0.3679, "step": 12727 }, { "epoch": 82.04979919678715, "grad_norm": 33.44609451293945, "learning_rate": 8.941935483870968e-06, "loss": 0.3747, "step": 12728 }, { "epoch": 82.05622489959839, "grad_norm": 17.88628387451172, "learning_rate": 8.938709677419354e-06, "loss": 0.4443, "step": 12729 }, { "epoch": 82.06265060240963, "grad_norm": 18.473651885986328, "learning_rate": 8.935483870967743e-06, "loss": 0.237, "step": 12730 }, { "epoch": 82.06907630522089, "grad_norm": 8.238988876342773, "learning_rate": 8.932258064516129e-06, "loss": 0.2025, "step": 12731 }, { "epoch": 82.07550200803213, "grad_norm": 7.565674781799316, "learning_rate": 8.929032258064516e-06, "loss": 0.14, "step": 12732 }, { "epoch": 82.08192771084337, "grad_norm": 11.21212387084961, "learning_rate": 8.925806451612904e-06, "loss": 0.189, "step": 12733 }, { "epoch": 82.08835341365462, "grad_norm": 7.757596015930176, "learning_rate": 8.922580645161291e-06, "loss": 0.4173, "step": 12734 }, { "epoch": 82.09477911646586, "grad_norm": 5.673405647277832, "learning_rate": 8.919354838709677e-06, "loss": 0.1794, "step": 12735 }, { "epoch": 82.10120481927711, "grad_norm": 6.0775299072265625, "learning_rate": 8.916129032258066e-06, "loss": 0.1512, "step": 12736 }, { "epoch": 82.10763052208836, "grad_norm": 6.582547187805176, "learning_rate": 8.912903225806452e-06, "loss": 0.3621, "step": 12737 }, { "epoch": 82.1140562248996, "grad_norm": 4.777718544006348, "learning_rate": 8.909677419354839e-06, "loss": 0.2305, "step": 12738 }, { "epoch": 82.12048192771084, "grad_norm": 36.973121643066406, "learning_rate": 8.906451612903226e-06, "loss": 0.3702, "step": 12739 }, { "epoch": 82.12690763052208, "grad_norm": 10.405068397521973, "learning_rate": 8.903225806451614e-06, "loss": 0.3537, "step": 12740 }, { "epoch": 82.13333333333334, "grad_norm": 4.857229232788086, "learning_rate": 8.9e-06, "loss": 0.435, "step": 12741 }, { "epoch": 82.13975903614458, "grad_norm": 6.5029706954956055, "learning_rate": 8.896774193548389e-06, "loss": 0.3162, "step": 12742 }, { "epoch": 82.14618473895582, "grad_norm": 10.055248260498047, "learning_rate": 8.893548387096774e-06, "loss": 0.297, "step": 12743 }, { "epoch": 82.15261044176707, "grad_norm": 22.37954330444336, "learning_rate": 8.890322580645162e-06, "loss": 0.5863, "step": 12744 }, { "epoch": 82.15903614457831, "grad_norm": 17.630718231201172, "learning_rate": 8.887096774193549e-06, "loss": 0.3199, "step": 12745 }, { "epoch": 82.16546184738957, "grad_norm": 24.661785125732422, "learning_rate": 8.883870967741936e-06, "loss": 0.1664, "step": 12746 }, { "epoch": 82.17188755020081, "grad_norm": 15.462448120117188, "learning_rate": 8.880645161290322e-06, "loss": 0.3718, "step": 12747 }, { "epoch": 82.17831325301205, "grad_norm": 19.298433303833008, "learning_rate": 8.877419354838711e-06, "loss": 0.2882, "step": 12748 }, { "epoch": 82.18473895582329, "grad_norm": 4.63648796081543, "learning_rate": 8.874193548387097e-06, "loss": 0.3129, "step": 12749 }, { "epoch": 82.19116465863453, "grad_norm": 60.76720428466797, "learning_rate": 8.870967741935484e-06, "loss": 0.3344, "step": 12750 }, { "epoch": 82.19759036144578, "grad_norm": 2.6844305992126465, "learning_rate": 8.867741935483872e-06, "loss": 0.3288, "step": 12751 }, { "epoch": 82.20401606425703, "grad_norm": 6.592637538909912, "learning_rate": 8.86451612903226e-06, "loss": 0.4376, "step": 12752 }, { "epoch": 82.21044176706827, "grad_norm": 19.428390502929688, "learning_rate": 8.861290322580645e-06, "loss": 0.4501, "step": 12753 }, { "epoch": 82.21686746987952, "grad_norm": 6.9482102394104, "learning_rate": 8.858064516129034e-06, "loss": 0.5822, "step": 12754 }, { "epoch": 82.22329317269076, "grad_norm": 12.113594055175781, "learning_rate": 8.85483870967742e-06, "loss": 0.485, "step": 12755 }, { "epoch": 82.229718875502, "grad_norm": 35.204994201660156, "learning_rate": 8.851612903225807e-06, "loss": 0.1846, "step": 12756 }, { "epoch": 82.23614457831326, "grad_norm": 64.13822174072266, "learning_rate": 8.848387096774195e-06, "loss": 0.4676, "step": 12757 }, { "epoch": 82.2425702811245, "grad_norm": 25.687471389770508, "learning_rate": 8.845161290322582e-06, "loss": 0.3043, "step": 12758 }, { "epoch": 82.24899598393574, "grad_norm": 29.24393653869629, "learning_rate": 8.841935483870968e-06, "loss": 0.339, "step": 12759 }, { "epoch": 82.25542168674698, "grad_norm": 11.886094093322754, "learning_rate": 8.838709677419357e-06, "loss": 0.3495, "step": 12760 }, { "epoch": 82.26184738955823, "grad_norm": 3.5608816146850586, "learning_rate": 8.835483870967742e-06, "loss": 0.1829, "step": 12761 }, { "epoch": 82.26827309236948, "grad_norm": 3.9207558631896973, "learning_rate": 8.83225806451613e-06, "loss": 0.2501, "step": 12762 }, { "epoch": 82.27469879518073, "grad_norm": 21.993759155273438, "learning_rate": 8.829032258064517e-06, "loss": 0.6129, "step": 12763 }, { "epoch": 82.28112449799197, "grad_norm": 3.7410061359405518, "learning_rate": 8.825806451612903e-06, "loss": 0.1855, "step": 12764 }, { "epoch": 82.28755020080321, "grad_norm": 14.947953224182129, "learning_rate": 8.82258064516129e-06, "loss": 0.2699, "step": 12765 }, { "epoch": 82.29397590361445, "grad_norm": 22.406055450439453, "learning_rate": 8.819354838709678e-06, "loss": 0.3665, "step": 12766 }, { "epoch": 82.30040160642571, "grad_norm": 13.103739738464355, "learning_rate": 8.816129032258065e-06, "loss": 0.5086, "step": 12767 }, { "epoch": 82.30682730923695, "grad_norm": 8.316972732543945, "learning_rate": 8.81290322580645e-06, "loss": 0.3099, "step": 12768 }, { "epoch": 82.3132530120482, "grad_norm": 27.43618392944336, "learning_rate": 8.80967741935484e-06, "loss": 0.44, "step": 12769 }, { "epoch": 82.31967871485944, "grad_norm": 10.621097564697266, "learning_rate": 8.806451612903226e-06, "loss": 0.2461, "step": 12770 }, { "epoch": 82.32610441767068, "grad_norm": 4.798941612243652, "learning_rate": 8.803225806451613e-06, "loss": 0.4887, "step": 12771 }, { "epoch": 82.33253012048193, "grad_norm": 10.58165454864502, "learning_rate": 8.8e-06, "loss": 0.2128, "step": 12772 }, { "epoch": 82.33895582329318, "grad_norm": 15.208868980407715, "learning_rate": 8.796774193548388e-06, "loss": 0.4214, "step": 12773 }, { "epoch": 82.34538152610442, "grad_norm": 17.727670669555664, "learning_rate": 8.793548387096774e-06, "loss": 0.235, "step": 12774 }, { "epoch": 82.35180722891566, "grad_norm": 7.989589691162109, "learning_rate": 8.790322580645163e-06, "loss": 0.188, "step": 12775 }, { "epoch": 82.3582329317269, "grad_norm": 31.532926559448242, "learning_rate": 8.787096774193548e-06, "loss": 0.3276, "step": 12776 }, { "epoch": 82.36465863453816, "grad_norm": 24.633073806762695, "learning_rate": 8.783870967741936e-06, "loss": 0.2329, "step": 12777 }, { "epoch": 82.3710843373494, "grad_norm": 11.901674270629883, "learning_rate": 8.780645161290323e-06, "loss": 0.2236, "step": 12778 }, { "epoch": 82.37751004016064, "grad_norm": 19.831666946411133, "learning_rate": 8.77741935483871e-06, "loss": 0.194, "step": 12779 }, { "epoch": 82.38393574297189, "grad_norm": 24.266199111938477, "learning_rate": 8.774193548387096e-06, "loss": 0.5555, "step": 12780 }, { "epoch": 82.39036144578313, "grad_norm": 8.892492294311523, "learning_rate": 8.770967741935485e-06, "loss": 0.4984, "step": 12781 }, { "epoch": 82.39678714859438, "grad_norm": 14.02311897277832, "learning_rate": 8.767741935483871e-06, "loss": 0.5951, "step": 12782 }, { "epoch": 82.40321285140563, "grad_norm": 5.579545974731445, "learning_rate": 8.764516129032258e-06, "loss": 0.4604, "step": 12783 }, { "epoch": 82.40963855421687, "grad_norm": 20.90677833557129, "learning_rate": 8.761290322580646e-06, "loss": 0.2744, "step": 12784 }, { "epoch": 82.41606425702811, "grad_norm": 23.819225311279297, "learning_rate": 8.758064516129033e-06, "loss": 0.3859, "step": 12785 }, { "epoch": 82.42248995983935, "grad_norm": 21.81110191345215, "learning_rate": 8.754838709677419e-06, "loss": 0.3813, "step": 12786 }, { "epoch": 82.4289156626506, "grad_norm": 21.538169860839844, "learning_rate": 8.751612903225808e-06, "loss": 0.4153, "step": 12787 }, { "epoch": 82.43534136546185, "grad_norm": 6.055915832519531, "learning_rate": 8.748387096774194e-06, "loss": 0.0767, "step": 12788 }, { "epoch": 82.4417670682731, "grad_norm": 15.381766319274902, "learning_rate": 8.745161290322581e-06, "loss": 0.5513, "step": 12789 }, { "epoch": 82.44819277108434, "grad_norm": 11.847177505493164, "learning_rate": 8.741935483870969e-06, "loss": 0.2871, "step": 12790 }, { "epoch": 82.45461847389558, "grad_norm": 8.682072639465332, "learning_rate": 8.738709677419356e-06, "loss": 0.1585, "step": 12791 }, { "epoch": 82.46104417670682, "grad_norm": 7.069615840911865, "learning_rate": 8.735483870967742e-06, "loss": 0.6152, "step": 12792 }, { "epoch": 82.46746987951808, "grad_norm": 8.93429946899414, "learning_rate": 8.73225806451613e-06, "loss": 0.5237, "step": 12793 }, { "epoch": 82.47389558232932, "grad_norm": 5.5154595375061035, "learning_rate": 8.729032258064516e-06, "loss": 0.3102, "step": 12794 }, { "epoch": 82.48032128514056, "grad_norm": 10.704414367675781, "learning_rate": 8.725806451612904e-06, "loss": 0.2825, "step": 12795 }, { "epoch": 82.4867469879518, "grad_norm": 12.758923530578613, "learning_rate": 8.722580645161291e-06, "loss": 0.3191, "step": 12796 }, { "epoch": 82.49317269076305, "grad_norm": 13.246789932250977, "learning_rate": 8.719354838709679e-06, "loss": 0.3787, "step": 12797 }, { "epoch": 82.4995983935743, "grad_norm": 30.774145126342773, "learning_rate": 8.716129032258064e-06, "loss": 0.353, "step": 12798 }, { "epoch": 82.50602409638554, "grad_norm": 10.646123886108398, "learning_rate": 8.712903225806452e-06, "loss": 0.2824, "step": 12799 }, { "epoch": 82.51244979919679, "grad_norm": 5.712855339050293, "learning_rate": 8.70967741935484e-06, "loss": 0.2619, "step": 12800 }, { "epoch": 82.51887550200803, "grad_norm": 75.17391967773438, "learning_rate": 8.706451612903225e-06, "loss": 0.2854, "step": 12801 }, { "epoch": 82.52530120481927, "grad_norm": 3.128732442855835, "learning_rate": 8.703225806451614e-06, "loss": 0.1797, "step": 12802 }, { "epoch": 82.53172690763053, "grad_norm": 10.956369400024414, "learning_rate": 8.7e-06, "loss": 0.3068, "step": 12803 }, { "epoch": 82.53815261044177, "grad_norm": 26.087955474853516, "learning_rate": 8.696774193548387e-06, "loss": 0.44, "step": 12804 }, { "epoch": 82.54457831325301, "grad_norm": 19.981386184692383, "learning_rate": 8.693548387096775e-06, "loss": 0.6481, "step": 12805 }, { "epoch": 82.55100401606425, "grad_norm": 15.232161521911621, "learning_rate": 8.690322580645162e-06, "loss": 0.1798, "step": 12806 }, { "epoch": 82.5574297188755, "grad_norm": 10.382444381713867, "learning_rate": 8.687096774193548e-06, "loss": 0.2702, "step": 12807 }, { "epoch": 82.56385542168675, "grad_norm": 10.00696849822998, "learning_rate": 8.683870967741937e-06, "loss": 0.2879, "step": 12808 }, { "epoch": 82.570281124498, "grad_norm": 27.074533462524414, "learning_rate": 8.680645161290322e-06, "loss": 0.4786, "step": 12809 }, { "epoch": 82.57670682730924, "grad_norm": 28.09942626953125, "learning_rate": 8.67741935483871e-06, "loss": 0.5038, "step": 12810 }, { "epoch": 82.58313253012048, "grad_norm": 6.452268600463867, "learning_rate": 8.674193548387097e-06, "loss": 0.3981, "step": 12811 }, { "epoch": 82.58955823293172, "grad_norm": 8.229849815368652, "learning_rate": 8.670967741935485e-06, "loss": 0.3276, "step": 12812 }, { "epoch": 82.59598393574298, "grad_norm": 4.183628082275391, "learning_rate": 8.66774193548387e-06, "loss": 0.4237, "step": 12813 }, { "epoch": 82.60240963855422, "grad_norm": 11.140013694763184, "learning_rate": 8.66451612903226e-06, "loss": 0.267, "step": 12814 }, { "epoch": 82.60883534136546, "grad_norm": 21.624954223632812, "learning_rate": 8.661290322580645e-06, "loss": 0.2582, "step": 12815 }, { "epoch": 82.6152610441767, "grad_norm": 24.79956817626953, "learning_rate": 8.658064516129033e-06, "loss": 0.3941, "step": 12816 }, { "epoch": 82.62168674698795, "grad_norm": 19.572187423706055, "learning_rate": 8.65483870967742e-06, "loss": 0.6089, "step": 12817 }, { "epoch": 82.6281124497992, "grad_norm": 22.435340881347656, "learning_rate": 8.651612903225807e-06, "loss": 0.4016, "step": 12818 }, { "epoch": 82.63453815261045, "grad_norm": 2.8561952114105225, "learning_rate": 8.648387096774193e-06, "loss": 0.3006, "step": 12819 }, { "epoch": 82.64096385542169, "grad_norm": 9.940181732177734, "learning_rate": 8.645161290322582e-06, "loss": 0.4544, "step": 12820 }, { "epoch": 82.64738955823293, "grad_norm": 52.68178176879883, "learning_rate": 8.641935483870968e-06, "loss": 0.3066, "step": 12821 }, { "epoch": 82.65381526104417, "grad_norm": 23.08588981628418, "learning_rate": 8.638709677419355e-06, "loss": 0.4406, "step": 12822 }, { "epoch": 82.66024096385541, "grad_norm": 20.672170639038086, "learning_rate": 8.635483870967743e-06, "loss": 0.3495, "step": 12823 }, { "epoch": 82.66666666666667, "grad_norm": 11.681784629821777, "learning_rate": 8.63225806451613e-06, "loss": 0.3927, "step": 12824 }, { "epoch": 82.67309236947791, "grad_norm": 11.138823509216309, "learning_rate": 8.629032258064516e-06, "loss": 0.2242, "step": 12825 }, { "epoch": 82.67951807228916, "grad_norm": 4.890313148498535, "learning_rate": 8.625806451612905e-06, "loss": 0.3135, "step": 12826 }, { "epoch": 82.6859437751004, "grad_norm": 6.590038299560547, "learning_rate": 8.62258064516129e-06, "loss": 0.2583, "step": 12827 }, { "epoch": 82.69236947791164, "grad_norm": 14.433467864990234, "learning_rate": 8.619354838709678e-06, "loss": 0.2411, "step": 12828 }, { "epoch": 82.6987951807229, "grad_norm": 73.41535949707031, "learning_rate": 8.616129032258065e-06, "loss": 0.3143, "step": 12829 }, { "epoch": 82.70522088353414, "grad_norm": 22.51445770263672, "learning_rate": 8.612903225806453e-06, "loss": 0.3988, "step": 12830 }, { "epoch": 82.71164658634538, "grad_norm": 5.343419075012207, "learning_rate": 8.609677419354838e-06, "loss": 0.2174, "step": 12831 }, { "epoch": 82.71807228915662, "grad_norm": 18.26007652282715, "learning_rate": 8.606451612903228e-06, "loss": 0.5522, "step": 12832 }, { "epoch": 82.72449799196787, "grad_norm": 21.66236114501953, "learning_rate": 8.603225806451613e-06, "loss": 0.2348, "step": 12833 }, { "epoch": 82.73092369477912, "grad_norm": 15.454244613647461, "learning_rate": 8.599999999999999e-06, "loss": 0.3564, "step": 12834 }, { "epoch": 82.73734939759036, "grad_norm": 15.268853187561035, "learning_rate": 8.596774193548388e-06, "loss": 0.3416, "step": 12835 }, { "epoch": 82.7437751004016, "grad_norm": 16.953575134277344, "learning_rate": 8.593548387096774e-06, "loss": 0.3804, "step": 12836 }, { "epoch": 82.75020080321285, "grad_norm": 8.519159317016602, "learning_rate": 8.590322580645161e-06, "loss": 0.5744, "step": 12837 }, { "epoch": 82.75662650602409, "grad_norm": 5.71144962310791, "learning_rate": 8.587096774193549e-06, "loss": 0.2177, "step": 12838 }, { "epoch": 82.76305220883535, "grad_norm": 15.328914642333984, "learning_rate": 8.583870967741936e-06, "loss": 0.4886, "step": 12839 }, { "epoch": 82.76947791164659, "grad_norm": 2.3537447452545166, "learning_rate": 8.580645161290322e-06, "loss": 0.0932, "step": 12840 }, { "epoch": 82.77590361445783, "grad_norm": 10.149913787841797, "learning_rate": 8.57741935483871e-06, "loss": 0.2787, "step": 12841 }, { "epoch": 82.78232931726907, "grad_norm": 7.688109397888184, "learning_rate": 8.574193548387096e-06, "loss": 0.2232, "step": 12842 }, { "epoch": 82.78875502008032, "grad_norm": 27.009984970092773, "learning_rate": 8.570967741935484e-06, "loss": 0.3197, "step": 12843 }, { "epoch": 82.79518072289157, "grad_norm": 13.066659927368164, "learning_rate": 8.567741935483871e-06, "loss": 0.1086, "step": 12844 }, { "epoch": 82.80160642570281, "grad_norm": 14.081171035766602, "learning_rate": 8.564516129032259e-06, "loss": 0.2821, "step": 12845 }, { "epoch": 82.80803212851406, "grad_norm": 5.786002159118652, "learning_rate": 8.561290322580644e-06, "loss": 0.2892, "step": 12846 }, { "epoch": 82.8144578313253, "grad_norm": 14.104249000549316, "learning_rate": 8.558064516129033e-06, "loss": 0.3104, "step": 12847 }, { "epoch": 82.82088353413654, "grad_norm": 11.139267921447754, "learning_rate": 8.55483870967742e-06, "loss": 0.2302, "step": 12848 }, { "epoch": 82.8273092369478, "grad_norm": 16.083383560180664, "learning_rate": 8.551612903225807e-06, "loss": 0.1017, "step": 12849 }, { "epoch": 82.83373493975904, "grad_norm": 3.2416274547576904, "learning_rate": 8.548387096774194e-06, "loss": 0.1129, "step": 12850 }, { "epoch": 82.84016064257028, "grad_norm": 18.63534927368164, "learning_rate": 8.545161290322581e-06, "loss": 0.4197, "step": 12851 }, { "epoch": 82.84658634538152, "grad_norm": 5.35771369934082, "learning_rate": 8.541935483870967e-06, "loss": 0.3365, "step": 12852 }, { "epoch": 82.85301204819277, "grad_norm": 8.055874824523926, "learning_rate": 8.538709677419356e-06, "loss": 0.2051, "step": 12853 }, { "epoch": 82.85943775100401, "grad_norm": 37.06861877441406, "learning_rate": 8.535483870967742e-06, "loss": 0.4056, "step": 12854 }, { "epoch": 82.86586345381527, "grad_norm": 21.274463653564453, "learning_rate": 8.53225806451613e-06, "loss": 0.5096, "step": 12855 }, { "epoch": 82.87228915662651, "grad_norm": 10.210789680480957, "learning_rate": 8.529032258064517e-06, "loss": 0.4261, "step": 12856 }, { "epoch": 82.87871485943775, "grad_norm": 18.750255584716797, "learning_rate": 8.525806451612904e-06, "loss": 0.6467, "step": 12857 }, { "epoch": 82.88514056224899, "grad_norm": 28.70277214050293, "learning_rate": 8.52258064516129e-06, "loss": 0.2767, "step": 12858 }, { "epoch": 82.89156626506023, "grad_norm": 6.516689300537109, "learning_rate": 8.519354838709679e-06, "loss": 0.1352, "step": 12859 }, { "epoch": 82.89799196787149, "grad_norm": 49.146446228027344, "learning_rate": 8.516129032258065e-06, "loss": 0.3236, "step": 12860 }, { "epoch": 82.90441767068273, "grad_norm": 10.073817253112793, "learning_rate": 8.512903225806452e-06, "loss": 0.4844, "step": 12861 }, { "epoch": 82.91084337349398, "grad_norm": 6.950591564178467, "learning_rate": 8.50967741935484e-06, "loss": 0.5201, "step": 12862 }, { "epoch": 82.91726907630522, "grad_norm": 18.90964698791504, "learning_rate": 8.506451612903227e-06, "loss": 0.3891, "step": 12863 }, { "epoch": 82.92369477911646, "grad_norm": 7.210976600646973, "learning_rate": 8.503225806451613e-06, "loss": 0.277, "step": 12864 }, { "epoch": 82.93012048192772, "grad_norm": 6.383068084716797, "learning_rate": 8.500000000000002e-06, "loss": 0.2965, "step": 12865 }, { "epoch": 82.93654618473896, "grad_norm": 21.475500106811523, "learning_rate": 8.496774193548387e-06, "loss": 0.2489, "step": 12866 }, { "epoch": 82.9429718875502, "grad_norm": 6.954483985900879, "learning_rate": 8.493548387096775e-06, "loss": 0.142, "step": 12867 }, { "epoch": 82.94939759036144, "grad_norm": 25.30748176574707, "learning_rate": 8.490322580645162e-06, "loss": 0.6495, "step": 12868 }, { "epoch": 82.95582329317268, "grad_norm": 18.12696075439453, "learning_rate": 8.48709677419355e-06, "loss": 0.3851, "step": 12869 }, { "epoch": 82.96224899598394, "grad_norm": 10.696881294250488, "learning_rate": 8.483870967741935e-06, "loss": 0.3127, "step": 12870 }, { "epoch": 82.96867469879518, "grad_norm": 21.874074935913086, "learning_rate": 8.480645161290323e-06, "loss": 0.5221, "step": 12871 }, { "epoch": 82.97510040160643, "grad_norm": 4.80658483505249, "learning_rate": 8.47741935483871e-06, "loss": 0.0991, "step": 12872 }, { "epoch": 82.98152610441767, "grad_norm": 18.379838943481445, "learning_rate": 8.474193548387096e-06, "loss": 0.3164, "step": 12873 }, { "epoch": 82.98795180722891, "grad_norm": 14.529796600341797, "learning_rate": 8.470967741935485e-06, "loss": 0.3216, "step": 12874 }, { "epoch": 82.99437751004017, "grad_norm": 7.872305393218994, "learning_rate": 8.46774193548387e-06, "loss": 0.3244, "step": 12875 }, { "epoch": 83.00401606425703, "grad_norm": 4.742885112762451, "learning_rate": 8.464516129032258e-06, "loss": 0.126, "step": 12876 }, { "epoch": 83.01044176706827, "grad_norm": 10.674004554748535, "learning_rate": 8.461290322580645e-06, "loss": 0.1489, "step": 12877 }, { "epoch": 83.01686746987951, "grad_norm": 12.6585111618042, "learning_rate": 8.458064516129033e-06, "loss": 0.2489, "step": 12878 }, { "epoch": 83.02329317269076, "grad_norm": 37.35628890991211, "learning_rate": 8.454838709677418e-06, "loss": 0.3607, "step": 12879 }, { "epoch": 83.02971887550201, "grad_norm": 11.746294021606445, "learning_rate": 8.451612903225808e-06, "loss": 0.429, "step": 12880 }, { "epoch": 83.03614457831326, "grad_norm": 30.71558952331543, "learning_rate": 8.448387096774193e-06, "loss": 0.3874, "step": 12881 }, { "epoch": 83.0425702811245, "grad_norm": 22.391223907470703, "learning_rate": 8.44516129032258e-06, "loss": 0.3979, "step": 12882 }, { "epoch": 83.04899598393574, "grad_norm": 17.441585540771484, "learning_rate": 8.441935483870968e-06, "loss": 0.2607, "step": 12883 }, { "epoch": 83.05542168674698, "grad_norm": 10.307107925415039, "learning_rate": 8.438709677419355e-06, "loss": 0.4046, "step": 12884 }, { "epoch": 83.06184738955824, "grad_norm": 41.84775924682617, "learning_rate": 8.435483870967741e-06, "loss": 0.3557, "step": 12885 }, { "epoch": 83.06827309236948, "grad_norm": 65.14762115478516, "learning_rate": 8.43225806451613e-06, "loss": 0.411, "step": 12886 }, { "epoch": 83.07469879518072, "grad_norm": 23.9814453125, "learning_rate": 8.429032258064516e-06, "loss": 0.3563, "step": 12887 }, { "epoch": 83.08112449799196, "grad_norm": 11.43112564086914, "learning_rate": 8.425806451612903e-06, "loss": 0.4894, "step": 12888 }, { "epoch": 83.08755020080321, "grad_norm": 10.109868049621582, "learning_rate": 8.42258064516129e-06, "loss": 0.1332, "step": 12889 }, { "epoch": 83.09397590361446, "grad_norm": 2.1427831649780273, "learning_rate": 8.419354838709678e-06, "loss": 0.0532, "step": 12890 }, { "epoch": 83.1004016064257, "grad_norm": 11.5997314453125, "learning_rate": 8.416129032258064e-06, "loss": 0.3879, "step": 12891 }, { "epoch": 83.10682730923695, "grad_norm": 12.717986106872559, "learning_rate": 8.412903225806453e-06, "loss": 0.1692, "step": 12892 }, { "epoch": 83.11325301204819, "grad_norm": 14.50624942779541, "learning_rate": 8.409677419354839e-06, "loss": 0.2379, "step": 12893 }, { "epoch": 83.11967871485943, "grad_norm": 7.730953216552734, "learning_rate": 8.406451612903226e-06, "loss": 0.3025, "step": 12894 }, { "epoch": 83.12610441767069, "grad_norm": 12.184335708618164, "learning_rate": 8.403225806451613e-06, "loss": 0.7161, "step": 12895 }, { "epoch": 83.13253012048193, "grad_norm": 11.406926155090332, "learning_rate": 8.400000000000001e-06, "loss": 0.2396, "step": 12896 }, { "epoch": 83.13895582329317, "grad_norm": 8.602312088012695, "learning_rate": 8.396774193548387e-06, "loss": 0.2769, "step": 12897 }, { "epoch": 83.14538152610442, "grad_norm": 11.687642097473145, "learning_rate": 8.393548387096776e-06, "loss": 0.1587, "step": 12898 }, { "epoch": 83.15180722891566, "grad_norm": 2.6813442707061768, "learning_rate": 8.390322580645161e-06, "loss": 0.1084, "step": 12899 }, { "epoch": 83.15823293172691, "grad_norm": 7.668830871582031, "learning_rate": 8.387096774193549e-06, "loss": 0.1578, "step": 12900 }, { "epoch": 83.16465863453816, "grad_norm": 5.627241134643555, "learning_rate": 8.383870967741936e-06, "loss": 0.1882, "step": 12901 }, { "epoch": 83.1710843373494, "grad_norm": 13.49813461303711, "learning_rate": 8.380645161290324e-06, "loss": 0.2507, "step": 12902 }, { "epoch": 83.17751004016064, "grad_norm": 12.877473831176758, "learning_rate": 8.37741935483871e-06, "loss": 0.1096, "step": 12903 }, { "epoch": 83.18393574297188, "grad_norm": 23.87272834777832, "learning_rate": 8.374193548387098e-06, "loss": 0.3418, "step": 12904 }, { "epoch": 83.19036144578314, "grad_norm": 18.52161407470703, "learning_rate": 8.370967741935484e-06, "loss": 0.1994, "step": 12905 }, { "epoch": 83.19678714859438, "grad_norm": 4.251863956451416, "learning_rate": 8.367741935483872e-06, "loss": 0.2558, "step": 12906 }, { "epoch": 83.20321285140562, "grad_norm": 7.596682548522949, "learning_rate": 8.364516129032259e-06, "loss": 0.273, "step": 12907 }, { "epoch": 83.20963855421687, "grad_norm": 12.932543754577637, "learning_rate": 8.361290322580645e-06, "loss": 0.1053, "step": 12908 }, { "epoch": 83.21606425702811, "grad_norm": 6.976655006408691, "learning_rate": 8.358064516129032e-06, "loss": 0.3574, "step": 12909 }, { "epoch": 83.22248995983935, "grad_norm": 21.97170639038086, "learning_rate": 8.35483870967742e-06, "loss": 0.4885, "step": 12910 }, { "epoch": 83.2289156626506, "grad_norm": 13.27502155303955, "learning_rate": 8.351612903225807e-06, "loss": 0.492, "step": 12911 }, { "epoch": 83.23534136546185, "grad_norm": 23.50288963317871, "learning_rate": 8.348387096774194e-06, "loss": 0.1943, "step": 12912 }, { "epoch": 83.24176706827309, "grad_norm": 18.896135330200195, "learning_rate": 8.345161290322582e-06, "loss": 0.4264, "step": 12913 }, { "epoch": 83.24819277108433, "grad_norm": 16.392614364624023, "learning_rate": 8.341935483870967e-06, "loss": 0.2778, "step": 12914 }, { "epoch": 83.25461847389558, "grad_norm": 4.7682414054870605, "learning_rate": 8.338709677419355e-06, "loss": 0.3045, "step": 12915 }, { "epoch": 83.26104417670683, "grad_norm": 10.637842178344727, "learning_rate": 8.335483870967742e-06, "loss": 0.4158, "step": 12916 }, { "epoch": 83.26746987951807, "grad_norm": 7.311986923217773, "learning_rate": 8.33225806451613e-06, "loss": 0.2883, "step": 12917 }, { "epoch": 83.27389558232932, "grad_norm": 8.242593765258789, "learning_rate": 8.329032258064517e-06, "loss": 0.2646, "step": 12918 }, { "epoch": 83.28032128514056, "grad_norm": 19.270366668701172, "learning_rate": 8.325806451612904e-06, "loss": 0.1042, "step": 12919 }, { "epoch": 83.2867469879518, "grad_norm": 23.281814575195312, "learning_rate": 8.32258064516129e-06, "loss": 0.3012, "step": 12920 }, { "epoch": 83.29317269076306, "grad_norm": 7.475740432739258, "learning_rate": 8.319354838709677e-06, "loss": 0.352, "step": 12921 }, { "epoch": 83.2995983935743, "grad_norm": 10.91305923461914, "learning_rate": 8.316129032258065e-06, "loss": 0.2123, "step": 12922 }, { "epoch": 83.30602409638554, "grad_norm": 2.0754785537719727, "learning_rate": 8.312903225806452e-06, "loss": 0.1837, "step": 12923 }, { "epoch": 83.31244979919678, "grad_norm": 11.098100662231445, "learning_rate": 8.309677419354838e-06, "loss": 0.3693, "step": 12924 }, { "epoch": 83.31887550200803, "grad_norm": 23.940872192382812, "learning_rate": 8.306451612903227e-06, "loss": 0.5528, "step": 12925 }, { "epoch": 83.32530120481928, "grad_norm": 5.233438968658447, "learning_rate": 8.303225806451613e-06, "loss": 0.1945, "step": 12926 }, { "epoch": 83.33172690763053, "grad_norm": 8.205334663391113, "learning_rate": 8.3e-06, "loss": 0.2446, "step": 12927 }, { "epoch": 83.33815261044177, "grad_norm": 26.205848693847656, "learning_rate": 8.296774193548388e-06, "loss": 0.3059, "step": 12928 }, { "epoch": 83.34457831325301, "grad_norm": 24.979185104370117, "learning_rate": 8.293548387096775e-06, "loss": 0.4236, "step": 12929 }, { "epoch": 83.35100401606425, "grad_norm": 5.806652545928955, "learning_rate": 8.29032258064516e-06, "loss": 0.4234, "step": 12930 }, { "epoch": 83.35742971887551, "grad_norm": 7.006274223327637, "learning_rate": 8.28709677419355e-06, "loss": 0.3259, "step": 12931 }, { "epoch": 83.36385542168675, "grad_norm": 4.3123931884765625, "learning_rate": 8.283870967741935e-06, "loss": 0.1561, "step": 12932 }, { "epoch": 83.37028112449799, "grad_norm": 7.815450668334961, "learning_rate": 8.280645161290323e-06, "loss": 0.709, "step": 12933 }, { "epoch": 83.37670682730923, "grad_norm": 45.77173614501953, "learning_rate": 8.27741935483871e-06, "loss": 0.2267, "step": 12934 }, { "epoch": 83.38313253012048, "grad_norm": 3.6178834438323975, "learning_rate": 8.274193548387098e-06, "loss": 0.4371, "step": 12935 }, { "epoch": 83.38955823293173, "grad_norm": 7.742133617401123, "learning_rate": 8.270967741935483e-06, "loss": 0.1139, "step": 12936 }, { "epoch": 83.39598393574298, "grad_norm": 17.63558578491211, "learning_rate": 8.267741935483872e-06, "loss": 0.2758, "step": 12937 }, { "epoch": 83.40240963855422, "grad_norm": 23.60877227783203, "learning_rate": 8.264516129032258e-06, "loss": 0.2924, "step": 12938 }, { "epoch": 83.40883534136546, "grad_norm": 7.974031925201416, "learning_rate": 8.261290322580646e-06, "loss": 0.3585, "step": 12939 }, { "epoch": 83.4152610441767, "grad_norm": 13.001275062561035, "learning_rate": 8.258064516129033e-06, "loss": 0.2058, "step": 12940 }, { "epoch": 83.42168674698796, "grad_norm": 9.659226417541504, "learning_rate": 8.254838709677419e-06, "loss": 0.3141, "step": 12941 }, { "epoch": 83.4281124497992, "grad_norm": 25.107284545898438, "learning_rate": 8.251612903225806e-06, "loss": 0.3616, "step": 12942 }, { "epoch": 83.43453815261044, "grad_norm": 22.974925994873047, "learning_rate": 8.248387096774193e-06, "loss": 0.3872, "step": 12943 }, { "epoch": 83.44096385542169, "grad_norm": 11.233532905578613, "learning_rate": 8.245161290322581e-06, "loss": 0.1526, "step": 12944 }, { "epoch": 83.44738955823293, "grad_norm": 13.053119659423828, "learning_rate": 8.241935483870968e-06, "loss": 0.5153, "step": 12945 }, { "epoch": 83.45381526104417, "grad_norm": 5.1252336502075195, "learning_rate": 8.238709677419356e-06, "loss": 0.1018, "step": 12946 }, { "epoch": 83.46024096385543, "grad_norm": 26.739395141601562, "learning_rate": 8.235483870967741e-06, "loss": 0.2159, "step": 12947 }, { "epoch": 83.46666666666667, "grad_norm": 17.54152488708496, "learning_rate": 8.232258064516129e-06, "loss": 0.4567, "step": 12948 }, { "epoch": 83.47309236947791, "grad_norm": 16.1575870513916, "learning_rate": 8.229032258064516e-06, "loss": 0.2818, "step": 12949 }, { "epoch": 83.47951807228915, "grad_norm": 4.982160568237305, "learning_rate": 8.225806451612904e-06, "loss": 0.3645, "step": 12950 }, { "epoch": 83.4859437751004, "grad_norm": 4.19518518447876, "learning_rate": 8.222580645161291e-06, "loss": 0.233, "step": 12951 }, { "epoch": 83.49236947791165, "grad_norm": 6.098033428192139, "learning_rate": 8.219354838709678e-06, "loss": 0.4176, "step": 12952 }, { "epoch": 83.4987951807229, "grad_norm": 4.08953857421875, "learning_rate": 8.216129032258064e-06, "loss": 0.3819, "step": 12953 }, { "epoch": 83.50522088353414, "grad_norm": 6.928491592407227, "learning_rate": 8.212903225806452e-06, "loss": 0.4784, "step": 12954 }, { "epoch": 83.51164658634538, "grad_norm": 7.389852046966553, "learning_rate": 8.209677419354839e-06, "loss": 0.2392, "step": 12955 }, { "epoch": 83.51807228915662, "grad_norm": 11.984957695007324, "learning_rate": 8.206451612903226e-06, "loss": 0.3781, "step": 12956 }, { "epoch": 83.52449799196788, "grad_norm": 13.292237281799316, "learning_rate": 8.203225806451614e-06, "loss": 0.5262, "step": 12957 }, { "epoch": 83.53092369477912, "grad_norm": 3.8465452194213867, "learning_rate": 8.200000000000001e-06, "loss": 0.304, "step": 12958 }, { "epoch": 83.53734939759036, "grad_norm": 7.190699100494385, "learning_rate": 8.196774193548387e-06, "loss": 0.161, "step": 12959 }, { "epoch": 83.5437751004016, "grad_norm": 40.61326599121094, "learning_rate": 8.193548387096774e-06, "loss": 0.428, "step": 12960 }, { "epoch": 83.55020080321285, "grad_norm": 8.694032669067383, "learning_rate": 8.190322580645162e-06, "loss": 0.5316, "step": 12961 }, { "epoch": 83.5566265060241, "grad_norm": 7.883370399475098, "learning_rate": 8.187096774193549e-06, "loss": 0.4383, "step": 12962 }, { "epoch": 83.56305220883534, "grad_norm": 5.991030216217041, "learning_rate": 8.183870967741936e-06, "loss": 0.315, "step": 12963 }, { "epoch": 83.56947791164659, "grad_norm": 12.921839714050293, "learning_rate": 8.180645161290324e-06, "loss": 0.4415, "step": 12964 }, { "epoch": 83.57590361445783, "grad_norm": 16.75126838684082, "learning_rate": 8.17741935483871e-06, "loss": 0.4496, "step": 12965 }, { "epoch": 83.58232931726907, "grad_norm": 28.71732521057129, "learning_rate": 8.174193548387097e-06, "loss": 0.359, "step": 12966 }, { "epoch": 83.58875502008033, "grad_norm": 18.486133575439453, "learning_rate": 8.170967741935484e-06, "loss": 0.3185, "step": 12967 }, { "epoch": 83.59518072289157, "grad_norm": 9.409573554992676, "learning_rate": 8.167741935483872e-06, "loss": 0.3388, "step": 12968 }, { "epoch": 83.60160642570281, "grad_norm": 11.386242866516113, "learning_rate": 8.164516129032259e-06, "loss": 0.4963, "step": 12969 }, { "epoch": 83.60803212851405, "grad_norm": 12.582253456115723, "learning_rate": 8.161290322580647e-06, "loss": 0.2811, "step": 12970 }, { "epoch": 83.6144578313253, "grad_norm": 7.386269569396973, "learning_rate": 8.158064516129032e-06, "loss": 0.0928, "step": 12971 }, { "epoch": 83.62088353413655, "grad_norm": 19.66320037841797, "learning_rate": 8.15483870967742e-06, "loss": 0.2287, "step": 12972 }, { "epoch": 83.6273092369478, "grad_norm": 35.19300842285156, "learning_rate": 8.151612903225807e-06, "loss": 0.307, "step": 12973 }, { "epoch": 83.63373493975904, "grad_norm": 17.134389877319336, "learning_rate": 8.148387096774194e-06, "loss": 0.3613, "step": 12974 }, { "epoch": 83.64016064257028, "grad_norm": 13.710550308227539, "learning_rate": 8.145161290322582e-06, "loss": 0.5008, "step": 12975 }, { "epoch": 83.64658634538152, "grad_norm": 9.223755836486816, "learning_rate": 8.141935483870968e-06, "loss": 0.373, "step": 12976 }, { "epoch": 83.65301204819278, "grad_norm": 14.092159271240234, "learning_rate": 8.138709677419355e-06, "loss": 0.3568, "step": 12977 }, { "epoch": 83.65943775100402, "grad_norm": 8.235334396362305, "learning_rate": 8.135483870967742e-06, "loss": 0.343, "step": 12978 }, { "epoch": 83.66586345381526, "grad_norm": 9.03912353515625, "learning_rate": 8.13225806451613e-06, "loss": 0.2346, "step": 12979 }, { "epoch": 83.6722891566265, "grad_norm": 10.489898681640625, "learning_rate": 8.129032258064515e-06, "loss": 0.3025, "step": 12980 }, { "epoch": 83.67871485943775, "grad_norm": 13.341700553894043, "learning_rate": 8.125806451612905e-06, "loss": 0.1386, "step": 12981 }, { "epoch": 83.68514056224899, "grad_norm": 16.51919937133789, "learning_rate": 8.12258064516129e-06, "loss": 0.5126, "step": 12982 }, { "epoch": 83.69156626506025, "grad_norm": 7.791223526000977, "learning_rate": 8.119354838709678e-06, "loss": 0.4407, "step": 12983 }, { "epoch": 83.69799196787149, "grad_norm": 4.462059497833252, "learning_rate": 8.116129032258065e-06, "loss": 0.2518, "step": 12984 }, { "epoch": 83.70441767068273, "grad_norm": 35.90179443359375, "learning_rate": 8.112903225806452e-06, "loss": 0.3888, "step": 12985 }, { "epoch": 83.71084337349397, "grad_norm": 5.8991498947143555, "learning_rate": 8.109677419354838e-06, "loss": 0.3297, "step": 12986 }, { "epoch": 83.71726907630521, "grad_norm": 6.383023262023926, "learning_rate": 8.106451612903226e-06, "loss": 0.4974, "step": 12987 }, { "epoch": 83.72369477911647, "grad_norm": 5.962385654449463, "learning_rate": 8.103225806451613e-06, "loss": 0.1472, "step": 12988 }, { "epoch": 83.73012048192771, "grad_norm": 2.1540870666503906, "learning_rate": 8.1e-06, "loss": 0.4238, "step": 12989 }, { "epoch": 83.73654618473896, "grad_norm": 18.608556747436523, "learning_rate": 8.096774193548388e-06, "loss": 0.3497, "step": 12990 }, { "epoch": 83.7429718875502, "grad_norm": 19.231184005737305, "learning_rate": 8.093548387096775e-06, "loss": 0.4253, "step": 12991 }, { "epoch": 83.74939759036144, "grad_norm": 17.490739822387695, "learning_rate": 8.090322580645161e-06, "loss": 0.4076, "step": 12992 }, { "epoch": 83.7558232931727, "grad_norm": 13.687716484069824, "learning_rate": 8.087096774193548e-06, "loss": 0.1805, "step": 12993 }, { "epoch": 83.76224899598394, "grad_norm": 13.039249420166016, "learning_rate": 8.083870967741936e-06, "loss": 0.4641, "step": 12994 }, { "epoch": 83.76867469879518, "grad_norm": 10.156686782836914, "learning_rate": 8.080645161290323e-06, "loss": 0.4743, "step": 12995 }, { "epoch": 83.77510040160642, "grad_norm": 9.591651916503906, "learning_rate": 8.07741935483871e-06, "loss": 0.4702, "step": 12996 }, { "epoch": 83.78152610441767, "grad_norm": 10.840063095092773, "learning_rate": 8.074193548387098e-06, "loss": 0.2747, "step": 12997 }, { "epoch": 83.78795180722892, "grad_norm": 9.592023849487305, "learning_rate": 8.070967741935484e-06, "loss": 0.3821, "step": 12998 }, { "epoch": 83.79437751004016, "grad_norm": 24.280025482177734, "learning_rate": 8.067741935483871e-06, "loss": 0.5218, "step": 12999 }, { "epoch": 83.8008032128514, "grad_norm": 35.62915802001953, "learning_rate": 8.064516129032258e-06, "loss": 0.211, "step": 13000 }, { "epoch": 83.80722891566265, "grad_norm": 1.9989758729934692, "learning_rate": 8.061290322580646e-06, "loss": 0.2082, "step": 13001 }, { "epoch": 83.81365461847389, "grad_norm": 11.50822925567627, "learning_rate": 8.058064516129033e-06, "loss": 0.2374, "step": 13002 }, { "epoch": 83.82008032128515, "grad_norm": 9.833263397216797, "learning_rate": 8.05483870967742e-06, "loss": 0.2891, "step": 13003 }, { "epoch": 83.82650602409639, "grad_norm": 6.046881198883057, "learning_rate": 8.051612903225806e-06, "loss": 0.4242, "step": 13004 }, { "epoch": 83.83293172690763, "grad_norm": 7.637340068817139, "learning_rate": 8.048387096774194e-06, "loss": 0.2475, "step": 13005 }, { "epoch": 83.83935742971887, "grad_norm": 7.3429694175720215, "learning_rate": 8.045161290322581e-06, "loss": 0.5211, "step": 13006 }, { "epoch": 83.84578313253012, "grad_norm": 9.063069343566895, "learning_rate": 8.041935483870969e-06, "loss": 0.2371, "step": 13007 }, { "epoch": 83.85220883534137, "grad_norm": 26.257368087768555, "learning_rate": 8.038709677419356e-06, "loss": 0.4661, "step": 13008 }, { "epoch": 83.85863453815261, "grad_norm": 5.364425182342529, "learning_rate": 8.035483870967743e-06, "loss": 0.3447, "step": 13009 }, { "epoch": 83.86506024096386, "grad_norm": 16.118284225463867, "learning_rate": 8.032258064516129e-06, "loss": 0.5323, "step": 13010 }, { "epoch": 83.8714859437751, "grad_norm": 29.680889129638672, "learning_rate": 8.029032258064516e-06, "loss": 0.4258, "step": 13011 }, { "epoch": 83.87791164658634, "grad_norm": 6.849820613861084, "learning_rate": 8.025806451612904e-06, "loss": 0.4242, "step": 13012 }, { "epoch": 83.88433734939758, "grad_norm": 6.900124549865723, "learning_rate": 8.02258064516129e-06, "loss": 0.4928, "step": 13013 }, { "epoch": 83.89076305220884, "grad_norm": 14.502230644226074, "learning_rate": 8.019354838709679e-06, "loss": 0.189, "step": 13014 }, { "epoch": 83.89718875502008, "grad_norm": 8.316871643066406, "learning_rate": 8.016129032258064e-06, "loss": 0.3469, "step": 13015 }, { "epoch": 83.90361445783132, "grad_norm": 33.29213333129883, "learning_rate": 8.012903225806452e-06, "loss": 0.4886, "step": 13016 }, { "epoch": 83.91004016064257, "grad_norm": 19.586740493774414, "learning_rate": 8.009677419354839e-06, "loss": 0.2768, "step": 13017 }, { "epoch": 83.91646586345381, "grad_norm": 13.91855525970459, "learning_rate": 8.006451612903227e-06, "loss": 0.5274, "step": 13018 }, { "epoch": 83.92289156626506, "grad_norm": 11.489862442016602, "learning_rate": 8.003225806451612e-06, "loss": 0.3119, "step": 13019 }, { "epoch": 83.92931726907631, "grad_norm": 10.716594696044922, "learning_rate": 8.000000000000001e-06, "loss": 0.2789, "step": 13020 }, { "epoch": 83.93574297188755, "grad_norm": 6.243762969970703, "learning_rate": 7.996774193548387e-06, "loss": 0.2431, "step": 13021 }, { "epoch": 83.94216867469879, "grad_norm": 9.240059852600098, "learning_rate": 7.993548387096774e-06, "loss": 0.116, "step": 13022 }, { "epoch": 83.94859437751003, "grad_norm": 7.991429328918457, "learning_rate": 7.990322580645162e-06, "loss": 0.3165, "step": 13023 }, { "epoch": 83.95502008032129, "grad_norm": 9.6337308883667, "learning_rate": 7.98709677419355e-06, "loss": 0.268, "step": 13024 }, { "epoch": 83.96144578313253, "grad_norm": 10.710724830627441, "learning_rate": 7.983870967741935e-06, "loss": 0.3524, "step": 13025 }, { "epoch": 83.96787148594377, "grad_norm": 10.890146255493164, "learning_rate": 7.980645161290324e-06, "loss": 0.3229, "step": 13026 }, { "epoch": 83.97429718875502, "grad_norm": 3.0137674808502197, "learning_rate": 7.97741935483871e-06, "loss": 0.3515, "step": 13027 }, { "epoch": 83.98072289156626, "grad_norm": 7.872539043426514, "learning_rate": 7.974193548387097e-06, "loss": 0.3404, "step": 13028 }, { "epoch": 83.98714859437752, "grad_norm": 7.9871602058410645, "learning_rate": 7.970967741935485e-06, "loss": 0.3329, "step": 13029 }, { "epoch": 83.99357429718876, "grad_norm": 10.729116439819336, "learning_rate": 7.967741935483872e-06, "loss": 0.5596, "step": 13030 }, { "epoch": 84.00321285140562, "grad_norm": 14.314599990844727, "learning_rate": 7.964516129032258e-06, "loss": 0.3318, "step": 13031 }, { "epoch": 84.00963855421686, "grad_norm": 10.693635940551758, "learning_rate": 7.961290322580647e-06, "loss": 0.1188, "step": 13032 }, { "epoch": 84.0160642570281, "grad_norm": 13.060120582580566, "learning_rate": 7.958064516129032e-06, "loss": 0.1306, "step": 13033 }, { "epoch": 84.02248995983936, "grad_norm": 5.50051736831665, "learning_rate": 7.95483870967742e-06, "loss": 0.1611, "step": 13034 }, { "epoch": 84.0289156626506, "grad_norm": 8.258410453796387, "learning_rate": 7.951612903225807e-06, "loss": 0.1532, "step": 13035 }, { "epoch": 84.03534136546185, "grad_norm": 17.092947006225586, "learning_rate": 7.948387096774195e-06, "loss": 0.3548, "step": 13036 }, { "epoch": 84.04176706827309, "grad_norm": 8.828580856323242, "learning_rate": 7.94516129032258e-06, "loss": 0.2899, "step": 13037 }, { "epoch": 84.04819277108433, "grad_norm": 21.379417419433594, "learning_rate": 7.94193548387097e-06, "loss": 0.1942, "step": 13038 }, { "epoch": 84.05461847389559, "grad_norm": 30.634008407592773, "learning_rate": 7.938709677419355e-06, "loss": 0.1498, "step": 13039 }, { "epoch": 84.06104417670683, "grad_norm": 13.113288879394531, "learning_rate": 7.935483870967743e-06, "loss": 0.2341, "step": 13040 }, { "epoch": 84.06746987951807, "grad_norm": 19.791095733642578, "learning_rate": 7.93225806451613e-06, "loss": 0.3627, "step": 13041 }, { "epoch": 84.07389558232931, "grad_norm": 18.922863006591797, "learning_rate": 7.929032258064517e-06, "loss": 0.4727, "step": 13042 }, { "epoch": 84.08032128514056, "grad_norm": 18.583818435668945, "learning_rate": 7.925806451612903e-06, "loss": 0.2144, "step": 13043 }, { "epoch": 84.08674698795181, "grad_norm": 19.241458892822266, "learning_rate": 7.922580645161292e-06, "loss": 0.5935, "step": 13044 }, { "epoch": 84.09317269076305, "grad_norm": 38.224586486816406, "learning_rate": 7.919354838709678e-06, "loss": 0.3809, "step": 13045 }, { "epoch": 84.0995983935743, "grad_norm": 9.21232795715332, "learning_rate": 7.916129032258065e-06, "loss": 0.4641, "step": 13046 }, { "epoch": 84.10602409638554, "grad_norm": 11.236682891845703, "learning_rate": 7.912903225806453e-06, "loss": 0.3187, "step": 13047 }, { "epoch": 84.11244979919678, "grad_norm": 8.16505241394043, "learning_rate": 7.909677419354838e-06, "loss": 0.3957, "step": 13048 }, { "epoch": 84.11887550200804, "grad_norm": 17.64716911315918, "learning_rate": 7.906451612903226e-06, "loss": 0.3128, "step": 13049 }, { "epoch": 84.12530120481928, "grad_norm": 5.523792266845703, "learning_rate": 7.903225806451613e-06, "loss": 0.2206, "step": 13050 }, { "epoch": 84.13172690763052, "grad_norm": 30.77695655822754, "learning_rate": 7.9e-06, "loss": 0.5435, "step": 13051 }, { "epoch": 84.13815261044176, "grad_norm": 15.274423599243164, "learning_rate": 7.896774193548386e-06, "loss": 0.2413, "step": 13052 }, { "epoch": 84.144578313253, "grad_norm": 3.9524149894714355, "learning_rate": 7.893548387096775e-06, "loss": 0.184, "step": 13053 }, { "epoch": 84.15100401606426, "grad_norm": 22.830135345458984, "learning_rate": 7.890322580645161e-06, "loss": 0.4362, "step": 13054 }, { "epoch": 84.1574297188755, "grad_norm": 6.39321756362915, "learning_rate": 7.887096774193549e-06, "loss": 0.2327, "step": 13055 }, { "epoch": 84.16385542168675, "grad_norm": 13.952515602111816, "learning_rate": 7.883870967741936e-06, "loss": 0.4009, "step": 13056 }, { "epoch": 84.17028112449799, "grad_norm": 12.58195686340332, "learning_rate": 7.880645161290323e-06, "loss": 0.2344, "step": 13057 }, { "epoch": 84.17670682730923, "grad_norm": 22.676145553588867, "learning_rate": 7.877419354838709e-06, "loss": 0.3172, "step": 13058 }, { "epoch": 84.18313253012049, "grad_norm": 17.887285232543945, "learning_rate": 7.874193548387098e-06, "loss": 0.2895, "step": 13059 }, { "epoch": 84.18955823293173, "grad_norm": 17.363933563232422, "learning_rate": 7.870967741935484e-06, "loss": 0.5354, "step": 13060 }, { "epoch": 84.19598393574297, "grad_norm": 2.7454416751861572, "learning_rate": 7.867741935483871e-06, "loss": 0.2473, "step": 13061 }, { "epoch": 84.20240963855422, "grad_norm": 21.531944274902344, "learning_rate": 7.864516129032259e-06, "loss": 0.2766, "step": 13062 }, { "epoch": 84.20883534136546, "grad_norm": 2.910982847213745, "learning_rate": 7.861290322580646e-06, "loss": 0.1631, "step": 13063 }, { "epoch": 84.21526104417671, "grad_norm": 5.103005409240723, "learning_rate": 7.858064516129032e-06, "loss": 0.2317, "step": 13064 }, { "epoch": 84.22168674698796, "grad_norm": 7.83416748046875, "learning_rate": 7.85483870967742e-06, "loss": 0.4115, "step": 13065 }, { "epoch": 84.2281124497992, "grad_norm": 7.304223537445068, "learning_rate": 7.851612903225807e-06, "loss": 0.3974, "step": 13066 }, { "epoch": 84.23453815261044, "grad_norm": 8.871739387512207, "learning_rate": 7.848387096774194e-06, "loss": 0.6207, "step": 13067 }, { "epoch": 84.24096385542168, "grad_norm": 8.125308990478516, "learning_rate": 7.845161290322581e-06, "loss": 0.3846, "step": 13068 }, { "epoch": 84.24738955823292, "grad_norm": 11.607333183288574, "learning_rate": 7.841935483870969e-06, "loss": 0.1466, "step": 13069 }, { "epoch": 84.25381526104418, "grad_norm": 15.661078453063965, "learning_rate": 7.838709677419354e-06, "loss": 0.346, "step": 13070 }, { "epoch": 84.26024096385542, "grad_norm": 12.503586769104004, "learning_rate": 7.835483870967744e-06, "loss": 0.5102, "step": 13071 }, { "epoch": 84.26666666666667, "grad_norm": 7.690601825714111, "learning_rate": 7.83225806451613e-06, "loss": 0.5569, "step": 13072 }, { "epoch": 84.27309236947791, "grad_norm": 28.51494598388672, "learning_rate": 7.829032258064517e-06, "loss": 0.2887, "step": 13073 }, { "epoch": 84.27951807228915, "grad_norm": 23.913328170776367, "learning_rate": 7.825806451612904e-06, "loss": 0.4335, "step": 13074 }, { "epoch": 84.2859437751004, "grad_norm": 10.240514755249023, "learning_rate": 7.822580645161291e-06, "loss": 0.3598, "step": 13075 }, { "epoch": 84.29236947791165, "grad_norm": 14.683364868164062, "learning_rate": 7.819354838709677e-06, "loss": 0.363, "step": 13076 }, { "epoch": 84.29879518072289, "grad_norm": 3.0118842124938965, "learning_rate": 7.816129032258066e-06, "loss": 0.5307, "step": 13077 }, { "epoch": 84.30522088353413, "grad_norm": 10.14458179473877, "learning_rate": 7.812903225806452e-06, "loss": 0.3507, "step": 13078 }, { "epoch": 84.31164658634538, "grad_norm": 10.196748733520508, "learning_rate": 7.80967741935484e-06, "loss": 0.3223, "step": 13079 }, { "epoch": 84.31807228915663, "grad_norm": 27.334306716918945, "learning_rate": 7.806451612903227e-06, "loss": 0.1846, "step": 13080 }, { "epoch": 84.32449799196787, "grad_norm": 11.205466270446777, "learning_rate": 7.803225806451614e-06, "loss": 0.3223, "step": 13081 }, { "epoch": 84.33092369477912, "grad_norm": 8.555020332336426, "learning_rate": 7.8e-06, "loss": 0.3196, "step": 13082 }, { "epoch": 84.33734939759036, "grad_norm": 14.334403991699219, "learning_rate": 7.796774193548387e-06, "loss": 0.299, "step": 13083 }, { "epoch": 84.3437751004016, "grad_norm": 6.610288143157959, "learning_rate": 7.793548387096775e-06, "loss": 0.4101, "step": 13084 }, { "epoch": 84.35020080321286, "grad_norm": 13.288288116455078, "learning_rate": 7.79032258064516e-06, "loss": 0.2682, "step": 13085 }, { "epoch": 84.3566265060241, "grad_norm": 8.321413040161133, "learning_rate": 7.78709677419355e-06, "loss": 0.1832, "step": 13086 }, { "epoch": 84.36305220883534, "grad_norm": 23.696063995361328, "learning_rate": 7.783870967741935e-06, "loss": 0.4314, "step": 13087 }, { "epoch": 84.36947791164658, "grad_norm": 9.500923156738281, "learning_rate": 7.780645161290323e-06, "loss": 0.2778, "step": 13088 }, { "epoch": 84.37590361445783, "grad_norm": 12.018815994262695, "learning_rate": 7.77741935483871e-06, "loss": 0.1958, "step": 13089 }, { "epoch": 84.38232931726908, "grad_norm": 7.247880935668945, "learning_rate": 7.774193548387097e-06, "loss": 0.1952, "step": 13090 }, { "epoch": 84.38875502008032, "grad_norm": 12.808196067810059, "learning_rate": 7.770967741935483e-06, "loss": 0.2909, "step": 13091 }, { "epoch": 84.39518072289157, "grad_norm": 134.1946258544922, "learning_rate": 7.767741935483872e-06, "loss": 0.1429, "step": 13092 }, { "epoch": 84.40160642570281, "grad_norm": 11.064109802246094, "learning_rate": 7.764516129032258e-06, "loss": 0.6218, "step": 13093 }, { "epoch": 84.40803212851405, "grad_norm": 23.35470962524414, "learning_rate": 7.761290322580645e-06, "loss": 0.4794, "step": 13094 }, { "epoch": 84.41445783132531, "grad_norm": 3.355945110321045, "learning_rate": 7.758064516129033e-06, "loss": 0.182, "step": 13095 }, { "epoch": 84.42088353413655, "grad_norm": 4.6531853675842285, "learning_rate": 7.75483870967742e-06, "loss": 0.1767, "step": 13096 }, { "epoch": 84.42730923694779, "grad_norm": 14.627321243286133, "learning_rate": 7.751612903225806e-06, "loss": 0.3186, "step": 13097 }, { "epoch": 84.43373493975903, "grad_norm": 29.362945556640625, "learning_rate": 7.748387096774195e-06, "loss": 0.4336, "step": 13098 }, { "epoch": 84.44016064257028, "grad_norm": 12.410299301147461, "learning_rate": 7.74516129032258e-06, "loss": 0.495, "step": 13099 }, { "epoch": 84.44658634538153, "grad_norm": 14.495796203613281, "learning_rate": 7.741935483870968e-06, "loss": 0.4988, "step": 13100 }, { "epoch": 84.45301204819278, "grad_norm": 6.388632297515869, "learning_rate": 7.738709677419355e-06, "loss": 0.3139, "step": 13101 }, { "epoch": 84.45943775100402, "grad_norm": 6.944838047027588, "learning_rate": 7.735483870967743e-06, "loss": 0.2714, "step": 13102 }, { "epoch": 84.46586345381526, "grad_norm": 12.623002052307129, "learning_rate": 7.732258064516129e-06, "loss": 0.3572, "step": 13103 }, { "epoch": 84.4722891566265, "grad_norm": 12.098716735839844, "learning_rate": 7.729032258064518e-06, "loss": 0.3123, "step": 13104 }, { "epoch": 84.47871485943774, "grad_norm": 12.368375778198242, "learning_rate": 7.725806451612903e-06, "loss": 0.3965, "step": 13105 }, { "epoch": 84.485140562249, "grad_norm": 8.600168228149414, "learning_rate": 7.72258064516129e-06, "loss": 0.5124, "step": 13106 }, { "epoch": 84.49156626506024, "grad_norm": 28.161659240722656, "learning_rate": 7.719354838709678e-06, "loss": 0.6579, "step": 13107 }, { "epoch": 84.49799196787149, "grad_norm": 8.736151695251465, "learning_rate": 7.716129032258066e-06, "loss": 0.2942, "step": 13108 }, { "epoch": 84.50441767068273, "grad_norm": 17.520015716552734, "learning_rate": 7.712903225806451e-06, "loss": 0.7302, "step": 13109 }, { "epoch": 84.51084337349397, "grad_norm": 11.432783126831055, "learning_rate": 7.70967741935484e-06, "loss": 0.2584, "step": 13110 }, { "epoch": 84.51726907630523, "grad_norm": 11.542115211486816, "learning_rate": 7.706451612903226e-06, "loss": 0.1727, "step": 13111 }, { "epoch": 84.52369477911647, "grad_norm": 9.645977020263672, "learning_rate": 7.703225806451613e-06, "loss": 0.4657, "step": 13112 }, { "epoch": 84.53012048192771, "grad_norm": 9.775972366333008, "learning_rate": 7.7e-06, "loss": 0.479, "step": 13113 }, { "epoch": 84.53654618473895, "grad_norm": 18.855504989624023, "learning_rate": 7.696774193548388e-06, "loss": 0.4435, "step": 13114 }, { "epoch": 84.5429718875502, "grad_norm": 39.36705780029297, "learning_rate": 7.693548387096774e-06, "loss": 0.5264, "step": 13115 }, { "epoch": 84.54939759036145, "grad_norm": 10.625813484191895, "learning_rate": 7.690322580645163e-06, "loss": 0.1571, "step": 13116 }, { "epoch": 84.5558232931727, "grad_norm": 24.97602081298828, "learning_rate": 7.687096774193549e-06, "loss": 0.2569, "step": 13117 }, { "epoch": 84.56224899598394, "grad_norm": 13.011868476867676, "learning_rate": 7.683870967741936e-06, "loss": 0.6412, "step": 13118 }, { "epoch": 84.56867469879518, "grad_norm": 10.474196434020996, "learning_rate": 7.680645161290324e-06, "loss": 0.3622, "step": 13119 }, { "epoch": 84.57510040160642, "grad_norm": 12.040019989013672, "learning_rate": 7.67741935483871e-06, "loss": 0.3879, "step": 13120 }, { "epoch": 84.58152610441768, "grad_norm": 17.880483627319336, "learning_rate": 7.674193548387097e-06, "loss": 0.204, "step": 13121 }, { "epoch": 84.58795180722892, "grad_norm": 8.235292434692383, "learning_rate": 7.670967741935484e-06, "loss": 0.2228, "step": 13122 }, { "epoch": 84.59437751004016, "grad_norm": 11.830750465393066, "learning_rate": 7.667741935483871e-06, "loss": 0.4986, "step": 13123 }, { "epoch": 84.6008032128514, "grad_norm": 15.800065040588379, "learning_rate": 7.664516129032257e-06, "loss": 0.4144, "step": 13124 }, { "epoch": 84.60722891566265, "grad_norm": 14.01159381866455, "learning_rate": 7.661290322580646e-06, "loss": 0.2676, "step": 13125 }, { "epoch": 84.6136546184739, "grad_norm": 13.011292457580566, "learning_rate": 7.658064516129032e-06, "loss": 0.2298, "step": 13126 }, { "epoch": 84.62008032128514, "grad_norm": 16.473669052124023, "learning_rate": 7.65483870967742e-06, "loss": 0.2362, "step": 13127 }, { "epoch": 84.62650602409639, "grad_norm": 5.045297622680664, "learning_rate": 7.651612903225807e-06, "loss": 0.1666, "step": 13128 }, { "epoch": 84.63293172690763, "grad_norm": 7.107117176055908, "learning_rate": 7.648387096774194e-06, "loss": 0.3585, "step": 13129 }, { "epoch": 84.63935742971887, "grad_norm": 18.672807693481445, "learning_rate": 7.64516129032258e-06, "loss": 0.3342, "step": 13130 }, { "epoch": 84.64578313253013, "grad_norm": 6.13191556930542, "learning_rate": 7.641935483870969e-06, "loss": 0.1347, "step": 13131 }, { "epoch": 84.65220883534137, "grad_norm": 6.064619064331055, "learning_rate": 7.638709677419355e-06, "loss": 0.144, "step": 13132 }, { "epoch": 84.65863453815261, "grad_norm": 2.8272485733032227, "learning_rate": 7.635483870967742e-06, "loss": 0.2333, "step": 13133 }, { "epoch": 84.66506024096385, "grad_norm": 10.864725112915039, "learning_rate": 7.63225806451613e-06, "loss": 0.4606, "step": 13134 }, { "epoch": 84.6714859437751, "grad_norm": 15.64476490020752, "learning_rate": 7.629032258064517e-06, "loss": 0.343, "step": 13135 }, { "epoch": 84.67791164658635, "grad_norm": 49.63121032714844, "learning_rate": 7.625806451612903e-06, "loss": 0.3937, "step": 13136 }, { "epoch": 84.6843373493976, "grad_norm": 12.943089485168457, "learning_rate": 7.622580645161291e-06, "loss": 0.4551, "step": 13137 }, { "epoch": 84.69076305220884, "grad_norm": 27.957605361938477, "learning_rate": 7.619354838709677e-06, "loss": 0.4896, "step": 13138 }, { "epoch": 84.69718875502008, "grad_norm": 7.717769145965576, "learning_rate": 7.616129032258066e-06, "loss": 0.3876, "step": 13139 }, { "epoch": 84.70361445783132, "grad_norm": 13.701664924621582, "learning_rate": 7.612903225806451e-06, "loss": 0.4052, "step": 13140 }, { "epoch": 84.71004016064256, "grad_norm": 15.714855194091797, "learning_rate": 7.6096774193548396e-06, "loss": 0.3506, "step": 13141 }, { "epoch": 84.71646586345382, "grad_norm": 9.349285125732422, "learning_rate": 7.606451612903226e-06, "loss": 0.2464, "step": 13142 }, { "epoch": 84.72289156626506, "grad_norm": 11.341621398925781, "learning_rate": 7.6032258064516135e-06, "loss": 0.2945, "step": 13143 }, { "epoch": 84.7293172690763, "grad_norm": 2.362778902053833, "learning_rate": 7.6e-06, "loss": 0.3742, "step": 13144 }, { "epoch": 84.73574297188755, "grad_norm": 11.236274719238281, "learning_rate": 7.596774193548388e-06, "loss": 0.345, "step": 13145 }, { "epoch": 84.74216867469879, "grad_norm": 10.290722846984863, "learning_rate": 7.593548387096774e-06, "loss": 0.2343, "step": 13146 }, { "epoch": 84.74859437751005, "grad_norm": 6.670712471008301, "learning_rate": 7.590322580645162e-06, "loss": 0.228, "step": 13147 }, { "epoch": 84.75502008032129, "grad_norm": 3.3725509643554688, "learning_rate": 7.587096774193549e-06, "loss": 0.2659, "step": 13148 }, { "epoch": 84.76144578313253, "grad_norm": 9.336834907531738, "learning_rate": 7.583870967741936e-06, "loss": 0.2734, "step": 13149 }, { "epoch": 84.76787148594377, "grad_norm": 8.693292617797852, "learning_rate": 7.580645161290323e-06, "loss": 0.1655, "step": 13150 }, { "epoch": 84.77429718875501, "grad_norm": 9.5253267288208, "learning_rate": 7.577419354838711e-06, "loss": 0.4038, "step": 13151 }, { "epoch": 84.78072289156627, "grad_norm": 22.689178466796875, "learning_rate": 7.574193548387097e-06, "loss": 0.1372, "step": 13152 }, { "epoch": 84.78714859437751, "grad_norm": 2.457313299179077, "learning_rate": 7.570967741935485e-06, "loss": 0.2239, "step": 13153 }, { "epoch": 84.79357429718875, "grad_norm": 12.222725868225098, "learning_rate": 7.5677419354838715e-06, "loss": 0.2963, "step": 13154 }, { "epoch": 84.8, "grad_norm": 4.587557792663574, "learning_rate": 7.564516129032258e-06, "loss": 0.3593, "step": 13155 }, { "epoch": 84.80642570281124, "grad_norm": 1.9200505018234253, "learning_rate": 7.5612903225806455e-06, "loss": 0.5035, "step": 13156 }, { "epoch": 84.8128514056225, "grad_norm": 20.116172790527344, "learning_rate": 7.558064516129032e-06, "loss": 0.3209, "step": 13157 }, { "epoch": 84.81927710843374, "grad_norm": 19.001672744750977, "learning_rate": 7.5548387096774195e-06, "loss": 0.2816, "step": 13158 }, { "epoch": 84.82570281124498, "grad_norm": 5.46907377243042, "learning_rate": 7.551612903225806e-06, "loss": 0.1214, "step": 13159 }, { "epoch": 84.83212851405622, "grad_norm": 15.59543514251709, "learning_rate": 7.548387096774194e-06, "loss": 0.185, "step": 13160 }, { "epoch": 84.83855421686746, "grad_norm": 18.166255950927734, "learning_rate": 7.545161290322581e-06, "loss": 0.5543, "step": 13161 }, { "epoch": 84.84497991967872, "grad_norm": 17.208524703979492, "learning_rate": 7.541935483870968e-06, "loss": 0.4524, "step": 13162 }, { "epoch": 84.85140562248996, "grad_norm": 7.3450164794921875, "learning_rate": 7.538709677419355e-06, "loss": 0.3296, "step": 13163 }, { "epoch": 84.8578313253012, "grad_norm": 8.123703956604004, "learning_rate": 7.535483870967742e-06, "loss": 0.4273, "step": 13164 }, { "epoch": 84.86425702811245, "grad_norm": 3.6593852043151855, "learning_rate": 7.532258064516129e-06, "loss": 0.2847, "step": 13165 }, { "epoch": 84.87068273092369, "grad_norm": 11.030529975891113, "learning_rate": 7.529032258064517e-06, "loss": 0.306, "step": 13166 }, { "epoch": 84.87710843373495, "grad_norm": 10.080367088317871, "learning_rate": 7.5258064516129035e-06, "loss": 0.1852, "step": 13167 }, { "epoch": 84.88353413654619, "grad_norm": 5.387946128845215, "learning_rate": 7.522580645161291e-06, "loss": 0.2292, "step": 13168 }, { "epoch": 84.88995983935743, "grad_norm": 6.399781227111816, "learning_rate": 7.5193548387096775e-06, "loss": 0.1547, "step": 13169 }, { "epoch": 84.89638554216867, "grad_norm": 5.467319011688232, "learning_rate": 7.516129032258065e-06, "loss": 0.2805, "step": 13170 }, { "epoch": 84.90281124497992, "grad_norm": 7.451640605926514, "learning_rate": 7.5129032258064514e-06, "loss": 0.3258, "step": 13171 }, { "epoch": 84.90923694779117, "grad_norm": 16.25223731994629, "learning_rate": 7.50967741935484e-06, "loss": 0.4849, "step": 13172 }, { "epoch": 84.91566265060241, "grad_norm": 4.618311405181885, "learning_rate": 7.506451612903226e-06, "loss": 0.2994, "step": 13173 }, { "epoch": 84.92208835341366, "grad_norm": 7.236945152282715, "learning_rate": 7.503225806451614e-06, "loss": 0.2967, "step": 13174 }, { "epoch": 84.9285140562249, "grad_norm": 8.428701400756836, "learning_rate": 7.5e-06, "loss": 0.4187, "step": 13175 }, { "epoch": 84.93493975903614, "grad_norm": 4.461762428283691, "learning_rate": 7.496774193548388e-06, "loss": 0.1338, "step": 13176 }, { "epoch": 84.94136546184738, "grad_norm": 14.638962745666504, "learning_rate": 7.493548387096774e-06, "loss": 0.156, "step": 13177 }, { "epoch": 84.94779116465864, "grad_norm": 7.052945137023926, "learning_rate": 7.490322580645162e-06, "loss": 0.2124, "step": 13178 }, { "epoch": 84.95421686746988, "grad_norm": 17.68760871887207, "learning_rate": 7.487096774193549e-06, "loss": 0.398, "step": 13179 }, { "epoch": 84.96064257028112, "grad_norm": 13.288509368896484, "learning_rate": 7.483870967741936e-06, "loss": 0.3644, "step": 13180 }, { "epoch": 84.96706827309237, "grad_norm": 15.638361930847168, "learning_rate": 7.480645161290323e-06, "loss": 0.474, "step": 13181 }, { "epoch": 84.97349397590361, "grad_norm": 13.978121757507324, "learning_rate": 7.47741935483871e-06, "loss": 0.4375, "step": 13182 }, { "epoch": 84.97991967871486, "grad_norm": 11.635990142822266, "learning_rate": 7.474193548387097e-06, "loss": 0.2129, "step": 13183 }, { "epoch": 84.9863453815261, "grad_norm": 9.571966171264648, "learning_rate": 7.470967741935485e-06, "loss": 0.2049, "step": 13184 }, { "epoch": 84.99277108433735, "grad_norm": 12.464406967163086, "learning_rate": 7.467741935483872e-06, "loss": 0.4634, "step": 13185 }, { "epoch": 85.00240963855421, "grad_norm": 15.207645416259766, "learning_rate": 7.464516129032259e-06, "loss": 0.1172, "step": 13186 }, { "epoch": 85.00883534136547, "grad_norm": 8.504337310791016, "learning_rate": 7.461290322580646e-06, "loss": 0.2736, "step": 13187 }, { "epoch": 85.01526104417671, "grad_norm": 8.245155334472656, "learning_rate": 7.458064516129033e-06, "loss": 0.256, "step": 13188 }, { "epoch": 85.02168674698795, "grad_norm": 18.836355209350586, "learning_rate": 7.4548387096774196e-06, "loss": 0.4919, "step": 13189 }, { "epoch": 85.0281124497992, "grad_norm": 8.312987327575684, "learning_rate": 7.451612903225806e-06, "loss": 0.2717, "step": 13190 }, { "epoch": 85.03453815261044, "grad_norm": 11.160734176635742, "learning_rate": 7.4483870967741935e-06, "loss": 0.1328, "step": 13191 }, { "epoch": 85.04096385542168, "grad_norm": 10.429062843322754, "learning_rate": 7.44516129032258e-06, "loss": 0.4401, "step": 13192 }, { "epoch": 85.04738955823294, "grad_norm": 21.730575561523438, "learning_rate": 7.441935483870968e-06, "loss": 0.5824, "step": 13193 }, { "epoch": 85.05381526104418, "grad_norm": 8.325961112976074, "learning_rate": 7.438709677419355e-06, "loss": 0.3008, "step": 13194 }, { "epoch": 85.06024096385542, "grad_norm": 7.539371967315674, "learning_rate": 7.435483870967742e-06, "loss": 0.3202, "step": 13195 }, { "epoch": 85.06666666666666, "grad_norm": 10.804649353027344, "learning_rate": 7.432258064516129e-06, "loss": 0.4285, "step": 13196 }, { "epoch": 85.0730923694779, "grad_norm": 7.570411682128906, "learning_rate": 7.429032258064516e-06, "loss": 0.3082, "step": 13197 }, { "epoch": 85.07951807228916, "grad_norm": 6.331821441650391, "learning_rate": 7.425806451612903e-06, "loss": 0.2114, "step": 13198 }, { "epoch": 85.0859437751004, "grad_norm": 5.767910003662109, "learning_rate": 7.422580645161291e-06, "loss": 0.3253, "step": 13199 }, { "epoch": 85.09236947791165, "grad_norm": 7.645781517028809, "learning_rate": 7.419354838709678e-06, "loss": 0.3598, "step": 13200 }, { "epoch": 85.09879518072289, "grad_norm": 10.629691123962402, "learning_rate": 7.416129032258065e-06, "loss": 0.3838, "step": 13201 }, { "epoch": 85.10522088353413, "grad_norm": 9.366730690002441, "learning_rate": 7.4129032258064515e-06, "loss": 0.3297, "step": 13202 }, { "epoch": 85.11164658634539, "grad_norm": 24.383024215698242, "learning_rate": 7.409677419354839e-06, "loss": 0.4762, "step": 13203 }, { "epoch": 85.11807228915663, "grad_norm": 6.619217395782471, "learning_rate": 7.4064516129032255e-06, "loss": 0.2333, "step": 13204 }, { "epoch": 85.12449799196787, "grad_norm": 8.382278442382812, "learning_rate": 7.403225806451614e-06, "loss": 0.5424, "step": 13205 }, { "epoch": 85.13092369477911, "grad_norm": 25.500638961791992, "learning_rate": 7.4e-06, "loss": 0.3164, "step": 13206 }, { "epoch": 85.13734939759036, "grad_norm": 7.587707042694092, "learning_rate": 7.396774193548388e-06, "loss": 0.2495, "step": 13207 }, { "epoch": 85.14377510040161, "grad_norm": 28.730562210083008, "learning_rate": 7.393548387096774e-06, "loss": 0.524, "step": 13208 }, { "epoch": 85.15020080321285, "grad_norm": 13.020840644836426, "learning_rate": 7.390322580645162e-06, "loss": 0.5312, "step": 13209 }, { "epoch": 85.1566265060241, "grad_norm": 13.158591270446777, "learning_rate": 7.387096774193548e-06, "loss": 0.3768, "step": 13210 }, { "epoch": 85.16305220883534, "grad_norm": 12.28628921508789, "learning_rate": 7.3838709677419365e-06, "loss": 0.2784, "step": 13211 }, { "epoch": 85.16947791164658, "grad_norm": 6.922232151031494, "learning_rate": 7.380645161290323e-06, "loss": 0.2252, "step": 13212 }, { "epoch": 85.17590361445784, "grad_norm": 12.250992774963379, "learning_rate": 7.37741935483871e-06, "loss": 0.3473, "step": 13213 }, { "epoch": 85.18232931726908, "grad_norm": 14.091384887695312, "learning_rate": 7.374193548387097e-06, "loss": 0.5431, "step": 13214 }, { "epoch": 85.18875502008032, "grad_norm": 14.847837448120117, "learning_rate": 7.370967741935484e-06, "loss": 0.3369, "step": 13215 }, { "epoch": 85.19518072289156, "grad_norm": 6.168142318725586, "learning_rate": 7.367741935483871e-06, "loss": 0.5451, "step": 13216 }, { "epoch": 85.2016064257028, "grad_norm": 24.88836097717285, "learning_rate": 7.364516129032259e-06, "loss": 0.5486, "step": 13217 }, { "epoch": 85.20803212851406, "grad_norm": 7.0736918449401855, "learning_rate": 7.361290322580646e-06, "loss": 0.2012, "step": 13218 }, { "epoch": 85.2144578313253, "grad_norm": 15.312277793884277, "learning_rate": 7.358064516129033e-06, "loss": 0.4655, "step": 13219 }, { "epoch": 85.22088353413655, "grad_norm": 22.55286407470703, "learning_rate": 7.35483870967742e-06, "loss": 0.3074, "step": 13220 }, { "epoch": 85.22730923694779, "grad_norm": 7.4463605880737305, "learning_rate": 7.351612903225807e-06, "loss": 0.3345, "step": 13221 }, { "epoch": 85.23373493975903, "grad_norm": 25.798423767089844, "learning_rate": 7.348387096774194e-06, "loss": 0.3829, "step": 13222 }, { "epoch": 85.24016064257029, "grad_norm": 6.844428539276123, "learning_rate": 7.345161290322582e-06, "loss": 0.2861, "step": 13223 }, { "epoch": 85.24658634538153, "grad_norm": 11.625588417053223, "learning_rate": 7.3419354838709684e-06, "loss": 0.3686, "step": 13224 }, { "epoch": 85.25301204819277, "grad_norm": 16.00362205505371, "learning_rate": 7.338709677419354e-06, "loss": 0.2794, "step": 13225 }, { "epoch": 85.25943775100401, "grad_norm": 19.562908172607422, "learning_rate": 7.335483870967742e-06, "loss": 0.5033, "step": 13226 }, { "epoch": 85.26586345381526, "grad_norm": 12.581562042236328, "learning_rate": 7.332258064516129e-06, "loss": 0.5027, "step": 13227 }, { "epoch": 85.2722891566265, "grad_norm": 18.140113830566406, "learning_rate": 7.329032258064516e-06, "loss": 0.3182, "step": 13228 }, { "epoch": 85.27871485943776, "grad_norm": 16.719493865966797, "learning_rate": 7.325806451612903e-06, "loss": 0.2805, "step": 13229 }, { "epoch": 85.285140562249, "grad_norm": 7.645970344543457, "learning_rate": 7.322580645161291e-06, "loss": 0.2044, "step": 13230 }, { "epoch": 85.29156626506024, "grad_norm": 12.548768043518066, "learning_rate": 7.319354838709677e-06, "loss": 0.5209, "step": 13231 }, { "epoch": 85.29799196787148, "grad_norm": 3.6853342056274414, "learning_rate": 7.316129032258065e-06, "loss": 0.2828, "step": 13232 }, { "epoch": 85.30441767068272, "grad_norm": 10.652060508728027, "learning_rate": 7.312903225806452e-06, "loss": 0.348, "step": 13233 }, { "epoch": 85.31084337349398, "grad_norm": 11.508254051208496, "learning_rate": 7.309677419354839e-06, "loss": 0.1792, "step": 13234 }, { "epoch": 85.31726907630522, "grad_norm": 19.933399200439453, "learning_rate": 7.306451612903226e-06, "loss": 0.6029, "step": 13235 }, { "epoch": 85.32369477911647, "grad_norm": 8.790851593017578, "learning_rate": 7.303225806451614e-06, "loss": 0.2181, "step": 13236 }, { "epoch": 85.33012048192771, "grad_norm": 28.997446060180664, "learning_rate": 7.2999999999999996e-06, "loss": 0.5495, "step": 13237 }, { "epoch": 85.33654618473895, "grad_norm": 22.576335906982422, "learning_rate": 7.296774193548388e-06, "loss": 0.2555, "step": 13238 }, { "epoch": 85.3429718875502, "grad_norm": 54.25862503051758, "learning_rate": 7.293548387096774e-06, "loss": 0.6233, "step": 13239 }, { "epoch": 85.34939759036145, "grad_norm": 7.432053565979004, "learning_rate": 7.290322580645162e-06, "loss": 0.4089, "step": 13240 }, { "epoch": 85.35582329317269, "grad_norm": 18.94207763671875, "learning_rate": 7.287096774193548e-06, "loss": 0.2841, "step": 13241 }, { "epoch": 85.36224899598393, "grad_norm": 38.87191390991211, "learning_rate": 7.2838709677419366e-06, "loss": 0.5436, "step": 13242 }, { "epoch": 85.36867469879518, "grad_norm": 12.53757381439209, "learning_rate": 7.280645161290322e-06, "loss": 0.6489, "step": 13243 }, { "epoch": 85.37510040160643, "grad_norm": 3.0300025939941406, "learning_rate": 7.2774193548387105e-06, "loss": 0.0672, "step": 13244 }, { "epoch": 85.38152610441767, "grad_norm": 8.67109489440918, "learning_rate": 7.274193548387097e-06, "loss": 0.2177, "step": 13245 }, { "epoch": 85.38795180722892, "grad_norm": 10.356417655944824, "learning_rate": 7.2709677419354845e-06, "loss": 0.4518, "step": 13246 }, { "epoch": 85.39437751004016, "grad_norm": 4.193113803863525, "learning_rate": 7.267741935483871e-06, "loss": 0.4107, "step": 13247 }, { "epoch": 85.4008032128514, "grad_norm": 9.561391830444336, "learning_rate": 7.264516129032259e-06, "loss": 0.3038, "step": 13248 }, { "epoch": 85.40722891566266, "grad_norm": 14.434520721435547, "learning_rate": 7.261290322580645e-06, "loss": 0.432, "step": 13249 }, { "epoch": 85.4136546184739, "grad_norm": 14.372011184692383, "learning_rate": 7.258064516129033e-06, "loss": 0.6046, "step": 13250 }, { "epoch": 85.42008032128514, "grad_norm": 11.741832733154297, "learning_rate": 7.25483870967742e-06, "loss": 0.2922, "step": 13251 }, { "epoch": 85.42650602409638, "grad_norm": 6.368126392364502, "learning_rate": 7.251612903225807e-06, "loss": 0.1878, "step": 13252 }, { "epoch": 85.43293172690763, "grad_norm": 17.53493881225586, "learning_rate": 7.248387096774194e-06, "loss": 0.5443, "step": 13253 }, { "epoch": 85.43935742971888, "grad_norm": 8.989279747009277, "learning_rate": 7.245161290322581e-06, "loss": 0.418, "step": 13254 }, { "epoch": 85.44578313253012, "grad_norm": 1.6947228908538818, "learning_rate": 7.241935483870968e-06, "loss": 0.1635, "step": 13255 }, { "epoch": 85.45220883534137, "grad_norm": 5.346325874328613, "learning_rate": 7.238709677419356e-06, "loss": 0.1255, "step": 13256 }, { "epoch": 85.45863453815261, "grad_norm": 16.09781837463379, "learning_rate": 7.2354838709677425e-06, "loss": 0.4849, "step": 13257 }, { "epoch": 85.46506024096385, "grad_norm": 4.2380452156066895, "learning_rate": 7.23225806451613e-06, "loss": 0.2047, "step": 13258 }, { "epoch": 85.47148594377511, "grad_norm": 5.009095668792725, "learning_rate": 7.2290322580645165e-06, "loss": 0.3564, "step": 13259 }, { "epoch": 85.47791164658635, "grad_norm": 21.68900489807129, "learning_rate": 7.225806451612904e-06, "loss": 0.3831, "step": 13260 }, { "epoch": 85.48433734939759, "grad_norm": 14.199094772338867, "learning_rate": 7.22258064516129e-06, "loss": 0.3555, "step": 13261 }, { "epoch": 85.49076305220883, "grad_norm": 2.5907647609710693, "learning_rate": 7.219354838709677e-06, "loss": 0.1414, "step": 13262 }, { "epoch": 85.49718875502008, "grad_norm": 12.253061294555664, "learning_rate": 7.216129032258065e-06, "loss": 0.2975, "step": 13263 }, { "epoch": 85.50361445783132, "grad_norm": 13.241004943847656, "learning_rate": 7.212903225806451e-06, "loss": 0.2423, "step": 13264 }, { "epoch": 85.51004016064257, "grad_norm": 14.058304786682129, "learning_rate": 7.209677419354839e-06, "loss": 0.3172, "step": 13265 }, { "epoch": 85.51646586345382, "grad_norm": 6.3158650398254395, "learning_rate": 7.206451612903226e-06, "loss": 0.1826, "step": 13266 }, { "epoch": 85.52289156626506, "grad_norm": 6.286923885345459, "learning_rate": 7.203225806451613e-06, "loss": 0.1948, "step": 13267 }, { "epoch": 85.5293172690763, "grad_norm": 11.573545455932617, "learning_rate": 7.2e-06, "loss": 0.3032, "step": 13268 }, { "epoch": 85.53574297188754, "grad_norm": 9.964654922485352, "learning_rate": 7.196774193548388e-06, "loss": 0.5311, "step": 13269 }, { "epoch": 85.5421686746988, "grad_norm": 36.12389373779297, "learning_rate": 7.193548387096774e-06, "loss": 0.4654, "step": 13270 }, { "epoch": 85.54859437751004, "grad_norm": 8.262939453125, "learning_rate": 7.190322580645162e-06, "loss": 0.5535, "step": 13271 }, { "epoch": 85.55502008032128, "grad_norm": 6.562072277069092, "learning_rate": 7.1870967741935484e-06, "loss": 0.3518, "step": 13272 }, { "epoch": 85.56144578313253, "grad_norm": 12.88209056854248, "learning_rate": 7.183870967741936e-06, "loss": 0.5224, "step": 13273 }, { "epoch": 85.56787148594377, "grad_norm": 16.707223892211914, "learning_rate": 7.180645161290322e-06, "loss": 0.4389, "step": 13274 }, { "epoch": 85.57429718875503, "grad_norm": 7.364062309265137, "learning_rate": 7.177419354838711e-06, "loss": 0.3988, "step": 13275 }, { "epoch": 85.58072289156627, "grad_norm": 16.493345260620117, "learning_rate": 7.174193548387096e-06, "loss": 0.2738, "step": 13276 }, { "epoch": 85.58714859437751, "grad_norm": 31.838163375854492, "learning_rate": 7.170967741935485e-06, "loss": 0.3895, "step": 13277 }, { "epoch": 85.59357429718875, "grad_norm": 20.369461059570312, "learning_rate": 7.167741935483871e-06, "loss": 0.4196, "step": 13278 }, { "epoch": 85.6, "grad_norm": 24.847816467285156, "learning_rate": 7.1645161290322585e-06, "loss": 0.2315, "step": 13279 }, { "epoch": 85.60642570281125, "grad_norm": 7.011425495147705, "learning_rate": 7.161290322580645e-06, "loss": 0.2012, "step": 13280 }, { "epoch": 85.6128514056225, "grad_norm": 7.010676860809326, "learning_rate": 7.158064516129033e-06, "loss": 0.1635, "step": 13281 }, { "epoch": 85.61927710843374, "grad_norm": 16.872461318969727, "learning_rate": 7.154838709677419e-06, "loss": 0.2524, "step": 13282 }, { "epoch": 85.62570281124498, "grad_norm": 14.016724586486816, "learning_rate": 7.151612903225807e-06, "loss": 0.2983, "step": 13283 }, { "epoch": 85.63212851405622, "grad_norm": 1.3213611841201782, "learning_rate": 7.148387096774194e-06, "loss": 0.2512, "step": 13284 }, { "epoch": 85.63855421686748, "grad_norm": 20.912185668945312, "learning_rate": 7.145161290322581e-06, "loss": 0.5252, "step": 13285 }, { "epoch": 85.64497991967872, "grad_norm": 13.457905769348145, "learning_rate": 7.141935483870968e-06, "loss": 0.484, "step": 13286 }, { "epoch": 85.65140562248996, "grad_norm": 24.40369987487793, "learning_rate": 7.138709677419356e-06, "loss": 0.4076, "step": 13287 }, { "epoch": 85.6578313253012, "grad_norm": 10.023138046264648, "learning_rate": 7.135483870967742e-06, "loss": 0.1518, "step": 13288 }, { "epoch": 85.66425702811244, "grad_norm": 4.681736946105957, "learning_rate": 7.13225806451613e-06, "loss": 0.568, "step": 13289 }, { "epoch": 85.6706827309237, "grad_norm": 25.603099822998047, "learning_rate": 7.1290322580645166e-06, "loss": 0.5188, "step": 13290 }, { "epoch": 85.67710843373494, "grad_norm": 6.579116344451904, "learning_rate": 7.125806451612904e-06, "loss": 0.4244, "step": 13291 }, { "epoch": 85.68353413654619, "grad_norm": 25.795732498168945, "learning_rate": 7.1225806451612905e-06, "loss": 0.4541, "step": 13292 }, { "epoch": 85.68995983935743, "grad_norm": 9.724305152893066, "learning_rate": 7.119354838709679e-06, "loss": 0.2618, "step": 13293 }, { "epoch": 85.69638554216867, "grad_norm": 6.680020332336426, "learning_rate": 7.1161290322580645e-06, "loss": 0.1614, "step": 13294 }, { "epoch": 85.70281124497993, "grad_norm": 8.340590476989746, "learning_rate": 7.112903225806453e-06, "loss": 0.425, "step": 13295 }, { "epoch": 85.70923694779117, "grad_norm": 19.551109313964844, "learning_rate": 7.109677419354839e-06, "loss": 0.5045, "step": 13296 }, { "epoch": 85.71566265060241, "grad_norm": 7.061267375946045, "learning_rate": 7.106451612903226e-06, "loss": 0.2302, "step": 13297 }, { "epoch": 85.72208835341365, "grad_norm": 4.905320644378662, "learning_rate": 7.103225806451613e-06, "loss": 0.3078, "step": 13298 }, { "epoch": 85.7285140562249, "grad_norm": 4.816975116729736, "learning_rate": 7.1e-06, "loss": 0.1443, "step": 13299 }, { "epoch": 85.73493975903614, "grad_norm": 8.891368865966797, "learning_rate": 7.096774193548387e-06, "loss": 0.2051, "step": 13300 }, { "epoch": 85.7413654618474, "grad_norm": 12.058900833129883, "learning_rate": 7.093548387096774e-06, "loss": 0.4609, "step": 13301 }, { "epoch": 85.74779116465864, "grad_norm": 7.74777889251709, "learning_rate": 7.090322580645162e-06, "loss": 0.2507, "step": 13302 }, { "epoch": 85.75421686746988, "grad_norm": 18.593843460083008, "learning_rate": 7.0870967741935485e-06, "loss": 0.5349, "step": 13303 }, { "epoch": 85.76064257028112, "grad_norm": 6.806259632110596, "learning_rate": 7.083870967741936e-06, "loss": 0.17, "step": 13304 }, { "epoch": 85.76706827309236, "grad_norm": 7.2449631690979, "learning_rate": 7.0806451612903225e-06, "loss": 0.2505, "step": 13305 }, { "epoch": 85.77349397590362, "grad_norm": 9.514148712158203, "learning_rate": 7.07741935483871e-06, "loss": 0.1636, "step": 13306 }, { "epoch": 85.77991967871486, "grad_norm": 38.262020111083984, "learning_rate": 7.0741935483870965e-06, "loss": 0.3755, "step": 13307 }, { "epoch": 85.7863453815261, "grad_norm": 18.07461166381836, "learning_rate": 7.070967741935485e-06, "loss": 0.5295, "step": 13308 }, { "epoch": 85.79277108433735, "grad_norm": 9.05273151397705, "learning_rate": 7.067741935483871e-06, "loss": 0.3177, "step": 13309 }, { "epoch": 85.79919678714859, "grad_norm": 8.912672996520996, "learning_rate": 7.064516129032259e-06, "loss": 0.2867, "step": 13310 }, { "epoch": 85.80562248995984, "grad_norm": 9.65764331817627, "learning_rate": 7.061290322580645e-06, "loss": 0.4796, "step": 13311 }, { "epoch": 85.81204819277109, "grad_norm": 6.218766689300537, "learning_rate": 7.058064516129033e-06, "loss": 0.3377, "step": 13312 }, { "epoch": 85.81847389558233, "grad_norm": 15.195438385009766, "learning_rate": 7.054838709677419e-06, "loss": 0.276, "step": 13313 }, { "epoch": 85.82489959839357, "grad_norm": 9.641221046447754, "learning_rate": 7.051612903225807e-06, "loss": 0.3906, "step": 13314 }, { "epoch": 85.83132530120481, "grad_norm": 12.545904159545898, "learning_rate": 7.048387096774194e-06, "loss": 0.3063, "step": 13315 }, { "epoch": 85.83775100401607, "grad_norm": 16.076229095458984, "learning_rate": 7.045161290322581e-06, "loss": 0.3147, "step": 13316 }, { "epoch": 85.84417670682731, "grad_norm": 11.364653587341309, "learning_rate": 7.041935483870968e-06, "loss": 0.4574, "step": 13317 }, { "epoch": 85.85060240963855, "grad_norm": 7.422155857086182, "learning_rate": 7.038709677419355e-06, "loss": 0.2984, "step": 13318 }, { "epoch": 85.8570281124498, "grad_norm": 22.94482421875, "learning_rate": 7.035483870967742e-06, "loss": 0.3791, "step": 13319 }, { "epoch": 85.86345381526104, "grad_norm": 21.876466751098633, "learning_rate": 7.03225806451613e-06, "loss": 0.7154, "step": 13320 }, { "epoch": 85.8698795180723, "grad_norm": 7.473025798797607, "learning_rate": 7.029032258064517e-06, "loss": 0.3931, "step": 13321 }, { "epoch": 85.87630522088354, "grad_norm": 10.850680351257324, "learning_rate": 7.025806451612904e-06, "loss": 0.2225, "step": 13322 }, { "epoch": 85.88273092369478, "grad_norm": 3.6538891792297363, "learning_rate": 7.022580645161291e-06, "loss": 0.2129, "step": 13323 }, { "epoch": 85.88915662650602, "grad_norm": 6.18935489654541, "learning_rate": 7.019354838709678e-06, "loss": 0.3948, "step": 13324 }, { "epoch": 85.89558232931726, "grad_norm": 10.302306175231934, "learning_rate": 7.016129032258065e-06, "loss": 0.5974, "step": 13325 }, { "epoch": 85.90200803212852, "grad_norm": 5.242063522338867, "learning_rate": 7.012903225806453e-06, "loss": 0.4024, "step": 13326 }, { "epoch": 85.90843373493976, "grad_norm": 8.535197257995605, "learning_rate": 7.0096774193548385e-06, "loss": 0.1314, "step": 13327 }, { "epoch": 85.914859437751, "grad_norm": 2.7177836894989014, "learning_rate": 7.006451612903227e-06, "loss": 0.1327, "step": 13328 }, { "epoch": 85.92128514056225, "grad_norm": 4.37748908996582, "learning_rate": 7.003225806451613e-06, "loss": 0.1654, "step": 13329 }, { "epoch": 85.92771084337349, "grad_norm": 9.8447265625, "learning_rate": 7.000000000000001e-06, "loss": 0.2989, "step": 13330 }, { "epoch": 85.93413654618475, "grad_norm": 33.359127044677734, "learning_rate": 6.996774193548387e-06, "loss": 0.299, "step": 13331 }, { "epoch": 85.94056224899599, "grad_norm": 10.780778884887695, "learning_rate": 6.993548387096774e-06, "loss": 0.1919, "step": 13332 }, { "epoch": 85.94698795180723, "grad_norm": 2.411113739013672, "learning_rate": 6.990322580645161e-06, "loss": 0.0706, "step": 13333 }, { "epoch": 85.95341365461847, "grad_norm": 28.667673110961914, "learning_rate": 6.987096774193548e-06, "loss": 0.1771, "step": 13334 }, { "epoch": 85.95983935742971, "grad_norm": 21.54433822631836, "learning_rate": 6.983870967741936e-06, "loss": 0.534, "step": 13335 }, { "epoch": 85.96626506024096, "grad_norm": 23.147098541259766, "learning_rate": 6.980645161290323e-06, "loss": 0.5853, "step": 13336 }, { "epoch": 85.97269076305221, "grad_norm": 7.287591934204102, "learning_rate": 6.97741935483871e-06, "loss": 0.2369, "step": 13337 }, { "epoch": 85.97911646586346, "grad_norm": 40.923858642578125, "learning_rate": 6.9741935483870966e-06, "loss": 0.4542, "step": 13338 }, { "epoch": 85.9855421686747, "grad_norm": 22.063167572021484, "learning_rate": 6.970967741935484e-06, "loss": 0.1983, "step": 13339 }, { "epoch": 85.99196787148594, "grad_norm": 17.346492767333984, "learning_rate": 6.9677419354838705e-06, "loss": 0.4261, "step": 13340 }, { "epoch": 86.00160642570282, "grad_norm": 14.099937438964844, "learning_rate": 6.964516129032259e-06, "loss": 0.3799, "step": 13341 }, { "epoch": 86.00803212851406, "grad_norm": 13.35822582244873, "learning_rate": 6.961290322580645e-06, "loss": 0.3524, "step": 13342 }, { "epoch": 86.0144578313253, "grad_norm": 5.811735153198242, "learning_rate": 6.958064516129033e-06, "loss": 0.1708, "step": 13343 }, { "epoch": 86.02088353413654, "grad_norm": 9.635345458984375, "learning_rate": 6.954838709677419e-06, "loss": 0.4394, "step": 13344 }, { "epoch": 86.02730923694779, "grad_norm": 7.680385589599609, "learning_rate": 6.951612903225807e-06, "loss": 0.3954, "step": 13345 }, { "epoch": 86.03373493975904, "grad_norm": 14.969892501831055, "learning_rate": 6.948387096774193e-06, "loss": 0.2099, "step": 13346 }, { "epoch": 86.04016064257029, "grad_norm": 28.568164825439453, "learning_rate": 6.9451612903225815e-06, "loss": 0.3718, "step": 13347 }, { "epoch": 86.04658634538153, "grad_norm": 12.610528945922852, "learning_rate": 6.941935483870968e-06, "loss": 0.4511, "step": 13348 }, { "epoch": 86.05301204819277, "grad_norm": 11.278131484985352, "learning_rate": 6.9387096774193554e-06, "loss": 0.1614, "step": 13349 }, { "epoch": 86.05943775100401, "grad_norm": 13.822565078735352, "learning_rate": 6.935483870967742e-06, "loss": 0.2619, "step": 13350 }, { "epoch": 86.06586345381525, "grad_norm": 5.4925408363342285, "learning_rate": 6.932258064516129e-06, "loss": 0.3779, "step": 13351 }, { "epoch": 86.07228915662651, "grad_norm": 10.440502166748047, "learning_rate": 6.929032258064516e-06, "loss": 0.2375, "step": 13352 }, { "epoch": 86.07871485943775, "grad_norm": 11.656556129455566, "learning_rate": 6.925806451612904e-06, "loss": 0.3242, "step": 13353 }, { "epoch": 86.085140562249, "grad_norm": 6.123196601867676, "learning_rate": 6.922580645161291e-06, "loss": 0.4484, "step": 13354 }, { "epoch": 86.09156626506024, "grad_norm": 4.676977157592773, "learning_rate": 6.919354838709678e-06, "loss": 0.2017, "step": 13355 }, { "epoch": 86.09799196787148, "grad_norm": 8.042160034179688, "learning_rate": 6.916129032258065e-06, "loss": 0.4511, "step": 13356 }, { "epoch": 86.10441767068274, "grad_norm": 10.040340423583984, "learning_rate": 6.912903225806452e-06, "loss": 0.0928, "step": 13357 }, { "epoch": 86.11084337349398, "grad_norm": 6.149263381958008, "learning_rate": 6.909677419354839e-06, "loss": 0.1964, "step": 13358 }, { "epoch": 86.11726907630522, "grad_norm": 11.805377006530762, "learning_rate": 6.906451612903227e-06, "loss": 0.3144, "step": 13359 }, { "epoch": 86.12369477911646, "grad_norm": 15.919102668762207, "learning_rate": 6.9032258064516135e-06, "loss": 0.5022, "step": 13360 }, { "epoch": 86.1301204819277, "grad_norm": 5.036407947540283, "learning_rate": 6.900000000000001e-06, "loss": 0.2187, "step": 13361 }, { "epoch": 86.13654618473896, "grad_norm": 10.017251014709473, "learning_rate": 6.896774193548387e-06, "loss": 0.1637, "step": 13362 }, { "epoch": 86.1429718875502, "grad_norm": 6.914827823638916, "learning_rate": 6.893548387096775e-06, "loss": 0.2775, "step": 13363 }, { "epoch": 86.14939759036145, "grad_norm": 10.949138641357422, "learning_rate": 6.890322580645161e-06, "loss": 0.5141, "step": 13364 }, { "epoch": 86.15582329317269, "grad_norm": 6.951726913452148, "learning_rate": 6.88709677419355e-06, "loss": 0.5723, "step": 13365 }, { "epoch": 86.16224899598393, "grad_norm": 9.50473403930664, "learning_rate": 6.883870967741936e-06, "loss": 0.2482, "step": 13366 }, { "epoch": 86.16867469879519, "grad_norm": 2.6377487182617188, "learning_rate": 6.880645161290322e-06, "loss": 0.0491, "step": 13367 }, { "epoch": 86.17510040160643, "grad_norm": 11.531567573547363, "learning_rate": 6.87741935483871e-06, "loss": 0.4179, "step": 13368 }, { "epoch": 86.18152610441767, "grad_norm": 7.991331577301025, "learning_rate": 6.874193548387097e-06, "loss": 0.3376, "step": 13369 }, { "epoch": 86.18795180722891, "grad_norm": 17.28472328186035, "learning_rate": 6.870967741935484e-06, "loss": 0.4162, "step": 13370 }, { "epoch": 86.19437751004016, "grad_norm": 6.411590099334717, "learning_rate": 6.867741935483871e-06, "loss": 0.1662, "step": 13371 }, { "epoch": 86.20080321285141, "grad_norm": 5.148918628692627, "learning_rate": 6.864516129032259e-06, "loss": 0.0884, "step": 13372 }, { "epoch": 86.20722891566265, "grad_norm": 5.605640888214111, "learning_rate": 6.861290322580645e-06, "loss": 0.3451, "step": 13373 }, { "epoch": 86.2136546184739, "grad_norm": 5.613945007324219, "learning_rate": 6.858064516129033e-06, "loss": 0.2703, "step": 13374 }, { "epoch": 86.22008032128514, "grad_norm": 2.4486165046691895, "learning_rate": 6.854838709677419e-06, "loss": 0.1914, "step": 13375 }, { "epoch": 86.22650602409638, "grad_norm": 12.15666389465332, "learning_rate": 6.851612903225807e-06, "loss": 0.2147, "step": 13376 }, { "epoch": 86.23293172690764, "grad_norm": 15.663439750671387, "learning_rate": 6.848387096774193e-06, "loss": 0.4471, "step": 13377 }, { "epoch": 86.23935742971888, "grad_norm": 13.427571296691895, "learning_rate": 6.845161290322582e-06, "loss": 0.5552, "step": 13378 }, { "epoch": 86.24578313253012, "grad_norm": 7.904725551605225, "learning_rate": 6.841935483870967e-06, "loss": 0.3024, "step": 13379 }, { "epoch": 86.25220883534136, "grad_norm": 5.348621368408203, "learning_rate": 6.8387096774193555e-06, "loss": 0.2527, "step": 13380 }, { "epoch": 86.2586345381526, "grad_norm": 6.682897090911865, "learning_rate": 6.835483870967742e-06, "loss": 0.229, "step": 13381 }, { "epoch": 86.26506024096386, "grad_norm": 8.856558799743652, "learning_rate": 6.8322580645161295e-06, "loss": 0.3892, "step": 13382 }, { "epoch": 86.2714859437751, "grad_norm": 13.862598419189453, "learning_rate": 6.829032258064516e-06, "loss": 0.2606, "step": 13383 }, { "epoch": 86.27791164658635, "grad_norm": 16.01692771911621, "learning_rate": 6.825806451612904e-06, "loss": 0.4545, "step": 13384 }, { "epoch": 86.28433734939759, "grad_norm": 10.953054428100586, "learning_rate": 6.82258064516129e-06, "loss": 0.4604, "step": 13385 }, { "epoch": 86.29076305220883, "grad_norm": 3.484339714050293, "learning_rate": 6.819354838709678e-06, "loss": 0.1503, "step": 13386 }, { "epoch": 86.29718875502007, "grad_norm": 6.4513020515441895, "learning_rate": 6.816129032258065e-06, "loss": 0.1087, "step": 13387 }, { "epoch": 86.30361445783133, "grad_norm": 14.606451988220215, "learning_rate": 6.812903225806452e-06, "loss": 0.4179, "step": 13388 }, { "epoch": 86.31004016064257, "grad_norm": 4.554305076599121, "learning_rate": 6.809677419354839e-06, "loss": 0.1725, "step": 13389 }, { "epoch": 86.31646586345381, "grad_norm": 11.833270072937012, "learning_rate": 6.806451612903226e-06, "loss": 0.4057, "step": 13390 }, { "epoch": 86.32289156626506, "grad_norm": 9.463199615478516, "learning_rate": 6.803225806451613e-06, "loss": 0.2433, "step": 13391 }, { "epoch": 86.3293172690763, "grad_norm": 13.986023902893066, "learning_rate": 6.800000000000001e-06, "loss": 0.3416, "step": 13392 }, { "epoch": 86.33574297188756, "grad_norm": 9.2688627243042, "learning_rate": 6.7967741935483875e-06, "loss": 0.2841, "step": 13393 }, { "epoch": 86.3421686746988, "grad_norm": 11.550835609436035, "learning_rate": 6.793548387096775e-06, "loss": 0.3332, "step": 13394 }, { "epoch": 86.34859437751004, "grad_norm": 10.844584465026855, "learning_rate": 6.7903225806451615e-06, "loss": 0.2324, "step": 13395 }, { "epoch": 86.35502008032128, "grad_norm": 5.114777565002441, "learning_rate": 6.787096774193549e-06, "loss": 0.1571, "step": 13396 }, { "epoch": 86.36144578313252, "grad_norm": 14.777974128723145, "learning_rate": 6.7838709677419354e-06, "loss": 0.1309, "step": 13397 }, { "epoch": 86.36787148594378, "grad_norm": 10.600785255432129, "learning_rate": 6.780645161290324e-06, "loss": 0.45, "step": 13398 }, { "epoch": 86.37429718875502, "grad_norm": 6.608964443206787, "learning_rate": 6.77741935483871e-06, "loss": 0.264, "step": 13399 }, { "epoch": 86.38072289156626, "grad_norm": 11.11927604675293, "learning_rate": 6.774193548387098e-06, "loss": 0.3495, "step": 13400 }, { "epoch": 86.3871485943775, "grad_norm": 6.740360736846924, "learning_rate": 6.770967741935484e-06, "loss": 0.3696, "step": 13401 }, { "epoch": 86.39357429718875, "grad_norm": 6.917623996734619, "learning_rate": 6.767741935483872e-06, "loss": 0.2108, "step": 13402 }, { "epoch": 86.4, "grad_norm": 10.733601570129395, "learning_rate": 6.764516129032258e-06, "loss": 0.1097, "step": 13403 }, { "epoch": 86.40642570281125, "grad_norm": 14.287664413452148, "learning_rate": 6.761290322580645e-06, "loss": 0.2226, "step": 13404 }, { "epoch": 86.41285140562249, "grad_norm": 18.69864273071289, "learning_rate": 6.758064516129033e-06, "loss": 0.8226, "step": 13405 }, { "epoch": 86.41927710843373, "grad_norm": 34.38224411010742, "learning_rate": 6.754838709677419e-06, "loss": 0.4459, "step": 13406 }, { "epoch": 86.42570281124497, "grad_norm": 10.51536750793457, "learning_rate": 6.751612903225807e-06, "loss": 0.523, "step": 13407 }, { "epoch": 86.43212851405623, "grad_norm": 11.59097957611084, "learning_rate": 6.7483870967741934e-06, "loss": 0.3286, "step": 13408 }, { "epoch": 86.43855421686747, "grad_norm": 5.7214035987854, "learning_rate": 6.745161290322581e-06, "loss": 0.2418, "step": 13409 }, { "epoch": 86.44497991967872, "grad_norm": 7.677597999572754, "learning_rate": 6.741935483870967e-06, "loss": 0.4065, "step": 13410 }, { "epoch": 86.45140562248996, "grad_norm": 12.36738395690918, "learning_rate": 6.738709677419356e-06, "loss": 0.3717, "step": 13411 }, { "epoch": 86.4578313253012, "grad_norm": 12.729636192321777, "learning_rate": 6.735483870967741e-06, "loss": 0.4982, "step": 13412 }, { "epoch": 86.46425702811246, "grad_norm": 11.584343910217285, "learning_rate": 6.73225806451613e-06, "loss": 0.1846, "step": 13413 }, { "epoch": 86.4706827309237, "grad_norm": 6.3899149894714355, "learning_rate": 6.729032258064516e-06, "loss": 0.2686, "step": 13414 }, { "epoch": 86.47710843373494, "grad_norm": 12.449111938476562, "learning_rate": 6.7258064516129036e-06, "loss": 0.4585, "step": 13415 }, { "epoch": 86.48353413654618, "grad_norm": 4.125480651855469, "learning_rate": 6.72258064516129e-06, "loss": 0.1057, "step": 13416 }, { "epoch": 86.48995983935743, "grad_norm": 5.362991809844971, "learning_rate": 6.719354838709678e-06, "loss": 0.2623, "step": 13417 }, { "epoch": 86.49638554216868, "grad_norm": 13.01168155670166, "learning_rate": 6.716129032258064e-06, "loss": 0.2423, "step": 13418 }, { "epoch": 86.50281124497992, "grad_norm": 8.951213836669922, "learning_rate": 6.712903225806452e-06, "loss": 0.22, "step": 13419 }, { "epoch": 86.50923694779117, "grad_norm": 10.283637046813965, "learning_rate": 6.709677419354839e-06, "loss": 0.5015, "step": 13420 }, { "epoch": 86.51566265060241, "grad_norm": 8.681588172912598, "learning_rate": 6.706451612903226e-06, "loss": 0.4515, "step": 13421 }, { "epoch": 86.52208835341365, "grad_norm": 7.768976211547852, "learning_rate": 6.703225806451613e-06, "loss": 0.3535, "step": 13422 }, { "epoch": 86.52851405622489, "grad_norm": 40.492340087890625, "learning_rate": 6.700000000000001e-06, "loss": 0.5381, "step": 13423 }, { "epoch": 86.53493975903615, "grad_norm": 8.18206787109375, "learning_rate": 6.696774193548387e-06, "loss": 0.2082, "step": 13424 }, { "epoch": 86.54136546184739, "grad_norm": 7.033740997314453, "learning_rate": 6.693548387096775e-06, "loss": 0.5775, "step": 13425 }, { "epoch": 86.54779116465863, "grad_norm": 11.954673767089844, "learning_rate": 6.690322580645162e-06, "loss": 0.2892, "step": 13426 }, { "epoch": 86.55421686746988, "grad_norm": 10.04866886138916, "learning_rate": 6.687096774193549e-06, "loss": 0.3587, "step": 13427 }, { "epoch": 86.56064257028112, "grad_norm": 9.057632446289062, "learning_rate": 6.6838709677419355e-06, "loss": 0.4318, "step": 13428 }, { "epoch": 86.56706827309237, "grad_norm": 5.972970485687256, "learning_rate": 6.680645161290324e-06, "loss": 0.1238, "step": 13429 }, { "epoch": 86.57349397590362, "grad_norm": 7.500702857971191, "learning_rate": 6.6774193548387095e-06, "loss": 0.1094, "step": 13430 }, { "epoch": 86.57991967871486, "grad_norm": 9.728418350219727, "learning_rate": 6.674193548387098e-06, "loss": 0.3091, "step": 13431 }, { "epoch": 86.5863453815261, "grad_norm": 7.668428897857666, "learning_rate": 6.670967741935484e-06, "loss": 0.3112, "step": 13432 }, { "epoch": 86.59277108433734, "grad_norm": 8.685026168823242, "learning_rate": 6.667741935483872e-06, "loss": 0.2568, "step": 13433 }, { "epoch": 86.5991967871486, "grad_norm": 5.665545463562012, "learning_rate": 6.664516129032258e-06, "loss": 0.3741, "step": 13434 }, { "epoch": 86.60562248995984, "grad_norm": 7.74843692779541, "learning_rate": 6.6612903225806465e-06, "loss": 0.4116, "step": 13435 }, { "epoch": 86.61204819277108, "grad_norm": 13.434791564941406, "learning_rate": 6.658064516129032e-06, "loss": 0.3407, "step": 13436 }, { "epoch": 86.61847389558233, "grad_norm": 8.095359802246094, "learning_rate": 6.6548387096774205e-06, "loss": 0.2629, "step": 13437 }, { "epoch": 86.62489959839357, "grad_norm": 17.9366512298584, "learning_rate": 6.651612903225807e-06, "loss": 0.4754, "step": 13438 }, { "epoch": 86.63132530120482, "grad_norm": 26.92389678955078, "learning_rate": 6.6483870967741936e-06, "loss": 0.3787, "step": 13439 }, { "epoch": 86.63775100401607, "grad_norm": 6.201456546783447, "learning_rate": 6.645161290322581e-06, "loss": 0.339, "step": 13440 }, { "epoch": 86.64417670682731, "grad_norm": 3.761131763458252, "learning_rate": 6.6419354838709675e-06, "loss": 0.093, "step": 13441 }, { "epoch": 86.65060240963855, "grad_norm": 8.023874282836914, "learning_rate": 6.638709677419355e-06, "loss": 0.3144, "step": 13442 }, { "epoch": 86.6570281124498, "grad_norm": 16.71082878112793, "learning_rate": 6.6354838709677415e-06, "loss": 0.5474, "step": 13443 }, { "epoch": 86.66345381526105, "grad_norm": 10.816415786743164, "learning_rate": 6.63225806451613e-06, "loss": 0.4285, "step": 13444 }, { "epoch": 86.66987951807229, "grad_norm": 22.495925903320312, "learning_rate": 6.629032258064516e-06, "loss": 0.3152, "step": 13445 }, { "epoch": 86.67630522088353, "grad_norm": 5.30946159362793, "learning_rate": 6.625806451612904e-06, "loss": 0.1804, "step": 13446 }, { "epoch": 86.68273092369478, "grad_norm": 23.23682403564453, "learning_rate": 6.62258064516129e-06, "loss": 0.5145, "step": 13447 }, { "epoch": 86.68915662650602, "grad_norm": 9.13611125946045, "learning_rate": 6.619354838709678e-06, "loss": 0.3467, "step": 13448 }, { "epoch": 86.69558232931728, "grad_norm": 15.35851001739502, "learning_rate": 6.616129032258064e-06, "loss": 0.3005, "step": 13449 }, { "epoch": 86.70200803212852, "grad_norm": 7.522763252258301, "learning_rate": 6.6129032258064524e-06, "loss": 0.2904, "step": 13450 }, { "epoch": 86.70843373493976, "grad_norm": 6.874068260192871, "learning_rate": 6.609677419354839e-06, "loss": 0.4019, "step": 13451 }, { "epoch": 86.714859437751, "grad_norm": 7.119190692901611, "learning_rate": 6.606451612903226e-06, "loss": 0.3168, "step": 13452 }, { "epoch": 86.72128514056224, "grad_norm": 4.398261547088623, "learning_rate": 6.603225806451613e-06, "loss": 0.1754, "step": 13453 }, { "epoch": 86.7277108433735, "grad_norm": 9.583515167236328, "learning_rate": 6.6e-06, "loss": 0.5024, "step": 13454 }, { "epoch": 86.73413654618474, "grad_norm": 12.374679565429688, "learning_rate": 6.596774193548387e-06, "loss": 0.3382, "step": 13455 }, { "epoch": 86.74056224899599, "grad_norm": 9.212234497070312, "learning_rate": 6.593548387096775e-06, "loss": 0.4902, "step": 13456 }, { "epoch": 86.74698795180723, "grad_norm": 5.624263286590576, "learning_rate": 6.590322580645162e-06, "loss": 0.241, "step": 13457 }, { "epoch": 86.75341365461847, "grad_norm": 15.306129455566406, "learning_rate": 6.587096774193549e-06, "loss": 0.596, "step": 13458 }, { "epoch": 86.75983935742971, "grad_norm": 10.756329536437988, "learning_rate": 6.583870967741936e-06, "loss": 0.3011, "step": 13459 }, { "epoch": 86.76626506024097, "grad_norm": 3.8677523136138916, "learning_rate": 6.580645161290323e-06, "loss": 0.2374, "step": 13460 }, { "epoch": 86.77269076305221, "grad_norm": 56.69096755981445, "learning_rate": 6.57741935483871e-06, "loss": 0.6376, "step": 13461 }, { "epoch": 86.77911646586345, "grad_norm": 16.284549713134766, "learning_rate": 6.574193548387098e-06, "loss": 0.2685, "step": 13462 }, { "epoch": 86.7855421686747, "grad_norm": 17.238094329833984, "learning_rate": 6.5709677419354836e-06, "loss": 0.3153, "step": 13463 }, { "epoch": 86.79196787148594, "grad_norm": 7.394012451171875, "learning_rate": 6.567741935483872e-06, "loss": 0.2577, "step": 13464 }, { "epoch": 86.7983935742972, "grad_norm": 7.140429973602295, "learning_rate": 6.564516129032258e-06, "loss": 0.1608, "step": 13465 }, { "epoch": 86.80481927710844, "grad_norm": 15.215216636657715, "learning_rate": 6.561290322580646e-06, "loss": 0.2831, "step": 13466 }, { "epoch": 86.81124497991968, "grad_norm": 16.723344802856445, "learning_rate": 6.558064516129032e-06, "loss": 0.4356, "step": 13467 }, { "epoch": 86.81767068273092, "grad_norm": 6.167455196380615, "learning_rate": 6.5548387096774206e-06, "loss": 0.392, "step": 13468 }, { "epoch": 86.82409638554216, "grad_norm": 10.271020889282227, "learning_rate": 6.551612903225806e-06, "loss": 0.1809, "step": 13469 }, { "epoch": 86.83052208835342, "grad_norm": 21.470638275146484, "learning_rate": 6.5483870967741945e-06, "loss": 0.3596, "step": 13470 }, { "epoch": 86.83694779116466, "grad_norm": 22.754411697387695, "learning_rate": 6.545161290322581e-06, "loss": 0.342, "step": 13471 }, { "epoch": 86.8433734939759, "grad_norm": 8.334784507751465, "learning_rate": 6.5419354838709685e-06, "loss": 0.1868, "step": 13472 }, { "epoch": 86.84979919678715, "grad_norm": 5.581734657287598, "learning_rate": 6.538709677419355e-06, "loss": 0.2813, "step": 13473 }, { "epoch": 86.85622489959839, "grad_norm": 14.238253593444824, "learning_rate": 6.535483870967742e-06, "loss": 0.244, "step": 13474 }, { "epoch": 86.86265060240964, "grad_norm": 15.513982772827148, "learning_rate": 6.532258064516129e-06, "loss": 0.2199, "step": 13475 }, { "epoch": 86.86907630522089, "grad_norm": 26.635421752929688, "learning_rate": 6.5290322580645155e-06, "loss": 0.6317, "step": 13476 }, { "epoch": 86.87550200803213, "grad_norm": 4.768005847930908, "learning_rate": 6.525806451612904e-06, "loss": 0.2777, "step": 13477 }, { "epoch": 86.88192771084337, "grad_norm": 12.071271896362305, "learning_rate": 6.52258064516129e-06, "loss": 0.4787, "step": 13478 }, { "epoch": 86.88835341365461, "grad_norm": 11.313007354736328, "learning_rate": 6.519354838709678e-06, "loss": 0.4178, "step": 13479 }, { "epoch": 86.89477911646587, "grad_norm": 13.71605110168457, "learning_rate": 6.516129032258064e-06, "loss": 0.5018, "step": 13480 }, { "epoch": 86.90120481927711, "grad_norm": 7.315127372741699, "learning_rate": 6.512903225806452e-06, "loss": 0.3002, "step": 13481 }, { "epoch": 86.90763052208835, "grad_norm": 8.502462387084961, "learning_rate": 6.509677419354838e-06, "loss": 0.3696, "step": 13482 }, { "epoch": 86.9140562248996, "grad_norm": 11.75753116607666, "learning_rate": 6.5064516129032265e-06, "loss": 0.4719, "step": 13483 }, { "epoch": 86.92048192771084, "grad_norm": 8.066452026367188, "learning_rate": 6.503225806451613e-06, "loss": 0.4269, "step": 13484 }, { "epoch": 86.9269076305221, "grad_norm": 15.186853408813477, "learning_rate": 6.5000000000000004e-06, "loss": 0.3029, "step": 13485 }, { "epoch": 86.93333333333334, "grad_norm": 7.251791000366211, "learning_rate": 6.496774193548387e-06, "loss": 0.3504, "step": 13486 }, { "epoch": 86.93975903614458, "grad_norm": 6.321578502655029, "learning_rate": 6.493548387096774e-06, "loss": 0.1491, "step": 13487 }, { "epoch": 86.94618473895582, "grad_norm": 7.118622303009033, "learning_rate": 6.490322580645161e-06, "loss": 0.3664, "step": 13488 }, { "epoch": 86.95261044176706, "grad_norm": 6.576611042022705, "learning_rate": 6.487096774193549e-06, "loss": 0.2682, "step": 13489 }, { "epoch": 86.95903614457832, "grad_norm": 34.11421203613281, "learning_rate": 6.483870967741936e-06, "loss": 0.5282, "step": 13490 }, { "epoch": 86.96546184738956, "grad_norm": 12.45678997039795, "learning_rate": 6.480645161290323e-06, "loss": 0.5363, "step": 13491 }, { "epoch": 86.9718875502008, "grad_norm": 7.633378028869629, "learning_rate": 6.47741935483871e-06, "loss": 0.5717, "step": 13492 }, { "epoch": 86.97831325301205, "grad_norm": 28.95498275756836, "learning_rate": 6.474193548387097e-06, "loss": 0.2708, "step": 13493 }, { "epoch": 86.98473895582329, "grad_norm": 8.326207160949707, "learning_rate": 6.470967741935484e-06, "loss": 0.2537, "step": 13494 }, { "epoch": 86.99116465863453, "grad_norm": 7.639054298400879, "learning_rate": 6.467741935483872e-06, "loss": 0.2809, "step": 13495 }, { "epoch": 87.00080321285141, "grad_norm": 4.773752212524414, "learning_rate": 6.4645161290322585e-06, "loss": 0.0922, "step": 13496 }, { "epoch": 87.00722891566265, "grad_norm": 6.968710899353027, "learning_rate": 6.461290322580646e-06, "loss": 0.2469, "step": 13497 }, { "epoch": 87.0136546184739, "grad_norm": 29.430442810058594, "learning_rate": 6.458064516129032e-06, "loss": 0.4586, "step": 13498 }, { "epoch": 87.02008032128514, "grad_norm": 10.925230026245117, "learning_rate": 6.45483870967742e-06, "loss": 0.4016, "step": 13499 }, { "epoch": 87.02650602409639, "grad_norm": 6.5890793800354, "learning_rate": 6.451612903225806e-06, "loss": 0.3724, "step": 13500 }, { "epoch": 87.03293172690763, "grad_norm": 16.448017120361328, "learning_rate": 6.448387096774195e-06, "loss": 0.5391, "step": 13501 }, { "epoch": 87.03935742971888, "grad_norm": 5.390415668487549, "learning_rate": 6.445161290322581e-06, "loss": 0.1761, "step": 13502 }, { "epoch": 87.04578313253012, "grad_norm": 8.339991569519043, "learning_rate": 6.441935483870969e-06, "loss": 0.3744, "step": 13503 }, { "epoch": 87.05220883534136, "grad_norm": 5.329864501953125, "learning_rate": 6.438709677419355e-06, "loss": 0.3275, "step": 13504 }, { "epoch": 87.05863453815262, "grad_norm": 9.400432586669922, "learning_rate": 6.4354838709677425e-06, "loss": 0.2217, "step": 13505 }, { "epoch": 87.06506024096386, "grad_norm": 9.327268600463867, "learning_rate": 6.432258064516129e-06, "loss": 0.2298, "step": 13506 }, { "epoch": 87.0714859437751, "grad_norm": 14.644146919250488, "learning_rate": 6.429032258064517e-06, "loss": 0.4171, "step": 13507 }, { "epoch": 87.07791164658634, "grad_norm": 16.775360107421875, "learning_rate": 6.425806451612904e-06, "loss": 0.1354, "step": 13508 }, { "epoch": 87.08433734939759, "grad_norm": 7.219648361206055, "learning_rate": 6.422580645161291e-06, "loss": 0.3113, "step": 13509 }, { "epoch": 87.09076305220883, "grad_norm": 12.677082061767578, "learning_rate": 6.419354838709678e-06, "loss": 0.1902, "step": 13510 }, { "epoch": 87.09718875502008, "grad_norm": 20.112918853759766, "learning_rate": 6.416129032258064e-06, "loss": 0.3, "step": 13511 }, { "epoch": 87.10361445783133, "grad_norm": 13.59380054473877, "learning_rate": 6.412903225806452e-06, "loss": 0.1847, "step": 13512 }, { "epoch": 87.11004016064257, "grad_norm": 8.168635368347168, "learning_rate": 6.409677419354838e-06, "loss": 0.3201, "step": 13513 }, { "epoch": 87.11646586345381, "grad_norm": 25.684616088867188, "learning_rate": 6.406451612903227e-06, "loss": 0.5939, "step": 13514 }, { "epoch": 87.12289156626505, "grad_norm": 7.727542877197266, "learning_rate": 6.403225806451612e-06, "loss": 0.3449, "step": 13515 }, { "epoch": 87.12931726907631, "grad_norm": 4.509312629699707, "learning_rate": 6.4000000000000006e-06, "loss": 0.2491, "step": 13516 }, { "epoch": 87.13574297188755, "grad_norm": 3.6642234325408936, "learning_rate": 6.396774193548387e-06, "loss": 0.1052, "step": 13517 }, { "epoch": 87.1421686746988, "grad_norm": 11.939581871032715, "learning_rate": 6.3935483870967745e-06, "loss": 0.6373, "step": 13518 }, { "epoch": 87.14859437751004, "grad_norm": 14.734682083129883, "learning_rate": 6.390322580645161e-06, "loss": 0.229, "step": 13519 }, { "epoch": 87.15502008032128, "grad_norm": 9.733906745910645, "learning_rate": 6.387096774193549e-06, "loss": 0.2399, "step": 13520 }, { "epoch": 87.16144578313254, "grad_norm": 12.670665740966797, "learning_rate": 6.383870967741935e-06, "loss": 0.4546, "step": 13521 }, { "epoch": 87.16787148594378, "grad_norm": 7.540274620056152, "learning_rate": 6.380645161290323e-06, "loss": 0.2605, "step": 13522 }, { "epoch": 87.17429718875502, "grad_norm": 11.960524559020996, "learning_rate": 6.37741935483871e-06, "loss": 0.4458, "step": 13523 }, { "epoch": 87.18072289156626, "grad_norm": 8.525282859802246, "learning_rate": 6.374193548387097e-06, "loss": 0.4047, "step": 13524 }, { "epoch": 87.1871485943775, "grad_norm": 11.950794219970703, "learning_rate": 6.370967741935484e-06, "loss": 0.3274, "step": 13525 }, { "epoch": 87.19357429718876, "grad_norm": 5.515535354614258, "learning_rate": 6.367741935483871e-06, "loss": 0.1978, "step": 13526 }, { "epoch": 87.2, "grad_norm": 16.670957565307617, "learning_rate": 6.364516129032258e-06, "loss": 0.7762, "step": 13527 }, { "epoch": 87.20642570281125, "grad_norm": 5.510606288909912, "learning_rate": 6.361290322580646e-06, "loss": 0.2113, "step": 13528 }, { "epoch": 87.21285140562249, "grad_norm": 10.558112144470215, "learning_rate": 6.3580645161290325e-06, "loss": 0.2428, "step": 13529 }, { "epoch": 87.21927710843373, "grad_norm": 18.0924129486084, "learning_rate": 6.35483870967742e-06, "loss": 0.4399, "step": 13530 }, { "epoch": 87.22570281124499, "grad_norm": 12.96668529510498, "learning_rate": 6.3516129032258065e-06, "loss": 0.6258, "step": 13531 }, { "epoch": 87.23212851405623, "grad_norm": 25.14198112487793, "learning_rate": 6.348387096774194e-06, "loss": 0.3372, "step": 13532 }, { "epoch": 87.23855421686747, "grad_norm": 8.929437637329102, "learning_rate": 6.3451612903225804e-06, "loss": 0.183, "step": 13533 }, { "epoch": 87.24497991967871, "grad_norm": 5.248844146728516, "learning_rate": 6.341935483870969e-06, "loss": 0.2784, "step": 13534 }, { "epoch": 87.25140562248995, "grad_norm": 9.194365501403809, "learning_rate": 6.338709677419355e-06, "loss": 0.3269, "step": 13535 }, { "epoch": 87.25783132530121, "grad_norm": 9.856157302856445, "learning_rate": 6.335483870967743e-06, "loss": 0.2441, "step": 13536 }, { "epoch": 87.26425702811245, "grad_norm": 7.325726509094238, "learning_rate": 6.332258064516129e-06, "loss": 0.3515, "step": 13537 }, { "epoch": 87.2706827309237, "grad_norm": 7.528064727783203, "learning_rate": 6.329032258064517e-06, "loss": 0.1877, "step": 13538 }, { "epoch": 87.27710843373494, "grad_norm": 14.162765502929688, "learning_rate": 6.325806451612903e-06, "loss": 0.5288, "step": 13539 }, { "epoch": 87.28353413654618, "grad_norm": 9.224968910217285, "learning_rate": 6.322580645161291e-06, "loss": 0.4165, "step": 13540 }, { "epoch": 87.28995983935744, "grad_norm": 11.592544555664062, "learning_rate": 6.319354838709678e-06, "loss": 0.3589, "step": 13541 }, { "epoch": 87.29638554216868, "grad_norm": 8.942456245422363, "learning_rate": 6.316129032258065e-06, "loss": 0.3929, "step": 13542 }, { "epoch": 87.30281124497992, "grad_norm": 20.485803604125977, "learning_rate": 6.312903225806452e-06, "loss": 0.4191, "step": 13543 }, { "epoch": 87.30923694779116, "grad_norm": 4.893963813781738, "learning_rate": 6.309677419354839e-06, "loss": 0.2733, "step": 13544 }, { "epoch": 87.3156626506024, "grad_norm": 10.114580154418945, "learning_rate": 6.306451612903226e-06, "loss": 0.3539, "step": 13545 }, { "epoch": 87.32208835341365, "grad_norm": 18.578630447387695, "learning_rate": 6.303225806451612e-06, "loss": 0.2701, "step": 13546 }, { "epoch": 87.3285140562249, "grad_norm": 14.747152328491211, "learning_rate": 6.300000000000001e-06, "loss": 0.3369, "step": 13547 }, { "epoch": 87.33493975903615, "grad_norm": 7.0214009284973145, "learning_rate": 6.296774193548386e-06, "loss": 0.4095, "step": 13548 }, { "epoch": 87.34136546184739, "grad_norm": 12.684633255004883, "learning_rate": 6.293548387096775e-06, "loss": 0.2934, "step": 13549 }, { "epoch": 87.34779116465863, "grad_norm": 28.356277465820312, "learning_rate": 6.290322580645161e-06, "loss": 0.5251, "step": 13550 }, { "epoch": 87.35421686746987, "grad_norm": 11.357434272766113, "learning_rate": 6.287096774193549e-06, "loss": 0.4095, "step": 13551 }, { "epoch": 87.36064257028113, "grad_norm": 7.746479034423828, "learning_rate": 6.283870967741935e-06, "loss": 0.4323, "step": 13552 }, { "epoch": 87.36706827309237, "grad_norm": 9.598017692565918, "learning_rate": 6.280645161290323e-06, "loss": 0.3316, "step": 13553 }, { "epoch": 87.37349397590361, "grad_norm": 18.470857620239258, "learning_rate": 6.277419354838709e-06, "loss": 0.347, "step": 13554 }, { "epoch": 87.37991967871486, "grad_norm": 11.7235746383667, "learning_rate": 6.274193548387097e-06, "loss": 0.5111, "step": 13555 }, { "epoch": 87.3863453815261, "grad_norm": 5.472783088684082, "learning_rate": 6.270967741935484e-06, "loss": 0.1438, "step": 13556 }, { "epoch": 87.39277108433735, "grad_norm": 10.11569881439209, "learning_rate": 6.267741935483871e-06, "loss": 0.2042, "step": 13557 }, { "epoch": 87.3991967871486, "grad_norm": 8.00730037689209, "learning_rate": 6.264516129032258e-06, "loss": 0.3279, "step": 13558 }, { "epoch": 87.40562248995984, "grad_norm": 10.669005393981934, "learning_rate": 6.261290322580646e-06, "loss": 0.5121, "step": 13559 }, { "epoch": 87.41204819277108, "grad_norm": 6.6489410400390625, "learning_rate": 6.258064516129032e-06, "loss": 0.1845, "step": 13560 }, { "epoch": 87.41847389558232, "grad_norm": 8.766242027282715, "learning_rate": 6.25483870967742e-06, "loss": 0.2879, "step": 13561 }, { "epoch": 87.42489959839358, "grad_norm": 11.357542991638184, "learning_rate": 6.251612903225807e-06, "loss": 0.3921, "step": 13562 }, { "epoch": 87.43132530120482, "grad_norm": 21.25037956237793, "learning_rate": 6.248387096774193e-06, "loss": 0.4222, "step": 13563 }, { "epoch": 87.43775100401606, "grad_norm": 8.351457595825195, "learning_rate": 6.2451612903225806e-06, "loss": 0.3952, "step": 13564 }, { "epoch": 87.4441767068273, "grad_norm": 8.362005233764648, "learning_rate": 6.241935483870968e-06, "loss": 0.1873, "step": 13565 }, { "epoch": 87.45060240963855, "grad_norm": 6.967912197113037, "learning_rate": 6.2387096774193545e-06, "loss": 0.4371, "step": 13566 }, { "epoch": 87.4570281124498, "grad_norm": 8.327869415283203, "learning_rate": 6.235483870967742e-06, "loss": 0.5942, "step": 13567 }, { "epoch": 87.46345381526105, "grad_norm": 5.207332611083984, "learning_rate": 6.232258064516129e-06, "loss": 0.1354, "step": 13568 }, { "epoch": 87.46987951807229, "grad_norm": 7.269742488861084, "learning_rate": 6.229032258064516e-06, "loss": 0.2596, "step": 13569 }, { "epoch": 87.47630522088353, "grad_norm": 11.516151428222656, "learning_rate": 6.225806451612903e-06, "loss": 0.4695, "step": 13570 }, { "epoch": 87.48273092369477, "grad_norm": 6.750399112701416, "learning_rate": 6.222580645161291e-06, "loss": 0.173, "step": 13571 }, { "epoch": 87.48915662650603, "grad_norm": 5.55545711517334, "learning_rate": 6.219354838709677e-06, "loss": 0.4017, "step": 13572 }, { "epoch": 87.49558232931727, "grad_norm": 4.363644123077393, "learning_rate": 6.216129032258065e-06, "loss": 0.178, "step": 13573 }, { "epoch": 87.50200803212851, "grad_norm": 8.231900215148926, "learning_rate": 6.212903225806452e-06, "loss": 0.3549, "step": 13574 }, { "epoch": 87.50843373493976, "grad_norm": 5.9098687171936035, "learning_rate": 6.2096774193548386e-06, "loss": 0.3161, "step": 13575 }, { "epoch": 87.514859437751, "grad_norm": 14.125438690185547, "learning_rate": 6.206451612903226e-06, "loss": 0.3927, "step": 13576 }, { "epoch": 87.52128514056226, "grad_norm": 11.117585182189941, "learning_rate": 6.203225806451613e-06, "loss": 0.3872, "step": 13577 }, { "epoch": 87.5277108433735, "grad_norm": 10.936565399169922, "learning_rate": 6.2e-06, "loss": 0.1834, "step": 13578 }, { "epoch": 87.53413654618474, "grad_norm": 9.158080101013184, "learning_rate": 6.196774193548387e-06, "loss": 0.3828, "step": 13579 }, { "epoch": 87.54056224899598, "grad_norm": 13.88725471496582, "learning_rate": 6.193548387096775e-06, "loss": 0.4337, "step": 13580 }, { "epoch": 87.54698795180722, "grad_norm": 17.613231658935547, "learning_rate": 6.190322580645161e-06, "loss": 0.325, "step": 13581 }, { "epoch": 87.55341365461847, "grad_norm": 6.102320671081543, "learning_rate": 6.187096774193549e-06, "loss": 0.1896, "step": 13582 }, { "epoch": 87.55983935742972, "grad_norm": 4.819173812866211, "learning_rate": 6.183870967741936e-06, "loss": 0.1472, "step": 13583 }, { "epoch": 87.56626506024097, "grad_norm": 3.8408093452453613, "learning_rate": 6.180645161290323e-06, "loss": 0.2223, "step": 13584 }, { "epoch": 87.57269076305221, "grad_norm": 18.505098342895508, "learning_rate": 6.17741935483871e-06, "loss": 0.611, "step": 13585 }, { "epoch": 87.57911646586345, "grad_norm": 7.325392246246338, "learning_rate": 6.1741935483870974e-06, "loss": 0.3043, "step": 13586 }, { "epoch": 87.58554216867469, "grad_norm": 10.20470142364502, "learning_rate": 6.170967741935484e-06, "loss": 0.4246, "step": 13587 }, { "epoch": 87.59196787148595, "grad_norm": 9.642539978027344, "learning_rate": 6.167741935483871e-06, "loss": 0.3071, "step": 13588 }, { "epoch": 87.59839357429719, "grad_norm": 4.616413116455078, "learning_rate": 6.164516129032259e-06, "loss": 0.2806, "step": 13589 }, { "epoch": 87.60481927710843, "grad_norm": 18.110437393188477, "learning_rate": 6.161290322580645e-06, "loss": 0.4004, "step": 13590 }, { "epoch": 87.61124497991968, "grad_norm": 8.548023223876953, "learning_rate": 6.158064516129033e-06, "loss": 0.3146, "step": 13591 }, { "epoch": 87.61767068273092, "grad_norm": 6.506291389465332, "learning_rate": 6.15483870967742e-06, "loss": 0.277, "step": 13592 }, { "epoch": 87.62409638554217, "grad_norm": 10.898289680480957, "learning_rate": 6.151612903225807e-06, "loss": 0.2193, "step": 13593 }, { "epoch": 87.63052208835342, "grad_norm": 8.241815567016602, "learning_rate": 6.148387096774194e-06, "loss": 0.2406, "step": 13594 }, { "epoch": 87.63694779116466, "grad_norm": 7.725975036621094, "learning_rate": 6.1451612903225815e-06, "loss": 0.2639, "step": 13595 }, { "epoch": 87.6433734939759, "grad_norm": 11.888050079345703, "learning_rate": 6.141935483870968e-06, "loss": 0.3466, "step": 13596 }, { "epoch": 87.64979919678714, "grad_norm": 30.140535354614258, "learning_rate": 6.1387096774193555e-06, "loss": 0.1689, "step": 13597 }, { "epoch": 87.6562248995984, "grad_norm": 5.915395736694336, "learning_rate": 6.135483870967743e-06, "loss": 0.4122, "step": 13598 }, { "epoch": 87.66265060240964, "grad_norm": 6.179309844970703, "learning_rate": 6.132258064516129e-06, "loss": 0.5155, "step": 13599 }, { "epoch": 87.66907630522088, "grad_norm": 41.159767150878906, "learning_rate": 6.129032258064516e-06, "loss": 0.5652, "step": 13600 }, { "epoch": 87.67550200803213, "grad_norm": 85.45641326904297, "learning_rate": 6.125806451612903e-06, "loss": 0.5244, "step": 13601 }, { "epoch": 87.68192771084337, "grad_norm": 15.055031776428223, "learning_rate": 6.12258064516129e-06, "loss": 0.1983, "step": 13602 }, { "epoch": 87.68835341365462, "grad_norm": 12.140835762023926, "learning_rate": 6.119354838709677e-06, "loss": 0.3477, "step": 13603 }, { "epoch": 87.69477911646587, "grad_norm": 14.988224029541016, "learning_rate": 6.116129032258065e-06, "loss": 0.2363, "step": 13604 }, { "epoch": 87.70120481927711, "grad_norm": 19.952844619750977, "learning_rate": 6.112903225806451e-06, "loss": 0.254, "step": 13605 }, { "epoch": 87.70763052208835, "grad_norm": 10.565441131591797, "learning_rate": 6.109677419354839e-06, "loss": 0.249, "step": 13606 }, { "epoch": 87.7140562248996, "grad_norm": 13.55928897857666, "learning_rate": 6.106451612903226e-06, "loss": 0.5451, "step": 13607 }, { "epoch": 87.72048192771085, "grad_norm": 13.66010856628418, "learning_rate": 6.103225806451613e-06, "loss": 0.2748, "step": 13608 }, { "epoch": 87.72690763052209, "grad_norm": 7.149771690368652, "learning_rate": 6.1e-06, "loss": 0.4617, "step": 13609 }, { "epoch": 87.73333333333333, "grad_norm": 10.452563285827637, "learning_rate": 6.0967741935483874e-06, "loss": 0.2464, "step": 13610 }, { "epoch": 87.73975903614458, "grad_norm": 24.76214027404785, "learning_rate": 6.093548387096774e-06, "loss": 0.2556, "step": 13611 }, { "epoch": 87.74618473895582, "grad_norm": 13.044961929321289, "learning_rate": 6.090322580645161e-06, "loss": 0.3946, "step": 13612 }, { "epoch": 87.75261044176708, "grad_norm": 8.549703598022461, "learning_rate": 6.087096774193549e-06, "loss": 0.3104, "step": 13613 }, { "epoch": 87.75903614457832, "grad_norm": 7.999269008636475, "learning_rate": 6.083870967741935e-06, "loss": 0.3409, "step": 13614 }, { "epoch": 87.76546184738956, "grad_norm": 4.971924304962158, "learning_rate": 6.080645161290323e-06, "loss": 0.2859, "step": 13615 }, { "epoch": 87.7718875502008, "grad_norm": 21.491323471069336, "learning_rate": 6.07741935483871e-06, "loss": 0.5707, "step": 13616 }, { "epoch": 87.77831325301204, "grad_norm": 4.892562389373779, "learning_rate": 6.074193548387097e-06, "loss": 0.1995, "step": 13617 }, { "epoch": 87.78473895582329, "grad_norm": 6.391353130340576, "learning_rate": 6.070967741935484e-06, "loss": 0.2765, "step": 13618 }, { "epoch": 87.79116465863454, "grad_norm": 6.518146514892578, "learning_rate": 6.0677419354838715e-06, "loss": 0.3607, "step": 13619 }, { "epoch": 87.79759036144578, "grad_norm": 10.553650856018066, "learning_rate": 6.064516129032258e-06, "loss": 0.2481, "step": 13620 }, { "epoch": 87.80401606425703, "grad_norm": 9.928726196289062, "learning_rate": 6.0612903225806455e-06, "loss": 0.2132, "step": 13621 }, { "epoch": 87.81044176706827, "grad_norm": 18.945987701416016, "learning_rate": 6.058064516129033e-06, "loss": 0.3727, "step": 13622 }, { "epoch": 87.81686746987951, "grad_norm": 4.066441059112549, "learning_rate": 6.054838709677419e-06, "loss": 0.06, "step": 13623 }, { "epoch": 87.82329317269077, "grad_norm": 10.537728309631348, "learning_rate": 6.051612903225807e-06, "loss": 0.5365, "step": 13624 }, { "epoch": 87.82971887550201, "grad_norm": 6.8489089012146, "learning_rate": 6.048387096774194e-06, "loss": 0.2906, "step": 13625 }, { "epoch": 87.83614457831325, "grad_norm": 7.171487808227539, "learning_rate": 6.045161290322581e-06, "loss": 0.2221, "step": 13626 }, { "epoch": 87.8425702811245, "grad_norm": 28.470766067504883, "learning_rate": 6.041935483870968e-06, "loss": 0.4825, "step": 13627 }, { "epoch": 87.84899598393574, "grad_norm": 13.79638671875, "learning_rate": 6.0387096774193556e-06, "loss": 0.4199, "step": 13628 }, { "epoch": 87.855421686747, "grad_norm": 19.291093826293945, "learning_rate": 6.035483870967742e-06, "loss": 0.2379, "step": 13629 }, { "epoch": 87.86184738955824, "grad_norm": 9.105409622192383, "learning_rate": 6.0322580645161295e-06, "loss": 0.3166, "step": 13630 }, { "epoch": 87.86827309236948, "grad_norm": 4.732435703277588, "learning_rate": 6.029032258064517e-06, "loss": 0.3729, "step": 13631 }, { "epoch": 87.87469879518072, "grad_norm": 10.889591217041016, "learning_rate": 6.0258064516129035e-06, "loss": 0.2939, "step": 13632 }, { "epoch": 87.88112449799196, "grad_norm": 16.13544464111328, "learning_rate": 6.022580645161291e-06, "loss": 0.4271, "step": 13633 }, { "epoch": 87.88755020080322, "grad_norm": 7.420890808105469, "learning_rate": 6.0193548387096774e-06, "loss": 0.3491, "step": 13634 }, { "epoch": 87.89397590361446, "grad_norm": 4.844944477081299, "learning_rate": 6.016129032258065e-06, "loss": 0.1262, "step": 13635 }, { "epoch": 87.9004016064257, "grad_norm": 19.982677459716797, "learning_rate": 6.012903225806451e-06, "loss": 0.201, "step": 13636 }, { "epoch": 87.90682730923695, "grad_norm": 15.019173622131348, "learning_rate": 6.009677419354839e-06, "loss": 0.2798, "step": 13637 }, { "epoch": 87.91325301204819, "grad_norm": 8.119112968444824, "learning_rate": 6.006451612903226e-06, "loss": 0.2674, "step": 13638 }, { "epoch": 87.91967871485944, "grad_norm": 24.237682342529297, "learning_rate": 6.003225806451613e-06, "loss": 0.7458, "step": 13639 }, { "epoch": 87.92610441767069, "grad_norm": 18.384323120117188, "learning_rate": 6e-06, "loss": 0.5045, "step": 13640 }, { "epoch": 87.93253012048193, "grad_norm": 18.326826095581055, "learning_rate": 5.9967741935483876e-06, "loss": 0.3314, "step": 13641 }, { "epoch": 87.93895582329317, "grad_norm": 5.978427410125732, "learning_rate": 5.993548387096774e-06, "loss": 0.2238, "step": 13642 }, { "epoch": 87.94538152610441, "grad_norm": 19.668106079101562, "learning_rate": 5.9903225806451615e-06, "loss": 0.5392, "step": 13643 }, { "epoch": 87.95180722891567, "grad_norm": 4.871905326843262, "learning_rate": 5.987096774193549e-06, "loss": 0.1193, "step": 13644 }, { "epoch": 87.95823293172691, "grad_norm": 11.10175895690918, "learning_rate": 5.9838709677419355e-06, "loss": 0.5112, "step": 13645 }, { "epoch": 87.96465863453815, "grad_norm": 10.128411293029785, "learning_rate": 5.980645161290323e-06, "loss": 0.227, "step": 13646 }, { "epoch": 87.9710843373494, "grad_norm": 10.802862167358398, "learning_rate": 5.97741935483871e-06, "loss": 0.3895, "step": 13647 }, { "epoch": 87.97751004016064, "grad_norm": 9.783493041992188, "learning_rate": 5.974193548387097e-06, "loss": 0.2988, "step": 13648 }, { "epoch": 87.9839357429719, "grad_norm": 6.278580188751221, "learning_rate": 5.970967741935484e-06, "loss": 0.2077, "step": 13649 }, { "epoch": 87.99036144578314, "grad_norm": 13.57073974609375, "learning_rate": 5.967741935483872e-06, "loss": 0.3034, "step": 13650 }, { "epoch": 87.99678714859438, "grad_norm": 11.997310638427734, "learning_rate": 5.964516129032258e-06, "loss": 0.3511, "step": 13651 }, { "epoch": 88.00642570281124, "grad_norm": 7.531935214996338, "learning_rate": 5.9612903225806456e-06, "loss": 0.3268, "step": 13652 }, { "epoch": 88.01285140562248, "grad_norm": 12.449915885925293, "learning_rate": 5.958064516129033e-06, "loss": 0.249, "step": 13653 }, { "epoch": 88.01927710843374, "grad_norm": 6.469474792480469, "learning_rate": 5.9548387096774195e-06, "loss": 0.2741, "step": 13654 }, { "epoch": 88.02570281124498, "grad_norm": 5.380314350128174, "learning_rate": 5.951612903225807e-06, "loss": 0.2602, "step": 13655 }, { "epoch": 88.03212851405623, "grad_norm": 9.320365905761719, "learning_rate": 5.948387096774194e-06, "loss": 0.5336, "step": 13656 }, { "epoch": 88.03855421686747, "grad_norm": 19.15413475036621, "learning_rate": 5.945161290322581e-06, "loss": 0.5715, "step": 13657 }, { "epoch": 88.04497991967871, "grad_norm": 8.420507431030273, "learning_rate": 5.941935483870968e-06, "loss": 0.22, "step": 13658 }, { "epoch": 88.05140562248997, "grad_norm": 11.744450569152832, "learning_rate": 5.938709677419356e-06, "loss": 0.4258, "step": 13659 }, { "epoch": 88.05783132530121, "grad_norm": 15.8191499710083, "learning_rate": 5.935483870967742e-06, "loss": 0.2292, "step": 13660 }, { "epoch": 88.06425702811245, "grad_norm": 10.208907127380371, "learning_rate": 5.93225806451613e-06, "loss": 0.2542, "step": 13661 }, { "epoch": 88.07068273092369, "grad_norm": 9.61933422088623, "learning_rate": 5.929032258064517e-06, "loss": 0.2725, "step": 13662 }, { "epoch": 88.07710843373494, "grad_norm": 8.13319206237793, "learning_rate": 5.925806451612904e-06, "loss": 0.4595, "step": 13663 }, { "epoch": 88.08353413654619, "grad_norm": 10.125794410705566, "learning_rate": 5.922580645161291e-06, "loss": 0.2534, "step": 13664 }, { "epoch": 88.08995983935743, "grad_norm": 9.64353084564209, "learning_rate": 5.9193548387096776e-06, "loss": 0.3507, "step": 13665 }, { "epoch": 88.09638554216868, "grad_norm": 9.583965301513672, "learning_rate": 5.916129032258065e-06, "loss": 0.3427, "step": 13666 }, { "epoch": 88.10281124497992, "grad_norm": 13.862837791442871, "learning_rate": 5.912903225806452e-06, "loss": 0.4071, "step": 13667 }, { "epoch": 88.10923694779116, "grad_norm": 7.392111778259277, "learning_rate": 5.909677419354839e-06, "loss": 0.2251, "step": 13668 }, { "epoch": 88.1156626506024, "grad_norm": 5.020719051361084, "learning_rate": 5.906451612903226e-06, "loss": 0.3253, "step": 13669 }, { "epoch": 88.12208835341366, "grad_norm": 9.492110252380371, "learning_rate": 5.903225806451613e-06, "loss": 0.2895, "step": 13670 }, { "epoch": 88.1285140562249, "grad_norm": 4.15389347076416, "learning_rate": 5.9e-06, "loss": 0.1905, "step": 13671 }, { "epoch": 88.13493975903614, "grad_norm": 12.407919883728027, "learning_rate": 5.896774193548387e-06, "loss": 0.4463, "step": 13672 }, { "epoch": 88.14136546184739, "grad_norm": 10.9640531539917, "learning_rate": 5.893548387096774e-06, "loss": 0.4587, "step": 13673 }, { "epoch": 88.14779116465863, "grad_norm": 9.172945022583008, "learning_rate": 5.890322580645162e-06, "loss": 0.3151, "step": 13674 }, { "epoch": 88.15421686746988, "grad_norm": 15.287590980529785, "learning_rate": 5.887096774193548e-06, "loss": 0.3718, "step": 13675 }, { "epoch": 88.16064257028113, "grad_norm": 6.948726177215576, "learning_rate": 5.8838709677419356e-06, "loss": 0.2779, "step": 13676 }, { "epoch": 88.16706827309237, "grad_norm": 16.978212356567383, "learning_rate": 5.880645161290323e-06, "loss": 0.4037, "step": 13677 }, { "epoch": 88.17349397590361, "grad_norm": 4.412450313568115, "learning_rate": 5.8774193548387095e-06, "loss": 0.0902, "step": 13678 }, { "epoch": 88.17991967871485, "grad_norm": 7.3686203956604, "learning_rate": 5.874193548387097e-06, "loss": 0.201, "step": 13679 }, { "epoch": 88.18634538152611, "grad_norm": 9.32132339477539, "learning_rate": 5.870967741935484e-06, "loss": 0.2901, "step": 13680 }, { "epoch": 88.19277108433735, "grad_norm": 9.322237968444824, "learning_rate": 5.867741935483871e-06, "loss": 0.377, "step": 13681 }, { "epoch": 88.1991967871486, "grad_norm": 5.688650608062744, "learning_rate": 5.864516129032258e-06, "loss": 0.3462, "step": 13682 }, { "epoch": 88.20562248995984, "grad_norm": 21.30672836303711, "learning_rate": 5.861290322580646e-06, "loss": 0.5711, "step": 13683 }, { "epoch": 88.21204819277108, "grad_norm": 33.329566955566406, "learning_rate": 5.858064516129032e-06, "loss": 0.3135, "step": 13684 }, { "epoch": 88.21847389558233, "grad_norm": 18.691097259521484, "learning_rate": 5.85483870967742e-06, "loss": 0.3447, "step": 13685 }, { "epoch": 88.22489959839358, "grad_norm": 7.495975494384766, "learning_rate": 5.851612903225807e-06, "loss": 0.3273, "step": 13686 }, { "epoch": 88.23132530120482, "grad_norm": 13.359444618225098, "learning_rate": 5.848387096774194e-06, "loss": 0.345, "step": 13687 }, { "epoch": 88.23775100401606, "grad_norm": 8.749173164367676, "learning_rate": 5.845161290322581e-06, "loss": 0.3763, "step": 13688 }, { "epoch": 88.2441767068273, "grad_norm": 8.325955390930176, "learning_rate": 5.841935483870968e-06, "loss": 0.3069, "step": 13689 }, { "epoch": 88.25060240963856, "grad_norm": 12.915409088134766, "learning_rate": 5.838709677419355e-06, "loss": 0.5347, "step": 13690 }, { "epoch": 88.2570281124498, "grad_norm": 19.162857055664062, "learning_rate": 5.835483870967742e-06, "loss": 0.2557, "step": 13691 }, { "epoch": 88.26345381526104, "grad_norm": 3.9110610485076904, "learning_rate": 5.83225806451613e-06, "loss": 0.092, "step": 13692 }, { "epoch": 88.26987951807229, "grad_norm": 11.186447143554688, "learning_rate": 5.829032258064516e-06, "loss": 0.4355, "step": 13693 }, { "epoch": 88.27630522088353, "grad_norm": 5.741999626159668, "learning_rate": 5.825806451612904e-06, "loss": 0.1087, "step": 13694 }, { "epoch": 88.28273092369479, "grad_norm": 11.915916442871094, "learning_rate": 5.822580645161291e-06, "loss": 0.4821, "step": 13695 }, { "epoch": 88.28915662650603, "grad_norm": 7.616482734680176, "learning_rate": 5.819354838709678e-06, "loss": 0.4224, "step": 13696 }, { "epoch": 88.29558232931727, "grad_norm": 6.462557315826416, "learning_rate": 5.816129032258065e-06, "loss": 0.3526, "step": 13697 }, { "epoch": 88.30200803212851, "grad_norm": 16.88758659362793, "learning_rate": 5.8129032258064525e-06, "loss": 0.5245, "step": 13698 }, { "epoch": 88.30843373493975, "grad_norm": 4.059993743896484, "learning_rate": 5.809677419354839e-06, "loss": 0.1206, "step": 13699 }, { "epoch": 88.31485943775101, "grad_norm": 15.894073486328125, "learning_rate": 5.806451612903226e-06, "loss": 0.3618, "step": 13700 }, { "epoch": 88.32128514056225, "grad_norm": 11.350177764892578, "learning_rate": 5.803225806451614e-06, "loss": 0.4531, "step": 13701 }, { "epoch": 88.3277108433735, "grad_norm": 8.500876426696777, "learning_rate": 5.8e-06, "loss": 0.1533, "step": 13702 }, { "epoch": 88.33413654618474, "grad_norm": 5.6377644538879395, "learning_rate": 5.796774193548388e-06, "loss": 0.1676, "step": 13703 }, { "epoch": 88.34056224899598, "grad_norm": 7.759264945983887, "learning_rate": 5.793548387096775e-06, "loss": 0.4606, "step": 13704 }, { "epoch": 88.34698795180722, "grad_norm": 12.150283813476562, "learning_rate": 5.790322580645161e-06, "loss": 0.5334, "step": 13705 }, { "epoch": 88.35341365461848, "grad_norm": 9.356378555297852, "learning_rate": 5.787096774193548e-06, "loss": 0.2415, "step": 13706 }, { "epoch": 88.35983935742972, "grad_norm": 11.362293243408203, "learning_rate": 5.783870967741936e-06, "loss": 0.6132, "step": 13707 }, { "epoch": 88.36626506024096, "grad_norm": 11.079670906066895, "learning_rate": 5.780645161290322e-06, "loss": 0.4214, "step": 13708 }, { "epoch": 88.3726907630522, "grad_norm": 7.7866363525390625, "learning_rate": 5.77741935483871e-06, "loss": 0.2066, "step": 13709 }, { "epoch": 88.37911646586345, "grad_norm": 9.151113510131836, "learning_rate": 5.774193548387097e-06, "loss": 0.3185, "step": 13710 }, { "epoch": 88.3855421686747, "grad_norm": 6.534732818603516, "learning_rate": 5.770967741935484e-06, "loss": 0.1863, "step": 13711 }, { "epoch": 88.39196787148595, "grad_norm": 8.009430885314941, "learning_rate": 5.767741935483871e-06, "loss": 0.1816, "step": 13712 }, { "epoch": 88.39839357429719, "grad_norm": 7.706727027893066, "learning_rate": 5.764516129032258e-06, "loss": 0.1383, "step": 13713 }, { "epoch": 88.40481927710843, "grad_norm": 8.101943016052246, "learning_rate": 5.761290322580645e-06, "loss": 0.2823, "step": 13714 }, { "epoch": 88.41124497991967, "grad_norm": 17.882165908813477, "learning_rate": 5.758064516129032e-06, "loss": 0.3974, "step": 13715 }, { "epoch": 88.41767068273093, "grad_norm": 16.100927352905273, "learning_rate": 5.75483870967742e-06, "loss": 0.5936, "step": 13716 }, { "epoch": 88.42409638554217, "grad_norm": 13.740479469299316, "learning_rate": 5.751612903225806e-06, "loss": 0.5131, "step": 13717 }, { "epoch": 88.43052208835341, "grad_norm": 7.733924388885498, "learning_rate": 5.748387096774194e-06, "loss": 0.4224, "step": 13718 }, { "epoch": 88.43694779116466, "grad_norm": 7.293237686157227, "learning_rate": 5.745161290322581e-06, "loss": 0.2809, "step": 13719 }, { "epoch": 88.4433734939759, "grad_norm": 8.05113697052002, "learning_rate": 5.741935483870968e-06, "loss": 0.4625, "step": 13720 }, { "epoch": 88.44979919678715, "grad_norm": 24.31340789794922, "learning_rate": 5.738709677419355e-06, "loss": 0.6097, "step": 13721 }, { "epoch": 88.4562248995984, "grad_norm": 8.74988842010498, "learning_rate": 5.7354838709677425e-06, "loss": 0.4597, "step": 13722 }, { "epoch": 88.46265060240964, "grad_norm": 10.10425853729248, "learning_rate": 5.732258064516129e-06, "loss": 0.2666, "step": 13723 }, { "epoch": 88.46907630522088, "grad_norm": 16.92888832092285, "learning_rate": 5.729032258064516e-06, "loss": 0.4217, "step": 13724 }, { "epoch": 88.47550200803212, "grad_norm": 18.232959747314453, "learning_rate": 5.725806451612904e-06, "loss": 0.9743, "step": 13725 }, { "epoch": 88.48192771084338, "grad_norm": 9.540701866149902, "learning_rate": 5.72258064516129e-06, "loss": 0.2183, "step": 13726 }, { "epoch": 88.48835341365462, "grad_norm": 10.152589797973633, "learning_rate": 5.719354838709678e-06, "loss": 0.2161, "step": 13727 }, { "epoch": 88.49477911646586, "grad_norm": 11.007783889770508, "learning_rate": 5.716129032258065e-06, "loss": 0.2364, "step": 13728 }, { "epoch": 88.5012048192771, "grad_norm": 7.868812084197998, "learning_rate": 5.712903225806452e-06, "loss": 0.227, "step": 13729 }, { "epoch": 88.50763052208835, "grad_norm": 9.1657075881958, "learning_rate": 5.709677419354839e-06, "loss": 0.4716, "step": 13730 }, { "epoch": 88.5140562248996, "grad_norm": 7.003095626831055, "learning_rate": 5.7064516129032265e-06, "loss": 0.5295, "step": 13731 }, { "epoch": 88.52048192771085, "grad_norm": 18.488040924072266, "learning_rate": 5.703225806451613e-06, "loss": 0.2865, "step": 13732 }, { "epoch": 88.52690763052209, "grad_norm": 10.210089683532715, "learning_rate": 5.7000000000000005e-06, "loss": 0.1354, "step": 13733 }, { "epoch": 88.53333333333333, "grad_norm": 7.754247188568115, "learning_rate": 5.696774193548388e-06, "loss": 0.2249, "step": 13734 }, { "epoch": 88.53975903614457, "grad_norm": 8.193336486816406, "learning_rate": 5.6935483870967744e-06, "loss": 0.286, "step": 13735 }, { "epoch": 88.54618473895583, "grad_norm": 20.844833374023438, "learning_rate": 5.690322580645162e-06, "loss": 0.4953, "step": 13736 }, { "epoch": 88.55261044176707, "grad_norm": 44.29685974121094, "learning_rate": 5.687096774193549e-06, "loss": 0.5039, "step": 13737 }, { "epoch": 88.55903614457831, "grad_norm": 9.953105926513672, "learning_rate": 5.683870967741936e-06, "loss": 0.317, "step": 13738 }, { "epoch": 88.56546184738956, "grad_norm": 7.311061382293701, "learning_rate": 5.680645161290323e-06, "loss": 0.1385, "step": 13739 }, { "epoch": 88.5718875502008, "grad_norm": 16.373844146728516, "learning_rate": 5.677419354838711e-06, "loss": 0.303, "step": 13740 }, { "epoch": 88.57831325301204, "grad_norm": 12.043750762939453, "learning_rate": 5.674193548387096e-06, "loss": 0.3992, "step": 13741 }, { "epoch": 88.5847389558233, "grad_norm": 7.199093341827393, "learning_rate": 5.670967741935484e-06, "loss": 0.2991, "step": 13742 }, { "epoch": 88.59116465863454, "grad_norm": 5.856044769287109, "learning_rate": 5.667741935483871e-06, "loss": 0.2193, "step": 13743 }, { "epoch": 88.59759036144578, "grad_norm": 15.639477729797363, "learning_rate": 5.664516129032258e-06, "loss": 0.419, "step": 13744 }, { "epoch": 88.60401606425702, "grad_norm": 4.876067638397217, "learning_rate": 5.661290322580645e-06, "loss": 0.2504, "step": 13745 }, { "epoch": 88.61044176706827, "grad_norm": 7.109385967254639, "learning_rate": 5.6580645161290325e-06, "loss": 0.2861, "step": 13746 }, { "epoch": 88.61686746987952, "grad_norm": 9.75068473815918, "learning_rate": 5.654838709677419e-06, "loss": 0.3308, "step": 13747 }, { "epoch": 88.62329317269077, "grad_norm": 13.014674186706543, "learning_rate": 5.651612903225806e-06, "loss": 0.4088, "step": 13748 }, { "epoch": 88.62971887550201, "grad_norm": 2.6176297664642334, "learning_rate": 5.648387096774194e-06, "loss": 0.1218, "step": 13749 }, { "epoch": 88.63614457831325, "grad_norm": 9.733858108520508, "learning_rate": 5.64516129032258e-06, "loss": 0.4602, "step": 13750 }, { "epoch": 88.64257028112449, "grad_norm": 6.119353294372559, "learning_rate": 5.641935483870968e-06, "loss": 0.3395, "step": 13751 }, { "epoch": 88.64899598393575, "grad_norm": 11.755963325500488, "learning_rate": 5.638709677419355e-06, "loss": 0.8102, "step": 13752 }, { "epoch": 88.65542168674699, "grad_norm": 12.045194625854492, "learning_rate": 5.635483870967742e-06, "loss": 0.5015, "step": 13753 }, { "epoch": 88.66184738955823, "grad_norm": 8.529099464416504, "learning_rate": 5.632258064516129e-06, "loss": 0.2174, "step": 13754 }, { "epoch": 88.66827309236947, "grad_norm": 19.661989212036133, "learning_rate": 5.6290322580645165e-06, "loss": 0.3396, "step": 13755 }, { "epoch": 88.67469879518072, "grad_norm": 10.769085884094238, "learning_rate": 5.625806451612903e-06, "loss": 0.4915, "step": 13756 }, { "epoch": 88.68112449799197, "grad_norm": 10.584556579589844, "learning_rate": 5.6225806451612905e-06, "loss": 0.2306, "step": 13757 }, { "epoch": 88.68755020080322, "grad_norm": 12.352609634399414, "learning_rate": 5.619354838709678e-06, "loss": 0.3188, "step": 13758 }, { "epoch": 88.69397590361446, "grad_norm": 22.864349365234375, "learning_rate": 5.6161290322580644e-06, "loss": 0.4119, "step": 13759 }, { "epoch": 88.7004016064257, "grad_norm": 5.443256378173828, "learning_rate": 5.612903225806452e-06, "loss": 0.185, "step": 13760 }, { "epoch": 88.70682730923694, "grad_norm": 14.995648384094238, "learning_rate": 5.609677419354839e-06, "loss": 0.3446, "step": 13761 }, { "epoch": 88.7132530120482, "grad_norm": 4.993779182434082, "learning_rate": 5.606451612903226e-06, "loss": 0.2536, "step": 13762 }, { "epoch": 88.71967871485944, "grad_norm": 9.435098648071289, "learning_rate": 5.603225806451613e-06, "loss": 0.3799, "step": 13763 }, { "epoch": 88.72610441767068, "grad_norm": 12.408164978027344, "learning_rate": 5.600000000000001e-06, "loss": 0.2323, "step": 13764 }, { "epoch": 88.73253012048193, "grad_norm": 17.216983795166016, "learning_rate": 5.596774193548387e-06, "loss": 0.3223, "step": 13765 }, { "epoch": 88.73895582329317, "grad_norm": 6.088422775268555, "learning_rate": 5.5935483870967745e-06, "loss": 0.279, "step": 13766 }, { "epoch": 88.74538152610442, "grad_norm": 14.260333061218262, "learning_rate": 5.590322580645162e-06, "loss": 0.7957, "step": 13767 }, { "epoch": 88.75180722891567, "grad_norm": 6.480499267578125, "learning_rate": 5.5870967741935485e-06, "loss": 0.2681, "step": 13768 }, { "epoch": 88.75823293172691, "grad_norm": 6.9371819496154785, "learning_rate": 5.583870967741936e-06, "loss": 0.3376, "step": 13769 }, { "epoch": 88.76465863453815, "grad_norm": 8.385942459106445, "learning_rate": 5.580645161290323e-06, "loss": 0.189, "step": 13770 }, { "epoch": 88.7710843373494, "grad_norm": 8.637503623962402, "learning_rate": 5.57741935483871e-06, "loss": 0.5252, "step": 13771 }, { "epoch": 88.77751004016065, "grad_norm": 17.071290969848633, "learning_rate": 5.574193548387097e-06, "loss": 0.4929, "step": 13772 }, { "epoch": 88.78393574297189, "grad_norm": 4.541753768920898, "learning_rate": 5.570967741935485e-06, "loss": 0.1956, "step": 13773 }, { "epoch": 88.79036144578313, "grad_norm": 7.252988338470459, "learning_rate": 5.567741935483871e-06, "loss": 0.3019, "step": 13774 }, { "epoch": 88.79678714859438, "grad_norm": 8.307382583618164, "learning_rate": 5.564516129032259e-06, "loss": 0.2493, "step": 13775 }, { "epoch": 88.80321285140562, "grad_norm": 10.809736251831055, "learning_rate": 5.561290322580645e-06, "loss": 0.263, "step": 13776 }, { "epoch": 88.80963855421686, "grad_norm": 10.989568710327148, "learning_rate": 5.5580645161290326e-06, "loss": 0.3534, "step": 13777 }, { "epoch": 88.81606425702812, "grad_norm": 8.818822860717773, "learning_rate": 5.554838709677419e-06, "loss": 0.3323, "step": 13778 }, { "epoch": 88.82248995983936, "grad_norm": 15.241700172424316, "learning_rate": 5.5516129032258065e-06, "loss": 0.3626, "step": 13779 }, { "epoch": 88.8289156626506, "grad_norm": 15.915928840637207, "learning_rate": 5.548387096774194e-06, "loss": 0.276, "step": 13780 }, { "epoch": 88.83534136546184, "grad_norm": 6.1391754150390625, "learning_rate": 5.5451612903225805e-06, "loss": 0.2395, "step": 13781 }, { "epoch": 88.84176706827309, "grad_norm": 13.804009437561035, "learning_rate": 5.541935483870968e-06, "loss": 0.356, "step": 13782 }, { "epoch": 88.84819277108434, "grad_norm": 9.342731475830078, "learning_rate": 5.538709677419355e-06, "loss": 0.4329, "step": 13783 }, { "epoch": 88.85461847389558, "grad_norm": 5.377484321594238, "learning_rate": 5.535483870967742e-06, "loss": 0.137, "step": 13784 }, { "epoch": 88.86104417670683, "grad_norm": 17.032690048217773, "learning_rate": 5.532258064516129e-06, "loss": 0.2545, "step": 13785 }, { "epoch": 88.86746987951807, "grad_norm": 5.938132286071777, "learning_rate": 5.529032258064517e-06, "loss": 0.3159, "step": 13786 }, { "epoch": 88.87389558232931, "grad_norm": 16.42367935180664, "learning_rate": 5.525806451612903e-06, "loss": 0.4026, "step": 13787 }, { "epoch": 88.88032128514057, "grad_norm": 11.633358001708984, "learning_rate": 5.522580645161291e-06, "loss": 0.3277, "step": 13788 }, { "epoch": 88.88674698795181, "grad_norm": 12.311053276062012, "learning_rate": 5.519354838709678e-06, "loss": 0.5398, "step": 13789 }, { "epoch": 88.89317269076305, "grad_norm": 7.010092258453369, "learning_rate": 5.5161290322580645e-06, "loss": 0.2892, "step": 13790 }, { "epoch": 88.8995983935743, "grad_norm": 10.096823692321777, "learning_rate": 5.512903225806452e-06, "loss": 0.3135, "step": 13791 }, { "epoch": 88.90602409638554, "grad_norm": 10.391105651855469, "learning_rate": 5.509677419354839e-06, "loss": 0.5201, "step": 13792 }, { "epoch": 88.91244979919679, "grad_norm": 11.024693489074707, "learning_rate": 5.506451612903226e-06, "loss": 0.4111, "step": 13793 }, { "epoch": 88.91887550200804, "grad_norm": 16.847272872924805, "learning_rate": 5.503225806451613e-06, "loss": 0.3279, "step": 13794 }, { "epoch": 88.92530120481928, "grad_norm": 11.445440292358398, "learning_rate": 5.500000000000001e-06, "loss": 0.6149, "step": 13795 }, { "epoch": 88.93172690763052, "grad_norm": 12.178791999816895, "learning_rate": 5.496774193548387e-06, "loss": 0.4161, "step": 13796 }, { "epoch": 88.93815261044176, "grad_norm": 12.89630126953125, "learning_rate": 5.493548387096775e-06, "loss": 0.4263, "step": 13797 }, { "epoch": 88.94457831325302, "grad_norm": 14.787447929382324, "learning_rate": 5.490322580645162e-06, "loss": 0.232, "step": 13798 }, { "epoch": 88.95100401606426, "grad_norm": 3.8669581413269043, "learning_rate": 5.487096774193549e-06, "loss": 0.1207, "step": 13799 }, { "epoch": 88.9574297188755, "grad_norm": 10.775385856628418, "learning_rate": 5.483870967741936e-06, "loss": 0.408, "step": 13800 }, { "epoch": 88.96385542168674, "grad_norm": 7.583415985107422, "learning_rate": 5.4806451612903226e-06, "loss": 0.2518, "step": 13801 }, { "epoch": 88.97028112449799, "grad_norm": 13.976825714111328, "learning_rate": 5.47741935483871e-06, "loss": 0.4349, "step": 13802 }, { "epoch": 88.97670682730924, "grad_norm": 6.276492118835449, "learning_rate": 5.474193548387097e-06, "loss": 0.3024, "step": 13803 }, { "epoch": 88.98313253012049, "grad_norm": 7.852107524871826, "learning_rate": 5.470967741935484e-06, "loss": 0.234, "step": 13804 }, { "epoch": 88.98955823293173, "grad_norm": 25.95674705505371, "learning_rate": 5.467741935483871e-06, "loss": 0.2074, "step": 13805 }, { "epoch": 88.99598393574297, "grad_norm": 10.988481521606445, "learning_rate": 5.464516129032259e-06, "loss": 0.1751, "step": 13806 }, { "epoch": 89.00562248995983, "grad_norm": 9.404327392578125, "learning_rate": 5.461290322580645e-06, "loss": 0.2096, "step": 13807 }, { "epoch": 89.01204819277109, "grad_norm": 17.108694076538086, "learning_rate": 5.458064516129033e-06, "loss": 0.229, "step": 13808 }, { "epoch": 89.01847389558233, "grad_norm": 23.933542251586914, "learning_rate": 5.45483870967742e-06, "loss": 0.4499, "step": 13809 }, { "epoch": 89.02489959839357, "grad_norm": 8.829254150390625, "learning_rate": 5.451612903225807e-06, "loss": 0.1972, "step": 13810 }, { "epoch": 89.03132530120482, "grad_norm": 25.9807186126709, "learning_rate": 5.448387096774194e-06, "loss": 0.6087, "step": 13811 }, { "epoch": 89.03775100401606, "grad_norm": 9.408805847167969, "learning_rate": 5.445161290322581e-06, "loss": 0.134, "step": 13812 }, { "epoch": 89.04417670682732, "grad_norm": 39.250118255615234, "learning_rate": 5.441935483870968e-06, "loss": 0.4408, "step": 13813 }, { "epoch": 89.05060240963856, "grad_norm": 31.03643226623535, "learning_rate": 5.4387096774193545e-06, "loss": 0.2108, "step": 13814 }, { "epoch": 89.0570281124498, "grad_norm": 17.42746353149414, "learning_rate": 5.435483870967742e-06, "loss": 0.4388, "step": 13815 }, { "epoch": 89.06345381526104, "grad_norm": 13.627771377563477, "learning_rate": 5.432258064516129e-06, "loss": 0.6221, "step": 13816 }, { "epoch": 89.06987951807228, "grad_norm": 29.215669631958008, "learning_rate": 5.429032258064516e-06, "loss": 0.515, "step": 13817 }, { "epoch": 89.07630522088354, "grad_norm": 19.37889289855957, "learning_rate": 5.425806451612903e-06, "loss": 0.3473, "step": 13818 }, { "epoch": 89.08273092369478, "grad_norm": 4.659404277801514, "learning_rate": 5.422580645161291e-06, "loss": 0.1031, "step": 13819 }, { "epoch": 89.08915662650602, "grad_norm": 23.05350112915039, "learning_rate": 5.419354838709677e-06, "loss": 0.4901, "step": 13820 }, { "epoch": 89.09558232931727, "grad_norm": 10.334497451782227, "learning_rate": 5.416129032258065e-06, "loss": 0.1962, "step": 13821 }, { "epoch": 89.10200803212851, "grad_norm": 13.521551132202148, "learning_rate": 5.412903225806452e-06, "loss": 0.4763, "step": 13822 }, { "epoch": 89.10843373493977, "grad_norm": 8.116397857666016, "learning_rate": 5.409677419354839e-06, "loss": 0.5319, "step": 13823 }, { "epoch": 89.11485943775101, "grad_norm": 6.236509799957275, "learning_rate": 5.406451612903226e-06, "loss": 0.1378, "step": 13824 }, { "epoch": 89.12128514056225, "grad_norm": 23.850149154663086, "learning_rate": 5.403225806451613e-06, "loss": 0.2395, "step": 13825 }, { "epoch": 89.12771084337349, "grad_norm": 9.581583976745605, "learning_rate": 5.4e-06, "loss": 0.2271, "step": 13826 }, { "epoch": 89.13413654618473, "grad_norm": 17.641159057617188, "learning_rate": 5.396774193548387e-06, "loss": 0.6998, "step": 13827 }, { "epoch": 89.14056224899599, "grad_norm": 1.698013186454773, "learning_rate": 5.393548387096775e-06, "loss": 0.2339, "step": 13828 }, { "epoch": 89.14698795180723, "grad_norm": 33.53544235229492, "learning_rate": 5.390322580645161e-06, "loss": 0.2985, "step": 13829 }, { "epoch": 89.15341365461848, "grad_norm": 5.019949913024902, "learning_rate": 5.387096774193549e-06, "loss": 0.3414, "step": 13830 }, { "epoch": 89.15983935742972, "grad_norm": 21.279159545898438, "learning_rate": 5.383870967741936e-06, "loss": 0.3437, "step": 13831 }, { "epoch": 89.16626506024096, "grad_norm": 21.73443603515625, "learning_rate": 5.380645161290323e-06, "loss": 0.5288, "step": 13832 }, { "epoch": 89.1726907630522, "grad_norm": 5.879932880401611, "learning_rate": 5.37741935483871e-06, "loss": 0.2373, "step": 13833 }, { "epoch": 89.17911646586346, "grad_norm": 5.052789688110352, "learning_rate": 5.3741935483870975e-06, "loss": 0.3923, "step": 13834 }, { "epoch": 89.1855421686747, "grad_norm": 23.82978057861328, "learning_rate": 5.370967741935484e-06, "loss": 0.5529, "step": 13835 }, { "epoch": 89.19196787148594, "grad_norm": 7.414181232452393, "learning_rate": 5.3677419354838714e-06, "loss": 0.2108, "step": 13836 }, { "epoch": 89.19839357429719, "grad_norm": 11.856797218322754, "learning_rate": 5.364516129032259e-06, "loss": 0.3005, "step": 13837 }, { "epoch": 89.20481927710843, "grad_norm": 3.6761980056762695, "learning_rate": 5.361290322580645e-06, "loss": 0.279, "step": 13838 }, { "epoch": 89.21124497991968, "grad_norm": 9.957139015197754, "learning_rate": 5.358064516129033e-06, "loss": 0.1953, "step": 13839 }, { "epoch": 89.21767068273093, "grad_norm": 13.068130493164062, "learning_rate": 5.35483870967742e-06, "loss": 0.1696, "step": 13840 }, { "epoch": 89.22409638554217, "grad_norm": 2.6888327598571777, "learning_rate": 5.351612903225807e-06, "loss": 0.2082, "step": 13841 }, { "epoch": 89.23052208835341, "grad_norm": 3.099316358566284, "learning_rate": 5.348387096774194e-06, "loss": 0.259, "step": 13842 }, { "epoch": 89.23694779116465, "grad_norm": 16.86345863342285, "learning_rate": 5.3451612903225815e-06, "loss": 0.5236, "step": 13843 }, { "epoch": 89.24337349397591, "grad_norm": 20.914384841918945, "learning_rate": 5.341935483870968e-06, "loss": 0.6007, "step": 13844 }, { "epoch": 89.24979919678715, "grad_norm": 6.168158054351807, "learning_rate": 5.3387096774193555e-06, "loss": 0.4607, "step": 13845 }, { "epoch": 89.2562248995984, "grad_norm": 17.725603103637695, "learning_rate": 5.335483870967743e-06, "loss": 0.6805, "step": 13846 }, { "epoch": 89.26265060240964, "grad_norm": 2.654547691345215, "learning_rate": 5.332258064516129e-06, "loss": 0.157, "step": 13847 }, { "epoch": 89.26907630522088, "grad_norm": 17.47589111328125, "learning_rate": 5.329032258064516e-06, "loss": 0.2386, "step": 13848 }, { "epoch": 89.27550200803213, "grad_norm": 18.50897979736328, "learning_rate": 5.325806451612903e-06, "loss": 0.2076, "step": 13849 }, { "epoch": 89.28192771084338, "grad_norm": 19.359472274780273, "learning_rate": 5.32258064516129e-06, "loss": 0.2844, "step": 13850 }, { "epoch": 89.28835341365462, "grad_norm": 12.508661270141602, "learning_rate": 5.319354838709677e-06, "loss": 0.3336, "step": 13851 }, { "epoch": 89.29477911646586, "grad_norm": 2.0557353496551514, "learning_rate": 5.316129032258065e-06, "loss": 0.3435, "step": 13852 }, { "epoch": 89.3012048192771, "grad_norm": 19.317228317260742, "learning_rate": 5.312903225806451e-06, "loss": 0.5262, "step": 13853 }, { "epoch": 89.30763052208836, "grad_norm": 11.493136405944824, "learning_rate": 5.309677419354839e-06, "loss": 0.3703, "step": 13854 }, { "epoch": 89.3140562248996, "grad_norm": 16.10757064819336, "learning_rate": 5.306451612903226e-06, "loss": 0.3552, "step": 13855 }, { "epoch": 89.32048192771084, "grad_norm": 4.583131313323975, "learning_rate": 5.303225806451613e-06, "loss": 0.3994, "step": 13856 }, { "epoch": 89.32690763052209, "grad_norm": 27.83868408203125, "learning_rate": 5.3e-06, "loss": 0.2468, "step": 13857 }, { "epoch": 89.33333333333333, "grad_norm": 40.16547775268555, "learning_rate": 5.2967741935483875e-06, "loss": 0.5709, "step": 13858 }, { "epoch": 89.33975903614459, "grad_norm": 28.63479232788086, "learning_rate": 5.293548387096774e-06, "loss": 0.4873, "step": 13859 }, { "epoch": 89.34618473895583, "grad_norm": 13.433250427246094, "learning_rate": 5.2903225806451614e-06, "loss": 0.4448, "step": 13860 }, { "epoch": 89.35261044176707, "grad_norm": 8.292299270629883, "learning_rate": 5.287096774193549e-06, "loss": 0.1299, "step": 13861 }, { "epoch": 89.35903614457831, "grad_norm": 8.239768028259277, "learning_rate": 5.283870967741935e-06, "loss": 0.1727, "step": 13862 }, { "epoch": 89.36546184738955, "grad_norm": 15.343035697937012, "learning_rate": 5.280645161290323e-06, "loss": 0.2885, "step": 13863 }, { "epoch": 89.3718875502008, "grad_norm": 5.6545610427856445, "learning_rate": 5.27741935483871e-06, "loss": 0.2654, "step": 13864 }, { "epoch": 89.37831325301205, "grad_norm": 7.685011386871338, "learning_rate": 5.274193548387097e-06, "loss": 0.1784, "step": 13865 }, { "epoch": 89.3847389558233, "grad_norm": 24.966182708740234, "learning_rate": 5.270967741935484e-06, "loss": 0.1932, "step": 13866 }, { "epoch": 89.39116465863454, "grad_norm": 10.639724731445312, "learning_rate": 5.2677419354838715e-06, "loss": 0.2256, "step": 13867 }, { "epoch": 89.39759036144578, "grad_norm": 3.101841449737549, "learning_rate": 5.264516129032258e-06, "loss": 0.4148, "step": 13868 }, { "epoch": 89.40401606425702, "grad_norm": 7.797054767608643, "learning_rate": 5.2612903225806455e-06, "loss": 0.55, "step": 13869 }, { "epoch": 89.41044176706828, "grad_norm": 21.097505569458008, "learning_rate": 5.258064516129033e-06, "loss": 0.4021, "step": 13870 }, { "epoch": 89.41686746987952, "grad_norm": 19.36062240600586, "learning_rate": 5.2548387096774195e-06, "loss": 0.6524, "step": 13871 }, { "epoch": 89.42329317269076, "grad_norm": 19.14179039001465, "learning_rate": 5.251612903225807e-06, "loss": 0.3493, "step": 13872 }, { "epoch": 89.429718875502, "grad_norm": 21.968345642089844, "learning_rate": 5.248387096774194e-06, "loss": 0.4638, "step": 13873 }, { "epoch": 89.43614457831325, "grad_norm": 3.359642505645752, "learning_rate": 5.245161290322581e-06, "loss": 0.3845, "step": 13874 }, { "epoch": 89.4425702811245, "grad_norm": 1.9314626455307007, "learning_rate": 5.241935483870968e-06, "loss": 0.1522, "step": 13875 }, { "epoch": 89.44899598393575, "grad_norm": 19.323745727539062, "learning_rate": 5.238709677419356e-06, "loss": 0.1241, "step": 13876 }, { "epoch": 89.45542168674699, "grad_norm": 10.26850700378418, "learning_rate": 5.235483870967742e-06, "loss": 0.0944, "step": 13877 }, { "epoch": 89.46184738955823, "grad_norm": 5.705228328704834, "learning_rate": 5.2322580645161296e-06, "loss": 0.2271, "step": 13878 }, { "epoch": 89.46827309236947, "grad_norm": 6.018062591552734, "learning_rate": 5.229032258064517e-06, "loss": 0.3384, "step": 13879 }, { "epoch": 89.47469879518073, "grad_norm": 20.619457244873047, "learning_rate": 5.2258064516129035e-06, "loss": 0.2952, "step": 13880 }, { "epoch": 89.48112449799197, "grad_norm": 12.788833618164062, "learning_rate": 5.222580645161291e-06, "loss": 0.3806, "step": 13881 }, { "epoch": 89.48755020080321, "grad_norm": 7.434150218963623, "learning_rate": 5.219354838709678e-06, "loss": 0.3201, "step": 13882 }, { "epoch": 89.49397590361446, "grad_norm": 16.01243782043457, "learning_rate": 5.216129032258064e-06, "loss": 0.7019, "step": 13883 }, { "epoch": 89.5004016064257, "grad_norm": 20.121973037719727, "learning_rate": 5.2129032258064514e-06, "loss": 0.1598, "step": 13884 }, { "epoch": 89.50682730923695, "grad_norm": 16.653018951416016, "learning_rate": 5.209677419354839e-06, "loss": 0.5713, "step": 13885 }, { "epoch": 89.5132530120482, "grad_norm": 32.797203063964844, "learning_rate": 5.206451612903225e-06, "loss": 0.4063, "step": 13886 }, { "epoch": 89.51967871485944, "grad_norm": 8.571011543273926, "learning_rate": 5.203225806451613e-06, "loss": 0.3347, "step": 13887 }, { "epoch": 89.52610441767068, "grad_norm": 9.303726196289062, "learning_rate": 5.2e-06, "loss": 0.2885, "step": 13888 }, { "epoch": 89.53253012048192, "grad_norm": 16.50235939025879, "learning_rate": 5.196774193548387e-06, "loss": 0.5993, "step": 13889 }, { "epoch": 89.53895582329318, "grad_norm": 4.802011966705322, "learning_rate": 5.193548387096774e-06, "loss": 0.4003, "step": 13890 }, { "epoch": 89.54538152610442, "grad_norm": 1.6000804901123047, "learning_rate": 5.1903225806451615e-06, "loss": 0.3306, "step": 13891 }, { "epoch": 89.55180722891566, "grad_norm": 6.281829833984375, "learning_rate": 5.187096774193548e-06, "loss": 0.2085, "step": 13892 }, { "epoch": 89.5582329317269, "grad_norm": 18.631303787231445, "learning_rate": 5.1838709677419355e-06, "loss": 0.2167, "step": 13893 }, { "epoch": 89.56465863453815, "grad_norm": 29.060001373291016, "learning_rate": 5.180645161290323e-06, "loss": 0.3893, "step": 13894 }, { "epoch": 89.5710843373494, "grad_norm": 16.299705505371094, "learning_rate": 5.1774193548387095e-06, "loss": 0.1982, "step": 13895 }, { "epoch": 89.57751004016065, "grad_norm": 64.01813507080078, "learning_rate": 5.174193548387097e-06, "loss": 0.4095, "step": 13896 }, { "epoch": 89.58393574297189, "grad_norm": 66.6383285522461, "learning_rate": 5.170967741935484e-06, "loss": 0.2911, "step": 13897 }, { "epoch": 89.59036144578313, "grad_norm": 8.802309036254883, "learning_rate": 5.167741935483871e-06, "loss": 0.507, "step": 13898 }, { "epoch": 89.59678714859437, "grad_norm": 64.8519058227539, "learning_rate": 5.164516129032258e-06, "loss": 0.3043, "step": 13899 }, { "epoch": 89.60321285140563, "grad_norm": 38.15950012207031, "learning_rate": 5.161290322580646e-06, "loss": 0.3699, "step": 13900 }, { "epoch": 89.60963855421687, "grad_norm": 21.372467041015625, "learning_rate": 5.158064516129032e-06, "loss": 0.3765, "step": 13901 }, { "epoch": 89.61606425702811, "grad_norm": 4.017727851867676, "learning_rate": 5.1548387096774196e-06, "loss": 0.1729, "step": 13902 }, { "epoch": 89.62248995983936, "grad_norm": 36.544559478759766, "learning_rate": 5.151612903225807e-06, "loss": 0.5393, "step": 13903 }, { "epoch": 89.6289156626506, "grad_norm": 16.547775268554688, "learning_rate": 5.1483870967741935e-06, "loss": 0.4572, "step": 13904 }, { "epoch": 89.63534136546184, "grad_norm": 9.368570327758789, "learning_rate": 5.145161290322581e-06, "loss": 0.3615, "step": 13905 }, { "epoch": 89.6417670682731, "grad_norm": 61.529449462890625, "learning_rate": 5.141935483870968e-06, "loss": 0.4216, "step": 13906 }, { "epoch": 89.64819277108434, "grad_norm": 24.11792755126953, "learning_rate": 5.138709677419355e-06, "loss": 0.56, "step": 13907 }, { "epoch": 89.65461847389558, "grad_norm": 2.644080877304077, "learning_rate": 5.135483870967742e-06, "loss": 0.1505, "step": 13908 }, { "epoch": 89.66104417670682, "grad_norm": 32.47414779663086, "learning_rate": 5.13225806451613e-06, "loss": 0.4223, "step": 13909 }, { "epoch": 89.66746987951807, "grad_norm": 15.668420791625977, "learning_rate": 5.129032258064516e-06, "loss": 0.4866, "step": 13910 }, { "epoch": 89.67389558232932, "grad_norm": 8.502876281738281, "learning_rate": 5.125806451612904e-06, "loss": 0.413, "step": 13911 }, { "epoch": 89.68032128514056, "grad_norm": 3.4488167762756348, "learning_rate": 5.122580645161291e-06, "loss": 0.4607, "step": 13912 }, { "epoch": 89.6867469879518, "grad_norm": 3.4301862716674805, "learning_rate": 5.119354838709678e-06, "loss": 0.1747, "step": 13913 }, { "epoch": 89.69317269076305, "grad_norm": 27.620607376098633, "learning_rate": 5.116129032258065e-06, "loss": 0.2929, "step": 13914 }, { "epoch": 89.69959839357429, "grad_norm": 8.128396034240723, "learning_rate": 5.112903225806452e-06, "loss": 0.2064, "step": 13915 }, { "epoch": 89.70602409638555, "grad_norm": 17.022876739501953, "learning_rate": 5.109677419354839e-06, "loss": 0.4473, "step": 13916 }, { "epoch": 89.71244979919679, "grad_norm": 10.213683128356934, "learning_rate": 5.106451612903226e-06, "loss": 0.1445, "step": 13917 }, { "epoch": 89.71887550200803, "grad_norm": 12.598165512084961, "learning_rate": 5.103225806451614e-06, "loss": 0.3774, "step": 13918 }, { "epoch": 89.72530120481927, "grad_norm": 20.672595977783203, "learning_rate": 5.1e-06, "loss": 0.3843, "step": 13919 }, { "epoch": 89.73172690763052, "grad_norm": 31.400882720947266, "learning_rate": 5.096774193548387e-06, "loss": 0.3403, "step": 13920 }, { "epoch": 89.73815261044177, "grad_norm": 8.797142028808594, "learning_rate": 5.093548387096774e-06, "loss": 0.4226, "step": 13921 }, { "epoch": 89.74457831325302, "grad_norm": 25.183486938476562, "learning_rate": 5.090322580645162e-06, "loss": 0.306, "step": 13922 }, { "epoch": 89.75100401606426, "grad_norm": 8.02104663848877, "learning_rate": 5.087096774193548e-06, "loss": 0.2187, "step": 13923 }, { "epoch": 89.7574297188755, "grad_norm": 26.355581283569336, "learning_rate": 5.083870967741936e-06, "loss": 0.4076, "step": 13924 }, { "epoch": 89.76385542168674, "grad_norm": 11.9255952835083, "learning_rate": 5.080645161290323e-06, "loss": 0.2687, "step": 13925 }, { "epoch": 89.770281124498, "grad_norm": 16.22026824951172, "learning_rate": 5.0774193548387096e-06, "loss": 0.2688, "step": 13926 }, { "epoch": 89.77670682730924, "grad_norm": 9.774345397949219, "learning_rate": 5.074193548387097e-06, "loss": 0.2109, "step": 13927 }, { "epoch": 89.78313253012048, "grad_norm": 14.524986267089844, "learning_rate": 5.070967741935484e-06, "loss": 0.2892, "step": 13928 }, { "epoch": 89.78955823293173, "grad_norm": 64.16797637939453, "learning_rate": 5.067741935483871e-06, "loss": 0.5114, "step": 13929 }, { "epoch": 89.79598393574297, "grad_norm": 8.912425994873047, "learning_rate": 5.064516129032258e-06, "loss": 0.4311, "step": 13930 }, { "epoch": 89.80240963855422, "grad_norm": 37.0676155090332, "learning_rate": 5.061290322580646e-06, "loss": 0.3582, "step": 13931 }, { "epoch": 89.80883534136547, "grad_norm": 21.65011978149414, "learning_rate": 5.058064516129032e-06, "loss": 0.4959, "step": 13932 }, { "epoch": 89.81526104417671, "grad_norm": 27.090295791625977, "learning_rate": 5.05483870967742e-06, "loss": 0.357, "step": 13933 }, { "epoch": 89.82168674698795, "grad_norm": 3.028554677963257, "learning_rate": 5.051612903225807e-06, "loss": 0.122, "step": 13934 }, { "epoch": 89.82811244979919, "grad_norm": 2.214151620864868, "learning_rate": 5.048387096774194e-06, "loss": 0.1496, "step": 13935 }, { "epoch": 89.83453815261043, "grad_norm": 15.824324607849121, "learning_rate": 5.045161290322581e-06, "loss": 0.3532, "step": 13936 }, { "epoch": 89.84096385542169, "grad_norm": 19.985340118408203, "learning_rate": 5.041935483870968e-06, "loss": 0.6049, "step": 13937 }, { "epoch": 89.84738955823293, "grad_norm": 9.276396751403809, "learning_rate": 5.038709677419355e-06, "loss": 0.2452, "step": 13938 }, { "epoch": 89.85381526104418, "grad_norm": 69.00469970703125, "learning_rate": 5.035483870967742e-06, "loss": 0.2672, "step": 13939 }, { "epoch": 89.86024096385542, "grad_norm": 13.269841194152832, "learning_rate": 5.032258064516129e-06, "loss": 0.27, "step": 13940 }, { "epoch": 89.86666666666666, "grad_norm": 3.358319044113159, "learning_rate": 5.029032258064516e-06, "loss": 0.3674, "step": 13941 }, { "epoch": 89.87309236947792, "grad_norm": 17.664749145507812, "learning_rate": 5.025806451612904e-06, "loss": 0.6387, "step": 13942 }, { "epoch": 89.87951807228916, "grad_norm": 15.844600677490234, "learning_rate": 5.02258064516129e-06, "loss": 0.3836, "step": 13943 }, { "epoch": 89.8859437751004, "grad_norm": 18.64333152770996, "learning_rate": 5.019354838709678e-06, "loss": 0.5322, "step": 13944 }, { "epoch": 89.89236947791164, "grad_norm": 11.575469017028809, "learning_rate": 5.016129032258065e-06, "loss": 0.3086, "step": 13945 }, { "epoch": 89.89879518072289, "grad_norm": 24.676280975341797, "learning_rate": 5.012903225806452e-06, "loss": 0.3468, "step": 13946 }, { "epoch": 89.90522088353414, "grad_norm": 1.7984217405319214, "learning_rate": 5.009677419354839e-06, "loss": 0.4048, "step": 13947 }, { "epoch": 89.91164658634538, "grad_norm": 15.379268646240234, "learning_rate": 5.0064516129032265e-06, "loss": 0.4392, "step": 13948 }, { "epoch": 89.91807228915663, "grad_norm": 31.31421661376953, "learning_rate": 5.003225806451613e-06, "loss": 0.6946, "step": 13949 }, { "epoch": 89.92449799196787, "grad_norm": 52.30035400390625, "learning_rate": 5e-06, "loss": 0.4745, "step": 13950 }, { "epoch": 89.93092369477911, "grad_norm": 7.303703308105469, "learning_rate": 4.996774193548388e-06, "loss": 0.1624, "step": 13951 }, { "epoch": 89.93734939759037, "grad_norm": 19.924409866333008, "learning_rate": 4.993548387096774e-06, "loss": 0.3681, "step": 13952 }, { "epoch": 89.94377510040161, "grad_norm": 17.233327865600586, "learning_rate": 4.990322580645162e-06, "loss": 0.3456, "step": 13953 }, { "epoch": 89.95020080321285, "grad_norm": 2.0011024475097656, "learning_rate": 4.987096774193548e-06, "loss": 0.3809, "step": 13954 }, { "epoch": 89.9566265060241, "grad_norm": 32.367183685302734, "learning_rate": 4.983870967741936e-06, "loss": 0.3733, "step": 13955 }, { "epoch": 89.96305220883534, "grad_norm": 19.511016845703125, "learning_rate": 4.980645161290322e-06, "loss": 0.4171, "step": 13956 }, { "epoch": 89.96947791164659, "grad_norm": 12.551291465759277, "learning_rate": 4.97741935483871e-06, "loss": 0.2568, "step": 13957 }, { "epoch": 89.97590361445783, "grad_norm": 11.27829647064209, "learning_rate": 4.974193548387097e-06, "loss": 0.1704, "step": 13958 }, { "epoch": 89.98232931726908, "grad_norm": 14.947091102600098, "learning_rate": 4.970967741935484e-06, "loss": 0.5842, "step": 13959 }, { "epoch": 89.98875502008032, "grad_norm": 3.9024465084075928, "learning_rate": 4.967741935483871e-06, "loss": 0.1429, "step": 13960 }, { "epoch": 89.99518072289156, "grad_norm": 5.722849369049072, "learning_rate": 4.9645161290322584e-06, "loss": 0.3216, "step": 13961 }, { "epoch": 90.00481927710844, "grad_norm": 14.946993827819824, "learning_rate": 4.961290322580645e-06, "loss": 0.2062, "step": 13962 }, { "epoch": 90.01124497991968, "grad_norm": 8.448301315307617, "learning_rate": 4.958064516129032e-06, "loss": 0.3982, "step": 13963 }, { "epoch": 90.01767068273092, "grad_norm": 14.642596244812012, "learning_rate": 4.95483870967742e-06, "loss": 0.3601, "step": 13964 }, { "epoch": 90.02409638554217, "grad_norm": 6.992004871368408, "learning_rate": 4.951612903225806e-06, "loss": 0.2398, "step": 13965 }, { "epoch": 90.03052208835341, "grad_norm": 11.90822982788086, "learning_rate": 4.948387096774194e-06, "loss": 0.4268, "step": 13966 }, { "epoch": 90.03694779116466, "grad_norm": 2.9642558097839355, "learning_rate": 4.945161290322581e-06, "loss": 0.2583, "step": 13967 }, { "epoch": 90.0433734939759, "grad_norm": 15.83539867401123, "learning_rate": 4.941935483870968e-06, "loss": 0.2694, "step": 13968 }, { "epoch": 90.04979919678715, "grad_norm": 21.65538787841797, "learning_rate": 4.938709677419355e-06, "loss": 0.3773, "step": 13969 }, { "epoch": 90.05622489959839, "grad_norm": 16.532493591308594, "learning_rate": 4.9354838709677425e-06, "loss": 0.5098, "step": 13970 }, { "epoch": 90.06265060240963, "grad_norm": 28.676362991333008, "learning_rate": 4.932258064516129e-06, "loss": 0.4363, "step": 13971 }, { "epoch": 90.06907630522089, "grad_norm": 10.619226455688477, "learning_rate": 4.9290322580645165e-06, "loss": 0.2327, "step": 13972 }, { "epoch": 90.07550200803213, "grad_norm": 45.59968185424805, "learning_rate": 4.925806451612904e-06, "loss": 0.4969, "step": 13973 }, { "epoch": 90.08192771084337, "grad_norm": 29.015708923339844, "learning_rate": 4.92258064516129e-06, "loss": 0.2818, "step": 13974 }, { "epoch": 90.08835341365462, "grad_norm": 20.480154037475586, "learning_rate": 4.919354838709678e-06, "loss": 0.6778, "step": 13975 }, { "epoch": 90.09477911646586, "grad_norm": 20.661876678466797, "learning_rate": 4.916129032258065e-06, "loss": 0.2054, "step": 13976 }, { "epoch": 90.10120481927711, "grad_norm": 9.077679634094238, "learning_rate": 4.912903225806452e-06, "loss": 0.4247, "step": 13977 }, { "epoch": 90.10763052208836, "grad_norm": 15.917296409606934, "learning_rate": 4.909677419354839e-06, "loss": 0.1953, "step": 13978 }, { "epoch": 90.1140562248996, "grad_norm": 3.2394139766693115, "learning_rate": 4.9064516129032266e-06, "loss": 0.2028, "step": 13979 }, { "epoch": 90.12048192771084, "grad_norm": 5.829947471618652, "learning_rate": 4.903225806451613e-06, "loss": 0.1992, "step": 13980 }, { "epoch": 90.12690763052208, "grad_norm": 3.098546266555786, "learning_rate": 4.9000000000000005e-06, "loss": 0.2869, "step": 13981 }, { "epoch": 90.13333333333334, "grad_norm": 22.07428550720215, "learning_rate": 4.896774193548388e-06, "loss": 0.4124, "step": 13982 }, { "epoch": 90.13975903614458, "grad_norm": 10.506928443908691, "learning_rate": 4.8935483870967745e-06, "loss": 0.445, "step": 13983 }, { "epoch": 90.14618473895582, "grad_norm": 11.592673301696777, "learning_rate": 4.890322580645162e-06, "loss": 0.3007, "step": 13984 }, { "epoch": 90.15261044176707, "grad_norm": 33.270660400390625, "learning_rate": 4.887096774193549e-06, "loss": 0.3061, "step": 13985 }, { "epoch": 90.15903614457831, "grad_norm": 10.916335105895996, "learning_rate": 4.883870967741936e-06, "loss": 0.2051, "step": 13986 }, { "epoch": 90.16546184738957, "grad_norm": 28.889678955078125, "learning_rate": 4.880645161290323e-06, "loss": 0.1809, "step": 13987 }, { "epoch": 90.17188755020081, "grad_norm": 4.863593578338623, "learning_rate": 4.877419354838711e-06, "loss": 0.3059, "step": 13988 }, { "epoch": 90.17831325301205, "grad_norm": 7.0752177238464355, "learning_rate": 4.874193548387097e-06, "loss": 0.2691, "step": 13989 }, { "epoch": 90.18473895582329, "grad_norm": 24.904033660888672, "learning_rate": 4.870967741935484e-06, "loss": 0.3491, "step": 13990 }, { "epoch": 90.19116465863453, "grad_norm": 7.811056613922119, "learning_rate": 4.867741935483871e-06, "loss": 0.3195, "step": 13991 }, { "epoch": 90.19759036144578, "grad_norm": 4.42833948135376, "learning_rate": 4.864516129032258e-06, "loss": 0.3214, "step": 13992 }, { "epoch": 90.20401606425703, "grad_norm": 3.3430678844451904, "learning_rate": 4.861290322580645e-06, "loss": 0.215, "step": 13993 }, { "epoch": 90.21044176706827, "grad_norm": 13.672191619873047, "learning_rate": 4.8580645161290325e-06, "loss": 0.1823, "step": 13994 }, { "epoch": 90.21686746987952, "grad_norm": 46.13875198364258, "learning_rate": 4.854838709677419e-06, "loss": 0.3504, "step": 13995 }, { "epoch": 90.22329317269076, "grad_norm": 2.832850694656372, "learning_rate": 4.8516129032258065e-06, "loss": 0.2097, "step": 13996 }, { "epoch": 90.229718875502, "grad_norm": 11.1462984085083, "learning_rate": 4.848387096774194e-06, "loss": 0.1002, "step": 13997 }, { "epoch": 90.23614457831326, "grad_norm": 18.404403686523438, "learning_rate": 4.84516129032258e-06, "loss": 0.5218, "step": 13998 }, { "epoch": 90.2425702811245, "grad_norm": 17.281829833984375, "learning_rate": 4.841935483870968e-06, "loss": 0.1956, "step": 13999 }, { "epoch": 90.24899598393574, "grad_norm": 6.286489963531494, "learning_rate": 4.838709677419355e-06, "loss": 0.3042, "step": 14000 }, { "epoch": 90.25542168674698, "grad_norm": 35.16960144042969, "learning_rate": 4.835483870967742e-06, "loss": 0.3464, "step": 14001 }, { "epoch": 90.26184738955823, "grad_norm": 8.339170455932617, "learning_rate": 4.832258064516129e-06, "loss": 0.2373, "step": 14002 }, { "epoch": 90.26827309236948, "grad_norm": 6.3005523681640625, "learning_rate": 4.8290322580645166e-06, "loss": 0.2675, "step": 14003 }, { "epoch": 90.27469879518073, "grad_norm": 96.38497924804688, "learning_rate": 4.825806451612903e-06, "loss": 0.4375, "step": 14004 }, { "epoch": 90.28112449799197, "grad_norm": 15.437843322753906, "learning_rate": 4.8225806451612905e-06, "loss": 0.2022, "step": 14005 }, { "epoch": 90.28755020080321, "grad_norm": 33.73817443847656, "learning_rate": 4.819354838709678e-06, "loss": 0.3829, "step": 14006 }, { "epoch": 90.29397590361445, "grad_norm": 9.206100463867188, "learning_rate": 4.8161290322580645e-06, "loss": 0.2925, "step": 14007 }, { "epoch": 90.30040160642571, "grad_norm": 6.379879951477051, "learning_rate": 4.812903225806452e-06, "loss": 0.2078, "step": 14008 }, { "epoch": 90.30682730923695, "grad_norm": 3.8442227840423584, "learning_rate": 4.809677419354839e-06, "loss": 0.2924, "step": 14009 }, { "epoch": 90.3132530120482, "grad_norm": 5.406546592712402, "learning_rate": 4.806451612903226e-06, "loss": 0.0798, "step": 14010 }, { "epoch": 90.31967871485944, "grad_norm": 6.84535026550293, "learning_rate": 4.803225806451613e-06, "loss": 0.1499, "step": 14011 }, { "epoch": 90.32610441767068, "grad_norm": 5.8504791259765625, "learning_rate": 4.800000000000001e-06, "loss": 0.4445, "step": 14012 }, { "epoch": 90.33253012048193, "grad_norm": 19.279630661010742, "learning_rate": 4.796774193548387e-06, "loss": 0.5224, "step": 14013 }, { "epoch": 90.33895582329318, "grad_norm": 2.927504301071167, "learning_rate": 4.793548387096775e-06, "loss": 0.2172, "step": 14014 }, { "epoch": 90.34538152610442, "grad_norm": 14.376469612121582, "learning_rate": 4.790322580645162e-06, "loss": 0.3682, "step": 14015 }, { "epoch": 90.35180722891566, "grad_norm": 20.584190368652344, "learning_rate": 4.7870967741935485e-06, "loss": 0.2796, "step": 14016 }, { "epoch": 90.3582329317269, "grad_norm": 3.119007110595703, "learning_rate": 4.783870967741936e-06, "loss": 0.3236, "step": 14017 }, { "epoch": 90.36465863453816, "grad_norm": 28.438447952270508, "learning_rate": 4.780645161290323e-06, "loss": 0.5157, "step": 14018 }, { "epoch": 90.3710843373494, "grad_norm": 29.027816772460938, "learning_rate": 4.77741935483871e-06, "loss": 0.4216, "step": 14019 }, { "epoch": 90.37751004016064, "grad_norm": 16.979230880737305, "learning_rate": 4.774193548387097e-06, "loss": 0.4975, "step": 14020 }, { "epoch": 90.38393574297189, "grad_norm": 4.983028888702393, "learning_rate": 4.770967741935485e-06, "loss": 0.3428, "step": 14021 }, { "epoch": 90.39036144578313, "grad_norm": 15.210982322692871, "learning_rate": 4.767741935483871e-06, "loss": 0.1619, "step": 14022 }, { "epoch": 90.39678714859438, "grad_norm": 9.958874702453613, "learning_rate": 4.764516129032259e-06, "loss": 0.2349, "step": 14023 }, { "epoch": 90.40321285140563, "grad_norm": 5.906524658203125, "learning_rate": 4.761290322580646e-06, "loss": 0.1982, "step": 14024 }, { "epoch": 90.40963855421687, "grad_norm": 15.672133445739746, "learning_rate": 4.758064516129032e-06, "loss": 0.4162, "step": 14025 }, { "epoch": 90.41606425702811, "grad_norm": 13.470648765563965, "learning_rate": 4.754838709677419e-06, "loss": 0.338, "step": 14026 }, { "epoch": 90.42248995983935, "grad_norm": 6.232745170593262, "learning_rate": 4.7516129032258066e-06, "loss": 0.273, "step": 14027 }, { "epoch": 90.4289156626506, "grad_norm": 12.586783409118652, "learning_rate": 4.748387096774193e-06, "loss": 0.1408, "step": 14028 }, { "epoch": 90.43534136546185, "grad_norm": 7.653289794921875, "learning_rate": 4.7451612903225805e-06, "loss": 0.3598, "step": 14029 }, { "epoch": 90.4417670682731, "grad_norm": 10.41759204864502, "learning_rate": 4.741935483870968e-06, "loss": 0.1464, "step": 14030 }, { "epoch": 90.44819277108434, "grad_norm": 16.712915420532227, "learning_rate": 4.7387096774193545e-06, "loss": 0.5598, "step": 14031 }, { "epoch": 90.45461847389558, "grad_norm": 20.647554397583008, "learning_rate": 4.735483870967742e-06, "loss": 0.2626, "step": 14032 }, { "epoch": 90.46104417670682, "grad_norm": 21.031240463256836, "learning_rate": 4.732258064516129e-06, "loss": 0.5268, "step": 14033 }, { "epoch": 90.46746987951808, "grad_norm": 101.02944946289062, "learning_rate": 4.729032258064516e-06, "loss": 0.4824, "step": 14034 }, { "epoch": 90.47389558232932, "grad_norm": 14.444235801696777, "learning_rate": 4.725806451612903e-06, "loss": 0.156, "step": 14035 }, { "epoch": 90.48032128514056, "grad_norm": 28.236724853515625, "learning_rate": 4.722580645161291e-06, "loss": 0.3121, "step": 14036 }, { "epoch": 90.4867469879518, "grad_norm": 6.349595069885254, "learning_rate": 4.719354838709677e-06, "loss": 0.1672, "step": 14037 }, { "epoch": 90.49317269076305, "grad_norm": 5.9939727783203125, "learning_rate": 4.716129032258065e-06, "loss": 0.2901, "step": 14038 }, { "epoch": 90.4995983935743, "grad_norm": 30.320528030395508, "learning_rate": 4.712903225806452e-06, "loss": 0.4274, "step": 14039 }, { "epoch": 90.50602409638554, "grad_norm": 16.803194046020508, "learning_rate": 4.7096774193548385e-06, "loss": 0.3228, "step": 14040 }, { "epoch": 90.51244979919679, "grad_norm": 20.250635147094727, "learning_rate": 4.706451612903226e-06, "loss": 0.4776, "step": 14041 }, { "epoch": 90.51887550200803, "grad_norm": 23.55064582824707, "learning_rate": 4.703225806451613e-06, "loss": 0.6075, "step": 14042 }, { "epoch": 90.52530120481927, "grad_norm": 9.697667121887207, "learning_rate": 4.7e-06, "loss": 0.1249, "step": 14043 }, { "epoch": 90.53172690763053, "grad_norm": 15.777759552001953, "learning_rate": 4.696774193548387e-06, "loss": 0.348, "step": 14044 }, { "epoch": 90.53815261044177, "grad_norm": 38.4803466796875, "learning_rate": 4.693548387096775e-06, "loss": 0.409, "step": 14045 }, { "epoch": 90.54457831325301, "grad_norm": 18.6412353515625, "learning_rate": 4.690322580645161e-06, "loss": 0.526, "step": 14046 }, { "epoch": 90.55100401606425, "grad_norm": 24.43244171142578, "learning_rate": 4.687096774193549e-06, "loss": 0.3091, "step": 14047 }, { "epoch": 90.5574297188755, "grad_norm": 9.426898002624512, "learning_rate": 4.683870967741936e-06, "loss": 0.3283, "step": 14048 }, { "epoch": 90.56385542168675, "grad_norm": 16.664735794067383, "learning_rate": 4.680645161290323e-06, "loss": 0.3515, "step": 14049 }, { "epoch": 90.570281124498, "grad_norm": 10.012286186218262, "learning_rate": 4.67741935483871e-06, "loss": 0.13, "step": 14050 }, { "epoch": 90.57670682730924, "grad_norm": 12.659531593322754, "learning_rate": 4.674193548387097e-06, "loss": 0.3392, "step": 14051 }, { "epoch": 90.58313253012048, "grad_norm": 35.54553985595703, "learning_rate": 4.670967741935484e-06, "loss": 0.3765, "step": 14052 }, { "epoch": 90.58955823293172, "grad_norm": 26.56236457824707, "learning_rate": 4.667741935483871e-06, "loss": 0.269, "step": 14053 }, { "epoch": 90.59598393574298, "grad_norm": 35.35313034057617, "learning_rate": 4.664516129032259e-06, "loss": 0.5989, "step": 14054 }, { "epoch": 90.60240963855422, "grad_norm": 6.164313316345215, "learning_rate": 4.661290322580645e-06, "loss": 0.282, "step": 14055 }, { "epoch": 90.60883534136546, "grad_norm": 7.630752086639404, "learning_rate": 4.658064516129033e-06, "loss": 0.4061, "step": 14056 }, { "epoch": 90.6152610441767, "grad_norm": 14.560248374938965, "learning_rate": 4.65483870967742e-06, "loss": 0.2757, "step": 14057 }, { "epoch": 90.62168674698795, "grad_norm": 3.5704572200775146, "learning_rate": 4.651612903225807e-06, "loss": 0.2434, "step": 14058 }, { "epoch": 90.6281124497992, "grad_norm": 4.925711154937744, "learning_rate": 4.648387096774194e-06, "loss": 0.3256, "step": 14059 }, { "epoch": 90.63453815261045, "grad_norm": 27.251794815063477, "learning_rate": 4.6451612903225815e-06, "loss": 0.4152, "step": 14060 }, { "epoch": 90.64096385542169, "grad_norm": 17.96341896057129, "learning_rate": 4.641935483870968e-06, "loss": 0.3618, "step": 14061 }, { "epoch": 90.64738955823293, "grad_norm": 15.135931968688965, "learning_rate": 4.638709677419355e-06, "loss": 0.7366, "step": 14062 }, { "epoch": 90.65381526104417, "grad_norm": 15.804286003112793, "learning_rate": 4.635483870967742e-06, "loss": 0.5792, "step": 14063 }, { "epoch": 90.66024096385541, "grad_norm": 8.618453025817871, "learning_rate": 4.632258064516129e-06, "loss": 0.3231, "step": 14064 }, { "epoch": 90.66666666666667, "grad_norm": 23.168500900268555, "learning_rate": 4.629032258064516e-06, "loss": 0.304, "step": 14065 }, { "epoch": 90.67309236947791, "grad_norm": 18.794816970825195, "learning_rate": 4.625806451612903e-06, "loss": 0.4861, "step": 14066 }, { "epoch": 90.67951807228916, "grad_norm": 36.4211540222168, "learning_rate": 4.622580645161291e-06, "loss": 0.5383, "step": 14067 }, { "epoch": 90.6859437751004, "grad_norm": 5.645369529724121, "learning_rate": 4.619354838709677e-06, "loss": 0.4364, "step": 14068 }, { "epoch": 90.69236947791164, "grad_norm": 4.409483909606934, "learning_rate": 4.616129032258065e-06, "loss": 0.5746, "step": 14069 }, { "epoch": 90.6987951807229, "grad_norm": 37.19992446899414, "learning_rate": 4.612903225806452e-06, "loss": 0.6291, "step": 14070 }, { "epoch": 90.70522088353414, "grad_norm": 10.02813720703125, "learning_rate": 4.609677419354839e-06, "loss": 0.3374, "step": 14071 }, { "epoch": 90.71164658634538, "grad_norm": 19.21245765686035, "learning_rate": 4.606451612903226e-06, "loss": 0.2401, "step": 14072 }, { "epoch": 90.71807228915662, "grad_norm": 8.344520568847656, "learning_rate": 4.6032258064516134e-06, "loss": 0.2585, "step": 14073 }, { "epoch": 90.72449799196787, "grad_norm": 14.566436767578125, "learning_rate": 4.6e-06, "loss": 0.3724, "step": 14074 }, { "epoch": 90.73092369477912, "grad_norm": 8.315403938293457, "learning_rate": 4.596774193548387e-06, "loss": 0.311, "step": 14075 }, { "epoch": 90.73734939759036, "grad_norm": 6.349186897277832, "learning_rate": 4.593548387096774e-06, "loss": 0.1455, "step": 14076 }, { "epoch": 90.7437751004016, "grad_norm": 23.848812103271484, "learning_rate": 4.590322580645161e-06, "loss": 0.4559, "step": 14077 }, { "epoch": 90.75020080321285, "grad_norm": 9.135514259338379, "learning_rate": 4.587096774193549e-06, "loss": 0.4335, "step": 14078 }, { "epoch": 90.75662650602409, "grad_norm": 7.801900863647461, "learning_rate": 4.583870967741935e-06, "loss": 0.5454, "step": 14079 }, { "epoch": 90.76305220883535, "grad_norm": 28.496604919433594, "learning_rate": 4.580645161290323e-06, "loss": 0.268, "step": 14080 }, { "epoch": 90.76947791164659, "grad_norm": 4.593417644500732, "learning_rate": 4.57741935483871e-06, "loss": 0.2524, "step": 14081 }, { "epoch": 90.77590361445783, "grad_norm": 30.884246826171875, "learning_rate": 4.574193548387097e-06, "loss": 0.5635, "step": 14082 }, { "epoch": 90.78232931726907, "grad_norm": 10.392555236816406, "learning_rate": 4.570967741935484e-06, "loss": 0.2258, "step": 14083 }, { "epoch": 90.78875502008032, "grad_norm": 37.270328521728516, "learning_rate": 4.5677419354838715e-06, "loss": 0.5206, "step": 14084 }, { "epoch": 90.79518072289157, "grad_norm": 13.360197067260742, "learning_rate": 4.564516129032258e-06, "loss": 0.4325, "step": 14085 }, { "epoch": 90.80160642570281, "grad_norm": 3.5635411739349365, "learning_rate": 4.5612903225806454e-06, "loss": 0.1417, "step": 14086 }, { "epoch": 90.80803212851406, "grad_norm": 9.344470977783203, "learning_rate": 4.558064516129033e-06, "loss": 0.3498, "step": 14087 }, { "epoch": 90.8144578313253, "grad_norm": 7.148070335388184, "learning_rate": 4.554838709677419e-06, "loss": 0.2085, "step": 14088 }, { "epoch": 90.82088353413654, "grad_norm": 37.572540283203125, "learning_rate": 4.551612903225807e-06, "loss": 0.4581, "step": 14089 }, { "epoch": 90.8273092369478, "grad_norm": 14.704938888549805, "learning_rate": 4.548387096774194e-06, "loss": 0.5645, "step": 14090 }, { "epoch": 90.83373493975904, "grad_norm": 11.992192268371582, "learning_rate": 4.545161290322581e-06, "loss": 0.2131, "step": 14091 }, { "epoch": 90.84016064257028, "grad_norm": 13.021437644958496, "learning_rate": 4.541935483870968e-06, "loss": 0.1911, "step": 14092 }, { "epoch": 90.84658634538152, "grad_norm": 53.99958038330078, "learning_rate": 4.5387096774193555e-06, "loss": 0.7096, "step": 14093 }, { "epoch": 90.85301204819277, "grad_norm": 6.4167094230651855, "learning_rate": 4.535483870967742e-06, "loss": 0.3203, "step": 14094 }, { "epoch": 90.85943775100401, "grad_norm": 15.420305252075195, "learning_rate": 4.5322580645161295e-06, "loss": 0.9039, "step": 14095 }, { "epoch": 90.86586345381527, "grad_norm": 12.812260627746582, "learning_rate": 4.529032258064516e-06, "loss": 0.3961, "step": 14096 }, { "epoch": 90.87228915662651, "grad_norm": 32.57374954223633, "learning_rate": 4.5258064516129034e-06, "loss": 0.3859, "step": 14097 }, { "epoch": 90.87871485943775, "grad_norm": 9.719368934631348, "learning_rate": 4.52258064516129e-06, "loss": 0.3049, "step": 14098 }, { "epoch": 90.88514056224899, "grad_norm": 41.69927978515625, "learning_rate": 4.519354838709677e-06, "loss": 0.1763, "step": 14099 }, { "epoch": 90.89156626506023, "grad_norm": 2.9921772480010986, "learning_rate": 4.516129032258065e-06, "loss": 0.2634, "step": 14100 }, { "epoch": 90.89799196787149, "grad_norm": 25.559696197509766, "learning_rate": 4.512903225806451e-06, "loss": 0.3739, "step": 14101 }, { "epoch": 90.90441767068273, "grad_norm": 9.751594543457031, "learning_rate": 4.509677419354839e-06, "loss": 0.668, "step": 14102 }, { "epoch": 90.91084337349398, "grad_norm": 13.769814491271973, "learning_rate": 4.506451612903226e-06, "loss": 0.3957, "step": 14103 }, { "epoch": 90.91726907630522, "grad_norm": 6.9859938621521, "learning_rate": 4.503225806451613e-06, "loss": 0.2415, "step": 14104 }, { "epoch": 90.92369477911646, "grad_norm": 6.8770880699157715, "learning_rate": 4.5e-06, "loss": 0.274, "step": 14105 }, { "epoch": 90.93012048192772, "grad_norm": 23.455705642700195, "learning_rate": 4.4967741935483875e-06, "loss": 0.1935, "step": 14106 }, { "epoch": 90.93654618473896, "grad_norm": 10.428261756896973, "learning_rate": 4.493548387096774e-06, "loss": 0.4325, "step": 14107 }, { "epoch": 90.9429718875502, "grad_norm": 4.569668292999268, "learning_rate": 4.4903225806451615e-06, "loss": 0.1964, "step": 14108 }, { "epoch": 90.94939759036144, "grad_norm": 12.818543434143066, "learning_rate": 4.487096774193549e-06, "loss": 0.3547, "step": 14109 }, { "epoch": 90.95582329317268, "grad_norm": 6.1389899253845215, "learning_rate": 4.4838709677419354e-06, "loss": 0.3005, "step": 14110 }, { "epoch": 90.96224899598394, "grad_norm": 11.836605072021484, "learning_rate": 4.480645161290323e-06, "loss": 0.2415, "step": 14111 }, { "epoch": 90.96867469879518, "grad_norm": 43.48137283325195, "learning_rate": 4.47741935483871e-06, "loss": 0.645, "step": 14112 }, { "epoch": 90.97510040160643, "grad_norm": 1.1225707530975342, "learning_rate": 4.474193548387097e-06, "loss": 0.2967, "step": 14113 }, { "epoch": 90.98152610441767, "grad_norm": 18.504018783569336, "learning_rate": 4.470967741935484e-06, "loss": 0.2532, "step": 14114 }, { "epoch": 90.98795180722891, "grad_norm": 11.069194793701172, "learning_rate": 4.467741935483872e-06, "loss": 0.3261, "step": 14115 }, { "epoch": 90.99437751004017, "grad_norm": 7.477406978607178, "learning_rate": 4.464516129032258e-06, "loss": 0.302, "step": 14116 }, { "epoch": 91.00401606425703, "grad_norm": 7.839364528656006, "learning_rate": 4.4612903225806455e-06, "loss": 0.1998, "step": 14117 }, { "epoch": 91.01044176706827, "grad_norm": 65.17717742919922, "learning_rate": 4.458064516129033e-06, "loss": 0.2728, "step": 14118 }, { "epoch": 91.01686746987951, "grad_norm": 11.029378890991211, "learning_rate": 4.4548387096774195e-06, "loss": 0.3835, "step": 14119 }, { "epoch": 91.02329317269076, "grad_norm": 2.180086851119995, "learning_rate": 4.451612903225807e-06, "loss": 0.3128, "step": 14120 }, { "epoch": 91.02971887550201, "grad_norm": 17.436521530151367, "learning_rate": 4.448387096774194e-06, "loss": 0.7563, "step": 14121 }, { "epoch": 91.03614457831326, "grad_norm": 5.643351078033447, "learning_rate": 4.445161290322581e-06, "loss": 0.3985, "step": 14122 }, { "epoch": 91.0425702811245, "grad_norm": 12.551058769226074, "learning_rate": 4.441935483870968e-06, "loss": 0.3164, "step": 14123 }, { "epoch": 91.04899598393574, "grad_norm": 21.79070281982422, "learning_rate": 4.438709677419356e-06, "loss": 0.182, "step": 14124 }, { "epoch": 91.05542168674698, "grad_norm": 23.451828002929688, "learning_rate": 4.435483870967742e-06, "loss": 0.3548, "step": 14125 }, { "epoch": 91.06184738955824, "grad_norm": 10.700026512145996, "learning_rate": 4.43225806451613e-06, "loss": 0.1344, "step": 14126 }, { "epoch": 91.06827309236948, "grad_norm": 6.370250225067139, "learning_rate": 4.429032258064517e-06, "loss": 0.3376, "step": 14127 }, { "epoch": 91.07469879518072, "grad_norm": 8.24059009552002, "learning_rate": 4.4258064516129036e-06, "loss": 0.3747, "step": 14128 }, { "epoch": 91.08112449799196, "grad_norm": 13.213296890258789, "learning_rate": 4.422580645161291e-06, "loss": 0.3552, "step": 14129 }, { "epoch": 91.08755020080321, "grad_norm": 16.53443717956543, "learning_rate": 4.419354838709678e-06, "loss": 0.2931, "step": 14130 }, { "epoch": 91.09397590361446, "grad_norm": 33.08927917480469, "learning_rate": 4.416129032258065e-06, "loss": 0.274, "step": 14131 }, { "epoch": 91.1004016064257, "grad_norm": 4.04755163192749, "learning_rate": 4.4129032258064515e-06, "loss": 0.3536, "step": 14132 }, { "epoch": 91.10682730923695, "grad_norm": 9.336751937866211, "learning_rate": 4.409677419354839e-06, "loss": 0.1243, "step": 14133 }, { "epoch": 91.11325301204819, "grad_norm": 11.420587539672852, "learning_rate": 4.406451612903225e-06, "loss": 0.3798, "step": 14134 }, { "epoch": 91.11967871485943, "grad_norm": 6.056922912597656, "learning_rate": 4.403225806451613e-06, "loss": 0.2289, "step": 14135 }, { "epoch": 91.12610441767069, "grad_norm": 9.258784294128418, "learning_rate": 4.4e-06, "loss": 0.2884, "step": 14136 }, { "epoch": 91.13253012048193, "grad_norm": 3.7595343589782715, "learning_rate": 4.396774193548387e-06, "loss": 0.0615, "step": 14137 }, { "epoch": 91.13895582329317, "grad_norm": 19.215919494628906, "learning_rate": 4.393548387096774e-06, "loss": 0.4376, "step": 14138 }, { "epoch": 91.14538152610442, "grad_norm": 17.443744659423828, "learning_rate": 4.390322580645162e-06, "loss": 0.585, "step": 14139 }, { "epoch": 91.15180722891566, "grad_norm": 12.602354049682617, "learning_rate": 4.387096774193548e-06, "loss": 0.3532, "step": 14140 }, { "epoch": 91.15823293172691, "grad_norm": 14.564062118530273, "learning_rate": 4.3838709677419355e-06, "loss": 0.2751, "step": 14141 }, { "epoch": 91.16465863453816, "grad_norm": 11.151795387268066, "learning_rate": 4.380645161290323e-06, "loss": 0.336, "step": 14142 }, { "epoch": 91.1710843373494, "grad_norm": 9.92762565612793, "learning_rate": 4.3774193548387095e-06, "loss": 0.6224, "step": 14143 }, { "epoch": 91.17751004016064, "grad_norm": 20.667232513427734, "learning_rate": 4.374193548387097e-06, "loss": 0.6921, "step": 14144 }, { "epoch": 91.18393574297188, "grad_norm": 5.632270812988281, "learning_rate": 4.370967741935484e-06, "loss": 0.1544, "step": 14145 }, { "epoch": 91.19036144578314, "grad_norm": 11.073857307434082, "learning_rate": 4.367741935483871e-06, "loss": 0.2373, "step": 14146 }, { "epoch": 91.19678714859438, "grad_norm": 67.44279479980469, "learning_rate": 4.364516129032258e-06, "loss": 0.4977, "step": 14147 }, { "epoch": 91.20321285140562, "grad_norm": 18.31694984436035, "learning_rate": 4.361290322580646e-06, "loss": 0.381, "step": 14148 }, { "epoch": 91.20963855421687, "grad_norm": 13.265922546386719, "learning_rate": 4.358064516129032e-06, "loss": 0.3314, "step": 14149 }, { "epoch": 91.21606425702811, "grad_norm": 43.19846725463867, "learning_rate": 4.35483870967742e-06, "loss": 0.3156, "step": 14150 }, { "epoch": 91.22248995983935, "grad_norm": 13.532369613647461, "learning_rate": 4.351612903225807e-06, "loss": 0.4665, "step": 14151 }, { "epoch": 91.2289156626506, "grad_norm": 8.436141014099121, "learning_rate": 4.3483870967741936e-06, "loss": 0.2278, "step": 14152 }, { "epoch": 91.23534136546185, "grad_norm": 17.714088439941406, "learning_rate": 4.345161290322581e-06, "loss": 0.5753, "step": 14153 }, { "epoch": 91.24176706827309, "grad_norm": 11.706573486328125, "learning_rate": 4.341935483870968e-06, "loss": 0.1644, "step": 14154 }, { "epoch": 91.24819277108433, "grad_norm": 14.073417663574219, "learning_rate": 4.338709677419355e-06, "loss": 0.2483, "step": 14155 }, { "epoch": 91.25461847389558, "grad_norm": 12.403079986572266, "learning_rate": 4.335483870967742e-06, "loss": 0.4641, "step": 14156 }, { "epoch": 91.26104417670683, "grad_norm": 14.330191612243652, "learning_rate": 4.33225806451613e-06, "loss": 0.3332, "step": 14157 }, { "epoch": 91.26746987951807, "grad_norm": 13.087742805480957, "learning_rate": 4.329032258064516e-06, "loss": 0.3135, "step": 14158 }, { "epoch": 91.27389558232932, "grad_norm": 4.399267673492432, "learning_rate": 4.325806451612904e-06, "loss": 0.0808, "step": 14159 }, { "epoch": 91.28032128514056, "grad_norm": 9.181187629699707, "learning_rate": 4.322580645161291e-06, "loss": 0.329, "step": 14160 }, { "epoch": 91.2867469879518, "grad_norm": 2.674663782119751, "learning_rate": 4.319354838709678e-06, "loss": 0.2577, "step": 14161 }, { "epoch": 91.29317269076306, "grad_norm": 12.671516418457031, "learning_rate": 4.316129032258065e-06, "loss": 0.2653, "step": 14162 }, { "epoch": 91.2995983935743, "grad_norm": 15.724082946777344, "learning_rate": 4.3129032258064524e-06, "loss": 0.3342, "step": 14163 }, { "epoch": 91.30602409638554, "grad_norm": 7.062438488006592, "learning_rate": 4.309677419354839e-06, "loss": 0.2436, "step": 14164 }, { "epoch": 91.31244979919678, "grad_norm": 20.066041946411133, "learning_rate": 4.306451612903226e-06, "loss": 0.4171, "step": 14165 }, { "epoch": 91.31887550200803, "grad_norm": 18.83283805847168, "learning_rate": 4.303225806451614e-06, "loss": 0.2004, "step": 14166 }, { "epoch": 91.32530120481928, "grad_norm": 5.017795562744141, "learning_rate": 4.2999999999999995e-06, "loss": 0.3967, "step": 14167 }, { "epoch": 91.33172690763053, "grad_norm": 15.485645294189453, "learning_rate": 4.296774193548387e-06, "loss": 0.3162, "step": 14168 }, { "epoch": 91.33815261044177, "grad_norm": 3.9840497970581055, "learning_rate": 4.293548387096774e-06, "loss": 0.2044, "step": 14169 }, { "epoch": 91.34457831325301, "grad_norm": 27.57086181640625, "learning_rate": 4.290322580645161e-06, "loss": 0.2196, "step": 14170 }, { "epoch": 91.35100401606425, "grad_norm": 17.205493927001953, "learning_rate": 4.287096774193548e-06, "loss": 0.5999, "step": 14171 }, { "epoch": 91.35742971887551, "grad_norm": 10.347908973693848, "learning_rate": 4.283870967741936e-06, "loss": 0.2607, "step": 14172 }, { "epoch": 91.36385542168675, "grad_norm": 10.309518814086914, "learning_rate": 4.280645161290322e-06, "loss": 0.2172, "step": 14173 }, { "epoch": 91.37028112449799, "grad_norm": 14.694273948669434, "learning_rate": 4.27741935483871e-06, "loss": 0.1574, "step": 14174 }, { "epoch": 91.37670682730923, "grad_norm": 13.761021614074707, "learning_rate": 4.274193548387097e-06, "loss": 0.4662, "step": 14175 }, { "epoch": 91.38313253012048, "grad_norm": 16.002849578857422, "learning_rate": 4.2709677419354836e-06, "loss": 0.3447, "step": 14176 }, { "epoch": 91.38955823293173, "grad_norm": 22.19718360900879, "learning_rate": 4.267741935483871e-06, "loss": 0.5165, "step": 14177 }, { "epoch": 91.39598393574298, "grad_norm": 12.716053009033203, "learning_rate": 4.264516129032258e-06, "loss": 0.254, "step": 14178 }, { "epoch": 91.40240963855422, "grad_norm": 17.17436408996582, "learning_rate": 4.261290322580645e-06, "loss": 0.3116, "step": 14179 }, { "epoch": 91.40883534136546, "grad_norm": 7.133237838745117, "learning_rate": 4.258064516129032e-06, "loss": 0.2667, "step": 14180 }, { "epoch": 91.4152610441767, "grad_norm": 14.686293601989746, "learning_rate": 4.25483870967742e-06, "loss": 0.4405, "step": 14181 }, { "epoch": 91.42168674698796, "grad_norm": 18.024660110473633, "learning_rate": 4.251612903225806e-06, "loss": 0.2533, "step": 14182 }, { "epoch": 91.4281124497992, "grad_norm": 11.884982109069824, "learning_rate": 4.248387096774194e-06, "loss": 0.4777, "step": 14183 }, { "epoch": 91.43453815261044, "grad_norm": 14.08698558807373, "learning_rate": 4.245161290322581e-06, "loss": 0.2911, "step": 14184 }, { "epoch": 91.44096385542169, "grad_norm": 22.976579666137695, "learning_rate": 4.241935483870968e-06, "loss": 0.6243, "step": 14185 }, { "epoch": 91.44738955823293, "grad_norm": 4.978161811828613, "learning_rate": 4.238709677419355e-06, "loss": 0.1952, "step": 14186 }, { "epoch": 91.45381526104417, "grad_norm": 5.42819356918335, "learning_rate": 4.235483870967742e-06, "loss": 0.3249, "step": 14187 }, { "epoch": 91.46024096385543, "grad_norm": 23.588611602783203, "learning_rate": 4.232258064516129e-06, "loss": 0.2199, "step": 14188 }, { "epoch": 91.46666666666667, "grad_norm": 10.144634246826172, "learning_rate": 4.229032258064516e-06, "loss": 0.4996, "step": 14189 }, { "epoch": 91.47309236947791, "grad_norm": 6.853063583374023, "learning_rate": 4.225806451612904e-06, "loss": 0.2935, "step": 14190 }, { "epoch": 91.47951807228915, "grad_norm": 8.278165817260742, "learning_rate": 4.22258064516129e-06, "loss": 0.2376, "step": 14191 }, { "epoch": 91.4859437751004, "grad_norm": 2.6753673553466797, "learning_rate": 4.219354838709678e-06, "loss": 0.202, "step": 14192 }, { "epoch": 91.49236947791165, "grad_norm": 15.635056495666504, "learning_rate": 4.216129032258065e-06, "loss": 0.342, "step": 14193 }, { "epoch": 91.4987951807229, "grad_norm": 2.9546215534210205, "learning_rate": 4.212903225806452e-06, "loss": 0.2953, "step": 14194 }, { "epoch": 91.50522088353414, "grad_norm": 10.775738716125488, "learning_rate": 4.209677419354839e-06, "loss": 0.2839, "step": 14195 }, { "epoch": 91.51164658634538, "grad_norm": 32.62097930908203, "learning_rate": 4.2064516129032265e-06, "loss": 0.3481, "step": 14196 }, { "epoch": 91.51807228915662, "grad_norm": 10.65744400024414, "learning_rate": 4.203225806451613e-06, "loss": 0.4002, "step": 14197 }, { "epoch": 91.52449799196788, "grad_norm": 2.740248918533325, "learning_rate": 4.2000000000000004e-06, "loss": 0.3133, "step": 14198 }, { "epoch": 91.53092369477912, "grad_norm": 8.104955673217773, "learning_rate": 4.196774193548388e-06, "loss": 0.1438, "step": 14199 }, { "epoch": 91.53734939759036, "grad_norm": 30.226123809814453, "learning_rate": 4.193548387096774e-06, "loss": 0.5306, "step": 14200 }, { "epoch": 91.5437751004016, "grad_norm": 11.538677215576172, "learning_rate": 4.190322580645162e-06, "loss": 0.3674, "step": 14201 }, { "epoch": 91.55020080321285, "grad_norm": 9.213540077209473, "learning_rate": 4.187096774193549e-06, "loss": 0.1937, "step": 14202 }, { "epoch": 91.5566265060241, "grad_norm": 12.027655601501465, "learning_rate": 4.183870967741936e-06, "loss": 0.3077, "step": 14203 }, { "epoch": 91.56305220883534, "grad_norm": 27.939523696899414, "learning_rate": 4.180645161290322e-06, "loss": 0.4245, "step": 14204 }, { "epoch": 91.56947791164659, "grad_norm": 4.910234451293945, "learning_rate": 4.17741935483871e-06, "loss": 0.0959, "step": 14205 }, { "epoch": 91.57590361445783, "grad_norm": 12.707109451293945, "learning_rate": 4.174193548387097e-06, "loss": 0.2351, "step": 14206 }, { "epoch": 91.58232931726907, "grad_norm": 4.470451831817627, "learning_rate": 4.170967741935484e-06, "loss": 0.3534, "step": 14207 }, { "epoch": 91.58875502008033, "grad_norm": 26.025951385498047, "learning_rate": 4.167741935483871e-06, "loss": 0.2856, "step": 14208 }, { "epoch": 91.59518072289157, "grad_norm": 20.566936492919922, "learning_rate": 4.1645161290322585e-06, "loss": 0.3757, "step": 14209 }, { "epoch": 91.60160642570281, "grad_norm": 13.160852432250977, "learning_rate": 4.161290322580645e-06, "loss": 0.4418, "step": 14210 }, { "epoch": 91.60803212851405, "grad_norm": 14.060074806213379, "learning_rate": 4.158064516129032e-06, "loss": 0.2966, "step": 14211 }, { "epoch": 91.6144578313253, "grad_norm": 8.631351470947266, "learning_rate": 4.154838709677419e-06, "loss": 0.3858, "step": 14212 }, { "epoch": 91.62088353413655, "grad_norm": 9.941097259521484, "learning_rate": 4.151612903225806e-06, "loss": 0.4277, "step": 14213 }, { "epoch": 91.6273092369478, "grad_norm": 10.16092586517334, "learning_rate": 4.148387096774194e-06, "loss": 0.377, "step": 14214 }, { "epoch": 91.63373493975904, "grad_norm": 17.04046630859375, "learning_rate": 4.14516129032258e-06, "loss": 0.2992, "step": 14215 }, { "epoch": 91.64016064257028, "grad_norm": 30.771438598632812, "learning_rate": 4.141935483870968e-06, "loss": 0.3604, "step": 14216 }, { "epoch": 91.64658634538152, "grad_norm": 19.484424591064453, "learning_rate": 4.138709677419355e-06, "loss": 0.4214, "step": 14217 }, { "epoch": 91.65301204819278, "grad_norm": 9.02022647857666, "learning_rate": 4.135483870967742e-06, "loss": 0.2262, "step": 14218 }, { "epoch": 91.65943775100402, "grad_norm": 17.402008056640625, "learning_rate": 4.132258064516129e-06, "loss": 0.329, "step": 14219 }, { "epoch": 91.66586345381526, "grad_norm": 32.541893005371094, "learning_rate": 4.1290322580645165e-06, "loss": 0.4363, "step": 14220 }, { "epoch": 91.6722891566265, "grad_norm": 3.803593635559082, "learning_rate": 4.125806451612903e-06, "loss": 0.2143, "step": 14221 }, { "epoch": 91.67871485943775, "grad_norm": 26.125436782836914, "learning_rate": 4.1225806451612904e-06, "loss": 0.3364, "step": 14222 }, { "epoch": 91.68514056224899, "grad_norm": 8.722736358642578, "learning_rate": 4.119354838709678e-06, "loss": 0.2801, "step": 14223 }, { "epoch": 91.69156626506025, "grad_norm": 11.112013816833496, "learning_rate": 4.116129032258064e-06, "loss": 0.4443, "step": 14224 }, { "epoch": 91.69799196787149, "grad_norm": 24.912012100219727, "learning_rate": 4.112903225806452e-06, "loss": 0.5429, "step": 14225 }, { "epoch": 91.70441767068273, "grad_norm": 10.228032112121582, "learning_rate": 4.109677419354839e-06, "loss": 0.4078, "step": 14226 }, { "epoch": 91.71084337349397, "grad_norm": 12.265101432800293, "learning_rate": 4.106451612903226e-06, "loss": 0.4884, "step": 14227 }, { "epoch": 91.71726907630521, "grad_norm": 23.57832145690918, "learning_rate": 4.103225806451613e-06, "loss": 0.3025, "step": 14228 }, { "epoch": 91.72369477911647, "grad_norm": 22.78856086730957, "learning_rate": 4.1000000000000006e-06, "loss": 0.2596, "step": 14229 }, { "epoch": 91.73012048192771, "grad_norm": 4.234719276428223, "learning_rate": 4.096774193548387e-06, "loss": 0.3645, "step": 14230 }, { "epoch": 91.73654618473896, "grad_norm": 7.208852291107178, "learning_rate": 4.0935483870967745e-06, "loss": 0.1824, "step": 14231 }, { "epoch": 91.7429718875502, "grad_norm": 5.017682075500488, "learning_rate": 4.090322580645162e-06, "loss": 0.1589, "step": 14232 }, { "epoch": 91.74939759036144, "grad_norm": 18.9503173828125, "learning_rate": 4.0870967741935485e-06, "loss": 0.3807, "step": 14233 }, { "epoch": 91.7558232931727, "grad_norm": 11.19483757019043, "learning_rate": 4.083870967741936e-06, "loss": 0.234, "step": 14234 }, { "epoch": 91.76224899598394, "grad_norm": 3.368717908859253, "learning_rate": 4.080645161290323e-06, "loss": 0.1309, "step": 14235 }, { "epoch": 91.76867469879518, "grad_norm": 9.589749336242676, "learning_rate": 4.07741935483871e-06, "loss": 0.3809, "step": 14236 }, { "epoch": 91.77510040160642, "grad_norm": 37.32256317138672, "learning_rate": 4.074193548387097e-06, "loss": 0.2997, "step": 14237 }, { "epoch": 91.78152610441767, "grad_norm": 9.817421913146973, "learning_rate": 4.070967741935484e-06, "loss": 0.2998, "step": 14238 }, { "epoch": 91.78795180722892, "grad_norm": 16.737693786621094, "learning_rate": 4.067741935483871e-06, "loss": 0.548, "step": 14239 }, { "epoch": 91.79437751004016, "grad_norm": 10.816320419311523, "learning_rate": 4.064516129032258e-06, "loss": 0.1747, "step": 14240 }, { "epoch": 91.8008032128514, "grad_norm": 6.652761936187744, "learning_rate": 4.061290322580645e-06, "loss": 0.4137, "step": 14241 }, { "epoch": 91.80722891566265, "grad_norm": 8.233898162841797, "learning_rate": 4.0580645161290325e-06, "loss": 0.5349, "step": 14242 }, { "epoch": 91.81365461847389, "grad_norm": 16.63419532775879, "learning_rate": 4.054838709677419e-06, "loss": 0.4353, "step": 14243 }, { "epoch": 91.82008032128515, "grad_norm": 21.746097564697266, "learning_rate": 4.0516129032258065e-06, "loss": 0.68, "step": 14244 }, { "epoch": 91.82650602409639, "grad_norm": 20.709335327148438, "learning_rate": 4.048387096774194e-06, "loss": 0.3923, "step": 14245 }, { "epoch": 91.83293172690763, "grad_norm": 18.33575439453125, "learning_rate": 4.0451612903225804e-06, "loss": 0.3182, "step": 14246 }, { "epoch": 91.83935742971887, "grad_norm": 22.44170379638672, "learning_rate": 4.041935483870968e-06, "loss": 0.3968, "step": 14247 }, { "epoch": 91.84578313253012, "grad_norm": 11.10117244720459, "learning_rate": 4.038709677419355e-06, "loss": 0.3124, "step": 14248 }, { "epoch": 91.85220883534137, "grad_norm": 5.0730180740356445, "learning_rate": 4.035483870967742e-06, "loss": 0.2178, "step": 14249 }, { "epoch": 91.85863453815261, "grad_norm": 14.472904205322266, "learning_rate": 4.032258064516129e-06, "loss": 0.415, "step": 14250 }, { "epoch": 91.86506024096386, "grad_norm": 24.940595626831055, "learning_rate": 4.029032258064517e-06, "loss": 0.5888, "step": 14251 }, { "epoch": 91.8714859437751, "grad_norm": 14.404158592224121, "learning_rate": 4.025806451612903e-06, "loss": 0.3788, "step": 14252 }, { "epoch": 91.87791164658634, "grad_norm": 6.7732954025268555, "learning_rate": 4.0225806451612906e-06, "loss": 0.2798, "step": 14253 }, { "epoch": 91.88433734939758, "grad_norm": 20.176435470581055, "learning_rate": 4.019354838709678e-06, "loss": 0.3618, "step": 14254 }, { "epoch": 91.89076305220884, "grad_norm": 5.170948028564453, "learning_rate": 4.0161290322580645e-06, "loss": 0.1276, "step": 14255 }, { "epoch": 91.89718875502008, "grad_norm": 7.660892963409424, "learning_rate": 4.012903225806452e-06, "loss": 0.3804, "step": 14256 }, { "epoch": 91.90361445783132, "grad_norm": 2.4577436447143555, "learning_rate": 4.009677419354839e-06, "loss": 0.4062, "step": 14257 }, { "epoch": 91.91004016064257, "grad_norm": 10.543505668640137, "learning_rate": 4.006451612903226e-06, "loss": 0.2641, "step": 14258 }, { "epoch": 91.91646586345381, "grad_norm": 12.36402416229248, "learning_rate": 4.003225806451613e-06, "loss": 0.1506, "step": 14259 }, { "epoch": 91.92289156626506, "grad_norm": 2.3087990283966064, "learning_rate": 4.000000000000001e-06, "loss": 0.2989, "step": 14260 }, { "epoch": 91.92931726907631, "grad_norm": 23.744476318359375, "learning_rate": 3.996774193548387e-06, "loss": 0.4057, "step": 14261 }, { "epoch": 91.93574297188755, "grad_norm": 14.107855796813965, "learning_rate": 3.993548387096775e-06, "loss": 0.2341, "step": 14262 }, { "epoch": 91.94216867469879, "grad_norm": 17.367605209350586, "learning_rate": 3.990322580645162e-06, "loss": 0.1439, "step": 14263 }, { "epoch": 91.94859437751003, "grad_norm": 12.912303924560547, "learning_rate": 3.9870967741935486e-06, "loss": 0.3517, "step": 14264 }, { "epoch": 91.95502008032129, "grad_norm": 10.90747356414795, "learning_rate": 3.983870967741936e-06, "loss": 0.3639, "step": 14265 }, { "epoch": 91.96144578313253, "grad_norm": 31.55292510986328, "learning_rate": 3.980645161290323e-06, "loss": 0.3896, "step": 14266 }, { "epoch": 91.96787148594377, "grad_norm": 9.654886245727539, "learning_rate": 3.97741935483871e-06, "loss": 0.2753, "step": 14267 }, { "epoch": 91.97429718875502, "grad_norm": 22.06874656677246, "learning_rate": 3.974193548387097e-06, "loss": 0.3925, "step": 14268 }, { "epoch": 91.98072289156626, "grad_norm": 17.42096519470215, "learning_rate": 3.970967741935485e-06, "loss": 0.5486, "step": 14269 }, { "epoch": 91.98714859437752, "grad_norm": 6.462732791900635, "learning_rate": 3.967741935483871e-06, "loss": 0.311, "step": 14270 }, { "epoch": 91.99357429718876, "grad_norm": 7.941625595092773, "learning_rate": 3.964516129032259e-06, "loss": 0.3795, "step": 14271 }, { "epoch": 92.00321285140562, "grad_norm": 3.6678550243377686, "learning_rate": 3.961290322580646e-06, "loss": 0.3274, "step": 14272 }, { "epoch": 92.00963855421686, "grad_norm": 7.393374919891357, "learning_rate": 3.958064516129033e-06, "loss": 0.4003, "step": 14273 }, { "epoch": 92.0160642570281, "grad_norm": 10.307401657104492, "learning_rate": 3.954838709677419e-06, "loss": 0.4484, "step": 14274 }, { "epoch": 92.02248995983936, "grad_norm": 133.59893798828125, "learning_rate": 3.951612903225807e-06, "loss": 0.427, "step": 14275 }, { "epoch": 92.0289156626506, "grad_norm": 13.749170303344727, "learning_rate": 3.948387096774193e-06, "loss": 0.2975, "step": 14276 }, { "epoch": 92.03534136546185, "grad_norm": 1.459378957748413, "learning_rate": 3.9451612903225806e-06, "loss": 0.1916, "step": 14277 }, { "epoch": 92.04176706827309, "grad_norm": 6.095083236694336, "learning_rate": 3.941935483870968e-06, "loss": 0.3698, "step": 14278 }, { "epoch": 92.04819277108433, "grad_norm": 76.7298812866211, "learning_rate": 3.9387096774193545e-06, "loss": 0.7562, "step": 14279 }, { "epoch": 92.05461847389559, "grad_norm": 15.496537208557129, "learning_rate": 3.935483870967742e-06, "loss": 0.1538, "step": 14280 }, { "epoch": 92.06104417670683, "grad_norm": 10.531099319458008, "learning_rate": 3.932258064516129e-06, "loss": 0.1992, "step": 14281 }, { "epoch": 92.06746987951807, "grad_norm": 34.12911605834961, "learning_rate": 3.929032258064516e-06, "loss": 0.3687, "step": 14282 }, { "epoch": 92.07389558232931, "grad_norm": 108.3017578125, "learning_rate": 3.925806451612903e-06, "loss": 0.3184, "step": 14283 }, { "epoch": 92.08032128514056, "grad_norm": 4.103860378265381, "learning_rate": 3.922580645161291e-06, "loss": 0.198, "step": 14284 }, { "epoch": 92.08674698795181, "grad_norm": 12.80871868133545, "learning_rate": 3.919354838709677e-06, "loss": 0.3616, "step": 14285 }, { "epoch": 92.09317269076305, "grad_norm": 16.79450798034668, "learning_rate": 3.916129032258065e-06, "loss": 0.5526, "step": 14286 }, { "epoch": 92.0995983935743, "grad_norm": 5.8526482582092285, "learning_rate": 3.912903225806452e-06, "loss": 0.2071, "step": 14287 }, { "epoch": 92.10602409638554, "grad_norm": 7.1265363693237305, "learning_rate": 3.9096774193548386e-06, "loss": 0.1884, "step": 14288 }, { "epoch": 92.11244979919678, "grad_norm": 3.2969794273376465, "learning_rate": 3.906451612903226e-06, "loss": 0.1779, "step": 14289 }, { "epoch": 92.11887550200804, "grad_norm": 11.95052719116211, "learning_rate": 3.903225806451613e-06, "loss": 0.6501, "step": 14290 }, { "epoch": 92.12530120481928, "grad_norm": 4.6020731925964355, "learning_rate": 3.9e-06, "loss": 0.3639, "step": 14291 }, { "epoch": 92.13172690763052, "grad_norm": 19.531761169433594, "learning_rate": 3.896774193548387e-06, "loss": 0.3535, "step": 14292 }, { "epoch": 92.13815261044176, "grad_norm": 9.177451133728027, "learning_rate": 3.893548387096775e-06, "loss": 0.2478, "step": 14293 }, { "epoch": 92.144578313253, "grad_norm": 4.43428373336792, "learning_rate": 3.890322580645161e-06, "loss": 0.3583, "step": 14294 }, { "epoch": 92.15100401606426, "grad_norm": 10.325054168701172, "learning_rate": 3.887096774193549e-06, "loss": 0.271, "step": 14295 }, { "epoch": 92.1574297188755, "grad_norm": 4.944028854370117, "learning_rate": 3.883870967741936e-06, "loss": 0.1143, "step": 14296 }, { "epoch": 92.16385542168675, "grad_norm": 11.607354164123535, "learning_rate": 3.880645161290323e-06, "loss": 0.4086, "step": 14297 }, { "epoch": 92.17028112449799, "grad_norm": 8.627546310424805, "learning_rate": 3.87741935483871e-06, "loss": 0.1994, "step": 14298 }, { "epoch": 92.17670682730923, "grad_norm": 14.029232025146484, "learning_rate": 3.8741935483870974e-06, "loss": 0.4983, "step": 14299 }, { "epoch": 92.18313253012049, "grad_norm": 5.691259384155273, "learning_rate": 3.870967741935484e-06, "loss": 0.4217, "step": 14300 }, { "epoch": 92.18955823293173, "grad_norm": 41.07063674926758, "learning_rate": 3.867741935483871e-06, "loss": 0.1649, "step": 14301 }, { "epoch": 92.19598393574297, "grad_norm": 24.35226058959961, "learning_rate": 3.864516129032259e-06, "loss": 0.6533, "step": 14302 }, { "epoch": 92.20240963855422, "grad_norm": 5.628444671630859, "learning_rate": 3.861290322580645e-06, "loss": 0.3338, "step": 14303 }, { "epoch": 92.20883534136546, "grad_norm": 4.589829921722412, "learning_rate": 3.858064516129033e-06, "loss": 0.2334, "step": 14304 }, { "epoch": 92.21526104417671, "grad_norm": 14.455399513244629, "learning_rate": 3.85483870967742e-06, "loss": 0.2679, "step": 14305 }, { "epoch": 92.22168674698796, "grad_norm": 9.893945693969727, "learning_rate": 3.851612903225807e-06, "loss": 0.3188, "step": 14306 }, { "epoch": 92.2281124497992, "grad_norm": 9.419306755065918, "learning_rate": 3.848387096774194e-06, "loss": 0.3628, "step": 14307 }, { "epoch": 92.23453815261044, "grad_norm": 9.912168502807617, "learning_rate": 3.8451612903225815e-06, "loss": 0.3122, "step": 14308 }, { "epoch": 92.24096385542168, "grad_norm": 9.742111206054688, "learning_rate": 3.841935483870968e-06, "loss": 0.4284, "step": 14309 }, { "epoch": 92.24738955823292, "grad_norm": 18.588119506835938, "learning_rate": 3.838709677419355e-06, "loss": 0.2088, "step": 14310 }, { "epoch": 92.25381526104418, "grad_norm": 17.72275161743164, "learning_rate": 3.835483870967742e-06, "loss": 0.3435, "step": 14311 }, { "epoch": 92.26024096385542, "grad_norm": 28.865001678466797, "learning_rate": 3.8322580645161286e-06, "loss": 0.3649, "step": 14312 }, { "epoch": 92.26666666666667, "grad_norm": 10.93815803527832, "learning_rate": 3.829032258064516e-06, "loss": 0.2024, "step": 14313 }, { "epoch": 92.27309236947791, "grad_norm": 3.848644733428955, "learning_rate": 3.825806451612903e-06, "loss": 0.2777, "step": 14314 }, { "epoch": 92.27951807228915, "grad_norm": 4.475855350494385, "learning_rate": 3.82258064516129e-06, "loss": 0.2225, "step": 14315 }, { "epoch": 92.2859437751004, "grad_norm": 15.604090690612793, "learning_rate": 3.819354838709677e-06, "loss": 0.3596, "step": 14316 }, { "epoch": 92.29236947791165, "grad_norm": 21.3431453704834, "learning_rate": 3.816129032258065e-06, "loss": 0.611, "step": 14317 }, { "epoch": 92.29879518072289, "grad_norm": 3.8068325519561768, "learning_rate": 3.8129032258064517e-06, "loss": 0.2127, "step": 14318 }, { "epoch": 92.30522088353413, "grad_norm": 18.813308715820312, "learning_rate": 3.8096774193548387e-06, "loss": 0.6138, "step": 14319 }, { "epoch": 92.31164658634538, "grad_norm": 4.446208953857422, "learning_rate": 3.8064516129032257e-06, "loss": 0.1393, "step": 14320 }, { "epoch": 92.31807228915663, "grad_norm": 15.544504165649414, "learning_rate": 3.803225806451613e-06, "loss": 0.5701, "step": 14321 }, { "epoch": 92.32449799196787, "grad_norm": 15.923378944396973, "learning_rate": 3.8e-06, "loss": 0.3501, "step": 14322 }, { "epoch": 92.33092369477912, "grad_norm": 7.871470928192139, "learning_rate": 3.796774193548387e-06, "loss": 0.3829, "step": 14323 }, { "epoch": 92.33734939759036, "grad_norm": 7.36602783203125, "learning_rate": 3.7935483870967744e-06, "loss": 0.3027, "step": 14324 }, { "epoch": 92.3437751004016, "grad_norm": 3.065108060836792, "learning_rate": 3.7903225806451614e-06, "loss": 0.251, "step": 14325 }, { "epoch": 92.35020080321286, "grad_norm": 6.109309673309326, "learning_rate": 3.7870967741935484e-06, "loss": 0.2097, "step": 14326 }, { "epoch": 92.3566265060241, "grad_norm": 18.634811401367188, "learning_rate": 3.7838709677419358e-06, "loss": 0.2963, "step": 14327 }, { "epoch": 92.36305220883534, "grad_norm": 4.677386283874512, "learning_rate": 3.7806451612903228e-06, "loss": 0.3244, "step": 14328 }, { "epoch": 92.36947791164658, "grad_norm": 13.771942138671875, "learning_rate": 3.7774193548387097e-06, "loss": 0.1736, "step": 14329 }, { "epoch": 92.37590361445783, "grad_norm": 3.4357409477233887, "learning_rate": 3.774193548387097e-06, "loss": 0.0679, "step": 14330 }, { "epoch": 92.38232931726908, "grad_norm": 4.268313407897949, "learning_rate": 3.770967741935484e-06, "loss": 0.1012, "step": 14331 }, { "epoch": 92.38875502008032, "grad_norm": 23.14268684387207, "learning_rate": 3.767741935483871e-06, "loss": 0.3574, "step": 14332 }, { "epoch": 92.39518072289157, "grad_norm": 11.24446964263916, "learning_rate": 3.7645161290322585e-06, "loss": 0.5051, "step": 14333 }, { "epoch": 92.40160642570281, "grad_norm": 9.296456336975098, "learning_rate": 3.7612903225806455e-06, "loss": 0.2329, "step": 14334 }, { "epoch": 92.40803212851405, "grad_norm": 19.529829025268555, "learning_rate": 3.7580645161290324e-06, "loss": 0.1963, "step": 14335 }, { "epoch": 92.41445783132531, "grad_norm": 2.818460464477539, "learning_rate": 3.75483870967742e-06, "loss": 0.1269, "step": 14336 }, { "epoch": 92.42088353413655, "grad_norm": 16.181245803833008, "learning_rate": 3.751612903225807e-06, "loss": 0.3297, "step": 14337 }, { "epoch": 92.42730923694779, "grad_norm": 6.723191738128662, "learning_rate": 3.748387096774194e-06, "loss": 0.1899, "step": 14338 }, { "epoch": 92.43373493975903, "grad_norm": 14.43718433380127, "learning_rate": 3.745161290322581e-06, "loss": 0.2695, "step": 14339 }, { "epoch": 92.44016064257028, "grad_norm": 14.276776313781738, "learning_rate": 3.741935483870968e-06, "loss": 0.5858, "step": 14340 }, { "epoch": 92.44658634538153, "grad_norm": 18.975322723388672, "learning_rate": 3.738709677419355e-06, "loss": 0.3541, "step": 14341 }, { "epoch": 92.45301204819278, "grad_norm": 8.420454978942871, "learning_rate": 3.7354838709677426e-06, "loss": 0.7162, "step": 14342 }, { "epoch": 92.45943775100402, "grad_norm": 14.72176456451416, "learning_rate": 3.7322580645161295e-06, "loss": 0.5147, "step": 14343 }, { "epoch": 92.46586345381526, "grad_norm": 9.09139633178711, "learning_rate": 3.7290322580645165e-06, "loss": 0.4582, "step": 14344 }, { "epoch": 92.4722891566265, "grad_norm": 22.247079849243164, "learning_rate": 3.725806451612903e-06, "loss": 0.3706, "step": 14345 }, { "epoch": 92.47871485943774, "grad_norm": 45.00745391845703, "learning_rate": 3.72258064516129e-06, "loss": 0.3285, "step": 14346 }, { "epoch": 92.485140562249, "grad_norm": 19.697616577148438, "learning_rate": 3.7193548387096774e-06, "loss": 0.3802, "step": 14347 }, { "epoch": 92.49156626506024, "grad_norm": 10.391802787780762, "learning_rate": 3.7161290322580644e-06, "loss": 0.4588, "step": 14348 }, { "epoch": 92.49799196787149, "grad_norm": 8.679645538330078, "learning_rate": 3.7129032258064514e-06, "loss": 0.2816, "step": 14349 }, { "epoch": 92.50441767068273, "grad_norm": 5.849952220916748, "learning_rate": 3.709677419354839e-06, "loss": 0.1561, "step": 14350 }, { "epoch": 92.51084337349397, "grad_norm": 3.5466620922088623, "learning_rate": 3.7064516129032258e-06, "loss": 0.1012, "step": 14351 }, { "epoch": 92.51726907630523, "grad_norm": 5.178757667541504, "learning_rate": 3.7032258064516127e-06, "loss": 0.1334, "step": 14352 }, { "epoch": 92.52369477911647, "grad_norm": 19.113183975219727, "learning_rate": 3.7e-06, "loss": 0.2787, "step": 14353 }, { "epoch": 92.53012048192771, "grad_norm": 19.521875381469727, "learning_rate": 3.696774193548387e-06, "loss": 0.6591, "step": 14354 }, { "epoch": 92.53654618473895, "grad_norm": 30.947389602661133, "learning_rate": 3.693548387096774e-06, "loss": 0.2372, "step": 14355 }, { "epoch": 92.5429718875502, "grad_norm": 10.338974952697754, "learning_rate": 3.6903225806451615e-06, "loss": 0.2777, "step": 14356 }, { "epoch": 92.54939759036145, "grad_norm": 11.429970741271973, "learning_rate": 3.6870967741935485e-06, "loss": 0.3938, "step": 14357 }, { "epoch": 92.5558232931727, "grad_norm": 10.206507682800293, "learning_rate": 3.6838709677419355e-06, "loss": 0.0984, "step": 14358 }, { "epoch": 92.56224899598394, "grad_norm": 9.863680839538574, "learning_rate": 3.680645161290323e-06, "loss": 0.1793, "step": 14359 }, { "epoch": 92.56867469879518, "grad_norm": 10.838850975036621, "learning_rate": 3.67741935483871e-06, "loss": 0.461, "step": 14360 }, { "epoch": 92.57510040160642, "grad_norm": 12.183812141418457, "learning_rate": 3.674193548387097e-06, "loss": 0.2491, "step": 14361 }, { "epoch": 92.58152610441768, "grad_norm": 8.237404823303223, "learning_rate": 3.6709677419354842e-06, "loss": 0.3917, "step": 14362 }, { "epoch": 92.58795180722892, "grad_norm": 27.899200439453125, "learning_rate": 3.667741935483871e-06, "loss": 0.4051, "step": 14363 }, { "epoch": 92.59437751004016, "grad_norm": 12.207844734191895, "learning_rate": 3.664516129032258e-06, "loss": 0.3729, "step": 14364 }, { "epoch": 92.6008032128514, "grad_norm": 6.476418972015381, "learning_rate": 3.6612903225806456e-06, "loss": 0.247, "step": 14365 }, { "epoch": 92.60722891566265, "grad_norm": 14.082891464233398, "learning_rate": 3.6580645161290326e-06, "loss": 0.3863, "step": 14366 }, { "epoch": 92.6136546184739, "grad_norm": 22.18630599975586, "learning_rate": 3.6548387096774195e-06, "loss": 0.3591, "step": 14367 }, { "epoch": 92.62008032128514, "grad_norm": 6.073556900024414, "learning_rate": 3.651612903225807e-06, "loss": 0.3501, "step": 14368 }, { "epoch": 92.62650602409639, "grad_norm": 16.47384262084961, "learning_rate": 3.648387096774194e-06, "loss": 0.3805, "step": 14369 }, { "epoch": 92.63293172690763, "grad_norm": 8.72144603729248, "learning_rate": 3.645161290322581e-06, "loss": 0.4319, "step": 14370 }, { "epoch": 92.63935742971887, "grad_norm": 8.715856552124023, "learning_rate": 3.6419354838709683e-06, "loss": 0.4475, "step": 14371 }, { "epoch": 92.64578313253013, "grad_norm": 3.2333836555480957, "learning_rate": 3.6387096774193553e-06, "loss": 0.4005, "step": 14372 }, { "epoch": 92.65220883534137, "grad_norm": 25.541584014892578, "learning_rate": 3.6354838709677422e-06, "loss": 0.4244, "step": 14373 }, { "epoch": 92.65863453815261, "grad_norm": 5.270405292510986, "learning_rate": 3.6322580645161296e-06, "loss": 0.2644, "step": 14374 }, { "epoch": 92.66506024096385, "grad_norm": 17.93520736694336, "learning_rate": 3.6290322580645166e-06, "loss": 0.3587, "step": 14375 }, { "epoch": 92.6714859437751, "grad_norm": 3.1089134216308594, "learning_rate": 3.6258064516129036e-06, "loss": 0.3961, "step": 14376 }, { "epoch": 92.67791164658635, "grad_norm": 6.137527942657471, "learning_rate": 3.6225806451612906e-06, "loss": 0.3304, "step": 14377 }, { "epoch": 92.6843373493976, "grad_norm": 12.834761619567871, "learning_rate": 3.619354838709678e-06, "loss": 0.2967, "step": 14378 }, { "epoch": 92.69076305220884, "grad_norm": 3.073209524154663, "learning_rate": 3.616129032258065e-06, "loss": 0.0847, "step": 14379 }, { "epoch": 92.69718875502008, "grad_norm": 7.210987091064453, "learning_rate": 3.612903225806452e-06, "loss": 0.2091, "step": 14380 }, { "epoch": 92.70361445783132, "grad_norm": 2.7032902240753174, "learning_rate": 3.6096774193548385e-06, "loss": 0.2943, "step": 14381 }, { "epoch": 92.71004016064256, "grad_norm": 8.248967170715332, "learning_rate": 3.6064516129032255e-06, "loss": 0.2506, "step": 14382 }, { "epoch": 92.71646586345382, "grad_norm": 8.033757209777832, "learning_rate": 3.603225806451613e-06, "loss": 0.3846, "step": 14383 }, { "epoch": 92.72289156626506, "grad_norm": 26.909568786621094, "learning_rate": 3.6e-06, "loss": 0.4432, "step": 14384 }, { "epoch": 92.7293172690763, "grad_norm": 14.941315650939941, "learning_rate": 3.596774193548387e-06, "loss": 0.5429, "step": 14385 }, { "epoch": 92.73574297188755, "grad_norm": 24.713228225708008, "learning_rate": 3.5935483870967742e-06, "loss": 0.3765, "step": 14386 }, { "epoch": 92.74216867469879, "grad_norm": 15.857542991638184, "learning_rate": 3.590322580645161e-06, "loss": 0.3897, "step": 14387 }, { "epoch": 92.74859437751005, "grad_norm": 12.864522933959961, "learning_rate": 3.587096774193548e-06, "loss": 0.1096, "step": 14388 }, { "epoch": 92.75502008032129, "grad_norm": 19.70058250427246, "learning_rate": 3.5838709677419356e-06, "loss": 0.2392, "step": 14389 }, { "epoch": 92.76144578313253, "grad_norm": 21.298431396484375, "learning_rate": 3.5806451612903225e-06, "loss": 0.3571, "step": 14390 }, { "epoch": 92.76787148594377, "grad_norm": 13.216983795166016, "learning_rate": 3.5774193548387095e-06, "loss": 0.3107, "step": 14391 }, { "epoch": 92.77429718875501, "grad_norm": 13.252659797668457, "learning_rate": 3.574193548387097e-06, "loss": 0.2557, "step": 14392 }, { "epoch": 92.78072289156627, "grad_norm": 16.318889617919922, "learning_rate": 3.570967741935484e-06, "loss": 0.4551, "step": 14393 }, { "epoch": 92.78714859437751, "grad_norm": 6.7481184005737305, "learning_rate": 3.567741935483871e-06, "loss": 0.3195, "step": 14394 }, { "epoch": 92.79357429718875, "grad_norm": 41.015098571777344, "learning_rate": 3.5645161290322583e-06, "loss": 0.3609, "step": 14395 }, { "epoch": 92.8, "grad_norm": 15.742323875427246, "learning_rate": 3.5612903225806453e-06, "loss": 0.359, "step": 14396 }, { "epoch": 92.80642570281124, "grad_norm": 10.627201080322266, "learning_rate": 3.5580645161290322e-06, "loss": 0.2662, "step": 14397 }, { "epoch": 92.8128514056225, "grad_norm": 27.398468017578125, "learning_rate": 3.5548387096774196e-06, "loss": 0.3083, "step": 14398 }, { "epoch": 92.81927710843374, "grad_norm": 38.34568405151367, "learning_rate": 3.5516129032258066e-06, "loss": 0.3267, "step": 14399 }, { "epoch": 92.82570281124498, "grad_norm": 14.429941177368164, "learning_rate": 3.5483870967741936e-06, "loss": 0.3705, "step": 14400 }, { "epoch": 92.83212851405622, "grad_norm": 19.839004516601562, "learning_rate": 3.545161290322581e-06, "loss": 0.313, "step": 14401 }, { "epoch": 92.83855421686746, "grad_norm": 5.584755897521973, "learning_rate": 3.541935483870968e-06, "loss": 0.3079, "step": 14402 }, { "epoch": 92.84497991967872, "grad_norm": 7.815654277801514, "learning_rate": 3.538709677419355e-06, "loss": 0.3904, "step": 14403 }, { "epoch": 92.85140562248996, "grad_norm": 12.787347793579102, "learning_rate": 3.5354838709677423e-06, "loss": 0.2578, "step": 14404 }, { "epoch": 92.8578313253012, "grad_norm": 14.511481285095215, "learning_rate": 3.5322580645161293e-06, "loss": 0.1498, "step": 14405 }, { "epoch": 92.86425702811245, "grad_norm": 8.079313278198242, "learning_rate": 3.5290322580645163e-06, "loss": 0.2948, "step": 14406 }, { "epoch": 92.87068273092369, "grad_norm": 17.396512985229492, "learning_rate": 3.5258064516129037e-06, "loss": 0.2536, "step": 14407 }, { "epoch": 92.87710843373495, "grad_norm": 7.1654133796691895, "learning_rate": 3.5225806451612907e-06, "loss": 0.2943, "step": 14408 }, { "epoch": 92.88353413654619, "grad_norm": 13.680750846862793, "learning_rate": 3.5193548387096777e-06, "loss": 0.3032, "step": 14409 }, { "epoch": 92.88995983935743, "grad_norm": 1.9425559043884277, "learning_rate": 3.516129032258065e-06, "loss": 0.0882, "step": 14410 }, { "epoch": 92.89638554216867, "grad_norm": 17.383760452270508, "learning_rate": 3.512903225806452e-06, "loss": 0.2265, "step": 14411 }, { "epoch": 92.90281124497992, "grad_norm": 18.282569885253906, "learning_rate": 3.509677419354839e-06, "loss": 0.4102, "step": 14412 }, { "epoch": 92.90923694779117, "grad_norm": 11.928258895874023, "learning_rate": 3.5064516129032264e-06, "loss": 0.3545, "step": 14413 }, { "epoch": 92.91566265060241, "grad_norm": 7.450326442718506, "learning_rate": 3.5032258064516134e-06, "loss": 0.3534, "step": 14414 }, { "epoch": 92.92208835341366, "grad_norm": 8.308058738708496, "learning_rate": 3.5000000000000004e-06, "loss": 0.2834, "step": 14415 }, { "epoch": 92.9285140562249, "grad_norm": 13.66911792755127, "learning_rate": 3.496774193548387e-06, "loss": 0.3673, "step": 14416 }, { "epoch": 92.93493975903614, "grad_norm": 20.979951858520508, "learning_rate": 3.493548387096774e-06, "loss": 0.2871, "step": 14417 }, { "epoch": 92.94136546184738, "grad_norm": 28.066129684448242, "learning_rate": 3.4903225806451613e-06, "loss": 0.5681, "step": 14418 }, { "epoch": 92.94779116465864, "grad_norm": 7.720947742462158, "learning_rate": 3.4870967741935483e-06, "loss": 0.2713, "step": 14419 }, { "epoch": 92.95421686746988, "grad_norm": 10.749557495117188, "learning_rate": 3.4838709677419353e-06, "loss": 0.7368, "step": 14420 }, { "epoch": 92.96064257028112, "grad_norm": 57.318702697753906, "learning_rate": 3.4806451612903227e-06, "loss": 0.4165, "step": 14421 }, { "epoch": 92.96706827309237, "grad_norm": 34.579315185546875, "learning_rate": 3.4774193548387096e-06, "loss": 0.5766, "step": 14422 }, { "epoch": 92.97349397590361, "grad_norm": 19.74081802368164, "learning_rate": 3.4741935483870966e-06, "loss": 0.3018, "step": 14423 }, { "epoch": 92.97991967871486, "grad_norm": 14.947749137878418, "learning_rate": 3.470967741935484e-06, "loss": 0.3674, "step": 14424 }, { "epoch": 92.9863453815261, "grad_norm": 25.177766799926758, "learning_rate": 3.467741935483871e-06, "loss": 0.3059, "step": 14425 }, { "epoch": 92.99277108433735, "grad_norm": 11.37771224975586, "learning_rate": 3.464516129032258e-06, "loss": 0.5428, "step": 14426 }, { "epoch": 93.00240963855421, "grad_norm": 16.770950317382812, "learning_rate": 3.4612903225806454e-06, "loss": 0.5427, "step": 14427 }, { "epoch": 93.00883534136547, "grad_norm": 23.70861053466797, "learning_rate": 3.4580645161290323e-06, "loss": 0.3968, "step": 14428 }, { "epoch": 93.01526104417671, "grad_norm": 12.754201889038086, "learning_rate": 3.4548387096774193e-06, "loss": 0.4811, "step": 14429 }, { "epoch": 93.02168674698795, "grad_norm": 6.073760986328125, "learning_rate": 3.4516129032258067e-06, "loss": 0.1477, "step": 14430 }, { "epoch": 93.0281124497992, "grad_norm": 4.4767889976501465, "learning_rate": 3.4483870967741937e-06, "loss": 0.1566, "step": 14431 }, { "epoch": 93.03453815261044, "grad_norm": 11.188156127929688, "learning_rate": 3.4451612903225807e-06, "loss": 0.2687, "step": 14432 }, { "epoch": 93.04096385542168, "grad_norm": 19.85924530029297, "learning_rate": 3.441935483870968e-06, "loss": 0.4243, "step": 14433 }, { "epoch": 93.04738955823294, "grad_norm": 14.22305679321289, "learning_rate": 3.438709677419355e-06, "loss": 0.2969, "step": 14434 }, { "epoch": 93.05381526104418, "grad_norm": 4.101675987243652, "learning_rate": 3.435483870967742e-06, "loss": 0.1655, "step": 14435 }, { "epoch": 93.06024096385542, "grad_norm": 4.178525447845459, "learning_rate": 3.4322580645161294e-06, "loss": 0.216, "step": 14436 }, { "epoch": 93.06666666666666, "grad_norm": 9.97656536102295, "learning_rate": 3.4290322580645164e-06, "loss": 0.2986, "step": 14437 }, { "epoch": 93.0730923694779, "grad_norm": 10.759495735168457, "learning_rate": 3.4258064516129034e-06, "loss": 0.6463, "step": 14438 }, { "epoch": 93.07951807228916, "grad_norm": 22.98220443725586, "learning_rate": 3.422580645161291e-06, "loss": 0.5235, "step": 14439 }, { "epoch": 93.0859437751004, "grad_norm": 7.596100330352783, "learning_rate": 3.4193548387096778e-06, "loss": 0.2941, "step": 14440 }, { "epoch": 93.09236947791165, "grad_norm": 11.023941040039062, "learning_rate": 3.4161290322580647e-06, "loss": 0.2291, "step": 14441 }, { "epoch": 93.09879518072289, "grad_norm": 8.807524681091309, "learning_rate": 3.412903225806452e-06, "loss": 0.4029, "step": 14442 }, { "epoch": 93.10522088353413, "grad_norm": 3.0386836528778076, "learning_rate": 3.409677419354839e-06, "loss": 0.0946, "step": 14443 }, { "epoch": 93.11164658634539, "grad_norm": 22.657100677490234, "learning_rate": 3.406451612903226e-06, "loss": 0.396, "step": 14444 }, { "epoch": 93.11807228915663, "grad_norm": 20.53308868408203, "learning_rate": 3.403225806451613e-06, "loss": 0.547, "step": 14445 }, { "epoch": 93.12449799196787, "grad_norm": 17.887191772460938, "learning_rate": 3.4000000000000005e-06, "loss": 0.1561, "step": 14446 }, { "epoch": 93.13092369477911, "grad_norm": 28.190185546875, "learning_rate": 3.3967741935483875e-06, "loss": 0.428, "step": 14447 }, { "epoch": 93.13734939759036, "grad_norm": 16.58814239501953, "learning_rate": 3.3935483870967744e-06, "loss": 0.6694, "step": 14448 }, { "epoch": 93.14377510040161, "grad_norm": 12.324519157409668, "learning_rate": 3.390322580645162e-06, "loss": 0.4328, "step": 14449 }, { "epoch": 93.15020080321285, "grad_norm": 3.0930802822113037, "learning_rate": 3.387096774193549e-06, "loss": 0.2484, "step": 14450 }, { "epoch": 93.1566265060241, "grad_norm": 6.046337604522705, "learning_rate": 3.383870967741936e-06, "loss": 0.3032, "step": 14451 }, { "epoch": 93.16305220883534, "grad_norm": 24.266456604003906, "learning_rate": 3.3806451612903223e-06, "loss": 0.3462, "step": 14452 }, { "epoch": 93.16947791164658, "grad_norm": 19.320707321166992, "learning_rate": 3.3774193548387093e-06, "loss": 0.7084, "step": 14453 }, { "epoch": 93.17590361445784, "grad_norm": 10.550289154052734, "learning_rate": 3.3741935483870967e-06, "loss": 0.2901, "step": 14454 }, { "epoch": 93.18232931726908, "grad_norm": 9.263237953186035, "learning_rate": 3.3709677419354837e-06, "loss": 0.3457, "step": 14455 }, { "epoch": 93.18875502008032, "grad_norm": 5.079331398010254, "learning_rate": 3.3677419354838707e-06, "loss": 0.1957, "step": 14456 }, { "epoch": 93.19518072289156, "grad_norm": 8.314498901367188, "learning_rate": 3.364516129032258e-06, "loss": 0.2887, "step": 14457 }, { "epoch": 93.2016064257028, "grad_norm": 10.623501777648926, "learning_rate": 3.361290322580645e-06, "loss": 0.22, "step": 14458 }, { "epoch": 93.20803212851406, "grad_norm": 11.298872947692871, "learning_rate": 3.358064516129032e-06, "loss": 0.1788, "step": 14459 }, { "epoch": 93.2144578313253, "grad_norm": 13.615965843200684, "learning_rate": 3.3548387096774194e-06, "loss": 0.3567, "step": 14460 }, { "epoch": 93.22088353413655, "grad_norm": 16.136579513549805, "learning_rate": 3.3516129032258064e-06, "loss": 0.4041, "step": 14461 }, { "epoch": 93.22730923694779, "grad_norm": 16.33208465576172, "learning_rate": 3.3483870967741934e-06, "loss": 0.491, "step": 14462 }, { "epoch": 93.23373493975903, "grad_norm": 25.483678817749023, "learning_rate": 3.345161290322581e-06, "loss": 0.2167, "step": 14463 }, { "epoch": 93.24016064257029, "grad_norm": 24.932729721069336, "learning_rate": 3.3419354838709678e-06, "loss": 0.134, "step": 14464 }, { "epoch": 93.24658634538153, "grad_norm": 5.4570465087890625, "learning_rate": 3.3387096774193547e-06, "loss": 0.2742, "step": 14465 }, { "epoch": 93.25301204819277, "grad_norm": 17.351757049560547, "learning_rate": 3.335483870967742e-06, "loss": 0.3416, "step": 14466 }, { "epoch": 93.25943775100401, "grad_norm": 10.88979721069336, "learning_rate": 3.332258064516129e-06, "loss": 0.4727, "step": 14467 }, { "epoch": 93.26586345381526, "grad_norm": 20.36587905883789, "learning_rate": 3.329032258064516e-06, "loss": 0.3958, "step": 14468 }, { "epoch": 93.2722891566265, "grad_norm": 14.801947593688965, "learning_rate": 3.3258064516129035e-06, "loss": 0.3796, "step": 14469 }, { "epoch": 93.27871485943776, "grad_norm": 12.47197151184082, "learning_rate": 3.3225806451612905e-06, "loss": 0.7571, "step": 14470 }, { "epoch": 93.285140562249, "grad_norm": 14.124266624450684, "learning_rate": 3.3193548387096775e-06, "loss": 0.3947, "step": 14471 }, { "epoch": 93.29156626506024, "grad_norm": 5.746678829193115, "learning_rate": 3.316129032258065e-06, "loss": 0.2638, "step": 14472 }, { "epoch": 93.29799196787148, "grad_norm": 7.4314117431640625, "learning_rate": 3.312903225806452e-06, "loss": 0.3457, "step": 14473 }, { "epoch": 93.30441767068272, "grad_norm": 19.41097640991211, "learning_rate": 3.309677419354839e-06, "loss": 0.1954, "step": 14474 }, { "epoch": 93.31084337349398, "grad_norm": 12.562381744384766, "learning_rate": 3.3064516129032262e-06, "loss": 0.5353, "step": 14475 }, { "epoch": 93.31726907630522, "grad_norm": 2.6381113529205322, "learning_rate": 3.303225806451613e-06, "loss": 0.0843, "step": 14476 }, { "epoch": 93.32369477911647, "grad_norm": 10.960721015930176, "learning_rate": 3.3e-06, "loss": 0.18, "step": 14477 }, { "epoch": 93.33012048192771, "grad_norm": 8.510088920593262, "learning_rate": 3.2967741935483876e-06, "loss": 0.4362, "step": 14478 }, { "epoch": 93.33654618473895, "grad_norm": 5.390844821929932, "learning_rate": 3.2935483870967745e-06, "loss": 0.1589, "step": 14479 }, { "epoch": 93.3429718875502, "grad_norm": 27.335681915283203, "learning_rate": 3.2903225806451615e-06, "loss": 0.4105, "step": 14480 }, { "epoch": 93.34939759036145, "grad_norm": 6.57299280166626, "learning_rate": 3.287096774193549e-06, "loss": 0.3197, "step": 14481 }, { "epoch": 93.35582329317269, "grad_norm": 11.135359764099121, "learning_rate": 3.283870967741936e-06, "loss": 0.2995, "step": 14482 }, { "epoch": 93.36224899598393, "grad_norm": 7.78124475479126, "learning_rate": 3.280645161290323e-06, "loss": 0.377, "step": 14483 }, { "epoch": 93.36867469879518, "grad_norm": 2.8129818439483643, "learning_rate": 3.2774193548387103e-06, "loss": 0.0961, "step": 14484 }, { "epoch": 93.37510040160643, "grad_norm": 6.60171365737915, "learning_rate": 3.2741935483870973e-06, "loss": 0.1992, "step": 14485 }, { "epoch": 93.38152610441767, "grad_norm": 10.859569549560547, "learning_rate": 3.2709677419354842e-06, "loss": 0.4175, "step": 14486 }, { "epoch": 93.38795180722892, "grad_norm": 4.039000511169434, "learning_rate": 3.267741935483871e-06, "loss": 0.1886, "step": 14487 }, { "epoch": 93.39437751004016, "grad_norm": 8.078177452087402, "learning_rate": 3.2645161290322578e-06, "loss": 0.3645, "step": 14488 }, { "epoch": 93.4008032128514, "grad_norm": 10.834370613098145, "learning_rate": 3.261290322580645e-06, "loss": 0.4491, "step": 14489 }, { "epoch": 93.40722891566266, "grad_norm": 5.824369430541992, "learning_rate": 3.258064516129032e-06, "loss": 0.1814, "step": 14490 }, { "epoch": 93.4136546184739, "grad_norm": 4.959320545196533, "learning_rate": 3.254838709677419e-06, "loss": 0.2024, "step": 14491 }, { "epoch": 93.42008032128514, "grad_norm": 12.433794975280762, "learning_rate": 3.2516129032258065e-06, "loss": 0.2654, "step": 14492 }, { "epoch": 93.42650602409638, "grad_norm": 10.823394775390625, "learning_rate": 3.2483870967741935e-06, "loss": 0.3493, "step": 14493 }, { "epoch": 93.43293172690763, "grad_norm": 4.484637260437012, "learning_rate": 3.2451612903225805e-06, "loss": 0.3013, "step": 14494 }, { "epoch": 93.43935742971888, "grad_norm": 6.963924407958984, "learning_rate": 3.241935483870968e-06, "loss": 0.2954, "step": 14495 }, { "epoch": 93.44578313253012, "grad_norm": 17.771343231201172, "learning_rate": 3.238709677419355e-06, "loss": 0.5116, "step": 14496 }, { "epoch": 93.45220883534137, "grad_norm": 24.807069778442383, "learning_rate": 3.235483870967742e-06, "loss": 0.3692, "step": 14497 }, { "epoch": 93.45863453815261, "grad_norm": 15.675200462341309, "learning_rate": 3.2322580645161292e-06, "loss": 0.5631, "step": 14498 }, { "epoch": 93.46506024096385, "grad_norm": 6.319476127624512, "learning_rate": 3.229032258064516e-06, "loss": 0.4868, "step": 14499 }, { "epoch": 93.47148594377511, "grad_norm": 18.60576629638672, "learning_rate": 3.225806451612903e-06, "loss": 0.5262, "step": 14500 }, { "epoch": 93.47791164658635, "grad_norm": 15.496716499328613, "learning_rate": 3.2225806451612906e-06, "loss": 0.247, "step": 14501 }, { "epoch": 93.48433734939759, "grad_norm": 17.949745178222656, "learning_rate": 3.2193548387096776e-06, "loss": 0.3662, "step": 14502 }, { "epoch": 93.49076305220883, "grad_norm": 5.529369354248047, "learning_rate": 3.2161290322580645e-06, "loss": 0.1419, "step": 14503 }, { "epoch": 93.49718875502008, "grad_norm": 17.221454620361328, "learning_rate": 3.212903225806452e-06, "loss": 0.2781, "step": 14504 }, { "epoch": 93.50361445783132, "grad_norm": 15.102007865905762, "learning_rate": 3.209677419354839e-06, "loss": 0.5945, "step": 14505 }, { "epoch": 93.51004016064257, "grad_norm": 9.64735221862793, "learning_rate": 3.206451612903226e-06, "loss": 0.1935, "step": 14506 }, { "epoch": 93.51646586345382, "grad_norm": 15.8397216796875, "learning_rate": 3.2032258064516133e-06, "loss": 0.4267, "step": 14507 }, { "epoch": 93.52289156626506, "grad_norm": 5.650059700012207, "learning_rate": 3.2000000000000003e-06, "loss": 0.1385, "step": 14508 }, { "epoch": 93.5293172690763, "grad_norm": 6.7883124351501465, "learning_rate": 3.1967741935483873e-06, "loss": 0.3659, "step": 14509 }, { "epoch": 93.53574297188754, "grad_norm": 8.808359146118164, "learning_rate": 3.1935483870967747e-06, "loss": 0.4235, "step": 14510 }, { "epoch": 93.5421686746988, "grad_norm": 24.914655685424805, "learning_rate": 3.1903225806451616e-06, "loss": 0.3616, "step": 14511 }, { "epoch": 93.54859437751004, "grad_norm": 20.77680778503418, "learning_rate": 3.1870967741935486e-06, "loss": 0.3115, "step": 14512 }, { "epoch": 93.55502008032128, "grad_norm": 7.507857322692871, "learning_rate": 3.1838709677419356e-06, "loss": 0.1551, "step": 14513 }, { "epoch": 93.56144578313253, "grad_norm": 14.99804401397705, "learning_rate": 3.180645161290323e-06, "loss": 0.5551, "step": 14514 }, { "epoch": 93.56787148594377, "grad_norm": 4.3658013343811035, "learning_rate": 3.17741935483871e-06, "loss": 0.2845, "step": 14515 }, { "epoch": 93.57429718875503, "grad_norm": 15.584761619567871, "learning_rate": 3.174193548387097e-06, "loss": 0.4028, "step": 14516 }, { "epoch": 93.58072289156627, "grad_norm": 18.528221130371094, "learning_rate": 3.1709677419354843e-06, "loss": 0.4274, "step": 14517 }, { "epoch": 93.58714859437751, "grad_norm": 14.575346946716309, "learning_rate": 3.1677419354838713e-06, "loss": 0.2097, "step": 14518 }, { "epoch": 93.59357429718875, "grad_norm": 8.880718231201172, "learning_rate": 3.1645161290322583e-06, "loss": 0.2678, "step": 14519 }, { "epoch": 93.6, "grad_norm": 8.134212493896484, "learning_rate": 3.1612903225806457e-06, "loss": 0.1209, "step": 14520 }, { "epoch": 93.60642570281125, "grad_norm": 8.9722318649292, "learning_rate": 3.1580645161290327e-06, "loss": 0.3806, "step": 14521 }, { "epoch": 93.6128514056225, "grad_norm": 5.1064372062683105, "learning_rate": 3.1548387096774197e-06, "loss": 0.1928, "step": 14522 }, { "epoch": 93.61927710843374, "grad_norm": 10.223021507263184, "learning_rate": 3.151612903225806e-06, "loss": 0.5603, "step": 14523 }, { "epoch": 93.62570281124498, "grad_norm": 9.223957061767578, "learning_rate": 3.148387096774193e-06, "loss": 0.2236, "step": 14524 }, { "epoch": 93.63212851405622, "grad_norm": 9.343523025512695, "learning_rate": 3.1451612903225806e-06, "loss": 0.2897, "step": 14525 }, { "epoch": 93.63855421686748, "grad_norm": 21.34851837158203, "learning_rate": 3.1419354838709676e-06, "loss": 0.661, "step": 14526 }, { "epoch": 93.64497991967872, "grad_norm": 12.438218116760254, "learning_rate": 3.1387096774193545e-06, "loss": 0.293, "step": 14527 }, { "epoch": 93.65140562248996, "grad_norm": 5.041207790374756, "learning_rate": 3.135483870967742e-06, "loss": 0.3787, "step": 14528 }, { "epoch": 93.6578313253012, "grad_norm": 10.161062240600586, "learning_rate": 3.132258064516129e-06, "loss": 0.1887, "step": 14529 }, { "epoch": 93.66425702811244, "grad_norm": 7.0593156814575195, "learning_rate": 3.129032258064516e-06, "loss": 0.3336, "step": 14530 }, { "epoch": 93.6706827309237, "grad_norm": 9.010624885559082, "learning_rate": 3.1258064516129033e-06, "loss": 0.4055, "step": 14531 }, { "epoch": 93.67710843373494, "grad_norm": 9.574532508850098, "learning_rate": 3.1225806451612903e-06, "loss": 0.5483, "step": 14532 }, { "epoch": 93.68353413654619, "grad_norm": 8.668885231018066, "learning_rate": 3.1193548387096773e-06, "loss": 0.4991, "step": 14533 }, { "epoch": 93.68995983935743, "grad_norm": 7.051167964935303, "learning_rate": 3.1161290322580647e-06, "loss": 0.1822, "step": 14534 }, { "epoch": 93.69638554216867, "grad_norm": 5.64151668548584, "learning_rate": 3.1129032258064516e-06, "loss": 0.1511, "step": 14535 }, { "epoch": 93.70281124497993, "grad_norm": 12.67479133605957, "learning_rate": 3.1096774193548386e-06, "loss": 0.5318, "step": 14536 }, { "epoch": 93.70923694779117, "grad_norm": 6.257437229156494, "learning_rate": 3.106451612903226e-06, "loss": 0.3529, "step": 14537 }, { "epoch": 93.71566265060241, "grad_norm": 18.051666259765625, "learning_rate": 3.103225806451613e-06, "loss": 0.341, "step": 14538 }, { "epoch": 93.72208835341365, "grad_norm": 28.05072784423828, "learning_rate": 3.1e-06, "loss": 0.5457, "step": 14539 }, { "epoch": 93.7285140562249, "grad_norm": 3.9112977981567383, "learning_rate": 3.0967741935483874e-06, "loss": 0.0754, "step": 14540 }, { "epoch": 93.73493975903614, "grad_norm": 6.300037860870361, "learning_rate": 3.0935483870967743e-06, "loss": 0.4164, "step": 14541 }, { "epoch": 93.7413654618474, "grad_norm": 17.910696029663086, "learning_rate": 3.0903225806451613e-06, "loss": 0.6926, "step": 14542 }, { "epoch": 93.74779116465864, "grad_norm": 11.279264450073242, "learning_rate": 3.0870967741935487e-06, "loss": 0.3611, "step": 14543 }, { "epoch": 93.75421686746988, "grad_norm": 25.7020320892334, "learning_rate": 3.0838709677419357e-06, "loss": 0.3708, "step": 14544 }, { "epoch": 93.76064257028112, "grad_norm": 19.51742172241211, "learning_rate": 3.0806451612903227e-06, "loss": 0.3175, "step": 14545 }, { "epoch": 93.76706827309236, "grad_norm": 10.32024097442627, "learning_rate": 3.07741935483871e-06, "loss": 0.2033, "step": 14546 }, { "epoch": 93.77349397590362, "grad_norm": 4.9239702224731445, "learning_rate": 3.074193548387097e-06, "loss": 0.1316, "step": 14547 }, { "epoch": 93.77991967871486, "grad_norm": 14.731832504272461, "learning_rate": 3.070967741935484e-06, "loss": 0.3658, "step": 14548 }, { "epoch": 93.7863453815261, "grad_norm": 9.511686325073242, "learning_rate": 3.0677419354838714e-06, "loss": 0.1551, "step": 14549 }, { "epoch": 93.79277108433735, "grad_norm": 8.150818824768066, "learning_rate": 3.064516129032258e-06, "loss": 0.1978, "step": 14550 }, { "epoch": 93.79919678714859, "grad_norm": 24.38567543029785, "learning_rate": 3.061290322580645e-06, "loss": 0.415, "step": 14551 }, { "epoch": 93.80562248995984, "grad_norm": 11.804464340209961, "learning_rate": 3.0580645161290324e-06, "loss": 0.4559, "step": 14552 }, { "epoch": 93.81204819277109, "grad_norm": 5.144803524017334, "learning_rate": 3.0548387096774193e-06, "loss": 0.2202, "step": 14553 }, { "epoch": 93.81847389558233, "grad_norm": 16.091812133789062, "learning_rate": 3.0516129032258063e-06, "loss": 0.2836, "step": 14554 }, { "epoch": 93.82489959839357, "grad_norm": 7.543270111083984, "learning_rate": 3.0483870967741937e-06, "loss": 0.2765, "step": 14555 }, { "epoch": 93.83132530120481, "grad_norm": 10.72646713256836, "learning_rate": 3.0451612903225807e-06, "loss": 0.2264, "step": 14556 }, { "epoch": 93.83775100401607, "grad_norm": 24.02692222595215, "learning_rate": 3.0419354838709677e-06, "loss": 0.5495, "step": 14557 }, { "epoch": 93.84417670682731, "grad_norm": 9.745963096618652, "learning_rate": 3.038709677419355e-06, "loss": 0.4168, "step": 14558 }, { "epoch": 93.85060240963855, "grad_norm": 24.504125595092773, "learning_rate": 3.035483870967742e-06, "loss": 0.3955, "step": 14559 }, { "epoch": 93.8570281124498, "grad_norm": 10.940658569335938, "learning_rate": 3.032258064516129e-06, "loss": 0.2256, "step": 14560 }, { "epoch": 93.86345381526104, "grad_norm": 15.825196266174316, "learning_rate": 3.0290322580645164e-06, "loss": 0.3898, "step": 14561 }, { "epoch": 93.8698795180723, "grad_norm": 8.952605247497559, "learning_rate": 3.0258064516129034e-06, "loss": 0.3819, "step": 14562 }, { "epoch": 93.87630522088354, "grad_norm": 12.44416332244873, "learning_rate": 3.0225806451612904e-06, "loss": 0.3967, "step": 14563 }, { "epoch": 93.88273092369478, "grad_norm": 15.800529479980469, "learning_rate": 3.0193548387096778e-06, "loss": 0.607, "step": 14564 }, { "epoch": 93.88915662650602, "grad_norm": 8.100207328796387, "learning_rate": 3.0161290322580648e-06, "loss": 0.3705, "step": 14565 }, { "epoch": 93.89558232931726, "grad_norm": 7.705081939697266, "learning_rate": 3.0129032258064517e-06, "loss": 0.3298, "step": 14566 }, { "epoch": 93.90200803212852, "grad_norm": 8.634535789489746, "learning_rate": 3.0096774193548387e-06, "loss": 0.2118, "step": 14567 }, { "epoch": 93.90843373493976, "grad_norm": 5.773900508880615, "learning_rate": 3.0064516129032257e-06, "loss": 0.1107, "step": 14568 }, { "epoch": 93.914859437751, "grad_norm": 10.405339241027832, "learning_rate": 3.003225806451613e-06, "loss": 0.2374, "step": 14569 }, { "epoch": 93.92128514056225, "grad_norm": 4.386605262756348, "learning_rate": 3e-06, "loss": 0.3074, "step": 14570 }, { "epoch": 93.92771084337349, "grad_norm": 10.079765319824219, "learning_rate": 2.996774193548387e-06, "loss": 0.3436, "step": 14571 }, { "epoch": 93.93413654618475, "grad_norm": 6.625021457672119, "learning_rate": 2.9935483870967745e-06, "loss": 0.31, "step": 14572 }, { "epoch": 93.94056224899599, "grad_norm": 12.235661506652832, "learning_rate": 2.9903225806451614e-06, "loss": 0.3464, "step": 14573 }, { "epoch": 93.94698795180723, "grad_norm": 8.063307762145996, "learning_rate": 2.9870967741935484e-06, "loss": 0.2564, "step": 14574 }, { "epoch": 93.95341365461847, "grad_norm": 15.74089241027832, "learning_rate": 2.983870967741936e-06, "loss": 0.2968, "step": 14575 }, { "epoch": 93.95983935742971, "grad_norm": 36.31997299194336, "learning_rate": 2.9806451612903228e-06, "loss": 0.1288, "step": 14576 }, { "epoch": 93.96626506024096, "grad_norm": 20.08683967590332, "learning_rate": 2.9774193548387098e-06, "loss": 0.4648, "step": 14577 }, { "epoch": 93.97269076305221, "grad_norm": 27.518596649169922, "learning_rate": 2.974193548387097e-06, "loss": 0.2554, "step": 14578 }, { "epoch": 93.97911646586346, "grad_norm": 7.441631317138672, "learning_rate": 2.970967741935484e-06, "loss": 0.2458, "step": 14579 }, { "epoch": 93.9855421686747, "grad_norm": 14.31341552734375, "learning_rate": 2.967741935483871e-06, "loss": 0.381, "step": 14580 }, { "epoch": 93.99196787148594, "grad_norm": 18.288122177124023, "learning_rate": 2.9645161290322585e-06, "loss": 0.2172, "step": 14581 }, { "epoch": 94.00160642570282, "grad_norm": 2.8844571113586426, "learning_rate": 2.9612903225806455e-06, "loss": 0.1248, "step": 14582 }, { "epoch": 94.00803212851406, "grad_norm": 6.673027992248535, "learning_rate": 2.9580645161290325e-06, "loss": 0.1282, "step": 14583 }, { "epoch": 94.0144578313253, "grad_norm": 9.683887481689453, "learning_rate": 2.9548387096774195e-06, "loss": 0.4647, "step": 14584 }, { "epoch": 94.02088353413654, "grad_norm": 9.02153205871582, "learning_rate": 2.9516129032258064e-06, "loss": 0.3265, "step": 14585 }, { "epoch": 94.02730923694779, "grad_norm": 10.853819847106934, "learning_rate": 2.9483870967741934e-06, "loss": 0.4645, "step": 14586 }, { "epoch": 94.03373493975904, "grad_norm": 54.35987091064453, "learning_rate": 2.945161290322581e-06, "loss": 0.4302, "step": 14587 }, { "epoch": 94.04016064257029, "grad_norm": 5.436905384063721, "learning_rate": 2.9419354838709678e-06, "loss": 0.1972, "step": 14588 }, { "epoch": 94.04658634538153, "grad_norm": 7.605123043060303, "learning_rate": 2.9387096774193548e-06, "loss": 0.2885, "step": 14589 }, { "epoch": 94.05301204819277, "grad_norm": 6.579733848571777, "learning_rate": 2.935483870967742e-06, "loss": 0.3096, "step": 14590 }, { "epoch": 94.05943775100401, "grad_norm": 6.677996635437012, "learning_rate": 2.932258064516129e-06, "loss": 0.377, "step": 14591 }, { "epoch": 94.06586345381525, "grad_norm": 12.86591911315918, "learning_rate": 2.929032258064516e-06, "loss": 0.4941, "step": 14592 }, { "epoch": 94.07228915662651, "grad_norm": 13.291104316711426, "learning_rate": 2.9258064516129035e-06, "loss": 0.2129, "step": 14593 }, { "epoch": 94.07871485943775, "grad_norm": 13.299737930297852, "learning_rate": 2.9225806451612905e-06, "loss": 0.2406, "step": 14594 }, { "epoch": 94.085140562249, "grad_norm": 12.54910945892334, "learning_rate": 2.9193548387096775e-06, "loss": 0.6361, "step": 14595 }, { "epoch": 94.09156626506024, "grad_norm": 8.547086715698242, "learning_rate": 2.916129032258065e-06, "loss": 0.0972, "step": 14596 }, { "epoch": 94.09799196787148, "grad_norm": 6.15116024017334, "learning_rate": 2.912903225806452e-06, "loss": 0.4427, "step": 14597 }, { "epoch": 94.10441767068274, "grad_norm": 10.610033988952637, "learning_rate": 2.909677419354839e-06, "loss": 0.328, "step": 14598 }, { "epoch": 94.11084337349398, "grad_norm": 5.719821453094482, "learning_rate": 2.9064516129032262e-06, "loss": 0.1218, "step": 14599 }, { "epoch": 94.11726907630522, "grad_norm": 16.231008529663086, "learning_rate": 2.903225806451613e-06, "loss": 0.4653, "step": 14600 }, { "epoch": 94.12369477911646, "grad_norm": 24.023983001708984, "learning_rate": 2.9e-06, "loss": 0.2807, "step": 14601 }, { "epoch": 94.1301204819277, "grad_norm": 24.110916137695312, "learning_rate": 2.8967741935483876e-06, "loss": 0.2432, "step": 14602 }, { "epoch": 94.13654618473896, "grad_norm": 6.951249122619629, "learning_rate": 2.893548387096774e-06, "loss": 0.2576, "step": 14603 }, { "epoch": 94.1429718875502, "grad_norm": 7.8462653160095215, "learning_rate": 2.890322580645161e-06, "loss": 0.241, "step": 14604 }, { "epoch": 94.14939759036145, "grad_norm": 10.958364486694336, "learning_rate": 2.8870967741935485e-06, "loss": 0.4578, "step": 14605 }, { "epoch": 94.15582329317269, "grad_norm": 7.1338629722595215, "learning_rate": 2.8838709677419355e-06, "loss": 0.1948, "step": 14606 }, { "epoch": 94.16224899598393, "grad_norm": 8.31074333190918, "learning_rate": 2.8806451612903225e-06, "loss": 0.3331, "step": 14607 }, { "epoch": 94.16867469879519, "grad_norm": 7.310613632202148, "learning_rate": 2.87741935483871e-06, "loss": 0.3174, "step": 14608 }, { "epoch": 94.17510040160643, "grad_norm": 9.633467674255371, "learning_rate": 2.874193548387097e-06, "loss": 0.2914, "step": 14609 }, { "epoch": 94.18152610441767, "grad_norm": 49.78889846801758, "learning_rate": 2.870967741935484e-06, "loss": 0.2226, "step": 14610 }, { "epoch": 94.18795180722891, "grad_norm": 13.889996528625488, "learning_rate": 2.8677419354838712e-06, "loss": 0.175, "step": 14611 }, { "epoch": 94.19437751004016, "grad_norm": 15.015727043151855, "learning_rate": 2.864516129032258e-06, "loss": 0.3349, "step": 14612 }, { "epoch": 94.20080321285141, "grad_norm": 9.078039169311523, "learning_rate": 2.861290322580645e-06, "loss": 0.3459, "step": 14613 }, { "epoch": 94.20722891566265, "grad_norm": 6.9566650390625, "learning_rate": 2.8580645161290326e-06, "loss": 0.2234, "step": 14614 }, { "epoch": 94.2136546184739, "grad_norm": 13.067700386047363, "learning_rate": 2.8548387096774196e-06, "loss": 0.3085, "step": 14615 }, { "epoch": 94.22008032128514, "grad_norm": 28.25640296936035, "learning_rate": 2.8516129032258065e-06, "loss": 0.3542, "step": 14616 }, { "epoch": 94.22650602409638, "grad_norm": 13.00396728515625, "learning_rate": 2.848387096774194e-06, "loss": 0.3982, "step": 14617 }, { "epoch": 94.23293172690764, "grad_norm": 3.8451550006866455, "learning_rate": 2.845161290322581e-06, "loss": 0.1329, "step": 14618 }, { "epoch": 94.23935742971888, "grad_norm": 50.07846450805664, "learning_rate": 2.841935483870968e-06, "loss": 0.4247, "step": 14619 }, { "epoch": 94.24578313253012, "grad_norm": 7.091166973114014, "learning_rate": 2.8387096774193553e-06, "loss": 0.1891, "step": 14620 }, { "epoch": 94.25220883534136, "grad_norm": 5.878461837768555, "learning_rate": 2.835483870967742e-06, "loss": 0.2626, "step": 14621 }, { "epoch": 94.2586345381526, "grad_norm": 22.12289810180664, "learning_rate": 2.832258064516129e-06, "loss": 0.2223, "step": 14622 }, { "epoch": 94.26506024096386, "grad_norm": 7.524128437042236, "learning_rate": 2.8290322580645162e-06, "loss": 0.3282, "step": 14623 }, { "epoch": 94.2714859437751, "grad_norm": 11.446085929870605, "learning_rate": 2.825806451612903e-06, "loss": 0.2979, "step": 14624 }, { "epoch": 94.27791164658635, "grad_norm": 8.852463722229004, "learning_rate": 2.82258064516129e-06, "loss": 0.4312, "step": 14625 }, { "epoch": 94.28433734939759, "grad_norm": 47.5977783203125, "learning_rate": 2.8193548387096776e-06, "loss": 0.365, "step": 14626 }, { "epoch": 94.29076305220883, "grad_norm": 5.43088436126709, "learning_rate": 2.8161290322580646e-06, "loss": 0.1348, "step": 14627 }, { "epoch": 94.29718875502007, "grad_norm": 17.831327438354492, "learning_rate": 2.8129032258064515e-06, "loss": 0.3154, "step": 14628 }, { "epoch": 94.30361445783133, "grad_norm": 11.68085765838623, "learning_rate": 2.809677419354839e-06, "loss": 0.4779, "step": 14629 }, { "epoch": 94.31004016064257, "grad_norm": 15.727706909179688, "learning_rate": 2.806451612903226e-06, "loss": 0.2685, "step": 14630 }, { "epoch": 94.31646586345381, "grad_norm": 13.484474182128906, "learning_rate": 2.803225806451613e-06, "loss": 0.3206, "step": 14631 }, { "epoch": 94.32289156626506, "grad_norm": 25.54563331604004, "learning_rate": 2.8000000000000003e-06, "loss": 0.2404, "step": 14632 }, { "epoch": 94.3293172690763, "grad_norm": 14.689909934997559, "learning_rate": 2.7967741935483873e-06, "loss": 0.5052, "step": 14633 }, { "epoch": 94.33574297188756, "grad_norm": 7.848817348480225, "learning_rate": 2.7935483870967743e-06, "loss": 0.2972, "step": 14634 }, { "epoch": 94.3421686746988, "grad_norm": 23.35886001586914, "learning_rate": 2.7903225806451617e-06, "loss": 0.3165, "step": 14635 }, { "epoch": 94.34859437751004, "grad_norm": 22.07911491394043, "learning_rate": 2.7870967741935486e-06, "loss": 0.4574, "step": 14636 }, { "epoch": 94.35502008032128, "grad_norm": 13.312625885009766, "learning_rate": 2.7838709677419356e-06, "loss": 0.3035, "step": 14637 }, { "epoch": 94.36144578313252, "grad_norm": 5.3651041984558105, "learning_rate": 2.7806451612903226e-06, "loss": 0.3402, "step": 14638 }, { "epoch": 94.36787148594378, "grad_norm": 6.567563533782959, "learning_rate": 2.7774193548387096e-06, "loss": 0.2889, "step": 14639 }, { "epoch": 94.37429718875502, "grad_norm": 10.607401847839355, "learning_rate": 2.774193548387097e-06, "loss": 0.4476, "step": 14640 }, { "epoch": 94.38072289156626, "grad_norm": 6.897310733795166, "learning_rate": 2.770967741935484e-06, "loss": 0.218, "step": 14641 }, { "epoch": 94.3871485943775, "grad_norm": 17.926687240600586, "learning_rate": 2.767741935483871e-06, "loss": 0.4639, "step": 14642 }, { "epoch": 94.39357429718875, "grad_norm": 19.459653854370117, "learning_rate": 2.7645161290322583e-06, "loss": 0.3508, "step": 14643 }, { "epoch": 94.4, "grad_norm": 5.426003456115723, "learning_rate": 2.7612903225806453e-06, "loss": 0.4256, "step": 14644 }, { "epoch": 94.40642570281125, "grad_norm": 4.052570343017578, "learning_rate": 2.7580645161290323e-06, "loss": 0.2313, "step": 14645 }, { "epoch": 94.41285140562249, "grad_norm": 8.077319145202637, "learning_rate": 2.7548387096774197e-06, "loss": 0.2754, "step": 14646 }, { "epoch": 94.41927710843373, "grad_norm": 4.101700782775879, "learning_rate": 2.7516129032258067e-06, "loss": 0.2046, "step": 14647 }, { "epoch": 94.42570281124497, "grad_norm": 41.299800872802734, "learning_rate": 2.7483870967741936e-06, "loss": 0.6297, "step": 14648 }, { "epoch": 94.43212851405623, "grad_norm": 11.070989608764648, "learning_rate": 2.745161290322581e-06, "loss": 0.2602, "step": 14649 }, { "epoch": 94.43855421686747, "grad_norm": 7.326385021209717, "learning_rate": 2.741935483870968e-06, "loss": 0.2479, "step": 14650 }, { "epoch": 94.44497991967872, "grad_norm": 11.192330360412598, "learning_rate": 2.738709677419355e-06, "loss": 0.8275, "step": 14651 }, { "epoch": 94.45140562248996, "grad_norm": 5.7510480880737305, "learning_rate": 2.735483870967742e-06, "loss": 0.1962, "step": 14652 }, { "epoch": 94.4578313253012, "grad_norm": 6.893479824066162, "learning_rate": 2.7322580645161294e-06, "loss": 0.3037, "step": 14653 }, { "epoch": 94.46425702811246, "grad_norm": 12.93110466003418, "learning_rate": 2.7290322580645163e-06, "loss": 0.3734, "step": 14654 }, { "epoch": 94.4706827309237, "grad_norm": 7.148435592651367, "learning_rate": 2.7258064516129033e-06, "loss": 0.1947, "step": 14655 }, { "epoch": 94.47710843373494, "grad_norm": 17.94976234436035, "learning_rate": 2.7225806451612903e-06, "loss": 0.2211, "step": 14656 }, { "epoch": 94.48353413654618, "grad_norm": 26.96068000793457, "learning_rate": 2.7193548387096773e-06, "loss": 0.6637, "step": 14657 }, { "epoch": 94.48995983935743, "grad_norm": 12.306183815002441, "learning_rate": 2.7161290322580647e-06, "loss": 0.3892, "step": 14658 }, { "epoch": 94.49638554216868, "grad_norm": 10.820113182067871, "learning_rate": 2.7129032258064517e-06, "loss": 0.3916, "step": 14659 }, { "epoch": 94.50281124497992, "grad_norm": 7.353091239929199, "learning_rate": 2.7096774193548386e-06, "loss": 0.1719, "step": 14660 }, { "epoch": 94.50923694779117, "grad_norm": 8.359915733337402, "learning_rate": 2.706451612903226e-06, "loss": 0.2073, "step": 14661 }, { "epoch": 94.51566265060241, "grad_norm": 13.341612815856934, "learning_rate": 2.703225806451613e-06, "loss": 0.372, "step": 14662 }, { "epoch": 94.52208835341365, "grad_norm": 6.243438243865967, "learning_rate": 2.7e-06, "loss": 0.2651, "step": 14663 }, { "epoch": 94.52851405622489, "grad_norm": 14.312261581420898, "learning_rate": 2.6967741935483874e-06, "loss": 0.4933, "step": 14664 }, { "epoch": 94.53493975903615, "grad_norm": 11.67774486541748, "learning_rate": 2.6935483870967744e-06, "loss": 0.5504, "step": 14665 }, { "epoch": 94.54136546184739, "grad_norm": 6.766262531280518, "learning_rate": 2.6903225806451613e-06, "loss": 0.4863, "step": 14666 }, { "epoch": 94.54779116465863, "grad_norm": 13.4021635055542, "learning_rate": 2.6870967741935487e-06, "loss": 0.337, "step": 14667 }, { "epoch": 94.55421686746988, "grad_norm": 6.9392194747924805, "learning_rate": 2.6838709677419357e-06, "loss": 0.2717, "step": 14668 }, { "epoch": 94.56064257028112, "grad_norm": 8.60723876953125, "learning_rate": 2.6806451612903227e-06, "loss": 0.2619, "step": 14669 }, { "epoch": 94.56706827309237, "grad_norm": 12.149943351745605, "learning_rate": 2.67741935483871e-06, "loss": 0.3793, "step": 14670 }, { "epoch": 94.57349397590362, "grad_norm": 12.902037620544434, "learning_rate": 2.674193548387097e-06, "loss": 0.2967, "step": 14671 }, { "epoch": 94.57991967871486, "grad_norm": 32.649288177490234, "learning_rate": 2.670967741935484e-06, "loss": 0.357, "step": 14672 }, { "epoch": 94.5863453815261, "grad_norm": 26.657995223999023, "learning_rate": 2.6677419354838715e-06, "loss": 0.1906, "step": 14673 }, { "epoch": 94.59277108433734, "grad_norm": 2.771854877471924, "learning_rate": 2.664516129032258e-06, "loss": 0.3016, "step": 14674 }, { "epoch": 94.5991967871486, "grad_norm": 9.128945350646973, "learning_rate": 2.661290322580645e-06, "loss": 0.2314, "step": 14675 }, { "epoch": 94.60562248995984, "grad_norm": 12.346458435058594, "learning_rate": 2.6580645161290324e-06, "loss": 0.2451, "step": 14676 }, { "epoch": 94.61204819277108, "grad_norm": 4.244537830352783, "learning_rate": 2.6548387096774194e-06, "loss": 0.239, "step": 14677 }, { "epoch": 94.61847389558233, "grad_norm": 10.167259216308594, "learning_rate": 2.6516129032258063e-06, "loss": 0.3853, "step": 14678 }, { "epoch": 94.62489959839357, "grad_norm": 9.960671424865723, "learning_rate": 2.6483870967741937e-06, "loss": 0.2088, "step": 14679 }, { "epoch": 94.63132530120482, "grad_norm": 20.79737663269043, "learning_rate": 2.6451612903225807e-06, "loss": 0.2676, "step": 14680 }, { "epoch": 94.63775100401607, "grad_norm": 20.717288970947266, "learning_rate": 2.6419354838709677e-06, "loss": 0.7568, "step": 14681 }, { "epoch": 94.64417670682731, "grad_norm": 4.6200175285339355, "learning_rate": 2.638709677419355e-06, "loss": 0.2576, "step": 14682 }, { "epoch": 94.65060240963855, "grad_norm": 9.834319114685059, "learning_rate": 2.635483870967742e-06, "loss": 0.4489, "step": 14683 }, { "epoch": 94.6570281124498, "grad_norm": 11.058889389038086, "learning_rate": 2.632258064516129e-06, "loss": 0.4929, "step": 14684 }, { "epoch": 94.66345381526105, "grad_norm": 15.0712251663208, "learning_rate": 2.6290322580645165e-06, "loss": 0.6294, "step": 14685 }, { "epoch": 94.66987951807229, "grad_norm": 16.813093185424805, "learning_rate": 2.6258064516129034e-06, "loss": 0.3348, "step": 14686 }, { "epoch": 94.67630522088353, "grad_norm": 14.25379753112793, "learning_rate": 2.6225806451612904e-06, "loss": 0.3331, "step": 14687 }, { "epoch": 94.68273092369478, "grad_norm": 28.348175048828125, "learning_rate": 2.619354838709678e-06, "loss": 0.3062, "step": 14688 }, { "epoch": 94.68915662650602, "grad_norm": 10.271477699279785, "learning_rate": 2.6161290322580648e-06, "loss": 0.3102, "step": 14689 }, { "epoch": 94.69558232931728, "grad_norm": 10.8275785446167, "learning_rate": 2.6129032258064518e-06, "loss": 0.3479, "step": 14690 }, { "epoch": 94.70200803212852, "grad_norm": 9.12473201751709, "learning_rate": 2.609677419354839e-06, "loss": 0.3674, "step": 14691 }, { "epoch": 94.70843373493976, "grad_norm": 22.28127670288086, "learning_rate": 2.6064516129032257e-06, "loss": 0.1565, "step": 14692 }, { "epoch": 94.714859437751, "grad_norm": 16.02029800415039, "learning_rate": 2.6032258064516127e-06, "loss": 0.3521, "step": 14693 }, { "epoch": 94.72128514056224, "grad_norm": 9.081704139709473, "learning_rate": 2.6e-06, "loss": 0.2403, "step": 14694 }, { "epoch": 94.7277108433735, "grad_norm": 11.082864761352539, "learning_rate": 2.596774193548387e-06, "loss": 0.3166, "step": 14695 }, { "epoch": 94.73413654618474, "grad_norm": 9.436758041381836, "learning_rate": 2.593548387096774e-06, "loss": 0.3096, "step": 14696 }, { "epoch": 94.74056224899599, "grad_norm": 8.340085983276367, "learning_rate": 2.5903225806451614e-06, "loss": 0.3427, "step": 14697 }, { "epoch": 94.74698795180723, "grad_norm": 8.495072364807129, "learning_rate": 2.5870967741935484e-06, "loss": 0.2065, "step": 14698 }, { "epoch": 94.75341365461847, "grad_norm": 4.830438137054443, "learning_rate": 2.5838709677419354e-06, "loss": 0.2262, "step": 14699 }, { "epoch": 94.75983935742971, "grad_norm": 3.229382276535034, "learning_rate": 2.580645161290323e-06, "loss": 0.2917, "step": 14700 }, { "epoch": 94.76626506024097, "grad_norm": 25.473636627197266, "learning_rate": 2.5774193548387098e-06, "loss": 0.3339, "step": 14701 }, { "epoch": 94.77269076305221, "grad_norm": 13.740643501281738, "learning_rate": 2.5741935483870968e-06, "loss": 0.4709, "step": 14702 }, { "epoch": 94.77911646586345, "grad_norm": 8.464680671691895, "learning_rate": 2.570967741935484e-06, "loss": 0.3206, "step": 14703 }, { "epoch": 94.7855421686747, "grad_norm": 15.241960525512695, "learning_rate": 2.567741935483871e-06, "loss": 0.229, "step": 14704 }, { "epoch": 94.79196787148594, "grad_norm": 13.263132095336914, "learning_rate": 2.564516129032258e-06, "loss": 0.3785, "step": 14705 }, { "epoch": 94.7983935742972, "grad_norm": 7.055060386657715, "learning_rate": 2.5612903225806455e-06, "loss": 0.3234, "step": 14706 }, { "epoch": 94.80481927710844, "grad_norm": 19.689311981201172, "learning_rate": 2.5580645161290325e-06, "loss": 0.1288, "step": 14707 }, { "epoch": 94.81124497991968, "grad_norm": 7.68066930770874, "learning_rate": 2.5548387096774195e-06, "loss": 0.1982, "step": 14708 }, { "epoch": 94.81767068273092, "grad_norm": 16.029130935668945, "learning_rate": 2.551612903225807e-06, "loss": 0.2929, "step": 14709 }, { "epoch": 94.82409638554216, "grad_norm": 11.858596801757812, "learning_rate": 2.5483870967741934e-06, "loss": 0.2772, "step": 14710 }, { "epoch": 94.83052208835342, "grad_norm": 14.527853965759277, "learning_rate": 2.545161290322581e-06, "loss": 0.149, "step": 14711 }, { "epoch": 94.83694779116466, "grad_norm": 16.02486801147461, "learning_rate": 2.541935483870968e-06, "loss": 0.6096, "step": 14712 }, { "epoch": 94.8433734939759, "grad_norm": 8.455117225646973, "learning_rate": 2.5387096774193548e-06, "loss": 0.211, "step": 14713 }, { "epoch": 94.84979919678715, "grad_norm": 9.53731632232666, "learning_rate": 2.535483870967742e-06, "loss": 0.151, "step": 14714 }, { "epoch": 94.85622489959839, "grad_norm": 17.92375946044922, "learning_rate": 2.532258064516129e-06, "loss": 0.6144, "step": 14715 }, { "epoch": 94.86265060240964, "grad_norm": 11.459017753601074, "learning_rate": 2.529032258064516e-06, "loss": 0.3339, "step": 14716 }, { "epoch": 94.86907630522089, "grad_norm": 6.305418968200684, "learning_rate": 2.5258064516129035e-06, "loss": 0.278, "step": 14717 }, { "epoch": 94.87550200803213, "grad_norm": 28.037654876708984, "learning_rate": 2.5225806451612905e-06, "loss": 0.5, "step": 14718 }, { "epoch": 94.88192771084337, "grad_norm": 9.358964920043945, "learning_rate": 2.5193548387096775e-06, "loss": 0.3064, "step": 14719 }, { "epoch": 94.88835341365461, "grad_norm": 22.705142974853516, "learning_rate": 2.5161290322580645e-06, "loss": 0.4141, "step": 14720 }, { "epoch": 94.89477911646587, "grad_norm": 11.423582077026367, "learning_rate": 2.512903225806452e-06, "loss": 0.2184, "step": 14721 }, { "epoch": 94.90120481927711, "grad_norm": 15.342230796813965, "learning_rate": 2.509677419354839e-06, "loss": 0.6119, "step": 14722 }, { "epoch": 94.90763052208835, "grad_norm": 12.639937400817871, "learning_rate": 2.506451612903226e-06, "loss": 0.2199, "step": 14723 }, { "epoch": 94.9140562248996, "grad_norm": 6.485754013061523, "learning_rate": 2.5032258064516132e-06, "loss": 0.29, "step": 14724 }, { "epoch": 94.92048192771084, "grad_norm": 8.724000930786133, "learning_rate": 2.5e-06, "loss": 0.2705, "step": 14725 }, { "epoch": 94.9269076305221, "grad_norm": 24.374502182006836, "learning_rate": 2.496774193548387e-06, "loss": 0.3684, "step": 14726 }, { "epoch": 94.93333333333334, "grad_norm": 18.395736694335938, "learning_rate": 2.493548387096774e-06, "loss": 0.5797, "step": 14727 }, { "epoch": 94.93975903614458, "grad_norm": 51.96238327026367, "learning_rate": 2.490322580645161e-06, "loss": 0.7307, "step": 14728 }, { "epoch": 94.94618473895582, "grad_norm": 4.665500164031982, "learning_rate": 2.4870967741935485e-06, "loss": 0.2414, "step": 14729 }, { "epoch": 94.95261044176706, "grad_norm": 8.727834701538086, "learning_rate": 2.4838709677419355e-06, "loss": 0.3786, "step": 14730 }, { "epoch": 94.95903614457832, "grad_norm": 14.213526725769043, "learning_rate": 2.4806451612903225e-06, "loss": 0.3932, "step": 14731 }, { "epoch": 94.96546184738956, "grad_norm": 29.11087417602539, "learning_rate": 2.47741935483871e-06, "loss": 0.4439, "step": 14732 }, { "epoch": 94.9718875502008, "grad_norm": 8.593742370605469, "learning_rate": 2.474193548387097e-06, "loss": 0.3173, "step": 14733 }, { "epoch": 94.97831325301205, "grad_norm": 11.949490547180176, "learning_rate": 2.470967741935484e-06, "loss": 0.3351, "step": 14734 }, { "epoch": 94.98473895582329, "grad_norm": 6.387451171875, "learning_rate": 2.4677419354838712e-06, "loss": 0.1118, "step": 14735 }, { "epoch": 94.99116465863453, "grad_norm": 15.960994720458984, "learning_rate": 2.4645161290322582e-06, "loss": 0.4868, "step": 14736 }, { "epoch": 95.00080321285141, "grad_norm": 49.35574722290039, "learning_rate": 2.461290322580645e-06, "loss": 0.5345, "step": 14737 }, { "epoch": 95.00722891566265, "grad_norm": 12.311368942260742, "learning_rate": 2.4580645161290326e-06, "loss": 0.487, "step": 14738 }, { "epoch": 95.0136546184739, "grad_norm": 10.348165512084961, "learning_rate": 2.4548387096774196e-06, "loss": 0.4036, "step": 14739 }, { "epoch": 95.02008032128514, "grad_norm": 20.955093383789062, "learning_rate": 2.4516129032258066e-06, "loss": 0.4564, "step": 14740 }, { "epoch": 95.02650602409639, "grad_norm": 6.7185187339782715, "learning_rate": 2.448387096774194e-06, "loss": 0.1979, "step": 14741 }, { "epoch": 95.03293172690763, "grad_norm": 7.3613057136535645, "learning_rate": 2.445161290322581e-06, "loss": 0.2696, "step": 14742 }, { "epoch": 95.03935742971888, "grad_norm": 7.196136951446533, "learning_rate": 2.441935483870968e-06, "loss": 0.3069, "step": 14743 }, { "epoch": 95.04578313253012, "grad_norm": 8.377595901489258, "learning_rate": 2.4387096774193553e-06, "loss": 0.3095, "step": 14744 }, { "epoch": 95.05220883534136, "grad_norm": 23.907556533813477, "learning_rate": 2.435483870967742e-06, "loss": 0.5985, "step": 14745 }, { "epoch": 95.05863453815262, "grad_norm": 7.451432228088379, "learning_rate": 2.432258064516129e-06, "loss": 0.1374, "step": 14746 }, { "epoch": 95.06506024096386, "grad_norm": 15.486669540405273, "learning_rate": 2.4290322580645162e-06, "loss": 0.8815, "step": 14747 }, { "epoch": 95.0714859437751, "grad_norm": 15.896920204162598, "learning_rate": 2.4258064516129032e-06, "loss": 0.3021, "step": 14748 }, { "epoch": 95.07791164658634, "grad_norm": 5.866889476776123, "learning_rate": 2.42258064516129e-06, "loss": 0.2755, "step": 14749 }, { "epoch": 95.08433734939759, "grad_norm": 11.269192695617676, "learning_rate": 2.4193548387096776e-06, "loss": 0.4622, "step": 14750 }, { "epoch": 95.09076305220883, "grad_norm": 8.407427787780762, "learning_rate": 2.4161290322580646e-06, "loss": 0.1271, "step": 14751 }, { "epoch": 95.09718875502008, "grad_norm": 10.680570602416992, "learning_rate": 2.4129032258064516e-06, "loss": 0.2678, "step": 14752 }, { "epoch": 95.10361445783133, "grad_norm": 18.62214469909668, "learning_rate": 2.409677419354839e-06, "loss": 0.4736, "step": 14753 }, { "epoch": 95.11004016064257, "grad_norm": 12.056745529174805, "learning_rate": 2.406451612903226e-06, "loss": 0.375, "step": 14754 }, { "epoch": 95.11646586345381, "grad_norm": 14.992303848266602, "learning_rate": 2.403225806451613e-06, "loss": 0.3027, "step": 14755 }, { "epoch": 95.12289156626505, "grad_norm": 2.9058752059936523, "learning_rate": 2.4000000000000003e-06, "loss": 0.1363, "step": 14756 }, { "epoch": 95.12931726907631, "grad_norm": 7.536345481872559, "learning_rate": 2.3967741935483873e-06, "loss": 0.2955, "step": 14757 }, { "epoch": 95.13574297188755, "grad_norm": 6.682382583618164, "learning_rate": 2.3935483870967743e-06, "loss": 0.1802, "step": 14758 }, { "epoch": 95.1421686746988, "grad_norm": 10.831971168518066, "learning_rate": 2.3903225806451617e-06, "loss": 0.3928, "step": 14759 }, { "epoch": 95.14859437751004, "grad_norm": 9.67151927947998, "learning_rate": 2.3870967741935486e-06, "loss": 0.3655, "step": 14760 }, { "epoch": 95.15502008032128, "grad_norm": 4.521610736846924, "learning_rate": 2.3838709677419356e-06, "loss": 0.1785, "step": 14761 }, { "epoch": 95.16144578313254, "grad_norm": 7.803110122680664, "learning_rate": 2.380645161290323e-06, "loss": 0.2495, "step": 14762 }, { "epoch": 95.16787148594378, "grad_norm": 7.444597244262695, "learning_rate": 2.3774193548387096e-06, "loss": 0.3581, "step": 14763 }, { "epoch": 95.17429718875502, "grad_norm": 16.45994758605957, "learning_rate": 2.3741935483870966e-06, "loss": 0.3099, "step": 14764 }, { "epoch": 95.18072289156626, "grad_norm": 7.053325176239014, "learning_rate": 2.370967741935484e-06, "loss": 0.2231, "step": 14765 }, { "epoch": 95.1871485943775, "grad_norm": 10.450323104858398, "learning_rate": 2.367741935483871e-06, "loss": 0.5144, "step": 14766 }, { "epoch": 95.19357429718876, "grad_norm": 7.60748291015625, "learning_rate": 2.364516129032258e-06, "loss": 0.3686, "step": 14767 }, { "epoch": 95.2, "grad_norm": 8.283978462219238, "learning_rate": 2.3612903225806453e-06, "loss": 0.2381, "step": 14768 }, { "epoch": 95.20642570281125, "grad_norm": 9.408421516418457, "learning_rate": 2.3580645161290323e-06, "loss": 0.2793, "step": 14769 }, { "epoch": 95.21285140562249, "grad_norm": 10.245034217834473, "learning_rate": 2.3548387096774193e-06, "loss": 0.2865, "step": 14770 }, { "epoch": 95.21927710843373, "grad_norm": 7.916601181030273, "learning_rate": 2.3516129032258067e-06, "loss": 0.224, "step": 14771 }, { "epoch": 95.22570281124499, "grad_norm": 12.289162635803223, "learning_rate": 2.3483870967741936e-06, "loss": 0.4175, "step": 14772 }, { "epoch": 95.23212851405623, "grad_norm": 9.449257850646973, "learning_rate": 2.3451612903225806e-06, "loss": 0.2946, "step": 14773 }, { "epoch": 95.23855421686747, "grad_norm": 14.009941101074219, "learning_rate": 2.341935483870968e-06, "loss": 0.3433, "step": 14774 }, { "epoch": 95.24497991967871, "grad_norm": 13.252552032470703, "learning_rate": 2.338709677419355e-06, "loss": 0.2191, "step": 14775 }, { "epoch": 95.25140562248995, "grad_norm": 8.759364128112793, "learning_rate": 2.335483870967742e-06, "loss": 0.4746, "step": 14776 }, { "epoch": 95.25783132530121, "grad_norm": 86.04593658447266, "learning_rate": 2.3322580645161294e-06, "loss": 0.659, "step": 14777 }, { "epoch": 95.26425702811245, "grad_norm": 5.364523410797119, "learning_rate": 2.3290322580645164e-06, "loss": 0.1794, "step": 14778 }, { "epoch": 95.2706827309237, "grad_norm": 14.919647216796875, "learning_rate": 2.3258064516129033e-06, "loss": 0.4549, "step": 14779 }, { "epoch": 95.27710843373494, "grad_norm": 12.128987312316895, "learning_rate": 2.3225806451612907e-06, "loss": 0.1805, "step": 14780 }, { "epoch": 95.28353413654618, "grad_norm": 7.867485523223877, "learning_rate": 2.3193548387096773e-06, "loss": 0.4078, "step": 14781 }, { "epoch": 95.28995983935744, "grad_norm": 10.849974632263184, "learning_rate": 2.3161290322580647e-06, "loss": 0.6687, "step": 14782 }, { "epoch": 95.29638554216868, "grad_norm": 10.896285057067871, "learning_rate": 2.3129032258064517e-06, "loss": 0.4323, "step": 14783 }, { "epoch": 95.30281124497992, "grad_norm": 6.577454090118408, "learning_rate": 2.3096774193548386e-06, "loss": 0.2936, "step": 14784 }, { "epoch": 95.30923694779116, "grad_norm": 9.222469329833984, "learning_rate": 2.306451612903226e-06, "loss": 0.2961, "step": 14785 }, { "epoch": 95.3156626506024, "grad_norm": 15.194891929626465, "learning_rate": 2.303225806451613e-06, "loss": 0.4647, "step": 14786 }, { "epoch": 95.32208835341365, "grad_norm": 8.394553184509277, "learning_rate": 2.3e-06, "loss": 0.2251, "step": 14787 }, { "epoch": 95.3285140562249, "grad_norm": 15.108988761901855, "learning_rate": 2.296774193548387e-06, "loss": 0.3551, "step": 14788 }, { "epoch": 95.33493975903615, "grad_norm": 13.112022399902344, "learning_rate": 2.2935483870967744e-06, "loss": 0.4511, "step": 14789 }, { "epoch": 95.34136546184739, "grad_norm": 7.42791748046875, "learning_rate": 2.2903225806451614e-06, "loss": 0.3888, "step": 14790 }, { "epoch": 95.34779116465863, "grad_norm": 18.920610427856445, "learning_rate": 2.2870967741935483e-06, "loss": 0.4691, "step": 14791 }, { "epoch": 95.35421686746987, "grad_norm": 11.860197067260742, "learning_rate": 2.2838709677419357e-06, "loss": 0.4558, "step": 14792 }, { "epoch": 95.36064257028113, "grad_norm": 11.896081924438477, "learning_rate": 2.2806451612903227e-06, "loss": 0.4366, "step": 14793 }, { "epoch": 95.36706827309237, "grad_norm": 21.627662658691406, "learning_rate": 2.2774193548387097e-06, "loss": 0.2806, "step": 14794 }, { "epoch": 95.37349397590361, "grad_norm": 15.212173461914062, "learning_rate": 2.274193548387097e-06, "loss": 0.3074, "step": 14795 }, { "epoch": 95.37991967871486, "grad_norm": 10.066771507263184, "learning_rate": 2.270967741935484e-06, "loss": 0.1336, "step": 14796 }, { "epoch": 95.3863453815261, "grad_norm": 8.521014213562012, "learning_rate": 2.267741935483871e-06, "loss": 0.4157, "step": 14797 }, { "epoch": 95.39277108433735, "grad_norm": 7.964627265930176, "learning_rate": 2.264516129032258e-06, "loss": 0.2818, "step": 14798 }, { "epoch": 95.3991967871486, "grad_norm": 15.100990295410156, "learning_rate": 2.261290322580645e-06, "loss": 0.218, "step": 14799 }, { "epoch": 95.40562248995984, "grad_norm": 18.25421905517578, "learning_rate": 2.2580645161290324e-06, "loss": 0.7036, "step": 14800 }, { "epoch": 95.41204819277108, "grad_norm": 21.249675750732422, "learning_rate": 2.2548387096774194e-06, "loss": 0.1963, "step": 14801 }, { "epoch": 95.41847389558232, "grad_norm": 11.172090530395508, "learning_rate": 2.2516129032258064e-06, "loss": 0.2433, "step": 14802 }, { "epoch": 95.42489959839358, "grad_norm": 7.941795825958252, "learning_rate": 2.2483870967741938e-06, "loss": 0.3671, "step": 14803 }, { "epoch": 95.43132530120482, "grad_norm": 10.016158103942871, "learning_rate": 2.2451612903225807e-06, "loss": 0.2098, "step": 14804 }, { "epoch": 95.43775100401606, "grad_norm": 11.61634635925293, "learning_rate": 2.2419354838709677e-06, "loss": 0.2938, "step": 14805 }, { "epoch": 95.4441767068273, "grad_norm": 7.121523380279541, "learning_rate": 2.238709677419355e-06, "loss": 0.2981, "step": 14806 }, { "epoch": 95.45060240963855, "grad_norm": 12.8706636428833, "learning_rate": 2.235483870967742e-06, "loss": 0.6346, "step": 14807 }, { "epoch": 95.4570281124498, "grad_norm": 7.306699275970459, "learning_rate": 2.232258064516129e-06, "loss": 0.4236, "step": 14808 }, { "epoch": 95.46345381526105, "grad_norm": 8.165836334228516, "learning_rate": 2.2290322580645165e-06, "loss": 0.3138, "step": 14809 }, { "epoch": 95.46987951807229, "grad_norm": 16.59922981262207, "learning_rate": 2.2258064516129034e-06, "loss": 0.5531, "step": 14810 }, { "epoch": 95.47630522088353, "grad_norm": 6.924594402313232, "learning_rate": 2.2225806451612904e-06, "loss": 0.3581, "step": 14811 }, { "epoch": 95.48273092369477, "grad_norm": 8.214510917663574, "learning_rate": 2.219354838709678e-06, "loss": 0.3733, "step": 14812 }, { "epoch": 95.48915662650603, "grad_norm": 15.697650909423828, "learning_rate": 2.216129032258065e-06, "loss": 0.3565, "step": 14813 }, { "epoch": 95.49558232931727, "grad_norm": 13.647473335266113, "learning_rate": 2.2129032258064518e-06, "loss": 0.628, "step": 14814 }, { "epoch": 95.50200803212851, "grad_norm": 11.961152076721191, "learning_rate": 2.209677419354839e-06, "loss": 0.5365, "step": 14815 }, { "epoch": 95.50843373493976, "grad_norm": 5.71742057800293, "learning_rate": 2.2064516129032257e-06, "loss": 0.1823, "step": 14816 }, { "epoch": 95.514859437751, "grad_norm": 5.702231407165527, "learning_rate": 2.2032258064516127e-06, "loss": 0.3108, "step": 14817 }, { "epoch": 95.52128514056226, "grad_norm": 8.204208374023438, "learning_rate": 2.2e-06, "loss": 0.3099, "step": 14818 }, { "epoch": 95.5277108433735, "grad_norm": 50.02544021606445, "learning_rate": 2.196774193548387e-06, "loss": 0.7179, "step": 14819 }, { "epoch": 95.53413654618474, "grad_norm": 10.45393180847168, "learning_rate": 2.193548387096774e-06, "loss": 0.2916, "step": 14820 }, { "epoch": 95.54056224899598, "grad_norm": 21.16009521484375, "learning_rate": 2.1903225806451615e-06, "loss": 0.359, "step": 14821 }, { "epoch": 95.54698795180722, "grad_norm": 6.342492580413818, "learning_rate": 2.1870967741935484e-06, "loss": 0.3999, "step": 14822 }, { "epoch": 95.55341365461847, "grad_norm": 5.143921375274658, "learning_rate": 2.1838709677419354e-06, "loss": 0.1722, "step": 14823 }, { "epoch": 95.55983935742972, "grad_norm": 6.9624481201171875, "learning_rate": 2.180645161290323e-06, "loss": 0.3628, "step": 14824 }, { "epoch": 95.56626506024097, "grad_norm": 6.599892616271973, "learning_rate": 2.17741935483871e-06, "loss": 0.1604, "step": 14825 }, { "epoch": 95.57269076305221, "grad_norm": 7.994511604309082, "learning_rate": 2.1741935483870968e-06, "loss": 0.3662, "step": 14826 }, { "epoch": 95.57911646586345, "grad_norm": 13.323959350585938, "learning_rate": 2.170967741935484e-06, "loss": 0.3547, "step": 14827 }, { "epoch": 95.58554216867469, "grad_norm": 7.158890724182129, "learning_rate": 2.167741935483871e-06, "loss": 0.2404, "step": 14828 }, { "epoch": 95.59196787148595, "grad_norm": 48.10688400268555, "learning_rate": 2.164516129032258e-06, "loss": 0.3464, "step": 14829 }, { "epoch": 95.59839357429719, "grad_norm": 7.250800132751465, "learning_rate": 2.1612903225806455e-06, "loss": 0.08, "step": 14830 }, { "epoch": 95.60481927710843, "grad_norm": 10.632303237915039, "learning_rate": 2.1580645161290325e-06, "loss": 0.3372, "step": 14831 }, { "epoch": 95.61124497991968, "grad_norm": 18.14885902404785, "learning_rate": 2.1548387096774195e-06, "loss": 0.2117, "step": 14832 }, { "epoch": 95.61767068273092, "grad_norm": 5.887693405151367, "learning_rate": 2.151612903225807e-06, "loss": 0.3027, "step": 14833 }, { "epoch": 95.62409638554217, "grad_norm": 10.97258472442627, "learning_rate": 2.1483870967741934e-06, "loss": 0.4259, "step": 14834 }, { "epoch": 95.63052208835342, "grad_norm": 11.915138244628906, "learning_rate": 2.1451612903225804e-06, "loss": 0.2576, "step": 14835 }, { "epoch": 95.63694779116466, "grad_norm": 11.628560066223145, "learning_rate": 2.141935483870968e-06, "loss": 0.5178, "step": 14836 }, { "epoch": 95.6433734939759, "grad_norm": 11.42341423034668, "learning_rate": 2.138709677419355e-06, "loss": 0.2602, "step": 14837 }, { "epoch": 95.64979919678714, "grad_norm": 6.2225422859191895, "learning_rate": 2.1354838709677418e-06, "loss": 0.3076, "step": 14838 }, { "epoch": 95.6562248995984, "grad_norm": 15.396835327148438, "learning_rate": 2.132258064516129e-06, "loss": 0.2776, "step": 14839 }, { "epoch": 95.66265060240964, "grad_norm": 11.804835319519043, "learning_rate": 2.129032258064516e-06, "loss": 0.356, "step": 14840 }, { "epoch": 95.66907630522088, "grad_norm": 9.225946426391602, "learning_rate": 2.125806451612903e-06, "loss": 0.3758, "step": 14841 }, { "epoch": 95.67550200803213, "grad_norm": 6.421280860900879, "learning_rate": 2.1225806451612905e-06, "loss": 0.1279, "step": 14842 }, { "epoch": 95.68192771084337, "grad_norm": 12.268692016601562, "learning_rate": 2.1193548387096775e-06, "loss": 0.4238, "step": 14843 }, { "epoch": 95.68835341365462, "grad_norm": 7.725805759429932, "learning_rate": 2.1161290322580645e-06, "loss": 0.1631, "step": 14844 }, { "epoch": 95.69477911646587, "grad_norm": 17.03683090209961, "learning_rate": 2.112903225806452e-06, "loss": 0.3116, "step": 14845 }, { "epoch": 95.70120481927711, "grad_norm": 7.099695205688477, "learning_rate": 2.109677419354839e-06, "loss": 0.3263, "step": 14846 }, { "epoch": 95.70763052208835, "grad_norm": 6.565580368041992, "learning_rate": 2.106451612903226e-06, "loss": 0.4115, "step": 14847 }, { "epoch": 95.7140562248996, "grad_norm": 11.5553560256958, "learning_rate": 2.1032258064516132e-06, "loss": 0.404, "step": 14848 }, { "epoch": 95.72048192771085, "grad_norm": 11.290911674499512, "learning_rate": 2.1000000000000002e-06, "loss": 0.3155, "step": 14849 }, { "epoch": 95.72690763052209, "grad_norm": 7.990855693817139, "learning_rate": 2.096774193548387e-06, "loss": 0.4122, "step": 14850 }, { "epoch": 95.73333333333333, "grad_norm": 5.546926021575928, "learning_rate": 2.0935483870967746e-06, "loss": 0.3012, "step": 14851 }, { "epoch": 95.73975903614458, "grad_norm": 18.03969383239746, "learning_rate": 2.090322580645161e-06, "loss": 0.3128, "step": 14852 }, { "epoch": 95.74618473895582, "grad_norm": 15.026175498962402, "learning_rate": 2.0870967741935486e-06, "loss": 0.3382, "step": 14853 }, { "epoch": 95.75261044176708, "grad_norm": 11.961773872375488, "learning_rate": 2.0838709677419355e-06, "loss": 0.4415, "step": 14854 }, { "epoch": 95.75903614457832, "grad_norm": 11.539888381958008, "learning_rate": 2.0806451612903225e-06, "loss": 0.6784, "step": 14855 }, { "epoch": 95.76546184738956, "grad_norm": 12.570387840270996, "learning_rate": 2.0774193548387095e-06, "loss": 0.3121, "step": 14856 }, { "epoch": 95.7718875502008, "grad_norm": 7.9238457679748535, "learning_rate": 2.074193548387097e-06, "loss": 0.1708, "step": 14857 }, { "epoch": 95.77831325301204, "grad_norm": 24.163114547729492, "learning_rate": 2.070967741935484e-06, "loss": 0.4849, "step": 14858 }, { "epoch": 95.78473895582329, "grad_norm": 10.565840721130371, "learning_rate": 2.067741935483871e-06, "loss": 0.3838, "step": 14859 }, { "epoch": 95.79116465863454, "grad_norm": 9.341994285583496, "learning_rate": 2.0645161290322582e-06, "loss": 0.5593, "step": 14860 }, { "epoch": 95.79759036144578, "grad_norm": 8.716296195983887, "learning_rate": 2.0612903225806452e-06, "loss": 0.2172, "step": 14861 }, { "epoch": 95.80401606425703, "grad_norm": 7.232272148132324, "learning_rate": 2.058064516129032e-06, "loss": 0.368, "step": 14862 }, { "epoch": 95.81044176706827, "grad_norm": 11.093317985534668, "learning_rate": 2.0548387096774196e-06, "loss": 0.5816, "step": 14863 }, { "epoch": 95.81686746987951, "grad_norm": 9.30703067779541, "learning_rate": 2.0516129032258066e-06, "loss": 0.2884, "step": 14864 }, { "epoch": 95.82329317269077, "grad_norm": 6.859528541564941, "learning_rate": 2.0483870967741936e-06, "loss": 0.3528, "step": 14865 }, { "epoch": 95.82971887550201, "grad_norm": 8.52696418762207, "learning_rate": 2.045161290322581e-06, "loss": 0.3825, "step": 14866 }, { "epoch": 95.83614457831325, "grad_norm": 5.0677571296691895, "learning_rate": 2.041935483870968e-06, "loss": 0.1237, "step": 14867 }, { "epoch": 95.8425702811245, "grad_norm": 7.433847427368164, "learning_rate": 2.038709677419355e-06, "loss": 0.3005, "step": 14868 }, { "epoch": 95.84899598393574, "grad_norm": 7.935870170593262, "learning_rate": 2.035483870967742e-06, "loss": 0.2383, "step": 14869 }, { "epoch": 95.855421686747, "grad_norm": 9.72923469543457, "learning_rate": 2.032258064516129e-06, "loss": 0.351, "step": 14870 }, { "epoch": 95.86184738955824, "grad_norm": 126.02049255371094, "learning_rate": 2.0290322580645163e-06, "loss": 0.3002, "step": 14871 }, { "epoch": 95.86827309236948, "grad_norm": 13.946846008300781, "learning_rate": 2.0258064516129032e-06, "loss": 0.3618, "step": 14872 }, { "epoch": 95.87469879518072, "grad_norm": 5.480673789978027, "learning_rate": 2.0225806451612902e-06, "loss": 0.0537, "step": 14873 }, { "epoch": 95.88112449799196, "grad_norm": 4.4695658683776855, "learning_rate": 2.0193548387096776e-06, "loss": 0.2676, "step": 14874 }, { "epoch": 95.88755020080322, "grad_norm": 5.982342720031738, "learning_rate": 2.0161290322580646e-06, "loss": 0.3908, "step": 14875 }, { "epoch": 95.89397590361446, "grad_norm": 15.103968620300293, "learning_rate": 2.0129032258064516e-06, "loss": 0.1165, "step": 14876 }, { "epoch": 95.9004016064257, "grad_norm": 9.046168327331543, "learning_rate": 2.009677419354839e-06, "loss": 0.241, "step": 14877 }, { "epoch": 95.90682730923695, "grad_norm": 5.6237287521362305, "learning_rate": 2.006451612903226e-06, "loss": 0.2218, "step": 14878 }, { "epoch": 95.91325301204819, "grad_norm": 11.49069881439209, "learning_rate": 2.003225806451613e-06, "loss": 0.2791, "step": 14879 }, { "epoch": 95.91967871485944, "grad_norm": 14.932774543762207, "learning_rate": 2.0000000000000003e-06, "loss": 0.2506, "step": 14880 }, { "epoch": 95.92610441767069, "grad_norm": 10.885221481323242, "learning_rate": 1.9967741935483873e-06, "loss": 0.4101, "step": 14881 }, { "epoch": 95.93253012048193, "grad_norm": 26.103662490844727, "learning_rate": 1.9935483870967743e-06, "loss": 0.8607, "step": 14882 }, { "epoch": 95.93895582329317, "grad_norm": 21.192869186401367, "learning_rate": 1.9903225806451617e-06, "loss": 0.4229, "step": 14883 }, { "epoch": 95.94538152610441, "grad_norm": 14.116283416748047, "learning_rate": 1.9870967741935487e-06, "loss": 0.422, "step": 14884 }, { "epoch": 95.95180722891567, "grad_norm": 11.69589614868164, "learning_rate": 1.9838709677419356e-06, "loss": 0.5316, "step": 14885 }, { "epoch": 95.95823293172691, "grad_norm": 4.769970893859863, "learning_rate": 1.980645161290323e-06, "loss": 0.3149, "step": 14886 }, { "epoch": 95.96465863453815, "grad_norm": 7.619920253753662, "learning_rate": 1.9774193548387096e-06, "loss": 0.1622, "step": 14887 }, { "epoch": 95.9710843373494, "grad_norm": 10.035475730895996, "learning_rate": 1.9741935483870966e-06, "loss": 0.2968, "step": 14888 }, { "epoch": 95.97751004016064, "grad_norm": 12.595285415649414, "learning_rate": 1.970967741935484e-06, "loss": 0.4286, "step": 14889 }, { "epoch": 95.9839357429719, "grad_norm": 13.332504272460938, "learning_rate": 1.967741935483871e-06, "loss": 0.2697, "step": 14890 }, { "epoch": 95.99036144578314, "grad_norm": 16.918275833129883, "learning_rate": 1.964516129032258e-06, "loss": 0.314, "step": 14891 }, { "epoch": 95.99678714859438, "grad_norm": 6.239960193634033, "learning_rate": 1.9612903225806453e-06, "loss": 0.1299, "step": 14892 }, { "epoch": 96.00642570281124, "grad_norm": 5.275424003601074, "learning_rate": 1.9580645161290323e-06, "loss": 0.1885, "step": 14893 }, { "epoch": 96.01285140562248, "grad_norm": 9.516412734985352, "learning_rate": 1.9548387096774193e-06, "loss": 0.3926, "step": 14894 }, { "epoch": 96.01927710843374, "grad_norm": 4.601276874542236, "learning_rate": 1.9516129032258067e-06, "loss": 0.1528, "step": 14895 }, { "epoch": 96.02570281124498, "grad_norm": 17.623262405395508, "learning_rate": 1.9483870967741937e-06, "loss": 0.3933, "step": 14896 }, { "epoch": 96.03212851405623, "grad_norm": 9.855977058410645, "learning_rate": 1.9451612903225806e-06, "loss": 0.3605, "step": 14897 }, { "epoch": 96.03855421686747, "grad_norm": 9.868617057800293, "learning_rate": 1.941935483870968e-06, "loss": 0.548, "step": 14898 }, { "epoch": 96.04497991967871, "grad_norm": 26.73898696899414, "learning_rate": 1.938709677419355e-06, "loss": 0.4998, "step": 14899 }, { "epoch": 96.05140562248997, "grad_norm": 6.411553859710693, "learning_rate": 1.935483870967742e-06, "loss": 0.2868, "step": 14900 }, { "epoch": 96.05783132530121, "grad_norm": 13.737231254577637, "learning_rate": 1.9322580645161294e-06, "loss": 0.4259, "step": 14901 }, { "epoch": 96.06425702811245, "grad_norm": 15.670063972473145, "learning_rate": 1.9290322580645164e-06, "loss": 0.1567, "step": 14902 }, { "epoch": 96.07068273092369, "grad_norm": 11.803284645080566, "learning_rate": 1.9258064516129034e-06, "loss": 0.4936, "step": 14903 }, { "epoch": 96.07710843373494, "grad_norm": 9.569406509399414, "learning_rate": 1.9225806451612908e-06, "loss": 0.3098, "step": 14904 }, { "epoch": 96.08353413654619, "grad_norm": 11.523085594177246, "learning_rate": 1.9193548387096773e-06, "loss": 0.5646, "step": 14905 }, { "epoch": 96.08995983935743, "grad_norm": 7.461269378662109, "learning_rate": 1.9161290322580643e-06, "loss": 0.2469, "step": 14906 }, { "epoch": 96.09638554216868, "grad_norm": 19.772293090820312, "learning_rate": 1.9129032258064517e-06, "loss": 0.2813, "step": 14907 }, { "epoch": 96.10281124497992, "grad_norm": 16.531503677368164, "learning_rate": 1.9096774193548387e-06, "loss": 0.2691, "step": 14908 }, { "epoch": 96.10923694779116, "grad_norm": 10.55507755279541, "learning_rate": 1.9064516129032259e-06, "loss": 0.2664, "step": 14909 }, { "epoch": 96.1156626506024, "grad_norm": 15.970447540283203, "learning_rate": 1.9032258064516128e-06, "loss": 0.5372, "step": 14910 }, { "epoch": 96.12208835341366, "grad_norm": 15.77552604675293, "learning_rate": 1.9e-06, "loss": 0.486, "step": 14911 }, { "epoch": 96.1285140562249, "grad_norm": 13.030672073364258, "learning_rate": 1.8967741935483872e-06, "loss": 0.5327, "step": 14912 }, { "epoch": 96.13493975903614, "grad_norm": 9.014286994934082, "learning_rate": 1.8935483870967742e-06, "loss": 0.3465, "step": 14913 }, { "epoch": 96.14136546184739, "grad_norm": 13.147271156311035, "learning_rate": 1.8903225806451614e-06, "loss": 0.4997, "step": 14914 }, { "epoch": 96.14779116465863, "grad_norm": 9.330957412719727, "learning_rate": 1.8870967741935486e-06, "loss": 0.3208, "step": 14915 }, { "epoch": 96.15421686746988, "grad_norm": 5.915245532989502, "learning_rate": 1.8838709677419355e-06, "loss": 0.2659, "step": 14916 }, { "epoch": 96.16064257028113, "grad_norm": 44.440635681152344, "learning_rate": 1.8806451612903227e-06, "loss": 0.4318, "step": 14917 }, { "epoch": 96.16706827309237, "grad_norm": 9.42365837097168, "learning_rate": 1.87741935483871e-06, "loss": 0.3576, "step": 14918 }, { "epoch": 96.17349397590361, "grad_norm": 4.526139736175537, "learning_rate": 1.874193548387097e-06, "loss": 0.1392, "step": 14919 }, { "epoch": 96.17991967871485, "grad_norm": 16.67943000793457, "learning_rate": 1.870967741935484e-06, "loss": 0.3562, "step": 14920 }, { "epoch": 96.18634538152611, "grad_norm": 9.532125473022461, "learning_rate": 1.8677419354838713e-06, "loss": 0.0599, "step": 14921 }, { "epoch": 96.19277108433735, "grad_norm": 16.461528778076172, "learning_rate": 1.8645161290322583e-06, "loss": 0.4634, "step": 14922 }, { "epoch": 96.1991967871486, "grad_norm": 4.2867536544799805, "learning_rate": 1.861290322580645e-06, "loss": 0.1205, "step": 14923 }, { "epoch": 96.20562248995984, "grad_norm": 18.905895233154297, "learning_rate": 1.8580645161290322e-06, "loss": 0.5647, "step": 14924 }, { "epoch": 96.21204819277108, "grad_norm": 9.48271369934082, "learning_rate": 1.8548387096774194e-06, "loss": 0.3432, "step": 14925 }, { "epoch": 96.21847389558233, "grad_norm": 8.13586711883545, "learning_rate": 1.8516129032258064e-06, "loss": 0.3571, "step": 14926 }, { "epoch": 96.22489959839358, "grad_norm": 17.159440994262695, "learning_rate": 1.8483870967741936e-06, "loss": 0.4759, "step": 14927 }, { "epoch": 96.23132530120482, "grad_norm": 11.45767879486084, "learning_rate": 1.8451612903225808e-06, "loss": 0.3551, "step": 14928 }, { "epoch": 96.23775100401606, "grad_norm": 13.32773208618164, "learning_rate": 1.8419354838709677e-06, "loss": 0.4732, "step": 14929 }, { "epoch": 96.2441767068273, "grad_norm": 13.631078720092773, "learning_rate": 1.838709677419355e-06, "loss": 0.4744, "step": 14930 }, { "epoch": 96.25060240963856, "grad_norm": 31.70009994506836, "learning_rate": 1.8354838709677421e-06, "loss": 0.5297, "step": 14931 }, { "epoch": 96.2570281124498, "grad_norm": 21.760726928710938, "learning_rate": 1.832258064516129e-06, "loss": 0.183, "step": 14932 }, { "epoch": 96.26345381526104, "grad_norm": 6.606301307678223, "learning_rate": 1.8290322580645163e-06, "loss": 0.4747, "step": 14933 }, { "epoch": 96.26987951807229, "grad_norm": 10.331121444702148, "learning_rate": 1.8258064516129035e-06, "loss": 0.3493, "step": 14934 }, { "epoch": 96.27630522088353, "grad_norm": 7.095942497253418, "learning_rate": 1.8225806451612904e-06, "loss": 0.1472, "step": 14935 }, { "epoch": 96.28273092369479, "grad_norm": 8.24691390991211, "learning_rate": 1.8193548387096776e-06, "loss": 0.3646, "step": 14936 }, { "epoch": 96.28915662650603, "grad_norm": 12.383768081665039, "learning_rate": 1.8161290322580648e-06, "loss": 0.326, "step": 14937 }, { "epoch": 96.29558232931727, "grad_norm": 11.908964157104492, "learning_rate": 1.8129032258064518e-06, "loss": 0.2115, "step": 14938 }, { "epoch": 96.30200803212851, "grad_norm": 9.055534362792969, "learning_rate": 1.809677419354839e-06, "loss": 0.4197, "step": 14939 }, { "epoch": 96.30843373493975, "grad_norm": 9.382303237915039, "learning_rate": 1.806451612903226e-06, "loss": 0.1663, "step": 14940 }, { "epoch": 96.31485943775101, "grad_norm": 21.950183868408203, "learning_rate": 1.8032258064516127e-06, "loss": 0.7557, "step": 14941 }, { "epoch": 96.32128514056225, "grad_norm": 16.332094192504883, "learning_rate": 1.8e-06, "loss": 0.6975, "step": 14942 }, { "epoch": 96.3277108433735, "grad_norm": 18.88175392150879, "learning_rate": 1.7967741935483871e-06, "loss": 0.2571, "step": 14943 }, { "epoch": 96.33413654618474, "grad_norm": 12.945661544799805, "learning_rate": 1.793548387096774e-06, "loss": 0.2874, "step": 14944 }, { "epoch": 96.34056224899598, "grad_norm": 9.819518089294434, "learning_rate": 1.7903225806451613e-06, "loss": 0.2911, "step": 14945 }, { "epoch": 96.34698795180722, "grad_norm": 9.97710132598877, "learning_rate": 1.7870967741935485e-06, "loss": 0.294, "step": 14946 }, { "epoch": 96.35341365461848, "grad_norm": 13.770003318786621, "learning_rate": 1.7838709677419354e-06, "loss": 0.4875, "step": 14947 }, { "epoch": 96.35983935742972, "grad_norm": 8.717634201049805, "learning_rate": 1.7806451612903226e-06, "loss": 0.5157, "step": 14948 }, { "epoch": 96.36626506024096, "grad_norm": 24.918270111083984, "learning_rate": 1.7774193548387098e-06, "loss": 0.6638, "step": 14949 }, { "epoch": 96.3726907630522, "grad_norm": 5.963072299957275, "learning_rate": 1.7741935483870968e-06, "loss": 0.3231, "step": 14950 }, { "epoch": 96.37911646586345, "grad_norm": 4.501736640930176, "learning_rate": 1.770967741935484e-06, "loss": 0.129, "step": 14951 }, { "epoch": 96.3855421686747, "grad_norm": 14.501899719238281, "learning_rate": 1.7677419354838712e-06, "loss": 0.3089, "step": 14952 }, { "epoch": 96.39196787148595, "grad_norm": 60.49264144897461, "learning_rate": 1.7645161290322582e-06, "loss": 0.3281, "step": 14953 }, { "epoch": 96.39839357429719, "grad_norm": 17.785554885864258, "learning_rate": 1.7612903225806453e-06, "loss": 0.4211, "step": 14954 }, { "epoch": 96.40481927710843, "grad_norm": 12.394969940185547, "learning_rate": 1.7580645161290325e-06, "loss": 0.426, "step": 14955 }, { "epoch": 96.41124497991967, "grad_norm": 8.594152450561523, "learning_rate": 1.7548387096774195e-06, "loss": 0.2892, "step": 14956 }, { "epoch": 96.41767068273093, "grad_norm": 23.217185974121094, "learning_rate": 1.7516129032258067e-06, "loss": 0.653, "step": 14957 }, { "epoch": 96.42409638554217, "grad_norm": 7.202144145965576, "learning_rate": 1.7483870967741935e-06, "loss": 0.253, "step": 14958 }, { "epoch": 96.43052208835341, "grad_norm": 6.962018013000488, "learning_rate": 1.7451612903225807e-06, "loss": 0.2641, "step": 14959 }, { "epoch": 96.43694779116466, "grad_norm": 11.253376007080078, "learning_rate": 1.7419354838709676e-06, "loss": 0.256, "step": 14960 }, { "epoch": 96.4433734939759, "grad_norm": 10.436906814575195, "learning_rate": 1.7387096774193548e-06, "loss": 0.4637, "step": 14961 }, { "epoch": 96.44979919678715, "grad_norm": 7.603058338165283, "learning_rate": 1.735483870967742e-06, "loss": 0.4318, "step": 14962 }, { "epoch": 96.4562248995984, "grad_norm": 9.219963073730469, "learning_rate": 1.732258064516129e-06, "loss": 0.2546, "step": 14963 }, { "epoch": 96.46265060240964, "grad_norm": 22.787900924682617, "learning_rate": 1.7290322580645162e-06, "loss": 0.3106, "step": 14964 }, { "epoch": 96.46907630522088, "grad_norm": 10.320220947265625, "learning_rate": 1.7258064516129034e-06, "loss": 0.3474, "step": 14965 }, { "epoch": 96.47550200803212, "grad_norm": 10.80875015258789, "learning_rate": 1.7225806451612903e-06, "loss": 0.2686, "step": 14966 }, { "epoch": 96.48192771084338, "grad_norm": 13.940255165100098, "learning_rate": 1.7193548387096775e-06, "loss": 0.3566, "step": 14967 }, { "epoch": 96.48835341365462, "grad_norm": 10.818912506103516, "learning_rate": 1.7161290322580647e-06, "loss": 0.465, "step": 14968 }, { "epoch": 96.49477911646586, "grad_norm": 9.591642379760742, "learning_rate": 1.7129032258064517e-06, "loss": 0.4988, "step": 14969 }, { "epoch": 96.5012048192771, "grad_norm": 23.087112426757812, "learning_rate": 1.7096774193548389e-06, "loss": 0.4231, "step": 14970 }, { "epoch": 96.50763052208835, "grad_norm": 11.104869842529297, "learning_rate": 1.706451612903226e-06, "loss": 0.2972, "step": 14971 }, { "epoch": 96.5140562248996, "grad_norm": 10.821382522583008, "learning_rate": 1.703225806451613e-06, "loss": 0.4472, "step": 14972 }, { "epoch": 96.52048192771085, "grad_norm": 14.037135124206543, "learning_rate": 1.7000000000000002e-06, "loss": 0.664, "step": 14973 }, { "epoch": 96.52690763052209, "grad_norm": 22.883270263671875, "learning_rate": 1.6967741935483872e-06, "loss": 0.3411, "step": 14974 }, { "epoch": 96.53333333333333, "grad_norm": 12.979292869567871, "learning_rate": 1.6935483870967744e-06, "loss": 0.4546, "step": 14975 }, { "epoch": 96.53975903614457, "grad_norm": 9.979979515075684, "learning_rate": 1.6903225806451612e-06, "loss": 0.4111, "step": 14976 }, { "epoch": 96.54618473895583, "grad_norm": 6.712916851043701, "learning_rate": 1.6870967741935484e-06, "loss": 0.2431, "step": 14977 }, { "epoch": 96.55261044176707, "grad_norm": 12.054630279541016, "learning_rate": 1.6838709677419353e-06, "loss": 0.3846, "step": 14978 }, { "epoch": 96.55903614457831, "grad_norm": 10.722052574157715, "learning_rate": 1.6806451612903225e-06, "loss": 0.6445, "step": 14979 }, { "epoch": 96.56546184738956, "grad_norm": 14.338276863098145, "learning_rate": 1.6774193548387097e-06, "loss": 0.3826, "step": 14980 }, { "epoch": 96.5718875502008, "grad_norm": 7.461144924163818, "learning_rate": 1.6741935483870967e-06, "loss": 0.3406, "step": 14981 }, { "epoch": 96.57831325301204, "grad_norm": 7.7982964515686035, "learning_rate": 1.6709677419354839e-06, "loss": 0.2307, "step": 14982 }, { "epoch": 96.5847389558233, "grad_norm": 11.979225158691406, "learning_rate": 1.667741935483871e-06, "loss": 0.4576, "step": 14983 }, { "epoch": 96.59116465863454, "grad_norm": 12.647335052490234, "learning_rate": 1.664516129032258e-06, "loss": 0.3063, "step": 14984 }, { "epoch": 96.59759036144578, "grad_norm": 15.088517189025879, "learning_rate": 1.6612903225806452e-06, "loss": 0.5689, "step": 14985 }, { "epoch": 96.60401606425702, "grad_norm": 7.210439682006836, "learning_rate": 1.6580645161290324e-06, "loss": 0.2314, "step": 14986 }, { "epoch": 96.61044176706827, "grad_norm": 12.674320220947266, "learning_rate": 1.6548387096774194e-06, "loss": 0.3733, "step": 14987 }, { "epoch": 96.61686746987952, "grad_norm": 5.3586530685424805, "learning_rate": 1.6516129032258066e-06, "loss": 0.0879, "step": 14988 }, { "epoch": 96.62329317269077, "grad_norm": 5.227388858795166, "learning_rate": 1.6483870967741938e-06, "loss": 0.1796, "step": 14989 }, { "epoch": 96.62971887550201, "grad_norm": 5.774913311004639, "learning_rate": 1.6451612903225808e-06, "loss": 0.2067, "step": 14990 }, { "epoch": 96.63614457831325, "grad_norm": 15.919702529907227, "learning_rate": 1.641935483870968e-06, "loss": 0.6731, "step": 14991 }, { "epoch": 96.64257028112449, "grad_norm": 14.357595443725586, "learning_rate": 1.6387096774193551e-06, "loss": 0.1967, "step": 14992 }, { "epoch": 96.64899598393575, "grad_norm": 9.516806602478027, "learning_rate": 1.6354838709677421e-06, "loss": 0.3947, "step": 14993 }, { "epoch": 96.65542168674699, "grad_norm": 10.765397071838379, "learning_rate": 1.6322580645161289e-06, "loss": 0.272, "step": 14994 }, { "epoch": 96.66184738955823, "grad_norm": 21.786584854125977, "learning_rate": 1.629032258064516e-06, "loss": 0.2525, "step": 14995 }, { "epoch": 96.66827309236947, "grad_norm": 5.802285671234131, "learning_rate": 1.6258064516129033e-06, "loss": 0.1105, "step": 14996 }, { "epoch": 96.67469879518072, "grad_norm": 11.619668960571289, "learning_rate": 1.6225806451612902e-06, "loss": 0.2704, "step": 14997 }, { "epoch": 96.68112449799197, "grad_norm": 5.939767360687256, "learning_rate": 1.6193548387096774e-06, "loss": 0.2462, "step": 14998 }, { "epoch": 96.68755020080322, "grad_norm": 10.353150367736816, "learning_rate": 1.6161290322580646e-06, "loss": 0.4627, "step": 14999 }, { "epoch": 96.69397590361446, "grad_norm": 6.035279273986816, "learning_rate": 1.6129032258064516e-06, "loss": 0.1564, "step": 15000 }, { "epoch": 96.7004016064257, "grad_norm": 8.526254653930664, "learning_rate": 1.6096774193548388e-06, "loss": 0.6383, "step": 15001 }, { "epoch": 96.70682730923694, "grad_norm": 6.847568988800049, "learning_rate": 1.606451612903226e-06, "loss": 0.3992, "step": 15002 }, { "epoch": 96.7132530120482, "grad_norm": 4.490616321563721, "learning_rate": 1.603225806451613e-06, "loss": 0.2751, "step": 15003 }, { "epoch": 96.71967871485944, "grad_norm": 8.904777526855469, "learning_rate": 1.6000000000000001e-06, "loss": 0.2463, "step": 15004 }, { "epoch": 96.72610441767068, "grad_norm": 3.5602688789367676, "learning_rate": 1.5967741935483873e-06, "loss": 0.156, "step": 15005 }, { "epoch": 96.73253012048193, "grad_norm": 12.774369239807129, "learning_rate": 1.5935483870967743e-06, "loss": 0.5111, "step": 15006 }, { "epoch": 96.73895582329317, "grad_norm": 14.432534217834473, "learning_rate": 1.5903225806451615e-06, "loss": 0.3148, "step": 15007 }, { "epoch": 96.74538152610442, "grad_norm": 3.8448328971862793, "learning_rate": 1.5870967741935485e-06, "loss": 0.1591, "step": 15008 }, { "epoch": 96.75180722891567, "grad_norm": 10.21217155456543, "learning_rate": 1.5838709677419357e-06, "loss": 0.4319, "step": 15009 }, { "epoch": 96.75823293172691, "grad_norm": 6.517330646514893, "learning_rate": 1.5806451612903229e-06, "loss": 0.3235, "step": 15010 }, { "epoch": 96.76465863453815, "grad_norm": 12.089818000793457, "learning_rate": 1.5774193548387098e-06, "loss": 0.3561, "step": 15011 }, { "epoch": 96.7710843373494, "grad_norm": 11.405375480651855, "learning_rate": 1.5741935483870966e-06, "loss": 0.3633, "step": 15012 }, { "epoch": 96.77751004016065, "grad_norm": 19.237838745117188, "learning_rate": 1.5709677419354838e-06, "loss": 0.3966, "step": 15013 }, { "epoch": 96.78393574297189, "grad_norm": 14.729240417480469, "learning_rate": 1.567741935483871e-06, "loss": 0.5952, "step": 15014 }, { "epoch": 96.79036144578313, "grad_norm": 4.09445333480835, "learning_rate": 1.564516129032258e-06, "loss": 0.1056, "step": 15015 }, { "epoch": 96.79678714859438, "grad_norm": 5.169845104217529, "learning_rate": 1.5612903225806451e-06, "loss": 0.1374, "step": 15016 }, { "epoch": 96.80321285140562, "grad_norm": 13.194511413574219, "learning_rate": 1.5580645161290323e-06, "loss": 0.5466, "step": 15017 }, { "epoch": 96.80963855421686, "grad_norm": 10.211921691894531, "learning_rate": 1.5548387096774193e-06, "loss": 0.4701, "step": 15018 }, { "epoch": 96.81606425702812, "grad_norm": 8.965376853942871, "learning_rate": 1.5516129032258065e-06, "loss": 0.3155, "step": 15019 }, { "epoch": 96.82248995983936, "grad_norm": 9.942517280578613, "learning_rate": 1.5483870967741937e-06, "loss": 0.362, "step": 15020 }, { "epoch": 96.8289156626506, "grad_norm": 14.708487510681152, "learning_rate": 1.5451612903225807e-06, "loss": 0.3004, "step": 15021 }, { "epoch": 96.83534136546184, "grad_norm": 7.684485912322998, "learning_rate": 1.5419354838709678e-06, "loss": 0.5095, "step": 15022 }, { "epoch": 96.84176706827309, "grad_norm": 13.98587417602539, "learning_rate": 1.538709677419355e-06, "loss": 0.4704, "step": 15023 }, { "epoch": 96.84819277108434, "grad_norm": 10.90545654296875, "learning_rate": 1.535483870967742e-06, "loss": 0.1501, "step": 15024 }, { "epoch": 96.85461847389558, "grad_norm": 3.0108673572540283, "learning_rate": 1.532258064516129e-06, "loss": 0.1305, "step": 15025 }, { "epoch": 96.86104417670683, "grad_norm": 25.687870025634766, "learning_rate": 1.5290322580645162e-06, "loss": 0.485, "step": 15026 }, { "epoch": 96.86746987951807, "grad_norm": 12.210180282592773, "learning_rate": 1.5258064516129032e-06, "loss": 0.1935, "step": 15027 }, { "epoch": 96.87389558232931, "grad_norm": 8.529200553894043, "learning_rate": 1.5225806451612903e-06, "loss": 0.3719, "step": 15028 }, { "epoch": 96.88032128514057, "grad_norm": 5.7114033699035645, "learning_rate": 1.5193548387096775e-06, "loss": 0.3117, "step": 15029 }, { "epoch": 96.88674698795181, "grad_norm": 9.120328903198242, "learning_rate": 1.5161290322580645e-06, "loss": 0.2963, "step": 15030 }, { "epoch": 96.89317269076305, "grad_norm": 11.351019859313965, "learning_rate": 1.5129032258064517e-06, "loss": 0.3537, "step": 15031 }, { "epoch": 96.8995983935743, "grad_norm": 12.867181777954102, "learning_rate": 1.5096774193548389e-06, "loss": 0.3804, "step": 15032 }, { "epoch": 96.90602409638554, "grad_norm": 11.55168628692627, "learning_rate": 1.5064516129032259e-06, "loss": 0.5307, "step": 15033 }, { "epoch": 96.91244979919679, "grad_norm": 7.042268753051758, "learning_rate": 1.5032258064516128e-06, "loss": 0.2356, "step": 15034 }, { "epoch": 96.91887550200804, "grad_norm": 11.440801620483398, "learning_rate": 1.5e-06, "loss": 0.558, "step": 15035 }, { "epoch": 96.92530120481928, "grad_norm": 6.938678741455078, "learning_rate": 1.4967741935483872e-06, "loss": 0.3411, "step": 15036 }, { "epoch": 96.93172690763052, "grad_norm": 9.277092933654785, "learning_rate": 1.4935483870967742e-06, "loss": 0.5137, "step": 15037 }, { "epoch": 96.93815261044176, "grad_norm": 5.9248127937316895, "learning_rate": 1.4903225806451614e-06, "loss": 0.1151, "step": 15038 }, { "epoch": 96.94457831325302, "grad_norm": 33.56766891479492, "learning_rate": 1.4870967741935486e-06, "loss": 0.5374, "step": 15039 }, { "epoch": 96.95100401606426, "grad_norm": 10.814526557922363, "learning_rate": 1.4838709677419356e-06, "loss": 0.1647, "step": 15040 }, { "epoch": 96.9574297188755, "grad_norm": 8.079103469848633, "learning_rate": 1.4806451612903227e-06, "loss": 0.269, "step": 15041 }, { "epoch": 96.96385542168674, "grad_norm": 7.611725330352783, "learning_rate": 1.4774193548387097e-06, "loss": 0.1936, "step": 15042 }, { "epoch": 96.97028112449799, "grad_norm": 10.845429420471191, "learning_rate": 1.4741935483870967e-06, "loss": 0.4728, "step": 15043 }, { "epoch": 96.97670682730924, "grad_norm": 14.283992767333984, "learning_rate": 1.4709677419354839e-06, "loss": 0.3466, "step": 15044 }, { "epoch": 96.98313253012049, "grad_norm": 10.132192611694336, "learning_rate": 1.467741935483871e-06, "loss": 0.2187, "step": 15045 }, { "epoch": 96.98955823293173, "grad_norm": 8.500137329101562, "learning_rate": 1.464516129032258e-06, "loss": 0.2999, "step": 15046 }, { "epoch": 96.99598393574297, "grad_norm": 59.35542678833008, "learning_rate": 1.4612903225806452e-06, "loss": 0.6985, "step": 15047 }, { "epoch": 97.00562248995983, "grad_norm": 6.8410820960998535, "learning_rate": 1.4580645161290324e-06, "loss": 0.4119, "step": 15048 }, { "epoch": 97.01204819277109, "grad_norm": 8.62179183959961, "learning_rate": 1.4548387096774194e-06, "loss": 0.2686, "step": 15049 }, { "epoch": 97.01847389558233, "grad_norm": 11.629165649414062, "learning_rate": 1.4516129032258066e-06, "loss": 0.2162, "step": 15050 }, { "epoch": 97.02489959839357, "grad_norm": 12.235464096069336, "learning_rate": 1.4483870967741938e-06, "loss": 0.4674, "step": 15051 }, { "epoch": 97.03132530120482, "grad_norm": 16.1197566986084, "learning_rate": 1.4451612903225806e-06, "loss": 0.2485, "step": 15052 }, { "epoch": 97.03775100401606, "grad_norm": 21.162641525268555, "learning_rate": 1.4419354838709677e-06, "loss": 0.5241, "step": 15053 }, { "epoch": 97.04417670682732, "grad_norm": 7.0500664710998535, "learning_rate": 1.438709677419355e-06, "loss": 0.2526, "step": 15054 }, { "epoch": 97.05060240963856, "grad_norm": 7.686862945556641, "learning_rate": 1.435483870967742e-06, "loss": 0.1994, "step": 15055 }, { "epoch": 97.0570281124498, "grad_norm": 8.888333320617676, "learning_rate": 1.432258064516129e-06, "loss": 0.5865, "step": 15056 }, { "epoch": 97.06345381526104, "grad_norm": 13.620329856872559, "learning_rate": 1.4290322580645163e-06, "loss": 0.4489, "step": 15057 }, { "epoch": 97.06987951807228, "grad_norm": 12.359652519226074, "learning_rate": 1.4258064516129033e-06, "loss": 0.465, "step": 15058 }, { "epoch": 97.07630522088354, "grad_norm": 5.604918956756592, "learning_rate": 1.4225806451612905e-06, "loss": 0.1758, "step": 15059 }, { "epoch": 97.08273092369478, "grad_norm": 8.304421424865723, "learning_rate": 1.4193548387096776e-06, "loss": 0.3937, "step": 15060 }, { "epoch": 97.08915662650602, "grad_norm": 12.160857200622559, "learning_rate": 1.4161290322580644e-06, "loss": 0.2546, "step": 15061 }, { "epoch": 97.09558232931727, "grad_norm": 51.41114807128906, "learning_rate": 1.4129032258064516e-06, "loss": 0.1772, "step": 15062 }, { "epoch": 97.10200803212851, "grad_norm": 11.36156177520752, "learning_rate": 1.4096774193548388e-06, "loss": 0.187, "step": 15063 }, { "epoch": 97.10843373493977, "grad_norm": 3.1286470890045166, "learning_rate": 1.4064516129032258e-06, "loss": 0.3149, "step": 15064 }, { "epoch": 97.11485943775101, "grad_norm": 6.964277744293213, "learning_rate": 1.403225806451613e-06, "loss": 0.4766, "step": 15065 }, { "epoch": 97.12128514056225, "grad_norm": 26.310829162597656, "learning_rate": 1.4000000000000001e-06, "loss": 0.4051, "step": 15066 }, { "epoch": 97.12771084337349, "grad_norm": 2.6700732707977295, "learning_rate": 1.3967741935483871e-06, "loss": 0.0805, "step": 15067 }, { "epoch": 97.13413654618473, "grad_norm": 12.265775680541992, "learning_rate": 1.3935483870967743e-06, "loss": 0.1595, "step": 15068 }, { "epoch": 97.14056224899599, "grad_norm": 2.3992486000061035, "learning_rate": 1.3903225806451613e-06, "loss": 0.3048, "step": 15069 }, { "epoch": 97.14698795180723, "grad_norm": 3.7208900451660156, "learning_rate": 1.3870967741935485e-06, "loss": 0.1639, "step": 15070 }, { "epoch": 97.15341365461848, "grad_norm": 38.659523010253906, "learning_rate": 1.3838709677419355e-06, "loss": 0.5307, "step": 15071 }, { "epoch": 97.15983935742972, "grad_norm": 13.15538215637207, "learning_rate": 1.3806451612903226e-06, "loss": 0.4247, "step": 15072 }, { "epoch": 97.16626506024096, "grad_norm": 4.177107334136963, "learning_rate": 1.3774193548387098e-06, "loss": 0.1274, "step": 15073 }, { "epoch": 97.1726907630522, "grad_norm": 64.75421905517578, "learning_rate": 1.3741935483870968e-06, "loss": 0.3828, "step": 15074 }, { "epoch": 97.17911646586346, "grad_norm": 8.608832359313965, "learning_rate": 1.370967741935484e-06, "loss": 0.3223, "step": 15075 }, { "epoch": 97.1855421686747, "grad_norm": 21.709529876708984, "learning_rate": 1.367741935483871e-06, "loss": 0.4603, "step": 15076 }, { "epoch": 97.19196787148594, "grad_norm": 5.156866550445557, "learning_rate": 1.3645161290322582e-06, "loss": 0.4532, "step": 15077 }, { "epoch": 97.19839357429719, "grad_norm": 22.616682052612305, "learning_rate": 1.3612903225806451e-06, "loss": 0.3932, "step": 15078 }, { "epoch": 97.20481927710843, "grad_norm": 41.04523849487305, "learning_rate": 1.3580645161290323e-06, "loss": 0.5138, "step": 15079 }, { "epoch": 97.21124497991968, "grad_norm": 4.693256378173828, "learning_rate": 1.3548387096774193e-06, "loss": 0.1563, "step": 15080 }, { "epoch": 97.21767068273093, "grad_norm": 16.007951736450195, "learning_rate": 1.3516129032258065e-06, "loss": 0.1898, "step": 15081 }, { "epoch": 97.22409638554217, "grad_norm": 17.432897567749023, "learning_rate": 1.3483870967741937e-06, "loss": 0.5109, "step": 15082 }, { "epoch": 97.23052208835341, "grad_norm": 11.918253898620605, "learning_rate": 1.3451612903225807e-06, "loss": 0.3241, "step": 15083 }, { "epoch": 97.23694779116465, "grad_norm": 21.64651107788086, "learning_rate": 1.3419354838709679e-06, "loss": 0.305, "step": 15084 }, { "epoch": 97.24337349397591, "grad_norm": 27.528533935546875, "learning_rate": 1.338709677419355e-06, "loss": 0.464, "step": 15085 }, { "epoch": 97.24979919678715, "grad_norm": 6.630945205688477, "learning_rate": 1.335483870967742e-06, "loss": 0.1773, "step": 15086 }, { "epoch": 97.2562248995984, "grad_norm": 5.991767883300781, "learning_rate": 1.332258064516129e-06, "loss": 0.4864, "step": 15087 }, { "epoch": 97.26265060240964, "grad_norm": 21.447786331176758, "learning_rate": 1.3290322580645162e-06, "loss": 0.2639, "step": 15088 }, { "epoch": 97.26907630522088, "grad_norm": 26.750709533691406, "learning_rate": 1.3258064516129032e-06, "loss": 0.5457, "step": 15089 }, { "epoch": 97.27550200803213, "grad_norm": 19.840341567993164, "learning_rate": 1.3225806451612904e-06, "loss": 0.2044, "step": 15090 }, { "epoch": 97.28192771084338, "grad_norm": 16.868907928466797, "learning_rate": 1.3193548387096775e-06, "loss": 0.2555, "step": 15091 }, { "epoch": 97.28835341365462, "grad_norm": 61.79794692993164, "learning_rate": 1.3161290322580645e-06, "loss": 0.4779, "step": 15092 }, { "epoch": 97.29477911646586, "grad_norm": 23.688194274902344, "learning_rate": 1.3129032258064517e-06, "loss": 0.2488, "step": 15093 }, { "epoch": 97.3012048192771, "grad_norm": 11.99560260772705, "learning_rate": 1.309677419354839e-06, "loss": 0.2166, "step": 15094 }, { "epoch": 97.30763052208836, "grad_norm": 24.367877960205078, "learning_rate": 1.3064516129032259e-06, "loss": 0.3612, "step": 15095 }, { "epoch": 97.3140562248996, "grad_norm": 16.680749893188477, "learning_rate": 1.3032258064516129e-06, "loss": 0.3168, "step": 15096 }, { "epoch": 97.32048192771084, "grad_norm": 16.597936630249023, "learning_rate": 1.3e-06, "loss": 0.226, "step": 15097 }, { "epoch": 97.32690763052209, "grad_norm": 24.47335433959961, "learning_rate": 1.296774193548387e-06, "loss": 0.3614, "step": 15098 }, { "epoch": 97.33333333333333, "grad_norm": 44.79481887817383, "learning_rate": 1.2935483870967742e-06, "loss": 0.4592, "step": 15099 }, { "epoch": 97.33975903614459, "grad_norm": 27.793420791625977, "learning_rate": 1.2903225806451614e-06, "loss": 0.3556, "step": 15100 }, { "epoch": 97.34618473895583, "grad_norm": 49.10464859008789, "learning_rate": 1.2870967741935484e-06, "loss": 0.1787, "step": 15101 }, { "epoch": 97.35261044176707, "grad_norm": 4.849086761474609, "learning_rate": 1.2838709677419356e-06, "loss": 0.517, "step": 15102 }, { "epoch": 97.35903614457831, "grad_norm": 3.9062561988830566, "learning_rate": 1.2806451612903228e-06, "loss": 0.2779, "step": 15103 }, { "epoch": 97.36546184738955, "grad_norm": 13.278464317321777, "learning_rate": 1.2774193548387097e-06, "loss": 0.4041, "step": 15104 }, { "epoch": 97.3718875502008, "grad_norm": 18.68195915222168, "learning_rate": 1.2741935483870967e-06, "loss": 0.5474, "step": 15105 }, { "epoch": 97.37831325301205, "grad_norm": 12.33204460144043, "learning_rate": 1.270967741935484e-06, "loss": 0.2672, "step": 15106 }, { "epoch": 97.3847389558233, "grad_norm": 15.926901817321777, "learning_rate": 1.267741935483871e-06, "loss": 0.3919, "step": 15107 }, { "epoch": 97.39116465863454, "grad_norm": 5.623376846313477, "learning_rate": 1.264516129032258e-06, "loss": 0.2105, "step": 15108 }, { "epoch": 97.39759036144578, "grad_norm": 25.48271369934082, "learning_rate": 1.2612903225806453e-06, "loss": 0.3093, "step": 15109 }, { "epoch": 97.40401606425702, "grad_norm": 22.6960391998291, "learning_rate": 1.2580645161290322e-06, "loss": 0.4374, "step": 15110 }, { "epoch": 97.41044176706828, "grad_norm": 9.826007843017578, "learning_rate": 1.2548387096774194e-06, "loss": 0.1621, "step": 15111 }, { "epoch": 97.41686746987952, "grad_norm": 7.114686012268066, "learning_rate": 1.2516129032258066e-06, "loss": 0.3011, "step": 15112 }, { "epoch": 97.42329317269076, "grad_norm": 27.07101821899414, "learning_rate": 1.2483870967741936e-06, "loss": 0.2424, "step": 15113 }, { "epoch": 97.429718875502, "grad_norm": 15.466545104980469, "learning_rate": 1.2451612903225806e-06, "loss": 0.5118, "step": 15114 }, { "epoch": 97.43614457831325, "grad_norm": 4.861606597900391, "learning_rate": 1.2419354838709678e-06, "loss": 0.1007, "step": 15115 }, { "epoch": 97.4425702811245, "grad_norm": 5.431557655334473, "learning_rate": 1.238709677419355e-06, "loss": 0.1044, "step": 15116 }, { "epoch": 97.44899598393575, "grad_norm": 5.098292827606201, "learning_rate": 1.235483870967742e-06, "loss": 0.2982, "step": 15117 }, { "epoch": 97.45542168674699, "grad_norm": 9.06811237335205, "learning_rate": 1.2322580645161291e-06, "loss": 0.1962, "step": 15118 }, { "epoch": 97.46184738955823, "grad_norm": 34.931732177734375, "learning_rate": 1.2290322580645163e-06, "loss": 0.323, "step": 15119 }, { "epoch": 97.46827309236947, "grad_norm": 5.641422271728516, "learning_rate": 1.2258064516129033e-06, "loss": 0.4301, "step": 15120 }, { "epoch": 97.47469879518073, "grad_norm": 18.1777400970459, "learning_rate": 1.2225806451612905e-06, "loss": 0.3843, "step": 15121 }, { "epoch": 97.48112449799197, "grad_norm": 28.354942321777344, "learning_rate": 1.2193548387096777e-06, "loss": 0.4865, "step": 15122 }, { "epoch": 97.48755020080321, "grad_norm": 10.817380905151367, "learning_rate": 1.2161290322580644e-06, "loss": 0.3629, "step": 15123 }, { "epoch": 97.49397590361446, "grad_norm": 15.782379150390625, "learning_rate": 1.2129032258064516e-06, "loss": 0.129, "step": 15124 }, { "epoch": 97.5004016064257, "grad_norm": 18.369218826293945, "learning_rate": 1.2096774193548388e-06, "loss": 0.2082, "step": 15125 }, { "epoch": 97.50682730923695, "grad_norm": 5.288506031036377, "learning_rate": 1.2064516129032258e-06, "loss": 0.316, "step": 15126 }, { "epoch": 97.5132530120482, "grad_norm": 5.827053546905518, "learning_rate": 1.203225806451613e-06, "loss": 0.3844, "step": 15127 }, { "epoch": 97.51967871485944, "grad_norm": 3.7579901218414307, "learning_rate": 1.2000000000000002e-06, "loss": 0.255, "step": 15128 }, { "epoch": 97.52610441767068, "grad_norm": 20.040653228759766, "learning_rate": 1.1967741935483871e-06, "loss": 0.3604, "step": 15129 }, { "epoch": 97.53253012048192, "grad_norm": 16.19903564453125, "learning_rate": 1.1935483870967743e-06, "loss": 0.461, "step": 15130 }, { "epoch": 97.53895582329318, "grad_norm": 4.548749923706055, "learning_rate": 1.1903225806451615e-06, "loss": 0.169, "step": 15131 }, { "epoch": 97.54538152610442, "grad_norm": 10.320395469665527, "learning_rate": 1.1870967741935483e-06, "loss": 0.3208, "step": 15132 }, { "epoch": 97.55180722891566, "grad_norm": 4.449441432952881, "learning_rate": 1.1838709677419355e-06, "loss": 0.2496, "step": 15133 }, { "epoch": 97.5582329317269, "grad_norm": 14.20630168914795, "learning_rate": 1.1806451612903227e-06, "loss": 0.3129, "step": 15134 }, { "epoch": 97.56465863453815, "grad_norm": 23.896753311157227, "learning_rate": 1.1774193548387096e-06, "loss": 0.3361, "step": 15135 }, { "epoch": 97.5710843373494, "grad_norm": 9.689318656921387, "learning_rate": 1.1741935483870968e-06, "loss": 0.2299, "step": 15136 }, { "epoch": 97.57751004016065, "grad_norm": 18.58542251586914, "learning_rate": 1.170967741935484e-06, "loss": 0.4024, "step": 15137 }, { "epoch": 97.58393574297189, "grad_norm": 16.21116828918457, "learning_rate": 1.167741935483871e-06, "loss": 0.281, "step": 15138 }, { "epoch": 97.59036144578313, "grad_norm": 24.7203426361084, "learning_rate": 1.1645161290322582e-06, "loss": 0.4106, "step": 15139 }, { "epoch": 97.59678714859437, "grad_norm": 2.9299399852752686, "learning_rate": 1.1612903225806454e-06, "loss": 0.4409, "step": 15140 }, { "epoch": 97.60321285140563, "grad_norm": 30.83014678955078, "learning_rate": 1.1580645161290323e-06, "loss": 0.3911, "step": 15141 }, { "epoch": 97.60963855421687, "grad_norm": 14.607195854187012, "learning_rate": 1.1548387096774193e-06, "loss": 0.3398, "step": 15142 }, { "epoch": 97.61606425702811, "grad_norm": 38.373504638671875, "learning_rate": 1.1516129032258065e-06, "loss": 0.5275, "step": 15143 }, { "epoch": 97.62248995983936, "grad_norm": 21.600812911987305, "learning_rate": 1.1483870967741935e-06, "loss": 0.6765, "step": 15144 }, { "epoch": 97.6289156626506, "grad_norm": 16.811933517456055, "learning_rate": 1.1451612903225807e-06, "loss": 0.3622, "step": 15145 }, { "epoch": 97.63534136546184, "grad_norm": 7.8754072189331055, "learning_rate": 1.1419354838709679e-06, "loss": 0.0793, "step": 15146 }, { "epoch": 97.6417670682731, "grad_norm": 31.727476119995117, "learning_rate": 1.1387096774193548e-06, "loss": 0.2895, "step": 15147 }, { "epoch": 97.64819277108434, "grad_norm": 47.35321807861328, "learning_rate": 1.135483870967742e-06, "loss": 0.309, "step": 15148 }, { "epoch": 97.65461847389558, "grad_norm": 21.813701629638672, "learning_rate": 1.132258064516129e-06, "loss": 0.2738, "step": 15149 }, { "epoch": 97.66104417670682, "grad_norm": 25.029199600219727, "learning_rate": 1.1290322580645162e-06, "loss": 0.595, "step": 15150 }, { "epoch": 97.66746987951807, "grad_norm": 5.872742176055908, "learning_rate": 1.1258064516129032e-06, "loss": 0.4708, "step": 15151 }, { "epoch": 97.67389558232932, "grad_norm": 21.220947265625, "learning_rate": 1.1225806451612904e-06, "loss": 0.6789, "step": 15152 }, { "epoch": 97.68032128514056, "grad_norm": 26.1301212310791, "learning_rate": 1.1193548387096776e-06, "loss": 0.2822, "step": 15153 }, { "epoch": 97.6867469879518, "grad_norm": 3.1953125, "learning_rate": 1.1161290322580645e-06, "loss": 0.1652, "step": 15154 }, { "epoch": 97.69317269076305, "grad_norm": 8.321951866149902, "learning_rate": 1.1129032258064517e-06, "loss": 0.3911, "step": 15155 }, { "epoch": 97.69959839357429, "grad_norm": 21.764738082885742, "learning_rate": 1.109677419354839e-06, "loss": 0.4503, "step": 15156 }, { "epoch": 97.70602409638555, "grad_norm": 18.861400604248047, "learning_rate": 1.1064516129032259e-06, "loss": 0.5009, "step": 15157 }, { "epoch": 97.71244979919679, "grad_norm": 20.495071411132812, "learning_rate": 1.1032258064516129e-06, "loss": 0.4485, "step": 15158 }, { "epoch": 97.71887550200803, "grad_norm": 4.603619575500488, "learning_rate": 1.1e-06, "loss": 0.1225, "step": 15159 }, { "epoch": 97.72530120481927, "grad_norm": 28.414098739624023, "learning_rate": 1.096774193548387e-06, "loss": 0.3088, "step": 15160 }, { "epoch": 97.73172690763052, "grad_norm": 8.97808837890625, "learning_rate": 1.0935483870967742e-06, "loss": 0.3676, "step": 15161 }, { "epoch": 97.73815261044177, "grad_norm": 35.39665222167969, "learning_rate": 1.0903225806451614e-06, "loss": 0.2244, "step": 15162 }, { "epoch": 97.74457831325302, "grad_norm": 3.720918893814087, "learning_rate": 1.0870967741935484e-06, "loss": 0.3835, "step": 15163 }, { "epoch": 97.75100401606426, "grad_norm": 18.671415328979492, "learning_rate": 1.0838709677419356e-06, "loss": 0.5891, "step": 15164 }, { "epoch": 97.7574297188755, "grad_norm": 3.2147998809814453, "learning_rate": 1.0806451612903228e-06, "loss": 0.3563, "step": 15165 }, { "epoch": 97.76385542168674, "grad_norm": 12.327836990356445, "learning_rate": 1.0774193548387097e-06, "loss": 0.2899, "step": 15166 }, { "epoch": 97.770281124498, "grad_norm": 11.392914772033691, "learning_rate": 1.0741935483870967e-06, "loss": 0.4759, "step": 15167 }, { "epoch": 97.77670682730924, "grad_norm": 12.641975402832031, "learning_rate": 1.070967741935484e-06, "loss": 0.1819, "step": 15168 }, { "epoch": 97.78313253012048, "grad_norm": 4.325752258300781, "learning_rate": 1.0677419354838709e-06, "loss": 0.2876, "step": 15169 }, { "epoch": 97.78955823293173, "grad_norm": 2.923379421234131, "learning_rate": 1.064516129032258e-06, "loss": 0.2483, "step": 15170 }, { "epoch": 97.79598393574297, "grad_norm": 4.224870681762695, "learning_rate": 1.0612903225806453e-06, "loss": 0.142, "step": 15171 }, { "epoch": 97.80240963855422, "grad_norm": 23.719898223876953, "learning_rate": 1.0580645161290322e-06, "loss": 0.5716, "step": 15172 }, { "epoch": 97.80883534136547, "grad_norm": 9.919095993041992, "learning_rate": 1.0548387096774194e-06, "loss": 0.2765, "step": 15173 }, { "epoch": 97.81526104417671, "grad_norm": 11.418497085571289, "learning_rate": 1.0516129032258066e-06, "loss": 0.2113, "step": 15174 }, { "epoch": 97.82168674698795, "grad_norm": 4.444021224975586, "learning_rate": 1.0483870967741936e-06, "loss": 0.2938, "step": 15175 }, { "epoch": 97.82811244979919, "grad_norm": 10.228302955627441, "learning_rate": 1.0451612903225806e-06, "loss": 0.4797, "step": 15176 }, { "epoch": 97.83453815261043, "grad_norm": 2.3045196533203125, "learning_rate": 1.0419354838709678e-06, "loss": 0.2032, "step": 15177 }, { "epoch": 97.84096385542169, "grad_norm": 17.738502502441406, "learning_rate": 1.0387096774193547e-06, "loss": 0.5095, "step": 15178 }, { "epoch": 97.84738955823293, "grad_norm": 6.256260395050049, "learning_rate": 1.035483870967742e-06, "loss": 0.4463, "step": 15179 }, { "epoch": 97.85381526104418, "grad_norm": 2.9046919345855713, "learning_rate": 1.0322580645161291e-06, "loss": 0.095, "step": 15180 }, { "epoch": 97.86024096385542, "grad_norm": 8.031854629516602, "learning_rate": 1.029032258064516e-06, "loss": 0.3064, "step": 15181 }, { "epoch": 97.86666666666666, "grad_norm": 8.976984977722168, "learning_rate": 1.0258064516129033e-06, "loss": 0.1955, "step": 15182 }, { "epoch": 97.87309236947792, "grad_norm": 6.4128899574279785, "learning_rate": 1.0225806451612905e-06, "loss": 0.2776, "step": 15183 }, { "epoch": 97.87951807228916, "grad_norm": 12.647272109985352, "learning_rate": 1.0193548387096775e-06, "loss": 0.2466, "step": 15184 }, { "epoch": 97.8859437751004, "grad_norm": 7.3373517990112305, "learning_rate": 1.0161290322580644e-06, "loss": 0.2976, "step": 15185 }, { "epoch": 97.89236947791164, "grad_norm": 12.446063995361328, "learning_rate": 1.0129032258064516e-06, "loss": 0.3974, "step": 15186 }, { "epoch": 97.89879518072289, "grad_norm": 11.135602951049805, "learning_rate": 1.0096774193548388e-06, "loss": 0.3585, "step": 15187 }, { "epoch": 97.90522088353414, "grad_norm": 22.88185691833496, "learning_rate": 1.0064516129032258e-06, "loss": 0.2066, "step": 15188 }, { "epoch": 97.91164658634538, "grad_norm": 22.455446243286133, "learning_rate": 1.003225806451613e-06, "loss": 0.521, "step": 15189 }, { "epoch": 97.91807228915663, "grad_norm": 36.25261306762695, "learning_rate": 1.0000000000000002e-06, "loss": 0.46, "step": 15190 }, { "epoch": 97.92449799196787, "grad_norm": 14.136408805847168, "learning_rate": 9.967741935483871e-07, "loss": 0.3132, "step": 15191 }, { "epoch": 97.93092369477911, "grad_norm": 11.798955917358398, "learning_rate": 9.935483870967743e-07, "loss": 0.2827, "step": 15192 }, { "epoch": 97.93734939759037, "grad_norm": 22.089527130126953, "learning_rate": 9.903225806451615e-07, "loss": 0.1255, "step": 15193 }, { "epoch": 97.94377510040161, "grad_norm": 6.1063642501831055, "learning_rate": 9.870967741935483e-07, "loss": 0.4431, "step": 15194 }, { "epoch": 97.95020080321285, "grad_norm": 18.768024444580078, "learning_rate": 9.838709677419355e-07, "loss": 0.6914, "step": 15195 }, { "epoch": 97.9566265060241, "grad_norm": 39.62160110473633, "learning_rate": 9.806451612903227e-07, "loss": 0.3536, "step": 15196 }, { "epoch": 97.96305220883534, "grad_norm": 14.619217872619629, "learning_rate": 9.774193548387096e-07, "loss": 0.2293, "step": 15197 }, { "epoch": 97.96947791164659, "grad_norm": 1.6305973529815674, "learning_rate": 9.741935483870968e-07, "loss": 0.2849, "step": 15198 }, { "epoch": 97.97590361445783, "grad_norm": 28.94073486328125, "learning_rate": 9.70967741935484e-07, "loss": 0.5149, "step": 15199 }, { "epoch": 97.98232931726908, "grad_norm": 6.176438331604004, "learning_rate": 9.67741935483871e-07, "loss": 0.2373, "step": 15200 }, { "epoch": 97.98875502008032, "grad_norm": 29.153385162353516, "learning_rate": 9.645161290322582e-07, "loss": 0.4478, "step": 15201 }, { "epoch": 97.99518072289156, "grad_norm": 1.146134376525879, "learning_rate": 9.612903225806454e-07, "loss": 0.2323, "step": 15202 }, { "epoch": 98.00481927710844, "grad_norm": 14.109060287475586, "learning_rate": 9.580645161290321e-07, "loss": 0.2343, "step": 15203 }, { "epoch": 98.01124497991968, "grad_norm": 22.032390594482422, "learning_rate": 9.548387096774193e-07, "loss": 0.4745, "step": 15204 }, { "epoch": 98.01767068273092, "grad_norm": 5.877330780029297, "learning_rate": 9.516129032258064e-07, "loss": 0.1459, "step": 15205 }, { "epoch": 98.02409638554217, "grad_norm": 8.074888229370117, "learning_rate": 9.483870967741936e-07, "loss": 0.2434, "step": 15206 }, { "epoch": 98.03052208835341, "grad_norm": 9.769190788269043, "learning_rate": 9.451612903225807e-07, "loss": 0.36, "step": 15207 }, { "epoch": 98.03694779116466, "grad_norm": 30.57162857055664, "learning_rate": 9.419354838709678e-07, "loss": 0.5381, "step": 15208 }, { "epoch": 98.0433734939759, "grad_norm": 5.595142364501953, "learning_rate": 9.38709677419355e-07, "loss": 0.1271, "step": 15209 }, { "epoch": 98.04979919678715, "grad_norm": 16.05998992919922, "learning_rate": 9.35483870967742e-07, "loss": 0.3513, "step": 15210 }, { "epoch": 98.05622489959839, "grad_norm": 11.453659057617188, "learning_rate": 9.322580645161291e-07, "loss": 0.3078, "step": 15211 }, { "epoch": 98.06265060240963, "grad_norm": 6.976768970489502, "learning_rate": 9.290322580645161e-07, "loss": 0.2047, "step": 15212 }, { "epoch": 98.06907630522089, "grad_norm": 2.420428514480591, "learning_rate": 9.258064516129032e-07, "loss": 0.1725, "step": 15213 }, { "epoch": 98.07550200803213, "grad_norm": 7.209744930267334, "learning_rate": 9.225806451612904e-07, "loss": 0.3027, "step": 15214 }, { "epoch": 98.08192771084337, "grad_norm": 3.709152936935425, "learning_rate": 9.193548387096775e-07, "loss": 0.305, "step": 15215 }, { "epoch": 98.08835341365462, "grad_norm": 5.746679306030273, "learning_rate": 9.161290322580645e-07, "loss": 0.2437, "step": 15216 }, { "epoch": 98.09477911646586, "grad_norm": 9.375129699707031, "learning_rate": 9.129032258064517e-07, "loss": 0.3715, "step": 15217 }, { "epoch": 98.10120481927711, "grad_norm": 5.618770122528076, "learning_rate": 9.096774193548388e-07, "loss": 0.144, "step": 15218 }, { "epoch": 98.10763052208836, "grad_norm": 6.37316370010376, "learning_rate": 9.064516129032259e-07, "loss": 0.3411, "step": 15219 }, { "epoch": 98.1140562248996, "grad_norm": 16.89469337463379, "learning_rate": 9.03225806451613e-07, "loss": 0.3689, "step": 15220 }, { "epoch": 98.12048192771084, "grad_norm": 20.827970504760742, "learning_rate": 9e-07, "loss": 0.4377, "step": 15221 }, { "epoch": 98.12690763052208, "grad_norm": 0.6666064858436584, "learning_rate": 8.96774193548387e-07, "loss": 0.211, "step": 15222 }, { "epoch": 98.13333333333334, "grad_norm": 3.5394651889801025, "learning_rate": 8.935483870967742e-07, "loss": 0.0399, "step": 15223 }, { "epoch": 98.13975903614458, "grad_norm": 7.035942554473877, "learning_rate": 8.903225806451613e-07, "loss": 0.3846, "step": 15224 }, { "epoch": 98.14618473895582, "grad_norm": 8.663742065429688, "learning_rate": 8.870967741935484e-07, "loss": 0.3416, "step": 15225 }, { "epoch": 98.15261044176707, "grad_norm": 8.970755577087402, "learning_rate": 8.838709677419356e-07, "loss": 0.3461, "step": 15226 }, { "epoch": 98.15903614457831, "grad_norm": 31.632356643676758, "learning_rate": 8.806451612903227e-07, "loss": 0.2318, "step": 15227 }, { "epoch": 98.16546184738957, "grad_norm": 13.697091102600098, "learning_rate": 8.774193548387098e-07, "loss": 0.3323, "step": 15228 }, { "epoch": 98.17188755020081, "grad_norm": 7.479915142059326, "learning_rate": 8.741935483870967e-07, "loss": 0.5631, "step": 15229 }, { "epoch": 98.17831325301205, "grad_norm": 53.54991149902344, "learning_rate": 8.709677419354838e-07, "loss": 0.3906, "step": 15230 }, { "epoch": 98.18473895582329, "grad_norm": 14.471731185913086, "learning_rate": 8.67741935483871e-07, "loss": 0.5455, "step": 15231 }, { "epoch": 98.19116465863453, "grad_norm": 24.4993839263916, "learning_rate": 8.645161290322581e-07, "loss": 0.3108, "step": 15232 }, { "epoch": 98.19759036144578, "grad_norm": 15.925439834594727, "learning_rate": 8.612903225806452e-07, "loss": 0.3765, "step": 15233 }, { "epoch": 98.20401606425703, "grad_norm": 21.7183780670166, "learning_rate": 8.580645161290324e-07, "loss": 0.3262, "step": 15234 }, { "epoch": 98.21044176706827, "grad_norm": 24.95956039428711, "learning_rate": 8.548387096774194e-07, "loss": 0.2596, "step": 15235 }, { "epoch": 98.21686746987952, "grad_norm": 22.298763275146484, "learning_rate": 8.516129032258065e-07, "loss": 0.2258, "step": 15236 }, { "epoch": 98.22329317269076, "grad_norm": 4.391711235046387, "learning_rate": 8.483870967741936e-07, "loss": 0.2249, "step": 15237 }, { "epoch": 98.229718875502, "grad_norm": 1.863879680633545, "learning_rate": 8.451612903225806e-07, "loss": 0.0742, "step": 15238 }, { "epoch": 98.23614457831326, "grad_norm": 19.111276626586914, "learning_rate": 8.419354838709677e-07, "loss": 0.4569, "step": 15239 }, { "epoch": 98.2425702811245, "grad_norm": 16.052614212036133, "learning_rate": 8.387096774193549e-07, "loss": 0.3021, "step": 15240 }, { "epoch": 98.24899598393574, "grad_norm": 17.795520782470703, "learning_rate": 8.354838709677419e-07, "loss": 0.457, "step": 15241 }, { "epoch": 98.25542168674698, "grad_norm": 21.601743698120117, "learning_rate": 8.32258064516129e-07, "loss": 0.4688, "step": 15242 }, { "epoch": 98.26184738955823, "grad_norm": 17.344547271728516, "learning_rate": 8.290322580645162e-07, "loss": 0.2395, "step": 15243 }, { "epoch": 98.26827309236948, "grad_norm": 20.48379898071289, "learning_rate": 8.258064516129033e-07, "loss": 0.2127, "step": 15244 }, { "epoch": 98.27469879518073, "grad_norm": 16.01972770690918, "learning_rate": 8.225806451612904e-07, "loss": 0.3608, "step": 15245 }, { "epoch": 98.28112449799197, "grad_norm": 24.73232650756836, "learning_rate": 8.193548387096776e-07, "loss": 0.2499, "step": 15246 }, { "epoch": 98.28755020080321, "grad_norm": 9.291291236877441, "learning_rate": 8.161290322580644e-07, "loss": 0.5965, "step": 15247 }, { "epoch": 98.29397590361445, "grad_norm": 2.1927337646484375, "learning_rate": 8.129032258064516e-07, "loss": 0.357, "step": 15248 }, { "epoch": 98.30040160642571, "grad_norm": 30.610095977783203, "learning_rate": 8.096774193548387e-07, "loss": 0.4069, "step": 15249 }, { "epoch": 98.30682730923695, "grad_norm": 27.193958282470703, "learning_rate": 8.064516129032258e-07, "loss": 0.4513, "step": 15250 }, { "epoch": 98.3132530120482, "grad_norm": 5.696734428405762, "learning_rate": 8.03225806451613e-07, "loss": 0.2501, "step": 15251 }, { "epoch": 98.31967871485944, "grad_norm": 7.905224323272705, "learning_rate": 8.000000000000001e-07, "loss": 0.2827, "step": 15252 }, { "epoch": 98.32610441767068, "grad_norm": 5.89695930480957, "learning_rate": 7.967741935483872e-07, "loss": 0.2025, "step": 15253 }, { "epoch": 98.33253012048193, "grad_norm": 38.44022750854492, "learning_rate": 7.935483870967742e-07, "loss": 0.3476, "step": 15254 }, { "epoch": 98.33895582329318, "grad_norm": 2.9784398078918457, "learning_rate": 7.903225806451614e-07, "loss": 0.1755, "step": 15255 }, { "epoch": 98.34538152610442, "grad_norm": 9.371220588684082, "learning_rate": 7.870967741935483e-07, "loss": 0.2188, "step": 15256 }, { "epoch": 98.35180722891566, "grad_norm": 51.421722412109375, "learning_rate": 7.838709677419355e-07, "loss": 0.3909, "step": 15257 }, { "epoch": 98.3582329317269, "grad_norm": 23.78407859802246, "learning_rate": 7.806451612903226e-07, "loss": 0.1367, "step": 15258 }, { "epoch": 98.36465863453816, "grad_norm": 32.410614013671875, "learning_rate": 7.774193548387097e-07, "loss": 0.3788, "step": 15259 }, { "epoch": 98.3710843373494, "grad_norm": 13.44857406616211, "learning_rate": 7.741935483870968e-07, "loss": 0.2378, "step": 15260 }, { "epoch": 98.37751004016064, "grad_norm": 5.59799861907959, "learning_rate": 7.709677419354839e-07, "loss": 0.279, "step": 15261 }, { "epoch": 98.38393574297189, "grad_norm": 18.157310485839844, "learning_rate": 7.67741935483871e-07, "loss": 0.5129, "step": 15262 }, { "epoch": 98.39036144578313, "grad_norm": 16.08034324645996, "learning_rate": 7.645161290322581e-07, "loss": 0.5815, "step": 15263 }, { "epoch": 98.39678714859438, "grad_norm": 8.222456932067871, "learning_rate": 7.612903225806452e-07, "loss": 0.1285, "step": 15264 }, { "epoch": 98.40321285140563, "grad_norm": 12.398505210876465, "learning_rate": 7.580645161290323e-07, "loss": 0.2828, "step": 15265 }, { "epoch": 98.40963855421687, "grad_norm": 5.3872222900390625, "learning_rate": 7.548387096774194e-07, "loss": 0.4379, "step": 15266 }, { "epoch": 98.41606425702811, "grad_norm": 22.133508682250977, "learning_rate": 7.516129032258064e-07, "loss": 0.2764, "step": 15267 }, { "epoch": 98.42248995983935, "grad_norm": 8.31542682647705, "learning_rate": 7.483870967741936e-07, "loss": 0.3058, "step": 15268 }, { "epoch": 98.4289156626506, "grad_norm": 5.472448348999023, "learning_rate": 7.451612903225807e-07, "loss": 0.5479, "step": 15269 }, { "epoch": 98.43534136546185, "grad_norm": 10.421720504760742, "learning_rate": 7.419354838709678e-07, "loss": 0.2938, "step": 15270 }, { "epoch": 98.4417670682731, "grad_norm": 20.384496688842773, "learning_rate": 7.387096774193549e-07, "loss": 0.4387, "step": 15271 }, { "epoch": 98.44819277108434, "grad_norm": 3.600383996963501, "learning_rate": 7.354838709677419e-07, "loss": 0.568, "step": 15272 }, { "epoch": 98.45461847389558, "grad_norm": 12.904111862182617, "learning_rate": 7.32258064516129e-07, "loss": 0.1309, "step": 15273 }, { "epoch": 98.46104417670682, "grad_norm": 14.62114429473877, "learning_rate": 7.290322580645162e-07, "loss": 0.2657, "step": 15274 }, { "epoch": 98.46746987951808, "grad_norm": 15.869316101074219, "learning_rate": 7.258064516129033e-07, "loss": 0.2741, "step": 15275 }, { "epoch": 98.47389558232932, "grad_norm": 3.6319127082824707, "learning_rate": 7.225806451612903e-07, "loss": 0.0966, "step": 15276 }, { "epoch": 98.48032128514056, "grad_norm": 24.370765686035156, "learning_rate": 7.193548387096775e-07, "loss": 0.6303, "step": 15277 }, { "epoch": 98.4867469879518, "grad_norm": 19.746335983276367, "learning_rate": 7.161290322580646e-07, "loss": 0.4635, "step": 15278 }, { "epoch": 98.49317269076305, "grad_norm": 16.114456176757812, "learning_rate": 7.129032258064516e-07, "loss": 0.5062, "step": 15279 }, { "epoch": 98.4995983935743, "grad_norm": 11.667956352233887, "learning_rate": 7.096774193548388e-07, "loss": 0.2436, "step": 15280 }, { "epoch": 98.50602409638554, "grad_norm": 12.552274703979492, "learning_rate": 7.064516129032258e-07, "loss": 0.1488, "step": 15281 }, { "epoch": 98.51244979919679, "grad_norm": 4.350343227386475, "learning_rate": 7.032258064516129e-07, "loss": 0.1174, "step": 15282 }, { "epoch": 98.51887550200803, "grad_norm": 8.921283721923828, "learning_rate": 7.000000000000001e-07, "loss": 0.216, "step": 15283 }, { "epoch": 98.52530120481927, "grad_norm": 27.331798553466797, "learning_rate": 6.967741935483872e-07, "loss": 0.2791, "step": 15284 }, { "epoch": 98.53172690763053, "grad_norm": 6.25811243057251, "learning_rate": 6.935483870967742e-07, "loss": 0.4385, "step": 15285 }, { "epoch": 98.53815261044177, "grad_norm": 34.64243698120117, "learning_rate": 6.903225806451613e-07, "loss": 0.3841, "step": 15286 }, { "epoch": 98.54457831325301, "grad_norm": 29.463634490966797, "learning_rate": 6.870967741935484e-07, "loss": 0.5096, "step": 15287 }, { "epoch": 98.55100401606425, "grad_norm": 10.894829750061035, "learning_rate": 6.838709677419355e-07, "loss": 0.3801, "step": 15288 }, { "epoch": 98.5574297188755, "grad_norm": 7.1650190353393555, "learning_rate": 6.806451612903226e-07, "loss": 0.2858, "step": 15289 }, { "epoch": 98.56385542168675, "grad_norm": 38.407474517822266, "learning_rate": 6.774193548387097e-07, "loss": 0.4681, "step": 15290 }, { "epoch": 98.570281124498, "grad_norm": 12.120704650878906, "learning_rate": 6.741935483870968e-07, "loss": 0.615, "step": 15291 }, { "epoch": 98.57670682730924, "grad_norm": 17.809968948364258, "learning_rate": 6.709677419354839e-07, "loss": 0.3124, "step": 15292 }, { "epoch": 98.58313253012048, "grad_norm": 17.363142013549805, "learning_rate": 6.67741935483871e-07, "loss": 0.3094, "step": 15293 }, { "epoch": 98.58955823293172, "grad_norm": 15.345542907714844, "learning_rate": 6.645161290322581e-07, "loss": 0.3828, "step": 15294 }, { "epoch": 98.59598393574298, "grad_norm": 8.651236534118652, "learning_rate": 6.612903225806452e-07, "loss": 0.3273, "step": 15295 }, { "epoch": 98.60240963855422, "grad_norm": 47.410369873046875, "learning_rate": 6.580645161290323e-07, "loss": 0.3912, "step": 15296 }, { "epoch": 98.60883534136546, "grad_norm": 4.885321617126465, "learning_rate": 6.548387096774195e-07, "loss": 0.1822, "step": 15297 }, { "epoch": 98.6152610441767, "grad_norm": 9.5950288772583, "learning_rate": 6.516129032258064e-07, "loss": 0.3281, "step": 15298 }, { "epoch": 98.62168674698795, "grad_norm": 7.971709251403809, "learning_rate": 6.483870967741935e-07, "loss": 0.3469, "step": 15299 }, { "epoch": 98.6281124497992, "grad_norm": 16.5406551361084, "learning_rate": 6.451612903225807e-07, "loss": 0.6725, "step": 15300 }, { "epoch": 98.63453815261045, "grad_norm": 12.565295219421387, "learning_rate": 6.419354838709678e-07, "loss": 0.435, "step": 15301 }, { "epoch": 98.64096385542169, "grad_norm": 3.951794147491455, "learning_rate": 6.387096774193549e-07, "loss": 0.3302, "step": 15302 }, { "epoch": 98.64738955823293, "grad_norm": 20.340030670166016, "learning_rate": 6.35483870967742e-07, "loss": 0.2903, "step": 15303 }, { "epoch": 98.65381526104417, "grad_norm": 13.124839782714844, "learning_rate": 6.32258064516129e-07, "loss": 0.2917, "step": 15304 }, { "epoch": 98.66024096385541, "grad_norm": 41.155242919921875, "learning_rate": 6.290322580645161e-07, "loss": 0.2266, "step": 15305 }, { "epoch": 98.66666666666667, "grad_norm": 10.623631477355957, "learning_rate": 6.258064516129033e-07, "loss": 0.4402, "step": 15306 }, { "epoch": 98.67309236947791, "grad_norm": 12.943266868591309, "learning_rate": 6.225806451612903e-07, "loss": 0.5434, "step": 15307 }, { "epoch": 98.67951807228916, "grad_norm": 14.606571197509766, "learning_rate": 6.193548387096775e-07, "loss": 0.1908, "step": 15308 }, { "epoch": 98.6859437751004, "grad_norm": 13.488358497619629, "learning_rate": 6.161290322580646e-07, "loss": 0.3607, "step": 15309 }, { "epoch": 98.69236947791164, "grad_norm": 16.005891799926758, "learning_rate": 6.129032258064516e-07, "loss": 0.3911, "step": 15310 }, { "epoch": 98.6987951807229, "grad_norm": 10.82468032836914, "learning_rate": 6.096774193548388e-07, "loss": 0.4445, "step": 15311 }, { "epoch": 98.70522088353414, "grad_norm": 6.000050067901611, "learning_rate": 6.064516129032258e-07, "loss": 0.2397, "step": 15312 }, { "epoch": 98.71164658634538, "grad_norm": 25.800281524658203, "learning_rate": 6.032258064516129e-07, "loss": 0.4259, "step": 15313 }, { "epoch": 98.71807228915662, "grad_norm": 9.29837417602539, "learning_rate": 6.000000000000001e-07, "loss": 0.3955, "step": 15314 }, { "epoch": 98.72449799196787, "grad_norm": 18.15039825439453, "learning_rate": 5.967741935483872e-07, "loss": 0.3774, "step": 15315 }, { "epoch": 98.73092369477912, "grad_norm": 17.59178352355957, "learning_rate": 5.935483870967741e-07, "loss": 0.3727, "step": 15316 }, { "epoch": 98.73734939759036, "grad_norm": 10.510601043701172, "learning_rate": 5.903225806451613e-07, "loss": 0.2598, "step": 15317 }, { "epoch": 98.7437751004016, "grad_norm": 8.770853996276855, "learning_rate": 5.870967741935484e-07, "loss": 0.2466, "step": 15318 }, { "epoch": 98.75020080321285, "grad_norm": 20.038118362426758, "learning_rate": 5.838709677419355e-07, "loss": 0.3128, "step": 15319 }, { "epoch": 98.75662650602409, "grad_norm": 12.495664596557617, "learning_rate": 5.806451612903227e-07, "loss": 0.3263, "step": 15320 }, { "epoch": 98.76305220883535, "grad_norm": 16.5413875579834, "learning_rate": 5.774193548387097e-07, "loss": 0.2631, "step": 15321 }, { "epoch": 98.76947791164659, "grad_norm": 5.593161582946777, "learning_rate": 5.741935483870967e-07, "loss": 0.3864, "step": 15322 }, { "epoch": 98.77590361445783, "grad_norm": 8.94122314453125, "learning_rate": 5.709677419354839e-07, "loss": 0.3233, "step": 15323 }, { "epoch": 98.78232931726907, "grad_norm": 8.648517608642578, "learning_rate": 5.67741935483871e-07, "loss": 0.2239, "step": 15324 }, { "epoch": 98.78875502008032, "grad_norm": 14.364272117614746, "learning_rate": 5.645161290322581e-07, "loss": 0.2619, "step": 15325 }, { "epoch": 98.79518072289157, "grad_norm": 7.57214879989624, "learning_rate": 5.612903225806452e-07, "loss": 0.2757, "step": 15326 }, { "epoch": 98.80160642570281, "grad_norm": 6.709228038787842, "learning_rate": 5.580645161290323e-07, "loss": 0.2372, "step": 15327 }, { "epoch": 98.80803212851406, "grad_norm": 12.801746368408203, "learning_rate": 5.548387096774195e-07, "loss": 0.5474, "step": 15328 }, { "epoch": 98.8144578313253, "grad_norm": 9.68685531616211, "learning_rate": 5.516129032258064e-07, "loss": 0.4192, "step": 15329 }, { "epoch": 98.82088353413654, "grad_norm": 5.846255779266357, "learning_rate": 5.483870967741935e-07, "loss": 0.3173, "step": 15330 }, { "epoch": 98.8273092369478, "grad_norm": 7.831781387329102, "learning_rate": 5.451612903225807e-07, "loss": 0.2131, "step": 15331 }, { "epoch": 98.83373493975904, "grad_norm": 12.51936149597168, "learning_rate": 5.419354838709678e-07, "loss": 0.3701, "step": 15332 }, { "epoch": 98.84016064257028, "grad_norm": 3.9581356048583984, "learning_rate": 5.387096774193549e-07, "loss": 0.3371, "step": 15333 }, { "epoch": 98.84658634538152, "grad_norm": 31.265277862548828, "learning_rate": 5.35483870967742e-07, "loss": 0.6441, "step": 15334 }, { "epoch": 98.85301204819277, "grad_norm": 2.5105388164520264, "learning_rate": 5.32258064516129e-07, "loss": 0.3526, "step": 15335 }, { "epoch": 98.85943775100401, "grad_norm": 11.966130256652832, "learning_rate": 5.290322580645161e-07, "loss": 0.2786, "step": 15336 }, { "epoch": 98.86586345381527, "grad_norm": 20.387311935424805, "learning_rate": 5.258064516129033e-07, "loss": 0.526, "step": 15337 }, { "epoch": 98.87228915662651, "grad_norm": 13.548686981201172, "learning_rate": 5.225806451612903e-07, "loss": 0.3369, "step": 15338 }, { "epoch": 98.87871485943775, "grad_norm": 9.935334205627441, "learning_rate": 5.193548387096774e-07, "loss": 0.3861, "step": 15339 }, { "epoch": 98.88514056224899, "grad_norm": 7.2365288734436035, "learning_rate": 5.161290322580646e-07, "loss": 0.2603, "step": 15340 }, { "epoch": 98.89156626506023, "grad_norm": 2.5359482765197754, "learning_rate": 5.129032258064516e-07, "loss": 0.3416, "step": 15341 }, { "epoch": 98.89799196787149, "grad_norm": 4.59909725189209, "learning_rate": 5.096774193548387e-07, "loss": 0.3281, "step": 15342 }, { "epoch": 98.90441767068273, "grad_norm": 2.349828004837036, "learning_rate": 5.064516129032258e-07, "loss": 0.3337, "step": 15343 }, { "epoch": 98.91084337349398, "grad_norm": 9.855101585388184, "learning_rate": 5.032258064516129e-07, "loss": 0.3021, "step": 15344 }, { "epoch": 98.91726907630522, "grad_norm": 30.444345474243164, "learning_rate": 5.000000000000001e-07, "loss": 0.1951, "step": 15345 }, { "epoch": 98.92369477911646, "grad_norm": 37.1721305847168, "learning_rate": 4.967741935483872e-07, "loss": 0.3472, "step": 15346 }, { "epoch": 98.93012048192772, "grad_norm": 7.7995476722717285, "learning_rate": 4.935483870967741e-07, "loss": 0.2574, "step": 15347 }, { "epoch": 98.93654618473896, "grad_norm": 30.090190887451172, "learning_rate": 4.903225806451613e-07, "loss": 0.2433, "step": 15348 }, { "epoch": 98.9429718875502, "grad_norm": 15.281509399414062, "learning_rate": 4.870967741935484e-07, "loss": 0.2099, "step": 15349 }, { "epoch": 98.94939759036144, "grad_norm": 26.01926040649414, "learning_rate": 4.838709677419355e-07, "loss": 0.6651, "step": 15350 }, { "epoch": 98.95582329317268, "grad_norm": 8.391522407531738, "learning_rate": 4.806451612903227e-07, "loss": 0.2113, "step": 15351 }, { "epoch": 98.96224899598394, "grad_norm": 16.952892303466797, "learning_rate": 4.774193548387097e-07, "loss": 0.3606, "step": 15352 }, { "epoch": 98.96867469879518, "grad_norm": 5.315958499908447, "learning_rate": 4.741935483870968e-07, "loss": 0.1893, "step": 15353 }, { "epoch": 98.97510040160643, "grad_norm": 8.628379821777344, "learning_rate": 4.709677419354839e-07, "loss": 0.3248, "step": 15354 }, { "epoch": 98.98152610441767, "grad_norm": 12.744010925292969, "learning_rate": 4.67741935483871e-07, "loss": 0.4012, "step": 15355 }, { "epoch": 98.98795180722891, "grad_norm": 72.77892303466797, "learning_rate": 4.6451612903225805e-07, "loss": 0.4865, "step": 15356 }, { "epoch": 98.99437751004017, "grad_norm": 14.045919418334961, "learning_rate": 4.612903225806452e-07, "loss": 0.3258, "step": 15357 }, { "epoch": 99.00401606425703, "grad_norm": 6.623345375061035, "learning_rate": 4.5806451612903227e-07, "loss": 0.1794, "step": 15358 }, { "epoch": 99.01044176706827, "grad_norm": 5.113718032836914, "learning_rate": 4.548387096774194e-07, "loss": 0.2264, "step": 15359 }, { "epoch": 99.01686746987951, "grad_norm": 28.877830505371094, "learning_rate": 4.516129032258065e-07, "loss": 0.3681, "step": 15360 }, { "epoch": 99.02329317269076, "grad_norm": 4.505036354064941, "learning_rate": 4.483870967741935e-07, "loss": 0.2299, "step": 15361 }, { "epoch": 99.02971887550201, "grad_norm": 21.129905700683594, "learning_rate": 4.4516129032258066e-07, "loss": 0.3739, "step": 15362 }, { "epoch": 99.03614457831326, "grad_norm": 5.796590805053711, "learning_rate": 4.419354838709678e-07, "loss": 0.3297, "step": 15363 }, { "epoch": 99.0425702811245, "grad_norm": 15.47343635559082, "learning_rate": 4.387096774193549e-07, "loss": 0.6878, "step": 15364 }, { "epoch": 99.04899598393574, "grad_norm": 10.525891304016113, "learning_rate": 4.354838709677419e-07, "loss": 0.4266, "step": 15365 }, { "epoch": 99.05542168674698, "grad_norm": 23.767757415771484, "learning_rate": 4.3225806451612904e-07, "loss": 0.5595, "step": 15366 }, { "epoch": 99.06184738955824, "grad_norm": 13.41415786743164, "learning_rate": 4.290322580645162e-07, "loss": 0.3824, "step": 15367 }, { "epoch": 99.06827309236948, "grad_norm": 8.276689529418945, "learning_rate": 4.2580645161290326e-07, "loss": 0.3042, "step": 15368 }, { "epoch": 99.07469879518072, "grad_norm": 24.429645538330078, "learning_rate": 4.225806451612903e-07, "loss": 0.2022, "step": 15369 }, { "epoch": 99.08112449799196, "grad_norm": 22.410715103149414, "learning_rate": 4.1935483870967743e-07, "loss": 0.421, "step": 15370 }, { "epoch": 99.08755020080321, "grad_norm": 7.133176803588867, "learning_rate": 4.161290322580645e-07, "loss": 0.3028, "step": 15371 }, { "epoch": 99.09397590361446, "grad_norm": 10.236547470092773, "learning_rate": 4.1290322580645165e-07, "loss": 0.5167, "step": 15372 }, { "epoch": 99.1004016064257, "grad_norm": 7.349006652832031, "learning_rate": 4.096774193548388e-07, "loss": 0.1978, "step": 15373 }, { "epoch": 99.10682730923695, "grad_norm": 12.36010456085205, "learning_rate": 4.064516129032258e-07, "loss": 0.2962, "step": 15374 }, { "epoch": 99.11325301204819, "grad_norm": 17.36210060119629, "learning_rate": 4.032258064516129e-07, "loss": 0.4802, "step": 15375 }, { "epoch": 99.11967871485943, "grad_norm": 7.5404052734375, "learning_rate": 4.0000000000000003e-07, "loss": 0.3743, "step": 15376 }, { "epoch": 99.12610441767069, "grad_norm": 9.086018562316895, "learning_rate": 3.967741935483871e-07, "loss": 0.2976, "step": 15377 }, { "epoch": 99.13253012048193, "grad_norm": 14.946887016296387, "learning_rate": 3.9354838709677415e-07, "loss": 0.3768, "step": 15378 }, { "epoch": 99.13895582329317, "grad_norm": 9.26294994354248, "learning_rate": 3.903225806451613e-07, "loss": 0.4757, "step": 15379 }, { "epoch": 99.14538152610442, "grad_norm": 13.884615898132324, "learning_rate": 3.870967741935484e-07, "loss": 0.4607, "step": 15380 }, { "epoch": 99.15180722891566, "grad_norm": 22.80859375, "learning_rate": 3.838709677419355e-07, "loss": 0.297, "step": 15381 }, { "epoch": 99.15823293172691, "grad_norm": 9.951404571533203, "learning_rate": 3.806451612903226e-07, "loss": 0.3147, "step": 15382 }, { "epoch": 99.16465863453816, "grad_norm": 8.918716430664062, "learning_rate": 3.774193548387097e-07, "loss": 0.1523, "step": 15383 }, { "epoch": 99.1710843373494, "grad_norm": 35.223289489746094, "learning_rate": 3.741935483870968e-07, "loss": 0.3665, "step": 15384 }, { "epoch": 99.17751004016064, "grad_norm": 9.470962524414062, "learning_rate": 3.709677419354839e-07, "loss": 0.3934, "step": 15385 }, { "epoch": 99.18393574297188, "grad_norm": 11.283510208129883, "learning_rate": 3.6774193548387097e-07, "loss": 0.4244, "step": 15386 }, { "epoch": 99.19036144578314, "grad_norm": 7.089020252227783, "learning_rate": 3.645161290322581e-07, "loss": 0.3291, "step": 15387 }, { "epoch": 99.19678714859438, "grad_norm": 19.878929138183594, "learning_rate": 3.6129032258064514e-07, "loss": 0.361, "step": 15388 }, { "epoch": 99.20321285140562, "grad_norm": 12.390937805175781, "learning_rate": 3.580645161290323e-07, "loss": 0.4149, "step": 15389 }, { "epoch": 99.20963855421687, "grad_norm": 7.20651388168335, "learning_rate": 3.548387096774194e-07, "loss": 0.4349, "step": 15390 }, { "epoch": 99.21606425702811, "grad_norm": 17.61990737915039, "learning_rate": 3.5161290322580644e-07, "loss": 0.1637, "step": 15391 }, { "epoch": 99.22248995983935, "grad_norm": 12.761890411376953, "learning_rate": 3.483870967741936e-07, "loss": 0.6446, "step": 15392 }, { "epoch": 99.2289156626506, "grad_norm": 16.394638061523438, "learning_rate": 3.4516129032258066e-07, "loss": 0.4967, "step": 15393 }, { "epoch": 99.23534136546185, "grad_norm": 12.727529525756836, "learning_rate": 3.4193548387096775e-07, "loss": 0.1446, "step": 15394 }, { "epoch": 99.24176706827309, "grad_norm": 10.830512046813965, "learning_rate": 3.3870967741935483e-07, "loss": 0.3665, "step": 15395 }, { "epoch": 99.24819277108433, "grad_norm": 15.891002655029297, "learning_rate": 3.3548387096774196e-07, "loss": 0.1424, "step": 15396 }, { "epoch": 99.25461847389558, "grad_norm": 19.600467681884766, "learning_rate": 3.3225806451612905e-07, "loss": 0.1481, "step": 15397 }, { "epoch": 99.26104417670683, "grad_norm": 4.54005241394043, "learning_rate": 3.2903225806451613e-07, "loss": 0.2757, "step": 15398 }, { "epoch": 99.26746987951807, "grad_norm": 9.61064338684082, "learning_rate": 3.258064516129032e-07, "loss": 0.2921, "step": 15399 }, { "epoch": 99.27389558232932, "grad_norm": 4.652872085571289, "learning_rate": 3.2258064516129035e-07, "loss": 0.2613, "step": 15400 }, { "epoch": 99.28032128514056, "grad_norm": 13.207401275634766, "learning_rate": 3.1935483870967743e-07, "loss": 0.4606, "step": 15401 }, { "epoch": 99.2867469879518, "grad_norm": 19.35040855407715, "learning_rate": 3.161290322580645e-07, "loss": 0.2796, "step": 15402 }, { "epoch": 99.29317269076306, "grad_norm": 6.521413326263428, "learning_rate": 3.1290322580645165e-07, "loss": 0.2837, "step": 15403 }, { "epoch": 99.2995983935743, "grad_norm": 22.484447479248047, "learning_rate": 3.0967741935483874e-07, "loss": 0.446, "step": 15404 }, { "epoch": 99.30602409638554, "grad_norm": 5.960504055023193, "learning_rate": 3.064516129032258e-07, "loss": 0.0743, "step": 15405 }, { "epoch": 99.31244979919678, "grad_norm": 8.011530876159668, "learning_rate": 3.032258064516129e-07, "loss": 0.4118, "step": 15406 }, { "epoch": 99.31887550200803, "grad_norm": 8.288139343261719, "learning_rate": 3.0000000000000004e-07, "loss": 0.2726, "step": 15407 }, { "epoch": 99.32530120481928, "grad_norm": 9.67909049987793, "learning_rate": 2.9677419354838707e-07, "loss": 0.1806, "step": 15408 }, { "epoch": 99.33172690763053, "grad_norm": 5.266911029815674, "learning_rate": 2.935483870967742e-07, "loss": 0.1518, "step": 15409 }, { "epoch": 99.33815261044177, "grad_norm": 5.11619758605957, "learning_rate": 2.9032258064516134e-07, "loss": 0.2421, "step": 15410 }, { "epoch": 99.34457831325301, "grad_norm": 2.699716567993164, "learning_rate": 2.8709677419354837e-07, "loss": 0.2873, "step": 15411 }, { "epoch": 99.35100401606425, "grad_norm": 12.987552642822266, "learning_rate": 2.838709677419355e-07, "loss": 0.4186, "step": 15412 }, { "epoch": 99.35742971887551, "grad_norm": 3.352811098098755, "learning_rate": 2.806451612903226e-07, "loss": 0.2069, "step": 15413 }, { "epoch": 99.36385542168675, "grad_norm": 13.124077796936035, "learning_rate": 2.7741935483870973e-07, "loss": 0.4917, "step": 15414 }, { "epoch": 99.37028112449799, "grad_norm": 8.050386428833008, "learning_rate": 2.7419354838709676e-07, "loss": 0.1277, "step": 15415 }, { "epoch": 99.37670682730923, "grad_norm": 14.552573204040527, "learning_rate": 2.709677419354839e-07, "loss": 0.2509, "step": 15416 }, { "epoch": 99.38313253012048, "grad_norm": 24.99256134033203, "learning_rate": 2.67741935483871e-07, "loss": 0.4684, "step": 15417 }, { "epoch": 99.38955823293173, "grad_norm": 15.262107849121094, "learning_rate": 2.6451612903225806e-07, "loss": 0.4659, "step": 15418 }, { "epoch": 99.39598393574298, "grad_norm": 14.499817848205566, "learning_rate": 2.6129032258064514e-07, "loss": 0.3453, "step": 15419 }, { "epoch": 99.40240963855422, "grad_norm": 3.9501001834869385, "learning_rate": 2.580645161290323e-07, "loss": 0.2384, "step": 15420 }, { "epoch": 99.40883534136546, "grad_norm": 22.217252731323242, "learning_rate": 2.5483870967741936e-07, "loss": 0.3968, "step": 15421 }, { "epoch": 99.4152610441767, "grad_norm": 11.646116256713867, "learning_rate": 2.5161290322580645e-07, "loss": 0.2169, "step": 15422 }, { "epoch": 99.42168674698796, "grad_norm": 13.997762680053711, "learning_rate": 2.483870967741936e-07, "loss": 0.4359, "step": 15423 }, { "epoch": 99.4281124497992, "grad_norm": 8.13096809387207, "learning_rate": 2.4516129032258067e-07, "loss": 0.1846, "step": 15424 }, { "epoch": 99.43453815261044, "grad_norm": 17.552989959716797, "learning_rate": 2.4193548387096775e-07, "loss": 0.4517, "step": 15425 }, { "epoch": 99.44096385542169, "grad_norm": 10.153552055358887, "learning_rate": 2.3870967741935483e-07, "loss": 0.3588, "step": 15426 }, { "epoch": 99.44738955823293, "grad_norm": 8.533988952636719, "learning_rate": 2.3548387096774194e-07, "loss": 0.3637, "step": 15427 }, { "epoch": 99.45381526104417, "grad_norm": 5.496722221374512, "learning_rate": 2.3225806451612903e-07, "loss": 0.3884, "step": 15428 }, { "epoch": 99.46024096385543, "grad_norm": 13.188508033752441, "learning_rate": 2.2903225806451614e-07, "loss": 0.4551, "step": 15429 }, { "epoch": 99.46666666666667, "grad_norm": 7.251683712005615, "learning_rate": 2.2580645161290325e-07, "loss": 0.1168, "step": 15430 }, { "epoch": 99.47309236947791, "grad_norm": 13.669005393981934, "learning_rate": 2.2258064516129033e-07, "loss": 0.4438, "step": 15431 }, { "epoch": 99.47951807228915, "grad_norm": 7.263370037078857, "learning_rate": 2.1935483870967744e-07, "loss": 0.2267, "step": 15432 }, { "epoch": 99.4859437751004, "grad_norm": 21.330184936523438, "learning_rate": 2.1612903225806452e-07, "loss": 0.4824, "step": 15433 }, { "epoch": 99.49236947791165, "grad_norm": 6.450384616851807, "learning_rate": 2.1290322580645163e-07, "loss": 0.3913, "step": 15434 }, { "epoch": 99.4987951807229, "grad_norm": 15.013589859008789, "learning_rate": 2.0967741935483871e-07, "loss": 0.3654, "step": 15435 }, { "epoch": 99.50522088353414, "grad_norm": 3.138206958770752, "learning_rate": 2.0645161290322582e-07, "loss": 0.2808, "step": 15436 }, { "epoch": 99.51164658634538, "grad_norm": 9.599662780761719, "learning_rate": 2.032258064516129e-07, "loss": 0.2998, "step": 15437 }, { "epoch": 99.51807228915662, "grad_norm": 6.996429443359375, "learning_rate": 2.0000000000000002e-07, "loss": 0.2966, "step": 15438 }, { "epoch": 99.52449799196788, "grad_norm": 12.14965534210205, "learning_rate": 1.9677419354838707e-07, "loss": 0.3258, "step": 15439 }, { "epoch": 99.53092369477912, "grad_norm": 41.24433135986328, "learning_rate": 1.935483870967742e-07, "loss": 0.5809, "step": 15440 }, { "epoch": 99.53734939759036, "grad_norm": 26.906421661376953, "learning_rate": 1.903225806451613e-07, "loss": 0.3501, "step": 15441 }, { "epoch": 99.5437751004016, "grad_norm": 11.177011489868164, "learning_rate": 1.870967741935484e-07, "loss": 0.5062, "step": 15442 }, { "epoch": 99.55020080321285, "grad_norm": 63.00579071044922, "learning_rate": 1.8387096774193549e-07, "loss": 0.2388, "step": 15443 }, { "epoch": 99.5566265060241, "grad_norm": 12.123581886291504, "learning_rate": 1.8064516129032257e-07, "loss": 0.3684, "step": 15444 }, { "epoch": 99.56305220883534, "grad_norm": 5.055211544036865, "learning_rate": 1.774193548387097e-07, "loss": 0.4265, "step": 15445 }, { "epoch": 99.56947791164659, "grad_norm": 18.01650047302246, "learning_rate": 1.741935483870968e-07, "loss": 0.6276, "step": 15446 }, { "epoch": 99.57590361445783, "grad_norm": 4.304084777832031, "learning_rate": 1.7096774193548387e-07, "loss": 0.2742, "step": 15447 }, { "epoch": 99.58232931726907, "grad_norm": 7.083930492401123, "learning_rate": 1.6774193548387098e-07, "loss": 0.3352, "step": 15448 }, { "epoch": 99.58875502008033, "grad_norm": 15.712194442749023, "learning_rate": 1.6451612903225807e-07, "loss": 0.2041, "step": 15449 }, { "epoch": 99.59518072289157, "grad_norm": 13.049651145935059, "learning_rate": 1.6129032258064518e-07, "loss": 0.27, "step": 15450 }, { "epoch": 99.60160642570281, "grad_norm": 20.419143676757812, "learning_rate": 1.5806451612903226e-07, "loss": 0.2978, "step": 15451 }, { "epoch": 99.60803212851405, "grad_norm": 22.406612396240234, "learning_rate": 1.5483870967741937e-07, "loss": 0.3242, "step": 15452 }, { "epoch": 99.6144578313253, "grad_norm": 9.895040512084961, "learning_rate": 1.5161290322580645e-07, "loss": 0.3307, "step": 15453 }, { "epoch": 99.62088353413655, "grad_norm": 32.06381607055664, "learning_rate": 1.4838709677419353e-07, "loss": 0.5382, "step": 15454 }, { "epoch": 99.6273092369478, "grad_norm": 11.830222129821777, "learning_rate": 1.4516129032258067e-07, "loss": 0.2923, "step": 15455 }, { "epoch": 99.63373493975904, "grad_norm": 15.68044662475586, "learning_rate": 1.4193548387096775e-07, "loss": 0.3514, "step": 15456 }, { "epoch": 99.64016064257028, "grad_norm": 7.695986747741699, "learning_rate": 1.3870967741935486e-07, "loss": 0.2389, "step": 15457 }, { "epoch": 99.64658634538152, "grad_norm": 17.781490325927734, "learning_rate": 1.3548387096774195e-07, "loss": 0.3711, "step": 15458 }, { "epoch": 99.65301204819278, "grad_norm": 7.585606098175049, "learning_rate": 1.3225806451612903e-07, "loss": 0.2876, "step": 15459 }, { "epoch": 99.65943775100402, "grad_norm": 18.438411712646484, "learning_rate": 1.2903225806451614e-07, "loss": 0.7171, "step": 15460 }, { "epoch": 99.66586345381526, "grad_norm": 5.26407527923584, "learning_rate": 1.2580645161290322e-07, "loss": 0.1649, "step": 15461 }, { "epoch": 99.6722891566265, "grad_norm": 7.035383701324463, "learning_rate": 1.2258064516129033e-07, "loss": 0.301, "step": 15462 }, { "epoch": 99.67871485943775, "grad_norm": 11.892929077148438, "learning_rate": 1.1935483870967742e-07, "loss": 0.1732, "step": 15463 }, { "epoch": 99.68514056224899, "grad_norm": 19.89870262145996, "learning_rate": 1.1612903225806451e-07, "loss": 0.395, "step": 15464 }, { "epoch": 99.69156626506025, "grad_norm": 10.943474769592285, "learning_rate": 1.1290322580645162e-07, "loss": 0.1846, "step": 15465 }, { "epoch": 99.69799196787149, "grad_norm": 34.40411376953125, "learning_rate": 1.0967741935483872e-07, "loss": 0.4285, "step": 15466 }, { "epoch": 99.70441767068273, "grad_norm": 21.49817657470703, "learning_rate": 1.0645161290322582e-07, "loss": 0.2507, "step": 15467 }, { "epoch": 99.71084337349397, "grad_norm": 35.558258056640625, "learning_rate": 1.0322580645161291e-07, "loss": 0.262, "step": 15468 }, { "epoch": 99.71726907630521, "grad_norm": 21.39742088317871, "learning_rate": 1.0000000000000001e-07, "loss": 0.2856, "step": 15469 }, { "epoch": 99.72369477911647, "grad_norm": 12.846513748168945, "learning_rate": 9.67741935483871e-08, "loss": 0.5471, "step": 15470 }, { "epoch": 99.73012048192771, "grad_norm": 14.43424129486084, "learning_rate": 9.35483870967742e-08, "loss": 0.2596, "step": 15471 }, { "epoch": 99.73654618473896, "grad_norm": 28.954753875732422, "learning_rate": 9.032258064516128e-08, "loss": 0.48, "step": 15472 }, { "epoch": 99.7429718875502, "grad_norm": 16.10736083984375, "learning_rate": 8.70967741935484e-08, "loss": 0.4297, "step": 15473 }, { "epoch": 99.74939759036144, "grad_norm": 11.360620498657227, "learning_rate": 8.387096774193549e-08, "loss": 0.3737, "step": 15474 }, { "epoch": 99.7558232931727, "grad_norm": 23.056142807006836, "learning_rate": 8.064516129032259e-08, "loss": 0.5579, "step": 15475 }, { "epoch": 99.76224899598394, "grad_norm": 10.47696304321289, "learning_rate": 7.741935483870968e-08, "loss": 0.4505, "step": 15476 }, { "epoch": 99.76867469879518, "grad_norm": 13.880757331848145, "learning_rate": 7.419354838709677e-08, "loss": 0.2464, "step": 15477 }, { "epoch": 99.77510040160642, "grad_norm": 7.167011737823486, "learning_rate": 7.096774193548388e-08, "loss": 0.2479, "step": 15478 }, { "epoch": 99.78152610441767, "grad_norm": 10.235090255737305, "learning_rate": 6.774193548387097e-08, "loss": 0.1338, "step": 15479 }, { "epoch": 99.78795180722892, "grad_norm": 21.611764907836914, "learning_rate": 6.451612903225807e-08, "loss": 0.6537, "step": 15480 }, { "epoch": 99.79437751004016, "grad_norm": 13.333406448364258, "learning_rate": 6.129032258064517e-08, "loss": 0.4588, "step": 15481 }, { "epoch": 99.8008032128514, "grad_norm": 8.313750267028809, "learning_rate": 5.8064516129032257e-08, "loss": 0.5842, "step": 15482 }, { "epoch": 99.80722891566265, "grad_norm": 23.91421890258789, "learning_rate": 5.483870967741936e-08, "loss": 0.4856, "step": 15483 }, { "epoch": 99.81365461847389, "grad_norm": 5.723372936248779, "learning_rate": 5.1612903225806456e-08, "loss": 0.1761, "step": 15484 }, { "epoch": 99.82008032128515, "grad_norm": 13.877432823181152, "learning_rate": 4.838709677419355e-08, "loss": 0.3409, "step": 15485 }, { "epoch": 99.82650602409639, "grad_norm": 7.728745937347412, "learning_rate": 4.516129032258064e-08, "loss": 0.3563, "step": 15486 }, { "epoch": 99.83293172690763, "grad_norm": 6.97348165512085, "learning_rate": 4.1935483870967746e-08, "loss": 0.2271, "step": 15487 }, { "epoch": 99.83935742971887, "grad_norm": 52.392677307128906, "learning_rate": 3.870967741935484e-08, "loss": 0.6968, "step": 15488 }, { "epoch": 99.84578313253012, "grad_norm": 15.29877758026123, "learning_rate": 3.548387096774194e-08, "loss": 0.2721, "step": 15489 }, { "epoch": 99.85220883534137, "grad_norm": 8.536388397216797, "learning_rate": 3.2258064516129035e-08, "loss": 0.3598, "step": 15490 }, { "epoch": 99.85863453815261, "grad_norm": 2.8071346282958984, "learning_rate": 2.9032258064516128e-08, "loss": 0.1882, "step": 15491 }, { "epoch": 99.86506024096386, "grad_norm": 21.088193893432617, "learning_rate": 2.5806451612903228e-08, "loss": 0.2586, "step": 15492 }, { "epoch": 99.8714859437751, "grad_norm": 21.127412796020508, "learning_rate": 2.258064516129032e-08, "loss": 0.3585, "step": 15493 }, { "epoch": 99.87791164658634, "grad_norm": 7.727818489074707, "learning_rate": 1.935483870967742e-08, "loss": 0.3485, "step": 15494 }, { "epoch": 99.88433734939758, "grad_norm": 7.015499591827393, "learning_rate": 1.6129032258064518e-08, "loss": 0.3639, "step": 15495 }, { "epoch": 99.89076305220884, "grad_norm": 5.213562965393066, "learning_rate": 1.2903225806451614e-08, "loss": 0.1982, "step": 15496 }, { "epoch": 99.89718875502008, "grad_norm": 16.288240432739258, "learning_rate": 9.67741935483871e-09, "loss": 0.2693, "step": 15497 }, { "epoch": 99.90361445783132, "grad_norm": 17.252452850341797, "learning_rate": 6.451612903225807e-09, "loss": 0.3272, "step": 15498 }, { "epoch": 99.91004016064257, "grad_norm": 19.998262405395508, "learning_rate": 3.2258064516129035e-09, "loss": 0.34, "step": 15499 }, { "epoch": 99.91646586345381, "grad_norm": 5.198437690734863, "learning_rate": 0.0, "loss": 0.3486, "step": 15500 } ], "logging_steps": 1, "max_steps": 15500, "num_input_tokens_seen": 0, "num_train_epochs": 100, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2.785107677864054e+19, "train_batch_size": 1, "trial_name": null, "trial_params": null }