{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9746588693957114, "eval_steps": 500, "global_step": 8500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00011466574934067194, "grad_norm": 9.3125, "learning_rate": 0.0, "loss": 1.7887, "step": 1 }, { "epoch": 0.00022933149868134388, "grad_norm": 25.625, "learning_rate": 3.816793893129771e-08, "loss": 1.9041, "step": 2 }, { "epoch": 0.0003439972480220158, "grad_norm": 9.5625, "learning_rate": 7.633587786259542e-08, "loss": 1.597, "step": 3 }, { "epoch": 0.00045866299736268775, "grad_norm": 23.75, "learning_rate": 1.1450381679389314e-07, "loss": 2.7009, "step": 4 }, { "epoch": 0.0005733287467033597, "grad_norm": 10.5625, "learning_rate": 1.5267175572519085e-07, "loss": 2.14, "step": 5 }, { "epoch": 0.0006879944960440316, "grad_norm": 9.0625, "learning_rate": 1.9083969465648858e-07, "loss": 2.0872, "step": 6 }, { "epoch": 0.0008026602453847036, "grad_norm": 10.875, "learning_rate": 2.2900763358778629e-07, "loss": 1.8268, "step": 7 }, { "epoch": 0.0009173259947253755, "grad_norm": 9.875, "learning_rate": 2.67175572519084e-07, "loss": 2.045, "step": 8 }, { "epoch": 0.0010319917440660474, "grad_norm": 8.125, "learning_rate": 3.053435114503817e-07, "loss": 1.6562, "step": 9 }, { "epoch": 0.0011466574934067195, "grad_norm": 34.25, "learning_rate": 3.4351145038167945e-07, "loss": 2.323, "step": 10 }, { "epoch": 0.0012613232427473914, "grad_norm": 10.625, "learning_rate": 3.8167938931297716e-07, "loss": 1.5942, "step": 11 }, { "epoch": 0.0013759889920880633, "grad_norm": 9.5625, "learning_rate": 4.1984732824427486e-07, "loss": 1.985, "step": 12 }, { "epoch": 0.0014906547414287351, "grad_norm": 10.5, "learning_rate": 4.5801526717557257e-07, "loss": 2.0963, "step": 13 }, { "epoch": 0.0016053204907694072, "grad_norm": 11.4375, "learning_rate": 4.961832061068702e-07, "loss": 1.6478, "step": 14 }, { "epoch": 0.0017199862401100791, "grad_norm": 11.5, "learning_rate": 5.34351145038168e-07, "loss": 1.4838, "step": 15 }, { "epoch": 0.001834651989450751, "grad_norm": 12.5, "learning_rate": 5.725190839694656e-07, "loss": 2.0602, "step": 16 }, { "epoch": 0.001949317738791423, "grad_norm": 11.75, "learning_rate": 6.106870229007634e-07, "loss": 2.0683, "step": 17 }, { "epoch": 0.0020639834881320948, "grad_norm": 9.375, "learning_rate": 6.48854961832061e-07, "loss": 1.6647, "step": 18 }, { "epoch": 0.002178649237472767, "grad_norm": 22.0, "learning_rate": 6.870229007633589e-07, "loss": 2.3647, "step": 19 }, { "epoch": 0.002293314986813439, "grad_norm": 10.1875, "learning_rate": 7.251908396946565e-07, "loss": 1.8912, "step": 20 }, { "epoch": 0.002407980736154111, "grad_norm": 10.625, "learning_rate": 7.633587786259543e-07, "loss": 1.97, "step": 21 }, { "epoch": 0.0025226464854947827, "grad_norm": 10.625, "learning_rate": 8.01526717557252e-07, "loss": 1.9024, "step": 22 }, { "epoch": 0.0026373122348354546, "grad_norm": 14.25, "learning_rate": 8.396946564885497e-07, "loss": 1.9355, "step": 23 }, { "epoch": 0.0027519779841761265, "grad_norm": 9.5625, "learning_rate": 8.778625954198474e-07, "loss": 1.7156, "step": 24 }, { "epoch": 0.0028666437335167984, "grad_norm": 11.9375, "learning_rate": 9.160305343511451e-07, "loss": 1.7582, "step": 25 }, { "epoch": 0.0029813094828574703, "grad_norm": 12.1875, "learning_rate": 9.54198473282443e-07, "loss": 2.4415, "step": 26 }, { "epoch": 0.0030959752321981426, "grad_norm": 11.6875, "learning_rate": 9.923664122137404e-07, "loss": 2.3691, "step": 27 }, { "epoch": 0.0032106409815388145, "grad_norm": 11.0625, "learning_rate": 1.0305343511450382e-06, "loss": 1.5541, "step": 28 }, { "epoch": 0.0033253067308794864, "grad_norm": 11.625, "learning_rate": 1.068702290076336e-06, "loss": 1.9681, "step": 29 }, { "epoch": 0.0034399724802201583, "grad_norm": 13.9375, "learning_rate": 1.1068702290076337e-06, "loss": 1.9642, "step": 30 }, { "epoch": 0.00355463822956083, "grad_norm": 12.0625, "learning_rate": 1.1450381679389313e-06, "loss": 1.6962, "step": 31 }, { "epoch": 0.003669303978901502, "grad_norm": 8.3125, "learning_rate": 1.1832061068702292e-06, "loss": 1.5854, "step": 32 }, { "epoch": 0.003783969728242174, "grad_norm": 11.9375, "learning_rate": 1.2213740458015268e-06, "loss": 1.8477, "step": 33 }, { "epoch": 0.003898635477582846, "grad_norm": 9.6875, "learning_rate": 1.2595419847328243e-06, "loss": 2.0313, "step": 34 }, { "epoch": 0.004013301226923518, "grad_norm": 9.4375, "learning_rate": 1.297709923664122e-06, "loss": 1.6543, "step": 35 }, { "epoch": 0.0041279669762641896, "grad_norm": 13.1875, "learning_rate": 1.33587786259542e-06, "loss": 1.6057, "step": 36 }, { "epoch": 0.0042426327256048614, "grad_norm": 11.625, "learning_rate": 1.3740458015267178e-06, "loss": 2.0333, "step": 37 }, { "epoch": 0.004357298474945534, "grad_norm": 9.375, "learning_rate": 1.4122137404580156e-06, "loss": 1.5231, "step": 38 }, { "epoch": 0.004471964224286206, "grad_norm": 12.0, "learning_rate": 1.450381679389313e-06, "loss": 1.5789, "step": 39 }, { "epoch": 0.004586629973626878, "grad_norm": 14.0, "learning_rate": 1.4885496183206109e-06, "loss": 1.7879, "step": 40 }, { "epoch": 0.00470129572296755, "grad_norm": 12.6875, "learning_rate": 1.5267175572519086e-06, "loss": 1.8417, "step": 41 }, { "epoch": 0.004815961472308222, "grad_norm": 10.1875, "learning_rate": 1.5648854961832064e-06, "loss": 1.5501, "step": 42 }, { "epoch": 0.004930627221648894, "grad_norm": 11.3125, "learning_rate": 1.603053435114504e-06, "loss": 2.1175, "step": 43 }, { "epoch": 0.0050452929709895655, "grad_norm": 12.9375, "learning_rate": 1.6412213740458017e-06, "loss": 2.0739, "step": 44 }, { "epoch": 0.005159958720330237, "grad_norm": 9.0, "learning_rate": 1.6793893129770995e-06, "loss": 2.1387, "step": 45 }, { "epoch": 0.005274624469670909, "grad_norm": 8.5, "learning_rate": 1.7175572519083972e-06, "loss": 1.6499, "step": 46 }, { "epoch": 0.005389290219011581, "grad_norm": 10.125, "learning_rate": 1.7557251908396948e-06, "loss": 1.8054, "step": 47 }, { "epoch": 0.005503955968352253, "grad_norm": 11.4375, "learning_rate": 1.7938931297709925e-06, "loss": 1.8223, "step": 48 }, { "epoch": 0.005618621717692925, "grad_norm": 12.625, "learning_rate": 1.8320610687022903e-06, "loss": 1.6984, "step": 49 }, { "epoch": 0.005733287467033597, "grad_norm": 11.9375, "learning_rate": 1.870229007633588e-06, "loss": 2.2691, "step": 50 }, { "epoch": 0.005847953216374269, "grad_norm": 9.375, "learning_rate": 1.908396946564886e-06, "loss": 2.0387, "step": 51 }, { "epoch": 0.0059626189657149406, "grad_norm": 11.8125, "learning_rate": 1.946564885496183e-06, "loss": 2.1778, "step": 52 }, { "epoch": 0.006077284715055613, "grad_norm": 11.6875, "learning_rate": 1.984732824427481e-06, "loss": 1.9465, "step": 53 }, { "epoch": 0.006191950464396285, "grad_norm": 10.9375, "learning_rate": 2.0229007633587786e-06, "loss": 1.9139, "step": 54 }, { "epoch": 0.006306616213736957, "grad_norm": 8.125, "learning_rate": 2.0610687022900764e-06, "loss": 1.8812, "step": 55 }, { "epoch": 0.006421281963077629, "grad_norm": 11.4375, "learning_rate": 2.099236641221374e-06, "loss": 1.4957, "step": 56 }, { "epoch": 0.006535947712418301, "grad_norm": 12.25, "learning_rate": 2.137404580152672e-06, "loss": 2.0828, "step": 57 }, { "epoch": 0.006650613461758973, "grad_norm": 10.0, "learning_rate": 2.1755725190839697e-06, "loss": 2.0118, "step": 58 }, { "epoch": 0.006765279211099645, "grad_norm": 10.375, "learning_rate": 2.2137404580152674e-06, "loss": 2.0232, "step": 59 }, { "epoch": 0.0068799449604403165, "grad_norm": 10.3125, "learning_rate": 2.2519083969465648e-06, "loss": 1.5616, "step": 60 }, { "epoch": 0.006994610709780988, "grad_norm": 9.875, "learning_rate": 2.2900763358778625e-06, "loss": 1.9874, "step": 61 }, { "epoch": 0.00710927645912166, "grad_norm": 8.6875, "learning_rate": 2.3282442748091603e-06, "loss": 1.6177, "step": 62 }, { "epoch": 0.007223942208462332, "grad_norm": 9.75, "learning_rate": 2.3664122137404585e-06, "loss": 1.9825, "step": 63 }, { "epoch": 0.007338607957803004, "grad_norm": 9.875, "learning_rate": 2.4045801526717562e-06, "loss": 2.1963, "step": 64 }, { "epoch": 0.007453273707143676, "grad_norm": 9.125, "learning_rate": 2.4427480916030536e-06, "loss": 1.7033, "step": 65 }, { "epoch": 0.007567939456484348, "grad_norm": 10.125, "learning_rate": 2.4809160305343513e-06, "loss": 1.8166, "step": 66 }, { "epoch": 0.00768260520582502, "grad_norm": 14.0, "learning_rate": 2.5190839694656487e-06, "loss": 2.4252, "step": 67 }, { "epoch": 0.007797270955165692, "grad_norm": 11.5, "learning_rate": 2.5572519083969464e-06, "loss": 2.0804, "step": 68 }, { "epoch": 0.007911936704506364, "grad_norm": 9.9375, "learning_rate": 2.595419847328244e-06, "loss": 1.9219, "step": 69 }, { "epoch": 0.008026602453847035, "grad_norm": 9.5, "learning_rate": 2.633587786259542e-06, "loss": 1.851, "step": 70 }, { "epoch": 0.008141268203187708, "grad_norm": 10.5, "learning_rate": 2.67175572519084e-06, "loss": 1.8046, "step": 71 }, { "epoch": 0.008255933952528379, "grad_norm": 8.9375, "learning_rate": 2.709923664122138e-06, "loss": 1.9477, "step": 72 }, { "epoch": 0.008370599701869052, "grad_norm": 10.1875, "learning_rate": 2.7480916030534356e-06, "loss": 1.6877, "step": 73 }, { "epoch": 0.008485265451209723, "grad_norm": 10.1875, "learning_rate": 2.7862595419847334e-06, "loss": 1.7432, "step": 74 }, { "epoch": 0.008599931200550396, "grad_norm": 11.1875, "learning_rate": 2.824427480916031e-06, "loss": 2.15, "step": 75 }, { "epoch": 0.008714596949891068, "grad_norm": 8.625, "learning_rate": 2.862595419847328e-06, "loss": 1.8203, "step": 76 }, { "epoch": 0.00882926269923174, "grad_norm": 9.375, "learning_rate": 2.900763358778626e-06, "loss": 1.9033, "step": 77 }, { "epoch": 0.008943928448572412, "grad_norm": 11.5, "learning_rate": 2.938931297709924e-06, "loss": 1.6601, "step": 78 }, { "epoch": 0.009058594197913083, "grad_norm": 9.875, "learning_rate": 2.9770992366412218e-06, "loss": 1.4963, "step": 79 }, { "epoch": 0.009173259947253756, "grad_norm": 10.3125, "learning_rate": 3.0152671755725195e-06, "loss": 1.9955, "step": 80 }, { "epoch": 0.009287925696594427, "grad_norm": 9.9375, "learning_rate": 3.0534351145038173e-06, "loss": 1.908, "step": 81 }, { "epoch": 0.0094025914459351, "grad_norm": 10.375, "learning_rate": 3.091603053435115e-06, "loss": 2.023, "step": 82 }, { "epoch": 0.00951725719527577, "grad_norm": 10.625, "learning_rate": 3.129770992366413e-06, "loss": 1.9262, "step": 83 }, { "epoch": 0.009631922944616443, "grad_norm": 9.4375, "learning_rate": 3.1679389312977097e-06, "loss": 1.7799, "step": 84 }, { "epoch": 0.009746588693957114, "grad_norm": 10.625, "learning_rate": 3.206106870229008e-06, "loss": 1.8976, "step": 85 }, { "epoch": 0.009861254443297787, "grad_norm": 11.3125, "learning_rate": 3.2442748091603056e-06, "loss": 1.5031, "step": 86 }, { "epoch": 0.009975920192638458, "grad_norm": 12.1875, "learning_rate": 3.2824427480916034e-06, "loss": 1.6742, "step": 87 }, { "epoch": 0.010090585941979131, "grad_norm": 11.75, "learning_rate": 3.320610687022901e-06, "loss": 1.7465, "step": 88 }, { "epoch": 0.010205251691319802, "grad_norm": 9.4375, "learning_rate": 3.358778625954199e-06, "loss": 1.7872, "step": 89 }, { "epoch": 0.010319917440660475, "grad_norm": 13.6875, "learning_rate": 3.3969465648854967e-06, "loss": 2.0216, "step": 90 }, { "epoch": 0.010434583190001148, "grad_norm": 9.375, "learning_rate": 3.4351145038167944e-06, "loss": 1.8331, "step": 91 }, { "epoch": 0.010549248939341819, "grad_norm": 9.0, "learning_rate": 3.473282442748092e-06, "loss": 1.951, "step": 92 }, { "epoch": 0.010663914688682491, "grad_norm": 11.875, "learning_rate": 3.5114503816793895e-06, "loss": 1.866, "step": 93 }, { "epoch": 0.010778580438023162, "grad_norm": 13.4375, "learning_rate": 3.5496183206106873e-06, "loss": 1.9245, "step": 94 }, { "epoch": 0.010893246187363835, "grad_norm": 11.5, "learning_rate": 3.587786259541985e-06, "loss": 1.9593, "step": 95 }, { "epoch": 0.011007911936704506, "grad_norm": 10.125, "learning_rate": 3.625954198473283e-06, "loss": 2.0351, "step": 96 }, { "epoch": 0.011122577686045179, "grad_norm": 13.3125, "learning_rate": 3.6641221374045806e-06, "loss": 1.8833, "step": 97 }, { "epoch": 0.01123724343538585, "grad_norm": 9.0625, "learning_rate": 3.7022900763358783e-06, "loss": 1.7671, "step": 98 }, { "epoch": 0.011351909184726523, "grad_norm": 8.3125, "learning_rate": 3.740458015267176e-06, "loss": 1.7366, "step": 99 }, { "epoch": 0.011466574934067194, "grad_norm": 12.75, "learning_rate": 3.778625954198474e-06, "loss": 2.2246, "step": 100 }, { "epoch": 0.011581240683407866, "grad_norm": 10.8125, "learning_rate": 3.816793893129772e-06, "loss": 1.6931, "step": 101 }, { "epoch": 0.011695906432748537, "grad_norm": 12.125, "learning_rate": 3.8549618320610685e-06, "loss": 1.8878, "step": 102 }, { "epoch": 0.01181057218208921, "grad_norm": 11.1875, "learning_rate": 3.893129770992366e-06, "loss": 1.9914, "step": 103 }, { "epoch": 0.011925237931429881, "grad_norm": 8.125, "learning_rate": 3.931297709923664e-06, "loss": 1.643, "step": 104 }, { "epoch": 0.012039903680770554, "grad_norm": 10.5, "learning_rate": 3.969465648854962e-06, "loss": 2.2117, "step": 105 }, { "epoch": 0.012154569430111227, "grad_norm": 10.25, "learning_rate": 4.0076335877862595e-06, "loss": 1.8666, "step": 106 }, { "epoch": 0.012269235179451898, "grad_norm": 8.3125, "learning_rate": 4.045801526717557e-06, "loss": 1.7322, "step": 107 }, { "epoch": 0.01238390092879257, "grad_norm": 8.625, "learning_rate": 4.083969465648855e-06, "loss": 1.9029, "step": 108 }, { "epoch": 0.012498566678133241, "grad_norm": 11.3125, "learning_rate": 4.122137404580153e-06, "loss": 1.7935, "step": 109 }, { "epoch": 0.012613232427473914, "grad_norm": 10.8125, "learning_rate": 4.1603053435114506e-06, "loss": 1.6538, "step": 110 }, { "epoch": 0.012727898176814585, "grad_norm": 12.875, "learning_rate": 4.198473282442748e-06, "loss": 1.3825, "step": 111 }, { "epoch": 0.012842563926155258, "grad_norm": 12.25, "learning_rate": 4.236641221374046e-06, "loss": 2.4639, "step": 112 }, { "epoch": 0.012957229675495929, "grad_norm": 24.75, "learning_rate": 4.274809160305344e-06, "loss": 1.876, "step": 113 }, { "epoch": 0.013071895424836602, "grad_norm": 12.125, "learning_rate": 4.312977099236642e-06, "loss": 1.8466, "step": 114 }, { "epoch": 0.013186561174177273, "grad_norm": 13.4375, "learning_rate": 4.351145038167939e-06, "loss": 2.1258, "step": 115 }, { "epoch": 0.013301226923517945, "grad_norm": 10.5, "learning_rate": 4.389312977099237e-06, "loss": 1.9982, "step": 116 }, { "epoch": 0.013415892672858616, "grad_norm": 9.375, "learning_rate": 4.427480916030535e-06, "loss": 2.086, "step": 117 }, { "epoch": 0.01353055842219929, "grad_norm": 11.125, "learning_rate": 4.465648854961833e-06, "loss": 1.9286, "step": 118 }, { "epoch": 0.01364522417153996, "grad_norm": 9.6875, "learning_rate": 4.5038167938931296e-06, "loss": 1.5366, "step": 119 }, { "epoch": 0.013759889920880633, "grad_norm": 10.6875, "learning_rate": 4.541984732824427e-06, "loss": 1.8931, "step": 120 }, { "epoch": 0.013874555670221306, "grad_norm": 11.0625, "learning_rate": 4.580152671755725e-06, "loss": 2.0263, "step": 121 }, { "epoch": 0.013989221419561977, "grad_norm": 12.0625, "learning_rate": 4.618320610687023e-06, "loss": 1.5788, "step": 122 }, { "epoch": 0.01410388716890265, "grad_norm": 11.3125, "learning_rate": 4.656488549618321e-06, "loss": 2.0352, "step": 123 }, { "epoch": 0.01421855291824332, "grad_norm": 14.0625, "learning_rate": 4.694656488549618e-06, "loss": 1.9577, "step": 124 }, { "epoch": 0.014333218667583993, "grad_norm": 9.3125, "learning_rate": 4.732824427480917e-06, "loss": 1.9097, "step": 125 }, { "epoch": 0.014447884416924664, "grad_norm": 26.375, "learning_rate": 4.770992366412215e-06, "loss": 2.0753, "step": 126 }, { "epoch": 0.014562550166265337, "grad_norm": 22.375, "learning_rate": 4.8091603053435125e-06, "loss": 1.8644, "step": 127 }, { "epoch": 0.014677215915606008, "grad_norm": 11.0, "learning_rate": 4.847328244274809e-06, "loss": 2.0931, "step": 128 }, { "epoch": 0.01479188166494668, "grad_norm": 8.6875, "learning_rate": 4.885496183206107e-06, "loss": 1.8406, "step": 129 }, { "epoch": 0.014906547414287352, "grad_norm": 9.6875, "learning_rate": 4.923664122137405e-06, "loss": 2.0837, "step": 130 }, { "epoch": 0.015021213163628025, "grad_norm": 9.0625, "learning_rate": 4.961832061068703e-06, "loss": 1.9349, "step": 131 }, { "epoch": 0.015135878912968696, "grad_norm": 11.75, "learning_rate": 5e-06, "loss": 1.8023, "step": 132 }, { "epoch": 0.015250544662309368, "grad_norm": 9.9375, "learning_rate": 5.038167938931297e-06, "loss": 1.675, "step": 133 }, { "epoch": 0.01536521041165004, "grad_norm": 9.875, "learning_rate": 5.076335877862596e-06, "loss": 1.9818, "step": 134 }, { "epoch": 0.015479876160990712, "grad_norm": 10.125, "learning_rate": 5.114503816793893e-06, "loss": 1.9492, "step": 135 }, { "epoch": 0.015594541910331383, "grad_norm": 9.625, "learning_rate": 5.1526717557251914e-06, "loss": 1.5018, "step": 136 }, { "epoch": 0.015709207659672056, "grad_norm": 9.875, "learning_rate": 5.190839694656488e-06, "loss": 1.9816, "step": 137 }, { "epoch": 0.01582387340901273, "grad_norm": 10.125, "learning_rate": 5.229007633587787e-06, "loss": 2.1126, "step": 138 }, { "epoch": 0.0159385391583534, "grad_norm": 8.5625, "learning_rate": 5.267175572519084e-06, "loss": 1.385, "step": 139 }, { "epoch": 0.01605320490769407, "grad_norm": 11.3125, "learning_rate": 5.3053435114503825e-06, "loss": 1.8642, "step": 140 }, { "epoch": 0.016167870657034743, "grad_norm": 10.6875, "learning_rate": 5.34351145038168e-06, "loss": 1.7833, "step": 141 }, { "epoch": 0.016282536406375416, "grad_norm": 10.1875, "learning_rate": 5.381679389312977e-06, "loss": 2.0467, "step": 142 }, { "epoch": 0.01639720215571609, "grad_norm": 9.625, "learning_rate": 5.419847328244276e-06, "loss": 1.8006, "step": 143 }, { "epoch": 0.016511867905056758, "grad_norm": 14.75, "learning_rate": 5.458015267175573e-06, "loss": 2.1622, "step": 144 }, { "epoch": 0.01662653365439743, "grad_norm": 8.875, "learning_rate": 5.496183206106871e-06, "loss": 2.0246, "step": 145 }, { "epoch": 0.016741199403738104, "grad_norm": 11.0, "learning_rate": 5.534351145038168e-06, "loss": 2.4745, "step": 146 }, { "epoch": 0.016855865153078776, "grad_norm": 57.25, "learning_rate": 5.572519083969467e-06, "loss": 2.113, "step": 147 }, { "epoch": 0.016970530902419446, "grad_norm": 27.375, "learning_rate": 5.610687022900764e-06, "loss": 1.96, "step": 148 }, { "epoch": 0.01708519665176012, "grad_norm": 9.875, "learning_rate": 5.648854961832062e-06, "loss": 1.9365, "step": 149 }, { "epoch": 0.01719986240110079, "grad_norm": 15.1875, "learning_rate": 5.687022900763359e-06, "loss": 1.7366, "step": 150 }, { "epoch": 0.017314528150441464, "grad_norm": 9.75, "learning_rate": 5.725190839694656e-06, "loss": 2.2346, "step": 151 }, { "epoch": 0.017429193899782137, "grad_norm": 11.125, "learning_rate": 5.763358778625955e-06, "loss": 2.1116, "step": 152 }, { "epoch": 0.017543859649122806, "grad_norm": 10.8125, "learning_rate": 5.801526717557252e-06, "loss": 1.6447, "step": 153 }, { "epoch": 0.01765852539846348, "grad_norm": 10.75, "learning_rate": 5.83969465648855e-06, "loss": 2.168, "step": 154 }, { "epoch": 0.01777319114780415, "grad_norm": 45.25, "learning_rate": 5.877862595419848e-06, "loss": 2.2209, "step": 155 }, { "epoch": 0.017887856897144824, "grad_norm": 9.75, "learning_rate": 5.916030534351146e-06, "loss": 1.835, "step": 156 }, { "epoch": 0.018002522646485494, "grad_norm": 11.0625, "learning_rate": 5.9541984732824435e-06, "loss": 1.7988, "step": 157 }, { "epoch": 0.018117188395826166, "grad_norm": 12.875, "learning_rate": 5.992366412213741e-06, "loss": 1.8944, "step": 158 }, { "epoch": 0.01823185414516684, "grad_norm": 9.375, "learning_rate": 6.030534351145039e-06, "loss": 1.8319, "step": 159 }, { "epoch": 0.018346519894507512, "grad_norm": 12.5, "learning_rate": 6.068702290076336e-06, "loss": 2.1422, "step": 160 }, { "epoch": 0.01846118564384818, "grad_norm": 14.3125, "learning_rate": 6.1068702290076346e-06, "loss": 1.5438, "step": 161 }, { "epoch": 0.018575851393188854, "grad_norm": 11.4375, "learning_rate": 6.1450381679389315e-06, "loss": 1.6895, "step": 162 }, { "epoch": 0.018690517142529527, "grad_norm": 10.75, "learning_rate": 6.18320610687023e-06, "loss": 1.5286, "step": 163 }, { "epoch": 0.0188051828918702, "grad_norm": 30.5, "learning_rate": 6.221374045801527e-06, "loss": 2.5311, "step": 164 }, { "epoch": 0.01891984864121087, "grad_norm": 9.6875, "learning_rate": 6.259541984732826e-06, "loss": 2.0003, "step": 165 }, { "epoch": 0.01903451439055154, "grad_norm": 14.75, "learning_rate": 6.2977099236641225e-06, "loss": 1.7126, "step": 166 }, { "epoch": 0.019149180139892214, "grad_norm": 10.8125, "learning_rate": 6.335877862595419e-06, "loss": 1.8691, "step": 167 }, { "epoch": 0.019263845889232887, "grad_norm": 9.0, "learning_rate": 6.374045801526718e-06, "loss": 1.7202, "step": 168 }, { "epoch": 0.01937851163857356, "grad_norm": 11.3125, "learning_rate": 6.412213740458016e-06, "loss": 1.779, "step": 169 }, { "epoch": 0.01949317738791423, "grad_norm": 14.3125, "learning_rate": 6.4503816793893135e-06, "loss": 2.173, "step": 170 }, { "epoch": 0.0196078431372549, "grad_norm": 34.75, "learning_rate": 6.488549618320611e-06, "loss": 1.7785, "step": 171 }, { "epoch": 0.019722508886595574, "grad_norm": 11.4375, "learning_rate": 6.526717557251909e-06, "loss": 2.1838, "step": 172 }, { "epoch": 0.019837174635936247, "grad_norm": 10.125, "learning_rate": 6.564885496183207e-06, "loss": 1.8227, "step": 173 }, { "epoch": 0.019951840385276916, "grad_norm": 10.4375, "learning_rate": 6.6030534351145046e-06, "loss": 1.7261, "step": 174 }, { "epoch": 0.02006650613461759, "grad_norm": 14.375, "learning_rate": 6.641221374045802e-06, "loss": 2.3964, "step": 175 }, { "epoch": 0.020181171883958262, "grad_norm": 13.4375, "learning_rate": 6.679389312977099e-06, "loss": 2.0484, "step": 176 }, { "epoch": 0.020295837633298935, "grad_norm": 11.5625, "learning_rate": 6.717557251908398e-06, "loss": 1.6068, "step": 177 }, { "epoch": 0.020410503382639604, "grad_norm": 8.3125, "learning_rate": 6.755725190839695e-06, "loss": 1.5096, "step": 178 }, { "epoch": 0.020525169131980277, "grad_norm": 11.4375, "learning_rate": 6.793893129770993e-06, "loss": 1.6544, "step": 179 }, { "epoch": 0.02063983488132095, "grad_norm": 10.75, "learning_rate": 6.83206106870229e-06, "loss": 1.9542, "step": 180 }, { "epoch": 0.020754500630661622, "grad_norm": 10.5, "learning_rate": 6.870229007633589e-06, "loss": 2.2447, "step": 181 }, { "epoch": 0.020869166380002295, "grad_norm": 9.1875, "learning_rate": 6.908396946564886e-06, "loss": 1.8911, "step": 182 }, { "epoch": 0.020983832129342964, "grad_norm": 9.9375, "learning_rate": 6.946564885496184e-06, "loss": 2.037, "step": 183 }, { "epoch": 0.021098497878683637, "grad_norm": 8.8125, "learning_rate": 6.984732824427481e-06, "loss": 1.5327, "step": 184 }, { "epoch": 0.02121316362802431, "grad_norm": 12.625, "learning_rate": 7.022900763358779e-06, "loss": 1.5419, "step": 185 }, { "epoch": 0.021327829377364983, "grad_norm": 11.125, "learning_rate": 7.061068702290077e-06, "loss": 1.6126, "step": 186 }, { "epoch": 0.021442495126705652, "grad_norm": 13.3125, "learning_rate": 7.0992366412213746e-06, "loss": 2.3405, "step": 187 }, { "epoch": 0.021557160876046325, "grad_norm": 12.0, "learning_rate": 7.137404580152672e-06, "loss": 1.7103, "step": 188 }, { "epoch": 0.021671826625386997, "grad_norm": 12.1875, "learning_rate": 7.17557251908397e-06, "loss": 2.0406, "step": 189 }, { "epoch": 0.02178649237472767, "grad_norm": 9.75, "learning_rate": 7.213740458015268e-06, "loss": 1.9133, "step": 190 }, { "epoch": 0.02190115812406834, "grad_norm": 10.3125, "learning_rate": 7.251908396946566e-06, "loss": 2.1114, "step": 191 }, { "epoch": 0.022015823873409012, "grad_norm": 10.5625, "learning_rate": 7.290076335877863e-06, "loss": 1.7909, "step": 192 }, { "epoch": 0.022130489622749685, "grad_norm": 11.125, "learning_rate": 7.328244274809161e-06, "loss": 1.9307, "step": 193 }, { "epoch": 0.022245155372090358, "grad_norm": 11.3125, "learning_rate": 7.366412213740458e-06, "loss": 2.0263, "step": 194 }, { "epoch": 0.022359821121431027, "grad_norm": 9.25, "learning_rate": 7.404580152671757e-06, "loss": 1.6658, "step": 195 }, { "epoch": 0.0224744868707717, "grad_norm": 10.875, "learning_rate": 7.4427480916030536e-06, "loss": 2.0997, "step": 196 }, { "epoch": 0.022589152620112372, "grad_norm": 9.5, "learning_rate": 7.480916030534352e-06, "loss": 1.5645, "step": 197 }, { "epoch": 0.022703818369453045, "grad_norm": 16.75, "learning_rate": 7.519083969465649e-06, "loss": 1.7767, "step": 198 }, { "epoch": 0.022818484118793718, "grad_norm": 9.625, "learning_rate": 7.557251908396948e-06, "loss": 1.8431, "step": 199 }, { "epoch": 0.022933149868134387, "grad_norm": 10.125, "learning_rate": 7.595419847328245e-06, "loss": 1.7448, "step": 200 }, { "epoch": 0.02304781561747506, "grad_norm": 9.125, "learning_rate": 7.633587786259543e-06, "loss": 1.8262, "step": 201 }, { "epoch": 0.023162481366815733, "grad_norm": 10.5, "learning_rate": 7.671755725190841e-06, "loss": 1.7657, "step": 202 }, { "epoch": 0.023277147116156405, "grad_norm": 12.1875, "learning_rate": 7.709923664122137e-06, "loss": 2.226, "step": 203 }, { "epoch": 0.023391812865497075, "grad_norm": 12.0625, "learning_rate": 7.748091603053436e-06, "loss": 2.2203, "step": 204 }, { "epoch": 0.023506478614837747, "grad_norm": 10.6875, "learning_rate": 7.786259541984733e-06, "loss": 1.9767, "step": 205 }, { "epoch": 0.02362114436417842, "grad_norm": 23.375, "learning_rate": 7.824427480916032e-06, "loss": 1.7781, "step": 206 }, { "epoch": 0.023735810113519093, "grad_norm": 10.125, "learning_rate": 7.862595419847328e-06, "loss": 1.7428, "step": 207 }, { "epoch": 0.023850475862859762, "grad_norm": 11.9375, "learning_rate": 7.900763358778627e-06, "loss": 1.963, "step": 208 }, { "epoch": 0.023965141612200435, "grad_norm": 10.875, "learning_rate": 7.938931297709924e-06, "loss": 1.61, "step": 209 }, { "epoch": 0.024079807361541108, "grad_norm": 9.75, "learning_rate": 7.977099236641223e-06, "loss": 1.94, "step": 210 }, { "epoch": 0.02419447311088178, "grad_norm": 9.0, "learning_rate": 8.015267175572519e-06, "loss": 1.8715, "step": 211 }, { "epoch": 0.024309138860222453, "grad_norm": 11.9375, "learning_rate": 8.053435114503817e-06, "loss": 1.7847, "step": 212 }, { "epoch": 0.024423804609563123, "grad_norm": 12.3125, "learning_rate": 8.091603053435115e-06, "loss": 2.2747, "step": 213 }, { "epoch": 0.024538470358903795, "grad_norm": 12.5, "learning_rate": 8.129770992366412e-06, "loss": 2.0973, "step": 214 }, { "epoch": 0.024653136108244468, "grad_norm": 13.25, "learning_rate": 8.16793893129771e-06, "loss": 1.8651, "step": 215 }, { "epoch": 0.02476780185758514, "grad_norm": 12.125, "learning_rate": 8.206106870229008e-06, "loss": 1.7578, "step": 216 }, { "epoch": 0.02488246760692581, "grad_norm": 11.625, "learning_rate": 8.244274809160306e-06, "loss": 1.6948, "step": 217 }, { "epoch": 0.024997133356266483, "grad_norm": 8.25, "learning_rate": 8.282442748091603e-06, "loss": 1.892, "step": 218 }, { "epoch": 0.025111799105607156, "grad_norm": 8.6875, "learning_rate": 8.320610687022901e-06, "loss": 2.007, "step": 219 }, { "epoch": 0.02522646485494783, "grad_norm": 14.6875, "learning_rate": 8.358778625954199e-06, "loss": 1.9615, "step": 220 }, { "epoch": 0.025341130604288498, "grad_norm": 25.625, "learning_rate": 8.396946564885497e-06, "loss": 1.8476, "step": 221 }, { "epoch": 0.02545579635362917, "grad_norm": 11.0, "learning_rate": 8.435114503816794e-06, "loss": 1.9152, "step": 222 }, { "epoch": 0.025570462102969843, "grad_norm": 12.25, "learning_rate": 8.473282442748092e-06, "loss": 1.6968, "step": 223 }, { "epoch": 0.025685127852310516, "grad_norm": 11.5625, "learning_rate": 8.51145038167939e-06, "loss": 1.8369, "step": 224 }, { "epoch": 0.025799793601651185, "grad_norm": 11.4375, "learning_rate": 8.549618320610688e-06, "loss": 2.4039, "step": 225 }, { "epoch": 0.025914459350991858, "grad_norm": 9.5625, "learning_rate": 8.587786259541985e-06, "loss": 1.7356, "step": 226 }, { "epoch": 0.02602912510033253, "grad_norm": 12.9375, "learning_rate": 8.625954198473283e-06, "loss": 2.0797, "step": 227 }, { "epoch": 0.026143790849673203, "grad_norm": 9.5625, "learning_rate": 8.664122137404581e-06, "loss": 1.9736, "step": 228 }, { "epoch": 0.026258456599013876, "grad_norm": 28.75, "learning_rate": 8.702290076335879e-06, "loss": 2.1437, "step": 229 }, { "epoch": 0.026373122348354545, "grad_norm": 10.5, "learning_rate": 8.740458015267176e-06, "loss": 1.9136, "step": 230 }, { "epoch": 0.026487788097695218, "grad_norm": 10.8125, "learning_rate": 8.778625954198474e-06, "loss": 1.8982, "step": 231 }, { "epoch": 0.02660245384703589, "grad_norm": 9.375, "learning_rate": 8.816793893129772e-06, "loss": 1.8118, "step": 232 }, { "epoch": 0.026717119596376564, "grad_norm": 9.25, "learning_rate": 8.85496183206107e-06, "loss": 1.8382, "step": 233 }, { "epoch": 0.026831785345717233, "grad_norm": 36.75, "learning_rate": 8.893129770992368e-06, "loss": 2.3755, "step": 234 }, { "epoch": 0.026946451095057906, "grad_norm": 11.0, "learning_rate": 8.931297709923665e-06, "loss": 1.9812, "step": 235 }, { "epoch": 0.02706111684439858, "grad_norm": 21.125, "learning_rate": 8.969465648854963e-06, "loss": 2.1709, "step": 236 }, { "epoch": 0.02717578259373925, "grad_norm": 11.0, "learning_rate": 9.007633587786259e-06, "loss": 1.756, "step": 237 }, { "epoch": 0.02729044834307992, "grad_norm": 10.4375, "learning_rate": 9.045801526717559e-06, "loss": 1.9527, "step": 238 }, { "epoch": 0.027405114092420593, "grad_norm": 22.0, "learning_rate": 9.083969465648855e-06, "loss": 2.0213, "step": 239 }, { "epoch": 0.027519779841761266, "grad_norm": 9.4375, "learning_rate": 9.122137404580154e-06, "loss": 2.1283, "step": 240 }, { "epoch": 0.02763444559110194, "grad_norm": 11.75, "learning_rate": 9.16030534351145e-06, "loss": 1.8396, "step": 241 }, { "epoch": 0.02774911134044261, "grad_norm": 9.8125, "learning_rate": 9.19847328244275e-06, "loss": 2.1474, "step": 242 }, { "epoch": 0.02786377708978328, "grad_norm": 10.3125, "learning_rate": 9.236641221374046e-06, "loss": 1.9196, "step": 243 }, { "epoch": 0.027978442839123954, "grad_norm": 26.75, "learning_rate": 9.274809160305345e-06, "loss": 1.8008, "step": 244 }, { "epoch": 0.028093108588464626, "grad_norm": 10.375, "learning_rate": 9.312977099236641e-06, "loss": 1.6378, "step": 245 }, { "epoch": 0.0282077743378053, "grad_norm": 12.0, "learning_rate": 9.351145038167939e-06, "loss": 1.9045, "step": 246 }, { "epoch": 0.02832244008714597, "grad_norm": 10.1875, "learning_rate": 9.389312977099237e-06, "loss": 2.0439, "step": 247 }, { "epoch": 0.02843710583648664, "grad_norm": 11.5, "learning_rate": 9.427480916030534e-06, "loss": 1.7712, "step": 248 }, { "epoch": 0.028551771585827314, "grad_norm": 9.0, "learning_rate": 9.465648854961834e-06, "loss": 1.8359, "step": 249 }, { "epoch": 0.028666437335167987, "grad_norm": 12.8125, "learning_rate": 9.50381679389313e-06, "loss": 1.5174, "step": 250 }, { "epoch": 0.028781103084508656, "grad_norm": 10.25, "learning_rate": 9.54198473282443e-06, "loss": 2.034, "step": 251 }, { "epoch": 0.02889576883384933, "grad_norm": 11.0, "learning_rate": 9.580152671755725e-06, "loss": 1.6431, "step": 252 }, { "epoch": 0.02901043458319, "grad_norm": 10.9375, "learning_rate": 9.618320610687025e-06, "loss": 1.7994, "step": 253 }, { "epoch": 0.029125100332530674, "grad_norm": 11.4375, "learning_rate": 9.656488549618321e-06, "loss": 1.8513, "step": 254 }, { "epoch": 0.029239766081871343, "grad_norm": 9.25, "learning_rate": 9.694656488549619e-06, "loss": 1.9269, "step": 255 }, { "epoch": 0.029354431831212016, "grad_norm": 10.5625, "learning_rate": 9.732824427480917e-06, "loss": 2.0572, "step": 256 }, { "epoch": 0.02946909758055269, "grad_norm": 12.0625, "learning_rate": 9.770992366412214e-06, "loss": 1.6371, "step": 257 }, { "epoch": 0.02958376332989336, "grad_norm": 12.75, "learning_rate": 9.809160305343512e-06, "loss": 2.068, "step": 258 }, { "epoch": 0.029698429079234034, "grad_norm": 11.4375, "learning_rate": 9.84732824427481e-06, "loss": 1.933, "step": 259 }, { "epoch": 0.029813094828574704, "grad_norm": 24.25, "learning_rate": 9.885496183206108e-06, "loss": 2.0847, "step": 260 }, { "epoch": 0.029927760577915376, "grad_norm": 10.3125, "learning_rate": 9.923664122137405e-06, "loss": 2.0812, "step": 261 }, { "epoch": 0.03004242632725605, "grad_norm": 12.3125, "learning_rate": 9.961832061068703e-06, "loss": 2.0512, "step": 262 }, { "epoch": 0.030157092076596722, "grad_norm": 11.0625, "learning_rate": 1e-05, "loss": 1.7728, "step": 263 }, { "epoch": 0.03027175782593739, "grad_norm": 10.4375, "learning_rate": 9.999999655172656e-06, "loss": 1.8603, "step": 264 }, { "epoch": 0.030386423575278064, "grad_norm": 11.125, "learning_rate": 9.999998620690665e-06, "loss": 1.8974, "step": 265 }, { "epoch": 0.030501089324618737, "grad_norm": 9.875, "learning_rate": 9.999996896554174e-06, "loss": 1.7763, "step": 266 }, { "epoch": 0.03061575507395941, "grad_norm": 10.875, "learning_rate": 9.999994482763422e-06, "loss": 1.7571, "step": 267 }, { "epoch": 0.03073042082330008, "grad_norm": 13.3125, "learning_rate": 9.999991379318738e-06, "loss": 1.6499, "step": 268 }, { "epoch": 0.03084508657264075, "grad_norm": 10.5, "learning_rate": 9.999987586220551e-06, "loss": 1.5847, "step": 269 }, { "epoch": 0.030959752321981424, "grad_norm": 13.5, "learning_rate": 9.999983103469387e-06, "loss": 1.9093, "step": 270 }, { "epoch": 0.031074418071322097, "grad_norm": 10.3125, "learning_rate": 9.999977931065859e-06, "loss": 2.2419, "step": 271 }, { "epoch": 0.031189083820662766, "grad_norm": 33.5, "learning_rate": 9.999972069010686e-06, "loss": 1.79, "step": 272 }, { "epoch": 0.03130374957000344, "grad_norm": 11.8125, "learning_rate": 9.999965517304673e-06, "loss": 2.0324, "step": 273 }, { "epoch": 0.03141841531934411, "grad_norm": 11.625, "learning_rate": 9.999958275948726e-06, "loss": 1.8671, "step": 274 }, { "epoch": 0.031533081068684785, "grad_norm": 10.25, "learning_rate": 9.999950344943843e-06, "loss": 1.8716, "step": 275 }, { "epoch": 0.03164774681802546, "grad_norm": 10.9375, "learning_rate": 9.999941724291115e-06, "loss": 1.5815, "step": 276 }, { "epoch": 0.03176241256736613, "grad_norm": 47.0, "learning_rate": 9.999932413991737e-06, "loss": 2.3216, "step": 277 }, { "epoch": 0.0318770783167068, "grad_norm": 9.5, "learning_rate": 9.999922414046988e-06, "loss": 1.5902, "step": 278 }, { "epoch": 0.03199174406604747, "grad_norm": 13.0, "learning_rate": 9.99991172445825e-06, "loss": 1.6091, "step": 279 }, { "epoch": 0.03210640981538814, "grad_norm": 10.5, "learning_rate": 9.999900345226996e-06, "loss": 1.9551, "step": 280 }, { "epoch": 0.032221075564728814, "grad_norm": 14.8125, "learning_rate": 9.999888276354797e-06, "loss": 1.655, "step": 281 }, { "epoch": 0.03233574131406949, "grad_norm": 9.1875, "learning_rate": 9.999875517843317e-06, "loss": 1.6096, "step": 282 }, { "epoch": 0.03245040706341016, "grad_norm": 9.5625, "learning_rate": 9.999862069694313e-06, "loss": 1.6491, "step": 283 }, { "epoch": 0.03256507281275083, "grad_norm": 12.25, "learning_rate": 9.999847931909645e-06, "loss": 1.9475, "step": 284 }, { "epoch": 0.032679738562091505, "grad_norm": 14.5, "learning_rate": 9.999833104491259e-06, "loss": 2.4504, "step": 285 }, { "epoch": 0.03279440431143218, "grad_norm": 10.875, "learning_rate": 9.999817587441203e-06, "loss": 2.1173, "step": 286 }, { "epoch": 0.032909070060772844, "grad_norm": 12.8125, "learning_rate": 9.999801380761616e-06, "loss": 2.2836, "step": 287 }, { "epoch": 0.033023735810113516, "grad_norm": 11.5, "learning_rate": 9.999784484454733e-06, "loss": 1.6493, "step": 288 }, { "epoch": 0.03313840155945419, "grad_norm": 9.625, "learning_rate": 9.999766898522885e-06, "loss": 2.0612, "step": 289 }, { "epoch": 0.03325306730879486, "grad_norm": 9.875, "learning_rate": 9.999748622968497e-06, "loss": 1.5319, "step": 290 }, { "epoch": 0.033367733058135535, "grad_norm": 11.0, "learning_rate": 9.99972965779409e-06, "loss": 1.7588, "step": 291 }, { "epoch": 0.03348239880747621, "grad_norm": 10.5625, "learning_rate": 9.999710003002282e-06, "loss": 1.9121, "step": 292 }, { "epoch": 0.03359706455681688, "grad_norm": 14.8125, "learning_rate": 9.99968965859578e-06, "loss": 2.1182, "step": 293 }, { "epoch": 0.03371173030615755, "grad_norm": 10.5625, "learning_rate": 9.999668624577395e-06, "loss": 1.9508, "step": 294 }, { "epoch": 0.033826396055498226, "grad_norm": 10.5, "learning_rate": 9.999646900950024e-06, "loss": 1.7532, "step": 295 }, { "epoch": 0.03394106180483889, "grad_norm": 10.875, "learning_rate": 9.999624487716666e-06, "loss": 1.8859, "step": 296 }, { "epoch": 0.034055727554179564, "grad_norm": 9.9375, "learning_rate": 9.99960138488041e-06, "loss": 1.6655, "step": 297 }, { "epoch": 0.03417039330352024, "grad_norm": 14.5, "learning_rate": 9.999577592444444e-06, "loss": 2.6149, "step": 298 }, { "epoch": 0.03428505905286091, "grad_norm": 9.625, "learning_rate": 9.99955311041205e-06, "loss": 1.8968, "step": 299 }, { "epoch": 0.03439972480220158, "grad_norm": 13.0, "learning_rate": 9.999527938786606e-06, "loss": 2.2929, "step": 300 }, { "epoch": 0.034514390551542255, "grad_norm": 13.8125, "learning_rate": 9.99950207757158e-06, "loss": 2.0613, "step": 301 }, { "epoch": 0.03462905630088293, "grad_norm": 10.8125, "learning_rate": 9.999475526770545e-06, "loss": 1.7524, "step": 302 }, { "epoch": 0.0347437220502236, "grad_norm": 14.1875, "learning_rate": 9.999448286387158e-06, "loss": 2.1728, "step": 303 }, { "epoch": 0.034858387799564274, "grad_norm": 9.125, "learning_rate": 9.999420356425178e-06, "loss": 2.0428, "step": 304 }, { "epoch": 0.03497305354890494, "grad_norm": 10.5, "learning_rate": 9.999391736888457e-06, "loss": 1.4197, "step": 305 }, { "epoch": 0.03508771929824561, "grad_norm": 9.125, "learning_rate": 9.999362427780943e-06, "loss": 1.601, "step": 306 }, { "epoch": 0.035202385047586285, "grad_norm": 12.375, "learning_rate": 9.99933242910668e-06, "loss": 1.901, "step": 307 }, { "epoch": 0.03531705079692696, "grad_norm": 11.25, "learning_rate": 9.999301740869802e-06, "loss": 1.8498, "step": 308 }, { "epoch": 0.03543171654626763, "grad_norm": 9.5625, "learning_rate": 9.999270363074548e-06, "loss": 1.7499, "step": 309 }, { "epoch": 0.0355463822956083, "grad_norm": 12.4375, "learning_rate": 9.999238295725238e-06, "loss": 1.6229, "step": 310 }, { "epoch": 0.035661048044948976, "grad_norm": 11.0, "learning_rate": 9.9992055388263e-06, "loss": 1.7539, "step": 311 }, { "epoch": 0.03577571379428965, "grad_norm": 14.375, "learning_rate": 9.999172092382254e-06, "loss": 1.9179, "step": 312 }, { "epoch": 0.035890379543630314, "grad_norm": 13.75, "learning_rate": 9.999137956397708e-06, "loss": 2.1377, "step": 313 }, { "epoch": 0.03600504529297099, "grad_norm": 11.4375, "learning_rate": 9.999103130877375e-06, "loss": 1.4768, "step": 314 }, { "epoch": 0.03611971104231166, "grad_norm": 9.875, "learning_rate": 9.999067615826055e-06, "loss": 1.7624, "step": 315 }, { "epoch": 0.03623437679165233, "grad_norm": 11.0625, "learning_rate": 9.99903141124865e-06, "loss": 1.7433, "step": 316 }, { "epoch": 0.036349042540993005, "grad_norm": 11.75, "learning_rate": 9.99899451715015e-06, "loss": 1.7933, "step": 317 }, { "epoch": 0.03646370829033368, "grad_norm": 28.625, "learning_rate": 9.998956933535648e-06, "loss": 1.8725, "step": 318 }, { "epoch": 0.03657837403967435, "grad_norm": 10.5, "learning_rate": 9.998918660410324e-06, "loss": 1.8906, "step": 319 }, { "epoch": 0.036693039789015024, "grad_norm": 12.1875, "learning_rate": 9.99887969777946e-06, "loss": 2.2342, "step": 320 }, { "epoch": 0.036807705538355696, "grad_norm": 13.0625, "learning_rate": 9.998840045648431e-06, "loss": 2.0835, "step": 321 }, { "epoch": 0.03692237128769636, "grad_norm": 15.25, "learning_rate": 9.998799704022702e-06, "loss": 2.1065, "step": 322 }, { "epoch": 0.037037037037037035, "grad_norm": 11.125, "learning_rate": 9.998758672907837e-06, "loss": 1.9103, "step": 323 }, { "epoch": 0.03715170278637771, "grad_norm": 10.4375, "learning_rate": 9.998716952309502e-06, "loss": 1.9575, "step": 324 }, { "epoch": 0.03726636853571838, "grad_norm": 13.625, "learning_rate": 9.998674542233446e-06, "loss": 2.0898, "step": 325 }, { "epoch": 0.03738103428505905, "grad_norm": 24.375, "learning_rate": 9.998631442685521e-06, "loss": 2.2672, "step": 326 }, { "epoch": 0.037495700034399726, "grad_norm": 11.0625, "learning_rate": 9.99858765367167e-06, "loss": 1.8253, "step": 327 }, { "epoch": 0.0376103657837404, "grad_norm": 11.0, "learning_rate": 9.998543175197935e-06, "loss": 1.6075, "step": 328 }, { "epoch": 0.03772503153308107, "grad_norm": 11.3125, "learning_rate": 9.99849800727045e-06, "loss": 2.5187, "step": 329 }, { "epoch": 0.03783969728242174, "grad_norm": 11.6875, "learning_rate": 9.998452149895444e-06, "loss": 1.7834, "step": 330 }, { "epoch": 0.03795436303176241, "grad_norm": 9.75, "learning_rate": 9.998405603079245e-06, "loss": 1.6472, "step": 331 }, { "epoch": 0.03806902878110308, "grad_norm": 30.0, "learning_rate": 9.99835836682827e-06, "loss": 2.1864, "step": 332 }, { "epoch": 0.038183694530443756, "grad_norm": 10.1875, "learning_rate": 9.998310441149034e-06, "loss": 1.7835, "step": 333 }, { "epoch": 0.03829836027978443, "grad_norm": 8.9375, "learning_rate": 9.998261826048152e-06, "loss": 1.7268, "step": 334 }, { "epoch": 0.0384130260291251, "grad_norm": 11.4375, "learning_rate": 9.998212521532324e-06, "loss": 2.0453, "step": 335 }, { "epoch": 0.038527691778465774, "grad_norm": 13.125, "learning_rate": 9.998162527608356e-06, "loss": 2.1078, "step": 336 }, { "epoch": 0.03864235752780645, "grad_norm": 10.9375, "learning_rate": 9.998111844283138e-06, "loss": 1.8703, "step": 337 }, { "epoch": 0.03875702327714712, "grad_norm": 8.3125, "learning_rate": 9.998060471563666e-06, "loss": 1.9393, "step": 338 }, { "epoch": 0.038871689026487785, "grad_norm": 13.0, "learning_rate": 9.998008409457023e-06, "loss": 2.3736, "step": 339 }, { "epoch": 0.03898635477582846, "grad_norm": 11.3125, "learning_rate": 9.99795565797039e-06, "loss": 2.0286, "step": 340 }, { "epoch": 0.03910102052516913, "grad_norm": 12.6875, "learning_rate": 9.997902217111045e-06, "loss": 2.17, "step": 341 }, { "epoch": 0.0392156862745098, "grad_norm": 10.375, "learning_rate": 9.997848086886357e-06, "loss": 1.9597, "step": 342 }, { "epoch": 0.039330352023850476, "grad_norm": 30.875, "learning_rate": 9.997793267303793e-06, "loss": 1.7293, "step": 343 }, { "epoch": 0.03944501777319115, "grad_norm": 11.3125, "learning_rate": 9.997737758370916e-06, "loss": 2.1488, "step": 344 }, { "epoch": 0.03955968352253182, "grad_norm": 10.875, "learning_rate": 9.997681560095378e-06, "loss": 1.8221, "step": 345 }, { "epoch": 0.039674349271872494, "grad_norm": 14.625, "learning_rate": 9.997624672484935e-06, "loss": 1.5912, "step": 346 }, { "epoch": 0.03978901502121316, "grad_norm": 11.0625, "learning_rate": 9.997567095547431e-06, "loss": 1.5531, "step": 347 }, { "epoch": 0.03990368077055383, "grad_norm": 12.6875, "learning_rate": 9.99750882929081e-06, "loss": 1.6229, "step": 348 }, { "epoch": 0.040018346519894506, "grad_norm": 11.3125, "learning_rate": 9.997449873723105e-06, "loss": 2.1418, "step": 349 }, { "epoch": 0.04013301226923518, "grad_norm": 11.75, "learning_rate": 9.99739022885245e-06, "loss": 2.3148, "step": 350 }, { "epoch": 0.04024767801857585, "grad_norm": 11.75, "learning_rate": 9.997329894687073e-06, "loss": 1.9163, "step": 351 }, { "epoch": 0.040362343767916524, "grad_norm": 11.1875, "learning_rate": 9.997268871235297e-06, "loss": 1.7382, "step": 352 }, { "epoch": 0.0404770095172572, "grad_norm": 11.4375, "learning_rate": 9.997207158505534e-06, "loss": 1.5633, "step": 353 }, { "epoch": 0.04059167526659787, "grad_norm": 10.5625, "learning_rate": 9.9971447565063e-06, "loss": 2.0312, "step": 354 }, { "epoch": 0.04070634101593854, "grad_norm": 11.3125, "learning_rate": 9.9970816652462e-06, "loss": 2.0914, "step": 355 }, { "epoch": 0.04082100676527921, "grad_norm": 13.5, "learning_rate": 9.997017884733937e-06, "loss": 2.0437, "step": 356 }, { "epoch": 0.04093567251461988, "grad_norm": 10.8125, "learning_rate": 9.99695341497831e-06, "loss": 1.9405, "step": 357 }, { "epoch": 0.041050338263960554, "grad_norm": 26.125, "learning_rate": 9.996888255988208e-06, "loss": 1.7734, "step": 358 }, { "epoch": 0.041165004013301226, "grad_norm": 9.125, "learning_rate": 9.996822407772624e-06, "loss": 1.6615, "step": 359 }, { "epoch": 0.0412796697626419, "grad_norm": 14.625, "learning_rate": 9.996755870340633e-06, "loss": 1.761, "step": 360 }, { "epoch": 0.04139433551198257, "grad_norm": 10.875, "learning_rate": 9.99668864370142e-06, "loss": 2.0534, "step": 361 }, { "epoch": 0.041509001261323245, "grad_norm": 11.75, "learning_rate": 9.996620727864253e-06, "loss": 1.9273, "step": 362 }, { "epoch": 0.04162366701066392, "grad_norm": 9.75, "learning_rate": 9.996552122838499e-06, "loss": 1.8703, "step": 363 }, { "epoch": 0.04173833276000459, "grad_norm": 12.25, "learning_rate": 9.996482828633625e-06, "loss": 1.5812, "step": 364 }, { "epoch": 0.041852998509345256, "grad_norm": 10.5625, "learning_rate": 9.996412845259185e-06, "loss": 1.817, "step": 365 }, { "epoch": 0.04196766425868593, "grad_norm": 13.1875, "learning_rate": 9.996342172724834e-06, "loss": 1.9002, "step": 366 }, { "epoch": 0.0420823300080266, "grad_norm": 12.625, "learning_rate": 9.99627081104032e-06, "loss": 1.9093, "step": 367 }, { "epoch": 0.042196995757367274, "grad_norm": 30.625, "learning_rate": 9.996198760215482e-06, "loss": 2.067, "step": 368 }, { "epoch": 0.04231166150670795, "grad_norm": 11.5, "learning_rate": 9.996126020260264e-06, "loss": 2.0494, "step": 369 }, { "epoch": 0.04242632725604862, "grad_norm": 9.9375, "learning_rate": 9.996052591184695e-06, "loss": 1.913, "step": 370 }, { "epoch": 0.04254099300538929, "grad_norm": 10.875, "learning_rate": 9.995978472998905e-06, "loss": 1.9757, "step": 371 }, { "epoch": 0.042655658754729965, "grad_norm": 13.5625, "learning_rate": 9.995903665713117e-06, "loss": 2.3389, "step": 372 }, { "epoch": 0.04277032450407063, "grad_norm": 9.5, "learning_rate": 9.99582816933765e-06, "loss": 1.7196, "step": 373 }, { "epoch": 0.042884990253411304, "grad_norm": 13.0, "learning_rate": 9.995751983882915e-06, "loss": 1.768, "step": 374 }, { "epoch": 0.042999656002751976, "grad_norm": 9.9375, "learning_rate": 9.99567510935942e-06, "loss": 1.453, "step": 375 }, { "epoch": 0.04311432175209265, "grad_norm": 10.125, "learning_rate": 9.995597545777771e-06, "loss": 1.9459, "step": 376 }, { "epoch": 0.04322898750143332, "grad_norm": 12.4375, "learning_rate": 9.995519293148667e-06, "loss": 2.5102, "step": 377 }, { "epoch": 0.043343653250773995, "grad_norm": 10.4375, "learning_rate": 9.995440351482898e-06, "loss": 1.9758, "step": 378 }, { "epoch": 0.04345831900011467, "grad_norm": 9.75, "learning_rate": 9.995360720791353e-06, "loss": 1.9289, "step": 379 }, { "epoch": 0.04357298474945534, "grad_norm": 9.875, "learning_rate": 9.99528040108502e-06, "loss": 1.6345, "step": 380 }, { "epoch": 0.04368765049879601, "grad_norm": 13.375, "learning_rate": 9.995199392374973e-06, "loss": 1.6882, "step": 381 }, { "epoch": 0.04380231624813668, "grad_norm": 10.625, "learning_rate": 9.995117694672388e-06, "loss": 1.7643, "step": 382 }, { "epoch": 0.04391698199747735, "grad_norm": 13.0625, "learning_rate": 9.99503530798853e-06, "loss": 2.2024, "step": 383 }, { "epoch": 0.044031647746818024, "grad_norm": 11.25, "learning_rate": 9.994952232334766e-06, "loss": 1.719, "step": 384 }, { "epoch": 0.0441463134961587, "grad_norm": 13.25, "learning_rate": 9.994868467722556e-06, "loss": 1.5611, "step": 385 }, { "epoch": 0.04426097924549937, "grad_norm": 10.875, "learning_rate": 9.99478401416345e-06, "loss": 1.5896, "step": 386 }, { "epoch": 0.04437564499484004, "grad_norm": 10.5, "learning_rate": 9.9946988716691e-06, "loss": 2.0887, "step": 387 }, { "epoch": 0.044490310744180715, "grad_norm": 9.9375, "learning_rate": 9.994613040251246e-06, "loss": 1.8936, "step": 388 }, { "epoch": 0.04460497649352139, "grad_norm": 13.875, "learning_rate": 9.994526519921732e-06, "loss": 2.0014, "step": 389 }, { "epoch": 0.044719642242862054, "grad_norm": 13.0, "learning_rate": 9.994439310692487e-06, "loss": 1.7695, "step": 390 }, { "epoch": 0.04483430799220273, "grad_norm": 9.125, "learning_rate": 9.994351412575542e-06, "loss": 1.9653, "step": 391 }, { "epoch": 0.0449489737415434, "grad_norm": 12.75, "learning_rate": 9.99426282558302e-06, "loss": 1.7951, "step": 392 }, { "epoch": 0.04506363949088407, "grad_norm": 14.1875, "learning_rate": 9.994173549727142e-06, "loss": 2.2271, "step": 393 }, { "epoch": 0.045178305240224745, "grad_norm": 10.0, "learning_rate": 9.994083585020218e-06, "loss": 1.6338, "step": 394 }, { "epoch": 0.04529297098956542, "grad_norm": 10.5625, "learning_rate": 9.993992931474662e-06, "loss": 1.6508, "step": 395 }, { "epoch": 0.04540763673890609, "grad_norm": 10.5, "learning_rate": 9.993901589102974e-06, "loss": 2.0675, "step": 396 }, { "epoch": 0.04552230248824676, "grad_norm": 9.1875, "learning_rate": 9.993809557917755e-06, "loss": 1.767, "step": 397 }, { "epoch": 0.045636968237587436, "grad_norm": 9.25, "learning_rate": 9.993716837931698e-06, "loss": 1.7164, "step": 398 }, { "epoch": 0.0457516339869281, "grad_norm": 13.5625, "learning_rate": 9.993623429157591e-06, "loss": 2.094, "step": 399 }, { "epoch": 0.045866299736268774, "grad_norm": 9.0, "learning_rate": 9.99352933160832e-06, "loss": 1.6744, "step": 400 }, { "epoch": 0.04598096548560945, "grad_norm": 13.0625, "learning_rate": 9.993434545296863e-06, "loss": 1.8149, "step": 401 }, { "epoch": 0.04609563123495012, "grad_norm": 9.5, "learning_rate": 9.993339070236294e-06, "loss": 1.5791, "step": 402 }, { "epoch": 0.04621029698429079, "grad_norm": 8.625, "learning_rate": 9.99324290643978e-06, "loss": 1.4621, "step": 403 }, { "epoch": 0.046324962733631465, "grad_norm": 11.6875, "learning_rate": 9.99314605392059e-06, "loss": 1.5301, "step": 404 }, { "epoch": 0.04643962848297214, "grad_norm": 10.0, "learning_rate": 9.993048512692079e-06, "loss": 1.8475, "step": 405 }, { "epoch": 0.04655429423231281, "grad_norm": 12.875, "learning_rate": 9.992950282767701e-06, "loss": 2.2196, "step": 406 }, { "epoch": 0.04666895998165348, "grad_norm": 11.5, "learning_rate": 9.992851364161006e-06, "loss": 1.7445, "step": 407 }, { "epoch": 0.04678362573099415, "grad_norm": 12.125, "learning_rate": 9.992751756885638e-06, "loss": 1.8614, "step": 408 }, { "epoch": 0.04689829148033482, "grad_norm": 9.4375, "learning_rate": 9.992651460955334e-06, "loss": 1.497, "step": 409 }, { "epoch": 0.047012957229675495, "grad_norm": 11.9375, "learning_rate": 9.992550476383931e-06, "loss": 2.2798, "step": 410 }, { "epoch": 0.04712762297901617, "grad_norm": 11.4375, "learning_rate": 9.992448803185357e-06, "loss": 1.9705, "step": 411 }, { "epoch": 0.04724228872835684, "grad_norm": 17.75, "learning_rate": 9.992346441373634e-06, "loss": 2.1785, "step": 412 }, { "epoch": 0.04735695447769751, "grad_norm": 10.9375, "learning_rate": 9.992243390962885e-06, "loss": 1.7301, "step": 413 }, { "epoch": 0.047471620227038186, "grad_norm": 8.5625, "learning_rate": 9.992139651967318e-06, "loss": 1.6393, "step": 414 }, { "epoch": 0.04758628597637886, "grad_norm": 11.0, "learning_rate": 9.992035224401246e-06, "loss": 2.132, "step": 415 }, { "epoch": 0.047700951725719525, "grad_norm": 12.5625, "learning_rate": 9.991930108279073e-06, "loss": 2.3096, "step": 416 }, { "epoch": 0.0478156174750602, "grad_norm": 11.625, "learning_rate": 9.991824303615294e-06, "loss": 1.9808, "step": 417 }, { "epoch": 0.04793028322440087, "grad_norm": 33.25, "learning_rate": 9.991717810424506e-06, "loss": 1.8749, "step": 418 }, { "epoch": 0.04804494897374154, "grad_norm": 11.375, "learning_rate": 9.991610628721398e-06, "loss": 1.6198, "step": 419 }, { "epoch": 0.048159614723082216, "grad_norm": 11.1875, "learning_rate": 9.991502758520752e-06, "loss": 1.8397, "step": 420 }, { "epoch": 0.04827428047242289, "grad_norm": 10.0625, "learning_rate": 9.991394199837445e-06, "loss": 1.9323, "step": 421 }, { "epoch": 0.04838894622176356, "grad_norm": 10.9375, "learning_rate": 9.991284952686455e-06, "loss": 1.9141, "step": 422 }, { "epoch": 0.048503611971104234, "grad_norm": 19.125, "learning_rate": 9.991175017082849e-06, "loss": 2.4725, "step": 423 }, { "epoch": 0.04861827772044491, "grad_norm": 29.375, "learning_rate": 9.991064393041787e-06, "loss": 1.8432, "step": 424 }, { "epoch": 0.04873294346978557, "grad_norm": 12.1875, "learning_rate": 9.990953080578533e-06, "loss": 2.0839, "step": 425 }, { "epoch": 0.048847609219126245, "grad_norm": 13.125, "learning_rate": 9.990841079708436e-06, "loss": 1.5958, "step": 426 }, { "epoch": 0.04896227496846692, "grad_norm": 9.3125, "learning_rate": 9.990728390446946e-06, "loss": 1.5192, "step": 427 }, { "epoch": 0.04907694071780759, "grad_norm": 10.4375, "learning_rate": 9.990615012809608e-06, "loss": 1.9536, "step": 428 }, { "epoch": 0.04919160646714826, "grad_norm": 16.0, "learning_rate": 9.990500946812058e-06, "loss": 1.687, "step": 429 }, { "epoch": 0.049306272216488936, "grad_norm": 10.9375, "learning_rate": 9.990386192470031e-06, "loss": 1.6868, "step": 430 }, { "epoch": 0.04942093796582961, "grad_norm": 10.25, "learning_rate": 9.990270749799353e-06, "loss": 1.7411, "step": 431 }, { "epoch": 0.04953560371517028, "grad_norm": 9.875, "learning_rate": 9.990154618815948e-06, "loss": 2.1048, "step": 432 }, { "epoch": 0.04965026946451095, "grad_norm": 9.8125, "learning_rate": 9.990037799535834e-06, "loss": 1.8389, "step": 433 }, { "epoch": 0.04976493521385162, "grad_norm": 10.5, "learning_rate": 9.989920291975125e-06, "loss": 2.1526, "step": 434 }, { "epoch": 0.04987960096319229, "grad_norm": 11.5625, "learning_rate": 9.98980209615003e-06, "loss": 1.9022, "step": 435 }, { "epoch": 0.049994266712532966, "grad_norm": 13.0, "learning_rate": 9.989683212076849e-06, "loss": 1.8956, "step": 436 }, { "epoch": 0.05010893246187364, "grad_norm": 15.875, "learning_rate": 9.98956363977198e-06, "loss": 2.0839, "step": 437 }, { "epoch": 0.05022359821121431, "grad_norm": 9.4375, "learning_rate": 9.989443379251918e-06, "loss": 1.607, "step": 438 }, { "epoch": 0.050338263960554984, "grad_norm": 10.375, "learning_rate": 9.989322430533247e-06, "loss": 1.8319, "step": 439 }, { "epoch": 0.05045292970989566, "grad_norm": 30.75, "learning_rate": 9.989200793632654e-06, "loss": 2.0955, "step": 440 }, { "epoch": 0.05056759545923633, "grad_norm": 10.375, "learning_rate": 9.989078468566912e-06, "loss": 1.7971, "step": 441 }, { "epoch": 0.050682261208576995, "grad_norm": 10.8125, "learning_rate": 9.988955455352897e-06, "loss": 2.0888, "step": 442 }, { "epoch": 0.05079692695791767, "grad_norm": 11.625, "learning_rate": 9.988831754007577e-06, "loss": 2.3194, "step": 443 }, { "epoch": 0.05091159270725834, "grad_norm": 12.125, "learning_rate": 9.98870736454801e-06, "loss": 1.4762, "step": 444 }, { "epoch": 0.051026258456599013, "grad_norm": 14.5, "learning_rate": 9.988582286991356e-06, "loss": 2.1649, "step": 445 }, { "epoch": 0.051140924205939686, "grad_norm": 11.375, "learning_rate": 9.988456521354868e-06, "loss": 2.1586, "step": 446 }, { "epoch": 0.05125558995528036, "grad_norm": 14.1875, "learning_rate": 9.988330067655889e-06, "loss": 2.2246, "step": 447 }, { "epoch": 0.05137025570462103, "grad_norm": 11.1875, "learning_rate": 9.988202925911865e-06, "loss": 1.6632, "step": 448 }, { "epoch": 0.051484921453961704, "grad_norm": 12.5, "learning_rate": 9.98807509614033e-06, "loss": 2.0423, "step": 449 }, { "epoch": 0.05159958720330237, "grad_norm": 10.625, "learning_rate": 9.987946578358918e-06, "loss": 1.718, "step": 450 }, { "epoch": 0.05171425295264304, "grad_norm": 10.625, "learning_rate": 9.987817372585354e-06, "loss": 1.4728, "step": 451 }, { "epoch": 0.051828918701983716, "grad_norm": 13.1875, "learning_rate": 9.98768747883746e-06, "loss": 2.0625, "step": 452 }, { "epoch": 0.05194358445132439, "grad_norm": 29.75, "learning_rate": 9.987556897133152e-06, "loss": 1.9275, "step": 453 }, { "epoch": 0.05205825020066506, "grad_norm": 16.0, "learning_rate": 9.987425627490442e-06, "loss": 2.193, "step": 454 }, { "epoch": 0.052172915950005734, "grad_norm": 8.9375, "learning_rate": 9.987293669927436e-06, "loss": 1.6979, "step": 455 }, { "epoch": 0.05228758169934641, "grad_norm": 13.25, "learning_rate": 9.987161024462333e-06, "loss": 1.6063, "step": 456 }, { "epoch": 0.05240224744868708, "grad_norm": 9.75, "learning_rate": 9.987027691113433e-06, "loss": 1.8876, "step": 457 }, { "epoch": 0.05251691319802775, "grad_norm": 13.1875, "learning_rate": 9.986893669899122e-06, "loss": 2.0059, "step": 458 }, { "epoch": 0.05263157894736842, "grad_norm": 10.8125, "learning_rate": 9.98675896083789e-06, "loss": 2.0189, "step": 459 }, { "epoch": 0.05274624469670909, "grad_norm": 11.1875, "learning_rate": 9.986623563948314e-06, "loss": 2.0079, "step": 460 }, { "epoch": 0.052860910446049764, "grad_norm": 10.4375, "learning_rate": 9.986487479249073e-06, "loss": 1.8427, "step": 461 }, { "epoch": 0.052975576195390436, "grad_norm": 11.5625, "learning_rate": 9.986350706758934e-06, "loss": 1.7834, "step": 462 }, { "epoch": 0.05309024194473111, "grad_norm": 23.0, "learning_rate": 9.986213246496764e-06, "loss": 2.1356, "step": 463 }, { "epoch": 0.05320490769407178, "grad_norm": 11.6875, "learning_rate": 9.986075098481521e-06, "loss": 1.9342, "step": 464 }, { "epoch": 0.053319573443412455, "grad_norm": 12.1875, "learning_rate": 9.985936262732263e-06, "loss": 1.6749, "step": 465 }, { "epoch": 0.05343423919275313, "grad_norm": 12.0625, "learning_rate": 9.985796739268138e-06, "loss": 1.7218, "step": 466 }, { "epoch": 0.05354890494209379, "grad_norm": 10.25, "learning_rate": 9.985656528108391e-06, "loss": 1.7632, "step": 467 }, { "epoch": 0.053663570691434466, "grad_norm": 11.0625, "learning_rate": 9.98551562927236e-06, "loss": 1.8978, "step": 468 }, { "epoch": 0.05377823644077514, "grad_norm": 11.5, "learning_rate": 9.98537404277948e-06, "loss": 1.8173, "step": 469 }, { "epoch": 0.05389290219011581, "grad_norm": 11.5625, "learning_rate": 9.985231768649284e-06, "loss": 2.2566, "step": 470 }, { "epoch": 0.054007567939456484, "grad_norm": 10.6875, "learning_rate": 9.985088806901391e-06, "loss": 1.7254, "step": 471 }, { "epoch": 0.05412223368879716, "grad_norm": 10.4375, "learning_rate": 9.98494515755552e-06, "loss": 1.6783, "step": 472 }, { "epoch": 0.05423689943813783, "grad_norm": 10.4375, "learning_rate": 9.984800820631487e-06, "loss": 1.8355, "step": 473 }, { "epoch": 0.0543515651874785, "grad_norm": 10.625, "learning_rate": 9.984655796149201e-06, "loss": 2.2732, "step": 474 }, { "epoch": 0.054466230936819175, "grad_norm": 12.8125, "learning_rate": 9.984510084128662e-06, "loss": 1.8638, "step": 475 }, { "epoch": 0.05458089668615984, "grad_norm": 10.9375, "learning_rate": 9.984363684589972e-06, "loss": 1.5064, "step": 476 }, { "epoch": 0.054695562435500514, "grad_norm": 12.9375, "learning_rate": 9.984216597553322e-06, "loss": 2.0861, "step": 477 }, { "epoch": 0.054810228184841187, "grad_norm": 11.875, "learning_rate": 9.984068823039e-06, "loss": 1.7857, "step": 478 }, { "epoch": 0.05492489393418186, "grad_norm": 10.75, "learning_rate": 9.983920361067388e-06, "loss": 1.8898, "step": 479 }, { "epoch": 0.05503955968352253, "grad_norm": 12.75, "learning_rate": 9.983771211658966e-06, "loss": 1.6008, "step": 480 }, { "epoch": 0.055154225432863205, "grad_norm": 11.8125, "learning_rate": 9.983621374834303e-06, "loss": 1.6922, "step": 481 }, { "epoch": 0.05526889118220388, "grad_norm": 9.75, "learning_rate": 9.98347085061407e-06, "loss": 1.811, "step": 482 }, { "epoch": 0.05538355693154455, "grad_norm": 12.75, "learning_rate": 9.983319639019025e-06, "loss": 1.7546, "step": 483 }, { "epoch": 0.05549822268088522, "grad_norm": 13.875, "learning_rate": 9.983167740070027e-06, "loss": 1.9747, "step": 484 }, { "epoch": 0.05561288843022589, "grad_norm": 10.6875, "learning_rate": 9.983015153788027e-06, "loss": 2.0572, "step": 485 }, { "epoch": 0.05572755417956656, "grad_norm": 11.5625, "learning_rate": 9.982861880194071e-06, "loss": 1.6853, "step": 486 }, { "epoch": 0.055842219928907234, "grad_norm": 12.5, "learning_rate": 9.9827079193093e-06, "loss": 1.9363, "step": 487 }, { "epoch": 0.05595688567824791, "grad_norm": 13.6875, "learning_rate": 9.982553271154953e-06, "loss": 1.774, "step": 488 }, { "epoch": 0.05607155142758858, "grad_norm": 10.3125, "learning_rate": 9.982397935752357e-06, "loss": 1.8513, "step": 489 }, { "epoch": 0.05618621717692925, "grad_norm": 15.4375, "learning_rate": 9.982241913122938e-06, "loss": 2.2838, "step": 490 }, { "epoch": 0.056300882926269925, "grad_norm": 9.3125, "learning_rate": 9.982085203288219e-06, "loss": 1.6388, "step": 491 }, { "epoch": 0.0564155486756106, "grad_norm": 11.6875, "learning_rate": 9.981927806269813e-06, "loss": 2.0311, "step": 492 }, { "epoch": 0.056530214424951264, "grad_norm": 12.75, "learning_rate": 9.98176972208943e-06, "loss": 2.0378, "step": 493 }, { "epoch": 0.05664488017429194, "grad_norm": 11.9375, "learning_rate": 9.981610950768874e-06, "loss": 2.1273, "step": 494 }, { "epoch": 0.05675954592363261, "grad_norm": 12.5625, "learning_rate": 9.981451492330047e-06, "loss": 1.9581, "step": 495 }, { "epoch": 0.05687421167297328, "grad_norm": 14.1875, "learning_rate": 9.98129134679494e-06, "loss": 2.0106, "step": 496 }, { "epoch": 0.056988877422313955, "grad_norm": 12.625, "learning_rate": 9.981130514185645e-06, "loss": 1.9209, "step": 497 }, { "epoch": 0.05710354317165463, "grad_norm": 12.4375, "learning_rate": 9.980968994524345e-06, "loss": 1.7053, "step": 498 }, { "epoch": 0.0572182089209953, "grad_norm": 11.8125, "learning_rate": 9.980806787833317e-06, "loss": 1.8568, "step": 499 }, { "epoch": 0.05733287467033597, "grad_norm": 30.25, "learning_rate": 9.980643894134937e-06, "loss": 1.8686, "step": 500 }, { "epoch": 0.057447540419676646, "grad_norm": 11.3125, "learning_rate": 9.98048031345167e-06, "loss": 1.8413, "step": 501 }, { "epoch": 0.05756220616901731, "grad_norm": 10.625, "learning_rate": 9.98031604580608e-06, "loss": 1.9459, "step": 502 }, { "epoch": 0.057676871918357984, "grad_norm": 14.75, "learning_rate": 9.980151091220826e-06, "loss": 1.6567, "step": 503 }, { "epoch": 0.05779153766769866, "grad_norm": 11.5, "learning_rate": 9.97998544971866e-06, "loss": 1.9606, "step": 504 }, { "epoch": 0.05790620341703933, "grad_norm": 10.5, "learning_rate": 9.979819121322427e-06, "loss": 1.9767, "step": 505 }, { "epoch": 0.05802086916638, "grad_norm": 10.6875, "learning_rate": 9.979652106055072e-06, "loss": 1.9541, "step": 506 }, { "epoch": 0.058135534915720675, "grad_norm": 11.5625, "learning_rate": 9.979484403939627e-06, "loss": 1.8699, "step": 507 }, { "epoch": 0.05825020066506135, "grad_norm": 10.5625, "learning_rate": 9.979316014999227e-06, "loss": 1.5196, "step": 508 }, { "epoch": 0.05836486641440202, "grad_norm": 10.5, "learning_rate": 9.979146939257098e-06, "loss": 1.8537, "step": 509 }, { "epoch": 0.05847953216374269, "grad_norm": 12.1875, "learning_rate": 9.97897717673656e-06, "loss": 1.6523, "step": 510 }, { "epoch": 0.05859419791308336, "grad_norm": 10.8125, "learning_rate": 9.978806727461028e-06, "loss": 1.648, "step": 511 }, { "epoch": 0.05870886366242403, "grad_norm": 10.5, "learning_rate": 9.978635591454012e-06, "loss": 1.7965, "step": 512 }, { "epoch": 0.058823529411764705, "grad_norm": 31.625, "learning_rate": 9.978463768739119e-06, "loss": 2.3536, "step": 513 }, { "epoch": 0.05893819516110538, "grad_norm": 11.1875, "learning_rate": 9.978291259340044e-06, "loss": 1.6261, "step": 514 }, { "epoch": 0.05905286091044605, "grad_norm": 12.25, "learning_rate": 9.97811806328059e-06, "loss": 1.7638, "step": 515 }, { "epoch": 0.05916752665978672, "grad_norm": 11.125, "learning_rate": 9.977944180584636e-06, "loss": 1.8232, "step": 516 }, { "epoch": 0.059282192409127396, "grad_norm": 14.0625, "learning_rate": 9.977769611276173e-06, "loss": 2.1356, "step": 517 }, { "epoch": 0.05939685815846807, "grad_norm": 9.0625, "learning_rate": 9.977594355379276e-06, "loss": 1.5501, "step": 518 }, { "epoch": 0.059511523907808735, "grad_norm": 10.4375, "learning_rate": 9.97741841291812e-06, "loss": 1.9987, "step": 519 }, { "epoch": 0.05962618965714941, "grad_norm": 13.8125, "learning_rate": 9.97724178391697e-06, "loss": 2.4986, "step": 520 }, { "epoch": 0.05974085540649008, "grad_norm": 10.75, "learning_rate": 9.977064468400193e-06, "loss": 1.9546, "step": 521 }, { "epoch": 0.05985552115583075, "grad_norm": 10.5, "learning_rate": 9.976886466392246e-06, "loss": 1.7046, "step": 522 }, { "epoch": 0.059970186905171426, "grad_norm": 13.125, "learning_rate": 9.976707777917676e-06, "loss": 1.8739, "step": 523 }, { "epoch": 0.0600848526545121, "grad_norm": 10.5625, "learning_rate": 9.976528403001134e-06, "loss": 1.8379, "step": 524 }, { "epoch": 0.06019951840385277, "grad_norm": 14.1875, "learning_rate": 9.976348341667357e-06, "loss": 1.7463, "step": 525 }, { "epoch": 0.060314184153193444, "grad_norm": 12.0, "learning_rate": 9.976167593941189e-06, "loss": 1.5953, "step": 526 }, { "epoch": 0.06042884990253411, "grad_norm": 13.0, "learning_rate": 9.975986159847552e-06, "loss": 1.4069, "step": 527 }, { "epoch": 0.06054351565187478, "grad_norm": 10.3125, "learning_rate": 9.975804039411477e-06, "loss": 2.1998, "step": 528 }, { "epoch": 0.060658181401215455, "grad_norm": 13.4375, "learning_rate": 9.97562123265808e-06, "loss": 1.6643, "step": 529 }, { "epoch": 0.06077284715055613, "grad_norm": 14.4375, "learning_rate": 9.975437739612582e-06, "loss": 2.0784, "step": 530 }, { "epoch": 0.0608875128998968, "grad_norm": 30.5, "learning_rate": 9.975253560300284e-06, "loss": 2.0125, "step": 531 }, { "epoch": 0.06100217864923747, "grad_norm": 11.6875, "learning_rate": 9.975068694746597e-06, "loss": 1.7426, "step": 532 }, { "epoch": 0.061116844398578146, "grad_norm": 13.125, "learning_rate": 9.974883142977017e-06, "loss": 1.5341, "step": 533 }, { "epoch": 0.06123151014791882, "grad_norm": 10.6875, "learning_rate": 9.974696905017137e-06, "loss": 2.0774, "step": 534 }, { "epoch": 0.06134617589725949, "grad_norm": 14.875, "learning_rate": 9.974509980892644e-06, "loss": 2.0798, "step": 535 }, { "epoch": 0.06146084164660016, "grad_norm": 12.375, "learning_rate": 9.974322370629323e-06, "loss": 1.5644, "step": 536 }, { "epoch": 0.06157550739594083, "grad_norm": 10.0625, "learning_rate": 9.97413407425305e-06, "loss": 1.8178, "step": 537 }, { "epoch": 0.0616901731452815, "grad_norm": 14.5, "learning_rate": 9.973945091789797e-06, "loss": 1.8419, "step": 538 }, { "epoch": 0.061804838894622176, "grad_norm": 12.125, "learning_rate": 9.97375542326563e-06, "loss": 1.5528, "step": 539 }, { "epoch": 0.06191950464396285, "grad_norm": 13.375, "learning_rate": 9.973565068706711e-06, "loss": 1.7369, "step": 540 }, { "epoch": 0.06203417039330352, "grad_norm": 10.3125, "learning_rate": 9.973374028139296e-06, "loss": 1.6429, "step": 541 }, { "epoch": 0.062148836142644194, "grad_norm": 11.625, "learning_rate": 9.973182301589735e-06, "loss": 1.8479, "step": 542 }, { "epoch": 0.06226350189198487, "grad_norm": 15.875, "learning_rate": 9.972989889084473e-06, "loss": 2.0476, "step": 543 }, { "epoch": 0.06237816764132553, "grad_norm": 12.5625, "learning_rate": 9.97279679065005e-06, "loss": 2.0793, "step": 544 }, { "epoch": 0.062492833390666205, "grad_norm": 12.6875, "learning_rate": 9.972603006313099e-06, "loss": 1.9345, "step": 545 }, { "epoch": 0.06260749914000688, "grad_norm": 11.125, "learning_rate": 9.97240853610035e-06, "loss": 1.9608, "step": 546 }, { "epoch": 0.06272216488934755, "grad_norm": 11.25, "learning_rate": 9.972213380038626e-06, "loss": 1.6997, "step": 547 }, { "epoch": 0.06283683063868822, "grad_norm": 11.125, "learning_rate": 9.972017538154846e-06, "loss": 1.9439, "step": 548 }, { "epoch": 0.0629514963880289, "grad_norm": 10.375, "learning_rate": 9.971821010476021e-06, "loss": 1.5731, "step": 549 }, { "epoch": 0.06306616213736957, "grad_norm": 30.75, "learning_rate": 9.97162379702926e-06, "loss": 1.8635, "step": 550 }, { "epoch": 0.06318082788671024, "grad_norm": 9.8125, "learning_rate": 9.971425897841764e-06, "loss": 2.0005, "step": 551 }, { "epoch": 0.06329549363605091, "grad_norm": 13.75, "learning_rate": 9.971227312940828e-06, "loss": 2.1071, "step": 552 }, { "epoch": 0.06341015938539159, "grad_norm": 11.8125, "learning_rate": 9.971028042353846e-06, "loss": 2.1505, "step": 553 }, { "epoch": 0.06352482513473226, "grad_norm": 11.9375, "learning_rate": 9.970828086108302e-06, "loss": 1.9455, "step": 554 }, { "epoch": 0.06363949088407293, "grad_norm": 13.4375, "learning_rate": 9.970627444231775e-06, "loss": 1.9742, "step": 555 }, { "epoch": 0.0637541566334136, "grad_norm": 13.5625, "learning_rate": 9.970426116751941e-06, "loss": 1.9272, "step": 556 }, { "epoch": 0.06386882238275426, "grad_norm": 13.125, "learning_rate": 9.970224103696569e-06, "loss": 1.5466, "step": 557 }, { "epoch": 0.06398348813209494, "grad_norm": 11.8125, "learning_rate": 9.970021405093524e-06, "loss": 1.719, "step": 558 }, { "epoch": 0.06409815388143561, "grad_norm": 31.0, "learning_rate": 9.969818020970762e-06, "loss": 2.3348, "step": 559 }, { "epoch": 0.06421281963077628, "grad_norm": 14.8125, "learning_rate": 9.969613951356338e-06, "loss": 1.6037, "step": 560 }, { "epoch": 0.06432748538011696, "grad_norm": 12.625, "learning_rate": 9.969409196278399e-06, "loss": 1.7875, "step": 561 }, { "epoch": 0.06444215112945763, "grad_norm": 15.3125, "learning_rate": 9.969203755765187e-06, "loss": 2.2653, "step": 562 }, { "epoch": 0.0645568168787983, "grad_norm": 9.5, "learning_rate": 9.968997629845039e-06, "loss": 1.6155, "step": 563 }, { "epoch": 0.06467148262813897, "grad_norm": 15.3125, "learning_rate": 9.968790818546385e-06, "loss": 2.1484, "step": 564 }, { "epoch": 0.06478614837747965, "grad_norm": 15.375, "learning_rate": 9.968583321897751e-06, "loss": 1.8584, "step": 565 }, { "epoch": 0.06490081412682032, "grad_norm": 14.375, "learning_rate": 9.968375139927757e-06, "loss": 1.7825, "step": 566 }, { "epoch": 0.06501547987616099, "grad_norm": 13.5625, "learning_rate": 9.968166272665121e-06, "loss": 1.4534, "step": 567 }, { "epoch": 0.06513014562550166, "grad_norm": 13.0, "learning_rate": 9.967956720138648e-06, "loss": 2.1553, "step": 568 }, { "epoch": 0.06524481137484234, "grad_norm": 12.5, "learning_rate": 9.967746482377242e-06, "loss": 2.0399, "step": 569 }, { "epoch": 0.06535947712418301, "grad_norm": 11.6875, "learning_rate": 9.967535559409904e-06, "loss": 1.9391, "step": 570 }, { "epoch": 0.06547414287352368, "grad_norm": 14.0625, "learning_rate": 9.967323951265727e-06, "loss": 1.9263, "step": 571 }, { "epoch": 0.06558880862286436, "grad_norm": 9.3125, "learning_rate": 9.967111657973894e-06, "loss": 1.8204, "step": 572 }, { "epoch": 0.06570347437220503, "grad_norm": 10.875, "learning_rate": 9.96689867956369e-06, "loss": 1.9118, "step": 573 }, { "epoch": 0.06581814012154569, "grad_norm": 14.4375, "learning_rate": 9.966685016064491e-06, "loss": 1.897, "step": 574 }, { "epoch": 0.06593280587088636, "grad_norm": 12.5625, "learning_rate": 9.966470667505767e-06, "loss": 2.1889, "step": 575 }, { "epoch": 0.06604747162022703, "grad_norm": 12.125, "learning_rate": 9.966255633917086e-06, "loss": 1.8603, "step": 576 }, { "epoch": 0.0661621373695677, "grad_norm": 9.9375, "learning_rate": 9.966039915328105e-06, "loss": 1.7495, "step": 577 }, { "epoch": 0.06627680311890838, "grad_norm": 11.875, "learning_rate": 9.965823511768579e-06, "loss": 1.8411, "step": 578 }, { "epoch": 0.06639146886824905, "grad_norm": 12.125, "learning_rate": 9.965606423268355e-06, "loss": 2.0404, "step": 579 }, { "epoch": 0.06650613461758972, "grad_norm": 11.6875, "learning_rate": 9.96538864985738e-06, "loss": 1.6162, "step": 580 }, { "epoch": 0.0666208003669304, "grad_norm": 28.625, "learning_rate": 9.96517019156569e-06, "loss": 1.9033, "step": 581 }, { "epoch": 0.06673546611627107, "grad_norm": 13.8125, "learning_rate": 9.964951048423416e-06, "loss": 1.9202, "step": 582 }, { "epoch": 0.06685013186561174, "grad_norm": 10.4375, "learning_rate": 9.964731220460785e-06, "loss": 1.7628, "step": 583 }, { "epoch": 0.06696479761495241, "grad_norm": 13.75, "learning_rate": 9.96451070770812e-06, "loss": 2.2524, "step": 584 }, { "epoch": 0.06707946336429309, "grad_norm": 9.9375, "learning_rate": 9.964289510195832e-06, "loss": 1.8657, "step": 585 }, { "epoch": 0.06719412911363376, "grad_norm": 12.5, "learning_rate": 9.964067627954436e-06, "loss": 1.6437, "step": 586 }, { "epoch": 0.06730879486297443, "grad_norm": 12.8125, "learning_rate": 9.963845061014535e-06, "loss": 2.1311, "step": 587 }, { "epoch": 0.0674234606123151, "grad_norm": 14.25, "learning_rate": 9.963621809406827e-06, "loss": 2.0989, "step": 588 }, { "epoch": 0.06753812636165578, "grad_norm": 14.125, "learning_rate": 9.963397873162106e-06, "loss": 1.9968, "step": 589 }, { "epoch": 0.06765279211099645, "grad_norm": 12.3125, "learning_rate": 9.963173252311258e-06, "loss": 1.8338, "step": 590 }, { "epoch": 0.06776745786033712, "grad_norm": 14.0625, "learning_rate": 9.962947946885268e-06, "loss": 1.8015, "step": 591 }, { "epoch": 0.06788212360967778, "grad_norm": 10.6875, "learning_rate": 9.96272195691521e-06, "loss": 1.7249, "step": 592 }, { "epoch": 0.06799678935901846, "grad_norm": 10.375, "learning_rate": 9.962495282432257e-06, "loss": 1.9883, "step": 593 }, { "epoch": 0.06811145510835913, "grad_norm": 12.625, "learning_rate": 9.962267923467673e-06, "loss": 1.9586, "step": 594 }, { "epoch": 0.0682261208576998, "grad_norm": 31.875, "learning_rate": 9.962039880052818e-06, "loss": 2.1534, "step": 595 }, { "epoch": 0.06834078660704047, "grad_norm": 11.75, "learning_rate": 9.961811152219148e-06, "loss": 1.7357, "step": 596 }, { "epoch": 0.06845545235638115, "grad_norm": 12.1875, "learning_rate": 9.96158173999821e-06, "loss": 1.8237, "step": 597 }, { "epoch": 0.06857011810572182, "grad_norm": 11.4375, "learning_rate": 9.961351643421646e-06, "loss": 1.7998, "step": 598 }, { "epoch": 0.06868478385506249, "grad_norm": 13.25, "learning_rate": 9.961120862521197e-06, "loss": 1.7733, "step": 599 }, { "epoch": 0.06879944960440317, "grad_norm": 9.5625, "learning_rate": 9.96088939732869e-06, "loss": 1.5393, "step": 600 }, { "epoch": 0.06891411535374384, "grad_norm": 12.875, "learning_rate": 9.960657247876055e-06, "loss": 1.8409, "step": 601 }, { "epoch": 0.06902878110308451, "grad_norm": 11.125, "learning_rate": 9.96042441419531e-06, "loss": 1.9938, "step": 602 }, { "epoch": 0.06914344685242518, "grad_norm": 13.75, "learning_rate": 9.960190896318573e-06, "loss": 1.5707, "step": 603 }, { "epoch": 0.06925811260176586, "grad_norm": 12.4375, "learning_rate": 9.959956694278052e-06, "loss": 1.4606, "step": 604 }, { "epoch": 0.06937277835110653, "grad_norm": 12.4375, "learning_rate": 9.95972180810605e-06, "loss": 1.6518, "step": 605 }, { "epoch": 0.0694874441004472, "grad_norm": 11.6875, "learning_rate": 9.959486237834966e-06, "loss": 2.112, "step": 606 }, { "epoch": 0.06960210984978787, "grad_norm": 11.8125, "learning_rate": 9.95924998349729e-06, "loss": 1.4188, "step": 607 }, { "epoch": 0.06971677559912855, "grad_norm": 26.875, "learning_rate": 9.959013045125612e-06, "loss": 2.2382, "step": 608 }, { "epoch": 0.0698314413484692, "grad_norm": 10.0625, "learning_rate": 9.958775422752611e-06, "loss": 1.6318, "step": 609 }, { "epoch": 0.06994610709780988, "grad_norm": 15.125, "learning_rate": 9.958537116411064e-06, "loss": 1.715, "step": 610 }, { "epoch": 0.07006077284715055, "grad_norm": 11.625, "learning_rate": 9.95829812613384e-06, "loss": 1.6774, "step": 611 }, { "epoch": 0.07017543859649122, "grad_norm": 12.3125, "learning_rate": 9.958058451953903e-06, "loss": 1.7144, "step": 612 }, { "epoch": 0.0702901043458319, "grad_norm": 12.0, "learning_rate": 9.957818093904313e-06, "loss": 1.8859, "step": 613 }, { "epoch": 0.07040477009517257, "grad_norm": 11.1875, "learning_rate": 9.957577052018221e-06, "loss": 1.7771, "step": 614 }, { "epoch": 0.07051943584451324, "grad_norm": 12.0625, "learning_rate": 9.957335326328875e-06, "loss": 1.6904, "step": 615 }, { "epoch": 0.07063410159385392, "grad_norm": 11.3125, "learning_rate": 9.957092916869614e-06, "loss": 1.6194, "step": 616 }, { "epoch": 0.07074876734319459, "grad_norm": 12.8125, "learning_rate": 9.956849823673878e-06, "loss": 1.8123, "step": 617 }, { "epoch": 0.07086343309253526, "grad_norm": 10.4375, "learning_rate": 9.956606046775194e-06, "loss": 1.7385, "step": 618 }, { "epoch": 0.07097809884187593, "grad_norm": 15.875, "learning_rate": 9.956361586207188e-06, "loss": 1.8263, "step": 619 }, { "epoch": 0.0710927645912166, "grad_norm": 10.6875, "learning_rate": 9.956116442003576e-06, "loss": 1.7632, "step": 620 }, { "epoch": 0.07120743034055728, "grad_norm": 14.0625, "learning_rate": 9.955870614198175e-06, "loss": 1.6177, "step": 621 }, { "epoch": 0.07132209608989795, "grad_norm": 11.75, "learning_rate": 9.95562410282489e-06, "loss": 1.9589, "step": 622 }, { "epoch": 0.07143676183923862, "grad_norm": 11.1875, "learning_rate": 9.955376907917724e-06, "loss": 1.7586, "step": 623 }, { "epoch": 0.0715514275885793, "grad_norm": 16.25, "learning_rate": 9.95512902951077e-06, "loss": 2.0637, "step": 624 }, { "epoch": 0.07166609333791997, "grad_norm": 11.9375, "learning_rate": 9.954880467638219e-06, "loss": 2.0466, "step": 625 }, { "epoch": 0.07178075908726063, "grad_norm": 9.0, "learning_rate": 9.954631222334357e-06, "loss": 1.9342, "step": 626 }, { "epoch": 0.0718954248366013, "grad_norm": 13.375, "learning_rate": 9.954381293633563e-06, "loss": 1.7622, "step": 627 }, { "epoch": 0.07201009058594197, "grad_norm": 13.875, "learning_rate": 9.954130681570306e-06, "loss": 2.1026, "step": 628 }, { "epoch": 0.07212475633528265, "grad_norm": 11.3125, "learning_rate": 9.953879386179159e-06, "loss": 1.9724, "step": 629 }, { "epoch": 0.07223942208462332, "grad_norm": 10.125, "learning_rate": 9.953627407494777e-06, "loss": 1.7253, "step": 630 }, { "epoch": 0.07235408783396399, "grad_norm": 28.875, "learning_rate": 9.953374745551921e-06, "loss": 1.8647, "step": 631 }, { "epoch": 0.07246875358330467, "grad_norm": 28.75, "learning_rate": 9.953121400385439e-06, "loss": 2.1089, "step": 632 }, { "epoch": 0.07258341933264534, "grad_norm": 13.0625, "learning_rate": 9.952867372030274e-06, "loss": 1.9414, "step": 633 }, { "epoch": 0.07269808508198601, "grad_norm": 13.6875, "learning_rate": 9.952612660521466e-06, "loss": 1.9478, "step": 634 }, { "epoch": 0.07281275083132668, "grad_norm": 12.6875, "learning_rate": 9.952357265894146e-06, "loss": 1.7925, "step": 635 }, { "epoch": 0.07292741658066736, "grad_norm": 11.0, "learning_rate": 9.95210118818354e-06, "loss": 1.7499, "step": 636 }, { "epoch": 0.07304208233000803, "grad_norm": 9.75, "learning_rate": 9.951844427424974e-06, "loss": 1.3464, "step": 637 }, { "epoch": 0.0731567480793487, "grad_norm": 11.5625, "learning_rate": 9.951586983653857e-06, "loss": 1.6767, "step": 638 }, { "epoch": 0.07327141382868937, "grad_norm": 12.75, "learning_rate": 9.951328856905705e-06, "loss": 1.8669, "step": 639 }, { "epoch": 0.07338607957803005, "grad_norm": 10.0625, "learning_rate": 9.951070047216115e-06, "loss": 1.7213, "step": 640 }, { "epoch": 0.07350074532737072, "grad_norm": 11.625, "learning_rate": 9.95081055462079e-06, "loss": 1.7676, "step": 641 }, { "epoch": 0.07361541107671139, "grad_norm": 9.3125, "learning_rate": 9.95055037915552e-06, "loss": 1.968, "step": 642 }, { "epoch": 0.07373007682605205, "grad_norm": 12.625, "learning_rate": 9.950289520856189e-06, "loss": 1.9067, "step": 643 }, { "epoch": 0.07384474257539272, "grad_norm": 11.1875, "learning_rate": 9.950027979758782e-06, "loss": 1.4936, "step": 644 }, { "epoch": 0.0739594083247334, "grad_norm": 10.875, "learning_rate": 9.94976575589937e-06, "loss": 2.0637, "step": 645 }, { "epoch": 0.07407407407407407, "grad_norm": 14.4375, "learning_rate": 9.949502849314123e-06, "loss": 1.409, "step": 646 }, { "epoch": 0.07418873982341474, "grad_norm": 15.6875, "learning_rate": 9.949239260039306e-06, "loss": 2.0513, "step": 647 }, { "epoch": 0.07430340557275542, "grad_norm": 10.3125, "learning_rate": 9.948974988111272e-06, "loss": 1.5231, "step": 648 }, { "epoch": 0.07441807132209609, "grad_norm": 10.1875, "learning_rate": 9.948710033566475e-06, "loss": 1.7257, "step": 649 }, { "epoch": 0.07453273707143676, "grad_norm": 10.9375, "learning_rate": 9.94844439644146e-06, "loss": 1.6799, "step": 650 }, { "epoch": 0.07464740282077743, "grad_norm": 12.1875, "learning_rate": 9.948178076772868e-06, "loss": 1.9141, "step": 651 }, { "epoch": 0.0747620685701181, "grad_norm": 12.8125, "learning_rate": 9.947911074597428e-06, "loss": 1.8081, "step": 652 }, { "epoch": 0.07487673431945878, "grad_norm": 11.5625, "learning_rate": 9.947643389951974e-06, "loss": 1.8019, "step": 653 }, { "epoch": 0.07499140006879945, "grad_norm": 10.75, "learning_rate": 9.947375022873422e-06, "loss": 1.8304, "step": 654 }, { "epoch": 0.07510606581814012, "grad_norm": 9.9375, "learning_rate": 9.947105973398795e-06, "loss": 1.7475, "step": 655 }, { "epoch": 0.0752207315674808, "grad_norm": 13.625, "learning_rate": 9.946836241565197e-06, "loss": 1.9689, "step": 656 }, { "epoch": 0.07533539731682147, "grad_norm": 13.9375, "learning_rate": 9.946565827409833e-06, "loss": 2.0997, "step": 657 }, { "epoch": 0.07545006306616214, "grad_norm": 11.3125, "learning_rate": 9.946294730970006e-06, "loss": 1.5449, "step": 658 }, { "epoch": 0.07556472881550282, "grad_norm": 13.75, "learning_rate": 9.946022952283106e-06, "loss": 1.7619, "step": 659 }, { "epoch": 0.07567939456484347, "grad_norm": 15.5625, "learning_rate": 9.945750491386617e-06, "loss": 1.6104, "step": 660 }, { "epoch": 0.07579406031418415, "grad_norm": 22.0, "learning_rate": 9.945477348318124e-06, "loss": 1.5747, "step": 661 }, { "epoch": 0.07590872606352482, "grad_norm": 12.125, "learning_rate": 9.945203523115301e-06, "loss": 2.4031, "step": 662 }, { "epoch": 0.07602339181286549, "grad_norm": 10.5625, "learning_rate": 9.944929015815914e-06, "loss": 1.6387, "step": 663 }, { "epoch": 0.07613805756220617, "grad_norm": 14.25, "learning_rate": 9.944653826457829e-06, "loss": 2.1497, "step": 664 }, { "epoch": 0.07625272331154684, "grad_norm": 11.9375, "learning_rate": 9.944377955079003e-06, "loss": 1.4971, "step": 665 }, { "epoch": 0.07636738906088751, "grad_norm": 14.5625, "learning_rate": 9.944101401717487e-06, "loss": 1.9554, "step": 666 }, { "epoch": 0.07648205481022818, "grad_norm": 10.75, "learning_rate": 9.943824166411424e-06, "loss": 2.38, "step": 667 }, { "epoch": 0.07659672055956886, "grad_norm": 11.5625, "learning_rate": 9.943546249199056e-06, "loss": 1.9202, "step": 668 }, { "epoch": 0.07671138630890953, "grad_norm": 22.25, "learning_rate": 9.943267650118717e-06, "loss": 2.0283, "step": 669 }, { "epoch": 0.0768260520582502, "grad_norm": 13.75, "learning_rate": 9.942988369208831e-06, "loss": 2.2903, "step": 670 }, { "epoch": 0.07694071780759087, "grad_norm": 14.1875, "learning_rate": 9.942708406507922e-06, "loss": 2.1468, "step": 671 }, { "epoch": 0.07705538355693155, "grad_norm": 12.75, "learning_rate": 9.942427762054604e-06, "loss": 2.0188, "step": 672 }, { "epoch": 0.07717004930627222, "grad_norm": 27.625, "learning_rate": 9.942146435887588e-06, "loss": 2.2767, "step": 673 }, { "epoch": 0.0772847150556129, "grad_norm": 11.875, "learning_rate": 9.941864428045678e-06, "loss": 2.0847, "step": 674 }, { "epoch": 0.07739938080495357, "grad_norm": 11.5625, "learning_rate": 9.941581738567769e-06, "loss": 1.633, "step": 675 }, { "epoch": 0.07751404655429424, "grad_norm": 10.6875, "learning_rate": 9.941298367492856e-06, "loss": 1.6818, "step": 676 }, { "epoch": 0.0776287123036349, "grad_norm": 13.1875, "learning_rate": 9.941014314860022e-06, "loss": 2.2683, "step": 677 }, { "epoch": 0.07774337805297557, "grad_norm": 13.5625, "learning_rate": 9.940729580708449e-06, "loss": 1.6607, "step": 678 }, { "epoch": 0.07785804380231624, "grad_norm": 10.5, "learning_rate": 9.940444165077407e-06, "loss": 1.717, "step": 679 }, { "epoch": 0.07797270955165692, "grad_norm": 11.125, "learning_rate": 9.940158068006266e-06, "loss": 1.9445, "step": 680 }, { "epoch": 0.07808737530099759, "grad_norm": 10.125, "learning_rate": 9.939871289534488e-06, "loss": 1.6939, "step": 681 }, { "epoch": 0.07820204105033826, "grad_norm": 9.375, "learning_rate": 9.939583829701629e-06, "loss": 1.4931, "step": 682 }, { "epoch": 0.07831670679967893, "grad_norm": 13.5, "learning_rate": 9.939295688547336e-06, "loss": 1.8477, "step": 683 }, { "epoch": 0.0784313725490196, "grad_norm": 11.1875, "learning_rate": 9.939006866111357e-06, "loss": 1.3672, "step": 684 }, { "epoch": 0.07854603829836028, "grad_norm": 12.1875, "learning_rate": 9.938717362433526e-06, "loss": 1.8827, "step": 685 }, { "epoch": 0.07866070404770095, "grad_norm": 11.0625, "learning_rate": 9.938427177553773e-06, "loss": 1.6818, "step": 686 }, { "epoch": 0.07877536979704163, "grad_norm": 14.0625, "learning_rate": 9.938136311512127e-06, "loss": 1.4832, "step": 687 }, { "epoch": 0.0788900355463823, "grad_norm": 16.375, "learning_rate": 9.937844764348707e-06, "loss": 2.1131, "step": 688 }, { "epoch": 0.07900470129572297, "grad_norm": 12.6875, "learning_rate": 9.937552536103727e-06, "loss": 2.0105, "step": 689 }, { "epoch": 0.07911936704506364, "grad_norm": 15.4375, "learning_rate": 9.93725962681749e-06, "loss": 2.3682, "step": 690 }, { "epoch": 0.07923403279440432, "grad_norm": 12.0625, "learning_rate": 9.936966036530402e-06, "loss": 1.8501, "step": 691 }, { "epoch": 0.07934869854374499, "grad_norm": 13.125, "learning_rate": 9.936671765282956e-06, "loss": 2.3594, "step": 692 }, { "epoch": 0.07946336429308566, "grad_norm": 12.375, "learning_rate": 9.936376813115742e-06, "loss": 1.8716, "step": 693 }, { "epoch": 0.07957803004242632, "grad_norm": 17.25, "learning_rate": 9.936081180069441e-06, "loss": 2.0732, "step": 694 }, { "epoch": 0.079692695791767, "grad_norm": 13.125, "learning_rate": 9.935784866184833e-06, "loss": 2.2817, "step": 695 }, { "epoch": 0.07980736154110767, "grad_norm": 10.25, "learning_rate": 9.935487871502787e-06, "loss": 1.6045, "step": 696 }, { "epoch": 0.07992202729044834, "grad_norm": 9.3125, "learning_rate": 9.935190196064268e-06, "loss": 1.8362, "step": 697 }, { "epoch": 0.08003669303978901, "grad_norm": 25.25, "learning_rate": 9.934891839910334e-06, "loss": 1.5552, "step": 698 }, { "epoch": 0.08015135878912968, "grad_norm": 12.875, "learning_rate": 9.934592803082138e-06, "loss": 1.7742, "step": 699 }, { "epoch": 0.08026602453847036, "grad_norm": 11.0, "learning_rate": 9.93429308562093e-06, "loss": 1.8961, "step": 700 }, { "epoch": 0.08038069028781103, "grad_norm": 10.6875, "learning_rate": 9.933992687568044e-06, "loss": 1.9908, "step": 701 }, { "epoch": 0.0804953560371517, "grad_norm": 11.8125, "learning_rate": 9.933691608964917e-06, "loss": 2.0557, "step": 702 }, { "epoch": 0.08061002178649238, "grad_norm": 11.5, "learning_rate": 9.93338984985308e-06, "loss": 1.4725, "step": 703 }, { "epoch": 0.08072468753583305, "grad_norm": 11.625, "learning_rate": 9.93308741027415e-06, "loss": 1.8192, "step": 704 }, { "epoch": 0.08083935328517372, "grad_norm": 7.6875, "learning_rate": 9.932784290269845e-06, "loss": 1.6705, "step": 705 }, { "epoch": 0.0809540190345144, "grad_norm": 11.25, "learning_rate": 9.932480489881975e-06, "loss": 2.0448, "step": 706 }, { "epoch": 0.08106868478385507, "grad_norm": 13.375, "learning_rate": 9.932176009152442e-06, "loss": 1.8885, "step": 707 }, { "epoch": 0.08118335053319574, "grad_norm": 10.375, "learning_rate": 9.931870848123247e-06, "loss": 1.9335, "step": 708 }, { "epoch": 0.08129801628253641, "grad_norm": 11.9375, "learning_rate": 9.931565006836475e-06, "loss": 1.6549, "step": 709 }, { "epoch": 0.08141268203187708, "grad_norm": 10.6875, "learning_rate": 9.931258485334316e-06, "loss": 1.9211, "step": 710 }, { "epoch": 0.08152734778121776, "grad_norm": 10.25, "learning_rate": 9.930951283659048e-06, "loss": 1.7819, "step": 711 }, { "epoch": 0.08164201353055842, "grad_norm": 13.4375, "learning_rate": 9.930643401853042e-06, "loss": 1.9528, "step": 712 }, { "epoch": 0.08175667927989909, "grad_norm": 12.5, "learning_rate": 9.930334839958767e-06, "loss": 1.6055, "step": 713 }, { "epoch": 0.08187134502923976, "grad_norm": 12.5, "learning_rate": 9.93002559801878e-06, "loss": 1.7837, "step": 714 }, { "epoch": 0.08198601077858043, "grad_norm": 9.9375, "learning_rate": 9.929715676075737e-06, "loss": 1.7498, "step": 715 }, { "epoch": 0.08210067652792111, "grad_norm": 13.25, "learning_rate": 9.929405074172385e-06, "loss": 2.0494, "step": 716 }, { "epoch": 0.08221534227726178, "grad_norm": 11.25, "learning_rate": 9.929093792351567e-06, "loss": 1.7821, "step": 717 }, { "epoch": 0.08233000802660245, "grad_norm": 13.4375, "learning_rate": 9.928781830656216e-06, "loss": 1.9111, "step": 718 }, { "epoch": 0.08244467377594313, "grad_norm": 31.875, "learning_rate": 9.928469189129364e-06, "loss": 1.9428, "step": 719 }, { "epoch": 0.0825593395252838, "grad_norm": 9.375, "learning_rate": 9.928155867814131e-06, "loss": 1.6083, "step": 720 }, { "epoch": 0.08267400527462447, "grad_norm": 11.9375, "learning_rate": 9.927841866753735e-06, "loss": 2.0548, "step": 721 }, { "epoch": 0.08278867102396514, "grad_norm": 10.875, "learning_rate": 9.92752718599149e-06, "loss": 1.8038, "step": 722 }, { "epoch": 0.08290333677330582, "grad_norm": 12.9375, "learning_rate": 9.927211825570793e-06, "loss": 2.1265, "step": 723 }, { "epoch": 0.08301800252264649, "grad_norm": 11.5, "learning_rate": 9.92689578553515e-06, "loss": 1.7912, "step": 724 }, { "epoch": 0.08313266827198716, "grad_norm": 14.5625, "learning_rate": 9.926579065928145e-06, "loss": 1.7986, "step": 725 }, { "epoch": 0.08324733402132783, "grad_norm": 9.1875, "learning_rate": 9.926261666793469e-06, "loss": 1.8359, "step": 726 }, { "epoch": 0.08336199977066851, "grad_norm": 12.25, "learning_rate": 9.925943588174898e-06, "loss": 2.1023, "step": 727 }, { "epoch": 0.08347666552000918, "grad_norm": 12.8125, "learning_rate": 9.925624830116304e-06, "loss": 1.9266, "step": 728 }, { "epoch": 0.08359133126934984, "grad_norm": 13.625, "learning_rate": 9.925305392661659e-06, "loss": 1.9049, "step": 729 }, { "epoch": 0.08370599701869051, "grad_norm": 13.3125, "learning_rate": 9.924985275855019e-06, "loss": 1.8787, "step": 730 }, { "epoch": 0.08382066276803118, "grad_norm": 14.625, "learning_rate": 9.92466447974054e-06, "loss": 1.7231, "step": 731 }, { "epoch": 0.08393532851737186, "grad_norm": 13.3125, "learning_rate": 9.924343004362467e-06, "loss": 1.7234, "step": 732 }, { "epoch": 0.08404999426671253, "grad_norm": 11.875, "learning_rate": 9.924020849765143e-06, "loss": 1.9803, "step": 733 }, { "epoch": 0.0841646600160532, "grad_norm": 10.25, "learning_rate": 9.923698015993004e-06, "loss": 1.7964, "step": 734 }, { "epoch": 0.08427932576539388, "grad_norm": 52.75, "learning_rate": 9.923374503090578e-06, "loss": 2.2585, "step": 735 }, { "epoch": 0.08439399151473455, "grad_norm": 9.5, "learning_rate": 9.923050311102487e-06, "loss": 1.8325, "step": 736 }, { "epoch": 0.08450865726407522, "grad_norm": 10.625, "learning_rate": 9.922725440073448e-06, "loss": 2.0818, "step": 737 }, { "epoch": 0.0846233230134159, "grad_norm": 13.75, "learning_rate": 9.92239989004827e-06, "loss": 1.7715, "step": 738 }, { "epoch": 0.08473798876275657, "grad_norm": 11.6875, "learning_rate": 9.922073661071857e-06, "loss": 1.6961, "step": 739 }, { "epoch": 0.08485265451209724, "grad_norm": 15.1875, "learning_rate": 9.921746753189202e-06, "loss": 2.201, "step": 740 }, { "epoch": 0.08496732026143791, "grad_norm": 27.375, "learning_rate": 9.921419166445405e-06, "loss": 1.6559, "step": 741 }, { "epoch": 0.08508198601077858, "grad_norm": 11.875, "learning_rate": 9.921090900885642e-06, "loss": 1.752, "step": 742 }, { "epoch": 0.08519665176011926, "grad_norm": 15.125, "learning_rate": 9.920761956555193e-06, "loss": 1.3775, "step": 743 }, { "epoch": 0.08531131750945993, "grad_norm": 12.375, "learning_rate": 9.920432333499433e-06, "loss": 1.855, "step": 744 }, { "epoch": 0.0854259832588006, "grad_norm": 14.1875, "learning_rate": 9.920102031763823e-06, "loss": 1.8976, "step": 745 }, { "epoch": 0.08554064900814126, "grad_norm": 12.0, "learning_rate": 9.919771051393922e-06, "loss": 1.7405, "step": 746 }, { "epoch": 0.08565531475748193, "grad_norm": 11.5625, "learning_rate": 9.919439392435386e-06, "loss": 1.9019, "step": 747 }, { "epoch": 0.08576998050682261, "grad_norm": 10.4375, "learning_rate": 9.919107054933958e-06, "loss": 1.9953, "step": 748 }, { "epoch": 0.08588464625616328, "grad_norm": 14.75, "learning_rate": 9.918774038935478e-06, "loss": 1.5472, "step": 749 }, { "epoch": 0.08599931200550395, "grad_norm": 11.25, "learning_rate": 9.91844034448588e-06, "loss": 1.4942, "step": 750 }, { "epoch": 0.08611397775484463, "grad_norm": 10.6875, "learning_rate": 9.918105971631192e-06, "loss": 1.6777, "step": 751 }, { "epoch": 0.0862286435041853, "grad_norm": 11.0, "learning_rate": 9.91777092041753e-06, "loss": 1.4665, "step": 752 }, { "epoch": 0.08634330925352597, "grad_norm": 13.0625, "learning_rate": 9.917435190891113e-06, "loss": 2.1315, "step": 753 }, { "epoch": 0.08645797500286664, "grad_norm": 13.625, "learning_rate": 9.917098783098245e-06, "loss": 2.311, "step": 754 }, { "epoch": 0.08657264075220732, "grad_norm": 12.5, "learning_rate": 9.916761697085328e-06, "loss": 1.7066, "step": 755 }, { "epoch": 0.08668730650154799, "grad_norm": 19.25, "learning_rate": 9.916423932898859e-06, "loss": 1.7103, "step": 756 }, { "epoch": 0.08680197225088866, "grad_norm": 12.0, "learning_rate": 9.916085490585424e-06, "loss": 1.7241, "step": 757 }, { "epoch": 0.08691663800022933, "grad_norm": 13.3125, "learning_rate": 9.915746370191702e-06, "loss": 1.8878, "step": 758 }, { "epoch": 0.08703130374957001, "grad_norm": 13.375, "learning_rate": 9.915406571764473e-06, "loss": 1.9504, "step": 759 }, { "epoch": 0.08714596949891068, "grad_norm": 8.875, "learning_rate": 9.915066095350604e-06, "loss": 1.539, "step": 760 }, { "epoch": 0.08726063524825135, "grad_norm": 9.4375, "learning_rate": 9.914724940997054e-06, "loss": 2.0397, "step": 761 }, { "epoch": 0.08737530099759203, "grad_norm": 10.0, "learning_rate": 9.914383108750883e-06, "loss": 2.1397, "step": 762 }, { "epoch": 0.08748996674693268, "grad_norm": 12.0, "learning_rate": 9.914040598659241e-06, "loss": 1.8368, "step": 763 }, { "epoch": 0.08760463249627336, "grad_norm": 13.1875, "learning_rate": 9.913697410769364e-06, "loss": 1.4995, "step": 764 }, { "epoch": 0.08771929824561403, "grad_norm": 12.6875, "learning_rate": 9.913353545128596e-06, "loss": 1.5319, "step": 765 }, { "epoch": 0.0878339639949547, "grad_norm": 12.0625, "learning_rate": 9.913009001784364e-06, "loss": 1.7464, "step": 766 }, { "epoch": 0.08794862974429538, "grad_norm": 11.125, "learning_rate": 9.91266378078419e-06, "loss": 1.9023, "step": 767 }, { "epoch": 0.08806329549363605, "grad_norm": 12.125, "learning_rate": 9.91231788217569e-06, "loss": 1.7141, "step": 768 }, { "epoch": 0.08817796124297672, "grad_norm": 13.9375, "learning_rate": 9.911971306006576e-06, "loss": 2.0365, "step": 769 }, { "epoch": 0.0882926269923174, "grad_norm": 24.125, "learning_rate": 9.911624052324651e-06, "loss": 2.1333, "step": 770 }, { "epoch": 0.08840729274165807, "grad_norm": 11.25, "learning_rate": 9.911276121177813e-06, "loss": 1.5783, "step": 771 }, { "epoch": 0.08852195849099874, "grad_norm": 12.375, "learning_rate": 9.910927512614052e-06, "loss": 1.7413, "step": 772 }, { "epoch": 0.08863662424033941, "grad_norm": 12.375, "learning_rate": 9.91057822668145e-06, "loss": 1.8985, "step": 773 }, { "epoch": 0.08875128998968008, "grad_norm": 10.1875, "learning_rate": 9.910228263428187e-06, "loss": 1.7264, "step": 774 }, { "epoch": 0.08886595573902076, "grad_norm": 10.3125, "learning_rate": 9.90987762290253e-06, "loss": 1.7397, "step": 775 }, { "epoch": 0.08898062148836143, "grad_norm": 11.6875, "learning_rate": 9.909526305152849e-06, "loss": 1.6624, "step": 776 }, { "epoch": 0.0890952872377021, "grad_norm": 10.625, "learning_rate": 9.909174310227596e-06, "loss": 1.626, "step": 777 }, { "epoch": 0.08920995298704278, "grad_norm": 32.0, "learning_rate": 9.908821638175325e-06, "loss": 2.1957, "step": 778 }, { "epoch": 0.08932461873638345, "grad_norm": 12.3125, "learning_rate": 9.90846828904468e-06, "loss": 1.59, "step": 779 }, { "epoch": 0.08943928448572411, "grad_norm": 12.6875, "learning_rate": 9.9081142628844e-06, "loss": 1.6171, "step": 780 }, { "epoch": 0.08955395023506478, "grad_norm": 24.875, "learning_rate": 9.90775955974331e-06, "loss": 1.7722, "step": 781 }, { "epoch": 0.08966861598440545, "grad_norm": 8.625, "learning_rate": 9.907404179670344e-06, "loss": 1.672, "step": 782 }, { "epoch": 0.08978328173374613, "grad_norm": 10.875, "learning_rate": 9.907048122714511e-06, "loss": 1.5255, "step": 783 }, { "epoch": 0.0898979474830868, "grad_norm": 11.0, "learning_rate": 9.906691388924929e-06, "loss": 1.7998, "step": 784 }, { "epoch": 0.09001261323242747, "grad_norm": 15.0625, "learning_rate": 9.9063339783508e-06, "loss": 1.9517, "step": 785 }, { "epoch": 0.09012727898176814, "grad_norm": 39.5, "learning_rate": 9.90597589104142e-06, "loss": 2.1514, "step": 786 }, { "epoch": 0.09024194473110882, "grad_norm": 14.0, "learning_rate": 9.905617127046182e-06, "loss": 1.9877, "step": 787 }, { "epoch": 0.09035661048044949, "grad_norm": 11.5625, "learning_rate": 9.905257686414573e-06, "loss": 1.7784, "step": 788 }, { "epoch": 0.09047127622979016, "grad_norm": 12.8125, "learning_rate": 9.904897569196168e-06, "loss": 2.0523, "step": 789 }, { "epoch": 0.09058594197913084, "grad_norm": 12.375, "learning_rate": 9.90453677544064e-06, "loss": 2.2256, "step": 790 }, { "epoch": 0.09070060772847151, "grad_norm": 11.125, "learning_rate": 9.904175305197752e-06, "loss": 1.8729, "step": 791 }, { "epoch": 0.09081527347781218, "grad_norm": 12.0, "learning_rate": 9.903813158517363e-06, "loss": 1.6559, "step": 792 }, { "epoch": 0.09092993922715285, "grad_norm": 22.0, "learning_rate": 9.903450335449424e-06, "loss": 1.7439, "step": 793 }, { "epoch": 0.09104460497649353, "grad_norm": 12.1875, "learning_rate": 9.903086836043979e-06, "loss": 1.542, "step": 794 }, { "epoch": 0.0911592707258342, "grad_norm": 13.8125, "learning_rate": 9.902722660351168e-06, "loss": 1.684, "step": 795 }, { "epoch": 0.09127393647517487, "grad_norm": 15.75, "learning_rate": 9.902357808421217e-06, "loss": 1.7065, "step": 796 }, { "epoch": 0.09138860222451553, "grad_norm": 13.25, "learning_rate": 9.901992280304457e-06, "loss": 1.9403, "step": 797 }, { "epoch": 0.0915032679738562, "grad_norm": 11.4375, "learning_rate": 9.9016260760513e-06, "loss": 1.6252, "step": 798 }, { "epoch": 0.09161793372319688, "grad_norm": 13.25, "learning_rate": 9.90125919571226e-06, "loss": 1.68, "step": 799 }, { "epoch": 0.09173259947253755, "grad_norm": 13.3125, "learning_rate": 9.900891639337942e-06, "loss": 1.7155, "step": 800 }, { "epoch": 0.09184726522187822, "grad_norm": 20.75, "learning_rate": 9.900523406979042e-06, "loss": 1.7646, "step": 801 }, { "epoch": 0.0919619309712189, "grad_norm": 13.0, "learning_rate": 9.900154498686348e-06, "loss": 1.7842, "step": 802 }, { "epoch": 0.09207659672055957, "grad_norm": 12.8125, "learning_rate": 9.899784914510749e-06, "loss": 1.9163, "step": 803 }, { "epoch": 0.09219126246990024, "grad_norm": 10.25, "learning_rate": 9.899414654503216e-06, "loss": 1.3768, "step": 804 }, { "epoch": 0.09230592821924091, "grad_norm": 19.875, "learning_rate": 9.899043718714825e-06, "loss": 2.1125, "step": 805 }, { "epoch": 0.09242059396858159, "grad_norm": 11.1875, "learning_rate": 9.898672107196738e-06, "loss": 1.9024, "step": 806 }, { "epoch": 0.09253525971792226, "grad_norm": 15.1875, "learning_rate": 9.89829982000021e-06, "loss": 1.6933, "step": 807 }, { "epoch": 0.09264992546726293, "grad_norm": 11.0, "learning_rate": 9.897926857176592e-06, "loss": 1.6883, "step": 808 }, { "epoch": 0.0927645912166036, "grad_norm": 13.125, "learning_rate": 9.897553218777327e-06, "loss": 1.7638, "step": 809 }, { "epoch": 0.09287925696594428, "grad_norm": 10.5625, "learning_rate": 9.897178904853951e-06, "loss": 1.6385, "step": 810 }, { "epoch": 0.09299392271528495, "grad_norm": 17.5, "learning_rate": 9.896803915458095e-06, "loss": 2.4461, "step": 811 }, { "epoch": 0.09310858846462562, "grad_norm": 12.5625, "learning_rate": 9.89642825064148e-06, "loss": 1.8764, "step": 812 }, { "epoch": 0.0932232542139663, "grad_norm": 18.375, "learning_rate": 9.89605191045592e-06, "loss": 1.9873, "step": 813 }, { "epoch": 0.09333791996330695, "grad_norm": 12.125, "learning_rate": 9.89567489495333e-06, "loss": 1.8222, "step": 814 }, { "epoch": 0.09345258571264763, "grad_norm": 28.75, "learning_rate": 9.895297204185705e-06, "loss": 1.8913, "step": 815 }, { "epoch": 0.0935672514619883, "grad_norm": 12.6875, "learning_rate": 9.894918838205146e-06, "loss": 2.0227, "step": 816 }, { "epoch": 0.09368191721132897, "grad_norm": 12.0625, "learning_rate": 9.894539797063838e-06, "loss": 1.6119, "step": 817 }, { "epoch": 0.09379658296066964, "grad_norm": 15.5625, "learning_rate": 9.894160080814061e-06, "loss": 1.906, "step": 818 }, { "epoch": 0.09391124871001032, "grad_norm": 10.875, "learning_rate": 9.893779689508195e-06, "loss": 1.8459, "step": 819 }, { "epoch": 0.09402591445935099, "grad_norm": 18.375, "learning_rate": 9.893398623198704e-06, "loss": 1.7479, "step": 820 }, { "epoch": 0.09414058020869166, "grad_norm": 12.5625, "learning_rate": 9.893016881938149e-06, "loss": 2.1851, "step": 821 }, { "epoch": 0.09425524595803234, "grad_norm": 15.25, "learning_rate": 9.892634465779185e-06, "loss": 2.3796, "step": 822 }, { "epoch": 0.09436991170737301, "grad_norm": 12.8125, "learning_rate": 9.89225137477456e-06, "loss": 1.7924, "step": 823 }, { "epoch": 0.09448457745671368, "grad_norm": 14.3125, "learning_rate": 9.89186760897711e-06, "loss": 1.7825, "step": 824 }, { "epoch": 0.09459924320605435, "grad_norm": 13.1875, "learning_rate": 9.891483168439773e-06, "loss": 1.9794, "step": 825 }, { "epoch": 0.09471390895539503, "grad_norm": 11.9375, "learning_rate": 9.89109805321557e-06, "loss": 2.0751, "step": 826 }, { "epoch": 0.0948285747047357, "grad_norm": 13.4375, "learning_rate": 9.890712263357627e-06, "loss": 2.0294, "step": 827 }, { "epoch": 0.09494324045407637, "grad_norm": 13.5, "learning_rate": 9.89032579891915e-06, "loss": 1.8756, "step": 828 }, { "epoch": 0.09505790620341704, "grad_norm": 11.25, "learning_rate": 9.88993865995345e-06, "loss": 1.8753, "step": 829 }, { "epoch": 0.09517257195275772, "grad_norm": 28.625, "learning_rate": 9.88955084651392e-06, "loss": 1.7181, "step": 830 }, { "epoch": 0.09528723770209838, "grad_norm": 19.375, "learning_rate": 9.889162358654056e-06, "loss": 2.2937, "step": 831 }, { "epoch": 0.09540190345143905, "grad_norm": 17.5, "learning_rate": 9.888773196427439e-06, "loss": 2.1523, "step": 832 }, { "epoch": 0.09551656920077972, "grad_norm": 10.8125, "learning_rate": 9.888383359887748e-06, "loss": 1.6819, "step": 833 }, { "epoch": 0.0956312349501204, "grad_norm": 12.75, "learning_rate": 9.887992849088754e-06, "loss": 1.9824, "step": 834 }, { "epoch": 0.09574590069946107, "grad_norm": 9.0625, "learning_rate": 9.887601664084321e-06, "loss": 1.8636, "step": 835 }, { "epoch": 0.09586056644880174, "grad_norm": 10.25, "learning_rate": 9.887209804928404e-06, "loss": 1.9221, "step": 836 }, { "epoch": 0.09597523219814241, "grad_norm": 14.375, "learning_rate": 9.886817271675054e-06, "loss": 1.8621, "step": 837 }, { "epoch": 0.09608989794748309, "grad_norm": 12.1875, "learning_rate": 9.88642406437841e-06, "loss": 1.8124, "step": 838 }, { "epoch": 0.09620456369682376, "grad_norm": 12.3125, "learning_rate": 9.886030183092713e-06, "loss": 1.7167, "step": 839 }, { "epoch": 0.09631922944616443, "grad_norm": 23.25, "learning_rate": 9.885635627872286e-06, "loss": 1.8925, "step": 840 }, { "epoch": 0.0964338951955051, "grad_norm": 10.8125, "learning_rate": 9.885240398771554e-06, "loss": 1.6938, "step": 841 }, { "epoch": 0.09654856094484578, "grad_norm": 13.375, "learning_rate": 9.88484449584503e-06, "loss": 1.6496, "step": 842 }, { "epoch": 0.09666322669418645, "grad_norm": 11.8125, "learning_rate": 9.884447919147322e-06, "loss": 1.8973, "step": 843 }, { "epoch": 0.09677789244352712, "grad_norm": 10.0, "learning_rate": 9.884050668733127e-06, "loss": 1.7394, "step": 844 }, { "epoch": 0.0968925581928678, "grad_norm": 15.25, "learning_rate": 9.883652744657244e-06, "loss": 2.7015, "step": 845 }, { "epoch": 0.09700722394220847, "grad_norm": 13.5625, "learning_rate": 9.883254146974555e-06, "loss": 1.9589, "step": 846 }, { "epoch": 0.09712188969154914, "grad_norm": 10.25, "learning_rate": 9.882854875740038e-06, "loss": 1.7673, "step": 847 }, { "epoch": 0.09723655544088981, "grad_norm": 14.5, "learning_rate": 9.882454931008768e-06, "loss": 1.7018, "step": 848 }, { "epoch": 0.09735122119023047, "grad_norm": 26.625, "learning_rate": 9.882054312835909e-06, "loss": 2.0728, "step": 849 }, { "epoch": 0.09746588693957114, "grad_norm": 12.625, "learning_rate": 9.881653021276716e-06, "loss": 1.9629, "step": 850 }, { "epoch": 0.09758055268891182, "grad_norm": 14.9375, "learning_rate": 9.881251056386542e-06, "loss": 2.022, "step": 851 }, { "epoch": 0.09769521843825249, "grad_norm": 13.1875, "learning_rate": 9.88084841822083e-06, "loss": 1.9566, "step": 852 }, { "epoch": 0.09780988418759316, "grad_norm": 11.125, "learning_rate": 9.880445106835116e-06, "loss": 1.6669, "step": 853 }, { "epoch": 0.09792454993693384, "grad_norm": 15.5, "learning_rate": 9.880041122285031e-06, "loss": 1.8021, "step": 854 }, { "epoch": 0.09803921568627451, "grad_norm": 10.875, "learning_rate": 9.879636464626294e-06, "loss": 1.7904, "step": 855 }, { "epoch": 0.09815388143561518, "grad_norm": 10.3125, "learning_rate": 9.879231133914721e-06, "loss": 1.6853, "step": 856 }, { "epoch": 0.09826854718495585, "grad_norm": 11.375, "learning_rate": 9.878825130206219e-06, "loss": 1.8292, "step": 857 }, { "epoch": 0.09838321293429653, "grad_norm": 25.625, "learning_rate": 9.878418453556791e-06, "loss": 1.6397, "step": 858 }, { "epoch": 0.0984978786836372, "grad_norm": 12.8125, "learning_rate": 9.878011104022527e-06, "loss": 1.6552, "step": 859 }, { "epoch": 0.09861254443297787, "grad_norm": 12.625, "learning_rate": 9.877603081659614e-06, "loss": 2.1377, "step": 860 }, { "epoch": 0.09872721018231854, "grad_norm": 10.4375, "learning_rate": 9.877194386524334e-06, "loss": 1.7511, "step": 861 }, { "epoch": 0.09884187593165922, "grad_norm": 11.625, "learning_rate": 9.876785018673054e-06, "loss": 1.7329, "step": 862 }, { "epoch": 0.09895654168099989, "grad_norm": 11.125, "learning_rate": 9.876374978162241e-06, "loss": 1.6056, "step": 863 }, { "epoch": 0.09907120743034056, "grad_norm": 11.75, "learning_rate": 9.875964265048452e-06, "loss": 1.7598, "step": 864 }, { "epoch": 0.09918587317968124, "grad_norm": 12.4375, "learning_rate": 9.875552879388336e-06, "loss": 1.6625, "step": 865 }, { "epoch": 0.0993005389290219, "grad_norm": 12.9375, "learning_rate": 9.87514082123864e-06, "loss": 1.6378, "step": 866 }, { "epoch": 0.09941520467836257, "grad_norm": 12.25, "learning_rate": 9.874728090656194e-06, "loss": 2.1577, "step": 867 }, { "epoch": 0.09952987042770324, "grad_norm": 11.875, "learning_rate": 9.874314687697928e-06, "loss": 2.0477, "step": 868 }, { "epoch": 0.09964453617704391, "grad_norm": 10.375, "learning_rate": 9.873900612420866e-06, "loss": 1.8378, "step": 869 }, { "epoch": 0.09975920192638459, "grad_norm": 13.25, "learning_rate": 9.873485864882117e-06, "loss": 1.6425, "step": 870 }, { "epoch": 0.09987386767572526, "grad_norm": 11.6875, "learning_rate": 9.873070445138892e-06, "loss": 1.7883, "step": 871 }, { "epoch": 0.09998853342506593, "grad_norm": 12.5, "learning_rate": 9.872654353248488e-06, "loss": 1.6057, "step": 872 }, { "epoch": 0.1001031991744066, "grad_norm": 12.375, "learning_rate": 9.872237589268295e-06, "loss": 1.5762, "step": 873 }, { "epoch": 0.10021786492374728, "grad_norm": 13.3125, "learning_rate": 9.871820153255803e-06, "loss": 2.0265, "step": 874 }, { "epoch": 0.10033253067308795, "grad_norm": 11.75, "learning_rate": 9.871402045268584e-06, "loss": 1.6662, "step": 875 }, { "epoch": 0.10044719642242862, "grad_norm": 24.0, "learning_rate": 9.87098326536431e-06, "loss": 1.9106, "step": 876 }, { "epoch": 0.1005618621717693, "grad_norm": 13.25, "learning_rate": 9.870563813600745e-06, "loss": 1.7769, "step": 877 }, { "epoch": 0.10067652792110997, "grad_norm": 13.25, "learning_rate": 9.870143690035743e-06, "loss": 1.9684, "step": 878 }, { "epoch": 0.10079119367045064, "grad_norm": 12.3125, "learning_rate": 9.869722894727253e-06, "loss": 1.6895, "step": 879 }, { "epoch": 0.10090585941979131, "grad_norm": 13.125, "learning_rate": 9.869301427733314e-06, "loss": 1.561, "step": 880 }, { "epoch": 0.10102052516913199, "grad_norm": 12.375, "learning_rate": 9.868879289112061e-06, "loss": 1.7705, "step": 881 }, { "epoch": 0.10113519091847266, "grad_norm": 12.875, "learning_rate": 9.86845647892172e-06, "loss": 1.963, "step": 882 }, { "epoch": 0.10124985666781332, "grad_norm": 14.125, "learning_rate": 9.868032997220608e-06, "loss": 1.5928, "step": 883 }, { "epoch": 0.10136452241715399, "grad_norm": 11.25, "learning_rate": 9.867608844067136e-06, "loss": 1.9542, "step": 884 }, { "epoch": 0.10147918816649466, "grad_norm": 11.625, "learning_rate": 9.86718401951981e-06, "loss": 1.9701, "step": 885 }, { "epoch": 0.10159385391583534, "grad_norm": 11.4375, "learning_rate": 9.866758523637227e-06, "loss": 1.4916, "step": 886 }, { "epoch": 0.10170851966517601, "grad_norm": 12.3125, "learning_rate": 9.866332356478075e-06, "loss": 2.3928, "step": 887 }, { "epoch": 0.10182318541451668, "grad_norm": 12.4375, "learning_rate": 9.865905518101131e-06, "loss": 1.6523, "step": 888 }, { "epoch": 0.10193785116385735, "grad_norm": 12.4375, "learning_rate": 9.865478008565277e-06, "loss": 1.7168, "step": 889 }, { "epoch": 0.10205251691319803, "grad_norm": 15.4375, "learning_rate": 9.865049827929476e-06, "loss": 1.8888, "step": 890 }, { "epoch": 0.1021671826625387, "grad_norm": 12.5, "learning_rate": 9.864620976252785e-06, "loss": 1.665, "step": 891 }, { "epoch": 0.10228184841187937, "grad_norm": 12.125, "learning_rate": 9.864191453594361e-06, "loss": 1.8333, "step": 892 }, { "epoch": 0.10239651416122005, "grad_norm": 13.9375, "learning_rate": 9.863761260013445e-06, "loss": 1.5713, "step": 893 }, { "epoch": 0.10251117991056072, "grad_norm": 27.5, "learning_rate": 9.863330395569375e-06, "loss": 2.227, "step": 894 }, { "epoch": 0.10262584565990139, "grad_norm": 12.8125, "learning_rate": 9.862898860321581e-06, "loss": 1.7127, "step": 895 }, { "epoch": 0.10274051140924206, "grad_norm": 13.5, "learning_rate": 9.862466654329583e-06, "loss": 1.8813, "step": 896 }, { "epoch": 0.10285517715858274, "grad_norm": 11.1875, "learning_rate": 9.862033777652997e-06, "loss": 2.0447, "step": 897 }, { "epoch": 0.10296984290792341, "grad_norm": 15.3125, "learning_rate": 9.86160023035153e-06, "loss": 1.4853, "step": 898 }, { "epoch": 0.10308450865726408, "grad_norm": 12.25, "learning_rate": 9.861166012484984e-06, "loss": 2.1858, "step": 899 }, { "epoch": 0.10319917440660474, "grad_norm": 10.5625, "learning_rate": 9.860731124113248e-06, "loss": 1.5828, "step": 900 }, { "epoch": 0.10331384015594541, "grad_norm": 11.25, "learning_rate": 9.860295565296307e-06, "loss": 2.0273, "step": 901 }, { "epoch": 0.10342850590528609, "grad_norm": 11.375, "learning_rate": 9.85985933609424e-06, "loss": 1.7435, "step": 902 }, { "epoch": 0.10354317165462676, "grad_norm": 9.9375, "learning_rate": 9.859422436567213e-06, "loss": 1.7165, "step": 903 }, { "epoch": 0.10365783740396743, "grad_norm": 13.875, "learning_rate": 9.858984866775489e-06, "loss": 2.0021, "step": 904 }, { "epoch": 0.1037725031533081, "grad_norm": 13.6875, "learning_rate": 9.858546626779425e-06, "loss": 1.6327, "step": 905 }, { "epoch": 0.10388716890264878, "grad_norm": 12.0, "learning_rate": 9.858107716639466e-06, "loss": 1.8884, "step": 906 }, { "epoch": 0.10400183465198945, "grad_norm": 12.875, "learning_rate": 9.857668136416152e-06, "loss": 1.8673, "step": 907 }, { "epoch": 0.10411650040133012, "grad_norm": 11.4375, "learning_rate": 9.857227886170113e-06, "loss": 2.0506, "step": 908 }, { "epoch": 0.1042311661506708, "grad_norm": 12.1875, "learning_rate": 9.856786965962074e-06, "loss": 1.7679, "step": 909 }, { "epoch": 0.10434583190001147, "grad_norm": 11.8125, "learning_rate": 9.856345375852853e-06, "loss": 1.6655, "step": 910 }, { "epoch": 0.10446049764935214, "grad_norm": 13.75, "learning_rate": 9.855903115903357e-06, "loss": 2.0138, "step": 911 }, { "epoch": 0.10457516339869281, "grad_norm": 10.875, "learning_rate": 9.855460186174587e-06, "loss": 1.7375, "step": 912 }, { "epoch": 0.10468982914803349, "grad_norm": 11.25, "learning_rate": 9.85501658672764e-06, "loss": 1.7699, "step": 913 }, { "epoch": 0.10480449489737416, "grad_norm": 11.6875, "learning_rate": 9.8545723176237e-06, "loss": 1.6281, "step": 914 }, { "epoch": 0.10491916064671483, "grad_norm": 13.0625, "learning_rate": 9.854127378924043e-06, "loss": 1.557, "step": 915 }, { "epoch": 0.1050338263960555, "grad_norm": 12.5625, "learning_rate": 9.853681770690044e-06, "loss": 1.495, "step": 916 }, { "epoch": 0.10514849214539616, "grad_norm": 18.125, "learning_rate": 9.853235492983163e-06, "loss": 1.8162, "step": 917 }, { "epoch": 0.10526315789473684, "grad_norm": 30.0, "learning_rate": 9.852788545864959e-06, "loss": 1.6333, "step": 918 }, { "epoch": 0.10537782364407751, "grad_norm": 25.125, "learning_rate": 9.852340929397076e-06, "loss": 1.9253, "step": 919 }, { "epoch": 0.10549248939341818, "grad_norm": 11.1875, "learning_rate": 9.851892643641256e-06, "loss": 1.645, "step": 920 }, { "epoch": 0.10560715514275885, "grad_norm": 11.6875, "learning_rate": 9.851443688659332e-06, "loss": 1.5811, "step": 921 }, { "epoch": 0.10572182089209953, "grad_norm": 14.6875, "learning_rate": 9.850994064513227e-06, "loss": 1.9013, "step": 922 }, { "epoch": 0.1058364866414402, "grad_norm": 28.875, "learning_rate": 9.85054377126496e-06, "loss": 1.9024, "step": 923 }, { "epoch": 0.10595115239078087, "grad_norm": 28.625, "learning_rate": 9.850092808976638e-06, "loss": 1.9886, "step": 924 }, { "epoch": 0.10606581814012155, "grad_norm": 11.125, "learning_rate": 9.849641177710467e-06, "loss": 1.5832, "step": 925 }, { "epoch": 0.10618048388946222, "grad_norm": 11.6875, "learning_rate": 9.849188877528736e-06, "loss": 1.8115, "step": 926 }, { "epoch": 0.10629514963880289, "grad_norm": 23.875, "learning_rate": 9.848735908493834e-06, "loss": 2.1686, "step": 927 }, { "epoch": 0.10640981538814356, "grad_norm": 29.125, "learning_rate": 9.84828227066824e-06, "loss": 2.0789, "step": 928 }, { "epoch": 0.10652448113748424, "grad_norm": 15.75, "learning_rate": 9.847827964114522e-06, "loss": 1.5565, "step": 929 }, { "epoch": 0.10663914688682491, "grad_norm": 23.0, "learning_rate": 9.847372988895343e-06, "loss": 1.8766, "step": 930 }, { "epoch": 0.10675381263616558, "grad_norm": 12.9375, "learning_rate": 9.846917345073461e-06, "loss": 1.8116, "step": 931 }, { "epoch": 0.10686847838550625, "grad_norm": 12.8125, "learning_rate": 9.846461032711723e-06, "loss": 2.0186, "step": 932 }, { "epoch": 0.10698314413484693, "grad_norm": 12.75, "learning_rate": 9.846004051873066e-06, "loss": 1.6542, "step": 933 }, { "epoch": 0.10709780988418759, "grad_norm": 12.375, "learning_rate": 9.845546402620523e-06, "loss": 1.8195, "step": 934 }, { "epoch": 0.10721247563352826, "grad_norm": 32.0, "learning_rate": 9.845088085017217e-06, "loss": 1.8511, "step": 935 }, { "epoch": 0.10732714138286893, "grad_norm": 10.875, "learning_rate": 9.84462909912637e-06, "loss": 1.8686, "step": 936 }, { "epoch": 0.1074418071322096, "grad_norm": 13.3125, "learning_rate": 9.844169445011283e-06, "loss": 2.1787, "step": 937 }, { "epoch": 0.10755647288155028, "grad_norm": 12.75, "learning_rate": 9.843709122735359e-06, "loss": 1.6049, "step": 938 }, { "epoch": 0.10767113863089095, "grad_norm": 13.0, "learning_rate": 9.84324813236209e-06, "loss": 1.7905, "step": 939 }, { "epoch": 0.10778580438023162, "grad_norm": 11.1875, "learning_rate": 9.842786473955063e-06, "loss": 2.0873, "step": 940 }, { "epoch": 0.1079004701295723, "grad_norm": 12.0, "learning_rate": 9.842324147577955e-06, "loss": 1.9355, "step": 941 }, { "epoch": 0.10801513587891297, "grad_norm": 11.625, "learning_rate": 9.841861153294533e-06, "loss": 1.8816, "step": 942 }, { "epoch": 0.10812980162825364, "grad_norm": 27.0, "learning_rate": 9.84139749116866e-06, "loss": 1.7986, "step": 943 }, { "epoch": 0.10824446737759431, "grad_norm": 18.25, "learning_rate": 9.840933161264288e-06, "loss": 1.7214, "step": 944 }, { "epoch": 0.10835913312693499, "grad_norm": 14.9375, "learning_rate": 9.840468163645462e-06, "loss": 1.7547, "step": 945 }, { "epoch": 0.10847379887627566, "grad_norm": 12.8125, "learning_rate": 9.840002498376322e-06, "loss": 1.8159, "step": 946 }, { "epoch": 0.10858846462561633, "grad_norm": 13.8125, "learning_rate": 9.839536165521095e-06, "loss": 1.8784, "step": 947 }, { "epoch": 0.108703130374957, "grad_norm": 13.5625, "learning_rate": 9.839069165144103e-06, "loss": 1.8081, "step": 948 }, { "epoch": 0.10881779612429768, "grad_norm": 13.625, "learning_rate": 9.838601497309762e-06, "loss": 1.8958, "step": 949 }, { "epoch": 0.10893246187363835, "grad_norm": 13.8125, "learning_rate": 9.838133162082579e-06, "loss": 1.5115, "step": 950 }, { "epoch": 0.10904712762297901, "grad_norm": 23.75, "learning_rate": 9.837664159527147e-06, "loss": 1.6277, "step": 951 }, { "epoch": 0.10916179337231968, "grad_norm": 11.5, "learning_rate": 9.837194489708158e-06, "loss": 1.8317, "step": 952 }, { "epoch": 0.10927645912166035, "grad_norm": 13.3125, "learning_rate": 9.836724152690395e-06, "loss": 1.9619, "step": 953 }, { "epoch": 0.10939112487100103, "grad_norm": 11.125, "learning_rate": 9.836253148538732e-06, "loss": 1.783, "step": 954 }, { "epoch": 0.1095057906203417, "grad_norm": 14.0, "learning_rate": 9.835781477318134e-06, "loss": 2.0408, "step": 955 }, { "epoch": 0.10962045636968237, "grad_norm": 12.5, "learning_rate": 9.835309139093659e-06, "loss": 1.9357, "step": 956 }, { "epoch": 0.10973512211902305, "grad_norm": 11.5625, "learning_rate": 9.834836133930459e-06, "loss": 1.8391, "step": 957 }, { "epoch": 0.10984978786836372, "grad_norm": 12.0, "learning_rate": 9.834362461893774e-06, "loss": 2.0489, "step": 958 }, { "epoch": 0.10996445361770439, "grad_norm": 13.5625, "learning_rate": 9.833888123048938e-06, "loss": 1.8279, "step": 959 }, { "epoch": 0.11007911936704506, "grad_norm": 13.6875, "learning_rate": 9.83341311746138e-06, "loss": 2.3529, "step": 960 }, { "epoch": 0.11019378511638574, "grad_norm": 10.875, "learning_rate": 9.832937445196614e-06, "loss": 1.8036, "step": 961 }, { "epoch": 0.11030845086572641, "grad_norm": 14.25, "learning_rate": 9.832461106320251e-06, "loss": 2.1211, "step": 962 }, { "epoch": 0.11042311661506708, "grad_norm": 13.0, "learning_rate": 9.831984100897995e-06, "loss": 1.8535, "step": 963 }, { "epoch": 0.11053778236440776, "grad_norm": 31.125, "learning_rate": 9.831506428995637e-06, "loss": 2.0687, "step": 964 }, { "epoch": 0.11065244811374843, "grad_norm": 12.625, "learning_rate": 9.831028090679064e-06, "loss": 1.8721, "step": 965 }, { "epoch": 0.1107671138630891, "grad_norm": 12.5, "learning_rate": 9.830549086014254e-06, "loss": 1.921, "step": 966 }, { "epoch": 0.11088177961242977, "grad_norm": 12.6875, "learning_rate": 9.830069415067275e-06, "loss": 1.9714, "step": 967 }, { "epoch": 0.11099644536177045, "grad_norm": 12.5, "learning_rate": 9.829589077904293e-06, "loss": 2.3597, "step": 968 }, { "epoch": 0.1111111111111111, "grad_norm": 14.1875, "learning_rate": 9.829108074591555e-06, "loss": 1.4499, "step": 969 }, { "epoch": 0.11122577686045178, "grad_norm": 13.25, "learning_rate": 9.828626405195412e-06, "loss": 1.8249, "step": 970 }, { "epoch": 0.11134044260979245, "grad_norm": 16.375, "learning_rate": 9.828144069782296e-06, "loss": 1.871, "step": 971 }, { "epoch": 0.11145510835913312, "grad_norm": 12.3125, "learning_rate": 9.82766106841874e-06, "loss": 1.9673, "step": 972 }, { "epoch": 0.1115697741084738, "grad_norm": 9.0625, "learning_rate": 9.827177401171362e-06, "loss": 1.7418, "step": 973 }, { "epoch": 0.11168443985781447, "grad_norm": 14.0, "learning_rate": 9.826693068106876e-06, "loss": 1.9722, "step": 974 }, { "epoch": 0.11179910560715514, "grad_norm": 13.875, "learning_rate": 9.826208069292087e-06, "loss": 1.9823, "step": 975 }, { "epoch": 0.11191377135649581, "grad_norm": 12.25, "learning_rate": 9.82572240479389e-06, "loss": 1.7991, "step": 976 }, { "epoch": 0.11202843710583649, "grad_norm": 11.9375, "learning_rate": 9.825236074679274e-06, "loss": 1.8331, "step": 977 }, { "epoch": 0.11214310285517716, "grad_norm": 10.875, "learning_rate": 9.82474907901532e-06, "loss": 1.8471, "step": 978 }, { "epoch": 0.11225776860451783, "grad_norm": 11.875, "learning_rate": 9.824261417869198e-06, "loss": 1.7291, "step": 979 }, { "epoch": 0.1123724343538585, "grad_norm": 10.4375, "learning_rate": 9.823773091308172e-06, "loss": 1.817, "step": 980 }, { "epoch": 0.11248710010319918, "grad_norm": 11.75, "learning_rate": 9.823284099399597e-06, "loss": 2.0141, "step": 981 }, { "epoch": 0.11260176585253985, "grad_norm": 11.75, "learning_rate": 9.822794442210922e-06, "loss": 1.8636, "step": 982 }, { "epoch": 0.11271643160188052, "grad_norm": 11.6875, "learning_rate": 9.822304119809683e-06, "loss": 1.576, "step": 983 }, { "epoch": 0.1128310973512212, "grad_norm": 11.4375, "learning_rate": 9.821813132263512e-06, "loss": 1.6, "step": 984 }, { "epoch": 0.11294576310056187, "grad_norm": 10.8125, "learning_rate": 9.821321479640134e-06, "loss": 1.6126, "step": 985 }, { "epoch": 0.11306042884990253, "grad_norm": 14.4375, "learning_rate": 9.820829162007359e-06, "loss": 1.543, "step": 986 }, { "epoch": 0.1131750945992432, "grad_norm": 13.0, "learning_rate": 9.820336179433093e-06, "loss": 1.7093, "step": 987 }, { "epoch": 0.11328976034858387, "grad_norm": 14.625, "learning_rate": 9.819842531985338e-06, "loss": 1.9614, "step": 988 }, { "epoch": 0.11340442609792455, "grad_norm": 15.625, "learning_rate": 9.819348219732179e-06, "loss": 1.7542, "step": 989 }, { "epoch": 0.11351909184726522, "grad_norm": 15.25, "learning_rate": 9.818853242741797e-06, "loss": 1.8681, "step": 990 }, { "epoch": 0.11363375759660589, "grad_norm": 11.25, "learning_rate": 9.818357601082467e-06, "loss": 1.5678, "step": 991 }, { "epoch": 0.11374842334594656, "grad_norm": 11.25, "learning_rate": 9.817861294822551e-06, "loss": 1.7298, "step": 992 }, { "epoch": 0.11386308909528724, "grad_norm": 17.375, "learning_rate": 9.817364324030508e-06, "loss": 2.3758, "step": 993 }, { "epoch": 0.11397775484462791, "grad_norm": 17.625, "learning_rate": 9.816866688774881e-06, "loss": 1.8775, "step": 994 }, { "epoch": 0.11409242059396858, "grad_norm": 9.9375, "learning_rate": 9.816368389124316e-06, "loss": 1.9537, "step": 995 }, { "epoch": 0.11420708634330926, "grad_norm": 12.8125, "learning_rate": 9.815869425147538e-06, "loss": 2.1564, "step": 996 }, { "epoch": 0.11432175209264993, "grad_norm": 23.125, "learning_rate": 9.815369796913372e-06, "loss": 2.005, "step": 997 }, { "epoch": 0.1144364178419906, "grad_norm": 18.25, "learning_rate": 9.814869504490731e-06, "loss": 1.9644, "step": 998 }, { "epoch": 0.11455108359133127, "grad_norm": 12.0625, "learning_rate": 9.814368547948623e-06, "loss": 1.3997, "step": 999 }, { "epoch": 0.11466574934067195, "grad_norm": 14.625, "learning_rate": 9.813866927356143e-06, "loss": 2.0303, "step": 1000 }, { "epoch": 0.11478041509001262, "grad_norm": 13.75, "learning_rate": 9.813364642782482e-06, "loss": 1.4571, "step": 1001 }, { "epoch": 0.11489508083935329, "grad_norm": 10.875, "learning_rate": 9.812861694296917e-06, "loss": 1.7579, "step": 1002 }, { "epoch": 0.11500974658869395, "grad_norm": 11.0625, "learning_rate": 9.812358081968826e-06, "loss": 1.6521, "step": 1003 }, { "epoch": 0.11512441233803462, "grad_norm": 13.625, "learning_rate": 9.811853805867668e-06, "loss": 1.5506, "step": 1004 }, { "epoch": 0.1152390780873753, "grad_norm": 12.6875, "learning_rate": 9.811348866063e-06, "loss": 1.8022, "step": 1005 }, { "epoch": 0.11535374383671597, "grad_norm": 10.6875, "learning_rate": 9.810843262624468e-06, "loss": 1.6386, "step": 1006 }, { "epoch": 0.11546840958605664, "grad_norm": 12.625, "learning_rate": 9.810336995621812e-06, "loss": 1.8808, "step": 1007 }, { "epoch": 0.11558307533539731, "grad_norm": 25.75, "learning_rate": 9.80983006512486e-06, "loss": 2.149, "step": 1008 }, { "epoch": 0.11569774108473799, "grad_norm": 12.5625, "learning_rate": 9.809322471203534e-06, "loss": 1.9521, "step": 1009 }, { "epoch": 0.11581240683407866, "grad_norm": 12.125, "learning_rate": 9.808814213927847e-06, "loss": 2.1767, "step": 1010 }, { "epoch": 0.11592707258341933, "grad_norm": 15.375, "learning_rate": 9.808305293367904e-06, "loss": 1.7987, "step": 1011 }, { "epoch": 0.11604173833276, "grad_norm": 11.5625, "learning_rate": 9.807795709593902e-06, "loss": 1.8677, "step": 1012 }, { "epoch": 0.11615640408210068, "grad_norm": 11.75, "learning_rate": 9.807285462676124e-06, "loss": 1.9701, "step": 1013 }, { "epoch": 0.11627106983144135, "grad_norm": 12.375, "learning_rate": 9.806774552684954e-06, "loss": 1.3533, "step": 1014 }, { "epoch": 0.11638573558078202, "grad_norm": 13.3125, "learning_rate": 9.806262979690859e-06, "loss": 1.824, "step": 1015 }, { "epoch": 0.1165004013301227, "grad_norm": 13.125, "learning_rate": 9.805750743764401e-06, "loss": 1.9514, "step": 1016 }, { "epoch": 0.11661506707946337, "grad_norm": 14.25, "learning_rate": 9.805237844976234e-06, "loss": 1.595, "step": 1017 }, { "epoch": 0.11672973282880404, "grad_norm": 11.125, "learning_rate": 9.804724283397102e-06, "loss": 1.9188, "step": 1018 }, { "epoch": 0.11684439857814471, "grad_norm": 11.9375, "learning_rate": 9.804210059097844e-06, "loss": 1.707, "step": 1019 }, { "epoch": 0.11695906432748537, "grad_norm": 17.375, "learning_rate": 9.803695172149382e-06, "loss": 1.7311, "step": 1020 }, { "epoch": 0.11707373007682605, "grad_norm": 12.9375, "learning_rate": 9.803179622622739e-06, "loss": 1.559, "step": 1021 }, { "epoch": 0.11718839582616672, "grad_norm": 12.5, "learning_rate": 9.802663410589024e-06, "loss": 1.7728, "step": 1022 }, { "epoch": 0.11730306157550739, "grad_norm": 13.4375, "learning_rate": 9.80214653611944e-06, "loss": 1.4086, "step": 1023 }, { "epoch": 0.11741772732484806, "grad_norm": 13.9375, "learning_rate": 9.801628999285276e-06, "loss": 1.4764, "step": 1024 }, { "epoch": 0.11753239307418874, "grad_norm": 10.5, "learning_rate": 9.801110800157921e-06, "loss": 1.7678, "step": 1025 }, { "epoch": 0.11764705882352941, "grad_norm": 9.625, "learning_rate": 9.800591938808846e-06, "loss": 1.571, "step": 1026 }, { "epoch": 0.11776172457287008, "grad_norm": 25.5, "learning_rate": 9.800072415309624e-06, "loss": 2.1658, "step": 1027 }, { "epoch": 0.11787639032221076, "grad_norm": 16.75, "learning_rate": 9.799552229731907e-06, "loss": 1.8223, "step": 1028 }, { "epoch": 0.11799105607155143, "grad_norm": 13.125, "learning_rate": 9.79903138214745e-06, "loss": 1.7347, "step": 1029 }, { "epoch": 0.1181057218208921, "grad_norm": 11.3125, "learning_rate": 9.79850987262809e-06, "loss": 1.988, "step": 1030 }, { "epoch": 0.11822038757023277, "grad_norm": 11.4375, "learning_rate": 9.797987701245761e-06, "loss": 2.0295, "step": 1031 }, { "epoch": 0.11833505331957345, "grad_norm": 10.8125, "learning_rate": 9.797464868072489e-06, "loss": 1.7432, "step": 1032 }, { "epoch": 0.11844971906891412, "grad_norm": 13.0, "learning_rate": 9.796941373180384e-06, "loss": 1.3647, "step": 1033 }, { "epoch": 0.11856438481825479, "grad_norm": 11.375, "learning_rate": 9.796417216641654e-06, "loss": 1.6037, "step": 1034 }, { "epoch": 0.11867905056759546, "grad_norm": 16.125, "learning_rate": 9.795892398528599e-06, "loss": 1.7027, "step": 1035 }, { "epoch": 0.11879371631693614, "grad_norm": 11.125, "learning_rate": 9.795366918913605e-06, "loss": 1.7636, "step": 1036 }, { "epoch": 0.1189083820662768, "grad_norm": 12.75, "learning_rate": 9.794840777869152e-06, "loss": 1.801, "step": 1037 }, { "epoch": 0.11902304781561747, "grad_norm": 12.5, "learning_rate": 9.794313975467814e-06, "loss": 1.4224, "step": 1038 }, { "epoch": 0.11913771356495814, "grad_norm": 13.5, "learning_rate": 9.793786511782248e-06, "loss": 1.7618, "step": 1039 }, { "epoch": 0.11925237931429881, "grad_norm": 12.625, "learning_rate": 9.79325838688521e-06, "loss": 1.8555, "step": 1040 }, { "epoch": 0.11936704506363949, "grad_norm": 12.375, "learning_rate": 9.792729600849548e-06, "loss": 2.3005, "step": 1041 }, { "epoch": 0.11948171081298016, "grad_norm": 13.5625, "learning_rate": 9.792200153748195e-06, "loss": 2.0441, "step": 1042 }, { "epoch": 0.11959637656232083, "grad_norm": 12.4375, "learning_rate": 9.791670045654177e-06, "loss": 1.6851, "step": 1043 }, { "epoch": 0.1197110423116615, "grad_norm": 23.375, "learning_rate": 9.791139276640615e-06, "loss": 1.8448, "step": 1044 }, { "epoch": 0.11982570806100218, "grad_norm": 11.125, "learning_rate": 9.790607846780718e-06, "loss": 1.5266, "step": 1045 }, { "epoch": 0.11994037381034285, "grad_norm": 11.4375, "learning_rate": 9.790075756147784e-06, "loss": 1.6233, "step": 1046 }, { "epoch": 0.12005503955968352, "grad_norm": 11.0, "learning_rate": 9.789543004815207e-06, "loss": 1.6566, "step": 1047 }, { "epoch": 0.1201697053090242, "grad_norm": 12.6875, "learning_rate": 9.789009592856471e-06, "loss": 1.8826, "step": 1048 }, { "epoch": 0.12028437105836487, "grad_norm": 20.25, "learning_rate": 9.788475520345146e-06, "loss": 2.2163, "step": 1049 }, { "epoch": 0.12039903680770554, "grad_norm": 10.4375, "learning_rate": 9.787940787354901e-06, "loss": 1.2909, "step": 1050 }, { "epoch": 0.12051370255704622, "grad_norm": 11.5, "learning_rate": 9.78740539395949e-06, "loss": 1.8991, "step": 1051 }, { "epoch": 0.12062836830638689, "grad_norm": 13.25, "learning_rate": 9.786869340232763e-06, "loss": 1.9696, "step": 1052 }, { "epoch": 0.12074303405572756, "grad_norm": 12.1875, "learning_rate": 9.786332626248655e-06, "loss": 1.6955, "step": 1053 }, { "epoch": 0.12085769980506822, "grad_norm": 9.9375, "learning_rate": 9.785795252081198e-06, "loss": 1.6897, "step": 1054 }, { "epoch": 0.12097236555440889, "grad_norm": 14.6875, "learning_rate": 9.78525721780451e-06, "loss": 1.9844, "step": 1055 }, { "epoch": 0.12108703130374956, "grad_norm": 16.125, "learning_rate": 9.784718523492805e-06, "loss": 1.9104, "step": 1056 }, { "epoch": 0.12120169705309024, "grad_norm": 13.125, "learning_rate": 9.784179169220384e-06, "loss": 1.8278, "step": 1057 }, { "epoch": 0.12131636280243091, "grad_norm": 12.5625, "learning_rate": 9.783639155061643e-06, "loss": 1.8259, "step": 1058 }, { "epoch": 0.12143102855177158, "grad_norm": 10.125, "learning_rate": 9.783098481091064e-06, "loss": 1.857, "step": 1059 }, { "epoch": 0.12154569430111226, "grad_norm": 11.625, "learning_rate": 9.782557147383225e-06, "loss": 1.8574, "step": 1060 }, { "epoch": 0.12166036005045293, "grad_norm": 13.5625, "learning_rate": 9.782015154012789e-06, "loss": 1.7411, "step": 1061 }, { "epoch": 0.1217750257997936, "grad_norm": 14.375, "learning_rate": 9.781472501054518e-06, "loss": 1.8005, "step": 1062 }, { "epoch": 0.12188969154913427, "grad_norm": 13.4375, "learning_rate": 9.780929188583257e-06, "loss": 2.1351, "step": 1063 }, { "epoch": 0.12200435729847495, "grad_norm": 14.25, "learning_rate": 9.78038521667395e-06, "loss": 1.8313, "step": 1064 }, { "epoch": 0.12211902304781562, "grad_norm": 14.125, "learning_rate": 9.77984058540162e-06, "loss": 1.7562, "step": 1065 }, { "epoch": 0.12223368879715629, "grad_norm": 27.625, "learning_rate": 9.779295294841397e-06, "loss": 2.2668, "step": 1066 }, { "epoch": 0.12234835454649697, "grad_norm": 12.8125, "learning_rate": 9.77874934506849e-06, "loss": 1.9452, "step": 1067 }, { "epoch": 0.12246302029583764, "grad_norm": 17.625, "learning_rate": 9.778202736158199e-06, "loss": 2.079, "step": 1068 }, { "epoch": 0.12257768604517831, "grad_norm": 12.625, "learning_rate": 9.777655468185925e-06, "loss": 1.7392, "step": 1069 }, { "epoch": 0.12269235179451898, "grad_norm": 12.875, "learning_rate": 9.777107541227147e-06, "loss": 1.4702, "step": 1070 }, { "epoch": 0.12280701754385964, "grad_norm": 14.4375, "learning_rate": 9.776558955357445e-06, "loss": 1.7694, "step": 1071 }, { "epoch": 0.12292168329320032, "grad_norm": 9.4375, "learning_rate": 9.776009710652483e-06, "loss": 1.7439, "step": 1072 }, { "epoch": 0.12303634904254099, "grad_norm": 15.75, "learning_rate": 9.775459807188022e-06, "loss": 2.2788, "step": 1073 }, { "epoch": 0.12315101479188166, "grad_norm": 14.0625, "learning_rate": 9.77490924503991e-06, "loss": 1.5959, "step": 1074 }, { "epoch": 0.12326568054122233, "grad_norm": 13.4375, "learning_rate": 9.774358024284083e-06, "loss": 1.7235, "step": 1075 }, { "epoch": 0.123380346290563, "grad_norm": 12.1875, "learning_rate": 9.773806144996576e-06, "loss": 1.724, "step": 1076 }, { "epoch": 0.12349501203990368, "grad_norm": 10.875, "learning_rate": 9.773253607253509e-06, "loss": 1.6259, "step": 1077 }, { "epoch": 0.12360967778924435, "grad_norm": 14.4375, "learning_rate": 9.77270041113109e-06, "loss": 2.2343, "step": 1078 }, { "epoch": 0.12372434353858502, "grad_norm": 29.5, "learning_rate": 9.772146556705628e-06, "loss": 1.6963, "step": 1079 }, { "epoch": 0.1238390092879257, "grad_norm": 16.75, "learning_rate": 9.771592044053512e-06, "loss": 1.9549, "step": 1080 }, { "epoch": 0.12395367503726637, "grad_norm": 14.8125, "learning_rate": 9.771036873251231e-06, "loss": 1.4991, "step": 1081 }, { "epoch": 0.12406834078660704, "grad_norm": 12.5, "learning_rate": 9.770481044375357e-06, "loss": 1.9496, "step": 1082 }, { "epoch": 0.12418300653594772, "grad_norm": 12.1875, "learning_rate": 9.769924557502554e-06, "loss": 1.8675, "step": 1083 }, { "epoch": 0.12429767228528839, "grad_norm": 28.5, "learning_rate": 9.769367412709585e-06, "loss": 1.9645, "step": 1084 }, { "epoch": 0.12441233803462906, "grad_norm": 19.25, "learning_rate": 9.768809610073292e-06, "loss": 1.5984, "step": 1085 }, { "epoch": 0.12452700378396973, "grad_norm": 12.9375, "learning_rate": 9.768251149670615e-06, "loss": 1.9562, "step": 1086 }, { "epoch": 0.1246416695333104, "grad_norm": 11.3125, "learning_rate": 9.767692031578582e-06, "loss": 1.9052, "step": 1087 }, { "epoch": 0.12475633528265107, "grad_norm": 16.75, "learning_rate": 9.767132255874316e-06, "loss": 1.7471, "step": 1088 }, { "epoch": 0.12487100103199174, "grad_norm": 30.375, "learning_rate": 9.766571822635022e-06, "loss": 1.9126, "step": 1089 }, { "epoch": 0.12498566678133241, "grad_norm": 12.75, "learning_rate": 9.766010731938007e-06, "loss": 1.8706, "step": 1090 }, { "epoch": 0.1251003325306731, "grad_norm": 18.0, "learning_rate": 9.765448983860658e-06, "loss": 2.0728, "step": 1091 }, { "epoch": 0.12521499828001376, "grad_norm": 13.75, "learning_rate": 9.764886578480462e-06, "loss": 1.757, "step": 1092 }, { "epoch": 0.12532966402935444, "grad_norm": 13.4375, "learning_rate": 9.764323515874986e-06, "loss": 2.0009, "step": 1093 }, { "epoch": 0.1254443297786951, "grad_norm": 11.875, "learning_rate": 9.7637597961219e-06, "loss": 1.5212, "step": 1094 }, { "epoch": 0.1255589955280358, "grad_norm": 9.5, "learning_rate": 9.763195419298956e-06, "loss": 1.6715, "step": 1095 }, { "epoch": 0.12567366127737645, "grad_norm": 14.5, "learning_rate": 9.762630385483997e-06, "loss": 2.2101, "step": 1096 }, { "epoch": 0.1257883270267171, "grad_norm": 15.0625, "learning_rate": 9.76206469475496e-06, "loss": 2.1927, "step": 1097 }, { "epoch": 0.1259029927760578, "grad_norm": 10.0, "learning_rate": 9.761498347189873e-06, "loss": 1.598, "step": 1098 }, { "epoch": 0.12601765852539845, "grad_norm": 13.9375, "learning_rate": 9.760931342866851e-06, "loss": 1.7702, "step": 1099 }, { "epoch": 0.12613232427473914, "grad_norm": 10.9375, "learning_rate": 9.760363681864102e-06, "loss": 1.5802, "step": 1100 }, { "epoch": 0.1262469900240798, "grad_norm": 12.5625, "learning_rate": 9.759795364259923e-06, "loss": 1.6105, "step": 1101 }, { "epoch": 0.12636165577342048, "grad_norm": 12.8125, "learning_rate": 9.759226390132704e-06, "loss": 1.6752, "step": 1102 }, { "epoch": 0.12647632152276114, "grad_norm": 16.125, "learning_rate": 9.758656759560924e-06, "loss": 1.929, "step": 1103 }, { "epoch": 0.12659098727210183, "grad_norm": 12.0, "learning_rate": 9.758086472623152e-06, "loss": 1.8807, "step": 1104 }, { "epoch": 0.1267056530214425, "grad_norm": 11.625, "learning_rate": 9.757515529398047e-06, "loss": 1.6892, "step": 1105 }, { "epoch": 0.12682031877078317, "grad_norm": 11.3125, "learning_rate": 9.756943929964365e-06, "loss": 1.7906, "step": 1106 }, { "epoch": 0.12693498452012383, "grad_norm": 15.3125, "learning_rate": 9.75637167440094e-06, "loss": 1.7359, "step": 1107 }, { "epoch": 0.12704965026946452, "grad_norm": 11.125, "learning_rate": 9.755798762786708e-06, "loss": 1.8182, "step": 1108 }, { "epoch": 0.12716431601880518, "grad_norm": 11.3125, "learning_rate": 9.75522519520069e-06, "loss": 1.669, "step": 1109 }, { "epoch": 0.12727898176814587, "grad_norm": 11.625, "learning_rate": 9.754650971722e-06, "loss": 1.5718, "step": 1110 }, { "epoch": 0.12739364751748652, "grad_norm": 11.6875, "learning_rate": 9.754076092429841e-06, "loss": 1.9882, "step": 1111 }, { "epoch": 0.1275083132668272, "grad_norm": 14.0, "learning_rate": 9.753500557403505e-06, "loss": 1.7954, "step": 1112 }, { "epoch": 0.12762297901616787, "grad_norm": 12.0625, "learning_rate": 9.752924366722378e-06, "loss": 2.0148, "step": 1113 }, { "epoch": 0.12773764476550853, "grad_norm": 12.375, "learning_rate": 9.752347520465932e-06, "loss": 1.4935, "step": 1114 }, { "epoch": 0.12785231051484922, "grad_norm": 10.625, "learning_rate": 9.751770018713734e-06, "loss": 1.8867, "step": 1115 }, { "epoch": 0.12796697626418987, "grad_norm": 15.1875, "learning_rate": 9.75119186154544e-06, "loss": 1.7092, "step": 1116 }, { "epoch": 0.12808164201353056, "grad_norm": 13.9375, "learning_rate": 9.750613049040794e-06, "loss": 2.2245, "step": 1117 }, { "epoch": 0.12819630776287122, "grad_norm": 13.3125, "learning_rate": 9.750033581279631e-06, "loss": 2.0795, "step": 1118 }, { "epoch": 0.1283109735122119, "grad_norm": 15.5625, "learning_rate": 9.749453458341883e-06, "loss": 2.1103, "step": 1119 }, { "epoch": 0.12842563926155257, "grad_norm": 11.375, "learning_rate": 9.74887268030756e-06, "loss": 1.8849, "step": 1120 }, { "epoch": 0.12854030501089325, "grad_norm": 14.6875, "learning_rate": 9.748291247256774e-06, "loss": 1.7647, "step": 1121 }, { "epoch": 0.1286549707602339, "grad_norm": 15.5625, "learning_rate": 9.74770915926972e-06, "loss": 1.7429, "step": 1122 }, { "epoch": 0.1287696365095746, "grad_norm": 11.75, "learning_rate": 9.747126416426689e-06, "loss": 1.9553, "step": 1123 }, { "epoch": 0.12888430225891526, "grad_norm": 14.3125, "learning_rate": 9.746543018808056e-06, "loss": 1.9828, "step": 1124 }, { "epoch": 0.12899896800825594, "grad_norm": 12.8125, "learning_rate": 9.745958966494291e-06, "loss": 1.9998, "step": 1125 }, { "epoch": 0.1291136337575966, "grad_norm": 11.875, "learning_rate": 9.745374259565954e-06, "loss": 1.7265, "step": 1126 }, { "epoch": 0.1292282995069373, "grad_norm": 12.9375, "learning_rate": 9.744788898103692e-06, "loss": 1.6228, "step": 1127 }, { "epoch": 0.12934296525627795, "grad_norm": 13.625, "learning_rate": 9.744202882188244e-06, "loss": 1.8119, "step": 1128 }, { "epoch": 0.12945763100561863, "grad_norm": 12.0, "learning_rate": 9.743616211900443e-06, "loss": 2.1222, "step": 1129 }, { "epoch": 0.1295722967549593, "grad_norm": 29.25, "learning_rate": 9.743028887321207e-06, "loss": 1.8354, "step": 1130 }, { "epoch": 0.12968696250429995, "grad_norm": 11.125, "learning_rate": 9.742440908531545e-06, "loss": 1.6608, "step": 1131 }, { "epoch": 0.12980162825364064, "grad_norm": 11.3125, "learning_rate": 9.74185227561256e-06, "loss": 1.892, "step": 1132 }, { "epoch": 0.1299162940029813, "grad_norm": 18.625, "learning_rate": 9.741262988645442e-06, "loss": 2.573, "step": 1133 }, { "epoch": 0.13003095975232198, "grad_norm": 14.0625, "learning_rate": 9.740673047711471e-06, "loss": 1.8951, "step": 1134 }, { "epoch": 0.13014562550166264, "grad_norm": 11.6875, "learning_rate": 9.740082452892018e-06, "loss": 1.9163, "step": 1135 }, { "epoch": 0.13026029125100333, "grad_norm": 11.9375, "learning_rate": 9.739491204268545e-06, "loss": 2.1957, "step": 1136 }, { "epoch": 0.130374957000344, "grad_norm": 14.1875, "learning_rate": 9.738899301922602e-06, "loss": 2.0105, "step": 1137 }, { "epoch": 0.13048962274968468, "grad_norm": 11.9375, "learning_rate": 9.738306745935834e-06, "loss": 1.6442, "step": 1138 }, { "epoch": 0.13060428849902533, "grad_norm": 11.9375, "learning_rate": 9.73771353638997e-06, "loss": 1.6511, "step": 1139 }, { "epoch": 0.13071895424836602, "grad_norm": 11.1875, "learning_rate": 9.737119673366833e-06, "loss": 1.7007, "step": 1140 }, { "epoch": 0.13083361999770668, "grad_norm": 12.5, "learning_rate": 9.736525156948333e-06, "loss": 1.7415, "step": 1141 }, { "epoch": 0.13094828574704737, "grad_norm": 15.3125, "learning_rate": 9.735929987216477e-06, "loss": 1.6196, "step": 1142 }, { "epoch": 0.13106295149638802, "grad_norm": 12.875, "learning_rate": 9.735334164253352e-06, "loss": 1.6852, "step": 1143 }, { "epoch": 0.1311776172457287, "grad_norm": 14.375, "learning_rate": 9.734737688141143e-06, "loss": 1.7498, "step": 1144 }, { "epoch": 0.13129228299506937, "grad_norm": 20.0, "learning_rate": 9.734140558962122e-06, "loss": 1.7909, "step": 1145 }, { "epoch": 0.13140694874441006, "grad_norm": 16.75, "learning_rate": 9.733542776798653e-06, "loss": 2.4597, "step": 1146 }, { "epoch": 0.13152161449375072, "grad_norm": 12.0, "learning_rate": 9.732944341733188e-06, "loss": 1.7012, "step": 1147 }, { "epoch": 0.13163628024309137, "grad_norm": 12.25, "learning_rate": 9.732345253848268e-06, "loss": 1.6934, "step": 1148 }, { "epoch": 0.13175094599243206, "grad_norm": 10.75, "learning_rate": 9.731745513226528e-06, "loss": 1.9332, "step": 1149 }, { "epoch": 0.13186561174177272, "grad_norm": 13.5625, "learning_rate": 9.731145119950688e-06, "loss": 1.7209, "step": 1150 }, { "epoch": 0.1319802774911134, "grad_norm": 10.3125, "learning_rate": 9.730544074103563e-06, "loss": 1.8702, "step": 1151 }, { "epoch": 0.13209494324045407, "grad_norm": 11.5625, "learning_rate": 9.729942375768055e-06, "loss": 1.5281, "step": 1152 }, { "epoch": 0.13220960898979475, "grad_norm": 10.5625, "learning_rate": 9.729340025027158e-06, "loss": 1.5383, "step": 1153 }, { "epoch": 0.1323242747391354, "grad_norm": 12.5625, "learning_rate": 9.728737021963954e-06, "loss": 1.6394, "step": 1154 }, { "epoch": 0.1324389404884761, "grad_norm": 14.1875, "learning_rate": 9.728133366661616e-06, "loss": 1.5466, "step": 1155 }, { "epoch": 0.13255360623781676, "grad_norm": 13.625, "learning_rate": 9.727529059203407e-06, "loss": 1.8431, "step": 1156 }, { "epoch": 0.13266827198715744, "grad_norm": 14.1875, "learning_rate": 9.726924099672677e-06, "loss": 2.122, "step": 1157 }, { "epoch": 0.1327829377364981, "grad_norm": 12.1875, "learning_rate": 9.726318488152873e-06, "loss": 1.7013, "step": 1158 }, { "epoch": 0.1328976034858388, "grad_norm": 13.625, "learning_rate": 9.725712224727524e-06, "loss": 1.7563, "step": 1159 }, { "epoch": 0.13301226923517945, "grad_norm": 13.1875, "learning_rate": 9.725105309480253e-06, "loss": 1.8408, "step": 1160 }, { "epoch": 0.13312693498452013, "grad_norm": 24.375, "learning_rate": 9.724497742494777e-06, "loss": 1.754, "step": 1161 }, { "epoch": 0.1332416007338608, "grad_norm": 11.3125, "learning_rate": 9.72388952385489e-06, "loss": 1.8337, "step": 1162 }, { "epoch": 0.13335626648320148, "grad_norm": 11.375, "learning_rate": 9.723280653644491e-06, "loss": 1.5215, "step": 1163 }, { "epoch": 0.13347093223254214, "grad_norm": 13.1875, "learning_rate": 9.72267113194756e-06, "loss": 2.038, "step": 1164 }, { "epoch": 0.13358559798188283, "grad_norm": 14.875, "learning_rate": 9.722060958848168e-06, "loss": 1.7595, "step": 1165 }, { "epoch": 0.13370026373122348, "grad_norm": 33.5, "learning_rate": 9.721450134430478e-06, "loss": 1.8814, "step": 1166 }, { "epoch": 0.13381492948056414, "grad_norm": 14.1875, "learning_rate": 9.72083865877874e-06, "loss": 1.542, "step": 1167 }, { "epoch": 0.13392959522990483, "grad_norm": 34.5, "learning_rate": 9.720226531977298e-06, "loss": 1.5978, "step": 1168 }, { "epoch": 0.1340442609792455, "grad_norm": 12.9375, "learning_rate": 9.71961375411058e-06, "loss": 1.9019, "step": 1169 }, { "epoch": 0.13415892672858618, "grad_norm": 15.375, "learning_rate": 9.71900032526311e-06, "loss": 2.0806, "step": 1170 }, { "epoch": 0.13427359247792683, "grad_norm": 13.6875, "learning_rate": 9.718386245519496e-06, "loss": 1.6474, "step": 1171 }, { "epoch": 0.13438825822726752, "grad_norm": 31.875, "learning_rate": 9.71777151496444e-06, "loss": 1.7631, "step": 1172 }, { "epoch": 0.13450292397660818, "grad_norm": 17.25, "learning_rate": 9.717156133682735e-06, "loss": 1.8292, "step": 1173 }, { "epoch": 0.13461758972594887, "grad_norm": 12.0, "learning_rate": 9.716540101759256e-06, "loss": 1.7814, "step": 1174 }, { "epoch": 0.13473225547528953, "grad_norm": 11.5625, "learning_rate": 9.715923419278977e-06, "loss": 1.7663, "step": 1175 }, { "epoch": 0.1348469212246302, "grad_norm": 16.375, "learning_rate": 9.715306086326955e-06, "loss": 2.0032, "step": 1176 }, { "epoch": 0.13496158697397087, "grad_norm": 15.1875, "learning_rate": 9.714688102988338e-06, "loss": 1.8871, "step": 1177 }, { "epoch": 0.13507625272331156, "grad_norm": 10.5, "learning_rate": 9.71406946934837e-06, "loss": 1.6114, "step": 1178 }, { "epoch": 0.13519091847265222, "grad_norm": 15.5, "learning_rate": 9.713450185492378e-06, "loss": 1.3613, "step": 1179 }, { "epoch": 0.1353055842219929, "grad_norm": 10.625, "learning_rate": 9.712830251505778e-06, "loss": 1.7763, "step": 1180 }, { "epoch": 0.13542024997133356, "grad_norm": 16.375, "learning_rate": 9.71220966747408e-06, "loss": 1.7473, "step": 1181 }, { "epoch": 0.13553491572067425, "grad_norm": 11.25, "learning_rate": 9.711588433482881e-06, "loss": 1.8088, "step": 1182 }, { "epoch": 0.1356495814700149, "grad_norm": 14.9375, "learning_rate": 9.710966549617868e-06, "loss": 1.5883, "step": 1183 }, { "epoch": 0.13576424721935557, "grad_norm": 13.25, "learning_rate": 9.71034401596482e-06, "loss": 1.8172, "step": 1184 }, { "epoch": 0.13587891296869625, "grad_norm": 11.6875, "learning_rate": 9.709720832609602e-06, "loss": 1.4234, "step": 1185 }, { "epoch": 0.1359935787180369, "grad_norm": 10.25, "learning_rate": 9.70909699963817e-06, "loss": 1.6412, "step": 1186 }, { "epoch": 0.1361082444673776, "grad_norm": 15.125, "learning_rate": 9.708472517136571e-06, "loss": 2.1891, "step": 1187 }, { "epoch": 0.13622291021671826, "grad_norm": 12.0, "learning_rate": 9.707847385190939e-06, "loss": 1.6598, "step": 1188 }, { "epoch": 0.13633757596605894, "grad_norm": 11.375, "learning_rate": 9.7072216038875e-06, "loss": 1.4161, "step": 1189 }, { "epoch": 0.1364522417153996, "grad_norm": 14.4375, "learning_rate": 9.70659517331257e-06, "loss": 1.6128, "step": 1190 }, { "epoch": 0.1365669074647403, "grad_norm": 13.0625, "learning_rate": 9.705968093552552e-06, "loss": 1.6519, "step": 1191 }, { "epoch": 0.13668157321408095, "grad_norm": 12.1875, "learning_rate": 9.705340364693936e-06, "loss": 1.7993, "step": 1192 }, { "epoch": 0.13679623896342163, "grad_norm": 16.5, "learning_rate": 9.704711986823311e-06, "loss": 2.2097, "step": 1193 }, { "epoch": 0.1369109047127623, "grad_norm": 12.0, "learning_rate": 9.704082960027347e-06, "loss": 1.8471, "step": 1194 }, { "epoch": 0.13702557046210298, "grad_norm": 29.25, "learning_rate": 9.703453284392807e-06, "loss": 1.9775, "step": 1195 }, { "epoch": 0.13714023621144364, "grad_norm": 13.8125, "learning_rate": 9.702822960006544e-06, "loss": 1.6039, "step": 1196 }, { "epoch": 0.13725490196078433, "grad_norm": 11.625, "learning_rate": 9.702191986955494e-06, "loss": 1.6354, "step": 1197 }, { "epoch": 0.13736956771012498, "grad_norm": 16.375, "learning_rate": 9.701560365326694e-06, "loss": 1.4687, "step": 1198 }, { "epoch": 0.13748423345946567, "grad_norm": 12.4375, "learning_rate": 9.700928095207261e-06, "loss": 1.6185, "step": 1199 }, { "epoch": 0.13759889920880633, "grad_norm": 13.6875, "learning_rate": 9.700295176684404e-06, "loss": 1.9454, "step": 1200 }, { "epoch": 0.137713564958147, "grad_norm": 11.625, "learning_rate": 9.699661609845426e-06, "loss": 1.7551, "step": 1201 }, { "epoch": 0.13782823070748768, "grad_norm": 13.9375, "learning_rate": 9.69902739477771e-06, "loss": 1.9537, "step": 1202 }, { "epoch": 0.13794289645682833, "grad_norm": 11.25, "learning_rate": 9.698392531568736e-06, "loss": 1.8202, "step": 1203 }, { "epoch": 0.13805756220616902, "grad_norm": 13.8125, "learning_rate": 9.697757020306073e-06, "loss": 2.119, "step": 1204 }, { "epoch": 0.13817222795550968, "grad_norm": 30.75, "learning_rate": 9.697120861077376e-06, "loss": 1.7106, "step": 1205 }, { "epoch": 0.13828689370485037, "grad_norm": 12.25, "learning_rate": 9.696484053970392e-06, "loss": 1.688, "step": 1206 }, { "epoch": 0.13840155945419103, "grad_norm": 14.125, "learning_rate": 9.695846599072956e-06, "loss": 1.5838, "step": 1207 }, { "epoch": 0.1385162252035317, "grad_norm": 10.9375, "learning_rate": 9.695208496472992e-06, "loss": 1.9953, "step": 1208 }, { "epoch": 0.13863089095287237, "grad_norm": 12.4375, "learning_rate": 9.694569746258516e-06, "loss": 1.5177, "step": 1209 }, { "epoch": 0.13874555670221306, "grad_norm": 10.8125, "learning_rate": 9.693930348517628e-06, "loss": 1.6677, "step": 1210 }, { "epoch": 0.13886022245155372, "grad_norm": 11.1875, "learning_rate": 9.693290303338524e-06, "loss": 1.6314, "step": 1211 }, { "epoch": 0.1389748882008944, "grad_norm": 15.5625, "learning_rate": 9.692649610809485e-06, "loss": 1.737, "step": 1212 }, { "epoch": 0.13908955395023506, "grad_norm": 14.6875, "learning_rate": 9.692008271018883e-06, "loss": 1.9788, "step": 1213 }, { "epoch": 0.13920421969957575, "grad_norm": 13.125, "learning_rate": 9.691366284055178e-06, "loss": 1.5723, "step": 1214 }, { "epoch": 0.1393188854489164, "grad_norm": 10.5, "learning_rate": 9.690723650006917e-06, "loss": 1.6236, "step": 1215 }, { "epoch": 0.1394335511982571, "grad_norm": 12.4375, "learning_rate": 9.690080368962745e-06, "loss": 1.5188, "step": 1216 }, { "epoch": 0.13954821694759775, "grad_norm": 10.125, "learning_rate": 9.689436441011384e-06, "loss": 1.6321, "step": 1217 }, { "epoch": 0.1396628826969384, "grad_norm": 12.3125, "learning_rate": 9.688791866241659e-06, "loss": 2.133, "step": 1218 }, { "epoch": 0.1397775484462791, "grad_norm": 10.625, "learning_rate": 9.688146644742469e-06, "loss": 2.0464, "step": 1219 }, { "epoch": 0.13989221419561976, "grad_norm": 11.375, "learning_rate": 9.687500776602813e-06, "loss": 1.7224, "step": 1220 }, { "epoch": 0.14000687994496044, "grad_norm": 13.3125, "learning_rate": 9.68685426191178e-06, "loss": 1.3901, "step": 1221 }, { "epoch": 0.1401215456943011, "grad_norm": 11.375, "learning_rate": 9.68620710075854e-06, "loss": 1.9295, "step": 1222 }, { "epoch": 0.1402362114436418, "grad_norm": 11.75, "learning_rate": 9.685559293232356e-06, "loss": 1.7929, "step": 1223 }, { "epoch": 0.14035087719298245, "grad_norm": 12.125, "learning_rate": 9.684910839422583e-06, "loss": 1.616, "step": 1224 }, { "epoch": 0.14046554294232313, "grad_norm": 13.3125, "learning_rate": 9.684261739418663e-06, "loss": 2.0385, "step": 1225 }, { "epoch": 0.1405802086916638, "grad_norm": 32.0, "learning_rate": 9.683611993310126e-06, "loss": 2.0676, "step": 1226 }, { "epoch": 0.14069487444100448, "grad_norm": 10.375, "learning_rate": 9.682961601186593e-06, "loss": 1.601, "step": 1227 }, { "epoch": 0.14080954019034514, "grad_norm": 15.4375, "learning_rate": 9.68231056313777e-06, "loss": 1.8776, "step": 1228 }, { "epoch": 0.14092420593968583, "grad_norm": 12.9375, "learning_rate": 9.681658879253461e-06, "loss": 1.6857, "step": 1229 }, { "epoch": 0.14103887168902648, "grad_norm": 13.3125, "learning_rate": 9.681006549623548e-06, "loss": 1.9781, "step": 1230 }, { "epoch": 0.14115353743836717, "grad_norm": 12.0625, "learning_rate": 9.680353574338011e-06, "loss": 1.9553, "step": 1231 }, { "epoch": 0.14126820318770783, "grad_norm": 10.875, "learning_rate": 9.679699953486913e-06, "loss": 1.6799, "step": 1232 }, { "epoch": 0.14138286893704852, "grad_norm": 11.8125, "learning_rate": 9.679045687160411e-06, "loss": 2.1809, "step": 1233 }, { "epoch": 0.14149753468638918, "grad_norm": 11.1875, "learning_rate": 9.678390775448744e-06, "loss": 1.8142, "step": 1234 }, { "epoch": 0.14161220043572983, "grad_norm": 12.25, "learning_rate": 9.677735218442253e-06, "loss": 1.6591, "step": 1235 }, { "epoch": 0.14172686618507052, "grad_norm": 11.3125, "learning_rate": 9.67707901623135e-06, "loss": 1.6666, "step": 1236 }, { "epoch": 0.14184153193441118, "grad_norm": 14.3125, "learning_rate": 9.67642216890655e-06, "loss": 2.1278, "step": 1237 }, { "epoch": 0.14195619768375187, "grad_norm": 11.5625, "learning_rate": 9.675764676558454e-06, "loss": 1.729, "step": 1238 }, { "epoch": 0.14207086343309253, "grad_norm": 14.625, "learning_rate": 9.67510653927775e-06, "loss": 2.2933, "step": 1239 }, { "epoch": 0.1421855291824332, "grad_norm": 14.375, "learning_rate": 9.674447757155213e-06, "loss": 2.4939, "step": 1240 }, { "epoch": 0.14230019493177387, "grad_norm": 26.375, "learning_rate": 9.67378833028171e-06, "loss": 1.9058, "step": 1241 }, { "epoch": 0.14241486068111456, "grad_norm": 12.5, "learning_rate": 9.6731282587482e-06, "loss": 1.7369, "step": 1242 }, { "epoch": 0.14252952643045522, "grad_norm": 12.3125, "learning_rate": 9.672467542645722e-06, "loss": 1.6755, "step": 1243 }, { "epoch": 0.1426441921797959, "grad_norm": 11.5625, "learning_rate": 9.671806182065415e-06, "loss": 1.3202, "step": 1244 }, { "epoch": 0.14275885792913656, "grad_norm": 10.875, "learning_rate": 9.671144177098496e-06, "loss": 1.6316, "step": 1245 }, { "epoch": 0.14287352367847725, "grad_norm": 12.875, "learning_rate": 9.670481527836277e-06, "loss": 1.7831, "step": 1246 }, { "epoch": 0.1429881894278179, "grad_norm": 13.25, "learning_rate": 9.66981823437016e-06, "loss": 1.7462, "step": 1247 }, { "epoch": 0.1431028551771586, "grad_norm": 12.25, "learning_rate": 9.669154296791632e-06, "loss": 1.8329, "step": 1248 }, { "epoch": 0.14321752092649925, "grad_norm": 13.5625, "learning_rate": 9.668489715192271e-06, "loss": 1.934, "step": 1249 }, { "epoch": 0.14333218667583994, "grad_norm": 11.0, "learning_rate": 9.667824489663743e-06, "loss": 1.7453, "step": 1250 }, { "epoch": 0.1434468524251806, "grad_norm": 13.5625, "learning_rate": 9.667158620297802e-06, "loss": 1.6791, "step": 1251 }, { "epoch": 0.14356151817452126, "grad_norm": 10.0, "learning_rate": 9.666492107186296e-06, "loss": 1.7159, "step": 1252 }, { "epoch": 0.14367618392386194, "grad_norm": 12.75, "learning_rate": 9.665824950421156e-06, "loss": 1.6662, "step": 1253 }, { "epoch": 0.1437908496732026, "grad_norm": 11.375, "learning_rate": 9.6651571500944e-06, "loss": 1.7748, "step": 1254 }, { "epoch": 0.1439055154225433, "grad_norm": 14.0625, "learning_rate": 9.664488706298143e-06, "loss": 1.8577, "step": 1255 }, { "epoch": 0.14402018117188395, "grad_norm": 28.625, "learning_rate": 9.663819619124581e-06, "loss": 2.4262, "step": 1256 }, { "epoch": 0.14413484692122464, "grad_norm": 12.5625, "learning_rate": 9.663149888666004e-06, "loss": 1.7734, "step": 1257 }, { "epoch": 0.1442495126705653, "grad_norm": 13.6875, "learning_rate": 9.662479515014787e-06, "loss": 1.6566, "step": 1258 }, { "epoch": 0.14436417841990598, "grad_norm": 13.8125, "learning_rate": 9.661808498263396e-06, "loss": 1.7734, "step": 1259 }, { "epoch": 0.14447884416924664, "grad_norm": 13.6875, "learning_rate": 9.661136838504386e-06, "loss": 2.0368, "step": 1260 }, { "epoch": 0.14459350991858733, "grad_norm": 14.625, "learning_rate": 9.660464535830396e-06, "loss": 1.6912, "step": 1261 }, { "epoch": 0.14470817566792799, "grad_norm": 15.0, "learning_rate": 9.659791590334162e-06, "loss": 1.7728, "step": 1262 }, { "epoch": 0.14482284141726867, "grad_norm": 14.5, "learning_rate": 9.6591180021085e-06, "loss": 1.5517, "step": 1263 }, { "epoch": 0.14493750716660933, "grad_norm": 15.375, "learning_rate": 9.658443771246322e-06, "loss": 2.3818, "step": 1264 }, { "epoch": 0.14505217291595002, "grad_norm": 11.6875, "learning_rate": 9.657768897840623e-06, "loss": 1.4692, "step": 1265 }, { "epoch": 0.14516683866529068, "grad_norm": 10.875, "learning_rate": 9.65709338198449e-06, "loss": 1.5032, "step": 1266 }, { "epoch": 0.14528150441463136, "grad_norm": 10.875, "learning_rate": 9.656417223771098e-06, "loss": 1.9481, "step": 1267 }, { "epoch": 0.14539617016397202, "grad_norm": 13.375, "learning_rate": 9.655740423293708e-06, "loss": 1.8455, "step": 1268 }, { "epoch": 0.14551083591331268, "grad_norm": 11.5, "learning_rate": 9.655062980645674e-06, "loss": 1.7029, "step": 1269 }, { "epoch": 0.14562550166265337, "grad_norm": 12.8125, "learning_rate": 9.654384895920434e-06, "loss": 1.8987, "step": 1270 }, { "epoch": 0.14574016741199403, "grad_norm": 13.5625, "learning_rate": 9.65370616921152e-06, "loss": 1.8245, "step": 1271 }, { "epoch": 0.1458548331613347, "grad_norm": 14.5, "learning_rate": 9.653026800612547e-06, "loss": 1.4818, "step": 1272 }, { "epoch": 0.14596949891067537, "grad_norm": 10.8125, "learning_rate": 9.652346790217221e-06, "loss": 1.4989, "step": 1273 }, { "epoch": 0.14608416466001606, "grad_norm": 12.125, "learning_rate": 9.651666138119339e-06, "loss": 1.3226, "step": 1274 }, { "epoch": 0.14619883040935672, "grad_norm": 12.8125, "learning_rate": 9.65098484441278e-06, "loss": 1.9623, "step": 1275 }, { "epoch": 0.1463134961586974, "grad_norm": 13.9375, "learning_rate": 9.650302909191518e-06, "loss": 1.4114, "step": 1276 }, { "epoch": 0.14642816190803806, "grad_norm": 15.4375, "learning_rate": 9.649620332549612e-06, "loss": 2.1656, "step": 1277 }, { "epoch": 0.14654282765737875, "grad_norm": 11.875, "learning_rate": 9.648937114581213e-06, "loss": 1.7312, "step": 1278 }, { "epoch": 0.1466574934067194, "grad_norm": 12.375, "learning_rate": 9.648253255380554e-06, "loss": 2.1255, "step": 1279 }, { "epoch": 0.1467721591560601, "grad_norm": 29.625, "learning_rate": 9.647568755041962e-06, "loss": 1.7735, "step": 1280 }, { "epoch": 0.14688682490540075, "grad_norm": 11.875, "learning_rate": 9.646883613659852e-06, "loss": 1.6713, "step": 1281 }, { "epoch": 0.14700149065474144, "grad_norm": 12.1875, "learning_rate": 9.646197831328725e-06, "loss": 1.6277, "step": 1282 }, { "epoch": 0.1471161564040821, "grad_norm": 31.25, "learning_rate": 9.645511408143172e-06, "loss": 1.6255, "step": 1283 }, { "epoch": 0.14723082215342279, "grad_norm": 13.5625, "learning_rate": 9.644824344197871e-06, "loss": 1.4824, "step": 1284 }, { "epoch": 0.14734548790276344, "grad_norm": 12.25, "learning_rate": 9.644136639587592e-06, "loss": 1.6866, "step": 1285 }, { "epoch": 0.1474601536521041, "grad_norm": 11.9375, "learning_rate": 9.643448294407186e-06, "loss": 1.8189, "step": 1286 }, { "epoch": 0.1475748194014448, "grad_norm": 11.5, "learning_rate": 9.642759308751602e-06, "loss": 1.8362, "step": 1287 }, { "epoch": 0.14768948515078545, "grad_norm": 11.5, "learning_rate": 9.64206968271587e-06, "loss": 1.5275, "step": 1288 }, { "epoch": 0.14780415090012614, "grad_norm": 13.6875, "learning_rate": 9.64137941639511e-06, "loss": 1.861, "step": 1289 }, { "epoch": 0.1479188166494668, "grad_norm": 12.75, "learning_rate": 9.640688509884533e-06, "loss": 1.7554, "step": 1290 }, { "epoch": 0.14803348239880748, "grad_norm": 12.75, "learning_rate": 9.639996963279437e-06, "loss": 1.6169, "step": 1291 }, { "epoch": 0.14814814814814814, "grad_norm": 12.625, "learning_rate": 9.639304776675204e-06, "loss": 1.9232, "step": 1292 }, { "epoch": 0.14826281389748883, "grad_norm": 18.75, "learning_rate": 9.638611950167311e-06, "loss": 1.725, "step": 1293 }, { "epoch": 0.14837747964682949, "grad_norm": 14.9375, "learning_rate": 9.63791848385132e-06, "loss": 1.9111, "step": 1294 }, { "epoch": 0.14849214539617017, "grad_norm": 17.125, "learning_rate": 9.637224377822882e-06, "loss": 2.0328, "step": 1295 }, { "epoch": 0.14860681114551083, "grad_norm": 15.375, "learning_rate": 9.636529632177732e-06, "loss": 1.7454, "step": 1296 }, { "epoch": 0.14872147689485152, "grad_norm": 15.625, "learning_rate": 9.635834247011702e-06, "loss": 2.2087, "step": 1297 }, { "epoch": 0.14883614264419218, "grad_norm": 13.875, "learning_rate": 9.635138222420704e-06, "loss": 2.1181, "step": 1298 }, { "epoch": 0.14895080839353286, "grad_norm": 13.9375, "learning_rate": 9.63444155850074e-06, "loss": 1.8491, "step": 1299 }, { "epoch": 0.14906547414287352, "grad_norm": 12.4375, "learning_rate": 9.633744255347905e-06, "loss": 1.8049, "step": 1300 }, { "epoch": 0.1491801398922142, "grad_norm": 15.25, "learning_rate": 9.633046313058378e-06, "loss": 1.581, "step": 1301 }, { "epoch": 0.14929480564155487, "grad_norm": 12.6875, "learning_rate": 9.632347731728423e-06, "loss": 1.7515, "step": 1302 }, { "epoch": 0.14940947139089553, "grad_norm": 13.6875, "learning_rate": 9.631648511454403e-06, "loss": 2.1483, "step": 1303 }, { "epoch": 0.1495241371402362, "grad_norm": 17.875, "learning_rate": 9.630948652332757e-06, "loss": 1.8035, "step": 1304 }, { "epoch": 0.14963880288957687, "grad_norm": 11.875, "learning_rate": 9.630248154460017e-06, "loss": 1.9476, "step": 1305 }, { "epoch": 0.14975346863891756, "grad_norm": 12.5, "learning_rate": 9.629547017932806e-06, "loss": 1.7522, "step": 1306 }, { "epoch": 0.14986813438825822, "grad_norm": 14.0625, "learning_rate": 9.62884524284783e-06, "loss": 1.8603, "step": 1307 }, { "epoch": 0.1499828001375989, "grad_norm": 14.1875, "learning_rate": 9.628142829301886e-06, "loss": 1.5544, "step": 1308 }, { "epoch": 0.15009746588693956, "grad_norm": 29.25, "learning_rate": 9.62743977739186e-06, "loss": 1.9828, "step": 1309 }, { "epoch": 0.15021213163628025, "grad_norm": 10.8125, "learning_rate": 9.626736087214724e-06, "loss": 1.7291, "step": 1310 }, { "epoch": 0.1503267973856209, "grad_norm": 12.875, "learning_rate": 9.626031758867538e-06, "loss": 1.8453, "step": 1311 }, { "epoch": 0.1504414631349616, "grad_norm": 13.625, "learning_rate": 9.62532679244745e-06, "loss": 2.0915, "step": 1312 }, { "epoch": 0.15055612888430225, "grad_norm": 14.6875, "learning_rate": 9.6246211880517e-06, "loss": 2.296, "step": 1313 }, { "epoch": 0.15067079463364294, "grad_norm": 17.75, "learning_rate": 9.62391494577761e-06, "loss": 2.3831, "step": 1314 }, { "epoch": 0.1507854603829836, "grad_norm": 18.125, "learning_rate": 9.623208065722592e-06, "loss": 1.5871, "step": 1315 }, { "epoch": 0.15090012613232429, "grad_norm": 21.125, "learning_rate": 9.622500547984148e-06, "loss": 2.4853, "step": 1316 }, { "epoch": 0.15101479188166494, "grad_norm": 18.0, "learning_rate": 9.621792392659866e-06, "loss": 1.902, "step": 1317 }, { "epoch": 0.15112945763100563, "grad_norm": 11.25, "learning_rate": 9.621083599847424e-06, "loss": 1.6003, "step": 1318 }, { "epoch": 0.1512441233803463, "grad_norm": 12.0625, "learning_rate": 9.620374169644584e-06, "loss": 2.0391, "step": 1319 }, { "epoch": 0.15135878912968695, "grad_norm": 14.5625, "learning_rate": 9.6196641021492e-06, "loss": 1.7461, "step": 1320 }, { "epoch": 0.15147345487902764, "grad_norm": 12.125, "learning_rate": 9.618953397459211e-06, "loss": 1.5988, "step": 1321 }, { "epoch": 0.1515881206283683, "grad_norm": 11.5, "learning_rate": 9.618242055672649e-06, "loss": 1.5787, "step": 1322 }, { "epoch": 0.15170278637770898, "grad_norm": 12.0, "learning_rate": 9.617530076887625e-06, "loss": 2.0692, "step": 1323 }, { "epoch": 0.15181745212704964, "grad_norm": 11.9375, "learning_rate": 9.616817461202346e-06, "loss": 1.8064, "step": 1324 }, { "epoch": 0.15193211787639033, "grad_norm": 14.4375, "learning_rate": 9.616104208715102e-06, "loss": 1.6715, "step": 1325 }, { "epoch": 0.15204678362573099, "grad_norm": 13.1875, "learning_rate": 9.615390319524272e-06, "loss": 2.1912, "step": 1326 }, { "epoch": 0.15216144937507167, "grad_norm": 17.5, "learning_rate": 9.614675793728328e-06, "loss": 1.7462, "step": 1327 }, { "epoch": 0.15227611512441233, "grad_norm": 10.8125, "learning_rate": 9.613960631425818e-06, "loss": 1.6391, "step": 1328 }, { "epoch": 0.15239078087375302, "grad_norm": 12.375, "learning_rate": 9.613244832715392e-06, "loss": 1.8139, "step": 1329 }, { "epoch": 0.15250544662309368, "grad_norm": 12.75, "learning_rate": 9.612528397695777e-06, "loss": 1.7977, "step": 1330 }, { "epoch": 0.15262011237243436, "grad_norm": 14.875, "learning_rate": 9.611811326465792e-06, "loss": 2.2173, "step": 1331 }, { "epoch": 0.15273477812177502, "grad_norm": 11.9375, "learning_rate": 9.611093619124344e-06, "loss": 1.7366, "step": 1332 }, { "epoch": 0.1528494438711157, "grad_norm": 12.5625, "learning_rate": 9.610375275770428e-06, "loss": 1.7073, "step": 1333 }, { "epoch": 0.15296410962045637, "grad_norm": 14.25, "learning_rate": 9.609656296503121e-06, "loss": 1.7786, "step": 1334 }, { "epoch": 0.15307877536979705, "grad_norm": 14.9375, "learning_rate": 9.6089366814216e-06, "loss": 1.9292, "step": 1335 }, { "epoch": 0.1531934411191377, "grad_norm": 12.5, "learning_rate": 9.608216430625114e-06, "loss": 1.7572, "step": 1336 }, { "epoch": 0.15330810686847837, "grad_norm": 11.625, "learning_rate": 9.607495544213014e-06, "loss": 1.6824, "step": 1337 }, { "epoch": 0.15342277261781906, "grad_norm": 30.375, "learning_rate": 9.60677402228473e-06, "loss": 1.9588, "step": 1338 }, { "epoch": 0.15353743836715972, "grad_norm": 14.75, "learning_rate": 9.606051864939785e-06, "loss": 1.9915, "step": 1339 }, { "epoch": 0.1536521041165004, "grad_norm": 10.75, "learning_rate": 9.605329072277783e-06, "loss": 1.604, "step": 1340 }, { "epoch": 0.15376676986584106, "grad_norm": 12.5, "learning_rate": 9.60460564439842e-06, "loss": 1.769, "step": 1341 }, { "epoch": 0.15388143561518175, "grad_norm": 10.9375, "learning_rate": 9.603881581401481e-06, "loss": 1.8654, "step": 1342 }, { "epoch": 0.1539961013645224, "grad_norm": 13.5625, "learning_rate": 9.603156883386837e-06, "loss": 1.7413, "step": 1343 }, { "epoch": 0.1541107671138631, "grad_norm": 16.0, "learning_rate": 9.602431550454443e-06, "loss": 1.6749, "step": 1344 }, { "epoch": 0.15422543286320375, "grad_norm": 13.625, "learning_rate": 9.601705582704348e-06, "loss": 1.9276, "step": 1345 }, { "epoch": 0.15434009861254444, "grad_norm": 10.3125, "learning_rate": 9.600978980236684e-06, "loss": 1.976, "step": 1346 }, { "epoch": 0.1544547643618851, "grad_norm": 11.0625, "learning_rate": 9.600251743151672e-06, "loss": 1.3372, "step": 1347 }, { "epoch": 0.1545694301112258, "grad_norm": 11.3125, "learning_rate": 9.599523871549623e-06, "loss": 1.7511, "step": 1348 }, { "epoch": 0.15468409586056645, "grad_norm": 11.0, "learning_rate": 9.598795365530929e-06, "loss": 1.9712, "step": 1349 }, { "epoch": 0.15479876160990713, "grad_norm": 11.0, "learning_rate": 9.598066225196075e-06, "loss": 1.8125, "step": 1350 }, { "epoch": 0.1549134273592478, "grad_norm": 14.875, "learning_rate": 9.597336450645633e-06, "loss": 1.8774, "step": 1351 }, { "epoch": 0.15502809310858848, "grad_norm": 17.375, "learning_rate": 9.596606041980261e-06, "loss": 2.1976, "step": 1352 }, { "epoch": 0.15514275885792914, "grad_norm": 12.875, "learning_rate": 9.595874999300704e-06, "loss": 1.6109, "step": 1353 }, { "epoch": 0.1552574246072698, "grad_norm": 14.1875, "learning_rate": 9.595143322707796e-06, "loss": 1.705, "step": 1354 }, { "epoch": 0.15537209035661048, "grad_norm": 14.0, "learning_rate": 9.59441101230246e-06, "loss": 1.739, "step": 1355 }, { "epoch": 0.15548675610595114, "grad_norm": 11.75, "learning_rate": 9.5936780681857e-06, "loss": 1.8954, "step": 1356 }, { "epoch": 0.15560142185529183, "grad_norm": 12.0625, "learning_rate": 9.592944490458615e-06, "loss": 1.9942, "step": 1357 }, { "epoch": 0.15571608760463249, "grad_norm": 12.6875, "learning_rate": 9.592210279222387e-06, "loss": 1.5335, "step": 1358 }, { "epoch": 0.15583075335397317, "grad_norm": 13.9375, "learning_rate": 9.591475434578286e-06, "loss": 1.6272, "step": 1359 }, { "epoch": 0.15594541910331383, "grad_norm": 14.8125, "learning_rate": 9.590739956627672e-06, "loss": 1.8869, "step": 1360 }, { "epoch": 0.15606008485265452, "grad_norm": 10.125, "learning_rate": 9.590003845471987e-06, "loss": 1.5469, "step": 1361 }, { "epoch": 0.15617475060199518, "grad_norm": 11.6875, "learning_rate": 9.589267101212764e-06, "loss": 1.7743, "step": 1362 }, { "epoch": 0.15628941635133586, "grad_norm": 11.75, "learning_rate": 9.588529723951626e-06, "loss": 1.668, "step": 1363 }, { "epoch": 0.15640408210067652, "grad_norm": 11.6875, "learning_rate": 9.587791713790277e-06, "loss": 1.7361, "step": 1364 }, { "epoch": 0.1565187478500172, "grad_norm": 12.5, "learning_rate": 9.587053070830513e-06, "loss": 1.4028, "step": 1365 }, { "epoch": 0.15663341359935787, "grad_norm": 16.125, "learning_rate": 9.586313795174213e-06, "loss": 1.4808, "step": 1366 }, { "epoch": 0.15674807934869855, "grad_norm": 26.875, "learning_rate": 9.585573886923349e-06, "loss": 1.9034, "step": 1367 }, { "epoch": 0.1568627450980392, "grad_norm": 11.6875, "learning_rate": 9.584833346179977e-06, "loss": 2.1225, "step": 1368 }, { "epoch": 0.1569774108473799, "grad_norm": 14.5, "learning_rate": 9.58409217304624e-06, "loss": 1.7538, "step": 1369 }, { "epoch": 0.15709207659672056, "grad_norm": 13.0, "learning_rate": 9.583350367624366e-06, "loss": 1.773, "step": 1370 }, { "epoch": 0.15720674234606122, "grad_norm": 14.6875, "learning_rate": 9.582607930016678e-06, "loss": 1.8415, "step": 1371 }, { "epoch": 0.1573214080954019, "grad_norm": 29.0, "learning_rate": 9.581864860325578e-06, "loss": 1.6952, "step": 1372 }, { "epoch": 0.15743607384474256, "grad_norm": 14.625, "learning_rate": 9.581121158653558e-06, "loss": 2.1146, "step": 1373 }, { "epoch": 0.15755073959408325, "grad_norm": 13.4375, "learning_rate": 9.5803768251032e-06, "loss": 1.6267, "step": 1374 }, { "epoch": 0.1576654053434239, "grad_norm": 14.625, "learning_rate": 9.579631859777168e-06, "loss": 1.6795, "step": 1375 }, { "epoch": 0.1577800710927646, "grad_norm": 12.5625, "learning_rate": 9.578886262778215e-06, "loss": 1.8449, "step": 1376 }, { "epoch": 0.15789473684210525, "grad_norm": 13.25, "learning_rate": 9.578140034209186e-06, "loss": 1.7466, "step": 1377 }, { "epoch": 0.15800940259144594, "grad_norm": 14.625, "learning_rate": 9.577393174173005e-06, "loss": 1.5783, "step": 1378 }, { "epoch": 0.1581240683407866, "grad_norm": 11.9375, "learning_rate": 9.57664568277269e-06, "loss": 1.5983, "step": 1379 }, { "epoch": 0.1582387340901273, "grad_norm": 11.1875, "learning_rate": 9.57589756011134e-06, "loss": 1.737, "step": 1380 }, { "epoch": 0.15835339983946795, "grad_norm": 13.6875, "learning_rate": 9.575148806292147e-06, "loss": 2.0276, "step": 1381 }, { "epoch": 0.15846806558880863, "grad_norm": 11.8125, "learning_rate": 9.574399421418388e-06, "loss": 1.6448, "step": 1382 }, { "epoch": 0.1585827313381493, "grad_norm": 11.875, "learning_rate": 9.573649405593422e-06, "loss": 1.4722, "step": 1383 }, { "epoch": 0.15869739708748998, "grad_norm": 11.6875, "learning_rate": 9.572898758920705e-06, "loss": 1.6673, "step": 1384 }, { "epoch": 0.15881206283683064, "grad_norm": 14.1875, "learning_rate": 9.572147481503771e-06, "loss": 1.8416, "step": 1385 }, { "epoch": 0.15892672858617132, "grad_norm": 11.625, "learning_rate": 9.571395573446243e-06, "loss": 1.595, "step": 1386 }, { "epoch": 0.15904139433551198, "grad_norm": 13.8125, "learning_rate": 9.570643034851835e-06, "loss": 1.7055, "step": 1387 }, { "epoch": 0.15915606008485264, "grad_norm": 13.125, "learning_rate": 9.569889865824346e-06, "loss": 1.6222, "step": 1388 }, { "epoch": 0.15927072583419333, "grad_norm": 11.375, "learning_rate": 9.56913606646766e-06, "loss": 1.7856, "step": 1389 }, { "epoch": 0.159385391583534, "grad_norm": 15.125, "learning_rate": 9.568381636885748e-06, "loss": 1.6637, "step": 1390 }, { "epoch": 0.15950005733287467, "grad_norm": 14.25, "learning_rate": 9.567626577182671e-06, "loss": 1.7867, "step": 1391 }, { "epoch": 0.15961472308221533, "grad_norm": 11.125, "learning_rate": 9.566870887462573e-06, "loss": 1.6357, "step": 1392 }, { "epoch": 0.15972938883155602, "grad_norm": 13.0625, "learning_rate": 9.566114567829691e-06, "loss": 1.7498, "step": 1393 }, { "epoch": 0.15984405458089668, "grad_norm": 12.875, "learning_rate": 9.565357618388342e-06, "loss": 1.6925, "step": 1394 }, { "epoch": 0.15995872033023736, "grad_norm": 12.6875, "learning_rate": 9.564600039242933e-06, "loss": 1.7409, "step": 1395 }, { "epoch": 0.16007338607957802, "grad_norm": 15.4375, "learning_rate": 9.563841830497958e-06, "loss": 1.7161, "step": 1396 }, { "epoch": 0.1601880518289187, "grad_norm": 14.6875, "learning_rate": 9.563082992257997e-06, "loss": 2.0338, "step": 1397 }, { "epoch": 0.16030271757825937, "grad_norm": 13.5, "learning_rate": 9.562323524627718e-06, "loss": 1.9106, "step": 1398 }, { "epoch": 0.16041738332760005, "grad_norm": 9.875, "learning_rate": 9.561563427711872e-06, "loss": 1.5442, "step": 1399 }, { "epoch": 0.1605320490769407, "grad_norm": 12.5, "learning_rate": 9.560802701615305e-06, "loss": 1.6619, "step": 1400 }, { "epoch": 0.1606467148262814, "grad_norm": 11.1875, "learning_rate": 9.56004134644294e-06, "loss": 1.8523, "step": 1401 }, { "epoch": 0.16076138057562206, "grad_norm": 14.375, "learning_rate": 9.559279362299796e-06, "loss": 1.8179, "step": 1402 }, { "epoch": 0.16087604632496275, "grad_norm": 13.125, "learning_rate": 9.558516749290972e-06, "loss": 1.8379, "step": 1403 }, { "epoch": 0.1609907120743034, "grad_norm": 13.0625, "learning_rate": 9.557753507521654e-06, "loss": 1.9272, "step": 1404 }, { "epoch": 0.16110537782364406, "grad_norm": 13.5, "learning_rate": 9.556989637097119e-06, "loss": 1.9107, "step": 1405 }, { "epoch": 0.16122004357298475, "grad_norm": 13.4375, "learning_rate": 9.556225138122727e-06, "loss": 1.7487, "step": 1406 }, { "epoch": 0.1613347093223254, "grad_norm": 13.75, "learning_rate": 9.555460010703927e-06, "loss": 1.9008, "step": 1407 }, { "epoch": 0.1614493750716661, "grad_norm": 13.6875, "learning_rate": 9.554694254946253e-06, "loss": 1.6617, "step": 1408 }, { "epoch": 0.16156404082100675, "grad_norm": 12.0625, "learning_rate": 9.553927870955328e-06, "loss": 1.8055, "step": 1409 }, { "epoch": 0.16167870657034744, "grad_norm": 15.4375, "learning_rate": 9.553160858836859e-06, "loss": 1.64, "step": 1410 }, { "epoch": 0.1617933723196881, "grad_norm": 10.75, "learning_rate": 9.55239321869664e-06, "loss": 1.7448, "step": 1411 }, { "epoch": 0.1619080380690288, "grad_norm": 15.5, "learning_rate": 9.551624950640553e-06, "loss": 1.9348, "step": 1412 }, { "epoch": 0.16202270381836945, "grad_norm": 10.0625, "learning_rate": 9.550856054774566e-06, "loss": 1.6023, "step": 1413 }, { "epoch": 0.16213736956771013, "grad_norm": 23.0, "learning_rate": 9.550086531204734e-06, "loss": 1.7037, "step": 1414 }, { "epoch": 0.1622520353170508, "grad_norm": 15.5, "learning_rate": 9.549316380037197e-06, "loss": 2.0024, "step": 1415 }, { "epoch": 0.16236670106639148, "grad_norm": 10.4375, "learning_rate": 9.548545601378183e-06, "loss": 1.7243, "step": 1416 }, { "epoch": 0.16248136681573214, "grad_norm": 17.375, "learning_rate": 9.547774195334008e-06, "loss": 1.7215, "step": 1417 }, { "epoch": 0.16259603256507282, "grad_norm": 9.625, "learning_rate": 9.54700216201107e-06, "loss": 1.7821, "step": 1418 }, { "epoch": 0.16271069831441348, "grad_norm": 13.125, "learning_rate": 9.546229501515856e-06, "loss": 1.9358, "step": 1419 }, { "epoch": 0.16282536406375417, "grad_norm": 17.375, "learning_rate": 9.545456213954944e-06, "loss": 1.9169, "step": 1420 }, { "epoch": 0.16294002981309483, "grad_norm": 14.75, "learning_rate": 9.544682299434991e-06, "loss": 1.9621, "step": 1421 }, { "epoch": 0.16305469556243551, "grad_norm": 14.0, "learning_rate": 9.543907758062742e-06, "loss": 2.0945, "step": 1422 }, { "epoch": 0.16316936131177617, "grad_norm": 12.6875, "learning_rate": 9.543132589945035e-06, "loss": 1.8022, "step": 1423 }, { "epoch": 0.16328402706111683, "grad_norm": 13.875, "learning_rate": 9.542356795188786e-06, "loss": 1.821, "step": 1424 }, { "epoch": 0.16339869281045752, "grad_norm": 13.3125, "learning_rate": 9.541580373901003e-06, "loss": 1.8428, "step": 1425 }, { "epoch": 0.16351335855979818, "grad_norm": 9.5625, "learning_rate": 9.540803326188778e-06, "loss": 1.7844, "step": 1426 }, { "epoch": 0.16362802430913886, "grad_norm": 14.0, "learning_rate": 9.540025652159288e-06, "loss": 1.7973, "step": 1427 }, { "epoch": 0.16374269005847952, "grad_norm": 11.75, "learning_rate": 9.539247351919803e-06, "loss": 2.0235, "step": 1428 }, { "epoch": 0.1638573558078202, "grad_norm": 13.5625, "learning_rate": 9.53846842557767e-06, "loss": 2.1746, "step": 1429 }, { "epoch": 0.16397202155716087, "grad_norm": 26.375, "learning_rate": 9.537688873240329e-06, "loss": 2.1984, "step": 1430 }, { "epoch": 0.16408668730650156, "grad_norm": 12.375, "learning_rate": 9.536908695015304e-06, "loss": 1.7548, "step": 1431 }, { "epoch": 0.16420135305584221, "grad_norm": 12.625, "learning_rate": 9.536127891010206e-06, "loss": 1.844, "step": 1432 }, { "epoch": 0.1643160188051829, "grad_norm": 13.3125, "learning_rate": 9.535346461332731e-06, "loss": 1.6944, "step": 1433 }, { "epoch": 0.16443068455452356, "grad_norm": 10.875, "learning_rate": 9.534564406090665e-06, "loss": 1.535, "step": 1434 }, { "epoch": 0.16454535030386425, "grad_norm": 14.375, "learning_rate": 9.533781725391874e-06, "loss": 1.8832, "step": 1435 }, { "epoch": 0.1646600160532049, "grad_norm": 13.9375, "learning_rate": 9.532998419344316e-06, "loss": 1.5788, "step": 1436 }, { "epoch": 0.1647746818025456, "grad_norm": 14.875, "learning_rate": 9.532214488056034e-06, "loss": 1.9203, "step": 1437 }, { "epoch": 0.16488934755188625, "grad_norm": 13.25, "learning_rate": 9.531429931635155e-06, "loss": 1.5835, "step": 1438 }, { "epoch": 0.16500401330122694, "grad_norm": 11.125, "learning_rate": 9.530644750189892e-06, "loss": 1.714, "step": 1439 }, { "epoch": 0.1651186790505676, "grad_norm": 12.125, "learning_rate": 9.529858943828549e-06, "loss": 1.523, "step": 1440 }, { "epoch": 0.16523334479990825, "grad_norm": 16.25, "learning_rate": 9.529072512659514e-06, "loss": 1.8888, "step": 1441 }, { "epoch": 0.16534801054924894, "grad_norm": 16.625, "learning_rate": 9.528285456791254e-06, "loss": 2.0465, "step": 1442 }, { "epoch": 0.1654626762985896, "grad_norm": 12.375, "learning_rate": 9.527497776332335e-06, "loss": 1.8652, "step": 1443 }, { "epoch": 0.1655773420479303, "grad_norm": 12.5625, "learning_rate": 9.526709471391397e-06, "loss": 1.2759, "step": 1444 }, { "epoch": 0.16569200779727095, "grad_norm": 13.125, "learning_rate": 9.525920542077176e-06, "loss": 1.7387, "step": 1445 }, { "epoch": 0.16580667354661163, "grad_norm": 12.125, "learning_rate": 9.525130988498488e-06, "loss": 1.448, "step": 1446 }, { "epoch": 0.1659213392959523, "grad_norm": 12.75, "learning_rate": 9.524340810764237e-06, "loss": 2.1273, "step": 1447 }, { "epoch": 0.16603600504529298, "grad_norm": 14.6875, "learning_rate": 9.523550008983413e-06, "loss": 1.6669, "step": 1448 }, { "epoch": 0.16615067079463364, "grad_norm": 17.5, "learning_rate": 9.522758583265093e-06, "loss": 2.1839, "step": 1449 }, { "epoch": 0.16626533654397432, "grad_norm": 12.625, "learning_rate": 9.521966533718437e-06, "loss": 1.6362, "step": 1450 }, { "epoch": 0.16638000229331498, "grad_norm": 15.5, "learning_rate": 9.521173860452694e-06, "loss": 2.1356, "step": 1451 }, { "epoch": 0.16649466804265567, "grad_norm": 11.8125, "learning_rate": 9.520380563577198e-06, "loss": 1.8749, "step": 1452 }, { "epoch": 0.16660933379199633, "grad_norm": 10.8125, "learning_rate": 9.519586643201371e-06, "loss": 1.6701, "step": 1453 }, { "epoch": 0.16672399954133701, "grad_norm": 13.4375, "learning_rate": 9.518792099434716e-06, "loss": 1.6679, "step": 1454 }, { "epoch": 0.16683866529067767, "grad_norm": 14.125, "learning_rate": 9.517996932386828e-06, "loss": 1.9644, "step": 1455 }, { "epoch": 0.16695333104001836, "grad_norm": 14.4375, "learning_rate": 9.517201142167384e-06, "loss": 1.7435, "step": 1456 }, { "epoch": 0.16706799678935902, "grad_norm": 12.0, "learning_rate": 9.516404728886148e-06, "loss": 1.9578, "step": 1457 }, { "epoch": 0.16718266253869968, "grad_norm": 15.375, "learning_rate": 9.515607692652971e-06, "loss": 1.8886, "step": 1458 }, { "epoch": 0.16729732828804036, "grad_norm": 12.375, "learning_rate": 9.514810033577787e-06, "loss": 1.9214, "step": 1459 }, { "epoch": 0.16741199403738102, "grad_norm": 9.25, "learning_rate": 9.514011751770619e-06, "loss": 1.339, "step": 1460 }, { "epoch": 0.1675266597867217, "grad_norm": 12.75, "learning_rate": 9.513212847341574e-06, "loss": 1.7193, "step": 1461 }, { "epoch": 0.16764132553606237, "grad_norm": 10.0, "learning_rate": 9.512413320400847e-06, "loss": 1.6918, "step": 1462 }, { "epoch": 0.16775599128540306, "grad_norm": 11.75, "learning_rate": 9.511613171058718e-06, "loss": 1.833, "step": 1463 }, { "epoch": 0.16787065703474371, "grad_norm": 14.6875, "learning_rate": 9.51081239942555e-06, "loss": 1.7963, "step": 1464 }, { "epoch": 0.1679853227840844, "grad_norm": 13.1875, "learning_rate": 9.510011005611797e-06, "loss": 1.7878, "step": 1465 }, { "epoch": 0.16809998853342506, "grad_norm": 14.1875, "learning_rate": 9.50920898972799e-06, "loss": 2.1258, "step": 1466 }, { "epoch": 0.16821465428276575, "grad_norm": 12.375, "learning_rate": 9.50840635188476e-06, "loss": 1.8086, "step": 1467 }, { "epoch": 0.1683293200321064, "grad_norm": 32.75, "learning_rate": 9.507603092192813e-06, "loss": 1.8962, "step": 1468 }, { "epoch": 0.1684439857814471, "grad_norm": 13.5625, "learning_rate": 9.50679921076294e-06, "loss": 1.9286, "step": 1469 }, { "epoch": 0.16855865153078775, "grad_norm": 10.5625, "learning_rate": 9.505994707706023e-06, "loss": 1.5591, "step": 1470 }, { "epoch": 0.16867331728012844, "grad_norm": 12.625, "learning_rate": 9.505189583133029e-06, "loss": 1.5203, "step": 1471 }, { "epoch": 0.1687879830294691, "grad_norm": 13.1875, "learning_rate": 9.504383837155008e-06, "loss": 1.542, "step": 1472 }, { "epoch": 0.16890264877880978, "grad_norm": 13.6875, "learning_rate": 9.503577469883098e-06, "loss": 1.7323, "step": 1473 }, { "epoch": 0.16901731452815044, "grad_norm": 10.375, "learning_rate": 9.502770481428526e-06, "loss": 1.5045, "step": 1474 }, { "epoch": 0.1691319802774911, "grad_norm": 12.125, "learning_rate": 9.501962871902591e-06, "loss": 1.5782, "step": 1475 }, { "epoch": 0.1692466460268318, "grad_norm": 12.4375, "learning_rate": 9.501154641416696e-06, "loss": 1.6031, "step": 1476 }, { "epoch": 0.16936131177617245, "grad_norm": 12.375, "learning_rate": 9.500345790082317e-06, "loss": 1.845, "step": 1477 }, { "epoch": 0.16947597752551313, "grad_norm": 11.6875, "learning_rate": 9.49953631801102e-06, "loss": 1.5139, "step": 1478 }, { "epoch": 0.1695906432748538, "grad_norm": 11.0625, "learning_rate": 9.49872622531446e-06, "loss": 1.6699, "step": 1479 }, { "epoch": 0.16970530902419448, "grad_norm": 12.625, "learning_rate": 9.497915512104367e-06, "loss": 1.8902, "step": 1480 }, { "epoch": 0.16981997477353514, "grad_norm": 15.0, "learning_rate": 9.497104178492568e-06, "loss": 1.6722, "step": 1481 }, { "epoch": 0.16993464052287582, "grad_norm": 10.8125, "learning_rate": 9.496292224590973e-06, "loss": 1.7091, "step": 1482 }, { "epoch": 0.17004930627221648, "grad_norm": 11.0625, "learning_rate": 9.49547965051157e-06, "loss": 1.6519, "step": 1483 }, { "epoch": 0.17016397202155717, "grad_norm": 11.75, "learning_rate": 9.494666456366442e-06, "loss": 1.7311, "step": 1484 }, { "epoch": 0.17027863777089783, "grad_norm": 13.875, "learning_rate": 9.493852642267751e-06, "loss": 1.4602, "step": 1485 }, { "epoch": 0.17039330352023851, "grad_norm": 12.3125, "learning_rate": 9.49303820832775e-06, "loss": 1.769, "step": 1486 }, { "epoch": 0.17050796926957917, "grad_norm": 15.4375, "learning_rate": 9.492223154658773e-06, "loss": 1.5085, "step": 1487 }, { "epoch": 0.17062263501891986, "grad_norm": 14.375, "learning_rate": 9.491407481373241e-06, "loss": 1.5327, "step": 1488 }, { "epoch": 0.17073730076826052, "grad_norm": 13.125, "learning_rate": 9.490591188583662e-06, "loss": 1.8789, "step": 1489 }, { "epoch": 0.1708519665176012, "grad_norm": 18.125, "learning_rate": 9.489774276402627e-06, "loss": 1.9776, "step": 1490 }, { "epoch": 0.17096663226694186, "grad_norm": 13.25, "learning_rate": 9.488956744942812e-06, "loss": 1.4229, "step": 1491 }, { "epoch": 0.17108129801628252, "grad_norm": 11.75, "learning_rate": 9.488138594316982e-06, "loss": 1.5008, "step": 1492 }, { "epoch": 0.1711959637656232, "grad_norm": 14.625, "learning_rate": 9.487319824637984e-06, "loss": 2.1677, "step": 1493 }, { "epoch": 0.17131062951496387, "grad_norm": 14.4375, "learning_rate": 9.486500436018752e-06, "loss": 2.0706, "step": 1494 }, { "epoch": 0.17142529526430456, "grad_norm": 11.3125, "learning_rate": 9.485680428572308e-06, "loss": 2.0094, "step": 1495 }, { "epoch": 0.17153996101364521, "grad_norm": 10.25, "learning_rate": 9.484859802411752e-06, "loss": 1.7147, "step": 1496 }, { "epoch": 0.1716546267629859, "grad_norm": 11.4375, "learning_rate": 9.484038557650275e-06, "loss": 1.6668, "step": 1497 }, { "epoch": 0.17176929251232656, "grad_norm": 13.0625, "learning_rate": 9.483216694401152e-06, "loss": 1.7786, "step": 1498 }, { "epoch": 0.17188395826166725, "grad_norm": 12.9375, "learning_rate": 9.482394212777745e-06, "loss": 1.8699, "step": 1499 }, { "epoch": 0.1719986240110079, "grad_norm": 15.75, "learning_rate": 9.4815711128935e-06, "loss": 2.0766, "step": 1500 }, { "epoch": 0.1721132897603486, "grad_norm": 13.875, "learning_rate": 9.480747394861944e-06, "loss": 1.7799, "step": 1501 }, { "epoch": 0.17222795550968925, "grad_norm": 13.75, "learning_rate": 9.479923058796696e-06, "loss": 1.6621, "step": 1502 }, { "epoch": 0.17234262125902994, "grad_norm": 14.6875, "learning_rate": 9.479098104811459e-06, "loss": 1.6569, "step": 1503 }, { "epoch": 0.1724572870083706, "grad_norm": 11.25, "learning_rate": 9.478272533020016e-06, "loss": 1.4285, "step": 1504 }, { "epoch": 0.17257195275771128, "grad_norm": 11.125, "learning_rate": 9.477446343536242e-06, "loss": 1.663, "step": 1505 }, { "epoch": 0.17268661850705194, "grad_norm": 14.25, "learning_rate": 9.476619536474093e-06, "loss": 1.4197, "step": 1506 }, { "epoch": 0.17280128425639263, "grad_norm": 18.375, "learning_rate": 9.475792111947608e-06, "loss": 1.9012, "step": 1507 }, { "epoch": 0.1729159500057333, "grad_norm": 13.25, "learning_rate": 9.47496407007092e-06, "loss": 1.6863, "step": 1508 }, { "epoch": 0.17303061575507395, "grad_norm": 13.0625, "learning_rate": 9.47413541095824e-06, "loss": 1.6165, "step": 1509 }, { "epoch": 0.17314528150441463, "grad_norm": 12.875, "learning_rate": 9.473306134723863e-06, "loss": 1.7524, "step": 1510 }, { "epoch": 0.1732599472537553, "grad_norm": 13.5625, "learning_rate": 9.472476241482174e-06, "loss": 1.5009, "step": 1511 }, { "epoch": 0.17337461300309598, "grad_norm": 12.1875, "learning_rate": 9.471645731347641e-06, "loss": 1.7962, "step": 1512 }, { "epoch": 0.17348927875243664, "grad_norm": 12.0, "learning_rate": 9.470814604434818e-06, "loss": 1.795, "step": 1513 }, { "epoch": 0.17360394450177732, "grad_norm": 12.5625, "learning_rate": 9.46998286085834e-06, "loss": 1.7181, "step": 1514 }, { "epoch": 0.17371861025111798, "grad_norm": 14.5625, "learning_rate": 9.469150500732933e-06, "loss": 1.9514, "step": 1515 }, { "epoch": 0.17383327600045867, "grad_norm": 13.1875, "learning_rate": 9.468317524173402e-06, "loss": 1.6696, "step": 1516 }, { "epoch": 0.17394794174979933, "grad_norm": 13.8125, "learning_rate": 9.467483931294644e-06, "loss": 1.8617, "step": 1517 }, { "epoch": 0.17406260749914002, "grad_norm": 14.75, "learning_rate": 9.466649722211635e-06, "loss": 1.5357, "step": 1518 }, { "epoch": 0.17417727324848067, "grad_norm": 13.375, "learning_rate": 9.46581489703944e-06, "loss": 1.7812, "step": 1519 }, { "epoch": 0.17429193899782136, "grad_norm": 18.75, "learning_rate": 9.464979455893206e-06, "loss": 1.8502, "step": 1520 }, { "epoch": 0.17440660474716202, "grad_norm": 15.625, "learning_rate": 9.464143398888166e-06, "loss": 1.7213, "step": 1521 }, { "epoch": 0.1745212704965027, "grad_norm": 14.9375, "learning_rate": 9.463306726139639e-06, "loss": 2.0317, "step": 1522 }, { "epoch": 0.17463593624584337, "grad_norm": 11.8125, "learning_rate": 9.462469437763027e-06, "loss": 1.8141, "step": 1523 }, { "epoch": 0.17475060199518405, "grad_norm": 14.5625, "learning_rate": 9.461631533873819e-06, "loss": 1.6099, "step": 1524 }, { "epoch": 0.1748652677445247, "grad_norm": 13.125, "learning_rate": 9.460793014587586e-06, "loss": 1.7277, "step": 1525 }, { "epoch": 0.17497993349386537, "grad_norm": 12.8125, "learning_rate": 9.459953880019987e-06, "loss": 1.6055, "step": 1526 }, { "epoch": 0.17509459924320606, "grad_norm": 14.0625, "learning_rate": 9.459114130286766e-06, "loss": 2.0125, "step": 1527 }, { "epoch": 0.17520926499254671, "grad_norm": 25.25, "learning_rate": 9.45827376550375e-06, "loss": 1.7028, "step": 1528 }, { "epoch": 0.1753239307418874, "grad_norm": 12.375, "learning_rate": 9.457432785786849e-06, "loss": 1.5205, "step": 1529 }, { "epoch": 0.17543859649122806, "grad_norm": 23.25, "learning_rate": 9.456591191252061e-06, "loss": 1.8271, "step": 1530 }, { "epoch": 0.17555326224056875, "grad_norm": 11.5, "learning_rate": 9.455748982015468e-06, "loss": 1.6701, "step": 1531 }, { "epoch": 0.1756679279899094, "grad_norm": 16.0, "learning_rate": 9.45490615819324e-06, "loss": 1.6374, "step": 1532 }, { "epoch": 0.1757825937392501, "grad_norm": 11.8125, "learning_rate": 9.454062719901624e-06, "loss": 1.9208, "step": 1533 }, { "epoch": 0.17589725948859075, "grad_norm": 15.0, "learning_rate": 9.453218667256959e-06, "loss": 1.6911, "step": 1534 }, { "epoch": 0.17601192523793144, "grad_norm": 11.9375, "learning_rate": 9.452374000375664e-06, "loss": 1.7891, "step": 1535 }, { "epoch": 0.1761265909872721, "grad_norm": 11.0625, "learning_rate": 9.451528719374246e-06, "loss": 1.7942, "step": 1536 }, { "epoch": 0.17624125673661278, "grad_norm": 11.5625, "learning_rate": 9.450682824369295e-06, "loss": 1.8566, "step": 1537 }, { "epoch": 0.17635592248595344, "grad_norm": 14.0, "learning_rate": 9.449836315477485e-06, "loss": 1.7913, "step": 1538 }, { "epoch": 0.17647058823529413, "grad_norm": 12.125, "learning_rate": 9.448989192815579e-06, "loss": 1.4313, "step": 1539 }, { "epoch": 0.1765852539846348, "grad_norm": 28.75, "learning_rate": 9.448141456500417e-06, "loss": 1.8567, "step": 1540 }, { "epoch": 0.17669991973397547, "grad_norm": 14.75, "learning_rate": 9.447293106648932e-06, "loss": 1.7436, "step": 1541 }, { "epoch": 0.17681458548331613, "grad_norm": 12.4375, "learning_rate": 9.446444143378135e-06, "loss": 1.691, "step": 1542 }, { "epoch": 0.1769292512326568, "grad_norm": 27.5, "learning_rate": 9.445594566805126e-06, "loss": 1.8251, "step": 1543 }, { "epoch": 0.17704391698199748, "grad_norm": 14.5625, "learning_rate": 9.444744377047087e-06, "loss": 2.0932, "step": 1544 }, { "epoch": 0.17715858273133814, "grad_norm": 10.5, "learning_rate": 9.443893574221287e-06, "loss": 1.8196, "step": 1545 }, { "epoch": 0.17727324848067882, "grad_norm": 20.25, "learning_rate": 9.443042158445074e-06, "loss": 1.981, "step": 1546 }, { "epoch": 0.17738791423001948, "grad_norm": 13.5, "learning_rate": 9.44219012983589e-06, "loss": 1.687, "step": 1547 }, { "epoch": 0.17750257997936017, "grad_norm": 13.8125, "learning_rate": 9.441337488511253e-06, "loss": 1.586, "step": 1548 }, { "epoch": 0.17761724572870083, "grad_norm": 14.8125, "learning_rate": 9.440484234588767e-06, "loss": 1.6103, "step": 1549 }, { "epoch": 0.17773191147804152, "grad_norm": 13.875, "learning_rate": 9.439630368186126e-06, "loss": 1.548, "step": 1550 }, { "epoch": 0.17784657722738217, "grad_norm": 13.9375, "learning_rate": 9.438775889421102e-06, "loss": 1.8418, "step": 1551 }, { "epoch": 0.17796124297672286, "grad_norm": 12.8125, "learning_rate": 9.437920798411555e-06, "loss": 1.9715, "step": 1552 }, { "epoch": 0.17807590872606352, "grad_norm": 12.0, "learning_rate": 9.437065095275429e-06, "loss": 1.926, "step": 1553 }, { "epoch": 0.1781905744754042, "grad_norm": 13.1875, "learning_rate": 9.436208780130751e-06, "loss": 2.0215, "step": 1554 }, { "epoch": 0.17830524022474487, "grad_norm": 14.0, "learning_rate": 9.435351853095633e-06, "loss": 1.9137, "step": 1555 }, { "epoch": 0.17841990597408555, "grad_norm": 13.875, "learning_rate": 9.434494314288274e-06, "loss": 1.9516, "step": 1556 }, { "epoch": 0.1785345717234262, "grad_norm": 12.0625, "learning_rate": 9.433636163826952e-06, "loss": 2.1608, "step": 1557 }, { "epoch": 0.1786492374727669, "grad_norm": 13.75, "learning_rate": 9.432777401830034e-06, "loss": 1.4252, "step": 1558 }, { "epoch": 0.17876390322210756, "grad_norm": 12.625, "learning_rate": 9.43191802841597e-06, "loss": 1.3633, "step": 1559 }, { "epoch": 0.17887856897144822, "grad_norm": 14.9375, "learning_rate": 9.431058043703294e-06, "loss": 1.8159, "step": 1560 }, { "epoch": 0.1789932347207889, "grad_norm": 13.8125, "learning_rate": 9.430197447810625e-06, "loss": 2.0832, "step": 1561 }, { "epoch": 0.17910790047012956, "grad_norm": 12.1875, "learning_rate": 9.429336240856664e-06, "loss": 1.5289, "step": 1562 }, { "epoch": 0.17922256621947025, "grad_norm": 16.875, "learning_rate": 9.428474422960202e-06, "loss": 1.9804, "step": 1563 }, { "epoch": 0.1793372319688109, "grad_norm": 11.75, "learning_rate": 9.427611994240105e-06, "loss": 1.7318, "step": 1564 }, { "epoch": 0.1794518977181516, "grad_norm": 13.0, "learning_rate": 9.426748954815332e-06, "loss": 1.9764, "step": 1565 }, { "epoch": 0.17956656346749225, "grad_norm": 9.6875, "learning_rate": 9.425885304804923e-06, "loss": 1.4297, "step": 1566 }, { "epoch": 0.17968122921683294, "grad_norm": 27.375, "learning_rate": 9.425021044328e-06, "loss": 1.763, "step": 1567 }, { "epoch": 0.1797958949661736, "grad_norm": 17.375, "learning_rate": 9.424156173503774e-06, "loss": 1.8911, "step": 1568 }, { "epoch": 0.17991056071551428, "grad_norm": 11.3125, "learning_rate": 9.423290692451536e-06, "loss": 1.6393, "step": 1569 }, { "epoch": 0.18002522646485494, "grad_norm": 11.625, "learning_rate": 9.422424601290661e-06, "loss": 1.543, "step": 1570 }, { "epoch": 0.18013989221419563, "grad_norm": 15.125, "learning_rate": 9.421557900140612e-06, "loss": 1.7386, "step": 1571 }, { "epoch": 0.1802545579635363, "grad_norm": 12.9375, "learning_rate": 9.420690589120933e-06, "loss": 1.449, "step": 1572 }, { "epoch": 0.18036922371287697, "grad_norm": 14.5, "learning_rate": 9.419822668351255e-06, "loss": 2.1209, "step": 1573 }, { "epoch": 0.18048388946221763, "grad_norm": 28.125, "learning_rate": 9.418954137951289e-06, "loss": 1.7951, "step": 1574 }, { "epoch": 0.18059855521155832, "grad_norm": 14.625, "learning_rate": 9.418084998040831e-06, "loss": 1.7249, "step": 1575 }, { "epoch": 0.18071322096089898, "grad_norm": 12.5625, "learning_rate": 9.417215248739764e-06, "loss": 1.723, "step": 1576 }, { "epoch": 0.18082788671023964, "grad_norm": 14.625, "learning_rate": 9.416344890168055e-06, "loss": 1.8035, "step": 1577 }, { "epoch": 0.18094255245958032, "grad_norm": 13.8125, "learning_rate": 9.41547392244575e-06, "loss": 1.6969, "step": 1578 }, { "epoch": 0.18105721820892098, "grad_norm": 10.6875, "learning_rate": 9.414602345692984e-06, "loss": 1.7383, "step": 1579 }, { "epoch": 0.18117188395826167, "grad_norm": 14.5625, "learning_rate": 9.413730160029975e-06, "loss": 1.8839, "step": 1580 }, { "epoch": 0.18128654970760233, "grad_norm": 16.625, "learning_rate": 9.412857365577025e-06, "loss": 2.2347, "step": 1581 }, { "epoch": 0.18140121545694302, "grad_norm": 11.25, "learning_rate": 9.411983962454514e-06, "loss": 1.7212, "step": 1582 }, { "epoch": 0.18151588120628367, "grad_norm": 11.875, "learning_rate": 9.41110995078292e-06, "loss": 1.7582, "step": 1583 }, { "epoch": 0.18163054695562436, "grad_norm": 15.4375, "learning_rate": 9.410235330682788e-06, "loss": 2.0489, "step": 1584 }, { "epoch": 0.18174521270496502, "grad_norm": 17.375, "learning_rate": 9.409360102274762e-06, "loss": 1.7509, "step": 1585 }, { "epoch": 0.1818598784543057, "grad_norm": 14.875, "learning_rate": 9.408484265679559e-06, "loss": 1.8178, "step": 1586 }, { "epoch": 0.18197454420364637, "grad_norm": 13.875, "learning_rate": 9.407607821017984e-06, "loss": 1.7771, "step": 1587 }, { "epoch": 0.18208920995298705, "grad_norm": 13.25, "learning_rate": 9.406730768410927e-06, "loss": 1.8311, "step": 1588 }, { "epoch": 0.1822038757023277, "grad_norm": 16.125, "learning_rate": 9.405853107979361e-06, "loss": 1.8239, "step": 1589 }, { "epoch": 0.1823185414516684, "grad_norm": 15.0625, "learning_rate": 9.404974839844342e-06, "loss": 1.6419, "step": 1590 }, { "epoch": 0.18243320720100906, "grad_norm": 14.5, "learning_rate": 9.404095964127009e-06, "loss": 1.8178, "step": 1591 }, { "epoch": 0.18254787295034974, "grad_norm": 10.0625, "learning_rate": 9.40321648094859e-06, "loss": 1.901, "step": 1592 }, { "epoch": 0.1826625386996904, "grad_norm": 12.0625, "learning_rate": 9.402336390430389e-06, "loss": 1.7265, "step": 1593 }, { "epoch": 0.18277720444903106, "grad_norm": 13.3125, "learning_rate": 9.401455692693798e-06, "loss": 1.8407, "step": 1594 }, { "epoch": 0.18289187019837175, "grad_norm": 13.125, "learning_rate": 9.400574387860295e-06, "loss": 1.8663, "step": 1595 }, { "epoch": 0.1830065359477124, "grad_norm": 12.5625, "learning_rate": 9.399692476051436e-06, "loss": 2.2237, "step": 1596 }, { "epoch": 0.1831212016970531, "grad_norm": 12.6875, "learning_rate": 9.398809957388869e-06, "loss": 1.587, "step": 1597 }, { "epoch": 0.18323586744639375, "grad_norm": 15.375, "learning_rate": 9.397926831994316e-06, "loss": 1.9037, "step": 1598 }, { "epoch": 0.18335053319573444, "grad_norm": 13.5, "learning_rate": 9.397043099989587e-06, "loss": 1.76, "step": 1599 }, { "epoch": 0.1834651989450751, "grad_norm": 13.25, "learning_rate": 9.396158761496577e-06, "loss": 1.605, "step": 1600 }, { "epoch": 0.18357986469441578, "grad_norm": 16.25, "learning_rate": 9.395273816637267e-06, "loss": 1.6698, "step": 1601 }, { "epoch": 0.18369453044375644, "grad_norm": 14.125, "learning_rate": 9.394388265533714e-06, "loss": 1.7572, "step": 1602 }, { "epoch": 0.18380919619309713, "grad_norm": 12.4375, "learning_rate": 9.393502108308065e-06, "loss": 1.4404, "step": 1603 }, { "epoch": 0.1839238619424378, "grad_norm": 12.5625, "learning_rate": 9.392615345082547e-06, "loss": 1.8142, "step": 1604 }, { "epoch": 0.18403852769177848, "grad_norm": 16.75, "learning_rate": 9.391727975979474e-06, "loss": 1.9086, "step": 1605 }, { "epoch": 0.18415319344111913, "grad_norm": 12.5, "learning_rate": 9.390840001121239e-06, "loss": 1.4202, "step": 1606 }, { "epoch": 0.18426785919045982, "grad_norm": 13.25, "learning_rate": 9.389951420630325e-06, "loss": 1.5151, "step": 1607 }, { "epoch": 0.18438252493980048, "grad_norm": 18.875, "learning_rate": 9.389062234629293e-06, "loss": 1.3822, "step": 1608 }, { "epoch": 0.18449719068914117, "grad_norm": 15.0625, "learning_rate": 9.388172443240789e-06, "loss": 1.9181, "step": 1609 }, { "epoch": 0.18461185643848183, "grad_norm": 15.875, "learning_rate": 9.38728204658754e-06, "loss": 1.7068, "step": 1610 }, { "epoch": 0.18472652218782248, "grad_norm": 13.375, "learning_rate": 9.386391044792363e-06, "loss": 1.7377, "step": 1611 }, { "epoch": 0.18484118793716317, "grad_norm": 13.6875, "learning_rate": 9.385499437978154e-06, "loss": 1.4145, "step": 1612 }, { "epoch": 0.18495585368650383, "grad_norm": 15.3125, "learning_rate": 9.384607226267892e-06, "loss": 1.9931, "step": 1613 }, { "epoch": 0.18507051943584452, "grad_norm": 14.75, "learning_rate": 9.383714409784643e-06, "loss": 1.7283, "step": 1614 }, { "epoch": 0.18518518518518517, "grad_norm": 13.8125, "learning_rate": 9.382820988651551e-06, "loss": 1.6234, "step": 1615 }, { "epoch": 0.18529985093452586, "grad_norm": 13.0625, "learning_rate": 9.381926962991848e-06, "loss": 1.8294, "step": 1616 }, { "epoch": 0.18541451668386652, "grad_norm": 11.6875, "learning_rate": 9.381032332928847e-06, "loss": 1.9315, "step": 1617 }, { "epoch": 0.1855291824332072, "grad_norm": 10.9375, "learning_rate": 9.380137098585946e-06, "loss": 1.5392, "step": 1618 }, { "epoch": 0.18564384818254787, "grad_norm": 18.625, "learning_rate": 9.379241260086627e-06, "loss": 1.8871, "step": 1619 }, { "epoch": 0.18575851393188855, "grad_norm": 14.5, "learning_rate": 9.37834481755445e-06, "loss": 1.7592, "step": 1620 }, { "epoch": 0.1858731796812292, "grad_norm": 13.0, "learning_rate": 9.377447771113067e-06, "loss": 1.9026, "step": 1621 }, { "epoch": 0.1859878454305699, "grad_norm": 12.5625, "learning_rate": 9.376550120886204e-06, "loss": 1.8976, "step": 1622 }, { "epoch": 0.18610251117991056, "grad_norm": 12.6875, "learning_rate": 9.375651866997675e-06, "loss": 1.9149, "step": 1623 }, { "epoch": 0.18621717692925124, "grad_norm": 13.3125, "learning_rate": 9.37475300957138e-06, "loss": 1.9172, "step": 1624 }, { "epoch": 0.1863318426785919, "grad_norm": 16.75, "learning_rate": 9.373853548731298e-06, "loss": 1.8215, "step": 1625 }, { "epoch": 0.1864465084279326, "grad_norm": 15.875, "learning_rate": 9.372953484601491e-06, "loss": 1.668, "step": 1626 }, { "epoch": 0.18656117417727325, "grad_norm": 11.5, "learning_rate": 9.372052817306106e-06, "loss": 1.5593, "step": 1627 }, { "epoch": 0.1866758399266139, "grad_norm": 12.25, "learning_rate": 9.371151546969376e-06, "loss": 2.0091, "step": 1628 }, { "epoch": 0.1867905056759546, "grad_norm": 9.9375, "learning_rate": 9.370249673715611e-06, "loss": 1.3983, "step": 1629 }, { "epoch": 0.18690517142529525, "grad_norm": 13.6875, "learning_rate": 9.369347197669208e-06, "loss": 2.3009, "step": 1630 }, { "epoch": 0.18701983717463594, "grad_norm": 16.125, "learning_rate": 9.368444118954648e-06, "loss": 2.1255, "step": 1631 }, { "epoch": 0.1871345029239766, "grad_norm": 13.3125, "learning_rate": 9.36754043769649e-06, "loss": 1.7388, "step": 1632 }, { "epoch": 0.18724916867331728, "grad_norm": 11.75, "learning_rate": 9.366636154019382e-06, "loss": 1.6331, "step": 1633 }, { "epoch": 0.18736383442265794, "grad_norm": 14.75, "learning_rate": 9.365731268048052e-06, "loss": 1.6323, "step": 1634 }, { "epoch": 0.18747850017199863, "grad_norm": 13.9375, "learning_rate": 9.364825779907311e-06, "loss": 1.9708, "step": 1635 }, { "epoch": 0.1875931659213393, "grad_norm": 12.25, "learning_rate": 9.363919689722057e-06, "loss": 1.8234, "step": 1636 }, { "epoch": 0.18770783167067998, "grad_norm": 12.25, "learning_rate": 9.363012997617265e-06, "loss": 1.7187, "step": 1637 }, { "epoch": 0.18782249742002063, "grad_norm": 14.0625, "learning_rate": 9.362105703717995e-06, "loss": 1.6981, "step": 1638 }, { "epoch": 0.18793716316936132, "grad_norm": 12.0625, "learning_rate": 9.361197808149395e-06, "loss": 1.9141, "step": 1639 }, { "epoch": 0.18805182891870198, "grad_norm": 26.75, "learning_rate": 9.360289311036687e-06, "loss": 1.818, "step": 1640 }, { "epoch": 0.18816649466804267, "grad_norm": 10.4375, "learning_rate": 9.359380212505185e-06, "loss": 1.634, "step": 1641 }, { "epoch": 0.18828116041738333, "grad_norm": 11.0, "learning_rate": 9.358470512680278e-06, "loss": 1.6521, "step": 1642 }, { "epoch": 0.188395826166724, "grad_norm": 14.6875, "learning_rate": 9.357560211687446e-06, "loss": 1.8909, "step": 1643 }, { "epoch": 0.18851049191606467, "grad_norm": 12.9375, "learning_rate": 9.356649309652244e-06, "loss": 2.1679, "step": 1644 }, { "epoch": 0.18862515766540533, "grad_norm": 10.6875, "learning_rate": 9.355737806700315e-06, "loss": 1.4733, "step": 1645 }, { "epoch": 0.18873982341474602, "grad_norm": 10.6875, "learning_rate": 9.354825702957384e-06, "loss": 1.493, "step": 1646 }, { "epoch": 0.18885448916408668, "grad_norm": 11.9375, "learning_rate": 9.353912998549259e-06, "loss": 1.7441, "step": 1647 }, { "epoch": 0.18896915491342736, "grad_norm": 12.125, "learning_rate": 9.352999693601827e-06, "loss": 1.8155, "step": 1648 }, { "epoch": 0.18908382066276802, "grad_norm": 24.125, "learning_rate": 9.352085788241065e-06, "loss": 1.8329, "step": 1649 }, { "epoch": 0.1891984864121087, "grad_norm": 11.1875, "learning_rate": 9.351171282593026e-06, "loss": 1.5668, "step": 1650 }, { "epoch": 0.18931315216144937, "grad_norm": 14.0, "learning_rate": 9.350256176783847e-06, "loss": 1.7144, "step": 1651 }, { "epoch": 0.18942781791079005, "grad_norm": 14.625, "learning_rate": 9.349340470939754e-06, "loss": 1.5696, "step": 1652 }, { "epoch": 0.1895424836601307, "grad_norm": 25.0, "learning_rate": 9.34842416518705e-06, "loss": 2.2715, "step": 1653 }, { "epoch": 0.1896571494094714, "grad_norm": 11.9375, "learning_rate": 9.34750725965212e-06, "loss": 1.8344, "step": 1654 }, { "epoch": 0.18977181515881206, "grad_norm": 12.25, "learning_rate": 9.346589754461434e-06, "loss": 1.872, "step": 1655 }, { "epoch": 0.18988648090815274, "grad_norm": 28.875, "learning_rate": 9.345671649741545e-06, "loss": 1.8284, "step": 1656 }, { "epoch": 0.1900011466574934, "grad_norm": 16.75, "learning_rate": 9.34475294561909e-06, "loss": 1.8869, "step": 1657 }, { "epoch": 0.1901158124068341, "grad_norm": 11.25, "learning_rate": 9.343833642220781e-06, "loss": 1.7219, "step": 1658 }, { "epoch": 0.19023047815617475, "grad_norm": 15.8125, "learning_rate": 9.342913739673424e-06, "loss": 1.8491, "step": 1659 }, { "epoch": 0.19034514390551543, "grad_norm": 13.4375, "learning_rate": 9.3419932381039e-06, "loss": 1.2566, "step": 1660 }, { "epoch": 0.1904598096548561, "grad_norm": 11.625, "learning_rate": 9.341072137639176e-06, "loss": 1.9245, "step": 1661 }, { "epoch": 0.19057447540419675, "grad_norm": 15.4375, "learning_rate": 9.340150438406297e-06, "loss": 1.9443, "step": 1662 }, { "epoch": 0.19068914115353744, "grad_norm": 10.375, "learning_rate": 9.339228140532397e-06, "loss": 1.5717, "step": 1663 }, { "epoch": 0.1908038069028781, "grad_norm": 14.0625, "learning_rate": 9.338305244144688e-06, "loss": 1.7192, "step": 1664 }, { "epoch": 0.19091847265221878, "grad_norm": 18.125, "learning_rate": 9.337381749370465e-06, "loss": 1.8552, "step": 1665 }, { "epoch": 0.19103313840155944, "grad_norm": 13.9375, "learning_rate": 9.33645765633711e-06, "loss": 2.0335, "step": 1666 }, { "epoch": 0.19114780415090013, "grad_norm": 16.375, "learning_rate": 9.33553296517208e-06, "loss": 2.3414, "step": 1667 }, { "epoch": 0.1912624699002408, "grad_norm": 16.875, "learning_rate": 9.33460767600292e-06, "loss": 2.0635, "step": 1668 }, { "epoch": 0.19137713564958148, "grad_norm": 13.125, "learning_rate": 9.333681788957256e-06, "loss": 1.9081, "step": 1669 }, { "epoch": 0.19149180139892213, "grad_norm": 14.75, "learning_rate": 9.332755304162798e-06, "loss": 1.9666, "step": 1670 }, { "epoch": 0.19160646714826282, "grad_norm": 10.625, "learning_rate": 9.331828221747336e-06, "loss": 1.7443, "step": 1671 }, { "epoch": 0.19172113289760348, "grad_norm": 29.875, "learning_rate": 9.33090054183874e-06, "loss": 1.7287, "step": 1672 }, { "epoch": 0.19183579864694417, "grad_norm": 12.4375, "learning_rate": 9.329972264564973e-06, "loss": 1.6628, "step": 1673 }, { "epoch": 0.19195046439628483, "grad_norm": 12.375, "learning_rate": 9.329043390054067e-06, "loss": 1.7984, "step": 1674 }, { "epoch": 0.1920651301456255, "grad_norm": 12.25, "learning_rate": 9.328113918434144e-06, "loss": 1.737, "step": 1675 }, { "epoch": 0.19217979589496617, "grad_norm": 11.625, "learning_rate": 9.327183849833407e-06, "loss": 1.7096, "step": 1676 }, { "epoch": 0.19229446164430686, "grad_norm": 12.9375, "learning_rate": 9.326253184380142e-06, "loss": 1.6965, "step": 1677 }, { "epoch": 0.19240912739364752, "grad_norm": 15.0625, "learning_rate": 9.325321922202716e-06, "loss": 1.6441, "step": 1678 }, { "epoch": 0.1925237931429882, "grad_norm": 12.8125, "learning_rate": 9.324390063429579e-06, "loss": 1.8116, "step": 1679 }, { "epoch": 0.19263845889232886, "grad_norm": 11.6875, "learning_rate": 9.323457608189264e-06, "loss": 2.0485, "step": 1680 }, { "epoch": 0.19275312464166952, "grad_norm": 12.5625, "learning_rate": 9.322524556610384e-06, "loss": 1.3535, "step": 1681 }, { "epoch": 0.1928677903910102, "grad_norm": 13.9375, "learning_rate": 9.321590908821636e-06, "loss": 1.8717, "step": 1682 }, { "epoch": 0.19298245614035087, "grad_norm": 14.875, "learning_rate": 9.320656664951799e-06, "loss": 1.8886, "step": 1683 }, { "epoch": 0.19309712188969155, "grad_norm": 13.75, "learning_rate": 9.319721825129735e-06, "loss": 1.4408, "step": 1684 }, { "epoch": 0.1932117876390322, "grad_norm": 15.4375, "learning_rate": 9.318786389484385e-06, "loss": 1.9899, "step": 1685 }, { "epoch": 0.1933264533883729, "grad_norm": 12.6875, "learning_rate": 9.317850358144777e-06, "loss": 1.7088, "step": 1686 }, { "epoch": 0.19344111913771356, "grad_norm": 22.875, "learning_rate": 9.316913731240017e-06, "loss": 1.8172, "step": 1687 }, { "epoch": 0.19355578488705424, "grad_norm": 11.5625, "learning_rate": 9.315976508899298e-06, "loss": 1.6731, "step": 1688 }, { "epoch": 0.1936704506363949, "grad_norm": 30.75, "learning_rate": 9.315038691251888e-06, "loss": 2.0733, "step": 1689 }, { "epoch": 0.1937851163857356, "grad_norm": 12.125, "learning_rate": 9.314100278427142e-06, "loss": 1.395, "step": 1690 }, { "epoch": 0.19389978213507625, "grad_norm": 19.625, "learning_rate": 9.313161270554498e-06, "loss": 1.9448, "step": 1691 }, { "epoch": 0.19401444788441694, "grad_norm": 12.8125, "learning_rate": 9.312221667763472e-06, "loss": 1.8761, "step": 1692 }, { "epoch": 0.1941291136337576, "grad_norm": 13.75, "learning_rate": 9.311281470183667e-06, "loss": 1.4925, "step": 1693 }, { "epoch": 0.19424377938309828, "grad_norm": 11.9375, "learning_rate": 9.310340677944763e-06, "loss": 1.8547, "step": 1694 }, { "epoch": 0.19435844513243894, "grad_norm": 15.125, "learning_rate": 9.309399291176524e-06, "loss": 1.9958, "step": 1695 }, { "epoch": 0.19447311088177963, "grad_norm": 15.6875, "learning_rate": 9.308457310008798e-06, "loss": 2.0707, "step": 1696 }, { "epoch": 0.19458777663112028, "grad_norm": 14.9375, "learning_rate": 9.307514734571514e-06, "loss": 1.9909, "step": 1697 }, { "epoch": 0.19470244238046094, "grad_norm": 12.9375, "learning_rate": 9.30657156499468e-06, "loss": 1.7059, "step": 1698 }, { "epoch": 0.19481710812980163, "grad_norm": 22.625, "learning_rate": 9.30562780140839e-06, "loss": 2.0458, "step": 1699 }, { "epoch": 0.1949317738791423, "grad_norm": 15.625, "learning_rate": 9.304683443942817e-06, "loss": 1.8943, "step": 1700 }, { "epoch": 0.19504643962848298, "grad_norm": 12.25, "learning_rate": 9.303738492728218e-06, "loss": 1.4514, "step": 1701 }, { "epoch": 0.19516110537782363, "grad_norm": 13.875, "learning_rate": 9.30279294789493e-06, "loss": 1.5558, "step": 1702 }, { "epoch": 0.19527577112716432, "grad_norm": 13.3125, "learning_rate": 9.301846809573373e-06, "loss": 1.6707, "step": 1703 }, { "epoch": 0.19539043687650498, "grad_norm": 31.75, "learning_rate": 9.30090007789405e-06, "loss": 1.86, "step": 1704 }, { "epoch": 0.19550510262584567, "grad_norm": 13.0, "learning_rate": 9.299952752987544e-06, "loss": 1.8056, "step": 1705 }, { "epoch": 0.19561976837518633, "grad_norm": 16.375, "learning_rate": 9.29900483498452e-06, "loss": 2.1704, "step": 1706 }, { "epoch": 0.195734434124527, "grad_norm": 17.125, "learning_rate": 9.298056324015725e-06, "loss": 1.6937, "step": 1707 }, { "epoch": 0.19584909987386767, "grad_norm": 35.25, "learning_rate": 9.29710722021199e-06, "loss": 1.976, "step": 1708 }, { "epoch": 0.19596376562320836, "grad_norm": 14.8125, "learning_rate": 9.296157523704223e-06, "loss": 1.9796, "step": 1709 }, { "epoch": 0.19607843137254902, "grad_norm": 15.125, "learning_rate": 9.295207234623418e-06, "loss": 1.8121, "step": 1710 }, { "epoch": 0.1961930971218897, "grad_norm": 14.4375, "learning_rate": 9.294256353100651e-06, "loss": 1.8378, "step": 1711 }, { "epoch": 0.19630776287123036, "grad_norm": 14.625, "learning_rate": 9.293304879267073e-06, "loss": 1.576, "step": 1712 }, { "epoch": 0.19642242862057105, "grad_norm": 11.5, "learning_rate": 9.292352813253927e-06, "loss": 1.7889, "step": 1713 }, { "epoch": 0.1965370943699117, "grad_norm": 15.0, "learning_rate": 9.29140015519253e-06, "loss": 1.8336, "step": 1714 }, { "epoch": 0.19665176011925237, "grad_norm": 11.875, "learning_rate": 9.290446905214281e-06, "loss": 1.5043, "step": 1715 }, { "epoch": 0.19676642586859305, "grad_norm": 16.75, "learning_rate": 9.289493063450666e-06, "loss": 1.8319, "step": 1716 }, { "epoch": 0.1968810916179337, "grad_norm": 15.0, "learning_rate": 9.288538630033248e-06, "loss": 1.6609, "step": 1717 }, { "epoch": 0.1969957573672744, "grad_norm": 12.625, "learning_rate": 9.287583605093674e-06, "loss": 1.4047, "step": 1718 }, { "epoch": 0.19711042311661506, "grad_norm": 18.875, "learning_rate": 9.28662798876367e-06, "loss": 1.9485, "step": 1719 }, { "epoch": 0.19722508886595574, "grad_norm": 16.875, "learning_rate": 9.285671781175045e-06, "loss": 1.8972, "step": 1720 }, { "epoch": 0.1973397546152964, "grad_norm": 11.5625, "learning_rate": 9.284714982459691e-06, "loss": 1.8678, "step": 1721 }, { "epoch": 0.1974544203646371, "grad_norm": 13.375, "learning_rate": 9.283757592749579e-06, "loss": 1.6971, "step": 1722 }, { "epoch": 0.19756908611397775, "grad_norm": 12.6875, "learning_rate": 9.282799612176762e-06, "loss": 1.8762, "step": 1723 }, { "epoch": 0.19768375186331844, "grad_norm": 16.0, "learning_rate": 9.281841040873377e-06, "loss": 1.6994, "step": 1724 }, { "epoch": 0.1977984176126591, "grad_norm": 17.5, "learning_rate": 9.280881878971638e-06, "loss": 1.8863, "step": 1725 }, { "epoch": 0.19791308336199978, "grad_norm": 10.5625, "learning_rate": 9.279922126603847e-06, "loss": 1.6297, "step": 1726 }, { "epoch": 0.19802774911134044, "grad_norm": 15.625, "learning_rate": 9.278961783902381e-06, "loss": 1.9244, "step": 1727 }, { "epoch": 0.19814241486068113, "grad_norm": 11.9375, "learning_rate": 9.2780008509997e-06, "loss": 1.5091, "step": 1728 }, { "epoch": 0.19825708061002179, "grad_norm": 14.1875, "learning_rate": 9.277039328028348e-06, "loss": 1.8596, "step": 1729 }, { "epoch": 0.19837174635936247, "grad_norm": 11.25, "learning_rate": 9.276077215120949e-06, "loss": 1.8486, "step": 1730 }, { "epoch": 0.19848641210870313, "grad_norm": 12.0625, "learning_rate": 9.275114512410208e-06, "loss": 1.6038, "step": 1731 }, { "epoch": 0.1986010778580438, "grad_norm": 10.5, "learning_rate": 9.274151220028912e-06, "loss": 1.5226, "step": 1732 }, { "epoch": 0.19871574360738448, "grad_norm": 12.125, "learning_rate": 9.273187338109925e-06, "loss": 1.7227, "step": 1733 }, { "epoch": 0.19883040935672514, "grad_norm": 11.75, "learning_rate": 9.272222866786202e-06, "loss": 1.6355, "step": 1734 }, { "epoch": 0.19894507510606582, "grad_norm": 11.5, "learning_rate": 9.271257806190768e-06, "loss": 1.4467, "step": 1735 }, { "epoch": 0.19905974085540648, "grad_norm": 18.875, "learning_rate": 9.27029215645674e-06, "loss": 1.9458, "step": 1736 }, { "epoch": 0.19917440660474717, "grad_norm": 19.5, "learning_rate": 9.269325917717308e-06, "loss": 1.9347, "step": 1737 }, { "epoch": 0.19928907235408783, "grad_norm": 12.0625, "learning_rate": 9.268359090105742e-06, "loss": 1.7182, "step": 1738 }, { "epoch": 0.1994037381034285, "grad_norm": 13.5, "learning_rate": 9.267391673755406e-06, "loss": 1.9269, "step": 1739 }, { "epoch": 0.19951840385276917, "grad_norm": 13.8125, "learning_rate": 9.266423668799732e-06, "loss": 2.2742, "step": 1740 }, { "epoch": 0.19963306960210986, "grad_norm": 17.0, "learning_rate": 9.265455075372237e-06, "loss": 1.6632, "step": 1741 }, { "epoch": 0.19974773535145052, "grad_norm": 25.375, "learning_rate": 9.264485893606523e-06, "loss": 1.9357, "step": 1742 }, { "epoch": 0.1998624011007912, "grad_norm": 13.3125, "learning_rate": 9.263516123636267e-06, "loss": 1.9552, "step": 1743 }, { "epoch": 0.19997706685013186, "grad_norm": 17.125, "learning_rate": 9.262545765595232e-06, "loss": 1.7969, "step": 1744 }, { "epoch": 0.20009173259947255, "grad_norm": 15.5, "learning_rate": 9.26157481961726e-06, "loss": 1.8245, "step": 1745 }, { "epoch": 0.2002063983488132, "grad_norm": 11.25, "learning_rate": 9.260603285836276e-06, "loss": 1.5395, "step": 1746 }, { "epoch": 0.2003210640981539, "grad_norm": 12.25, "learning_rate": 9.259631164386283e-06, "loss": 1.7079, "step": 1747 }, { "epoch": 0.20043572984749455, "grad_norm": 11.875, "learning_rate": 9.258658455401365e-06, "loss": 1.7159, "step": 1748 }, { "epoch": 0.2005503955968352, "grad_norm": 10.8125, "learning_rate": 9.257685159015693e-06, "loss": 1.5904, "step": 1749 }, { "epoch": 0.2006650613461759, "grad_norm": 13.4375, "learning_rate": 9.25671127536351e-06, "loss": 1.6221, "step": 1750 }, { "epoch": 0.20077972709551656, "grad_norm": 38.75, "learning_rate": 9.255736804579148e-06, "loss": 2.1308, "step": 1751 }, { "epoch": 0.20089439284485724, "grad_norm": 27.75, "learning_rate": 9.254761746797015e-06, "loss": 2.1034, "step": 1752 }, { "epoch": 0.2010090585941979, "grad_norm": 13.625, "learning_rate": 9.253786102151604e-06, "loss": 1.9055, "step": 1753 }, { "epoch": 0.2011237243435386, "grad_norm": 15.4375, "learning_rate": 9.252809870777482e-06, "loss": 1.5009, "step": 1754 }, { "epoch": 0.20123839009287925, "grad_norm": 15.875, "learning_rate": 9.251833052809305e-06, "loss": 1.8554, "step": 1755 }, { "epoch": 0.20135305584221994, "grad_norm": 14.5625, "learning_rate": 9.250855648381805e-06, "loss": 1.6307, "step": 1756 }, { "epoch": 0.2014677215915606, "grad_norm": 13.9375, "learning_rate": 9.249877657629799e-06, "loss": 1.728, "step": 1757 }, { "epoch": 0.20158238734090128, "grad_norm": 12.3125, "learning_rate": 9.248899080688178e-06, "loss": 1.738, "step": 1758 }, { "epoch": 0.20169705309024194, "grad_norm": 14.3125, "learning_rate": 9.247919917691923e-06, "loss": 1.6147, "step": 1759 }, { "epoch": 0.20181171883958263, "grad_norm": 16.75, "learning_rate": 9.246940168776086e-06, "loss": 2.0024, "step": 1760 }, { "epoch": 0.20192638458892329, "grad_norm": 56.0, "learning_rate": 9.245959834075807e-06, "loss": 2.3987, "step": 1761 }, { "epoch": 0.20204105033826397, "grad_norm": 14.1875, "learning_rate": 9.244978913726304e-06, "loss": 1.6905, "step": 1762 }, { "epoch": 0.20215571608760463, "grad_norm": 13.0625, "learning_rate": 9.243997407862878e-06, "loss": 1.4694, "step": 1763 }, { "epoch": 0.20227038183694532, "grad_norm": 14.6875, "learning_rate": 9.243015316620908e-06, "loss": 1.5886, "step": 1764 }, { "epoch": 0.20238504758628598, "grad_norm": 33.0, "learning_rate": 9.242032640135851e-06, "loss": 1.9508, "step": 1765 }, { "epoch": 0.20249971333562664, "grad_norm": 13.125, "learning_rate": 9.241049378543255e-06, "loss": 1.946, "step": 1766 }, { "epoch": 0.20261437908496732, "grad_norm": 13.125, "learning_rate": 9.240065531978737e-06, "loss": 2.0751, "step": 1767 }, { "epoch": 0.20272904483430798, "grad_norm": 11.0625, "learning_rate": 9.239081100578002e-06, "loss": 1.7466, "step": 1768 }, { "epoch": 0.20284371058364867, "grad_norm": 12.9375, "learning_rate": 9.238096084476833e-06, "loss": 1.8838, "step": 1769 }, { "epoch": 0.20295837633298933, "grad_norm": 13.375, "learning_rate": 9.237110483811095e-06, "loss": 2.0641, "step": 1770 }, { "epoch": 0.20307304208233, "grad_norm": 11.9375, "learning_rate": 9.236124298716735e-06, "loss": 2.0152, "step": 1771 }, { "epoch": 0.20318770783167067, "grad_norm": 13.1875, "learning_rate": 9.235137529329773e-06, "loss": 1.8721, "step": 1772 }, { "epoch": 0.20330237358101136, "grad_norm": 13.75, "learning_rate": 9.234150175786318e-06, "loss": 1.7678, "step": 1773 }, { "epoch": 0.20341703933035202, "grad_norm": 12.375, "learning_rate": 9.233162238222557e-06, "loss": 1.7053, "step": 1774 }, { "epoch": 0.2035317050796927, "grad_norm": 23.25, "learning_rate": 9.232173716774758e-06, "loss": 1.6657, "step": 1775 }, { "epoch": 0.20364637082903336, "grad_norm": 13.0625, "learning_rate": 9.231184611579266e-06, "loss": 1.8458, "step": 1776 }, { "epoch": 0.20376103657837405, "grad_norm": 11.6875, "learning_rate": 9.230194922772511e-06, "loss": 1.5023, "step": 1777 }, { "epoch": 0.2038757023277147, "grad_norm": 16.875, "learning_rate": 9.229204650491002e-06, "loss": 1.4148, "step": 1778 }, { "epoch": 0.2039903680770554, "grad_norm": 13.0625, "learning_rate": 9.228213794871327e-06, "loss": 1.9854, "step": 1779 }, { "epoch": 0.20410503382639605, "grad_norm": 14.375, "learning_rate": 9.227222356050154e-06, "loss": 1.7865, "step": 1780 }, { "epoch": 0.20421969957573674, "grad_norm": 16.125, "learning_rate": 9.226230334164237e-06, "loss": 2.2086, "step": 1781 }, { "epoch": 0.2043343653250774, "grad_norm": 13.6875, "learning_rate": 9.225237729350403e-06, "loss": 1.4502, "step": 1782 }, { "epoch": 0.20444903107441806, "grad_norm": 12.0625, "learning_rate": 9.224244541745565e-06, "loss": 1.8208, "step": 1783 }, { "epoch": 0.20456369682375874, "grad_norm": 12.375, "learning_rate": 9.223250771486718e-06, "loss": 2.0797, "step": 1784 }, { "epoch": 0.2046783625730994, "grad_norm": 13.125, "learning_rate": 9.222256418710925e-06, "loss": 1.6659, "step": 1785 }, { "epoch": 0.2047930283224401, "grad_norm": 14.125, "learning_rate": 9.221261483555344e-06, "loss": 1.7962, "step": 1786 }, { "epoch": 0.20490769407178075, "grad_norm": 16.125, "learning_rate": 9.220265966157206e-06, "loss": 2.2354, "step": 1787 }, { "epoch": 0.20502235982112144, "grad_norm": 14.375, "learning_rate": 9.219269866653824e-06, "loss": 1.9232, "step": 1788 }, { "epoch": 0.2051370255704621, "grad_norm": 12.4375, "learning_rate": 9.218273185182589e-06, "loss": 1.8849, "step": 1789 }, { "epoch": 0.20525169131980278, "grad_norm": 15.375, "learning_rate": 9.217275921880977e-06, "loss": 1.8804, "step": 1790 }, { "epoch": 0.20536635706914344, "grad_norm": 12.25, "learning_rate": 9.21627807688654e-06, "loss": 2.0685, "step": 1791 }, { "epoch": 0.20548102281848413, "grad_norm": 16.0, "learning_rate": 9.215279650336912e-06, "loss": 2.0307, "step": 1792 }, { "epoch": 0.20559568856782479, "grad_norm": 9.8125, "learning_rate": 9.214280642369807e-06, "loss": 1.4873, "step": 1793 }, { "epoch": 0.20571035431716547, "grad_norm": 16.5, "learning_rate": 9.213281053123018e-06, "loss": 1.8901, "step": 1794 }, { "epoch": 0.20582502006650613, "grad_norm": 13.625, "learning_rate": 9.21228088273442e-06, "loss": 2.2112, "step": 1795 }, { "epoch": 0.20593968581584682, "grad_norm": 14.1875, "learning_rate": 9.211280131341969e-06, "loss": 1.6971, "step": 1796 }, { "epoch": 0.20605435156518748, "grad_norm": 13.875, "learning_rate": 9.210278799083697e-06, "loss": 1.7841, "step": 1797 }, { "epoch": 0.20616901731452816, "grad_norm": 13.1875, "learning_rate": 9.20927688609772e-06, "loss": 1.83, "step": 1798 }, { "epoch": 0.20628368306386882, "grad_norm": 12.0, "learning_rate": 9.208274392522232e-06, "loss": 1.6407, "step": 1799 }, { "epoch": 0.20639834881320948, "grad_norm": 14.375, "learning_rate": 9.20727131849551e-06, "loss": 1.9705, "step": 1800 }, { "epoch": 0.20651301456255017, "grad_norm": 15.6875, "learning_rate": 9.206267664155906e-06, "loss": 1.7096, "step": 1801 }, { "epoch": 0.20662768031189083, "grad_norm": 18.375, "learning_rate": 9.205263429641857e-06, "loss": 1.8617, "step": 1802 }, { "epoch": 0.2067423460612315, "grad_norm": 12.1875, "learning_rate": 9.204258615091878e-06, "loss": 1.6905, "step": 1803 }, { "epoch": 0.20685701181057217, "grad_norm": 14.875, "learning_rate": 9.203253220644565e-06, "loss": 1.8659, "step": 1804 }, { "epoch": 0.20697167755991286, "grad_norm": 15.1875, "learning_rate": 9.20224724643859e-06, "loss": 1.6629, "step": 1805 }, { "epoch": 0.20708634330925352, "grad_norm": 13.3125, "learning_rate": 9.201240692612709e-06, "loss": 1.7595, "step": 1806 }, { "epoch": 0.2072010090585942, "grad_norm": 14.1875, "learning_rate": 9.200233559305758e-06, "loss": 1.6871, "step": 1807 }, { "epoch": 0.20731567480793486, "grad_norm": 13.875, "learning_rate": 9.19922584665665e-06, "loss": 1.7201, "step": 1808 }, { "epoch": 0.20743034055727555, "grad_norm": 14.5, "learning_rate": 9.198217554804383e-06, "loss": 1.9388, "step": 1809 }, { "epoch": 0.2075450063066162, "grad_norm": 13.75, "learning_rate": 9.197208683888028e-06, "loss": 1.7363, "step": 1810 }, { "epoch": 0.2076596720559569, "grad_norm": 17.375, "learning_rate": 9.196199234046742e-06, "loss": 2.3725, "step": 1811 }, { "epoch": 0.20777433780529755, "grad_norm": 16.625, "learning_rate": 9.195189205419758e-06, "loss": 1.7422, "step": 1812 }, { "epoch": 0.20788900355463824, "grad_norm": 14.0625, "learning_rate": 9.19417859814639e-06, "loss": 1.8012, "step": 1813 }, { "epoch": 0.2080036693039789, "grad_norm": 11.8125, "learning_rate": 9.193167412366035e-06, "loss": 1.6233, "step": 1814 }, { "epoch": 0.2081183350533196, "grad_norm": 13.9375, "learning_rate": 9.192155648218163e-06, "loss": 2.1434, "step": 1815 }, { "epoch": 0.20823300080266025, "grad_norm": 12.3125, "learning_rate": 9.19114330584233e-06, "loss": 1.4268, "step": 1816 }, { "epoch": 0.2083476665520009, "grad_norm": 13.0, "learning_rate": 9.190130385378167e-06, "loss": 1.5077, "step": 1817 }, { "epoch": 0.2084623323013416, "grad_norm": 12.625, "learning_rate": 9.18911688696539e-06, "loss": 1.8516, "step": 1818 }, { "epoch": 0.20857699805068225, "grad_norm": 11.8125, "learning_rate": 9.188102810743789e-06, "loss": 1.8396, "step": 1819 }, { "epoch": 0.20869166380002294, "grad_norm": 19.25, "learning_rate": 9.187088156853237e-06, "loss": 1.8139, "step": 1820 }, { "epoch": 0.2088063295493636, "grad_norm": 15.8125, "learning_rate": 9.186072925433688e-06, "loss": 2.2504, "step": 1821 }, { "epoch": 0.20892099529870428, "grad_norm": 14.0625, "learning_rate": 9.185057116625173e-06, "loss": 1.5051, "step": 1822 }, { "epoch": 0.20903566104804494, "grad_norm": 13.5, "learning_rate": 9.184040730567804e-06, "loss": 2.0116, "step": 1823 }, { "epoch": 0.20915032679738563, "grad_norm": 12.4375, "learning_rate": 9.18302376740177e-06, "loss": 1.6383, "step": 1824 }, { "epoch": 0.2092649925467263, "grad_norm": 14.4375, "learning_rate": 9.182006227267343e-06, "loss": 1.8447, "step": 1825 }, { "epoch": 0.20937965829606697, "grad_norm": 13.625, "learning_rate": 9.180988110304874e-06, "loss": 1.5221, "step": 1826 }, { "epoch": 0.20949432404540763, "grad_norm": 17.125, "learning_rate": 9.17996941665479e-06, "loss": 2.1402, "step": 1827 }, { "epoch": 0.20960898979474832, "grad_norm": 14.1875, "learning_rate": 9.178950146457605e-06, "loss": 1.8489, "step": 1828 }, { "epoch": 0.20972365554408898, "grad_norm": 29.375, "learning_rate": 9.177930299853905e-06, "loss": 1.7762, "step": 1829 }, { "epoch": 0.20983832129342966, "grad_norm": 17.0, "learning_rate": 9.176909876984357e-06, "loss": 1.789, "step": 1830 }, { "epoch": 0.20995298704277032, "grad_norm": 21.875, "learning_rate": 9.175888877989713e-06, "loss": 1.7464, "step": 1831 }, { "epoch": 0.210067652792111, "grad_norm": 15.75, "learning_rate": 9.174867303010796e-06, "loss": 1.8448, "step": 1832 }, { "epoch": 0.21018231854145167, "grad_norm": 11.125, "learning_rate": 9.173845152188517e-06, "loss": 1.723, "step": 1833 }, { "epoch": 0.21029698429079233, "grad_norm": 14.5625, "learning_rate": 9.172822425663856e-06, "loss": 2.0908, "step": 1834 }, { "epoch": 0.210411650040133, "grad_norm": 13.0625, "learning_rate": 9.171799123577885e-06, "loss": 1.7026, "step": 1835 }, { "epoch": 0.21052631578947367, "grad_norm": 13.6875, "learning_rate": 9.170775246071748e-06, "loss": 1.6636, "step": 1836 }, { "epoch": 0.21064098153881436, "grad_norm": 16.375, "learning_rate": 9.169750793286667e-06, "loss": 1.9389, "step": 1837 }, { "epoch": 0.21075564728815502, "grad_norm": 14.75, "learning_rate": 9.168725765363946e-06, "loss": 2.0194, "step": 1838 }, { "epoch": 0.2108703130374957, "grad_norm": 14.1875, "learning_rate": 9.167700162444968e-06, "loss": 1.545, "step": 1839 }, { "epoch": 0.21098497878683636, "grad_norm": 11.75, "learning_rate": 9.166673984671198e-06, "loss": 1.7961, "step": 1840 }, { "epoch": 0.21109964453617705, "grad_norm": 13.8125, "learning_rate": 9.165647232184176e-06, "loss": 2.0905, "step": 1841 }, { "epoch": 0.2112143102855177, "grad_norm": 13.5, "learning_rate": 9.164619905125522e-06, "loss": 1.5432, "step": 1842 }, { "epoch": 0.2113289760348584, "grad_norm": 16.5, "learning_rate": 9.163592003636937e-06, "loss": 1.4894, "step": 1843 }, { "epoch": 0.21144364178419905, "grad_norm": 13.0625, "learning_rate": 9.1625635278602e-06, "loss": 2.2027, "step": 1844 }, { "epoch": 0.21155830753353974, "grad_norm": 17.125, "learning_rate": 9.16153447793717e-06, "loss": 2.2982, "step": 1845 }, { "epoch": 0.2116729732828804, "grad_norm": 14.5, "learning_rate": 9.160504854009786e-06, "loss": 1.8932, "step": 1846 }, { "epoch": 0.2117876390322211, "grad_norm": 12.625, "learning_rate": 9.159474656220063e-06, "loss": 1.7852, "step": 1847 }, { "epoch": 0.21190230478156175, "grad_norm": 14.5, "learning_rate": 9.158443884710097e-06, "loss": 1.5993, "step": 1848 }, { "epoch": 0.21201697053090243, "grad_norm": 12.8125, "learning_rate": 9.157412539622066e-06, "loss": 1.7198, "step": 1849 }, { "epoch": 0.2121316362802431, "grad_norm": 13.0, "learning_rate": 9.156380621098223e-06, "loss": 1.7713, "step": 1850 }, { "epoch": 0.21224630202958375, "grad_norm": 11.75, "learning_rate": 9.155348129280899e-06, "loss": 1.5294, "step": 1851 }, { "epoch": 0.21236096777892444, "grad_norm": 14.25, "learning_rate": 9.15431506431251e-06, "loss": 1.6469, "step": 1852 }, { "epoch": 0.2124756335282651, "grad_norm": 12.875, "learning_rate": 9.153281426335547e-06, "loss": 1.7832, "step": 1853 }, { "epoch": 0.21259029927760578, "grad_norm": 12.5625, "learning_rate": 9.152247215492579e-06, "loss": 2.0409, "step": 1854 }, { "epoch": 0.21270496502694644, "grad_norm": 13.5625, "learning_rate": 9.151212431926256e-06, "loss": 1.5529, "step": 1855 }, { "epoch": 0.21281963077628713, "grad_norm": 16.75, "learning_rate": 9.15017707577931e-06, "loss": 1.8059, "step": 1856 }, { "epoch": 0.2129342965256278, "grad_norm": 13.25, "learning_rate": 9.149141147194542e-06, "loss": 2.1329, "step": 1857 }, { "epoch": 0.21304896227496847, "grad_norm": 36.25, "learning_rate": 9.148104646314845e-06, "loss": 1.9621, "step": 1858 }, { "epoch": 0.21316362802430913, "grad_norm": 13.0, "learning_rate": 9.147067573283182e-06, "loss": 1.2965, "step": 1859 }, { "epoch": 0.21327829377364982, "grad_norm": 17.75, "learning_rate": 9.146029928242596e-06, "loss": 1.863, "step": 1860 }, { "epoch": 0.21339295952299048, "grad_norm": 11.6875, "learning_rate": 9.144991711336216e-06, "loss": 1.834, "step": 1861 }, { "epoch": 0.21350762527233116, "grad_norm": 14.375, "learning_rate": 9.143952922707236e-06, "loss": 1.596, "step": 1862 }, { "epoch": 0.21362229102167182, "grad_norm": 10.125, "learning_rate": 9.142913562498943e-06, "loss": 1.7416, "step": 1863 }, { "epoch": 0.2137369567710125, "grad_norm": 10.3125, "learning_rate": 9.141873630854693e-06, "loss": 1.4771, "step": 1864 }, { "epoch": 0.21385162252035317, "grad_norm": 13.6875, "learning_rate": 9.140833127917929e-06, "loss": 1.7682, "step": 1865 }, { "epoch": 0.21396628826969386, "grad_norm": 15.5625, "learning_rate": 9.139792053832166e-06, "loss": 1.7122, "step": 1866 }, { "epoch": 0.21408095401903451, "grad_norm": 13.1875, "learning_rate": 9.138750408741002e-06, "loss": 1.6755, "step": 1867 }, { "epoch": 0.21419561976837517, "grad_norm": 13.375, "learning_rate": 9.13770819278811e-06, "loss": 1.7776, "step": 1868 }, { "epoch": 0.21431028551771586, "grad_norm": 21.25, "learning_rate": 9.136665406117244e-06, "loss": 1.4203, "step": 1869 }, { "epoch": 0.21442495126705652, "grad_norm": 10.6875, "learning_rate": 9.135622048872239e-06, "loss": 1.7134, "step": 1870 }, { "epoch": 0.2145396170163972, "grad_norm": 13.125, "learning_rate": 9.134578121197002e-06, "loss": 1.9466, "step": 1871 }, { "epoch": 0.21465428276573786, "grad_norm": 9.75, "learning_rate": 9.133533623235526e-06, "loss": 1.768, "step": 1872 }, { "epoch": 0.21476894851507855, "grad_norm": 12.0, "learning_rate": 9.132488555131882e-06, "loss": 1.5997, "step": 1873 }, { "epoch": 0.2148836142644192, "grad_norm": 16.0, "learning_rate": 9.13144291703021e-06, "loss": 1.6249, "step": 1874 }, { "epoch": 0.2149982800137599, "grad_norm": 11.125, "learning_rate": 9.130396709074742e-06, "loss": 1.7923, "step": 1875 }, { "epoch": 0.21511294576310055, "grad_norm": 14.9375, "learning_rate": 9.129349931409782e-06, "loss": 1.6828, "step": 1876 }, { "epoch": 0.21522761151244124, "grad_norm": 17.625, "learning_rate": 9.128302584179709e-06, "loss": 2.2248, "step": 1877 }, { "epoch": 0.2153422772617819, "grad_norm": 16.0, "learning_rate": 9.127254667528989e-06, "loss": 1.8546, "step": 1878 }, { "epoch": 0.2154569430111226, "grad_norm": 13.0625, "learning_rate": 9.126206181602159e-06, "loss": 1.7717, "step": 1879 }, { "epoch": 0.21557160876046325, "grad_norm": 14.625, "learning_rate": 9.125157126543838e-06, "loss": 1.9963, "step": 1880 }, { "epoch": 0.21568627450980393, "grad_norm": 11.9375, "learning_rate": 9.124107502498725e-06, "loss": 1.6508, "step": 1881 }, { "epoch": 0.2158009402591446, "grad_norm": 16.5, "learning_rate": 9.123057309611596e-06, "loss": 1.4135, "step": 1882 }, { "epoch": 0.21591560600848528, "grad_norm": 13.375, "learning_rate": 9.122006548027303e-06, "loss": 1.5731, "step": 1883 }, { "epoch": 0.21603027175782594, "grad_norm": 16.375, "learning_rate": 9.120955217890779e-06, "loss": 1.9927, "step": 1884 }, { "epoch": 0.2161449375071666, "grad_norm": 16.75, "learning_rate": 9.119903319347034e-06, "loss": 1.7697, "step": 1885 }, { "epoch": 0.21625960325650728, "grad_norm": 16.25, "learning_rate": 9.118850852541162e-06, "loss": 2.0828, "step": 1886 }, { "epoch": 0.21637426900584794, "grad_norm": 13.3125, "learning_rate": 9.117797817618323e-06, "loss": 1.5217, "step": 1887 }, { "epoch": 0.21648893475518863, "grad_norm": 14.8125, "learning_rate": 9.116744214723771e-06, "loss": 2.3751, "step": 1888 }, { "epoch": 0.2166036005045293, "grad_norm": 14.1875, "learning_rate": 9.115690044002825e-06, "loss": 1.9584, "step": 1889 }, { "epoch": 0.21671826625386997, "grad_norm": 21.625, "learning_rate": 9.11463530560089e-06, "loss": 2.0269, "step": 1890 }, { "epoch": 0.21683293200321063, "grad_norm": 13.0625, "learning_rate": 9.113579999663447e-06, "loss": 1.8271, "step": 1891 }, { "epoch": 0.21694759775255132, "grad_norm": 21.125, "learning_rate": 9.112524126336055e-06, "loss": 1.6491, "step": 1892 }, { "epoch": 0.21706226350189198, "grad_norm": 11.3125, "learning_rate": 9.111467685764351e-06, "loss": 1.7463, "step": 1893 }, { "epoch": 0.21717692925123266, "grad_norm": 16.5, "learning_rate": 9.110410678094051e-06, "loss": 1.9464, "step": 1894 }, { "epoch": 0.21729159500057332, "grad_norm": 12.625, "learning_rate": 9.109353103470952e-06, "loss": 1.7503, "step": 1895 }, { "epoch": 0.217406260749914, "grad_norm": 16.875, "learning_rate": 9.108294962040922e-06, "loss": 2.1328, "step": 1896 }, { "epoch": 0.21752092649925467, "grad_norm": 14.875, "learning_rate": 9.107236253949912e-06, "loss": 1.6692, "step": 1897 }, { "epoch": 0.21763559224859536, "grad_norm": 13.9375, "learning_rate": 9.106176979343954e-06, "loss": 2.2426, "step": 1898 }, { "epoch": 0.21775025799793601, "grad_norm": 15.1875, "learning_rate": 9.105117138369151e-06, "loss": 1.5414, "step": 1899 }, { "epoch": 0.2178649237472767, "grad_norm": 13.3125, "learning_rate": 9.10405673117169e-06, "loss": 1.8663, "step": 1900 }, { "epoch": 0.21797958949661736, "grad_norm": 15.125, "learning_rate": 9.102995757897834e-06, "loss": 2.1302, "step": 1901 }, { "epoch": 0.21809425524595802, "grad_norm": 16.5, "learning_rate": 9.101934218693924e-06, "loss": 2.1158, "step": 1902 }, { "epoch": 0.2182089209952987, "grad_norm": 10.5, "learning_rate": 9.100872113706377e-06, "loss": 1.5176, "step": 1903 }, { "epoch": 0.21832358674463936, "grad_norm": 13.625, "learning_rate": 9.099809443081691e-06, "loss": 1.5355, "step": 1904 }, { "epoch": 0.21843825249398005, "grad_norm": 15.1875, "learning_rate": 9.098746206966443e-06, "loss": 2.2372, "step": 1905 }, { "epoch": 0.2185529182433207, "grad_norm": 13.0625, "learning_rate": 9.097682405507286e-06, "loss": 1.8242, "step": 1906 }, { "epoch": 0.2186675839926614, "grad_norm": 11.8125, "learning_rate": 9.096618038850948e-06, "loss": 1.9084, "step": 1907 }, { "epoch": 0.21878224974200206, "grad_norm": 17.75, "learning_rate": 9.095553107144241e-06, "loss": 1.6233, "step": 1908 }, { "epoch": 0.21889691549134274, "grad_norm": 16.25, "learning_rate": 9.094487610534052e-06, "loss": 1.8448, "step": 1909 }, { "epoch": 0.2190115812406834, "grad_norm": 12.0625, "learning_rate": 9.093421549167344e-06, "loss": 1.6829, "step": 1910 }, { "epoch": 0.2191262469900241, "grad_norm": 31.5, "learning_rate": 9.092354923191162e-06, "loss": 1.5719, "step": 1911 }, { "epoch": 0.21924091273936475, "grad_norm": 25.625, "learning_rate": 9.091287732752626e-06, "loss": 1.7354, "step": 1912 }, { "epoch": 0.21935557848870543, "grad_norm": 11.625, "learning_rate": 9.090219977998934e-06, "loss": 1.7523, "step": 1913 }, { "epoch": 0.2194702442380461, "grad_norm": 15.125, "learning_rate": 9.08915165907736e-06, "loss": 1.7186, "step": 1914 }, { "epoch": 0.21958490998738678, "grad_norm": 13.5, "learning_rate": 9.088082776135264e-06, "loss": 1.9243, "step": 1915 }, { "epoch": 0.21969957573672744, "grad_norm": 18.0, "learning_rate": 9.087013329320074e-06, "loss": 1.7324, "step": 1916 }, { "epoch": 0.21981424148606812, "grad_norm": 11.0625, "learning_rate": 9.085943318779301e-06, "loss": 1.5442, "step": 1917 }, { "epoch": 0.21992890723540878, "grad_norm": 15.0625, "learning_rate": 9.084872744660532e-06, "loss": 2.2153, "step": 1918 }, { "epoch": 0.22004357298474944, "grad_norm": 14.6875, "learning_rate": 9.083801607111433e-06, "loss": 1.7925, "step": 1919 }, { "epoch": 0.22015823873409013, "grad_norm": 13.3125, "learning_rate": 9.082729906279747e-06, "loss": 1.9554, "step": 1920 }, { "epoch": 0.2202729044834308, "grad_norm": 12.5, "learning_rate": 9.081657642313292e-06, "loss": 1.5908, "step": 1921 }, { "epoch": 0.22038757023277147, "grad_norm": 12.625, "learning_rate": 9.080584815359972e-06, "loss": 1.9894, "step": 1922 }, { "epoch": 0.22050223598211213, "grad_norm": 15.3125, "learning_rate": 9.079511425567759e-06, "loss": 1.6047, "step": 1923 }, { "epoch": 0.22061690173145282, "grad_norm": 19.75, "learning_rate": 9.078437473084707e-06, "loss": 2.3174, "step": 1924 }, { "epoch": 0.22073156748079348, "grad_norm": 12.125, "learning_rate": 9.077362958058946e-06, "loss": 1.7078, "step": 1925 }, { "epoch": 0.22084623323013416, "grad_norm": 12.75, "learning_rate": 9.076287880638689e-06, "loss": 2.0806, "step": 1926 }, { "epoch": 0.22096089897947482, "grad_norm": 14.0625, "learning_rate": 9.075212240972219e-06, "loss": 1.305, "step": 1927 }, { "epoch": 0.2210755647288155, "grad_norm": 17.375, "learning_rate": 9.074136039207902e-06, "loss": 1.9779, "step": 1928 }, { "epoch": 0.22119023047815617, "grad_norm": 14.375, "learning_rate": 9.073059275494175e-06, "loss": 1.5954, "step": 1929 }, { "epoch": 0.22130489622749686, "grad_norm": 14.0625, "learning_rate": 9.071981949979565e-06, "loss": 1.7899, "step": 1930 }, { "epoch": 0.22141956197683751, "grad_norm": 13.75, "learning_rate": 9.07090406281266e-06, "loss": 1.6768, "step": 1931 }, { "epoch": 0.2215342277261782, "grad_norm": 11.0, "learning_rate": 9.06982561414214e-06, "loss": 1.5808, "step": 1932 }, { "epoch": 0.22164889347551886, "grad_norm": 13.3125, "learning_rate": 9.068746604116755e-06, "loss": 2.0701, "step": 1933 }, { "epoch": 0.22176355922485955, "grad_norm": 13.0625, "learning_rate": 9.067667032885335e-06, "loss": 2.0954, "step": 1934 }, { "epoch": 0.2218782249742002, "grad_norm": 13.9375, "learning_rate": 9.06658690059678e-06, "loss": 1.9755, "step": 1935 }, { "epoch": 0.2219928907235409, "grad_norm": 12.125, "learning_rate": 9.06550620740008e-06, "loss": 1.3554, "step": 1936 }, { "epoch": 0.22210755647288155, "grad_norm": 12.9375, "learning_rate": 9.064424953444297e-06, "loss": 1.769, "step": 1937 }, { "epoch": 0.2222222222222222, "grad_norm": 16.625, "learning_rate": 9.063343138878564e-06, "loss": 1.9907, "step": 1938 }, { "epoch": 0.2223368879715629, "grad_norm": 12.875, "learning_rate": 9.0622607638521e-06, "loss": 1.585, "step": 1939 }, { "epoch": 0.22245155372090356, "grad_norm": 14.8125, "learning_rate": 9.061177828514197e-06, "loss": 1.7266, "step": 1940 }, { "epoch": 0.22256621947024424, "grad_norm": 12.0625, "learning_rate": 9.060094333014227e-06, "loss": 1.9909, "step": 1941 }, { "epoch": 0.2226808852195849, "grad_norm": 14.75, "learning_rate": 9.059010277501634e-06, "loss": 1.9098, "step": 1942 }, { "epoch": 0.2227955509689256, "grad_norm": 12.6875, "learning_rate": 9.057925662125946e-06, "loss": 1.8824, "step": 1943 }, { "epoch": 0.22291021671826625, "grad_norm": 14.125, "learning_rate": 9.056840487036764e-06, "loss": 1.9891, "step": 1944 }, { "epoch": 0.22302488246760693, "grad_norm": 12.9375, "learning_rate": 9.05575475238377e-06, "loss": 1.4922, "step": 1945 }, { "epoch": 0.2231395482169476, "grad_norm": 12.4375, "learning_rate": 9.054668458316713e-06, "loss": 1.5908, "step": 1946 }, { "epoch": 0.22325421396628828, "grad_norm": 17.5, "learning_rate": 9.053581604985433e-06, "loss": 1.9377, "step": 1947 }, { "epoch": 0.22336887971562894, "grad_norm": 15.4375, "learning_rate": 9.05249419253984e-06, "loss": 1.6555, "step": 1948 }, { "epoch": 0.22348354546496962, "grad_norm": 12.9375, "learning_rate": 9.051406221129919e-06, "loss": 1.7617, "step": 1949 }, { "epoch": 0.22359821121431028, "grad_norm": 12.1875, "learning_rate": 9.050317690905737e-06, "loss": 1.6737, "step": 1950 }, { "epoch": 0.22371287696365097, "grad_norm": 16.25, "learning_rate": 9.049228602017437e-06, "loss": 2.3386, "step": 1951 }, { "epoch": 0.22382754271299163, "grad_norm": 15.25, "learning_rate": 9.048138954615235e-06, "loss": 1.4539, "step": 1952 }, { "epoch": 0.22394220846233232, "grad_norm": 12.625, "learning_rate": 9.04704874884943e-06, "loss": 1.7132, "step": 1953 }, { "epoch": 0.22405687421167297, "grad_norm": 13.875, "learning_rate": 9.045957984870394e-06, "loss": 1.4791, "step": 1954 }, { "epoch": 0.22417153996101363, "grad_norm": 15.1875, "learning_rate": 9.044866662828576e-06, "loss": 1.5771, "step": 1955 }, { "epoch": 0.22428620571035432, "grad_norm": 19.125, "learning_rate": 9.043774782874503e-06, "loss": 1.7681, "step": 1956 }, { "epoch": 0.22440087145969498, "grad_norm": 16.875, "learning_rate": 9.042682345158782e-06, "loss": 1.8073, "step": 1957 }, { "epoch": 0.22451553720903566, "grad_norm": 12.3125, "learning_rate": 9.041589349832091e-06, "loss": 1.901, "step": 1958 }, { "epoch": 0.22463020295837632, "grad_norm": 17.625, "learning_rate": 9.04049579704519e-06, "loss": 1.7092, "step": 1959 }, { "epoch": 0.224744868707717, "grad_norm": 12.625, "learning_rate": 9.039401686948912e-06, "loss": 1.5572, "step": 1960 }, { "epoch": 0.22485953445705767, "grad_norm": 13.75, "learning_rate": 9.03830701969417e-06, "loss": 1.9463, "step": 1961 }, { "epoch": 0.22497420020639836, "grad_norm": 24.75, "learning_rate": 9.03721179543195e-06, "loss": 1.4991, "step": 1962 }, { "epoch": 0.22508886595573901, "grad_norm": 11.625, "learning_rate": 9.03611601431332e-06, "loss": 1.5579, "step": 1963 }, { "epoch": 0.2252035317050797, "grad_norm": 13.1875, "learning_rate": 9.035019676489422e-06, "loss": 1.6585, "step": 1964 }, { "epoch": 0.22531819745442036, "grad_norm": 27.0, "learning_rate": 9.033922782111474e-06, "loss": 1.657, "step": 1965 }, { "epoch": 0.22543286320376105, "grad_norm": 13.6875, "learning_rate": 9.03282533133077e-06, "loss": 1.8763, "step": 1966 }, { "epoch": 0.2255475289531017, "grad_norm": 15.1875, "learning_rate": 9.031727324298687e-06, "loss": 1.6223, "step": 1967 }, { "epoch": 0.2256621947024424, "grad_norm": 14.0625, "learning_rate": 9.030628761166669e-06, "loss": 1.757, "step": 1968 }, { "epoch": 0.22577686045178305, "grad_norm": 14.0, "learning_rate": 9.029529642086246e-06, "loss": 2.0612, "step": 1969 }, { "epoch": 0.22589152620112374, "grad_norm": 11.625, "learning_rate": 9.028429967209015e-06, "loss": 1.656, "step": 1970 }, { "epoch": 0.2260061919504644, "grad_norm": 15.75, "learning_rate": 9.027329736686664e-06, "loss": 1.8427, "step": 1971 }, { "epoch": 0.22612085769980506, "grad_norm": 16.875, "learning_rate": 9.02622895067094e-06, "loss": 2.2811, "step": 1972 }, { "epoch": 0.22623552344914574, "grad_norm": 14.125, "learning_rate": 9.02512760931368e-06, "loss": 1.5319, "step": 1973 }, { "epoch": 0.2263501891984864, "grad_norm": 13.9375, "learning_rate": 9.024025712766792e-06, "loss": 1.6673, "step": 1974 }, { "epoch": 0.2264648549478271, "grad_norm": 16.125, "learning_rate": 9.022923261182263e-06, "loss": 1.7235, "step": 1975 }, { "epoch": 0.22657952069716775, "grad_norm": 12.875, "learning_rate": 9.021820254712154e-06, "loss": 1.7576, "step": 1976 }, { "epoch": 0.22669418644650843, "grad_norm": 13.8125, "learning_rate": 9.020716693508602e-06, "loss": 1.4927, "step": 1977 }, { "epoch": 0.2268088521958491, "grad_norm": 16.25, "learning_rate": 9.019612577723826e-06, "loss": 1.5885, "step": 1978 }, { "epoch": 0.22692351794518978, "grad_norm": 17.125, "learning_rate": 9.018507907510115e-06, "loss": 1.9538, "step": 1979 }, { "epoch": 0.22703818369453044, "grad_norm": 13.25, "learning_rate": 9.017402683019838e-06, "loss": 1.9618, "step": 1980 }, { "epoch": 0.22715284944387112, "grad_norm": 13.25, "learning_rate": 9.01629690440544e-06, "loss": 1.5646, "step": 1981 }, { "epoch": 0.22726751519321178, "grad_norm": 18.625, "learning_rate": 9.01519057181944e-06, "loss": 1.5485, "step": 1982 }, { "epoch": 0.22738218094255247, "grad_norm": 31.25, "learning_rate": 9.014083685414437e-06, "loss": 2.076, "step": 1983 }, { "epoch": 0.22749684669189313, "grad_norm": 11.5625, "learning_rate": 9.012976245343106e-06, "loss": 1.7708, "step": 1984 }, { "epoch": 0.22761151244123382, "grad_norm": 30.625, "learning_rate": 9.011868251758195e-06, "loss": 1.7176, "step": 1985 }, { "epoch": 0.22772617819057447, "grad_norm": 15.8125, "learning_rate": 9.010759704812534e-06, "loss": 1.7022, "step": 1986 }, { "epoch": 0.22784084393991516, "grad_norm": 12.6875, "learning_rate": 9.009650604659024e-06, "loss": 1.4825, "step": 1987 }, { "epoch": 0.22795550968925582, "grad_norm": 13.0625, "learning_rate": 9.008540951450642e-06, "loss": 1.6673, "step": 1988 }, { "epoch": 0.22807017543859648, "grad_norm": 13.875, "learning_rate": 9.007430745340447e-06, "loss": 1.8256, "step": 1989 }, { "epoch": 0.22818484118793717, "grad_norm": 12.5, "learning_rate": 9.006319986481568e-06, "loss": 1.9064, "step": 1990 }, { "epoch": 0.22829950693727782, "grad_norm": 16.125, "learning_rate": 9.005208675027215e-06, "loss": 1.8077, "step": 1991 }, { "epoch": 0.2284141726866185, "grad_norm": 14.4375, "learning_rate": 9.00409681113067e-06, "loss": 1.3712, "step": 1992 }, { "epoch": 0.22852883843595917, "grad_norm": 13.625, "learning_rate": 9.002984394945298e-06, "loss": 1.6914, "step": 1993 }, { "epoch": 0.22864350418529986, "grad_norm": 13.0, "learning_rate": 9.001871426624529e-06, "loss": 1.8094, "step": 1994 }, { "epoch": 0.22875816993464052, "grad_norm": 17.0, "learning_rate": 9.000757906321882e-06, "loss": 1.5314, "step": 1995 }, { "epoch": 0.2288728356839812, "grad_norm": 12.8125, "learning_rate": 8.99964383419094e-06, "loss": 1.9294, "step": 1996 }, { "epoch": 0.22898750143332186, "grad_norm": 13.25, "learning_rate": 8.998529210385376e-06, "loss": 1.616, "step": 1997 }, { "epoch": 0.22910216718266255, "grad_norm": 13.8125, "learning_rate": 8.997414035058924e-06, "loss": 1.6481, "step": 1998 }, { "epoch": 0.2292168329320032, "grad_norm": 15.4375, "learning_rate": 8.996298308365402e-06, "loss": 1.6009, "step": 1999 }, { "epoch": 0.2293314986813439, "grad_norm": 12.6875, "learning_rate": 8.995182030458707e-06, "loss": 1.9426, "step": 2000 }, { "epoch": 0.22944616443068455, "grad_norm": 21.625, "learning_rate": 8.994065201492804e-06, "loss": 1.5293, "step": 2001 }, { "epoch": 0.22956083018002524, "grad_norm": 26.5, "learning_rate": 8.992947821621742e-06, "loss": 1.9446, "step": 2002 }, { "epoch": 0.2296754959293659, "grad_norm": 14.75, "learning_rate": 8.99182989099964e-06, "loss": 1.7684, "step": 2003 }, { "epoch": 0.22979016167870658, "grad_norm": 15.5625, "learning_rate": 8.990711409780695e-06, "loss": 1.7342, "step": 2004 }, { "epoch": 0.22990482742804724, "grad_norm": 11.375, "learning_rate": 8.989592378119181e-06, "loss": 1.9667, "step": 2005 }, { "epoch": 0.2300194931773879, "grad_norm": 12.9375, "learning_rate": 8.988472796169448e-06, "loss": 1.7343, "step": 2006 }, { "epoch": 0.2301341589267286, "grad_norm": 13.6875, "learning_rate": 8.98735266408592e-06, "loss": 1.5809, "step": 2007 }, { "epoch": 0.23024882467606925, "grad_norm": 12.5, "learning_rate": 8.986231982023098e-06, "loss": 1.6485, "step": 2008 }, { "epoch": 0.23036349042540993, "grad_norm": 14.3125, "learning_rate": 8.985110750135557e-06, "loss": 1.8944, "step": 2009 }, { "epoch": 0.2304781561747506, "grad_norm": 13.3125, "learning_rate": 8.983988968577951e-06, "loss": 1.6161, "step": 2010 }, { "epoch": 0.23059282192409128, "grad_norm": 12.5625, "learning_rate": 8.98286663750501e-06, "loss": 1.7153, "step": 2011 }, { "epoch": 0.23070748767343194, "grad_norm": 17.875, "learning_rate": 8.981743757071536e-06, "loss": 1.8735, "step": 2012 }, { "epoch": 0.23082215342277262, "grad_norm": 17.25, "learning_rate": 8.98062032743241e-06, "loss": 1.7914, "step": 2013 }, { "epoch": 0.23093681917211328, "grad_norm": 18.75, "learning_rate": 8.979496348742587e-06, "loss": 1.7242, "step": 2014 }, { "epoch": 0.23105148492145397, "grad_norm": 16.375, "learning_rate": 8.9783718211571e-06, "loss": 1.4087, "step": 2015 }, { "epoch": 0.23116615067079463, "grad_norm": 13.5625, "learning_rate": 8.977246744831053e-06, "loss": 1.9235, "step": 2016 }, { "epoch": 0.23128081642013532, "grad_norm": 14.4375, "learning_rate": 8.976121119919633e-06, "loss": 1.8122, "step": 2017 }, { "epoch": 0.23139548216947597, "grad_norm": 12.875, "learning_rate": 8.974994946578094e-06, "loss": 1.8919, "step": 2018 }, { "epoch": 0.23151014791881666, "grad_norm": 16.125, "learning_rate": 8.973868224961773e-06, "loss": 1.709, "step": 2019 }, { "epoch": 0.23162481366815732, "grad_norm": 18.375, "learning_rate": 8.97274095522608e-06, "loss": 2.3454, "step": 2020 }, { "epoch": 0.231739479417498, "grad_norm": 15.4375, "learning_rate": 8.971613137526498e-06, "loss": 1.8582, "step": 2021 }, { "epoch": 0.23185414516683867, "grad_norm": 13.375, "learning_rate": 8.97048477201859e-06, "loss": 1.6778, "step": 2022 }, { "epoch": 0.23196881091617932, "grad_norm": 10.9375, "learning_rate": 8.969355858857994e-06, "loss": 1.748, "step": 2023 }, { "epoch": 0.23208347666552, "grad_norm": 14.75, "learning_rate": 8.968226398200418e-06, "loss": 2.2155, "step": 2024 }, { "epoch": 0.23219814241486067, "grad_norm": 15.5625, "learning_rate": 8.967096390201653e-06, "loss": 2.267, "step": 2025 }, { "epoch": 0.23231280816420136, "grad_norm": 11.8125, "learning_rate": 8.96596583501756e-06, "loss": 1.9413, "step": 2026 }, { "epoch": 0.23242747391354202, "grad_norm": 14.0, "learning_rate": 8.964834732804079e-06, "loss": 1.9592, "step": 2027 }, { "epoch": 0.2325421396628827, "grad_norm": 16.125, "learning_rate": 8.963703083717223e-06, "loss": 1.6745, "step": 2028 }, { "epoch": 0.23265680541222336, "grad_norm": 15.125, "learning_rate": 8.96257088791308e-06, "loss": 1.8467, "step": 2029 }, { "epoch": 0.23277147116156405, "grad_norm": 14.5, "learning_rate": 8.961438145547819e-06, "loss": 2.0357, "step": 2030 }, { "epoch": 0.2328861369109047, "grad_norm": 18.75, "learning_rate": 8.960304856777676e-06, "loss": 2.0004, "step": 2031 }, { "epoch": 0.2330008026602454, "grad_norm": 13.25, "learning_rate": 8.959171021758968e-06, "loss": 1.6014, "step": 2032 }, { "epoch": 0.23311546840958605, "grad_norm": 14.5, "learning_rate": 8.958036640648087e-06, "loss": 1.8077, "step": 2033 }, { "epoch": 0.23323013415892674, "grad_norm": 17.125, "learning_rate": 8.956901713601499e-06, "loss": 1.9695, "step": 2034 }, { "epoch": 0.2333447999082674, "grad_norm": 15.0, "learning_rate": 8.955766240775744e-06, "loss": 2.0812, "step": 2035 }, { "epoch": 0.23345946565760808, "grad_norm": 27.25, "learning_rate": 8.95463022232744e-06, "loss": 1.7835, "step": 2036 }, { "epoch": 0.23357413140694874, "grad_norm": 19.5, "learning_rate": 8.95349365841328e-06, "loss": 1.7408, "step": 2037 }, { "epoch": 0.23368879715628943, "grad_norm": 14.75, "learning_rate": 8.95235654919003e-06, "loss": 1.7172, "step": 2038 }, { "epoch": 0.2338034629056301, "grad_norm": 15.25, "learning_rate": 8.95121889481453e-06, "loss": 2.2248, "step": 2039 }, { "epoch": 0.23391812865497075, "grad_norm": 12.9375, "learning_rate": 8.950080695443704e-06, "loss": 1.702, "step": 2040 }, { "epoch": 0.23403279440431143, "grad_norm": 15.8125, "learning_rate": 8.94894195123454e-06, "loss": 1.9728, "step": 2041 }, { "epoch": 0.2341474601536521, "grad_norm": 27.375, "learning_rate": 8.947802662344108e-06, "loss": 2.3873, "step": 2042 }, { "epoch": 0.23426212590299278, "grad_norm": 12.75, "learning_rate": 8.94666282892955e-06, "loss": 1.6386, "step": 2043 }, { "epoch": 0.23437679165233344, "grad_norm": 15.375, "learning_rate": 8.945522451148087e-06, "loss": 1.5876, "step": 2044 }, { "epoch": 0.23449145740167412, "grad_norm": 13.0625, "learning_rate": 8.944381529157008e-06, "loss": 1.6, "step": 2045 }, { "epoch": 0.23460612315101478, "grad_norm": 14.6875, "learning_rate": 8.943240063113686e-06, "loss": 1.6069, "step": 2046 }, { "epoch": 0.23472078890035547, "grad_norm": 13.375, "learning_rate": 8.94209805317556e-06, "loss": 1.7934, "step": 2047 }, { "epoch": 0.23483545464969613, "grad_norm": 14.5625, "learning_rate": 8.940955499500152e-06, "loss": 2.0254, "step": 2048 }, { "epoch": 0.23495012039903682, "grad_norm": 18.75, "learning_rate": 8.939812402245054e-06, "loss": 1.9024, "step": 2049 }, { "epoch": 0.23506478614837747, "grad_norm": 9.9375, "learning_rate": 8.938668761567935e-06, "loss": 1.4213, "step": 2050 }, { "epoch": 0.23517945189771816, "grad_norm": 16.625, "learning_rate": 8.937524577626537e-06, "loss": 1.6192, "step": 2051 }, { "epoch": 0.23529411764705882, "grad_norm": 11.3125, "learning_rate": 8.93637985057868e-06, "loss": 1.5907, "step": 2052 }, { "epoch": 0.2354087833963995, "grad_norm": 11.9375, "learning_rate": 8.935234580582259e-06, "loss": 1.6993, "step": 2053 }, { "epoch": 0.23552344914574017, "grad_norm": 17.875, "learning_rate": 8.934088767795236e-06, "loss": 1.9525, "step": 2054 }, { "epoch": 0.23563811489508085, "grad_norm": 13.8125, "learning_rate": 8.932942412375663e-06, "loss": 1.7081, "step": 2055 }, { "epoch": 0.2357527806444215, "grad_norm": 10.1875, "learning_rate": 8.931795514481649e-06, "loss": 1.3414, "step": 2056 }, { "epoch": 0.23586744639376217, "grad_norm": 11.625, "learning_rate": 8.930648074271391e-06, "loss": 1.6504, "step": 2057 }, { "epoch": 0.23598211214310286, "grad_norm": 14.875, "learning_rate": 8.929500091903158e-06, "loss": 2.0289, "step": 2058 }, { "epoch": 0.23609677789244352, "grad_norm": 13.1875, "learning_rate": 8.92835156753529e-06, "loss": 1.6404, "step": 2059 }, { "epoch": 0.2362114436417842, "grad_norm": 18.625, "learning_rate": 8.927202501326205e-06, "loss": 1.7722, "step": 2060 }, { "epoch": 0.23632610939112486, "grad_norm": 13.4375, "learning_rate": 8.92605289343439e-06, "loss": 1.6179, "step": 2061 }, { "epoch": 0.23644077514046555, "grad_norm": 11.4375, "learning_rate": 8.92490274401842e-06, "loss": 1.5276, "step": 2062 }, { "epoch": 0.2365554408898062, "grad_norm": 14.5625, "learning_rate": 8.923752053236932e-06, "loss": 2.2315, "step": 2063 }, { "epoch": 0.2366701066391469, "grad_norm": 11.9375, "learning_rate": 8.922600821248639e-06, "loss": 1.7431, "step": 2064 }, { "epoch": 0.23678477238848755, "grad_norm": 13.125, "learning_rate": 8.921449048212336e-06, "loss": 1.6915, "step": 2065 }, { "epoch": 0.23689943813782824, "grad_norm": 15.375, "learning_rate": 8.920296734286886e-06, "loss": 1.9313, "step": 2066 }, { "epoch": 0.2370141038871689, "grad_norm": 13.5, "learning_rate": 8.91914387963123e-06, "loss": 2.0753, "step": 2067 }, { "epoch": 0.23712876963650958, "grad_norm": 10.6875, "learning_rate": 8.917990484404383e-06, "loss": 1.2579, "step": 2068 }, { "epoch": 0.23724343538585024, "grad_norm": 12.9375, "learning_rate": 8.916836548765429e-06, "loss": 1.8321, "step": 2069 }, { "epoch": 0.23735810113519093, "grad_norm": 13.8125, "learning_rate": 8.915682072873537e-06, "loss": 1.7845, "step": 2070 }, { "epoch": 0.2374727668845316, "grad_norm": 12.0625, "learning_rate": 8.91452705688794e-06, "loss": 1.8145, "step": 2071 }, { "epoch": 0.23758743263387228, "grad_norm": 13.4375, "learning_rate": 8.913371500967956e-06, "loss": 1.9326, "step": 2072 }, { "epoch": 0.23770209838321293, "grad_norm": 13.9375, "learning_rate": 8.912215405272968e-06, "loss": 1.5377, "step": 2073 }, { "epoch": 0.2378167641325536, "grad_norm": 17.5, "learning_rate": 8.911058769962438e-06, "loss": 1.373, "step": 2074 }, { "epoch": 0.23793142988189428, "grad_norm": 13.125, "learning_rate": 8.909901595195902e-06, "loss": 1.9962, "step": 2075 }, { "epoch": 0.23804609563123494, "grad_norm": 15.75, "learning_rate": 8.908743881132973e-06, "loss": 1.9551, "step": 2076 }, { "epoch": 0.23816076138057563, "grad_norm": 13.625, "learning_rate": 8.90758562793333e-06, "loss": 1.776, "step": 2077 }, { "epoch": 0.23827542712991628, "grad_norm": 16.75, "learning_rate": 8.906426835756736e-06, "loss": 1.5995, "step": 2078 }, { "epoch": 0.23839009287925697, "grad_norm": 15.1875, "learning_rate": 8.905267504763024e-06, "loss": 1.9871, "step": 2079 }, { "epoch": 0.23850475862859763, "grad_norm": 11.8125, "learning_rate": 8.9041076351121e-06, "loss": 1.698, "step": 2080 }, { "epoch": 0.23861942437793832, "grad_norm": 14.5625, "learning_rate": 8.902947226963948e-06, "loss": 1.9222, "step": 2081 }, { "epoch": 0.23873409012727898, "grad_norm": 15.4375, "learning_rate": 8.901786280478621e-06, "loss": 1.9437, "step": 2082 }, { "epoch": 0.23884875587661966, "grad_norm": 13.5625, "learning_rate": 8.900624795816253e-06, "loss": 2.0668, "step": 2083 }, { "epoch": 0.23896342162596032, "grad_norm": 10.75, "learning_rate": 8.899462773137047e-06, "loss": 1.8702, "step": 2084 }, { "epoch": 0.239078087375301, "grad_norm": 12.75, "learning_rate": 8.898300212601281e-06, "loss": 1.6048, "step": 2085 }, { "epoch": 0.23919275312464167, "grad_norm": 12.5, "learning_rate": 8.89713711436931e-06, "loss": 1.6364, "step": 2086 }, { "epoch": 0.23930741887398235, "grad_norm": 12.5, "learning_rate": 8.895973478601561e-06, "loss": 1.6495, "step": 2087 }, { "epoch": 0.239422084623323, "grad_norm": 13.9375, "learning_rate": 8.894809305458534e-06, "loss": 1.6712, "step": 2088 }, { "epoch": 0.2395367503726637, "grad_norm": 11.8125, "learning_rate": 8.893644595100803e-06, "loss": 1.7443, "step": 2089 }, { "epoch": 0.23965141612200436, "grad_norm": 18.75, "learning_rate": 8.892479347689023e-06, "loss": 2.0331, "step": 2090 }, { "epoch": 0.23976608187134502, "grad_norm": 16.125, "learning_rate": 8.891313563383912e-06, "loss": 1.6266, "step": 2091 }, { "epoch": 0.2398807476206857, "grad_norm": 15.5625, "learning_rate": 8.890147242346272e-06, "loss": 1.6862, "step": 2092 }, { "epoch": 0.23999541337002636, "grad_norm": 15.0625, "learning_rate": 8.888980384736972e-06, "loss": 1.8511, "step": 2093 }, { "epoch": 0.24011007911936705, "grad_norm": 19.0, "learning_rate": 8.887812990716958e-06, "loss": 1.5903, "step": 2094 }, { "epoch": 0.2402247448687077, "grad_norm": 13.0625, "learning_rate": 8.88664506044725e-06, "loss": 1.8766, "step": 2095 }, { "epoch": 0.2403394106180484, "grad_norm": 11.3125, "learning_rate": 8.885476594088944e-06, "loss": 1.6461, "step": 2096 }, { "epoch": 0.24045407636738905, "grad_norm": 13.1875, "learning_rate": 8.884307591803204e-06, "loss": 1.5643, "step": 2097 }, { "epoch": 0.24056874211672974, "grad_norm": 17.875, "learning_rate": 8.883138053751275e-06, "loss": 1.7991, "step": 2098 }, { "epoch": 0.2406834078660704, "grad_norm": 13.5625, "learning_rate": 8.88196798009447e-06, "loss": 2.2975, "step": 2099 }, { "epoch": 0.24079807361541108, "grad_norm": 19.25, "learning_rate": 8.880797370994178e-06, "loss": 2.0813, "step": 2100 }, { "epoch": 0.24091273936475174, "grad_norm": 28.0, "learning_rate": 8.879626226611864e-06, "loss": 1.8406, "step": 2101 }, { "epoch": 0.24102740511409243, "grad_norm": 12.5, "learning_rate": 8.878454547109066e-06, "loss": 1.9862, "step": 2102 }, { "epoch": 0.2411420708634331, "grad_norm": 15.3125, "learning_rate": 8.877282332647392e-06, "loss": 1.8796, "step": 2103 }, { "epoch": 0.24125673661277378, "grad_norm": 14.375, "learning_rate": 8.876109583388527e-06, "loss": 1.9784, "step": 2104 }, { "epoch": 0.24137140236211443, "grad_norm": 12.4375, "learning_rate": 8.874936299494232e-06, "loss": 1.6288, "step": 2105 }, { "epoch": 0.24148606811145512, "grad_norm": 14.125, "learning_rate": 8.873762481126338e-06, "loss": 1.9047, "step": 2106 }, { "epoch": 0.24160073386079578, "grad_norm": 13.25, "learning_rate": 8.872588128446749e-06, "loss": 1.9549, "step": 2107 }, { "epoch": 0.24171539961013644, "grad_norm": 13.0625, "learning_rate": 8.871413241617446e-06, "loss": 1.4413, "step": 2108 }, { "epoch": 0.24183006535947713, "grad_norm": 12.875, "learning_rate": 8.870237820800482e-06, "loss": 2.1624, "step": 2109 }, { "epoch": 0.24194473110881778, "grad_norm": 13.625, "learning_rate": 8.869061866157985e-06, "loss": 2.2014, "step": 2110 }, { "epoch": 0.24205939685815847, "grad_norm": 13.9375, "learning_rate": 8.867885377852154e-06, "loss": 2.0527, "step": 2111 }, { "epoch": 0.24217406260749913, "grad_norm": 13.1875, "learning_rate": 8.866708356045264e-06, "loss": 2.1631, "step": 2112 }, { "epoch": 0.24228872835683982, "grad_norm": 13.3125, "learning_rate": 8.865530800899662e-06, "loss": 1.9287, "step": 2113 }, { "epoch": 0.24240339410618048, "grad_norm": 13.9375, "learning_rate": 8.864352712577771e-06, "loss": 1.907, "step": 2114 }, { "epoch": 0.24251805985552116, "grad_norm": 13.4375, "learning_rate": 8.863174091242083e-06, "loss": 1.8103, "step": 2115 }, { "epoch": 0.24263272560486182, "grad_norm": 11.1875, "learning_rate": 8.861994937055168e-06, "loss": 1.5109, "step": 2116 }, { "epoch": 0.2427473913542025, "grad_norm": 12.125, "learning_rate": 8.86081525017967e-06, "loss": 1.9665, "step": 2117 }, { "epoch": 0.24286205710354317, "grad_norm": 14.6875, "learning_rate": 8.859635030778299e-06, "loss": 1.8938, "step": 2118 }, { "epoch": 0.24297672285288385, "grad_norm": 14.5, "learning_rate": 8.858454279013848e-06, "loss": 1.6492, "step": 2119 }, { "epoch": 0.2430913886022245, "grad_norm": 27.125, "learning_rate": 8.857272995049179e-06, "loss": 1.8055, "step": 2120 }, { "epoch": 0.2432060543515652, "grad_norm": 11.875, "learning_rate": 8.856091179047226e-06, "loss": 1.5781, "step": 2121 }, { "epoch": 0.24332072010090586, "grad_norm": 12.75, "learning_rate": 8.854908831170999e-06, "loss": 1.8058, "step": 2122 }, { "epoch": 0.24343538585024654, "grad_norm": 14.0, "learning_rate": 8.853725951583579e-06, "loss": 2.0445, "step": 2123 }, { "epoch": 0.2435500515995872, "grad_norm": 11.9375, "learning_rate": 8.852542540448124e-06, "loss": 1.9413, "step": 2124 }, { "epoch": 0.24366471734892786, "grad_norm": 14.0, "learning_rate": 8.851358597927859e-06, "loss": 1.6424, "step": 2125 }, { "epoch": 0.24377938309826855, "grad_norm": 14.8125, "learning_rate": 8.850174124186092e-06, "loss": 1.9653, "step": 2126 }, { "epoch": 0.2438940488476092, "grad_norm": 13.8125, "learning_rate": 8.848989119386193e-06, "loss": 1.8355, "step": 2127 }, { "epoch": 0.2440087145969499, "grad_norm": 14.625, "learning_rate": 8.847803583691615e-06, "loss": 1.8676, "step": 2128 }, { "epoch": 0.24412338034629055, "grad_norm": 12.0, "learning_rate": 8.846617517265878e-06, "loss": 1.8884, "step": 2129 }, { "epoch": 0.24423804609563124, "grad_norm": 13.6875, "learning_rate": 8.845430920272578e-06, "loss": 2.094, "step": 2130 }, { "epoch": 0.2443527118449719, "grad_norm": 29.0, "learning_rate": 8.844243792875384e-06, "loss": 1.6124, "step": 2131 }, { "epoch": 0.24446737759431258, "grad_norm": 11.4375, "learning_rate": 8.843056135238036e-06, "loss": 1.5078, "step": 2132 }, { "epoch": 0.24458204334365324, "grad_norm": 14.125, "learning_rate": 8.841867947524349e-06, "loss": 2.1791, "step": 2133 }, { "epoch": 0.24469670909299393, "grad_norm": 14.875, "learning_rate": 8.840679229898212e-06, "loss": 2.0801, "step": 2134 }, { "epoch": 0.2448113748423346, "grad_norm": 12.6875, "learning_rate": 8.839489982523584e-06, "loss": 1.5154, "step": 2135 }, { "epoch": 0.24492604059167528, "grad_norm": 10.6875, "learning_rate": 8.838300205564501e-06, "loss": 1.4544, "step": 2136 }, { "epoch": 0.24504070634101593, "grad_norm": 13.0, "learning_rate": 8.83710989918507e-06, "loss": 1.952, "step": 2137 }, { "epoch": 0.24515537209035662, "grad_norm": 13.3125, "learning_rate": 8.83591906354947e-06, "loss": 1.7707, "step": 2138 }, { "epoch": 0.24527003783969728, "grad_norm": 14.375, "learning_rate": 8.834727698821954e-06, "loss": 1.9633, "step": 2139 }, { "epoch": 0.24538470358903797, "grad_norm": 9.9375, "learning_rate": 8.833535805166848e-06, "loss": 1.7866, "step": 2140 }, { "epoch": 0.24549936933837863, "grad_norm": 15.25, "learning_rate": 8.832343382748552e-06, "loss": 1.6828, "step": 2141 }, { "epoch": 0.24561403508771928, "grad_norm": 13.1875, "learning_rate": 8.831150431731537e-06, "loss": 1.6787, "step": 2142 }, { "epoch": 0.24572870083705997, "grad_norm": 24.5, "learning_rate": 8.82995695228035e-06, "loss": 1.9251, "step": 2143 }, { "epoch": 0.24584336658640063, "grad_norm": 14.3125, "learning_rate": 8.828762944559606e-06, "loss": 1.979, "step": 2144 }, { "epoch": 0.24595803233574132, "grad_norm": 16.25, "learning_rate": 8.827568408733996e-06, "loss": 1.5983, "step": 2145 }, { "epoch": 0.24607269808508198, "grad_norm": 27.125, "learning_rate": 8.826373344968285e-06, "loss": 1.6854, "step": 2146 }, { "epoch": 0.24618736383442266, "grad_norm": 12.375, "learning_rate": 8.82517775342731e-06, "loss": 1.9301, "step": 2147 }, { "epoch": 0.24630202958376332, "grad_norm": 11.625, "learning_rate": 8.823981634275976e-06, "loss": 1.3811, "step": 2148 }, { "epoch": 0.246416695333104, "grad_norm": 15.125, "learning_rate": 8.822784987679267e-06, "loss": 2.0056, "step": 2149 }, { "epoch": 0.24653136108244467, "grad_norm": 11.125, "learning_rate": 8.821587813802239e-06, "loss": 1.5299, "step": 2150 }, { "epoch": 0.24664602683178535, "grad_norm": 28.75, "learning_rate": 8.820390112810018e-06, "loss": 1.9226, "step": 2151 }, { "epoch": 0.246760692581126, "grad_norm": 13.1875, "learning_rate": 8.819191884867804e-06, "loss": 1.6917, "step": 2152 }, { "epoch": 0.2468753583304667, "grad_norm": 14.4375, "learning_rate": 8.81799313014087e-06, "loss": 1.8425, "step": 2153 }, { "epoch": 0.24699002407980736, "grad_norm": 13.5625, "learning_rate": 8.81679384879456e-06, "loss": 2.2053, "step": 2154 }, { "epoch": 0.24710468982914804, "grad_norm": 20.25, "learning_rate": 8.815594040994295e-06, "loss": 2.0197, "step": 2155 }, { "epoch": 0.2472193555784887, "grad_norm": 13.0625, "learning_rate": 8.814393706905563e-06, "loss": 1.7477, "step": 2156 }, { "epoch": 0.2473340213278294, "grad_norm": 12.375, "learning_rate": 8.81319284669393e-06, "loss": 1.9985, "step": 2157 }, { "epoch": 0.24744868707717005, "grad_norm": 14.75, "learning_rate": 8.811991460525027e-06, "loss": 2.1224, "step": 2158 }, { "epoch": 0.2475633528265107, "grad_norm": 17.375, "learning_rate": 8.810789548564566e-06, "loss": 1.5374, "step": 2159 }, { "epoch": 0.2476780185758514, "grad_norm": 13.375, "learning_rate": 8.809587110978328e-06, "loss": 2.0211, "step": 2160 }, { "epoch": 0.24779268432519205, "grad_norm": 12.9375, "learning_rate": 8.808384147932164e-06, "loss": 1.5997, "step": 2161 }, { "epoch": 0.24790735007453274, "grad_norm": 13.1875, "learning_rate": 8.807180659592004e-06, "loss": 1.8655, "step": 2162 }, { "epoch": 0.2480220158238734, "grad_norm": 15.75, "learning_rate": 8.805976646123842e-06, "loss": 1.7183, "step": 2163 }, { "epoch": 0.24813668157321409, "grad_norm": 29.0, "learning_rate": 8.804772107693748e-06, "loss": 1.8291, "step": 2164 }, { "epoch": 0.24825134732255474, "grad_norm": 11.5625, "learning_rate": 8.80356704446787e-06, "loss": 1.3961, "step": 2165 }, { "epoch": 0.24836601307189543, "grad_norm": 16.625, "learning_rate": 8.80236145661242e-06, "loss": 1.9074, "step": 2166 }, { "epoch": 0.2484806788212361, "grad_norm": 15.1875, "learning_rate": 8.801155344293687e-06, "loss": 2.0436, "step": 2167 }, { "epoch": 0.24859534457057678, "grad_norm": 12.3125, "learning_rate": 8.799948707678031e-06, "loss": 2.0606, "step": 2168 }, { "epoch": 0.24871001031991744, "grad_norm": 14.1875, "learning_rate": 8.798741546931884e-06, "loss": 1.7837, "step": 2169 }, { "epoch": 0.24882467606925812, "grad_norm": 14.125, "learning_rate": 8.79753386222175e-06, "loss": 2.088, "step": 2170 }, { "epoch": 0.24893934181859878, "grad_norm": 30.375, "learning_rate": 8.796325653714209e-06, "loss": 1.5082, "step": 2171 }, { "epoch": 0.24905400756793947, "grad_norm": 11.75, "learning_rate": 8.795116921575908e-06, "loss": 1.75, "step": 2172 }, { "epoch": 0.24916867331728013, "grad_norm": 11.625, "learning_rate": 8.79390766597357e-06, "loss": 1.9516, "step": 2173 }, { "epoch": 0.2492833390666208, "grad_norm": 17.0, "learning_rate": 8.792697887073987e-06, "loss": 2.1755, "step": 2174 }, { "epoch": 0.24939800481596147, "grad_norm": 33.25, "learning_rate": 8.791487585044025e-06, "loss": 2.0565, "step": 2175 }, { "epoch": 0.24951267056530213, "grad_norm": 13.6875, "learning_rate": 8.790276760050625e-06, "loss": 1.9436, "step": 2176 }, { "epoch": 0.24962733631464282, "grad_norm": 17.25, "learning_rate": 8.789065412260793e-06, "loss": 1.8287, "step": 2177 }, { "epoch": 0.24974200206398348, "grad_norm": 13.5625, "learning_rate": 8.787853541841614e-06, "loss": 1.7278, "step": 2178 }, { "epoch": 0.24985666781332416, "grad_norm": 14.1875, "learning_rate": 8.786641148960243e-06, "loss": 1.9335, "step": 2179 }, { "epoch": 0.24997133356266482, "grad_norm": 27.75, "learning_rate": 8.785428233783906e-06, "loss": 1.9168, "step": 2180 }, { "epoch": 0.2500859993120055, "grad_norm": 14.875, "learning_rate": 8.7842147964799e-06, "loss": 1.6712, "step": 2181 }, { "epoch": 0.2502006650613462, "grad_norm": 15.9375, "learning_rate": 8.783000837215597e-06, "loss": 1.6125, "step": 2182 }, { "epoch": 0.25031533081068685, "grad_norm": 15.0625, "learning_rate": 8.781786356158438e-06, "loss": 1.9194, "step": 2183 }, { "epoch": 0.2504299965600275, "grad_norm": 13.25, "learning_rate": 8.78057135347594e-06, "loss": 1.9081, "step": 2184 }, { "epoch": 0.25054466230936817, "grad_norm": 35.0, "learning_rate": 8.779355829335686e-06, "loss": 2.0989, "step": 2185 }, { "epoch": 0.2506593280587089, "grad_norm": 17.75, "learning_rate": 8.778139783905337e-06, "loss": 2.575, "step": 2186 }, { "epoch": 0.25077399380804954, "grad_norm": 34.75, "learning_rate": 8.776923217352624e-06, "loss": 1.5712, "step": 2187 }, { "epoch": 0.2508886595573902, "grad_norm": 15.0625, "learning_rate": 8.775706129845348e-06, "loss": 1.6165, "step": 2188 }, { "epoch": 0.25100332530673086, "grad_norm": 20.75, "learning_rate": 8.77448852155138e-06, "loss": 1.7161, "step": 2189 }, { "epoch": 0.2511179910560716, "grad_norm": 15.3125, "learning_rate": 8.773270392638673e-06, "loss": 1.8901, "step": 2190 }, { "epoch": 0.25123265680541224, "grad_norm": 14.0625, "learning_rate": 8.772051743275238e-06, "loss": 1.8066, "step": 2191 }, { "epoch": 0.2513473225547529, "grad_norm": 14.0625, "learning_rate": 8.770832573629167e-06, "loss": 1.4909, "step": 2192 }, { "epoch": 0.25146198830409355, "grad_norm": 14.8125, "learning_rate": 8.76961288386862e-06, "loss": 2.0239, "step": 2193 }, { "epoch": 0.2515766540534342, "grad_norm": 31.375, "learning_rate": 8.768392674161833e-06, "loss": 1.3775, "step": 2194 }, { "epoch": 0.2516913198027749, "grad_norm": 16.5, "learning_rate": 8.767171944677108e-06, "loss": 1.4939, "step": 2195 }, { "epoch": 0.2518059855521156, "grad_norm": 19.125, "learning_rate": 8.765950695582821e-06, "loss": 1.7164, "step": 2196 }, { "epoch": 0.25192065130145624, "grad_norm": 11.4375, "learning_rate": 8.764728927047424e-06, "loss": 1.4572, "step": 2197 }, { "epoch": 0.2520353170507969, "grad_norm": 17.375, "learning_rate": 8.763506639239433e-06, "loss": 2.2827, "step": 2198 }, { "epoch": 0.2521499828001376, "grad_norm": 16.625, "learning_rate": 8.762283832327436e-06, "loss": 1.761, "step": 2199 }, { "epoch": 0.2522646485494783, "grad_norm": 13.8125, "learning_rate": 8.761060506480104e-06, "loss": 1.7102, "step": 2200 }, { "epoch": 0.25237931429881894, "grad_norm": 14.8125, "learning_rate": 8.759836661866165e-06, "loss": 2.3041, "step": 2201 }, { "epoch": 0.2524939800481596, "grad_norm": 11.625, "learning_rate": 8.75861229865443e-06, "loss": 1.8028, "step": 2202 }, { "epoch": 0.2526086457975003, "grad_norm": 13.8125, "learning_rate": 8.757387417013773e-06, "loss": 1.5229, "step": 2203 }, { "epoch": 0.25272331154684097, "grad_norm": 12.6875, "learning_rate": 8.756162017113144e-06, "loss": 1.6533, "step": 2204 }, { "epoch": 0.2528379772961816, "grad_norm": 14.0625, "learning_rate": 8.754936099121564e-06, "loss": 1.7243, "step": 2205 }, { "epoch": 0.2529526430455223, "grad_norm": 12.75, "learning_rate": 8.753709663208126e-06, "loss": 1.8174, "step": 2206 }, { "epoch": 0.253067308794863, "grad_norm": 11.5625, "learning_rate": 8.752482709541989e-06, "loss": 1.6816, "step": 2207 }, { "epoch": 0.25318197454420366, "grad_norm": 20.75, "learning_rate": 8.751255238292393e-06, "loss": 1.8291, "step": 2208 }, { "epoch": 0.2532966402935443, "grad_norm": 19.875, "learning_rate": 8.750027249628643e-06, "loss": 2.0412, "step": 2209 }, { "epoch": 0.253411306042885, "grad_norm": 17.125, "learning_rate": 8.748798743720115e-06, "loss": 2.0291, "step": 2210 }, { "epoch": 0.25352597179222564, "grad_norm": 11.4375, "learning_rate": 8.747569720736258e-06, "loss": 1.7157, "step": 2211 }, { "epoch": 0.25364063754156635, "grad_norm": 12.1875, "learning_rate": 8.746340180846595e-06, "loss": 1.9483, "step": 2212 }, { "epoch": 0.253755303290907, "grad_norm": 43.25, "learning_rate": 8.745110124220715e-06, "loss": 2.3036, "step": 2213 }, { "epoch": 0.25386996904024767, "grad_norm": 24.5, "learning_rate": 8.74387955102828e-06, "loss": 1.8202, "step": 2214 }, { "epoch": 0.2539846347895883, "grad_norm": 12.4375, "learning_rate": 8.742648461439028e-06, "loss": 1.9325, "step": 2215 }, { "epoch": 0.25409930053892904, "grad_norm": 26.125, "learning_rate": 8.741416855622762e-06, "loss": 2.1224, "step": 2216 }, { "epoch": 0.2542139662882697, "grad_norm": 14.0, "learning_rate": 8.740184733749358e-06, "loss": 1.8687, "step": 2217 }, { "epoch": 0.25432863203761036, "grad_norm": 10.9375, "learning_rate": 8.738952095988765e-06, "loss": 1.6888, "step": 2218 }, { "epoch": 0.254443297786951, "grad_norm": 15.6875, "learning_rate": 8.737718942511e-06, "loss": 1.7008, "step": 2219 }, { "epoch": 0.25455796353629173, "grad_norm": 14.3125, "learning_rate": 8.736485273486155e-06, "loss": 1.6106, "step": 2220 }, { "epoch": 0.2546726292856324, "grad_norm": 13.8125, "learning_rate": 8.735251089084392e-06, "loss": 1.9442, "step": 2221 }, { "epoch": 0.25478729503497305, "grad_norm": 31.5, "learning_rate": 8.734016389475939e-06, "loss": 1.9167, "step": 2222 }, { "epoch": 0.2549019607843137, "grad_norm": 12.4375, "learning_rate": 8.732781174831104e-06, "loss": 1.5571, "step": 2223 }, { "epoch": 0.2550166265336544, "grad_norm": 15.875, "learning_rate": 8.731545445320258e-06, "loss": 1.9599, "step": 2224 }, { "epoch": 0.2551312922829951, "grad_norm": 12.875, "learning_rate": 8.730309201113846e-06, "loss": 1.8535, "step": 2225 }, { "epoch": 0.25524595803233574, "grad_norm": 13.25, "learning_rate": 8.729072442382389e-06, "loss": 1.6074, "step": 2226 }, { "epoch": 0.2553606237816764, "grad_norm": 32.25, "learning_rate": 8.727835169296469e-06, "loss": 1.7947, "step": 2227 }, { "epoch": 0.25547528953101706, "grad_norm": 14.625, "learning_rate": 8.726597382026747e-06, "loss": 1.7424, "step": 2228 }, { "epoch": 0.2555899552803578, "grad_norm": 14.125, "learning_rate": 8.725359080743952e-06, "loss": 1.6266, "step": 2229 }, { "epoch": 0.25570462102969843, "grad_norm": 28.0, "learning_rate": 8.724120265618883e-06, "loss": 1.7068, "step": 2230 }, { "epoch": 0.2558192867790391, "grad_norm": 11.9375, "learning_rate": 8.722880936822412e-06, "loss": 1.9883, "step": 2231 }, { "epoch": 0.25593395252837975, "grad_norm": 14.75, "learning_rate": 8.721641094525482e-06, "loss": 1.7218, "step": 2232 }, { "epoch": 0.25604861827772046, "grad_norm": 15.8125, "learning_rate": 8.7204007388991e-06, "loss": 1.7461, "step": 2233 }, { "epoch": 0.2561632840270611, "grad_norm": 13.1875, "learning_rate": 8.719159870114357e-06, "loss": 1.6606, "step": 2234 }, { "epoch": 0.2562779497764018, "grad_norm": 15.4375, "learning_rate": 8.717918488342401e-06, "loss": 2.413, "step": 2235 }, { "epoch": 0.25639261552574244, "grad_norm": 10.0625, "learning_rate": 8.71667659375446e-06, "loss": 1.6659, "step": 2236 }, { "epoch": 0.25650728127508315, "grad_norm": 12.625, "learning_rate": 8.715434186521831e-06, "loss": 1.611, "step": 2237 }, { "epoch": 0.2566219470244238, "grad_norm": 12.0625, "learning_rate": 8.714191266815877e-06, "loss": 1.4516, "step": 2238 }, { "epoch": 0.25673661277376447, "grad_norm": 11.5625, "learning_rate": 8.712947834808037e-06, "loss": 1.3297, "step": 2239 }, { "epoch": 0.25685127852310513, "grad_norm": 13.3125, "learning_rate": 8.711703890669818e-06, "loss": 1.8298, "step": 2240 }, { "epoch": 0.25696594427244585, "grad_norm": 13.125, "learning_rate": 8.7104594345728e-06, "loss": 2.0098, "step": 2241 }, { "epoch": 0.2570806100217865, "grad_norm": 13.0, "learning_rate": 8.70921446668863e-06, "loss": 2.3761, "step": 2242 }, { "epoch": 0.25719527577112716, "grad_norm": 13.3125, "learning_rate": 8.707968987189029e-06, "loss": 1.6482, "step": 2243 }, { "epoch": 0.2573099415204678, "grad_norm": 15.0625, "learning_rate": 8.706722996245784e-06, "loss": 1.4036, "step": 2244 }, { "epoch": 0.2574246072698085, "grad_norm": 18.5, "learning_rate": 8.705476494030763e-06, "loss": 1.7635, "step": 2245 }, { "epoch": 0.2575392730191492, "grad_norm": 12.875, "learning_rate": 8.704229480715888e-06, "loss": 1.8825, "step": 2246 }, { "epoch": 0.25765393876848985, "grad_norm": 17.0, "learning_rate": 8.702981956473166e-06, "loss": 1.8044, "step": 2247 }, { "epoch": 0.2577686045178305, "grad_norm": 19.875, "learning_rate": 8.70173392147467e-06, "loss": 1.6499, "step": 2248 }, { "epoch": 0.25788327026717117, "grad_norm": 13.5625, "learning_rate": 8.70048537589254e-06, "loss": 1.7332, "step": 2249 }, { "epoch": 0.2579979360165119, "grad_norm": 15.0625, "learning_rate": 8.69923631989899e-06, "loss": 1.7768, "step": 2250 }, { "epoch": 0.25811260176585255, "grad_norm": 18.25, "learning_rate": 8.697986753666305e-06, "loss": 1.9162, "step": 2251 }, { "epoch": 0.2582272675151932, "grad_norm": 18.5, "learning_rate": 8.696736677366835e-06, "loss": 2.1631, "step": 2252 }, { "epoch": 0.25834193326453386, "grad_norm": 17.125, "learning_rate": 8.695486091173009e-06, "loss": 1.7182, "step": 2253 }, { "epoch": 0.2584565990138746, "grad_norm": 27.5, "learning_rate": 8.694234995257318e-06, "loss": 1.5468, "step": 2254 }, { "epoch": 0.25857126476321524, "grad_norm": 13.75, "learning_rate": 8.692983389792326e-06, "loss": 1.7742, "step": 2255 }, { "epoch": 0.2586859305125559, "grad_norm": 16.0, "learning_rate": 8.691731274950672e-06, "loss": 1.9366, "step": 2256 }, { "epoch": 0.25880059626189655, "grad_norm": 14.9375, "learning_rate": 8.69047865090506e-06, "loss": 1.5338, "step": 2257 }, { "epoch": 0.25891526201123727, "grad_norm": 11.5625, "learning_rate": 8.689225517828264e-06, "loss": 2.1541, "step": 2258 }, { "epoch": 0.2590299277605779, "grad_norm": 11.625, "learning_rate": 8.687971875893131e-06, "loss": 1.7756, "step": 2259 }, { "epoch": 0.2591445935099186, "grad_norm": 15.0625, "learning_rate": 8.686717725272577e-06, "loss": 1.5459, "step": 2260 }, { "epoch": 0.25925925925925924, "grad_norm": 15.3125, "learning_rate": 8.685463066139587e-06, "loss": 2.0542, "step": 2261 }, { "epoch": 0.2593739250085999, "grad_norm": 15.8125, "learning_rate": 8.68420789866722e-06, "loss": 1.5884, "step": 2262 }, { "epoch": 0.2594885907579406, "grad_norm": 14.5, "learning_rate": 8.6829522230286e-06, "loss": 1.8867, "step": 2263 }, { "epoch": 0.2596032565072813, "grad_norm": 13.5, "learning_rate": 8.681696039396925e-06, "loss": 2.1951, "step": 2264 }, { "epoch": 0.25971792225662194, "grad_norm": 16.75, "learning_rate": 8.68043934794546e-06, "loss": 2.0217, "step": 2265 }, { "epoch": 0.2598325880059626, "grad_norm": 13.8125, "learning_rate": 8.679182148847542e-06, "loss": 1.82, "step": 2266 }, { "epoch": 0.2599472537553033, "grad_norm": 11.75, "learning_rate": 8.677924442276579e-06, "loss": 1.681, "step": 2267 }, { "epoch": 0.26006191950464397, "grad_norm": 12.5, "learning_rate": 8.676666228406047e-06, "loss": 1.4699, "step": 2268 }, { "epoch": 0.2601765852539846, "grad_norm": 13.9375, "learning_rate": 8.675407507409492e-06, "loss": 1.6756, "step": 2269 }, { "epoch": 0.2602912510033253, "grad_norm": 13.9375, "learning_rate": 8.674148279460532e-06, "loss": 1.7939, "step": 2270 }, { "epoch": 0.260405916752666, "grad_norm": 13.25, "learning_rate": 8.672888544732852e-06, "loss": 1.7138, "step": 2271 }, { "epoch": 0.26052058250200666, "grad_norm": 13.875, "learning_rate": 8.671628303400209e-06, "loss": 1.4609, "step": 2272 }, { "epoch": 0.2606352482513473, "grad_norm": 14.0, "learning_rate": 8.670367555636428e-06, "loss": 1.7427, "step": 2273 }, { "epoch": 0.260749914000688, "grad_norm": 26.875, "learning_rate": 8.669106301615406e-06, "loss": 1.9094, "step": 2274 }, { "epoch": 0.2608645797500287, "grad_norm": 15.75, "learning_rate": 8.66784454151111e-06, "loss": 1.7631, "step": 2275 }, { "epoch": 0.26097924549936935, "grad_norm": 13.6875, "learning_rate": 8.666582275497576e-06, "loss": 1.6355, "step": 2276 }, { "epoch": 0.26109391124871, "grad_norm": 14.8125, "learning_rate": 8.665319503748908e-06, "loss": 2.0608, "step": 2277 }, { "epoch": 0.26120857699805067, "grad_norm": 14.0625, "learning_rate": 8.664056226439281e-06, "loss": 1.7756, "step": 2278 }, { "epoch": 0.2613232427473913, "grad_norm": 15.3125, "learning_rate": 8.662792443742942e-06, "loss": 1.8908, "step": 2279 }, { "epoch": 0.26143790849673204, "grad_norm": 13.8125, "learning_rate": 8.661528155834203e-06, "loss": 1.616, "step": 2280 }, { "epoch": 0.2615525742460727, "grad_norm": 14.5625, "learning_rate": 8.66026336288745e-06, "loss": 1.6646, "step": 2281 }, { "epoch": 0.26166723999541336, "grad_norm": 14.4375, "learning_rate": 8.65899806507714e-06, "loss": 2.0915, "step": 2282 }, { "epoch": 0.261781905744754, "grad_norm": 13.3125, "learning_rate": 8.657732262577792e-06, "loss": 1.5806, "step": 2283 }, { "epoch": 0.26189657149409473, "grad_norm": 12.0625, "learning_rate": 8.656465955563999e-06, "loss": 1.836, "step": 2284 }, { "epoch": 0.2620112372434354, "grad_norm": 17.375, "learning_rate": 8.65519914421043e-06, "loss": 2.3605, "step": 2285 }, { "epoch": 0.26212590299277605, "grad_norm": 11.625, "learning_rate": 8.65393182869181e-06, "loss": 1.4939, "step": 2286 }, { "epoch": 0.2622405687421167, "grad_norm": 15.125, "learning_rate": 8.652664009182944e-06, "loss": 1.6926, "step": 2287 }, { "epoch": 0.2623552344914574, "grad_norm": 11.8125, "learning_rate": 8.651395685858708e-06, "loss": 1.7226, "step": 2288 }, { "epoch": 0.2624699002407981, "grad_norm": 16.625, "learning_rate": 8.650126858894036e-06, "loss": 1.7577, "step": 2289 }, { "epoch": 0.26258456599013874, "grad_norm": 14.625, "learning_rate": 8.648857528463942e-06, "loss": 1.6583, "step": 2290 }, { "epoch": 0.2626992317394794, "grad_norm": 14.125, "learning_rate": 8.647587694743507e-06, "loss": 1.6022, "step": 2291 }, { "epoch": 0.2628138974888201, "grad_norm": 13.1875, "learning_rate": 8.646317357907877e-06, "loss": 1.6641, "step": 2292 }, { "epoch": 0.2629285632381608, "grad_norm": 15.5, "learning_rate": 8.645046518132273e-06, "loss": 1.5996, "step": 2293 }, { "epoch": 0.26304322898750143, "grad_norm": 16.75, "learning_rate": 8.643775175591983e-06, "loss": 1.8754, "step": 2294 }, { "epoch": 0.2631578947368421, "grad_norm": 15.0625, "learning_rate": 8.642503330462365e-06, "loss": 2.1475, "step": 2295 }, { "epoch": 0.26327256048618275, "grad_norm": 16.125, "learning_rate": 8.641230982918845e-06, "loss": 1.2242, "step": 2296 }, { "epoch": 0.26338722623552346, "grad_norm": 18.0, "learning_rate": 8.639958133136919e-06, "loss": 1.8027, "step": 2297 }, { "epoch": 0.2635018919848641, "grad_norm": 13.0, "learning_rate": 8.63868478129215e-06, "loss": 1.7502, "step": 2298 }, { "epoch": 0.2636165577342048, "grad_norm": 10.375, "learning_rate": 8.637410927560177e-06, "loss": 1.6527, "step": 2299 }, { "epoch": 0.26373122348354544, "grad_norm": 14.9375, "learning_rate": 8.636136572116702e-06, "loss": 1.6909, "step": 2300 }, { "epoch": 0.26384588923288615, "grad_norm": 15.4375, "learning_rate": 8.634861715137497e-06, "loss": 1.7968, "step": 2301 }, { "epoch": 0.2639605549822268, "grad_norm": 26.875, "learning_rate": 8.633586356798407e-06, "loss": 1.8516, "step": 2302 }, { "epoch": 0.2640752207315675, "grad_norm": 12.25, "learning_rate": 8.63231049727534e-06, "loss": 2.1002, "step": 2303 }, { "epoch": 0.26418988648090813, "grad_norm": 17.875, "learning_rate": 8.631034136744278e-06, "loss": 1.7696, "step": 2304 }, { "epoch": 0.26430455223024885, "grad_norm": 17.75, "learning_rate": 8.629757275381272e-06, "loss": 1.5726, "step": 2305 }, { "epoch": 0.2644192179795895, "grad_norm": 12.6875, "learning_rate": 8.628479913362439e-06, "loss": 2.0546, "step": 2306 }, { "epoch": 0.26453388372893016, "grad_norm": 28.875, "learning_rate": 8.627202050863966e-06, "loss": 1.9934, "step": 2307 }, { "epoch": 0.2646485494782708, "grad_norm": 13.375, "learning_rate": 8.625923688062111e-06, "loss": 1.5976, "step": 2308 }, { "epoch": 0.26476321522761154, "grad_norm": 14.6875, "learning_rate": 8.624644825133201e-06, "loss": 1.9017, "step": 2309 }, { "epoch": 0.2648778809769522, "grad_norm": 13.125, "learning_rate": 8.623365462253629e-06, "loss": 1.8323, "step": 2310 }, { "epoch": 0.26499254672629285, "grad_norm": 17.25, "learning_rate": 8.622085599599857e-06, "loss": 2.1248, "step": 2311 }, { "epoch": 0.2651072124756335, "grad_norm": 15.6875, "learning_rate": 8.620805237348423e-06, "loss": 1.9501, "step": 2312 }, { "epoch": 0.26522187822497423, "grad_norm": 12.375, "learning_rate": 8.619524375675922e-06, "loss": 2.0242, "step": 2313 }, { "epoch": 0.2653365439743149, "grad_norm": 18.125, "learning_rate": 8.618243014759028e-06, "loss": 1.9466, "step": 2314 }, { "epoch": 0.26545120972365555, "grad_norm": 13.625, "learning_rate": 8.616961154774484e-06, "loss": 1.5742, "step": 2315 }, { "epoch": 0.2655658754729962, "grad_norm": 12.625, "learning_rate": 8.615678795899092e-06, "loss": 1.7984, "step": 2316 }, { "epoch": 0.26568054122233686, "grad_norm": 15.125, "learning_rate": 8.614395938309729e-06, "loss": 1.9544, "step": 2317 }, { "epoch": 0.2657952069716776, "grad_norm": 17.25, "learning_rate": 8.613112582183346e-06, "loss": 1.7335, "step": 2318 }, { "epoch": 0.26590987272101824, "grad_norm": 12.75, "learning_rate": 8.611828727696953e-06, "loss": 1.3701, "step": 2319 }, { "epoch": 0.2660245384703589, "grad_norm": 11.875, "learning_rate": 8.610544375027636e-06, "loss": 1.9125, "step": 2320 }, { "epoch": 0.26613920421969955, "grad_norm": 12.1875, "learning_rate": 8.609259524352544e-06, "loss": 1.2165, "step": 2321 }, { "epoch": 0.26625386996904027, "grad_norm": 14.8125, "learning_rate": 8.6079741758489e-06, "loss": 1.3283, "step": 2322 }, { "epoch": 0.2663685357183809, "grad_norm": 14.1875, "learning_rate": 8.606688329693994e-06, "loss": 1.6213, "step": 2323 }, { "epoch": 0.2664832014677216, "grad_norm": 12.375, "learning_rate": 8.605401986065181e-06, "loss": 1.7476, "step": 2324 }, { "epoch": 0.26659786721706225, "grad_norm": 13.375, "learning_rate": 8.604115145139889e-06, "loss": 1.8099, "step": 2325 }, { "epoch": 0.26671253296640296, "grad_norm": 14.1875, "learning_rate": 8.602827807095615e-06, "loss": 1.5986, "step": 2326 }, { "epoch": 0.2668271987157436, "grad_norm": 23.125, "learning_rate": 8.60153997210992e-06, "loss": 1.6088, "step": 2327 }, { "epoch": 0.2669418644650843, "grad_norm": 12.3125, "learning_rate": 8.60025164036044e-06, "loss": 1.591, "step": 2328 }, { "epoch": 0.26705653021442494, "grad_norm": 13.0625, "learning_rate": 8.598962812024868e-06, "loss": 1.6275, "step": 2329 }, { "epoch": 0.26717119596376565, "grad_norm": 13.1875, "learning_rate": 8.597673487280984e-06, "loss": 1.7591, "step": 2330 }, { "epoch": 0.2672858617131063, "grad_norm": 15.25, "learning_rate": 8.596383666306618e-06, "loss": 1.5556, "step": 2331 }, { "epoch": 0.26740052746244697, "grad_norm": 14.0625, "learning_rate": 8.595093349279678e-06, "loss": 1.7532, "step": 2332 }, { "epoch": 0.2675151932117876, "grad_norm": 14.0, "learning_rate": 8.59380253637814e-06, "loss": 1.5925, "step": 2333 }, { "epoch": 0.2676298589611283, "grad_norm": 11.1875, "learning_rate": 8.592511227780044e-06, "loss": 1.3601, "step": 2334 }, { "epoch": 0.267744524710469, "grad_norm": 10.9375, "learning_rate": 8.591219423663505e-06, "loss": 1.8825, "step": 2335 }, { "epoch": 0.26785919045980966, "grad_norm": 17.5, "learning_rate": 8.589927124206703e-06, "loss": 2.0975, "step": 2336 }, { "epoch": 0.2679738562091503, "grad_norm": 11.6875, "learning_rate": 8.588634329587885e-06, "loss": 1.4597, "step": 2337 }, { "epoch": 0.268088521958491, "grad_norm": 30.0, "learning_rate": 8.587341039985365e-06, "loss": 1.6616, "step": 2338 }, { "epoch": 0.2682031877078317, "grad_norm": 14.3125, "learning_rate": 8.586047255577527e-06, "loss": 1.5323, "step": 2339 }, { "epoch": 0.26831785345717235, "grad_norm": 13.3125, "learning_rate": 8.584752976542831e-06, "loss": 1.8292, "step": 2340 }, { "epoch": 0.268432519206513, "grad_norm": 14.0625, "learning_rate": 8.583458203059791e-06, "loss": 1.8209, "step": 2341 }, { "epoch": 0.26854718495585367, "grad_norm": 16.625, "learning_rate": 8.582162935306999e-06, "loss": 1.8795, "step": 2342 }, { "epoch": 0.2686618507051944, "grad_norm": 19.75, "learning_rate": 8.580867173463114e-06, "loss": 1.6595, "step": 2343 }, { "epoch": 0.26877651645453504, "grad_norm": 13.375, "learning_rate": 8.579570917706857e-06, "loss": 1.7303, "step": 2344 }, { "epoch": 0.2688911822038757, "grad_norm": 13.1875, "learning_rate": 8.578274168217026e-06, "loss": 1.9004, "step": 2345 }, { "epoch": 0.26900584795321636, "grad_norm": 14.375, "learning_rate": 8.576976925172482e-06, "loss": 1.8977, "step": 2346 }, { "epoch": 0.2691205137025571, "grad_norm": 15.0625, "learning_rate": 8.575679188752155e-06, "loss": 2.2096, "step": 2347 }, { "epoch": 0.26923517945189773, "grad_norm": 16.875, "learning_rate": 8.574380959135043e-06, "loss": 1.9762, "step": 2348 }, { "epoch": 0.2693498452012384, "grad_norm": 15.0625, "learning_rate": 8.57308223650021e-06, "loss": 2.0823, "step": 2349 }, { "epoch": 0.26946451095057905, "grad_norm": 13.25, "learning_rate": 8.571783021026792e-06, "loss": 1.9963, "step": 2350 }, { "epoch": 0.2695791766999197, "grad_norm": 12.0625, "learning_rate": 8.570483312893992e-06, "loss": 1.6845, "step": 2351 }, { "epoch": 0.2696938424492604, "grad_norm": 29.0, "learning_rate": 8.569183112281076e-06, "loss": 1.6645, "step": 2352 }, { "epoch": 0.2698085081986011, "grad_norm": 12.8125, "learning_rate": 8.567882419367387e-06, "loss": 1.603, "step": 2353 }, { "epoch": 0.26992317394794174, "grad_norm": 12.1875, "learning_rate": 8.566581234332327e-06, "loss": 1.6861, "step": 2354 }, { "epoch": 0.2700378396972824, "grad_norm": 12.375, "learning_rate": 8.56527955735537e-06, "loss": 2.1148, "step": 2355 }, { "epoch": 0.2701525054466231, "grad_norm": 13.625, "learning_rate": 8.56397738861606e-06, "loss": 1.9862, "step": 2356 }, { "epoch": 0.2702671711959638, "grad_norm": 26.375, "learning_rate": 8.562674728294004e-06, "loss": 2.5079, "step": 2357 }, { "epoch": 0.27038183694530443, "grad_norm": 15.875, "learning_rate": 8.561371576568881e-06, "loss": 1.7444, "step": 2358 }, { "epoch": 0.2704965026946451, "grad_norm": 14.1875, "learning_rate": 8.560067933620436e-06, "loss": 1.6376, "step": 2359 }, { "epoch": 0.2706111684439858, "grad_norm": 26.875, "learning_rate": 8.558763799628478e-06, "loss": 1.6936, "step": 2360 }, { "epoch": 0.27072583419332646, "grad_norm": 13.75, "learning_rate": 8.55745917477289e-06, "loss": 1.5367, "step": 2361 }, { "epoch": 0.2708404999426671, "grad_norm": 17.0, "learning_rate": 8.556154059233623e-06, "loss": 1.7874, "step": 2362 }, { "epoch": 0.2709551656920078, "grad_norm": 15.1875, "learning_rate": 8.554848453190687e-06, "loss": 1.6333, "step": 2363 }, { "epoch": 0.2710698314413485, "grad_norm": 16.0, "learning_rate": 8.55354235682417e-06, "loss": 1.8517, "step": 2364 }, { "epoch": 0.27118449719068916, "grad_norm": 16.375, "learning_rate": 8.55223577031422e-06, "loss": 1.9762, "step": 2365 }, { "epoch": 0.2712991629400298, "grad_norm": 17.75, "learning_rate": 8.550928693841058e-06, "loss": 1.4932, "step": 2366 }, { "epoch": 0.2714138286893705, "grad_norm": 12.75, "learning_rate": 8.549621127584972e-06, "loss": 1.7679, "step": 2367 }, { "epoch": 0.27152849443871113, "grad_norm": 12.8125, "learning_rate": 8.548313071726311e-06, "loss": 1.8691, "step": 2368 }, { "epoch": 0.27164316018805185, "grad_norm": 12.75, "learning_rate": 8.547004526445499e-06, "loss": 1.8035, "step": 2369 }, { "epoch": 0.2717578259373925, "grad_norm": 18.25, "learning_rate": 8.545695491923025e-06, "loss": 1.6452, "step": 2370 }, { "epoch": 0.27187249168673316, "grad_norm": 20.75, "learning_rate": 8.544385968339444e-06, "loss": 1.6578, "step": 2371 }, { "epoch": 0.2719871574360738, "grad_norm": 32.25, "learning_rate": 8.543075955875381e-06, "loss": 1.7891, "step": 2372 }, { "epoch": 0.27210182318541454, "grad_norm": 14.0, "learning_rate": 8.541765454711528e-06, "loss": 1.4716, "step": 2373 }, { "epoch": 0.2722164889347552, "grad_norm": 13.5, "learning_rate": 8.540454465028643e-06, "loss": 2.0825, "step": 2374 }, { "epoch": 0.27233115468409586, "grad_norm": 11.9375, "learning_rate": 8.539142987007552e-06, "loss": 1.6481, "step": 2375 }, { "epoch": 0.2724458204334365, "grad_norm": 13.875, "learning_rate": 8.537831020829146e-06, "loss": 1.5723, "step": 2376 }, { "epoch": 0.27256048618277723, "grad_norm": 13.4375, "learning_rate": 8.536518566674389e-06, "loss": 1.5021, "step": 2377 }, { "epoch": 0.2726751519321179, "grad_norm": 12.625, "learning_rate": 8.53520562472431e-06, "loss": 1.7985, "step": 2378 }, { "epoch": 0.27278981768145855, "grad_norm": 14.6875, "learning_rate": 8.533892195159999e-06, "loss": 1.944, "step": 2379 }, { "epoch": 0.2729044834307992, "grad_norm": 13.5, "learning_rate": 8.532578278162625e-06, "loss": 1.9445, "step": 2380 }, { "epoch": 0.2730191491801399, "grad_norm": 15.0, "learning_rate": 8.531263873913412e-06, "loss": 1.7444, "step": 2381 }, { "epoch": 0.2731338149294806, "grad_norm": 14.0, "learning_rate": 8.52994898259366e-06, "loss": 1.5042, "step": 2382 }, { "epoch": 0.27324848067882124, "grad_norm": 16.125, "learning_rate": 8.528633604384734e-06, "loss": 1.7385, "step": 2383 }, { "epoch": 0.2733631464281619, "grad_norm": 14.375, "learning_rate": 8.527317739468063e-06, "loss": 1.8761, "step": 2384 }, { "epoch": 0.27347781217750256, "grad_norm": 11.4375, "learning_rate": 8.526001388025146e-06, "loss": 1.7006, "step": 2385 }, { "epoch": 0.27359247792684327, "grad_norm": 13.1875, "learning_rate": 8.52468455023755e-06, "loss": 1.8648, "step": 2386 }, { "epoch": 0.27370714367618393, "grad_norm": 15.9375, "learning_rate": 8.523367226286907e-06, "loss": 1.9915, "step": 2387 }, { "epoch": 0.2738218094255246, "grad_norm": 13.125, "learning_rate": 8.522049416354917e-06, "loss": 1.5868, "step": 2388 }, { "epoch": 0.27393647517486525, "grad_norm": 16.875, "learning_rate": 8.520731120623344e-06, "loss": 1.8821, "step": 2389 }, { "epoch": 0.27405114092420596, "grad_norm": 28.375, "learning_rate": 8.519412339274027e-06, "loss": 1.82, "step": 2390 }, { "epoch": 0.2741658066735466, "grad_norm": 13.1875, "learning_rate": 8.518093072488863e-06, "loss": 1.9288, "step": 2391 }, { "epoch": 0.2742804724228873, "grad_norm": 13.25, "learning_rate": 8.51677332044982e-06, "loss": 1.8059, "step": 2392 }, { "epoch": 0.27439513817222794, "grad_norm": 14.375, "learning_rate": 8.515453083338935e-06, "loss": 1.8939, "step": 2393 }, { "epoch": 0.27450980392156865, "grad_norm": 15.5625, "learning_rate": 8.514132361338307e-06, "loss": 1.861, "step": 2394 }, { "epoch": 0.2746244696709093, "grad_norm": 11.625, "learning_rate": 8.512811154630106e-06, "loss": 1.7205, "step": 2395 }, { "epoch": 0.27473913542024997, "grad_norm": 14.25, "learning_rate": 8.511489463396563e-06, "loss": 1.9337, "step": 2396 }, { "epoch": 0.27485380116959063, "grad_norm": 13.3125, "learning_rate": 8.510167287819986e-06, "loss": 1.7777, "step": 2397 }, { "epoch": 0.27496846691893134, "grad_norm": 14.5625, "learning_rate": 8.508844628082743e-06, "loss": 1.975, "step": 2398 }, { "epoch": 0.275083132668272, "grad_norm": 13.9375, "learning_rate": 8.507521484367267e-06, "loss": 1.5575, "step": 2399 }, { "epoch": 0.27519779841761266, "grad_norm": 17.5, "learning_rate": 8.50619785685606e-06, "loss": 1.9552, "step": 2400 }, { "epoch": 0.2753124641669533, "grad_norm": 14.25, "learning_rate": 8.504873745731695e-06, "loss": 1.3765, "step": 2401 }, { "epoch": 0.275427129916294, "grad_norm": 17.5, "learning_rate": 8.503549151176804e-06, "loss": 1.5761, "step": 2402 }, { "epoch": 0.2755417956656347, "grad_norm": 14.625, "learning_rate": 8.502224073374092e-06, "loss": 1.4025, "step": 2403 }, { "epoch": 0.27565646141497535, "grad_norm": 11.9375, "learning_rate": 8.500898512506327e-06, "loss": 1.5061, "step": 2404 }, { "epoch": 0.275771127164316, "grad_norm": 19.125, "learning_rate": 8.499572468756347e-06, "loss": 1.7838, "step": 2405 }, { "epoch": 0.27588579291365667, "grad_norm": 15.375, "learning_rate": 8.498245942307053e-06, "loss": 1.8001, "step": 2406 }, { "epoch": 0.2760004586629974, "grad_norm": 16.75, "learning_rate": 8.496918933341414e-06, "loss": 1.4561, "step": 2407 }, { "epoch": 0.27611512441233804, "grad_norm": 14.4375, "learning_rate": 8.495591442042464e-06, "loss": 1.9605, "step": 2408 }, { "epoch": 0.2762297901616787, "grad_norm": 23.625, "learning_rate": 8.494263468593307e-06, "loss": 1.9478, "step": 2409 }, { "epoch": 0.27634445591101936, "grad_norm": 14.4375, "learning_rate": 8.49293501317711e-06, "loss": 1.5872, "step": 2410 }, { "epoch": 0.2764591216603601, "grad_norm": 13.875, "learning_rate": 8.491606075977112e-06, "loss": 1.7436, "step": 2411 }, { "epoch": 0.27657378740970073, "grad_norm": 13.125, "learning_rate": 8.49027665717661e-06, "loss": 1.7928, "step": 2412 }, { "epoch": 0.2766884531590414, "grad_norm": 11.9375, "learning_rate": 8.488946756958974e-06, "loss": 1.8963, "step": 2413 }, { "epoch": 0.27680311890838205, "grad_norm": 12.0, "learning_rate": 8.48761637550764e-06, "loss": 1.7796, "step": 2414 }, { "epoch": 0.27691778465772277, "grad_norm": 14.0, "learning_rate": 8.486285513006105e-06, "loss": 1.6874, "step": 2415 }, { "epoch": 0.2770324504070634, "grad_norm": 13.5, "learning_rate": 8.484954169637938e-06, "loss": 1.5541, "step": 2416 }, { "epoch": 0.2771471161564041, "grad_norm": 15.75, "learning_rate": 8.483622345586774e-06, "loss": 1.5761, "step": 2417 }, { "epoch": 0.27726178190574474, "grad_norm": 13.1875, "learning_rate": 8.48229004103631e-06, "loss": 1.3186, "step": 2418 }, { "epoch": 0.2773764476550854, "grad_norm": 13.375, "learning_rate": 8.480957256170315e-06, "loss": 1.6161, "step": 2419 }, { "epoch": 0.2774911134044261, "grad_norm": 14.1875, "learning_rate": 8.479623991172619e-06, "loss": 1.7649, "step": 2420 }, { "epoch": 0.2776057791537668, "grad_norm": 33.0, "learning_rate": 8.47829024622712e-06, "loss": 1.723, "step": 2421 }, { "epoch": 0.27772044490310743, "grad_norm": 14.5, "learning_rate": 8.476956021517784e-06, "loss": 1.4308, "step": 2422 }, { "epoch": 0.2778351106524481, "grad_norm": 18.125, "learning_rate": 8.475621317228641e-06, "loss": 2.0722, "step": 2423 }, { "epoch": 0.2779497764017888, "grad_norm": 26.375, "learning_rate": 8.474286133543792e-06, "loss": 1.7381, "step": 2424 }, { "epoch": 0.27806444215112947, "grad_norm": 12.9375, "learning_rate": 8.472950470647394e-06, "loss": 1.6478, "step": 2425 }, { "epoch": 0.2781791079004701, "grad_norm": 12.25, "learning_rate": 8.471614328723678e-06, "loss": 1.7919, "step": 2426 }, { "epoch": 0.2782937736498108, "grad_norm": 14.375, "learning_rate": 8.470277707956944e-06, "loss": 1.7323, "step": 2427 }, { "epoch": 0.2784084393991515, "grad_norm": 13.25, "learning_rate": 8.468940608531546e-06, "loss": 1.7826, "step": 2428 }, { "epoch": 0.27852310514849216, "grad_norm": 13.125, "learning_rate": 8.467603030631916e-06, "loss": 1.7845, "step": 2429 }, { "epoch": 0.2786377708978328, "grad_norm": 16.75, "learning_rate": 8.46626497444255e-06, "loss": 1.672, "step": 2430 }, { "epoch": 0.2787524366471735, "grad_norm": 14.3125, "learning_rate": 8.464926440148e-06, "loss": 1.6397, "step": 2431 }, { "epoch": 0.2788671023965142, "grad_norm": 14.75, "learning_rate": 8.463587427932896e-06, "loss": 2.1596, "step": 2432 }, { "epoch": 0.27898176814585485, "grad_norm": 15.125, "learning_rate": 8.462247937981928e-06, "loss": 2.0684, "step": 2433 }, { "epoch": 0.2790964338951955, "grad_norm": 14.5625, "learning_rate": 8.460907970479854e-06, "loss": 1.6081, "step": 2434 }, { "epoch": 0.27921109964453616, "grad_norm": 16.625, "learning_rate": 8.459567525611497e-06, "loss": 1.8995, "step": 2435 }, { "epoch": 0.2793257653938768, "grad_norm": 11.75, "learning_rate": 8.458226603561744e-06, "loss": 1.4105, "step": 2436 }, { "epoch": 0.27944043114321754, "grad_norm": 17.0, "learning_rate": 8.45688520451555e-06, "loss": 1.8393, "step": 2437 }, { "epoch": 0.2795550968925582, "grad_norm": 17.75, "learning_rate": 8.455543328657938e-06, "loss": 2.0116, "step": 2438 }, { "epoch": 0.27966976264189886, "grad_norm": 12.375, "learning_rate": 8.454200976173992e-06, "loss": 1.4551, "step": 2439 }, { "epoch": 0.2797844283912395, "grad_norm": 15.25, "learning_rate": 8.452858147248863e-06, "loss": 1.7804, "step": 2440 }, { "epoch": 0.27989909414058023, "grad_norm": 16.25, "learning_rate": 8.451514842067773e-06, "loss": 1.8943, "step": 2441 }, { "epoch": 0.2800137598899209, "grad_norm": 13.5625, "learning_rate": 8.450171060816e-06, "loss": 1.5395, "step": 2442 }, { "epoch": 0.28012842563926155, "grad_norm": 21.5, "learning_rate": 8.448826803678896e-06, "loss": 1.7314, "step": 2443 }, { "epoch": 0.2802430913886022, "grad_norm": 12.875, "learning_rate": 8.447482070841874e-06, "loss": 1.9094, "step": 2444 }, { "epoch": 0.2803577571379429, "grad_norm": 17.5, "learning_rate": 8.446136862490416e-06, "loss": 1.7447, "step": 2445 }, { "epoch": 0.2804724228872836, "grad_norm": 12.75, "learning_rate": 8.44479117881007e-06, "loss": 1.5967, "step": 2446 }, { "epoch": 0.28058708863662424, "grad_norm": 14.3125, "learning_rate": 8.443445019986441e-06, "loss": 1.5388, "step": 2447 }, { "epoch": 0.2807017543859649, "grad_norm": 17.5, "learning_rate": 8.44209838620521e-06, "loss": 1.8288, "step": 2448 }, { "epoch": 0.2808164201353056, "grad_norm": 14.0, "learning_rate": 8.440751277652121e-06, "loss": 1.6803, "step": 2449 }, { "epoch": 0.28093108588464627, "grad_norm": 11.75, "learning_rate": 8.439403694512978e-06, "loss": 1.9224, "step": 2450 }, { "epoch": 0.28104575163398693, "grad_norm": 19.25, "learning_rate": 8.438055636973657e-06, "loss": 2.1345, "step": 2451 }, { "epoch": 0.2811604173833276, "grad_norm": 12.1875, "learning_rate": 8.436707105220097e-06, "loss": 1.6257, "step": 2452 }, { "epoch": 0.28127508313266825, "grad_norm": 15.0625, "learning_rate": 8.4353580994383e-06, "loss": 1.9988, "step": 2453 }, { "epoch": 0.28138974888200896, "grad_norm": 14.375, "learning_rate": 8.434008619814338e-06, "loss": 1.9441, "step": 2454 }, { "epoch": 0.2815044146313496, "grad_norm": 12.5, "learning_rate": 8.432658666534346e-06, "loss": 1.6852, "step": 2455 }, { "epoch": 0.2816190803806903, "grad_norm": 15.625, "learning_rate": 8.431308239784521e-06, "loss": 2.0338, "step": 2456 }, { "epoch": 0.28173374613003094, "grad_norm": 15.1875, "learning_rate": 8.429957339751132e-06, "loss": 2.0282, "step": 2457 }, { "epoch": 0.28184841187937165, "grad_norm": 16.25, "learning_rate": 8.428605966620509e-06, "loss": 1.7925, "step": 2458 }, { "epoch": 0.2819630776287123, "grad_norm": 93.5, "learning_rate": 8.427254120579049e-06, "loss": 2.2267, "step": 2459 }, { "epoch": 0.28207774337805297, "grad_norm": 15.3125, "learning_rate": 8.425901801813212e-06, "loss": 1.7156, "step": 2460 }, { "epoch": 0.28219240912739363, "grad_norm": 13.3125, "learning_rate": 8.424549010509523e-06, "loss": 1.9732, "step": 2461 }, { "epoch": 0.28230707487673434, "grad_norm": 12.9375, "learning_rate": 8.423195746854578e-06, "loss": 1.6979, "step": 2462 }, { "epoch": 0.282421740626075, "grad_norm": 15.125, "learning_rate": 8.421842011035031e-06, "loss": 1.9821, "step": 2463 }, { "epoch": 0.28253640637541566, "grad_norm": 14.1875, "learning_rate": 8.420487803237605e-06, "loss": 1.5679, "step": 2464 }, { "epoch": 0.2826510721247563, "grad_norm": 15.8125, "learning_rate": 8.419133123649088e-06, "loss": 1.6087, "step": 2465 }, { "epoch": 0.28276573787409703, "grad_norm": 11.6875, "learning_rate": 8.417777972456329e-06, "loss": 1.3367, "step": 2466 }, { "epoch": 0.2828804036234377, "grad_norm": 18.5, "learning_rate": 8.416422349846249e-06, "loss": 1.8548, "step": 2467 }, { "epoch": 0.28299506937277835, "grad_norm": 15.1875, "learning_rate": 8.415066256005829e-06, "loss": 1.6569, "step": 2468 }, { "epoch": 0.283109735122119, "grad_norm": 18.375, "learning_rate": 8.413709691122117e-06, "loss": 1.6614, "step": 2469 }, { "epoch": 0.28322440087145967, "grad_norm": 13.5, "learning_rate": 8.412352655382222e-06, "loss": 1.962, "step": 2470 }, { "epoch": 0.2833390666208004, "grad_norm": 15.5625, "learning_rate": 8.410995148973325e-06, "loss": 1.9921, "step": 2471 }, { "epoch": 0.28345373237014104, "grad_norm": 20.875, "learning_rate": 8.409637172082666e-06, "loss": 1.6763, "step": 2472 }, { "epoch": 0.2835683981194817, "grad_norm": 12.875, "learning_rate": 8.408278724897552e-06, "loss": 1.4707, "step": 2473 }, { "epoch": 0.28368306386882236, "grad_norm": 14.8125, "learning_rate": 8.406919807605355e-06, "loss": 1.8506, "step": 2474 }, { "epoch": 0.2837977296181631, "grad_norm": 14.125, "learning_rate": 8.405560420393516e-06, "loss": 1.9344, "step": 2475 }, { "epoch": 0.28391239536750373, "grad_norm": 15.875, "learning_rate": 8.40420056344953e-06, "loss": 1.7771, "step": 2476 }, { "epoch": 0.2840270611168444, "grad_norm": 21.375, "learning_rate": 8.402840236960967e-06, "loss": 1.7356, "step": 2477 }, { "epoch": 0.28414172686618505, "grad_norm": 14.1875, "learning_rate": 8.401479441115457e-06, "loss": 1.6877, "step": 2478 }, { "epoch": 0.28425639261552577, "grad_norm": 14.3125, "learning_rate": 8.400118176100698e-06, "loss": 1.67, "step": 2479 }, { "epoch": 0.2843710583648664, "grad_norm": 12.5, "learning_rate": 8.398756442104447e-06, "loss": 1.5292, "step": 2480 }, { "epoch": 0.2844857241142071, "grad_norm": 14.75, "learning_rate": 8.39739423931453e-06, "loss": 1.9464, "step": 2481 }, { "epoch": 0.28460038986354774, "grad_norm": 14.0625, "learning_rate": 8.39603156791884e-06, "loss": 2.0164, "step": 2482 }, { "epoch": 0.28471505561288846, "grad_norm": 15.5625, "learning_rate": 8.394668428105327e-06, "loss": 1.6427, "step": 2483 }, { "epoch": 0.2848297213622291, "grad_norm": 14.3125, "learning_rate": 8.393304820062016e-06, "loss": 1.8458, "step": 2484 }, { "epoch": 0.2849443871115698, "grad_norm": 15.1875, "learning_rate": 8.391940743976986e-06, "loss": 1.7427, "step": 2485 }, { "epoch": 0.28505905286091043, "grad_norm": 12.4375, "learning_rate": 8.390576200038386e-06, "loss": 1.8659, "step": 2486 }, { "epoch": 0.2851737186102511, "grad_norm": 18.875, "learning_rate": 8.38921118843443e-06, "loss": 1.9638, "step": 2487 }, { "epoch": 0.2852883843595918, "grad_norm": 13.625, "learning_rate": 8.387845709353392e-06, "loss": 1.7775, "step": 2488 }, { "epoch": 0.28540305010893247, "grad_norm": 15.0625, "learning_rate": 8.38647976298362e-06, "loss": 1.7787, "step": 2489 }, { "epoch": 0.2855177158582731, "grad_norm": 36.5, "learning_rate": 8.385113349513517e-06, "loss": 1.8471, "step": 2490 }, { "epoch": 0.2856323816076138, "grad_norm": 16.375, "learning_rate": 8.383746469131552e-06, "loss": 1.8642, "step": 2491 }, { "epoch": 0.2857470473569545, "grad_norm": 12.6875, "learning_rate": 8.382379122026262e-06, "loss": 2.0245, "step": 2492 }, { "epoch": 0.28586171310629516, "grad_norm": 14.9375, "learning_rate": 8.381011308386246e-06, "loss": 1.6125, "step": 2493 }, { "epoch": 0.2859763788556358, "grad_norm": 11.5, "learning_rate": 8.379643028400169e-06, "loss": 1.5793, "step": 2494 }, { "epoch": 0.2860910446049765, "grad_norm": 12.5625, "learning_rate": 8.378274282256758e-06, "loss": 2.0011, "step": 2495 }, { "epoch": 0.2862057103543172, "grad_norm": 14.25, "learning_rate": 8.376905070144804e-06, "loss": 1.7102, "step": 2496 }, { "epoch": 0.28632037610365785, "grad_norm": 17.375, "learning_rate": 8.375535392253167e-06, "loss": 2.0294, "step": 2497 }, { "epoch": 0.2864350418529985, "grad_norm": 12.9375, "learning_rate": 8.374165248770764e-06, "loss": 1.9228, "step": 2498 }, { "epoch": 0.28654970760233917, "grad_norm": 13.4375, "learning_rate": 8.372794639886584e-06, "loss": 1.9564, "step": 2499 }, { "epoch": 0.2866643733516799, "grad_norm": 15.5, "learning_rate": 8.371423565789674e-06, "loss": 1.4938, "step": 2500 }, { "epoch": 0.28677903910102054, "grad_norm": 19.625, "learning_rate": 8.37005202666915e-06, "loss": 1.9131, "step": 2501 }, { "epoch": 0.2868937048503612, "grad_norm": 13.8125, "learning_rate": 8.368680022714187e-06, "loss": 1.595, "step": 2502 }, { "epoch": 0.28700837059970186, "grad_norm": 23.5, "learning_rate": 8.367307554114025e-06, "loss": 1.7699, "step": 2503 }, { "epoch": 0.2871230363490425, "grad_norm": 13.375, "learning_rate": 8.365934621057976e-06, "loss": 1.9449, "step": 2504 }, { "epoch": 0.28723770209838323, "grad_norm": 14.4375, "learning_rate": 8.364561223735406e-06, "loss": 1.7332, "step": 2505 }, { "epoch": 0.2873523678477239, "grad_norm": 14.75, "learning_rate": 8.363187362335749e-06, "loss": 2.0139, "step": 2506 }, { "epoch": 0.28746703359706455, "grad_norm": 13.8125, "learning_rate": 8.361813037048504e-06, "loss": 1.474, "step": 2507 }, { "epoch": 0.2875816993464052, "grad_norm": 17.875, "learning_rate": 8.360438248063232e-06, "loss": 1.929, "step": 2508 }, { "epoch": 0.2876963650957459, "grad_norm": 15.25, "learning_rate": 8.35906299556956e-06, "loss": 1.9273, "step": 2509 }, { "epoch": 0.2878110308450866, "grad_norm": 34.5, "learning_rate": 8.357687279757177e-06, "loss": 2.0388, "step": 2510 }, { "epoch": 0.28792569659442724, "grad_norm": 18.5, "learning_rate": 8.356311100815837e-06, "loss": 1.6571, "step": 2511 }, { "epoch": 0.2880403623437679, "grad_norm": 13.6875, "learning_rate": 8.354934458935358e-06, "loss": 1.9047, "step": 2512 }, { "epoch": 0.2881550280931086, "grad_norm": 12.1875, "learning_rate": 8.353557354305621e-06, "loss": 1.9124, "step": 2513 }, { "epoch": 0.28826969384244927, "grad_norm": 15.0, "learning_rate": 8.352179787116573e-06, "loss": 2.1758, "step": 2514 }, { "epoch": 0.28838435959178993, "grad_norm": 13.9375, "learning_rate": 8.35080175755822e-06, "loss": 1.6022, "step": 2515 }, { "epoch": 0.2884990253411306, "grad_norm": 15.9375, "learning_rate": 8.349423265820636e-06, "loss": 1.7723, "step": 2516 }, { "epoch": 0.2886136910904713, "grad_norm": 12.9375, "learning_rate": 8.348044312093959e-06, "loss": 1.3611, "step": 2517 }, { "epoch": 0.28872835683981196, "grad_norm": 12.625, "learning_rate": 8.34666489656839e-06, "loss": 1.4869, "step": 2518 }, { "epoch": 0.2888430225891526, "grad_norm": 28.625, "learning_rate": 8.34528501943419e-06, "loss": 1.8004, "step": 2519 }, { "epoch": 0.2889576883384933, "grad_norm": 17.5, "learning_rate": 8.343904680881691e-06, "loss": 1.6077, "step": 2520 }, { "epoch": 0.28907235408783394, "grad_norm": 14.375, "learning_rate": 8.342523881101278e-06, "loss": 1.7751, "step": 2521 }, { "epoch": 0.28918701983717465, "grad_norm": 13.3125, "learning_rate": 8.341142620283413e-06, "loss": 1.5334, "step": 2522 }, { "epoch": 0.2893016855865153, "grad_norm": 15.1875, "learning_rate": 8.33976089861861e-06, "loss": 2.4715, "step": 2523 }, { "epoch": 0.28941635133585597, "grad_norm": 14.75, "learning_rate": 8.338378716297455e-06, "loss": 1.5401, "step": 2524 }, { "epoch": 0.28953101708519663, "grad_norm": 13.6875, "learning_rate": 8.336996073510589e-06, "loss": 1.7057, "step": 2525 }, { "epoch": 0.28964568283453734, "grad_norm": 13.375, "learning_rate": 8.335612970448724e-06, "loss": 1.692, "step": 2526 }, { "epoch": 0.289760348583878, "grad_norm": 15.0, "learning_rate": 8.334229407302633e-06, "loss": 1.8337, "step": 2527 }, { "epoch": 0.28987501433321866, "grad_norm": 14.8125, "learning_rate": 8.332845384263151e-06, "loss": 2.0041, "step": 2528 }, { "epoch": 0.2899896800825593, "grad_norm": 16.375, "learning_rate": 8.331460901521178e-06, "loss": 1.7749, "step": 2529 }, { "epoch": 0.29010434583190003, "grad_norm": 12.9375, "learning_rate": 8.330075959267677e-06, "loss": 1.9053, "step": 2530 }, { "epoch": 0.2902190115812407, "grad_norm": 16.25, "learning_rate": 8.328690557693675e-06, "loss": 2.1105, "step": 2531 }, { "epoch": 0.29033367733058135, "grad_norm": 13.4375, "learning_rate": 8.327304696990261e-06, "loss": 1.6346, "step": 2532 }, { "epoch": 0.290448343079922, "grad_norm": 12.4375, "learning_rate": 8.325918377348587e-06, "loss": 1.8064, "step": 2533 }, { "epoch": 0.2905630088292627, "grad_norm": 30.375, "learning_rate": 8.324531598959871e-06, "loss": 1.474, "step": 2534 }, { "epoch": 0.2906776745786034, "grad_norm": 18.125, "learning_rate": 8.323144362015394e-06, "loss": 2.2366, "step": 2535 }, { "epoch": 0.29079234032794404, "grad_norm": 17.125, "learning_rate": 8.321756666706496e-06, "loss": 2.4086, "step": 2536 }, { "epoch": 0.2909070060772847, "grad_norm": 15.8125, "learning_rate": 8.320368513224584e-06, "loss": 1.6681, "step": 2537 }, { "epoch": 0.29102167182662536, "grad_norm": 14.4375, "learning_rate": 8.318979901761129e-06, "loss": 1.6313, "step": 2538 }, { "epoch": 0.2911363375759661, "grad_norm": 16.375, "learning_rate": 8.317590832507661e-06, "loss": 1.7739, "step": 2539 }, { "epoch": 0.29125100332530673, "grad_norm": 15.75, "learning_rate": 8.316201305655776e-06, "loss": 1.9668, "step": 2540 }, { "epoch": 0.2913656690746474, "grad_norm": 30.5, "learning_rate": 8.314811321397133e-06, "loss": 1.6808, "step": 2541 }, { "epoch": 0.29148033482398805, "grad_norm": 17.875, "learning_rate": 8.313420879923456e-06, "loss": 1.5776, "step": 2542 }, { "epoch": 0.29159500057332877, "grad_norm": 12.0625, "learning_rate": 8.312029981426529e-06, "loss": 1.7276, "step": 2543 }, { "epoch": 0.2917096663226694, "grad_norm": 14.9375, "learning_rate": 8.310638626098196e-06, "loss": 1.8279, "step": 2544 }, { "epoch": 0.2918243320720101, "grad_norm": 15.1875, "learning_rate": 8.309246814130373e-06, "loss": 1.6306, "step": 2545 }, { "epoch": 0.29193899782135074, "grad_norm": 16.125, "learning_rate": 8.307854545715032e-06, "loss": 1.9128, "step": 2546 }, { "epoch": 0.29205366357069146, "grad_norm": 14.9375, "learning_rate": 8.306461821044209e-06, "loss": 2.0847, "step": 2547 }, { "epoch": 0.2921683293200321, "grad_norm": 16.375, "learning_rate": 8.305068640310005e-06, "loss": 1.8273, "step": 2548 }, { "epoch": 0.2922829950693728, "grad_norm": 14.25, "learning_rate": 8.303675003704583e-06, "loss": 1.9941, "step": 2549 }, { "epoch": 0.29239766081871343, "grad_norm": 16.25, "learning_rate": 8.302280911420167e-06, "loss": 2.3209, "step": 2550 }, { "epoch": 0.29251232656805415, "grad_norm": 11.375, "learning_rate": 8.300886363649048e-06, "loss": 1.9598, "step": 2551 }, { "epoch": 0.2926269923173948, "grad_norm": 14.5625, "learning_rate": 8.299491360583573e-06, "loss": 2.0382, "step": 2552 }, { "epoch": 0.29274165806673547, "grad_norm": 15.625, "learning_rate": 8.298095902416162e-06, "loss": 1.6527, "step": 2553 }, { "epoch": 0.2928563238160761, "grad_norm": 13.3125, "learning_rate": 8.296699989339287e-06, "loss": 1.9126, "step": 2554 }, { "epoch": 0.2929709895654168, "grad_norm": 20.75, "learning_rate": 8.29530362154549e-06, "loss": 1.8128, "step": 2555 }, { "epoch": 0.2930856553147575, "grad_norm": 12.5625, "learning_rate": 8.293906799227372e-06, "loss": 1.5801, "step": 2556 }, { "epoch": 0.29320032106409816, "grad_norm": 13.1875, "learning_rate": 8.292509522577599e-06, "loss": 1.8628, "step": 2557 }, { "epoch": 0.2933149868134388, "grad_norm": 16.0, "learning_rate": 8.291111791788897e-06, "loss": 1.7384, "step": 2558 }, { "epoch": 0.2934296525627795, "grad_norm": 11.8125, "learning_rate": 8.28971360705406e-06, "loss": 2.0675, "step": 2559 }, { "epoch": 0.2935443183121202, "grad_norm": 19.375, "learning_rate": 8.288314968565937e-06, "loss": 2.2299, "step": 2560 }, { "epoch": 0.29365898406146085, "grad_norm": 14.625, "learning_rate": 8.286915876517445e-06, "loss": 1.7798, "step": 2561 }, { "epoch": 0.2937736498108015, "grad_norm": 14.6875, "learning_rate": 8.285516331101563e-06, "loss": 1.9299, "step": 2562 }, { "epoch": 0.29388831556014217, "grad_norm": 12.8125, "learning_rate": 8.28411633251133e-06, "loss": 1.5007, "step": 2563 }, { "epoch": 0.2940029813094829, "grad_norm": 13.4375, "learning_rate": 8.282715880939851e-06, "loss": 1.8545, "step": 2564 }, { "epoch": 0.29411764705882354, "grad_norm": 13.1875, "learning_rate": 8.281314976580289e-06, "loss": 1.5143, "step": 2565 }, { "epoch": 0.2942323128081642, "grad_norm": 13.625, "learning_rate": 8.279913619625874e-06, "loss": 1.8858, "step": 2566 }, { "epoch": 0.29434697855750486, "grad_norm": 14.875, "learning_rate": 8.278511810269897e-06, "loss": 1.6854, "step": 2567 }, { "epoch": 0.29446164430684557, "grad_norm": 13.8125, "learning_rate": 8.277109548705709e-06, "loss": 1.6783, "step": 2568 }, { "epoch": 0.29457631005618623, "grad_norm": 15.375, "learning_rate": 8.275706835126726e-06, "loss": 2.1286, "step": 2569 }, { "epoch": 0.2946909758055269, "grad_norm": 13.4375, "learning_rate": 8.274303669726427e-06, "loss": 1.7464, "step": 2570 }, { "epoch": 0.29480564155486755, "grad_norm": 17.5, "learning_rate": 8.27290005269835e-06, "loss": 1.8662, "step": 2571 }, { "epoch": 0.2949203073042082, "grad_norm": 12.6875, "learning_rate": 8.271495984236096e-06, "loss": 1.7258, "step": 2572 }, { "epoch": 0.2950349730535489, "grad_norm": 12.875, "learning_rate": 8.270091464533334e-06, "loss": 1.6126, "step": 2573 }, { "epoch": 0.2951496388028896, "grad_norm": 14.0625, "learning_rate": 8.268686493783787e-06, "loss": 1.6017, "step": 2574 }, { "epoch": 0.29526430455223024, "grad_norm": 14.5, "learning_rate": 8.267281072181245e-06, "loss": 1.5577, "step": 2575 }, { "epoch": 0.2953789703015709, "grad_norm": 13.75, "learning_rate": 8.265875199919559e-06, "loss": 1.772, "step": 2576 }, { "epoch": 0.2954936360509116, "grad_norm": 16.0, "learning_rate": 8.264468877192642e-06, "loss": 1.7474, "step": 2577 }, { "epoch": 0.29560830180025227, "grad_norm": 11.25, "learning_rate": 8.26306210419447e-06, "loss": 1.6075, "step": 2578 }, { "epoch": 0.29572296754959293, "grad_norm": 13.125, "learning_rate": 8.261654881119081e-06, "loss": 1.7727, "step": 2579 }, { "epoch": 0.2958376332989336, "grad_norm": 13.8125, "learning_rate": 8.260247208160574e-06, "loss": 1.3857, "step": 2580 }, { "epoch": 0.2959522990482743, "grad_norm": 13.1875, "learning_rate": 8.25883908551311e-06, "loss": 1.8865, "step": 2581 }, { "epoch": 0.29606696479761496, "grad_norm": 18.375, "learning_rate": 8.257430513370915e-06, "loss": 1.9711, "step": 2582 }, { "epoch": 0.2961816305469556, "grad_norm": 12.8125, "learning_rate": 8.25602149192827e-06, "loss": 1.3735, "step": 2583 }, { "epoch": 0.2962962962962963, "grad_norm": 14.625, "learning_rate": 8.254612021379527e-06, "loss": 2.0284, "step": 2584 }, { "epoch": 0.296410962045637, "grad_norm": 11.6875, "learning_rate": 8.253202101919095e-06, "loss": 1.7306, "step": 2585 }, { "epoch": 0.29652562779497765, "grad_norm": 16.375, "learning_rate": 8.251791733741443e-06, "loss": 1.7959, "step": 2586 }, { "epoch": 0.2966402935443183, "grad_norm": 14.4375, "learning_rate": 8.250380917041108e-06, "loss": 1.6073, "step": 2587 }, { "epoch": 0.29675495929365897, "grad_norm": 14.0625, "learning_rate": 8.248969652012682e-06, "loss": 1.5354, "step": 2588 }, { "epoch": 0.29686962504299963, "grad_norm": 12.75, "learning_rate": 8.247557938850824e-06, "loss": 1.8636, "step": 2589 }, { "epoch": 0.29698429079234034, "grad_norm": 14.125, "learning_rate": 8.246145777750254e-06, "loss": 1.7468, "step": 2590 }, { "epoch": 0.297098956541681, "grad_norm": 12.625, "learning_rate": 8.244733168905748e-06, "loss": 1.9111, "step": 2591 }, { "epoch": 0.29721362229102166, "grad_norm": 14.375, "learning_rate": 8.243320112512155e-06, "loss": 1.6815, "step": 2592 }, { "epoch": 0.2973282880403623, "grad_norm": 18.875, "learning_rate": 8.241906608764373e-06, "loss": 2.0533, "step": 2593 }, { "epoch": 0.29744295378970304, "grad_norm": 13.0, "learning_rate": 8.240492657857371e-06, "loss": 1.6026, "step": 2594 }, { "epoch": 0.2975576195390437, "grad_norm": 11.5625, "learning_rate": 8.239078259986177e-06, "loss": 1.6387, "step": 2595 }, { "epoch": 0.29767228528838435, "grad_norm": 15.25, "learning_rate": 8.23766341534588e-06, "loss": 2.1664, "step": 2596 }, { "epoch": 0.297786951037725, "grad_norm": 12.25, "learning_rate": 8.236248124131629e-06, "loss": 1.493, "step": 2597 }, { "epoch": 0.2979016167870657, "grad_norm": 13.8125, "learning_rate": 8.234832386538639e-06, "loss": 1.6992, "step": 2598 }, { "epoch": 0.2980162825364064, "grad_norm": 15.0, "learning_rate": 8.233416202762183e-06, "loss": 1.7958, "step": 2599 }, { "epoch": 0.29813094828574704, "grad_norm": 33.25, "learning_rate": 8.231999572997596e-06, "loss": 1.707, "step": 2600 }, { "epoch": 0.2982456140350877, "grad_norm": 14.0, "learning_rate": 8.230582497440274e-06, "loss": 1.9877, "step": 2601 }, { "epoch": 0.2983602797844284, "grad_norm": 11.1875, "learning_rate": 8.229164976285678e-06, "loss": 1.6035, "step": 2602 }, { "epoch": 0.2984749455337691, "grad_norm": 16.0, "learning_rate": 8.227747009729326e-06, "loss": 2.1685, "step": 2603 }, { "epoch": 0.29858961128310973, "grad_norm": 11.75, "learning_rate": 8.226328597966803e-06, "loss": 1.8416, "step": 2604 }, { "epoch": 0.2987042770324504, "grad_norm": 13.625, "learning_rate": 8.224909741193747e-06, "loss": 1.6931, "step": 2605 }, { "epoch": 0.29881894278179105, "grad_norm": 14.3125, "learning_rate": 8.223490439605866e-06, "loss": 1.8113, "step": 2606 }, { "epoch": 0.29893360853113177, "grad_norm": 12.625, "learning_rate": 8.222070693398924e-06, "loss": 1.7091, "step": 2607 }, { "epoch": 0.2990482742804724, "grad_norm": 14.4375, "learning_rate": 8.220650502768748e-06, "loss": 1.6031, "step": 2608 }, { "epoch": 0.2991629400298131, "grad_norm": 18.125, "learning_rate": 8.219229867911225e-06, "loss": 1.8178, "step": 2609 }, { "epoch": 0.29927760577915374, "grad_norm": 14.9375, "learning_rate": 8.217808789022308e-06, "loss": 2.3194, "step": 2610 }, { "epoch": 0.29939227152849446, "grad_norm": 13.0625, "learning_rate": 8.216387266298003e-06, "loss": 1.755, "step": 2611 }, { "epoch": 0.2995069372778351, "grad_norm": 16.75, "learning_rate": 8.214965299934386e-06, "loss": 2.028, "step": 2612 }, { "epoch": 0.2996216030271758, "grad_norm": 15.3125, "learning_rate": 8.21354289012759e-06, "loss": 1.6483, "step": 2613 }, { "epoch": 0.29973626877651643, "grad_norm": 14.25, "learning_rate": 8.212120037073806e-06, "loss": 1.4756, "step": 2614 }, { "epoch": 0.29985093452585715, "grad_norm": 16.625, "learning_rate": 8.210696740969292e-06, "loss": 1.4406, "step": 2615 }, { "epoch": 0.2999656002751978, "grad_norm": 17.0, "learning_rate": 8.209273002010364e-06, "loss": 1.5666, "step": 2616 }, { "epoch": 0.30008026602453847, "grad_norm": 13.6875, "learning_rate": 8.207848820393401e-06, "loss": 1.9704, "step": 2617 }, { "epoch": 0.3001949317738791, "grad_norm": 13.8125, "learning_rate": 8.206424196314838e-06, "loss": 1.4698, "step": 2618 }, { "epoch": 0.30030959752321984, "grad_norm": 12.625, "learning_rate": 8.204999129971179e-06, "loss": 1.6966, "step": 2619 }, { "epoch": 0.3004242632725605, "grad_norm": 27.125, "learning_rate": 8.203573621558982e-06, "loss": 1.8217, "step": 2620 }, { "epoch": 0.30053892902190116, "grad_norm": 12.875, "learning_rate": 8.202147671274869e-06, "loss": 1.7007, "step": 2621 }, { "epoch": 0.3006535947712418, "grad_norm": 14.4375, "learning_rate": 8.200721279315525e-06, "loss": 1.9111, "step": 2622 }, { "epoch": 0.3007682605205825, "grad_norm": 14.5, "learning_rate": 8.199294445877691e-06, "loss": 1.4752, "step": 2623 }, { "epoch": 0.3008829262699232, "grad_norm": 17.75, "learning_rate": 8.197867171158172e-06, "loss": 1.9827, "step": 2624 }, { "epoch": 0.30099759201926385, "grad_norm": 12.5625, "learning_rate": 8.196439455353833e-06, "loss": 2.136, "step": 2625 }, { "epoch": 0.3011122577686045, "grad_norm": 17.0, "learning_rate": 8.195011298661602e-06, "loss": 1.6043, "step": 2626 }, { "epoch": 0.30122692351794517, "grad_norm": 14.6875, "learning_rate": 8.193582701278463e-06, "loss": 1.7959, "step": 2627 }, { "epoch": 0.3013415892672859, "grad_norm": 13.375, "learning_rate": 8.192153663401467e-06, "loss": 1.6643, "step": 2628 }, { "epoch": 0.30145625501662654, "grad_norm": 11.5, "learning_rate": 8.190724185227723e-06, "loss": 1.6647, "step": 2629 }, { "epoch": 0.3015709207659672, "grad_norm": 13.5625, "learning_rate": 8.189294266954397e-06, "loss": 2.0066, "step": 2630 }, { "epoch": 0.30168558651530786, "grad_norm": 16.75, "learning_rate": 8.187863908778719e-06, "loss": 1.7997, "step": 2631 }, { "epoch": 0.30180025226464857, "grad_norm": 35.25, "learning_rate": 8.186433110897982e-06, "loss": 2.271, "step": 2632 }, { "epoch": 0.30191491801398923, "grad_norm": 15.0, "learning_rate": 8.185001873509535e-06, "loss": 1.6845, "step": 2633 }, { "epoch": 0.3020295837633299, "grad_norm": 15.0625, "learning_rate": 8.183570196810792e-06, "loss": 1.7091, "step": 2634 }, { "epoch": 0.30214424951267055, "grad_norm": 15.5, "learning_rate": 8.182138080999226e-06, "loss": 1.4161, "step": 2635 }, { "epoch": 0.30225891526201126, "grad_norm": 17.125, "learning_rate": 8.180705526272368e-06, "loss": 1.8558, "step": 2636 }, { "epoch": 0.3023735810113519, "grad_norm": 13.8125, "learning_rate": 8.179272532827812e-06, "loss": 2.0132, "step": 2637 }, { "epoch": 0.3024882467606926, "grad_norm": 12.1875, "learning_rate": 8.177839100863211e-06, "loss": 1.6082, "step": 2638 }, { "epoch": 0.30260291251003324, "grad_norm": 14.25, "learning_rate": 8.176405230576286e-06, "loss": 1.779, "step": 2639 }, { "epoch": 0.3027175782593739, "grad_norm": 14.875, "learning_rate": 8.174970922164804e-06, "loss": 1.6115, "step": 2640 }, { "epoch": 0.3028322440087146, "grad_norm": 13.875, "learning_rate": 8.173536175826604e-06, "loss": 2.0243, "step": 2641 }, { "epoch": 0.30294690975805527, "grad_norm": 13.625, "learning_rate": 8.172100991759581e-06, "loss": 1.9081, "step": 2642 }, { "epoch": 0.30306157550739593, "grad_norm": 16.375, "learning_rate": 8.170665370161693e-06, "loss": 1.4182, "step": 2643 }, { "epoch": 0.3031762412567366, "grad_norm": 16.125, "learning_rate": 8.169229311230954e-06, "loss": 1.5881, "step": 2644 }, { "epoch": 0.3032909070060773, "grad_norm": 20.25, "learning_rate": 8.167792815165443e-06, "loss": 1.7628, "step": 2645 }, { "epoch": 0.30340557275541796, "grad_norm": 15.5, "learning_rate": 8.166355882163297e-06, "loss": 2.4304, "step": 2646 }, { "epoch": 0.3035202385047586, "grad_norm": 20.125, "learning_rate": 8.164918512422715e-06, "loss": 1.6882, "step": 2647 }, { "epoch": 0.3036349042540993, "grad_norm": 13.9375, "learning_rate": 8.16348070614195e-06, "loss": 1.7678, "step": 2648 }, { "epoch": 0.30374957000344, "grad_norm": 13.75, "learning_rate": 8.162042463519326e-06, "loss": 1.9468, "step": 2649 }, { "epoch": 0.30386423575278065, "grad_norm": 13.8125, "learning_rate": 8.160603784753218e-06, "loss": 1.7509, "step": 2650 }, { "epoch": 0.3039789015021213, "grad_norm": 12.8125, "learning_rate": 8.159164670042062e-06, "loss": 1.8452, "step": 2651 }, { "epoch": 0.30409356725146197, "grad_norm": 11.5, "learning_rate": 8.15772511958436e-06, "loss": 1.6829, "step": 2652 }, { "epoch": 0.3042082330008027, "grad_norm": 12.3125, "learning_rate": 8.15628513357867e-06, "loss": 1.4341, "step": 2653 }, { "epoch": 0.30432289875014334, "grad_norm": 13.0, "learning_rate": 8.15484471222361e-06, "loss": 1.7399, "step": 2654 }, { "epoch": 0.304437564499484, "grad_norm": 16.25, "learning_rate": 8.153403855717858e-06, "loss": 1.6564, "step": 2655 }, { "epoch": 0.30455223024882466, "grad_norm": 14.0, "learning_rate": 8.151962564260155e-06, "loss": 1.7127, "step": 2656 }, { "epoch": 0.3046668959981653, "grad_norm": 14.5625, "learning_rate": 8.150520838049298e-06, "loss": 1.5322, "step": 2657 }, { "epoch": 0.30478156174750604, "grad_norm": 12.4375, "learning_rate": 8.149078677284144e-06, "loss": 1.5804, "step": 2658 }, { "epoch": 0.3048962274968467, "grad_norm": 12.125, "learning_rate": 8.147636082163615e-06, "loss": 1.9908, "step": 2659 }, { "epoch": 0.30501089324618735, "grad_norm": 14.3125, "learning_rate": 8.146193052886687e-06, "loss": 1.5289, "step": 2660 }, { "epoch": 0.305125558995528, "grad_norm": 13.4375, "learning_rate": 8.144749589652399e-06, "loss": 1.9099, "step": 2661 }, { "epoch": 0.3052402247448687, "grad_norm": 12.625, "learning_rate": 8.14330569265985e-06, "loss": 1.6016, "step": 2662 }, { "epoch": 0.3053548904942094, "grad_norm": 12.1875, "learning_rate": 8.141861362108197e-06, "loss": 1.5478, "step": 2663 }, { "epoch": 0.30546955624355004, "grad_norm": 15.1875, "learning_rate": 8.14041659819666e-06, "loss": 1.6879, "step": 2664 }, { "epoch": 0.3055842219928907, "grad_norm": 14.1875, "learning_rate": 8.138971401124514e-06, "loss": 1.6575, "step": 2665 }, { "epoch": 0.3056988877422314, "grad_norm": 10.9375, "learning_rate": 8.137525771091097e-06, "loss": 1.5581, "step": 2666 }, { "epoch": 0.3058135534915721, "grad_norm": 14.1875, "learning_rate": 8.136079708295807e-06, "loss": 1.863, "step": 2667 }, { "epoch": 0.30592821924091274, "grad_norm": 30.125, "learning_rate": 8.1346332129381e-06, "loss": 2.1813, "step": 2668 }, { "epoch": 0.3060428849902534, "grad_norm": 13.6875, "learning_rate": 8.133186285217493e-06, "loss": 2.1045, "step": 2669 }, { "epoch": 0.3061575507395941, "grad_norm": 18.75, "learning_rate": 8.131738925333563e-06, "loss": 1.7598, "step": 2670 }, { "epoch": 0.30627221648893477, "grad_norm": 12.5625, "learning_rate": 8.130291133485944e-06, "loss": 1.5697, "step": 2671 }, { "epoch": 0.3063868822382754, "grad_norm": 15.25, "learning_rate": 8.12884290987433e-06, "loss": 1.9453, "step": 2672 }, { "epoch": 0.3065015479876161, "grad_norm": 26.375, "learning_rate": 8.12739425469848e-06, "loss": 1.8167, "step": 2673 }, { "epoch": 0.30661621373695674, "grad_norm": 17.125, "learning_rate": 8.125945168158206e-06, "loss": 1.9527, "step": 2674 }, { "epoch": 0.30673087948629746, "grad_norm": 16.125, "learning_rate": 8.124495650453381e-06, "loss": 1.541, "step": 2675 }, { "epoch": 0.3068455452356381, "grad_norm": 17.25, "learning_rate": 8.12304570178394e-06, "loss": 2.0182, "step": 2676 }, { "epoch": 0.3069602109849788, "grad_norm": 13.75, "learning_rate": 8.121595322349875e-06, "loss": 2.165, "step": 2677 }, { "epoch": 0.30707487673431944, "grad_norm": 14.0625, "learning_rate": 8.120144512351238e-06, "loss": 1.4741, "step": 2678 }, { "epoch": 0.30718954248366015, "grad_norm": 26.375, "learning_rate": 8.118693271988142e-06, "loss": 2.0331, "step": 2679 }, { "epoch": 0.3073042082330008, "grad_norm": 12.375, "learning_rate": 8.117241601460755e-06, "loss": 1.6745, "step": 2680 }, { "epoch": 0.30741887398234147, "grad_norm": 14.0625, "learning_rate": 8.11578950096931e-06, "loss": 1.7347, "step": 2681 }, { "epoch": 0.3075335397316821, "grad_norm": 12.6875, "learning_rate": 8.114336970714096e-06, "loss": 1.5991, "step": 2682 }, { "epoch": 0.30764820548102284, "grad_norm": 13.6875, "learning_rate": 8.11288401089546e-06, "loss": 1.5061, "step": 2683 }, { "epoch": 0.3077628712303635, "grad_norm": 16.0, "learning_rate": 8.111430621713813e-06, "loss": 1.7709, "step": 2684 }, { "epoch": 0.30787753697970416, "grad_norm": 11.875, "learning_rate": 8.109976803369623e-06, "loss": 1.8559, "step": 2685 }, { "epoch": 0.3079922027290448, "grad_norm": 14.125, "learning_rate": 8.108522556063412e-06, "loss": 1.6197, "step": 2686 }, { "epoch": 0.30810686847838553, "grad_norm": 14.25, "learning_rate": 8.107067879995768e-06, "loss": 1.9211, "step": 2687 }, { "epoch": 0.3082215342277262, "grad_norm": 13.25, "learning_rate": 8.105612775367338e-06, "loss": 1.488, "step": 2688 }, { "epoch": 0.30833619997706685, "grad_norm": 13.4375, "learning_rate": 8.104157242378822e-06, "loss": 1.5857, "step": 2689 }, { "epoch": 0.3084508657264075, "grad_norm": 13.625, "learning_rate": 8.102701281230985e-06, "loss": 2.0696, "step": 2690 }, { "epoch": 0.30856553147574817, "grad_norm": 14.8125, "learning_rate": 8.101244892124651e-06, "loss": 1.8689, "step": 2691 }, { "epoch": 0.3086801972250889, "grad_norm": 13.5625, "learning_rate": 8.099788075260699e-06, "loss": 2.0817, "step": 2692 }, { "epoch": 0.30879486297442954, "grad_norm": 14.4375, "learning_rate": 8.098330830840066e-06, "loss": 1.6091, "step": 2693 }, { "epoch": 0.3089095287237702, "grad_norm": 13.3125, "learning_rate": 8.096873159063758e-06, "loss": 1.9207, "step": 2694 }, { "epoch": 0.30902419447311086, "grad_norm": 17.25, "learning_rate": 8.09541506013283e-06, "loss": 1.8102, "step": 2695 }, { "epoch": 0.3091388602224516, "grad_norm": 19.25, "learning_rate": 8.093956534248395e-06, "loss": 1.9661, "step": 2696 }, { "epoch": 0.30925352597179223, "grad_norm": 12.4375, "learning_rate": 8.092497581611635e-06, "loss": 1.6706, "step": 2697 }, { "epoch": 0.3093681917211329, "grad_norm": 12.9375, "learning_rate": 8.091038202423781e-06, "loss": 1.7043, "step": 2698 }, { "epoch": 0.30948285747047355, "grad_norm": 16.375, "learning_rate": 8.089578396886128e-06, "loss": 1.7793, "step": 2699 }, { "epoch": 0.30959752321981426, "grad_norm": 12.9375, "learning_rate": 8.088118165200028e-06, "loss": 1.6722, "step": 2700 }, { "epoch": 0.3097121889691549, "grad_norm": 12.5, "learning_rate": 8.086657507566891e-06, "loss": 1.5488, "step": 2701 }, { "epoch": 0.3098268547184956, "grad_norm": 14.1875, "learning_rate": 8.085196424188189e-06, "loss": 1.6259, "step": 2702 }, { "epoch": 0.30994152046783624, "grad_norm": 16.125, "learning_rate": 8.083734915265447e-06, "loss": 1.5632, "step": 2703 }, { "epoch": 0.31005618621717695, "grad_norm": 14.5625, "learning_rate": 8.082272981000258e-06, "loss": 1.6561, "step": 2704 }, { "epoch": 0.3101708519665176, "grad_norm": 13.3125, "learning_rate": 8.080810621594264e-06, "loss": 2.0796, "step": 2705 }, { "epoch": 0.31028551771585827, "grad_norm": 14.5625, "learning_rate": 8.079347837249169e-06, "loss": 1.962, "step": 2706 }, { "epoch": 0.31040018346519893, "grad_norm": 11.125, "learning_rate": 8.07788462816674e-06, "loss": 1.5756, "step": 2707 }, { "epoch": 0.3105148492145396, "grad_norm": 15.125, "learning_rate": 8.076420994548793e-06, "loss": 1.7799, "step": 2708 }, { "epoch": 0.3106295149638803, "grad_norm": 16.25, "learning_rate": 8.074956936597214e-06, "loss": 1.6691, "step": 2709 }, { "epoch": 0.31074418071322096, "grad_norm": 27.875, "learning_rate": 8.073492454513939e-06, "loss": 1.5917, "step": 2710 }, { "epoch": 0.3108588464625616, "grad_norm": 18.75, "learning_rate": 8.072027548500964e-06, "loss": 2.0228, "step": 2711 }, { "epoch": 0.3109735122119023, "grad_norm": 18.125, "learning_rate": 8.07056221876035e-06, "loss": 2.2114, "step": 2712 }, { "epoch": 0.311088177961243, "grad_norm": 13.6875, "learning_rate": 8.069096465494206e-06, "loss": 1.8426, "step": 2713 }, { "epoch": 0.31120284371058365, "grad_norm": 14.125, "learning_rate": 8.06763028890471e-06, "loss": 1.9544, "step": 2714 }, { "epoch": 0.3113175094599243, "grad_norm": 16.875, "learning_rate": 8.066163689194086e-06, "loss": 1.6764, "step": 2715 }, { "epoch": 0.31143217520926497, "grad_norm": 16.25, "learning_rate": 8.064696666564631e-06, "loss": 2.0001, "step": 2716 }, { "epoch": 0.3115468409586057, "grad_norm": 17.5, "learning_rate": 8.063229221218687e-06, "loss": 1.523, "step": 2717 }, { "epoch": 0.31166150670794635, "grad_norm": 11.0, "learning_rate": 8.061761353358664e-06, "loss": 1.2457, "step": 2718 }, { "epoch": 0.311776172457287, "grad_norm": 13.4375, "learning_rate": 8.060293063187024e-06, "loss": 1.7405, "step": 2719 }, { "epoch": 0.31189083820662766, "grad_norm": 15.1875, "learning_rate": 8.058824350906289e-06, "loss": 1.8042, "step": 2720 }, { "epoch": 0.3120055039559684, "grad_norm": 16.75, "learning_rate": 8.057355216719043e-06, "loss": 1.8854, "step": 2721 }, { "epoch": 0.31212016970530904, "grad_norm": 13.25, "learning_rate": 8.055885660827922e-06, "loss": 1.9556, "step": 2722 }, { "epoch": 0.3122348354546497, "grad_norm": 13.9375, "learning_rate": 8.054415683435626e-06, "loss": 1.4622, "step": 2723 }, { "epoch": 0.31234950120399035, "grad_norm": 14.0, "learning_rate": 8.052945284744908e-06, "loss": 1.5117, "step": 2724 }, { "epoch": 0.312464166953331, "grad_norm": 15.625, "learning_rate": 8.051474464958583e-06, "loss": 1.7475, "step": 2725 }, { "epoch": 0.3125788327026717, "grad_norm": 11.9375, "learning_rate": 8.05000322427952e-06, "loss": 1.7932, "step": 2726 }, { "epoch": 0.3126934984520124, "grad_norm": 18.25, "learning_rate": 8.048531562910655e-06, "loss": 1.731, "step": 2727 }, { "epoch": 0.31280816420135305, "grad_norm": 13.25, "learning_rate": 8.047059481054967e-06, "loss": 1.9622, "step": 2728 }, { "epoch": 0.3129228299506937, "grad_norm": 27.875, "learning_rate": 8.045586978915508e-06, "loss": 1.8285, "step": 2729 }, { "epoch": 0.3130374957000344, "grad_norm": 13.5, "learning_rate": 8.04411405669538e-06, "loss": 1.831, "step": 2730 }, { "epoch": 0.3131521614493751, "grad_norm": 12.9375, "learning_rate": 8.042640714597741e-06, "loss": 1.7788, "step": 2731 }, { "epoch": 0.31326682719871574, "grad_norm": 14.8125, "learning_rate": 8.041166952825816e-06, "loss": 1.6916, "step": 2732 }, { "epoch": 0.3133814929480564, "grad_norm": 14.4375, "learning_rate": 8.03969277158288e-06, "loss": 1.8114, "step": 2733 }, { "epoch": 0.3134961586973971, "grad_norm": 13.875, "learning_rate": 8.038218171072266e-06, "loss": 1.9445, "step": 2734 }, { "epoch": 0.31361082444673777, "grad_norm": 12.0625, "learning_rate": 8.03674315149737e-06, "loss": 1.581, "step": 2735 }, { "epoch": 0.3137254901960784, "grad_norm": 13.9375, "learning_rate": 8.035267713061641e-06, "loss": 1.8506, "step": 2736 }, { "epoch": 0.3138401559454191, "grad_norm": 12.0625, "learning_rate": 8.033791855968589e-06, "loss": 1.9211, "step": 2737 }, { "epoch": 0.3139548216947598, "grad_norm": 14.125, "learning_rate": 8.03231558042178e-06, "loss": 1.8277, "step": 2738 }, { "epoch": 0.31406948744410046, "grad_norm": 12.4375, "learning_rate": 8.03083888662484e-06, "loss": 1.6347, "step": 2739 }, { "epoch": 0.3141841531934411, "grad_norm": 16.625, "learning_rate": 8.029361774781445e-06, "loss": 1.7272, "step": 2740 }, { "epoch": 0.3142988189427818, "grad_norm": 13.8125, "learning_rate": 8.027884245095339e-06, "loss": 1.7358, "step": 2741 }, { "epoch": 0.31441348469212244, "grad_norm": 12.3125, "learning_rate": 8.026406297770319e-06, "loss": 1.7328, "step": 2742 }, { "epoch": 0.31452815044146315, "grad_norm": 12.6875, "learning_rate": 8.024927933010238e-06, "loss": 1.6987, "step": 2743 }, { "epoch": 0.3146428161908038, "grad_norm": 14.6875, "learning_rate": 8.02344915101901e-06, "loss": 1.8232, "step": 2744 }, { "epoch": 0.31475748194014447, "grad_norm": 11.5, "learning_rate": 8.021969952000603e-06, "loss": 1.6885, "step": 2745 }, { "epoch": 0.3148721476894851, "grad_norm": 15.0, "learning_rate": 8.020490336159045e-06, "loss": 1.7789, "step": 2746 }, { "epoch": 0.31498681343882584, "grad_norm": 12.0625, "learning_rate": 8.019010303698423e-06, "loss": 1.8335, "step": 2747 }, { "epoch": 0.3151014791881665, "grad_norm": 13.25, "learning_rate": 8.017529854822874e-06, "loss": 1.6253, "step": 2748 }, { "epoch": 0.31521614493750716, "grad_norm": 13.75, "learning_rate": 8.016048989736602e-06, "loss": 1.7991, "step": 2749 }, { "epoch": 0.3153308106868478, "grad_norm": 31.5, "learning_rate": 8.014567708643865e-06, "loss": 1.9912, "step": 2750 }, { "epoch": 0.31544547643618853, "grad_norm": 12.1875, "learning_rate": 8.013086011748973e-06, "loss": 1.9028, "step": 2751 }, { "epoch": 0.3155601421855292, "grad_norm": 14.1875, "learning_rate": 8.011603899256302e-06, "loss": 1.8452, "step": 2752 }, { "epoch": 0.31567480793486985, "grad_norm": 12.875, "learning_rate": 8.010121371370278e-06, "loss": 1.8623, "step": 2753 }, { "epoch": 0.3157894736842105, "grad_norm": 13.125, "learning_rate": 8.00863842829539e-06, "loss": 1.5289, "step": 2754 }, { "epoch": 0.3159041394335512, "grad_norm": 12.5, "learning_rate": 8.00715507023618e-06, "loss": 1.8719, "step": 2755 }, { "epoch": 0.3160188051828919, "grad_norm": 19.25, "learning_rate": 8.005671297397248e-06, "loss": 1.4504, "step": 2756 }, { "epoch": 0.31613347093223254, "grad_norm": 14.5625, "learning_rate": 8.004187109983257e-06, "loss": 1.7992, "step": 2757 }, { "epoch": 0.3162481366815732, "grad_norm": 14.75, "learning_rate": 8.002702508198919e-06, "loss": 1.6839, "step": 2758 }, { "epoch": 0.31636280243091386, "grad_norm": 14.625, "learning_rate": 8.001217492249005e-06, "loss": 1.5638, "step": 2759 }, { "epoch": 0.3164774681802546, "grad_norm": 14.5, "learning_rate": 7.999732062338348e-06, "loss": 1.9698, "step": 2760 }, { "epoch": 0.31659213392959523, "grad_norm": 19.75, "learning_rate": 7.99824621867183e-06, "loss": 2.3607, "step": 2761 }, { "epoch": 0.3167067996789359, "grad_norm": 12.75, "learning_rate": 7.9967599614544e-06, "loss": 1.6668, "step": 2762 }, { "epoch": 0.31682146542827655, "grad_norm": 15.625, "learning_rate": 7.995273290891057e-06, "loss": 1.8813, "step": 2763 }, { "epoch": 0.31693613117761726, "grad_norm": 13.75, "learning_rate": 7.993786207186859e-06, "loss": 2.0122, "step": 2764 }, { "epoch": 0.3170507969269579, "grad_norm": 13.6875, "learning_rate": 7.99229871054692e-06, "loss": 1.7141, "step": 2765 }, { "epoch": 0.3171654626762986, "grad_norm": 13.1875, "learning_rate": 7.990810801176412e-06, "loss": 1.916, "step": 2766 }, { "epoch": 0.31728012842563924, "grad_norm": 13.75, "learning_rate": 7.989322479280564e-06, "loss": 1.8086, "step": 2767 }, { "epoch": 0.31739479417497996, "grad_norm": 15.5, "learning_rate": 7.987833745064662e-06, "loss": 1.7799, "step": 2768 }, { "epoch": 0.3175094599243206, "grad_norm": 31.875, "learning_rate": 7.986344598734048e-06, "loss": 2.1178, "step": 2769 }, { "epoch": 0.3176241256736613, "grad_norm": 12.9375, "learning_rate": 7.984855040494122e-06, "loss": 1.7957, "step": 2770 }, { "epoch": 0.31773879142300193, "grad_norm": 29.5, "learning_rate": 7.98336507055034e-06, "loss": 2.0391, "step": 2771 }, { "epoch": 0.31785345717234265, "grad_norm": 16.0, "learning_rate": 7.981874689108212e-06, "loss": 1.6029, "step": 2772 }, { "epoch": 0.3179681229216833, "grad_norm": 19.25, "learning_rate": 7.980383896373312e-06, "loss": 2.0965, "step": 2773 }, { "epoch": 0.31808278867102396, "grad_norm": 10.8125, "learning_rate": 7.978892692551265e-06, "loss": 1.465, "step": 2774 }, { "epoch": 0.3181974544203646, "grad_norm": 14.9375, "learning_rate": 7.977401077847754e-06, "loss": 1.6704, "step": 2775 }, { "epoch": 0.3183121201697053, "grad_norm": 17.375, "learning_rate": 7.97590905246852e-06, "loss": 1.6595, "step": 2776 }, { "epoch": 0.318426785919046, "grad_norm": 14.1875, "learning_rate": 7.974416616619355e-06, "loss": 1.5871, "step": 2777 }, { "epoch": 0.31854145166838665, "grad_norm": 35.25, "learning_rate": 7.972923770506118e-06, "loss": 1.9243, "step": 2778 }, { "epoch": 0.3186561174177273, "grad_norm": 23.875, "learning_rate": 7.971430514334714e-06, "loss": 2.0147, "step": 2779 }, { "epoch": 0.318770783167068, "grad_norm": 12.625, "learning_rate": 7.969936848311113e-06, "loss": 1.6784, "step": 2780 }, { "epoch": 0.3188854489164087, "grad_norm": 14.1875, "learning_rate": 7.968442772641335e-06, "loss": 1.7871, "step": 2781 }, { "epoch": 0.31900011466574935, "grad_norm": 14.375, "learning_rate": 7.966948287531461e-06, "loss": 1.7887, "step": 2782 }, { "epoch": 0.31911478041509, "grad_norm": 13.875, "learning_rate": 7.965453393187624e-06, "loss": 1.6083, "step": 2783 }, { "epoch": 0.31922944616443066, "grad_norm": 14.875, "learning_rate": 7.96395808981602e-06, "loss": 1.7423, "step": 2784 }, { "epoch": 0.3193441119137714, "grad_norm": 13.8125, "learning_rate": 7.962462377622895e-06, "loss": 1.6623, "step": 2785 }, { "epoch": 0.31945877766311204, "grad_norm": 12.0625, "learning_rate": 7.960966256814554e-06, "loss": 1.6115, "step": 2786 }, { "epoch": 0.3195734434124527, "grad_norm": 18.625, "learning_rate": 7.95946972759736e-06, "loss": 2.2075, "step": 2787 }, { "epoch": 0.31968810916179335, "grad_norm": 17.75, "learning_rate": 7.95797279017773e-06, "loss": 2.2438, "step": 2788 }, { "epoch": 0.31980277491113407, "grad_norm": 13.125, "learning_rate": 7.956475444762138e-06, "loss": 1.9911, "step": 2789 }, { "epoch": 0.31991744066047473, "grad_norm": 13.6875, "learning_rate": 7.954977691557112e-06, "loss": 1.6861, "step": 2790 }, { "epoch": 0.3200321064098154, "grad_norm": 15.375, "learning_rate": 7.95347953076924e-06, "loss": 1.6032, "step": 2791 }, { "epoch": 0.32014677215915605, "grad_norm": 14.125, "learning_rate": 7.951980962605168e-06, "loss": 2.446, "step": 2792 }, { "epoch": 0.3202614379084967, "grad_norm": 11.5625, "learning_rate": 7.95048198727159e-06, "loss": 1.6682, "step": 2793 }, { "epoch": 0.3203761036578374, "grad_norm": 12.4375, "learning_rate": 7.948982604975263e-06, "loss": 1.8963, "step": 2794 }, { "epoch": 0.3204907694071781, "grad_norm": 14.625, "learning_rate": 7.947482815923001e-06, "loss": 1.8057, "step": 2795 }, { "epoch": 0.32060543515651874, "grad_norm": 13.6875, "learning_rate": 7.945982620321666e-06, "loss": 2.139, "step": 2796 }, { "epoch": 0.3207201009058594, "grad_norm": 14.0625, "learning_rate": 7.944482018378186e-06, "loss": 2.1072, "step": 2797 }, { "epoch": 0.3208347666552001, "grad_norm": 11.75, "learning_rate": 7.942981010299537e-06, "loss": 1.403, "step": 2798 }, { "epoch": 0.32094943240454077, "grad_norm": 18.0, "learning_rate": 7.941479596292756e-06, "loss": 1.4798, "step": 2799 }, { "epoch": 0.3210640981538814, "grad_norm": 12.5, "learning_rate": 7.939977776564936e-06, "loss": 1.9662, "step": 2800 }, { "epoch": 0.3211787639032221, "grad_norm": 14.9375, "learning_rate": 7.938475551323222e-06, "loss": 1.9195, "step": 2801 }, { "epoch": 0.3212934296525628, "grad_norm": 12.75, "learning_rate": 7.936972920774816e-06, "loss": 1.8852, "step": 2802 }, { "epoch": 0.32140809540190346, "grad_norm": 12.3125, "learning_rate": 7.93546988512698e-06, "loss": 1.6937, "step": 2803 }, { "epoch": 0.3215227611512441, "grad_norm": 13.9375, "learning_rate": 7.93396644458703e-06, "loss": 1.6935, "step": 2804 }, { "epoch": 0.3216374269005848, "grad_norm": 15.5, "learning_rate": 7.932462599362336e-06, "loss": 1.8207, "step": 2805 }, { "epoch": 0.3217520926499255, "grad_norm": 21.0, "learning_rate": 7.930958349660323e-06, "loss": 1.5474, "step": 2806 }, { "epoch": 0.32186675839926615, "grad_norm": 14.0, "learning_rate": 7.929453695688475e-06, "loss": 1.8658, "step": 2807 }, { "epoch": 0.3219814241486068, "grad_norm": 14.0, "learning_rate": 7.92794863765433e-06, "loss": 1.7296, "step": 2808 }, { "epoch": 0.32209608989794747, "grad_norm": 15.75, "learning_rate": 7.926443175765483e-06, "loss": 1.4524, "step": 2809 }, { "epoch": 0.3222107556472881, "grad_norm": 12.625, "learning_rate": 7.924937310229583e-06, "loss": 1.8154, "step": 2810 }, { "epoch": 0.32232542139662884, "grad_norm": 13.25, "learning_rate": 7.923431041254334e-06, "loss": 1.7106, "step": 2811 }, { "epoch": 0.3224400871459695, "grad_norm": 14.0625, "learning_rate": 7.9219243690475e-06, "loss": 1.6179, "step": 2812 }, { "epoch": 0.32255475289531016, "grad_norm": 14.0, "learning_rate": 7.920417293816896e-06, "loss": 1.5702, "step": 2813 }, { "epoch": 0.3226694186446508, "grad_norm": 18.5, "learning_rate": 7.918909815770395e-06, "loss": 1.9617, "step": 2814 }, { "epoch": 0.32278408439399153, "grad_norm": 12.0, "learning_rate": 7.917401935115923e-06, "loss": 1.4623, "step": 2815 }, { "epoch": 0.3228987501433322, "grad_norm": 14.1875, "learning_rate": 7.915893652061466e-06, "loss": 1.6765, "step": 2816 }, { "epoch": 0.32301341589267285, "grad_norm": 14.125, "learning_rate": 7.91438496681506e-06, "loss": 1.8329, "step": 2817 }, { "epoch": 0.3231280816420135, "grad_norm": 15.5625, "learning_rate": 7.912875879584803e-06, "loss": 2.2237, "step": 2818 }, { "epoch": 0.3232427473913542, "grad_norm": 13.4375, "learning_rate": 7.911366390578842e-06, "loss": 1.7299, "step": 2819 }, { "epoch": 0.3233574131406949, "grad_norm": 13.0, "learning_rate": 7.909856500005383e-06, "loss": 1.7623, "step": 2820 }, { "epoch": 0.32347207889003554, "grad_norm": 14.4375, "learning_rate": 7.908346208072686e-06, "loss": 2.0709, "step": 2821 }, { "epoch": 0.3235867446393762, "grad_norm": 14.4375, "learning_rate": 7.906835514989069e-06, "loss": 2.0736, "step": 2822 }, { "epoch": 0.3237014103887169, "grad_norm": 15.8125, "learning_rate": 7.9053244209629e-06, "loss": 1.7811, "step": 2823 }, { "epoch": 0.3238160761380576, "grad_norm": 17.25, "learning_rate": 7.903812926202611e-06, "loss": 1.7635, "step": 2824 }, { "epoch": 0.32393074188739823, "grad_norm": 14.625, "learning_rate": 7.902301030916679e-06, "loss": 1.8441, "step": 2825 }, { "epoch": 0.3240454076367389, "grad_norm": 12.25, "learning_rate": 7.900788735313643e-06, "loss": 1.5468, "step": 2826 }, { "epoch": 0.3241600733860796, "grad_norm": 17.125, "learning_rate": 7.899276039602094e-06, "loss": 1.9909, "step": 2827 }, { "epoch": 0.32427473913542026, "grad_norm": 15.4375, "learning_rate": 7.897762943990681e-06, "loss": 1.8197, "step": 2828 }, { "epoch": 0.3243894048847609, "grad_norm": 33.75, "learning_rate": 7.896249448688107e-06, "loss": 2.2309, "step": 2829 }, { "epoch": 0.3245040706341016, "grad_norm": 11.6875, "learning_rate": 7.894735553903128e-06, "loss": 1.9244, "step": 2830 }, { "epoch": 0.32461873638344224, "grad_norm": 12.0, "learning_rate": 7.893221259844558e-06, "loss": 1.8385, "step": 2831 }, { "epoch": 0.32473340213278296, "grad_norm": 13.8125, "learning_rate": 7.891706566721267e-06, "loss": 1.8121, "step": 2832 }, { "epoch": 0.3248480678821236, "grad_norm": 16.0, "learning_rate": 7.890191474742173e-06, "loss": 1.6405, "step": 2833 }, { "epoch": 0.3249627336314643, "grad_norm": 15.125, "learning_rate": 7.88867598411626e-06, "loss": 1.6376, "step": 2834 }, { "epoch": 0.32507739938080493, "grad_norm": 13.8125, "learning_rate": 7.887160095052555e-06, "loss": 1.7953, "step": 2835 }, { "epoch": 0.32519206513014565, "grad_norm": 15.6875, "learning_rate": 7.88564380776015e-06, "loss": 1.9071, "step": 2836 }, { "epoch": 0.3253067308794863, "grad_norm": 16.75, "learning_rate": 7.88412712244819e-06, "loss": 1.7398, "step": 2837 }, { "epoch": 0.32542139662882696, "grad_norm": 12.4375, "learning_rate": 7.882610039325868e-06, "loss": 1.7563, "step": 2838 }, { "epoch": 0.3255360623781676, "grad_norm": 36.5, "learning_rate": 7.881092558602438e-06, "loss": 1.5449, "step": 2839 }, { "epoch": 0.32565072812750834, "grad_norm": 15.125, "learning_rate": 7.87957468048721e-06, "loss": 2.0597, "step": 2840 }, { "epoch": 0.325765393876849, "grad_norm": 12.625, "learning_rate": 7.878056405189543e-06, "loss": 1.9063, "step": 2841 }, { "epoch": 0.32588005962618966, "grad_norm": 13.5, "learning_rate": 7.876537732918855e-06, "loss": 1.7967, "step": 2842 }, { "epoch": 0.3259947253755303, "grad_norm": 14.5625, "learning_rate": 7.87501866388462e-06, "loss": 1.8429, "step": 2843 }, { "epoch": 0.32610939112487103, "grad_norm": 15.0625, "learning_rate": 7.873499198296362e-06, "loss": 1.8504, "step": 2844 }, { "epoch": 0.3262240568742117, "grad_norm": 13.9375, "learning_rate": 7.871979336363664e-06, "loss": 1.7752, "step": 2845 }, { "epoch": 0.32633872262355235, "grad_norm": 15.0, "learning_rate": 7.870459078296163e-06, "loss": 2.215, "step": 2846 }, { "epoch": 0.326453388372893, "grad_norm": 11.6875, "learning_rate": 7.868938424303546e-06, "loss": 1.506, "step": 2847 }, { "epoch": 0.32656805412223366, "grad_norm": 16.25, "learning_rate": 7.86741737459556e-06, "loss": 1.55, "step": 2848 }, { "epoch": 0.3266827198715744, "grad_norm": 11.75, "learning_rate": 7.865895929382007e-06, "loss": 1.6048, "step": 2849 }, { "epoch": 0.32679738562091504, "grad_norm": 14.125, "learning_rate": 7.86437408887274e-06, "loss": 1.5587, "step": 2850 }, { "epoch": 0.3269120513702557, "grad_norm": 14.5, "learning_rate": 7.862851853277664e-06, "loss": 1.7263, "step": 2851 }, { "epoch": 0.32702671711959636, "grad_norm": 29.125, "learning_rate": 7.861329222806748e-06, "loss": 2.2004, "step": 2852 }, { "epoch": 0.32714138286893707, "grad_norm": 13.625, "learning_rate": 7.859806197670007e-06, "loss": 1.7022, "step": 2853 }, { "epoch": 0.32725604861827773, "grad_norm": 13.3125, "learning_rate": 7.858282778077514e-06, "loss": 1.4545, "step": 2854 }, { "epoch": 0.3273707143676184, "grad_norm": 13.4375, "learning_rate": 7.856758964239397e-06, "loss": 1.8023, "step": 2855 }, { "epoch": 0.32748538011695905, "grad_norm": 12.75, "learning_rate": 7.855234756365833e-06, "loss": 1.769, "step": 2856 }, { "epoch": 0.32760004586629976, "grad_norm": 13.25, "learning_rate": 7.853710154667062e-06, "loss": 1.8873, "step": 2857 }, { "epoch": 0.3277147116156404, "grad_norm": 16.125, "learning_rate": 7.852185159353372e-06, "loss": 1.8865, "step": 2858 }, { "epoch": 0.3278293773649811, "grad_norm": 16.625, "learning_rate": 7.850659770635106e-06, "loss": 1.5913, "step": 2859 }, { "epoch": 0.32794404311432174, "grad_norm": 16.75, "learning_rate": 7.849133988722663e-06, "loss": 1.4918, "step": 2860 }, { "epoch": 0.32805870886366245, "grad_norm": 12.875, "learning_rate": 7.847607813826496e-06, "loss": 1.6829, "step": 2861 }, { "epoch": 0.3281733746130031, "grad_norm": 17.0, "learning_rate": 7.846081246157111e-06, "loss": 2.0232, "step": 2862 }, { "epoch": 0.32828804036234377, "grad_norm": 17.75, "learning_rate": 7.84455428592507e-06, "loss": 1.6571, "step": 2863 }, { "epoch": 0.32840270611168443, "grad_norm": 15.125, "learning_rate": 7.843026933340989e-06, "loss": 1.9082, "step": 2864 }, { "epoch": 0.3285173718610251, "grad_norm": 12.4375, "learning_rate": 7.841499188615533e-06, "loss": 1.8618, "step": 2865 }, { "epoch": 0.3286320376103658, "grad_norm": 14.875, "learning_rate": 7.839971051959427e-06, "loss": 2.0395, "step": 2866 }, { "epoch": 0.32874670335970646, "grad_norm": 13.8125, "learning_rate": 7.838442523583451e-06, "loss": 1.7334, "step": 2867 }, { "epoch": 0.3288613691090471, "grad_norm": 16.375, "learning_rate": 7.836913603698435e-06, "loss": 1.9301, "step": 2868 }, { "epoch": 0.3289760348583878, "grad_norm": 12.75, "learning_rate": 7.835384292515263e-06, "loss": 1.6856, "step": 2869 }, { "epoch": 0.3290907006077285, "grad_norm": 15.0, "learning_rate": 7.833854590244876e-06, "loss": 1.593, "step": 2870 }, { "epoch": 0.32920536635706915, "grad_norm": 18.375, "learning_rate": 7.832324497098267e-06, "loss": 1.7777, "step": 2871 }, { "epoch": 0.3293200321064098, "grad_norm": 14.875, "learning_rate": 7.83079401328648e-06, "loss": 2.2257, "step": 2872 }, { "epoch": 0.32943469785575047, "grad_norm": 15.0625, "learning_rate": 7.82926313902062e-06, "loss": 2.2475, "step": 2873 }, { "epoch": 0.3295493636050912, "grad_norm": 14.625, "learning_rate": 7.82773187451184e-06, "loss": 1.7885, "step": 2874 }, { "epoch": 0.32966402935443184, "grad_norm": 14.6875, "learning_rate": 7.826200219971351e-06, "loss": 1.7906, "step": 2875 }, { "epoch": 0.3297786951037725, "grad_norm": 16.0, "learning_rate": 7.824668175610412e-06, "loss": 1.8364, "step": 2876 }, { "epoch": 0.32989336085311316, "grad_norm": 12.4375, "learning_rate": 7.823135741640345e-06, "loss": 1.6954, "step": 2877 }, { "epoch": 0.3300080266024539, "grad_norm": 16.125, "learning_rate": 7.821602918272513e-06, "loss": 1.954, "step": 2878 }, { "epoch": 0.33012269235179453, "grad_norm": 12.0, "learning_rate": 7.820069705718343e-06, "loss": 1.6377, "step": 2879 }, { "epoch": 0.3302373581011352, "grad_norm": 16.5, "learning_rate": 7.818536104189313e-06, "loss": 1.9795, "step": 2880 }, { "epoch": 0.33035202385047585, "grad_norm": 14.25, "learning_rate": 7.817002113896955e-06, "loss": 1.9839, "step": 2881 }, { "epoch": 0.3304666895998165, "grad_norm": 13.3125, "learning_rate": 7.81546773505285e-06, "loss": 1.9385, "step": 2882 }, { "epoch": 0.3305813553491572, "grad_norm": 11.75, "learning_rate": 7.81393296786864e-06, "loss": 1.8939, "step": 2883 }, { "epoch": 0.3306960210984979, "grad_norm": 12.5625, "learning_rate": 7.812397812556016e-06, "loss": 1.9891, "step": 2884 }, { "epoch": 0.33081068684783854, "grad_norm": 13.4375, "learning_rate": 7.810862269326722e-06, "loss": 1.6762, "step": 2885 }, { "epoch": 0.3309253525971792, "grad_norm": 14.9375, "learning_rate": 7.809326338392557e-06, "loss": 2.137, "step": 2886 }, { "epoch": 0.3310400183465199, "grad_norm": 12.6875, "learning_rate": 7.807790019965377e-06, "loss": 1.9945, "step": 2887 }, { "epoch": 0.3311546840958606, "grad_norm": 12.125, "learning_rate": 7.806253314257082e-06, "loss": 1.8879, "step": 2888 }, { "epoch": 0.33126934984520123, "grad_norm": 13.3125, "learning_rate": 7.804716221479638e-06, "loss": 1.6368, "step": 2889 }, { "epoch": 0.3313840155945419, "grad_norm": 13.125, "learning_rate": 7.803178741845052e-06, "loss": 1.5803, "step": 2890 }, { "epoch": 0.3314986813438826, "grad_norm": 27.75, "learning_rate": 7.80164087556539e-06, "loss": 1.7796, "step": 2891 }, { "epoch": 0.33161334709322327, "grad_norm": 13.9375, "learning_rate": 7.800102622852776e-06, "loss": 1.7554, "step": 2892 }, { "epoch": 0.3317280128425639, "grad_norm": 13.75, "learning_rate": 7.79856398391938e-06, "loss": 1.7047, "step": 2893 }, { "epoch": 0.3318426785919046, "grad_norm": 13.125, "learning_rate": 7.797024958977425e-06, "loss": 1.6283, "step": 2894 }, { "epoch": 0.3319573443412453, "grad_norm": 12.5, "learning_rate": 7.795485548239196e-06, "loss": 1.7509, "step": 2895 }, { "epoch": 0.33207201009058596, "grad_norm": 15.375, "learning_rate": 7.793945751917022e-06, "loss": 1.8093, "step": 2896 }, { "epoch": 0.3321866758399266, "grad_norm": 14.1875, "learning_rate": 7.792405570223289e-06, "loss": 1.8702, "step": 2897 }, { "epoch": 0.3323013415892673, "grad_norm": 15.375, "learning_rate": 7.790865003370435e-06, "loss": 1.7352, "step": 2898 }, { "epoch": 0.33241600733860793, "grad_norm": 18.375, "learning_rate": 7.789324051570953e-06, "loss": 1.7602, "step": 2899 }, { "epoch": 0.33253067308794865, "grad_norm": 12.5625, "learning_rate": 7.787782715037387e-06, "loss": 1.5397, "step": 2900 }, { "epoch": 0.3326453388372893, "grad_norm": 13.9375, "learning_rate": 7.786240993982335e-06, "loss": 1.8307, "step": 2901 }, { "epoch": 0.33276000458662996, "grad_norm": 14.8125, "learning_rate": 7.78469888861845e-06, "loss": 1.5913, "step": 2902 }, { "epoch": 0.3328746703359706, "grad_norm": 13.25, "learning_rate": 7.783156399158433e-06, "loss": 1.6687, "step": 2903 }, { "epoch": 0.33298933608531134, "grad_norm": 12.125, "learning_rate": 7.781613525815044e-06, "loss": 1.5979, "step": 2904 }, { "epoch": 0.333104001834652, "grad_norm": 14.1875, "learning_rate": 7.78007026880109e-06, "loss": 1.5402, "step": 2905 }, { "epoch": 0.33321866758399266, "grad_norm": 15.0625, "learning_rate": 7.778526628329436e-06, "loss": 1.8905, "step": 2906 }, { "epoch": 0.3333333333333333, "grad_norm": 13.4375, "learning_rate": 7.776982604612996e-06, "loss": 1.3923, "step": 2907 }, { "epoch": 0.33344799908267403, "grad_norm": 15.3125, "learning_rate": 7.775438197864742e-06, "loss": 2.0288, "step": 2908 }, { "epoch": 0.3335626648320147, "grad_norm": 12.875, "learning_rate": 7.773893408297693e-06, "loss": 1.7245, "step": 2909 }, { "epoch": 0.33367733058135535, "grad_norm": 15.4375, "learning_rate": 7.772348236124923e-06, "loss": 2.2501, "step": 2910 }, { "epoch": 0.333791996330696, "grad_norm": 13.0625, "learning_rate": 7.770802681559558e-06, "loss": 1.8221, "step": 2911 }, { "epoch": 0.3339066620800367, "grad_norm": 15.125, "learning_rate": 7.769256744814781e-06, "loss": 1.4374, "step": 2912 }, { "epoch": 0.3340213278293774, "grad_norm": 15.1875, "learning_rate": 7.767710426103823e-06, "loss": 1.4154, "step": 2913 }, { "epoch": 0.33413599357871804, "grad_norm": 24.75, "learning_rate": 7.766163725639969e-06, "loss": 1.9876, "step": 2914 }, { "epoch": 0.3342506593280587, "grad_norm": 12.25, "learning_rate": 7.764616643636556e-06, "loss": 1.9985, "step": 2915 }, { "epoch": 0.33436532507739936, "grad_norm": 11.75, "learning_rate": 7.763069180306977e-06, "loss": 1.8282, "step": 2916 }, { "epoch": 0.33447999082674007, "grad_norm": 13.1875, "learning_rate": 7.761521335864672e-06, "loss": 1.5959, "step": 2917 }, { "epoch": 0.33459465657608073, "grad_norm": 29.375, "learning_rate": 7.759973110523137e-06, "loss": 2.4111, "step": 2918 }, { "epoch": 0.3347093223254214, "grad_norm": 13.0625, "learning_rate": 7.758424504495924e-06, "loss": 1.2617, "step": 2919 }, { "epoch": 0.33482398807476205, "grad_norm": 13.25, "learning_rate": 7.756875517996631e-06, "loss": 2.0493, "step": 2920 }, { "epoch": 0.33493865382410276, "grad_norm": 14.0625, "learning_rate": 7.75532615123891e-06, "loss": 1.4152, "step": 2921 }, { "epoch": 0.3350533195734434, "grad_norm": 12.8125, "learning_rate": 7.753776404436466e-06, "loss": 1.7395, "step": 2922 }, { "epoch": 0.3351679853227841, "grad_norm": 13.75, "learning_rate": 7.75222627780306e-06, "loss": 1.5387, "step": 2923 }, { "epoch": 0.33528265107212474, "grad_norm": 15.625, "learning_rate": 7.750675771552503e-06, "loss": 1.9151, "step": 2924 }, { "epoch": 0.33539731682146545, "grad_norm": 12.9375, "learning_rate": 7.749124885898655e-06, "loss": 1.8871, "step": 2925 }, { "epoch": 0.3355119825708061, "grad_norm": 14.625, "learning_rate": 7.74757362105543e-06, "loss": 1.7312, "step": 2926 }, { "epoch": 0.33562664832014677, "grad_norm": 12.8125, "learning_rate": 7.746021977236801e-06, "loss": 1.7028, "step": 2927 }, { "epoch": 0.33574131406948743, "grad_norm": 13.6875, "learning_rate": 7.744469954656781e-06, "loss": 1.5234, "step": 2928 }, { "epoch": 0.33585597981882814, "grad_norm": 13.3125, "learning_rate": 7.742917553529448e-06, "loss": 1.9333, "step": 2929 }, { "epoch": 0.3359706455681688, "grad_norm": 11.6875, "learning_rate": 7.741364774068921e-06, "loss": 1.6694, "step": 2930 }, { "epoch": 0.33608531131750946, "grad_norm": 13.6875, "learning_rate": 7.739811616489378e-06, "loss": 1.594, "step": 2931 }, { "epoch": 0.3361999770668501, "grad_norm": 11.5625, "learning_rate": 7.73825808100505e-06, "loss": 1.8548, "step": 2932 }, { "epoch": 0.3363146428161908, "grad_norm": 14.0, "learning_rate": 7.736704167830216e-06, "loss": 1.7777, "step": 2933 }, { "epoch": 0.3364293085655315, "grad_norm": 11.75, "learning_rate": 7.735149877179206e-06, "loss": 1.9656, "step": 2934 }, { "epoch": 0.33654397431487215, "grad_norm": 16.125, "learning_rate": 7.73359520926641e-06, "loss": 1.8996, "step": 2935 }, { "epoch": 0.3366586400642128, "grad_norm": 19.625, "learning_rate": 7.732040164306261e-06, "loss": 2.0582, "step": 2936 }, { "epoch": 0.33677330581355347, "grad_norm": 30.25, "learning_rate": 7.730484742513247e-06, "loss": 1.7477, "step": 2937 }, { "epoch": 0.3368879715628942, "grad_norm": 13.6875, "learning_rate": 7.728928944101911e-06, "loss": 1.6389, "step": 2938 }, { "epoch": 0.33700263731223484, "grad_norm": 15.375, "learning_rate": 7.727372769286847e-06, "loss": 1.567, "step": 2939 }, { "epoch": 0.3371173030615755, "grad_norm": 10.1875, "learning_rate": 7.725816218282698e-06, "loss": 1.5032, "step": 2940 }, { "epoch": 0.33723196881091616, "grad_norm": 12.25, "learning_rate": 7.724259291304159e-06, "loss": 2.1525, "step": 2941 }, { "epoch": 0.3373466345602569, "grad_norm": 14.375, "learning_rate": 7.722701988565979e-06, "loss": 1.9074, "step": 2942 }, { "epoch": 0.33746130030959753, "grad_norm": 13.375, "learning_rate": 7.721144310282961e-06, "loss": 1.6323, "step": 2943 }, { "epoch": 0.3375759660589382, "grad_norm": 14.4375, "learning_rate": 7.719586256669955e-06, "loss": 1.5763, "step": 2944 }, { "epoch": 0.33769063180827885, "grad_norm": 14.875, "learning_rate": 7.718027827941865e-06, "loss": 1.6104, "step": 2945 }, { "epoch": 0.33780529755761957, "grad_norm": 13.9375, "learning_rate": 7.716469024313645e-06, "loss": 1.9587, "step": 2946 }, { "epoch": 0.3379199633069602, "grad_norm": 14.8125, "learning_rate": 7.714909846000305e-06, "loss": 1.9156, "step": 2947 }, { "epoch": 0.3380346290563009, "grad_norm": 13.875, "learning_rate": 7.713350293216904e-06, "loss": 1.877, "step": 2948 }, { "epoch": 0.33814929480564154, "grad_norm": 28.625, "learning_rate": 7.711790366178549e-06, "loss": 1.9226, "step": 2949 }, { "epoch": 0.3382639605549822, "grad_norm": 21.0, "learning_rate": 7.710230065100404e-06, "loss": 1.9975, "step": 2950 }, { "epoch": 0.3383786263043229, "grad_norm": 12.75, "learning_rate": 7.708669390197684e-06, "loss": 1.6836, "step": 2951 }, { "epoch": 0.3384932920536636, "grad_norm": 16.625, "learning_rate": 7.707108341685654e-06, "loss": 1.7136, "step": 2952 }, { "epoch": 0.33860795780300423, "grad_norm": 15.5, "learning_rate": 7.70554691977963e-06, "loss": 1.9353, "step": 2953 }, { "epoch": 0.3387226235523449, "grad_norm": 13.5625, "learning_rate": 7.703985124694981e-06, "loss": 1.9081, "step": 2954 }, { "epoch": 0.3388372893016856, "grad_norm": 30.375, "learning_rate": 7.702422956647126e-06, "loss": 1.9949, "step": 2955 }, { "epoch": 0.33895195505102627, "grad_norm": 12.8125, "learning_rate": 7.700860415851539e-06, "loss": 1.7749, "step": 2956 }, { "epoch": 0.3390666208003669, "grad_norm": 14.9375, "learning_rate": 7.699297502523739e-06, "loss": 1.9899, "step": 2957 }, { "epoch": 0.3391812865497076, "grad_norm": 18.125, "learning_rate": 7.697734216879302e-06, "loss": 2.0669, "step": 2958 }, { "epoch": 0.3392959522990483, "grad_norm": 11.9375, "learning_rate": 7.696170559133854e-06, "loss": 1.7808, "step": 2959 }, { "epoch": 0.33941061804838896, "grad_norm": 14.8125, "learning_rate": 7.694606529503071e-06, "loss": 1.7279, "step": 2960 }, { "epoch": 0.3395252837977296, "grad_norm": 15.5, "learning_rate": 7.693042128202679e-06, "loss": 1.6042, "step": 2961 }, { "epoch": 0.3396399495470703, "grad_norm": 16.375, "learning_rate": 7.691477355448462e-06, "loss": 1.8729, "step": 2962 }, { "epoch": 0.339754615296411, "grad_norm": 18.0, "learning_rate": 7.689912211456248e-06, "loss": 1.9828, "step": 2963 }, { "epoch": 0.33986928104575165, "grad_norm": 14.25, "learning_rate": 7.688346696441917e-06, "loss": 1.9245, "step": 2964 }, { "epoch": 0.3399839467950923, "grad_norm": 10.4375, "learning_rate": 7.686780810621406e-06, "loss": 1.3596, "step": 2965 }, { "epoch": 0.34009861254443297, "grad_norm": 16.25, "learning_rate": 7.685214554210693e-06, "loss": 2.0492, "step": 2966 }, { "epoch": 0.3402132782937736, "grad_norm": 13.125, "learning_rate": 7.68364792742582e-06, "loss": 1.5892, "step": 2967 }, { "epoch": 0.34032794404311434, "grad_norm": 14.1875, "learning_rate": 7.682080930482871e-06, "loss": 1.7416, "step": 2968 }, { "epoch": 0.340442609792455, "grad_norm": 13.875, "learning_rate": 7.680513563597982e-06, "loss": 1.8511, "step": 2969 }, { "epoch": 0.34055727554179566, "grad_norm": 15.3125, "learning_rate": 7.678945826987343e-06, "loss": 1.8883, "step": 2970 }, { "epoch": 0.3406719412911363, "grad_norm": 13.875, "learning_rate": 7.677377720867189e-06, "loss": 1.7575, "step": 2971 }, { "epoch": 0.34078660704047703, "grad_norm": 15.0, "learning_rate": 7.675809245453817e-06, "loss": 1.8651, "step": 2972 }, { "epoch": 0.3409012727898177, "grad_norm": 12.9375, "learning_rate": 7.674240400963567e-06, "loss": 1.2487, "step": 2973 }, { "epoch": 0.34101593853915835, "grad_norm": 12.9375, "learning_rate": 7.672671187612825e-06, "loss": 1.5857, "step": 2974 }, { "epoch": 0.341130604288499, "grad_norm": 15.0625, "learning_rate": 7.671101605618042e-06, "loss": 1.5236, "step": 2975 }, { "epoch": 0.3412452700378397, "grad_norm": 18.25, "learning_rate": 7.669531655195705e-06, "loss": 1.812, "step": 2976 }, { "epoch": 0.3413599357871804, "grad_norm": 32.0, "learning_rate": 7.667961336562363e-06, "loss": 1.9573, "step": 2977 }, { "epoch": 0.34147460153652104, "grad_norm": 12.375, "learning_rate": 7.666390649934612e-06, "loss": 1.875, "step": 2978 }, { "epoch": 0.3415892672858617, "grad_norm": 16.875, "learning_rate": 7.664819595529096e-06, "loss": 2.0157, "step": 2979 }, { "epoch": 0.3417039330352024, "grad_norm": 16.625, "learning_rate": 7.663248173562514e-06, "loss": 1.74, "step": 2980 }, { "epoch": 0.34181859878454307, "grad_norm": 12.375, "learning_rate": 7.66167638425161e-06, "loss": 1.6329, "step": 2981 }, { "epoch": 0.34193326453388373, "grad_norm": 11.875, "learning_rate": 7.660104227813188e-06, "loss": 1.6007, "step": 2982 }, { "epoch": 0.3420479302832244, "grad_norm": 13.875, "learning_rate": 7.658531704464093e-06, "loss": 1.8859, "step": 2983 }, { "epoch": 0.34216259603256505, "grad_norm": 13.375, "learning_rate": 7.656958814421225e-06, "loss": 1.7149, "step": 2984 }, { "epoch": 0.34227726178190576, "grad_norm": 13.3125, "learning_rate": 7.655385557901534e-06, "loss": 1.7412, "step": 2985 }, { "epoch": 0.3423919275312464, "grad_norm": 13.0625, "learning_rate": 7.653811935122024e-06, "loss": 1.6261, "step": 2986 }, { "epoch": 0.3425065932805871, "grad_norm": 15.4375, "learning_rate": 7.652237946299742e-06, "loss": 2.1098, "step": 2987 }, { "epoch": 0.34262125902992774, "grad_norm": 13.1875, "learning_rate": 7.650663591651792e-06, "loss": 1.8534, "step": 2988 }, { "epoch": 0.34273592477926845, "grad_norm": 17.25, "learning_rate": 7.649088871395326e-06, "loss": 1.9291, "step": 2989 }, { "epoch": 0.3428505905286091, "grad_norm": 14.1875, "learning_rate": 7.647513785747545e-06, "loss": 2.3614, "step": 2990 }, { "epoch": 0.34296525627794977, "grad_norm": 16.0, "learning_rate": 7.645938334925704e-06, "loss": 1.5055, "step": 2991 }, { "epoch": 0.34307992202729043, "grad_norm": 15.5, "learning_rate": 7.644362519147107e-06, "loss": 1.7376, "step": 2992 }, { "epoch": 0.34319458777663114, "grad_norm": 15.4375, "learning_rate": 7.642786338629108e-06, "loss": 1.8174, "step": 2993 }, { "epoch": 0.3433092535259718, "grad_norm": 15.8125, "learning_rate": 7.641209793589106e-06, "loss": 1.5575, "step": 2994 }, { "epoch": 0.34342391927531246, "grad_norm": 13.8125, "learning_rate": 7.639632884244561e-06, "loss": 1.5696, "step": 2995 }, { "epoch": 0.3435385850246531, "grad_norm": 14.125, "learning_rate": 7.638055610812974e-06, "loss": 1.8939, "step": 2996 }, { "epoch": 0.34365325077399383, "grad_norm": 19.375, "learning_rate": 7.636477973511903e-06, "loss": 1.9928, "step": 2997 }, { "epoch": 0.3437679165233345, "grad_norm": 12.375, "learning_rate": 7.634899972558952e-06, "loss": 1.5525, "step": 2998 }, { "epoch": 0.34388258227267515, "grad_norm": 14.3125, "learning_rate": 7.633321608171775e-06, "loss": 1.7411, "step": 2999 }, { "epoch": 0.3439972480220158, "grad_norm": 12.75, "learning_rate": 7.631742880568076e-06, "loss": 2.0157, "step": 3000 }, { "epoch": 0.34411191377135647, "grad_norm": 14.625, "learning_rate": 7.630163789965614e-06, "loss": 1.5273, "step": 3001 }, { "epoch": 0.3442265795206972, "grad_norm": 15.1875, "learning_rate": 7.628584336582192e-06, "loss": 2.1128, "step": 3002 }, { "epoch": 0.34434124527003784, "grad_norm": 12.5625, "learning_rate": 7.627004520635667e-06, "loss": 1.5797, "step": 3003 }, { "epoch": 0.3444559110193785, "grad_norm": 12.9375, "learning_rate": 7.625424342343943e-06, "loss": 1.6774, "step": 3004 }, { "epoch": 0.34457057676871916, "grad_norm": 35.75, "learning_rate": 7.6238438019249775e-06, "loss": 1.9168, "step": 3005 }, { "epoch": 0.3446852425180599, "grad_norm": 13.625, "learning_rate": 7.622262899596773e-06, "loss": 1.4798, "step": 3006 }, { "epoch": 0.34479990826740053, "grad_norm": 16.875, "learning_rate": 7.620681635577386e-06, "loss": 1.3377, "step": 3007 }, { "epoch": 0.3449145740167412, "grad_norm": 11.75, "learning_rate": 7.619100010084923e-06, "loss": 1.8817, "step": 3008 }, { "epoch": 0.34502923976608185, "grad_norm": 14.1875, "learning_rate": 7.617518023337538e-06, "loss": 1.7043, "step": 3009 }, { "epoch": 0.34514390551542257, "grad_norm": 12.9375, "learning_rate": 7.615935675553436e-06, "loss": 1.5942, "step": 3010 }, { "epoch": 0.3452585712647632, "grad_norm": 15.5625, "learning_rate": 7.614352966950871e-06, "loss": 1.6524, "step": 3011 }, { "epoch": 0.3453732370141039, "grad_norm": 16.125, "learning_rate": 7.612769897748151e-06, "loss": 1.9823, "step": 3012 }, { "epoch": 0.34548790276344454, "grad_norm": 16.375, "learning_rate": 7.611186468163625e-06, "loss": 1.9007, "step": 3013 }, { "epoch": 0.34560256851278526, "grad_norm": 13.1875, "learning_rate": 7.6096026784157e-06, "loss": 1.4213, "step": 3014 }, { "epoch": 0.3457172342621259, "grad_norm": 21.125, "learning_rate": 7.608018528722829e-06, "loss": 2.1691, "step": 3015 }, { "epoch": 0.3458319000114666, "grad_norm": 17.875, "learning_rate": 7.606434019303514e-06, "loss": 1.6927, "step": 3016 }, { "epoch": 0.34594656576080723, "grad_norm": 15.625, "learning_rate": 7.604849150376312e-06, "loss": 1.6829, "step": 3017 }, { "epoch": 0.3460612315101479, "grad_norm": 11.0, "learning_rate": 7.603263922159822e-06, "loss": 1.6831, "step": 3018 }, { "epoch": 0.3461758972594886, "grad_norm": 12.25, "learning_rate": 7.6016783348726955e-06, "loss": 1.8965, "step": 3019 }, { "epoch": 0.34629056300882927, "grad_norm": 15.25, "learning_rate": 7.600092388733637e-06, "loss": 1.4425, "step": 3020 }, { "epoch": 0.3464052287581699, "grad_norm": 16.0, "learning_rate": 7.598506083961395e-06, "loss": 1.6929, "step": 3021 }, { "epoch": 0.3465198945075106, "grad_norm": 13.375, "learning_rate": 7.596919420774769e-06, "loss": 1.5797, "step": 3022 }, { "epoch": 0.3466345602568513, "grad_norm": 15.875, "learning_rate": 7.595332399392612e-06, "loss": 1.9654, "step": 3023 }, { "epoch": 0.34674922600619196, "grad_norm": 16.375, "learning_rate": 7.593745020033823e-06, "loss": 1.7768, "step": 3024 }, { "epoch": 0.3468638917555326, "grad_norm": 13.9375, "learning_rate": 7.5921572829173474e-06, "loss": 1.9561, "step": 3025 }, { "epoch": 0.3469785575048733, "grad_norm": 12.8125, "learning_rate": 7.590569188262187e-06, "loss": 1.8412, "step": 3026 }, { "epoch": 0.347093223254214, "grad_norm": 14.4375, "learning_rate": 7.588980736287389e-06, "loss": 1.761, "step": 3027 }, { "epoch": 0.34720788900355465, "grad_norm": 15.125, "learning_rate": 7.587391927212047e-06, "loss": 1.5956, "step": 3028 }, { "epoch": 0.3473225547528953, "grad_norm": 18.375, "learning_rate": 7.58580276125531e-06, "loss": 1.7316, "step": 3029 }, { "epoch": 0.34743722050223597, "grad_norm": 17.125, "learning_rate": 7.584213238636372e-06, "loss": 1.8816, "step": 3030 }, { "epoch": 0.3475518862515767, "grad_norm": 12.5625, "learning_rate": 7.582623359574477e-06, "loss": 1.7225, "step": 3031 }, { "epoch": 0.34766655200091734, "grad_norm": 13.5625, "learning_rate": 7.581033124288918e-06, "loss": 1.7862, "step": 3032 }, { "epoch": 0.347781217750258, "grad_norm": 14.625, "learning_rate": 7.579442532999039e-06, "loss": 1.7793, "step": 3033 }, { "epoch": 0.34789588349959866, "grad_norm": 14.3125, "learning_rate": 7.5778515859242315e-06, "loss": 1.9814, "step": 3034 }, { "epoch": 0.3480105492489393, "grad_norm": 36.75, "learning_rate": 7.576260283283936e-06, "loss": 1.9678, "step": 3035 }, { "epoch": 0.34812521499828003, "grad_norm": 14.75, "learning_rate": 7.574668625297642e-06, "loss": 1.7442, "step": 3036 }, { "epoch": 0.3482398807476207, "grad_norm": 12.625, "learning_rate": 7.57307661218489e-06, "loss": 1.6306, "step": 3037 }, { "epoch": 0.34835454649696135, "grad_norm": 13.0, "learning_rate": 7.571484244165266e-06, "loss": 1.7164, "step": 3038 }, { "epoch": 0.348469212246302, "grad_norm": 18.875, "learning_rate": 7.569891521458406e-06, "loss": 1.483, "step": 3039 }, { "epoch": 0.3485838779956427, "grad_norm": 13.875, "learning_rate": 7.568298444283999e-06, "loss": 1.5514, "step": 3040 }, { "epoch": 0.3486985437449834, "grad_norm": 13.25, "learning_rate": 7.566705012861777e-06, "loss": 1.7083, "step": 3041 }, { "epoch": 0.34881320949432404, "grad_norm": 12.8125, "learning_rate": 7.565111227411525e-06, "loss": 1.4956, "step": 3042 }, { "epoch": 0.3489278752436647, "grad_norm": 12.3125, "learning_rate": 7.563517088153074e-06, "loss": 1.7895, "step": 3043 }, { "epoch": 0.3490425409930054, "grad_norm": 11.625, "learning_rate": 7.561922595306306e-06, "loss": 1.8075, "step": 3044 }, { "epoch": 0.34915720674234607, "grad_norm": 14.1875, "learning_rate": 7.56032774909115e-06, "loss": 1.7913, "step": 3045 }, { "epoch": 0.34927187249168673, "grad_norm": 13.6875, "learning_rate": 7.5587325497275864e-06, "loss": 1.9325, "step": 3046 }, { "epoch": 0.3493865382410274, "grad_norm": 13.75, "learning_rate": 7.557136997435641e-06, "loss": 1.6051, "step": 3047 }, { "epoch": 0.3495012039903681, "grad_norm": 15.4375, "learning_rate": 7.55554109243539e-06, "loss": 1.6046, "step": 3048 }, { "epoch": 0.34961586973970876, "grad_norm": 14.25, "learning_rate": 7.553944834946959e-06, "loss": 1.7839, "step": 3049 }, { "epoch": 0.3497305354890494, "grad_norm": 15.3125, "learning_rate": 7.55234822519052e-06, "loss": 1.6834, "step": 3050 }, { "epoch": 0.3498452012383901, "grad_norm": 19.375, "learning_rate": 7.550751263386296e-06, "loss": 2.0535, "step": 3051 }, { "epoch": 0.34995986698773074, "grad_norm": 28.25, "learning_rate": 7.5491539497545575e-06, "loss": 1.6411, "step": 3052 }, { "epoch": 0.35007453273707145, "grad_norm": 16.5, "learning_rate": 7.547556284515622e-06, "loss": 2.0821, "step": 3053 }, { "epoch": 0.3501891984864121, "grad_norm": 17.25, "learning_rate": 7.545958267889857e-06, "loss": 1.4889, "step": 3054 }, { "epoch": 0.35030386423575277, "grad_norm": 19.75, "learning_rate": 7.544359900097681e-06, "loss": 1.6943, "step": 3055 }, { "epoch": 0.35041852998509343, "grad_norm": 15.125, "learning_rate": 7.542761181359556e-06, "loss": 1.8734, "step": 3056 }, { "epoch": 0.35053319573443414, "grad_norm": 11.75, "learning_rate": 7.541162111895994e-06, "loss": 1.4733, "step": 3057 }, { "epoch": 0.3506478614837748, "grad_norm": 12.1875, "learning_rate": 7.539562691927561e-06, "loss": 1.4715, "step": 3058 }, { "epoch": 0.35076252723311546, "grad_norm": 14.3125, "learning_rate": 7.537962921674862e-06, "loss": 2.0216, "step": 3059 }, { "epoch": 0.3508771929824561, "grad_norm": 27.5, "learning_rate": 7.5363628013585546e-06, "loss": 2.0123, "step": 3060 }, { "epoch": 0.35099185873179684, "grad_norm": 16.875, "learning_rate": 7.534762331199346e-06, "loss": 1.8057, "step": 3061 }, { "epoch": 0.3511065244811375, "grad_norm": 16.125, "learning_rate": 7.533161511417992e-06, "loss": 1.9723, "step": 3062 }, { "epoch": 0.35122119023047815, "grad_norm": 13.6875, "learning_rate": 7.531560342235293e-06, "loss": 1.7726, "step": 3063 }, { "epoch": 0.3513358559798188, "grad_norm": 13.125, "learning_rate": 7.529958823872101e-06, "loss": 1.8035, "step": 3064 }, { "epoch": 0.3514505217291595, "grad_norm": 18.375, "learning_rate": 7.528356956549316e-06, "loss": 1.9445, "step": 3065 }, { "epoch": 0.3515651874785002, "grad_norm": 13.75, "learning_rate": 7.526754740487882e-06, "loss": 1.4841, "step": 3066 }, { "epoch": 0.35167985322784084, "grad_norm": 31.625, "learning_rate": 7.525152175908796e-06, "loss": 1.6768, "step": 3067 }, { "epoch": 0.3517945189771815, "grad_norm": 18.5, "learning_rate": 7.5235492630331035e-06, "loss": 1.9627, "step": 3068 }, { "epoch": 0.35190918472652216, "grad_norm": 14.9375, "learning_rate": 7.52194600208189e-06, "loss": 1.7, "step": 3069 }, { "epoch": 0.3520238504758629, "grad_norm": 17.75, "learning_rate": 7.520342393276301e-06, "loss": 1.4345, "step": 3070 }, { "epoch": 0.35213851622520354, "grad_norm": 12.375, "learning_rate": 7.518738436837518e-06, "loss": 1.6509, "step": 3071 }, { "epoch": 0.3522531819745442, "grad_norm": 13.1875, "learning_rate": 7.517134132986781e-06, "loss": 1.5312, "step": 3072 }, { "epoch": 0.35236784772388485, "grad_norm": 13.4375, "learning_rate": 7.515529481945372e-06, "loss": 1.5915, "step": 3073 }, { "epoch": 0.35248251347322557, "grad_norm": 12.5625, "learning_rate": 7.5139244839346205e-06, "loss": 1.9707, "step": 3074 }, { "epoch": 0.3525971792225662, "grad_norm": 17.125, "learning_rate": 7.512319139175905e-06, "loss": 1.8454, "step": 3075 }, { "epoch": 0.3527118449719069, "grad_norm": 14.25, "learning_rate": 7.510713447890654e-06, "loss": 1.5967, "step": 3076 }, { "epoch": 0.35282651072124754, "grad_norm": 10.875, "learning_rate": 7.509107410300343e-06, "loss": 1.4895, "step": 3077 }, { "epoch": 0.35294117647058826, "grad_norm": 13.875, "learning_rate": 7.507501026626492e-06, "loss": 1.6639, "step": 3078 }, { "epoch": 0.3530558422199289, "grad_norm": 16.25, "learning_rate": 7.50589429709067e-06, "loss": 2.1339, "step": 3079 }, { "epoch": 0.3531705079692696, "grad_norm": 13.75, "learning_rate": 7.504287221914499e-06, "loss": 1.9342, "step": 3080 }, { "epoch": 0.35328517371861023, "grad_norm": 13.125, "learning_rate": 7.5026798013196415e-06, "loss": 1.5644, "step": 3081 }, { "epoch": 0.35339983946795095, "grad_norm": 13.875, "learning_rate": 7.501072035527808e-06, "loss": 1.7492, "step": 3082 }, { "epoch": 0.3535145052172916, "grad_norm": 29.0, "learning_rate": 7.499463924760765e-06, "loss": 1.8399, "step": 3083 }, { "epoch": 0.35362917096663227, "grad_norm": 12.1875, "learning_rate": 7.497855469240316e-06, "loss": 1.7119, "step": 3084 }, { "epoch": 0.3537438367159729, "grad_norm": 14.625, "learning_rate": 7.496246669188318e-06, "loss": 1.7864, "step": 3085 }, { "epoch": 0.3538585024653136, "grad_norm": 15.9375, "learning_rate": 7.494637524826677e-06, "loss": 1.8738, "step": 3086 }, { "epoch": 0.3539731682146543, "grad_norm": 14.5, "learning_rate": 7.493028036377341e-06, "loss": 1.4998, "step": 3087 }, { "epoch": 0.35408783396399496, "grad_norm": 15.25, "learning_rate": 7.491418204062307e-06, "loss": 1.7566, "step": 3088 }, { "epoch": 0.3542024997133356, "grad_norm": 12.875, "learning_rate": 7.489808028103623e-06, "loss": 1.5848, "step": 3089 }, { "epoch": 0.3543171654626763, "grad_norm": 12.5, "learning_rate": 7.4881975087233835e-06, "loss": 1.7057, "step": 3090 }, { "epoch": 0.354431831212017, "grad_norm": 21.875, "learning_rate": 7.486586646143725e-06, "loss": 1.8447, "step": 3091 }, { "epoch": 0.35454649696135765, "grad_norm": 14.5, "learning_rate": 7.484975440586839e-06, "loss": 1.6141, "step": 3092 }, { "epoch": 0.3546611627106983, "grad_norm": 15.125, "learning_rate": 7.483363892274959e-06, "loss": 1.661, "step": 3093 }, { "epoch": 0.35477582846003897, "grad_norm": 12.75, "learning_rate": 7.481752001430365e-06, "loss": 1.5765, "step": 3094 }, { "epoch": 0.3548904942093797, "grad_norm": 12.3125, "learning_rate": 7.48013976827539e-06, "loss": 1.831, "step": 3095 }, { "epoch": 0.35500515995872034, "grad_norm": 13.6875, "learning_rate": 7.4785271930324095e-06, "loss": 1.7832, "step": 3096 }, { "epoch": 0.355119825708061, "grad_norm": 10.9375, "learning_rate": 7.4769142759238476e-06, "loss": 1.8321, "step": 3097 }, { "epoch": 0.35523449145740166, "grad_norm": 14.1875, "learning_rate": 7.475301017172177e-06, "loss": 2.3485, "step": 3098 }, { "epoch": 0.35534915720674237, "grad_norm": 17.125, "learning_rate": 7.473687416999914e-06, "loss": 2.0988, "step": 3099 }, { "epoch": 0.35546382295608303, "grad_norm": 15.9375, "learning_rate": 7.472073475629625e-06, "loss": 2.0065, "step": 3100 }, { "epoch": 0.3555784887054237, "grad_norm": 14.3125, "learning_rate": 7.47045919328392e-06, "loss": 2.0963, "step": 3101 }, { "epoch": 0.35569315445476435, "grad_norm": 18.0, "learning_rate": 7.468844570185463e-06, "loss": 1.7721, "step": 3102 }, { "epoch": 0.355807820204105, "grad_norm": 17.25, "learning_rate": 7.467229606556956e-06, "loss": 1.7675, "step": 3103 }, { "epoch": 0.3559224859534457, "grad_norm": 17.625, "learning_rate": 7.465614302621156e-06, "loss": 2.187, "step": 3104 }, { "epoch": 0.3560371517027864, "grad_norm": 14.0625, "learning_rate": 7.4639986586008615e-06, "loss": 1.6409, "step": 3105 }, { "epoch": 0.35615181745212704, "grad_norm": 13.0, "learning_rate": 7.46238267471892e-06, "loss": 1.7466, "step": 3106 }, { "epoch": 0.3562664832014677, "grad_norm": 14.75, "learning_rate": 7.460766351198225e-06, "loss": 1.8377, "step": 3107 }, { "epoch": 0.3563811489508084, "grad_norm": 15.3125, "learning_rate": 7.45914968826172e-06, "loss": 2.0118, "step": 3108 }, { "epoch": 0.35649581470014907, "grad_norm": 13.0, "learning_rate": 7.457532686132389e-06, "loss": 1.6495, "step": 3109 }, { "epoch": 0.35661048044948973, "grad_norm": 13.375, "learning_rate": 7.455915345033271e-06, "loss": 1.7525, "step": 3110 }, { "epoch": 0.3567251461988304, "grad_norm": 13.875, "learning_rate": 7.4542976651874436e-06, "loss": 1.7059, "step": 3111 }, { "epoch": 0.3568398119481711, "grad_norm": 15.0, "learning_rate": 7.452679646818037e-06, "loss": 1.7435, "step": 3112 }, { "epoch": 0.35695447769751176, "grad_norm": 15.3125, "learning_rate": 7.4510612901482245e-06, "loss": 1.8015, "step": 3113 }, { "epoch": 0.3570691434468524, "grad_norm": 26.25, "learning_rate": 7.449442595401231e-06, "loss": 1.8807, "step": 3114 }, { "epoch": 0.3571838091961931, "grad_norm": 14.75, "learning_rate": 7.4478235628003206e-06, "loss": 1.6057, "step": 3115 }, { "epoch": 0.3572984749455338, "grad_norm": 11.625, "learning_rate": 7.446204192568808e-06, "loss": 1.8985, "step": 3116 }, { "epoch": 0.35741314069487445, "grad_norm": 15.4375, "learning_rate": 7.444584484930057e-06, "loss": 1.3893, "step": 3117 }, { "epoch": 0.3575278064442151, "grad_norm": 15.9375, "learning_rate": 7.442964440107476e-06, "loss": 1.8573, "step": 3118 }, { "epoch": 0.35764247219355577, "grad_norm": 14.5625, "learning_rate": 7.441344058324516e-06, "loss": 1.9322, "step": 3119 }, { "epoch": 0.35775713794289643, "grad_norm": 11.6875, "learning_rate": 7.43972333980468e-06, "loss": 1.35, "step": 3120 }, { "epoch": 0.35787180369223714, "grad_norm": 12.5625, "learning_rate": 7.438102284771513e-06, "loss": 1.4554, "step": 3121 }, { "epoch": 0.3579864694415778, "grad_norm": 12.375, "learning_rate": 7.436480893448612e-06, "loss": 1.7899, "step": 3122 }, { "epoch": 0.35810113519091846, "grad_norm": 19.25, "learning_rate": 7.434859166059616e-06, "loss": 1.7001, "step": 3123 }, { "epoch": 0.3582158009402591, "grad_norm": 15.1875, "learning_rate": 7.433237102828209e-06, "loss": 1.8953, "step": 3124 }, { "epoch": 0.35833046668959984, "grad_norm": 17.25, "learning_rate": 7.431614703978125e-06, "loss": 1.665, "step": 3125 }, { "epoch": 0.3584451324389405, "grad_norm": 15.3125, "learning_rate": 7.429991969733144e-06, "loss": 1.6173, "step": 3126 }, { "epoch": 0.35855979818828115, "grad_norm": 13.0, "learning_rate": 7.428368900317092e-06, "loss": 2.0085, "step": 3127 }, { "epoch": 0.3586744639376218, "grad_norm": 16.75, "learning_rate": 7.426745495953838e-06, "loss": 1.6267, "step": 3128 }, { "epoch": 0.3587891296869625, "grad_norm": 15.25, "learning_rate": 7.425121756867301e-06, "loss": 1.8045, "step": 3129 }, { "epoch": 0.3589037954363032, "grad_norm": 14.25, "learning_rate": 7.423497683281444e-06, "loss": 1.7895, "step": 3130 }, { "epoch": 0.35901846118564384, "grad_norm": 13.75, "learning_rate": 7.421873275420278e-06, "loss": 2.0309, "step": 3131 }, { "epoch": 0.3591331269349845, "grad_norm": 13.625, "learning_rate": 7.420248533507859e-06, "loss": 1.5656, "step": 3132 }, { "epoch": 0.3592477926843252, "grad_norm": 14.3125, "learning_rate": 7.4186234577682905e-06, "loss": 1.76, "step": 3133 }, { "epoch": 0.3593624584336659, "grad_norm": 11.25, "learning_rate": 7.416998048425716e-06, "loss": 1.5711, "step": 3134 }, { "epoch": 0.35947712418300654, "grad_norm": 18.875, "learning_rate": 7.415372305704335e-06, "loss": 2.0967, "step": 3135 }, { "epoch": 0.3595917899323472, "grad_norm": 15.5625, "learning_rate": 7.413746229828384e-06, "loss": 1.9477, "step": 3136 }, { "epoch": 0.35970645568168785, "grad_norm": 12.25, "learning_rate": 7.412119821022154e-06, "loss": 1.6112, "step": 3137 }, { "epoch": 0.35982112143102857, "grad_norm": 14.375, "learning_rate": 7.41049307950997e-06, "loss": 2.2977, "step": 3138 }, { "epoch": 0.3599357871803692, "grad_norm": 13.5625, "learning_rate": 7.4088660055162155e-06, "loss": 1.9388, "step": 3139 }, { "epoch": 0.3600504529297099, "grad_norm": 15.1875, "learning_rate": 7.407238599265314e-06, "loss": 1.7029, "step": 3140 }, { "epoch": 0.36016511867905054, "grad_norm": 15.5, "learning_rate": 7.405610860981731e-06, "loss": 1.8924, "step": 3141 }, { "epoch": 0.36027978442839126, "grad_norm": 14.1875, "learning_rate": 7.403982790889988e-06, "loss": 1.8179, "step": 3142 }, { "epoch": 0.3603944501777319, "grad_norm": 15.125, "learning_rate": 7.4023543892146425e-06, "loss": 1.9491, "step": 3143 }, { "epoch": 0.3605091159270726, "grad_norm": 11.4375, "learning_rate": 7.400725656180299e-06, "loss": 1.5608, "step": 3144 }, { "epoch": 0.36062378167641324, "grad_norm": 15.6875, "learning_rate": 7.3990965920116166e-06, "loss": 1.6007, "step": 3145 }, { "epoch": 0.36073844742575395, "grad_norm": 12.4375, "learning_rate": 7.397467196933288e-06, "loss": 1.6177, "step": 3146 }, { "epoch": 0.3608531131750946, "grad_norm": 14.625, "learning_rate": 7.395837471170061e-06, "loss": 1.6193, "step": 3147 }, { "epoch": 0.36096777892443527, "grad_norm": 13.25, "learning_rate": 7.394207414946724e-06, "loss": 1.7783, "step": 3148 }, { "epoch": 0.3610824446737759, "grad_norm": 15.5625, "learning_rate": 7.392577028488111e-06, "loss": 2.0704, "step": 3149 }, { "epoch": 0.36119711042311664, "grad_norm": 14.3125, "learning_rate": 7.390946312019102e-06, "loss": 1.8435, "step": 3150 }, { "epoch": 0.3613117761724573, "grad_norm": 12.25, "learning_rate": 7.389315265764627e-06, "loss": 1.4544, "step": 3151 }, { "epoch": 0.36142644192179796, "grad_norm": 14.0625, "learning_rate": 7.387683889949655e-06, "loss": 1.6137, "step": 3152 }, { "epoch": 0.3615411076711386, "grad_norm": 14.375, "learning_rate": 7.386052184799204e-06, "loss": 1.6117, "step": 3153 }, { "epoch": 0.3616557734204793, "grad_norm": 13.4375, "learning_rate": 7.3844201505383365e-06, "loss": 1.84, "step": 3154 }, { "epoch": 0.36177043916982, "grad_norm": 12.8125, "learning_rate": 7.382787787392161e-06, "loss": 1.39, "step": 3155 }, { "epoch": 0.36188510491916065, "grad_norm": 16.125, "learning_rate": 7.381155095585828e-06, "loss": 1.7776, "step": 3156 }, { "epoch": 0.3619997706685013, "grad_norm": 32.75, "learning_rate": 7.379522075344541e-06, "loss": 1.9311, "step": 3157 }, { "epoch": 0.36211443641784197, "grad_norm": 11.625, "learning_rate": 7.37788872689354e-06, "loss": 2.071, "step": 3158 }, { "epoch": 0.3622291021671827, "grad_norm": 14.3125, "learning_rate": 7.376255050458116e-06, "loss": 2.3252, "step": 3159 }, { "epoch": 0.36234376791652334, "grad_norm": 35.5, "learning_rate": 7.374621046263603e-06, "loss": 1.9547, "step": 3160 }, { "epoch": 0.362458433665864, "grad_norm": 14.75, "learning_rate": 7.372986714535382e-06, "loss": 2.0164, "step": 3161 }, { "epoch": 0.36257309941520466, "grad_norm": 14.5, "learning_rate": 7.3713520554988775e-06, "loss": 1.8021, "step": 3162 }, { "epoch": 0.3626877651645454, "grad_norm": 14.0, "learning_rate": 7.369717069379558e-06, "loss": 2.0745, "step": 3163 }, { "epoch": 0.36280243091388603, "grad_norm": 12.4375, "learning_rate": 7.36808175640294e-06, "loss": 1.4981, "step": 3164 }, { "epoch": 0.3629170966632267, "grad_norm": 12.9375, "learning_rate": 7.366446116794583e-06, "loss": 1.6413, "step": 3165 }, { "epoch": 0.36303176241256735, "grad_norm": 16.25, "learning_rate": 7.364810150780092e-06, "loss": 1.9228, "step": 3166 }, { "epoch": 0.36314642816190806, "grad_norm": 14.3125, "learning_rate": 7.36317385858512e-06, "loss": 1.8478, "step": 3167 }, { "epoch": 0.3632610939112487, "grad_norm": 15.625, "learning_rate": 7.36153724043536e-06, "loss": 1.7916, "step": 3168 }, { "epoch": 0.3633757596605894, "grad_norm": 14.1875, "learning_rate": 7.359900296556551e-06, "loss": 1.8, "step": 3169 }, { "epoch": 0.36349042540993004, "grad_norm": 15.375, "learning_rate": 7.358263027174481e-06, "loss": 1.5198, "step": 3170 }, { "epoch": 0.3636050911592707, "grad_norm": 15.625, "learning_rate": 7.35662543251498e-06, "loss": 1.6746, "step": 3171 }, { "epoch": 0.3637197569086114, "grad_norm": 15.5625, "learning_rate": 7.35498751280392e-06, "loss": 2.0675, "step": 3172 }, { "epoch": 0.3638344226579521, "grad_norm": 30.875, "learning_rate": 7.3533492682672245e-06, "loss": 1.7013, "step": 3173 }, { "epoch": 0.36394908840729273, "grad_norm": 15.5, "learning_rate": 7.351710699130856e-06, "loss": 2.1303, "step": 3174 }, { "epoch": 0.3640637541566334, "grad_norm": 36.25, "learning_rate": 7.350071805620823e-06, "loss": 1.6751, "step": 3175 }, { "epoch": 0.3641784199059741, "grad_norm": 12.5625, "learning_rate": 7.348432587963181e-06, "loss": 1.3954, "step": 3176 }, { "epoch": 0.36429308565531476, "grad_norm": 13.9375, "learning_rate": 7.346793046384031e-06, "loss": 1.9825, "step": 3177 }, { "epoch": 0.3644077514046554, "grad_norm": 26.875, "learning_rate": 7.345153181109512e-06, "loss": 2.0764, "step": 3178 }, { "epoch": 0.3645224171539961, "grad_norm": 35.0, "learning_rate": 7.3435129923658155e-06, "loss": 1.7405, "step": 3179 }, { "epoch": 0.3646370829033368, "grad_norm": 13.6875, "learning_rate": 7.341872480379173e-06, "loss": 1.6589, "step": 3180 }, { "epoch": 0.36475174865267745, "grad_norm": 38.75, "learning_rate": 7.3402316453758615e-06, "loss": 2.038, "step": 3181 }, { "epoch": 0.3648664144020181, "grad_norm": 13.625, "learning_rate": 7.338590487582202e-06, "loss": 1.8398, "step": 3182 }, { "epoch": 0.36498108015135877, "grad_norm": 13.75, "learning_rate": 7.336949007224565e-06, "loss": 1.4953, "step": 3183 }, { "epoch": 0.3650957459006995, "grad_norm": 19.75, "learning_rate": 7.335307204529357e-06, "loss": 1.7229, "step": 3184 }, { "epoch": 0.36521041165004015, "grad_norm": 11.125, "learning_rate": 7.333665079723036e-06, "loss": 1.5011, "step": 3185 }, { "epoch": 0.3653250773993808, "grad_norm": 14.75, "learning_rate": 7.3320226330321005e-06, "loss": 1.8446, "step": 3186 }, { "epoch": 0.36543974314872146, "grad_norm": 11.5, "learning_rate": 7.330379864683096e-06, "loss": 1.5418, "step": 3187 }, { "epoch": 0.3655544088980621, "grad_norm": 30.375, "learning_rate": 7.32873677490261e-06, "loss": 1.5984, "step": 3188 }, { "epoch": 0.36566907464740284, "grad_norm": 13.875, "learning_rate": 7.327093363917275e-06, "loss": 1.5302, "step": 3189 }, { "epoch": 0.3657837403967435, "grad_norm": 11.5, "learning_rate": 7.32544963195377e-06, "loss": 1.4019, "step": 3190 }, { "epoch": 0.36589840614608415, "grad_norm": 15.0, "learning_rate": 7.323805579238812e-06, "loss": 1.5373, "step": 3191 }, { "epoch": 0.3660130718954248, "grad_norm": 15.375, "learning_rate": 7.322161205999173e-06, "loss": 1.4716, "step": 3192 }, { "epoch": 0.3661277376447655, "grad_norm": 12.5625, "learning_rate": 7.32051651246166e-06, "loss": 1.7023, "step": 3193 }, { "epoch": 0.3662424033941062, "grad_norm": 18.375, "learning_rate": 7.318871498853127e-06, "loss": 1.7628, "step": 3194 }, { "epoch": 0.36635706914344685, "grad_norm": 13.625, "learning_rate": 7.317226165400473e-06, "loss": 2.0057, "step": 3195 }, { "epoch": 0.3664717348927875, "grad_norm": 15.875, "learning_rate": 7.315580512330638e-06, "loss": 1.3622, "step": 3196 }, { "epoch": 0.3665864006421282, "grad_norm": 13.3125, "learning_rate": 7.3139345398706105e-06, "loss": 1.7, "step": 3197 }, { "epoch": 0.3667010663914689, "grad_norm": 15.75, "learning_rate": 7.312288248247423e-06, "loss": 2.1008, "step": 3198 }, { "epoch": 0.36681573214080954, "grad_norm": 13.625, "learning_rate": 7.3106416376881475e-06, "loss": 1.4895, "step": 3199 }, { "epoch": 0.3669303978901502, "grad_norm": 29.75, "learning_rate": 7.308994708419901e-06, "loss": 1.7375, "step": 3200 }, { "epoch": 0.3670450636394909, "grad_norm": 12.875, "learning_rate": 7.307347460669849e-06, "loss": 1.9692, "step": 3201 }, { "epoch": 0.36715972938883157, "grad_norm": 11.3125, "learning_rate": 7.305699894665197e-06, "loss": 1.6386, "step": 3202 }, { "epoch": 0.3672743951381722, "grad_norm": 13.5, "learning_rate": 7.304052010633195e-06, "loss": 2.2538, "step": 3203 }, { "epoch": 0.3673890608875129, "grad_norm": 18.25, "learning_rate": 7.302403808801137e-06, "loss": 1.9073, "step": 3204 }, { "epoch": 0.36750372663685354, "grad_norm": 17.125, "learning_rate": 7.300755289396362e-06, "loss": 2.0479, "step": 3205 }, { "epoch": 0.36761839238619426, "grad_norm": 13.8125, "learning_rate": 7.299106452646251e-06, "loss": 1.7223, "step": 3206 }, { "epoch": 0.3677330581355349, "grad_norm": 14.3125, "learning_rate": 7.29745729877823e-06, "loss": 1.8265, "step": 3207 }, { "epoch": 0.3678477238848756, "grad_norm": 21.375, "learning_rate": 7.295807828019768e-06, "loss": 2.0605, "step": 3208 }, { "epoch": 0.36796238963421624, "grad_norm": 13.5, "learning_rate": 7.294158040598378e-06, "loss": 1.8383, "step": 3209 }, { "epoch": 0.36807705538355695, "grad_norm": 17.125, "learning_rate": 7.292507936741616e-06, "loss": 2.0944, "step": 3210 }, { "epoch": 0.3681917211328976, "grad_norm": 13.6875, "learning_rate": 7.290857516677085e-06, "loss": 1.6654, "step": 3211 }, { "epoch": 0.36830638688223827, "grad_norm": 14.5, "learning_rate": 7.289206780632427e-06, "loss": 1.8738, "step": 3212 }, { "epoch": 0.3684210526315789, "grad_norm": 17.25, "learning_rate": 7.287555728835329e-06, "loss": 1.7319, "step": 3213 }, { "epoch": 0.36853571838091964, "grad_norm": 12.3125, "learning_rate": 7.285904361513523e-06, "loss": 1.3318, "step": 3214 }, { "epoch": 0.3686503841302603, "grad_norm": 12.875, "learning_rate": 7.284252678894786e-06, "loss": 1.6713, "step": 3215 }, { "epoch": 0.36876504987960096, "grad_norm": 14.4375, "learning_rate": 7.282600681206929e-06, "loss": 1.9564, "step": 3216 }, { "epoch": 0.3688797156289416, "grad_norm": 14.0, "learning_rate": 7.280948368677823e-06, "loss": 1.8518, "step": 3217 }, { "epoch": 0.36899438137828233, "grad_norm": 14.1875, "learning_rate": 7.279295741535367e-06, "loss": 1.8536, "step": 3218 }, { "epoch": 0.369109047127623, "grad_norm": 15.875, "learning_rate": 7.27764280000751e-06, "loss": 1.4035, "step": 3219 }, { "epoch": 0.36922371287696365, "grad_norm": 12.375, "learning_rate": 7.275989544322245e-06, "loss": 1.7958, "step": 3220 }, { "epoch": 0.3693383786263043, "grad_norm": 34.75, "learning_rate": 7.274335974707606e-06, "loss": 1.8018, "step": 3221 }, { "epoch": 0.36945304437564497, "grad_norm": 15.9375, "learning_rate": 7.272682091391672e-06, "loss": 1.8612, "step": 3222 }, { "epoch": 0.3695677101249857, "grad_norm": 13.0625, "learning_rate": 7.271027894602567e-06, "loss": 1.6398, "step": 3223 }, { "epoch": 0.36968237587432634, "grad_norm": 14.5625, "learning_rate": 7.269373384568451e-06, "loss": 1.768, "step": 3224 }, { "epoch": 0.369797041623667, "grad_norm": 13.875, "learning_rate": 7.267718561517536e-06, "loss": 1.6337, "step": 3225 }, { "epoch": 0.36991170737300766, "grad_norm": 13.625, "learning_rate": 7.2660634256780714e-06, "loss": 1.4305, "step": 3226 }, { "epoch": 0.3700263731223484, "grad_norm": 11.5625, "learning_rate": 7.2644079772783535e-06, "loss": 1.8722, "step": 3227 }, { "epoch": 0.37014103887168903, "grad_norm": 13.0625, "learning_rate": 7.262752216546718e-06, "loss": 2.066, "step": 3228 }, { "epoch": 0.3702557046210297, "grad_norm": 14.5, "learning_rate": 7.261096143711546e-06, "loss": 1.7721, "step": 3229 }, { "epoch": 0.37037037037037035, "grad_norm": 39.25, "learning_rate": 7.259439759001262e-06, "loss": 2.154, "step": 3230 }, { "epoch": 0.37048503611971106, "grad_norm": 15.25, "learning_rate": 7.257783062644332e-06, "loss": 1.9925, "step": 3231 }, { "epoch": 0.3705997018690517, "grad_norm": 13.8125, "learning_rate": 7.256126054869266e-06, "loss": 1.8567, "step": 3232 }, { "epoch": 0.3707143676183924, "grad_norm": 15.0625, "learning_rate": 7.254468735904616e-06, "loss": 1.8666, "step": 3233 }, { "epoch": 0.37082903336773304, "grad_norm": 12.9375, "learning_rate": 7.252811105978977e-06, "loss": 1.6483, "step": 3234 }, { "epoch": 0.37094369911707376, "grad_norm": 11.8125, "learning_rate": 7.25115316532099e-06, "loss": 1.7761, "step": 3235 }, { "epoch": 0.3710583648664144, "grad_norm": 13.3125, "learning_rate": 7.249494914159333e-06, "loss": 1.7207, "step": 3236 }, { "epoch": 0.3711730306157551, "grad_norm": 26.125, "learning_rate": 7.2478363527227334e-06, "loss": 1.9387, "step": 3237 }, { "epoch": 0.37128769636509573, "grad_norm": 12.9375, "learning_rate": 7.246177481239957e-06, "loss": 1.7654, "step": 3238 }, { "epoch": 0.3714023621144364, "grad_norm": 29.375, "learning_rate": 7.244518299939811e-06, "loss": 1.6532, "step": 3239 }, { "epoch": 0.3715170278637771, "grad_norm": 15.375, "learning_rate": 7.242858809051152e-06, "loss": 1.7235, "step": 3240 }, { "epoch": 0.37163169361311776, "grad_norm": 23.0, "learning_rate": 7.24119900880287e-06, "loss": 1.848, "step": 3241 }, { "epoch": 0.3717463593624584, "grad_norm": 16.125, "learning_rate": 7.23953889942391e-06, "loss": 1.865, "step": 3242 }, { "epoch": 0.3718610251117991, "grad_norm": 13.8125, "learning_rate": 7.237878481143247e-06, "loss": 2.0015, "step": 3243 }, { "epoch": 0.3719756908611398, "grad_norm": 37.5, "learning_rate": 7.236217754189904e-06, "loss": 1.9194, "step": 3244 }, { "epoch": 0.37209035661048045, "grad_norm": 15.9375, "learning_rate": 7.234556718792949e-06, "loss": 1.6357, "step": 3245 }, { "epoch": 0.3722050223598211, "grad_norm": 14.1875, "learning_rate": 7.232895375181489e-06, "loss": 1.8089, "step": 3246 }, { "epoch": 0.3723196881091618, "grad_norm": 16.375, "learning_rate": 7.231233723584674e-06, "loss": 1.839, "step": 3247 }, { "epoch": 0.3724343538585025, "grad_norm": 19.0, "learning_rate": 7.2295717642317e-06, "loss": 1.83, "step": 3248 }, { "epoch": 0.37254901960784315, "grad_norm": 17.5, "learning_rate": 7.227909497351799e-06, "loss": 1.6101, "step": 3249 }, { "epoch": 0.3726636853571838, "grad_norm": 17.5, "learning_rate": 7.22624692317425e-06, "loss": 2.002, "step": 3250 }, { "epoch": 0.37277835110652446, "grad_norm": 14.875, "learning_rate": 7.2245840419283755e-06, "loss": 1.4881, "step": 3251 }, { "epoch": 0.3728930168558652, "grad_norm": 14.0, "learning_rate": 7.222920853843538e-06, "loss": 1.8584, "step": 3252 }, { "epoch": 0.37300768260520584, "grad_norm": 13.75, "learning_rate": 7.221257359149139e-06, "loss": 1.6236, "step": 3253 }, { "epoch": 0.3731223483545465, "grad_norm": 13.8125, "learning_rate": 7.219593558074629e-06, "loss": 2.0745, "step": 3254 }, { "epoch": 0.37323701410388715, "grad_norm": 14.0, "learning_rate": 7.217929450849498e-06, "loss": 1.6203, "step": 3255 }, { "epoch": 0.3733516798532278, "grad_norm": 14.75, "learning_rate": 7.216265037703276e-06, "loss": 2.1936, "step": 3256 }, { "epoch": 0.37346634560256853, "grad_norm": 13.5625, "learning_rate": 7.214600318865538e-06, "loss": 1.5537, "step": 3257 }, { "epoch": 0.3735810113519092, "grad_norm": 13.375, "learning_rate": 7.2129352945659005e-06, "loss": 1.4815, "step": 3258 }, { "epoch": 0.37369567710124985, "grad_norm": 16.0, "learning_rate": 7.211269965034021e-06, "loss": 1.6052, "step": 3259 }, { "epoch": 0.3738103428505905, "grad_norm": 13.875, "learning_rate": 7.2096043304996e-06, "loss": 1.5757, "step": 3260 }, { "epoch": 0.3739250085999312, "grad_norm": 15.4375, "learning_rate": 7.207938391192381e-06, "loss": 1.9239, "step": 3261 }, { "epoch": 0.3740396743492719, "grad_norm": 16.875, "learning_rate": 7.206272147342149e-06, "loss": 2.1384, "step": 3262 }, { "epoch": 0.37415434009861254, "grad_norm": 16.875, "learning_rate": 7.204605599178729e-06, "loss": 1.9898, "step": 3263 }, { "epoch": 0.3742690058479532, "grad_norm": 16.125, "learning_rate": 7.2029387469319885e-06, "loss": 1.7973, "step": 3264 }, { "epoch": 0.3743836715972939, "grad_norm": 36.25, "learning_rate": 7.201271590831842e-06, "loss": 1.6706, "step": 3265 }, { "epoch": 0.37449833734663457, "grad_norm": 18.75, "learning_rate": 7.199604131108238e-06, "loss": 1.7551, "step": 3266 }, { "epoch": 0.3746130030959752, "grad_norm": 12.625, "learning_rate": 7.197936367991174e-06, "loss": 1.6198, "step": 3267 }, { "epoch": 0.3747276688453159, "grad_norm": 51.75, "learning_rate": 7.1962683017106845e-06, "loss": 1.6013, "step": 3268 }, { "epoch": 0.3748423345946566, "grad_norm": 12.375, "learning_rate": 7.194599932496845e-06, "loss": 1.6277, "step": 3269 }, { "epoch": 0.37495700034399726, "grad_norm": 13.125, "learning_rate": 7.192931260579781e-06, "loss": 1.5412, "step": 3270 }, { "epoch": 0.3750716660933379, "grad_norm": 14.375, "learning_rate": 7.191262286189649e-06, "loss": 1.6317, "step": 3271 }, { "epoch": 0.3751863318426786, "grad_norm": 15.5, "learning_rate": 7.189593009556652e-06, "loss": 1.8796, "step": 3272 }, { "epoch": 0.37530099759201924, "grad_norm": 12.8125, "learning_rate": 7.187923430911041e-06, "loss": 1.7583, "step": 3273 }, { "epoch": 0.37541566334135995, "grad_norm": 14.75, "learning_rate": 7.186253550483095e-06, "loss": 1.9272, "step": 3274 }, { "epoch": 0.3755303290907006, "grad_norm": 14.5, "learning_rate": 7.1845833685031465e-06, "loss": 2.0528, "step": 3275 }, { "epoch": 0.37564499484004127, "grad_norm": 15.5, "learning_rate": 7.182912885201563e-06, "loss": 1.6127, "step": 3276 }, { "epoch": 0.3757596605893819, "grad_norm": 14.9375, "learning_rate": 7.1812421008087595e-06, "loss": 1.4969, "step": 3277 }, { "epoch": 0.37587432633872264, "grad_norm": 13.0, "learning_rate": 7.179571015555185e-06, "loss": 1.7038, "step": 3278 }, { "epoch": 0.3759889920880633, "grad_norm": 14.0625, "learning_rate": 7.177899629671335e-06, "loss": 1.8634, "step": 3279 }, { "epoch": 0.37610365783740396, "grad_norm": 31.75, "learning_rate": 7.1762279433877465e-06, "loss": 1.8896, "step": 3280 }, { "epoch": 0.3762183235867446, "grad_norm": 22.625, "learning_rate": 7.174555956934996e-06, "loss": 1.9729, "step": 3281 }, { "epoch": 0.37633298933608533, "grad_norm": 12.75, "learning_rate": 7.172883670543701e-06, "loss": 1.456, "step": 3282 }, { "epoch": 0.376447655085426, "grad_norm": 16.25, "learning_rate": 7.171211084444525e-06, "loss": 1.8764, "step": 3283 }, { "epoch": 0.37656232083476665, "grad_norm": 13.5, "learning_rate": 7.169538198868164e-06, "loss": 1.899, "step": 3284 }, { "epoch": 0.3766769865841073, "grad_norm": 13.1875, "learning_rate": 7.167865014045365e-06, "loss": 1.5351, "step": 3285 }, { "epoch": 0.376791652333448, "grad_norm": 33.0, "learning_rate": 7.1661915302069095e-06, "loss": 1.9424, "step": 3286 }, { "epoch": 0.3769063180827887, "grad_norm": 14.8125, "learning_rate": 7.164517747583626e-06, "loss": 1.6589, "step": 3287 }, { "epoch": 0.37702098383212934, "grad_norm": 17.0, "learning_rate": 7.162843666406377e-06, "loss": 1.503, "step": 3288 }, { "epoch": 0.37713564958147, "grad_norm": 13.125, "learning_rate": 7.161169286906071e-06, "loss": 1.6412, "step": 3289 }, { "epoch": 0.37725031533081066, "grad_norm": 18.0, "learning_rate": 7.15949460931366e-06, "loss": 1.8735, "step": 3290 }, { "epoch": 0.3773649810801514, "grad_norm": 13.8125, "learning_rate": 7.157819633860129e-06, "loss": 2.0366, "step": 3291 }, { "epoch": 0.37747964682949203, "grad_norm": 13.125, "learning_rate": 7.156144360776513e-06, "loss": 1.8286, "step": 3292 }, { "epoch": 0.3775943125788327, "grad_norm": 14.6875, "learning_rate": 7.154468790293883e-06, "loss": 1.5999, "step": 3293 }, { "epoch": 0.37770897832817335, "grad_norm": 13.5625, "learning_rate": 7.152792922643348e-06, "loss": 1.6299, "step": 3294 }, { "epoch": 0.37782364407751406, "grad_norm": 12.625, "learning_rate": 7.151116758056067e-06, "loss": 1.6173, "step": 3295 }, { "epoch": 0.3779383098268547, "grad_norm": 14.4375, "learning_rate": 7.149440296763234e-06, "loss": 2.0361, "step": 3296 }, { "epoch": 0.3780529755761954, "grad_norm": 14.6875, "learning_rate": 7.147763538996083e-06, "loss": 1.5973, "step": 3297 }, { "epoch": 0.37816764132553604, "grad_norm": 13.9375, "learning_rate": 7.146086484985893e-06, "loss": 2.0637, "step": 3298 }, { "epoch": 0.37828230707487676, "grad_norm": 12.875, "learning_rate": 7.14440913496398e-06, "loss": 1.4729, "step": 3299 }, { "epoch": 0.3783969728242174, "grad_norm": 12.4375, "learning_rate": 7.142731489161702e-06, "loss": 1.5576, "step": 3300 }, { "epoch": 0.3785116385735581, "grad_norm": 12.75, "learning_rate": 7.14105354781046e-06, "loss": 2.2043, "step": 3301 }, { "epoch": 0.37862630432289873, "grad_norm": 12.375, "learning_rate": 7.139375311141694e-06, "loss": 1.8377, "step": 3302 }, { "epoch": 0.37874097007223945, "grad_norm": 16.0, "learning_rate": 7.137696779386884e-06, "loss": 2.0563, "step": 3303 }, { "epoch": 0.3788556358215801, "grad_norm": 17.5, "learning_rate": 7.13601795277755e-06, "loss": 1.6039, "step": 3304 }, { "epoch": 0.37897030157092076, "grad_norm": 13.6875, "learning_rate": 7.1343388315452575e-06, "loss": 2.0536, "step": 3305 }, { "epoch": 0.3790849673202614, "grad_norm": 12.3125, "learning_rate": 7.132659415921606e-06, "loss": 1.878, "step": 3306 }, { "epoch": 0.3791996330696021, "grad_norm": 14.3125, "learning_rate": 7.1309797061382416e-06, "loss": 1.7414, "step": 3307 }, { "epoch": 0.3793142988189428, "grad_norm": 12.875, "learning_rate": 7.129299702426846e-06, "loss": 1.9254, "step": 3308 }, { "epoch": 0.37942896456828346, "grad_norm": 14.75, "learning_rate": 7.127619405019144e-06, "loss": 1.683, "step": 3309 }, { "epoch": 0.3795436303176241, "grad_norm": 14.5, "learning_rate": 7.125938814146902e-06, "loss": 1.9731, "step": 3310 }, { "epoch": 0.3796582960669648, "grad_norm": 12.6875, "learning_rate": 7.124257930041925e-06, "loss": 2.0267, "step": 3311 }, { "epoch": 0.3797729618163055, "grad_norm": 14.375, "learning_rate": 7.1225767529360575e-06, "loss": 1.4039, "step": 3312 }, { "epoch": 0.37988762756564615, "grad_norm": 13.8125, "learning_rate": 7.120895283061188e-06, "loss": 1.5331, "step": 3313 }, { "epoch": 0.3800022933149868, "grad_norm": 12.5625, "learning_rate": 7.119213520649241e-06, "loss": 1.3749, "step": 3314 }, { "epoch": 0.38011695906432746, "grad_norm": 18.875, "learning_rate": 7.117531465932186e-06, "loss": 1.5865, "step": 3315 }, { "epoch": 0.3802316248136682, "grad_norm": 16.0, "learning_rate": 7.115849119142027e-06, "loss": 1.4084, "step": 3316 }, { "epoch": 0.38034629056300884, "grad_norm": 15.0, "learning_rate": 7.114166480510816e-06, "loss": 1.8192, "step": 3317 }, { "epoch": 0.3804609563123495, "grad_norm": 17.125, "learning_rate": 7.112483550270639e-06, "loss": 1.2616, "step": 3318 }, { "epoch": 0.38057562206169016, "grad_norm": 12.4375, "learning_rate": 7.110800328653621e-06, "loss": 1.467, "step": 3319 }, { "epoch": 0.38069028781103087, "grad_norm": 14.8125, "learning_rate": 7.109116815891936e-06, "loss": 2.0345, "step": 3320 }, { "epoch": 0.38080495356037153, "grad_norm": 28.875, "learning_rate": 7.107433012217788e-06, "loss": 1.8263, "step": 3321 }, { "epoch": 0.3809196193097122, "grad_norm": 15.625, "learning_rate": 7.105748917863427e-06, "loss": 2.5177, "step": 3322 }, { "epoch": 0.38103428505905285, "grad_norm": 14.0, "learning_rate": 7.104064533061144e-06, "loss": 1.6009, "step": 3323 }, { "epoch": 0.3811489508083935, "grad_norm": 31.25, "learning_rate": 7.102379858043264e-06, "loss": 2.1751, "step": 3324 }, { "epoch": 0.3812636165577342, "grad_norm": 29.25, "learning_rate": 7.100694893042159e-06, "loss": 1.8406, "step": 3325 }, { "epoch": 0.3813782823070749, "grad_norm": 14.3125, "learning_rate": 7.0990096382902356e-06, "loss": 1.6657, "step": 3326 }, { "epoch": 0.38149294805641554, "grad_norm": 16.125, "learning_rate": 7.0973240940199436e-06, "loss": 1.6417, "step": 3327 }, { "epoch": 0.3816076138057562, "grad_norm": 16.875, "learning_rate": 7.095638260463771e-06, "loss": 1.7081, "step": 3328 }, { "epoch": 0.3817222795550969, "grad_norm": 14.875, "learning_rate": 7.093952137854248e-06, "loss": 1.9134, "step": 3329 }, { "epoch": 0.38183694530443757, "grad_norm": 14.625, "learning_rate": 7.092265726423941e-06, "loss": 1.7384, "step": 3330 }, { "epoch": 0.38195161105377823, "grad_norm": 26.0, "learning_rate": 7.090579026405459e-06, "loss": 1.9432, "step": 3331 }, { "epoch": 0.3820662768031189, "grad_norm": 12.9375, "learning_rate": 7.08889203803145e-06, "loss": 1.6684, "step": 3332 }, { "epoch": 0.3821809425524596, "grad_norm": 33.0, "learning_rate": 7.087204761534604e-06, "loss": 1.9138, "step": 3333 }, { "epoch": 0.38229560830180026, "grad_norm": 12.75, "learning_rate": 7.085517197147645e-06, "loss": 1.949, "step": 3334 }, { "epoch": 0.3824102740511409, "grad_norm": 13.9375, "learning_rate": 7.083829345103344e-06, "loss": 1.6185, "step": 3335 }, { "epoch": 0.3825249398004816, "grad_norm": 16.125, "learning_rate": 7.082141205634505e-06, "loss": 1.7823, "step": 3336 }, { "epoch": 0.3826396055498223, "grad_norm": 14.6875, "learning_rate": 7.080452778973976e-06, "loss": 1.8178, "step": 3337 }, { "epoch": 0.38275427129916295, "grad_norm": 13.375, "learning_rate": 7.0787640653546445e-06, "loss": 1.4295, "step": 3338 }, { "epoch": 0.3828689370485036, "grad_norm": 13.4375, "learning_rate": 7.0770750650094335e-06, "loss": 1.4672, "step": 3339 }, { "epoch": 0.38298360279784427, "grad_norm": 28.125, "learning_rate": 7.075385778171311e-06, "loss": 2.0967, "step": 3340 }, { "epoch": 0.383098268547185, "grad_norm": 13.5625, "learning_rate": 7.073696205073278e-06, "loss": 1.9853, "step": 3341 }, { "epoch": 0.38321293429652564, "grad_norm": 17.75, "learning_rate": 7.072006345948385e-06, "loss": 1.6075, "step": 3342 }, { "epoch": 0.3833276000458663, "grad_norm": 15.0625, "learning_rate": 7.070316201029712e-06, "loss": 1.4142, "step": 3343 }, { "epoch": 0.38344226579520696, "grad_norm": 16.375, "learning_rate": 7.068625770550381e-06, "loss": 1.8957, "step": 3344 }, { "epoch": 0.3835569315445476, "grad_norm": 15.125, "learning_rate": 7.06693505474356e-06, "loss": 1.9495, "step": 3345 }, { "epoch": 0.38367159729388833, "grad_norm": 12.125, "learning_rate": 7.065244053842445e-06, "loss": 1.4321, "step": 3346 }, { "epoch": 0.383786263043229, "grad_norm": 12.9375, "learning_rate": 7.0635527680802794e-06, "loss": 1.5685, "step": 3347 }, { "epoch": 0.38390092879256965, "grad_norm": 14.4375, "learning_rate": 7.061861197690347e-06, "loss": 1.6537, "step": 3348 }, { "epoch": 0.3840155945419103, "grad_norm": 11.875, "learning_rate": 7.0601693429059624e-06, "loss": 1.4236, "step": 3349 }, { "epoch": 0.384130260291251, "grad_norm": 15.6875, "learning_rate": 7.058477203960489e-06, "loss": 1.6897, "step": 3350 }, { "epoch": 0.3842449260405917, "grad_norm": 12.0625, "learning_rate": 7.056784781087323e-06, "loss": 1.5926, "step": 3351 }, { "epoch": 0.38435959178993234, "grad_norm": 14.125, "learning_rate": 7.055092074519902e-06, "loss": 2.165, "step": 3352 }, { "epoch": 0.384474257539273, "grad_norm": 24.625, "learning_rate": 7.053399084491703e-06, "loss": 1.7497, "step": 3353 }, { "epoch": 0.3845889232886137, "grad_norm": 15.0, "learning_rate": 7.0517058112362415e-06, "loss": 1.8569, "step": 3354 }, { "epoch": 0.3847035890379544, "grad_norm": 13.0625, "learning_rate": 7.050012254987073e-06, "loss": 1.6428, "step": 3355 }, { "epoch": 0.38481825478729503, "grad_norm": 14.9375, "learning_rate": 7.048318415977791e-06, "loss": 1.7688, "step": 3356 }, { "epoch": 0.3849329205366357, "grad_norm": 17.25, "learning_rate": 7.046624294442027e-06, "loss": 1.872, "step": 3357 }, { "epoch": 0.3850475862859764, "grad_norm": 14.25, "learning_rate": 7.044929890613454e-06, "loss": 1.619, "step": 3358 }, { "epoch": 0.38516225203531707, "grad_norm": 11.8125, "learning_rate": 7.0432352047257835e-06, "loss": 1.6704, "step": 3359 }, { "epoch": 0.3852769177846577, "grad_norm": 12.875, "learning_rate": 7.041540237012763e-06, "loss": 2.0055, "step": 3360 }, { "epoch": 0.3853915835339984, "grad_norm": 16.0, "learning_rate": 7.039844987708182e-06, "loss": 1.403, "step": 3361 }, { "epoch": 0.38550624928333904, "grad_norm": 14.0, "learning_rate": 7.038149457045868e-06, "loss": 1.748, "step": 3362 }, { "epoch": 0.38562091503267976, "grad_norm": 14.8125, "learning_rate": 7.036453645259689e-06, "loss": 2.1142, "step": 3363 }, { "epoch": 0.3857355807820204, "grad_norm": 12.6875, "learning_rate": 7.034757552583546e-06, "loss": 1.6196, "step": 3364 }, { "epoch": 0.3858502465313611, "grad_norm": 16.0, "learning_rate": 7.033061179251386e-06, "loss": 2.0463, "step": 3365 }, { "epoch": 0.38596491228070173, "grad_norm": 19.0, "learning_rate": 7.031364525497187e-06, "loss": 1.8262, "step": 3366 }, { "epoch": 0.38607957803004245, "grad_norm": 12.8125, "learning_rate": 7.0296675915549765e-06, "loss": 1.7783, "step": 3367 }, { "epoch": 0.3861942437793831, "grad_norm": 12.8125, "learning_rate": 7.027970377658809e-06, "loss": 2.1564, "step": 3368 }, { "epoch": 0.38630890952872377, "grad_norm": 27.625, "learning_rate": 7.026272884042784e-06, "loss": 1.9538, "step": 3369 }, { "epoch": 0.3864235752780644, "grad_norm": 15.0, "learning_rate": 7.024575110941041e-06, "loss": 1.7099, "step": 3370 }, { "epoch": 0.38653824102740514, "grad_norm": 15.75, "learning_rate": 7.022877058587752e-06, "loss": 1.8312, "step": 3371 }, { "epoch": 0.3866529067767458, "grad_norm": 17.125, "learning_rate": 7.021178727217132e-06, "loss": 2.1008, "step": 3372 }, { "epoch": 0.38676757252608646, "grad_norm": 12.625, "learning_rate": 7.019480117063434e-06, "loss": 1.5858, "step": 3373 }, { "epoch": 0.3868822382754271, "grad_norm": 12.1875, "learning_rate": 7.01778122836095e-06, "loss": 1.6486, "step": 3374 }, { "epoch": 0.38699690402476783, "grad_norm": 13.5625, "learning_rate": 7.016082061344005e-06, "loss": 1.4758, "step": 3375 }, { "epoch": 0.3871115697741085, "grad_norm": 13.9375, "learning_rate": 7.014382616246971e-06, "loss": 1.8782, "step": 3376 }, { "epoch": 0.38722623552344915, "grad_norm": 14.5, "learning_rate": 7.0126828933042545e-06, "loss": 2.0184, "step": 3377 }, { "epoch": 0.3873409012727898, "grad_norm": 17.875, "learning_rate": 7.010982892750296e-06, "loss": 1.8648, "step": 3378 }, { "epoch": 0.38745556702213046, "grad_norm": 15.4375, "learning_rate": 7.009282614819581e-06, "loss": 1.7005, "step": 3379 }, { "epoch": 0.3875702327714712, "grad_norm": 18.875, "learning_rate": 7.007582059746631e-06, "loss": 2.1039, "step": 3380 }, { "epoch": 0.38768489852081184, "grad_norm": 14.375, "learning_rate": 7.0058812277660014e-06, "loss": 1.8937, "step": 3381 }, { "epoch": 0.3877995642701525, "grad_norm": 15.75, "learning_rate": 7.004180119112294e-06, "loss": 1.6737, "step": 3382 }, { "epoch": 0.38791423001949316, "grad_norm": 12.875, "learning_rate": 7.002478734020142e-06, "loss": 1.5061, "step": 3383 }, { "epoch": 0.38802889576883387, "grad_norm": 14.9375, "learning_rate": 7.000777072724219e-06, "loss": 1.4783, "step": 3384 }, { "epoch": 0.38814356151817453, "grad_norm": 14.0625, "learning_rate": 6.999075135459236e-06, "loss": 2.0073, "step": 3385 }, { "epoch": 0.3882582272675152, "grad_norm": 14.0, "learning_rate": 6.997372922459944e-06, "loss": 1.8117, "step": 3386 }, { "epoch": 0.38837289301685585, "grad_norm": 12.6875, "learning_rate": 6.995670433961133e-06, "loss": 1.9024, "step": 3387 }, { "epoch": 0.38848755876619656, "grad_norm": 13.9375, "learning_rate": 6.993967670197625e-06, "loss": 2.0514, "step": 3388 }, { "epoch": 0.3886022245155372, "grad_norm": 18.875, "learning_rate": 6.992264631404284e-06, "loss": 1.3539, "step": 3389 }, { "epoch": 0.3887168902648779, "grad_norm": 14.25, "learning_rate": 6.990561317816016e-06, "loss": 1.6605, "step": 3390 }, { "epoch": 0.38883155601421854, "grad_norm": 16.125, "learning_rate": 6.9888577296677554e-06, "loss": 1.8415, "step": 3391 }, { "epoch": 0.38894622176355925, "grad_norm": 13.375, "learning_rate": 6.987153867194484e-06, "loss": 1.7488, "step": 3392 }, { "epoch": 0.3890608875128999, "grad_norm": 24.75, "learning_rate": 6.9854497306312156e-06, "loss": 1.686, "step": 3393 }, { "epoch": 0.38917555326224057, "grad_norm": 15.875, "learning_rate": 6.983745320212999e-06, "loss": 1.7412, "step": 3394 }, { "epoch": 0.38929021901158123, "grad_norm": 13.75, "learning_rate": 6.9820406361749325e-06, "loss": 2.2007, "step": 3395 }, { "epoch": 0.3894048847609219, "grad_norm": 14.0625, "learning_rate": 6.9803356787521415e-06, "loss": 1.596, "step": 3396 }, { "epoch": 0.3895195505102626, "grad_norm": 13.0625, "learning_rate": 6.97863044817979e-06, "loss": 1.7633, "step": 3397 }, { "epoch": 0.38963421625960326, "grad_norm": 14.625, "learning_rate": 6.976924944693086e-06, "loss": 1.572, "step": 3398 }, { "epoch": 0.3897488820089439, "grad_norm": 18.875, "learning_rate": 6.975219168527269e-06, "loss": 1.6912, "step": 3399 }, { "epoch": 0.3898635477582846, "grad_norm": 12.9375, "learning_rate": 6.97351311991762e-06, "loss": 1.8434, "step": 3400 }, { "epoch": 0.3899782135076253, "grad_norm": 14.0625, "learning_rate": 6.971806799099453e-06, "loss": 1.823, "step": 3401 }, { "epoch": 0.39009287925696595, "grad_norm": 19.75, "learning_rate": 6.970100206308126e-06, "loss": 1.7505, "step": 3402 }, { "epoch": 0.3902075450063066, "grad_norm": 13.625, "learning_rate": 6.968393341779028e-06, "loss": 1.5818, "step": 3403 }, { "epoch": 0.39032221075564727, "grad_norm": 18.0, "learning_rate": 6.966686205747588e-06, "loss": 1.5065, "step": 3404 }, { "epoch": 0.390436876504988, "grad_norm": 12.9375, "learning_rate": 6.964978798449277e-06, "loss": 1.4615, "step": 3405 }, { "epoch": 0.39055154225432864, "grad_norm": 15.8125, "learning_rate": 6.9632711201195944e-06, "loss": 1.7178, "step": 3406 }, { "epoch": 0.3906662080036693, "grad_norm": 12.1875, "learning_rate": 6.961563170994086e-06, "loss": 2.1467, "step": 3407 }, { "epoch": 0.39078087375300996, "grad_norm": 18.375, "learning_rate": 6.959854951308328e-06, "loss": 1.5624, "step": 3408 }, { "epoch": 0.3908955395023507, "grad_norm": 15.8125, "learning_rate": 6.958146461297937e-06, "loss": 1.4028, "step": 3409 }, { "epoch": 0.39101020525169133, "grad_norm": 15.5, "learning_rate": 6.9564377011985685e-06, "loss": 1.8255, "step": 3410 }, { "epoch": 0.391124871001032, "grad_norm": 18.25, "learning_rate": 6.9547286712459115e-06, "loss": 1.8474, "step": 3411 }, { "epoch": 0.39123953675037265, "grad_norm": 15.25, "learning_rate": 6.953019371675695e-06, "loss": 1.6648, "step": 3412 }, { "epoch": 0.3913542024997133, "grad_norm": 15.5625, "learning_rate": 6.951309802723685e-06, "loss": 1.9555, "step": 3413 }, { "epoch": 0.391468868249054, "grad_norm": 16.5, "learning_rate": 6.949599964625682e-06, "loss": 1.827, "step": 3414 }, { "epoch": 0.3915835339983947, "grad_norm": 12.9375, "learning_rate": 6.947889857617527e-06, "loss": 1.8627, "step": 3415 }, { "epoch": 0.39169819974773534, "grad_norm": 15.8125, "learning_rate": 6.946179481935096e-06, "loss": 1.8129, "step": 3416 }, { "epoch": 0.391812865497076, "grad_norm": 16.5, "learning_rate": 6.944468837814303e-06, "loss": 1.8376, "step": 3417 }, { "epoch": 0.3919275312464167, "grad_norm": 15.6875, "learning_rate": 6.942757925491101e-06, "loss": 1.2789, "step": 3418 }, { "epoch": 0.3920421969957574, "grad_norm": 13.0, "learning_rate": 6.941046745201472e-06, "loss": 1.531, "step": 3419 }, { "epoch": 0.39215686274509803, "grad_norm": 14.4375, "learning_rate": 6.9393352971814466e-06, "loss": 1.6438, "step": 3420 }, { "epoch": 0.3922715284944387, "grad_norm": 11.5, "learning_rate": 6.937623581667083e-06, "loss": 1.5942, "step": 3421 }, { "epoch": 0.3923861942437794, "grad_norm": 17.75, "learning_rate": 6.93591159889448e-06, "loss": 1.908, "step": 3422 }, { "epoch": 0.39250085999312007, "grad_norm": 14.4375, "learning_rate": 6.934199349099775e-06, "loss": 1.6305, "step": 3423 }, { "epoch": 0.3926155257424607, "grad_norm": 14.5625, "learning_rate": 6.932486832519139e-06, "loss": 2.2526, "step": 3424 }, { "epoch": 0.3927301914918014, "grad_norm": 12.5, "learning_rate": 6.93077404938878e-06, "loss": 1.6223, "step": 3425 }, { "epoch": 0.3928448572411421, "grad_norm": 12.625, "learning_rate": 6.929060999944946e-06, "loss": 1.8682, "step": 3426 }, { "epoch": 0.39295952299048276, "grad_norm": 15.5625, "learning_rate": 6.927347684423918e-06, "loss": 1.8233, "step": 3427 }, { "epoch": 0.3930741887398234, "grad_norm": 26.25, "learning_rate": 6.9256341030620155e-06, "loss": 1.672, "step": 3428 }, { "epoch": 0.3931888544891641, "grad_norm": 13.0625, "learning_rate": 6.923920256095594e-06, "loss": 1.5677, "step": 3429 }, { "epoch": 0.39330352023850473, "grad_norm": 12.75, "learning_rate": 6.922206143761048e-06, "loss": 1.4559, "step": 3430 }, { "epoch": 0.39341818598784545, "grad_norm": 13.5, "learning_rate": 6.920491766294803e-06, "loss": 1.5519, "step": 3431 }, { "epoch": 0.3935328517371861, "grad_norm": 20.625, "learning_rate": 6.918777123933327e-06, "loss": 2.513, "step": 3432 }, { "epoch": 0.39364751748652677, "grad_norm": 13.6875, "learning_rate": 6.917062216913123e-06, "loss": 1.9803, "step": 3433 }, { "epoch": 0.3937621832358674, "grad_norm": 17.25, "learning_rate": 6.915347045470728e-06, "loss": 1.6703, "step": 3434 }, { "epoch": 0.39387684898520814, "grad_norm": 12.625, "learning_rate": 6.913631609842718e-06, "loss": 1.8472, "step": 3435 }, { "epoch": 0.3939915147345488, "grad_norm": 15.125, "learning_rate": 6.911915910265704e-06, "loss": 1.763, "step": 3436 }, { "epoch": 0.39410618048388946, "grad_norm": 17.375, "learning_rate": 6.910199946976337e-06, "loss": 2.1239, "step": 3437 }, { "epoch": 0.3942208462332301, "grad_norm": 13.75, "learning_rate": 6.908483720211297e-06, "loss": 1.7743, "step": 3438 }, { "epoch": 0.39433551198257083, "grad_norm": 15.8125, "learning_rate": 6.906767230207307e-06, "loss": 1.6592, "step": 3439 }, { "epoch": 0.3944501777319115, "grad_norm": 14.5, "learning_rate": 6.905050477201124e-06, "loss": 1.6951, "step": 3440 }, { "epoch": 0.39456484348125215, "grad_norm": 25.875, "learning_rate": 6.90333346142954e-06, "loss": 1.717, "step": 3441 }, { "epoch": 0.3946795092305928, "grad_norm": 16.0, "learning_rate": 6.901616183129387e-06, "loss": 2.0023, "step": 3442 }, { "epoch": 0.3947941749799335, "grad_norm": 14.75, "learning_rate": 6.899898642537531e-06, "loss": 1.7576, "step": 3443 }, { "epoch": 0.3949088407292742, "grad_norm": 13.25, "learning_rate": 6.89818083989087e-06, "loss": 1.9682, "step": 3444 }, { "epoch": 0.39502350647861484, "grad_norm": 20.375, "learning_rate": 6.8964627754263465e-06, "loss": 2.0484, "step": 3445 }, { "epoch": 0.3951381722279555, "grad_norm": 17.875, "learning_rate": 6.894744449380933e-06, "loss": 1.7348, "step": 3446 }, { "epoch": 0.39525283797729616, "grad_norm": 15.0625, "learning_rate": 6.893025861991639e-06, "loss": 1.7707, "step": 3447 }, { "epoch": 0.39536750372663687, "grad_norm": 32.25, "learning_rate": 6.891307013495513e-06, "loss": 2.0205, "step": 3448 }, { "epoch": 0.39548216947597753, "grad_norm": 15.25, "learning_rate": 6.889587904129634e-06, "loss": 1.9089, "step": 3449 }, { "epoch": 0.3955968352253182, "grad_norm": 14.5, "learning_rate": 6.887868534131123e-06, "loss": 1.9677, "step": 3450 }, { "epoch": 0.39571150097465885, "grad_norm": 13.0, "learning_rate": 6.8861489037371356e-06, "loss": 1.5479, "step": 3451 }, { "epoch": 0.39582616672399956, "grad_norm": 18.75, "learning_rate": 6.884429013184859e-06, "loss": 2.0195, "step": 3452 }, { "epoch": 0.3959408324733402, "grad_norm": 13.6875, "learning_rate": 6.88270886271152e-06, "loss": 1.6101, "step": 3453 }, { "epoch": 0.3960554982226809, "grad_norm": 18.5, "learning_rate": 6.880988452554382e-06, "loss": 2.4063, "step": 3454 }, { "epoch": 0.39617016397202154, "grad_norm": 15.375, "learning_rate": 6.8792677829507426e-06, "loss": 2.2503, "step": 3455 }, { "epoch": 0.39628482972136225, "grad_norm": 18.0, "learning_rate": 6.877546854137933e-06, "loss": 1.5566, "step": 3456 }, { "epoch": 0.3963994954707029, "grad_norm": 12.875, "learning_rate": 6.875825666353324e-06, "loss": 1.5403, "step": 3457 }, { "epoch": 0.39651416122004357, "grad_norm": 15.125, "learning_rate": 6.8741042198343225e-06, "loss": 2.1043, "step": 3458 }, { "epoch": 0.39662882696938423, "grad_norm": 11.5, "learning_rate": 6.8723825148183655e-06, "loss": 1.5936, "step": 3459 }, { "epoch": 0.39674349271872494, "grad_norm": 22.375, "learning_rate": 6.870660551542933e-06, "loss": 1.6411, "step": 3460 }, { "epoch": 0.3968581584680656, "grad_norm": 13.8125, "learning_rate": 6.868938330245534e-06, "loss": 1.4971, "step": 3461 }, { "epoch": 0.39697282421740626, "grad_norm": 15.625, "learning_rate": 6.86721585116372e-06, "loss": 2.0599, "step": 3462 }, { "epoch": 0.3970874899667469, "grad_norm": 13.3125, "learning_rate": 6.865493114535069e-06, "loss": 1.8036, "step": 3463 }, { "epoch": 0.3972021557160876, "grad_norm": 15.4375, "learning_rate": 6.863770120597204e-06, "loss": 2.1319, "step": 3464 }, { "epoch": 0.3973168214654283, "grad_norm": 13.625, "learning_rate": 6.8620468695877775e-06, "loss": 1.5372, "step": 3465 }, { "epoch": 0.39743148721476895, "grad_norm": 14.9375, "learning_rate": 6.860323361744477e-06, "loss": 1.6228, "step": 3466 }, { "epoch": 0.3975461529641096, "grad_norm": 14.0, "learning_rate": 6.8585995973050335e-06, "loss": 1.3981, "step": 3467 }, { "epoch": 0.39766081871345027, "grad_norm": 13.4375, "learning_rate": 6.8568755765072015e-06, "loss": 1.8167, "step": 3468 }, { "epoch": 0.397775484462791, "grad_norm": 11.9375, "learning_rate": 6.855151299588778e-06, "loss": 1.5761, "step": 3469 }, { "epoch": 0.39789015021213164, "grad_norm": 29.375, "learning_rate": 6.853426766787598e-06, "loss": 1.9837, "step": 3470 }, { "epoch": 0.3980048159614723, "grad_norm": 26.25, "learning_rate": 6.851701978341523e-06, "loss": 1.7313, "step": 3471 }, { "epoch": 0.39811948171081296, "grad_norm": 12.0625, "learning_rate": 6.849976934488456e-06, "loss": 1.3553, "step": 3472 }, { "epoch": 0.3982341474601537, "grad_norm": 15.625, "learning_rate": 6.848251635466337e-06, "loss": 1.8786, "step": 3473 }, { "epoch": 0.39834881320949433, "grad_norm": 17.375, "learning_rate": 6.846526081513134e-06, "loss": 2.1216, "step": 3474 }, { "epoch": 0.398463478958835, "grad_norm": 14.125, "learning_rate": 6.844800272866856e-06, "loss": 2.1818, "step": 3475 }, { "epoch": 0.39857814470817565, "grad_norm": 14.75, "learning_rate": 6.8430742097655455e-06, "loss": 1.8798, "step": 3476 }, { "epoch": 0.39869281045751637, "grad_norm": 13.5, "learning_rate": 6.841347892447282e-06, "loss": 1.7032, "step": 3477 }, { "epoch": 0.398807476206857, "grad_norm": 14.75, "learning_rate": 6.839621321150174e-06, "loss": 1.517, "step": 3478 }, { "epoch": 0.3989221419561977, "grad_norm": 12.8125, "learning_rate": 6.837894496112371e-06, "loss": 1.4668, "step": 3479 }, { "epoch": 0.39903680770553834, "grad_norm": 13.1875, "learning_rate": 6.836167417572057e-06, "loss": 1.8331, "step": 3480 }, { "epoch": 0.399151473454879, "grad_norm": 17.625, "learning_rate": 6.834440085767447e-06, "loss": 1.9654, "step": 3481 }, { "epoch": 0.3992661392042197, "grad_norm": 14.8125, "learning_rate": 6.832712500936795e-06, "loss": 1.8808, "step": 3482 }, { "epoch": 0.3993808049535604, "grad_norm": 12.375, "learning_rate": 6.830984663318391e-06, "loss": 1.4305, "step": 3483 }, { "epoch": 0.39949547070290103, "grad_norm": 19.375, "learning_rate": 6.8292565731505515e-06, "loss": 1.5643, "step": 3484 }, { "epoch": 0.3996101364522417, "grad_norm": 13.75, "learning_rate": 6.8275282306716365e-06, "loss": 1.9696, "step": 3485 }, { "epoch": 0.3997248022015824, "grad_norm": 18.75, "learning_rate": 6.825799636120039e-06, "loss": 1.9647, "step": 3486 }, { "epoch": 0.39983946795092307, "grad_norm": 17.875, "learning_rate": 6.824070789734184e-06, "loss": 1.7533, "step": 3487 }, { "epoch": 0.3999541337002637, "grad_norm": 15.0625, "learning_rate": 6.822341691752533e-06, "loss": 1.5451, "step": 3488 }, { "epoch": 0.4000687994496044, "grad_norm": 17.0, "learning_rate": 6.820612342413583e-06, "loss": 1.7127, "step": 3489 }, { "epoch": 0.4001834651989451, "grad_norm": 11.0, "learning_rate": 6.818882741955866e-06, "loss": 1.5882, "step": 3490 }, { "epoch": 0.40029813094828576, "grad_norm": 14.4375, "learning_rate": 6.817152890617943e-06, "loss": 1.5687, "step": 3491 }, { "epoch": 0.4004127966976264, "grad_norm": 16.125, "learning_rate": 6.815422788638418e-06, "loss": 1.8818, "step": 3492 }, { "epoch": 0.4005274624469671, "grad_norm": 32.0, "learning_rate": 6.8136924362559265e-06, "loss": 1.8213, "step": 3493 }, { "epoch": 0.4006421281963078, "grad_norm": 15.0, "learning_rate": 6.811961833709132e-06, "loss": 1.9841, "step": 3494 }, { "epoch": 0.40075679394564845, "grad_norm": 17.75, "learning_rate": 6.810230981236744e-06, "loss": 1.8233, "step": 3495 }, { "epoch": 0.4008714596949891, "grad_norm": 13.125, "learning_rate": 6.808499879077497e-06, "loss": 1.9595, "step": 3496 }, { "epoch": 0.40098612544432977, "grad_norm": 16.375, "learning_rate": 6.806768527470166e-06, "loss": 1.7463, "step": 3497 }, { "epoch": 0.4011007911936704, "grad_norm": 13.5625, "learning_rate": 6.805036926653557e-06, "loss": 1.6259, "step": 3498 }, { "epoch": 0.40121545694301114, "grad_norm": 14.3125, "learning_rate": 6.8033050768665096e-06, "loss": 1.6247, "step": 3499 }, { "epoch": 0.4013301226923518, "grad_norm": 24.625, "learning_rate": 6.801572978347901e-06, "loss": 1.6707, "step": 3500 }, { "epoch": 0.40144478844169246, "grad_norm": 34.0, "learning_rate": 6.7998406313366424e-06, "loss": 1.6919, "step": 3501 }, { "epoch": 0.4015594541910331, "grad_norm": 20.625, "learning_rate": 6.798108036071677e-06, "loss": 1.8488, "step": 3502 }, { "epoch": 0.40167411994037383, "grad_norm": 13.0, "learning_rate": 6.796375192791982e-06, "loss": 1.5844, "step": 3503 }, { "epoch": 0.4017887856897145, "grad_norm": 17.875, "learning_rate": 6.794642101736571e-06, "loss": 1.6802, "step": 3504 }, { "epoch": 0.40190345143905515, "grad_norm": 12.5, "learning_rate": 6.792908763144492e-06, "loss": 1.7522, "step": 3505 }, { "epoch": 0.4020181171883958, "grad_norm": 15.4375, "learning_rate": 6.791175177254825e-06, "loss": 1.5923, "step": 3506 }, { "epoch": 0.4021327829377365, "grad_norm": 13.375, "learning_rate": 6.789441344306684e-06, "loss": 1.6504, "step": 3507 }, { "epoch": 0.4022474486870772, "grad_norm": 14.75, "learning_rate": 6.7877072645392215e-06, "loss": 1.5701, "step": 3508 }, { "epoch": 0.40236211443641784, "grad_norm": 13.875, "learning_rate": 6.785972938191617e-06, "loss": 1.5617, "step": 3509 }, { "epoch": 0.4024767801857585, "grad_norm": 30.375, "learning_rate": 6.78423836550309e-06, "loss": 2.0663, "step": 3510 }, { "epoch": 0.4025914459350992, "grad_norm": 14.5625, "learning_rate": 6.782503546712891e-06, "loss": 1.798, "step": 3511 }, { "epoch": 0.40270611168443987, "grad_norm": 13.125, "learning_rate": 6.780768482060305e-06, "loss": 1.7792, "step": 3512 }, { "epoch": 0.40282077743378053, "grad_norm": 16.5, "learning_rate": 6.779033171784652e-06, "loss": 1.7679, "step": 3513 }, { "epoch": 0.4029354431831212, "grad_norm": 15.25, "learning_rate": 6.777297616125283e-06, "loss": 1.8133, "step": 3514 }, { "epoch": 0.40305010893246185, "grad_norm": 16.75, "learning_rate": 6.77556181532159e-06, "loss": 1.8551, "step": 3515 }, { "epoch": 0.40316477468180256, "grad_norm": 16.625, "learning_rate": 6.773825769612985e-06, "loss": 1.5897, "step": 3516 }, { "epoch": 0.4032794404311432, "grad_norm": 11.9375, "learning_rate": 6.772089479238931e-06, "loss": 1.4337, "step": 3517 }, { "epoch": 0.4033941061804839, "grad_norm": 12.9375, "learning_rate": 6.770352944438911e-06, "loss": 1.557, "step": 3518 }, { "epoch": 0.40350877192982454, "grad_norm": 13.5, "learning_rate": 6.768616165452448e-06, "loss": 1.7189, "step": 3519 }, { "epoch": 0.40362343767916525, "grad_norm": 13.1875, "learning_rate": 6.7668791425190985e-06, "loss": 1.891, "step": 3520 }, { "epoch": 0.4037381034285059, "grad_norm": 16.375, "learning_rate": 6.76514187587845e-06, "loss": 1.8422, "step": 3521 }, { "epoch": 0.40385276917784657, "grad_norm": 15.4375, "learning_rate": 6.7634043657701266e-06, "loss": 1.6605, "step": 3522 }, { "epoch": 0.40396743492718723, "grad_norm": 12.75, "learning_rate": 6.761666612433786e-06, "loss": 1.7351, "step": 3523 }, { "epoch": 0.40408210067652794, "grad_norm": 15.0625, "learning_rate": 6.759928616109115e-06, "loss": 1.5726, "step": 3524 }, { "epoch": 0.4041967664258686, "grad_norm": 12.75, "learning_rate": 6.758190377035839e-06, "loss": 1.6057, "step": 3525 }, { "epoch": 0.40431143217520926, "grad_norm": 11.875, "learning_rate": 6.756451895453715e-06, "loss": 1.5086, "step": 3526 }, { "epoch": 0.4044260979245499, "grad_norm": 14.375, "learning_rate": 6.754713171602534e-06, "loss": 1.7212, "step": 3527 }, { "epoch": 0.40454076367389064, "grad_norm": 15.6875, "learning_rate": 6.752974205722117e-06, "loss": 2.195, "step": 3528 }, { "epoch": 0.4046554294232313, "grad_norm": 14.125, "learning_rate": 6.751234998052324e-06, "loss": 2.157, "step": 3529 }, { "epoch": 0.40477009517257195, "grad_norm": 11.5625, "learning_rate": 6.749495548833045e-06, "loss": 1.5148, "step": 3530 }, { "epoch": 0.4048847609219126, "grad_norm": 12.5625, "learning_rate": 6.747755858304203e-06, "loss": 1.4494, "step": 3531 }, { "epoch": 0.40499942667125327, "grad_norm": 14.5, "learning_rate": 6.746015926705755e-06, "loss": 1.4342, "step": 3532 }, { "epoch": 0.405114092420594, "grad_norm": 13.4375, "learning_rate": 6.744275754277694e-06, "loss": 1.8281, "step": 3533 }, { "epoch": 0.40522875816993464, "grad_norm": 16.875, "learning_rate": 6.7425353412600395e-06, "loss": 1.7218, "step": 3534 }, { "epoch": 0.4053434239192753, "grad_norm": 11.625, "learning_rate": 6.740794687892851e-06, "loss": 1.5151, "step": 3535 }, { "epoch": 0.40545808966861596, "grad_norm": 14.625, "learning_rate": 6.739053794416217e-06, "loss": 2.0544, "step": 3536 }, { "epoch": 0.4055727554179567, "grad_norm": 13.8125, "learning_rate": 6.737312661070263e-06, "loss": 1.6124, "step": 3537 }, { "epoch": 0.40568742116729734, "grad_norm": 12.875, "learning_rate": 6.735571288095144e-06, "loss": 1.7117, "step": 3538 }, { "epoch": 0.405802086916638, "grad_norm": 14.875, "learning_rate": 6.733829675731047e-06, "loss": 1.8724, "step": 3539 }, { "epoch": 0.40591675266597865, "grad_norm": 35.5, "learning_rate": 6.7320878242181976e-06, "loss": 2.3695, "step": 3540 }, { "epoch": 0.40603141841531937, "grad_norm": 15.25, "learning_rate": 6.730345733796848e-06, "loss": 1.739, "step": 3541 }, { "epoch": 0.40614608416466, "grad_norm": 13.9375, "learning_rate": 6.728603404707289e-06, "loss": 1.7893, "step": 3542 }, { "epoch": 0.4062607499140007, "grad_norm": 37.0, "learning_rate": 6.726860837189839e-06, "loss": 1.5343, "step": 3543 }, { "epoch": 0.40637541566334134, "grad_norm": 16.0, "learning_rate": 6.725118031484855e-06, "loss": 2.1855, "step": 3544 }, { "epoch": 0.40649008141268206, "grad_norm": 13.25, "learning_rate": 6.723374987832722e-06, "loss": 1.8813, "step": 3545 }, { "epoch": 0.4066047471620227, "grad_norm": 15.5625, "learning_rate": 6.7216317064738605e-06, "loss": 1.6857, "step": 3546 }, { "epoch": 0.4067194129113634, "grad_norm": 14.625, "learning_rate": 6.719888187648721e-06, "loss": 1.6384, "step": 3547 }, { "epoch": 0.40683407866070403, "grad_norm": 13.25, "learning_rate": 6.718144431597791e-06, "loss": 1.7728, "step": 3548 }, { "epoch": 0.4069487444100447, "grad_norm": 13.1875, "learning_rate": 6.716400438561588e-06, "loss": 2.1424, "step": 3549 }, { "epoch": 0.4070634101593854, "grad_norm": 15.25, "learning_rate": 6.71465620878066e-06, "loss": 1.6088, "step": 3550 }, { "epoch": 0.40717807590872607, "grad_norm": 33.5, "learning_rate": 6.712911742495593e-06, "loss": 1.6673, "step": 3551 }, { "epoch": 0.4072927416580667, "grad_norm": 13.0625, "learning_rate": 6.711167039947004e-06, "loss": 1.7967, "step": 3552 }, { "epoch": 0.4074074074074074, "grad_norm": 12.5, "learning_rate": 6.709422101375538e-06, "loss": 1.5874, "step": 3553 }, { "epoch": 0.4075220731567481, "grad_norm": 13.25, "learning_rate": 6.707676927021878e-06, "loss": 1.6622, "step": 3554 }, { "epoch": 0.40763673890608876, "grad_norm": 14.8125, "learning_rate": 6.7059315171267385e-06, "loss": 1.6882, "step": 3555 }, { "epoch": 0.4077514046554294, "grad_norm": 15.5, "learning_rate": 6.7041858719308615e-06, "loss": 1.7238, "step": 3556 }, { "epoch": 0.4078660704047701, "grad_norm": 17.25, "learning_rate": 6.70243999167503e-06, "loss": 2.1611, "step": 3557 }, { "epoch": 0.4079807361541108, "grad_norm": 12.1875, "learning_rate": 6.700693876600053e-06, "loss": 1.4956, "step": 3558 }, { "epoch": 0.40809540190345145, "grad_norm": 12.8125, "learning_rate": 6.698947526946774e-06, "loss": 1.4931, "step": 3559 }, { "epoch": 0.4082100676527921, "grad_norm": 17.0, "learning_rate": 6.697200942956068e-06, "loss": 2.0333, "step": 3560 }, { "epoch": 0.40832473340213277, "grad_norm": 16.25, "learning_rate": 6.695454124868843e-06, "loss": 2.0511, "step": 3561 }, { "epoch": 0.4084393991514735, "grad_norm": 19.125, "learning_rate": 6.693707072926041e-06, "loss": 1.6335, "step": 3562 }, { "epoch": 0.40855406490081414, "grad_norm": 18.125, "learning_rate": 6.691959787368633e-06, "loss": 1.8449, "step": 3563 }, { "epoch": 0.4086687306501548, "grad_norm": 14.4375, "learning_rate": 6.690212268437624e-06, "loss": 1.7624, "step": 3564 }, { "epoch": 0.40878339639949546, "grad_norm": 16.625, "learning_rate": 6.688464516374051e-06, "loss": 2.0421, "step": 3565 }, { "epoch": 0.4088980621488361, "grad_norm": 12.8125, "learning_rate": 6.686716531418982e-06, "loss": 1.5336, "step": 3566 }, { "epoch": 0.40901272789817683, "grad_norm": 12.6875, "learning_rate": 6.68496831381352e-06, "loss": 1.7958, "step": 3567 }, { "epoch": 0.4091273936475175, "grad_norm": 28.125, "learning_rate": 6.683219863798797e-06, "loss": 1.7431, "step": 3568 }, { "epoch": 0.40924205939685815, "grad_norm": 22.625, "learning_rate": 6.681471181615979e-06, "loss": 1.7346, "step": 3569 }, { "epoch": 0.4093567251461988, "grad_norm": 15.375, "learning_rate": 6.679722267506266e-06, "loss": 1.7984, "step": 3570 }, { "epoch": 0.4094713908955395, "grad_norm": 14.875, "learning_rate": 6.677973121710881e-06, "loss": 1.7286, "step": 3571 }, { "epoch": 0.4095860566448802, "grad_norm": 14.3125, "learning_rate": 6.676223744471091e-06, "loss": 1.6912, "step": 3572 }, { "epoch": 0.40970072239422084, "grad_norm": 16.25, "learning_rate": 6.674474136028186e-06, "loss": 1.7252, "step": 3573 }, { "epoch": 0.4098153881435615, "grad_norm": 16.125, "learning_rate": 6.672724296623493e-06, "loss": 1.9151, "step": 3574 }, { "epoch": 0.4099300538929022, "grad_norm": 15.1875, "learning_rate": 6.6709742264983676e-06, "loss": 1.6545, "step": 3575 }, { "epoch": 0.41004471964224287, "grad_norm": 21.5, "learning_rate": 6.669223925894199e-06, "loss": 2.0642, "step": 3576 }, { "epoch": 0.41015938539158353, "grad_norm": 12.875, "learning_rate": 6.6674733950524104e-06, "loss": 2.015, "step": 3577 }, { "epoch": 0.4102740511409242, "grad_norm": 16.125, "learning_rate": 6.66572263421445e-06, "loss": 1.8091, "step": 3578 }, { "epoch": 0.4103887168902649, "grad_norm": 10.9375, "learning_rate": 6.663971643621804e-06, "loss": 1.4297, "step": 3579 }, { "epoch": 0.41050338263960556, "grad_norm": 16.625, "learning_rate": 6.662220423515988e-06, "loss": 1.5625, "step": 3580 }, { "epoch": 0.4106180483889462, "grad_norm": 15.625, "learning_rate": 6.6604689741385495e-06, "loss": 1.4388, "step": 3581 }, { "epoch": 0.4107327141382869, "grad_norm": 12.875, "learning_rate": 6.658717295731067e-06, "loss": 1.4115, "step": 3582 }, { "epoch": 0.41084737988762754, "grad_norm": 11.625, "learning_rate": 6.656965388535154e-06, "loss": 1.7332, "step": 3583 }, { "epoch": 0.41096204563696825, "grad_norm": 14.25, "learning_rate": 6.655213252792446e-06, "loss": 1.7211, "step": 3584 }, { "epoch": 0.4110767113863089, "grad_norm": 14.25, "learning_rate": 6.653460888744625e-06, "loss": 1.4622, "step": 3585 }, { "epoch": 0.41119137713564957, "grad_norm": 13.4375, "learning_rate": 6.65170829663339e-06, "loss": 1.7217, "step": 3586 }, { "epoch": 0.41130604288499023, "grad_norm": 13.875, "learning_rate": 6.649955476700481e-06, "loss": 1.5842, "step": 3587 }, { "epoch": 0.41142070863433094, "grad_norm": 15.0625, "learning_rate": 6.648202429187665e-06, "loss": 1.838, "step": 3588 }, { "epoch": 0.4115353743836716, "grad_norm": 15.625, "learning_rate": 6.6464491543367394e-06, "loss": 1.8686, "step": 3589 }, { "epoch": 0.41165004013301226, "grad_norm": 34.5, "learning_rate": 6.644695652389539e-06, "loss": 1.5266, "step": 3590 }, { "epoch": 0.4117647058823529, "grad_norm": 30.0, "learning_rate": 6.642941923587923e-06, "loss": 1.7928, "step": 3591 }, { "epoch": 0.41187937163169364, "grad_norm": 40.0, "learning_rate": 6.641187968173788e-06, "loss": 2.2762, "step": 3592 }, { "epoch": 0.4119940373810343, "grad_norm": 11.875, "learning_rate": 6.639433786389055e-06, "loss": 1.5639, "step": 3593 }, { "epoch": 0.41210870313037495, "grad_norm": 19.375, "learning_rate": 6.637679378475681e-06, "loss": 2.0079, "step": 3594 }, { "epoch": 0.4122233688797156, "grad_norm": 13.25, "learning_rate": 6.6359247446756556e-06, "loss": 1.8699, "step": 3595 }, { "epoch": 0.4123380346290563, "grad_norm": 15.6875, "learning_rate": 6.634169885230994e-06, "loss": 2.1254, "step": 3596 }, { "epoch": 0.412452700378397, "grad_norm": 13.0625, "learning_rate": 6.632414800383747e-06, "loss": 1.7148, "step": 3597 }, { "epoch": 0.41256736612773764, "grad_norm": 15.9375, "learning_rate": 6.630659490375997e-06, "loss": 1.4923, "step": 3598 }, { "epoch": 0.4126820318770783, "grad_norm": 16.0, "learning_rate": 6.62890395544985e-06, "loss": 2.1783, "step": 3599 }, { "epoch": 0.41279669762641896, "grad_norm": 15.8125, "learning_rate": 6.627148195847455e-06, "loss": 1.5556, "step": 3600 }, { "epoch": 0.4129113633757597, "grad_norm": 12.5, "learning_rate": 6.625392211810981e-06, "loss": 1.6068, "step": 3601 }, { "epoch": 0.41302602912510034, "grad_norm": 13.0, "learning_rate": 6.623636003582634e-06, "loss": 1.8515, "step": 3602 }, { "epoch": 0.413140694874441, "grad_norm": 15.75, "learning_rate": 6.62187957140465e-06, "loss": 1.999, "step": 3603 }, { "epoch": 0.41325536062378165, "grad_norm": 11.375, "learning_rate": 6.620122915519296e-06, "loss": 1.5799, "step": 3604 }, { "epoch": 0.41337002637312237, "grad_norm": 12.5, "learning_rate": 6.618366036168867e-06, "loss": 1.7053, "step": 3605 }, { "epoch": 0.413484692122463, "grad_norm": 13.75, "learning_rate": 6.616608933595692e-06, "loss": 1.9363, "step": 3606 }, { "epoch": 0.4135993578718037, "grad_norm": 12.5, "learning_rate": 6.614851608042131e-06, "loss": 1.5646, "step": 3607 }, { "epoch": 0.41371402362114434, "grad_norm": 19.5, "learning_rate": 6.613094059750574e-06, "loss": 2.2077, "step": 3608 }, { "epoch": 0.41382868937048506, "grad_norm": 12.8125, "learning_rate": 6.611336288963436e-06, "loss": 2.2354, "step": 3609 }, { "epoch": 0.4139433551198257, "grad_norm": 15.125, "learning_rate": 6.6095782959231735e-06, "loss": 1.6602, "step": 3610 }, { "epoch": 0.4140580208691664, "grad_norm": 13.9375, "learning_rate": 6.607820080872267e-06, "loss": 1.7156, "step": 3611 }, { "epoch": 0.41417268661850704, "grad_norm": 13.75, "learning_rate": 6.606061644053228e-06, "loss": 1.7914, "step": 3612 }, { "epoch": 0.41428735236784775, "grad_norm": 13.25, "learning_rate": 6.6043029857085995e-06, "loss": 1.7303, "step": 3613 }, { "epoch": 0.4144020181171884, "grad_norm": 12.875, "learning_rate": 6.602544106080955e-06, "loss": 1.4129, "step": 3614 }, { "epoch": 0.41451668386652907, "grad_norm": 33.25, "learning_rate": 6.600785005412897e-06, "loss": 1.593, "step": 3615 }, { "epoch": 0.4146313496158697, "grad_norm": 13.8125, "learning_rate": 6.599025683947062e-06, "loss": 1.422, "step": 3616 }, { "epoch": 0.4147460153652104, "grad_norm": 16.5, "learning_rate": 6.5972661419261156e-06, "loss": 1.5149, "step": 3617 }, { "epoch": 0.4148606811145511, "grad_norm": 23.375, "learning_rate": 6.59550637959275e-06, "loss": 1.7158, "step": 3618 }, { "epoch": 0.41497534686389176, "grad_norm": 17.125, "learning_rate": 6.593746397189692e-06, "loss": 1.6159, "step": 3619 }, { "epoch": 0.4150900126132324, "grad_norm": 15.0, "learning_rate": 6.5919861949597e-06, "loss": 1.7413, "step": 3620 }, { "epoch": 0.4152046783625731, "grad_norm": 13.0625, "learning_rate": 6.590225773145557e-06, "loss": 1.4218, "step": 3621 }, { "epoch": 0.4153193441119138, "grad_norm": 15.5, "learning_rate": 6.58846513199008e-06, "loss": 1.7667, "step": 3622 }, { "epoch": 0.41543400986125445, "grad_norm": 15.0, "learning_rate": 6.58670427173612e-06, "loss": 1.9516, "step": 3623 }, { "epoch": 0.4155486756105951, "grad_norm": 15.5, "learning_rate": 6.58494319262655e-06, "loss": 2.0586, "step": 3624 }, { "epoch": 0.41566334135993577, "grad_norm": 11.9375, "learning_rate": 6.583181894904277e-06, "loss": 1.7139, "step": 3625 }, { "epoch": 0.4157780071092765, "grad_norm": 17.25, "learning_rate": 6.58142037881224e-06, "loss": 1.9602, "step": 3626 }, { "epoch": 0.41589267285861714, "grad_norm": 16.5, "learning_rate": 6.579658644593408e-06, "loss": 1.6914, "step": 3627 }, { "epoch": 0.4160073386079578, "grad_norm": 13.75, "learning_rate": 6.577896692490775e-06, "loss": 1.557, "step": 3628 }, { "epoch": 0.41612200435729846, "grad_norm": 12.8125, "learning_rate": 6.576134522747372e-06, "loss": 1.5801, "step": 3629 }, { "epoch": 0.4162366701066392, "grad_norm": 16.875, "learning_rate": 6.574372135606255e-06, "loss": 1.4811, "step": 3630 }, { "epoch": 0.41635133585597983, "grad_norm": 15.875, "learning_rate": 6.572609531310511e-06, "loss": 1.8998, "step": 3631 }, { "epoch": 0.4164660016053205, "grad_norm": 15.0, "learning_rate": 6.570846710103261e-06, "loss": 1.5496, "step": 3632 }, { "epoch": 0.41658066735466115, "grad_norm": 11.9375, "learning_rate": 6.56908367222765e-06, "loss": 1.8714, "step": 3633 }, { "epoch": 0.4166953331040018, "grad_norm": 13.5, "learning_rate": 6.567320417926854e-06, "loss": 1.4693, "step": 3634 }, { "epoch": 0.4168099988533425, "grad_norm": 12.75, "learning_rate": 6.565556947444085e-06, "loss": 1.7857, "step": 3635 }, { "epoch": 0.4169246646026832, "grad_norm": 15.8125, "learning_rate": 6.563793261022575e-06, "loss": 1.8141, "step": 3636 }, { "epoch": 0.41703933035202384, "grad_norm": 13.1875, "learning_rate": 6.562029358905594e-06, "loss": 1.6745, "step": 3637 }, { "epoch": 0.4171539961013645, "grad_norm": 14.5, "learning_rate": 6.560265241336439e-06, "loss": 1.5878, "step": 3638 }, { "epoch": 0.4172686618507052, "grad_norm": 11.3125, "learning_rate": 6.558500908558434e-06, "loss": 1.7147, "step": 3639 }, { "epoch": 0.4173833276000459, "grad_norm": 33.0, "learning_rate": 6.5567363608149355e-06, "loss": 2.2711, "step": 3640 }, { "epoch": 0.41749799334938653, "grad_norm": 15.4375, "learning_rate": 6.554971598349331e-06, "loss": 1.7823, "step": 3641 }, { "epoch": 0.4176126590987272, "grad_norm": 14.5625, "learning_rate": 6.553206621405037e-06, "loss": 1.6839, "step": 3642 }, { "epoch": 0.4177273248480679, "grad_norm": 27.625, "learning_rate": 6.551441430225494e-06, "loss": 1.5111, "step": 3643 }, { "epoch": 0.41784199059740856, "grad_norm": 14.625, "learning_rate": 6.5496760250541795e-06, "loss": 1.846, "step": 3644 }, { "epoch": 0.4179566563467492, "grad_norm": 14.3125, "learning_rate": 6.547910406134598e-06, "loss": 1.6858, "step": 3645 }, { "epoch": 0.4180713220960899, "grad_norm": 13.5, "learning_rate": 6.54614457371028e-06, "loss": 1.8339, "step": 3646 }, { "epoch": 0.4181859878454306, "grad_norm": 32.0, "learning_rate": 6.5443785280247905e-06, "loss": 2.0429, "step": 3647 }, { "epoch": 0.41830065359477125, "grad_norm": 18.875, "learning_rate": 6.542612269321723e-06, "loss": 2.1466, "step": 3648 }, { "epoch": 0.4184153193441119, "grad_norm": 14.75, "learning_rate": 6.540845797844697e-06, "loss": 1.5626, "step": 3649 }, { "epoch": 0.4185299850934526, "grad_norm": 15.9375, "learning_rate": 6.539079113837363e-06, "loss": 1.7961, "step": 3650 }, { "epoch": 0.41864465084279323, "grad_norm": 15.8125, "learning_rate": 6.5373122175434045e-06, "loss": 1.8397, "step": 3651 }, { "epoch": 0.41875931659213395, "grad_norm": 15.375, "learning_rate": 6.535545109206529e-06, "loss": 1.9434, "step": 3652 }, { "epoch": 0.4188739823414746, "grad_norm": 14.125, "learning_rate": 6.5337777890704755e-06, "loss": 1.8168, "step": 3653 }, { "epoch": 0.41898864809081526, "grad_norm": 14.5, "learning_rate": 6.532010257379013e-06, "loss": 1.693, "step": 3654 }, { "epoch": 0.4191033138401559, "grad_norm": 15.875, "learning_rate": 6.530242514375938e-06, "loss": 1.4838, "step": 3655 }, { "epoch": 0.41921797958949664, "grad_norm": 16.625, "learning_rate": 6.528474560305077e-06, "loss": 1.9626, "step": 3656 }, { "epoch": 0.4193326453388373, "grad_norm": 13.125, "learning_rate": 6.526706395410287e-06, "loss": 1.6279, "step": 3657 }, { "epoch": 0.41944731108817795, "grad_norm": 13.625, "learning_rate": 6.52493801993545e-06, "loss": 1.7193, "step": 3658 }, { "epoch": 0.4195619768375186, "grad_norm": 26.75, "learning_rate": 6.523169434124481e-06, "loss": 2.0768, "step": 3659 }, { "epoch": 0.4196766425868593, "grad_norm": 12.625, "learning_rate": 6.5214006382213245e-06, "loss": 1.7757, "step": 3660 }, { "epoch": 0.4197913083362, "grad_norm": 25.625, "learning_rate": 6.51963163246995e-06, "loss": 1.9391, "step": 3661 }, { "epoch": 0.41990597408554065, "grad_norm": 27.375, "learning_rate": 6.5178624171143565e-06, "loss": 1.9745, "step": 3662 }, { "epoch": 0.4200206398348813, "grad_norm": 14.1875, "learning_rate": 6.516092992398578e-06, "loss": 2.1429, "step": 3663 }, { "epoch": 0.420135305584222, "grad_norm": 16.5, "learning_rate": 6.51432335856667e-06, "loss": 1.9127, "step": 3664 }, { "epoch": 0.4202499713335627, "grad_norm": 31.125, "learning_rate": 6.5125535158627185e-06, "loss": 1.6508, "step": 3665 }, { "epoch": 0.42036463708290334, "grad_norm": 13.625, "learning_rate": 6.510783464530843e-06, "loss": 2.1912, "step": 3666 }, { "epoch": 0.420479302832244, "grad_norm": 17.625, "learning_rate": 6.509013204815188e-06, "loss": 1.778, "step": 3667 }, { "epoch": 0.42059396858158465, "grad_norm": 15.875, "learning_rate": 6.507242736959923e-06, "loss": 1.892, "step": 3668 }, { "epoch": 0.42070863433092537, "grad_norm": 16.75, "learning_rate": 6.505472061209254e-06, "loss": 1.3963, "step": 3669 }, { "epoch": 0.420823300080266, "grad_norm": 16.25, "learning_rate": 6.503701177807413e-06, "loss": 1.8162, "step": 3670 }, { "epoch": 0.4209379658296067, "grad_norm": 14.625, "learning_rate": 6.501930086998656e-06, "loss": 1.6202, "step": 3671 }, { "epoch": 0.42105263157894735, "grad_norm": 12.0, "learning_rate": 6.500158789027273e-06, "loss": 1.7846, "step": 3672 }, { "epoch": 0.42116729732828806, "grad_norm": 16.875, "learning_rate": 6.498387284137581e-06, "loss": 1.7212, "step": 3673 }, { "epoch": 0.4212819630776287, "grad_norm": 15.6875, "learning_rate": 6.496615572573926e-06, "loss": 1.8779, "step": 3674 }, { "epoch": 0.4213966288269694, "grad_norm": 11.125, "learning_rate": 6.494843654580678e-06, "loss": 1.8058, "step": 3675 }, { "epoch": 0.42151129457631004, "grad_norm": 50.5, "learning_rate": 6.493071530402244e-06, "loss": 2.0255, "step": 3676 }, { "epoch": 0.42162596032565075, "grad_norm": 15.625, "learning_rate": 6.491299200283054e-06, "loss": 1.745, "step": 3677 }, { "epoch": 0.4217406260749914, "grad_norm": 16.125, "learning_rate": 6.489526664467564e-06, "loss": 1.9898, "step": 3678 }, { "epoch": 0.42185529182433207, "grad_norm": 15.8125, "learning_rate": 6.487753923200264e-06, "loss": 1.8106, "step": 3679 }, { "epoch": 0.4219699575736727, "grad_norm": 21.125, "learning_rate": 6.485980976725671e-06, "loss": 1.9303, "step": 3680 }, { "epoch": 0.42208462332301344, "grad_norm": 13.8125, "learning_rate": 6.484207825288325e-06, "loss": 1.5492, "step": 3681 }, { "epoch": 0.4221992890723541, "grad_norm": 17.125, "learning_rate": 6.482434469132803e-06, "loss": 2.0337, "step": 3682 }, { "epoch": 0.42231395482169476, "grad_norm": 14.9375, "learning_rate": 6.480660908503704e-06, "loss": 2.552, "step": 3683 }, { "epoch": 0.4224286205710354, "grad_norm": 13.125, "learning_rate": 6.4788871436456555e-06, "loss": 1.4304, "step": 3684 }, { "epoch": 0.4225432863203761, "grad_norm": 15.1875, "learning_rate": 6.477113174803317e-06, "loss": 1.7638, "step": 3685 }, { "epoch": 0.4226579520697168, "grad_norm": 15.5, "learning_rate": 6.4753390022213715e-06, "loss": 1.7913, "step": 3686 }, { "epoch": 0.42277261781905745, "grad_norm": 13.8125, "learning_rate": 6.473564626144533e-06, "loss": 1.8381, "step": 3687 }, { "epoch": 0.4228872835683981, "grad_norm": 14.0, "learning_rate": 6.471790046817543e-06, "loss": 1.9886, "step": 3688 }, { "epoch": 0.42300194931773877, "grad_norm": 13.9375, "learning_rate": 6.470015264485172e-06, "loss": 1.636, "step": 3689 }, { "epoch": 0.4231166150670795, "grad_norm": 14.9375, "learning_rate": 6.468240279392215e-06, "loss": 1.6468, "step": 3690 }, { "epoch": 0.42323128081642014, "grad_norm": 13.125, "learning_rate": 6.466465091783498e-06, "loss": 1.5892, "step": 3691 }, { "epoch": 0.4233459465657608, "grad_norm": 11.6875, "learning_rate": 6.464689701903876e-06, "loss": 1.4223, "step": 3692 }, { "epoch": 0.42346061231510146, "grad_norm": 26.625, "learning_rate": 6.462914109998231e-06, "loss": 1.8121, "step": 3693 }, { "epoch": 0.4235752780644422, "grad_norm": 14.0, "learning_rate": 6.461138316311467e-06, "loss": 1.7654, "step": 3694 }, { "epoch": 0.42368994381378283, "grad_norm": 17.25, "learning_rate": 6.459362321088527e-06, "loss": 1.8992, "step": 3695 }, { "epoch": 0.4238046095631235, "grad_norm": 13.4375, "learning_rate": 6.457586124574371e-06, "loss": 1.564, "step": 3696 }, { "epoch": 0.42391927531246415, "grad_norm": 14.0, "learning_rate": 6.455809727013993e-06, "loss": 1.4349, "step": 3697 }, { "epoch": 0.42403394106180486, "grad_norm": 15.25, "learning_rate": 6.454033128652415e-06, "loss": 1.5074, "step": 3698 }, { "epoch": 0.4241486068111455, "grad_norm": 16.0, "learning_rate": 6.452256329734682e-06, "loss": 2.2741, "step": 3699 }, { "epoch": 0.4242632725604862, "grad_norm": 16.25, "learning_rate": 6.45047933050587e-06, "loss": 2.0194, "step": 3700 }, { "epoch": 0.42437793830982684, "grad_norm": 12.3125, "learning_rate": 6.448702131211084e-06, "loss": 1.5167, "step": 3701 }, { "epoch": 0.4244926040591675, "grad_norm": 14.3125, "learning_rate": 6.446924732095455e-06, "loss": 1.5982, "step": 3702 }, { "epoch": 0.4246072698085082, "grad_norm": 13.375, "learning_rate": 6.445147133404139e-06, "loss": 1.8159, "step": 3703 }, { "epoch": 0.4247219355578489, "grad_norm": 13.6875, "learning_rate": 6.443369335382322e-06, "loss": 2.1083, "step": 3704 }, { "epoch": 0.42483660130718953, "grad_norm": 17.125, "learning_rate": 6.4415913382752205e-06, "loss": 1.6691, "step": 3705 }, { "epoch": 0.4249512670565302, "grad_norm": 16.5, "learning_rate": 6.4398131423280704e-06, "loss": 1.8686, "step": 3706 }, { "epoch": 0.4250659328058709, "grad_norm": 12.625, "learning_rate": 6.438034747786145e-06, "loss": 1.7727, "step": 3707 }, { "epoch": 0.42518059855521156, "grad_norm": 21.125, "learning_rate": 6.4362561548947375e-06, "loss": 1.5015, "step": 3708 }, { "epoch": 0.4252952643045522, "grad_norm": 12.5, "learning_rate": 6.434477363899169e-06, "loss": 1.8305, "step": 3709 }, { "epoch": 0.4254099300538929, "grad_norm": 13.75, "learning_rate": 6.432698375044794e-06, "loss": 1.7592, "step": 3710 }, { "epoch": 0.4255245958032336, "grad_norm": 13.1875, "learning_rate": 6.430919188576986e-06, "loss": 1.8389, "step": 3711 }, { "epoch": 0.42563926155257426, "grad_norm": 12.75, "learning_rate": 6.429139804741152e-06, "loss": 1.791, "step": 3712 }, { "epoch": 0.4257539273019149, "grad_norm": 16.25, "learning_rate": 6.427360223782726e-06, "loss": 1.6685, "step": 3713 }, { "epoch": 0.4258685930512556, "grad_norm": 15.25, "learning_rate": 6.4255804459471625e-06, "loss": 1.7256, "step": 3714 }, { "epoch": 0.4259832588005963, "grad_norm": 21.5, "learning_rate": 6.423800471479951e-06, "loss": 1.8317, "step": 3715 }, { "epoch": 0.42609792454993695, "grad_norm": 12.5625, "learning_rate": 6.4220203006266045e-06, "loss": 1.33, "step": 3716 }, { "epoch": 0.4262125902992776, "grad_norm": 15.1875, "learning_rate": 6.420239933632666e-06, "loss": 2.0763, "step": 3717 }, { "epoch": 0.42632725604861826, "grad_norm": 14.0625, "learning_rate": 6.4184593707436986e-06, "loss": 1.9186, "step": 3718 }, { "epoch": 0.4264419217979589, "grad_norm": 13.875, "learning_rate": 6.4166786122052994e-06, "loss": 1.7927, "step": 3719 }, { "epoch": 0.42655658754729964, "grad_norm": 20.0, "learning_rate": 6.414897658263091e-06, "loss": 2.1536, "step": 3720 }, { "epoch": 0.4266712532966403, "grad_norm": 12.9375, "learning_rate": 6.41311650916272e-06, "loss": 1.7836, "step": 3721 }, { "epoch": 0.42678591904598095, "grad_norm": 12.25, "learning_rate": 6.411335165149863e-06, "loss": 1.5013, "step": 3722 }, { "epoch": 0.4269005847953216, "grad_norm": 14.125, "learning_rate": 6.409553626470224e-06, "loss": 1.7224, "step": 3723 }, { "epoch": 0.42701525054466233, "grad_norm": 14.6875, "learning_rate": 6.40777189336953e-06, "loss": 1.5997, "step": 3724 }, { "epoch": 0.427129916294003, "grad_norm": 13.75, "learning_rate": 6.4059899660935366e-06, "loss": 1.6868, "step": 3725 }, { "epoch": 0.42724458204334365, "grad_norm": 11.875, "learning_rate": 6.404207844888028e-06, "loss": 1.7533, "step": 3726 }, { "epoch": 0.4273592477926843, "grad_norm": 30.875, "learning_rate": 6.402425529998817e-06, "loss": 1.7424, "step": 3727 }, { "epoch": 0.427473913542025, "grad_norm": 13.5, "learning_rate": 6.400643021671734e-06, "loss": 1.6568, "step": 3728 }, { "epoch": 0.4275885792913657, "grad_norm": 21.75, "learning_rate": 6.3988603201526446e-06, "loss": 1.5152, "step": 3729 }, { "epoch": 0.42770324504070634, "grad_norm": 13.0, "learning_rate": 6.397077425687442e-06, "loss": 1.5953, "step": 3730 }, { "epoch": 0.427817910790047, "grad_norm": 15.5625, "learning_rate": 6.395294338522035e-06, "loss": 1.7132, "step": 3731 }, { "epoch": 0.4279325765393877, "grad_norm": 14.625, "learning_rate": 6.393511058902373e-06, "loss": 1.6914, "step": 3732 }, { "epoch": 0.42804724228872837, "grad_norm": 16.125, "learning_rate": 6.391727587074423e-06, "loss": 1.9801, "step": 3733 }, { "epoch": 0.42816190803806903, "grad_norm": 13.3125, "learning_rate": 6.38994392328418e-06, "loss": 1.732, "step": 3734 }, { "epoch": 0.4282765737874097, "grad_norm": 14.625, "learning_rate": 6.388160067777669e-06, "loss": 1.598, "step": 3735 }, { "epoch": 0.42839123953675035, "grad_norm": 12.9375, "learning_rate": 6.386376020800936e-06, "loss": 1.8332, "step": 3736 }, { "epoch": 0.42850590528609106, "grad_norm": 12.9375, "learning_rate": 6.384591782600058e-06, "loss": 1.8259, "step": 3737 }, { "epoch": 0.4286205710354317, "grad_norm": 14.0, "learning_rate": 6.382807353421138e-06, "loss": 1.7712, "step": 3738 }, { "epoch": 0.4287352367847724, "grad_norm": 16.125, "learning_rate": 6.381022733510301e-06, "loss": 1.7365, "step": 3739 }, { "epoch": 0.42884990253411304, "grad_norm": 13.25, "learning_rate": 6.379237923113702e-06, "loss": 1.4987, "step": 3740 }, { "epoch": 0.42896456828345375, "grad_norm": 14.375, "learning_rate": 6.377452922477523e-06, "loss": 1.6636, "step": 3741 }, { "epoch": 0.4290792340327944, "grad_norm": 12.25, "learning_rate": 6.3756677318479695e-06, "loss": 1.618, "step": 3742 }, { "epoch": 0.42919389978213507, "grad_norm": 12.5, "learning_rate": 6.373882351471276e-06, "loss": 1.8378, "step": 3743 }, { "epoch": 0.4293085655314757, "grad_norm": 11.1875, "learning_rate": 6.372096781593699e-06, "loss": 1.6589, "step": 3744 }, { "epoch": 0.42942323128081644, "grad_norm": 13.75, "learning_rate": 6.370311022461528e-06, "loss": 1.6058, "step": 3745 }, { "epoch": 0.4295378970301571, "grad_norm": 16.375, "learning_rate": 6.368525074321071e-06, "loss": 1.9451, "step": 3746 }, { "epoch": 0.42965256277949776, "grad_norm": 17.375, "learning_rate": 6.3667389374186635e-06, "loss": 1.7748, "step": 3747 }, { "epoch": 0.4297672285288384, "grad_norm": 11.25, "learning_rate": 6.3649526120006764e-06, "loss": 1.6239, "step": 3748 }, { "epoch": 0.42988189427817913, "grad_norm": 13.9375, "learning_rate": 6.363166098313493e-06, "loss": 1.5925, "step": 3749 }, { "epoch": 0.4299965600275198, "grad_norm": 28.75, "learning_rate": 6.36137939660353e-06, "loss": 1.8434, "step": 3750 }, { "epoch": 0.43011122577686045, "grad_norm": 14.5625, "learning_rate": 6.3595925071172295e-06, "loss": 1.3121, "step": 3751 }, { "epoch": 0.4302258915262011, "grad_norm": 16.625, "learning_rate": 6.35780543010106e-06, "loss": 2.2962, "step": 3752 }, { "epoch": 0.43034055727554177, "grad_norm": 13.8125, "learning_rate": 6.356018165801512e-06, "loss": 1.4989, "step": 3753 }, { "epoch": 0.4304552230248825, "grad_norm": 12.8125, "learning_rate": 6.354230714465107e-06, "loss": 1.7888, "step": 3754 }, { "epoch": 0.43056988877422314, "grad_norm": 14.375, "learning_rate": 6.35244307633839e-06, "loss": 1.9128, "step": 3755 }, { "epoch": 0.4306845545235638, "grad_norm": 11.375, "learning_rate": 6.350655251667927e-06, "loss": 1.432, "step": 3756 }, { "epoch": 0.43079922027290446, "grad_norm": 15.5, "learning_rate": 6.348867240700322e-06, "loss": 1.6147, "step": 3757 }, { "epoch": 0.4309138860222452, "grad_norm": 37.25, "learning_rate": 6.347079043682192e-06, "loss": 2.5028, "step": 3758 }, { "epoch": 0.43102855177158583, "grad_norm": 14.1875, "learning_rate": 6.345290660860184e-06, "loss": 1.6206, "step": 3759 }, { "epoch": 0.4311432175209265, "grad_norm": 14.4375, "learning_rate": 6.343502092480973e-06, "loss": 1.6841, "step": 3760 }, { "epoch": 0.43125788327026715, "grad_norm": 18.0, "learning_rate": 6.341713338791259e-06, "loss": 1.6717, "step": 3761 }, { "epoch": 0.43137254901960786, "grad_norm": 14.8125, "learning_rate": 6.339924400037765e-06, "loss": 1.382, "step": 3762 }, { "epoch": 0.4314872147689485, "grad_norm": 13.375, "learning_rate": 6.338135276467241e-06, "loss": 1.6569, "step": 3763 }, { "epoch": 0.4316018805182892, "grad_norm": 19.625, "learning_rate": 6.336345968326463e-06, "loss": 1.8409, "step": 3764 }, { "epoch": 0.43171654626762984, "grad_norm": 13.6875, "learning_rate": 6.334556475862231e-06, "loss": 1.5994, "step": 3765 }, { "epoch": 0.43183121201697056, "grad_norm": 15.9375, "learning_rate": 6.332766799321373e-06, "loss": 1.5775, "step": 3766 }, { "epoch": 0.4319458777663112, "grad_norm": 15.5, "learning_rate": 6.33097693895074e-06, "loss": 1.7032, "step": 3767 }, { "epoch": 0.4320605435156519, "grad_norm": 16.375, "learning_rate": 6.329186894997208e-06, "loss": 1.7084, "step": 3768 }, { "epoch": 0.43217520926499253, "grad_norm": 15.6875, "learning_rate": 6.327396667707681e-06, "loss": 2.1154, "step": 3769 }, { "epoch": 0.4322898750143332, "grad_norm": 14.0, "learning_rate": 6.325606257329087e-06, "loss": 1.4025, "step": 3770 }, { "epoch": 0.4324045407636739, "grad_norm": 15.25, "learning_rate": 6.323815664108377e-06, "loss": 2.0579, "step": 3771 }, { "epoch": 0.43251920651301456, "grad_norm": 13.75, "learning_rate": 6.322024888292531e-06, "loss": 2.311, "step": 3772 }, { "epoch": 0.4326338722623552, "grad_norm": 16.875, "learning_rate": 6.320233930128552e-06, "loss": 1.6215, "step": 3773 }, { "epoch": 0.4327485380116959, "grad_norm": 15.0625, "learning_rate": 6.318442789863467e-06, "loss": 1.9308, "step": 3774 }, { "epoch": 0.4328632037610366, "grad_norm": 17.25, "learning_rate": 6.316651467744332e-06, "loss": 1.5918, "step": 3775 }, { "epoch": 0.43297786951037726, "grad_norm": 12.875, "learning_rate": 6.314859964018224e-06, "loss": 1.9996, "step": 3776 }, { "epoch": 0.4330925352597179, "grad_norm": 14.875, "learning_rate": 6.313068278932249e-06, "loss": 1.3987, "step": 3777 }, { "epoch": 0.4332072010090586, "grad_norm": 13.375, "learning_rate": 6.311276412733533e-06, "loss": 1.6763, "step": 3778 }, { "epoch": 0.4333218667583993, "grad_norm": 12.625, "learning_rate": 6.309484365669231e-06, "loss": 1.4813, "step": 3779 }, { "epoch": 0.43343653250773995, "grad_norm": 14.375, "learning_rate": 6.307692137986522e-06, "loss": 1.8661, "step": 3780 }, { "epoch": 0.4335511982570806, "grad_norm": 16.0, "learning_rate": 6.305899729932609e-06, "loss": 1.7383, "step": 3781 }, { "epoch": 0.43366586400642126, "grad_norm": 12.5, "learning_rate": 6.3041071417547215e-06, "loss": 1.5742, "step": 3782 }, { "epoch": 0.433780529755762, "grad_norm": 16.875, "learning_rate": 6.3023143737001135e-06, "loss": 2.122, "step": 3783 }, { "epoch": 0.43389519550510264, "grad_norm": 17.125, "learning_rate": 6.300521426016059e-06, "loss": 1.7999, "step": 3784 }, { "epoch": 0.4340098612544433, "grad_norm": 14.625, "learning_rate": 6.2987282989498664e-06, "loss": 1.5383, "step": 3785 }, { "epoch": 0.43412452700378396, "grad_norm": 15.1875, "learning_rate": 6.29693499274886e-06, "loss": 1.6027, "step": 3786 }, { "epoch": 0.4342391927531246, "grad_norm": 17.5, "learning_rate": 6.295141507660394e-06, "loss": 1.9468, "step": 3787 }, { "epoch": 0.43435385850246533, "grad_norm": 15.75, "learning_rate": 6.293347843931844e-06, "loss": 1.63, "step": 3788 }, { "epoch": 0.434468524251806, "grad_norm": 11.5625, "learning_rate": 6.291554001810613e-06, "loss": 1.8515, "step": 3789 }, { "epoch": 0.43458319000114665, "grad_norm": 21.625, "learning_rate": 6.289759981544125e-06, "loss": 1.9227, "step": 3790 }, { "epoch": 0.4346978557504873, "grad_norm": 13.0, "learning_rate": 6.287965783379835e-06, "loss": 1.6269, "step": 3791 }, { "epoch": 0.434812521499828, "grad_norm": 13.75, "learning_rate": 6.286171407565215e-06, "loss": 1.7389, "step": 3792 }, { "epoch": 0.4349271872491687, "grad_norm": 12.6875, "learning_rate": 6.284376854347766e-06, "loss": 1.643, "step": 3793 }, { "epoch": 0.43504185299850934, "grad_norm": 12.625, "learning_rate": 6.282582123975011e-06, "loss": 1.4857, "step": 3794 }, { "epoch": 0.43515651874785, "grad_norm": 14.4375, "learning_rate": 6.280787216694503e-06, "loss": 1.7025, "step": 3795 }, { "epoch": 0.4352711844971907, "grad_norm": 12.75, "learning_rate": 6.27899213275381e-06, "loss": 1.6753, "step": 3796 }, { "epoch": 0.43538585024653137, "grad_norm": 16.5, "learning_rate": 6.277196872400534e-06, "loss": 2.0161, "step": 3797 }, { "epoch": 0.43550051599587203, "grad_norm": 14.5, "learning_rate": 6.275401435882294e-06, "loss": 1.988, "step": 3798 }, { "epoch": 0.4356151817452127, "grad_norm": 17.25, "learning_rate": 6.273605823446738e-06, "loss": 1.5168, "step": 3799 }, { "epoch": 0.4357298474945534, "grad_norm": 20.125, "learning_rate": 6.271810035341535e-06, "loss": 2.4927, "step": 3800 }, { "epoch": 0.43584451324389406, "grad_norm": 13.875, "learning_rate": 6.27001407181438e-06, "loss": 1.9094, "step": 3801 }, { "epoch": 0.4359591789932347, "grad_norm": 14.1875, "learning_rate": 6.268217933112994e-06, "loss": 1.532, "step": 3802 }, { "epoch": 0.4360738447425754, "grad_norm": 17.125, "learning_rate": 6.2664216194851165e-06, "loss": 1.8133, "step": 3803 }, { "epoch": 0.43618851049191604, "grad_norm": 13.0, "learning_rate": 6.264625131178519e-06, "loss": 1.7505, "step": 3804 }, { "epoch": 0.43630317624125675, "grad_norm": 13.375, "learning_rate": 6.26282846844099e-06, "loss": 1.9588, "step": 3805 }, { "epoch": 0.4364178419905974, "grad_norm": 15.125, "learning_rate": 6.261031631520344e-06, "loss": 2.0653, "step": 3806 }, { "epoch": 0.43653250773993807, "grad_norm": 13.375, "learning_rate": 6.259234620664424e-06, "loss": 1.9006, "step": 3807 }, { "epoch": 0.43664717348927873, "grad_norm": 18.0, "learning_rate": 6.257437436121091e-06, "loss": 1.7624, "step": 3808 }, { "epoch": 0.43676183923861944, "grad_norm": 15.25, "learning_rate": 6.25564007813823e-06, "loss": 1.9378, "step": 3809 }, { "epoch": 0.4368765049879601, "grad_norm": 16.25, "learning_rate": 6.253842546963758e-06, "loss": 1.818, "step": 3810 }, { "epoch": 0.43699117073730076, "grad_norm": 11.875, "learning_rate": 6.252044842845606e-06, "loss": 1.8197, "step": 3811 }, { "epoch": 0.4371058364866414, "grad_norm": 15.5, "learning_rate": 6.2502469660317335e-06, "loss": 2.0774, "step": 3812 }, { "epoch": 0.43722050223598213, "grad_norm": 15.0625, "learning_rate": 6.248448916770124e-06, "loss": 1.6419, "step": 3813 }, { "epoch": 0.4373351679853228, "grad_norm": 14.625, "learning_rate": 6.246650695308784e-06, "loss": 1.6621, "step": 3814 }, { "epoch": 0.43744983373466345, "grad_norm": 12.9375, "learning_rate": 6.244852301895746e-06, "loss": 1.5166, "step": 3815 }, { "epoch": 0.4375644994840041, "grad_norm": 16.0, "learning_rate": 6.243053736779059e-06, "loss": 1.7489, "step": 3816 }, { "epoch": 0.4376791652333448, "grad_norm": 12.0, "learning_rate": 6.241255000206807e-06, "loss": 1.3384, "step": 3817 }, { "epoch": 0.4377938309826855, "grad_norm": 17.75, "learning_rate": 6.2394560924270856e-06, "loss": 1.4566, "step": 3818 }, { "epoch": 0.43790849673202614, "grad_norm": 13.75, "learning_rate": 6.237657013688022e-06, "loss": 1.566, "step": 3819 }, { "epoch": 0.4380231624813668, "grad_norm": 17.0, "learning_rate": 6.235857764237767e-06, "loss": 1.6864, "step": 3820 }, { "epoch": 0.43813782823070746, "grad_norm": 15.5625, "learning_rate": 6.2340583443244905e-06, "loss": 1.9344, "step": 3821 }, { "epoch": 0.4382524939800482, "grad_norm": 17.0, "learning_rate": 6.2322587541963885e-06, "loss": 1.9264, "step": 3822 }, { "epoch": 0.43836715972938883, "grad_norm": 14.5, "learning_rate": 6.230458994101681e-06, "loss": 1.4699, "step": 3823 }, { "epoch": 0.4384818254787295, "grad_norm": 17.75, "learning_rate": 6.2286590642886095e-06, "loss": 2.0904, "step": 3824 }, { "epoch": 0.43859649122807015, "grad_norm": 13.6875, "learning_rate": 6.22685896500544e-06, "loss": 1.4577, "step": 3825 }, { "epoch": 0.43871115697741087, "grad_norm": 13.0, "learning_rate": 6.225058696500463e-06, "loss": 1.3541, "step": 3826 }, { "epoch": 0.4388258227267515, "grad_norm": 15.5, "learning_rate": 6.22325825902199e-06, "loss": 1.9687, "step": 3827 }, { "epoch": 0.4389404884760922, "grad_norm": 12.75, "learning_rate": 6.221457652818356e-06, "loss": 1.9733, "step": 3828 }, { "epoch": 0.43905515422543284, "grad_norm": 12.8125, "learning_rate": 6.219656878137925e-06, "loss": 2.0051, "step": 3829 }, { "epoch": 0.43916981997477356, "grad_norm": 13.3125, "learning_rate": 6.217855935229076e-06, "loss": 1.6445, "step": 3830 }, { "epoch": 0.4392844857241142, "grad_norm": 14.75, "learning_rate": 6.2160548243402134e-06, "loss": 1.7102, "step": 3831 }, { "epoch": 0.4393991514734549, "grad_norm": 14.5625, "learning_rate": 6.214253545719768e-06, "loss": 2.0403, "step": 3832 }, { "epoch": 0.43951381722279553, "grad_norm": 15.3125, "learning_rate": 6.212452099616194e-06, "loss": 1.7202, "step": 3833 }, { "epoch": 0.43962848297213625, "grad_norm": 15.1875, "learning_rate": 6.210650486277961e-06, "loss": 1.6609, "step": 3834 }, { "epoch": 0.4397431487214769, "grad_norm": 12.375, "learning_rate": 6.208848705953574e-06, "loss": 1.4043, "step": 3835 }, { "epoch": 0.43985781447081757, "grad_norm": 16.5, "learning_rate": 6.207046758891548e-06, "loss": 1.9026, "step": 3836 }, { "epoch": 0.4399724802201582, "grad_norm": 13.375, "learning_rate": 6.205244645340431e-06, "loss": 1.6386, "step": 3837 }, { "epoch": 0.4400871459694989, "grad_norm": 14.5, "learning_rate": 6.203442365548791e-06, "loss": 1.8116, "step": 3838 }, { "epoch": 0.4402018117188396, "grad_norm": 14.8125, "learning_rate": 6.201639919765214e-06, "loss": 1.905, "step": 3839 }, { "epoch": 0.44031647746818026, "grad_norm": 15.0625, "learning_rate": 6.199837308238316e-06, "loss": 1.504, "step": 3840 }, { "epoch": 0.4404311432175209, "grad_norm": 22.25, "learning_rate": 6.198034531216733e-06, "loss": 2.0194, "step": 3841 }, { "epoch": 0.4405458089668616, "grad_norm": 19.625, "learning_rate": 6.196231588949122e-06, "loss": 1.7888, "step": 3842 }, { "epoch": 0.4406604747162023, "grad_norm": 14.3125, "learning_rate": 6.194428481684166e-06, "loss": 1.868, "step": 3843 }, { "epoch": 0.44077514046554295, "grad_norm": 12.625, "learning_rate": 6.192625209670569e-06, "loss": 1.7834, "step": 3844 }, { "epoch": 0.4408898062148836, "grad_norm": 13.875, "learning_rate": 6.190821773157059e-06, "loss": 1.4402, "step": 3845 }, { "epoch": 0.44100447196422426, "grad_norm": 12.1875, "learning_rate": 6.189018172392383e-06, "loss": 1.4954, "step": 3846 }, { "epoch": 0.441119137713565, "grad_norm": 15.9375, "learning_rate": 6.187214407625314e-06, "loss": 2.352, "step": 3847 }, { "epoch": 0.44123380346290564, "grad_norm": 14.625, "learning_rate": 6.18541047910465e-06, "loss": 1.439, "step": 3848 }, { "epoch": 0.4413484692122463, "grad_norm": 16.25, "learning_rate": 6.183606387079205e-06, "loss": 1.4651, "step": 3849 }, { "epoch": 0.44146313496158696, "grad_norm": 11.8125, "learning_rate": 6.18180213179782e-06, "loss": 1.6168, "step": 3850 }, { "epoch": 0.44157780071092767, "grad_norm": 15.1875, "learning_rate": 6.179997713509359e-06, "loss": 1.9476, "step": 3851 }, { "epoch": 0.44169246646026833, "grad_norm": 15.4375, "learning_rate": 6.178193132462707e-06, "loss": 1.8075, "step": 3852 }, { "epoch": 0.441807132209609, "grad_norm": 18.375, "learning_rate": 6.17638838890677e-06, "loss": 1.7581, "step": 3853 }, { "epoch": 0.44192179795894965, "grad_norm": 16.625, "learning_rate": 6.174583483090478e-06, "loss": 1.646, "step": 3854 }, { "epoch": 0.44203646370829036, "grad_norm": 13.875, "learning_rate": 6.172778415262786e-06, "loss": 1.9026, "step": 3855 }, { "epoch": 0.442151129457631, "grad_norm": 16.625, "learning_rate": 6.170973185672664e-06, "loss": 1.7904, "step": 3856 }, { "epoch": 0.4422657952069717, "grad_norm": 14.8125, "learning_rate": 6.169167794569114e-06, "loss": 1.2769, "step": 3857 }, { "epoch": 0.44238046095631234, "grad_norm": 14.4375, "learning_rate": 6.167362242201154e-06, "loss": 1.8459, "step": 3858 }, { "epoch": 0.442495126705653, "grad_norm": 15.3125, "learning_rate": 6.165556528817821e-06, "loss": 1.8966, "step": 3859 }, { "epoch": 0.4426097924549937, "grad_norm": 13.625, "learning_rate": 6.1637506546681854e-06, "loss": 2.2783, "step": 3860 }, { "epoch": 0.44272445820433437, "grad_norm": 17.875, "learning_rate": 6.161944620001329e-06, "loss": 2.2239, "step": 3861 }, { "epoch": 0.44283912395367503, "grad_norm": 14.1875, "learning_rate": 6.160138425066361e-06, "loss": 1.7924, "step": 3862 }, { "epoch": 0.4429537897030157, "grad_norm": 15.125, "learning_rate": 6.158332070112413e-06, "loss": 1.7704, "step": 3863 }, { "epoch": 0.4430684554523564, "grad_norm": 18.25, "learning_rate": 6.156525555388634e-06, "loss": 2.0239, "step": 3864 }, { "epoch": 0.44318312120169706, "grad_norm": 13.6875, "learning_rate": 6.1547188811442e-06, "loss": 1.5979, "step": 3865 }, { "epoch": 0.4432977869510377, "grad_norm": 14.875, "learning_rate": 6.152912047628308e-06, "loss": 1.7457, "step": 3866 }, { "epoch": 0.4434124527003784, "grad_norm": 13.875, "learning_rate": 6.151105055090175e-06, "loss": 1.7312, "step": 3867 }, { "epoch": 0.4435271184497191, "grad_norm": 14.3125, "learning_rate": 6.149297903779043e-06, "loss": 1.8464, "step": 3868 }, { "epoch": 0.44364178419905975, "grad_norm": 13.75, "learning_rate": 6.14749059394417e-06, "loss": 1.7089, "step": 3869 }, { "epoch": 0.4437564499484004, "grad_norm": 14.1875, "learning_rate": 6.145683125834846e-06, "loss": 1.796, "step": 3870 }, { "epoch": 0.44387111569774107, "grad_norm": 14.8125, "learning_rate": 6.143875499700372e-06, "loss": 1.9499, "step": 3871 }, { "epoch": 0.4439857814470818, "grad_norm": 13.5, "learning_rate": 6.142067715790077e-06, "loss": 1.6764, "step": 3872 }, { "epoch": 0.44410044719642244, "grad_norm": 14.5, "learning_rate": 6.14025977435331e-06, "loss": 1.8181, "step": 3873 }, { "epoch": 0.4442151129457631, "grad_norm": 13.5625, "learning_rate": 6.138451675639443e-06, "loss": 1.762, "step": 3874 }, { "epoch": 0.44432977869510376, "grad_norm": 19.625, "learning_rate": 6.1366434198978675e-06, "loss": 2.2738, "step": 3875 }, { "epoch": 0.4444444444444444, "grad_norm": 14.6875, "learning_rate": 6.134835007377999e-06, "loss": 2.1393, "step": 3876 }, { "epoch": 0.44455911019378513, "grad_norm": 12.9375, "learning_rate": 6.133026438329275e-06, "loss": 1.7618, "step": 3877 }, { "epoch": 0.4446737759431258, "grad_norm": 11.75, "learning_rate": 6.131217713001149e-06, "loss": 1.5109, "step": 3878 }, { "epoch": 0.44478844169246645, "grad_norm": 36.0, "learning_rate": 6.129408831643102e-06, "loss": 1.9116, "step": 3879 }, { "epoch": 0.4449031074418071, "grad_norm": 16.625, "learning_rate": 6.127599794504637e-06, "loss": 1.8316, "step": 3880 }, { "epoch": 0.4450177731911478, "grad_norm": 22.625, "learning_rate": 6.125790601835273e-06, "loss": 1.3837, "step": 3881 }, { "epoch": 0.4451324389404885, "grad_norm": 13.4375, "learning_rate": 6.123981253884556e-06, "loss": 1.4726, "step": 3882 }, { "epoch": 0.44524710468982914, "grad_norm": 16.75, "learning_rate": 6.122171750902051e-06, "loss": 1.4774, "step": 3883 }, { "epoch": 0.4453617704391698, "grad_norm": 34.75, "learning_rate": 6.120362093137341e-06, "loss": 1.9987, "step": 3884 }, { "epoch": 0.4454764361885105, "grad_norm": 12.875, "learning_rate": 6.118552280840038e-06, "loss": 1.6906, "step": 3885 }, { "epoch": 0.4455911019378512, "grad_norm": 16.375, "learning_rate": 6.116742314259769e-06, "loss": 1.8455, "step": 3886 }, { "epoch": 0.44570576768719183, "grad_norm": 26.625, "learning_rate": 6.114932193646184e-06, "loss": 1.5425, "step": 3887 }, { "epoch": 0.4458204334365325, "grad_norm": 13.5625, "learning_rate": 6.113121919248958e-06, "loss": 1.688, "step": 3888 }, { "epoch": 0.4459350991858732, "grad_norm": 17.875, "learning_rate": 6.1113114913177785e-06, "loss": 1.7693, "step": 3889 }, { "epoch": 0.44604976493521387, "grad_norm": 12.8125, "learning_rate": 6.109500910102363e-06, "loss": 1.9274, "step": 3890 }, { "epoch": 0.4461644306845545, "grad_norm": 21.25, "learning_rate": 6.107690175852445e-06, "loss": 1.7395, "step": 3891 }, { "epoch": 0.4462790964338952, "grad_norm": 12.125, "learning_rate": 6.105879288817785e-06, "loss": 1.7728, "step": 3892 }, { "epoch": 0.44639376218323584, "grad_norm": 15.3125, "learning_rate": 6.104068249248154e-06, "loss": 1.5198, "step": 3893 }, { "epoch": 0.44650842793257656, "grad_norm": 17.125, "learning_rate": 6.102257057393355e-06, "loss": 1.809, "step": 3894 }, { "epoch": 0.4466230936819172, "grad_norm": 19.625, "learning_rate": 6.1004457135032055e-06, "loss": 1.8557, "step": 3895 }, { "epoch": 0.4467377594312579, "grad_norm": 14.25, "learning_rate": 6.0986342178275455e-06, "loss": 1.6501, "step": 3896 }, { "epoch": 0.44685242518059853, "grad_norm": 13.3125, "learning_rate": 6.096822570616238e-06, "loss": 1.6244, "step": 3897 }, { "epoch": 0.44696709092993925, "grad_norm": 15.75, "learning_rate": 6.095010772119165e-06, "loss": 1.7915, "step": 3898 }, { "epoch": 0.4470817566792799, "grad_norm": 15.25, "learning_rate": 6.093198822586226e-06, "loss": 1.8307, "step": 3899 }, { "epoch": 0.44719642242862057, "grad_norm": 15.875, "learning_rate": 6.09138672226735e-06, "loss": 1.993, "step": 3900 }, { "epoch": 0.4473110881779612, "grad_norm": 11.875, "learning_rate": 6.089574471412478e-06, "loss": 1.8681, "step": 3901 }, { "epoch": 0.44742575392730194, "grad_norm": 19.875, "learning_rate": 6.087762070271578e-06, "loss": 1.3949, "step": 3902 }, { "epoch": 0.4475404196766426, "grad_norm": 11.8125, "learning_rate": 6.085949519094634e-06, "loss": 1.5298, "step": 3903 }, { "epoch": 0.44765508542598326, "grad_norm": 11.625, "learning_rate": 6.084136818131654e-06, "loss": 1.5992, "step": 3904 }, { "epoch": 0.4477697511753239, "grad_norm": 17.75, "learning_rate": 6.082323967632667e-06, "loss": 1.7739, "step": 3905 }, { "epoch": 0.44788441692466463, "grad_norm": 12.6875, "learning_rate": 6.080510967847717e-06, "loss": 1.517, "step": 3906 }, { "epoch": 0.4479990826740053, "grad_norm": 15.5, "learning_rate": 6.078697819026879e-06, "loss": 1.6926, "step": 3907 }, { "epoch": 0.44811374842334595, "grad_norm": 12.0625, "learning_rate": 6.076884521420236e-06, "loss": 1.511, "step": 3908 }, { "epoch": 0.4482284141726866, "grad_norm": 12.6875, "learning_rate": 6.0750710752779e-06, "loss": 1.9273, "step": 3909 }, { "epoch": 0.44834307992202727, "grad_norm": 13.375, "learning_rate": 6.073257480850004e-06, "loss": 2.0205, "step": 3910 }, { "epoch": 0.448457745671368, "grad_norm": 32.25, "learning_rate": 6.071443738386696e-06, "loss": 1.9338, "step": 3911 }, { "epoch": 0.44857241142070864, "grad_norm": 14.0, "learning_rate": 6.069629848138148e-06, "loss": 1.7439, "step": 3912 }, { "epoch": 0.4486870771700493, "grad_norm": 13.6875, "learning_rate": 6.067815810354551e-06, "loss": 1.7425, "step": 3913 }, { "epoch": 0.44880174291938996, "grad_norm": 16.75, "learning_rate": 6.066001625286118e-06, "loss": 2.3245, "step": 3914 }, { "epoch": 0.44891640866873067, "grad_norm": 13.125, "learning_rate": 6.064187293183079e-06, "loss": 1.4488, "step": 3915 }, { "epoch": 0.44903107441807133, "grad_norm": 15.375, "learning_rate": 6.0623728142956895e-06, "loss": 1.6252, "step": 3916 }, { "epoch": 0.449145740167412, "grad_norm": 17.125, "learning_rate": 6.0605581888742214e-06, "loss": 2.0116, "step": 3917 }, { "epoch": 0.44926040591675265, "grad_norm": 14.375, "learning_rate": 6.0587434171689665e-06, "loss": 1.9164, "step": 3918 }, { "epoch": 0.44937507166609336, "grad_norm": 17.375, "learning_rate": 6.056928499430238e-06, "loss": 1.6257, "step": 3919 }, { "epoch": 0.449489737415434, "grad_norm": 22.75, "learning_rate": 6.055113435908372e-06, "loss": 1.8316, "step": 3920 }, { "epoch": 0.4496044031647747, "grad_norm": 14.25, "learning_rate": 6.053298226853718e-06, "loss": 2.0082, "step": 3921 }, { "epoch": 0.44971906891411534, "grad_norm": 15.6875, "learning_rate": 6.051482872516652e-06, "loss": 2.0965, "step": 3922 }, { "epoch": 0.44983373466345605, "grad_norm": 19.625, "learning_rate": 6.049667373147567e-06, "loss": 2.074, "step": 3923 }, { "epoch": 0.4499484004127967, "grad_norm": 14.25, "learning_rate": 6.047851728996875e-06, "loss": 1.6542, "step": 3924 }, { "epoch": 0.45006306616213737, "grad_norm": 13.4375, "learning_rate": 6.046035940315011e-06, "loss": 1.6296, "step": 3925 }, { "epoch": 0.45017773191147803, "grad_norm": 12.5, "learning_rate": 6.044220007352429e-06, "loss": 1.3277, "step": 3926 }, { "epoch": 0.4502923976608187, "grad_norm": 18.375, "learning_rate": 6.042403930359602e-06, "loss": 1.3695, "step": 3927 }, { "epoch": 0.4504070634101594, "grad_norm": 14.1875, "learning_rate": 6.040587709587024e-06, "loss": 1.8494, "step": 3928 }, { "epoch": 0.45052172915950006, "grad_norm": 11.4375, "learning_rate": 6.038771345285205e-06, "loss": 1.6602, "step": 3929 }, { "epoch": 0.4506363949088407, "grad_norm": 13.5625, "learning_rate": 6.03695483770468e-06, "loss": 1.8684, "step": 3930 }, { "epoch": 0.4507510606581814, "grad_norm": 15.125, "learning_rate": 6.035138187096e-06, "loss": 1.3663, "step": 3931 }, { "epoch": 0.4508657264075221, "grad_norm": 14.375, "learning_rate": 6.033321393709741e-06, "loss": 1.5454, "step": 3932 }, { "epoch": 0.45098039215686275, "grad_norm": 12.375, "learning_rate": 6.031504457796493e-06, "loss": 1.4466, "step": 3933 }, { "epoch": 0.4510950579062034, "grad_norm": 12.0, "learning_rate": 6.0296873796068645e-06, "loss": 1.6968, "step": 3934 }, { "epoch": 0.45120972365554407, "grad_norm": 14.1875, "learning_rate": 6.027870159391491e-06, "loss": 1.7676, "step": 3935 }, { "epoch": 0.4513243894048848, "grad_norm": 12.0625, "learning_rate": 6.026052797401022e-06, "loss": 1.4158, "step": 3936 }, { "epoch": 0.45143905515422544, "grad_norm": 14.1875, "learning_rate": 6.0242352938861285e-06, "loss": 1.6714, "step": 3937 }, { "epoch": 0.4515537209035661, "grad_norm": 11.9375, "learning_rate": 6.0224176490974995e-06, "loss": 1.6988, "step": 3938 }, { "epoch": 0.45166838665290676, "grad_norm": 13.0, "learning_rate": 6.020599863285845e-06, "loss": 1.6926, "step": 3939 }, { "epoch": 0.4517830524022475, "grad_norm": 11.8125, "learning_rate": 6.0187819367018926e-06, "loss": 1.4936, "step": 3940 }, { "epoch": 0.45189771815158813, "grad_norm": 13.5625, "learning_rate": 6.016963869596393e-06, "loss": 1.7966, "step": 3941 }, { "epoch": 0.4520123839009288, "grad_norm": 16.25, "learning_rate": 6.0151456622201134e-06, "loss": 1.5762, "step": 3942 }, { "epoch": 0.45212704965026945, "grad_norm": 13.25, "learning_rate": 6.013327314823838e-06, "loss": 1.6135, "step": 3943 }, { "epoch": 0.4522417153996101, "grad_norm": 14.3125, "learning_rate": 6.011508827658377e-06, "loss": 1.4952, "step": 3944 }, { "epoch": 0.4523563811489508, "grad_norm": 16.875, "learning_rate": 6.009690200974554e-06, "loss": 2.2435, "step": 3945 }, { "epoch": 0.4524710468982915, "grad_norm": 12.3125, "learning_rate": 6.007871435023213e-06, "loss": 2.1171, "step": 3946 }, { "epoch": 0.45258571264763214, "grad_norm": 13.5625, "learning_rate": 6.006052530055221e-06, "loss": 2.128, "step": 3947 }, { "epoch": 0.4527003783969728, "grad_norm": 15.125, "learning_rate": 6.004233486321461e-06, "loss": 1.5916, "step": 3948 }, { "epoch": 0.4528150441463135, "grad_norm": 20.625, "learning_rate": 6.00241430407283e-06, "loss": 1.8428, "step": 3949 }, { "epoch": 0.4529297098956542, "grad_norm": 12.625, "learning_rate": 6.0005949835602554e-06, "loss": 1.5594, "step": 3950 }, { "epoch": 0.45304437564499483, "grad_norm": 13.125, "learning_rate": 5.998775525034676e-06, "loss": 1.618, "step": 3951 }, { "epoch": 0.4531590413943355, "grad_norm": 12.25, "learning_rate": 5.9969559287470514e-06, "loss": 1.5694, "step": 3952 }, { "epoch": 0.4532737071436762, "grad_norm": 19.5, "learning_rate": 5.99513619494836e-06, "loss": 1.8308, "step": 3953 }, { "epoch": 0.45338837289301687, "grad_norm": 12.3125, "learning_rate": 5.993316323889599e-06, "loss": 2.005, "step": 3954 }, { "epoch": 0.4535030386423575, "grad_norm": 14.8125, "learning_rate": 5.991496315821787e-06, "loss": 2.3332, "step": 3955 }, { "epoch": 0.4536177043916982, "grad_norm": 13.8125, "learning_rate": 5.989676170995955e-06, "loss": 1.8046, "step": 3956 }, { "epoch": 0.4537323701410389, "grad_norm": 13.9375, "learning_rate": 5.9878558896631635e-06, "loss": 1.8852, "step": 3957 }, { "epoch": 0.45384703589037956, "grad_norm": 14.375, "learning_rate": 5.98603547207448e-06, "loss": 1.7756, "step": 3958 }, { "epoch": 0.4539617016397202, "grad_norm": 25.375, "learning_rate": 5.984214918481e-06, "loss": 1.6022, "step": 3959 }, { "epoch": 0.4540763673890609, "grad_norm": 21.5, "learning_rate": 5.982394229133832e-06, "loss": 1.8739, "step": 3960 }, { "epoch": 0.45419103313840153, "grad_norm": 30.5, "learning_rate": 5.980573404284107e-06, "loss": 2.4304, "step": 3961 }, { "epoch": 0.45430569888774225, "grad_norm": 34.75, "learning_rate": 5.97875244418297e-06, "loss": 2.1405, "step": 3962 }, { "epoch": 0.4544203646370829, "grad_norm": 16.25, "learning_rate": 5.976931349081594e-06, "loss": 1.6795, "step": 3963 }, { "epoch": 0.45453503038642357, "grad_norm": 13.3125, "learning_rate": 5.9751101192311576e-06, "loss": 1.6356, "step": 3964 }, { "epoch": 0.4546496961357642, "grad_norm": 13.3125, "learning_rate": 5.973288754882868e-06, "loss": 1.6165, "step": 3965 }, { "epoch": 0.45476436188510494, "grad_norm": 17.0, "learning_rate": 5.971467256287947e-06, "loss": 2.0361, "step": 3966 }, { "epoch": 0.4548790276344456, "grad_norm": 18.375, "learning_rate": 5.969645623697636e-06, "loss": 1.7095, "step": 3967 }, { "epoch": 0.45499369338378626, "grad_norm": 28.625, "learning_rate": 5.967823857363195e-06, "loss": 1.8033, "step": 3968 }, { "epoch": 0.4551083591331269, "grad_norm": 14.5625, "learning_rate": 5.9660019575359e-06, "loss": 1.7254, "step": 3969 }, { "epoch": 0.45522302488246763, "grad_norm": 15.5625, "learning_rate": 5.964179924467051e-06, "loss": 1.8053, "step": 3970 }, { "epoch": 0.4553376906318083, "grad_norm": 15.6875, "learning_rate": 5.962357758407958e-06, "loss": 1.5821, "step": 3971 }, { "epoch": 0.45545235638114895, "grad_norm": 14.0, "learning_rate": 5.960535459609958e-06, "loss": 2.0899, "step": 3972 }, { "epoch": 0.4555670221304896, "grad_norm": 16.25, "learning_rate": 5.9587130283244e-06, "loss": 2.0336, "step": 3973 }, { "epoch": 0.4556816878798303, "grad_norm": 14.9375, "learning_rate": 5.956890464802655e-06, "loss": 2.0357, "step": 3974 }, { "epoch": 0.455796353629171, "grad_norm": 13.25, "learning_rate": 5.9550677692961086e-06, "loss": 2.2991, "step": 3975 }, { "epoch": 0.45591101937851164, "grad_norm": 14.4375, "learning_rate": 5.953244942056171e-06, "loss": 1.9542, "step": 3976 }, { "epoch": 0.4560256851278523, "grad_norm": 15.5625, "learning_rate": 5.951421983334264e-06, "loss": 1.898, "step": 3977 }, { "epoch": 0.45614035087719296, "grad_norm": 12.4375, "learning_rate": 5.949598893381828e-06, "loss": 2.0501, "step": 3978 }, { "epoch": 0.45625501662653367, "grad_norm": 14.9375, "learning_rate": 5.947775672450327e-06, "loss": 2.1118, "step": 3979 }, { "epoch": 0.45636968237587433, "grad_norm": 16.75, "learning_rate": 5.94595232079124e-06, "loss": 1.6082, "step": 3980 }, { "epoch": 0.456484348125215, "grad_norm": 16.625, "learning_rate": 5.944128838656059e-06, "loss": 1.7945, "step": 3981 }, { "epoch": 0.45659901387455565, "grad_norm": 15.4375, "learning_rate": 5.942305226296302e-06, "loss": 1.9504, "step": 3982 }, { "epoch": 0.45671367962389636, "grad_norm": 11.6875, "learning_rate": 5.9404814839635025e-06, "loss": 1.8196, "step": 3983 }, { "epoch": 0.456828345373237, "grad_norm": 14.0625, "learning_rate": 5.938657611909207e-06, "loss": 1.8543, "step": 3984 }, { "epoch": 0.4569430111225777, "grad_norm": 10.375, "learning_rate": 5.9368336103849886e-06, "loss": 1.5819, "step": 3985 }, { "epoch": 0.45705767687191834, "grad_norm": 24.375, "learning_rate": 5.9350094796424306e-06, "loss": 1.6082, "step": 3986 }, { "epoch": 0.45717234262125905, "grad_norm": 14.1875, "learning_rate": 5.933185219933137e-06, "loss": 1.9739, "step": 3987 }, { "epoch": 0.4572870083705997, "grad_norm": 16.875, "learning_rate": 5.931360831508733e-06, "loss": 1.7167, "step": 3988 }, { "epoch": 0.45740167411994037, "grad_norm": 14.6875, "learning_rate": 5.929536314620853e-06, "loss": 2.1884, "step": 3989 }, { "epoch": 0.45751633986928103, "grad_norm": 15.3125, "learning_rate": 5.927711669521156e-06, "loss": 1.5499, "step": 3990 }, { "epoch": 0.45763100561862174, "grad_norm": 19.875, "learning_rate": 5.9258868964613214e-06, "loss": 2.1172, "step": 3991 }, { "epoch": 0.4577456713679624, "grad_norm": 13.0, "learning_rate": 5.9240619956930365e-06, "loss": 1.5683, "step": 3992 }, { "epoch": 0.45786033711730306, "grad_norm": 11.0625, "learning_rate": 5.922236967468013e-06, "loss": 1.4651, "step": 3993 }, { "epoch": 0.4579750028666437, "grad_norm": 12.1875, "learning_rate": 5.92041181203798e-06, "loss": 1.6517, "step": 3994 }, { "epoch": 0.4580896686159844, "grad_norm": 15.125, "learning_rate": 5.918586529654684e-06, "loss": 2.0193, "step": 3995 }, { "epoch": 0.4582043343653251, "grad_norm": 16.75, "learning_rate": 5.916761120569884e-06, "loss": 1.6676, "step": 3996 }, { "epoch": 0.45831900011466575, "grad_norm": 15.4375, "learning_rate": 5.914935585035363e-06, "loss": 1.7675, "step": 3997 }, { "epoch": 0.4584336658640064, "grad_norm": 11.6875, "learning_rate": 5.913109923302919e-06, "loss": 1.8174, "step": 3998 }, { "epoch": 0.45854833161334707, "grad_norm": 15.25, "learning_rate": 5.911284135624365e-06, "loss": 1.7285, "step": 3999 }, { "epoch": 0.4586629973626878, "grad_norm": 12.9375, "learning_rate": 5.909458222251536e-06, "loss": 1.5502, "step": 4000 }, { "epoch": 0.45877766311202844, "grad_norm": 13.5, "learning_rate": 5.907632183436282e-06, "loss": 1.8227, "step": 4001 }, { "epoch": 0.4588923288613691, "grad_norm": 14.4375, "learning_rate": 5.905806019430467e-06, "loss": 1.7189, "step": 4002 }, { "epoch": 0.45900699461070976, "grad_norm": 24.625, "learning_rate": 5.903979730485979e-06, "loss": 1.8511, "step": 4003 }, { "epoch": 0.4591216603600505, "grad_norm": 13.75, "learning_rate": 5.9021533168547186e-06, "loss": 1.6879, "step": 4004 }, { "epoch": 0.45923632610939114, "grad_norm": 16.625, "learning_rate": 5.900326778788605e-06, "loss": 1.9365, "step": 4005 }, { "epoch": 0.4593509918587318, "grad_norm": 32.5, "learning_rate": 5.898500116539575e-06, "loss": 1.9839, "step": 4006 }, { "epoch": 0.45946565760807245, "grad_norm": 15.0, "learning_rate": 5.896673330359579e-06, "loss": 2.045, "step": 4007 }, { "epoch": 0.45958032335741317, "grad_norm": 13.5, "learning_rate": 5.894846420500591e-06, "loss": 1.6562, "step": 4008 }, { "epoch": 0.4596949891067538, "grad_norm": 14.125, "learning_rate": 5.893019387214594e-06, "loss": 1.7937, "step": 4009 }, { "epoch": 0.4598096548560945, "grad_norm": 14.375, "learning_rate": 5.891192230753597e-06, "loss": 1.3465, "step": 4010 }, { "epoch": 0.45992432060543514, "grad_norm": 12.6875, "learning_rate": 5.88936495136962e-06, "loss": 1.854, "step": 4011 }, { "epoch": 0.4600389863547758, "grad_norm": 18.125, "learning_rate": 5.8875375493146995e-06, "loss": 1.6228, "step": 4012 }, { "epoch": 0.4601536521041165, "grad_norm": 13.5625, "learning_rate": 5.885710024840894e-06, "loss": 2.0314, "step": 4013 }, { "epoch": 0.4602683178534572, "grad_norm": 15.125, "learning_rate": 5.883882378200271e-06, "loss": 1.7012, "step": 4014 }, { "epoch": 0.46038298360279784, "grad_norm": 13.5, "learning_rate": 5.882054609644924e-06, "loss": 1.4303, "step": 4015 }, { "epoch": 0.4604976493521385, "grad_norm": 11.0, "learning_rate": 5.880226719426957e-06, "loss": 1.2912, "step": 4016 }, { "epoch": 0.4606123151014792, "grad_norm": 13.75, "learning_rate": 5.878398707798492e-06, "loss": 1.7046, "step": 4017 }, { "epoch": 0.46072698085081987, "grad_norm": 16.5, "learning_rate": 5.876570575011669e-06, "loss": 1.6271, "step": 4018 }, { "epoch": 0.4608416466001605, "grad_norm": 13.125, "learning_rate": 5.874742321318644e-06, "loss": 1.8522, "step": 4019 }, { "epoch": 0.4609563123495012, "grad_norm": 15.25, "learning_rate": 5.872913946971591e-06, "loss": 1.6758, "step": 4020 }, { "epoch": 0.4610709780988419, "grad_norm": 11.8125, "learning_rate": 5.871085452222698e-06, "loss": 1.8042, "step": 4021 }, { "epoch": 0.46118564384818256, "grad_norm": 12.875, "learning_rate": 5.869256837324171e-06, "loss": 1.9285, "step": 4022 }, { "epoch": 0.4613003095975232, "grad_norm": 13.75, "learning_rate": 5.867428102528233e-06, "loss": 1.7588, "step": 4023 }, { "epoch": 0.4614149753468639, "grad_norm": 13.4375, "learning_rate": 5.8655992480871225e-06, "loss": 1.324, "step": 4024 }, { "epoch": 0.4615296410962046, "grad_norm": 25.75, "learning_rate": 5.863770274253095e-06, "loss": 1.6204, "step": 4025 }, { "epoch": 0.46164430684554525, "grad_norm": 16.375, "learning_rate": 5.861941181278426e-06, "loss": 1.8271, "step": 4026 }, { "epoch": 0.4617589725948859, "grad_norm": 14.6875, "learning_rate": 5.860111969415397e-06, "loss": 1.7295, "step": 4027 }, { "epoch": 0.46187363834422657, "grad_norm": 13.125, "learning_rate": 5.85828263891632e-06, "loss": 1.5838, "step": 4028 }, { "epoch": 0.4619883040935672, "grad_norm": 12.0625, "learning_rate": 5.8564531900335116e-06, "loss": 1.7207, "step": 4029 }, { "epoch": 0.46210296984290794, "grad_norm": 15.5625, "learning_rate": 5.854623623019313e-06, "loss": 1.8365, "step": 4030 }, { "epoch": 0.4622176355922486, "grad_norm": 12.5, "learning_rate": 5.852793938126075e-06, "loss": 1.6088, "step": 4031 }, { "epoch": 0.46233230134158926, "grad_norm": 13.8125, "learning_rate": 5.8509641356061685e-06, "loss": 1.7292, "step": 4032 }, { "epoch": 0.4624469670909299, "grad_norm": 12.875, "learning_rate": 5.849134215711981e-06, "loss": 2.0748, "step": 4033 }, { "epoch": 0.46256163284027063, "grad_norm": 12.8125, "learning_rate": 5.847304178695914e-06, "loss": 1.7358, "step": 4034 }, { "epoch": 0.4626762985896113, "grad_norm": 14.1875, "learning_rate": 5.845474024810387e-06, "loss": 2.0721, "step": 4035 }, { "epoch": 0.46279096433895195, "grad_norm": 13.875, "learning_rate": 5.843643754307835e-06, "loss": 1.5861, "step": 4036 }, { "epoch": 0.4629056300882926, "grad_norm": 12.0, "learning_rate": 5.8418133674407074e-06, "loss": 1.9678, "step": 4037 }, { "epoch": 0.4630202958376333, "grad_norm": 12.3125, "learning_rate": 5.839982864461473e-06, "loss": 1.4869, "step": 4038 }, { "epoch": 0.463134961586974, "grad_norm": 19.25, "learning_rate": 5.838152245622614e-06, "loss": 1.6976, "step": 4039 }, { "epoch": 0.46324962733631464, "grad_norm": 11.9375, "learning_rate": 5.836321511176628e-06, "loss": 1.8423, "step": 4040 }, { "epoch": 0.4633642930856553, "grad_norm": 14.0625, "learning_rate": 5.834490661376033e-06, "loss": 2.172, "step": 4041 }, { "epoch": 0.463478958834996, "grad_norm": 13.8125, "learning_rate": 5.832659696473357e-06, "loss": 1.6042, "step": 4042 }, { "epoch": 0.46359362458433667, "grad_norm": 12.9375, "learning_rate": 5.830828616721148e-06, "loss": 1.6692, "step": 4043 }, { "epoch": 0.46370829033367733, "grad_norm": 17.125, "learning_rate": 5.828997422371968e-06, "loss": 1.801, "step": 4044 }, { "epoch": 0.463822956083018, "grad_norm": 20.25, "learning_rate": 5.827166113678397e-06, "loss": 1.7737, "step": 4045 }, { "epoch": 0.46393762183235865, "grad_norm": 16.25, "learning_rate": 5.825334690893028e-06, "loss": 1.7791, "step": 4046 }, { "epoch": 0.46405228758169936, "grad_norm": 14.625, "learning_rate": 5.823503154268468e-06, "loss": 2.048, "step": 4047 }, { "epoch": 0.46416695333104, "grad_norm": 15.25, "learning_rate": 5.821671504057349e-06, "loss": 1.9309, "step": 4048 }, { "epoch": 0.4642816190803807, "grad_norm": 13.8125, "learning_rate": 5.819839740512305e-06, "loss": 1.7065, "step": 4049 }, { "epoch": 0.46439628482972134, "grad_norm": 15.4375, "learning_rate": 5.8180078638859984e-06, "loss": 1.8679, "step": 4050 }, { "epoch": 0.46451095057906205, "grad_norm": 14.625, "learning_rate": 5.816175874431101e-06, "loss": 1.8843, "step": 4051 }, { "epoch": 0.4646256163284027, "grad_norm": 22.5, "learning_rate": 5.814343772400296e-06, "loss": 1.6869, "step": 4052 }, { "epoch": 0.46474028207774337, "grad_norm": 26.5, "learning_rate": 5.812511558046292e-06, "loss": 1.7207, "step": 4053 }, { "epoch": 0.46485494782708403, "grad_norm": 15.5, "learning_rate": 5.810679231621807e-06, "loss": 1.9184, "step": 4054 }, { "epoch": 0.46496961357642475, "grad_norm": 15.25, "learning_rate": 5.808846793379574e-06, "loss": 1.8591, "step": 4055 }, { "epoch": 0.4650842793257654, "grad_norm": 17.0, "learning_rate": 5.807014243572344e-06, "loss": 2.2795, "step": 4056 }, { "epoch": 0.46519894507510606, "grad_norm": 13.25, "learning_rate": 5.8051815824528825e-06, "loss": 1.6967, "step": 4057 }, { "epoch": 0.4653136108244467, "grad_norm": 14.5, "learning_rate": 5.80334881027397e-06, "loss": 1.8784, "step": 4058 }, { "epoch": 0.46542827657378744, "grad_norm": 29.625, "learning_rate": 5.801515927288401e-06, "loss": 1.9125, "step": 4059 }, { "epoch": 0.4655429423231281, "grad_norm": 15.0, "learning_rate": 5.79968293374899e-06, "loss": 1.5534, "step": 4060 }, { "epoch": 0.46565760807246875, "grad_norm": 13.0, "learning_rate": 5.7978498299085614e-06, "loss": 1.7649, "step": 4061 }, { "epoch": 0.4657722738218094, "grad_norm": 26.75, "learning_rate": 5.796016616019956e-06, "loss": 2.1587, "step": 4062 }, { "epoch": 0.46588693957115007, "grad_norm": 15.0625, "learning_rate": 5.794183292336033e-06, "loss": 1.5194, "step": 4063 }, { "epoch": 0.4660016053204908, "grad_norm": 14.75, "learning_rate": 5.792349859109663e-06, "loss": 1.9413, "step": 4064 }, { "epoch": 0.46611627106983144, "grad_norm": 11.5, "learning_rate": 5.790516316593733e-06, "loss": 1.7638, "step": 4065 }, { "epoch": 0.4662309368191721, "grad_norm": 14.0625, "learning_rate": 5.788682665041147e-06, "loss": 1.9102, "step": 4066 }, { "epoch": 0.46634560256851276, "grad_norm": 14.6875, "learning_rate": 5.786848904704819e-06, "loss": 1.7571, "step": 4067 }, { "epoch": 0.4664602683178535, "grad_norm": 16.375, "learning_rate": 5.785015035837684e-06, "loss": 2.0451, "step": 4068 }, { "epoch": 0.46657493406719414, "grad_norm": 14.9375, "learning_rate": 5.78318105869269e-06, "loss": 1.6346, "step": 4069 }, { "epoch": 0.4666895998165348, "grad_norm": 13.625, "learning_rate": 5.781346973522796e-06, "loss": 1.4511, "step": 4070 }, { "epoch": 0.46680426556587545, "grad_norm": 15.6875, "learning_rate": 5.779512780580981e-06, "loss": 1.5583, "step": 4071 }, { "epoch": 0.46691893131521617, "grad_norm": 15.5625, "learning_rate": 5.777678480120238e-06, "loss": 1.5508, "step": 4072 }, { "epoch": 0.4670335970645568, "grad_norm": 15.6875, "learning_rate": 5.775844072393573e-06, "loss": 1.7723, "step": 4073 }, { "epoch": 0.4671482628138975, "grad_norm": 13.4375, "learning_rate": 5.774009557654006e-06, "loss": 1.5565, "step": 4074 }, { "epoch": 0.46726292856323814, "grad_norm": 15.5, "learning_rate": 5.7721749361545735e-06, "loss": 1.9408, "step": 4075 }, { "epoch": 0.46737759431257886, "grad_norm": 12.125, "learning_rate": 5.77034020814833e-06, "loss": 1.6644, "step": 4076 }, { "epoch": 0.4674922600619195, "grad_norm": 14.3125, "learning_rate": 5.768505373888337e-06, "loss": 1.5057, "step": 4077 }, { "epoch": 0.4676069258112602, "grad_norm": 13.5, "learning_rate": 5.766670433627677e-06, "loss": 2.2248, "step": 4078 }, { "epoch": 0.46772159156060084, "grad_norm": 15.0, "learning_rate": 5.764835387619444e-06, "loss": 1.6242, "step": 4079 }, { "epoch": 0.4678362573099415, "grad_norm": 13.4375, "learning_rate": 5.76300023611675e-06, "loss": 1.9499, "step": 4080 }, { "epoch": 0.4679509230592822, "grad_norm": 13.5, "learning_rate": 5.761164979372717e-06, "loss": 1.3433, "step": 4081 }, { "epoch": 0.46806558880862287, "grad_norm": 19.875, "learning_rate": 5.759329617640483e-06, "loss": 1.8715, "step": 4082 }, { "epoch": 0.4681802545579635, "grad_norm": 19.5, "learning_rate": 5.757494151173204e-06, "loss": 1.624, "step": 4083 }, { "epoch": 0.4682949203073042, "grad_norm": 17.75, "learning_rate": 5.7556585802240435e-06, "loss": 1.9294, "step": 4084 }, { "epoch": 0.4684095860566449, "grad_norm": 13.1875, "learning_rate": 5.753822905046189e-06, "loss": 1.8012, "step": 4085 }, { "epoch": 0.46852425180598556, "grad_norm": 14.5, "learning_rate": 5.751987125892834e-06, "loss": 1.9264, "step": 4086 }, { "epoch": 0.4686389175553262, "grad_norm": 14.8125, "learning_rate": 5.750151243017187e-06, "loss": 1.7245, "step": 4087 }, { "epoch": 0.4687535833046669, "grad_norm": 14.1875, "learning_rate": 5.748315256672476e-06, "loss": 1.6599, "step": 4088 }, { "epoch": 0.4688682490540076, "grad_norm": 22.5, "learning_rate": 5.74647916711194e-06, "loss": 1.7755, "step": 4089 }, { "epoch": 0.46898291480334825, "grad_norm": 13.75, "learning_rate": 5.744642974588831e-06, "loss": 2.0055, "step": 4090 }, { "epoch": 0.4690975805526889, "grad_norm": 11.6875, "learning_rate": 5.742806679356419e-06, "loss": 1.6325, "step": 4091 }, { "epoch": 0.46921224630202957, "grad_norm": 13.25, "learning_rate": 5.740970281667984e-06, "loss": 1.6391, "step": 4092 }, { "epoch": 0.4693269120513703, "grad_norm": 13.9375, "learning_rate": 5.739133781776824e-06, "loss": 1.6275, "step": 4093 }, { "epoch": 0.46944157780071094, "grad_norm": 16.875, "learning_rate": 5.737297179936247e-06, "loss": 1.5005, "step": 4094 }, { "epoch": 0.4695562435500516, "grad_norm": 13.375, "learning_rate": 5.735460476399579e-06, "loss": 1.8342, "step": 4095 }, { "epoch": 0.46967090929939226, "grad_norm": 15.5625, "learning_rate": 5.733623671420157e-06, "loss": 1.87, "step": 4096 }, { "epoch": 0.4697855750487329, "grad_norm": 13.4375, "learning_rate": 5.7317867652513325e-06, "loss": 1.4571, "step": 4097 }, { "epoch": 0.46990024079807363, "grad_norm": 15.1875, "learning_rate": 5.729949758146474e-06, "loss": 1.7665, "step": 4098 }, { "epoch": 0.4700149065474143, "grad_norm": 14.125, "learning_rate": 5.728112650358961e-06, "loss": 1.603, "step": 4099 }, { "epoch": 0.47012957229675495, "grad_norm": 11.8125, "learning_rate": 5.7262754421421865e-06, "loss": 1.7235, "step": 4100 }, { "epoch": 0.4702442380460956, "grad_norm": 17.375, "learning_rate": 5.724438133749559e-06, "loss": 1.7356, "step": 4101 }, { "epoch": 0.4703589037954363, "grad_norm": 16.625, "learning_rate": 5.722600725434499e-06, "loss": 1.8229, "step": 4102 }, { "epoch": 0.470473569544777, "grad_norm": 17.0, "learning_rate": 5.720763217450444e-06, "loss": 2.3585, "step": 4103 }, { "epoch": 0.47058823529411764, "grad_norm": 16.0, "learning_rate": 5.7189256100508405e-06, "loss": 1.5474, "step": 4104 }, { "epoch": 0.4707029010434583, "grad_norm": 19.25, "learning_rate": 5.717087903489156e-06, "loss": 2.1154, "step": 4105 }, { "epoch": 0.470817566792799, "grad_norm": 15.25, "learning_rate": 5.715250098018861e-06, "loss": 2.0645, "step": 4106 }, { "epoch": 0.4709322325421397, "grad_norm": 13.75, "learning_rate": 5.71341219389345e-06, "loss": 1.6257, "step": 4107 }, { "epoch": 0.47104689829148033, "grad_norm": 13.1875, "learning_rate": 5.711574191366427e-06, "loss": 1.6829, "step": 4108 }, { "epoch": 0.471161564040821, "grad_norm": 14.4375, "learning_rate": 5.709736090691306e-06, "loss": 1.7536, "step": 4109 }, { "epoch": 0.4712762297901617, "grad_norm": 11.625, "learning_rate": 5.7078978921216225e-06, "loss": 1.5389, "step": 4110 }, { "epoch": 0.47139089553950236, "grad_norm": 13.75, "learning_rate": 5.706059595910917e-06, "loss": 2.1928, "step": 4111 }, { "epoch": 0.471505561288843, "grad_norm": 13.375, "learning_rate": 5.704221202312748e-06, "loss": 1.4595, "step": 4112 }, { "epoch": 0.4716202270381837, "grad_norm": 17.125, "learning_rate": 5.7023827115806905e-06, "loss": 1.7508, "step": 4113 }, { "epoch": 0.47173489278752434, "grad_norm": 16.625, "learning_rate": 5.700544123968325e-06, "loss": 1.6046, "step": 4114 }, { "epoch": 0.47184955853686505, "grad_norm": 19.5, "learning_rate": 5.698705439729251e-06, "loss": 1.69, "step": 4115 }, { "epoch": 0.4719642242862057, "grad_norm": 15.75, "learning_rate": 5.696866659117081e-06, "loss": 1.8527, "step": 4116 }, { "epoch": 0.4720788900355464, "grad_norm": 15.0, "learning_rate": 5.695027782385438e-06, "loss": 1.6425, "step": 4117 }, { "epoch": 0.47219355578488703, "grad_norm": 17.0, "learning_rate": 5.69318880978796e-06, "loss": 1.7055, "step": 4118 }, { "epoch": 0.47230822153422775, "grad_norm": 14.5625, "learning_rate": 5.691349741578301e-06, "loss": 1.3121, "step": 4119 }, { "epoch": 0.4724228872835684, "grad_norm": 18.25, "learning_rate": 5.689510578010123e-06, "loss": 1.6449, "step": 4120 }, { "epoch": 0.47253755303290906, "grad_norm": 14.375, "learning_rate": 5.6876713193371035e-06, "loss": 1.8636, "step": 4121 }, { "epoch": 0.4726522187822497, "grad_norm": 16.5, "learning_rate": 5.685831965812934e-06, "loss": 1.9011, "step": 4122 }, { "epoch": 0.47276688453159044, "grad_norm": 17.0, "learning_rate": 5.6839925176913195e-06, "loss": 1.7414, "step": 4123 }, { "epoch": 0.4728815502809311, "grad_norm": 12.4375, "learning_rate": 5.682152975225973e-06, "loss": 1.709, "step": 4124 }, { "epoch": 0.47299621603027175, "grad_norm": 13.125, "learning_rate": 5.680313338670627e-06, "loss": 1.5267, "step": 4125 }, { "epoch": 0.4731108817796124, "grad_norm": 18.5, "learning_rate": 5.678473608279024e-06, "loss": 1.808, "step": 4126 }, { "epoch": 0.4732255475289531, "grad_norm": 13.75, "learning_rate": 5.676633784304921e-06, "loss": 1.5101, "step": 4127 }, { "epoch": 0.4733402132782938, "grad_norm": 13.75, "learning_rate": 5.674793867002083e-06, "loss": 1.6472, "step": 4128 }, { "epoch": 0.47345487902763445, "grad_norm": 13.375, "learning_rate": 5.672953856624295e-06, "loss": 1.5956, "step": 4129 }, { "epoch": 0.4735695447769751, "grad_norm": 17.75, "learning_rate": 5.67111375342535e-06, "loss": 1.8074, "step": 4130 }, { "epoch": 0.47368421052631576, "grad_norm": 12.9375, "learning_rate": 5.669273557659055e-06, "loss": 1.7578, "step": 4131 }, { "epoch": 0.4737988762756565, "grad_norm": 14.5, "learning_rate": 5.66743326957923e-06, "loss": 1.5785, "step": 4132 }, { "epoch": 0.47391354202499714, "grad_norm": 17.5, "learning_rate": 5.6655928894397096e-06, "loss": 2.2002, "step": 4133 }, { "epoch": 0.4740282077743378, "grad_norm": 14.125, "learning_rate": 5.663752417494335e-06, "loss": 1.6919, "step": 4134 }, { "epoch": 0.47414287352367845, "grad_norm": 17.625, "learning_rate": 5.6619118539969684e-06, "loss": 1.9399, "step": 4135 }, { "epoch": 0.47425753927301917, "grad_norm": 14.875, "learning_rate": 5.660071199201479e-06, "loss": 2.0062, "step": 4136 }, { "epoch": 0.4743722050223598, "grad_norm": 12.25, "learning_rate": 5.658230453361748e-06, "loss": 2.1126, "step": 4137 }, { "epoch": 0.4744868707717005, "grad_norm": 13.25, "learning_rate": 5.656389616731675e-06, "loss": 2.1174, "step": 4138 }, { "epoch": 0.47460153652104115, "grad_norm": 14.875, "learning_rate": 5.654548689565165e-06, "loss": 1.4493, "step": 4139 }, { "epoch": 0.47471620227038186, "grad_norm": 23.125, "learning_rate": 5.65270767211614e-06, "loss": 1.6937, "step": 4140 }, { "epoch": 0.4748308680197225, "grad_norm": 17.75, "learning_rate": 5.650866564638536e-06, "loss": 2.0825, "step": 4141 }, { "epoch": 0.4749455337690632, "grad_norm": 14.1875, "learning_rate": 5.649025367386292e-06, "loss": 1.9932, "step": 4142 }, { "epoch": 0.47506019951840384, "grad_norm": 27.25, "learning_rate": 5.647184080613372e-06, "loss": 2.0398, "step": 4143 }, { "epoch": 0.47517486526774455, "grad_norm": 21.375, "learning_rate": 5.645342704573744e-06, "loss": 1.8208, "step": 4144 }, { "epoch": 0.4752895310170852, "grad_norm": 15.5, "learning_rate": 5.643501239521393e-06, "loss": 1.8241, "step": 4145 }, { "epoch": 0.47540419676642587, "grad_norm": 12.625, "learning_rate": 5.64165968571031e-06, "loss": 1.4186, "step": 4146 }, { "epoch": 0.4755188625157665, "grad_norm": 15.1875, "learning_rate": 5.639818043394506e-06, "loss": 1.7243, "step": 4147 }, { "epoch": 0.4756335282651072, "grad_norm": 17.875, "learning_rate": 5.637976312827998e-06, "loss": 1.828, "step": 4148 }, { "epoch": 0.4757481940144479, "grad_norm": 13.75, "learning_rate": 5.63613449426482e-06, "loss": 1.7166, "step": 4149 }, { "epoch": 0.47586285976378856, "grad_norm": 14.625, "learning_rate": 5.634292587959012e-06, "loss": 1.7943, "step": 4150 }, { "epoch": 0.4759775255131292, "grad_norm": 12.5625, "learning_rate": 5.632450594164636e-06, "loss": 1.6227, "step": 4151 }, { "epoch": 0.4760921912624699, "grad_norm": 18.75, "learning_rate": 5.630608513135752e-06, "loss": 1.5397, "step": 4152 }, { "epoch": 0.4762068570118106, "grad_norm": 12.75, "learning_rate": 5.628766345126446e-06, "loss": 1.9387, "step": 4153 }, { "epoch": 0.47632152276115125, "grad_norm": 16.5, "learning_rate": 5.6269240903908085e-06, "loss": 2.0848, "step": 4154 }, { "epoch": 0.4764361885104919, "grad_norm": 14.6875, "learning_rate": 5.6250817491829425e-06, "loss": 1.4381, "step": 4155 }, { "epoch": 0.47655085425983257, "grad_norm": 16.0, "learning_rate": 5.623239321756963e-06, "loss": 1.9035, "step": 4156 }, { "epoch": 0.4766655200091733, "grad_norm": 15.8125, "learning_rate": 5.621396808367001e-06, "loss": 1.9831, "step": 4157 }, { "epoch": 0.47678018575851394, "grad_norm": 17.125, "learning_rate": 5.619554209267194e-06, "loss": 2.1742, "step": 4158 }, { "epoch": 0.4768948515078546, "grad_norm": 15.5625, "learning_rate": 5.617711524711692e-06, "loss": 1.843, "step": 4159 }, { "epoch": 0.47700951725719526, "grad_norm": 15.3125, "learning_rate": 5.615868754954662e-06, "loss": 1.8076, "step": 4160 }, { "epoch": 0.477124183006536, "grad_norm": 13.5, "learning_rate": 5.6140259002502765e-06, "loss": 1.7994, "step": 4161 }, { "epoch": 0.47723884875587663, "grad_norm": 17.625, "learning_rate": 5.61218296085272e-06, "loss": 1.6157, "step": 4162 }, { "epoch": 0.4773535145052173, "grad_norm": 12.5, "learning_rate": 5.610339937016196e-06, "loss": 1.5693, "step": 4163 }, { "epoch": 0.47746818025455795, "grad_norm": 13.75, "learning_rate": 5.60849682899491e-06, "loss": 2.035, "step": 4164 }, { "epoch": 0.4775828460038986, "grad_norm": 25.125, "learning_rate": 5.606653637043087e-06, "loss": 1.7549, "step": 4165 }, { "epoch": 0.4776975117532393, "grad_norm": 33.0, "learning_rate": 5.604810361414959e-06, "loss": 1.936, "step": 4166 }, { "epoch": 0.47781217750258, "grad_norm": 14.4375, "learning_rate": 5.602967002364769e-06, "loss": 1.6327, "step": 4167 }, { "epoch": 0.47792684325192064, "grad_norm": 28.125, "learning_rate": 5.601123560146774e-06, "loss": 1.5968, "step": 4168 }, { "epoch": 0.4780415090012613, "grad_norm": 15.75, "learning_rate": 5.599280035015243e-06, "loss": 1.6482, "step": 4169 }, { "epoch": 0.478156174750602, "grad_norm": 14.6875, "learning_rate": 5.597436427224455e-06, "loss": 1.5571, "step": 4170 }, { "epoch": 0.4782708404999427, "grad_norm": 15.3125, "learning_rate": 5.5955927370287e-06, "loss": 1.8166, "step": 4171 }, { "epoch": 0.47838550624928333, "grad_norm": 17.75, "learning_rate": 5.59374896468228e-06, "loss": 1.7498, "step": 4172 }, { "epoch": 0.478500171998624, "grad_norm": 13.5625, "learning_rate": 5.591905110439508e-06, "loss": 1.6793, "step": 4173 }, { "epoch": 0.4786148377479647, "grad_norm": 27.625, "learning_rate": 5.590061174554708e-06, "loss": 1.8306, "step": 4174 }, { "epoch": 0.47872950349730536, "grad_norm": 12.625, "learning_rate": 5.588217157282217e-06, "loss": 1.7479, "step": 4175 }, { "epoch": 0.478844169246646, "grad_norm": 13.875, "learning_rate": 5.586373058876383e-06, "loss": 1.8635, "step": 4176 }, { "epoch": 0.4789588349959867, "grad_norm": 33.5, "learning_rate": 5.584528879591562e-06, "loss": 1.8378, "step": 4177 }, { "epoch": 0.4790735007453274, "grad_norm": 15.0, "learning_rate": 5.5826846196821226e-06, "loss": 1.723, "step": 4178 }, { "epoch": 0.47918816649466806, "grad_norm": 13.8125, "learning_rate": 5.580840279402448e-06, "loss": 1.6737, "step": 4179 }, { "epoch": 0.4793028322440087, "grad_norm": 13.625, "learning_rate": 5.57899585900693e-06, "loss": 1.9776, "step": 4180 }, { "epoch": 0.4794174979933494, "grad_norm": 12.4375, "learning_rate": 5.5771513587499686e-06, "loss": 1.6618, "step": 4181 }, { "epoch": 0.47953216374269003, "grad_norm": 12.75, "learning_rate": 5.575306778885978e-06, "loss": 1.8096, "step": 4182 }, { "epoch": 0.47964682949203075, "grad_norm": 14.5625, "learning_rate": 5.573462119669386e-06, "loss": 2.0126, "step": 4183 }, { "epoch": 0.4797614952413714, "grad_norm": 13.6875, "learning_rate": 5.571617381354623e-06, "loss": 1.5252, "step": 4184 }, { "epoch": 0.47987616099071206, "grad_norm": 13.0625, "learning_rate": 5.569772564196139e-06, "loss": 1.7128, "step": 4185 }, { "epoch": 0.4799908267400527, "grad_norm": 17.0, "learning_rate": 5.567927668448393e-06, "loss": 1.9017, "step": 4186 }, { "epoch": 0.48010549248939344, "grad_norm": 12.6875, "learning_rate": 5.566082694365847e-06, "loss": 1.6882, "step": 4187 }, { "epoch": 0.4802201582387341, "grad_norm": 14.4375, "learning_rate": 5.564237642202987e-06, "loss": 1.673, "step": 4188 }, { "epoch": 0.48033482398807475, "grad_norm": 13.625, "learning_rate": 5.562392512214299e-06, "loss": 1.7664, "step": 4189 }, { "epoch": 0.4804494897374154, "grad_norm": 20.0, "learning_rate": 5.560547304654282e-06, "loss": 1.602, "step": 4190 }, { "epoch": 0.48056415548675613, "grad_norm": 13.25, "learning_rate": 5.558702019777452e-06, "loss": 1.9192, "step": 4191 }, { "epoch": 0.4806788212360968, "grad_norm": 17.25, "learning_rate": 5.556856657838328e-06, "loss": 1.6506, "step": 4192 }, { "epoch": 0.48079348698543745, "grad_norm": 16.25, "learning_rate": 5.555011219091441e-06, "loss": 1.9522, "step": 4193 }, { "epoch": 0.4809081527347781, "grad_norm": 28.25, "learning_rate": 5.553165703791336e-06, "loss": 1.6339, "step": 4194 }, { "epoch": 0.4810228184841188, "grad_norm": 14.125, "learning_rate": 5.551320112192568e-06, "loss": 2.1347, "step": 4195 }, { "epoch": 0.4811374842334595, "grad_norm": 14.4375, "learning_rate": 5.549474444549698e-06, "loss": 1.8277, "step": 4196 }, { "epoch": 0.48125214998280014, "grad_norm": 14.8125, "learning_rate": 5.547628701117302e-06, "loss": 1.7571, "step": 4197 }, { "epoch": 0.4813668157321408, "grad_norm": 15.4375, "learning_rate": 5.545782882149968e-06, "loss": 1.6703, "step": 4198 }, { "epoch": 0.48148148148148145, "grad_norm": 13.75, "learning_rate": 5.5439369879022874e-06, "loss": 1.861, "step": 4199 }, { "epoch": 0.48159614723082217, "grad_norm": 16.875, "learning_rate": 5.542091018628867e-06, "loss": 1.8052, "step": 4200 }, { "epoch": 0.48171081298016283, "grad_norm": 12.75, "learning_rate": 5.5402449745843255e-06, "loss": 1.6898, "step": 4201 }, { "epoch": 0.4818254787295035, "grad_norm": 15.1875, "learning_rate": 5.538398856023286e-06, "loss": 1.8069, "step": 4202 }, { "epoch": 0.48194014447884415, "grad_norm": 16.25, "learning_rate": 5.536552663200387e-06, "loss": 1.5531, "step": 4203 }, { "epoch": 0.48205481022818486, "grad_norm": 14.875, "learning_rate": 5.5347063963702775e-06, "loss": 1.7866, "step": 4204 }, { "epoch": 0.4821694759775255, "grad_norm": 13.25, "learning_rate": 5.532860055787612e-06, "loss": 1.9361, "step": 4205 }, { "epoch": 0.4822841417268662, "grad_norm": 15.6875, "learning_rate": 5.531013641707059e-06, "loss": 2.3278, "step": 4206 }, { "epoch": 0.48239880747620684, "grad_norm": 12.3125, "learning_rate": 5.529167154383297e-06, "loss": 1.6555, "step": 4207 }, { "epoch": 0.48251347322554755, "grad_norm": 34.5, "learning_rate": 5.527320594071013e-06, "loss": 2.0403, "step": 4208 }, { "epoch": 0.4826281389748882, "grad_norm": 13.75, "learning_rate": 5.525473961024902e-06, "loss": 1.5823, "step": 4209 }, { "epoch": 0.48274280472422887, "grad_norm": 14.0625, "learning_rate": 5.523627255499678e-06, "loss": 1.7376, "step": 4210 }, { "epoch": 0.48285747047356953, "grad_norm": 13.6875, "learning_rate": 5.521780477750054e-06, "loss": 1.9833, "step": 4211 }, { "epoch": 0.48297213622291024, "grad_norm": 14.1875, "learning_rate": 5.519933628030758e-06, "loss": 1.8015, "step": 4212 }, { "epoch": 0.4830868019722509, "grad_norm": 14.75, "learning_rate": 5.51808670659653e-06, "loss": 1.4032, "step": 4213 }, { "epoch": 0.48320146772159156, "grad_norm": 16.75, "learning_rate": 5.516239713702116e-06, "loss": 1.7959, "step": 4214 }, { "epoch": 0.4833161334709322, "grad_norm": 16.0, "learning_rate": 5.5143926496022735e-06, "loss": 1.6696, "step": 4215 }, { "epoch": 0.4834307992202729, "grad_norm": 19.625, "learning_rate": 5.512545514551772e-06, "loss": 1.6017, "step": 4216 }, { "epoch": 0.4835454649696136, "grad_norm": 15.5, "learning_rate": 5.510698308805386e-06, "loss": 2.1864, "step": 4217 }, { "epoch": 0.48366013071895425, "grad_norm": 16.375, "learning_rate": 5.508851032617902e-06, "loss": 1.7312, "step": 4218 }, { "epoch": 0.4837747964682949, "grad_norm": 11.75, "learning_rate": 5.507003686244119e-06, "loss": 1.5702, "step": 4219 }, { "epoch": 0.48388946221763557, "grad_norm": 13.625, "learning_rate": 5.505156269938842e-06, "loss": 1.6367, "step": 4220 }, { "epoch": 0.4840041279669763, "grad_norm": 13.5, "learning_rate": 5.503308783956885e-06, "loss": 1.6934, "step": 4221 }, { "epoch": 0.48411879371631694, "grad_norm": 15.25, "learning_rate": 5.501461228553075e-06, "loss": 1.6124, "step": 4222 }, { "epoch": 0.4842334594656576, "grad_norm": 13.9375, "learning_rate": 5.49961360398225e-06, "loss": 1.8027, "step": 4223 }, { "epoch": 0.48434812521499826, "grad_norm": 15.1875, "learning_rate": 5.49776591049925e-06, "loss": 1.4357, "step": 4224 }, { "epoch": 0.484462790964339, "grad_norm": 13.625, "learning_rate": 5.495918148358931e-06, "loss": 1.7813, "step": 4225 }, { "epoch": 0.48457745671367963, "grad_norm": 14.5625, "learning_rate": 5.494070317816157e-06, "loss": 1.899, "step": 4226 }, { "epoch": 0.4846921224630203, "grad_norm": 12.0, "learning_rate": 5.4922224191258e-06, "loss": 1.8604, "step": 4227 }, { "epoch": 0.48480678821236095, "grad_norm": 12.4375, "learning_rate": 5.490374452542742e-06, "loss": 1.4568, "step": 4228 }, { "epoch": 0.48492145396170167, "grad_norm": 15.0625, "learning_rate": 5.4885264183218776e-06, "loss": 1.6482, "step": 4229 }, { "epoch": 0.4850361197110423, "grad_norm": 32.5, "learning_rate": 5.486678316718106e-06, "loss": 1.8465, "step": 4230 }, { "epoch": 0.485150785460383, "grad_norm": 17.875, "learning_rate": 5.484830147986337e-06, "loss": 1.9675, "step": 4231 }, { "epoch": 0.48526545120972364, "grad_norm": 22.875, "learning_rate": 5.482981912381489e-06, "loss": 1.9066, "step": 4232 }, { "epoch": 0.4853801169590643, "grad_norm": 16.75, "learning_rate": 5.481133610158495e-06, "loss": 1.9922, "step": 4233 }, { "epoch": 0.485494782708405, "grad_norm": 14.9375, "learning_rate": 5.479285241572289e-06, "loss": 1.6325, "step": 4234 }, { "epoch": 0.4856094484577457, "grad_norm": 13.8125, "learning_rate": 5.47743680687782e-06, "loss": 2.0203, "step": 4235 }, { "epoch": 0.48572411420708633, "grad_norm": 13.9375, "learning_rate": 5.475588306330047e-06, "loss": 1.6925, "step": 4236 }, { "epoch": 0.485838779956427, "grad_norm": 15.5625, "learning_rate": 5.4737397401839285e-06, "loss": 1.7337, "step": 4237 }, { "epoch": 0.4859534457057677, "grad_norm": 14.75, "learning_rate": 5.471891108694446e-06, "loss": 1.6549, "step": 4238 }, { "epoch": 0.48606811145510836, "grad_norm": 13.5, "learning_rate": 5.47004241211658e-06, "loss": 2.1978, "step": 4239 }, { "epoch": 0.486182777204449, "grad_norm": 27.625, "learning_rate": 5.468193650705321e-06, "loss": 2.0483, "step": 4240 }, { "epoch": 0.4862974429537897, "grad_norm": 15.8125, "learning_rate": 5.466344824715675e-06, "loss": 1.5942, "step": 4241 }, { "epoch": 0.4864121087031304, "grad_norm": 14.0625, "learning_rate": 5.464495934402648e-06, "loss": 1.5243, "step": 4242 }, { "epoch": 0.48652677445247106, "grad_norm": 20.75, "learning_rate": 5.462646980021262e-06, "loss": 1.9907, "step": 4243 }, { "epoch": 0.4866414402018117, "grad_norm": 15.375, "learning_rate": 5.460797961826542e-06, "loss": 1.9361, "step": 4244 }, { "epoch": 0.4867561059511524, "grad_norm": 13.8125, "learning_rate": 5.45894888007353e-06, "loss": 1.9418, "step": 4245 }, { "epoch": 0.4868707717004931, "grad_norm": 13.9375, "learning_rate": 5.457099735017267e-06, "loss": 1.6444, "step": 4246 }, { "epoch": 0.48698543744983375, "grad_norm": 14.125, "learning_rate": 5.455250526912807e-06, "loss": 1.7587, "step": 4247 }, { "epoch": 0.4871001031991744, "grad_norm": 19.25, "learning_rate": 5.453401256015218e-06, "loss": 1.766, "step": 4248 }, { "epoch": 0.48721476894851506, "grad_norm": 14.0, "learning_rate": 5.4515519225795665e-06, "loss": 2.0099, "step": 4249 }, { "epoch": 0.4873294346978557, "grad_norm": 13.0, "learning_rate": 5.449702526860934e-06, "loss": 1.6335, "step": 4250 }, { "epoch": 0.48744410044719644, "grad_norm": 12.4375, "learning_rate": 5.447853069114412e-06, "loss": 1.4858, "step": 4251 }, { "epoch": 0.4875587661965371, "grad_norm": 18.125, "learning_rate": 5.446003549595093e-06, "loss": 1.7377, "step": 4252 }, { "epoch": 0.48767343194587776, "grad_norm": 12.3125, "learning_rate": 5.444153968558088e-06, "loss": 1.7682, "step": 4253 }, { "epoch": 0.4877880976952184, "grad_norm": 13.3125, "learning_rate": 5.442304326258509e-06, "loss": 1.5561, "step": 4254 }, { "epoch": 0.48790276344455913, "grad_norm": 12.5, "learning_rate": 5.4404546229514795e-06, "loss": 1.4334, "step": 4255 }, { "epoch": 0.4880174291938998, "grad_norm": 12.3125, "learning_rate": 5.43860485889213e-06, "loss": 1.6822, "step": 4256 }, { "epoch": 0.48813209494324045, "grad_norm": 13.25, "learning_rate": 5.4367550343356e-06, "loss": 2.0087, "step": 4257 }, { "epoch": 0.4882467606925811, "grad_norm": 13.5625, "learning_rate": 5.434905149537041e-06, "loss": 1.4142, "step": 4258 }, { "epoch": 0.4883614264419218, "grad_norm": 14.125, "learning_rate": 5.433055204751604e-06, "loss": 1.9429, "step": 4259 }, { "epoch": 0.4884760921912625, "grad_norm": 15.5625, "learning_rate": 5.431205200234457e-06, "loss": 1.9342, "step": 4260 }, { "epoch": 0.48859075794060314, "grad_norm": 19.75, "learning_rate": 5.429355136240773e-06, "loss": 1.7554, "step": 4261 }, { "epoch": 0.4887054236899438, "grad_norm": 21.5, "learning_rate": 5.427505013025732e-06, "loss": 1.7812, "step": 4262 }, { "epoch": 0.4888200894392845, "grad_norm": 17.875, "learning_rate": 5.4256548308445234e-06, "loss": 1.7055, "step": 4263 }, { "epoch": 0.48893475518862517, "grad_norm": 11.8125, "learning_rate": 5.423804589952344e-06, "loss": 1.6915, "step": 4264 }, { "epoch": 0.48904942093796583, "grad_norm": 11.0625, "learning_rate": 5.4219542906043994e-06, "loss": 1.4508, "step": 4265 }, { "epoch": 0.4891640866873065, "grad_norm": 18.125, "learning_rate": 5.420103933055906e-06, "loss": 2.0866, "step": 4266 }, { "epoch": 0.48927875243664715, "grad_norm": 32.5, "learning_rate": 5.41825351756208e-06, "loss": 2.1639, "step": 4267 }, { "epoch": 0.48939341818598786, "grad_norm": 13.5625, "learning_rate": 5.416403044378155e-06, "loss": 1.7451, "step": 4268 }, { "epoch": 0.4895080839353285, "grad_norm": 13.5, "learning_rate": 5.414552513759368e-06, "loss": 1.696, "step": 4269 }, { "epoch": 0.4896227496846692, "grad_norm": 12.9375, "learning_rate": 5.412701925960964e-06, "loss": 1.4966, "step": 4270 }, { "epoch": 0.48973741543400984, "grad_norm": 12.375, "learning_rate": 5.410851281238195e-06, "loss": 1.6227, "step": 4271 }, { "epoch": 0.48985208118335055, "grad_norm": 11.625, "learning_rate": 5.409000579846324e-06, "loss": 1.9278, "step": 4272 }, { "epoch": 0.4899667469326912, "grad_norm": 13.75, "learning_rate": 5.4071498220406194e-06, "loss": 1.6383, "step": 4273 }, { "epoch": 0.49008141268203187, "grad_norm": 12.6875, "learning_rate": 5.405299008076358e-06, "loss": 1.7141, "step": 4274 }, { "epoch": 0.49019607843137253, "grad_norm": 26.25, "learning_rate": 5.403448138208823e-06, "loss": 1.9762, "step": 4275 }, { "epoch": 0.49031074418071324, "grad_norm": 11.125, "learning_rate": 5.401597212693309e-06, "loss": 1.7454, "step": 4276 }, { "epoch": 0.4904254099300539, "grad_norm": 13.75, "learning_rate": 5.399746231785114e-06, "loss": 1.4584, "step": 4277 }, { "epoch": 0.49054007567939456, "grad_norm": 13.625, "learning_rate": 5.397895195739545e-06, "loss": 1.7364, "step": 4278 }, { "epoch": 0.4906547414287352, "grad_norm": 15.1875, "learning_rate": 5.3960441048119205e-06, "loss": 2.0976, "step": 4279 }, { "epoch": 0.49076940717807593, "grad_norm": 17.75, "learning_rate": 5.39419295925756e-06, "loss": 1.4719, "step": 4280 }, { "epoch": 0.4908840729274166, "grad_norm": 13.625, "learning_rate": 5.392341759331795e-06, "loss": 1.5521, "step": 4281 }, { "epoch": 0.49099873867675725, "grad_norm": 22.75, "learning_rate": 5.390490505289962e-06, "loss": 1.6455, "step": 4282 }, { "epoch": 0.4911134044260979, "grad_norm": 17.0, "learning_rate": 5.388639197387409e-06, "loss": 2.0493, "step": 4283 }, { "epoch": 0.49122807017543857, "grad_norm": 14.375, "learning_rate": 5.386787835879486e-06, "loss": 1.8302, "step": 4284 }, { "epoch": 0.4913427359247793, "grad_norm": 32.25, "learning_rate": 5.384936421021553e-06, "loss": 1.9221, "step": 4285 }, { "epoch": 0.49145740167411994, "grad_norm": 13.8125, "learning_rate": 5.383084953068981e-06, "loss": 1.5072, "step": 4286 }, { "epoch": 0.4915720674234606, "grad_norm": 13.8125, "learning_rate": 5.3812334322771386e-06, "loss": 1.8769, "step": 4287 }, { "epoch": 0.49168673317280126, "grad_norm": 13.75, "learning_rate": 5.379381858901414e-06, "loss": 1.8746, "step": 4288 }, { "epoch": 0.491801398922142, "grad_norm": 31.25, "learning_rate": 5.3775302331971914e-06, "loss": 2.1122, "step": 4289 }, { "epoch": 0.49191606467148263, "grad_norm": 16.875, "learning_rate": 5.375678555419871e-06, "loss": 1.3506, "step": 4290 }, { "epoch": 0.4920307304208233, "grad_norm": 16.25, "learning_rate": 5.3738268258248546e-06, "loss": 1.8252, "step": 4291 }, { "epoch": 0.49214539617016395, "grad_norm": 14.4375, "learning_rate": 5.3719750446675535e-06, "loss": 1.4296, "step": 4292 }, { "epoch": 0.49226006191950467, "grad_norm": 13.875, "learning_rate": 5.370123212203384e-06, "loss": 1.6212, "step": 4293 }, { "epoch": 0.4923747276688453, "grad_norm": 15.4375, "learning_rate": 5.368271328687774e-06, "loss": 1.6442, "step": 4294 }, { "epoch": 0.492489393418186, "grad_norm": 26.125, "learning_rate": 5.3664193943761544e-06, "loss": 1.9592, "step": 4295 }, { "epoch": 0.49260405916752664, "grad_norm": 13.625, "learning_rate": 5.364567409523963e-06, "loss": 1.6106, "step": 4296 }, { "epoch": 0.49271872491686736, "grad_norm": 15.5625, "learning_rate": 5.362715374386646e-06, "loss": 1.7843, "step": 4297 }, { "epoch": 0.492833390666208, "grad_norm": 13.125, "learning_rate": 5.36086328921966e-06, "loss": 1.6936, "step": 4298 }, { "epoch": 0.4929480564155487, "grad_norm": 14.25, "learning_rate": 5.35901115427846e-06, "loss": 1.7815, "step": 4299 }, { "epoch": 0.49306272216488933, "grad_norm": 14.125, "learning_rate": 5.357158969818514e-06, "loss": 1.9246, "step": 4300 }, { "epoch": 0.49317738791423, "grad_norm": 13.9375, "learning_rate": 5.355306736095298e-06, "loss": 1.5657, "step": 4301 }, { "epoch": 0.4932920536635707, "grad_norm": 17.125, "learning_rate": 5.35345445336429e-06, "loss": 2.1625, "step": 4302 }, { "epoch": 0.49340671941291137, "grad_norm": 13.4375, "learning_rate": 5.351602121880976e-06, "loss": 1.4376, "step": 4303 }, { "epoch": 0.493521385162252, "grad_norm": 13.4375, "learning_rate": 5.349749741900852e-06, "loss": 1.685, "step": 4304 }, { "epoch": 0.4936360509115927, "grad_norm": 12.625, "learning_rate": 5.3478973136794195e-06, "loss": 1.8728, "step": 4305 }, { "epoch": 0.4937507166609334, "grad_norm": 13.5625, "learning_rate": 5.346044837472183e-06, "loss": 1.6342, "step": 4306 }, { "epoch": 0.49386538241027406, "grad_norm": 15.625, "learning_rate": 5.344192313534657e-06, "loss": 2.1515, "step": 4307 }, { "epoch": 0.4939800481596147, "grad_norm": 16.5, "learning_rate": 5.342339742122363e-06, "loss": 2.123, "step": 4308 }, { "epoch": 0.4940947139089554, "grad_norm": 11.375, "learning_rate": 5.340487123490826e-06, "loss": 1.5895, "step": 4309 }, { "epoch": 0.4942093796582961, "grad_norm": 15.6875, "learning_rate": 5.338634457895583e-06, "loss": 1.718, "step": 4310 }, { "epoch": 0.49432404540763675, "grad_norm": 14.0, "learning_rate": 5.33678174559217e-06, "loss": 1.6209, "step": 4311 }, { "epoch": 0.4944387111569774, "grad_norm": 13.9375, "learning_rate": 5.334928986836133e-06, "loss": 1.6946, "step": 4312 }, { "epoch": 0.49455337690631807, "grad_norm": 13.9375, "learning_rate": 5.33307618188303e-06, "loss": 1.7887, "step": 4313 }, { "epoch": 0.4946680426556588, "grad_norm": 12.3125, "learning_rate": 5.331223330988414e-06, "loss": 1.9251, "step": 4314 }, { "epoch": 0.49478270840499944, "grad_norm": 15.6875, "learning_rate": 5.329370434407854e-06, "loss": 1.8652, "step": 4315 }, { "epoch": 0.4948973741543401, "grad_norm": 14.25, "learning_rate": 5.327517492396923e-06, "loss": 1.5445, "step": 4316 }, { "epoch": 0.49501203990368076, "grad_norm": 14.125, "learning_rate": 5.325664505211194e-06, "loss": 2.2251, "step": 4317 }, { "epoch": 0.4951267056530214, "grad_norm": 12.8125, "learning_rate": 5.323811473106256e-06, "loss": 1.9589, "step": 4318 }, { "epoch": 0.49524137140236213, "grad_norm": 17.125, "learning_rate": 5.321958396337696e-06, "loss": 1.8929, "step": 4319 }, { "epoch": 0.4953560371517028, "grad_norm": 15.0625, "learning_rate": 5.320105275161115e-06, "loss": 2.0478, "step": 4320 }, { "epoch": 0.49547070290104345, "grad_norm": 13.125, "learning_rate": 5.318252109832112e-06, "loss": 1.6653, "step": 4321 }, { "epoch": 0.4955853686503841, "grad_norm": 17.75, "learning_rate": 5.3163989006062955e-06, "loss": 1.9474, "step": 4322 }, { "epoch": 0.4957000343997248, "grad_norm": 15.5, "learning_rate": 5.314545647739283e-06, "loss": 1.8876, "step": 4323 }, { "epoch": 0.4958147001490655, "grad_norm": 19.125, "learning_rate": 5.312692351486693e-06, "loss": 2.0093, "step": 4324 }, { "epoch": 0.49592936589840614, "grad_norm": 15.5, "learning_rate": 5.310839012104155e-06, "loss": 2.1205, "step": 4325 }, { "epoch": 0.4960440316477468, "grad_norm": 15.6875, "learning_rate": 5.308985629847301e-06, "loss": 2.0379, "step": 4326 }, { "epoch": 0.4961586973970875, "grad_norm": 33.0, "learning_rate": 5.307132204971768e-06, "loss": 2.3851, "step": 4327 }, { "epoch": 0.49627336314642817, "grad_norm": 17.75, "learning_rate": 5.3052787377332e-06, "loss": 1.8879, "step": 4328 }, { "epoch": 0.49638802889576883, "grad_norm": 12.1875, "learning_rate": 5.303425228387251e-06, "loss": 2.1231, "step": 4329 }, { "epoch": 0.4965026946451095, "grad_norm": 15.375, "learning_rate": 5.301571677189577e-06, "loss": 1.9846, "step": 4330 }, { "epoch": 0.4966173603944502, "grad_norm": 14.4375, "learning_rate": 5.299718084395837e-06, "loss": 1.6333, "step": 4331 }, { "epoch": 0.49673202614379086, "grad_norm": 14.0625, "learning_rate": 5.2978644502617e-06, "loss": 1.7323, "step": 4332 }, { "epoch": 0.4968466918931315, "grad_norm": 12.4375, "learning_rate": 5.296010775042842e-06, "loss": 1.4782, "step": 4333 }, { "epoch": 0.4969613576424722, "grad_norm": 14.5625, "learning_rate": 5.294157058994937e-06, "loss": 2.0106, "step": 4334 }, { "epoch": 0.49707602339181284, "grad_norm": 14.0625, "learning_rate": 5.292303302373675e-06, "loss": 1.4316, "step": 4335 }, { "epoch": 0.49719068914115355, "grad_norm": 15.4375, "learning_rate": 5.290449505434745e-06, "loss": 2.0938, "step": 4336 }, { "epoch": 0.4973053548904942, "grad_norm": 30.0, "learning_rate": 5.288595668433839e-06, "loss": 1.9588, "step": 4337 }, { "epoch": 0.49742002063983487, "grad_norm": 16.375, "learning_rate": 5.286741791626664e-06, "loss": 1.5937, "step": 4338 }, { "epoch": 0.49753468638917553, "grad_norm": 31.25, "learning_rate": 5.284887875268925e-06, "loss": 1.5726, "step": 4339 }, { "epoch": 0.49764935213851624, "grad_norm": 13.8125, "learning_rate": 5.283033919616332e-06, "loss": 1.8102, "step": 4340 }, { "epoch": 0.4977640178878569, "grad_norm": 15.25, "learning_rate": 5.2811799249246076e-06, "loss": 1.7621, "step": 4341 }, { "epoch": 0.49787868363719756, "grad_norm": 30.625, "learning_rate": 5.279325891449471e-06, "loss": 1.7785, "step": 4342 }, { "epoch": 0.4979933493865382, "grad_norm": 21.125, "learning_rate": 5.277471819446651e-06, "loss": 1.8468, "step": 4343 }, { "epoch": 0.49810801513587893, "grad_norm": 15.5, "learning_rate": 5.275617709171883e-06, "loss": 2.0438, "step": 4344 }, { "epoch": 0.4982226808852196, "grad_norm": 14.625, "learning_rate": 5.273763560880908e-06, "loss": 1.5164, "step": 4345 }, { "epoch": 0.49833734663456025, "grad_norm": 12.625, "learning_rate": 5.271909374829466e-06, "loss": 1.6211, "step": 4346 }, { "epoch": 0.4984520123839009, "grad_norm": 15.1875, "learning_rate": 5.270055151273309e-06, "loss": 1.7665, "step": 4347 }, { "epoch": 0.4985666781332416, "grad_norm": 16.75, "learning_rate": 5.268200890468193e-06, "loss": 1.9491, "step": 4348 }, { "epoch": 0.4986813438825823, "grad_norm": 13.0625, "learning_rate": 5.2663465926698755e-06, "loss": 1.929, "step": 4349 }, { "epoch": 0.49879600963192294, "grad_norm": 16.875, "learning_rate": 5.2644922581341216e-06, "loss": 1.8166, "step": 4350 }, { "epoch": 0.4989106753812636, "grad_norm": 42.0, "learning_rate": 5.2626378871167035e-06, "loss": 1.9589, "step": 4351 }, { "epoch": 0.49902534113060426, "grad_norm": 13.3125, "learning_rate": 5.2607834798733956e-06, "loss": 1.7667, "step": 4352 }, { "epoch": 0.499140006879945, "grad_norm": 18.25, "learning_rate": 5.258929036659976e-06, "loss": 1.6613, "step": 4353 }, { "epoch": 0.49925467262928563, "grad_norm": 10.875, "learning_rate": 5.257074557732232e-06, "loss": 1.5418, "step": 4354 }, { "epoch": 0.4993693383786263, "grad_norm": 15.4375, "learning_rate": 5.2552200433459554e-06, "loss": 1.991, "step": 4355 }, { "epoch": 0.49948400412796695, "grad_norm": 12.0625, "learning_rate": 5.253365493756937e-06, "loss": 1.7406, "step": 4356 }, { "epoch": 0.49959866987730767, "grad_norm": 16.875, "learning_rate": 5.2515109092209775e-06, "loss": 1.7893, "step": 4357 }, { "epoch": 0.4997133356266483, "grad_norm": 15.0, "learning_rate": 5.249656289993884e-06, "loss": 1.4754, "step": 4358 }, { "epoch": 0.499828001375989, "grad_norm": 12.0, "learning_rate": 5.247801636331462e-06, "loss": 1.7008, "step": 4359 }, { "epoch": 0.49994266712532964, "grad_norm": 15.3125, "learning_rate": 5.24594694848953e-06, "loss": 1.6925, "step": 4360 }, { "epoch": 0.5000573328746704, "grad_norm": 17.125, "learning_rate": 5.244092226723904e-06, "loss": 1.4475, "step": 4361 }, { "epoch": 0.500171998624011, "grad_norm": 13.625, "learning_rate": 5.2422374712904066e-06, "loss": 1.877, "step": 4362 }, { "epoch": 0.5002866643733517, "grad_norm": 18.125, "learning_rate": 5.240382682444869e-06, "loss": 1.9626, "step": 4363 }, { "epoch": 0.5004013301226924, "grad_norm": 26.75, "learning_rate": 5.2385278604431225e-06, "loss": 1.9593, "step": 4364 }, { "epoch": 0.500515995872033, "grad_norm": 14.0625, "learning_rate": 5.236673005541003e-06, "loss": 1.4845, "step": 4365 }, { "epoch": 0.5006306616213737, "grad_norm": 14.6875, "learning_rate": 5.2348181179943555e-06, "loss": 1.6622, "step": 4366 }, { "epoch": 0.5007453273707144, "grad_norm": 13.3125, "learning_rate": 5.232963198059025e-06, "loss": 1.8415, "step": 4367 }, { "epoch": 0.500859993120055, "grad_norm": 15.25, "learning_rate": 5.23110824599086e-06, "loss": 1.959, "step": 4368 }, { "epoch": 0.5009746588693957, "grad_norm": 13.8125, "learning_rate": 5.22925326204572e-06, "loss": 1.6031, "step": 4369 }, { "epoch": 0.5010893246187363, "grad_norm": 13.5625, "learning_rate": 5.22739824647946e-06, "loss": 2.0594, "step": 4370 }, { "epoch": 0.5012039903680771, "grad_norm": 12.8125, "learning_rate": 5.225543199547949e-06, "loss": 1.6742, "step": 4371 }, { "epoch": 0.5013186561174178, "grad_norm": 16.125, "learning_rate": 5.223688121507051e-06, "loss": 1.7012, "step": 4372 }, { "epoch": 0.5014333218667584, "grad_norm": 15.375, "learning_rate": 5.221833012612642e-06, "loss": 1.8353, "step": 4373 }, { "epoch": 0.5015479876160991, "grad_norm": 15.0, "learning_rate": 5.219977873120596e-06, "loss": 2.0695, "step": 4374 }, { "epoch": 0.5016626533654397, "grad_norm": 15.8125, "learning_rate": 5.218122703286797e-06, "loss": 1.8473, "step": 4375 }, { "epoch": 0.5017773191147804, "grad_norm": 12.5, "learning_rate": 5.216267503367128e-06, "loss": 1.5351, "step": 4376 }, { "epoch": 0.5018919848641211, "grad_norm": 15.5625, "learning_rate": 5.214412273617478e-06, "loss": 1.3548, "step": 4377 }, { "epoch": 0.5020066506134617, "grad_norm": 12.5625, "learning_rate": 5.212557014293744e-06, "loss": 1.876, "step": 4378 }, { "epoch": 0.5021213163628024, "grad_norm": 17.625, "learning_rate": 5.210701725651821e-06, "loss": 1.5481, "step": 4379 }, { "epoch": 0.5022359821121432, "grad_norm": 16.5, "learning_rate": 5.208846407947612e-06, "loss": 1.6643, "step": 4380 }, { "epoch": 0.5023506478614838, "grad_norm": 18.0, "learning_rate": 5.20699106143702e-06, "loss": 1.8799, "step": 4381 }, { "epoch": 0.5024653136108245, "grad_norm": 18.0, "learning_rate": 5.2051356863759585e-06, "loss": 1.9457, "step": 4382 }, { "epoch": 0.5025799793601651, "grad_norm": 13.625, "learning_rate": 5.2032802830203385e-06, "loss": 1.8083, "step": 4383 }, { "epoch": 0.5026946451095058, "grad_norm": 16.75, "learning_rate": 5.201424851626078e-06, "loss": 1.6299, "step": 4384 }, { "epoch": 0.5028093108588465, "grad_norm": 14.5625, "learning_rate": 5.1995693924490995e-06, "loss": 1.8414, "step": 4385 }, { "epoch": 0.5029239766081871, "grad_norm": 32.75, "learning_rate": 5.197713905745328e-06, "loss": 1.9075, "step": 4386 }, { "epoch": 0.5030386423575278, "grad_norm": 13.25, "learning_rate": 5.19585839177069e-06, "loss": 1.8519, "step": 4387 }, { "epoch": 0.5031533081068684, "grad_norm": 16.25, "learning_rate": 5.1940028507811225e-06, "loss": 1.9985, "step": 4388 }, { "epoch": 0.5032679738562091, "grad_norm": 15.375, "learning_rate": 5.192147283032558e-06, "loss": 1.9158, "step": 4389 }, { "epoch": 0.5033826396055499, "grad_norm": 16.25, "learning_rate": 5.190291688780939e-06, "loss": 1.7512, "step": 4390 }, { "epoch": 0.5034973053548905, "grad_norm": 15.125, "learning_rate": 5.18843606828221e-06, "loss": 1.7471, "step": 4391 }, { "epoch": 0.5036119711042312, "grad_norm": 15.0625, "learning_rate": 5.186580421792315e-06, "loss": 1.6386, "step": 4392 }, { "epoch": 0.5037266368535719, "grad_norm": 14.125, "learning_rate": 5.184724749567209e-06, "loss": 1.607, "step": 4393 }, { "epoch": 0.5038413026029125, "grad_norm": 12.625, "learning_rate": 5.182869051862844e-06, "loss": 1.5275, "step": 4394 }, { "epoch": 0.5039559683522532, "grad_norm": 12.25, "learning_rate": 5.18101332893518e-06, "loss": 1.9015, "step": 4395 }, { "epoch": 0.5040706341015938, "grad_norm": 12.375, "learning_rate": 5.179157581040179e-06, "loss": 1.7309, "step": 4396 }, { "epoch": 0.5041852998509345, "grad_norm": 16.125, "learning_rate": 5.177301808433802e-06, "loss": 1.4355, "step": 4397 }, { "epoch": 0.5042999656002752, "grad_norm": 12.5, "learning_rate": 5.175446011372022e-06, "loss": 1.686, "step": 4398 }, { "epoch": 0.5044146313496158, "grad_norm": 15.75, "learning_rate": 5.173590190110809e-06, "loss": 1.6653, "step": 4399 }, { "epoch": 0.5045292970989566, "grad_norm": 12.3125, "learning_rate": 5.171734344906136e-06, "loss": 1.7423, "step": 4400 }, { "epoch": 0.5046439628482973, "grad_norm": 15.0625, "learning_rate": 5.169878476013987e-06, "loss": 2.0851, "step": 4401 }, { "epoch": 0.5047586285976379, "grad_norm": 12.6875, "learning_rate": 5.168022583690338e-06, "loss": 1.6266, "step": 4402 }, { "epoch": 0.5048732943469786, "grad_norm": 14.875, "learning_rate": 5.1661666681911774e-06, "loss": 1.807, "step": 4403 }, { "epoch": 0.5049879600963192, "grad_norm": 14.6875, "learning_rate": 5.164310729772492e-06, "loss": 1.6446, "step": 4404 }, { "epoch": 0.5051026258456599, "grad_norm": 13.4375, "learning_rate": 5.162454768690274e-06, "loss": 1.4196, "step": 4405 }, { "epoch": 0.5052172915950006, "grad_norm": 14.5, "learning_rate": 5.160598785200516e-06, "loss": 1.791, "step": 4406 }, { "epoch": 0.5053319573443412, "grad_norm": 14.1875, "learning_rate": 5.158742779559217e-06, "loss": 1.8662, "step": 4407 }, { "epoch": 0.5054466230936819, "grad_norm": 14.5625, "learning_rate": 5.156886752022379e-06, "loss": 1.8209, "step": 4408 }, { "epoch": 0.5055612888430225, "grad_norm": 14.0625, "learning_rate": 5.1550307028460025e-06, "loss": 1.753, "step": 4409 }, { "epoch": 0.5056759545923633, "grad_norm": 14.875, "learning_rate": 5.153174632286098e-06, "loss": 1.7468, "step": 4410 }, { "epoch": 0.505790620341704, "grad_norm": 19.375, "learning_rate": 5.151318540598671e-06, "loss": 1.7453, "step": 4411 }, { "epoch": 0.5059052860910446, "grad_norm": 11.6875, "learning_rate": 5.149462428039734e-06, "loss": 1.8496, "step": 4412 }, { "epoch": 0.5060199518403853, "grad_norm": 29.0, "learning_rate": 5.1476062948653075e-06, "loss": 1.8054, "step": 4413 }, { "epoch": 0.506134617589726, "grad_norm": 13.5, "learning_rate": 5.145750141331404e-06, "loss": 1.5566, "step": 4414 }, { "epoch": 0.5062492833390666, "grad_norm": 15.0625, "learning_rate": 5.1438939676940485e-06, "loss": 1.7416, "step": 4415 }, { "epoch": 0.5063639490884073, "grad_norm": 14.4375, "learning_rate": 5.142037774209262e-06, "loss": 1.6433, "step": 4416 }, { "epoch": 0.5064786148377479, "grad_norm": 21.25, "learning_rate": 5.140181561133072e-06, "loss": 1.9827, "step": 4417 }, { "epoch": 0.5065932805870886, "grad_norm": 17.125, "learning_rate": 5.138325328721508e-06, "loss": 1.8011, "step": 4418 }, { "epoch": 0.5067079463364293, "grad_norm": 12.9375, "learning_rate": 5.1364690772306e-06, "loss": 1.5757, "step": 4419 }, { "epoch": 0.50682261208577, "grad_norm": 14.0625, "learning_rate": 5.134612806916387e-06, "loss": 1.9918, "step": 4420 }, { "epoch": 0.5069372778351107, "grad_norm": 12.9375, "learning_rate": 5.132756518034901e-06, "loss": 1.5006, "step": 4421 }, { "epoch": 0.5070519435844513, "grad_norm": 13.5625, "learning_rate": 5.1309002108421845e-06, "loss": 1.5938, "step": 4422 }, { "epoch": 0.507166609333792, "grad_norm": 20.25, "learning_rate": 5.12904388559428e-06, "loss": 2.0016, "step": 4423 }, { "epoch": 0.5072812750831327, "grad_norm": 17.125, "learning_rate": 5.1271875425472304e-06, "loss": 1.9894, "step": 4424 }, { "epoch": 0.5073959408324733, "grad_norm": 34.75, "learning_rate": 5.125331181957084e-06, "loss": 1.8046, "step": 4425 }, { "epoch": 0.507510606581814, "grad_norm": 13.0, "learning_rate": 5.123474804079891e-06, "loss": 1.6434, "step": 4426 }, { "epoch": 0.5076252723311547, "grad_norm": 12.75, "learning_rate": 5.1216184091717e-06, "loss": 1.5069, "step": 4427 }, { "epoch": 0.5077399380804953, "grad_norm": 17.375, "learning_rate": 5.119761997488569e-06, "loss": 1.8847, "step": 4428 }, { "epoch": 0.507854603829836, "grad_norm": 10.8125, "learning_rate": 5.117905569286553e-06, "loss": 1.41, "step": 4429 }, { "epoch": 0.5079692695791767, "grad_norm": 15.125, "learning_rate": 5.116049124821714e-06, "loss": 1.7341, "step": 4430 }, { "epoch": 0.5080839353285174, "grad_norm": 16.875, "learning_rate": 5.114192664350107e-06, "loss": 1.6093, "step": 4431 }, { "epoch": 0.5081986010778581, "grad_norm": 14.4375, "learning_rate": 5.112336188127799e-06, "loss": 1.5151, "step": 4432 }, { "epoch": 0.5083132668271987, "grad_norm": 13.6875, "learning_rate": 5.110479696410857e-06, "loss": 1.8506, "step": 4433 }, { "epoch": 0.5084279325765394, "grad_norm": 16.125, "learning_rate": 5.1086231894553436e-06, "loss": 1.7548, "step": 4434 }, { "epoch": 0.5085425983258801, "grad_norm": 13.0, "learning_rate": 5.106766667517336e-06, "loss": 1.6111, "step": 4435 }, { "epoch": 0.5086572640752207, "grad_norm": 14.875, "learning_rate": 5.104910130852899e-06, "loss": 2.2369, "step": 4436 }, { "epoch": 0.5087719298245614, "grad_norm": 14.5, "learning_rate": 5.103053579718109e-06, "loss": 1.602, "step": 4437 }, { "epoch": 0.508886595573902, "grad_norm": 11.625, "learning_rate": 5.101197014369044e-06, "loss": 1.7907, "step": 4438 }, { "epoch": 0.5090012613232427, "grad_norm": 15.4375, "learning_rate": 5.099340435061778e-06, "loss": 1.8261, "step": 4439 }, { "epoch": 0.5091159270725835, "grad_norm": 13.875, "learning_rate": 5.097483842052393e-06, "loss": 1.9191, "step": 4440 }, { "epoch": 0.5092305928219241, "grad_norm": 15.9375, "learning_rate": 5.095627235596971e-06, "loss": 1.5979, "step": 4441 }, { "epoch": 0.5093452585712648, "grad_norm": 12.625, "learning_rate": 5.093770615951594e-06, "loss": 1.6813, "step": 4442 }, { "epoch": 0.5094599243206054, "grad_norm": 14.4375, "learning_rate": 5.091913983372347e-06, "loss": 1.7907, "step": 4443 }, { "epoch": 0.5095745900699461, "grad_norm": 12.5625, "learning_rate": 5.09005733811532e-06, "loss": 1.6847, "step": 4444 }, { "epoch": 0.5096892558192868, "grad_norm": 16.125, "learning_rate": 5.0882006804365995e-06, "loss": 2.0573, "step": 4445 }, { "epoch": 0.5098039215686274, "grad_norm": 15.125, "learning_rate": 5.086344010592276e-06, "loss": 1.9961, "step": 4446 }, { "epoch": 0.5099185873179681, "grad_norm": 16.125, "learning_rate": 5.0844873288384415e-06, "loss": 2.0966, "step": 4447 }, { "epoch": 0.5100332530673088, "grad_norm": 16.75, "learning_rate": 5.082630635431192e-06, "loss": 1.7447, "step": 4448 }, { "epoch": 0.5101479188166494, "grad_norm": 14.9375, "learning_rate": 5.080773930626619e-06, "loss": 2.1917, "step": 4449 }, { "epoch": 0.5102625845659902, "grad_norm": 13.8125, "learning_rate": 5.078917214680823e-06, "loss": 1.6265, "step": 4450 }, { "epoch": 0.5103772503153308, "grad_norm": 15.125, "learning_rate": 5.077060487849903e-06, "loss": 1.7989, "step": 4451 }, { "epoch": 0.5104919160646715, "grad_norm": 15.6875, "learning_rate": 5.0752037503899566e-06, "loss": 2.0955, "step": 4452 }, { "epoch": 0.5106065818140122, "grad_norm": 11.875, "learning_rate": 5.073347002557085e-06, "loss": 1.7274, "step": 4453 }, { "epoch": 0.5107212475633528, "grad_norm": 16.25, "learning_rate": 5.071490244607394e-06, "loss": 1.8822, "step": 4454 }, { "epoch": 0.5108359133126935, "grad_norm": 16.375, "learning_rate": 5.0696334767969864e-06, "loss": 1.9065, "step": 4455 }, { "epoch": 0.5109505790620341, "grad_norm": 13.0625, "learning_rate": 5.067776699381969e-06, "loss": 1.5846, "step": 4456 }, { "epoch": 0.5110652448113748, "grad_norm": 13.375, "learning_rate": 5.065919912618445e-06, "loss": 1.4821, "step": 4457 }, { "epoch": 0.5111799105607155, "grad_norm": 15.4375, "learning_rate": 5.064063116762529e-06, "loss": 1.7428, "step": 4458 }, { "epoch": 0.5112945763100561, "grad_norm": 13.0, "learning_rate": 5.062206312070324e-06, "loss": 1.9624, "step": 4459 }, { "epoch": 0.5114092420593969, "grad_norm": 14.6875, "learning_rate": 5.060349498797946e-06, "loss": 1.7491, "step": 4460 }, { "epoch": 0.5115239078087376, "grad_norm": 13.4375, "learning_rate": 5.058492677201505e-06, "loss": 1.8379, "step": 4461 }, { "epoch": 0.5116385735580782, "grad_norm": 16.625, "learning_rate": 5.056635847537112e-06, "loss": 2.3333, "step": 4462 }, { "epoch": 0.5117532393074189, "grad_norm": 37.5, "learning_rate": 5.054779010060887e-06, "loss": 2.7167, "step": 4463 }, { "epoch": 0.5118679050567595, "grad_norm": 15.125, "learning_rate": 5.05292216502894e-06, "loss": 1.7035, "step": 4464 }, { "epoch": 0.5119825708061002, "grad_norm": 12.3125, "learning_rate": 5.051065312697388e-06, "loss": 1.8378, "step": 4465 }, { "epoch": 0.5120972365554409, "grad_norm": 13.5625, "learning_rate": 5.049208453322352e-06, "loss": 1.7983, "step": 4466 }, { "epoch": 0.5122119023047815, "grad_norm": 20.0, "learning_rate": 5.047351587159945e-06, "loss": 1.532, "step": 4467 }, { "epoch": 0.5123265680541222, "grad_norm": 14.5, "learning_rate": 5.04549471446629e-06, "loss": 1.806, "step": 4468 }, { "epoch": 0.512441233803463, "grad_norm": 13.625, "learning_rate": 5.043637835497507e-06, "loss": 1.5748, "step": 4469 }, { "epoch": 0.5125558995528036, "grad_norm": 20.5, "learning_rate": 5.041780950509717e-06, "loss": 2.3475, "step": 4470 }, { "epoch": 0.5126705653021443, "grad_norm": 13.875, "learning_rate": 5.0399240597590406e-06, "loss": 1.3226, "step": 4471 }, { "epoch": 0.5127852310514849, "grad_norm": 15.375, "learning_rate": 5.038067163501601e-06, "loss": 1.8417, "step": 4472 }, { "epoch": 0.5128998968008256, "grad_norm": 12.0625, "learning_rate": 5.036210261993523e-06, "loss": 1.6296, "step": 4473 }, { "epoch": 0.5130145625501663, "grad_norm": 19.625, "learning_rate": 5.0343533554909275e-06, "loss": 1.515, "step": 4474 }, { "epoch": 0.5131292282995069, "grad_norm": 14.25, "learning_rate": 5.032496444249942e-06, "loss": 1.8447, "step": 4475 }, { "epoch": 0.5132438940488476, "grad_norm": 11.125, "learning_rate": 5.030639528526693e-06, "loss": 1.4915, "step": 4476 }, { "epoch": 0.5133585597981882, "grad_norm": 18.375, "learning_rate": 5.028782608577304e-06, "loss": 1.7246, "step": 4477 }, { "epoch": 0.5134732255475289, "grad_norm": 12.9375, "learning_rate": 5.026925684657902e-06, "loss": 1.6995, "step": 4478 }, { "epoch": 0.5135878912968697, "grad_norm": 15.75, "learning_rate": 5.025068757024616e-06, "loss": 1.7801, "step": 4479 }, { "epoch": 0.5137025570462103, "grad_norm": 13.875, "learning_rate": 5.023211825933573e-06, "loss": 1.8207, "step": 4480 }, { "epoch": 0.513817222795551, "grad_norm": 13.0625, "learning_rate": 5.0213548916409006e-06, "loss": 2.114, "step": 4481 }, { "epoch": 0.5139318885448917, "grad_norm": 16.125, "learning_rate": 5.019497954402728e-06, "loss": 1.9397, "step": 4482 }, { "epoch": 0.5140465542942323, "grad_norm": 15.375, "learning_rate": 5.017641014475184e-06, "loss": 1.5385, "step": 4483 }, { "epoch": 0.514161220043573, "grad_norm": 14.75, "learning_rate": 5.015784072114398e-06, "loss": 1.547, "step": 4484 }, { "epoch": 0.5142758857929136, "grad_norm": 15.8125, "learning_rate": 5.013927127576501e-06, "loss": 1.8428, "step": 4485 }, { "epoch": 0.5143905515422543, "grad_norm": 31.5, "learning_rate": 5.012070181117622e-06, "loss": 2.2425, "step": 4486 }, { "epoch": 0.514505217291595, "grad_norm": 12.875, "learning_rate": 5.010213232993891e-06, "loss": 1.5164, "step": 4487 }, { "epoch": 0.5146198830409356, "grad_norm": 14.875, "learning_rate": 5.008356283461441e-06, "loss": 1.6736, "step": 4488 }, { "epoch": 0.5147345487902764, "grad_norm": 12.375, "learning_rate": 5.0064993327764e-06, "loss": 1.5742, "step": 4489 }, { "epoch": 0.514849214539617, "grad_norm": 14.0625, "learning_rate": 5.0046423811948995e-06, "loss": 1.7618, "step": 4490 }, { "epoch": 0.5149638802889577, "grad_norm": 12.3125, "learning_rate": 5.002785428973071e-06, "loss": 1.5649, "step": 4491 }, { "epoch": 0.5150785460382984, "grad_norm": 14.625, "learning_rate": 5.000928476367046e-06, "loss": 1.5336, "step": 4492 }, { "epoch": 0.515193211787639, "grad_norm": 14.5, "learning_rate": 4.999071523632955e-06, "loss": 1.6345, "step": 4493 }, { "epoch": 0.5153078775369797, "grad_norm": 37.25, "learning_rate": 4.9972145710269295e-06, "loss": 2.0725, "step": 4494 }, { "epoch": 0.5154225432863204, "grad_norm": 16.625, "learning_rate": 4.995357618805102e-06, "loss": 1.7238, "step": 4495 }, { "epoch": 0.515537209035661, "grad_norm": 14.375, "learning_rate": 4.993500667223602e-06, "loss": 1.75, "step": 4496 }, { "epoch": 0.5156518747850017, "grad_norm": 12.8125, "learning_rate": 4.991643716538561e-06, "loss": 1.7809, "step": 4497 }, { "epoch": 0.5157665405343423, "grad_norm": 13.125, "learning_rate": 4.989786767006109e-06, "loss": 2.0362, "step": 4498 }, { "epoch": 0.5158812062836831, "grad_norm": 18.875, "learning_rate": 4.987929818882379e-06, "loss": 1.6277, "step": 4499 }, { "epoch": 0.5159958720330238, "grad_norm": 11.25, "learning_rate": 4.9860728724235e-06, "loss": 1.7243, "step": 4500 }, { "epoch": 0.5161105377823644, "grad_norm": 16.25, "learning_rate": 4.984215927885604e-06, "loss": 1.7674, "step": 4501 }, { "epoch": 0.5162252035317051, "grad_norm": 13.3125, "learning_rate": 4.982358985524818e-06, "loss": 1.6143, "step": 4502 }, { "epoch": 0.5163398692810458, "grad_norm": 16.75, "learning_rate": 4.980502045597273e-06, "loss": 1.8966, "step": 4503 }, { "epoch": 0.5164545350303864, "grad_norm": 13.25, "learning_rate": 4.978645108359101e-06, "loss": 1.8591, "step": 4504 }, { "epoch": 0.5165692007797271, "grad_norm": 12.875, "learning_rate": 4.976788174066429e-06, "loss": 1.462, "step": 4505 }, { "epoch": 0.5166838665290677, "grad_norm": 13.1875, "learning_rate": 4.974931242975386e-06, "loss": 1.9434, "step": 4506 }, { "epoch": 0.5167985322784084, "grad_norm": 13.9375, "learning_rate": 4.9730743153421005e-06, "loss": 1.335, "step": 4507 }, { "epoch": 0.5169131980277492, "grad_norm": 16.0, "learning_rate": 4.971217391422698e-06, "loss": 1.6549, "step": 4508 }, { "epoch": 0.5170278637770898, "grad_norm": 13.375, "learning_rate": 4.969360471473308e-06, "loss": 1.6347, "step": 4509 }, { "epoch": 0.5171425295264305, "grad_norm": 13.0, "learning_rate": 4.967503555750059e-06, "loss": 1.7804, "step": 4510 }, { "epoch": 0.5172571952757711, "grad_norm": 13.1875, "learning_rate": 4.965646644509074e-06, "loss": 1.626, "step": 4511 }, { "epoch": 0.5173718610251118, "grad_norm": 14.9375, "learning_rate": 4.96378973800648e-06, "loss": 1.882, "step": 4512 }, { "epoch": 0.5174865267744525, "grad_norm": 21.0, "learning_rate": 4.9619328364984e-06, "loss": 2.465, "step": 4513 }, { "epoch": 0.5176011925237931, "grad_norm": 17.875, "learning_rate": 4.960075940240961e-06, "loss": 1.7351, "step": 4514 }, { "epoch": 0.5177158582731338, "grad_norm": 14.5625, "learning_rate": 4.958219049490284e-06, "loss": 1.9466, "step": 4515 }, { "epoch": 0.5178305240224745, "grad_norm": 13.375, "learning_rate": 4.956362164502495e-06, "loss": 1.4938, "step": 4516 }, { "epoch": 0.5179451897718151, "grad_norm": 14.8125, "learning_rate": 4.954505285533711e-06, "loss": 1.5778, "step": 4517 }, { "epoch": 0.5180598555211559, "grad_norm": 14.0, "learning_rate": 4.952648412840056e-06, "loss": 1.8403, "step": 4518 }, { "epoch": 0.5181745212704965, "grad_norm": 17.625, "learning_rate": 4.95079154667765e-06, "loss": 1.7955, "step": 4519 }, { "epoch": 0.5182891870198372, "grad_norm": 32.25, "learning_rate": 4.948934687302613e-06, "loss": 2.0564, "step": 4520 }, { "epoch": 0.5184038527691779, "grad_norm": 13.5625, "learning_rate": 4.947077834971062e-06, "loss": 1.958, "step": 4521 }, { "epoch": 0.5185185185185185, "grad_norm": 13.625, "learning_rate": 4.945220989939116e-06, "loss": 1.6273, "step": 4522 }, { "epoch": 0.5186331842678592, "grad_norm": 15.5, "learning_rate": 4.943364152462887e-06, "loss": 2.2222, "step": 4523 }, { "epoch": 0.5187478500171998, "grad_norm": 15.8125, "learning_rate": 4.9415073227984955e-06, "loss": 1.6354, "step": 4524 }, { "epoch": 0.5188625157665405, "grad_norm": 12.375, "learning_rate": 4.939650501202055e-06, "loss": 1.932, "step": 4525 }, { "epoch": 0.5189771815158812, "grad_norm": 15.4375, "learning_rate": 4.937793687929677e-06, "loss": 1.5224, "step": 4526 }, { "epoch": 0.5190918472652218, "grad_norm": 14.3125, "learning_rate": 4.935936883237474e-06, "loss": 1.3934, "step": 4527 }, { "epoch": 0.5192065130145626, "grad_norm": 11.625, "learning_rate": 4.934080087381555e-06, "loss": 1.7521, "step": 4528 }, { "epoch": 0.5193211787639033, "grad_norm": 11.9375, "learning_rate": 4.932223300618033e-06, "loss": 1.7438, "step": 4529 }, { "epoch": 0.5194358445132439, "grad_norm": 28.25, "learning_rate": 4.930366523203014e-06, "loss": 1.799, "step": 4530 }, { "epoch": 0.5195505102625846, "grad_norm": 17.375, "learning_rate": 4.928509755392608e-06, "loss": 2.1931, "step": 4531 }, { "epoch": 0.5196651760119252, "grad_norm": 13.3125, "learning_rate": 4.926652997442916e-06, "loss": 2.1442, "step": 4532 }, { "epoch": 0.5197798417612659, "grad_norm": 18.0, "learning_rate": 4.924796249610045e-06, "loss": 1.3421, "step": 4533 }, { "epoch": 0.5198945075106066, "grad_norm": 13.375, "learning_rate": 4.9229395121500986e-06, "loss": 1.4256, "step": 4534 }, { "epoch": 0.5200091732599472, "grad_norm": 14.1875, "learning_rate": 4.921082785319178e-06, "loss": 1.703, "step": 4535 }, { "epoch": 0.5201238390092879, "grad_norm": 14.75, "learning_rate": 4.919226069373382e-06, "loss": 1.761, "step": 4536 }, { "epoch": 0.5202385047586287, "grad_norm": 14.9375, "learning_rate": 4.917369364568811e-06, "loss": 1.6649, "step": 4537 }, { "epoch": 0.5203531705079693, "grad_norm": 14.25, "learning_rate": 4.915512671161559e-06, "loss": 2.0094, "step": 4538 }, { "epoch": 0.52046783625731, "grad_norm": 13.1875, "learning_rate": 4.9136559894077265e-06, "loss": 1.6328, "step": 4539 }, { "epoch": 0.5205825020066506, "grad_norm": 13.875, "learning_rate": 4.911799319563402e-06, "loss": 1.9492, "step": 4540 }, { "epoch": 0.5206971677559913, "grad_norm": 13.0625, "learning_rate": 4.909942661884682e-06, "loss": 1.9178, "step": 4541 }, { "epoch": 0.520811833505332, "grad_norm": 13.75, "learning_rate": 4.908086016627654e-06, "loss": 2.1214, "step": 4542 }, { "epoch": 0.5209264992546726, "grad_norm": 14.0625, "learning_rate": 4.9062293840484075e-06, "loss": 1.2841, "step": 4543 }, { "epoch": 0.5210411650040133, "grad_norm": 26.25, "learning_rate": 4.90437276440303e-06, "loss": 1.5778, "step": 4544 }, { "epoch": 0.5211558307533539, "grad_norm": 15.0625, "learning_rate": 4.902516157947609e-06, "loss": 1.6171, "step": 4545 }, { "epoch": 0.5212704965026946, "grad_norm": 14.75, "learning_rate": 4.9006595649382235e-06, "loss": 1.8007, "step": 4546 }, { "epoch": 0.5213851622520354, "grad_norm": 15.375, "learning_rate": 4.89880298563096e-06, "loss": 2.2496, "step": 4547 }, { "epoch": 0.521499828001376, "grad_norm": 14.25, "learning_rate": 4.896946420281892e-06, "loss": 1.5276, "step": 4548 }, { "epoch": 0.5216144937507167, "grad_norm": 14.8125, "learning_rate": 4.895089869147102e-06, "loss": 1.5025, "step": 4549 }, { "epoch": 0.5217291595000574, "grad_norm": 14.3125, "learning_rate": 4.893233332482667e-06, "loss": 1.4869, "step": 4550 }, { "epoch": 0.521843825249398, "grad_norm": 15.1875, "learning_rate": 4.891376810544657e-06, "loss": 1.5232, "step": 4551 }, { "epoch": 0.5219584909987387, "grad_norm": 16.625, "learning_rate": 4.889520303589146e-06, "loss": 1.8956, "step": 4552 }, { "epoch": 0.5220731567480793, "grad_norm": 13.375, "learning_rate": 4.887663811872202e-06, "loss": 1.6835, "step": 4553 }, { "epoch": 0.52218782249742, "grad_norm": 16.125, "learning_rate": 4.885807335649894e-06, "loss": 1.4417, "step": 4554 }, { "epoch": 0.5223024882467607, "grad_norm": 16.625, "learning_rate": 4.8839508751782885e-06, "loss": 1.72, "step": 4555 }, { "epoch": 0.5224171539961013, "grad_norm": 14.0625, "learning_rate": 4.882094430713448e-06, "loss": 1.2834, "step": 4556 }, { "epoch": 0.522531819745442, "grad_norm": 21.375, "learning_rate": 4.880238002511433e-06, "loss": 1.9054, "step": 4557 }, { "epoch": 0.5226464854947827, "grad_norm": 15.0, "learning_rate": 4.878381590828301e-06, "loss": 2.0266, "step": 4558 }, { "epoch": 0.5227611512441234, "grad_norm": 15.375, "learning_rate": 4.876525195920111e-06, "loss": 1.8138, "step": 4559 }, { "epoch": 0.5228758169934641, "grad_norm": 17.5, "learning_rate": 4.874668818042918e-06, "loss": 1.4647, "step": 4560 }, { "epoch": 0.5229904827428047, "grad_norm": 19.25, "learning_rate": 4.87281245745277e-06, "loss": 1.6858, "step": 4561 }, { "epoch": 0.5231051484921454, "grad_norm": 14.125, "learning_rate": 4.870956114405722e-06, "loss": 1.8304, "step": 4562 }, { "epoch": 0.5232198142414861, "grad_norm": 14.3125, "learning_rate": 4.8690997891578154e-06, "loss": 1.9555, "step": 4563 }, { "epoch": 0.5233344799908267, "grad_norm": 17.375, "learning_rate": 4.8672434819651e-06, "loss": 1.7203, "step": 4564 }, { "epoch": 0.5234491457401674, "grad_norm": 15.25, "learning_rate": 4.865387193083615e-06, "loss": 1.8985, "step": 4565 }, { "epoch": 0.523563811489508, "grad_norm": 18.75, "learning_rate": 4.863530922769401e-06, "loss": 1.8672, "step": 4566 }, { "epoch": 0.5236784772388487, "grad_norm": 12.0, "learning_rate": 4.861674671278494e-06, "loss": 1.8726, "step": 4567 }, { "epoch": 0.5237931429881895, "grad_norm": 14.5, "learning_rate": 4.859818438866929e-06, "loss": 1.6251, "step": 4568 }, { "epoch": 0.5239078087375301, "grad_norm": 16.75, "learning_rate": 4.8579622257907396e-06, "loss": 1.8527, "step": 4569 }, { "epoch": 0.5240224744868708, "grad_norm": 13.125, "learning_rate": 4.856106032305954e-06, "loss": 1.5581, "step": 4570 }, { "epoch": 0.5241371402362115, "grad_norm": 15.3125, "learning_rate": 4.854249858668597e-06, "loss": 1.5142, "step": 4571 }, { "epoch": 0.5242518059855521, "grad_norm": 14.3125, "learning_rate": 4.852393705134695e-06, "loss": 2.1838, "step": 4572 }, { "epoch": 0.5243664717348928, "grad_norm": 13.625, "learning_rate": 4.850537571960266e-06, "loss": 1.9544, "step": 4573 }, { "epoch": 0.5244811374842334, "grad_norm": 13.1875, "learning_rate": 4.848681459401331e-06, "loss": 1.3205, "step": 4574 }, { "epoch": 0.5245958032335741, "grad_norm": 16.375, "learning_rate": 4.846825367713904e-06, "loss": 1.5779, "step": 4575 }, { "epoch": 0.5247104689829148, "grad_norm": 17.125, "learning_rate": 4.844969297153999e-06, "loss": 2.3932, "step": 4576 }, { "epoch": 0.5248251347322554, "grad_norm": 35.0, "learning_rate": 4.843113247977623e-06, "loss": 1.9734, "step": 4577 }, { "epoch": 0.5249398004815962, "grad_norm": 12.875, "learning_rate": 4.841257220440782e-06, "loss": 1.5309, "step": 4578 }, { "epoch": 0.5250544662309368, "grad_norm": 12.875, "learning_rate": 4.839401214799486e-06, "loss": 1.5247, "step": 4579 }, { "epoch": 0.5251691319802775, "grad_norm": 14.5625, "learning_rate": 4.837545231309728e-06, "loss": 2.0085, "step": 4580 }, { "epoch": 0.5252837977296182, "grad_norm": 17.5, "learning_rate": 4.83568927022751e-06, "loss": 2.1383, "step": 4581 }, { "epoch": 0.5253984634789588, "grad_norm": 12.8125, "learning_rate": 4.833833331808826e-06, "loss": 1.7202, "step": 4582 }, { "epoch": 0.5255131292282995, "grad_norm": 17.125, "learning_rate": 4.831977416309663e-06, "loss": 1.5435, "step": 4583 }, { "epoch": 0.5256277949776402, "grad_norm": 14.0625, "learning_rate": 4.830121523986015e-06, "loss": 1.8575, "step": 4584 }, { "epoch": 0.5257424607269808, "grad_norm": 15.875, "learning_rate": 4.828265655093865e-06, "loss": 1.7851, "step": 4585 }, { "epoch": 0.5258571264763215, "grad_norm": 13.3125, "learning_rate": 4.8264098098891936e-06, "loss": 1.5366, "step": 4586 }, { "epoch": 0.5259717922256621, "grad_norm": 13.1875, "learning_rate": 4.82455398862798e-06, "loss": 1.746, "step": 4587 }, { "epoch": 0.5260864579750029, "grad_norm": 15.0, "learning_rate": 4.822698191566199e-06, "loss": 1.3096, "step": 4588 }, { "epoch": 0.5262011237243436, "grad_norm": 19.75, "learning_rate": 4.820842418959824e-06, "loss": 1.9113, "step": 4589 }, { "epoch": 0.5263157894736842, "grad_norm": 55.25, "learning_rate": 4.8189866710648204e-06, "loss": 1.8285, "step": 4590 }, { "epoch": 0.5264304552230249, "grad_norm": 13.0, "learning_rate": 4.817130948137157e-06, "loss": 1.5432, "step": 4591 }, { "epoch": 0.5265451209723655, "grad_norm": 15.5625, "learning_rate": 4.815275250432792e-06, "loss": 1.5797, "step": 4592 }, { "epoch": 0.5266597867217062, "grad_norm": 16.75, "learning_rate": 4.813419578207684e-06, "loss": 1.6761, "step": 4593 }, { "epoch": 0.5267744524710469, "grad_norm": 13.1875, "learning_rate": 4.811563931717791e-06, "loss": 1.6877, "step": 4594 }, { "epoch": 0.5268891182203875, "grad_norm": 14.5625, "learning_rate": 4.809708311219062e-06, "loss": 1.7775, "step": 4595 }, { "epoch": 0.5270037839697282, "grad_norm": 20.875, "learning_rate": 4.807852716967443e-06, "loss": 2.1665, "step": 4596 }, { "epoch": 0.527118449719069, "grad_norm": 29.625, "learning_rate": 4.80599714921888e-06, "loss": 2.3181, "step": 4597 }, { "epoch": 0.5272331154684096, "grad_norm": 12.875, "learning_rate": 4.80414160822931e-06, "loss": 1.597, "step": 4598 }, { "epoch": 0.5273477812177503, "grad_norm": 20.375, "learning_rate": 4.8022860942546735e-06, "loss": 1.9187, "step": 4599 }, { "epoch": 0.5274624469670909, "grad_norm": 14.3125, "learning_rate": 4.800430607550901e-06, "loss": 1.785, "step": 4600 }, { "epoch": 0.5275771127164316, "grad_norm": 13.8125, "learning_rate": 4.798575148373924e-06, "loss": 1.8135, "step": 4601 }, { "epoch": 0.5276917784657723, "grad_norm": 16.375, "learning_rate": 4.796719716979664e-06, "loss": 1.8877, "step": 4602 }, { "epoch": 0.5278064442151129, "grad_norm": 12.5625, "learning_rate": 4.794864313624042e-06, "loss": 1.5459, "step": 4603 }, { "epoch": 0.5279211099644536, "grad_norm": 12.6875, "learning_rate": 4.793008938562981e-06, "loss": 1.362, "step": 4604 }, { "epoch": 0.5280357757137943, "grad_norm": 13.25, "learning_rate": 4.79115359205239e-06, "loss": 1.6636, "step": 4605 }, { "epoch": 0.528150441463135, "grad_norm": 13.5, "learning_rate": 4.789298274348181e-06, "loss": 1.7731, "step": 4606 }, { "epoch": 0.5282651072124757, "grad_norm": 15.9375, "learning_rate": 4.787442985706259e-06, "loss": 2.0604, "step": 4607 }, { "epoch": 0.5283797729618163, "grad_norm": 13.4375, "learning_rate": 4.785587726382523e-06, "loss": 1.9172, "step": 4608 }, { "epoch": 0.528494438711157, "grad_norm": 15.75, "learning_rate": 4.783732496632873e-06, "loss": 1.8132, "step": 4609 }, { "epoch": 0.5286091044604977, "grad_norm": 16.25, "learning_rate": 4.781877296713206e-06, "loss": 1.5278, "step": 4610 }, { "epoch": 0.5287237702098383, "grad_norm": 16.375, "learning_rate": 4.780022126879406e-06, "loss": 1.8086, "step": 4611 }, { "epoch": 0.528838435959179, "grad_norm": 13.875, "learning_rate": 4.778166987387361e-06, "loss": 1.32, "step": 4612 }, { "epoch": 0.5289531017085196, "grad_norm": 13.625, "learning_rate": 4.77631187849295e-06, "loss": 1.6425, "step": 4613 }, { "epoch": 0.5290677674578603, "grad_norm": 29.25, "learning_rate": 4.774456800452053e-06, "loss": 1.6381, "step": 4614 }, { "epoch": 0.529182433207201, "grad_norm": 14.6875, "learning_rate": 4.7726017535205405e-06, "loss": 1.6402, "step": 4615 }, { "epoch": 0.5292970989565416, "grad_norm": 14.25, "learning_rate": 4.770746737954283e-06, "loss": 1.6272, "step": 4616 }, { "epoch": 0.5294117647058824, "grad_norm": 14.8125, "learning_rate": 4.768891754009141e-06, "loss": 2.1675, "step": 4617 }, { "epoch": 0.5295264304552231, "grad_norm": 14.25, "learning_rate": 4.767036801940976e-06, "loss": 1.4029, "step": 4618 }, { "epoch": 0.5296410962045637, "grad_norm": 13.875, "learning_rate": 4.765181882005645e-06, "loss": 1.8554, "step": 4619 }, { "epoch": 0.5297557619539044, "grad_norm": 12.375, "learning_rate": 4.763326994458998e-06, "loss": 1.7038, "step": 4620 }, { "epoch": 0.529870427703245, "grad_norm": 16.75, "learning_rate": 4.761472139556879e-06, "loss": 1.407, "step": 4621 }, { "epoch": 0.5299850934525857, "grad_norm": 18.0, "learning_rate": 4.759617317555133e-06, "loss": 1.8881, "step": 4622 }, { "epoch": 0.5300997592019264, "grad_norm": 13.3125, "learning_rate": 4.757762528709594e-06, "loss": 1.5367, "step": 4623 }, { "epoch": 0.530214424951267, "grad_norm": 13.875, "learning_rate": 4.755907773276097e-06, "loss": 1.8488, "step": 4624 }, { "epoch": 0.5303290907006077, "grad_norm": 14.375, "learning_rate": 4.754053051510472e-06, "loss": 1.8963, "step": 4625 }, { "epoch": 0.5304437564499485, "grad_norm": 15.25, "learning_rate": 4.7521983636685395e-06, "loss": 1.7358, "step": 4626 }, { "epoch": 0.5305584221992891, "grad_norm": 20.875, "learning_rate": 4.7503437100061186e-06, "loss": 2.0249, "step": 4627 }, { "epoch": 0.5306730879486298, "grad_norm": 18.0, "learning_rate": 4.7484890907790225e-06, "loss": 2.1258, "step": 4628 }, { "epoch": 0.5307877536979704, "grad_norm": 13.625, "learning_rate": 4.7466345062430654e-06, "loss": 2.034, "step": 4629 }, { "epoch": 0.5309024194473111, "grad_norm": 13.5625, "learning_rate": 4.744779956654046e-06, "loss": 1.6578, "step": 4630 }, { "epoch": 0.5310170851966518, "grad_norm": 12.375, "learning_rate": 4.742925442267769e-06, "loss": 1.5863, "step": 4631 }, { "epoch": 0.5311317509459924, "grad_norm": 15.0, "learning_rate": 4.741070963340026e-06, "loss": 1.9831, "step": 4632 }, { "epoch": 0.5312464166953331, "grad_norm": 15.625, "learning_rate": 4.739216520126606e-06, "loss": 1.7495, "step": 4633 }, { "epoch": 0.5313610824446737, "grad_norm": 16.625, "learning_rate": 4.737362112883297e-06, "loss": 1.9206, "step": 4634 }, { "epoch": 0.5314757481940144, "grad_norm": 16.375, "learning_rate": 4.735507741865879e-06, "loss": 1.3949, "step": 4635 }, { "epoch": 0.5315904139433552, "grad_norm": 13.75, "learning_rate": 4.733653407330126e-06, "loss": 1.6627, "step": 4636 }, { "epoch": 0.5317050796926958, "grad_norm": 16.875, "learning_rate": 4.73179910953181e-06, "loss": 1.6578, "step": 4637 }, { "epoch": 0.5318197454420365, "grad_norm": 11.9375, "learning_rate": 4.729944848726691e-06, "loss": 1.8603, "step": 4638 }, { "epoch": 0.5319344111913772, "grad_norm": 13.375, "learning_rate": 4.7280906251705356e-06, "loss": 1.5217, "step": 4639 }, { "epoch": 0.5320490769407178, "grad_norm": 16.125, "learning_rate": 4.726236439119094e-06, "loss": 1.7298, "step": 4640 }, { "epoch": 0.5321637426900585, "grad_norm": 13.8125, "learning_rate": 4.724382290828119e-06, "loss": 1.7747, "step": 4641 }, { "epoch": 0.5322784084393991, "grad_norm": 14.8125, "learning_rate": 4.722528180553351e-06, "loss": 1.7863, "step": 4642 }, { "epoch": 0.5323930741887398, "grad_norm": 14.75, "learning_rate": 4.72067410855053e-06, "loss": 1.7792, "step": 4643 }, { "epoch": 0.5325077399380805, "grad_norm": 15.25, "learning_rate": 4.718820075075394e-06, "loss": 1.9346, "step": 4644 }, { "epoch": 0.5326224056874211, "grad_norm": 20.0, "learning_rate": 4.716966080383669e-06, "loss": 1.7928, "step": 4645 }, { "epoch": 0.5327370714367619, "grad_norm": 12.875, "learning_rate": 4.715112124731077e-06, "loss": 1.8687, "step": 4646 }, { "epoch": 0.5328517371861025, "grad_norm": 28.75, "learning_rate": 4.713258208373338e-06, "loss": 1.9269, "step": 4647 }, { "epoch": 0.5329664029354432, "grad_norm": 13.0, "learning_rate": 4.711404331566161e-06, "loss": 1.7837, "step": 4648 }, { "epoch": 0.5330810686847839, "grad_norm": 16.875, "learning_rate": 4.709550494565257e-06, "loss": 1.7298, "step": 4649 }, { "epoch": 0.5331957344341245, "grad_norm": 14.375, "learning_rate": 4.707696697626327e-06, "loss": 1.595, "step": 4650 }, { "epoch": 0.5333104001834652, "grad_norm": 13.4375, "learning_rate": 4.705842941005065e-06, "loss": 1.6835, "step": 4651 }, { "epoch": 0.5334250659328059, "grad_norm": 15.5625, "learning_rate": 4.703989224957161e-06, "loss": 1.5754, "step": 4652 }, { "epoch": 0.5335397316821465, "grad_norm": 15.0, "learning_rate": 4.7021355497383005e-06, "loss": 1.6564, "step": 4653 }, { "epoch": 0.5336543974314872, "grad_norm": 12.3125, "learning_rate": 4.700281915604164e-06, "loss": 1.7264, "step": 4654 }, { "epoch": 0.5337690631808278, "grad_norm": 24.875, "learning_rate": 4.698428322810424e-06, "loss": 1.9109, "step": 4655 }, { "epoch": 0.5338837289301686, "grad_norm": 13.9375, "learning_rate": 4.6965747716127505e-06, "loss": 1.7053, "step": 4656 }, { "epoch": 0.5339983946795093, "grad_norm": 13.1875, "learning_rate": 4.694721262266802e-06, "loss": 1.7836, "step": 4657 }, { "epoch": 0.5341130604288499, "grad_norm": 12.625, "learning_rate": 4.692867795028234e-06, "loss": 1.7564, "step": 4658 }, { "epoch": 0.5342277261781906, "grad_norm": 12.75, "learning_rate": 4.691014370152701e-06, "loss": 2.0028, "step": 4659 }, { "epoch": 0.5343423919275313, "grad_norm": 24.375, "learning_rate": 4.689160987895847e-06, "loss": 2.0489, "step": 4660 }, { "epoch": 0.5344570576768719, "grad_norm": 19.625, "learning_rate": 4.687307648513308e-06, "loss": 1.6073, "step": 4661 }, { "epoch": 0.5345717234262126, "grad_norm": 11.25, "learning_rate": 4.685454352260719e-06, "loss": 1.4806, "step": 4662 }, { "epoch": 0.5346863891755532, "grad_norm": 19.0, "learning_rate": 4.683601099393705e-06, "loss": 1.78, "step": 4663 }, { "epoch": 0.5348010549248939, "grad_norm": 14.875, "learning_rate": 4.681747890167891e-06, "loss": 1.7899, "step": 4664 }, { "epoch": 0.5349157206742347, "grad_norm": 14.25, "learning_rate": 4.679894724838887e-06, "loss": 1.6506, "step": 4665 }, { "epoch": 0.5350303864235753, "grad_norm": 11.4375, "learning_rate": 4.6780416036623046e-06, "loss": 1.6071, "step": 4666 }, { "epoch": 0.535145052172916, "grad_norm": 12.8125, "learning_rate": 4.6761885268937454e-06, "loss": 1.5406, "step": 4667 }, { "epoch": 0.5352597179222566, "grad_norm": 26.375, "learning_rate": 4.674335494788806e-06, "loss": 1.4721, "step": 4668 }, { "epoch": 0.5353743836715973, "grad_norm": 17.625, "learning_rate": 4.672482507603079e-06, "loss": 1.577, "step": 4669 }, { "epoch": 0.535489049420938, "grad_norm": 30.0, "learning_rate": 4.6706295655921466e-06, "loss": 1.7593, "step": 4670 }, { "epoch": 0.5356037151702786, "grad_norm": 12.9375, "learning_rate": 4.668776669011587e-06, "loss": 2.0833, "step": 4671 }, { "epoch": 0.5357183809196193, "grad_norm": 14.1875, "learning_rate": 4.666923818116973e-06, "loss": 1.508, "step": 4672 }, { "epoch": 0.53583304666896, "grad_norm": 12.875, "learning_rate": 4.665071013163866e-06, "loss": 1.8246, "step": 4673 }, { "epoch": 0.5359477124183006, "grad_norm": 13.125, "learning_rate": 4.663218254407831e-06, "loss": 1.5667, "step": 4674 }, { "epoch": 0.5360623781676414, "grad_norm": 13.25, "learning_rate": 4.661365542104419e-06, "loss": 1.9114, "step": 4675 }, { "epoch": 0.536177043916982, "grad_norm": 15.0625, "learning_rate": 4.659512876509175e-06, "loss": 1.739, "step": 4676 }, { "epoch": 0.5362917096663227, "grad_norm": 14.375, "learning_rate": 4.657660257877638e-06, "loss": 1.4673, "step": 4677 }, { "epoch": 0.5364063754156634, "grad_norm": 12.25, "learning_rate": 4.6558076864653435e-06, "loss": 1.5898, "step": 4678 }, { "epoch": 0.536521041165004, "grad_norm": 14.1875, "learning_rate": 4.653955162527818e-06, "loss": 1.6824, "step": 4679 }, { "epoch": 0.5366357069143447, "grad_norm": 15.6875, "learning_rate": 4.652102686320581e-06, "loss": 2.2123, "step": 4680 }, { "epoch": 0.5367503726636853, "grad_norm": 17.875, "learning_rate": 4.6502502580991485e-06, "loss": 2.1254, "step": 4681 }, { "epoch": 0.536865038413026, "grad_norm": 13.5, "learning_rate": 4.648397878119026e-06, "loss": 1.8317, "step": 4682 }, { "epoch": 0.5369797041623667, "grad_norm": 14.25, "learning_rate": 4.646545546635712e-06, "loss": 1.7702, "step": 4683 }, { "epoch": 0.5370943699117073, "grad_norm": 12.125, "learning_rate": 4.644693263904704e-06, "loss": 1.5665, "step": 4684 }, { "epoch": 0.537209035661048, "grad_norm": 13.8125, "learning_rate": 4.642841030181487e-06, "loss": 1.802, "step": 4685 }, { "epoch": 0.5373237014103888, "grad_norm": 13.5625, "learning_rate": 4.640988845721541e-06, "loss": 1.7148, "step": 4686 }, { "epoch": 0.5374383671597294, "grad_norm": 12.1875, "learning_rate": 4.639136710780342e-06, "loss": 1.6814, "step": 4687 }, { "epoch": 0.5375530329090701, "grad_norm": 16.25, "learning_rate": 4.637284625613353e-06, "loss": 1.8906, "step": 4688 }, { "epoch": 0.5376676986584107, "grad_norm": 14.5, "learning_rate": 4.635432590476038e-06, "loss": 2.1522, "step": 4689 }, { "epoch": 0.5377823644077514, "grad_norm": 11.9375, "learning_rate": 4.633580605623847e-06, "loss": 1.5673, "step": 4690 }, { "epoch": 0.5378970301570921, "grad_norm": 14.625, "learning_rate": 4.631728671312228e-06, "loss": 2.067, "step": 4691 }, { "epoch": 0.5380116959064327, "grad_norm": 14.0, "learning_rate": 4.629876787796617e-06, "loss": 1.7885, "step": 4692 }, { "epoch": 0.5381263616557734, "grad_norm": 14.4375, "learning_rate": 4.628024955332447e-06, "loss": 1.798, "step": 4693 }, { "epoch": 0.5382410274051141, "grad_norm": 14.0625, "learning_rate": 4.626173174175147e-06, "loss": 2.154, "step": 4694 }, { "epoch": 0.5383556931544548, "grad_norm": 16.25, "learning_rate": 4.624321444580131e-06, "loss": 1.7561, "step": 4695 }, { "epoch": 0.5384703589037955, "grad_norm": 16.875, "learning_rate": 4.622469766802809e-06, "loss": 2.2608, "step": 4696 }, { "epoch": 0.5385850246531361, "grad_norm": 14.5625, "learning_rate": 4.620618141098589e-06, "loss": 2.0658, "step": 4697 }, { "epoch": 0.5386996904024768, "grad_norm": 13.0, "learning_rate": 4.618766567722861e-06, "loss": 1.5731, "step": 4698 }, { "epoch": 0.5388143561518175, "grad_norm": 16.625, "learning_rate": 4.61691504693102e-06, "loss": 2.3459, "step": 4699 }, { "epoch": 0.5389290219011581, "grad_norm": 15.0, "learning_rate": 4.6150635789784475e-06, "loss": 2.2883, "step": 4700 }, { "epoch": 0.5390436876504988, "grad_norm": 14.375, "learning_rate": 4.6132121641205165e-06, "loss": 1.9069, "step": 4701 }, { "epoch": 0.5391583533998394, "grad_norm": 15.0625, "learning_rate": 4.611360802612593e-06, "loss": 1.8441, "step": 4702 }, { "epoch": 0.5392730191491801, "grad_norm": 13.375, "learning_rate": 4.609509494710038e-06, "loss": 1.7998, "step": 4703 }, { "epoch": 0.5393876848985208, "grad_norm": 16.375, "learning_rate": 4.607658240668207e-06, "loss": 1.7452, "step": 4704 }, { "epoch": 0.5395023506478615, "grad_norm": 15.75, "learning_rate": 4.605807040742441e-06, "loss": 2.0309, "step": 4705 }, { "epoch": 0.5396170163972022, "grad_norm": 13.0625, "learning_rate": 4.603955895188081e-06, "loss": 1.606, "step": 4706 }, { "epoch": 0.5397316821465429, "grad_norm": 13.5625, "learning_rate": 4.602104804260456e-06, "loss": 1.5295, "step": 4707 }, { "epoch": 0.5398463478958835, "grad_norm": 16.625, "learning_rate": 4.600253768214887e-06, "loss": 1.7011, "step": 4708 }, { "epoch": 0.5399610136452242, "grad_norm": 20.75, "learning_rate": 4.598402787306692e-06, "loss": 2.3454, "step": 4709 }, { "epoch": 0.5400756793945648, "grad_norm": 15.875, "learning_rate": 4.596551861791179e-06, "loss": 2.0726, "step": 4710 }, { "epoch": 0.5401903451439055, "grad_norm": 13.875, "learning_rate": 4.594700991923644e-06, "loss": 1.578, "step": 4711 }, { "epoch": 0.5403050108932462, "grad_norm": 13.375, "learning_rate": 4.592850177959383e-06, "loss": 1.7918, "step": 4712 }, { "epoch": 0.5404196766425868, "grad_norm": 15.8125, "learning_rate": 4.590999420153677e-06, "loss": 1.6819, "step": 4713 }, { "epoch": 0.5405343423919275, "grad_norm": 16.0, "learning_rate": 4.589148718761806e-06, "loss": 1.8898, "step": 4714 }, { "epoch": 0.5406490081412682, "grad_norm": 12.3125, "learning_rate": 4.587298074039038e-06, "loss": 1.7204, "step": 4715 }, { "epoch": 0.5407636738906089, "grad_norm": 14.0, "learning_rate": 4.585447486240634e-06, "loss": 1.77, "step": 4716 }, { "epoch": 0.5408783396399496, "grad_norm": 14.5625, "learning_rate": 4.583596955621846e-06, "loss": 1.6438, "step": 4717 }, { "epoch": 0.5409930053892902, "grad_norm": 11.375, "learning_rate": 4.58174648243792e-06, "loss": 1.4255, "step": 4718 }, { "epoch": 0.5411076711386309, "grad_norm": 14.9375, "learning_rate": 4.579896066944096e-06, "loss": 1.7593, "step": 4719 }, { "epoch": 0.5412223368879716, "grad_norm": 15.3125, "learning_rate": 4.578045709395602e-06, "loss": 1.9705, "step": 4720 }, { "epoch": 0.5413370026373122, "grad_norm": 23.5, "learning_rate": 4.576195410047658e-06, "loss": 1.9872, "step": 4721 }, { "epoch": 0.5414516683866529, "grad_norm": 12.4375, "learning_rate": 4.57434516915548e-06, "loss": 1.7679, "step": 4722 }, { "epoch": 0.5415663341359935, "grad_norm": 16.625, "learning_rate": 4.572494986974269e-06, "loss": 2.3165, "step": 4723 }, { "epoch": 0.5416809998853342, "grad_norm": 19.75, "learning_rate": 4.5706448637592275e-06, "loss": 1.8013, "step": 4724 }, { "epoch": 0.541795665634675, "grad_norm": 14.0625, "learning_rate": 4.568794799765544e-06, "loss": 1.6188, "step": 4725 }, { "epoch": 0.5419103313840156, "grad_norm": 14.6875, "learning_rate": 4.566944795248398e-06, "loss": 1.8462, "step": 4726 }, { "epoch": 0.5420249971333563, "grad_norm": 13.625, "learning_rate": 4.565094850462961e-06, "loss": 1.7756, "step": 4727 }, { "epoch": 0.542139662882697, "grad_norm": 14.5625, "learning_rate": 4.563244965664399e-06, "loss": 1.5889, "step": 4728 }, { "epoch": 0.5422543286320376, "grad_norm": 12.0, "learning_rate": 4.5613951411078715e-06, "loss": 1.5751, "step": 4729 }, { "epoch": 0.5423689943813783, "grad_norm": 15.8125, "learning_rate": 4.559545377048522e-06, "loss": 2.09, "step": 4730 }, { "epoch": 0.5424836601307189, "grad_norm": 12.3125, "learning_rate": 4.557695673741493e-06, "loss": 1.3063, "step": 4731 }, { "epoch": 0.5425983258800596, "grad_norm": 15.625, "learning_rate": 4.555846031441915e-06, "loss": 2.0268, "step": 4732 }, { "epoch": 0.5427129916294003, "grad_norm": 14.9375, "learning_rate": 4.5539964504049076e-06, "loss": 2.0507, "step": 4733 }, { "epoch": 0.542827657378741, "grad_norm": 16.25, "learning_rate": 4.5521469308855905e-06, "loss": 1.8574, "step": 4734 }, { "epoch": 0.5429423231280817, "grad_norm": 30.0, "learning_rate": 4.5502974731390675e-06, "loss": 1.6136, "step": 4735 }, { "epoch": 0.5430569888774223, "grad_norm": 12.5625, "learning_rate": 4.548448077420435e-06, "loss": 1.41, "step": 4736 }, { "epoch": 0.543171654626763, "grad_norm": 13.4375, "learning_rate": 4.546598743984785e-06, "loss": 1.8836, "step": 4737 }, { "epoch": 0.5432863203761037, "grad_norm": 13.0625, "learning_rate": 4.544749473087192e-06, "loss": 1.7797, "step": 4738 }, { "epoch": 0.5434009861254443, "grad_norm": 15.125, "learning_rate": 4.542900264982735e-06, "loss": 2.1267, "step": 4739 }, { "epoch": 0.543515651874785, "grad_norm": 14.6875, "learning_rate": 4.541051119926471e-06, "loss": 1.4497, "step": 4740 }, { "epoch": 0.5436303176241257, "grad_norm": 14.75, "learning_rate": 4.5392020381734584e-06, "loss": 1.9128, "step": 4741 }, { "epoch": 0.5437449833734663, "grad_norm": 13.6875, "learning_rate": 4.53735301997874e-06, "loss": 2.1835, "step": 4742 }, { "epoch": 0.543859649122807, "grad_norm": 14.25, "learning_rate": 4.535504065597351e-06, "loss": 1.4751, "step": 4743 }, { "epoch": 0.5439743148721476, "grad_norm": 12.0, "learning_rate": 4.533655175284326e-06, "loss": 1.6515, "step": 4744 }, { "epoch": 0.5440889806214884, "grad_norm": 18.375, "learning_rate": 4.53180634929468e-06, "loss": 1.6493, "step": 4745 }, { "epoch": 0.5442036463708291, "grad_norm": 13.375, "learning_rate": 4.5299575878834226e-06, "loss": 2.2631, "step": 4746 }, { "epoch": 0.5443183121201697, "grad_norm": 13.875, "learning_rate": 4.528108891305556e-06, "loss": 1.7075, "step": 4747 }, { "epoch": 0.5444329778695104, "grad_norm": 11.9375, "learning_rate": 4.5262602598160715e-06, "loss": 1.8135, "step": 4748 }, { "epoch": 0.544547643618851, "grad_norm": 14.0, "learning_rate": 4.524411693669955e-06, "loss": 1.771, "step": 4749 }, { "epoch": 0.5446623093681917, "grad_norm": 16.625, "learning_rate": 4.522563193122181e-06, "loss": 1.8493, "step": 4750 }, { "epoch": 0.5447769751175324, "grad_norm": 17.125, "learning_rate": 4.520714758427714e-06, "loss": 1.8135, "step": 4751 }, { "epoch": 0.544891640866873, "grad_norm": 13.4375, "learning_rate": 4.518866389841508e-06, "loss": 1.7657, "step": 4752 }, { "epoch": 0.5450063066162137, "grad_norm": 14.625, "learning_rate": 4.517018087618511e-06, "loss": 2.0473, "step": 4753 }, { "epoch": 0.5451209723655545, "grad_norm": 18.125, "learning_rate": 4.515169852013666e-06, "loss": 2.1342, "step": 4754 }, { "epoch": 0.5452356381148951, "grad_norm": 14.4375, "learning_rate": 4.513321683281896e-06, "loss": 1.3002, "step": 4755 }, { "epoch": 0.5453503038642358, "grad_norm": 14.125, "learning_rate": 4.511473581678124e-06, "loss": 1.6097, "step": 4756 }, { "epoch": 0.5454649696135764, "grad_norm": 12.6875, "learning_rate": 4.50962554745726e-06, "loss": 1.5554, "step": 4757 }, { "epoch": 0.5455796353629171, "grad_norm": 13.5, "learning_rate": 4.507777580874201e-06, "loss": 1.7222, "step": 4758 }, { "epoch": 0.5456943011122578, "grad_norm": 11.8125, "learning_rate": 4.505929682183844e-06, "loss": 1.3549, "step": 4759 }, { "epoch": 0.5458089668615984, "grad_norm": 12.6875, "learning_rate": 4.50408185164107e-06, "loss": 1.6758, "step": 4760 }, { "epoch": 0.5459236326109391, "grad_norm": 13.5, "learning_rate": 4.502234089500752e-06, "loss": 1.7129, "step": 4761 }, { "epoch": 0.5460382983602798, "grad_norm": 15.5, "learning_rate": 4.5003863960177525e-06, "loss": 1.9691, "step": 4762 }, { "epoch": 0.5461529641096204, "grad_norm": 33.75, "learning_rate": 4.498538771446925e-06, "loss": 1.8755, "step": 4763 }, { "epoch": 0.5462676298589612, "grad_norm": 14.3125, "learning_rate": 4.4966912160431165e-06, "loss": 1.5628, "step": 4764 }, { "epoch": 0.5463822956083018, "grad_norm": 13.4375, "learning_rate": 4.494843730061161e-06, "loss": 1.8463, "step": 4765 }, { "epoch": 0.5464969613576425, "grad_norm": 30.0, "learning_rate": 4.4929963137558825e-06, "loss": 2.2558, "step": 4766 }, { "epoch": 0.5466116271069832, "grad_norm": 28.125, "learning_rate": 4.4911489673820995e-06, "loss": 1.9098, "step": 4767 }, { "epoch": 0.5467262928563238, "grad_norm": 13.375, "learning_rate": 4.489301691194615e-06, "loss": 1.4722, "step": 4768 }, { "epoch": 0.5468409586056645, "grad_norm": 15.1875, "learning_rate": 4.4874544854482284e-06, "loss": 1.805, "step": 4769 }, { "epoch": 0.5469556243550051, "grad_norm": 20.5, "learning_rate": 4.485607350397727e-06, "loss": 1.8473, "step": 4770 }, { "epoch": 0.5470702901043458, "grad_norm": 16.125, "learning_rate": 4.483760286297885e-06, "loss": 1.4623, "step": 4771 }, { "epoch": 0.5471849558536865, "grad_norm": 15.125, "learning_rate": 4.481913293403472e-06, "loss": 1.5222, "step": 4772 }, { "epoch": 0.5472996216030271, "grad_norm": 14.5625, "learning_rate": 4.480066371969243e-06, "loss": 1.7624, "step": 4773 }, { "epoch": 0.5474142873523679, "grad_norm": 16.375, "learning_rate": 4.478219522249947e-06, "loss": 1.7977, "step": 4774 }, { "epoch": 0.5475289531017086, "grad_norm": 39.75, "learning_rate": 4.476372744500323e-06, "loss": 1.6782, "step": 4775 }, { "epoch": 0.5476436188510492, "grad_norm": 19.5, "learning_rate": 4.474526038975099e-06, "loss": 1.5996, "step": 4776 }, { "epoch": 0.5477582846003899, "grad_norm": 26.125, "learning_rate": 4.4726794059289894e-06, "loss": 1.6465, "step": 4777 }, { "epoch": 0.5478729503497305, "grad_norm": 14.75, "learning_rate": 4.470832845616704e-06, "loss": 1.6683, "step": 4778 }, { "epoch": 0.5479876160990712, "grad_norm": 13.5, "learning_rate": 4.468986358292942e-06, "loss": 1.7806, "step": 4779 }, { "epoch": 0.5481022818484119, "grad_norm": 14.4375, "learning_rate": 4.4671399442123885e-06, "loss": 1.8694, "step": 4780 }, { "epoch": 0.5482169475977525, "grad_norm": 15.6875, "learning_rate": 4.465293603629724e-06, "loss": 1.9356, "step": 4781 }, { "epoch": 0.5483316133470932, "grad_norm": 17.125, "learning_rate": 4.4634473367996144e-06, "loss": 1.7719, "step": 4782 }, { "epoch": 0.5484462790964338, "grad_norm": 15.0625, "learning_rate": 4.461601143976715e-06, "loss": 1.4589, "step": 4783 }, { "epoch": 0.5485609448457746, "grad_norm": 14.9375, "learning_rate": 4.459755025415675e-06, "loss": 1.9742, "step": 4784 }, { "epoch": 0.5486756105951153, "grad_norm": 20.125, "learning_rate": 4.457908981371134e-06, "loss": 2.1173, "step": 4785 }, { "epoch": 0.5487902763444559, "grad_norm": 14.125, "learning_rate": 4.456063012097714e-06, "loss": 1.6127, "step": 4786 }, { "epoch": 0.5489049420937966, "grad_norm": 27.625, "learning_rate": 4.454217117850035e-06, "loss": 2.1169, "step": 4787 }, { "epoch": 0.5490196078431373, "grad_norm": 16.875, "learning_rate": 4.452371298882697e-06, "loss": 1.7826, "step": 4788 }, { "epoch": 0.5491342735924779, "grad_norm": 11.5, "learning_rate": 4.4505255554503035e-06, "loss": 1.8026, "step": 4789 }, { "epoch": 0.5492489393418186, "grad_norm": 14.625, "learning_rate": 4.448679887807434e-06, "loss": 1.5978, "step": 4790 }, { "epoch": 0.5493636050911592, "grad_norm": 22.125, "learning_rate": 4.446834296208665e-06, "loss": 1.7475, "step": 4791 }, { "epoch": 0.5494782708404999, "grad_norm": 16.75, "learning_rate": 4.444988780908561e-06, "loss": 1.879, "step": 4792 }, { "epoch": 0.5495929365898407, "grad_norm": 15.4375, "learning_rate": 4.443143342161673e-06, "loss": 1.7789, "step": 4793 }, { "epoch": 0.5497076023391813, "grad_norm": 14.125, "learning_rate": 4.4412979802225495e-06, "loss": 2.005, "step": 4794 }, { "epoch": 0.549822268088522, "grad_norm": 11.625, "learning_rate": 4.439452695345719e-06, "loss": 1.8691, "step": 4795 }, { "epoch": 0.5499369338378627, "grad_norm": 11.75, "learning_rate": 4.437607487785703e-06, "loss": 1.8616, "step": 4796 }, { "epoch": 0.5500515995872033, "grad_norm": 14.1875, "learning_rate": 4.435762357797015e-06, "loss": 1.7288, "step": 4797 }, { "epoch": 0.550166265336544, "grad_norm": 16.25, "learning_rate": 4.433917305634153e-06, "loss": 1.6835, "step": 4798 }, { "epoch": 0.5502809310858846, "grad_norm": 13.0625, "learning_rate": 4.432072331551608e-06, "loss": 1.8893, "step": 4799 }, { "epoch": 0.5503955968352253, "grad_norm": 14.875, "learning_rate": 4.4302274358038614e-06, "loss": 1.671, "step": 4800 }, { "epoch": 0.550510262584566, "grad_norm": 14.0, "learning_rate": 4.428382618645379e-06, "loss": 1.5439, "step": 4801 }, { "epoch": 0.5506249283339066, "grad_norm": 12.1875, "learning_rate": 4.4265378803306165e-06, "loss": 1.6769, "step": 4802 }, { "epoch": 0.5507395940832474, "grad_norm": 14.1875, "learning_rate": 4.424693221114022e-06, "loss": 1.8623, "step": 4803 }, { "epoch": 0.550854259832588, "grad_norm": 15.8125, "learning_rate": 4.422848641250032e-06, "loss": 1.7334, "step": 4804 }, { "epoch": 0.5509689255819287, "grad_norm": 25.125, "learning_rate": 4.4210041409930715e-06, "loss": 1.8972, "step": 4805 }, { "epoch": 0.5510835913312694, "grad_norm": 16.75, "learning_rate": 4.4191597205975525e-06, "loss": 1.5444, "step": 4806 }, { "epoch": 0.55119825708061, "grad_norm": 14.4375, "learning_rate": 4.417315380317879e-06, "loss": 1.7073, "step": 4807 }, { "epoch": 0.5513129228299507, "grad_norm": 16.375, "learning_rate": 4.41547112040844e-06, "loss": 1.7871, "step": 4808 }, { "epoch": 0.5514275885792914, "grad_norm": 13.75, "learning_rate": 4.413626941123618e-06, "loss": 2.0663, "step": 4809 }, { "epoch": 0.551542254328632, "grad_norm": 14.125, "learning_rate": 4.411782842717784e-06, "loss": 1.6947, "step": 4810 }, { "epoch": 0.5516569200779727, "grad_norm": 14.6875, "learning_rate": 4.409938825445293e-06, "loss": 1.6988, "step": 4811 }, { "epoch": 0.5517715858273133, "grad_norm": 14.0, "learning_rate": 4.408094889560494e-06, "loss": 1.8127, "step": 4812 }, { "epoch": 0.551886251576654, "grad_norm": 12.4375, "learning_rate": 4.40625103531772e-06, "loss": 1.6609, "step": 4813 }, { "epoch": 0.5520009173259948, "grad_norm": 32.0, "learning_rate": 4.404407262971301e-06, "loss": 1.8133, "step": 4814 }, { "epoch": 0.5521155830753354, "grad_norm": 13.9375, "learning_rate": 4.402563572775546e-06, "loss": 1.5667, "step": 4815 }, { "epoch": 0.5522302488246761, "grad_norm": 13.125, "learning_rate": 4.400719964984759e-06, "loss": 2.2128, "step": 4816 }, { "epoch": 0.5523449145740167, "grad_norm": 13.375, "learning_rate": 4.398876439853227e-06, "loss": 1.4623, "step": 4817 }, { "epoch": 0.5524595803233574, "grad_norm": 17.875, "learning_rate": 4.397032997635232e-06, "loss": 1.9639, "step": 4818 }, { "epoch": 0.5525742460726981, "grad_norm": 17.625, "learning_rate": 4.395189638585043e-06, "loss": 1.8411, "step": 4819 }, { "epoch": 0.5526889118220387, "grad_norm": 14.625, "learning_rate": 4.393346362956915e-06, "loss": 2.0522, "step": 4820 }, { "epoch": 0.5528035775713794, "grad_norm": 13.3125, "learning_rate": 4.391503171005091e-06, "loss": 1.6198, "step": 4821 }, { "epoch": 0.5529182433207201, "grad_norm": 17.0, "learning_rate": 4.389660062983807e-06, "loss": 1.7308, "step": 4822 }, { "epoch": 0.5530329090700608, "grad_norm": 17.75, "learning_rate": 4.38781703914728e-06, "loss": 1.7487, "step": 4823 }, { "epoch": 0.5531475748194015, "grad_norm": 13.1875, "learning_rate": 4.385974099749725e-06, "loss": 1.862, "step": 4824 }, { "epoch": 0.5532622405687421, "grad_norm": 14.0, "learning_rate": 4.38413124504534e-06, "loss": 1.9316, "step": 4825 }, { "epoch": 0.5533769063180828, "grad_norm": 11.875, "learning_rate": 4.382288475288309e-06, "loss": 1.2106, "step": 4826 }, { "epoch": 0.5534915720674235, "grad_norm": 16.5, "learning_rate": 4.380445790732808e-06, "loss": 2.0532, "step": 4827 }, { "epoch": 0.5536062378167641, "grad_norm": 14.125, "learning_rate": 4.378603191632999e-06, "loss": 1.6639, "step": 4828 }, { "epoch": 0.5537209035661048, "grad_norm": 12.875, "learning_rate": 4.3767606782430375e-06, "loss": 1.8491, "step": 4829 }, { "epoch": 0.5538355693154455, "grad_norm": 14.3125, "learning_rate": 4.374918250817059e-06, "loss": 1.6844, "step": 4830 }, { "epoch": 0.5539502350647861, "grad_norm": 14.1875, "learning_rate": 4.373075909609193e-06, "loss": 1.7722, "step": 4831 }, { "epoch": 0.5540649008141268, "grad_norm": 14.0625, "learning_rate": 4.371233654873556e-06, "loss": 1.6822, "step": 4832 }, { "epoch": 0.5541795665634675, "grad_norm": 16.375, "learning_rate": 4.369391486864249e-06, "loss": 1.601, "step": 4833 }, { "epoch": 0.5542942323128082, "grad_norm": 13.25, "learning_rate": 4.367549405835366e-06, "loss": 1.9129, "step": 4834 }, { "epoch": 0.5544088980621489, "grad_norm": 13.125, "learning_rate": 4.3657074120409885e-06, "loss": 1.5064, "step": 4835 }, { "epoch": 0.5545235638114895, "grad_norm": 15.8125, "learning_rate": 4.363865505735182e-06, "loss": 1.7466, "step": 4836 }, { "epoch": 0.5546382295608302, "grad_norm": 30.0, "learning_rate": 4.3620236871720036e-06, "loss": 1.7921, "step": 4837 }, { "epoch": 0.5547528953101708, "grad_norm": 25.0, "learning_rate": 4.360181956605495e-06, "loss": 1.715, "step": 4838 }, { "epoch": 0.5548675610595115, "grad_norm": 15.5, "learning_rate": 4.358340314289691e-06, "loss": 1.8136, "step": 4839 }, { "epoch": 0.5549822268088522, "grad_norm": 18.375, "learning_rate": 4.3564987604786095e-06, "loss": 1.6691, "step": 4840 }, { "epoch": 0.5550968925581928, "grad_norm": 15.8125, "learning_rate": 4.354657295426257e-06, "loss": 1.6221, "step": 4841 }, { "epoch": 0.5552115583075335, "grad_norm": 12.375, "learning_rate": 4.352815919386629e-06, "loss": 1.584, "step": 4842 }, { "epoch": 0.5553262240568743, "grad_norm": 16.0, "learning_rate": 4.350974632613708e-06, "loss": 1.6152, "step": 4843 }, { "epoch": 0.5554408898062149, "grad_norm": 15.0, "learning_rate": 4.349133435361467e-06, "loss": 1.5569, "step": 4844 }, { "epoch": 0.5555555555555556, "grad_norm": 14.625, "learning_rate": 4.347292327883862e-06, "loss": 1.873, "step": 4845 }, { "epoch": 0.5556702213048962, "grad_norm": 14.0625, "learning_rate": 4.345451310434837e-06, "loss": 1.7687, "step": 4846 }, { "epoch": 0.5557848870542369, "grad_norm": 13.375, "learning_rate": 4.343610383268328e-06, "loss": 1.5353, "step": 4847 }, { "epoch": 0.5558995528035776, "grad_norm": 13.9375, "learning_rate": 4.341769546638253e-06, "loss": 1.4342, "step": 4848 }, { "epoch": 0.5560142185529182, "grad_norm": 16.125, "learning_rate": 4.339928800798522e-06, "loss": 1.964, "step": 4849 }, { "epoch": 0.5561288843022589, "grad_norm": 15.0, "learning_rate": 4.338088146003032e-06, "loss": 1.6914, "step": 4850 }, { "epoch": 0.5562435500515995, "grad_norm": 14.625, "learning_rate": 4.336247582505667e-06, "loss": 1.5839, "step": 4851 }, { "epoch": 0.5563582158009402, "grad_norm": 16.75, "learning_rate": 4.334407110560293e-06, "loss": 2.011, "step": 4852 }, { "epoch": 0.556472881550281, "grad_norm": 13.75, "learning_rate": 4.33256673042077e-06, "loss": 1.9634, "step": 4853 }, { "epoch": 0.5565875472996216, "grad_norm": 16.5, "learning_rate": 4.330726442340946e-06, "loss": 1.7375, "step": 4854 }, { "epoch": 0.5567022130489623, "grad_norm": 16.875, "learning_rate": 4.328886246574651e-06, "loss": 1.8807, "step": 4855 }, { "epoch": 0.556816878798303, "grad_norm": 16.0, "learning_rate": 4.327046143375707e-06, "loss": 1.7609, "step": 4856 }, { "epoch": 0.5569315445476436, "grad_norm": 14.3125, "learning_rate": 4.32520613299792e-06, "loss": 1.5729, "step": 4857 }, { "epoch": 0.5570462102969843, "grad_norm": 15.5, "learning_rate": 4.323366215695081e-06, "loss": 1.6001, "step": 4858 }, { "epoch": 0.5571608760463249, "grad_norm": 12.75, "learning_rate": 4.321526391720977e-06, "loss": 1.5558, "step": 4859 }, { "epoch": 0.5572755417956656, "grad_norm": 14.125, "learning_rate": 4.319686661329375e-06, "loss": 1.6299, "step": 4860 }, { "epoch": 0.5573902075450063, "grad_norm": 13.0, "learning_rate": 4.317847024774029e-06, "loss": 1.6345, "step": 4861 }, { "epoch": 0.557504873294347, "grad_norm": 15.875, "learning_rate": 4.316007482308684e-06, "loss": 1.6409, "step": 4862 }, { "epoch": 0.5576195390436877, "grad_norm": 14.3125, "learning_rate": 4.314168034187066e-06, "loss": 1.8883, "step": 4863 }, { "epoch": 0.5577342047930284, "grad_norm": 13.375, "learning_rate": 4.312328680662897e-06, "loss": 1.3682, "step": 4864 }, { "epoch": 0.557848870542369, "grad_norm": 13.8125, "learning_rate": 4.3104894219898776e-06, "loss": 1.4882, "step": 4865 }, { "epoch": 0.5579635362917097, "grad_norm": 11.9375, "learning_rate": 4.3086502584217e-06, "loss": 1.5823, "step": 4866 }, { "epoch": 0.5580782020410503, "grad_norm": 14.375, "learning_rate": 4.3068111902120404e-06, "loss": 1.4767, "step": 4867 }, { "epoch": 0.558192867790391, "grad_norm": 20.75, "learning_rate": 4.304972217614563e-06, "loss": 1.6269, "step": 4868 }, { "epoch": 0.5583075335397317, "grad_norm": 18.5, "learning_rate": 4.30313334088292e-06, "loss": 2.0669, "step": 4869 }, { "epoch": 0.5584221992890723, "grad_norm": 17.0, "learning_rate": 4.30129456027075e-06, "loss": 1.75, "step": 4870 }, { "epoch": 0.558536865038413, "grad_norm": 20.25, "learning_rate": 4.299455876031677e-06, "loss": 2.022, "step": 4871 }, { "epoch": 0.5586515307877536, "grad_norm": 18.75, "learning_rate": 4.297617288419312e-06, "loss": 1.8477, "step": 4872 }, { "epoch": 0.5587661965370944, "grad_norm": 11.375, "learning_rate": 4.295778797687252e-06, "loss": 1.487, "step": 4873 }, { "epoch": 0.5588808622864351, "grad_norm": 14.5, "learning_rate": 4.293940404089084e-06, "loss": 1.6555, "step": 4874 }, { "epoch": 0.5589955280357757, "grad_norm": 16.375, "learning_rate": 4.29210210787838e-06, "loss": 2.2706, "step": 4875 }, { "epoch": 0.5591101937851164, "grad_norm": 17.0, "learning_rate": 4.290263909308696e-06, "loss": 1.4955, "step": 4876 }, { "epoch": 0.5592248595344571, "grad_norm": 16.5, "learning_rate": 4.2884258086335755e-06, "loss": 1.5897, "step": 4877 }, { "epoch": 0.5593395252837977, "grad_norm": 15.5, "learning_rate": 4.286587806106551e-06, "loss": 1.8616, "step": 4878 }, { "epoch": 0.5594541910331384, "grad_norm": 15.125, "learning_rate": 4.28474990198114e-06, "loss": 1.9631, "step": 4879 }, { "epoch": 0.559568856782479, "grad_norm": 16.25, "learning_rate": 4.282912096510846e-06, "loss": 1.9755, "step": 4880 }, { "epoch": 0.5596835225318197, "grad_norm": 15.4375, "learning_rate": 4.281074389949161e-06, "loss": 1.6772, "step": 4881 }, { "epoch": 0.5597981882811605, "grad_norm": 16.625, "learning_rate": 4.2792367825495595e-06, "loss": 2.2997, "step": 4882 }, { "epoch": 0.5599128540305011, "grad_norm": 12.75, "learning_rate": 4.2773992745655025e-06, "loss": 1.7262, "step": 4883 }, { "epoch": 0.5600275197798418, "grad_norm": 13.75, "learning_rate": 4.275561866250443e-06, "loss": 1.6938, "step": 4884 }, { "epoch": 0.5601421855291824, "grad_norm": 13.6875, "learning_rate": 4.273724557857816e-06, "loss": 1.6581, "step": 4885 }, { "epoch": 0.5602568512785231, "grad_norm": 14.6875, "learning_rate": 4.271887349641041e-06, "loss": 1.5842, "step": 4886 }, { "epoch": 0.5603715170278638, "grad_norm": 17.125, "learning_rate": 4.270050241853527e-06, "loss": 1.984, "step": 4887 }, { "epoch": 0.5604861827772044, "grad_norm": 16.75, "learning_rate": 4.2682132347486674e-06, "loss": 1.629, "step": 4888 }, { "epoch": 0.5606008485265451, "grad_norm": 13.375, "learning_rate": 4.266376328579845e-06, "loss": 2.1916, "step": 4889 }, { "epoch": 0.5607155142758858, "grad_norm": 18.375, "learning_rate": 4.264539523600423e-06, "loss": 1.7145, "step": 4890 }, { "epoch": 0.5608301800252264, "grad_norm": 25.0, "learning_rate": 4.262702820063754e-06, "loss": 1.4671, "step": 4891 }, { "epoch": 0.5609448457745672, "grad_norm": 15.9375, "learning_rate": 4.260866218223177e-06, "loss": 1.7018, "step": 4892 }, { "epoch": 0.5610595115239078, "grad_norm": 13.4375, "learning_rate": 4.2590297183320156e-06, "loss": 1.7926, "step": 4893 }, { "epoch": 0.5611741772732485, "grad_norm": 13.75, "learning_rate": 4.2571933206435816e-06, "loss": 1.5657, "step": 4894 }, { "epoch": 0.5612888430225892, "grad_norm": 11.75, "learning_rate": 4.2553570254111695e-06, "loss": 1.9433, "step": 4895 }, { "epoch": 0.5614035087719298, "grad_norm": 16.25, "learning_rate": 4.253520832888062e-06, "loss": 1.7897, "step": 4896 }, { "epoch": 0.5615181745212705, "grad_norm": 14.5625, "learning_rate": 4.251684743327526e-06, "loss": 1.8102, "step": 4897 }, { "epoch": 0.5616328402706112, "grad_norm": 14.0625, "learning_rate": 4.249848756982814e-06, "loss": 1.4214, "step": 4898 }, { "epoch": 0.5617475060199518, "grad_norm": 12.875, "learning_rate": 4.248012874107167e-06, "loss": 1.9615, "step": 4899 }, { "epoch": 0.5618621717692925, "grad_norm": 16.25, "learning_rate": 4.246177094953812e-06, "loss": 1.7284, "step": 4900 }, { "epoch": 0.5619768375186331, "grad_norm": 16.625, "learning_rate": 4.244341419775957e-06, "loss": 1.8255, "step": 4901 }, { "epoch": 0.5620915032679739, "grad_norm": 13.375, "learning_rate": 4.242505848826799e-06, "loss": 1.8642, "step": 4902 }, { "epoch": 0.5622061690173146, "grad_norm": 14.875, "learning_rate": 4.2406703823595175e-06, "loss": 2.2225, "step": 4903 }, { "epoch": 0.5623208347666552, "grad_norm": 12.5625, "learning_rate": 4.238835020627285e-06, "loss": 1.5742, "step": 4904 }, { "epoch": 0.5624355005159959, "grad_norm": 18.0, "learning_rate": 4.236999763883252e-06, "loss": 2.2488, "step": 4905 }, { "epoch": 0.5625501662653365, "grad_norm": 14.375, "learning_rate": 4.235164612380557e-06, "loss": 1.6605, "step": 4906 }, { "epoch": 0.5626648320146772, "grad_norm": 15.6875, "learning_rate": 4.2333295663723265e-06, "loss": 1.9347, "step": 4907 }, { "epoch": 0.5627794977640179, "grad_norm": 15.0625, "learning_rate": 4.231494626111665e-06, "loss": 1.9053, "step": 4908 }, { "epoch": 0.5628941635133585, "grad_norm": 12.875, "learning_rate": 4.229659791851672e-06, "loss": 1.5726, "step": 4909 }, { "epoch": 0.5630088292626992, "grad_norm": 12.625, "learning_rate": 4.227825063845427e-06, "loss": 1.4692, "step": 4910 }, { "epoch": 0.56312349501204, "grad_norm": 11.5, "learning_rate": 4.2259904423459966e-06, "loss": 1.5156, "step": 4911 }, { "epoch": 0.5632381607613806, "grad_norm": 15.3125, "learning_rate": 4.22415592760643e-06, "loss": 1.4533, "step": 4912 }, { "epoch": 0.5633528265107213, "grad_norm": 14.25, "learning_rate": 4.222321519879762e-06, "loss": 1.9515, "step": 4913 }, { "epoch": 0.5634674922600619, "grad_norm": 16.125, "learning_rate": 4.2204872194190196e-06, "loss": 1.9488, "step": 4914 }, { "epoch": 0.5635821580094026, "grad_norm": 13.6875, "learning_rate": 4.218653026477204e-06, "loss": 1.7225, "step": 4915 }, { "epoch": 0.5636968237587433, "grad_norm": 13.5625, "learning_rate": 4.216818941307312e-06, "loss": 2.4625, "step": 4916 }, { "epoch": 0.5638114895080839, "grad_norm": 18.25, "learning_rate": 4.214984964162317e-06, "loss": 2.2665, "step": 4917 }, { "epoch": 0.5639261552574246, "grad_norm": 13.5, "learning_rate": 4.213151095295181e-06, "loss": 2.0946, "step": 4918 }, { "epoch": 0.5640408210067652, "grad_norm": 12.4375, "learning_rate": 4.211317334958855e-06, "loss": 1.9989, "step": 4919 }, { "epoch": 0.5641554867561059, "grad_norm": 14.0, "learning_rate": 4.2094836834062695e-06, "loss": 1.7987, "step": 4920 }, { "epoch": 0.5642701525054467, "grad_norm": 13.8125, "learning_rate": 4.2076501408903396e-06, "loss": 1.9949, "step": 4921 }, { "epoch": 0.5643848182547873, "grad_norm": 12.625, "learning_rate": 4.20581670766397e-06, "loss": 2.0031, "step": 4922 }, { "epoch": 0.564499484004128, "grad_norm": 12.8125, "learning_rate": 4.2039833839800445e-06, "loss": 1.6289, "step": 4923 }, { "epoch": 0.5646141497534687, "grad_norm": 26.375, "learning_rate": 4.202150170091439e-06, "loss": 1.8984, "step": 4924 }, { "epoch": 0.5647288155028093, "grad_norm": 16.625, "learning_rate": 4.2003170662510116e-06, "loss": 1.698, "step": 4925 }, { "epoch": 0.56484348125215, "grad_norm": 14.5625, "learning_rate": 4.1984840727116e-06, "loss": 1.5708, "step": 4926 }, { "epoch": 0.5649581470014906, "grad_norm": 14.8125, "learning_rate": 4.1966511897260324e-06, "loss": 1.6029, "step": 4927 }, { "epoch": 0.5650728127508313, "grad_norm": 15.9375, "learning_rate": 4.194818417547118e-06, "loss": 1.8109, "step": 4928 }, { "epoch": 0.565187478500172, "grad_norm": 14.8125, "learning_rate": 4.1929857564276575e-06, "loss": 1.8227, "step": 4929 }, { "epoch": 0.5653021442495126, "grad_norm": 13.375, "learning_rate": 4.1911532066204276e-06, "loss": 2.0485, "step": 4930 }, { "epoch": 0.5654168099988534, "grad_norm": 20.5, "learning_rate": 4.189320768378196e-06, "loss": 2.0447, "step": 4931 }, { "epoch": 0.5655314757481941, "grad_norm": 12.3125, "learning_rate": 4.187488441953711e-06, "loss": 1.7719, "step": 4932 }, { "epoch": 0.5656461414975347, "grad_norm": 15.5625, "learning_rate": 4.185656227599705e-06, "loss": 1.6952, "step": 4933 }, { "epoch": 0.5657608072468754, "grad_norm": 13.6875, "learning_rate": 4.183824125568902e-06, "loss": 1.7649, "step": 4934 }, { "epoch": 0.565875472996216, "grad_norm": 18.75, "learning_rate": 4.181992136114003e-06, "loss": 2.047, "step": 4935 }, { "epoch": 0.5659901387455567, "grad_norm": 16.25, "learning_rate": 4.1801602594876955e-06, "loss": 1.4042, "step": 4936 }, { "epoch": 0.5661048044948974, "grad_norm": 14.4375, "learning_rate": 4.178328495942654e-06, "loss": 1.7245, "step": 4937 }, { "epoch": 0.566219470244238, "grad_norm": 14.5625, "learning_rate": 4.176496845731532e-06, "loss": 2.119, "step": 4938 }, { "epoch": 0.5663341359935787, "grad_norm": 14.5625, "learning_rate": 4.174665309106975e-06, "loss": 1.9526, "step": 4939 }, { "epoch": 0.5664488017429193, "grad_norm": 13.875, "learning_rate": 4.1728338863216036e-06, "loss": 2.0484, "step": 4940 }, { "epoch": 0.56656346749226, "grad_norm": 15.5, "learning_rate": 4.1710025776280325e-06, "loss": 1.6554, "step": 4941 }, { "epoch": 0.5666781332416008, "grad_norm": 13.5625, "learning_rate": 4.169171383278853e-06, "loss": 1.4492, "step": 4942 }, { "epoch": 0.5667927989909414, "grad_norm": 12.0625, "learning_rate": 4.167340303526643e-06, "loss": 1.5832, "step": 4943 }, { "epoch": 0.5669074647402821, "grad_norm": 36.0, "learning_rate": 4.1655093386239685e-06, "loss": 2.1026, "step": 4944 }, { "epoch": 0.5670221304896228, "grad_norm": 12.0625, "learning_rate": 4.163678488823373e-06, "loss": 1.7342, "step": 4945 }, { "epoch": 0.5671367962389634, "grad_norm": 14.875, "learning_rate": 4.161847754377388e-06, "loss": 1.4901, "step": 4946 }, { "epoch": 0.5672514619883041, "grad_norm": 12.875, "learning_rate": 4.160017135538529e-06, "loss": 1.9097, "step": 4947 }, { "epoch": 0.5673661277376447, "grad_norm": 35.5, "learning_rate": 4.158186632559293e-06, "loss": 2.0244, "step": 4948 }, { "epoch": 0.5674807934869854, "grad_norm": 16.625, "learning_rate": 4.156356245692166e-06, "loss": 1.6513, "step": 4949 }, { "epoch": 0.5675954592363261, "grad_norm": 14.5625, "learning_rate": 4.154525975189614e-06, "loss": 2.0671, "step": 4950 }, { "epoch": 0.5677101249856668, "grad_norm": 14.5, "learning_rate": 4.152695821304088e-06, "loss": 1.8036, "step": 4951 }, { "epoch": 0.5678247907350075, "grad_norm": 16.375, "learning_rate": 4.150865784288021e-06, "loss": 1.8917, "step": 4952 }, { "epoch": 0.5679394564843481, "grad_norm": 14.8125, "learning_rate": 4.149035864393832e-06, "loss": 1.7071, "step": 4953 }, { "epoch": 0.5680541222336888, "grad_norm": 15.25, "learning_rate": 4.147206061873927e-06, "loss": 2.0367, "step": 4954 }, { "epoch": 0.5681687879830295, "grad_norm": 19.75, "learning_rate": 4.145376376980689e-06, "loss": 1.4755, "step": 4955 }, { "epoch": 0.5682834537323701, "grad_norm": 12.5625, "learning_rate": 4.14354680996649e-06, "loss": 1.4392, "step": 4956 }, { "epoch": 0.5683981194817108, "grad_norm": 15.0625, "learning_rate": 4.141717361083683e-06, "loss": 1.9135, "step": 4957 }, { "epoch": 0.5685127852310515, "grad_norm": 11.75, "learning_rate": 4.1398880305846034e-06, "loss": 1.5359, "step": 4958 }, { "epoch": 0.5686274509803921, "grad_norm": 12.1875, "learning_rate": 4.138058818721576e-06, "loss": 1.7342, "step": 4959 }, { "epoch": 0.5687421167297328, "grad_norm": 17.375, "learning_rate": 4.136229725746906e-06, "loss": 1.6019, "step": 4960 }, { "epoch": 0.5688567824790735, "grad_norm": 24.625, "learning_rate": 4.134400751912879e-06, "loss": 1.8111, "step": 4961 }, { "epoch": 0.5689714482284142, "grad_norm": 13.5625, "learning_rate": 4.1325718974717695e-06, "loss": 1.4908, "step": 4962 }, { "epoch": 0.5690861139777549, "grad_norm": 15.8125, "learning_rate": 4.1307431626758296e-06, "loss": 1.7517, "step": 4963 }, { "epoch": 0.5692007797270955, "grad_norm": 14.375, "learning_rate": 4.128914547777303e-06, "loss": 1.8181, "step": 4964 }, { "epoch": 0.5693154454764362, "grad_norm": 12.125, "learning_rate": 4.12708605302841e-06, "loss": 1.2947, "step": 4965 }, { "epoch": 0.5694301112257769, "grad_norm": 11.5, "learning_rate": 4.125257678681357e-06, "loss": 1.6798, "step": 4966 }, { "epoch": 0.5695447769751175, "grad_norm": 14.0625, "learning_rate": 4.123429424988332e-06, "loss": 1.7018, "step": 4967 }, { "epoch": 0.5696594427244582, "grad_norm": 15.8125, "learning_rate": 4.121601292201508e-06, "loss": 1.6945, "step": 4968 }, { "epoch": 0.5697741084737988, "grad_norm": 13.4375, "learning_rate": 4.1197732805730445e-06, "loss": 1.7106, "step": 4969 }, { "epoch": 0.5698887742231395, "grad_norm": 16.5, "learning_rate": 4.1179453903550784e-06, "loss": 2.1972, "step": 4970 }, { "epoch": 0.5700034399724803, "grad_norm": 16.375, "learning_rate": 4.11611762179973e-06, "loss": 1.6984, "step": 4971 }, { "epoch": 0.5701181057218209, "grad_norm": 17.0, "learning_rate": 4.1142899751591095e-06, "loss": 1.7065, "step": 4972 }, { "epoch": 0.5702327714711616, "grad_norm": 14.6875, "learning_rate": 4.1124624506853004e-06, "loss": 1.5664, "step": 4973 }, { "epoch": 0.5703474372205022, "grad_norm": 11.9375, "learning_rate": 4.110635048630381e-06, "loss": 1.6614, "step": 4974 }, { "epoch": 0.5704621029698429, "grad_norm": 14.375, "learning_rate": 4.108807769246404e-06, "loss": 1.6795, "step": 4975 }, { "epoch": 0.5705767687191836, "grad_norm": 14.5625, "learning_rate": 4.106980612785407e-06, "loss": 1.7478, "step": 4976 }, { "epoch": 0.5706914344685242, "grad_norm": 12.75, "learning_rate": 4.105153579499411e-06, "loss": 1.8129, "step": 4977 }, { "epoch": 0.5708061002178649, "grad_norm": 12.125, "learning_rate": 4.1033266696404215e-06, "loss": 1.7179, "step": 4978 }, { "epoch": 0.5709207659672056, "grad_norm": 13.3125, "learning_rate": 4.101499883460427e-06, "loss": 1.9753, "step": 4979 }, { "epoch": 0.5710354317165462, "grad_norm": 15.1875, "learning_rate": 4.099673221211396e-06, "loss": 1.6178, "step": 4980 }, { "epoch": 0.571150097465887, "grad_norm": 16.875, "learning_rate": 4.097846683145282e-06, "loss": 2.0016, "step": 4981 }, { "epoch": 0.5712647632152276, "grad_norm": 29.75, "learning_rate": 4.096020269514024e-06, "loss": 1.9019, "step": 4982 }, { "epoch": 0.5713794289645683, "grad_norm": 30.75, "learning_rate": 4.094193980569534e-06, "loss": 1.7783, "step": 4983 }, { "epoch": 0.571494094713909, "grad_norm": 16.75, "learning_rate": 4.09236781656372e-06, "loss": 1.8042, "step": 4984 }, { "epoch": 0.5716087604632496, "grad_norm": 15.25, "learning_rate": 4.090541777748466e-06, "loss": 1.8298, "step": 4985 }, { "epoch": 0.5717234262125903, "grad_norm": 13.6875, "learning_rate": 4.0887158643756365e-06, "loss": 1.7598, "step": 4986 }, { "epoch": 0.5718380919619309, "grad_norm": 23.75, "learning_rate": 4.086890076697083e-06, "loss": 1.7549, "step": 4987 }, { "epoch": 0.5719527577112716, "grad_norm": 13.5, "learning_rate": 4.0850644149646375e-06, "loss": 1.8044, "step": 4988 }, { "epoch": 0.5720674234606123, "grad_norm": 14.1875, "learning_rate": 4.083238879430117e-06, "loss": 2.2391, "step": 4989 }, { "epoch": 0.572182089209953, "grad_norm": 25.375, "learning_rate": 4.081413470345317e-06, "loss": 1.8844, "step": 4990 }, { "epoch": 0.5722967549592937, "grad_norm": 15.5625, "learning_rate": 4.0795881879620205e-06, "loss": 1.8334, "step": 4991 }, { "epoch": 0.5724114207086344, "grad_norm": 14.5625, "learning_rate": 4.077763032531987e-06, "loss": 1.9067, "step": 4992 }, { "epoch": 0.572526086457975, "grad_norm": 13.25, "learning_rate": 4.0759380043069635e-06, "loss": 1.7666, "step": 4993 }, { "epoch": 0.5726407522073157, "grad_norm": 12.75, "learning_rate": 4.074113103538679e-06, "loss": 1.7937, "step": 4994 }, { "epoch": 0.5727554179566563, "grad_norm": 18.25, "learning_rate": 4.0722883304788444e-06, "loss": 2.076, "step": 4995 }, { "epoch": 0.572870083705997, "grad_norm": 14.25, "learning_rate": 4.0704636853791494e-06, "loss": 1.6905, "step": 4996 }, { "epoch": 0.5729847494553377, "grad_norm": 13.625, "learning_rate": 4.068639168491271e-06, "loss": 1.5226, "step": 4997 }, { "epoch": 0.5730994152046783, "grad_norm": 14.875, "learning_rate": 4.0668147800668635e-06, "loss": 1.6906, "step": 4998 }, { "epoch": 0.573214080954019, "grad_norm": 14.5, "learning_rate": 4.06499052035757e-06, "loss": 1.6505, "step": 4999 }, { "epoch": 0.5733287467033598, "grad_norm": 13.6875, "learning_rate": 4.063166389615013e-06, "loss": 1.7491, "step": 5000 }, { "epoch": 0.5734434124527004, "grad_norm": 15.0625, "learning_rate": 4.061342388090794e-06, "loss": 1.7983, "step": 5001 }, { "epoch": 0.5735580782020411, "grad_norm": 16.25, "learning_rate": 4.0595185160365e-06, "loss": 1.7143, "step": 5002 }, { "epoch": 0.5736727439513817, "grad_norm": 16.375, "learning_rate": 4.057694773703699e-06, "loss": 1.8646, "step": 5003 }, { "epoch": 0.5737874097007224, "grad_norm": 14.6875, "learning_rate": 4.055871161343943e-06, "loss": 2.0058, "step": 5004 }, { "epoch": 0.5739020754500631, "grad_norm": 13.875, "learning_rate": 4.054047679208763e-06, "loss": 1.676, "step": 5005 }, { "epoch": 0.5740167411994037, "grad_norm": 11.75, "learning_rate": 4.052224327549674e-06, "loss": 1.8777, "step": 5006 }, { "epoch": 0.5741314069487444, "grad_norm": 14.3125, "learning_rate": 4.050401106618175e-06, "loss": 1.6619, "step": 5007 }, { "epoch": 0.574246072698085, "grad_norm": 14.8125, "learning_rate": 4.048578016665739e-06, "loss": 1.8373, "step": 5008 }, { "epoch": 0.5743607384474257, "grad_norm": 13.75, "learning_rate": 4.04675505794383e-06, "loss": 1.7221, "step": 5009 }, { "epoch": 0.5744754041967665, "grad_norm": 18.625, "learning_rate": 4.044932230703892e-06, "loss": 2.0445, "step": 5010 }, { "epoch": 0.5745900699461071, "grad_norm": 13.8125, "learning_rate": 4.043109535197347e-06, "loss": 2.1985, "step": 5011 }, { "epoch": 0.5747047356954478, "grad_norm": 13.1875, "learning_rate": 4.041286971675602e-06, "loss": 1.836, "step": 5012 }, { "epoch": 0.5748194014447885, "grad_norm": 26.375, "learning_rate": 4.039464540390043e-06, "loss": 1.7893, "step": 5013 }, { "epoch": 0.5749340671941291, "grad_norm": 15.625, "learning_rate": 4.037642241592043e-06, "loss": 1.5837, "step": 5014 }, { "epoch": 0.5750487329434698, "grad_norm": 17.0, "learning_rate": 4.035820075532951e-06, "loss": 1.9575, "step": 5015 }, { "epoch": 0.5751633986928104, "grad_norm": 13.8125, "learning_rate": 4.033998042464101e-06, "loss": 1.9248, "step": 5016 }, { "epoch": 0.5752780644421511, "grad_norm": 14.875, "learning_rate": 4.032176142636806e-06, "loss": 1.5989, "step": 5017 }, { "epoch": 0.5753927301914918, "grad_norm": 15.6875, "learning_rate": 4.030354376302364e-06, "loss": 1.9005, "step": 5018 }, { "epoch": 0.5755073959408324, "grad_norm": 16.625, "learning_rate": 4.0285327437120535e-06, "loss": 1.9479, "step": 5019 }, { "epoch": 0.5756220616901732, "grad_norm": 13.1875, "learning_rate": 4.0267112451171344e-06, "loss": 1.7823, "step": 5020 }, { "epoch": 0.5757367274395138, "grad_norm": 12.375, "learning_rate": 4.024889880768844e-06, "loss": 1.6997, "step": 5021 }, { "epoch": 0.5758513931888545, "grad_norm": 15.5, "learning_rate": 4.023068650918409e-06, "loss": 1.8029, "step": 5022 }, { "epoch": 0.5759660589381952, "grad_norm": 12.8125, "learning_rate": 4.02124755581703e-06, "loss": 1.4752, "step": 5023 }, { "epoch": 0.5760807246875358, "grad_norm": 16.25, "learning_rate": 4.019426595715894e-06, "loss": 1.9863, "step": 5024 }, { "epoch": 0.5761953904368765, "grad_norm": 15.4375, "learning_rate": 4.017605770866169e-06, "loss": 1.8825, "step": 5025 }, { "epoch": 0.5763100561862172, "grad_norm": 12.375, "learning_rate": 4.015785081519003e-06, "loss": 1.6498, "step": 5026 }, { "epoch": 0.5764247219355578, "grad_norm": 14.4375, "learning_rate": 4.013964527925521e-06, "loss": 1.6481, "step": 5027 }, { "epoch": 0.5765393876848985, "grad_norm": 13.875, "learning_rate": 4.012144110336837e-06, "loss": 1.6053, "step": 5028 }, { "epoch": 0.5766540534342391, "grad_norm": 17.375, "learning_rate": 4.010323829004045e-06, "loss": 1.7952, "step": 5029 }, { "epoch": 0.5767687191835799, "grad_norm": 13.75, "learning_rate": 4.008503684178214e-06, "loss": 1.6152, "step": 5030 }, { "epoch": 0.5768833849329206, "grad_norm": 15.125, "learning_rate": 4.006683676110402e-06, "loss": 1.8167, "step": 5031 }, { "epoch": 0.5769980506822612, "grad_norm": 14.8125, "learning_rate": 4.004863805051642e-06, "loss": 1.9247, "step": 5032 }, { "epoch": 0.5771127164316019, "grad_norm": 15.6875, "learning_rate": 4.003044071252949e-06, "loss": 2.0768, "step": 5033 }, { "epoch": 0.5772273821809426, "grad_norm": 16.875, "learning_rate": 4.001224474965325e-06, "loss": 1.6128, "step": 5034 }, { "epoch": 0.5773420479302832, "grad_norm": 16.375, "learning_rate": 3.999405016439745e-06, "loss": 1.957, "step": 5035 }, { "epoch": 0.5774567136796239, "grad_norm": 13.1875, "learning_rate": 3.997585695927171e-06, "loss": 1.7582, "step": 5036 }, { "epoch": 0.5775713794289645, "grad_norm": 12.9375, "learning_rate": 3.995766513678543e-06, "loss": 1.5341, "step": 5037 }, { "epoch": 0.5776860451783052, "grad_norm": 15.0625, "learning_rate": 3.99394746994478e-06, "loss": 2.0923, "step": 5038 }, { "epoch": 0.577800710927646, "grad_norm": 17.0, "learning_rate": 3.9921285649767875e-06, "loss": 1.6296, "step": 5039 }, { "epoch": 0.5779153766769866, "grad_norm": 15.1875, "learning_rate": 3.990309799025447e-06, "loss": 1.6395, "step": 5040 }, { "epoch": 0.5780300424263273, "grad_norm": 15.4375, "learning_rate": 3.988491172341625e-06, "loss": 1.9226, "step": 5041 }, { "epoch": 0.5781447081756679, "grad_norm": 14.4375, "learning_rate": 3.986672685176163e-06, "loss": 1.7493, "step": 5042 }, { "epoch": 0.5782593739250086, "grad_norm": 16.25, "learning_rate": 3.984854337779887e-06, "loss": 1.5625, "step": 5043 }, { "epoch": 0.5783740396743493, "grad_norm": 15.25, "learning_rate": 3.983036130403607e-06, "loss": 1.8524, "step": 5044 }, { "epoch": 0.5784887054236899, "grad_norm": 14.75, "learning_rate": 3.981218063298108e-06, "loss": 1.7605, "step": 5045 }, { "epoch": 0.5786033711730306, "grad_norm": 15.0625, "learning_rate": 3.979400136714156e-06, "loss": 1.7131, "step": 5046 }, { "epoch": 0.5787180369223713, "grad_norm": 14.625, "learning_rate": 3.977582350902502e-06, "loss": 1.7204, "step": 5047 }, { "epoch": 0.5788327026717119, "grad_norm": 15.0, "learning_rate": 3.975764706113872e-06, "loss": 1.7486, "step": 5048 }, { "epoch": 0.5789473684210527, "grad_norm": 12.0, "learning_rate": 3.973947202598978e-06, "loss": 1.523, "step": 5049 }, { "epoch": 0.5790620341703933, "grad_norm": 12.6875, "learning_rate": 3.97212984060851e-06, "loss": 1.9203, "step": 5050 }, { "epoch": 0.579176699919734, "grad_norm": 14.1875, "learning_rate": 3.970312620393137e-06, "loss": 1.8934, "step": 5051 }, { "epoch": 0.5792913656690747, "grad_norm": 14.8125, "learning_rate": 3.96849554220351e-06, "loss": 1.6989, "step": 5052 }, { "epoch": 0.5794060314184153, "grad_norm": 35.25, "learning_rate": 3.966678606290259e-06, "loss": 2.2106, "step": 5053 }, { "epoch": 0.579520697167756, "grad_norm": 14.8125, "learning_rate": 3.964861812904e-06, "loss": 1.684, "step": 5054 }, { "epoch": 0.5796353629170966, "grad_norm": 12.4375, "learning_rate": 3.963045162295321e-06, "loss": 1.7207, "step": 5055 }, { "epoch": 0.5797500286664373, "grad_norm": 13.0, "learning_rate": 3.961228654714797e-06, "loss": 1.5263, "step": 5056 }, { "epoch": 0.579864694415778, "grad_norm": 14.5625, "learning_rate": 3.9594122904129794e-06, "loss": 1.7074, "step": 5057 }, { "epoch": 0.5799793601651186, "grad_norm": 13.875, "learning_rate": 3.9575960696403985e-06, "loss": 1.4753, "step": 5058 }, { "epoch": 0.5800940259144594, "grad_norm": 11.0625, "learning_rate": 3.955779992647571e-06, "loss": 1.4784, "step": 5059 }, { "epoch": 0.5802086916638001, "grad_norm": 16.875, "learning_rate": 3.9539640596849895e-06, "loss": 1.4501, "step": 5060 }, { "epoch": 0.5803233574131407, "grad_norm": 15.3125, "learning_rate": 3.9521482710031265e-06, "loss": 1.4337, "step": 5061 }, { "epoch": 0.5804380231624814, "grad_norm": 15.8125, "learning_rate": 3.950332626852436e-06, "loss": 1.7382, "step": 5062 }, { "epoch": 0.580552688911822, "grad_norm": 16.375, "learning_rate": 3.948517127483349e-06, "loss": 2.138, "step": 5063 }, { "epoch": 0.5806673546611627, "grad_norm": 14.375, "learning_rate": 3.946701773146284e-06, "loss": 1.8435, "step": 5064 }, { "epoch": 0.5807820204105034, "grad_norm": 17.0, "learning_rate": 3.9448865640916296e-06, "loss": 1.9225, "step": 5065 }, { "epoch": 0.580896686159844, "grad_norm": 15.3125, "learning_rate": 3.943071500569763e-06, "loss": 1.7692, "step": 5066 }, { "epoch": 0.5810113519091847, "grad_norm": 14.625, "learning_rate": 3.941256582831035e-06, "loss": 1.788, "step": 5067 }, { "epoch": 0.5811260176585255, "grad_norm": 21.0, "learning_rate": 3.9394418111257785e-06, "loss": 2.2795, "step": 5068 }, { "epoch": 0.581240683407866, "grad_norm": 17.375, "learning_rate": 3.9376271857043105e-06, "loss": 1.8825, "step": 5069 }, { "epoch": 0.5813553491572068, "grad_norm": 13.375, "learning_rate": 3.9358127068169216e-06, "loss": 1.7046, "step": 5070 }, { "epoch": 0.5814700149065474, "grad_norm": 15.5625, "learning_rate": 3.933998374713884e-06, "loss": 1.8224, "step": 5071 }, { "epoch": 0.5815846806558881, "grad_norm": 12.375, "learning_rate": 3.932184189645451e-06, "loss": 1.5967, "step": 5072 }, { "epoch": 0.5816993464052288, "grad_norm": 15.25, "learning_rate": 3.930370151861853e-06, "loss": 1.5899, "step": 5073 }, { "epoch": 0.5818140121545694, "grad_norm": 12.0625, "learning_rate": 3.928556261613305e-06, "loss": 1.6608, "step": 5074 }, { "epoch": 0.5819286779039101, "grad_norm": 13.5, "learning_rate": 3.926742519149997e-06, "loss": 1.4369, "step": 5075 }, { "epoch": 0.5820433436532507, "grad_norm": 12.1875, "learning_rate": 3.9249289247221005e-06, "loss": 1.5315, "step": 5076 }, { "epoch": 0.5821580094025914, "grad_norm": 14.8125, "learning_rate": 3.9231154785797655e-06, "loss": 1.6205, "step": 5077 }, { "epoch": 0.5822726751519322, "grad_norm": 16.375, "learning_rate": 3.921302180973122e-06, "loss": 1.8186, "step": 5078 }, { "epoch": 0.5823873409012728, "grad_norm": 15.3125, "learning_rate": 3.9194890321522825e-06, "loss": 1.8972, "step": 5079 }, { "epoch": 0.5825020066506135, "grad_norm": 15.5, "learning_rate": 3.917676032367334e-06, "loss": 2.0401, "step": 5080 }, { "epoch": 0.5826166723999542, "grad_norm": 14.1875, "learning_rate": 3.915863181868347e-06, "loss": 1.8215, "step": 5081 }, { "epoch": 0.5827313381492948, "grad_norm": 13.5, "learning_rate": 3.914050480905368e-06, "loss": 2.4745, "step": 5082 }, { "epoch": 0.5828460038986355, "grad_norm": 16.625, "learning_rate": 3.912237929728423e-06, "loss": 1.5937, "step": 5083 }, { "epoch": 0.5829606696479761, "grad_norm": 16.25, "learning_rate": 3.9104255285875225e-06, "loss": 1.906, "step": 5084 }, { "epoch": 0.5830753353973168, "grad_norm": 15.1875, "learning_rate": 3.908613277732652e-06, "loss": 1.8189, "step": 5085 }, { "epoch": 0.5831900011466575, "grad_norm": 17.375, "learning_rate": 3.906801177413775e-06, "loss": 1.8351, "step": 5086 }, { "epoch": 0.5833046668959981, "grad_norm": 13.0625, "learning_rate": 3.904989227880838e-06, "loss": 1.749, "step": 5087 }, { "epoch": 0.5834193326453389, "grad_norm": 14.8125, "learning_rate": 3.903177429383763e-06, "loss": 1.6529, "step": 5088 }, { "epoch": 0.5835339983946795, "grad_norm": 16.375, "learning_rate": 3.901365782172456e-06, "loss": 2.1445, "step": 5089 }, { "epoch": 0.5836486641440202, "grad_norm": 12.8125, "learning_rate": 3.899554286496796e-06, "loss": 1.8122, "step": 5090 }, { "epoch": 0.5837633298933609, "grad_norm": 14.0, "learning_rate": 3.897742942606647e-06, "loss": 1.7224, "step": 5091 }, { "epoch": 0.5838779956427015, "grad_norm": 16.875, "learning_rate": 3.895931750751847e-06, "loss": 2.034, "step": 5092 }, { "epoch": 0.5839926613920422, "grad_norm": 12.875, "learning_rate": 3.894120711182216e-06, "loss": 1.8437, "step": 5093 }, { "epoch": 0.5841073271413829, "grad_norm": 14.0, "learning_rate": 3.892309824147554e-06, "loss": 1.9002, "step": 5094 }, { "epoch": 0.5842219928907235, "grad_norm": 13.3125, "learning_rate": 3.890499089897639e-06, "loss": 1.8981, "step": 5095 }, { "epoch": 0.5843366586400642, "grad_norm": 15.6875, "learning_rate": 3.888688508682223e-06, "loss": 1.8748, "step": 5096 }, { "epoch": 0.5844513243894048, "grad_norm": 16.125, "learning_rate": 3.886878080751046e-06, "loss": 1.7479, "step": 5097 }, { "epoch": 0.5845659901387456, "grad_norm": 15.3125, "learning_rate": 3.8850678063538164e-06, "loss": 1.7775, "step": 5098 }, { "epoch": 0.5846806558880863, "grad_norm": 17.25, "learning_rate": 3.883257685740232e-06, "loss": 1.5282, "step": 5099 }, { "epoch": 0.5847953216374269, "grad_norm": 16.25, "learning_rate": 3.881447719159963e-06, "loss": 1.7246, "step": 5100 }, { "epoch": 0.5849099873867676, "grad_norm": 12.1875, "learning_rate": 3.879637906862661e-06, "loss": 1.9079, "step": 5101 }, { "epoch": 0.5850246531361083, "grad_norm": 13.125, "learning_rate": 3.8778282490979516e-06, "loss": 1.7292, "step": 5102 }, { "epoch": 0.5851393188854489, "grad_norm": 16.375, "learning_rate": 3.876018746115444e-06, "loss": 1.6749, "step": 5103 }, { "epoch": 0.5852539846347896, "grad_norm": 20.875, "learning_rate": 3.874209398164727e-06, "loss": 2.1107, "step": 5104 }, { "epoch": 0.5853686503841302, "grad_norm": 16.25, "learning_rate": 3.872400205495364e-06, "loss": 1.8244, "step": 5105 }, { "epoch": 0.5854833161334709, "grad_norm": 14.125, "learning_rate": 3.870591168356899e-06, "loss": 2.1071, "step": 5106 }, { "epoch": 0.5855979818828116, "grad_norm": 14.625, "learning_rate": 3.868782286998853e-06, "loss": 1.4535, "step": 5107 }, { "epoch": 0.5857126476321523, "grad_norm": 15.125, "learning_rate": 3.866973561670727e-06, "loss": 1.5733, "step": 5108 }, { "epoch": 0.585827313381493, "grad_norm": 13.1875, "learning_rate": 3.865164992622001e-06, "loss": 1.6797, "step": 5109 }, { "epoch": 0.5859419791308336, "grad_norm": 17.5, "learning_rate": 3.863356580102133e-06, "loss": 1.8771, "step": 5110 }, { "epoch": 0.5860566448801743, "grad_norm": 14.5, "learning_rate": 3.861548324360558e-06, "loss": 2.0103, "step": 5111 }, { "epoch": 0.586171310629515, "grad_norm": 14.625, "learning_rate": 3.859740225646692e-06, "loss": 1.7533, "step": 5112 }, { "epoch": 0.5862859763788556, "grad_norm": 14.4375, "learning_rate": 3.857932284209924e-06, "loss": 1.9665, "step": 5113 }, { "epoch": 0.5864006421281963, "grad_norm": 16.75, "learning_rate": 3.85612450029963e-06, "loss": 1.6362, "step": 5114 }, { "epoch": 0.586515307877537, "grad_norm": 12.875, "learning_rate": 3.854316874165156e-06, "loss": 1.8815, "step": 5115 }, { "epoch": 0.5866299736268776, "grad_norm": 17.375, "learning_rate": 3.852509406055831e-06, "loss": 2.0311, "step": 5116 }, { "epoch": 0.5867446393762183, "grad_norm": 15.3125, "learning_rate": 3.85070209622096e-06, "loss": 1.8009, "step": 5117 }, { "epoch": 0.586859305125559, "grad_norm": 19.125, "learning_rate": 3.848894944909825e-06, "loss": 2.3575, "step": 5118 }, { "epoch": 0.5869739708748997, "grad_norm": 15.0625, "learning_rate": 3.847087952371693e-06, "loss": 2.1153, "step": 5119 }, { "epoch": 0.5870886366242404, "grad_norm": 58.25, "learning_rate": 3.845281118855801e-06, "loss": 1.7975, "step": 5120 }, { "epoch": 0.587203302373581, "grad_norm": 14.0625, "learning_rate": 3.843474444611368e-06, "loss": 1.6647, "step": 5121 }, { "epoch": 0.5873179681229217, "grad_norm": 15.375, "learning_rate": 3.841667929887591e-06, "loss": 1.596, "step": 5122 }, { "epoch": 0.5874326338722623, "grad_norm": 14.625, "learning_rate": 3.8398615749336395e-06, "loss": 1.4904, "step": 5123 }, { "epoch": 0.587547299621603, "grad_norm": 17.125, "learning_rate": 3.838055379998671e-06, "loss": 1.5446, "step": 5124 }, { "epoch": 0.5876619653709437, "grad_norm": 12.125, "learning_rate": 3.836249345331816e-06, "loss": 1.5349, "step": 5125 }, { "epoch": 0.5877766311202843, "grad_norm": 14.0, "learning_rate": 3.834443471182181e-06, "loss": 1.7381, "step": 5126 }, { "epoch": 0.587891296869625, "grad_norm": 14.0, "learning_rate": 3.8326377577988495e-06, "loss": 1.7844, "step": 5127 }, { "epoch": 0.5880059626189658, "grad_norm": 16.125, "learning_rate": 3.830832205430887e-06, "loss": 1.8643, "step": 5128 }, { "epoch": 0.5881206283683064, "grad_norm": 11.75, "learning_rate": 3.829026814327337e-06, "loss": 1.5845, "step": 5129 }, { "epoch": 0.5882352941176471, "grad_norm": 12.0, "learning_rate": 3.827221584737216e-06, "loss": 1.541, "step": 5130 }, { "epoch": 0.5883499598669877, "grad_norm": 32.5, "learning_rate": 3.825416516909524e-06, "loss": 1.9679, "step": 5131 }, { "epoch": 0.5884646256163284, "grad_norm": 13.4375, "learning_rate": 3.823611611093233e-06, "loss": 1.6653, "step": 5132 }, { "epoch": 0.5885792913656691, "grad_norm": 12.75, "learning_rate": 3.821806867537294e-06, "loss": 1.5283, "step": 5133 }, { "epoch": 0.5886939571150097, "grad_norm": 18.375, "learning_rate": 3.820002286490642e-06, "loss": 2.1615, "step": 5134 }, { "epoch": 0.5888086228643504, "grad_norm": 14.4375, "learning_rate": 3.8181978682021804e-06, "loss": 2.2717, "step": 5135 }, { "epoch": 0.5889232886136911, "grad_norm": 16.125, "learning_rate": 3.8163936129207965e-06, "loss": 1.902, "step": 5136 }, { "epoch": 0.5890379543630317, "grad_norm": 13.0, "learning_rate": 3.8145895208953528e-06, "loss": 1.7033, "step": 5137 }, { "epoch": 0.5891526201123725, "grad_norm": 14.8125, "learning_rate": 3.812785592374686e-06, "loss": 1.8852, "step": 5138 }, { "epoch": 0.5892672858617131, "grad_norm": 14.6875, "learning_rate": 3.8109818276076194e-06, "loss": 1.7235, "step": 5139 }, { "epoch": 0.5893819516110538, "grad_norm": 14.3125, "learning_rate": 3.8091782268429434e-06, "loss": 1.9763, "step": 5140 }, { "epoch": 0.5894966173603945, "grad_norm": 24.75, "learning_rate": 3.8073747903294323e-06, "loss": 1.4435, "step": 5141 }, { "epoch": 0.5896112831097351, "grad_norm": 11.625, "learning_rate": 3.8055715183158356e-06, "loss": 1.7353, "step": 5142 }, { "epoch": 0.5897259488590758, "grad_norm": 19.125, "learning_rate": 3.8037684110508776e-06, "loss": 1.6724, "step": 5143 }, { "epoch": 0.5898406146084164, "grad_norm": 12.6875, "learning_rate": 3.801965468783268e-06, "loss": 1.399, "step": 5144 }, { "epoch": 0.5899552803577571, "grad_norm": 14.625, "learning_rate": 3.800162691761685e-06, "loss": 1.5663, "step": 5145 }, { "epoch": 0.5900699461070978, "grad_norm": 13.625, "learning_rate": 3.7983600802347873e-06, "loss": 1.4547, "step": 5146 }, { "epoch": 0.5901846118564384, "grad_norm": 12.125, "learning_rate": 3.7965576344512115e-06, "loss": 2.0121, "step": 5147 }, { "epoch": 0.5902992776057792, "grad_norm": 17.625, "learning_rate": 3.794755354659569e-06, "loss": 1.884, "step": 5148 }, { "epoch": 0.5904139433551199, "grad_norm": 30.875, "learning_rate": 3.792953241108452e-06, "loss": 1.6387, "step": 5149 }, { "epoch": 0.5905286091044605, "grad_norm": 15.5625, "learning_rate": 3.791151294046428e-06, "loss": 2.1112, "step": 5150 }, { "epoch": 0.5906432748538012, "grad_norm": 13.5, "learning_rate": 3.78934951372204e-06, "loss": 1.7029, "step": 5151 }, { "epoch": 0.5907579406031418, "grad_norm": 15.0625, "learning_rate": 3.7875479003838083e-06, "loss": 2.0332, "step": 5152 }, { "epoch": 0.5908726063524825, "grad_norm": 11.5625, "learning_rate": 3.7857464542802313e-06, "loss": 1.6364, "step": 5153 }, { "epoch": 0.5909872721018232, "grad_norm": 12.75, "learning_rate": 3.783945175659788e-06, "loss": 1.901, "step": 5154 }, { "epoch": 0.5911019378511638, "grad_norm": 14.4375, "learning_rate": 3.7821440647709257e-06, "loss": 1.685, "step": 5155 }, { "epoch": 0.5912166036005045, "grad_norm": 27.5, "learning_rate": 3.7803431218620768e-06, "loss": 1.6155, "step": 5156 }, { "epoch": 0.5913312693498453, "grad_norm": 14.375, "learning_rate": 3.7785423471816453e-06, "loss": 1.4173, "step": 5157 }, { "epoch": 0.5914459350991859, "grad_norm": 15.625, "learning_rate": 3.7767417409780118e-06, "loss": 1.7574, "step": 5158 }, { "epoch": 0.5915606008485266, "grad_norm": 18.875, "learning_rate": 3.7749413034995386e-06, "loss": 1.6809, "step": 5159 }, { "epoch": 0.5916752665978672, "grad_norm": 14.9375, "learning_rate": 3.773141034994562e-06, "loss": 1.3052, "step": 5160 }, { "epoch": 0.5917899323472079, "grad_norm": 13.375, "learning_rate": 3.771340935711392e-06, "loss": 1.6808, "step": 5161 }, { "epoch": 0.5919045980965486, "grad_norm": 11.75, "learning_rate": 3.7695410058983217e-06, "loss": 1.7823, "step": 5162 }, { "epoch": 0.5920192638458892, "grad_norm": 12.8125, "learning_rate": 3.7677412458036123e-06, "loss": 1.546, "step": 5163 }, { "epoch": 0.5921339295952299, "grad_norm": 15.3125, "learning_rate": 3.765941655675511e-06, "loss": 2.4057, "step": 5164 }, { "epoch": 0.5922485953445705, "grad_norm": 13.9375, "learning_rate": 3.7641422357622345e-06, "loss": 2.0024, "step": 5165 }, { "epoch": 0.5923632610939112, "grad_norm": 13.1875, "learning_rate": 3.7623429863119794e-06, "loss": 1.9267, "step": 5166 }, { "epoch": 0.592477926843252, "grad_norm": 15.0, "learning_rate": 3.7605439075729165e-06, "loss": 1.7092, "step": 5167 }, { "epoch": 0.5925925925925926, "grad_norm": 14.5625, "learning_rate": 3.7587449997931947e-06, "loss": 1.7054, "step": 5168 }, { "epoch": 0.5927072583419333, "grad_norm": 13.9375, "learning_rate": 3.7569462632209415e-06, "loss": 1.9171, "step": 5169 }, { "epoch": 0.592821924091274, "grad_norm": 21.625, "learning_rate": 3.755147698104256e-06, "loss": 2.4774, "step": 5170 }, { "epoch": 0.5929365898406146, "grad_norm": 16.0, "learning_rate": 3.753349304691216e-06, "loss": 1.9791, "step": 5171 }, { "epoch": 0.5930512555899553, "grad_norm": 16.5, "learning_rate": 3.7515510832298774e-06, "loss": 1.7614, "step": 5172 }, { "epoch": 0.5931659213392959, "grad_norm": 33.5, "learning_rate": 3.749753033968267e-06, "loss": 2.1144, "step": 5173 }, { "epoch": 0.5932805870886366, "grad_norm": 17.125, "learning_rate": 3.747955157154395e-06, "loss": 1.707, "step": 5174 }, { "epoch": 0.5933952528379773, "grad_norm": 14.0, "learning_rate": 3.7461574530362432e-06, "loss": 1.5074, "step": 5175 }, { "epoch": 0.5935099185873179, "grad_norm": 11.75, "learning_rate": 3.744359921861771e-06, "loss": 1.6424, "step": 5176 }, { "epoch": 0.5936245843366587, "grad_norm": 14.125, "learning_rate": 3.7425625638789117e-06, "loss": 1.9301, "step": 5177 }, { "epoch": 0.5937392500859993, "grad_norm": 12.8125, "learning_rate": 3.740765379335577e-06, "loss": 1.4849, "step": 5178 }, { "epoch": 0.59385391583534, "grad_norm": 14.0, "learning_rate": 3.7389683684796563e-06, "loss": 2.1043, "step": 5179 }, { "epoch": 0.5939685815846807, "grad_norm": 15.25, "learning_rate": 3.7371715315590114e-06, "loss": 1.8663, "step": 5180 }, { "epoch": 0.5940832473340213, "grad_norm": 16.125, "learning_rate": 3.735374868821483e-06, "loss": 2.1982, "step": 5181 }, { "epoch": 0.594197913083362, "grad_norm": 13.0, "learning_rate": 3.733578380514885e-06, "loss": 1.6134, "step": 5182 }, { "epoch": 0.5943125788327027, "grad_norm": 13.1875, "learning_rate": 3.7317820668870073e-06, "loss": 1.6173, "step": 5183 }, { "epoch": 0.5944272445820433, "grad_norm": 15.5, "learning_rate": 3.7299859281856206e-06, "loss": 1.6519, "step": 5184 }, { "epoch": 0.594541910331384, "grad_norm": 14.6875, "learning_rate": 3.7281899646584666e-06, "loss": 1.8923, "step": 5185 }, { "epoch": 0.5946565760807246, "grad_norm": 16.5, "learning_rate": 3.726394176553264e-06, "loss": 1.73, "step": 5186 }, { "epoch": 0.5947712418300654, "grad_norm": 13.0625, "learning_rate": 3.7245985641177085e-06, "loss": 1.8046, "step": 5187 }, { "epoch": 0.5948859075794061, "grad_norm": 16.125, "learning_rate": 3.722803127599467e-06, "loss": 2.2335, "step": 5188 }, { "epoch": 0.5950005733287467, "grad_norm": 15.5, "learning_rate": 3.721007867246191e-06, "loss": 1.6555, "step": 5189 }, { "epoch": 0.5951152390780874, "grad_norm": 15.3125, "learning_rate": 3.7192127833054987e-06, "loss": 2.2423, "step": 5190 }, { "epoch": 0.5952299048274281, "grad_norm": 14.875, "learning_rate": 3.7174178760249902e-06, "loss": 1.7237, "step": 5191 }, { "epoch": 0.5953445705767687, "grad_norm": 20.25, "learning_rate": 3.7156231456522357e-06, "loss": 1.6857, "step": 5192 }, { "epoch": 0.5954592363261094, "grad_norm": 13.375, "learning_rate": 3.713828592434785e-06, "loss": 1.6374, "step": 5193 }, { "epoch": 0.59557390207545, "grad_norm": 17.5, "learning_rate": 3.712034216620166e-06, "loss": 1.8097, "step": 5194 }, { "epoch": 0.5956885678247907, "grad_norm": 15.75, "learning_rate": 3.710240018455876e-06, "loss": 2.2043, "step": 5195 }, { "epoch": 0.5958032335741315, "grad_norm": 17.125, "learning_rate": 3.7084459981893882e-06, "loss": 1.6765, "step": 5196 }, { "epoch": 0.5959178993234721, "grad_norm": 15.0, "learning_rate": 3.706652156068158e-06, "loss": 1.9472, "step": 5197 }, { "epoch": 0.5960325650728128, "grad_norm": 14.0625, "learning_rate": 3.704858492339607e-06, "loss": 1.8542, "step": 5198 }, { "epoch": 0.5961472308221534, "grad_norm": 15.25, "learning_rate": 3.7030650072511397e-06, "loss": 1.8598, "step": 5199 }, { "epoch": 0.5962618965714941, "grad_norm": 13.5625, "learning_rate": 3.701271701050135e-06, "loss": 1.486, "step": 5200 }, { "epoch": 0.5963765623208348, "grad_norm": 16.0, "learning_rate": 3.6994785739839423e-06, "loss": 1.924, "step": 5201 }, { "epoch": 0.5964912280701754, "grad_norm": 14.125, "learning_rate": 3.6976856262998894e-06, "loss": 1.7448, "step": 5202 }, { "epoch": 0.5966058938195161, "grad_norm": 12.125, "learning_rate": 3.695892858245279e-06, "loss": 1.3856, "step": 5203 }, { "epoch": 0.5967205595688568, "grad_norm": 14.875, "learning_rate": 3.6941002700673923e-06, "loss": 1.2914, "step": 5204 }, { "epoch": 0.5968352253181974, "grad_norm": 16.375, "learning_rate": 3.692307862013479e-06, "loss": 2.1275, "step": 5205 }, { "epoch": 0.5969498910675382, "grad_norm": 16.125, "learning_rate": 3.6905156343307713e-06, "loss": 1.7309, "step": 5206 }, { "epoch": 0.5970645568168788, "grad_norm": 14.0, "learning_rate": 3.6887235872664703e-06, "loss": 1.2839, "step": 5207 }, { "epoch": 0.5971792225662195, "grad_norm": 13.1875, "learning_rate": 3.6869317210677525e-06, "loss": 1.5556, "step": 5208 }, { "epoch": 0.5972938883155602, "grad_norm": 12.75, "learning_rate": 3.685140035981776e-06, "loss": 1.8777, "step": 5209 }, { "epoch": 0.5974085540649008, "grad_norm": 19.375, "learning_rate": 3.6833485322556695e-06, "loss": 1.9478, "step": 5210 }, { "epoch": 0.5975232198142415, "grad_norm": 11.6875, "learning_rate": 3.6815572101365337e-06, "loss": 1.515, "step": 5211 }, { "epoch": 0.5976378855635821, "grad_norm": 13.4375, "learning_rate": 3.6797660698714503e-06, "loss": 1.6556, "step": 5212 }, { "epoch": 0.5977525513129228, "grad_norm": 15.0, "learning_rate": 3.6779751117074697e-06, "loss": 1.8579, "step": 5213 }, { "epoch": 0.5978672170622635, "grad_norm": 16.0, "learning_rate": 3.676184335891624e-06, "loss": 1.5251, "step": 5214 }, { "epoch": 0.5979818828116041, "grad_norm": 15.3125, "learning_rate": 3.6743937426709143e-06, "loss": 1.7899, "step": 5215 }, { "epoch": 0.5980965485609449, "grad_norm": 13.6875, "learning_rate": 3.6726033322923204e-06, "loss": 2.0687, "step": 5216 }, { "epoch": 0.5982112143102856, "grad_norm": 15.625, "learning_rate": 3.6708131050027927e-06, "loss": 1.6074, "step": 5217 }, { "epoch": 0.5983258800596262, "grad_norm": 16.25, "learning_rate": 3.6690230610492605e-06, "loss": 1.606, "step": 5218 }, { "epoch": 0.5984405458089669, "grad_norm": 13.5625, "learning_rate": 3.6672332006786275e-06, "loss": 2.0606, "step": 5219 }, { "epoch": 0.5985552115583075, "grad_norm": 14.25, "learning_rate": 3.6654435241377708e-06, "loss": 1.9893, "step": 5220 }, { "epoch": 0.5986698773076482, "grad_norm": 19.25, "learning_rate": 3.6636540316735385e-06, "loss": 1.9915, "step": 5221 }, { "epoch": 0.5987845430569889, "grad_norm": 38.5, "learning_rate": 3.661864723532761e-06, "loss": 1.8068, "step": 5222 }, { "epoch": 0.5988992088063295, "grad_norm": 12.8125, "learning_rate": 3.6600755999622362e-06, "loss": 1.7801, "step": 5223 }, { "epoch": 0.5990138745556702, "grad_norm": 13.6875, "learning_rate": 3.6582866612087415e-06, "loss": 2.0157, "step": 5224 }, { "epoch": 0.599128540305011, "grad_norm": 14.375, "learning_rate": 3.6564979075190277e-06, "loss": 1.7292, "step": 5225 }, { "epoch": 0.5992432060543516, "grad_norm": 34.75, "learning_rate": 3.654709339139818e-06, "loss": 1.8594, "step": 5226 }, { "epoch": 0.5993578718036923, "grad_norm": 16.5, "learning_rate": 3.65292095631781e-06, "loss": 2.0294, "step": 5227 }, { "epoch": 0.5994725375530329, "grad_norm": 13.5, "learning_rate": 3.651132759299678e-06, "loss": 2.2565, "step": 5228 }, { "epoch": 0.5995872033023736, "grad_norm": 13.125, "learning_rate": 3.649344748332072e-06, "loss": 1.6611, "step": 5229 }, { "epoch": 0.5997018690517143, "grad_norm": 11.6875, "learning_rate": 3.6475569236616117e-06, "loss": 1.4833, "step": 5230 }, { "epoch": 0.5998165348010549, "grad_norm": 16.0, "learning_rate": 3.6457692855348943e-06, "loss": 1.8121, "step": 5231 }, { "epoch": 0.5999312005503956, "grad_norm": 20.125, "learning_rate": 3.64398183419849e-06, "loss": 1.8871, "step": 5232 }, { "epoch": 0.6000458662997362, "grad_norm": 12.6875, "learning_rate": 3.6421945698989414e-06, "loss": 1.6151, "step": 5233 }, { "epoch": 0.6001605320490769, "grad_norm": 16.75, "learning_rate": 3.640407492882771e-06, "loss": 1.3732, "step": 5234 }, { "epoch": 0.6002751977984176, "grad_norm": 17.625, "learning_rate": 3.638620603396472e-06, "loss": 1.92, "step": 5235 }, { "epoch": 0.6003898635477583, "grad_norm": 12.75, "learning_rate": 3.636833901686509e-06, "loss": 1.6098, "step": 5236 }, { "epoch": 0.600504529297099, "grad_norm": 14.0, "learning_rate": 3.6350473879993265e-06, "loss": 1.9172, "step": 5237 }, { "epoch": 0.6006191950464397, "grad_norm": 15.9375, "learning_rate": 3.6332610625813357e-06, "loss": 1.9661, "step": 5238 }, { "epoch": 0.6007338607957803, "grad_norm": 14.4375, "learning_rate": 3.6314749256789315e-06, "loss": 1.8895, "step": 5239 }, { "epoch": 0.600848526545121, "grad_norm": 15.75, "learning_rate": 3.6296889775384735e-06, "loss": 1.7292, "step": 5240 }, { "epoch": 0.6009631922944616, "grad_norm": 18.5, "learning_rate": 3.6279032184063023e-06, "loss": 1.4088, "step": 5241 }, { "epoch": 0.6010778580438023, "grad_norm": 12.5625, "learning_rate": 3.626117648528726e-06, "loss": 1.7245, "step": 5242 }, { "epoch": 0.601192523793143, "grad_norm": 30.125, "learning_rate": 3.62433226815203e-06, "loss": 1.8949, "step": 5243 }, { "epoch": 0.6013071895424836, "grad_norm": 13.5625, "learning_rate": 3.622547077522478e-06, "loss": 1.7503, "step": 5244 }, { "epoch": 0.6014218552918243, "grad_norm": 20.25, "learning_rate": 3.6207620768862995e-06, "loss": 1.5455, "step": 5245 }, { "epoch": 0.601536521041165, "grad_norm": 14.5, "learning_rate": 3.6189772664897006e-06, "loss": 1.6558, "step": 5246 }, { "epoch": 0.6016511867905057, "grad_norm": 15.875, "learning_rate": 3.617192646578864e-06, "loss": 1.9514, "step": 5247 }, { "epoch": 0.6017658525398464, "grad_norm": 29.75, "learning_rate": 3.6154082173999416e-06, "loss": 1.822, "step": 5248 }, { "epoch": 0.601880518289187, "grad_norm": 12.1875, "learning_rate": 3.613623979199064e-06, "loss": 1.4544, "step": 5249 }, { "epoch": 0.6019951840385277, "grad_norm": 13.25, "learning_rate": 3.6118399322223325e-06, "loss": 1.9518, "step": 5250 }, { "epoch": 0.6021098497878684, "grad_norm": 13.25, "learning_rate": 3.6100560767158215e-06, "loss": 1.6783, "step": 5251 }, { "epoch": 0.602224515537209, "grad_norm": 17.5, "learning_rate": 3.6082724129255793e-06, "loss": 1.5505, "step": 5252 }, { "epoch": 0.6023391812865497, "grad_norm": 22.25, "learning_rate": 3.606488941097628e-06, "loss": 2.2827, "step": 5253 }, { "epoch": 0.6024538470358903, "grad_norm": 17.25, "learning_rate": 3.6047056614779664e-06, "loss": 2.1669, "step": 5254 }, { "epoch": 0.602568512785231, "grad_norm": 18.875, "learning_rate": 3.6029225743125604e-06, "loss": 2.2258, "step": 5255 }, { "epoch": 0.6026831785345718, "grad_norm": 16.125, "learning_rate": 3.6011396798473563e-06, "loss": 1.6435, "step": 5256 }, { "epoch": 0.6027978442839124, "grad_norm": 14.3125, "learning_rate": 3.599356978328269e-06, "loss": 1.6817, "step": 5257 }, { "epoch": 0.6029125100332531, "grad_norm": 11.0, "learning_rate": 3.597574470001185e-06, "loss": 1.7506, "step": 5258 }, { "epoch": 0.6030271757825938, "grad_norm": 13.875, "learning_rate": 3.5957921551119714e-06, "loss": 1.6892, "step": 5259 }, { "epoch": 0.6031418415319344, "grad_norm": 17.375, "learning_rate": 3.5940100339064643e-06, "loss": 1.5074, "step": 5260 }, { "epoch": 0.6032565072812751, "grad_norm": 12.8125, "learning_rate": 3.5922281066304722e-06, "loss": 1.2116, "step": 5261 }, { "epoch": 0.6033711730306157, "grad_norm": 15.8125, "learning_rate": 3.5904463735297784e-06, "loss": 1.8858, "step": 5262 }, { "epoch": 0.6034858387799564, "grad_norm": 14.5625, "learning_rate": 3.5886648348501366e-06, "loss": 1.648, "step": 5263 }, { "epoch": 0.6036005045292971, "grad_norm": 15.0, "learning_rate": 3.5868834908372814e-06, "loss": 1.5899, "step": 5264 }, { "epoch": 0.6037151702786377, "grad_norm": 15.1875, "learning_rate": 3.5851023417369103e-06, "loss": 1.4786, "step": 5265 }, { "epoch": 0.6038298360279785, "grad_norm": 12.5625, "learning_rate": 3.5833213877947022e-06, "loss": 1.3949, "step": 5266 }, { "epoch": 0.6039445017773191, "grad_norm": 16.5, "learning_rate": 3.581540629256303e-06, "loss": 1.9696, "step": 5267 }, { "epoch": 0.6040591675266598, "grad_norm": 13.3125, "learning_rate": 3.5797600663673347e-06, "loss": 1.7772, "step": 5268 }, { "epoch": 0.6041738332760005, "grad_norm": 14.625, "learning_rate": 3.577979699373395e-06, "loss": 1.8404, "step": 5269 }, { "epoch": 0.6042884990253411, "grad_norm": 15.0, "learning_rate": 3.57619952852005e-06, "loss": 1.8162, "step": 5270 }, { "epoch": 0.6044031647746818, "grad_norm": 14.8125, "learning_rate": 3.574419554052839e-06, "loss": 1.8376, "step": 5271 }, { "epoch": 0.6045178305240225, "grad_norm": 13.875, "learning_rate": 3.5726397762172773e-06, "loss": 1.873, "step": 5272 }, { "epoch": 0.6046324962733631, "grad_norm": 17.375, "learning_rate": 3.5708601952588483e-06, "loss": 1.7196, "step": 5273 }, { "epoch": 0.6047471620227038, "grad_norm": 16.125, "learning_rate": 3.569080811423015e-06, "loss": 1.5909, "step": 5274 }, { "epoch": 0.6048618277720444, "grad_norm": 13.5625, "learning_rate": 3.5673016249552085e-06, "loss": 1.3108, "step": 5275 }, { "epoch": 0.6049764935213852, "grad_norm": 14.25, "learning_rate": 3.565522636100833e-06, "loss": 2.0953, "step": 5276 }, { "epoch": 0.6050911592707259, "grad_norm": 18.0, "learning_rate": 3.5637438451052654e-06, "loss": 1.6587, "step": 5277 }, { "epoch": 0.6052058250200665, "grad_norm": 16.375, "learning_rate": 3.561965252213855e-06, "loss": 2.0633, "step": 5278 }, { "epoch": 0.6053204907694072, "grad_norm": 15.0, "learning_rate": 3.56018685767193e-06, "loss": 1.6863, "step": 5279 }, { "epoch": 0.6054351565187478, "grad_norm": 16.25, "learning_rate": 3.5584086617247808e-06, "loss": 1.8107, "step": 5280 }, { "epoch": 0.6055498222680885, "grad_norm": 16.375, "learning_rate": 3.556630664617679e-06, "loss": 1.6331, "step": 5281 }, { "epoch": 0.6056644880174292, "grad_norm": 13.0625, "learning_rate": 3.5548528665958635e-06, "loss": 1.5488, "step": 5282 }, { "epoch": 0.6057791537667698, "grad_norm": 12.0625, "learning_rate": 3.5530752679045466e-06, "loss": 2.1671, "step": 5283 }, { "epoch": 0.6058938195161105, "grad_norm": 15.0, "learning_rate": 3.551297868788916e-06, "loss": 2.004, "step": 5284 }, { "epoch": 0.6060084852654513, "grad_norm": 16.75, "learning_rate": 3.549520669494131e-06, "loss": 1.9316, "step": 5285 }, { "epoch": 0.6061231510147919, "grad_norm": 11.5625, "learning_rate": 3.5477436702653197e-06, "loss": 1.7503, "step": 5286 }, { "epoch": 0.6062378167641326, "grad_norm": 17.625, "learning_rate": 3.5459668713475874e-06, "loss": 1.868, "step": 5287 }, { "epoch": 0.6063524825134732, "grad_norm": 14.8125, "learning_rate": 3.544190272986007e-06, "loss": 1.3967, "step": 5288 }, { "epoch": 0.6064671482628139, "grad_norm": 14.5, "learning_rate": 3.542413875425631e-06, "loss": 1.6448, "step": 5289 }, { "epoch": 0.6065818140121546, "grad_norm": 13.875, "learning_rate": 3.540637678911475e-06, "loss": 1.742, "step": 5290 }, { "epoch": 0.6066964797614952, "grad_norm": 16.875, "learning_rate": 3.5388616836885337e-06, "loss": 2.0823, "step": 5291 }, { "epoch": 0.6068111455108359, "grad_norm": 15.1875, "learning_rate": 3.5370858900017714e-06, "loss": 1.7992, "step": 5292 }, { "epoch": 0.6069258112601766, "grad_norm": 12.6875, "learning_rate": 3.5353102980961227e-06, "loss": 1.8326, "step": 5293 }, { "epoch": 0.6070404770095172, "grad_norm": 14.875, "learning_rate": 3.5335349082165018e-06, "loss": 2.1007, "step": 5294 }, { "epoch": 0.607155142758858, "grad_norm": 13.9375, "learning_rate": 3.531759720607787e-06, "loss": 1.7861, "step": 5295 }, { "epoch": 0.6072698085081986, "grad_norm": 17.625, "learning_rate": 3.52998473551483e-06, "loss": 1.6807, "step": 5296 }, { "epoch": 0.6073844742575393, "grad_norm": 15.625, "learning_rate": 3.528209953182459e-06, "loss": 1.7489, "step": 5297 }, { "epoch": 0.60749914000688, "grad_norm": 14.875, "learning_rate": 3.526435373855468e-06, "loss": 2.0735, "step": 5298 }, { "epoch": 0.6076138057562206, "grad_norm": 13.0, "learning_rate": 3.5246609977786293e-06, "loss": 1.6777, "step": 5299 }, { "epoch": 0.6077284715055613, "grad_norm": 17.0, "learning_rate": 3.522886825196684e-06, "loss": 1.651, "step": 5300 }, { "epoch": 0.6078431372549019, "grad_norm": 35.25, "learning_rate": 3.521112856354346e-06, "loss": 2.1655, "step": 5301 }, { "epoch": 0.6079578030042426, "grad_norm": 15.6875, "learning_rate": 3.5193390914962977e-06, "loss": 1.744, "step": 5302 }, { "epoch": 0.6080724687535833, "grad_norm": 13.75, "learning_rate": 3.5175655308671963e-06, "loss": 1.3737, "step": 5303 }, { "epoch": 0.6081871345029239, "grad_norm": 15.25, "learning_rate": 3.515792174711675e-06, "loss": 2.1594, "step": 5304 }, { "epoch": 0.6083018002522647, "grad_norm": 14.25, "learning_rate": 3.5140190232743305e-06, "loss": 1.7533, "step": 5305 }, { "epoch": 0.6084164660016054, "grad_norm": 34.0, "learning_rate": 3.512246076799737e-06, "loss": 2.2862, "step": 5306 }, { "epoch": 0.608531131750946, "grad_norm": 30.25, "learning_rate": 3.510473335532438e-06, "loss": 2.165, "step": 5307 }, { "epoch": 0.6086457975002867, "grad_norm": 13.9375, "learning_rate": 3.5087007997169478e-06, "loss": 1.4875, "step": 5308 }, { "epoch": 0.6087604632496273, "grad_norm": 14.75, "learning_rate": 3.5069284695977563e-06, "loss": 1.7789, "step": 5309 }, { "epoch": 0.608875128998968, "grad_norm": 13.3125, "learning_rate": 3.5051563454193227e-06, "loss": 1.893, "step": 5310 }, { "epoch": 0.6089897947483087, "grad_norm": 14.875, "learning_rate": 3.5033844274260764e-06, "loss": 1.8216, "step": 5311 }, { "epoch": 0.6091044604976493, "grad_norm": 14.875, "learning_rate": 3.5016127158624207e-06, "loss": 1.9769, "step": 5312 }, { "epoch": 0.60921912624699, "grad_norm": 13.125, "learning_rate": 3.4998412109727277e-06, "loss": 1.5549, "step": 5313 }, { "epoch": 0.6093337919963306, "grad_norm": 12.1875, "learning_rate": 3.498069913001345e-06, "loss": 1.6815, "step": 5314 }, { "epoch": 0.6094484577456714, "grad_norm": 11.875, "learning_rate": 3.4962988221925882e-06, "loss": 1.8108, "step": 5315 }, { "epoch": 0.6095631234950121, "grad_norm": 11.375, "learning_rate": 3.4945279387907464e-06, "loss": 1.7797, "step": 5316 }, { "epoch": 0.6096777892443527, "grad_norm": 35.0, "learning_rate": 3.4927572630400773e-06, "loss": 2.0408, "step": 5317 }, { "epoch": 0.6097924549936934, "grad_norm": 15.875, "learning_rate": 3.490986795184813e-06, "loss": 1.5962, "step": 5318 }, { "epoch": 0.6099071207430341, "grad_norm": 17.125, "learning_rate": 3.4892165354691575e-06, "loss": 1.7316, "step": 5319 }, { "epoch": 0.6100217864923747, "grad_norm": 14.125, "learning_rate": 3.4874464841372824e-06, "loss": 1.579, "step": 5320 }, { "epoch": 0.6101364522417154, "grad_norm": 16.375, "learning_rate": 3.485676641433332e-06, "loss": 2.3445, "step": 5321 }, { "epoch": 0.610251117991056, "grad_norm": 20.5, "learning_rate": 3.4839070076014246e-06, "loss": 1.7752, "step": 5322 }, { "epoch": 0.6103657837403967, "grad_norm": 17.0, "learning_rate": 3.482137582885644e-06, "loss": 2.0323, "step": 5323 }, { "epoch": 0.6104804494897375, "grad_norm": 10.6875, "learning_rate": 3.4803683675300515e-06, "loss": 1.5917, "step": 5324 }, { "epoch": 0.6105951152390781, "grad_norm": 45.25, "learning_rate": 3.478599361778677e-06, "loss": 1.9531, "step": 5325 }, { "epoch": 0.6107097809884188, "grad_norm": 13.375, "learning_rate": 3.47683056587552e-06, "loss": 1.6625, "step": 5326 }, { "epoch": 0.6108244467377595, "grad_norm": 13.8125, "learning_rate": 3.4750619800645512e-06, "loss": 1.3746, "step": 5327 }, { "epoch": 0.6109391124871001, "grad_norm": 29.25, "learning_rate": 3.4732936045897136e-06, "loss": 2.1142, "step": 5328 }, { "epoch": 0.6110537782364408, "grad_norm": 15.3125, "learning_rate": 3.471525439694924e-06, "loss": 2.0933, "step": 5329 }, { "epoch": 0.6111684439857814, "grad_norm": 14.5, "learning_rate": 3.469757485624062e-06, "loss": 1.3991, "step": 5330 }, { "epoch": 0.6112831097351221, "grad_norm": 14.9375, "learning_rate": 3.4679897426209886e-06, "loss": 1.5745, "step": 5331 }, { "epoch": 0.6113977754844628, "grad_norm": 16.375, "learning_rate": 3.466222210929526e-06, "loss": 1.8799, "step": 5332 }, { "epoch": 0.6115124412338034, "grad_norm": 18.875, "learning_rate": 3.4644548907934718e-06, "loss": 1.5762, "step": 5333 }, { "epoch": 0.6116271069831442, "grad_norm": 16.75, "learning_rate": 3.4626877824565963e-06, "loss": 1.7534, "step": 5334 }, { "epoch": 0.6117417727324848, "grad_norm": 12.6875, "learning_rate": 3.4609208861626375e-06, "loss": 1.5674, "step": 5335 }, { "epoch": 0.6118564384818255, "grad_norm": 13.0625, "learning_rate": 3.4591542021553047e-06, "loss": 1.6329, "step": 5336 }, { "epoch": 0.6119711042311662, "grad_norm": 12.625, "learning_rate": 3.4573877306782793e-06, "loss": 1.6102, "step": 5337 }, { "epoch": 0.6120857699805068, "grad_norm": 12.9375, "learning_rate": 3.4556214719752103e-06, "loss": 1.7011, "step": 5338 }, { "epoch": 0.6122004357298475, "grad_norm": 13.25, "learning_rate": 3.4538554262897217e-06, "loss": 1.6048, "step": 5339 }, { "epoch": 0.6123151014791882, "grad_norm": 13.3125, "learning_rate": 3.452089593865404e-06, "loss": 1.5962, "step": 5340 }, { "epoch": 0.6124297672285288, "grad_norm": 18.75, "learning_rate": 3.450323974945822e-06, "loss": 2.0069, "step": 5341 }, { "epoch": 0.6125444329778695, "grad_norm": 12.4375, "learning_rate": 3.4485585697745073e-06, "loss": 1.8255, "step": 5342 }, { "epoch": 0.6126590987272101, "grad_norm": 14.875, "learning_rate": 3.4467933785949637e-06, "loss": 1.888, "step": 5343 }, { "epoch": 0.6127737644765509, "grad_norm": 14.3125, "learning_rate": 3.4450284016506687e-06, "loss": 1.6935, "step": 5344 }, { "epoch": 0.6128884302258916, "grad_norm": 12.75, "learning_rate": 3.4432636391850653e-06, "loss": 1.8959, "step": 5345 }, { "epoch": 0.6130030959752322, "grad_norm": 15.0, "learning_rate": 3.4414990914415682e-06, "loss": 1.6937, "step": 5346 }, { "epoch": 0.6131177617245729, "grad_norm": 18.125, "learning_rate": 3.4397347586635643e-06, "loss": 2.0624, "step": 5347 }, { "epoch": 0.6132324274739135, "grad_norm": 14.8125, "learning_rate": 3.437970641094407e-06, "loss": 1.6945, "step": 5348 }, { "epoch": 0.6133470932232542, "grad_norm": 16.5, "learning_rate": 3.4362067389774255e-06, "loss": 1.5188, "step": 5349 }, { "epoch": 0.6134617589725949, "grad_norm": 13.5625, "learning_rate": 3.434443052555917e-06, "loss": 1.5827, "step": 5350 }, { "epoch": 0.6135764247219355, "grad_norm": 14.375, "learning_rate": 3.4326795820731472e-06, "loss": 2.0048, "step": 5351 }, { "epoch": 0.6136910904712762, "grad_norm": 12.5625, "learning_rate": 3.4309163277723523e-06, "loss": 1.6186, "step": 5352 }, { "epoch": 0.613805756220617, "grad_norm": 14.125, "learning_rate": 3.4291532898967394e-06, "loss": 1.8069, "step": 5353 }, { "epoch": 0.6139204219699576, "grad_norm": 12.875, "learning_rate": 3.4273904686894888e-06, "loss": 1.7628, "step": 5354 }, { "epoch": 0.6140350877192983, "grad_norm": 16.375, "learning_rate": 3.4256278643937455e-06, "loss": 1.8452, "step": 5355 }, { "epoch": 0.6141497534686389, "grad_norm": 14.375, "learning_rate": 3.42386547725263e-06, "loss": 1.7833, "step": 5356 }, { "epoch": 0.6142644192179796, "grad_norm": 15.0, "learning_rate": 3.422103307509227e-06, "loss": 1.602, "step": 5357 }, { "epoch": 0.6143790849673203, "grad_norm": 11.375, "learning_rate": 3.4203413554065935e-06, "loss": 2.205, "step": 5358 }, { "epoch": 0.6144937507166609, "grad_norm": 14.75, "learning_rate": 3.4185796211877604e-06, "loss": 1.6567, "step": 5359 }, { "epoch": 0.6146084164660016, "grad_norm": 16.375, "learning_rate": 3.416818105095725e-06, "loss": 1.498, "step": 5360 }, { "epoch": 0.6147230822153423, "grad_norm": 13.6875, "learning_rate": 3.4150568073734524e-06, "loss": 1.6839, "step": 5361 }, { "epoch": 0.6148377479646829, "grad_norm": 15.0, "learning_rate": 3.4132957282638823e-06, "loss": 1.7945, "step": 5362 }, { "epoch": 0.6149524137140236, "grad_norm": 21.25, "learning_rate": 3.4115348680099193e-06, "loss": 1.7874, "step": 5363 }, { "epoch": 0.6150670794633643, "grad_norm": 14.375, "learning_rate": 3.4097742268544443e-06, "loss": 1.8798, "step": 5364 }, { "epoch": 0.615181745212705, "grad_norm": 15.3125, "learning_rate": 3.4080138050403012e-06, "loss": 1.6617, "step": 5365 }, { "epoch": 0.6152964109620457, "grad_norm": 15.8125, "learning_rate": 3.406253602810309e-06, "loss": 1.43, "step": 5366 }, { "epoch": 0.6154110767113863, "grad_norm": 17.875, "learning_rate": 3.404493620407252e-06, "loss": 2.2413, "step": 5367 }, { "epoch": 0.615525742460727, "grad_norm": 14.5, "learning_rate": 3.4027338580738857e-06, "loss": 1.7084, "step": 5368 }, { "epoch": 0.6156404082100676, "grad_norm": 12.75, "learning_rate": 3.4009743160529384e-06, "loss": 1.6087, "step": 5369 }, { "epoch": 0.6157550739594083, "grad_norm": 15.25, "learning_rate": 3.3992149945871043e-06, "loss": 2.0109, "step": 5370 }, { "epoch": 0.615869739708749, "grad_norm": 15.875, "learning_rate": 3.3974558939190473e-06, "loss": 1.9087, "step": 5371 }, { "epoch": 0.6159844054580896, "grad_norm": 12.4375, "learning_rate": 3.395697014291403e-06, "loss": 1.4697, "step": 5372 }, { "epoch": 0.6160990712074303, "grad_norm": 17.375, "learning_rate": 3.3939383559467733e-06, "loss": 1.7729, "step": 5373 }, { "epoch": 0.6162137369567711, "grad_norm": 13.625, "learning_rate": 3.392179919127734e-06, "loss": 1.7441, "step": 5374 }, { "epoch": 0.6163284027061117, "grad_norm": 14.5625, "learning_rate": 3.3904217040768273e-06, "loss": 1.7304, "step": 5375 }, { "epoch": 0.6164430684554524, "grad_norm": 13.375, "learning_rate": 3.388663711036566e-06, "loss": 1.4887, "step": 5376 }, { "epoch": 0.616557734204793, "grad_norm": 15.0, "learning_rate": 3.3869059402494297e-06, "loss": 1.7045, "step": 5377 }, { "epoch": 0.6166723999541337, "grad_norm": 28.5, "learning_rate": 3.385148391957869e-06, "loss": 1.8487, "step": 5378 }, { "epoch": 0.6167870657034744, "grad_norm": 30.375, "learning_rate": 3.3833910664043084e-06, "loss": 2.0413, "step": 5379 }, { "epoch": 0.616901731452815, "grad_norm": 14.9375, "learning_rate": 3.3816339638311337e-06, "loss": 1.8803, "step": 5380 }, { "epoch": 0.6170163972021557, "grad_norm": 15.625, "learning_rate": 3.379877084480706e-06, "loss": 1.7196, "step": 5381 }, { "epoch": 0.6171310629514963, "grad_norm": 14.5625, "learning_rate": 3.378120428595352e-06, "loss": 1.8201, "step": 5382 }, { "epoch": 0.617245728700837, "grad_norm": 15.5625, "learning_rate": 3.376363996417367e-06, "loss": 1.5685, "step": 5383 }, { "epoch": 0.6173603944501778, "grad_norm": 13.1875, "learning_rate": 3.3746077881890198e-06, "loss": 1.457, "step": 5384 }, { "epoch": 0.6174750601995184, "grad_norm": 16.5, "learning_rate": 3.3728518041525473e-06, "loss": 1.5885, "step": 5385 }, { "epoch": 0.6175897259488591, "grad_norm": 12.375, "learning_rate": 3.3710960445501507e-06, "loss": 1.6823, "step": 5386 }, { "epoch": 0.6177043916981998, "grad_norm": 12.875, "learning_rate": 3.369340509624006e-06, "loss": 1.7644, "step": 5387 }, { "epoch": 0.6178190574475404, "grad_norm": 18.625, "learning_rate": 3.367585199616253e-06, "loss": 1.3357, "step": 5388 }, { "epoch": 0.6179337231968811, "grad_norm": 12.875, "learning_rate": 3.365830114769007e-06, "loss": 1.9474, "step": 5389 }, { "epoch": 0.6180483889462217, "grad_norm": 17.25, "learning_rate": 3.3640752553243457e-06, "loss": 1.8802, "step": 5390 }, { "epoch": 0.6181630546955624, "grad_norm": 14.4375, "learning_rate": 3.36232062152432e-06, "loss": 1.7087, "step": 5391 }, { "epoch": 0.6182777204449031, "grad_norm": 13.1875, "learning_rate": 3.360566213610947e-06, "loss": 1.7298, "step": 5392 }, { "epoch": 0.6183923861942437, "grad_norm": 18.5, "learning_rate": 3.358812031826213e-06, "loss": 2.104, "step": 5393 }, { "epoch": 0.6185070519435845, "grad_norm": 12.1875, "learning_rate": 3.3570580764120777e-06, "loss": 1.4531, "step": 5394 }, { "epoch": 0.6186217176929252, "grad_norm": 18.0, "learning_rate": 3.355304347610463e-06, "loss": 1.3699, "step": 5395 }, { "epoch": 0.6187363834422658, "grad_norm": 14.75, "learning_rate": 3.3535508456632614e-06, "loss": 1.2051, "step": 5396 }, { "epoch": 0.6188510491916065, "grad_norm": 19.375, "learning_rate": 3.3517975708123386e-06, "loss": 2.3698, "step": 5397 }, { "epoch": 0.6189657149409471, "grad_norm": 15.8125, "learning_rate": 3.3500445232995205e-06, "loss": 1.3408, "step": 5398 }, { "epoch": 0.6190803806902878, "grad_norm": 33.0, "learning_rate": 3.3482917033666105e-06, "loss": 1.9647, "step": 5399 }, { "epoch": 0.6191950464396285, "grad_norm": 13.4375, "learning_rate": 3.3465391112553768e-06, "loss": 1.5961, "step": 5400 }, { "epoch": 0.6193097121889691, "grad_norm": 22.75, "learning_rate": 3.3447867472075553e-06, "loss": 1.6969, "step": 5401 }, { "epoch": 0.6194243779383098, "grad_norm": 12.6875, "learning_rate": 3.3430346114648494e-06, "loss": 1.4706, "step": 5402 }, { "epoch": 0.6195390436876504, "grad_norm": 16.25, "learning_rate": 3.3412827042689326e-06, "loss": 1.8919, "step": 5403 }, { "epoch": 0.6196537094369912, "grad_norm": 12.5, "learning_rate": 3.339531025861451e-06, "loss": 1.4061, "step": 5404 }, { "epoch": 0.6197683751863319, "grad_norm": 15.0, "learning_rate": 3.3377795764840125e-06, "loss": 1.5267, "step": 5405 }, { "epoch": 0.6198830409356725, "grad_norm": 15.0625, "learning_rate": 3.3360283563781975e-06, "loss": 1.6372, "step": 5406 }, { "epoch": 0.6199977066850132, "grad_norm": 14.875, "learning_rate": 3.3342773657855514e-06, "loss": 1.7661, "step": 5407 }, { "epoch": 0.6201123724343539, "grad_norm": 15.5, "learning_rate": 3.3325266049475912e-06, "loss": 2.3098, "step": 5408 }, { "epoch": 0.6202270381836945, "grad_norm": 11.75, "learning_rate": 3.3307760741058005e-06, "loss": 1.5681, "step": 5409 }, { "epoch": 0.6203417039330352, "grad_norm": 16.25, "learning_rate": 3.329025773501634e-06, "loss": 1.6881, "step": 5410 }, { "epoch": 0.6204563696823758, "grad_norm": 15.4375, "learning_rate": 3.3272757033765086e-06, "loss": 1.8182, "step": 5411 }, { "epoch": 0.6205710354317165, "grad_norm": 23.875, "learning_rate": 3.325525863971816e-06, "loss": 1.7412, "step": 5412 }, { "epoch": 0.6206857011810573, "grad_norm": 13.0, "learning_rate": 3.32377625552891e-06, "loss": 1.9462, "step": 5413 }, { "epoch": 0.6208003669303979, "grad_norm": 15.625, "learning_rate": 3.3220268782891195e-06, "loss": 1.5582, "step": 5414 }, { "epoch": 0.6209150326797386, "grad_norm": 13.75, "learning_rate": 3.3202777324937363e-06, "loss": 1.6977, "step": 5415 }, { "epoch": 0.6210296984290792, "grad_norm": 17.5, "learning_rate": 3.3185288183840212e-06, "loss": 1.915, "step": 5416 }, { "epoch": 0.6211443641784199, "grad_norm": 13.5625, "learning_rate": 3.3167801362012035e-06, "loss": 1.7317, "step": 5417 }, { "epoch": 0.6212590299277606, "grad_norm": 14.25, "learning_rate": 3.3150316861864794e-06, "loss": 1.6344, "step": 5418 }, { "epoch": 0.6213736956771012, "grad_norm": 17.75, "learning_rate": 3.3132834685810187e-06, "loss": 2.1515, "step": 5419 }, { "epoch": 0.6214883614264419, "grad_norm": 11.8125, "learning_rate": 3.3115354836259515e-06, "loss": 1.8073, "step": 5420 }, { "epoch": 0.6216030271757826, "grad_norm": 17.75, "learning_rate": 3.309787731562378e-06, "loss": 1.8395, "step": 5421 }, { "epoch": 0.6217176929251232, "grad_norm": 13.25, "learning_rate": 3.308040212631369e-06, "loss": 1.6405, "step": 5422 }, { "epoch": 0.621832358674464, "grad_norm": 16.0, "learning_rate": 3.30629292707396e-06, "loss": 1.6881, "step": 5423 }, { "epoch": 0.6219470244238046, "grad_norm": 12.3125, "learning_rate": 3.304545875131157e-06, "loss": 1.7204, "step": 5424 }, { "epoch": 0.6220616901731453, "grad_norm": 14.4375, "learning_rate": 3.3027990570439337e-06, "loss": 1.7859, "step": 5425 }, { "epoch": 0.622176355922486, "grad_norm": 12.1875, "learning_rate": 3.301052473053228e-06, "loss": 1.8203, "step": 5426 }, { "epoch": 0.6222910216718266, "grad_norm": 12.4375, "learning_rate": 3.2993061233999493e-06, "loss": 1.6447, "step": 5427 }, { "epoch": 0.6224056874211673, "grad_norm": 11.4375, "learning_rate": 3.2975600083249703e-06, "loss": 1.4029, "step": 5428 }, { "epoch": 0.622520353170508, "grad_norm": 18.875, "learning_rate": 3.2958141280691393e-06, "loss": 1.7037, "step": 5429 }, { "epoch": 0.6226350189198486, "grad_norm": 15.375, "learning_rate": 3.294068482873264e-06, "loss": 2.2352, "step": 5430 }, { "epoch": 0.6227496846691893, "grad_norm": 14.5625, "learning_rate": 3.2923230729781236e-06, "loss": 2.0318, "step": 5431 }, { "epoch": 0.6228643504185299, "grad_norm": 27.625, "learning_rate": 3.2905778986244636e-06, "loss": 2.1801, "step": 5432 }, { "epoch": 0.6229790161678707, "grad_norm": 21.125, "learning_rate": 3.288832960052998e-06, "loss": 1.8292, "step": 5433 }, { "epoch": 0.6230936819172114, "grad_norm": 12.9375, "learning_rate": 3.287088257504407e-06, "loss": 1.1588, "step": 5434 }, { "epoch": 0.623208347666552, "grad_norm": 14.1875, "learning_rate": 3.2853437912193416e-06, "loss": 1.413, "step": 5435 }, { "epoch": 0.6233230134158927, "grad_norm": 13.125, "learning_rate": 3.2835995614384143e-06, "loss": 1.6769, "step": 5436 }, { "epoch": 0.6234376791652333, "grad_norm": 17.0, "learning_rate": 3.281855568402211e-06, "loss": 1.72, "step": 5437 }, { "epoch": 0.623552344914574, "grad_norm": 15.5625, "learning_rate": 3.28011181235128e-06, "loss": 1.5918, "step": 5438 }, { "epoch": 0.6236670106639147, "grad_norm": 13.375, "learning_rate": 3.278368293526142e-06, "loss": 1.8259, "step": 5439 }, { "epoch": 0.6237816764132553, "grad_norm": 17.25, "learning_rate": 3.2766250121672786e-06, "loss": 1.7586, "step": 5440 }, { "epoch": 0.623896342162596, "grad_norm": 13.5625, "learning_rate": 3.274881968515147e-06, "loss": 1.4591, "step": 5441 }, { "epoch": 0.6240110079119368, "grad_norm": 13.6875, "learning_rate": 3.2731391628101616e-06, "loss": 1.7853, "step": 5442 }, { "epoch": 0.6241256736612774, "grad_norm": 14.375, "learning_rate": 3.271396595292712e-06, "loss": 1.8721, "step": 5443 }, { "epoch": 0.6242403394106181, "grad_norm": 11.375, "learning_rate": 3.269654266203153e-06, "loss": 1.3929, "step": 5444 }, { "epoch": 0.6243550051599587, "grad_norm": 15.375, "learning_rate": 3.2679121757818033e-06, "loss": 1.7949, "step": 5445 }, { "epoch": 0.6244696709092994, "grad_norm": 15.375, "learning_rate": 3.2661703242689542e-06, "loss": 2.1147, "step": 5446 }, { "epoch": 0.6245843366586401, "grad_norm": 14.8125, "learning_rate": 3.264428711904859e-06, "loss": 1.7932, "step": 5447 }, { "epoch": 0.6246990024079807, "grad_norm": 16.5, "learning_rate": 3.2626873389297374e-06, "loss": 1.998, "step": 5448 }, { "epoch": 0.6248136681573214, "grad_norm": 16.875, "learning_rate": 3.2609462055837825e-06, "loss": 2.254, "step": 5449 }, { "epoch": 0.624928333906662, "grad_norm": 14.1875, "learning_rate": 3.2592053121071504e-06, "loss": 1.7089, "step": 5450 }, { "epoch": 0.6250429996560027, "grad_norm": 14.0, "learning_rate": 3.2574646587399617e-06, "loss": 1.9571, "step": 5451 }, { "epoch": 0.6251576654053435, "grad_norm": 19.125, "learning_rate": 3.255724245722309e-06, "loss": 2.1625, "step": 5452 }, { "epoch": 0.6252723311546841, "grad_norm": 13.125, "learning_rate": 3.253984073294245e-06, "loss": 1.4896, "step": 5453 }, { "epoch": 0.6253869969040248, "grad_norm": 14.0, "learning_rate": 3.2522441416957985e-06, "loss": 1.8149, "step": 5454 }, { "epoch": 0.6255016626533655, "grad_norm": 13.875, "learning_rate": 3.250504451166956e-06, "loss": 1.7718, "step": 5455 }, { "epoch": 0.6256163284027061, "grad_norm": 17.125, "learning_rate": 3.2487650019476775e-06, "loss": 1.4211, "step": 5456 }, { "epoch": 0.6257309941520468, "grad_norm": 18.5, "learning_rate": 3.2470257942778837e-06, "loss": 1.9363, "step": 5457 }, { "epoch": 0.6258456599013874, "grad_norm": 13.125, "learning_rate": 3.2452868283974674e-06, "loss": 1.9313, "step": 5458 }, { "epoch": 0.6259603256507281, "grad_norm": 13.625, "learning_rate": 3.243548104546286e-06, "loss": 2.0218, "step": 5459 }, { "epoch": 0.6260749914000688, "grad_norm": 14.8125, "learning_rate": 3.241809622964162e-06, "loss": 1.603, "step": 5460 }, { "epoch": 0.6261896571494094, "grad_norm": 12.5625, "learning_rate": 3.240071383890886e-06, "loss": 1.887, "step": 5461 }, { "epoch": 0.6263043228987502, "grad_norm": 16.875, "learning_rate": 3.2383333875662166e-06, "loss": 2.2058, "step": 5462 }, { "epoch": 0.6264189886480909, "grad_norm": 14.5625, "learning_rate": 3.236595634229874e-06, "loss": 1.4772, "step": 5463 }, { "epoch": 0.6265336543974315, "grad_norm": 12.6875, "learning_rate": 3.2348581241215516e-06, "loss": 1.5257, "step": 5464 }, { "epoch": 0.6266483201467722, "grad_norm": 15.0, "learning_rate": 3.2331208574809036e-06, "loss": 1.7748, "step": 5465 }, { "epoch": 0.6267629858961128, "grad_norm": 13.5625, "learning_rate": 3.2313838345475544e-06, "loss": 1.8571, "step": 5466 }, { "epoch": 0.6268776516454535, "grad_norm": 15.125, "learning_rate": 3.2296470555610916e-06, "loss": 1.9315, "step": 5467 }, { "epoch": 0.6269923173947942, "grad_norm": 13.6875, "learning_rate": 3.22791052076107e-06, "loss": 1.9875, "step": 5468 }, { "epoch": 0.6271069831441348, "grad_norm": 15.6875, "learning_rate": 3.2261742303870157e-06, "loss": 2.0041, "step": 5469 }, { "epoch": 0.6272216488934755, "grad_norm": 14.3125, "learning_rate": 3.2244381846784124e-06, "loss": 1.6896, "step": 5470 }, { "epoch": 0.6273363146428161, "grad_norm": 15.9375, "learning_rate": 3.2227023838747173e-06, "loss": 1.9048, "step": 5471 }, { "epoch": 0.6274509803921569, "grad_norm": 18.0, "learning_rate": 3.2209668282153506e-06, "loss": 1.8256, "step": 5472 }, { "epoch": 0.6275656461414976, "grad_norm": 11.5625, "learning_rate": 3.219231517939696e-06, "loss": 1.5811, "step": 5473 }, { "epoch": 0.6276803118908382, "grad_norm": 16.5, "learning_rate": 3.21749645328711e-06, "loss": 2.0255, "step": 5474 }, { "epoch": 0.6277949776401789, "grad_norm": 15.75, "learning_rate": 3.2157616344969116e-06, "loss": 2.1358, "step": 5475 }, { "epoch": 0.6279096433895196, "grad_norm": 14.375, "learning_rate": 3.214027061808385e-06, "loss": 2.0956, "step": 5476 }, { "epoch": 0.6280243091388602, "grad_norm": 15.25, "learning_rate": 3.2122927354607814e-06, "loss": 1.8683, "step": 5477 }, { "epoch": 0.6281389748882009, "grad_norm": 15.625, "learning_rate": 3.2105586556933162e-06, "loss": 1.5784, "step": 5478 }, { "epoch": 0.6282536406375415, "grad_norm": 14.25, "learning_rate": 3.208824822745177e-06, "loss": 2.2478, "step": 5479 }, { "epoch": 0.6283683063868822, "grad_norm": 14.3125, "learning_rate": 3.207091236855509e-06, "loss": 1.6772, "step": 5480 }, { "epoch": 0.628482972136223, "grad_norm": 11.8125, "learning_rate": 3.2053578982634304e-06, "loss": 1.5805, "step": 5481 }, { "epoch": 0.6285976378855636, "grad_norm": 16.875, "learning_rate": 3.2036248072080194e-06, "loss": 1.7268, "step": 5482 }, { "epoch": 0.6287123036349043, "grad_norm": 12.4375, "learning_rate": 3.2018919639283248e-06, "loss": 1.6995, "step": 5483 }, { "epoch": 0.6288269693842449, "grad_norm": 15.0, "learning_rate": 3.200159368663358e-06, "loss": 1.6948, "step": 5484 }, { "epoch": 0.6289416351335856, "grad_norm": 15.0, "learning_rate": 3.1984270216520996e-06, "loss": 1.6004, "step": 5485 }, { "epoch": 0.6290563008829263, "grad_norm": 15.25, "learning_rate": 3.1966949231334917e-06, "loss": 1.9674, "step": 5486 }, { "epoch": 0.6291709666322669, "grad_norm": 12.0625, "learning_rate": 3.1949630733464455e-06, "loss": 1.9795, "step": 5487 }, { "epoch": 0.6292856323816076, "grad_norm": 17.75, "learning_rate": 3.1932314725298348e-06, "loss": 1.6722, "step": 5488 }, { "epoch": 0.6294002981309483, "grad_norm": 30.5, "learning_rate": 3.1915001209225026e-06, "loss": 1.953, "step": 5489 }, { "epoch": 0.6295149638802889, "grad_norm": 15.5625, "learning_rate": 3.1897690187632576e-06, "loss": 1.4763, "step": 5490 }, { "epoch": 0.6296296296296297, "grad_norm": 14.1875, "learning_rate": 3.1880381662908694e-06, "loss": 1.6846, "step": 5491 }, { "epoch": 0.6297442953789703, "grad_norm": 18.25, "learning_rate": 3.1863075637440764e-06, "loss": 2.2151, "step": 5492 }, { "epoch": 0.629858961128311, "grad_norm": 17.625, "learning_rate": 3.1845772113615814e-06, "loss": 1.9645, "step": 5493 }, { "epoch": 0.6299736268776517, "grad_norm": 12.1875, "learning_rate": 3.182847109382058e-06, "loss": 1.6117, "step": 5494 }, { "epoch": 0.6300882926269923, "grad_norm": 26.5, "learning_rate": 3.1811172580441356e-06, "loss": 1.9905, "step": 5495 }, { "epoch": 0.630202958376333, "grad_norm": 13.5, "learning_rate": 3.179387657586418e-06, "loss": 1.959, "step": 5496 }, { "epoch": 0.6303176241256737, "grad_norm": 16.25, "learning_rate": 3.1776583082474688e-06, "loss": 1.7985, "step": 5497 }, { "epoch": 0.6304322898750143, "grad_norm": 16.0, "learning_rate": 3.1759292102658174e-06, "loss": 1.475, "step": 5498 }, { "epoch": 0.630546955624355, "grad_norm": 13.0, "learning_rate": 3.1742003638799624e-06, "loss": 1.6457, "step": 5499 }, { "epoch": 0.6306616213736956, "grad_norm": 12.4375, "learning_rate": 3.1724717693283647e-06, "loss": 1.6014, "step": 5500 }, { "epoch": 0.6307762871230363, "grad_norm": 14.9375, "learning_rate": 3.17074342684945e-06, "loss": 1.7949, "step": 5501 }, { "epoch": 0.6308909528723771, "grad_norm": 15.3125, "learning_rate": 3.1690153366816123e-06, "loss": 1.9598, "step": 5502 }, { "epoch": 0.6310056186217177, "grad_norm": 16.625, "learning_rate": 3.167287499063204e-06, "loss": 1.7214, "step": 5503 }, { "epoch": 0.6311202843710584, "grad_norm": 15.0, "learning_rate": 3.165559914232553e-06, "loss": 1.4121, "step": 5504 }, { "epoch": 0.631234950120399, "grad_norm": 13.4375, "learning_rate": 3.163832582427944e-06, "loss": 1.6268, "step": 5505 }, { "epoch": 0.6313496158697397, "grad_norm": 15.1875, "learning_rate": 3.1621055038876303e-06, "loss": 1.5984, "step": 5506 }, { "epoch": 0.6314642816190804, "grad_norm": 12.625, "learning_rate": 3.160378678849827e-06, "loss": 1.6314, "step": 5507 }, { "epoch": 0.631578947368421, "grad_norm": 16.375, "learning_rate": 3.1586521075527186e-06, "loss": 1.583, "step": 5508 }, { "epoch": 0.6316936131177617, "grad_norm": 13.0625, "learning_rate": 3.156925790234454e-06, "loss": 1.8674, "step": 5509 }, { "epoch": 0.6318082788671024, "grad_norm": 15.375, "learning_rate": 3.1551997271331445e-06, "loss": 1.7926, "step": 5510 }, { "epoch": 0.631922944616443, "grad_norm": 13.875, "learning_rate": 3.1534739184868673e-06, "loss": 1.8233, "step": 5511 }, { "epoch": 0.6320376103657838, "grad_norm": 12.3125, "learning_rate": 3.1517483645336654e-06, "loss": 1.7788, "step": 5512 }, { "epoch": 0.6321522761151244, "grad_norm": 13.25, "learning_rate": 3.150023065511544e-06, "loss": 1.7275, "step": 5513 }, { "epoch": 0.6322669418644651, "grad_norm": 15.4375, "learning_rate": 3.148298021658478e-06, "loss": 1.8948, "step": 5514 }, { "epoch": 0.6323816076138058, "grad_norm": 14.4375, "learning_rate": 3.1465732332124044e-06, "loss": 1.8251, "step": 5515 }, { "epoch": 0.6324962733631464, "grad_norm": 13.9375, "learning_rate": 3.144848700411223e-06, "loss": 1.6572, "step": 5516 }, { "epoch": 0.6326109391124871, "grad_norm": 13.0, "learning_rate": 3.1431244234928006e-06, "loss": 1.6137, "step": 5517 }, { "epoch": 0.6327256048618277, "grad_norm": 10.6875, "learning_rate": 3.1414004026949673e-06, "loss": 1.6434, "step": 5518 }, { "epoch": 0.6328402706111684, "grad_norm": 13.3125, "learning_rate": 3.1396766382555227e-06, "loss": 1.9826, "step": 5519 }, { "epoch": 0.6329549363605091, "grad_norm": 18.0, "learning_rate": 3.1379531304122233e-06, "loss": 1.9654, "step": 5520 }, { "epoch": 0.6330696021098497, "grad_norm": 11.3125, "learning_rate": 3.136229879402798e-06, "loss": 1.7373, "step": 5521 }, { "epoch": 0.6331842678591905, "grad_norm": 16.5, "learning_rate": 3.134506885464933e-06, "loss": 1.9653, "step": 5522 }, { "epoch": 0.6332989336085312, "grad_norm": 12.6875, "learning_rate": 3.132784148836282e-06, "loss": 2.0253, "step": 5523 }, { "epoch": 0.6334135993578718, "grad_norm": 14.5, "learning_rate": 3.131061669754466e-06, "loss": 1.5727, "step": 5524 }, { "epoch": 0.6335282651072125, "grad_norm": 13.25, "learning_rate": 3.129339448457069e-06, "loss": 1.7402, "step": 5525 }, { "epoch": 0.6336429308565531, "grad_norm": 12.6875, "learning_rate": 3.1276174851816353e-06, "loss": 1.6811, "step": 5526 }, { "epoch": 0.6337575966058938, "grad_norm": 13.1875, "learning_rate": 3.1258957801656804e-06, "loss": 1.4989, "step": 5527 }, { "epoch": 0.6338722623552345, "grad_norm": 32.0, "learning_rate": 3.1241743336466767e-06, "loss": 1.7036, "step": 5528 }, { "epoch": 0.6339869281045751, "grad_norm": 17.75, "learning_rate": 3.122453145862069e-06, "loss": 1.7658, "step": 5529 }, { "epoch": 0.6341015938539158, "grad_norm": 21.0, "learning_rate": 3.1207322170492595e-06, "loss": 1.5807, "step": 5530 }, { "epoch": 0.6342162596032566, "grad_norm": 12.8125, "learning_rate": 3.1190115474456196e-06, "loss": 1.8115, "step": 5531 }, { "epoch": 0.6343309253525972, "grad_norm": 14.0, "learning_rate": 3.1172911372884805e-06, "loss": 1.747, "step": 5532 }, { "epoch": 0.6344455911019379, "grad_norm": 11.0, "learning_rate": 3.115570986815141e-06, "loss": 1.5478, "step": 5533 }, { "epoch": 0.6345602568512785, "grad_norm": 12.5625, "learning_rate": 3.1138510962628652e-06, "loss": 1.8823, "step": 5534 }, { "epoch": 0.6346749226006192, "grad_norm": 15.75, "learning_rate": 3.1121314658688774e-06, "loss": 1.5035, "step": 5535 }, { "epoch": 0.6347895883499599, "grad_norm": 13.375, "learning_rate": 3.1104120958703664e-06, "loss": 1.621, "step": 5536 }, { "epoch": 0.6349042540993005, "grad_norm": 15.25, "learning_rate": 3.1086929865044896e-06, "loss": 1.5275, "step": 5537 }, { "epoch": 0.6350189198486412, "grad_norm": 17.625, "learning_rate": 3.106974138008362e-06, "loss": 2.0018, "step": 5538 }, { "epoch": 0.6351335855979818, "grad_norm": 14.6875, "learning_rate": 3.1052555506190672e-06, "loss": 1.5182, "step": 5539 }, { "epoch": 0.6352482513473225, "grad_norm": 12.4375, "learning_rate": 3.1035372245736543e-06, "loss": 1.8313, "step": 5540 }, { "epoch": 0.6353629170966633, "grad_norm": 17.375, "learning_rate": 3.1018191601091314e-06, "loss": 1.9141, "step": 5541 }, { "epoch": 0.6354775828460039, "grad_norm": 36.75, "learning_rate": 3.100101357462472e-06, "loss": 2.1581, "step": 5542 }, { "epoch": 0.6355922485953446, "grad_norm": 13.1875, "learning_rate": 3.0983838168706127e-06, "loss": 1.7046, "step": 5543 }, { "epoch": 0.6357069143446853, "grad_norm": 28.375, "learning_rate": 3.096666538570461e-06, "loss": 1.7143, "step": 5544 }, { "epoch": 0.6358215800940259, "grad_norm": 12.125, "learning_rate": 3.0949495227988775e-06, "loss": 1.7758, "step": 5545 }, { "epoch": 0.6359362458433666, "grad_norm": 14.0625, "learning_rate": 3.093232769792695e-06, "loss": 1.9162, "step": 5546 }, { "epoch": 0.6360509115927072, "grad_norm": 11.25, "learning_rate": 3.091516279788706e-06, "loss": 1.7996, "step": 5547 }, { "epoch": 0.6361655773420479, "grad_norm": 39.0, "learning_rate": 3.089800053023665e-06, "loss": 2.0553, "step": 5548 }, { "epoch": 0.6362802430913886, "grad_norm": 15.625, "learning_rate": 3.0880840897342957e-06, "loss": 1.8769, "step": 5549 }, { "epoch": 0.6363949088407292, "grad_norm": 11.8125, "learning_rate": 3.086368390157283e-06, "loss": 1.7359, "step": 5550 }, { "epoch": 0.63650957459007, "grad_norm": 13.125, "learning_rate": 3.0846529545292737e-06, "loss": 1.916, "step": 5551 }, { "epoch": 0.6366242403394106, "grad_norm": 12.25, "learning_rate": 3.082937783086879e-06, "loss": 1.7721, "step": 5552 }, { "epoch": 0.6367389060887513, "grad_norm": 30.25, "learning_rate": 3.0812228760666753e-06, "loss": 1.8177, "step": 5553 }, { "epoch": 0.636853571838092, "grad_norm": 15.875, "learning_rate": 3.0795082337051986e-06, "loss": 1.8294, "step": 5554 }, { "epoch": 0.6369682375874326, "grad_norm": 12.25, "learning_rate": 3.077793856238954e-06, "loss": 1.4474, "step": 5555 }, { "epoch": 0.6370829033367733, "grad_norm": 13.6875, "learning_rate": 3.0760797439044075e-06, "loss": 1.8852, "step": 5556 }, { "epoch": 0.637197569086114, "grad_norm": 13.1875, "learning_rate": 3.074365896937986e-06, "loss": 1.4163, "step": 5557 }, { "epoch": 0.6373122348354546, "grad_norm": 17.375, "learning_rate": 3.0726523155760842e-06, "loss": 1.8453, "step": 5558 }, { "epoch": 0.6374269005847953, "grad_norm": 14.0, "learning_rate": 3.0709390000550543e-06, "loss": 1.6874, "step": 5559 }, { "epoch": 0.637541566334136, "grad_norm": 17.5, "learning_rate": 3.0692259506112208e-06, "loss": 2.0603, "step": 5560 }, { "epoch": 0.6376562320834767, "grad_norm": 17.875, "learning_rate": 3.067513167480862e-06, "loss": 1.8359, "step": 5561 }, { "epoch": 0.6377708978328174, "grad_norm": 15.375, "learning_rate": 3.065800650900227e-06, "loss": 1.5822, "step": 5562 }, { "epoch": 0.637885563582158, "grad_norm": 14.75, "learning_rate": 3.0640884011055214e-06, "loss": 1.6397, "step": 5563 }, { "epoch": 0.6380002293314987, "grad_norm": 18.25, "learning_rate": 3.062376418332918e-06, "loss": 1.6291, "step": 5564 }, { "epoch": 0.6381148950808394, "grad_norm": 15.6875, "learning_rate": 3.0606647028185555e-06, "loss": 1.9057, "step": 5565 }, { "epoch": 0.63822956083018, "grad_norm": 13.0, "learning_rate": 3.05895325479853e-06, "loss": 1.7139, "step": 5566 }, { "epoch": 0.6383442265795207, "grad_norm": 14.25, "learning_rate": 3.0572420745089026e-06, "loss": 1.9197, "step": 5567 }, { "epoch": 0.6384588923288613, "grad_norm": 16.375, "learning_rate": 3.055531162185699e-06, "loss": 1.4636, "step": 5568 }, { "epoch": 0.638573558078202, "grad_norm": 13.9375, "learning_rate": 3.053820518064905e-06, "loss": 1.8164, "step": 5569 }, { "epoch": 0.6386882238275428, "grad_norm": 16.5, "learning_rate": 3.052110142382474e-06, "loss": 1.605, "step": 5570 }, { "epoch": 0.6388028895768834, "grad_norm": 14.5, "learning_rate": 3.05040003537432e-06, "loss": 1.5615, "step": 5571 }, { "epoch": 0.6389175553262241, "grad_norm": 13.25, "learning_rate": 3.048690197276318e-06, "loss": 1.6936, "step": 5572 }, { "epoch": 0.6390322210755647, "grad_norm": 16.75, "learning_rate": 3.0469806283243063e-06, "loss": 2.1794, "step": 5573 }, { "epoch": 0.6391468868249054, "grad_norm": 11.5, "learning_rate": 3.0452713287540893e-06, "loss": 1.6456, "step": 5574 }, { "epoch": 0.6392615525742461, "grad_norm": 15.4375, "learning_rate": 3.0435622988014336e-06, "loss": 1.5482, "step": 5575 }, { "epoch": 0.6393762183235867, "grad_norm": 14.6875, "learning_rate": 3.0418535387020644e-06, "loss": 1.7249, "step": 5576 }, { "epoch": 0.6394908840729274, "grad_norm": 19.875, "learning_rate": 3.0401450486916747e-06, "loss": 2.1514, "step": 5577 }, { "epoch": 0.6396055498222681, "grad_norm": 13.6875, "learning_rate": 3.0384368290059175e-06, "loss": 1.5284, "step": 5578 }, { "epoch": 0.6397202155716087, "grad_norm": 14.9375, "learning_rate": 3.036728879880407e-06, "loss": 1.9347, "step": 5579 }, { "epoch": 0.6398348813209495, "grad_norm": 11.75, "learning_rate": 3.035021201550725e-06, "loss": 1.4489, "step": 5580 }, { "epoch": 0.6399495470702901, "grad_norm": 13.6875, "learning_rate": 3.0333137942524134e-06, "loss": 1.8113, "step": 5581 }, { "epoch": 0.6400642128196308, "grad_norm": 13.375, "learning_rate": 3.0316066582209745e-06, "loss": 1.4881, "step": 5582 }, { "epoch": 0.6401788785689715, "grad_norm": 14.0, "learning_rate": 3.0298997936918772e-06, "loss": 1.9472, "step": 5583 }, { "epoch": 0.6402935443183121, "grad_norm": 14.1875, "learning_rate": 3.0281932009005476e-06, "loss": 1.6699, "step": 5584 }, { "epoch": 0.6404082100676528, "grad_norm": 16.0, "learning_rate": 3.026486880082382e-06, "loss": 1.6809, "step": 5585 }, { "epoch": 0.6405228758169934, "grad_norm": 18.375, "learning_rate": 3.024780831472732e-06, "loss": 1.8958, "step": 5586 }, { "epoch": 0.6406375415663341, "grad_norm": 13.5, "learning_rate": 3.0230750553069153e-06, "loss": 1.5044, "step": 5587 }, { "epoch": 0.6407522073156748, "grad_norm": 15.3125, "learning_rate": 3.0213695518202114e-06, "loss": 1.7028, "step": 5588 }, { "epoch": 0.6408668730650154, "grad_norm": 12.4375, "learning_rate": 3.0196643212478593e-06, "loss": 1.2562, "step": 5589 }, { "epoch": 0.6409815388143562, "grad_norm": 15.6875, "learning_rate": 3.0179593638250683e-06, "loss": 1.522, "step": 5590 }, { "epoch": 0.6410962045636969, "grad_norm": 14.4375, "learning_rate": 3.016254679787002e-06, "loss": 1.701, "step": 5591 }, { "epoch": 0.6412108703130375, "grad_norm": 15.1875, "learning_rate": 3.014550269368788e-06, "loss": 1.9589, "step": 5592 }, { "epoch": 0.6413255360623782, "grad_norm": 18.25, "learning_rate": 3.0128461328055193e-06, "loss": 1.9528, "step": 5593 }, { "epoch": 0.6414402018117188, "grad_norm": 13.1875, "learning_rate": 3.011142270332245e-06, "loss": 1.7931, "step": 5594 }, { "epoch": 0.6415548675610595, "grad_norm": 22.0, "learning_rate": 3.0094386821839847e-06, "loss": 2.0069, "step": 5595 }, { "epoch": 0.6416695333104002, "grad_norm": 12.9375, "learning_rate": 3.0077353685957166e-06, "loss": 1.4226, "step": 5596 }, { "epoch": 0.6417841990597408, "grad_norm": 15.1875, "learning_rate": 3.006032329802378e-06, "loss": 1.9095, "step": 5597 }, { "epoch": 0.6418988648090815, "grad_norm": 16.0, "learning_rate": 3.00432956603887e-06, "loss": 1.6266, "step": 5598 }, { "epoch": 0.6420135305584223, "grad_norm": 14.4375, "learning_rate": 3.002627077540056e-06, "loss": 2.0529, "step": 5599 }, { "epoch": 0.6421281963077629, "grad_norm": 12.25, "learning_rate": 3.0009248645407657e-06, "loss": 1.9337, "step": 5600 }, { "epoch": 0.6422428620571036, "grad_norm": 12.3125, "learning_rate": 2.999222927275783e-06, "loss": 1.9317, "step": 5601 }, { "epoch": 0.6423575278064442, "grad_norm": 13.125, "learning_rate": 2.997521265979861e-06, "loss": 1.8804, "step": 5602 }, { "epoch": 0.6424721935557849, "grad_norm": 14.9375, "learning_rate": 2.9958198808877095e-06, "loss": 1.7519, "step": 5603 }, { "epoch": 0.6425868593051256, "grad_norm": 13.3125, "learning_rate": 2.9941187722339994e-06, "loss": 1.9171, "step": 5604 }, { "epoch": 0.6427015250544662, "grad_norm": 13.5625, "learning_rate": 2.992417940253371e-06, "loss": 1.928, "step": 5605 }, { "epoch": 0.6428161908038069, "grad_norm": 10.3125, "learning_rate": 2.9907173851804204e-06, "loss": 1.5256, "step": 5606 }, { "epoch": 0.6429308565531475, "grad_norm": 16.375, "learning_rate": 2.9890171072497053e-06, "loss": 1.9126, "step": 5607 }, { "epoch": 0.6430455223024882, "grad_norm": 14.625, "learning_rate": 2.987317106695748e-06, "loss": 1.6653, "step": 5608 }, { "epoch": 0.643160188051829, "grad_norm": 14.5625, "learning_rate": 2.9856173837530287e-06, "loss": 1.8715, "step": 5609 }, { "epoch": 0.6432748538011696, "grad_norm": 11.75, "learning_rate": 2.9839179386559957e-06, "loss": 1.5794, "step": 5610 }, { "epoch": 0.6433895195505103, "grad_norm": 13.6875, "learning_rate": 2.982218771639053e-06, "loss": 2.0788, "step": 5611 }, { "epoch": 0.643504185299851, "grad_norm": 13.0, "learning_rate": 2.9805198829365677e-06, "loss": 1.5781, "step": 5612 }, { "epoch": 0.6436188510491916, "grad_norm": 18.625, "learning_rate": 2.9788212727828705e-06, "loss": 1.9259, "step": 5613 }, { "epoch": 0.6437335167985323, "grad_norm": 26.375, "learning_rate": 2.977122941412249e-06, "loss": 1.853, "step": 5614 }, { "epoch": 0.6438481825478729, "grad_norm": 15.8125, "learning_rate": 2.975424889058961e-06, "loss": 1.9023, "step": 5615 }, { "epoch": 0.6439628482972136, "grad_norm": 15.5, "learning_rate": 2.9737271159572177e-06, "loss": 1.9361, "step": 5616 }, { "epoch": 0.6440775140465543, "grad_norm": 15.5, "learning_rate": 2.972029622341193e-06, "loss": 1.8738, "step": 5617 }, { "epoch": 0.6441921797958949, "grad_norm": 16.625, "learning_rate": 2.970332408445027e-06, "loss": 2.0795, "step": 5618 }, { "epoch": 0.6443068455452357, "grad_norm": 34.5, "learning_rate": 2.9686354745028136e-06, "loss": 1.8153, "step": 5619 }, { "epoch": 0.6444215112945763, "grad_norm": 15.1875, "learning_rate": 2.966938820748616e-06, "loss": 1.6797, "step": 5620 }, { "epoch": 0.644536177043917, "grad_norm": 31.125, "learning_rate": 2.965242447416456e-06, "loss": 2.0878, "step": 5621 }, { "epoch": 0.6446508427932577, "grad_norm": 18.25, "learning_rate": 2.9635463547403143e-06, "loss": 1.8966, "step": 5622 }, { "epoch": 0.6447655085425983, "grad_norm": 11.9375, "learning_rate": 2.9618505429541334e-06, "loss": 1.7896, "step": 5623 }, { "epoch": 0.644880174291939, "grad_norm": 21.125, "learning_rate": 2.9601550122918184e-06, "loss": 1.8483, "step": 5624 }, { "epoch": 0.6449948400412797, "grad_norm": 14.1875, "learning_rate": 2.9584597629872386e-06, "loss": 1.8602, "step": 5625 }, { "epoch": 0.6451095057906203, "grad_norm": 13.4375, "learning_rate": 2.9567647952742177e-06, "loss": 1.533, "step": 5626 }, { "epoch": 0.645224171539961, "grad_norm": 11.9375, "learning_rate": 2.9550701093865475e-06, "loss": 1.6316, "step": 5627 }, { "epoch": 0.6453388372893016, "grad_norm": 14.25, "learning_rate": 2.9533757055579756e-06, "loss": 1.8455, "step": 5628 }, { "epoch": 0.6454535030386424, "grad_norm": 16.125, "learning_rate": 2.9516815840222106e-06, "loss": 1.823, "step": 5629 }, { "epoch": 0.6455681687879831, "grad_norm": 33.75, "learning_rate": 2.949987745012928e-06, "loss": 1.8347, "step": 5630 }, { "epoch": 0.6456828345373237, "grad_norm": 17.875, "learning_rate": 2.9482941887637593e-06, "loss": 1.5614, "step": 5631 }, { "epoch": 0.6457975002866644, "grad_norm": 12.6875, "learning_rate": 2.946600915508298e-06, "loss": 1.5716, "step": 5632 }, { "epoch": 0.6459121660360051, "grad_norm": 15.25, "learning_rate": 2.9449079254800998e-06, "loss": 1.8193, "step": 5633 }, { "epoch": 0.6460268317853457, "grad_norm": 14.375, "learning_rate": 2.9432152189126784e-06, "loss": 1.7501, "step": 5634 }, { "epoch": 0.6461414975346864, "grad_norm": 13.5, "learning_rate": 2.9415227960395126e-06, "loss": 1.5839, "step": 5635 }, { "epoch": 0.646256163284027, "grad_norm": 14.25, "learning_rate": 2.9398306570940384e-06, "loss": 1.9513, "step": 5636 }, { "epoch": 0.6463708290333677, "grad_norm": 15.5, "learning_rate": 2.9381388023096556e-06, "loss": 2.0757, "step": 5637 }, { "epoch": 0.6464854947827084, "grad_norm": 11.875, "learning_rate": 2.9364472319197214e-06, "loss": 1.5319, "step": 5638 }, { "epoch": 0.646600160532049, "grad_norm": 12.6875, "learning_rate": 2.934755946157556e-06, "loss": 1.5935, "step": 5639 }, { "epoch": 0.6467148262813898, "grad_norm": 11.4375, "learning_rate": 2.933064945256442e-06, "loss": 1.5378, "step": 5640 }, { "epoch": 0.6468294920307304, "grad_norm": 13.75, "learning_rate": 2.9313742294496195e-06, "loss": 1.3973, "step": 5641 }, { "epoch": 0.6469441577800711, "grad_norm": 13.375, "learning_rate": 2.9296837989702898e-06, "loss": 1.6617, "step": 5642 }, { "epoch": 0.6470588235294118, "grad_norm": 12.4375, "learning_rate": 2.9279936540516176e-06, "loss": 1.5141, "step": 5643 }, { "epoch": 0.6471734892787524, "grad_norm": 18.75, "learning_rate": 2.926303794926722e-06, "loss": 1.658, "step": 5644 }, { "epoch": 0.6472881550280931, "grad_norm": 16.25, "learning_rate": 2.924614221828691e-06, "loss": 1.8963, "step": 5645 }, { "epoch": 0.6474028207774338, "grad_norm": 14.75, "learning_rate": 2.9229249349905686e-06, "loss": 2.1902, "step": 5646 }, { "epoch": 0.6475174865267744, "grad_norm": 16.75, "learning_rate": 2.921235934645359e-06, "loss": 1.9987, "step": 5647 }, { "epoch": 0.6476321522761151, "grad_norm": 13.9375, "learning_rate": 2.9195472210260258e-06, "loss": 1.5321, "step": 5648 }, { "epoch": 0.6477468180254558, "grad_norm": 14.0, "learning_rate": 2.9178587943654967e-06, "loss": 1.8993, "step": 5649 }, { "epoch": 0.6478614837747965, "grad_norm": 17.875, "learning_rate": 2.916170654896658e-06, "loss": 1.4585, "step": 5650 }, { "epoch": 0.6479761495241372, "grad_norm": 14.125, "learning_rate": 2.914482802852356e-06, "loss": 1.6207, "step": 5651 }, { "epoch": 0.6480908152734778, "grad_norm": 12.5625, "learning_rate": 2.912795238465399e-06, "loss": 1.9231, "step": 5652 }, { "epoch": 0.6482054810228185, "grad_norm": 13.75, "learning_rate": 2.9111079619685525e-06, "loss": 1.5986, "step": 5653 }, { "epoch": 0.6483201467721592, "grad_norm": 13.875, "learning_rate": 2.909420973594541e-06, "loss": 1.9966, "step": 5654 }, { "epoch": 0.6484348125214998, "grad_norm": 14.0625, "learning_rate": 2.9077342735760614e-06, "loss": 1.5523, "step": 5655 }, { "epoch": 0.6485494782708405, "grad_norm": 14.3125, "learning_rate": 2.9060478621457544e-06, "loss": 1.9674, "step": 5656 }, { "epoch": 0.6486641440201811, "grad_norm": 16.625, "learning_rate": 2.90436173953623e-06, "loss": 2.1949, "step": 5657 }, { "epoch": 0.6487788097695218, "grad_norm": 14.4375, "learning_rate": 2.90267590598006e-06, "loss": 2.0064, "step": 5658 }, { "epoch": 0.6488934755188626, "grad_norm": 12.6875, "learning_rate": 2.9009903617097652e-06, "loss": 1.575, "step": 5659 }, { "epoch": 0.6490081412682032, "grad_norm": 13.375, "learning_rate": 2.8993051069578414e-06, "loss": 1.6597, "step": 5660 }, { "epoch": 0.6491228070175439, "grad_norm": 17.875, "learning_rate": 2.8976201419567374e-06, "loss": 1.8872, "step": 5661 }, { "epoch": 0.6492374727668845, "grad_norm": 13.5625, "learning_rate": 2.8959354669388583e-06, "loss": 1.6037, "step": 5662 }, { "epoch": 0.6493521385162252, "grad_norm": 13.0, "learning_rate": 2.8942510821365737e-06, "loss": 2.0304, "step": 5663 }, { "epoch": 0.6494668042655659, "grad_norm": 15.6875, "learning_rate": 2.892566987782213e-06, "loss": 1.6952, "step": 5664 }, { "epoch": 0.6495814700149065, "grad_norm": 12.6875, "learning_rate": 2.8908831841080654e-06, "loss": 1.6907, "step": 5665 }, { "epoch": 0.6496961357642472, "grad_norm": 12.9375, "learning_rate": 2.8891996713463793e-06, "loss": 1.9392, "step": 5666 }, { "epoch": 0.6498108015135879, "grad_norm": 22.75, "learning_rate": 2.887516449729365e-06, "loss": 1.9654, "step": 5667 }, { "epoch": 0.6499254672629285, "grad_norm": 13.0625, "learning_rate": 2.885833519489186e-06, "loss": 1.5971, "step": 5668 }, { "epoch": 0.6500401330122693, "grad_norm": 15.1875, "learning_rate": 2.884150880857972e-06, "loss": 1.876, "step": 5669 }, { "epoch": 0.6501547987616099, "grad_norm": 13.1875, "learning_rate": 2.8824685340678168e-06, "loss": 1.6768, "step": 5670 }, { "epoch": 0.6502694645109506, "grad_norm": 37.5, "learning_rate": 2.8807864793507608e-06, "loss": 1.8421, "step": 5671 }, { "epoch": 0.6503841302602913, "grad_norm": 13.75, "learning_rate": 2.8791047169388143e-06, "loss": 1.8636, "step": 5672 }, { "epoch": 0.6504987960096319, "grad_norm": 12.6875, "learning_rate": 2.877423247063946e-06, "loss": 1.5627, "step": 5673 }, { "epoch": 0.6506134617589726, "grad_norm": 16.125, "learning_rate": 2.8757420699580763e-06, "loss": 1.5709, "step": 5674 }, { "epoch": 0.6507281275083132, "grad_norm": 14.5, "learning_rate": 2.874061185853099e-06, "loss": 1.8268, "step": 5675 }, { "epoch": 0.6508427932576539, "grad_norm": 10.9375, "learning_rate": 2.872380594980858e-06, "loss": 1.3891, "step": 5676 }, { "epoch": 0.6509574590069946, "grad_norm": 17.75, "learning_rate": 2.8707002975731567e-06, "loss": 1.8035, "step": 5677 }, { "epoch": 0.6510721247563352, "grad_norm": 14.375, "learning_rate": 2.8690202938617605e-06, "loss": 1.8272, "step": 5678 }, { "epoch": 0.651186790505676, "grad_norm": 14.0, "learning_rate": 2.8673405840783953e-06, "loss": 1.5965, "step": 5679 }, { "epoch": 0.6513014562550167, "grad_norm": 14.1875, "learning_rate": 2.865661168454744e-06, "loss": 1.5889, "step": 5680 }, { "epoch": 0.6514161220043573, "grad_norm": 13.8125, "learning_rate": 2.86398204722245e-06, "loss": 1.3791, "step": 5681 }, { "epoch": 0.651530787753698, "grad_norm": 13.8125, "learning_rate": 2.8623032206131184e-06, "loss": 2.1577, "step": 5682 }, { "epoch": 0.6516454535030386, "grad_norm": 36.25, "learning_rate": 2.8606246888583078e-06, "loss": 2.1177, "step": 5683 }, { "epoch": 0.6517601192523793, "grad_norm": 15.75, "learning_rate": 2.8589464521895416e-06, "loss": 2.0689, "step": 5684 }, { "epoch": 0.65187478500172, "grad_norm": 16.625, "learning_rate": 2.8572685108382993e-06, "loss": 1.4664, "step": 5685 }, { "epoch": 0.6519894507510606, "grad_norm": 12.9375, "learning_rate": 2.855590865036022e-06, "loss": 1.6727, "step": 5686 }, { "epoch": 0.6521041165004013, "grad_norm": 14.1875, "learning_rate": 2.8539135150141085e-06, "loss": 1.7268, "step": 5687 }, { "epoch": 0.6522187822497421, "grad_norm": 13.0625, "learning_rate": 2.8522364610039192e-06, "loss": 1.4045, "step": 5688 }, { "epoch": 0.6523334479990827, "grad_norm": 14.125, "learning_rate": 2.8505597032367664e-06, "loss": 1.6503, "step": 5689 }, { "epoch": 0.6524481137484234, "grad_norm": 14.375, "learning_rate": 2.848883241943935e-06, "loss": 1.7725, "step": 5690 }, { "epoch": 0.652562779497764, "grad_norm": 12.8125, "learning_rate": 2.847207077356654e-06, "loss": 1.6484, "step": 5691 }, { "epoch": 0.6526774452471047, "grad_norm": 14.4375, "learning_rate": 2.8455312097061207e-06, "loss": 1.9098, "step": 5692 }, { "epoch": 0.6527921109964454, "grad_norm": 13.3125, "learning_rate": 2.8438556392234885e-06, "loss": 1.9597, "step": 5693 }, { "epoch": 0.652906776745786, "grad_norm": 15.8125, "learning_rate": 2.8421803661398716e-06, "loss": 1.8415, "step": 5694 }, { "epoch": 0.6530214424951267, "grad_norm": 11.75, "learning_rate": 2.840505390686341e-06, "loss": 1.8186, "step": 5695 }, { "epoch": 0.6531361082444673, "grad_norm": 16.625, "learning_rate": 2.8388307130939306e-06, "loss": 1.8803, "step": 5696 }, { "epoch": 0.653250773993808, "grad_norm": 14.375, "learning_rate": 2.8371563335936253e-06, "loss": 1.6172, "step": 5697 }, { "epoch": 0.6533654397431488, "grad_norm": 14.25, "learning_rate": 2.835482252416376e-06, "loss": 1.7428, "step": 5698 }, { "epoch": 0.6534801054924894, "grad_norm": 13.8125, "learning_rate": 2.8338084697930913e-06, "loss": 1.6892, "step": 5699 }, { "epoch": 0.6535947712418301, "grad_norm": 13.1875, "learning_rate": 2.8321349859546368e-06, "loss": 1.7366, "step": 5700 }, { "epoch": 0.6537094369911708, "grad_norm": 17.75, "learning_rate": 2.8304618011318374e-06, "loss": 1.5331, "step": 5701 }, { "epoch": 0.6538241027405114, "grad_norm": 15.875, "learning_rate": 2.828788915555479e-06, "loss": 1.8607, "step": 5702 }, { "epoch": 0.6539387684898521, "grad_norm": 16.375, "learning_rate": 2.8271163294563007e-06, "loss": 1.4132, "step": 5703 }, { "epoch": 0.6540534342391927, "grad_norm": 22.5, "learning_rate": 2.825444043065004e-06, "loss": 1.8012, "step": 5704 }, { "epoch": 0.6541680999885334, "grad_norm": 13.1875, "learning_rate": 2.823772056612255e-06, "loss": 1.5697, "step": 5705 }, { "epoch": 0.6542827657378741, "grad_norm": 14.6875, "learning_rate": 2.8221003703286665e-06, "loss": 1.6573, "step": 5706 }, { "epoch": 0.6543974314872147, "grad_norm": 19.375, "learning_rate": 2.820428984444816e-06, "loss": 1.5121, "step": 5707 }, { "epoch": 0.6545120972365555, "grad_norm": 15.125, "learning_rate": 2.818757899191244e-06, "loss": 1.8052, "step": 5708 }, { "epoch": 0.6546267629858961, "grad_norm": 15.875, "learning_rate": 2.8170871147984367e-06, "loss": 1.871, "step": 5709 }, { "epoch": 0.6547414287352368, "grad_norm": 15.0, "learning_rate": 2.8154166314968543e-06, "loss": 1.5962, "step": 5710 }, { "epoch": 0.6548560944845775, "grad_norm": 16.0, "learning_rate": 2.8137464495169075e-06, "loss": 1.4159, "step": 5711 }, { "epoch": 0.6549707602339181, "grad_norm": 15.1875, "learning_rate": 2.8120765690889623e-06, "loss": 1.627, "step": 5712 }, { "epoch": 0.6550854259832588, "grad_norm": 14.8125, "learning_rate": 2.8104069904433484e-06, "loss": 1.7899, "step": 5713 }, { "epoch": 0.6552000917325995, "grad_norm": 15.0, "learning_rate": 2.8087377138103532e-06, "loss": 1.9991, "step": 5714 }, { "epoch": 0.6553147574819401, "grad_norm": 14.3125, "learning_rate": 2.807068739420221e-06, "loss": 1.6188, "step": 5715 }, { "epoch": 0.6554294232312808, "grad_norm": 16.5, "learning_rate": 2.805400067503155e-06, "loss": 1.8611, "step": 5716 }, { "epoch": 0.6555440889806214, "grad_norm": 16.875, "learning_rate": 2.8037316982893193e-06, "loss": 2.0407, "step": 5717 }, { "epoch": 0.6556587547299622, "grad_norm": 13.625, "learning_rate": 2.8020636320088287e-06, "loss": 1.7157, "step": 5718 }, { "epoch": 0.6557734204793029, "grad_norm": 14.5625, "learning_rate": 2.8003958688917615e-06, "loss": 2.1019, "step": 5719 }, { "epoch": 0.6558880862286435, "grad_norm": 12.9375, "learning_rate": 2.7987284091681598e-06, "loss": 1.5944, "step": 5720 }, { "epoch": 0.6560027519779842, "grad_norm": 12.125, "learning_rate": 2.797061253068012e-06, "loss": 1.4991, "step": 5721 }, { "epoch": 0.6561174177273249, "grad_norm": 12.75, "learning_rate": 2.795394400821273e-06, "loss": 1.7186, "step": 5722 }, { "epoch": 0.6562320834766655, "grad_norm": 12.5625, "learning_rate": 2.7937278526578548e-06, "loss": 1.752, "step": 5723 }, { "epoch": 0.6563467492260062, "grad_norm": 17.75, "learning_rate": 2.7920616088076192e-06, "loss": 1.6758, "step": 5724 }, { "epoch": 0.6564614149753468, "grad_norm": 16.375, "learning_rate": 2.7903956695004002e-06, "loss": 1.7043, "step": 5725 }, { "epoch": 0.6565760807246875, "grad_norm": 15.0625, "learning_rate": 2.7887300349659818e-06, "loss": 1.6517, "step": 5726 }, { "epoch": 0.6566907464740283, "grad_norm": 17.0, "learning_rate": 2.7870647054341016e-06, "loss": 1.5058, "step": 5727 }, { "epoch": 0.6568054122233689, "grad_norm": 30.625, "learning_rate": 2.785399681134464e-06, "loss": 1.7955, "step": 5728 }, { "epoch": 0.6569200779727096, "grad_norm": 16.375, "learning_rate": 2.783734962296726e-06, "loss": 2.0296, "step": 5729 }, { "epoch": 0.6570347437220502, "grad_norm": 14.9375, "learning_rate": 2.7820705491505027e-06, "loss": 1.4795, "step": 5730 }, { "epoch": 0.6571494094713909, "grad_norm": 16.875, "learning_rate": 2.7804064419253713e-06, "loss": 1.6444, "step": 5731 }, { "epoch": 0.6572640752207316, "grad_norm": 14.5625, "learning_rate": 2.7787426408508635e-06, "loss": 1.9772, "step": 5732 }, { "epoch": 0.6573787409700722, "grad_norm": 11.875, "learning_rate": 2.777079146156465e-06, "loss": 1.4887, "step": 5733 }, { "epoch": 0.6574934067194129, "grad_norm": 16.875, "learning_rate": 2.7754159580716257e-06, "loss": 2.1331, "step": 5734 }, { "epoch": 0.6576080724687536, "grad_norm": 24.625, "learning_rate": 2.7737530768257505e-06, "loss": 1.8801, "step": 5735 }, { "epoch": 0.6577227382180942, "grad_norm": 14.1875, "learning_rate": 2.7720905026482027e-06, "loss": 1.5644, "step": 5736 }, { "epoch": 0.657837403967435, "grad_norm": 13.75, "learning_rate": 2.7704282357683015e-06, "loss": 1.8776, "step": 5737 }, { "epoch": 0.6579520697167756, "grad_norm": 12.875, "learning_rate": 2.768766276415328e-06, "loss": 1.5057, "step": 5738 }, { "epoch": 0.6580667354661163, "grad_norm": 16.5, "learning_rate": 2.7671046248185116e-06, "loss": 1.5367, "step": 5739 }, { "epoch": 0.658181401215457, "grad_norm": 17.875, "learning_rate": 2.765443281207053e-06, "loss": 1.676, "step": 5740 }, { "epoch": 0.6582960669647976, "grad_norm": 14.6875, "learning_rate": 2.7637822458100976e-06, "loss": 1.7636, "step": 5741 }, { "epoch": 0.6584107327141383, "grad_norm": 12.5625, "learning_rate": 2.7621215188567553e-06, "loss": 1.59, "step": 5742 }, { "epoch": 0.6585253984634789, "grad_norm": 11.6875, "learning_rate": 2.7604611005760917e-06, "loss": 1.3912, "step": 5743 }, { "epoch": 0.6586400642128196, "grad_norm": 11.3125, "learning_rate": 2.758800991197129e-06, "loss": 1.7329, "step": 5744 }, { "epoch": 0.6587547299621603, "grad_norm": 13.125, "learning_rate": 2.757141190948849e-06, "loss": 2.0227, "step": 5745 }, { "epoch": 0.6588693957115009, "grad_norm": 17.5, "learning_rate": 2.755481700060191e-06, "loss": 1.9691, "step": 5746 }, { "epoch": 0.6589840614608417, "grad_norm": 14.9375, "learning_rate": 2.753822518760046e-06, "loss": 1.7408, "step": 5747 }, { "epoch": 0.6590987272101824, "grad_norm": 12.9375, "learning_rate": 2.7521636472772682e-06, "loss": 1.6572, "step": 5748 }, { "epoch": 0.659213392959523, "grad_norm": 16.75, "learning_rate": 2.750505085840668e-06, "loss": 2.181, "step": 5749 }, { "epoch": 0.6593280587088637, "grad_norm": 14.25, "learning_rate": 2.7488468346790122e-06, "loss": 1.5672, "step": 5750 }, { "epoch": 0.6594427244582043, "grad_norm": 15.0625, "learning_rate": 2.747188894021024e-06, "loss": 2.0737, "step": 5751 }, { "epoch": 0.659557390207545, "grad_norm": 15.0, "learning_rate": 2.7455312640953875e-06, "loss": 1.7568, "step": 5752 }, { "epoch": 0.6596720559568857, "grad_norm": 27.25, "learning_rate": 2.7438739451307373e-06, "loss": 1.5783, "step": 5753 }, { "epoch": 0.6597867217062263, "grad_norm": 15.25, "learning_rate": 2.7422169373556684e-06, "loss": 1.4587, "step": 5754 }, { "epoch": 0.659901387455567, "grad_norm": 16.125, "learning_rate": 2.7405602409987398e-06, "loss": 1.6795, "step": 5755 }, { "epoch": 0.6600160532049077, "grad_norm": 14.5, "learning_rate": 2.738903856288455e-06, "loss": 1.6685, "step": 5756 }, { "epoch": 0.6601307189542484, "grad_norm": 16.125, "learning_rate": 2.737247783453283e-06, "loss": 1.8511, "step": 5757 }, { "epoch": 0.6602453847035891, "grad_norm": 18.625, "learning_rate": 2.7355920227216494e-06, "loss": 2.1223, "step": 5758 }, { "epoch": 0.6603600504529297, "grad_norm": 15.8125, "learning_rate": 2.7339365743219285e-06, "loss": 1.7557, "step": 5759 }, { "epoch": 0.6604747162022704, "grad_norm": 16.75, "learning_rate": 2.7322814384824648e-06, "loss": 1.7045, "step": 5760 }, { "epoch": 0.6605893819516111, "grad_norm": 17.625, "learning_rate": 2.730626615431551e-06, "loss": 1.8348, "step": 5761 }, { "epoch": 0.6607040477009517, "grad_norm": 12.9375, "learning_rate": 2.728972105397436e-06, "loss": 1.6543, "step": 5762 }, { "epoch": 0.6608187134502924, "grad_norm": 12.9375, "learning_rate": 2.727317908608329e-06, "loss": 1.8648, "step": 5763 }, { "epoch": 0.660933379199633, "grad_norm": 13.9375, "learning_rate": 2.7256640252923956e-06, "loss": 1.9177, "step": 5764 }, { "epoch": 0.6610480449489737, "grad_norm": 14.8125, "learning_rate": 2.724010455677757e-06, "loss": 1.4196, "step": 5765 }, { "epoch": 0.6611627106983144, "grad_norm": 12.5, "learning_rate": 2.722357199992491e-06, "loss": 1.8457, "step": 5766 }, { "epoch": 0.661277376447655, "grad_norm": 17.375, "learning_rate": 2.7207042584646366e-06, "loss": 1.8496, "step": 5767 }, { "epoch": 0.6613920421969958, "grad_norm": 12.8125, "learning_rate": 2.71905163132218e-06, "loss": 1.9323, "step": 5768 }, { "epoch": 0.6615067079463365, "grad_norm": 13.1875, "learning_rate": 2.7173993187930696e-06, "loss": 2.1355, "step": 5769 }, { "epoch": 0.6616213736956771, "grad_norm": 14.5625, "learning_rate": 2.7157473211052177e-06, "loss": 1.8416, "step": 5770 }, { "epoch": 0.6617360394450178, "grad_norm": 13.375, "learning_rate": 2.714095638486478e-06, "loss": 1.6587, "step": 5771 }, { "epoch": 0.6618507051943584, "grad_norm": 13.875, "learning_rate": 2.7124442711646728e-06, "loss": 2.0246, "step": 5772 }, { "epoch": 0.6619653709436991, "grad_norm": 33.25, "learning_rate": 2.7107932193675767e-06, "loss": 2.1513, "step": 5773 }, { "epoch": 0.6620800366930398, "grad_norm": 13.75, "learning_rate": 2.7091424833229157e-06, "loss": 1.7733, "step": 5774 }, { "epoch": 0.6621947024423804, "grad_norm": 12.5625, "learning_rate": 2.707492063258384e-06, "loss": 1.6617, "step": 5775 }, { "epoch": 0.6623093681917211, "grad_norm": 14.1875, "learning_rate": 2.705841959401625e-06, "loss": 1.6367, "step": 5776 }, { "epoch": 0.6624240339410618, "grad_norm": 14.125, "learning_rate": 2.7041921719802343e-06, "loss": 1.6362, "step": 5777 }, { "epoch": 0.6625386996904025, "grad_norm": 31.75, "learning_rate": 2.702542701221772e-06, "loss": 1.5794, "step": 5778 }, { "epoch": 0.6626533654397432, "grad_norm": 17.75, "learning_rate": 2.70089354735375e-06, "loss": 2.0848, "step": 5779 }, { "epoch": 0.6627680311890838, "grad_norm": 12.1875, "learning_rate": 2.699244710603639e-06, "loss": 1.544, "step": 5780 }, { "epoch": 0.6628826969384245, "grad_norm": 13.9375, "learning_rate": 2.6975961911988635e-06, "loss": 1.2948, "step": 5781 }, { "epoch": 0.6629973626877652, "grad_norm": 12.125, "learning_rate": 2.6959479893668074e-06, "loss": 1.6014, "step": 5782 }, { "epoch": 0.6631120284371058, "grad_norm": 12.9375, "learning_rate": 2.694300105334805e-06, "loss": 1.6454, "step": 5783 }, { "epoch": 0.6632266941864465, "grad_norm": 36.0, "learning_rate": 2.6926525393301527e-06, "loss": 1.9582, "step": 5784 }, { "epoch": 0.6633413599357871, "grad_norm": 17.375, "learning_rate": 2.6910052915801007e-06, "loss": 2.011, "step": 5785 }, { "epoch": 0.6634560256851278, "grad_norm": 15.1875, "learning_rate": 2.6893583623118546e-06, "loss": 1.5055, "step": 5786 }, { "epoch": 0.6635706914344686, "grad_norm": 14.125, "learning_rate": 2.687711751752578e-06, "loss": 1.92, "step": 5787 }, { "epoch": 0.6636853571838092, "grad_norm": 14.9375, "learning_rate": 2.686065460129391e-06, "loss": 1.595, "step": 5788 }, { "epoch": 0.6638000229331499, "grad_norm": 10.8125, "learning_rate": 2.684419487669362e-06, "loss": 1.3515, "step": 5789 }, { "epoch": 0.6639146886824906, "grad_norm": 17.25, "learning_rate": 2.68277383459953e-06, "loss": 2.0112, "step": 5790 }, { "epoch": 0.6640293544318312, "grad_norm": 12.0625, "learning_rate": 2.6811285011468747e-06, "loss": 1.6712, "step": 5791 }, { "epoch": 0.6641440201811719, "grad_norm": 11.4375, "learning_rate": 2.6794834875383414e-06, "loss": 1.5557, "step": 5792 }, { "epoch": 0.6642586859305125, "grad_norm": 14.0625, "learning_rate": 2.677838794000828e-06, "loss": 1.9179, "step": 5793 }, { "epoch": 0.6643733516798532, "grad_norm": 11.875, "learning_rate": 2.676194420761188e-06, "loss": 1.9099, "step": 5794 }, { "epoch": 0.664488017429194, "grad_norm": 16.125, "learning_rate": 2.6745503680462325e-06, "loss": 1.893, "step": 5795 }, { "epoch": 0.6646026831785345, "grad_norm": 14.875, "learning_rate": 2.6729066360827282e-06, "loss": 1.5702, "step": 5796 }, { "epoch": 0.6647173489278753, "grad_norm": 31.75, "learning_rate": 2.671263225097393e-06, "loss": 2.1178, "step": 5797 }, { "epoch": 0.6648320146772159, "grad_norm": 13.0, "learning_rate": 2.6696201353169065e-06, "loss": 1.5415, "step": 5798 }, { "epoch": 0.6649466804265566, "grad_norm": 12.4375, "learning_rate": 2.667977366967901e-06, "loss": 1.872, "step": 5799 }, { "epoch": 0.6650613461758973, "grad_norm": 16.125, "learning_rate": 2.666334920276965e-06, "loss": 2.0085, "step": 5800 }, { "epoch": 0.6651760119252379, "grad_norm": 13.6875, "learning_rate": 2.6646927954706436e-06, "loss": 1.5197, "step": 5801 }, { "epoch": 0.6652906776745786, "grad_norm": 14.8125, "learning_rate": 2.663050992775438e-06, "loss": 1.8812, "step": 5802 }, { "epoch": 0.6654053434239193, "grad_norm": 16.0, "learning_rate": 2.6614095124177997e-06, "loss": 1.9157, "step": 5803 }, { "epoch": 0.6655200091732599, "grad_norm": 14.875, "learning_rate": 2.6597683546241393e-06, "loss": 1.7432, "step": 5804 }, { "epoch": 0.6656346749226006, "grad_norm": 12.1875, "learning_rate": 2.6581275196208294e-06, "loss": 1.6561, "step": 5805 }, { "epoch": 0.6657493406719412, "grad_norm": 29.0, "learning_rate": 2.6564870076341866e-06, "loss": 1.9163, "step": 5806 }, { "epoch": 0.665864006421282, "grad_norm": 13.875, "learning_rate": 2.6548468188904893e-06, "loss": 1.9586, "step": 5807 }, { "epoch": 0.6659786721706227, "grad_norm": 14.9375, "learning_rate": 2.653206953615972e-06, "loss": 1.6514, "step": 5808 }, { "epoch": 0.6660933379199633, "grad_norm": 14.8125, "learning_rate": 2.6515674120368185e-06, "loss": 1.9941, "step": 5809 }, { "epoch": 0.666208003669304, "grad_norm": 28.5, "learning_rate": 2.649928194379177e-06, "loss": 2.0725, "step": 5810 }, { "epoch": 0.6663226694186446, "grad_norm": 12.5, "learning_rate": 2.6482893008691467e-06, "loss": 1.5026, "step": 5811 }, { "epoch": 0.6664373351679853, "grad_norm": 17.5, "learning_rate": 2.646650731732777e-06, "loss": 1.7784, "step": 5812 }, { "epoch": 0.666552000917326, "grad_norm": 17.375, "learning_rate": 2.6450124871960807e-06, "loss": 2.1371, "step": 5813 }, { "epoch": 0.6666666666666666, "grad_norm": 34.0, "learning_rate": 2.643374567485022e-06, "loss": 1.8509, "step": 5814 }, { "epoch": 0.6667813324160073, "grad_norm": 14.6875, "learning_rate": 2.6417369728255195e-06, "loss": 1.8988, "step": 5815 }, { "epoch": 0.6668959981653481, "grad_norm": 17.125, "learning_rate": 2.6400997034434495e-06, "loss": 2.3326, "step": 5816 }, { "epoch": 0.6670106639146887, "grad_norm": 12.625, "learning_rate": 2.6384627595646427e-06, "loss": 1.6034, "step": 5817 }, { "epoch": 0.6671253296640294, "grad_norm": 14.5, "learning_rate": 2.636826141414882e-06, "loss": 2.0688, "step": 5818 }, { "epoch": 0.66723999541337, "grad_norm": 13.1875, "learning_rate": 2.6351898492199072e-06, "loss": 1.5396, "step": 5819 }, { "epoch": 0.6673546611627107, "grad_norm": 14.4375, "learning_rate": 2.6335538832054192e-06, "loss": 1.6038, "step": 5820 }, { "epoch": 0.6674693269120514, "grad_norm": 14.125, "learning_rate": 2.6319182435970625e-06, "loss": 1.8735, "step": 5821 }, { "epoch": 0.667583992661392, "grad_norm": 11.5625, "learning_rate": 2.6302829306204438e-06, "loss": 1.4612, "step": 5822 }, { "epoch": 0.6676986584107327, "grad_norm": 13.4375, "learning_rate": 2.628647944501126e-06, "loss": 1.8313, "step": 5823 }, { "epoch": 0.6678133241600734, "grad_norm": 18.875, "learning_rate": 2.627013285464618e-06, "loss": 2.1726, "step": 5824 }, { "epoch": 0.667927989909414, "grad_norm": 15.6875, "learning_rate": 2.6253789537363966e-06, "loss": 1.7508, "step": 5825 }, { "epoch": 0.6680426556587548, "grad_norm": 27.75, "learning_rate": 2.623744949541886e-06, "loss": 1.9299, "step": 5826 }, { "epoch": 0.6681573214080954, "grad_norm": 16.5, "learning_rate": 2.6221112731064623e-06, "loss": 1.7763, "step": 5827 }, { "epoch": 0.6682719871574361, "grad_norm": 17.0, "learning_rate": 2.6204779246554613e-06, "loss": 1.7296, "step": 5828 }, { "epoch": 0.6683866529067768, "grad_norm": 16.75, "learning_rate": 2.618844904414173e-06, "loss": 1.726, "step": 5829 }, { "epoch": 0.6685013186561174, "grad_norm": 14.6875, "learning_rate": 2.617212212607841e-06, "loss": 2.0005, "step": 5830 }, { "epoch": 0.6686159844054581, "grad_norm": 13.9375, "learning_rate": 2.6155798494616648e-06, "loss": 1.8973, "step": 5831 }, { "epoch": 0.6687306501547987, "grad_norm": 15.875, "learning_rate": 2.613947815200798e-06, "loss": 1.7109, "step": 5832 }, { "epoch": 0.6688453159041394, "grad_norm": 12.0, "learning_rate": 2.612316110050347e-06, "loss": 1.7828, "step": 5833 }, { "epoch": 0.6689599816534801, "grad_norm": 19.75, "learning_rate": 2.6106847342353745e-06, "loss": 1.8936, "step": 5834 }, { "epoch": 0.6690746474028207, "grad_norm": 17.125, "learning_rate": 2.609053687980899e-06, "loss": 1.7356, "step": 5835 }, { "epoch": 0.6691893131521615, "grad_norm": 12.8125, "learning_rate": 2.607422971511891e-06, "loss": 1.8114, "step": 5836 }, { "epoch": 0.6693039789015022, "grad_norm": 15.875, "learning_rate": 2.6057925850532777e-06, "loss": 1.6929, "step": 5837 }, { "epoch": 0.6694186446508428, "grad_norm": 16.25, "learning_rate": 2.6041625288299415e-06, "loss": 1.7607, "step": 5838 }, { "epoch": 0.6695333104001835, "grad_norm": 16.125, "learning_rate": 2.602532803066712e-06, "loss": 2.0769, "step": 5839 }, { "epoch": 0.6696479761495241, "grad_norm": 32.25, "learning_rate": 2.600903407988386e-06, "loss": 2.2298, "step": 5840 }, { "epoch": 0.6697626418988648, "grad_norm": 16.125, "learning_rate": 2.599274343819702e-06, "loss": 1.9125, "step": 5841 }, { "epoch": 0.6698773076482055, "grad_norm": 31.875, "learning_rate": 2.5976456107853605e-06, "loss": 1.6815, "step": 5842 }, { "epoch": 0.6699919733975461, "grad_norm": 14.75, "learning_rate": 2.5960172091100133e-06, "loss": 1.5844, "step": 5843 }, { "epoch": 0.6701066391468868, "grad_norm": 17.0, "learning_rate": 2.594389139018269e-06, "loss": 1.9247, "step": 5844 }, { "epoch": 0.6702213048962274, "grad_norm": 16.375, "learning_rate": 2.5927614007346873e-06, "loss": 1.8134, "step": 5845 }, { "epoch": 0.6703359706455682, "grad_norm": 12.5625, "learning_rate": 2.5911339944837866e-06, "loss": 1.8316, "step": 5846 }, { "epoch": 0.6704506363949089, "grad_norm": 12.9375, "learning_rate": 2.5895069204900315e-06, "loss": 1.6585, "step": 5847 }, { "epoch": 0.6705653021442495, "grad_norm": 16.5, "learning_rate": 2.587880178977849e-06, "loss": 1.4274, "step": 5848 }, { "epoch": 0.6706799678935902, "grad_norm": 22.125, "learning_rate": 2.586253770171617e-06, "loss": 2.1346, "step": 5849 }, { "epoch": 0.6707946336429309, "grad_norm": 13.75, "learning_rate": 2.5846276942956662e-06, "loss": 1.7972, "step": 5850 }, { "epoch": 0.6709092993922715, "grad_norm": 13.5625, "learning_rate": 2.583001951574284e-06, "loss": 1.7366, "step": 5851 }, { "epoch": 0.6710239651416122, "grad_norm": 14.0, "learning_rate": 2.5813765422317133e-06, "loss": 1.7374, "step": 5852 }, { "epoch": 0.6711386308909528, "grad_norm": 15.125, "learning_rate": 2.579751466492143e-06, "loss": 1.5108, "step": 5853 }, { "epoch": 0.6712532966402935, "grad_norm": 13.0, "learning_rate": 2.5781267245797215e-06, "loss": 1.707, "step": 5854 }, { "epoch": 0.6713679623896343, "grad_norm": 14.8125, "learning_rate": 2.5765023167185577e-06, "loss": 1.6158, "step": 5855 }, { "epoch": 0.6714826281389749, "grad_norm": 14.875, "learning_rate": 2.5748782431327013e-06, "loss": 1.9341, "step": 5856 }, { "epoch": 0.6715972938883156, "grad_norm": 14.75, "learning_rate": 2.573254504046163e-06, "loss": 1.7375, "step": 5857 }, { "epoch": 0.6717119596376563, "grad_norm": 10.8125, "learning_rate": 2.5716310996829096e-06, "loss": 1.7552, "step": 5858 }, { "epoch": 0.6718266253869969, "grad_norm": 18.75, "learning_rate": 2.570008030266856e-06, "loss": 1.9924, "step": 5859 }, { "epoch": 0.6719412911363376, "grad_norm": 15.1875, "learning_rate": 2.5683852960218745e-06, "loss": 1.4688, "step": 5860 }, { "epoch": 0.6720559568856782, "grad_norm": 15.125, "learning_rate": 2.566762897171794e-06, "loss": 1.906, "step": 5861 }, { "epoch": 0.6721706226350189, "grad_norm": 14.4375, "learning_rate": 2.565140833940387e-06, "loss": 1.8253, "step": 5862 }, { "epoch": 0.6722852883843596, "grad_norm": 15.8125, "learning_rate": 2.56351910655139e-06, "loss": 2.1074, "step": 5863 }, { "epoch": 0.6723999541337002, "grad_norm": 18.375, "learning_rate": 2.5618977152284874e-06, "loss": 2.0683, "step": 5864 }, { "epoch": 0.672514619883041, "grad_norm": 16.5, "learning_rate": 2.5602766601953223e-06, "loss": 1.4748, "step": 5865 }, { "epoch": 0.6726292856323816, "grad_norm": 14.1875, "learning_rate": 2.5586559416754854e-06, "loss": 1.9289, "step": 5866 }, { "epoch": 0.6727439513817223, "grad_norm": 15.6875, "learning_rate": 2.557035559892527e-06, "loss": 1.4243, "step": 5867 }, { "epoch": 0.672858617131063, "grad_norm": 13.0625, "learning_rate": 2.5554155150699444e-06, "loss": 1.5493, "step": 5868 }, { "epoch": 0.6729732828804036, "grad_norm": 17.125, "learning_rate": 2.553795807431192e-06, "loss": 2.1385, "step": 5869 }, { "epoch": 0.6730879486297443, "grad_norm": 13.625, "learning_rate": 2.552176437199682e-06, "loss": 1.6165, "step": 5870 }, { "epoch": 0.673202614379085, "grad_norm": 17.25, "learning_rate": 2.5505574045987712e-06, "loss": 2.0949, "step": 5871 }, { "epoch": 0.6733172801284256, "grad_norm": 14.1875, "learning_rate": 2.548938709851776e-06, "loss": 1.675, "step": 5872 }, { "epoch": 0.6734319458777663, "grad_norm": 33.5, "learning_rate": 2.5473203531819657e-06, "loss": 1.714, "step": 5873 }, { "epoch": 0.6735466116271069, "grad_norm": 12.25, "learning_rate": 2.545702334812557e-06, "loss": 1.6524, "step": 5874 }, { "epoch": 0.6736612773764477, "grad_norm": 17.625, "learning_rate": 2.5440846549667297e-06, "loss": 1.4224, "step": 5875 }, { "epoch": 0.6737759431257884, "grad_norm": 14.1875, "learning_rate": 2.5424673138676126e-06, "loss": 2.1378, "step": 5876 }, { "epoch": 0.673890608875129, "grad_norm": 12.0625, "learning_rate": 2.540850311738282e-06, "loss": 1.6124, "step": 5877 }, { "epoch": 0.6740052746244697, "grad_norm": 12.875, "learning_rate": 2.5392336488017767e-06, "loss": 1.7771, "step": 5878 }, { "epoch": 0.6741199403738103, "grad_norm": 12.0625, "learning_rate": 2.5376173252810817e-06, "loss": 1.6252, "step": 5879 }, { "epoch": 0.674234606123151, "grad_norm": 15.8125, "learning_rate": 2.5360013413991398e-06, "loss": 1.8808, "step": 5880 }, { "epoch": 0.6743492718724917, "grad_norm": 13.5, "learning_rate": 2.534385697378845e-06, "loss": 1.926, "step": 5881 }, { "epoch": 0.6744639376218323, "grad_norm": 15.875, "learning_rate": 2.532770393443046e-06, "loss": 1.7449, "step": 5882 }, { "epoch": 0.674578603371173, "grad_norm": 17.375, "learning_rate": 2.531155429814539e-06, "loss": 1.7851, "step": 5883 }, { "epoch": 0.6746932691205138, "grad_norm": 14.1875, "learning_rate": 2.529540806716081e-06, "loss": 1.9947, "step": 5884 }, { "epoch": 0.6748079348698544, "grad_norm": 29.625, "learning_rate": 2.5279265243703775e-06, "loss": 1.8697, "step": 5885 }, { "epoch": 0.6749226006191951, "grad_norm": 13.625, "learning_rate": 2.526312583000087e-06, "loss": 1.7325, "step": 5886 }, { "epoch": 0.6750372663685357, "grad_norm": 13.625, "learning_rate": 2.524698982827824e-06, "loss": 1.7147, "step": 5887 }, { "epoch": 0.6751519321178764, "grad_norm": 18.5, "learning_rate": 2.523085724076154e-06, "loss": 1.9136, "step": 5888 }, { "epoch": 0.6752665978672171, "grad_norm": 13.0, "learning_rate": 2.521472806967591e-06, "loss": 1.8967, "step": 5889 }, { "epoch": 0.6753812636165577, "grad_norm": 19.0, "learning_rate": 2.519860231724612e-06, "loss": 2.0324, "step": 5890 }, { "epoch": 0.6754959293658984, "grad_norm": 14.875, "learning_rate": 2.5182479985696374e-06, "loss": 1.5331, "step": 5891 }, { "epoch": 0.6756105951152391, "grad_norm": 17.625, "learning_rate": 2.5166361077250445e-06, "loss": 1.7121, "step": 5892 }, { "epoch": 0.6757252608645797, "grad_norm": 12.5625, "learning_rate": 2.5150245594131627e-06, "loss": 1.5804, "step": 5893 }, { "epoch": 0.6758399266139204, "grad_norm": 15.125, "learning_rate": 2.5134133538562753e-06, "loss": 1.9277, "step": 5894 }, { "epoch": 0.675954592363261, "grad_norm": 12.75, "learning_rate": 2.5118024912766177e-06, "loss": 1.979, "step": 5895 }, { "epoch": 0.6760692581126018, "grad_norm": 17.0, "learning_rate": 2.5101919718963763e-06, "loss": 1.7365, "step": 5896 }, { "epoch": 0.6761839238619425, "grad_norm": 19.625, "learning_rate": 2.5085817959376944e-06, "loss": 1.6632, "step": 5897 }, { "epoch": 0.6762985896112831, "grad_norm": 13.625, "learning_rate": 2.506971963622662e-06, "loss": 1.7791, "step": 5898 }, { "epoch": 0.6764132553606238, "grad_norm": 17.125, "learning_rate": 2.505362475173325e-06, "loss": 1.5481, "step": 5899 }, { "epoch": 0.6765279211099644, "grad_norm": 13.4375, "learning_rate": 2.5037533308116824e-06, "loss": 1.634, "step": 5900 }, { "epoch": 0.6766425868593051, "grad_norm": 11.9375, "learning_rate": 2.502144530759685e-06, "loss": 1.6775, "step": 5901 }, { "epoch": 0.6767572526086458, "grad_norm": 15.375, "learning_rate": 2.500536075239236e-06, "loss": 2.1, "step": 5902 }, { "epoch": 0.6768719183579864, "grad_norm": 12.5625, "learning_rate": 2.4989279644721935e-06, "loss": 1.4776, "step": 5903 }, { "epoch": 0.6769865841073271, "grad_norm": 13.9375, "learning_rate": 2.49732019868036e-06, "loss": 1.6121, "step": 5904 }, { "epoch": 0.6771012498566679, "grad_norm": 13.1875, "learning_rate": 2.4957127780855034e-06, "loss": 1.7256, "step": 5905 }, { "epoch": 0.6772159156060085, "grad_norm": 14.3125, "learning_rate": 2.4941057029093306e-06, "loss": 1.7276, "step": 5906 }, { "epoch": 0.6773305813553492, "grad_norm": 12.75, "learning_rate": 2.4924989733735096e-06, "loss": 1.7037, "step": 5907 }, { "epoch": 0.6774452471046898, "grad_norm": 52.0, "learning_rate": 2.4908925896996584e-06, "loss": 2.5616, "step": 5908 }, { "epoch": 0.6775599128540305, "grad_norm": 16.75, "learning_rate": 2.4892865521093456e-06, "loss": 1.7346, "step": 5909 }, { "epoch": 0.6776745786033712, "grad_norm": 13.0625, "learning_rate": 2.487680860824095e-06, "loss": 1.7924, "step": 5910 }, { "epoch": 0.6777892443527118, "grad_norm": 14.0625, "learning_rate": 2.486075516065382e-06, "loss": 1.6701, "step": 5911 }, { "epoch": 0.6779039101020525, "grad_norm": 18.625, "learning_rate": 2.4844705180546303e-06, "loss": 1.6084, "step": 5912 }, { "epoch": 0.6780185758513931, "grad_norm": 12.5625, "learning_rate": 2.48286586701322e-06, "loss": 1.7323, "step": 5913 }, { "epoch": 0.6781332416007338, "grad_norm": 14.1875, "learning_rate": 2.4812615631624824e-06, "loss": 1.6381, "step": 5914 }, { "epoch": 0.6782479073500746, "grad_norm": 11.6875, "learning_rate": 2.479657606723701e-06, "loss": 1.8937, "step": 5915 }, { "epoch": 0.6783625730994152, "grad_norm": 12.6875, "learning_rate": 2.478053997918111e-06, "loss": 1.5112, "step": 5916 }, { "epoch": 0.6784772388487559, "grad_norm": 17.875, "learning_rate": 2.4764507369669e-06, "loss": 1.5107, "step": 5917 }, { "epoch": 0.6785919045980966, "grad_norm": 14.3125, "learning_rate": 2.4748478240912054e-06, "loss": 1.402, "step": 5918 }, { "epoch": 0.6787065703474372, "grad_norm": 31.375, "learning_rate": 2.473245259512118e-06, "loss": 2.1851, "step": 5919 }, { "epoch": 0.6788212360967779, "grad_norm": 24.75, "learning_rate": 2.471643043450686e-06, "loss": 1.7748, "step": 5920 }, { "epoch": 0.6789359018461185, "grad_norm": 14.5625, "learning_rate": 2.4700411761279e-06, "loss": 1.8153, "step": 5921 }, { "epoch": 0.6790505675954592, "grad_norm": 13.5, "learning_rate": 2.468439657764708e-06, "loss": 1.8314, "step": 5922 }, { "epoch": 0.6791652333448, "grad_norm": 13.5625, "learning_rate": 2.4668384885820113e-06, "loss": 1.7876, "step": 5923 }, { "epoch": 0.6792798990941405, "grad_norm": 15.625, "learning_rate": 2.4652376688006545e-06, "loss": 1.8669, "step": 5924 }, { "epoch": 0.6793945648434813, "grad_norm": 13.875, "learning_rate": 2.4636371986414462e-06, "loss": 2.0801, "step": 5925 }, { "epoch": 0.679509230592822, "grad_norm": 18.375, "learning_rate": 2.462037078325141e-06, "loss": 1.8617, "step": 5926 }, { "epoch": 0.6796238963421626, "grad_norm": 10.9375, "learning_rate": 2.4604373080724412e-06, "loss": 1.7742, "step": 5927 }, { "epoch": 0.6797385620915033, "grad_norm": 18.25, "learning_rate": 2.4588378881040066e-06, "loss": 2.0543, "step": 5928 }, { "epoch": 0.6798532278408439, "grad_norm": 15.0, "learning_rate": 2.4572388186404454e-06, "loss": 1.8535, "step": 5929 }, { "epoch": 0.6799678935901846, "grad_norm": 28.0, "learning_rate": 2.45564009990232e-06, "loss": 1.5936, "step": 5930 }, { "epoch": 0.6800825593395253, "grad_norm": 16.125, "learning_rate": 2.4540417321101434e-06, "loss": 1.9118, "step": 5931 }, { "epoch": 0.6801972250888659, "grad_norm": 11.9375, "learning_rate": 2.452443715484381e-06, "loss": 1.8929, "step": 5932 }, { "epoch": 0.6803118908382066, "grad_norm": 14.8125, "learning_rate": 2.4508460502454454e-06, "loss": 1.9864, "step": 5933 }, { "epoch": 0.6804265565875472, "grad_norm": 21.375, "learning_rate": 2.449248736613706e-06, "loss": 1.901, "step": 5934 }, { "epoch": 0.680541222336888, "grad_norm": 16.0, "learning_rate": 2.4476517748094815e-06, "loss": 1.7715, "step": 5935 }, { "epoch": 0.6806558880862287, "grad_norm": 16.125, "learning_rate": 2.446055165053042e-06, "loss": 1.5596, "step": 5936 }, { "epoch": 0.6807705538355693, "grad_norm": 14.4375, "learning_rate": 2.4444589075646107e-06, "loss": 1.7307, "step": 5937 }, { "epoch": 0.68088521958491, "grad_norm": 13.0, "learning_rate": 2.4428630025643617e-06, "loss": 1.591, "step": 5938 }, { "epoch": 0.6809998853342507, "grad_norm": 14.125, "learning_rate": 2.4412674502724144e-06, "loss": 1.5036, "step": 5939 }, { "epoch": 0.6811145510835913, "grad_norm": 15.3125, "learning_rate": 2.43967225090885e-06, "loss": 1.369, "step": 5940 }, { "epoch": 0.681229216832932, "grad_norm": 15.375, "learning_rate": 2.438077404693696e-06, "loss": 1.8132, "step": 5941 }, { "epoch": 0.6813438825822726, "grad_norm": 18.5, "learning_rate": 2.436482911846928e-06, "loss": 1.8128, "step": 5942 }, { "epoch": 0.6814585483316133, "grad_norm": 15.5, "learning_rate": 2.4348887725884767e-06, "loss": 2.0, "step": 5943 }, { "epoch": 0.6815732140809541, "grad_norm": 14.875, "learning_rate": 2.4332949871382237e-06, "loss": 2.1472, "step": 5944 }, { "epoch": 0.6816878798302947, "grad_norm": 18.875, "learning_rate": 2.4317015557160022e-06, "loss": 1.8833, "step": 5945 }, { "epoch": 0.6818025455796354, "grad_norm": 12.8125, "learning_rate": 2.430108478541594e-06, "loss": 1.5475, "step": 5946 }, { "epoch": 0.681917211328976, "grad_norm": 17.625, "learning_rate": 2.428515755834737e-06, "loss": 1.7961, "step": 5947 }, { "epoch": 0.6820318770783167, "grad_norm": 12.5625, "learning_rate": 2.4269233878151128e-06, "loss": 1.8254, "step": 5948 }, { "epoch": 0.6821465428276574, "grad_norm": 15.0625, "learning_rate": 2.4253313747023593e-06, "loss": 1.6447, "step": 5949 }, { "epoch": 0.682261208576998, "grad_norm": 25.875, "learning_rate": 2.4237397167160654e-06, "loss": 1.499, "step": 5950 }, { "epoch": 0.6823758743263387, "grad_norm": 23.375, "learning_rate": 2.4221484140757694e-06, "loss": 2.1016, "step": 5951 }, { "epoch": 0.6824905400756794, "grad_norm": 17.5, "learning_rate": 2.420557467000962e-06, "loss": 2.2321, "step": 5952 }, { "epoch": 0.68260520582502, "grad_norm": 13.8125, "learning_rate": 2.4189668757110844e-06, "loss": 1.5581, "step": 5953 }, { "epoch": 0.6827198715743608, "grad_norm": 15.8125, "learning_rate": 2.417376640425524e-06, "loss": 2.0052, "step": 5954 }, { "epoch": 0.6828345373237014, "grad_norm": 14.5, "learning_rate": 2.4157867613636304e-06, "loss": 1.713, "step": 5955 }, { "epoch": 0.6829492030730421, "grad_norm": 13.0, "learning_rate": 2.4141972387446915e-06, "loss": 1.5022, "step": 5956 }, { "epoch": 0.6830638688223828, "grad_norm": 14.625, "learning_rate": 2.412608072787954e-06, "loss": 1.7289, "step": 5957 }, { "epoch": 0.6831785345717234, "grad_norm": 15.875, "learning_rate": 2.411019263712612e-06, "loss": 1.7213, "step": 5958 }, { "epoch": 0.6832932003210641, "grad_norm": 16.5, "learning_rate": 2.409430811737813e-06, "loss": 1.7475, "step": 5959 }, { "epoch": 0.6834078660704048, "grad_norm": 14.0625, "learning_rate": 2.4078427170826525e-06, "loss": 1.7305, "step": 5960 }, { "epoch": 0.6835225318197454, "grad_norm": 13.9375, "learning_rate": 2.40625497996618e-06, "loss": 1.928, "step": 5961 }, { "epoch": 0.6836371975690861, "grad_norm": 17.0, "learning_rate": 2.4046676006073898e-06, "loss": 2.0493, "step": 5962 }, { "epoch": 0.6837518633184267, "grad_norm": 20.0, "learning_rate": 2.4030805792252327e-06, "loss": 1.9569, "step": 5963 }, { "epoch": 0.6838665290677675, "grad_norm": 21.25, "learning_rate": 2.4014939160386075e-06, "loss": 2.248, "step": 5964 }, { "epoch": 0.6839811948171082, "grad_norm": 15.1875, "learning_rate": 2.399907611266365e-06, "loss": 1.6831, "step": 5965 }, { "epoch": 0.6840958605664488, "grad_norm": 14.875, "learning_rate": 2.3983216651273045e-06, "loss": 1.8426, "step": 5966 }, { "epoch": 0.6842105263157895, "grad_norm": 15.6875, "learning_rate": 2.3967360778401806e-06, "loss": 2.1047, "step": 5967 }, { "epoch": 0.6843251920651301, "grad_norm": 13.8125, "learning_rate": 2.3951508496236897e-06, "loss": 1.7669, "step": 5968 }, { "epoch": 0.6844398578144708, "grad_norm": 15.125, "learning_rate": 2.3935659806964844e-06, "loss": 1.568, "step": 5969 }, { "epoch": 0.6845545235638115, "grad_norm": 14.1875, "learning_rate": 2.391981471277173e-06, "loss": 1.7479, "step": 5970 }, { "epoch": 0.6846691893131521, "grad_norm": 11.75, "learning_rate": 2.3903973215843015e-06, "loss": 1.4276, "step": 5971 }, { "epoch": 0.6847838550624928, "grad_norm": 13.4375, "learning_rate": 2.3888135318363764e-06, "loss": 1.9936, "step": 5972 }, { "epoch": 0.6848985208118336, "grad_norm": 16.375, "learning_rate": 2.387230102251853e-06, "loss": 2.0112, "step": 5973 }, { "epoch": 0.6850131865611742, "grad_norm": 21.25, "learning_rate": 2.3856470330491287e-06, "loss": 1.7977, "step": 5974 }, { "epoch": 0.6851278523105149, "grad_norm": 13.125, "learning_rate": 2.3840643244465646e-06, "loss": 1.7035, "step": 5975 }, { "epoch": 0.6852425180598555, "grad_norm": 14.125, "learning_rate": 2.382481976662464e-06, "loss": 1.9385, "step": 5976 }, { "epoch": 0.6853571838091962, "grad_norm": 13.875, "learning_rate": 2.380899989915079e-06, "loss": 1.4906, "step": 5977 }, { "epoch": 0.6854718495585369, "grad_norm": 15.625, "learning_rate": 2.379318364422615e-06, "loss": 1.657, "step": 5978 }, { "epoch": 0.6855865153078775, "grad_norm": 33.0, "learning_rate": 2.377737100403229e-06, "loss": 1.9458, "step": 5979 }, { "epoch": 0.6857011810572182, "grad_norm": 18.125, "learning_rate": 2.376156198075024e-06, "loss": 2.0545, "step": 5980 }, { "epoch": 0.6858158468065588, "grad_norm": 15.6875, "learning_rate": 2.3745756576560573e-06, "loss": 1.5147, "step": 5981 }, { "epoch": 0.6859305125558995, "grad_norm": 14.9375, "learning_rate": 2.372995479364335e-06, "loss": 2.1325, "step": 5982 }, { "epoch": 0.6860451783052403, "grad_norm": 15.75, "learning_rate": 2.371415663417809e-06, "loss": 1.811, "step": 5983 }, { "epoch": 0.6861598440545809, "grad_norm": 18.25, "learning_rate": 2.3698362100343857e-06, "loss": 1.879, "step": 5984 }, { "epoch": 0.6862745098039216, "grad_norm": 19.75, "learning_rate": 2.368257119431925e-06, "loss": 1.631, "step": 5985 }, { "epoch": 0.6863891755532623, "grad_norm": 12.0625, "learning_rate": 2.366678391828227e-06, "loss": 1.8732, "step": 5986 }, { "epoch": 0.6865038413026029, "grad_norm": 16.125, "learning_rate": 2.3651000274410494e-06, "loss": 1.4422, "step": 5987 }, { "epoch": 0.6866185070519436, "grad_norm": 15.4375, "learning_rate": 2.3635220264880993e-06, "loss": 2.0267, "step": 5988 }, { "epoch": 0.6867331728012842, "grad_norm": 14.1875, "learning_rate": 2.361944389187026e-06, "loss": 1.8745, "step": 5989 }, { "epoch": 0.6868478385506249, "grad_norm": 17.375, "learning_rate": 2.3603671157554396e-06, "loss": 1.8196, "step": 5990 }, { "epoch": 0.6869625042999656, "grad_norm": 12.0, "learning_rate": 2.3587902064108965e-06, "loss": 1.6024, "step": 5991 }, { "epoch": 0.6870771700493062, "grad_norm": 14.0625, "learning_rate": 2.3572136613708958e-06, "loss": 2.0546, "step": 5992 }, { "epoch": 0.687191835798647, "grad_norm": 14.0625, "learning_rate": 2.3556374808528944e-06, "loss": 1.5692, "step": 5993 }, { "epoch": 0.6873065015479877, "grad_norm": 17.125, "learning_rate": 2.3540616650742964e-06, "loss": 2.0977, "step": 5994 }, { "epoch": 0.6874211672973283, "grad_norm": 15.75, "learning_rate": 2.3524862142524557e-06, "loss": 1.7159, "step": 5995 }, { "epoch": 0.687535833046669, "grad_norm": 13.75, "learning_rate": 2.3509111286046756e-06, "loss": 2.2318, "step": 5996 }, { "epoch": 0.6876504987960096, "grad_norm": 13.75, "learning_rate": 2.3493364083482106e-06, "loss": 1.8833, "step": 5997 }, { "epoch": 0.6877651645453503, "grad_norm": 14.75, "learning_rate": 2.3477620537002604e-06, "loss": 1.7936, "step": 5998 }, { "epoch": 0.687879830294691, "grad_norm": 18.5, "learning_rate": 2.3461880648779784e-06, "loss": 1.9481, "step": 5999 }, { "epoch": 0.6879944960440316, "grad_norm": 18.625, "learning_rate": 2.3446144420984668e-06, "loss": 1.8127, "step": 6000 }, { "epoch": 0.6881091617933723, "grad_norm": 15.875, "learning_rate": 2.3430411855787764e-06, "loss": 1.8342, "step": 6001 }, { "epoch": 0.6882238275427129, "grad_norm": 12.9375, "learning_rate": 2.3414682955359087e-06, "loss": 1.9247, "step": 6002 }, { "epoch": 0.6883384932920537, "grad_norm": 14.75, "learning_rate": 2.339895772186815e-06, "loss": 1.6563, "step": 6003 }, { "epoch": 0.6884531590413944, "grad_norm": 15.4375, "learning_rate": 2.338323615748389e-06, "loss": 2.1495, "step": 6004 }, { "epoch": 0.688567824790735, "grad_norm": 20.375, "learning_rate": 2.3367518264374885e-06, "loss": 2.3286, "step": 6005 }, { "epoch": 0.6886824905400757, "grad_norm": 14.1875, "learning_rate": 2.335180404470905e-06, "loss": 1.6844, "step": 6006 }, { "epoch": 0.6887971562894164, "grad_norm": 13.875, "learning_rate": 2.333609350065389e-06, "loss": 1.734, "step": 6007 }, { "epoch": 0.688911822038757, "grad_norm": 15.0625, "learning_rate": 2.332038663437637e-06, "loss": 1.5326, "step": 6008 }, { "epoch": 0.6890264877880977, "grad_norm": 34.0, "learning_rate": 2.330468344804296e-06, "loss": 2.1948, "step": 6009 }, { "epoch": 0.6891411535374383, "grad_norm": 17.0, "learning_rate": 2.32889839438196e-06, "loss": 1.6638, "step": 6010 }, { "epoch": 0.689255819286779, "grad_norm": 14.0625, "learning_rate": 2.3273288123871767e-06, "loss": 2.0121, "step": 6011 }, { "epoch": 0.6893704850361198, "grad_norm": 16.875, "learning_rate": 2.3257595990364367e-06, "loss": 1.9411, "step": 6012 }, { "epoch": 0.6894851507854604, "grad_norm": 17.0, "learning_rate": 2.324190754546184e-06, "loss": 1.7209, "step": 6013 }, { "epoch": 0.6895998165348011, "grad_norm": 12.0, "learning_rate": 2.3226222791328114e-06, "loss": 1.715, "step": 6014 }, { "epoch": 0.6897144822841417, "grad_norm": 15.125, "learning_rate": 2.321054173012659e-06, "loss": 1.7772, "step": 6015 }, { "epoch": 0.6898291480334824, "grad_norm": 15.375, "learning_rate": 2.319486436402019e-06, "loss": 1.5227, "step": 6016 }, { "epoch": 0.6899438137828231, "grad_norm": 13.875, "learning_rate": 2.3179190695171317e-06, "loss": 1.9642, "step": 6017 }, { "epoch": 0.6900584795321637, "grad_norm": 12.1875, "learning_rate": 2.3163520725741814e-06, "loss": 1.6622, "step": 6018 }, { "epoch": 0.6901731452815044, "grad_norm": 16.5, "learning_rate": 2.314785445789306e-06, "loss": 1.7521, "step": 6019 }, { "epoch": 0.6902878110308451, "grad_norm": 13.3125, "learning_rate": 2.3132191893785972e-06, "loss": 2.0322, "step": 6020 }, { "epoch": 0.6904024767801857, "grad_norm": 17.375, "learning_rate": 2.3116533035580845e-06, "loss": 1.5498, "step": 6021 }, { "epoch": 0.6905171425295265, "grad_norm": 13.125, "learning_rate": 2.3100877885437544e-06, "loss": 1.4367, "step": 6022 }, { "epoch": 0.690631808278867, "grad_norm": 16.25, "learning_rate": 2.3085226445515413e-06, "loss": 1.9502, "step": 6023 }, { "epoch": 0.6907464740282078, "grad_norm": 14.5625, "learning_rate": 2.3069578717973213e-06, "loss": 2.1114, "step": 6024 }, { "epoch": 0.6908611397775485, "grad_norm": 13.3125, "learning_rate": 2.3053934704969303e-06, "loss": 1.5252, "step": 6025 }, { "epoch": 0.6909758055268891, "grad_norm": 12.4375, "learning_rate": 2.3038294408661487e-06, "loss": 1.635, "step": 6026 }, { "epoch": 0.6910904712762298, "grad_norm": 14.4375, "learning_rate": 2.3022657831207e-06, "loss": 2.0399, "step": 6027 }, { "epoch": 0.6912051370255705, "grad_norm": 17.375, "learning_rate": 2.300702497476263e-06, "loss": 1.8462, "step": 6028 }, { "epoch": 0.6913198027749111, "grad_norm": 12.125, "learning_rate": 2.299139584148463e-06, "loss": 1.9207, "step": 6029 }, { "epoch": 0.6914344685242518, "grad_norm": 14.125, "learning_rate": 2.2975770433528744e-06, "loss": 1.7979, "step": 6030 }, { "epoch": 0.6915491342735924, "grad_norm": 26.5, "learning_rate": 2.29601487530502e-06, "loss": 1.6255, "step": 6031 }, { "epoch": 0.6916638000229332, "grad_norm": 13.9375, "learning_rate": 2.2944530802203724e-06, "loss": 2.2709, "step": 6032 }, { "epoch": 0.6917784657722739, "grad_norm": 18.75, "learning_rate": 2.2928916583143485e-06, "loss": 1.7617, "step": 6033 }, { "epoch": 0.6918931315216145, "grad_norm": 12.9375, "learning_rate": 2.291330609802316e-06, "loss": 1.6312, "step": 6034 }, { "epoch": 0.6920077972709552, "grad_norm": 14.3125, "learning_rate": 2.2897699348995978e-06, "loss": 2.1166, "step": 6035 }, { "epoch": 0.6921224630202958, "grad_norm": 12.25, "learning_rate": 2.288209633821453e-06, "loss": 1.7607, "step": 6036 }, { "epoch": 0.6922371287696365, "grad_norm": 15.0625, "learning_rate": 2.2866497067830984e-06, "loss": 1.7923, "step": 6037 }, { "epoch": 0.6923517945189772, "grad_norm": 14.4375, "learning_rate": 2.285090153999697e-06, "loss": 1.9717, "step": 6038 }, { "epoch": 0.6924664602683178, "grad_norm": 16.75, "learning_rate": 2.283530975686355e-06, "loss": 1.8336, "step": 6039 }, { "epoch": 0.6925811260176585, "grad_norm": 16.5, "learning_rate": 2.2819721720581355e-06, "loss": 2.2958, "step": 6040 }, { "epoch": 0.6926957917669992, "grad_norm": 14.9375, "learning_rate": 2.2804137433300466e-06, "loss": 2.0537, "step": 6041 }, { "epoch": 0.6928104575163399, "grad_norm": 11.4375, "learning_rate": 2.27885568971704e-06, "loss": 1.525, "step": 6042 }, { "epoch": 0.6929251232656806, "grad_norm": 11.9375, "learning_rate": 2.277298011434021e-06, "loss": 1.6069, "step": 6043 }, { "epoch": 0.6930397890150212, "grad_norm": 12.5, "learning_rate": 2.275740708695842e-06, "loss": 1.7441, "step": 6044 }, { "epoch": 0.6931544547643619, "grad_norm": 17.25, "learning_rate": 2.2741837817173036e-06, "loss": 1.6003, "step": 6045 }, { "epoch": 0.6932691205137026, "grad_norm": 15.25, "learning_rate": 2.272627230713153e-06, "loss": 1.732, "step": 6046 }, { "epoch": 0.6933837862630432, "grad_norm": 14.6875, "learning_rate": 2.27107105589809e-06, "loss": 1.546, "step": 6047 }, { "epoch": 0.6934984520123839, "grad_norm": 19.5, "learning_rate": 2.2695152574867545e-06, "loss": 1.6868, "step": 6048 }, { "epoch": 0.6936131177617245, "grad_norm": 13.0625, "learning_rate": 2.2679598356937414e-06, "loss": 2.0185, "step": 6049 }, { "epoch": 0.6937277835110652, "grad_norm": 13.5, "learning_rate": 2.2664047907335918e-06, "loss": 1.8318, "step": 6050 }, { "epoch": 0.693842449260406, "grad_norm": 20.875, "learning_rate": 2.2648501228207942e-06, "loss": 1.7985, "step": 6051 }, { "epoch": 0.6939571150097466, "grad_norm": 13.0625, "learning_rate": 2.263295832169785e-06, "loss": 1.2584, "step": 6052 }, { "epoch": 0.6940717807590873, "grad_norm": 33.25, "learning_rate": 2.2617419189949517e-06, "loss": 2.086, "step": 6053 }, { "epoch": 0.694186446508428, "grad_norm": 12.4375, "learning_rate": 2.260188383510621e-06, "loss": 1.4135, "step": 6054 }, { "epoch": 0.6943011122577686, "grad_norm": 15.6875, "learning_rate": 2.258635225931081e-06, "loss": 1.7432, "step": 6055 }, { "epoch": 0.6944157780071093, "grad_norm": 15.0625, "learning_rate": 2.2570824464705543e-06, "loss": 1.7199, "step": 6056 }, { "epoch": 0.6945304437564499, "grad_norm": 12.5625, "learning_rate": 2.2555300453432193e-06, "loss": 1.7561, "step": 6057 }, { "epoch": 0.6946451095057906, "grad_norm": 18.625, "learning_rate": 2.253978022763201e-06, "loss": 2.1546, "step": 6058 }, { "epoch": 0.6947597752551313, "grad_norm": 13.9375, "learning_rate": 2.252426378944569e-06, "loss": 1.5942, "step": 6059 }, { "epoch": 0.6948744410044719, "grad_norm": 15.0, "learning_rate": 2.2508751141013463e-06, "loss": 1.4611, "step": 6060 }, { "epoch": 0.6949891067538126, "grad_norm": 17.5, "learning_rate": 2.2493242284474994e-06, "loss": 2.0096, "step": 6061 }, { "epoch": 0.6951037725031534, "grad_norm": 12.125, "learning_rate": 2.2477737221969407e-06, "loss": 1.7818, "step": 6062 }, { "epoch": 0.695218438252494, "grad_norm": 15.375, "learning_rate": 2.2462235955635347e-06, "loss": 1.5236, "step": 6063 }, { "epoch": 0.6953331040018347, "grad_norm": 15.0, "learning_rate": 2.2446738487610924e-06, "loss": 1.6864, "step": 6064 }, { "epoch": 0.6954477697511753, "grad_norm": 15.5625, "learning_rate": 2.243124482003371e-06, "loss": 1.9831, "step": 6065 }, { "epoch": 0.695562435500516, "grad_norm": 16.125, "learning_rate": 2.2415754955040765e-06, "loss": 1.6657, "step": 6066 }, { "epoch": 0.6956771012498567, "grad_norm": 12.4375, "learning_rate": 2.240026889476864e-06, "loss": 1.7085, "step": 6067 }, { "epoch": 0.6957917669991973, "grad_norm": 13.4375, "learning_rate": 2.23847866413533e-06, "loss": 1.585, "step": 6068 }, { "epoch": 0.695906432748538, "grad_norm": 11.75, "learning_rate": 2.236930819693024e-06, "loss": 1.5604, "step": 6069 }, { "epoch": 0.6960210984978786, "grad_norm": 17.875, "learning_rate": 2.2353833563634453e-06, "loss": 1.9655, "step": 6070 }, { "epoch": 0.6961357642472193, "grad_norm": 14.0625, "learning_rate": 2.233836274360033e-06, "loss": 1.8782, "step": 6071 }, { "epoch": 0.6962504299965601, "grad_norm": 16.125, "learning_rate": 2.232289573896178e-06, "loss": 1.5843, "step": 6072 }, { "epoch": 0.6963650957459007, "grad_norm": 10.75, "learning_rate": 2.2307432551852215e-06, "loss": 1.608, "step": 6073 }, { "epoch": 0.6964797614952414, "grad_norm": 17.375, "learning_rate": 2.229197318440442e-06, "loss": 1.7116, "step": 6074 }, { "epoch": 0.6965944272445821, "grad_norm": 16.125, "learning_rate": 2.227651763875078e-06, "loss": 1.6801, "step": 6075 }, { "epoch": 0.6967090929939227, "grad_norm": 13.5, "learning_rate": 2.2261065917023094e-06, "loss": 1.6233, "step": 6076 }, { "epoch": 0.6968237587432634, "grad_norm": 13.375, "learning_rate": 2.2245618021352593e-06, "loss": 1.4218, "step": 6077 }, { "epoch": 0.696938424492604, "grad_norm": 13.9375, "learning_rate": 2.2230173953870045e-06, "loss": 1.9047, "step": 6078 }, { "epoch": 0.6970530902419447, "grad_norm": 31.125, "learning_rate": 2.221473371670565e-06, "loss": 1.9893, "step": 6079 }, { "epoch": 0.6971677559912854, "grad_norm": 14.8125, "learning_rate": 2.219929731198911e-06, "loss": 1.5419, "step": 6080 }, { "epoch": 0.697282421740626, "grad_norm": 17.5, "learning_rate": 2.218386474184957e-06, "loss": 1.6201, "step": 6081 }, { "epoch": 0.6973970874899668, "grad_norm": 17.375, "learning_rate": 2.216843600841569e-06, "loss": 1.6575, "step": 6082 }, { "epoch": 0.6975117532393074, "grad_norm": 12.8125, "learning_rate": 2.215301111381552e-06, "loss": 1.3993, "step": 6083 }, { "epoch": 0.6976264189886481, "grad_norm": 12.5625, "learning_rate": 2.213759006017664e-06, "loss": 1.4141, "step": 6084 }, { "epoch": 0.6977410847379888, "grad_norm": 12.6875, "learning_rate": 2.2122172849626146e-06, "loss": 1.7289, "step": 6085 }, { "epoch": 0.6978557504873294, "grad_norm": 15.8125, "learning_rate": 2.2106759484290487e-06, "loss": 2.1209, "step": 6086 }, { "epoch": 0.6979704162366701, "grad_norm": 13.0, "learning_rate": 2.209134996629566e-06, "loss": 2.0811, "step": 6087 }, { "epoch": 0.6980850819860108, "grad_norm": 13.8125, "learning_rate": 2.207594429776714e-06, "loss": 1.8861, "step": 6088 }, { "epoch": 0.6981997477353514, "grad_norm": 14.3125, "learning_rate": 2.2060542480829783e-06, "loss": 1.6351, "step": 6089 }, { "epoch": 0.6983144134846921, "grad_norm": 14.375, "learning_rate": 2.2045144517608035e-06, "loss": 1.8876, "step": 6090 }, { "epoch": 0.6984290792340327, "grad_norm": 16.625, "learning_rate": 2.2029750410225754e-06, "loss": 2.1545, "step": 6091 }, { "epoch": 0.6985437449833735, "grad_norm": 17.5, "learning_rate": 2.201436016080622e-06, "loss": 1.9716, "step": 6092 }, { "epoch": 0.6986584107327142, "grad_norm": 16.625, "learning_rate": 2.199897377147225e-06, "loss": 1.5653, "step": 6093 }, { "epoch": 0.6987730764820548, "grad_norm": 13.5, "learning_rate": 2.1983591244346096e-06, "loss": 1.7608, "step": 6094 }, { "epoch": 0.6988877422313955, "grad_norm": 15.375, "learning_rate": 2.1968212581549495e-06, "loss": 1.762, "step": 6095 }, { "epoch": 0.6990024079807362, "grad_norm": 14.125, "learning_rate": 2.1952837785203635e-06, "loss": 1.6296, "step": 6096 }, { "epoch": 0.6991170737300768, "grad_norm": 16.75, "learning_rate": 2.1937466857429183e-06, "loss": 1.7048, "step": 6097 }, { "epoch": 0.6992317394794175, "grad_norm": 15.1875, "learning_rate": 2.1922099800346247e-06, "loss": 1.7014, "step": 6098 }, { "epoch": 0.6993464052287581, "grad_norm": 12.8125, "learning_rate": 2.190673661607443e-06, "loss": 1.6474, "step": 6099 }, { "epoch": 0.6994610709780988, "grad_norm": 16.25, "learning_rate": 2.189137730673279e-06, "loss": 1.9689, "step": 6100 }, { "epoch": 0.6995757367274396, "grad_norm": 15.25, "learning_rate": 2.187602187443985e-06, "loss": 1.7861, "step": 6101 }, { "epoch": 0.6996904024767802, "grad_norm": 20.875, "learning_rate": 2.1860670321313604e-06, "loss": 1.8024, "step": 6102 }, { "epoch": 0.6998050682261209, "grad_norm": 12.0, "learning_rate": 2.1845322649471515e-06, "loss": 1.5402, "step": 6103 }, { "epoch": 0.6999197339754615, "grad_norm": 12.875, "learning_rate": 2.182997886103046e-06, "loss": 1.8667, "step": 6104 }, { "epoch": 0.7000343997248022, "grad_norm": 13.1875, "learning_rate": 2.1814638958106882e-06, "loss": 1.6816, "step": 6105 }, { "epoch": 0.7001490654741429, "grad_norm": 12.5, "learning_rate": 2.179930294281658e-06, "loss": 1.9486, "step": 6106 }, { "epoch": 0.7002637312234835, "grad_norm": 14.375, "learning_rate": 2.1783970817274887e-06, "loss": 1.772, "step": 6107 }, { "epoch": 0.7003783969728242, "grad_norm": 15.1875, "learning_rate": 2.1768642583596574e-06, "loss": 1.6967, "step": 6108 }, { "epoch": 0.7004930627221649, "grad_norm": 12.0625, "learning_rate": 2.1753318243895873e-06, "loss": 1.5484, "step": 6109 }, { "epoch": 0.7006077284715055, "grad_norm": 15.0625, "learning_rate": 2.173799780028649e-06, "loss": 1.6013, "step": 6110 }, { "epoch": 0.7007223942208463, "grad_norm": 14.4375, "learning_rate": 2.1722681254881606e-06, "loss": 1.814, "step": 6111 }, { "epoch": 0.7008370599701869, "grad_norm": 12.4375, "learning_rate": 2.1707368609793815e-06, "loss": 1.9101, "step": 6112 }, { "epoch": 0.7009517257195276, "grad_norm": 13.625, "learning_rate": 2.1692059867135213e-06, "loss": 2.1032, "step": 6113 }, { "epoch": 0.7010663914688683, "grad_norm": 13.9375, "learning_rate": 2.1676755029017355e-06, "loss": 1.4566, "step": 6114 }, { "epoch": 0.7011810572182089, "grad_norm": 13.375, "learning_rate": 2.166145409755125e-06, "loss": 1.9531, "step": 6115 }, { "epoch": 0.7012957229675496, "grad_norm": 13.4375, "learning_rate": 2.1646157074847374e-06, "loss": 2.0879, "step": 6116 }, { "epoch": 0.7014103887168902, "grad_norm": 12.8125, "learning_rate": 2.163086396301567e-06, "loss": 1.7366, "step": 6117 }, { "epoch": 0.7015250544662309, "grad_norm": 13.9375, "learning_rate": 2.1615574764165497e-06, "loss": 1.9978, "step": 6118 }, { "epoch": 0.7016397202155716, "grad_norm": 19.875, "learning_rate": 2.1600289480405717e-06, "loss": 1.5259, "step": 6119 }, { "epoch": 0.7017543859649122, "grad_norm": 11.9375, "learning_rate": 2.158500811384469e-06, "loss": 1.5666, "step": 6120 }, { "epoch": 0.701869051714253, "grad_norm": 13.8125, "learning_rate": 2.1569730666590134e-06, "loss": 1.8023, "step": 6121 }, { "epoch": 0.7019837174635937, "grad_norm": 12.4375, "learning_rate": 2.1554457140749304e-06, "loss": 1.4346, "step": 6122 }, { "epoch": 0.7020983832129343, "grad_norm": 12.875, "learning_rate": 2.153918753842891e-06, "loss": 2.0295, "step": 6123 }, { "epoch": 0.702213048962275, "grad_norm": 14.875, "learning_rate": 2.1523921861735043e-06, "loss": 1.8982, "step": 6124 }, { "epoch": 0.7023277147116156, "grad_norm": 15.25, "learning_rate": 2.1508660112773373e-06, "loss": 2.0839, "step": 6125 }, { "epoch": 0.7024423804609563, "grad_norm": 14.125, "learning_rate": 2.1493402293648963e-06, "loss": 1.8103, "step": 6126 }, { "epoch": 0.702557046210297, "grad_norm": 16.5, "learning_rate": 2.1478148406466305e-06, "loss": 2.0844, "step": 6127 }, { "epoch": 0.7026717119596376, "grad_norm": 17.625, "learning_rate": 2.1462898453329394e-06, "loss": 1.5926, "step": 6128 }, { "epoch": 0.7027863777089783, "grad_norm": 15.6875, "learning_rate": 2.1447652436341676e-06, "loss": 1.6518, "step": 6129 }, { "epoch": 0.702901043458319, "grad_norm": 16.875, "learning_rate": 2.1432410357606046e-06, "loss": 1.8326, "step": 6130 }, { "epoch": 0.7030157092076597, "grad_norm": 16.0, "learning_rate": 2.1417172219224865e-06, "loss": 2.175, "step": 6131 }, { "epoch": 0.7031303749570004, "grad_norm": 14.0, "learning_rate": 2.140193802329995e-06, "loss": 1.7748, "step": 6132 }, { "epoch": 0.703245040706341, "grad_norm": 17.375, "learning_rate": 2.138670777193254e-06, "loss": 1.6486, "step": 6133 }, { "epoch": 0.7033597064556817, "grad_norm": 13.6875, "learning_rate": 2.137148146722336e-06, "loss": 1.6056, "step": 6134 }, { "epoch": 0.7034743722050224, "grad_norm": 19.0, "learning_rate": 2.1356259111272633e-06, "loss": 2.234, "step": 6135 }, { "epoch": 0.703589037954363, "grad_norm": 14.0, "learning_rate": 2.1341040706179946e-06, "loss": 1.9448, "step": 6136 }, { "epoch": 0.7037037037037037, "grad_norm": 14.0625, "learning_rate": 2.13258262540444e-06, "loss": 1.5168, "step": 6137 }, { "epoch": 0.7038183694530443, "grad_norm": 14.4375, "learning_rate": 2.131061575696457e-06, "loss": 1.6721, "step": 6138 }, { "epoch": 0.703933035202385, "grad_norm": 17.25, "learning_rate": 2.129540921703838e-06, "loss": 1.5328, "step": 6139 }, { "epoch": 0.7040477009517258, "grad_norm": 15.8125, "learning_rate": 2.128020663636336e-06, "loss": 1.8426, "step": 6140 }, { "epoch": 0.7041623667010664, "grad_norm": 13.625, "learning_rate": 2.1265008017036393e-06, "loss": 1.8842, "step": 6141 }, { "epoch": 0.7042770324504071, "grad_norm": 15.5625, "learning_rate": 2.124981336115382e-06, "loss": 1.5274, "step": 6142 }, { "epoch": 0.7043916981997478, "grad_norm": 13.375, "learning_rate": 2.1234622670811463e-06, "loss": 2.0838, "step": 6143 }, { "epoch": 0.7045063639490884, "grad_norm": 13.9375, "learning_rate": 2.121943594810459e-06, "loss": 1.9567, "step": 6144 }, { "epoch": 0.7046210296984291, "grad_norm": 12.875, "learning_rate": 2.1204253195127917e-06, "loss": 1.6961, "step": 6145 }, { "epoch": 0.7047356954477697, "grad_norm": 41.0, "learning_rate": 2.1189074413975625e-06, "loss": 1.8572, "step": 6146 }, { "epoch": 0.7048503611971104, "grad_norm": 14.1875, "learning_rate": 2.1173899606741343e-06, "loss": 1.8104, "step": 6147 }, { "epoch": 0.7049650269464511, "grad_norm": 17.5, "learning_rate": 2.1158728775518124e-06, "loss": 2.1541, "step": 6148 }, { "epoch": 0.7050796926957917, "grad_norm": 13.0625, "learning_rate": 2.1143561922398496e-06, "loss": 1.7124, "step": 6149 }, { "epoch": 0.7051943584451325, "grad_norm": 17.5, "learning_rate": 2.1128399049474456e-06, "loss": 1.5459, "step": 6150 }, { "epoch": 0.705309024194473, "grad_norm": 14.5, "learning_rate": 2.111324015883742e-06, "loss": 1.3532, "step": 6151 }, { "epoch": 0.7054236899438138, "grad_norm": 16.625, "learning_rate": 2.1098085252578275e-06, "loss": 2.0128, "step": 6152 }, { "epoch": 0.7055383556931545, "grad_norm": 13.375, "learning_rate": 2.108293433278737e-06, "loss": 1.848, "step": 6153 }, { "epoch": 0.7056530214424951, "grad_norm": 15.0625, "learning_rate": 2.1067787401554423e-06, "loss": 1.5693, "step": 6154 }, { "epoch": 0.7057676871918358, "grad_norm": 31.0, "learning_rate": 2.1052644460968743e-06, "loss": 2.0075, "step": 6155 }, { "epoch": 0.7058823529411765, "grad_norm": 14.8125, "learning_rate": 2.1037505513118957e-06, "loss": 1.5499, "step": 6156 }, { "epoch": 0.7059970186905171, "grad_norm": 15.4375, "learning_rate": 2.1022370560093205e-06, "loss": 1.9822, "step": 6157 }, { "epoch": 0.7061116844398578, "grad_norm": 17.0, "learning_rate": 2.1007239603979074e-06, "loss": 1.668, "step": 6158 }, { "epoch": 0.7062263501891984, "grad_norm": 13.25, "learning_rate": 2.0992112646863584e-06, "loss": 1.5551, "step": 6159 }, { "epoch": 0.7063410159385392, "grad_norm": 13.8125, "learning_rate": 2.0976989690833216e-06, "loss": 1.5522, "step": 6160 }, { "epoch": 0.7064556816878799, "grad_norm": 13.3125, "learning_rate": 2.0961870737973913e-06, "loss": 1.8466, "step": 6161 }, { "epoch": 0.7065703474372205, "grad_norm": 17.125, "learning_rate": 2.0946755790371005e-06, "loss": 1.9331, "step": 6162 }, { "epoch": 0.7066850131865612, "grad_norm": 17.375, "learning_rate": 2.093164485010932e-06, "loss": 1.5038, "step": 6163 }, { "epoch": 0.7067996789359019, "grad_norm": 32.0, "learning_rate": 2.091653791927315e-06, "loss": 2.0883, "step": 6164 }, { "epoch": 0.7069143446852425, "grad_norm": 14.5625, "learning_rate": 2.0901434999946183e-06, "loss": 1.7668, "step": 6165 }, { "epoch": 0.7070290104345832, "grad_norm": 14.4375, "learning_rate": 2.0886336094211597e-06, "loss": 1.5784, "step": 6166 }, { "epoch": 0.7071436761839238, "grad_norm": 16.0, "learning_rate": 2.0871241204152e-06, "loss": 1.9021, "step": 6167 }, { "epoch": 0.7072583419332645, "grad_norm": 15.25, "learning_rate": 2.085615033184942e-06, "loss": 1.8626, "step": 6168 }, { "epoch": 0.7073730076826052, "grad_norm": 15.9375, "learning_rate": 2.084106347938535e-06, "loss": 1.6423, "step": 6169 }, { "epoch": 0.7074876734319459, "grad_norm": 14.9375, "learning_rate": 2.0825980648840787e-06, "loss": 2.0902, "step": 6170 }, { "epoch": 0.7076023391812866, "grad_norm": 17.0, "learning_rate": 2.081090184229607e-06, "loss": 1.9079, "step": 6171 }, { "epoch": 0.7077170049306272, "grad_norm": 19.0, "learning_rate": 2.079582706183105e-06, "loss": 2.0481, "step": 6172 }, { "epoch": 0.7078316706799679, "grad_norm": 13.4375, "learning_rate": 2.0780756309525025e-06, "loss": 1.866, "step": 6173 }, { "epoch": 0.7079463364293086, "grad_norm": 14.0, "learning_rate": 2.0765689587456655e-06, "loss": 1.8015, "step": 6174 }, { "epoch": 0.7080610021786492, "grad_norm": 19.375, "learning_rate": 2.0750626897704175e-06, "loss": 1.8352, "step": 6175 }, { "epoch": 0.7081756679279899, "grad_norm": 14.6875, "learning_rate": 2.0735568242345184e-06, "loss": 1.8474, "step": 6176 }, { "epoch": 0.7082903336773306, "grad_norm": 16.125, "learning_rate": 2.0720513623456708e-06, "loss": 1.7533, "step": 6177 }, { "epoch": 0.7084049994266712, "grad_norm": 15.25, "learning_rate": 2.0705463043115253e-06, "loss": 2.0768, "step": 6178 }, { "epoch": 0.708519665176012, "grad_norm": 13.125, "learning_rate": 2.0690416503396777e-06, "loss": 1.7677, "step": 6179 }, { "epoch": 0.7086343309253526, "grad_norm": 15.875, "learning_rate": 2.067537400637665e-06, "loss": 1.6627, "step": 6180 }, { "epoch": 0.7087489966746933, "grad_norm": 13.75, "learning_rate": 2.0660335554129697e-06, "loss": 1.5688, "step": 6181 }, { "epoch": 0.708863662424034, "grad_norm": 14.5, "learning_rate": 2.064530114873021e-06, "loss": 2.0796, "step": 6182 }, { "epoch": 0.7089783281733746, "grad_norm": 16.75, "learning_rate": 2.0630270792251855e-06, "loss": 2.1023, "step": 6183 }, { "epoch": 0.7090929939227153, "grad_norm": 14.375, "learning_rate": 2.0615244486767796e-06, "loss": 1.6916, "step": 6184 }, { "epoch": 0.709207659672056, "grad_norm": 15.3125, "learning_rate": 2.0600222234350666e-06, "loss": 2.2009, "step": 6185 }, { "epoch": 0.7093223254213966, "grad_norm": 14.4375, "learning_rate": 2.058520403707245e-06, "loss": 1.8261, "step": 6186 }, { "epoch": 0.7094369911707373, "grad_norm": 16.625, "learning_rate": 2.0570189897004638e-06, "loss": 1.8282, "step": 6187 }, { "epoch": 0.7095516569200779, "grad_norm": 17.375, "learning_rate": 2.055517981621817e-06, "loss": 1.9853, "step": 6188 }, { "epoch": 0.7096663226694186, "grad_norm": 15.5, "learning_rate": 2.0540173796783343e-06, "loss": 1.4496, "step": 6189 }, { "epoch": 0.7097809884187594, "grad_norm": 17.0, "learning_rate": 2.052517184077e-06, "loss": 1.2935, "step": 6190 }, { "epoch": 0.7098956541681, "grad_norm": 13.0625, "learning_rate": 2.0510173950247376e-06, "loss": 1.4668, "step": 6191 }, { "epoch": 0.7100103199174407, "grad_norm": 15.6875, "learning_rate": 2.0495180127284108e-06, "loss": 1.9599, "step": 6192 }, { "epoch": 0.7101249856667813, "grad_norm": 16.5, "learning_rate": 2.0480190373948334e-06, "loss": 1.8457, "step": 6193 }, { "epoch": 0.710239651416122, "grad_norm": 33.25, "learning_rate": 2.04652046923076e-06, "loss": 1.647, "step": 6194 }, { "epoch": 0.7103543171654627, "grad_norm": 13.1875, "learning_rate": 2.045022308442889e-06, "loss": 1.7121, "step": 6195 }, { "epoch": 0.7104689829148033, "grad_norm": 13.0625, "learning_rate": 2.0435245552378636e-06, "loss": 1.8641, "step": 6196 }, { "epoch": 0.710583648664144, "grad_norm": 11.5, "learning_rate": 2.0420272098222725e-06, "loss": 1.5041, "step": 6197 }, { "epoch": 0.7106983144134847, "grad_norm": 29.625, "learning_rate": 2.0405302724026417e-06, "loss": 1.6839, "step": 6198 }, { "epoch": 0.7108129801628253, "grad_norm": 16.625, "learning_rate": 2.039033743185447e-06, "loss": 1.7753, "step": 6199 }, { "epoch": 0.7109276459121661, "grad_norm": 15.4375, "learning_rate": 2.0375376223771066e-06, "loss": 1.7348, "step": 6200 }, { "epoch": 0.7110423116615067, "grad_norm": 15.9375, "learning_rate": 2.0360419101839813e-06, "loss": 1.97, "step": 6201 }, { "epoch": 0.7111569774108474, "grad_norm": 15.125, "learning_rate": 2.0345466068123758e-06, "loss": 1.8998, "step": 6202 }, { "epoch": 0.7112716431601881, "grad_norm": 13.125, "learning_rate": 2.0330517124685416e-06, "loss": 1.5675, "step": 6203 }, { "epoch": 0.7113863089095287, "grad_norm": 13.5, "learning_rate": 2.0315572273586648e-06, "loss": 1.8673, "step": 6204 }, { "epoch": 0.7115009746588694, "grad_norm": 21.875, "learning_rate": 2.0300631516888886e-06, "loss": 2.0734, "step": 6205 }, { "epoch": 0.71161564040821, "grad_norm": 13.6875, "learning_rate": 2.0285694856652865e-06, "loss": 1.7327, "step": 6206 }, { "epoch": 0.7117303061575507, "grad_norm": 15.9375, "learning_rate": 2.0270762294938834e-06, "loss": 1.718, "step": 6207 }, { "epoch": 0.7118449719068914, "grad_norm": 14.6875, "learning_rate": 2.0255833833806453e-06, "loss": 1.7518, "step": 6208 }, { "epoch": 0.711959637656232, "grad_norm": 13.8125, "learning_rate": 2.024090947531482e-06, "loss": 2.2188, "step": 6209 }, { "epoch": 0.7120743034055728, "grad_norm": 13.9375, "learning_rate": 2.022598922152246e-06, "loss": 1.7699, "step": 6210 }, { "epoch": 0.7121889691549135, "grad_norm": 35.0, "learning_rate": 2.0211073074487364e-06, "loss": 2.1487, "step": 6211 }, { "epoch": 0.7123036349042541, "grad_norm": 14.4375, "learning_rate": 2.0196161036266894e-06, "loss": 2.1416, "step": 6212 }, { "epoch": 0.7124183006535948, "grad_norm": 15.5625, "learning_rate": 2.0181253108917892e-06, "loss": 1.6877, "step": 6213 }, { "epoch": 0.7125329664029354, "grad_norm": 18.25, "learning_rate": 2.016634929449662e-06, "loss": 2.1369, "step": 6214 }, { "epoch": 0.7126476321522761, "grad_norm": 14.5625, "learning_rate": 2.015144959505879e-06, "loss": 1.8196, "step": 6215 }, { "epoch": 0.7127622979016168, "grad_norm": 32.25, "learning_rate": 2.0136554012659525e-06, "loss": 2.1272, "step": 6216 }, { "epoch": 0.7128769636509574, "grad_norm": 30.75, "learning_rate": 2.0121662549353405e-06, "loss": 1.9747, "step": 6217 }, { "epoch": 0.7129916294002981, "grad_norm": 13.75, "learning_rate": 2.010677520719438e-06, "loss": 1.7691, "step": 6218 }, { "epoch": 0.7131062951496389, "grad_norm": 14.9375, "learning_rate": 2.0091891988235877e-06, "loss": 2.0942, "step": 6219 }, { "epoch": 0.7132209608989795, "grad_norm": 14.25, "learning_rate": 2.007701289453082e-06, "loss": 1.6587, "step": 6220 }, { "epoch": 0.7133356266483202, "grad_norm": 16.0, "learning_rate": 2.006213792813142e-06, "loss": 1.7286, "step": 6221 }, { "epoch": 0.7134502923976608, "grad_norm": 20.5, "learning_rate": 2.0047267091089435e-06, "loss": 2.2335, "step": 6222 }, { "epoch": 0.7135649581470015, "grad_norm": 32.5, "learning_rate": 2.0032400385456018e-06, "loss": 2.1954, "step": 6223 }, { "epoch": 0.7136796238963422, "grad_norm": 16.875, "learning_rate": 2.0017537813281694e-06, "loss": 1.4422, "step": 6224 }, { "epoch": 0.7137942896456828, "grad_norm": 12.5625, "learning_rate": 2.0002679376616536e-06, "loss": 1.7962, "step": 6225 }, { "epoch": 0.7139089553950235, "grad_norm": 19.25, "learning_rate": 1.998782507750997e-06, "loss": 1.941, "step": 6226 }, { "epoch": 0.7140236211443641, "grad_norm": 14.25, "learning_rate": 1.9972974918010833e-06, "loss": 2.2484, "step": 6227 }, { "epoch": 0.7141382868937048, "grad_norm": 13.375, "learning_rate": 1.995812890016744e-06, "loss": 1.6147, "step": 6228 }, { "epoch": 0.7142529526430456, "grad_norm": 14.5625, "learning_rate": 1.994328702602752e-06, "loss": 1.6398, "step": 6229 }, { "epoch": 0.7143676183923862, "grad_norm": 14.5625, "learning_rate": 1.9928449297638213e-06, "loss": 1.6351, "step": 6230 }, { "epoch": 0.7144822841417269, "grad_norm": 19.25, "learning_rate": 1.9913615717046113e-06, "loss": 1.8912, "step": 6231 }, { "epoch": 0.7145969498910676, "grad_norm": 15.1875, "learning_rate": 1.9898786286297243e-06, "loss": 1.8637, "step": 6232 }, { "epoch": 0.7147116156404082, "grad_norm": 14.8125, "learning_rate": 1.988396100743701e-06, "loss": 1.583, "step": 6233 }, { "epoch": 0.7148262813897489, "grad_norm": 13.8125, "learning_rate": 1.986913988251027e-06, "loss": 1.6, "step": 6234 }, { "epoch": 0.7149409471390895, "grad_norm": 28.625, "learning_rate": 1.9854322913561376e-06, "loss": 1.669, "step": 6235 }, { "epoch": 0.7150556128884302, "grad_norm": 16.125, "learning_rate": 1.9839510102633984e-06, "loss": 1.9065, "step": 6236 }, { "epoch": 0.7151702786377709, "grad_norm": 15.625, "learning_rate": 1.9824701451771265e-06, "loss": 1.8431, "step": 6237 }, { "epoch": 0.7152849443871115, "grad_norm": 13.0625, "learning_rate": 1.9809896963015806e-06, "loss": 1.7652, "step": 6238 }, { "epoch": 0.7153996101364523, "grad_norm": 19.25, "learning_rate": 1.979509663840955e-06, "loss": 1.8267, "step": 6239 }, { "epoch": 0.7155142758857929, "grad_norm": 13.0, "learning_rate": 1.978030047999397e-06, "loss": 1.5928, "step": 6240 }, { "epoch": 0.7156289416351336, "grad_norm": 12.375, "learning_rate": 1.9765508489809915e-06, "loss": 1.9984, "step": 6241 }, { "epoch": 0.7157436073844743, "grad_norm": 13.625, "learning_rate": 1.9750720669897627e-06, "loss": 1.5493, "step": 6242 }, { "epoch": 0.7158582731338149, "grad_norm": 23.5, "learning_rate": 1.9735937022296817e-06, "loss": 2.2265, "step": 6243 }, { "epoch": 0.7159729388831556, "grad_norm": 13.875, "learning_rate": 1.9721157549046616e-06, "loss": 1.8136, "step": 6244 }, { "epoch": 0.7160876046324963, "grad_norm": 12.75, "learning_rate": 1.9706382252185556e-06, "loss": 1.6961, "step": 6245 }, { "epoch": 0.7162022703818369, "grad_norm": 18.625, "learning_rate": 1.969161113375162e-06, "loss": 2.2397, "step": 6246 }, { "epoch": 0.7163169361311776, "grad_norm": 13.25, "learning_rate": 1.9676844195782217e-06, "loss": 1.5355, "step": 6247 }, { "epoch": 0.7164316018805182, "grad_norm": 15.75, "learning_rate": 1.9662081440314124e-06, "loss": 2.1193, "step": 6248 }, { "epoch": 0.716546267629859, "grad_norm": 18.0, "learning_rate": 1.96473228693836e-06, "loss": 1.8658, "step": 6249 }, { "epoch": 0.7166609333791997, "grad_norm": 14.6875, "learning_rate": 1.9632568485026315e-06, "loss": 1.7766, "step": 6250 }, { "epoch": 0.7167755991285403, "grad_norm": 13.9375, "learning_rate": 1.961781828927736e-06, "loss": 1.5035, "step": 6251 }, { "epoch": 0.716890264877881, "grad_norm": 32.0, "learning_rate": 1.960307228417122e-06, "loss": 1.905, "step": 6252 }, { "epoch": 0.7170049306272217, "grad_norm": 14.875, "learning_rate": 1.9588330471741867e-06, "loss": 1.8453, "step": 6253 }, { "epoch": 0.7171195963765623, "grad_norm": 17.25, "learning_rate": 1.957359285402259e-06, "loss": 1.7383, "step": 6254 }, { "epoch": 0.717234262125903, "grad_norm": 13.375, "learning_rate": 1.955885943304623e-06, "loss": 1.8434, "step": 6255 }, { "epoch": 0.7173489278752436, "grad_norm": 14.1875, "learning_rate": 1.9544130210844936e-06, "loss": 1.7058, "step": 6256 }, { "epoch": 0.7174635936245843, "grad_norm": 13.875, "learning_rate": 1.9529405189450336e-06, "loss": 2.0045, "step": 6257 }, { "epoch": 0.717578259373925, "grad_norm": 14.25, "learning_rate": 1.9514684370893468e-06, "loss": 1.758, "step": 6258 }, { "epoch": 0.7176929251232657, "grad_norm": 17.75, "learning_rate": 1.9499967757204793e-06, "loss": 1.9112, "step": 6259 }, { "epoch": 0.7178075908726064, "grad_norm": 13.875, "learning_rate": 1.9485255350414174e-06, "loss": 1.5996, "step": 6260 }, { "epoch": 0.717922256621947, "grad_norm": 14.375, "learning_rate": 1.947054715255094e-06, "loss": 2.0352, "step": 6261 }, { "epoch": 0.7180369223712877, "grad_norm": 14.125, "learning_rate": 1.9455843165643754e-06, "loss": 1.6278, "step": 6262 }, { "epoch": 0.7181515881206284, "grad_norm": 18.625, "learning_rate": 1.944114339172079e-06, "loss": 1.9172, "step": 6263 }, { "epoch": 0.718266253869969, "grad_norm": 14.625, "learning_rate": 1.9426447832809585e-06, "loss": 1.7237, "step": 6264 }, { "epoch": 0.7183809196193097, "grad_norm": 15.5625, "learning_rate": 1.9411756490937117e-06, "loss": 1.4568, "step": 6265 }, { "epoch": 0.7184955853686504, "grad_norm": 13.25, "learning_rate": 1.939706936812978e-06, "loss": 1.9407, "step": 6266 }, { "epoch": 0.718610251117991, "grad_norm": 15.6875, "learning_rate": 1.9382386466413396e-06, "loss": 1.3855, "step": 6267 }, { "epoch": 0.7187249168673318, "grad_norm": 15.125, "learning_rate": 1.936770778781315e-06, "loss": 1.4654, "step": 6268 }, { "epoch": 0.7188395826166724, "grad_norm": 13.8125, "learning_rate": 1.93530333343537e-06, "loss": 1.6723, "step": 6269 }, { "epoch": 0.7189542483660131, "grad_norm": 15.125, "learning_rate": 1.9338363108059144e-06, "loss": 1.7318, "step": 6270 }, { "epoch": 0.7190689141153538, "grad_norm": 15.75, "learning_rate": 1.9323697110952923e-06, "loss": 2.0532, "step": 6271 }, { "epoch": 0.7191835798646944, "grad_norm": 15.25, "learning_rate": 1.930903534505794e-06, "loss": 1.787, "step": 6272 }, { "epoch": 0.7192982456140351, "grad_norm": 14.625, "learning_rate": 1.9294377812396526e-06, "loss": 1.518, "step": 6273 }, { "epoch": 0.7194129113633757, "grad_norm": 11.6875, "learning_rate": 1.9279724514990355e-06, "loss": 1.3803, "step": 6274 }, { "epoch": 0.7195275771127164, "grad_norm": 13.0, "learning_rate": 1.926507545486062e-06, "loss": 1.8015, "step": 6275 }, { "epoch": 0.7196422428620571, "grad_norm": 13.125, "learning_rate": 1.9250430634027885e-06, "loss": 1.8291, "step": 6276 }, { "epoch": 0.7197569086113977, "grad_norm": 18.375, "learning_rate": 1.9235790054512086e-06, "loss": 2.2424, "step": 6277 }, { "epoch": 0.7198715743607385, "grad_norm": 15.9375, "learning_rate": 1.922115371833263e-06, "loss": 1.5957, "step": 6278 }, { "epoch": 0.7199862401100792, "grad_norm": 17.125, "learning_rate": 1.9206521627508324e-06, "loss": 1.9563, "step": 6279 }, { "epoch": 0.7201009058594198, "grad_norm": 18.25, "learning_rate": 1.9191893784057377e-06, "loss": 2.0813, "step": 6280 }, { "epoch": 0.7202155716087605, "grad_norm": 15.125, "learning_rate": 1.9177270189997427e-06, "loss": 2.0032, "step": 6281 }, { "epoch": 0.7203302373581011, "grad_norm": 17.0, "learning_rate": 1.9162650847345543e-06, "loss": 1.5657, "step": 6282 }, { "epoch": 0.7204449031074418, "grad_norm": 14.625, "learning_rate": 1.914803575811814e-06, "loss": 2.1541, "step": 6283 }, { "epoch": 0.7205595688567825, "grad_norm": 13.6875, "learning_rate": 1.913342492433109e-06, "loss": 1.8736, "step": 6284 }, { "epoch": 0.7206742346061231, "grad_norm": 13.375, "learning_rate": 1.911881834799974e-06, "loss": 1.7186, "step": 6285 }, { "epoch": 0.7207889003554638, "grad_norm": 15.5625, "learning_rate": 1.9104216031138735e-06, "loss": 1.9776, "step": 6286 }, { "epoch": 0.7209035661048045, "grad_norm": 33.5, "learning_rate": 1.9089617975762196e-06, "loss": 1.941, "step": 6287 }, { "epoch": 0.7210182318541452, "grad_norm": 17.5, "learning_rate": 1.907502418388367e-06, "loss": 1.7583, "step": 6288 }, { "epoch": 0.7211328976034859, "grad_norm": 15.375, "learning_rate": 1.9060434657516048e-06, "loss": 1.7193, "step": 6289 }, { "epoch": 0.7212475633528265, "grad_norm": 13.75, "learning_rate": 1.9045849398671712e-06, "loss": 1.948, "step": 6290 }, { "epoch": 0.7213622291021672, "grad_norm": 13.8125, "learning_rate": 1.903126840936243e-06, "loss": 1.9785, "step": 6291 }, { "epoch": 0.7214768948515079, "grad_norm": 24.0, "learning_rate": 1.9016691691599338e-06, "loss": 1.7441, "step": 6292 }, { "epoch": 0.7215915606008485, "grad_norm": 16.75, "learning_rate": 1.9002119247393032e-06, "loss": 1.7174, "step": 6293 }, { "epoch": 0.7217062263501892, "grad_norm": 21.875, "learning_rate": 1.8987551078753501e-06, "loss": 1.784, "step": 6294 }, { "epoch": 0.7218208920995298, "grad_norm": 12.9375, "learning_rate": 1.8972987187690151e-06, "loss": 1.4386, "step": 6295 }, { "epoch": 0.7219355578488705, "grad_norm": 12.125, "learning_rate": 1.8958427576211785e-06, "loss": 1.5033, "step": 6296 }, { "epoch": 0.7220502235982112, "grad_norm": 14.6875, "learning_rate": 1.8943872246326644e-06, "loss": 1.7064, "step": 6297 }, { "epoch": 0.7221648893475519, "grad_norm": 38.75, "learning_rate": 1.8929321200042332e-06, "loss": 1.9006, "step": 6298 }, { "epoch": 0.7222795550968926, "grad_norm": 13.625, "learning_rate": 1.8914774439365896e-06, "loss": 1.7076, "step": 6299 }, { "epoch": 0.7223942208462333, "grad_norm": 14.8125, "learning_rate": 1.8900231966303788e-06, "loss": 1.6468, "step": 6300 }, { "epoch": 0.7225088865955739, "grad_norm": 21.625, "learning_rate": 1.888569378286187e-06, "loss": 1.6895, "step": 6301 }, { "epoch": 0.7226235523449146, "grad_norm": 17.0, "learning_rate": 1.8871159891045393e-06, "loss": 1.591, "step": 6302 }, { "epoch": 0.7227382180942552, "grad_norm": 18.125, "learning_rate": 1.8856630292859063e-06, "loss": 2.0351, "step": 6303 }, { "epoch": 0.7228528838435959, "grad_norm": 14.25, "learning_rate": 1.8842104990306897e-06, "loss": 1.789, "step": 6304 }, { "epoch": 0.7229675495929366, "grad_norm": 13.5625, "learning_rate": 1.8827583985392467e-06, "loss": 1.5737, "step": 6305 }, { "epoch": 0.7230822153422772, "grad_norm": 13.5, "learning_rate": 1.8813067280118603e-06, "loss": 1.6811, "step": 6306 }, { "epoch": 0.723196881091618, "grad_norm": 16.375, "learning_rate": 1.8798554876487629e-06, "loss": 1.8861, "step": 6307 }, { "epoch": 0.7233115468409586, "grad_norm": 13.0625, "learning_rate": 1.8784046776501264e-06, "loss": 1.4691, "step": 6308 }, { "epoch": 0.7234262125902993, "grad_norm": 17.125, "learning_rate": 1.876954298216061e-06, "loss": 1.346, "step": 6309 }, { "epoch": 0.72354087833964, "grad_norm": 17.625, "learning_rate": 1.8755043495466197e-06, "loss": 1.9646, "step": 6310 }, { "epoch": 0.7236555440889806, "grad_norm": 11.875, "learning_rate": 1.874054831841796e-06, "loss": 2.0511, "step": 6311 }, { "epoch": 0.7237702098383213, "grad_norm": 13.9375, "learning_rate": 1.8726057453015218e-06, "loss": 2.1494, "step": 6312 }, { "epoch": 0.723884875587662, "grad_norm": 14.5625, "learning_rate": 1.8711570901256709e-06, "loss": 1.6367, "step": 6313 }, { "epoch": 0.7239995413370026, "grad_norm": 13.5625, "learning_rate": 1.8697088665140584e-06, "loss": 1.5818, "step": 6314 }, { "epoch": 0.7241142070863433, "grad_norm": 12.9375, "learning_rate": 1.8682610746664388e-06, "loss": 1.8454, "step": 6315 }, { "epoch": 0.7242288728356839, "grad_norm": 14.5625, "learning_rate": 1.8668137147825077e-06, "loss": 1.8725, "step": 6316 }, { "epoch": 0.7243435385850246, "grad_norm": 14.625, "learning_rate": 1.8653667870619025e-06, "loss": 1.8912, "step": 6317 }, { "epoch": 0.7244582043343654, "grad_norm": 14.5625, "learning_rate": 1.8639202917041953e-06, "loss": 1.9257, "step": 6318 }, { "epoch": 0.724572870083706, "grad_norm": 13.125, "learning_rate": 1.8624742289089033e-06, "loss": 1.8447, "step": 6319 }, { "epoch": 0.7246875358330467, "grad_norm": 17.625, "learning_rate": 1.8610285988754884e-06, "loss": 1.9269, "step": 6320 }, { "epoch": 0.7248022015823874, "grad_norm": 18.375, "learning_rate": 1.8595834018033421e-06, "loss": 1.3517, "step": 6321 }, { "epoch": 0.724916867331728, "grad_norm": 13.9375, "learning_rate": 1.8581386378918036e-06, "loss": 1.8312, "step": 6322 }, { "epoch": 0.7250315330810687, "grad_norm": 13.0625, "learning_rate": 1.8566943073401527e-06, "loss": 1.4148, "step": 6323 }, { "epoch": 0.7251461988304093, "grad_norm": 16.125, "learning_rate": 1.8552504103476015e-06, "loss": 2.1546, "step": 6324 }, { "epoch": 0.72526086457975, "grad_norm": 14.0625, "learning_rate": 1.8538069471133142e-06, "loss": 1.7785, "step": 6325 }, { "epoch": 0.7253755303290907, "grad_norm": 18.125, "learning_rate": 1.8523639178363878e-06, "loss": 2.1762, "step": 6326 }, { "epoch": 0.7254901960784313, "grad_norm": 16.375, "learning_rate": 1.8509213227158578e-06, "loss": 2.1025, "step": 6327 }, { "epoch": 0.7256048618277721, "grad_norm": 16.75, "learning_rate": 1.8494791619507046e-06, "loss": 2.44, "step": 6328 }, { "epoch": 0.7257195275771127, "grad_norm": 20.75, "learning_rate": 1.848037435739846e-06, "loss": 1.6205, "step": 6329 }, { "epoch": 0.7258341933264534, "grad_norm": 12.75, "learning_rate": 1.8465961442821422e-06, "loss": 2.1832, "step": 6330 }, { "epoch": 0.7259488590757941, "grad_norm": 30.125, "learning_rate": 1.8451552877763907e-06, "loss": 1.6678, "step": 6331 }, { "epoch": 0.7260635248251347, "grad_norm": 17.25, "learning_rate": 1.8437148664213317e-06, "loss": 1.8445, "step": 6332 }, { "epoch": 0.7261781905744754, "grad_norm": 11.8125, "learning_rate": 1.8422748804156415e-06, "loss": 1.9037, "step": 6333 }, { "epoch": 0.7262928563238161, "grad_norm": 11.9375, "learning_rate": 1.8408353299579379e-06, "loss": 1.8257, "step": 6334 }, { "epoch": 0.7264075220731567, "grad_norm": 13.9375, "learning_rate": 1.8393962152467848e-06, "loss": 1.7003, "step": 6335 }, { "epoch": 0.7265221878224974, "grad_norm": 15.5, "learning_rate": 1.837957536480675e-06, "loss": 2.0705, "step": 6336 }, { "epoch": 0.726636853571838, "grad_norm": 17.25, "learning_rate": 1.8365192938580496e-06, "loss": 1.719, "step": 6337 }, { "epoch": 0.7267515193211788, "grad_norm": 12.625, "learning_rate": 1.8350814875772876e-06, "loss": 1.6147, "step": 6338 }, { "epoch": 0.7268661850705195, "grad_norm": 18.25, "learning_rate": 1.8336441178367027e-06, "loss": 1.5763, "step": 6339 }, { "epoch": 0.7269808508198601, "grad_norm": 13.4375, "learning_rate": 1.8322071848345568e-06, "loss": 1.4002, "step": 6340 }, { "epoch": 0.7270955165692008, "grad_norm": 13.875, "learning_rate": 1.830770688769048e-06, "loss": 1.5845, "step": 6341 }, { "epoch": 0.7272101823185414, "grad_norm": 12.125, "learning_rate": 1.8293346298383097e-06, "loss": 1.8346, "step": 6342 }, { "epoch": 0.7273248480678821, "grad_norm": 14.6875, "learning_rate": 1.8278990082404207e-06, "loss": 1.457, "step": 6343 }, { "epoch": 0.7274395138172228, "grad_norm": 14.0, "learning_rate": 1.8264638241733983e-06, "loss": 1.4092, "step": 6344 }, { "epoch": 0.7275541795665634, "grad_norm": 13.5, "learning_rate": 1.8250290778351982e-06, "loss": 1.6319, "step": 6345 }, { "epoch": 0.7276688453159041, "grad_norm": 15.9375, "learning_rate": 1.8235947694237161e-06, "loss": 1.7242, "step": 6346 }, { "epoch": 0.7277835110652449, "grad_norm": 13.125, "learning_rate": 1.8221608991367889e-06, "loss": 1.9044, "step": 6347 }, { "epoch": 0.7278981768145855, "grad_norm": 12.5625, "learning_rate": 1.8207274671721897e-06, "loss": 1.3802, "step": 6348 }, { "epoch": 0.7280128425639262, "grad_norm": 15.75, "learning_rate": 1.8192944737276335e-06, "loss": 1.4284, "step": 6349 }, { "epoch": 0.7281275083132668, "grad_norm": 14.0, "learning_rate": 1.817861919000775e-06, "loss": 1.6913, "step": 6350 }, { "epoch": 0.7282421740626075, "grad_norm": 15.6875, "learning_rate": 1.816429803189208e-06, "loss": 1.8306, "step": 6351 }, { "epoch": 0.7283568398119482, "grad_norm": 14.9375, "learning_rate": 1.8149981264904653e-06, "loss": 1.897, "step": 6352 }, { "epoch": 0.7284715055612888, "grad_norm": 17.125, "learning_rate": 1.8135668891020208e-06, "loss": 2.2005, "step": 6353 }, { "epoch": 0.7285861713106295, "grad_norm": 17.25, "learning_rate": 1.8121360912212816e-06, "loss": 2.1909, "step": 6354 }, { "epoch": 0.7287008370599702, "grad_norm": 13.3125, "learning_rate": 1.8107057330456058e-06, "loss": 1.972, "step": 6355 }, { "epoch": 0.7288155028093108, "grad_norm": 13.6875, "learning_rate": 1.8092758147722795e-06, "loss": 1.7909, "step": 6356 }, { "epoch": 0.7289301685586516, "grad_norm": 14.4375, "learning_rate": 1.8078463365985332e-06, "loss": 1.8779, "step": 6357 }, { "epoch": 0.7290448343079922, "grad_norm": 17.625, "learning_rate": 1.8064172987215368e-06, "loss": 1.9694, "step": 6358 }, { "epoch": 0.7291595000573329, "grad_norm": 14.375, "learning_rate": 1.8049887013383988e-06, "loss": 1.8928, "step": 6359 }, { "epoch": 0.7292741658066736, "grad_norm": 13.5, "learning_rate": 1.8035605446461673e-06, "loss": 1.483, "step": 6360 }, { "epoch": 0.7293888315560142, "grad_norm": 14.25, "learning_rate": 1.8021328288418304e-06, "loss": 1.7073, "step": 6361 }, { "epoch": 0.7295034973053549, "grad_norm": 12.6875, "learning_rate": 1.8007055541223111e-06, "loss": 1.7754, "step": 6362 }, { "epoch": 0.7296181630546955, "grad_norm": 16.5, "learning_rate": 1.799278720684477e-06, "loss": 1.805, "step": 6363 }, { "epoch": 0.7297328288040362, "grad_norm": 17.75, "learning_rate": 1.7978523287251314e-06, "loss": 2.4812, "step": 6364 }, { "epoch": 0.7298474945533769, "grad_norm": 14.6875, "learning_rate": 1.7964263784410192e-06, "loss": 1.674, "step": 6365 }, { "epoch": 0.7299621603027175, "grad_norm": 13.5625, "learning_rate": 1.7950008700288224e-06, "loss": 2.0569, "step": 6366 }, { "epoch": 0.7300768260520583, "grad_norm": 15.8125, "learning_rate": 1.7935758036851642e-06, "loss": 1.7585, "step": 6367 }, { "epoch": 0.730191491801399, "grad_norm": 14.25, "learning_rate": 1.792151179606602e-06, "loss": 1.5552, "step": 6368 }, { "epoch": 0.7303061575507396, "grad_norm": 15.9375, "learning_rate": 1.7907269979896359e-06, "loss": 1.9862, "step": 6369 }, { "epoch": 0.7304208233000803, "grad_norm": 15.4375, "learning_rate": 1.7893032590307096e-06, "loss": 2.0623, "step": 6370 }, { "epoch": 0.7305354890494209, "grad_norm": 13.625, "learning_rate": 1.7878799629261956e-06, "loss": 2.0886, "step": 6371 }, { "epoch": 0.7306501547987616, "grad_norm": 14.1875, "learning_rate": 1.786457109872412e-06, "loss": 1.6088, "step": 6372 }, { "epoch": 0.7307648205481023, "grad_norm": 14.3125, "learning_rate": 1.785034700065616e-06, "loss": 1.6242, "step": 6373 }, { "epoch": 0.7308794862974429, "grad_norm": 13.8125, "learning_rate": 1.783612733701997e-06, "loss": 1.6896, "step": 6374 }, { "epoch": 0.7309941520467836, "grad_norm": 13.25, "learning_rate": 1.7821912109776928e-06, "loss": 1.4727, "step": 6375 }, { "epoch": 0.7311088177961242, "grad_norm": 14.375, "learning_rate": 1.7807701320887765e-06, "loss": 1.8194, "step": 6376 }, { "epoch": 0.731223483545465, "grad_norm": 19.0, "learning_rate": 1.7793494972312541e-06, "loss": 1.4742, "step": 6377 }, { "epoch": 0.7313381492948057, "grad_norm": 21.0, "learning_rate": 1.7779293066010778e-06, "loss": 1.751, "step": 6378 }, { "epoch": 0.7314528150441463, "grad_norm": 13.8125, "learning_rate": 1.7765095603941352e-06, "loss": 1.5971, "step": 6379 }, { "epoch": 0.731567480793487, "grad_norm": 12.375, "learning_rate": 1.7750902588062536e-06, "loss": 1.5967, "step": 6380 }, { "epoch": 0.7316821465428277, "grad_norm": 13.875, "learning_rate": 1.773671402033198e-06, "loss": 1.6698, "step": 6381 }, { "epoch": 0.7317968122921683, "grad_norm": 17.25, "learning_rate": 1.7722529902706748e-06, "loss": 1.6129, "step": 6382 }, { "epoch": 0.731911478041509, "grad_norm": 16.0, "learning_rate": 1.7708350237143236e-06, "loss": 1.8706, "step": 6383 }, { "epoch": 0.7320261437908496, "grad_norm": 15.0625, "learning_rate": 1.769417502559726e-06, "loss": 1.772, "step": 6384 }, { "epoch": 0.7321408095401903, "grad_norm": 15.0, "learning_rate": 1.7680004270024064e-06, "loss": 1.6534, "step": 6385 }, { "epoch": 0.732255475289531, "grad_norm": 17.125, "learning_rate": 1.7665837972378186e-06, "loss": 1.7005, "step": 6386 }, { "epoch": 0.7323701410388717, "grad_norm": 11.4375, "learning_rate": 1.7651676134613621e-06, "loss": 1.789, "step": 6387 }, { "epoch": 0.7324848067882124, "grad_norm": 11.25, "learning_rate": 1.763751875868373e-06, "loss": 1.4947, "step": 6388 }, { "epoch": 0.7325994725375531, "grad_norm": 16.0, "learning_rate": 1.7623365846541208e-06, "loss": 1.4596, "step": 6389 }, { "epoch": 0.7327141382868937, "grad_norm": 23.0, "learning_rate": 1.760921740013823e-06, "loss": 1.845, "step": 6390 }, { "epoch": 0.7328288040362344, "grad_norm": 16.625, "learning_rate": 1.75950734214263e-06, "loss": 1.754, "step": 6391 }, { "epoch": 0.732943469785575, "grad_norm": 18.625, "learning_rate": 1.7580933912356285e-06, "loss": 1.5805, "step": 6392 }, { "epoch": 0.7330581355349157, "grad_norm": 14.375, "learning_rate": 1.7566798874878472e-06, "loss": 1.7378, "step": 6393 }, { "epoch": 0.7331728012842564, "grad_norm": 15.5, "learning_rate": 1.7552668310942521e-06, "loss": 1.5979, "step": 6394 }, { "epoch": 0.733287467033597, "grad_norm": 15.8125, "learning_rate": 1.7538542222497479e-06, "loss": 1.3873, "step": 6395 }, { "epoch": 0.7334021327829378, "grad_norm": 14.1875, "learning_rate": 1.752442061149176e-06, "loss": 2.002, "step": 6396 }, { "epoch": 0.7335167985322784, "grad_norm": 10.875, "learning_rate": 1.751030347987319e-06, "loss": 1.4899, "step": 6397 }, { "epoch": 0.7336314642816191, "grad_norm": 32.25, "learning_rate": 1.7496190829588937e-06, "loss": 1.8899, "step": 6398 }, { "epoch": 0.7337461300309598, "grad_norm": 15.5625, "learning_rate": 1.7482082662585576e-06, "loss": 1.398, "step": 6399 }, { "epoch": 0.7338607957803004, "grad_norm": 14.5, "learning_rate": 1.7467978980809064e-06, "loss": 1.9283, "step": 6400 }, { "epoch": 0.7339754615296411, "grad_norm": 13.8125, "learning_rate": 1.7453879786204735e-06, "loss": 1.8805, "step": 6401 }, { "epoch": 0.7340901272789818, "grad_norm": 20.875, "learning_rate": 1.7439785080717303e-06, "loss": 1.3969, "step": 6402 }, { "epoch": 0.7342047930283224, "grad_norm": 17.625, "learning_rate": 1.7425694866290887e-06, "loss": 2.1716, "step": 6403 }, { "epoch": 0.7343194587776631, "grad_norm": 31.375, "learning_rate": 1.7411609144868902e-06, "loss": 1.8965, "step": 6404 }, { "epoch": 0.7344341245270037, "grad_norm": 15.125, "learning_rate": 1.739752791839428e-06, "loss": 1.9197, "step": 6405 }, { "epoch": 0.7345487902763445, "grad_norm": 17.75, "learning_rate": 1.7383451188809203e-06, "loss": 2.1362, "step": 6406 }, { "epoch": 0.7346634560256852, "grad_norm": 32.25, "learning_rate": 1.736937895805531e-06, "loss": 2.2399, "step": 6407 }, { "epoch": 0.7347781217750258, "grad_norm": 11.625, "learning_rate": 1.735531122807359e-06, "loss": 1.6279, "step": 6408 }, { "epoch": 0.7348927875243665, "grad_norm": 18.375, "learning_rate": 1.7341248000804423e-06, "loss": 1.8035, "step": 6409 }, { "epoch": 0.7350074532737071, "grad_norm": 19.75, "learning_rate": 1.7327189278187561e-06, "loss": 1.8918, "step": 6410 }, { "epoch": 0.7351221190230478, "grad_norm": 11.3125, "learning_rate": 1.7313135062162157e-06, "loss": 1.1825, "step": 6411 }, { "epoch": 0.7352367847723885, "grad_norm": 13.625, "learning_rate": 1.7299085354666685e-06, "loss": 1.5692, "step": 6412 }, { "epoch": 0.7353514505217291, "grad_norm": 12.875, "learning_rate": 1.7285040157639055e-06, "loss": 1.9502, "step": 6413 }, { "epoch": 0.7354661162710698, "grad_norm": 13.1875, "learning_rate": 1.7270999473016524e-06, "loss": 1.5212, "step": 6414 }, { "epoch": 0.7355807820204106, "grad_norm": 29.25, "learning_rate": 1.7256963302735752e-06, "loss": 1.5468, "step": 6415 }, { "epoch": 0.7356954477697512, "grad_norm": 16.625, "learning_rate": 1.7242931648732746e-06, "loss": 2.4258, "step": 6416 }, { "epoch": 0.7358101135190919, "grad_norm": 13.875, "learning_rate": 1.7228904512942934e-06, "loss": 1.8072, "step": 6417 }, { "epoch": 0.7359247792684325, "grad_norm": 14.5, "learning_rate": 1.7214881897301055e-06, "loss": 1.6813, "step": 6418 }, { "epoch": 0.7360394450177732, "grad_norm": 19.125, "learning_rate": 1.7200863803741258e-06, "loss": 1.7928, "step": 6419 }, { "epoch": 0.7361541107671139, "grad_norm": 13.3125, "learning_rate": 1.7186850234197122e-06, "loss": 1.9915, "step": 6420 }, { "epoch": 0.7362687765164545, "grad_norm": 13.5625, "learning_rate": 1.7172841190601507e-06, "loss": 1.6464, "step": 6421 }, { "epoch": 0.7363834422657952, "grad_norm": 13.1875, "learning_rate": 1.7158836674886708e-06, "loss": 1.5327, "step": 6422 }, { "epoch": 0.7364981080151359, "grad_norm": 16.25, "learning_rate": 1.7144836688984395e-06, "loss": 1.9079, "step": 6423 }, { "epoch": 0.7366127737644765, "grad_norm": 15.9375, "learning_rate": 1.7130841234825552e-06, "loss": 1.6345, "step": 6424 }, { "epoch": 0.7367274395138173, "grad_norm": 19.375, "learning_rate": 1.7116850314340634e-06, "loss": 1.5495, "step": 6425 }, { "epoch": 0.7368421052631579, "grad_norm": 15.25, "learning_rate": 1.7102863929459424e-06, "loss": 1.2912, "step": 6426 }, { "epoch": 0.7369567710124986, "grad_norm": 14.0, "learning_rate": 1.7088882082111035e-06, "loss": 1.6104, "step": 6427 }, { "epoch": 0.7370714367618393, "grad_norm": 14.4375, "learning_rate": 1.7074904774224027e-06, "loss": 1.4933, "step": 6428 }, { "epoch": 0.7371861025111799, "grad_norm": 16.75, "learning_rate": 1.7060932007726293e-06, "loss": 1.8426, "step": 6429 }, { "epoch": 0.7373007682605206, "grad_norm": 17.5, "learning_rate": 1.7046963784545113e-06, "loss": 1.7626, "step": 6430 }, { "epoch": 0.7374154340098612, "grad_norm": 14.375, "learning_rate": 1.7033000106607133e-06, "loss": 1.3275, "step": 6431 }, { "epoch": 0.7375300997592019, "grad_norm": 15.6875, "learning_rate": 1.7019040975838402e-06, "loss": 1.8862, "step": 6432 }, { "epoch": 0.7376447655085426, "grad_norm": 13.5625, "learning_rate": 1.7005086394164278e-06, "loss": 1.7083, "step": 6433 }, { "epoch": 0.7377594312578832, "grad_norm": 14.3125, "learning_rate": 1.6991136363509525e-06, "loss": 1.7618, "step": 6434 }, { "epoch": 0.737874097007224, "grad_norm": 25.75, "learning_rate": 1.697719088579834e-06, "loss": 1.7105, "step": 6435 }, { "epoch": 0.7379887627565647, "grad_norm": 25.75, "learning_rate": 1.6963249962954186e-06, "loss": 1.7605, "step": 6436 }, { "epoch": 0.7381034285059053, "grad_norm": 16.375, "learning_rate": 1.6949313596899958e-06, "loss": 2.2399, "step": 6437 }, { "epoch": 0.738218094255246, "grad_norm": 23.5, "learning_rate": 1.6935381789557936e-06, "loss": 1.691, "step": 6438 }, { "epoch": 0.7383327600045866, "grad_norm": 16.5, "learning_rate": 1.6921454542849692e-06, "loss": 1.887, "step": 6439 }, { "epoch": 0.7384474257539273, "grad_norm": 13.25, "learning_rate": 1.6907531858696274e-06, "loss": 1.5905, "step": 6440 }, { "epoch": 0.738562091503268, "grad_norm": 13.375, "learning_rate": 1.6893613739018055e-06, "loss": 1.5613, "step": 6441 }, { "epoch": 0.7386767572526086, "grad_norm": 12.875, "learning_rate": 1.6879700185734738e-06, "loss": 1.7403, "step": 6442 }, { "epoch": 0.7387914230019493, "grad_norm": 13.625, "learning_rate": 1.6865791200765452e-06, "loss": 1.9715, "step": 6443 }, { "epoch": 0.7389060887512899, "grad_norm": 12.375, "learning_rate": 1.685188678602867e-06, "loss": 1.5933, "step": 6444 }, { "epoch": 0.7390207545006306, "grad_norm": 15.375, "learning_rate": 1.6837986943442253e-06, "loss": 2.0241, "step": 6445 }, { "epoch": 0.7391354202499714, "grad_norm": 16.375, "learning_rate": 1.6824091674923403e-06, "loss": 1.8089, "step": 6446 }, { "epoch": 0.739250085999312, "grad_norm": 13.1875, "learning_rate": 1.6810200982388736e-06, "loss": 1.6147, "step": 6447 }, { "epoch": 0.7393647517486527, "grad_norm": 14.25, "learning_rate": 1.6796314867754176e-06, "loss": 1.7008, "step": 6448 }, { "epoch": 0.7394794174979934, "grad_norm": 18.125, "learning_rate": 1.6782433332935054e-06, "loss": 2.4772, "step": 6449 }, { "epoch": 0.739594083247334, "grad_norm": 14.75, "learning_rate": 1.6768556379846073e-06, "loss": 1.5811, "step": 6450 }, { "epoch": 0.7397087489966747, "grad_norm": 13.9375, "learning_rate": 1.675468401040129e-06, "loss": 1.3662, "step": 6451 }, { "epoch": 0.7398234147460153, "grad_norm": 15.375, "learning_rate": 1.6740816226514135e-06, "loss": 1.8899, "step": 6452 }, { "epoch": 0.739938080495356, "grad_norm": 14.5, "learning_rate": 1.672695303009742e-06, "loss": 1.7117, "step": 6453 }, { "epoch": 0.7400527462446967, "grad_norm": 16.625, "learning_rate": 1.6713094423063258e-06, "loss": 1.451, "step": 6454 }, { "epoch": 0.7401674119940373, "grad_norm": 13.5, "learning_rate": 1.6699240407323246e-06, "loss": 1.5678, "step": 6455 }, { "epoch": 0.7402820777433781, "grad_norm": 15.0, "learning_rate": 1.6685390984788236e-06, "loss": 1.838, "step": 6456 }, { "epoch": 0.7403967434927188, "grad_norm": 12.75, "learning_rate": 1.6671546157368506e-06, "loss": 1.7085, "step": 6457 }, { "epoch": 0.7405114092420594, "grad_norm": 15.75, "learning_rate": 1.6657705926973682e-06, "loss": 1.4369, "step": 6458 }, { "epoch": 0.7406260749914001, "grad_norm": 14.1875, "learning_rate": 1.6643870295512765e-06, "loss": 2.0743, "step": 6459 }, { "epoch": 0.7407407407407407, "grad_norm": 15.5, "learning_rate": 1.6630039264894115e-06, "loss": 1.656, "step": 6460 }, { "epoch": 0.7408554064900814, "grad_norm": 14.875, "learning_rate": 1.6616212837025475e-06, "loss": 1.6584, "step": 6461 }, { "epoch": 0.7409700722394221, "grad_norm": 13.4375, "learning_rate": 1.6602391013813906e-06, "loss": 1.5897, "step": 6462 }, { "epoch": 0.7410847379887627, "grad_norm": 15.75, "learning_rate": 1.658857379716588e-06, "loss": 2.2658, "step": 6463 }, { "epoch": 0.7411994037381034, "grad_norm": 26.25, "learning_rate": 1.657476118898722e-06, "loss": 1.7421, "step": 6464 }, { "epoch": 0.741314069487444, "grad_norm": 13.75, "learning_rate": 1.656095319118311e-06, "loss": 1.8796, "step": 6465 }, { "epoch": 0.7414287352367848, "grad_norm": 14.9375, "learning_rate": 1.6547149805658103e-06, "loss": 1.6804, "step": 6466 }, { "epoch": 0.7415434009861255, "grad_norm": 13.375, "learning_rate": 1.6533351034316125e-06, "loss": 1.86, "step": 6467 }, { "epoch": 0.7416580667354661, "grad_norm": 46.75, "learning_rate": 1.6519556879060422e-06, "loss": 1.7322, "step": 6468 }, { "epoch": 0.7417727324848068, "grad_norm": 18.0, "learning_rate": 1.650576734179364e-06, "loss": 2.2473, "step": 6469 }, { "epoch": 0.7418873982341475, "grad_norm": 14.3125, "learning_rate": 1.6491982424417823e-06, "loss": 1.5845, "step": 6470 }, { "epoch": 0.7420020639834881, "grad_norm": 12.875, "learning_rate": 1.6478202128834297e-06, "loss": 1.5102, "step": 6471 }, { "epoch": 0.7421167297328288, "grad_norm": 10.6875, "learning_rate": 1.6464426456943799e-06, "loss": 1.348, "step": 6472 }, { "epoch": 0.7422313954821694, "grad_norm": 14.5, "learning_rate": 1.6450655410646449e-06, "loss": 1.8868, "step": 6473 }, { "epoch": 0.7423460612315101, "grad_norm": 13.625, "learning_rate": 1.6436888991841637e-06, "loss": 1.5963, "step": 6474 }, { "epoch": 0.7424607269808509, "grad_norm": 15.1875, "learning_rate": 1.6423127202428235e-06, "loss": 1.5973, "step": 6475 }, { "epoch": 0.7425753927301915, "grad_norm": 16.5, "learning_rate": 1.6409370044304418e-06, "loss": 1.731, "step": 6476 }, { "epoch": 0.7426900584795322, "grad_norm": 14.3125, "learning_rate": 1.6395617519367695e-06, "loss": 1.7547, "step": 6477 }, { "epoch": 0.7428047242288728, "grad_norm": 13.5625, "learning_rate": 1.6381869629514973e-06, "loss": 1.6586, "step": 6478 }, { "epoch": 0.7429193899782135, "grad_norm": 12.3125, "learning_rate": 1.6368126376642519e-06, "loss": 1.4087, "step": 6479 }, { "epoch": 0.7430340557275542, "grad_norm": 12.5625, "learning_rate": 1.6354387762645952e-06, "loss": 1.4923, "step": 6480 }, { "epoch": 0.7431487214768948, "grad_norm": 16.75, "learning_rate": 1.6340653789420247e-06, "loss": 1.467, "step": 6481 }, { "epoch": 0.7432633872262355, "grad_norm": 13.5625, "learning_rate": 1.632692445885976e-06, "loss": 1.5244, "step": 6482 }, { "epoch": 0.7433780529755762, "grad_norm": 14.8125, "learning_rate": 1.6313199772858163e-06, "loss": 1.9815, "step": 6483 }, { "epoch": 0.7434927187249168, "grad_norm": 12.125, "learning_rate": 1.629947973330851e-06, "loss": 1.614, "step": 6484 }, { "epoch": 0.7436073844742576, "grad_norm": 14.625, "learning_rate": 1.6285764342103272e-06, "loss": 2.1037, "step": 6485 }, { "epoch": 0.7437220502235982, "grad_norm": 13.1875, "learning_rate": 1.6272053601134169e-06, "loss": 1.6896, "step": 6486 }, { "epoch": 0.7438367159729389, "grad_norm": 14.8125, "learning_rate": 1.6258347512292367e-06, "loss": 2.017, "step": 6487 }, { "epoch": 0.7439513817222796, "grad_norm": 12.75, "learning_rate": 1.6244646077468363e-06, "loss": 2.1097, "step": 6488 }, { "epoch": 0.7440660474716202, "grad_norm": 21.125, "learning_rate": 1.6230949298551968e-06, "loss": 1.9998, "step": 6489 }, { "epoch": 0.7441807132209609, "grad_norm": 16.625, "learning_rate": 1.6217257177432433e-06, "loss": 1.7485, "step": 6490 }, { "epoch": 0.7442953789703016, "grad_norm": 16.25, "learning_rate": 1.620356971599833e-06, "loss": 2.1267, "step": 6491 }, { "epoch": 0.7444100447196422, "grad_norm": 14.6875, "learning_rate": 1.6189886916137548e-06, "loss": 1.8695, "step": 6492 }, { "epoch": 0.7445247104689829, "grad_norm": 17.0, "learning_rate": 1.617620877973739e-06, "loss": 2.222, "step": 6493 }, { "epoch": 0.7446393762183235, "grad_norm": 13.25, "learning_rate": 1.616253530868449e-06, "loss": 1.5002, "step": 6494 }, { "epoch": 0.7447540419676643, "grad_norm": 16.125, "learning_rate": 1.6148866504864841e-06, "loss": 2.1354, "step": 6495 }, { "epoch": 0.744868707717005, "grad_norm": 13.6875, "learning_rate": 1.6135202370163798e-06, "loss": 1.5831, "step": 6496 }, { "epoch": 0.7449833734663456, "grad_norm": 14.375, "learning_rate": 1.6121542906466086e-06, "loss": 1.8672, "step": 6497 }, { "epoch": 0.7450980392156863, "grad_norm": 13.25, "learning_rate": 1.6107888115655729e-06, "loss": 1.6093, "step": 6498 }, { "epoch": 0.7452127049650269, "grad_norm": 14.75, "learning_rate": 1.6094237999616164e-06, "loss": 1.709, "step": 6499 }, { "epoch": 0.7453273707143676, "grad_norm": 15.5625, "learning_rate": 1.6080592560230163e-06, "loss": 1.9677, "step": 6500 }, { "epoch": 0.7454420364637083, "grad_norm": 15.4375, "learning_rate": 1.6066951799379854e-06, "loss": 1.5942, "step": 6501 }, { "epoch": 0.7455567022130489, "grad_norm": 13.4375, "learning_rate": 1.6053315718946728e-06, "loss": 1.8267, "step": 6502 }, { "epoch": 0.7456713679623896, "grad_norm": 14.4375, "learning_rate": 1.6039684320811628e-06, "loss": 2.2005, "step": 6503 }, { "epoch": 0.7457860337117304, "grad_norm": 17.375, "learning_rate": 1.60260576068547e-06, "loss": 1.931, "step": 6504 }, { "epoch": 0.745900699461071, "grad_norm": 15.0625, "learning_rate": 1.6012435578955554e-06, "loss": 1.7057, "step": 6505 }, { "epoch": 0.7460153652104117, "grad_norm": 17.75, "learning_rate": 1.5998818238993048e-06, "loss": 1.526, "step": 6506 }, { "epoch": 0.7461300309597523, "grad_norm": 11.4375, "learning_rate": 1.5985205588845437e-06, "loss": 1.4665, "step": 6507 }, { "epoch": 0.746244696709093, "grad_norm": 12.4375, "learning_rate": 1.597159763039034e-06, "loss": 1.8487, "step": 6508 }, { "epoch": 0.7463593624584337, "grad_norm": 12.9375, "learning_rate": 1.595799436550471e-06, "loss": 1.5561, "step": 6509 }, { "epoch": 0.7464740282077743, "grad_norm": 13.5625, "learning_rate": 1.594439579606485e-06, "loss": 1.6816, "step": 6510 }, { "epoch": 0.746588693957115, "grad_norm": 15.0, "learning_rate": 1.5930801923946453e-06, "loss": 1.4788, "step": 6511 }, { "epoch": 0.7467033597064556, "grad_norm": 18.0, "learning_rate": 1.5917212751024496e-06, "loss": 1.9394, "step": 6512 }, { "epoch": 0.7468180254557963, "grad_norm": 15.4375, "learning_rate": 1.590362827917336e-06, "loss": 1.6083, "step": 6513 }, { "epoch": 0.7469326912051371, "grad_norm": 15.5, "learning_rate": 1.589004851026677e-06, "loss": 1.9362, "step": 6514 }, { "epoch": 0.7470473569544777, "grad_norm": 19.5, "learning_rate": 1.5876473446177793e-06, "loss": 1.5935, "step": 6515 }, { "epoch": 0.7471620227038184, "grad_norm": 12.0625, "learning_rate": 1.5862903088778852e-06, "loss": 1.6369, "step": 6516 }, { "epoch": 0.7472766884531591, "grad_norm": 14.25, "learning_rate": 1.5849337439941731e-06, "loss": 1.4521, "step": 6517 }, { "epoch": 0.7473913542024997, "grad_norm": 18.25, "learning_rate": 1.5835776501537526e-06, "loss": 1.4328, "step": 6518 }, { "epoch": 0.7475060199518404, "grad_norm": 14.1875, "learning_rate": 1.58222202754367e-06, "loss": 1.4578, "step": 6519 }, { "epoch": 0.747620685701181, "grad_norm": 13.8125, "learning_rate": 1.5808668763509138e-06, "loss": 1.7768, "step": 6520 }, { "epoch": 0.7477353514505217, "grad_norm": 17.125, "learning_rate": 1.5795121967623956e-06, "loss": 2.3013, "step": 6521 }, { "epoch": 0.7478500171998624, "grad_norm": 14.5625, "learning_rate": 1.5781579889649694e-06, "loss": 2.0045, "step": 6522 }, { "epoch": 0.747964682949203, "grad_norm": 14.25, "learning_rate": 1.576804253145424e-06, "loss": 1.6384, "step": 6523 }, { "epoch": 0.7480793486985438, "grad_norm": 16.25, "learning_rate": 1.5754509894904767e-06, "loss": 1.5441, "step": 6524 }, { "epoch": 0.7481940144478845, "grad_norm": 15.4375, "learning_rate": 1.5740981981867893e-06, "loss": 1.7318, "step": 6525 }, { "epoch": 0.7483086801972251, "grad_norm": 14.875, "learning_rate": 1.5727458794209532e-06, "loss": 2.0563, "step": 6526 }, { "epoch": 0.7484233459465658, "grad_norm": 24.0, "learning_rate": 1.5713940333794925e-06, "loss": 1.5742, "step": 6527 }, { "epoch": 0.7485380116959064, "grad_norm": 14.625, "learning_rate": 1.5700426602488695e-06, "loss": 1.6704, "step": 6528 }, { "epoch": 0.7486526774452471, "grad_norm": 16.125, "learning_rate": 1.56869176021548e-06, "loss": 1.7225, "step": 6529 }, { "epoch": 0.7487673431945878, "grad_norm": 32.5, "learning_rate": 1.5673413334656563e-06, "loss": 1.8726, "step": 6530 }, { "epoch": 0.7488820089439284, "grad_norm": 12.3125, "learning_rate": 1.5659913801856624e-06, "loss": 1.4623, "step": 6531 }, { "epoch": 0.7489966746932691, "grad_norm": 15.125, "learning_rate": 1.5646419005617014e-06, "loss": 2.1357, "step": 6532 }, { "epoch": 0.7491113404426097, "grad_norm": 12.375, "learning_rate": 1.5632928947799047e-06, "loss": 1.9089, "step": 6533 }, { "epoch": 0.7492260061919505, "grad_norm": 16.125, "learning_rate": 1.5619443630263426e-06, "loss": 1.6992, "step": 6534 }, { "epoch": 0.7493406719412912, "grad_norm": 15.875, "learning_rate": 1.5605963054870227e-06, "loss": 1.62, "step": 6535 }, { "epoch": 0.7494553376906318, "grad_norm": 17.75, "learning_rate": 1.5592487223478802e-06, "loss": 2.0462, "step": 6536 }, { "epoch": 0.7495700034399725, "grad_norm": 15.25, "learning_rate": 1.55790161379479e-06, "loss": 1.7956, "step": 6537 }, { "epoch": 0.7496846691893132, "grad_norm": 16.625, "learning_rate": 1.556554980013561e-06, "loss": 1.4476, "step": 6538 }, { "epoch": 0.7497993349386538, "grad_norm": 14.125, "learning_rate": 1.5552088211899313e-06, "loss": 1.8845, "step": 6539 }, { "epoch": 0.7499140006879945, "grad_norm": 12.625, "learning_rate": 1.5538631375095825e-06, "loss": 1.8576, "step": 6540 }, { "epoch": 0.7500286664373351, "grad_norm": 15.75, "learning_rate": 1.5525179291581265e-06, "loss": 1.5355, "step": 6541 }, { "epoch": 0.7501433321866758, "grad_norm": 13.875, "learning_rate": 1.5511731963211058e-06, "loss": 1.598, "step": 6542 }, { "epoch": 0.7502579979360166, "grad_norm": 12.25, "learning_rate": 1.5498289391840016e-06, "loss": 1.407, "step": 6543 }, { "epoch": 0.7503726636853572, "grad_norm": 26.875, "learning_rate": 1.5484851579322291e-06, "loss": 1.8714, "step": 6544 }, { "epoch": 0.7504873294346979, "grad_norm": 14.9375, "learning_rate": 1.5471418527511367e-06, "loss": 1.6483, "step": 6545 }, { "epoch": 0.7506019951840385, "grad_norm": 12.75, "learning_rate": 1.545799023826009e-06, "loss": 1.562, "step": 6546 }, { "epoch": 0.7507166609333792, "grad_norm": 15.625, "learning_rate": 1.5444566713420644e-06, "loss": 1.7754, "step": 6547 }, { "epoch": 0.7508313266827199, "grad_norm": 14.375, "learning_rate": 1.5431147954844517e-06, "loss": 1.7266, "step": 6548 }, { "epoch": 0.7509459924320605, "grad_norm": 13.125, "learning_rate": 1.5417733964382587e-06, "loss": 1.6139, "step": 6549 }, { "epoch": 0.7510606581814012, "grad_norm": 15.5625, "learning_rate": 1.5404324743885058e-06, "loss": 1.8464, "step": 6550 }, { "epoch": 0.7511753239307419, "grad_norm": 13.375, "learning_rate": 1.5390920295201473e-06, "loss": 1.7156, "step": 6551 }, { "epoch": 0.7512899896800825, "grad_norm": 14.0, "learning_rate": 1.537752062018073e-06, "loss": 2.032, "step": 6552 }, { "epoch": 0.7514046554294233, "grad_norm": 13.25, "learning_rate": 1.5364125720671063e-06, "loss": 1.6944, "step": 6553 }, { "epoch": 0.7515193211787639, "grad_norm": 14.125, "learning_rate": 1.5350735598520006e-06, "loss": 1.8676, "step": 6554 }, { "epoch": 0.7516339869281046, "grad_norm": 15.875, "learning_rate": 1.5337350255574528e-06, "loss": 1.694, "step": 6555 }, { "epoch": 0.7517486526774453, "grad_norm": 14.125, "learning_rate": 1.532396969368084e-06, "loss": 1.6793, "step": 6556 }, { "epoch": 0.7518633184267859, "grad_norm": 15.3125, "learning_rate": 1.5310593914684546e-06, "loss": 1.6469, "step": 6557 }, { "epoch": 0.7519779841761266, "grad_norm": 21.375, "learning_rate": 1.5297222920430576e-06, "loss": 1.6113, "step": 6558 }, { "epoch": 0.7520926499254673, "grad_norm": 13.5625, "learning_rate": 1.5283856712763217e-06, "loss": 1.5113, "step": 6559 }, { "epoch": 0.7522073156748079, "grad_norm": 15.75, "learning_rate": 1.5270495293526073e-06, "loss": 1.83, "step": 6560 }, { "epoch": 0.7523219814241486, "grad_norm": 14.1875, "learning_rate": 1.525713866456211e-06, "loss": 1.641, "step": 6561 }, { "epoch": 0.7524366471734892, "grad_norm": 14.125, "learning_rate": 1.5243786827713592e-06, "loss": 1.8962, "step": 6562 }, { "epoch": 0.75255131292283, "grad_norm": 14.5, "learning_rate": 1.523043978482217e-06, "loss": 1.8589, "step": 6563 }, { "epoch": 0.7526659786721707, "grad_norm": 14.0, "learning_rate": 1.5217097537728814e-06, "loss": 1.8593, "step": 6564 }, { "epoch": 0.7527806444215113, "grad_norm": 23.75, "learning_rate": 1.5203760088273827e-06, "loss": 1.6979, "step": 6565 }, { "epoch": 0.752895310170852, "grad_norm": 18.375, "learning_rate": 1.5190427438296863e-06, "loss": 1.6258, "step": 6566 }, { "epoch": 0.7530099759201926, "grad_norm": 12.9375, "learning_rate": 1.5177099589636918e-06, "loss": 1.5072, "step": 6567 }, { "epoch": 0.7531246416695333, "grad_norm": 13.625, "learning_rate": 1.516377654413228e-06, "loss": 1.7084, "step": 6568 }, { "epoch": 0.753239307418874, "grad_norm": 14.625, "learning_rate": 1.5150458303620618e-06, "loss": 1.7554, "step": 6569 }, { "epoch": 0.7533539731682146, "grad_norm": 15.3125, "learning_rate": 1.5137144869938969e-06, "loss": 1.6618, "step": 6570 }, { "epoch": 0.7534686389175553, "grad_norm": 15.375, "learning_rate": 1.5123836244923623e-06, "loss": 1.4051, "step": 6571 }, { "epoch": 0.753583304666896, "grad_norm": 14.875, "learning_rate": 1.5110532430410268e-06, "loss": 1.7846, "step": 6572 }, { "epoch": 0.7536979704162367, "grad_norm": 15.6875, "learning_rate": 1.5097233428233926e-06, "loss": 1.5331, "step": 6573 }, { "epoch": 0.7538126361655774, "grad_norm": 13.3125, "learning_rate": 1.508393924022889e-06, "loss": 1.5446, "step": 6574 }, { "epoch": 0.753927301914918, "grad_norm": 12.8125, "learning_rate": 1.5070649868228897e-06, "loss": 1.4412, "step": 6575 }, { "epoch": 0.7540419676642587, "grad_norm": 12.875, "learning_rate": 1.5057365314066951e-06, "loss": 1.9117, "step": 6576 }, { "epoch": 0.7541566334135994, "grad_norm": 31.875, "learning_rate": 1.5044085579575379e-06, "loss": 1.8795, "step": 6577 }, { "epoch": 0.75427129916294, "grad_norm": 34.75, "learning_rate": 1.5030810666585882e-06, "loss": 1.6865, "step": 6578 }, { "epoch": 0.7543859649122807, "grad_norm": 14.25, "learning_rate": 1.501754057692948e-06, "loss": 1.2858, "step": 6579 }, { "epoch": 0.7545006306616213, "grad_norm": 16.75, "learning_rate": 1.5004275312436527e-06, "loss": 1.561, "step": 6580 }, { "epoch": 0.754615296410962, "grad_norm": 12.6875, "learning_rate": 1.4991014874936722e-06, "loss": 1.9776, "step": 6581 }, { "epoch": 0.7547299621603027, "grad_norm": 16.0, "learning_rate": 1.4977759266259095e-06, "loss": 1.9717, "step": 6582 }, { "epoch": 0.7548446279096434, "grad_norm": 14.5, "learning_rate": 1.4964508488231972e-06, "loss": 1.523, "step": 6583 }, { "epoch": 0.7549592936589841, "grad_norm": 13.8125, "learning_rate": 1.4951262542683053e-06, "loss": 1.696, "step": 6584 }, { "epoch": 0.7550739594083248, "grad_norm": 14.5, "learning_rate": 1.4938021431439408e-06, "loss": 1.8028, "step": 6585 }, { "epoch": 0.7551886251576654, "grad_norm": 19.125, "learning_rate": 1.4924785156327353e-06, "loss": 2.1491, "step": 6586 }, { "epoch": 0.7553032909070061, "grad_norm": 12.9375, "learning_rate": 1.491155371917259e-06, "loss": 1.4281, "step": 6587 }, { "epoch": 0.7554179566563467, "grad_norm": 14.25, "learning_rate": 1.4898327121800155e-06, "loss": 1.972, "step": 6588 }, { "epoch": 0.7555326224056874, "grad_norm": 15.5625, "learning_rate": 1.488510536603437e-06, "loss": 2.0538, "step": 6589 }, { "epoch": 0.7556472881550281, "grad_norm": 12.25, "learning_rate": 1.4871888453698957e-06, "loss": 1.6081, "step": 6590 }, { "epoch": 0.7557619539043687, "grad_norm": 13.875, "learning_rate": 1.485867638661695e-06, "loss": 1.5239, "step": 6591 }, { "epoch": 0.7558766196537094, "grad_norm": 15.1875, "learning_rate": 1.4845469166610665e-06, "loss": 2.1662, "step": 6592 }, { "epoch": 0.7559912854030502, "grad_norm": 12.875, "learning_rate": 1.4832266795501805e-06, "loss": 1.7736, "step": 6593 }, { "epoch": 0.7561059511523908, "grad_norm": 14.75, "learning_rate": 1.4819069275111376e-06, "loss": 1.6287, "step": 6594 }, { "epoch": 0.7562206169017315, "grad_norm": 16.375, "learning_rate": 1.4805876607259733e-06, "loss": 2.248, "step": 6595 }, { "epoch": 0.7563352826510721, "grad_norm": 15.125, "learning_rate": 1.4792688793766553e-06, "loss": 1.8402, "step": 6596 }, { "epoch": 0.7564499484004128, "grad_norm": 13.3125, "learning_rate": 1.4779505836450858e-06, "loss": 1.5767, "step": 6597 }, { "epoch": 0.7565646141497535, "grad_norm": 14.75, "learning_rate": 1.4766327737130954e-06, "loss": 1.8119, "step": 6598 }, { "epoch": 0.7566792798990941, "grad_norm": 14.0625, "learning_rate": 1.4753154497624517e-06, "loss": 1.7873, "step": 6599 }, { "epoch": 0.7567939456484348, "grad_norm": 12.3125, "learning_rate": 1.4739986119748557e-06, "loss": 1.673, "step": 6600 }, { "epoch": 0.7569086113977754, "grad_norm": 14.4375, "learning_rate": 1.4726822605319396e-06, "loss": 2.104, "step": 6601 }, { "epoch": 0.7570232771471161, "grad_norm": 13.25, "learning_rate": 1.4713663956152686e-06, "loss": 1.4326, "step": 6602 }, { "epoch": 0.7571379428964569, "grad_norm": 12.1875, "learning_rate": 1.4700510174063427e-06, "loss": 1.5033, "step": 6603 }, { "epoch": 0.7572526086457975, "grad_norm": 15.125, "learning_rate": 1.4687361260865896e-06, "loss": 1.8787, "step": 6604 }, { "epoch": 0.7573672743951382, "grad_norm": 15.4375, "learning_rate": 1.467421721837378e-06, "loss": 1.8992, "step": 6605 }, { "epoch": 0.7574819401444789, "grad_norm": 12.375, "learning_rate": 1.4661078048400023e-06, "loss": 1.5045, "step": 6606 }, { "epoch": 0.7575966058938195, "grad_norm": 14.625, "learning_rate": 1.464794375275692e-06, "loss": 1.9496, "step": 6607 }, { "epoch": 0.7577112716431602, "grad_norm": 17.375, "learning_rate": 1.4634814333256108e-06, "loss": 1.9041, "step": 6608 }, { "epoch": 0.7578259373925008, "grad_norm": 12.9375, "learning_rate": 1.4621689791708537e-06, "loss": 1.7025, "step": 6609 }, { "epoch": 0.7579406031418415, "grad_norm": 13.6875, "learning_rate": 1.4608570129924492e-06, "loss": 1.7341, "step": 6610 }, { "epoch": 0.7580552688911822, "grad_norm": 15.375, "learning_rate": 1.4595455349713583e-06, "loss": 2.1047, "step": 6611 }, { "epoch": 0.7581699346405228, "grad_norm": 12.1875, "learning_rate": 1.4582345452884727e-06, "loss": 1.6134, "step": 6612 }, { "epoch": 0.7582846003898636, "grad_norm": 14.5625, "learning_rate": 1.456924044124619e-06, "loss": 1.9392, "step": 6613 }, { "epoch": 0.7583992661392042, "grad_norm": 15.0, "learning_rate": 1.4556140316605566e-06, "loss": 1.7053, "step": 6614 }, { "epoch": 0.7585139318885449, "grad_norm": 18.625, "learning_rate": 1.4543045080769763e-06, "loss": 1.8576, "step": 6615 }, { "epoch": 0.7586285976378856, "grad_norm": 14.0625, "learning_rate": 1.4529954735545015e-06, "loss": 1.7047, "step": 6616 }, { "epoch": 0.7587432633872262, "grad_norm": 14.125, "learning_rate": 1.4516869282736912e-06, "loss": 1.5569, "step": 6617 }, { "epoch": 0.7588579291365669, "grad_norm": 15.6875, "learning_rate": 1.4503788724150302e-06, "loss": 1.8255, "step": 6618 }, { "epoch": 0.7589725948859076, "grad_norm": 15.0625, "learning_rate": 1.4490713061589406e-06, "loss": 1.796, "step": 6619 }, { "epoch": 0.7590872606352482, "grad_norm": 15.25, "learning_rate": 1.4477642296857808e-06, "loss": 2.077, "step": 6620 }, { "epoch": 0.7592019263845889, "grad_norm": 14.125, "learning_rate": 1.4464576431758316e-06, "loss": 1.9366, "step": 6621 }, { "epoch": 0.7593165921339295, "grad_norm": 17.375, "learning_rate": 1.445151546809314e-06, "loss": 1.7951, "step": 6622 }, { "epoch": 0.7594312578832703, "grad_norm": 15.8125, "learning_rate": 1.4438459407663808e-06, "loss": 1.8101, "step": 6623 }, { "epoch": 0.759545923632611, "grad_norm": 14.375, "learning_rate": 1.44254082522711e-06, "loss": 1.5832, "step": 6624 }, { "epoch": 0.7596605893819516, "grad_norm": 11.9375, "learning_rate": 1.441236200371523e-06, "loss": 1.6295, "step": 6625 }, { "epoch": 0.7597752551312923, "grad_norm": 11.75, "learning_rate": 1.4399320663795674e-06, "loss": 1.6717, "step": 6626 }, { "epoch": 0.759889920880633, "grad_norm": 11.75, "learning_rate": 1.4386284234311205e-06, "loss": 1.6147, "step": 6627 }, { "epoch": 0.7600045866299736, "grad_norm": 15.375, "learning_rate": 1.4373252717059965e-06, "loss": 1.7706, "step": 6628 }, { "epoch": 0.7601192523793143, "grad_norm": 13.1875, "learning_rate": 1.436022611383941e-06, "loss": 1.7719, "step": 6629 }, { "epoch": 0.7602339181286549, "grad_norm": 15.8125, "learning_rate": 1.4347204426446304e-06, "loss": 1.8477, "step": 6630 }, { "epoch": 0.7603485838779956, "grad_norm": 14.875, "learning_rate": 1.433418765667674e-06, "loss": 1.8202, "step": 6631 }, { "epoch": 0.7604632496273364, "grad_norm": 18.75, "learning_rate": 1.432117580632616e-06, "loss": 2.2874, "step": 6632 }, { "epoch": 0.760577915376677, "grad_norm": 11.8125, "learning_rate": 1.4308168877189254e-06, "loss": 1.5877, "step": 6633 }, { "epoch": 0.7606925811260177, "grad_norm": 17.125, "learning_rate": 1.429516687106009e-06, "loss": 1.8049, "step": 6634 }, { "epoch": 0.7608072468753583, "grad_norm": 13.8125, "learning_rate": 1.4282169789732092e-06, "loss": 1.7741, "step": 6635 }, { "epoch": 0.760921912624699, "grad_norm": 14.5625, "learning_rate": 1.4269177634997916e-06, "loss": 1.6485, "step": 6636 }, { "epoch": 0.7610365783740397, "grad_norm": 15.4375, "learning_rate": 1.425619040864959e-06, "loss": 1.8378, "step": 6637 }, { "epoch": 0.7611512441233803, "grad_norm": 18.0, "learning_rate": 1.4243208112478474e-06, "loss": 1.9558, "step": 6638 }, { "epoch": 0.761265909872721, "grad_norm": 13.8125, "learning_rate": 1.4230230748275182e-06, "loss": 1.6456, "step": 6639 }, { "epoch": 0.7613805756220617, "grad_norm": 28.75, "learning_rate": 1.4217258317829736e-06, "loss": 2.0354, "step": 6640 }, { "epoch": 0.7614952413714023, "grad_norm": 15.625, "learning_rate": 1.4204290822931444e-06, "loss": 1.7496, "step": 6641 }, { "epoch": 0.7616099071207431, "grad_norm": 13.125, "learning_rate": 1.4191328265368886e-06, "loss": 1.6737, "step": 6642 }, { "epoch": 0.7617245728700837, "grad_norm": 16.125, "learning_rate": 1.4178370646930017e-06, "loss": 2.2352, "step": 6643 }, { "epoch": 0.7618392386194244, "grad_norm": 15.125, "learning_rate": 1.41654179694021e-06, "loss": 2.0085, "step": 6644 }, { "epoch": 0.7619539043687651, "grad_norm": 12.3125, "learning_rate": 1.4152470234571702e-06, "loss": 1.582, "step": 6645 }, { "epoch": 0.7620685701181057, "grad_norm": 15.125, "learning_rate": 1.4139527444224721e-06, "loss": 1.5752, "step": 6646 }, { "epoch": 0.7621832358674464, "grad_norm": 28.25, "learning_rate": 1.4126589600146378e-06, "loss": 1.9027, "step": 6647 }, { "epoch": 0.762297901616787, "grad_norm": 15.5625, "learning_rate": 1.411365670412118e-06, "loss": 2.0346, "step": 6648 }, { "epoch": 0.7624125673661277, "grad_norm": 14.4375, "learning_rate": 1.410072875793298e-06, "loss": 1.635, "step": 6649 }, { "epoch": 0.7625272331154684, "grad_norm": 13.1875, "learning_rate": 1.4087805763364949e-06, "loss": 1.9118, "step": 6650 }, { "epoch": 0.762641898864809, "grad_norm": 21.875, "learning_rate": 1.407488772219956e-06, "loss": 1.8883, "step": 6651 }, { "epoch": 0.7627565646141498, "grad_norm": 34.75, "learning_rate": 1.4061974636218617e-06, "loss": 2.2252, "step": 6652 }, { "epoch": 0.7628712303634905, "grad_norm": 13.625, "learning_rate": 1.4049066507203245e-06, "loss": 1.9778, "step": 6653 }, { "epoch": 0.7629858961128311, "grad_norm": 16.625, "learning_rate": 1.4036163336933833e-06, "loss": 1.8662, "step": 6654 }, { "epoch": 0.7631005618621718, "grad_norm": 17.5, "learning_rate": 1.4023265127190189e-06, "loss": 1.469, "step": 6655 }, { "epoch": 0.7632152276115124, "grad_norm": 12.625, "learning_rate": 1.4010371879751322e-06, "loss": 1.669, "step": 6656 }, { "epoch": 0.7633298933608531, "grad_norm": 12.5625, "learning_rate": 1.3997483596395627e-06, "loss": 1.6162, "step": 6657 }, { "epoch": 0.7634445591101938, "grad_norm": 16.375, "learning_rate": 1.3984600278900806e-06, "loss": 1.5116, "step": 6658 }, { "epoch": 0.7635592248595344, "grad_norm": 13.9375, "learning_rate": 1.3971721929043858e-06, "loss": 1.5086, "step": 6659 }, { "epoch": 0.7636738906088751, "grad_norm": 15.4375, "learning_rate": 1.3958848548601107e-06, "loss": 1.959, "step": 6660 }, { "epoch": 0.7637885563582159, "grad_norm": 14.25, "learning_rate": 1.3945980139348208e-06, "loss": 1.4938, "step": 6661 }, { "epoch": 0.7639032221075565, "grad_norm": 13.3125, "learning_rate": 1.3933116703060079e-06, "loss": 1.5771, "step": 6662 }, { "epoch": 0.7640178878568972, "grad_norm": 17.0, "learning_rate": 1.3920258241511008e-06, "loss": 1.842, "step": 6663 }, { "epoch": 0.7641325536062378, "grad_norm": 12.25, "learning_rate": 1.3907404756474568e-06, "loss": 1.5934, "step": 6664 }, { "epoch": 0.7642472193555785, "grad_norm": 14.25, "learning_rate": 1.3894556249723656e-06, "loss": 1.7529, "step": 6665 }, { "epoch": 0.7643618851049192, "grad_norm": 17.375, "learning_rate": 1.388171272303047e-06, "loss": 1.7028, "step": 6666 }, { "epoch": 0.7644765508542598, "grad_norm": 11.125, "learning_rate": 1.3868874178166564e-06, "loss": 1.5631, "step": 6667 }, { "epoch": 0.7645912166036005, "grad_norm": 14.6875, "learning_rate": 1.3856040616902716e-06, "loss": 1.4552, "step": 6668 }, { "epoch": 0.7647058823529411, "grad_norm": 16.0, "learning_rate": 1.384321204100909e-06, "loss": 2.065, "step": 6669 }, { "epoch": 0.7648205481022818, "grad_norm": 17.625, "learning_rate": 1.3830388452255184e-06, "loss": 1.7439, "step": 6670 }, { "epoch": 0.7649352138516226, "grad_norm": 17.125, "learning_rate": 1.3817569852409713e-06, "loss": 1.9193, "step": 6671 }, { "epoch": 0.7650498796009632, "grad_norm": 12.25, "learning_rate": 1.3804756243240785e-06, "loss": 1.9004, "step": 6672 }, { "epoch": 0.7651645453503039, "grad_norm": 11.625, "learning_rate": 1.3791947626515806e-06, "loss": 1.5701, "step": 6673 }, { "epoch": 0.7652792110996446, "grad_norm": 15.5625, "learning_rate": 1.3779144004001431e-06, "loss": 1.9923, "step": 6674 }, { "epoch": 0.7653938768489852, "grad_norm": 15.0, "learning_rate": 1.3766345377463725e-06, "loss": 1.8367, "step": 6675 }, { "epoch": 0.7655085425983259, "grad_norm": 15.25, "learning_rate": 1.375355174866801e-06, "loss": 1.9265, "step": 6676 }, { "epoch": 0.7656232083476665, "grad_norm": 15.5, "learning_rate": 1.3740763119378898e-06, "loss": 1.9378, "step": 6677 }, { "epoch": 0.7657378740970072, "grad_norm": 28.625, "learning_rate": 1.3727979491360349e-06, "loss": 1.8468, "step": 6678 }, { "epoch": 0.7658525398463479, "grad_norm": 15.5, "learning_rate": 1.3715200866375627e-06, "loss": 2.3319, "step": 6679 }, { "epoch": 0.7659672055956885, "grad_norm": 18.75, "learning_rate": 1.3702427246187288e-06, "loss": 1.6125, "step": 6680 }, { "epoch": 0.7660818713450293, "grad_norm": 16.125, "learning_rate": 1.3689658632557219e-06, "loss": 2.0455, "step": 6681 }, { "epoch": 0.76619653709437, "grad_norm": 15.6875, "learning_rate": 1.3676895027246618e-06, "loss": 1.644, "step": 6682 }, { "epoch": 0.7663112028437106, "grad_norm": 14.75, "learning_rate": 1.3664136432015945e-06, "loss": 1.7527, "step": 6683 }, { "epoch": 0.7664258685930513, "grad_norm": 14.9375, "learning_rate": 1.3651382848625021e-06, "loss": 2.0838, "step": 6684 }, { "epoch": 0.7665405343423919, "grad_norm": 15.1875, "learning_rate": 1.363863427883299e-06, "loss": 1.7242, "step": 6685 }, { "epoch": 0.7666552000917326, "grad_norm": 13.6875, "learning_rate": 1.3625890724398238e-06, "loss": 1.7325, "step": 6686 }, { "epoch": 0.7667698658410733, "grad_norm": 16.25, "learning_rate": 1.3613152187078505e-06, "loss": 1.8598, "step": 6687 }, { "epoch": 0.7668845315904139, "grad_norm": 15.4375, "learning_rate": 1.3600418668630844e-06, "loss": 2.0573, "step": 6688 }, { "epoch": 0.7669991973397546, "grad_norm": 14.8125, "learning_rate": 1.3587690170811563e-06, "loss": 1.4829, "step": 6689 }, { "epoch": 0.7671138630890952, "grad_norm": 12.5, "learning_rate": 1.3574966695376351e-06, "loss": 1.4667, "step": 6690 }, { "epoch": 0.767228528838436, "grad_norm": 12.3125, "learning_rate": 1.3562248244080178e-06, "loss": 1.4166, "step": 6691 }, { "epoch": 0.7673431945877767, "grad_norm": 17.5, "learning_rate": 1.354953481867728e-06, "loss": 2.2114, "step": 6692 }, { "epoch": 0.7674578603371173, "grad_norm": 15.125, "learning_rate": 1.353682642092124e-06, "loss": 1.6525, "step": 6693 }, { "epoch": 0.767572526086458, "grad_norm": 17.875, "learning_rate": 1.3524123052564947e-06, "loss": 2.0202, "step": 6694 }, { "epoch": 0.7676871918357987, "grad_norm": 17.0, "learning_rate": 1.3511424715360582e-06, "loss": 1.6905, "step": 6695 }, { "epoch": 0.7678018575851393, "grad_norm": 16.625, "learning_rate": 1.3498731411059645e-06, "loss": 1.8156, "step": 6696 }, { "epoch": 0.76791652333448, "grad_norm": 14.6875, "learning_rate": 1.3486043141412947e-06, "loss": 1.8322, "step": 6697 }, { "epoch": 0.7680311890838206, "grad_norm": 13.125, "learning_rate": 1.3473359908170564e-06, "loss": 1.6235, "step": 6698 }, { "epoch": 0.7681458548331613, "grad_norm": 33.5, "learning_rate": 1.3460681713081918e-06, "loss": 1.7117, "step": 6699 }, { "epoch": 0.768260520582502, "grad_norm": 13.8125, "learning_rate": 1.344800855789573e-06, "loss": 1.9471, "step": 6700 }, { "epoch": 0.7683751863318427, "grad_norm": 41.25, "learning_rate": 1.3435340444360017e-06, "loss": 1.9141, "step": 6701 }, { "epoch": 0.7684898520811834, "grad_norm": 13.6875, "learning_rate": 1.34226773742221e-06, "loss": 1.8242, "step": 6702 }, { "epoch": 0.768604517830524, "grad_norm": 14.875, "learning_rate": 1.3410019349228631e-06, "loss": 1.4887, "step": 6703 }, { "epoch": 0.7687191835798647, "grad_norm": 16.625, "learning_rate": 1.3397366371125492e-06, "loss": 1.9827, "step": 6704 }, { "epoch": 0.7688338493292054, "grad_norm": 19.875, "learning_rate": 1.3384718441657983e-06, "loss": 2.2614, "step": 6705 }, { "epoch": 0.768948515078546, "grad_norm": 13.25, "learning_rate": 1.3372075562570603e-06, "loss": 1.9581, "step": 6706 }, { "epoch": 0.7690631808278867, "grad_norm": 13.625, "learning_rate": 1.33594377356072e-06, "loss": 1.734, "step": 6707 }, { "epoch": 0.7691778465772274, "grad_norm": 15.875, "learning_rate": 1.3346804962510935e-06, "loss": 2.116, "step": 6708 }, { "epoch": 0.769292512326568, "grad_norm": 15.0, "learning_rate": 1.333417724502425e-06, "loss": 2.2344, "step": 6709 }, { "epoch": 0.7694071780759087, "grad_norm": 13.25, "learning_rate": 1.3321554584888896e-06, "loss": 1.827, "step": 6710 }, { "epoch": 0.7695218438252494, "grad_norm": 14.375, "learning_rate": 1.330893698384595e-06, "loss": 1.4736, "step": 6711 }, { "epoch": 0.7696365095745901, "grad_norm": 21.0, "learning_rate": 1.329632444363574e-06, "loss": 1.8542, "step": 6712 }, { "epoch": 0.7697511753239308, "grad_norm": 15.1875, "learning_rate": 1.3283716965997934e-06, "loss": 1.7733, "step": 6713 }, { "epoch": 0.7698658410732714, "grad_norm": 15.875, "learning_rate": 1.3271114552671494e-06, "loss": 1.4282, "step": 6714 }, { "epoch": 0.7699805068226121, "grad_norm": 14.5625, "learning_rate": 1.3258517205394688e-06, "loss": 1.7463, "step": 6715 }, { "epoch": 0.7700951725719528, "grad_norm": 17.875, "learning_rate": 1.324592492590508e-06, "loss": 1.8822, "step": 6716 }, { "epoch": 0.7702098383212934, "grad_norm": 13.5625, "learning_rate": 1.3233337715939542e-06, "loss": 1.5632, "step": 6717 }, { "epoch": 0.7703245040706341, "grad_norm": 14.25, "learning_rate": 1.3220755577234218e-06, "loss": 1.4877, "step": 6718 }, { "epoch": 0.7704391698199747, "grad_norm": 17.0, "learning_rate": 1.3208178511524572e-06, "loss": 1.6318, "step": 6719 }, { "epoch": 0.7705538355693154, "grad_norm": 15.125, "learning_rate": 1.3195606520545411e-06, "loss": 1.8218, "step": 6720 }, { "epoch": 0.7706685013186562, "grad_norm": 11.3125, "learning_rate": 1.3183039606030761e-06, "loss": 1.6717, "step": 6721 }, { "epoch": 0.7707831670679968, "grad_norm": 34.5, "learning_rate": 1.3170477769714008e-06, "loss": 1.8149, "step": 6722 }, { "epoch": 0.7708978328173375, "grad_norm": 12.5625, "learning_rate": 1.3157921013327819e-06, "loss": 1.6104, "step": 6723 }, { "epoch": 0.7710124985666781, "grad_norm": 29.25, "learning_rate": 1.3145369338604126e-06, "loss": 1.9924, "step": 6724 }, { "epoch": 0.7711271643160188, "grad_norm": 14.8125, "learning_rate": 1.3132822747274233e-06, "loss": 1.8251, "step": 6725 }, { "epoch": 0.7712418300653595, "grad_norm": 25.5, "learning_rate": 1.3120281241068706e-06, "loss": 1.5021, "step": 6726 }, { "epoch": 0.7713564958147001, "grad_norm": 11.375, "learning_rate": 1.3107744821717378e-06, "loss": 1.3114, "step": 6727 }, { "epoch": 0.7714711615640408, "grad_norm": 17.125, "learning_rate": 1.3095213490949416e-06, "loss": 1.8133, "step": 6728 }, { "epoch": 0.7715858273133815, "grad_norm": 22.75, "learning_rate": 1.3082687250493286e-06, "loss": 2.0771, "step": 6729 }, { "epoch": 0.7717004930627221, "grad_norm": 14.625, "learning_rate": 1.3070166102076743e-06, "loss": 2.1549, "step": 6730 }, { "epoch": 0.7718151588120629, "grad_norm": 12.125, "learning_rate": 1.3057650047426833e-06, "loss": 1.7139, "step": 6731 }, { "epoch": 0.7719298245614035, "grad_norm": 14.125, "learning_rate": 1.3045139088269936e-06, "loss": 1.7937, "step": 6732 }, { "epoch": 0.7720444903107442, "grad_norm": 14.75, "learning_rate": 1.3032633226331665e-06, "loss": 2.0309, "step": 6733 }, { "epoch": 0.7721591560600849, "grad_norm": 14.8125, "learning_rate": 1.302013246333696e-06, "loss": 1.9372, "step": 6734 }, { "epoch": 0.7722738218094255, "grad_norm": 14.5, "learning_rate": 1.3007636801010109e-06, "loss": 1.85, "step": 6735 }, { "epoch": 0.7723884875587662, "grad_norm": 33.75, "learning_rate": 1.2995146241074609e-06, "loss": 2.0296, "step": 6736 }, { "epoch": 0.7725031533081068, "grad_norm": 14.8125, "learning_rate": 1.298266078525331e-06, "loss": 2.0941, "step": 6737 }, { "epoch": 0.7726178190574475, "grad_norm": 15.375, "learning_rate": 1.2970180435268354e-06, "loss": 1.9829, "step": 6738 }, { "epoch": 0.7727324848067882, "grad_norm": 14.125, "learning_rate": 1.2957705192841125e-06, "loss": 1.5292, "step": 6739 }, { "epoch": 0.7728471505561288, "grad_norm": 13.875, "learning_rate": 1.2945235059692391e-06, "loss": 1.7297, "step": 6740 }, { "epoch": 0.7729618163054696, "grad_norm": 12.625, "learning_rate": 1.2932770037542164e-06, "loss": 1.7349, "step": 6741 }, { "epoch": 0.7730764820548103, "grad_norm": 15.3125, "learning_rate": 1.2920310128109732e-06, "loss": 1.8066, "step": 6742 }, { "epoch": 0.7731911478041509, "grad_norm": 12.375, "learning_rate": 1.2907855333113711e-06, "loss": 1.6567, "step": 6743 }, { "epoch": 0.7733058135534916, "grad_norm": 15.3125, "learning_rate": 1.2895405654272015e-06, "loss": 1.8188, "step": 6744 }, { "epoch": 0.7734204793028322, "grad_norm": 14.8125, "learning_rate": 1.2882961093301821e-06, "loss": 1.4786, "step": 6745 }, { "epoch": 0.7735351450521729, "grad_norm": 13.75, "learning_rate": 1.2870521651919637e-06, "loss": 1.7882, "step": 6746 }, { "epoch": 0.7736498108015136, "grad_norm": 14.4375, "learning_rate": 1.2858087331841245e-06, "loss": 1.9992, "step": 6747 }, { "epoch": 0.7737644765508542, "grad_norm": 34.5, "learning_rate": 1.2845658134781708e-06, "loss": 1.9136, "step": 6748 }, { "epoch": 0.7738791423001949, "grad_norm": 16.5, "learning_rate": 1.2833234062455407e-06, "loss": 2.0227, "step": 6749 }, { "epoch": 0.7739938080495357, "grad_norm": 19.5, "learning_rate": 1.2820815116576e-06, "loss": 1.8501, "step": 6750 }, { "epoch": 0.7741084737988763, "grad_norm": 15.25, "learning_rate": 1.2808401298856448e-06, "loss": 1.7419, "step": 6751 }, { "epoch": 0.774223139548217, "grad_norm": 14.5625, "learning_rate": 1.2795992611009e-06, "loss": 1.7695, "step": 6752 }, { "epoch": 0.7743378052975576, "grad_norm": 15.375, "learning_rate": 1.2783589054745216e-06, "loss": 1.3957, "step": 6753 }, { "epoch": 0.7744524710468983, "grad_norm": 16.25, "learning_rate": 1.2771190631775881e-06, "loss": 1.6265, "step": 6754 }, { "epoch": 0.774567136796239, "grad_norm": 15.875, "learning_rate": 1.2758797343811186e-06, "loss": 2.4583, "step": 6755 }, { "epoch": 0.7746818025455796, "grad_norm": 14.0625, "learning_rate": 1.2746409192560493e-06, "loss": 1.899, "step": 6756 }, { "epoch": 0.7747964682949203, "grad_norm": 14.375, "learning_rate": 1.273402617973254e-06, "loss": 1.7982, "step": 6757 }, { "epoch": 0.7749111340442609, "grad_norm": 14.4375, "learning_rate": 1.2721648307035322e-06, "loss": 1.5205, "step": 6758 }, { "epoch": 0.7750257997936016, "grad_norm": 11.5625, "learning_rate": 1.2709275576176127e-06, "loss": 1.481, "step": 6759 }, { "epoch": 0.7751404655429424, "grad_norm": 16.0, "learning_rate": 1.269690798886154e-06, "loss": 1.7423, "step": 6760 }, { "epoch": 0.775255131292283, "grad_norm": 14.8125, "learning_rate": 1.268454554679745e-06, "loss": 2.0089, "step": 6761 }, { "epoch": 0.7753697970416237, "grad_norm": 12.125, "learning_rate": 1.267218825168899e-06, "loss": 1.7416, "step": 6762 }, { "epoch": 0.7754844627909644, "grad_norm": 13.0625, "learning_rate": 1.265983610524063e-06, "loss": 1.9152, "step": 6763 }, { "epoch": 0.775599128540305, "grad_norm": 17.75, "learning_rate": 1.2647489109156108e-06, "loss": 2.2079, "step": 6764 }, { "epoch": 0.7757137942896457, "grad_norm": 14.5625, "learning_rate": 1.2635147265138459e-06, "loss": 1.8575, "step": 6765 }, { "epoch": 0.7758284600389863, "grad_norm": 18.125, "learning_rate": 1.262281057489001e-06, "loss": 1.5668, "step": 6766 }, { "epoch": 0.775943125788327, "grad_norm": 13.6875, "learning_rate": 1.2610479040112378e-06, "loss": 1.6737, "step": 6767 }, { "epoch": 0.7760577915376677, "grad_norm": 15.125, "learning_rate": 1.2598152662506441e-06, "loss": 1.6131, "step": 6768 }, { "epoch": 0.7761724572870083, "grad_norm": 11.8125, "learning_rate": 1.2585831443772384e-06, "loss": 1.4213, "step": 6769 }, { "epoch": 0.7762871230363491, "grad_norm": 13.5625, "learning_rate": 1.2573515385609736e-06, "loss": 1.502, "step": 6770 }, { "epoch": 0.7764017887856897, "grad_norm": 13.4375, "learning_rate": 1.2561204489717204e-06, "loss": 1.7634, "step": 6771 }, { "epoch": 0.7765164545350304, "grad_norm": 11.125, "learning_rate": 1.254889875779287e-06, "loss": 1.8185, "step": 6772 }, { "epoch": 0.7766311202843711, "grad_norm": 15.4375, "learning_rate": 1.2536598191534083e-06, "loss": 2.0105, "step": 6773 }, { "epoch": 0.7767457860337117, "grad_norm": 20.75, "learning_rate": 1.2524302792637427e-06, "loss": 1.9465, "step": 6774 }, { "epoch": 0.7768604517830524, "grad_norm": 14.6875, "learning_rate": 1.2512012562798865e-06, "loss": 1.8678, "step": 6775 }, { "epoch": 0.7769751175323931, "grad_norm": 11.9375, "learning_rate": 1.2499727503713593e-06, "loss": 1.3355, "step": 6776 }, { "epoch": 0.7770897832817337, "grad_norm": 17.5, "learning_rate": 1.2487447617076082e-06, "loss": 1.4105, "step": 6777 }, { "epoch": 0.7772044490310744, "grad_norm": 12.0, "learning_rate": 1.2475172904580118e-06, "loss": 1.4031, "step": 6778 }, { "epoch": 0.777319114780415, "grad_norm": 15.3125, "learning_rate": 1.2462903367918766e-06, "loss": 2.1578, "step": 6779 }, { "epoch": 0.7774337805297558, "grad_norm": 14.1875, "learning_rate": 1.2450639008784366e-06, "loss": 1.5068, "step": 6780 }, { "epoch": 0.7775484462790965, "grad_norm": 14.625, "learning_rate": 1.2438379828868564e-06, "loss": 1.7431, "step": 6781 }, { "epoch": 0.7776631120284371, "grad_norm": 16.0, "learning_rate": 1.242612582986229e-06, "loss": 2.0378, "step": 6782 }, { "epoch": 0.7777777777777778, "grad_norm": 14.9375, "learning_rate": 1.2413877013455717e-06, "loss": 1.5582, "step": 6783 }, { "epoch": 0.7778924435271185, "grad_norm": 14.0625, "learning_rate": 1.240163338133834e-06, "loss": 1.5081, "step": 6784 }, { "epoch": 0.7780071092764591, "grad_norm": 15.8125, "learning_rate": 1.2389394935198973e-06, "loss": 1.9252, "step": 6785 }, { "epoch": 0.7781217750257998, "grad_norm": 14.25, "learning_rate": 1.2377161676725642e-06, "loss": 2.0346, "step": 6786 }, { "epoch": 0.7782364407751404, "grad_norm": 12.0625, "learning_rate": 1.23649336076057e-06, "loss": 1.5265, "step": 6787 }, { "epoch": 0.7783511065244811, "grad_norm": 29.0, "learning_rate": 1.2352710729525791e-06, "loss": 1.641, "step": 6788 }, { "epoch": 0.7784657722738219, "grad_norm": 15.0, "learning_rate": 1.2340493044171786e-06, "loss": 1.7811, "step": 6789 }, { "epoch": 0.7785804380231625, "grad_norm": 11.5625, "learning_rate": 1.2328280553228922e-06, "loss": 1.4794, "step": 6790 }, { "epoch": 0.7786951037725032, "grad_norm": 14.3125, "learning_rate": 1.2316073258381683e-06, "loss": 1.945, "step": 6791 }, { "epoch": 0.7788097695218438, "grad_norm": 13.375, "learning_rate": 1.2303871161313807e-06, "loss": 1.8169, "step": 6792 }, { "epoch": 0.7789244352711845, "grad_norm": 33.75, "learning_rate": 1.2291674263708346e-06, "loss": 2.2058, "step": 6793 }, { "epoch": 0.7790391010205252, "grad_norm": 15.1875, "learning_rate": 1.2279482567247636e-06, "loss": 1.5824, "step": 6794 }, { "epoch": 0.7791537667698658, "grad_norm": 12.75, "learning_rate": 1.2267296073613288e-06, "loss": 1.7668, "step": 6795 }, { "epoch": 0.7792684325192065, "grad_norm": 16.0, "learning_rate": 1.2255114784486194e-06, "loss": 1.8633, "step": 6796 }, { "epoch": 0.7793830982685472, "grad_norm": 14.0, "learning_rate": 1.2242938701546547e-06, "loss": 1.6866, "step": 6797 }, { "epoch": 0.7794977640178878, "grad_norm": 29.125, "learning_rate": 1.2230767826473778e-06, "loss": 1.6327, "step": 6798 }, { "epoch": 0.7796124297672286, "grad_norm": 12.875, "learning_rate": 1.2218602160946642e-06, "loss": 2.0935, "step": 6799 }, { "epoch": 0.7797270955165692, "grad_norm": 15.8125, "learning_rate": 1.2206441706643152e-06, "loss": 1.728, "step": 6800 }, { "epoch": 0.7798417612659099, "grad_norm": 13.3125, "learning_rate": 1.2194286465240623e-06, "loss": 1.4082, "step": 6801 }, { "epoch": 0.7799564270152506, "grad_norm": 14.625, "learning_rate": 1.2182136438415631e-06, "loss": 1.6501, "step": 6802 }, { "epoch": 0.7800710927645912, "grad_norm": 15.3125, "learning_rate": 1.2169991627844064e-06, "loss": 2.0247, "step": 6803 }, { "epoch": 0.7801857585139319, "grad_norm": 18.5, "learning_rate": 1.2157852035201012e-06, "loss": 2.0789, "step": 6804 }, { "epoch": 0.7803004242632725, "grad_norm": 18.375, "learning_rate": 1.2145717662160966e-06, "loss": 1.6091, "step": 6805 }, { "epoch": 0.7804150900126132, "grad_norm": 13.0, "learning_rate": 1.213358851039758e-06, "loss": 1.8928, "step": 6806 }, { "epoch": 0.7805297557619539, "grad_norm": 14.375, "learning_rate": 1.2121464581583869e-06, "loss": 1.5273, "step": 6807 }, { "epoch": 0.7806444215112945, "grad_norm": 12.8125, "learning_rate": 1.2109345877392081e-06, "loss": 1.5829, "step": 6808 }, { "epoch": 0.7807590872606353, "grad_norm": 12.75, "learning_rate": 1.209723239949377e-06, "loss": 1.6043, "step": 6809 }, { "epoch": 0.780873753009976, "grad_norm": 17.125, "learning_rate": 1.2085124149559757e-06, "loss": 2.0747, "step": 6810 }, { "epoch": 0.7809884187593166, "grad_norm": 19.0, "learning_rate": 1.207302112926016e-06, "loss": 1.9274, "step": 6811 }, { "epoch": 0.7811030845086573, "grad_norm": 16.5, "learning_rate": 1.206092334026433e-06, "loss": 1.807, "step": 6812 }, { "epoch": 0.7812177502579979, "grad_norm": 14.9375, "learning_rate": 1.2048830784240939e-06, "loss": 1.593, "step": 6813 }, { "epoch": 0.7813324160073386, "grad_norm": 14.0625, "learning_rate": 1.2036743462857924e-06, "loss": 1.9062, "step": 6814 }, { "epoch": 0.7814470817566793, "grad_norm": 11.5625, "learning_rate": 1.2024661377782503e-06, "loss": 1.653, "step": 6815 }, { "epoch": 0.7815617475060199, "grad_norm": 12.5625, "learning_rate": 1.2012584530681175e-06, "loss": 1.6389, "step": 6816 }, { "epoch": 0.7816764132553606, "grad_norm": 13.25, "learning_rate": 1.2000512923219715e-06, "loss": 1.4447, "step": 6817 }, { "epoch": 0.7817910790047014, "grad_norm": 15.5625, "learning_rate": 1.1988446557063149e-06, "loss": 2.1227, "step": 6818 }, { "epoch": 0.781905744754042, "grad_norm": 22.125, "learning_rate": 1.1976385433875798e-06, "loss": 2.0988, "step": 6819 }, { "epoch": 0.7820204105033827, "grad_norm": 14.875, "learning_rate": 1.1964329555321308e-06, "loss": 1.6011, "step": 6820 }, { "epoch": 0.7821350762527233, "grad_norm": 15.0, "learning_rate": 1.1952278923062521e-06, "loss": 1.7743, "step": 6821 }, { "epoch": 0.782249742002064, "grad_norm": 13.375, "learning_rate": 1.1940233538761597e-06, "loss": 1.5437, "step": 6822 }, { "epoch": 0.7823644077514047, "grad_norm": 13.125, "learning_rate": 1.1928193404079968e-06, "loss": 1.8159, "step": 6823 }, { "epoch": 0.7824790735007453, "grad_norm": 15.5, "learning_rate": 1.1916158520678351e-06, "loss": 1.6458, "step": 6824 }, { "epoch": 0.782593739250086, "grad_norm": 14.0, "learning_rate": 1.1904128890216715e-06, "loss": 1.6306, "step": 6825 }, { "epoch": 0.7827084049994266, "grad_norm": 17.375, "learning_rate": 1.1892104514354347e-06, "loss": 2.1266, "step": 6826 }, { "epoch": 0.7828230707487673, "grad_norm": 15.75, "learning_rate": 1.188008539474974e-06, "loss": 1.6809, "step": 6827 }, { "epoch": 0.782937736498108, "grad_norm": 13.375, "learning_rate": 1.1868071533060722e-06, "loss": 1.5425, "step": 6828 }, { "epoch": 0.7830524022474487, "grad_norm": 14.0, "learning_rate": 1.1856062930944374e-06, "loss": 1.4734, "step": 6829 }, { "epoch": 0.7831670679967894, "grad_norm": 12.6875, "learning_rate": 1.1844059590057056e-06, "loss": 1.8502, "step": 6830 }, { "epoch": 0.7832817337461301, "grad_norm": 17.875, "learning_rate": 1.18320615120544e-06, "loss": 1.4622, "step": 6831 }, { "epoch": 0.7833963994954707, "grad_norm": 13.1875, "learning_rate": 1.1820068698591325e-06, "loss": 1.7168, "step": 6832 }, { "epoch": 0.7835110652448114, "grad_norm": 15.375, "learning_rate": 1.1808081151321982e-06, "loss": 1.8193, "step": 6833 }, { "epoch": 0.783625730994152, "grad_norm": 27.0, "learning_rate": 1.1796098871899826e-06, "loss": 1.6744, "step": 6834 }, { "epoch": 0.7837403967434927, "grad_norm": 16.375, "learning_rate": 1.1784121861977627e-06, "loss": 1.5698, "step": 6835 }, { "epoch": 0.7838550624928334, "grad_norm": 18.625, "learning_rate": 1.1772150123207343e-06, "loss": 2.2624, "step": 6836 }, { "epoch": 0.783969728242174, "grad_norm": 12.75, "learning_rate": 1.176018365724026e-06, "loss": 1.6638, "step": 6837 }, { "epoch": 0.7840843939915147, "grad_norm": 14.9375, "learning_rate": 1.1748222465726938e-06, "loss": 1.6247, "step": 6838 }, { "epoch": 0.7841990597408554, "grad_norm": 12.3125, "learning_rate": 1.1736266550317156e-06, "loss": 1.5318, "step": 6839 }, { "epoch": 0.7843137254901961, "grad_norm": 14.625, "learning_rate": 1.1724315912660038e-06, "loss": 1.7019, "step": 6840 }, { "epoch": 0.7844283912395368, "grad_norm": 18.875, "learning_rate": 1.1712370554403957e-06, "loss": 1.8044, "step": 6841 }, { "epoch": 0.7845430569888774, "grad_norm": 13.125, "learning_rate": 1.1700430477196517e-06, "loss": 1.5568, "step": 6842 }, { "epoch": 0.7846577227382181, "grad_norm": 12.25, "learning_rate": 1.168849568268463e-06, "loss": 1.92, "step": 6843 }, { "epoch": 0.7847723884875588, "grad_norm": 15.0, "learning_rate": 1.1676566172514486e-06, "loss": 1.7374, "step": 6844 }, { "epoch": 0.7848870542368994, "grad_norm": 14.3125, "learning_rate": 1.166464194833153e-06, "loss": 1.5703, "step": 6845 }, { "epoch": 0.7850017199862401, "grad_norm": 29.25, "learning_rate": 1.165272301178047e-06, "loss": 2.0905, "step": 6846 }, { "epoch": 0.7851163857355807, "grad_norm": 17.75, "learning_rate": 1.1640809364505329e-06, "loss": 2.0714, "step": 6847 }, { "epoch": 0.7852310514849214, "grad_norm": 16.375, "learning_rate": 1.1628901008149323e-06, "loss": 1.824, "step": 6848 }, { "epoch": 0.7853457172342622, "grad_norm": 16.375, "learning_rate": 1.1616997944355002e-06, "loss": 1.5192, "step": 6849 }, { "epoch": 0.7854603829836028, "grad_norm": 17.5, "learning_rate": 1.1605100174764172e-06, "loss": 2.0875, "step": 6850 }, { "epoch": 0.7855750487329435, "grad_norm": 23.75, "learning_rate": 1.1593207701017895e-06, "loss": 2.3036, "step": 6851 }, { "epoch": 0.7856897144822842, "grad_norm": 34.25, "learning_rate": 1.1581320524756518e-06, "loss": 2.4428, "step": 6852 }, { "epoch": 0.7858043802316248, "grad_norm": 14.8125, "learning_rate": 1.1569438647619662e-06, "loss": 1.5622, "step": 6853 }, { "epoch": 0.7859190459809655, "grad_norm": 15.3125, "learning_rate": 1.1557562071246164e-06, "loss": 1.7057, "step": 6854 }, { "epoch": 0.7860337117303061, "grad_norm": 14.0625, "learning_rate": 1.1545690797274229e-06, "loss": 1.5875, "step": 6855 }, { "epoch": 0.7861483774796468, "grad_norm": 16.125, "learning_rate": 1.1533824827341228e-06, "loss": 1.6739, "step": 6856 }, { "epoch": 0.7862630432289875, "grad_norm": 15.0, "learning_rate": 1.1521964163083855e-06, "loss": 1.4722, "step": 6857 }, { "epoch": 0.7863777089783281, "grad_norm": 14.125, "learning_rate": 1.151010880613807e-06, "loss": 1.5876, "step": 6858 }, { "epoch": 0.7864923747276689, "grad_norm": 15.25, "learning_rate": 1.1498258758139097e-06, "loss": 1.6249, "step": 6859 }, { "epoch": 0.7866070404770095, "grad_norm": 16.375, "learning_rate": 1.1486414020721409e-06, "loss": 1.9136, "step": 6860 }, { "epoch": 0.7867217062263502, "grad_norm": 10.75, "learning_rate": 1.1474574595518777e-06, "loss": 1.6688, "step": 6861 }, { "epoch": 0.7868363719756909, "grad_norm": 17.25, "learning_rate": 1.146274048416423e-06, "loss": 1.7269, "step": 6862 }, { "epoch": 0.7869510377250315, "grad_norm": 17.625, "learning_rate": 1.1450911688290034e-06, "loss": 1.811, "step": 6863 }, { "epoch": 0.7870657034743722, "grad_norm": 14.0625, "learning_rate": 1.1439088209527755e-06, "loss": 1.5307, "step": 6864 }, { "epoch": 0.7871803692237129, "grad_norm": 15.9375, "learning_rate": 1.142727004950822e-06, "loss": 1.9586, "step": 6865 }, { "epoch": 0.7872950349730535, "grad_norm": 33.0, "learning_rate": 1.1415457209861524e-06, "loss": 1.8753, "step": 6866 }, { "epoch": 0.7874097007223942, "grad_norm": 16.625, "learning_rate": 1.1403649692217011e-06, "loss": 1.9744, "step": 6867 }, { "epoch": 0.7875243664717348, "grad_norm": 14.6875, "learning_rate": 1.1391847498203329e-06, "loss": 1.5354, "step": 6868 }, { "epoch": 0.7876390322210756, "grad_norm": 12.6875, "learning_rate": 1.1380050629448315e-06, "loss": 1.6925, "step": 6869 }, { "epoch": 0.7877536979704163, "grad_norm": 17.875, "learning_rate": 1.1368259087579181e-06, "loss": 1.7329, "step": 6870 }, { "epoch": 0.7878683637197569, "grad_norm": 13.375, "learning_rate": 1.1356472874222308e-06, "loss": 1.7321, "step": 6871 }, { "epoch": 0.7879830294690976, "grad_norm": 17.0, "learning_rate": 1.1344691991003388e-06, "loss": 2.181, "step": 6872 }, { "epoch": 0.7880976952184382, "grad_norm": 13.4375, "learning_rate": 1.1332916439547365e-06, "loss": 1.6028, "step": 6873 }, { "epoch": 0.7882123609677789, "grad_norm": 11.875, "learning_rate": 1.1321146221478462e-06, "loss": 1.6831, "step": 6874 }, { "epoch": 0.7883270267171196, "grad_norm": 14.1875, "learning_rate": 1.130938133842015e-06, "loss": 1.6259, "step": 6875 }, { "epoch": 0.7884416924664602, "grad_norm": 13.625, "learning_rate": 1.129762179199519e-06, "loss": 1.5244, "step": 6876 }, { "epoch": 0.788556358215801, "grad_norm": 14.4375, "learning_rate": 1.128586758382555e-06, "loss": 2.031, "step": 6877 }, { "epoch": 0.7886710239651417, "grad_norm": 14.8125, "learning_rate": 1.1274118715532522e-06, "loss": 1.8111, "step": 6878 }, { "epoch": 0.7887856897144823, "grad_norm": 15.8125, "learning_rate": 1.1262375188736635e-06, "loss": 2.1251, "step": 6879 }, { "epoch": 0.788900355463823, "grad_norm": 18.625, "learning_rate": 1.1250637005057685e-06, "loss": 1.8858, "step": 6880 }, { "epoch": 0.7890150212131636, "grad_norm": 11.6875, "learning_rate": 1.1238904166114728e-06, "loss": 1.6737, "step": 6881 }, { "epoch": 0.7891296869625043, "grad_norm": 13.125, "learning_rate": 1.1227176673526102e-06, "loss": 1.5907, "step": 6882 }, { "epoch": 0.789244352711845, "grad_norm": 15.1875, "learning_rate": 1.1215454528909358e-06, "loss": 1.4747, "step": 6883 }, { "epoch": 0.7893590184611856, "grad_norm": 15.5, "learning_rate": 1.1203737733881354e-06, "loss": 1.6241, "step": 6884 }, { "epoch": 0.7894736842105263, "grad_norm": 15.25, "learning_rate": 1.1192026290058233e-06, "loss": 2.1449, "step": 6885 }, { "epoch": 0.789588349959867, "grad_norm": 12.625, "learning_rate": 1.1180320199055321e-06, "loss": 1.6829, "step": 6886 }, { "epoch": 0.7897030157092076, "grad_norm": 14.9375, "learning_rate": 1.1168619462487269e-06, "loss": 1.7374, "step": 6887 }, { "epoch": 0.7898176814585484, "grad_norm": 13.9375, "learning_rate": 1.115692408196798e-06, "loss": 1.6121, "step": 6888 }, { "epoch": 0.789932347207889, "grad_norm": 12.5, "learning_rate": 1.1145234059110572e-06, "loss": 1.6382, "step": 6889 }, { "epoch": 0.7900470129572297, "grad_norm": 16.625, "learning_rate": 1.11335493955275e-06, "loss": 2.0167, "step": 6890 }, { "epoch": 0.7901616787065704, "grad_norm": 14.9375, "learning_rate": 1.112187009283044e-06, "loss": 2.0725, "step": 6891 }, { "epoch": 0.790276344455911, "grad_norm": 13.625, "learning_rate": 1.1110196152630303e-06, "loss": 1.7701, "step": 6892 }, { "epoch": 0.7903910102052517, "grad_norm": 12.5625, "learning_rate": 1.1098527576537299e-06, "loss": 1.4798, "step": 6893 }, { "epoch": 0.7905056759545923, "grad_norm": 18.25, "learning_rate": 1.1086864366160888e-06, "loss": 2.0132, "step": 6894 }, { "epoch": 0.790620341703933, "grad_norm": 13.875, "learning_rate": 1.1075206523109783e-06, "loss": 1.9309, "step": 6895 }, { "epoch": 0.7907350074532737, "grad_norm": 17.0, "learning_rate": 1.1063554048991965e-06, "loss": 1.7412, "step": 6896 }, { "epoch": 0.7908496732026143, "grad_norm": 14.125, "learning_rate": 1.1051906945414687e-06, "loss": 1.5621, "step": 6897 }, { "epoch": 0.7909643389519551, "grad_norm": 25.5, "learning_rate": 1.1040265213984408e-06, "loss": 2.3434, "step": 6898 }, { "epoch": 0.7910790047012958, "grad_norm": 14.0625, "learning_rate": 1.1028628856306906e-06, "loss": 1.9133, "step": 6899 }, { "epoch": 0.7911936704506364, "grad_norm": 19.5, "learning_rate": 1.1016997873987189e-06, "loss": 1.8822, "step": 6900 }, { "epoch": 0.7913083361999771, "grad_norm": 14.25, "learning_rate": 1.1005372268629538e-06, "loss": 1.6948, "step": 6901 }, { "epoch": 0.7914230019493177, "grad_norm": 16.0, "learning_rate": 1.0993752041837473e-06, "loss": 1.3692, "step": 6902 }, { "epoch": 0.7915376676986584, "grad_norm": 20.125, "learning_rate": 1.09821371952138e-06, "loss": 1.4919, "step": 6903 }, { "epoch": 0.7916523334479991, "grad_norm": 22.75, "learning_rate": 1.0970527730360524e-06, "loss": 2.1109, "step": 6904 }, { "epoch": 0.7917669991973397, "grad_norm": 12.125, "learning_rate": 1.0958923648878993e-06, "loss": 2.112, "step": 6905 }, { "epoch": 0.7918816649466804, "grad_norm": 12.0, "learning_rate": 1.094732495236977e-06, "loss": 1.7903, "step": 6906 }, { "epoch": 0.791996330696021, "grad_norm": 18.125, "learning_rate": 1.0935731642432645e-06, "loss": 1.5522, "step": 6907 }, { "epoch": 0.7921109964453618, "grad_norm": 14.625, "learning_rate": 1.0924143720666708e-06, "loss": 1.5516, "step": 6908 }, { "epoch": 0.7922256621947025, "grad_norm": 12.625, "learning_rate": 1.0912561188670285e-06, "loss": 1.5385, "step": 6909 }, { "epoch": 0.7923403279440431, "grad_norm": 15.875, "learning_rate": 1.0900984048040974e-06, "loss": 1.8326, "step": 6910 }, { "epoch": 0.7924549936933838, "grad_norm": 10.75, "learning_rate": 1.0889412300375624e-06, "loss": 1.6331, "step": 6911 }, { "epoch": 0.7925696594427245, "grad_norm": 11.1875, "learning_rate": 1.0877845947270338e-06, "loss": 1.6017, "step": 6912 }, { "epoch": 0.7926843251920651, "grad_norm": 11.9375, "learning_rate": 1.0866284990320458e-06, "loss": 1.9195, "step": 6913 }, { "epoch": 0.7927989909414058, "grad_norm": 13.6875, "learning_rate": 1.0854729431120608e-06, "loss": 1.5996, "step": 6914 }, { "epoch": 0.7929136566907464, "grad_norm": 15.5, "learning_rate": 1.0843179271264653e-06, "loss": 2.0787, "step": 6915 }, { "epoch": 0.7930283224400871, "grad_norm": 12.6875, "learning_rate": 1.0831634512345724e-06, "loss": 1.7559, "step": 6916 }, { "epoch": 0.7931429881894279, "grad_norm": 10.625, "learning_rate": 1.0820095155956195e-06, "loss": 1.6196, "step": 6917 }, { "epoch": 0.7932576539387685, "grad_norm": 13.5, "learning_rate": 1.0808561203687712e-06, "loss": 1.9312, "step": 6918 }, { "epoch": 0.7933723196881092, "grad_norm": 14.8125, "learning_rate": 1.0797032657131135e-06, "loss": 1.4604, "step": 6919 }, { "epoch": 0.7934869854374499, "grad_norm": 15.25, "learning_rate": 1.0785509517876646e-06, "loss": 1.8923, "step": 6920 }, { "epoch": 0.7936016511867905, "grad_norm": 24.5, "learning_rate": 1.0773991787513617e-06, "loss": 2.3543, "step": 6921 }, { "epoch": 0.7937163169361312, "grad_norm": 17.0, "learning_rate": 1.0762479467630698e-06, "loss": 1.6938, "step": 6922 }, { "epoch": 0.7938309826854718, "grad_norm": 17.375, "learning_rate": 1.0750972559815803e-06, "loss": 1.5475, "step": 6923 }, { "epoch": 0.7939456484348125, "grad_norm": 17.5, "learning_rate": 1.0739471065656092e-06, "loss": 1.6767, "step": 6924 }, { "epoch": 0.7940603141841532, "grad_norm": 13.9375, "learning_rate": 1.072797498673797e-06, "loss": 1.5205, "step": 6925 }, { "epoch": 0.7941749799334938, "grad_norm": 14.25, "learning_rate": 1.0716484324647119e-06, "loss": 2.0168, "step": 6926 }, { "epoch": 0.7942896456828346, "grad_norm": 23.625, "learning_rate": 1.070499908096843e-06, "loss": 1.5013, "step": 6927 }, { "epoch": 0.7944043114321752, "grad_norm": 14.3125, "learning_rate": 1.0693519257286096e-06, "loss": 1.4811, "step": 6928 }, { "epoch": 0.7945189771815159, "grad_norm": 16.0, "learning_rate": 1.068204485518352e-06, "loss": 1.836, "step": 6929 }, { "epoch": 0.7946336429308566, "grad_norm": 20.375, "learning_rate": 1.0670575876243388e-06, "loss": 1.9757, "step": 6930 }, { "epoch": 0.7947483086801972, "grad_norm": 16.75, "learning_rate": 1.0659112322047631e-06, "loss": 2.0332, "step": 6931 }, { "epoch": 0.7948629744295379, "grad_norm": 14.25, "learning_rate": 1.0647654194177437e-06, "loss": 1.6995, "step": 6932 }, { "epoch": 0.7949776401788786, "grad_norm": 15.6875, "learning_rate": 1.0636201494213212e-06, "loss": 1.875, "step": 6933 }, { "epoch": 0.7950923059282192, "grad_norm": 15.9375, "learning_rate": 1.062475422373463e-06, "loss": 1.4721, "step": 6934 }, { "epoch": 0.7952069716775599, "grad_norm": 14.25, "learning_rate": 1.0613312384320674e-06, "loss": 1.6554, "step": 6935 }, { "epoch": 0.7953216374269005, "grad_norm": 16.125, "learning_rate": 1.0601875977549476e-06, "loss": 2.0964, "step": 6936 }, { "epoch": 0.7954363031762413, "grad_norm": 19.75, "learning_rate": 1.0590445004998496e-06, "loss": 1.7477, "step": 6937 }, { "epoch": 0.795550968925582, "grad_norm": 13.3125, "learning_rate": 1.0579019468244423e-06, "loss": 1.7047, "step": 6938 }, { "epoch": 0.7956656346749226, "grad_norm": 14.3125, "learning_rate": 1.0567599368863157e-06, "loss": 1.9351, "step": 6939 }, { "epoch": 0.7957803004242633, "grad_norm": 16.625, "learning_rate": 1.0556184708429924e-06, "loss": 1.6752, "step": 6940 }, { "epoch": 0.7958949661736039, "grad_norm": 12.4375, "learning_rate": 1.0544775488519154e-06, "loss": 1.587, "step": 6941 }, { "epoch": 0.7960096319229446, "grad_norm": 12.5625, "learning_rate": 1.0533371710704505e-06, "loss": 1.4895, "step": 6942 }, { "epoch": 0.7961242976722853, "grad_norm": 13.125, "learning_rate": 1.0521973376558926e-06, "loss": 1.6118, "step": 6943 }, { "epoch": 0.7962389634216259, "grad_norm": 16.75, "learning_rate": 1.0510580487654603e-06, "loss": 1.7671, "step": 6944 }, { "epoch": 0.7963536291709666, "grad_norm": 32.75, "learning_rate": 1.0499193045562965e-06, "loss": 1.8871, "step": 6945 }, { "epoch": 0.7964682949203074, "grad_norm": 13.25, "learning_rate": 1.048781105185469e-06, "loss": 1.72, "step": 6946 }, { "epoch": 0.796582960669648, "grad_norm": 12.25, "learning_rate": 1.0476434508099725e-06, "loss": 1.6414, "step": 6947 }, { "epoch": 0.7966976264189887, "grad_norm": 14.875, "learning_rate": 1.0465063415867216e-06, "loss": 1.8667, "step": 6948 }, { "epoch": 0.7968122921683293, "grad_norm": 15.25, "learning_rate": 1.0453697776725595e-06, "loss": 1.4848, "step": 6949 }, { "epoch": 0.79692695791767, "grad_norm": 12.9375, "learning_rate": 1.044233759224257e-06, "loss": 2.0521, "step": 6950 }, { "epoch": 0.7970416236670107, "grad_norm": 21.375, "learning_rate": 1.0430982863985023e-06, "loss": 1.5996, "step": 6951 }, { "epoch": 0.7971562894163513, "grad_norm": 13.9375, "learning_rate": 1.0419633593519135e-06, "loss": 1.4751, "step": 6952 }, { "epoch": 0.797270955165692, "grad_norm": 14.5625, "learning_rate": 1.0408289782410335e-06, "loss": 1.8055, "step": 6953 }, { "epoch": 0.7973856209150327, "grad_norm": 20.25, "learning_rate": 1.0396951432223246e-06, "loss": 2.0692, "step": 6954 }, { "epoch": 0.7975002866643733, "grad_norm": 20.25, "learning_rate": 1.038561854452182e-06, "loss": 2.1144, "step": 6955 }, { "epoch": 0.797614952413714, "grad_norm": 14.125, "learning_rate": 1.0374291120869206e-06, "loss": 1.9075, "step": 6956 }, { "epoch": 0.7977296181630547, "grad_norm": 25.625, "learning_rate": 1.0362969162827785e-06, "loss": 1.7409, "step": 6957 }, { "epoch": 0.7978442839123954, "grad_norm": 16.625, "learning_rate": 1.0351652671959223e-06, "loss": 1.7978, "step": 6958 }, { "epoch": 0.7979589496617361, "grad_norm": 11.8125, "learning_rate": 1.0340341649824403e-06, "loss": 1.712, "step": 6959 }, { "epoch": 0.7980736154110767, "grad_norm": 16.875, "learning_rate": 1.0329036097983476e-06, "loss": 1.3277, "step": 6960 }, { "epoch": 0.7981882811604174, "grad_norm": 13.1875, "learning_rate": 1.0317736017995816e-06, "loss": 1.771, "step": 6961 }, { "epoch": 0.798302946909758, "grad_norm": 22.0, "learning_rate": 1.0306441411420077e-06, "loss": 1.8829, "step": 6962 }, { "epoch": 0.7984176126590987, "grad_norm": 13.0, "learning_rate": 1.0295152279814098e-06, "loss": 1.5322, "step": 6963 }, { "epoch": 0.7985322784084394, "grad_norm": 17.0, "learning_rate": 1.0283868624735022e-06, "loss": 1.4513, "step": 6964 }, { "epoch": 0.79864694415778, "grad_norm": 15.5625, "learning_rate": 1.027259044773921e-06, "loss": 1.7578, "step": 6965 }, { "epoch": 0.7987616099071208, "grad_norm": 15.25, "learning_rate": 1.0261317750382276e-06, "loss": 2.096, "step": 6966 }, { "epoch": 0.7988762756564615, "grad_norm": 13.0625, "learning_rate": 1.0250050534219068e-06, "loss": 1.8802, "step": 6967 }, { "epoch": 0.7989909414058021, "grad_norm": 27.75, "learning_rate": 1.02387888008037e-06, "loss": 1.955, "step": 6968 }, { "epoch": 0.7991056071551428, "grad_norm": 17.0, "learning_rate": 1.0227532551689472e-06, "loss": 1.7589, "step": 6969 }, { "epoch": 0.7992202729044834, "grad_norm": 14.25, "learning_rate": 1.0216281788429022e-06, "loss": 1.6935, "step": 6970 }, { "epoch": 0.7993349386538241, "grad_norm": 12.8125, "learning_rate": 1.020503651257414e-06, "loss": 1.4312, "step": 6971 }, { "epoch": 0.7994496044031648, "grad_norm": 15.625, "learning_rate": 1.019379672567591e-06, "loss": 1.5572, "step": 6972 }, { "epoch": 0.7995642701525054, "grad_norm": 14.3125, "learning_rate": 1.0182562429284647e-06, "loss": 1.8863, "step": 6973 }, { "epoch": 0.7996789359018461, "grad_norm": 11.8125, "learning_rate": 1.0171333624949909e-06, "loss": 1.4807, "step": 6974 }, { "epoch": 0.7997936016511867, "grad_norm": 14.75, "learning_rate": 1.016011031422049e-06, "loss": 1.7185, "step": 6975 }, { "epoch": 0.7999082674005275, "grad_norm": 16.125, "learning_rate": 1.0148892498644453e-06, "loss": 1.7802, "step": 6976 }, { "epoch": 0.8000229331498682, "grad_norm": 30.125, "learning_rate": 1.013768017976905e-06, "loss": 1.8067, "step": 6977 }, { "epoch": 0.8001375988992088, "grad_norm": 23.75, "learning_rate": 1.012647335914082e-06, "loss": 2.092, "step": 6978 }, { "epoch": 0.8002522646485495, "grad_norm": 12.8125, "learning_rate": 1.011527203830553e-06, "loss": 1.568, "step": 6979 }, { "epoch": 0.8003669303978902, "grad_norm": 15.4375, "learning_rate": 1.0104076218808195e-06, "loss": 1.8508, "step": 6980 }, { "epoch": 0.8004815961472308, "grad_norm": 16.0, "learning_rate": 1.0092885902193056e-06, "loss": 1.7379, "step": 6981 }, { "epoch": 0.8005962618965715, "grad_norm": 16.75, "learning_rate": 1.0081701090003626e-06, "loss": 1.7868, "step": 6982 }, { "epoch": 0.8007109276459121, "grad_norm": 16.75, "learning_rate": 1.00705217837826e-06, "loss": 1.8581, "step": 6983 }, { "epoch": 0.8008255933952528, "grad_norm": 13.4375, "learning_rate": 1.005934798507196e-06, "loss": 1.8077, "step": 6984 }, { "epoch": 0.8009402591445935, "grad_norm": 14.8125, "learning_rate": 1.004817969541295e-06, "loss": 1.5208, "step": 6985 }, { "epoch": 0.8010549248939342, "grad_norm": 13.125, "learning_rate": 1.0037016916345987e-06, "loss": 1.3426, "step": 6986 }, { "epoch": 0.8011695906432749, "grad_norm": 15.625, "learning_rate": 1.002585964941078e-06, "loss": 1.5929, "step": 6987 }, { "epoch": 0.8012842563926156, "grad_norm": 13.6875, "learning_rate": 1.0014707896146269e-06, "loss": 2.0188, "step": 6988 }, { "epoch": 0.8013989221419562, "grad_norm": 19.5, "learning_rate": 1.0003561658090589e-06, "loss": 1.8517, "step": 6989 }, { "epoch": 0.8015135878912969, "grad_norm": 15.25, "learning_rate": 9.99242093678119e-07, "loss": 1.3143, "step": 6990 }, { "epoch": 0.8016282536406375, "grad_norm": 25.375, "learning_rate": 9.98128573375472e-07, "loss": 1.8913, "step": 6991 }, { "epoch": 0.8017429193899782, "grad_norm": 12.75, "learning_rate": 9.970156050547047e-07, "loss": 1.7669, "step": 6992 }, { "epoch": 0.8018575851393189, "grad_norm": 17.5, "learning_rate": 9.959031888693304e-07, "loss": 1.5954, "step": 6993 }, { "epoch": 0.8019722508886595, "grad_norm": 15.875, "learning_rate": 9.947913249727865e-07, "loss": 1.8437, "step": 6994 }, { "epoch": 0.8020869166380002, "grad_norm": 15.4375, "learning_rate": 9.936800135184332e-07, "loss": 2.0959, "step": 6995 }, { "epoch": 0.8022015823873409, "grad_norm": 16.625, "learning_rate": 9.925692546595544e-07, "loss": 1.7533, "step": 6996 }, { "epoch": 0.8023162481366816, "grad_norm": 14.375, "learning_rate": 9.914590485493598e-07, "loss": 1.9003, "step": 6997 }, { "epoch": 0.8024309138860223, "grad_norm": 32.75, "learning_rate": 9.903493953409782e-07, "loss": 2.0463, "step": 6998 }, { "epoch": 0.8025455796353629, "grad_norm": 14.0, "learning_rate": 9.892402951874657e-07, "loss": 1.6758, "step": 6999 }, { "epoch": 0.8026602453847036, "grad_norm": 14.9375, "learning_rate": 9.881317482418045e-07, "loss": 1.915, "step": 7000 }, { "epoch": 0.8027749111340443, "grad_norm": 14.4375, "learning_rate": 9.87023754656895e-07, "loss": 1.7013, "step": 7001 }, { "epoch": 0.8028895768833849, "grad_norm": 15.8125, "learning_rate": 9.859163145855632e-07, "loss": 1.5586, "step": 7002 }, { "epoch": 0.8030042426327256, "grad_norm": 12.375, "learning_rate": 9.848094281805626e-07, "loss": 1.7728, "step": 7003 }, { "epoch": 0.8031189083820662, "grad_norm": 16.25, "learning_rate": 9.83703095594562e-07, "loss": 2.0591, "step": 7004 }, { "epoch": 0.803233574131407, "grad_norm": 17.125, "learning_rate": 9.825973169801627e-07, "loss": 1.8327, "step": 7005 }, { "epoch": 0.8033482398807477, "grad_norm": 14.9375, "learning_rate": 9.81492092489887e-07, "loss": 1.5677, "step": 7006 }, { "epoch": 0.8034629056300883, "grad_norm": 16.875, "learning_rate": 9.80387422276175e-07, "loss": 1.7157, "step": 7007 }, { "epoch": 0.803577571379429, "grad_norm": 17.625, "learning_rate": 9.792833064913982e-07, "loss": 1.8252, "step": 7008 }, { "epoch": 0.8036922371287696, "grad_norm": 12.5, "learning_rate": 9.781797452878471e-07, "loss": 1.5275, "step": 7009 }, { "epoch": 0.8038069028781103, "grad_norm": 12.6875, "learning_rate": 9.77076738817737e-07, "loss": 1.5935, "step": 7010 }, { "epoch": 0.803921568627451, "grad_norm": 14.875, "learning_rate": 9.759742872332073e-07, "loss": 1.9115, "step": 7011 }, { "epoch": 0.8040362343767916, "grad_norm": 13.0, "learning_rate": 9.748723906863211e-07, "loss": 1.9318, "step": 7012 }, { "epoch": 0.8041509001261323, "grad_norm": 16.25, "learning_rate": 9.737710493290615e-07, "loss": 1.5515, "step": 7013 }, { "epoch": 0.804265565875473, "grad_norm": 15.9375, "learning_rate": 9.726702633133383e-07, "loss": 2.0472, "step": 7014 }, { "epoch": 0.8043802316248136, "grad_norm": 14.5, "learning_rate": 9.715700327909844e-07, "loss": 1.6102, "step": 7015 }, { "epoch": 0.8044948973741544, "grad_norm": 11.25, "learning_rate": 9.704703579137559e-07, "loss": 1.2474, "step": 7016 }, { "epoch": 0.804609563123495, "grad_norm": 13.9375, "learning_rate": 9.693712388333316e-07, "loss": 1.5112, "step": 7017 }, { "epoch": 0.8047242288728357, "grad_norm": 13.9375, "learning_rate": 9.682726757013155e-07, "loss": 1.609, "step": 7018 }, { "epoch": 0.8048388946221764, "grad_norm": 16.25, "learning_rate": 9.671746686692296e-07, "loss": 1.7258, "step": 7019 }, { "epoch": 0.804953560371517, "grad_norm": 14.8125, "learning_rate": 9.660772178885274e-07, "loss": 1.7816, "step": 7020 }, { "epoch": 0.8050682261208577, "grad_norm": 15.5, "learning_rate": 9.649803235105787e-07, "loss": 1.5831, "step": 7021 }, { "epoch": 0.8051828918701984, "grad_norm": 14.1875, "learning_rate": 9.6388398568668e-07, "loss": 2.0293, "step": 7022 }, { "epoch": 0.805297557619539, "grad_norm": 13.125, "learning_rate": 9.627882045680498e-07, "loss": 1.7143, "step": 7023 }, { "epoch": 0.8054122233688797, "grad_norm": 26.875, "learning_rate": 9.616929803058306e-07, "loss": 1.952, "step": 7024 }, { "epoch": 0.8055268891182203, "grad_norm": 18.75, "learning_rate": 9.605983130510876e-07, "loss": 1.9376, "step": 7025 }, { "epoch": 0.8056415548675611, "grad_norm": 15.4375, "learning_rate": 9.595042029548112e-07, "loss": 1.4863, "step": 7026 }, { "epoch": 0.8057562206169018, "grad_norm": 22.5, "learning_rate": 9.584106501679091e-07, "loss": 1.6201, "step": 7027 }, { "epoch": 0.8058708863662424, "grad_norm": 31.375, "learning_rate": 9.573176548412183e-07, "loss": 1.8978, "step": 7028 }, { "epoch": 0.8059855521155831, "grad_norm": 14.8125, "learning_rate": 9.562252171254966e-07, "loss": 1.6546, "step": 7029 }, { "epoch": 0.8061002178649237, "grad_norm": 15.25, "learning_rate": 9.55133337171425e-07, "loss": 1.6144, "step": 7030 }, { "epoch": 0.8062148836142644, "grad_norm": 18.0, "learning_rate": 9.54042015129607e-07, "loss": 1.9887, "step": 7031 }, { "epoch": 0.8063295493636051, "grad_norm": 12.75, "learning_rate": 9.529512511505718e-07, "loss": 1.6298, "step": 7032 }, { "epoch": 0.8064442151129457, "grad_norm": 14.875, "learning_rate": 9.51861045384766e-07, "loss": 1.8003, "step": 7033 }, { "epoch": 0.8065588808622864, "grad_norm": 14.75, "learning_rate": 9.507713979825628e-07, "loss": 1.8741, "step": 7034 }, { "epoch": 0.8066735466116272, "grad_norm": 13.3125, "learning_rate": 9.496823090942631e-07, "loss": 1.6178, "step": 7035 }, { "epoch": 0.8067882123609678, "grad_norm": 13.75, "learning_rate": 9.485937788700817e-07, "loss": 1.511, "step": 7036 }, { "epoch": 0.8069028781103085, "grad_norm": 13.9375, "learning_rate": 9.475058074601612e-07, "loss": 1.6843, "step": 7037 }, { "epoch": 0.8070175438596491, "grad_norm": 16.875, "learning_rate": 9.464183950145683e-07, "loss": 1.6396, "step": 7038 }, { "epoch": 0.8071322096089898, "grad_norm": 14.375, "learning_rate": 9.45331541683287e-07, "loss": 1.7099, "step": 7039 }, { "epoch": 0.8072468753583305, "grad_norm": 14.5, "learning_rate": 9.442452476162323e-07, "loss": 1.6592, "step": 7040 }, { "epoch": 0.8073615411076711, "grad_norm": 16.25, "learning_rate": 9.431595129632365e-07, "loss": 2.2528, "step": 7041 }, { "epoch": 0.8074762068570118, "grad_norm": 11.5, "learning_rate": 9.420743378740544e-07, "loss": 1.5392, "step": 7042 }, { "epoch": 0.8075908726063524, "grad_norm": 15.5, "learning_rate": 9.409897224983666e-07, "loss": 1.5322, "step": 7043 }, { "epoch": 0.8077055383556931, "grad_norm": 13.0, "learning_rate": 9.399056669857748e-07, "loss": 1.6574, "step": 7044 }, { "epoch": 0.8078202041050339, "grad_norm": 14.3125, "learning_rate": 9.388221714858031e-07, "loss": 2.0313, "step": 7045 }, { "epoch": 0.8079348698543745, "grad_norm": 15.3125, "learning_rate": 9.377392361479004e-07, "loss": 1.9871, "step": 7046 }, { "epoch": 0.8080495356037152, "grad_norm": 13.6875, "learning_rate": 9.366568611214377e-07, "loss": 1.8514, "step": 7047 }, { "epoch": 0.8081642013530559, "grad_norm": 15.0625, "learning_rate": 9.355750465557051e-07, "loss": 1.788, "step": 7048 }, { "epoch": 0.8082788671023965, "grad_norm": 13.9375, "learning_rate": 9.344937925999186e-07, "loss": 1.8327, "step": 7049 }, { "epoch": 0.8083935328517372, "grad_norm": 13.9375, "learning_rate": 9.334130994032203e-07, "loss": 1.561, "step": 7050 }, { "epoch": 0.8085081986010778, "grad_norm": 13.9375, "learning_rate": 9.323329671146675e-07, "loss": 1.5907, "step": 7051 }, { "epoch": 0.8086228643504185, "grad_norm": 13.3125, "learning_rate": 9.312533958832454e-07, "loss": 1.5801, "step": 7052 }, { "epoch": 0.8087375300997592, "grad_norm": 15.6875, "learning_rate": 9.30174385857861e-07, "loss": 1.2879, "step": 7053 }, { "epoch": 0.8088521958490998, "grad_norm": 12.5625, "learning_rate": 9.290959371873392e-07, "loss": 1.7009, "step": 7054 }, { "epoch": 0.8089668615984406, "grad_norm": 13.875, "learning_rate": 9.280180500204361e-07, "loss": 1.9715, "step": 7055 }, { "epoch": 0.8090815273477813, "grad_norm": 13.75, "learning_rate": 9.269407245058254e-07, "loss": 2.1806, "step": 7056 }, { "epoch": 0.8091961930971219, "grad_norm": 15.5, "learning_rate": 9.258639607921005e-07, "loss": 1.556, "step": 7057 }, { "epoch": 0.8093108588464626, "grad_norm": 16.125, "learning_rate": 9.24787759027782e-07, "loss": 1.921, "step": 7058 }, { "epoch": 0.8094255245958032, "grad_norm": 15.0625, "learning_rate": 9.237121193613119e-07, "loss": 1.6796, "step": 7059 }, { "epoch": 0.8095401903451439, "grad_norm": 15.125, "learning_rate": 9.226370419410541e-07, "loss": 1.6357, "step": 7060 }, { "epoch": 0.8096548560944846, "grad_norm": 11.625, "learning_rate": 9.21562526915294e-07, "loss": 1.6663, "step": 7061 }, { "epoch": 0.8097695218438252, "grad_norm": 13.5625, "learning_rate": 9.204885744322428e-07, "loss": 1.2978, "step": 7062 }, { "epoch": 0.8098841875931659, "grad_norm": 9.75, "learning_rate": 9.19415184640029e-07, "loss": 1.8527, "step": 7063 }, { "epoch": 0.8099988533425065, "grad_norm": 17.0, "learning_rate": 9.183423576867079e-07, "loss": 1.6675, "step": 7064 }, { "epoch": 0.8101135190918473, "grad_norm": 14.9375, "learning_rate": 9.172700937202545e-07, "loss": 1.6765, "step": 7065 }, { "epoch": 0.810228184841188, "grad_norm": 18.75, "learning_rate": 9.161983928885676e-07, "loss": 1.5035, "step": 7066 }, { "epoch": 0.8103428505905286, "grad_norm": 12.375, "learning_rate": 9.151272553394686e-07, "loss": 1.5261, "step": 7067 }, { "epoch": 0.8104575163398693, "grad_norm": 14.0, "learning_rate": 9.140566812207008e-07, "loss": 1.7351, "step": 7068 }, { "epoch": 0.81057218208921, "grad_norm": 20.5, "learning_rate": 9.129866706799262e-07, "loss": 1.4674, "step": 7069 }, { "epoch": 0.8106868478385506, "grad_norm": 12.625, "learning_rate": 9.119172238647377e-07, "loss": 1.9023, "step": 7070 }, { "epoch": 0.8108015135878913, "grad_norm": 13.5, "learning_rate": 9.108483409226404e-07, "loss": 2.0091, "step": 7071 }, { "epoch": 0.8109161793372319, "grad_norm": 12.125, "learning_rate": 9.097800220010689e-07, "loss": 1.5505, "step": 7072 }, { "epoch": 0.8110308450865726, "grad_norm": 14.9375, "learning_rate": 9.087122672473759e-07, "loss": 1.736, "step": 7073 }, { "epoch": 0.8111455108359134, "grad_norm": 14.25, "learning_rate": 9.076450768088391e-07, "loss": 1.6284, "step": 7074 }, { "epoch": 0.811260176585254, "grad_norm": 14.0625, "learning_rate": 9.065784508326564e-07, "loss": 1.7377, "step": 7075 }, { "epoch": 0.8113748423345947, "grad_norm": 12.75, "learning_rate": 9.055123894659501e-07, "loss": 1.4129, "step": 7076 }, { "epoch": 0.8114895080839353, "grad_norm": 13.8125, "learning_rate": 9.044468928557604e-07, "loss": 1.5851, "step": 7077 }, { "epoch": 0.811604173833276, "grad_norm": 14.9375, "learning_rate": 9.033819611490529e-07, "loss": 1.6323, "step": 7078 }, { "epoch": 0.8117188395826167, "grad_norm": 13.5625, "learning_rate": 9.023175944927159e-07, "loss": 1.5937, "step": 7079 }, { "epoch": 0.8118335053319573, "grad_norm": 16.5, "learning_rate": 9.012537930335574e-07, "loss": 1.5557, "step": 7080 }, { "epoch": 0.811948171081298, "grad_norm": 15.1875, "learning_rate": 9.001905569183089e-07, "loss": 1.6301, "step": 7081 }, { "epoch": 0.8120628368306387, "grad_norm": 13.4375, "learning_rate": 8.991278862936254e-07, "loss": 1.7295, "step": 7082 }, { "epoch": 0.8121775025799793, "grad_norm": 18.875, "learning_rate": 8.980657813060783e-07, "loss": 1.5884, "step": 7083 }, { "epoch": 0.81229216832932, "grad_norm": 18.375, "learning_rate": 8.970042421021663e-07, "loss": 2.0605, "step": 7084 }, { "epoch": 0.8124068340786607, "grad_norm": 27.25, "learning_rate": 8.959432688283104e-07, "loss": 1.7258, "step": 7085 }, { "epoch": 0.8125214998280014, "grad_norm": 20.125, "learning_rate": 8.948828616308497e-07, "loss": 1.7857, "step": 7086 }, { "epoch": 0.8126361655773421, "grad_norm": 13.4375, "learning_rate": 8.938230206560472e-07, "loss": 1.3048, "step": 7087 }, { "epoch": 0.8127508313266827, "grad_norm": 12.9375, "learning_rate": 8.927637460500893e-07, "loss": 1.5038, "step": 7088 }, { "epoch": 0.8128654970760234, "grad_norm": 11.625, "learning_rate": 8.917050379590791e-07, "loss": 1.6808, "step": 7089 }, { "epoch": 0.8129801628253641, "grad_norm": 17.125, "learning_rate": 8.906468965290493e-07, "loss": 2.0477, "step": 7090 }, { "epoch": 0.8130948285747047, "grad_norm": 15.125, "learning_rate": 8.895893219059498e-07, "loss": 1.5679, "step": 7091 }, { "epoch": 0.8132094943240454, "grad_norm": 15.5, "learning_rate": 8.885323142356506e-07, "loss": 1.545, "step": 7092 }, { "epoch": 0.813324160073386, "grad_norm": 12.5, "learning_rate": 8.874758736639466e-07, "loss": 1.5807, "step": 7093 }, { "epoch": 0.8134388258227268, "grad_norm": 26.0, "learning_rate": 8.864200003365542e-07, "loss": 2.2479, "step": 7094 }, { "epoch": 0.8135534915720675, "grad_norm": 14.4375, "learning_rate": 8.853646943991107e-07, "loss": 1.7816, "step": 7095 }, { "epoch": 0.8136681573214081, "grad_norm": 14.1875, "learning_rate": 8.843099559971757e-07, "loss": 1.7584, "step": 7096 }, { "epoch": 0.8137828230707488, "grad_norm": 33.0, "learning_rate": 8.832557852762313e-07, "loss": 2.0363, "step": 7097 }, { "epoch": 0.8138974888200894, "grad_norm": 15.8125, "learning_rate": 8.822021823816779e-07, "loss": 1.9313, "step": 7098 }, { "epoch": 0.8140121545694301, "grad_norm": 15.6875, "learning_rate": 8.811491474588397e-07, "loss": 1.9991, "step": 7099 }, { "epoch": 0.8141268203187708, "grad_norm": 14.875, "learning_rate": 8.800966806529664e-07, "loss": 1.8238, "step": 7100 }, { "epoch": 0.8142414860681114, "grad_norm": 14.5625, "learning_rate": 8.790447821092224e-07, "loss": 1.4552, "step": 7101 }, { "epoch": 0.8143561518174521, "grad_norm": 14.0, "learning_rate": 8.779934519726985e-07, "loss": 1.9118, "step": 7102 }, { "epoch": 0.8144708175667928, "grad_norm": 15.9375, "learning_rate": 8.76942690388406e-07, "loss": 2.0435, "step": 7103 }, { "epoch": 0.8145854833161335, "grad_norm": 13.25, "learning_rate": 8.758924975012745e-07, "loss": 1.9452, "step": 7104 }, { "epoch": 0.8147001490654742, "grad_norm": 14.4375, "learning_rate": 8.748428734561615e-07, "loss": 1.588, "step": 7105 }, { "epoch": 0.8148148148148148, "grad_norm": 13.125, "learning_rate": 8.737938183978428e-07, "loss": 1.9655, "step": 7106 }, { "epoch": 0.8149294805641555, "grad_norm": 13.75, "learning_rate": 8.727453324710134e-07, "loss": 1.6109, "step": 7107 }, { "epoch": 0.8150441463134962, "grad_norm": 14.375, "learning_rate": 8.716974158202923e-07, "loss": 1.8397, "step": 7108 }, { "epoch": 0.8151588120628368, "grad_norm": 11.375, "learning_rate": 8.706500685902198e-07, "loss": 1.6497, "step": 7109 }, { "epoch": 0.8152734778121775, "grad_norm": 12.625, "learning_rate": 8.696032909252583e-07, "loss": 1.6672, "step": 7110 }, { "epoch": 0.8153881435615181, "grad_norm": 13.5, "learning_rate": 8.685570829697898e-07, "loss": 1.5801, "step": 7111 }, { "epoch": 0.8155028093108588, "grad_norm": 29.5, "learning_rate": 8.67511444868121e-07, "loss": 1.9883, "step": 7112 }, { "epoch": 0.8156174750601995, "grad_norm": 11.8125, "learning_rate": 8.664663767644743e-07, "loss": 1.5797, "step": 7113 }, { "epoch": 0.8157321408095402, "grad_norm": 16.625, "learning_rate": 8.654218788029988e-07, "loss": 2.0261, "step": 7114 }, { "epoch": 0.8158468065588809, "grad_norm": 14.875, "learning_rate": 8.643779511277628e-07, "loss": 1.8689, "step": 7115 }, { "epoch": 0.8159614723082216, "grad_norm": 12.625, "learning_rate": 8.633345938827564e-07, "loss": 1.5458, "step": 7116 }, { "epoch": 0.8160761380575622, "grad_norm": 16.0, "learning_rate": 8.622918072118908e-07, "loss": 1.6036, "step": 7117 }, { "epoch": 0.8161908038069029, "grad_norm": 15.625, "learning_rate": 8.612495912589996e-07, "loss": 1.7539, "step": 7118 }, { "epoch": 0.8163054695562435, "grad_norm": 15.6875, "learning_rate": 8.602079461678331e-07, "loss": 1.6229, "step": 7119 }, { "epoch": 0.8164201353055842, "grad_norm": 14.0625, "learning_rate": 8.591668720820717e-07, "loss": 1.5752, "step": 7120 }, { "epoch": 0.8165348010549249, "grad_norm": 13.5625, "learning_rate": 8.581263691453074e-07, "loss": 1.5938, "step": 7121 }, { "epoch": 0.8166494668042655, "grad_norm": 13.125, "learning_rate": 8.57086437501059e-07, "loss": 2.0305, "step": 7122 }, { "epoch": 0.8167641325536062, "grad_norm": 15.125, "learning_rate": 8.560470772927653e-07, "loss": 1.374, "step": 7123 }, { "epoch": 0.816878798302947, "grad_norm": 18.25, "learning_rate": 8.550082886637861e-07, "loss": 1.8913, "step": 7124 }, { "epoch": 0.8169934640522876, "grad_norm": 12.4375, "learning_rate": 8.539700717574035e-07, "loss": 2.1927, "step": 7125 }, { "epoch": 0.8171081298016283, "grad_norm": 28.75, "learning_rate": 8.529324267168199e-07, "loss": 1.9049, "step": 7126 }, { "epoch": 0.8172227955509689, "grad_norm": 12.6875, "learning_rate": 8.518953536851565e-07, "loss": 1.44, "step": 7127 }, { "epoch": 0.8173374613003096, "grad_norm": 31.0, "learning_rate": 8.50858852805459e-07, "loss": 1.7231, "step": 7128 }, { "epoch": 0.8174521270496503, "grad_norm": 11.75, "learning_rate": 8.498229242206929e-07, "loss": 1.5938, "step": 7129 }, { "epoch": 0.8175667927989909, "grad_norm": 13.8125, "learning_rate": 8.487875680737445e-07, "loss": 1.7995, "step": 7130 }, { "epoch": 0.8176814585483316, "grad_norm": 16.25, "learning_rate": 8.477527845074224e-07, "loss": 1.7059, "step": 7131 }, { "epoch": 0.8177961242976722, "grad_norm": 11.3125, "learning_rate": 8.467185736644556e-07, "loss": 1.5211, "step": 7132 }, { "epoch": 0.817910790047013, "grad_norm": 15.8125, "learning_rate": 8.456849356874913e-07, "loss": 1.8342, "step": 7133 }, { "epoch": 0.8180254557963537, "grad_norm": 13.0625, "learning_rate": 8.446518707191009e-07, "loss": 1.8667, "step": 7134 }, { "epoch": 0.8181401215456943, "grad_norm": 11.8125, "learning_rate": 8.436193789017789e-07, "loss": 1.6816, "step": 7135 }, { "epoch": 0.818254787295035, "grad_norm": 17.25, "learning_rate": 8.425874603779349e-07, "loss": 2.0711, "step": 7136 }, { "epoch": 0.8183694530443757, "grad_norm": 13.125, "learning_rate": 8.415561152899027e-07, "loss": 1.8117, "step": 7137 }, { "epoch": 0.8184841187937163, "grad_norm": 21.75, "learning_rate": 8.405253437799388e-07, "loss": 1.9746, "step": 7138 }, { "epoch": 0.818598784543057, "grad_norm": 13.1875, "learning_rate": 8.394951459902145e-07, "loss": 1.7278, "step": 7139 }, { "epoch": 0.8187134502923976, "grad_norm": 13.875, "learning_rate": 8.384655220628296e-07, "loss": 1.8142, "step": 7140 }, { "epoch": 0.8188281160417383, "grad_norm": 12.6875, "learning_rate": 8.374364721398015e-07, "loss": 1.6842, "step": 7141 }, { "epoch": 0.818942781791079, "grad_norm": 15.875, "learning_rate": 8.364079963630645e-07, "loss": 1.9498, "step": 7142 }, { "epoch": 0.8190574475404196, "grad_norm": 15.1875, "learning_rate": 8.353800948744794e-07, "loss": 1.7802, "step": 7143 }, { "epoch": 0.8191721132897604, "grad_norm": 13.3125, "learning_rate": 8.343527678158253e-07, "loss": 1.9089, "step": 7144 }, { "epoch": 0.819286779039101, "grad_norm": 10.1875, "learning_rate": 8.333260153288026e-07, "loss": 1.7856, "step": 7145 }, { "epoch": 0.8194014447884417, "grad_norm": 18.125, "learning_rate": 8.322998375550318e-07, "loss": 1.5548, "step": 7146 }, { "epoch": 0.8195161105377824, "grad_norm": 13.5, "learning_rate": 8.312742346360558e-07, "loss": 1.8384, "step": 7147 }, { "epoch": 0.819630776287123, "grad_norm": 18.5, "learning_rate": 8.302492067133349e-07, "loss": 1.8642, "step": 7148 }, { "epoch": 0.8197454420364637, "grad_norm": 12.625, "learning_rate": 8.292247539282522e-07, "loss": 1.6184, "step": 7149 }, { "epoch": 0.8198601077858044, "grad_norm": 19.875, "learning_rate": 8.28200876422115e-07, "loss": 2.2999, "step": 7150 }, { "epoch": 0.819974773535145, "grad_norm": 14.6875, "learning_rate": 8.271775743361444e-07, "loss": 1.6463, "step": 7151 }, { "epoch": 0.8200894392844857, "grad_norm": 16.125, "learning_rate": 8.261548478114855e-07, "loss": 1.7137, "step": 7152 }, { "epoch": 0.8202041050338263, "grad_norm": 13.8125, "learning_rate": 8.251326969892059e-07, "loss": 1.7221, "step": 7153 }, { "epoch": 0.8203187707831671, "grad_norm": 14.5, "learning_rate": 8.241111220102881e-07, "loss": 1.8592, "step": 7154 }, { "epoch": 0.8204334365325078, "grad_norm": 18.0, "learning_rate": 8.23090123015643e-07, "loss": 1.7728, "step": 7155 }, { "epoch": 0.8205481022818484, "grad_norm": 15.1875, "learning_rate": 8.22069700146097e-07, "loss": 1.6116, "step": 7156 }, { "epoch": 0.8206627680311891, "grad_norm": 13.0, "learning_rate": 8.210498535423961e-07, "loss": 1.6156, "step": 7157 }, { "epoch": 0.8207774337805298, "grad_norm": 15.3125, "learning_rate": 8.200305833452099e-07, "loss": 1.9553, "step": 7158 }, { "epoch": 0.8208920995298704, "grad_norm": 17.625, "learning_rate": 8.190118896951272e-07, "loss": 2.0261, "step": 7159 }, { "epoch": 0.8210067652792111, "grad_norm": 14.375, "learning_rate": 8.179937727326576e-07, "loss": 1.7321, "step": 7160 }, { "epoch": 0.8211214310285517, "grad_norm": 14.4375, "learning_rate": 8.169762325982311e-07, "loss": 1.6824, "step": 7161 }, { "epoch": 0.8212360967778924, "grad_norm": 13.5, "learning_rate": 8.159592694321983e-07, "loss": 1.6932, "step": 7162 }, { "epoch": 0.8213507625272332, "grad_norm": 15.1875, "learning_rate": 8.14942883374828e-07, "loss": 1.6621, "step": 7163 }, { "epoch": 0.8214654282765738, "grad_norm": 18.75, "learning_rate": 8.13927074566313e-07, "loss": 2.0129, "step": 7164 }, { "epoch": 0.8215800940259145, "grad_norm": 14.4375, "learning_rate": 8.129118431467637e-07, "loss": 1.8291, "step": 7165 }, { "epoch": 0.8216947597752551, "grad_norm": 16.0, "learning_rate": 8.118971892562127e-07, "loss": 2.0787, "step": 7166 }, { "epoch": 0.8218094255245958, "grad_norm": 13.8125, "learning_rate": 8.108831130346118e-07, "loss": 1.5978, "step": 7167 }, { "epoch": 0.8219240912739365, "grad_norm": 21.25, "learning_rate": 8.098696146218354e-07, "loss": 1.6924, "step": 7168 }, { "epoch": 0.8220387570232771, "grad_norm": 13.4375, "learning_rate": 8.088566941576714e-07, "loss": 1.6043, "step": 7169 }, { "epoch": 0.8221534227726178, "grad_norm": 13.0625, "learning_rate": 8.078443517818385e-07, "loss": 1.7064, "step": 7170 }, { "epoch": 0.8222680885219585, "grad_norm": 16.375, "learning_rate": 8.068325876339667e-07, "loss": 1.6558, "step": 7171 }, { "epoch": 0.8223827542712991, "grad_norm": 14.8125, "learning_rate": 8.058214018536104e-07, "loss": 1.7375, "step": 7172 }, { "epoch": 0.8224974200206399, "grad_norm": 13.5, "learning_rate": 8.048107945802431e-07, "loss": 1.5346, "step": 7173 }, { "epoch": 0.8226120857699805, "grad_norm": 13.8125, "learning_rate": 8.038007659532588e-07, "loss": 1.7075, "step": 7174 }, { "epoch": 0.8227267515193212, "grad_norm": 14.5, "learning_rate": 8.027913161119727e-07, "loss": 1.4888, "step": 7175 }, { "epoch": 0.8228414172686619, "grad_norm": 17.125, "learning_rate": 8.017824451956191e-07, "loss": 1.7709, "step": 7176 }, { "epoch": 0.8229560830180025, "grad_norm": 15.1875, "learning_rate": 8.00774153343351e-07, "loss": 1.7362, "step": 7177 }, { "epoch": 0.8230707487673432, "grad_norm": 14.6875, "learning_rate": 7.997664406942441e-07, "loss": 1.9208, "step": 7178 }, { "epoch": 0.8231854145166838, "grad_norm": 11.125, "learning_rate": 7.987593073872923e-07, "loss": 1.6343, "step": 7179 }, { "epoch": 0.8233000802660245, "grad_norm": 16.375, "learning_rate": 7.977527535614116e-07, "loss": 2.0256, "step": 7180 }, { "epoch": 0.8234147460153652, "grad_norm": 16.25, "learning_rate": 7.967467793554362e-07, "loss": 1.8969, "step": 7181 }, { "epoch": 0.8235294117647058, "grad_norm": 12.125, "learning_rate": 7.95741384908123e-07, "loss": 1.5169, "step": 7182 }, { "epoch": 0.8236440775140466, "grad_norm": 30.125, "learning_rate": 7.947365703581433e-07, "loss": 1.7077, "step": 7183 }, { "epoch": 0.8237587432633873, "grad_norm": 11.0, "learning_rate": 7.937323358440935e-07, "loss": 1.4216, "step": 7184 }, { "epoch": 0.8238734090127279, "grad_norm": 13.125, "learning_rate": 7.927286815044916e-07, "loss": 1.5046, "step": 7185 }, { "epoch": 0.8239880747620686, "grad_norm": 13.75, "learning_rate": 7.91725607477769e-07, "loss": 1.9811, "step": 7186 }, { "epoch": 0.8241027405114092, "grad_norm": 16.5, "learning_rate": 7.907231139022814e-07, "loss": 1.7826, "step": 7187 }, { "epoch": 0.8242174062607499, "grad_norm": 12.6875, "learning_rate": 7.897212009163058e-07, "loss": 2.033, "step": 7188 }, { "epoch": 0.8243320720100906, "grad_norm": 11.375, "learning_rate": 7.887198686580327e-07, "loss": 1.6348, "step": 7189 }, { "epoch": 0.8244467377594312, "grad_norm": 23.0, "learning_rate": 7.877191172655807e-07, "loss": 1.8119, "step": 7190 }, { "epoch": 0.8245614035087719, "grad_norm": 15.9375, "learning_rate": 7.867189468769843e-07, "loss": 2.1894, "step": 7191 }, { "epoch": 0.8246760692581127, "grad_norm": 14.0, "learning_rate": 7.857193576301952e-07, "loss": 1.9085, "step": 7192 }, { "epoch": 0.8247907350074533, "grad_norm": 12.5, "learning_rate": 7.847203496630895e-07, "loss": 1.7564, "step": 7193 }, { "epoch": 0.824905400756794, "grad_norm": 14.9375, "learning_rate": 7.837219231134612e-07, "loss": 1.4818, "step": 7194 }, { "epoch": 0.8250200665061346, "grad_norm": 16.375, "learning_rate": 7.827240781190237e-07, "loss": 2.02, "step": 7195 }, { "epoch": 0.8251347322554753, "grad_norm": 14.9375, "learning_rate": 7.817268148174112e-07, "loss": 1.5926, "step": 7196 }, { "epoch": 0.825249398004816, "grad_norm": 14.3125, "learning_rate": 7.807301333461781e-07, "loss": 1.5236, "step": 7197 }, { "epoch": 0.8253640637541566, "grad_norm": 12.3125, "learning_rate": 7.797340338427956e-07, "loss": 1.8649, "step": 7198 }, { "epoch": 0.8254787295034973, "grad_norm": 17.5, "learning_rate": 7.787385164446559e-07, "loss": 1.682, "step": 7199 }, { "epoch": 0.8255933952528379, "grad_norm": 13.75, "learning_rate": 7.77743581289076e-07, "loss": 1.439, "step": 7200 }, { "epoch": 0.8257080610021786, "grad_norm": 27.25, "learning_rate": 7.767492285132844e-07, "loss": 1.8102, "step": 7201 }, { "epoch": 0.8258227267515194, "grad_norm": 18.0, "learning_rate": 7.757554582544341e-07, "loss": 1.8226, "step": 7202 }, { "epoch": 0.82593739250086, "grad_norm": 18.25, "learning_rate": 7.747622706495983e-07, "loss": 2.1999, "step": 7203 }, { "epoch": 0.8260520582502007, "grad_norm": 15.0625, "learning_rate": 7.737696658357641e-07, "loss": 1.8489, "step": 7204 }, { "epoch": 0.8261667239995414, "grad_norm": 14.75, "learning_rate": 7.727776439498464e-07, "loss": 2.0442, "step": 7205 }, { "epoch": 0.826281389748882, "grad_norm": 14.5625, "learning_rate": 7.717862051286762e-07, "loss": 1.576, "step": 7206 }, { "epoch": 0.8263960554982227, "grad_norm": 17.25, "learning_rate": 7.707953495090004e-07, "loss": 1.9192, "step": 7207 }, { "epoch": 0.8265107212475633, "grad_norm": 14.6875, "learning_rate": 7.698050772274901e-07, "loss": 2.1497, "step": 7208 }, { "epoch": 0.826625386996904, "grad_norm": 14.125, "learning_rate": 7.68815388420735e-07, "loss": 1.7036, "step": 7209 }, { "epoch": 0.8267400527462447, "grad_norm": 12.8125, "learning_rate": 7.678262832252431e-07, "loss": 1.7355, "step": 7210 }, { "epoch": 0.8268547184955853, "grad_norm": 14.4375, "learning_rate": 7.668377617774425e-07, "loss": 1.6302, "step": 7211 }, { "epoch": 0.826969384244926, "grad_norm": 15.6875, "learning_rate": 7.658498242136825e-07, "loss": 1.4702, "step": 7212 }, { "epoch": 0.8270840499942668, "grad_norm": 13.9375, "learning_rate": 7.648624706702285e-07, "loss": 1.8732, "step": 7213 }, { "epoch": 0.8271987157436074, "grad_norm": 16.125, "learning_rate": 7.63875701283267e-07, "loss": 2.0645, "step": 7214 }, { "epoch": 0.8273133814929481, "grad_norm": 15.6875, "learning_rate": 7.62889516188905e-07, "loss": 1.598, "step": 7215 }, { "epoch": 0.8274280472422887, "grad_norm": 15.9375, "learning_rate": 7.619039155231672e-07, "loss": 2.045, "step": 7216 }, { "epoch": 0.8275427129916294, "grad_norm": 13.6875, "learning_rate": 7.609188994219991e-07, "loss": 1.8536, "step": 7217 }, { "epoch": 0.8276573787409701, "grad_norm": 20.0, "learning_rate": 7.599344680212656e-07, "loss": 1.4699, "step": 7218 }, { "epoch": 0.8277720444903107, "grad_norm": 13.4375, "learning_rate": 7.589506214567466e-07, "loss": 1.6872, "step": 7219 }, { "epoch": 0.8278867102396514, "grad_norm": 14.8125, "learning_rate": 7.579673598641502e-07, "loss": 1.6466, "step": 7220 }, { "epoch": 0.828001375988992, "grad_norm": 14.0625, "learning_rate": 7.56984683379095e-07, "loss": 1.512, "step": 7221 }, { "epoch": 0.8281160417383328, "grad_norm": 11.3125, "learning_rate": 7.560025921371233e-07, "loss": 1.6689, "step": 7222 }, { "epoch": 0.8282307074876735, "grad_norm": 14.5, "learning_rate": 7.550210862736962e-07, "loss": 1.3976, "step": 7223 }, { "epoch": 0.8283453732370141, "grad_norm": 15.3125, "learning_rate": 7.540401659241936e-07, "loss": 1.612, "step": 7224 }, { "epoch": 0.8284600389863548, "grad_norm": 10.875, "learning_rate": 7.530598312239146e-07, "loss": 1.7303, "step": 7225 }, { "epoch": 0.8285747047356955, "grad_norm": 12.3125, "learning_rate": 7.520800823080793e-07, "loss": 1.8077, "step": 7226 }, { "epoch": 0.8286893704850361, "grad_norm": 13.75, "learning_rate": 7.511009193118224e-07, "loss": 1.9501, "step": 7227 }, { "epoch": 0.8288040362343768, "grad_norm": 14.5, "learning_rate": 7.501223423702025e-07, "loss": 2.0188, "step": 7228 }, { "epoch": 0.8289187019837174, "grad_norm": 13.6875, "learning_rate": 7.491443516181951e-07, "loss": 1.9954, "step": 7229 }, { "epoch": 0.8290333677330581, "grad_norm": 12.4375, "learning_rate": 7.481669471906961e-07, "loss": 1.6891, "step": 7230 }, { "epoch": 0.8291480334823988, "grad_norm": 17.25, "learning_rate": 7.471901292225198e-07, "loss": 2.2625, "step": 7231 }, { "epoch": 0.8292626992317395, "grad_norm": 11.75, "learning_rate": 7.462138978483996e-07, "loss": 1.4836, "step": 7232 }, { "epoch": 0.8293773649810802, "grad_norm": 16.0, "learning_rate": 7.452382532029867e-07, "loss": 1.5419, "step": 7233 }, { "epoch": 0.8294920307304208, "grad_norm": 14.5625, "learning_rate": 7.442631954208529e-07, "loss": 2.0365, "step": 7234 }, { "epoch": 0.8296066964797615, "grad_norm": 20.25, "learning_rate": 7.432887246364912e-07, "loss": 2.2374, "step": 7235 }, { "epoch": 0.8297213622291022, "grad_norm": 14.5625, "learning_rate": 7.423148409843089e-07, "loss": 1.8235, "step": 7236 }, { "epoch": 0.8298360279784428, "grad_norm": 14.8125, "learning_rate": 7.413415445986355e-07, "loss": 1.8593, "step": 7237 }, { "epoch": 0.8299506937277835, "grad_norm": 11.0625, "learning_rate": 7.403688356137195e-07, "loss": 1.2583, "step": 7238 }, { "epoch": 0.8300653594771242, "grad_norm": 17.125, "learning_rate": 7.393967141637248e-07, "loss": 2.3957, "step": 7239 }, { "epoch": 0.8301800252264648, "grad_norm": 15.0, "learning_rate": 7.384251803827397e-07, "loss": 1.6942, "step": 7240 }, { "epoch": 0.8302946909758055, "grad_norm": 15.9375, "learning_rate": 7.374542344047691e-07, "loss": 1.9951, "step": 7241 }, { "epoch": 0.8304093567251462, "grad_norm": 15.8125, "learning_rate": 7.364838763637344e-07, "loss": 1.6103, "step": 7242 }, { "epoch": 0.8305240224744869, "grad_norm": 13.5625, "learning_rate": 7.35514106393479e-07, "loss": 1.7286, "step": 7243 }, { "epoch": 0.8306386882238276, "grad_norm": 12.875, "learning_rate": 7.345449246277636e-07, "loss": 1.9169, "step": 7244 }, { "epoch": 0.8307533539731682, "grad_norm": 14.75, "learning_rate": 7.335763312002691e-07, "loss": 1.6516, "step": 7245 }, { "epoch": 0.8308680197225089, "grad_norm": 16.625, "learning_rate": 7.326083262445943e-07, "loss": 1.8773, "step": 7246 }, { "epoch": 0.8309826854718496, "grad_norm": 18.0, "learning_rate": 7.316409098942584e-07, "loss": 1.5927, "step": 7247 }, { "epoch": 0.8310973512211902, "grad_norm": 14.6875, "learning_rate": 7.306740822826957e-07, "loss": 1.6266, "step": 7248 }, { "epoch": 0.8312120169705309, "grad_norm": 14.5625, "learning_rate": 7.297078435432608e-07, "loss": 1.6029, "step": 7249 }, { "epoch": 0.8313266827198715, "grad_norm": 28.375, "learning_rate": 7.287421938092326e-07, "loss": 1.9399, "step": 7250 }, { "epoch": 0.8314413484692122, "grad_norm": 15.5625, "learning_rate": 7.277771332137995e-07, "loss": 2.2236, "step": 7251 }, { "epoch": 0.831556014218553, "grad_norm": 12.5, "learning_rate": 7.268126618900751e-07, "loss": 1.6568, "step": 7252 }, { "epoch": 0.8316706799678936, "grad_norm": 14.5625, "learning_rate": 7.258487799710911e-07, "loss": 1.9185, "step": 7253 }, { "epoch": 0.8317853457172343, "grad_norm": 20.875, "learning_rate": 7.248854875897926e-07, "loss": 2.1029, "step": 7254 }, { "epoch": 0.8319000114665749, "grad_norm": 11.75, "learning_rate": 7.239227848790509e-07, "loss": 1.9127, "step": 7255 }, { "epoch": 0.8320146772159156, "grad_norm": 14.4375, "learning_rate": 7.22960671971653e-07, "loss": 1.9016, "step": 7256 }, { "epoch": 0.8321293429652563, "grad_norm": 14.3125, "learning_rate": 7.219991490003015e-07, "loss": 1.9317, "step": 7257 }, { "epoch": 0.8322440087145969, "grad_norm": 16.125, "learning_rate": 7.21038216097621e-07, "loss": 1.4832, "step": 7258 }, { "epoch": 0.8323586744639376, "grad_norm": 13.5625, "learning_rate": 7.200778733961544e-07, "loss": 1.6713, "step": 7259 }, { "epoch": 0.8324733402132783, "grad_norm": 14.375, "learning_rate": 7.191181210283627e-07, "loss": 1.7568, "step": 7260 }, { "epoch": 0.832588005962619, "grad_norm": 16.25, "learning_rate": 7.181589591266247e-07, "loss": 1.9766, "step": 7261 }, { "epoch": 0.8327026717119597, "grad_norm": 13.25, "learning_rate": 7.172003878232403e-07, "loss": 1.8403, "step": 7262 }, { "epoch": 0.8328173374613003, "grad_norm": 13.6875, "learning_rate": 7.162424072504237e-07, "loss": 1.8173, "step": 7263 }, { "epoch": 0.832932003210641, "grad_norm": 16.625, "learning_rate": 7.152850175403109e-07, "loss": 1.6996, "step": 7264 }, { "epoch": 0.8330466689599817, "grad_norm": 14.75, "learning_rate": 7.143282188249562e-07, "loss": 2.0402, "step": 7265 }, { "epoch": 0.8331613347093223, "grad_norm": 11.75, "learning_rate": 7.133720112363312e-07, "loss": 1.6041, "step": 7266 }, { "epoch": 0.833276000458663, "grad_norm": 13.5625, "learning_rate": 7.124163949063268e-07, "loss": 1.5587, "step": 7267 }, { "epoch": 0.8333906662080036, "grad_norm": 13.4375, "learning_rate": 7.114613699667527e-07, "loss": 1.638, "step": 7268 }, { "epoch": 0.8335053319573443, "grad_norm": 15.8125, "learning_rate": 7.105069365493339e-07, "loss": 1.4645, "step": 7269 }, { "epoch": 0.833619997706685, "grad_norm": 10.5, "learning_rate": 7.095530947857199e-07, "loss": 1.5437, "step": 7270 }, { "epoch": 0.8337346634560256, "grad_norm": 18.0, "learning_rate": 7.085998448074721e-07, "loss": 2.0873, "step": 7271 }, { "epoch": 0.8338493292053664, "grad_norm": 14.0, "learning_rate": 7.076471867460743e-07, "loss": 2.0446, "step": 7272 }, { "epoch": 0.8339639949547071, "grad_norm": 12.9375, "learning_rate": 7.066951207329276e-07, "loss": 1.7966, "step": 7273 }, { "epoch": 0.8340786607040477, "grad_norm": 12.0625, "learning_rate": 7.057436468993506e-07, "loss": 1.5306, "step": 7274 }, { "epoch": 0.8341933264533884, "grad_norm": 15.75, "learning_rate": 7.047927653765818e-07, "loss": 2.0756, "step": 7275 }, { "epoch": 0.834307992202729, "grad_norm": 11.9375, "learning_rate": 7.038424762957779e-07, "loss": 1.5534, "step": 7276 }, { "epoch": 0.8344226579520697, "grad_norm": 31.625, "learning_rate": 7.028927797880114e-07, "loss": 2.1406, "step": 7277 }, { "epoch": 0.8345373237014104, "grad_norm": 13.6875, "learning_rate": 7.019436759842757e-07, "loss": 1.8293, "step": 7278 }, { "epoch": 0.834651989450751, "grad_norm": 13.0625, "learning_rate": 7.009951650154812e-07, "loss": 1.7976, "step": 7279 }, { "epoch": 0.8347666552000917, "grad_norm": 14.5, "learning_rate": 7.000472470124569e-07, "loss": 1.361, "step": 7280 }, { "epoch": 0.8348813209494325, "grad_norm": 15.3125, "learning_rate": 6.990999221059508e-07, "loss": 1.7817, "step": 7281 }, { "epoch": 0.8349959866987731, "grad_norm": 19.75, "learning_rate": 6.981531904266287e-07, "loss": 1.4576, "step": 7282 }, { "epoch": 0.8351106524481138, "grad_norm": 11.875, "learning_rate": 6.972070521050722e-07, "loss": 1.5839, "step": 7283 }, { "epoch": 0.8352253181974544, "grad_norm": 18.625, "learning_rate": 6.962615072717833e-07, "loss": 2.0182, "step": 7284 }, { "epoch": 0.8353399839467951, "grad_norm": 18.0, "learning_rate": 6.953165560571851e-07, "loss": 1.5359, "step": 7285 }, { "epoch": 0.8354546496961358, "grad_norm": 14.0625, "learning_rate": 6.943721985916113e-07, "loss": 1.7961, "step": 7286 }, { "epoch": 0.8355693154454764, "grad_norm": 14.5, "learning_rate": 6.934284350053211e-07, "loss": 2.0046, "step": 7287 }, { "epoch": 0.8356839811948171, "grad_norm": 17.625, "learning_rate": 6.924852654284881e-07, "loss": 1.607, "step": 7288 }, { "epoch": 0.8357986469441577, "grad_norm": 14.5625, "learning_rate": 6.915426899912015e-07, "loss": 1.6989, "step": 7289 }, { "epoch": 0.8359133126934984, "grad_norm": 17.75, "learning_rate": 6.906007088234756e-07, "loss": 1.7354, "step": 7290 }, { "epoch": 0.8360279784428392, "grad_norm": 11.125, "learning_rate": 6.896593220552384e-07, "loss": 1.4999, "step": 7291 }, { "epoch": 0.8361426441921798, "grad_norm": 13.6875, "learning_rate": 6.88718529816334e-07, "loss": 1.5688, "step": 7292 }, { "epoch": 0.8362573099415205, "grad_norm": 17.0, "learning_rate": 6.877783322365283e-07, "loss": 1.5719, "step": 7293 }, { "epoch": 0.8363719756908612, "grad_norm": 11.875, "learning_rate": 6.868387294455026e-07, "loss": 1.4557, "step": 7294 }, { "epoch": 0.8364866414402018, "grad_norm": 12.3125, "learning_rate": 6.85899721572858e-07, "loss": 1.666, "step": 7295 }, { "epoch": 0.8366013071895425, "grad_norm": 14.25, "learning_rate": 6.849613087481128e-07, "loss": 2.1492, "step": 7296 }, { "epoch": 0.8367159729388831, "grad_norm": 11.25, "learning_rate": 6.840234911007043e-07, "loss": 1.5772, "step": 7297 }, { "epoch": 0.8368306386882238, "grad_norm": 16.25, "learning_rate": 6.830862687599837e-07, "loss": 1.5811, "step": 7298 }, { "epoch": 0.8369453044375645, "grad_norm": 12.875, "learning_rate": 6.821496418552231e-07, "loss": 1.4483, "step": 7299 }, { "epoch": 0.8370599701869051, "grad_norm": 27.625, "learning_rate": 6.812136105156164e-07, "loss": 1.7682, "step": 7300 }, { "epoch": 0.8371746359362459, "grad_norm": 13.9375, "learning_rate": 6.802781748702675e-07, "loss": 1.9641, "step": 7301 }, { "epoch": 0.8372893016855865, "grad_norm": 14.5, "learning_rate": 6.793433350482026e-07, "loss": 1.5423, "step": 7302 }, { "epoch": 0.8374039674349272, "grad_norm": 38.75, "learning_rate": 6.784090911783664e-07, "loss": 2.2132, "step": 7303 }, { "epoch": 0.8375186331842679, "grad_norm": 13.125, "learning_rate": 6.774754433896174e-07, "loss": 1.5914, "step": 7304 }, { "epoch": 0.8376332989336085, "grad_norm": 14.3125, "learning_rate": 6.765423918107372e-07, "loss": 1.9624, "step": 7305 }, { "epoch": 0.8377479646829492, "grad_norm": 14.0625, "learning_rate": 6.756099365704221e-07, "loss": 1.5953, "step": 7306 }, { "epoch": 0.8378626304322899, "grad_norm": 14.0625, "learning_rate": 6.74678077797285e-07, "loss": 1.6788, "step": 7307 }, { "epoch": 0.8379772961816305, "grad_norm": 26.625, "learning_rate": 6.73746815619859e-07, "loss": 2.2351, "step": 7308 }, { "epoch": 0.8380919619309712, "grad_norm": 14.5625, "learning_rate": 6.728161501665942e-07, "loss": 1.8147, "step": 7309 }, { "epoch": 0.8382066276803118, "grad_norm": 15.25, "learning_rate": 6.718860815658573e-07, "loss": 1.7994, "step": 7310 }, { "epoch": 0.8383212934296526, "grad_norm": 15.9375, "learning_rate": 6.709566099459342e-07, "loss": 1.9973, "step": 7311 }, { "epoch": 0.8384359591789933, "grad_norm": 22.875, "learning_rate": 6.700277354350288e-07, "loss": 1.6717, "step": 7312 }, { "epoch": 0.8385506249283339, "grad_norm": 15.625, "learning_rate": 6.690994581612597e-07, "loss": 1.4475, "step": 7313 }, { "epoch": 0.8386652906776746, "grad_norm": 15.375, "learning_rate": 6.681717782526659e-07, "loss": 1.7795, "step": 7314 }, { "epoch": 0.8387799564270153, "grad_norm": 22.125, "learning_rate": 6.672446958372026e-07, "loss": 1.8301, "step": 7315 }, { "epoch": 0.8388946221763559, "grad_norm": 13.3125, "learning_rate": 6.663182110427441e-07, "loss": 1.804, "step": 7316 }, { "epoch": 0.8390092879256966, "grad_norm": 18.25, "learning_rate": 6.653923239970811e-07, "loss": 1.9036, "step": 7317 }, { "epoch": 0.8391239536750372, "grad_norm": 16.5, "learning_rate": 6.644670348279225e-07, "loss": 2.0891, "step": 7318 }, { "epoch": 0.8392386194243779, "grad_norm": 14.8125, "learning_rate": 6.635423436628918e-07, "loss": 1.5274, "step": 7319 }, { "epoch": 0.8393532851737187, "grad_norm": 15.75, "learning_rate": 6.626182506295364e-07, "loss": 2.1341, "step": 7320 }, { "epoch": 0.8394679509230593, "grad_norm": 35.75, "learning_rate": 6.616947558553138e-07, "loss": 2.1382, "step": 7321 }, { "epoch": 0.8395826166724, "grad_norm": 19.875, "learning_rate": 6.607718594676049e-07, "loss": 1.8187, "step": 7322 }, { "epoch": 0.8396972824217406, "grad_norm": 17.5, "learning_rate": 6.598495615937039e-07, "loss": 1.9434, "step": 7323 }, { "epoch": 0.8398119481710813, "grad_norm": 16.375, "learning_rate": 6.589278623608259e-07, "loss": 1.5747, "step": 7324 }, { "epoch": 0.839926613920422, "grad_norm": 17.25, "learning_rate": 6.580067618961006e-07, "loss": 2.0674, "step": 7325 }, { "epoch": 0.8400412796697626, "grad_norm": 13.9375, "learning_rate": 6.570862603265771e-07, "loss": 1.7215, "step": 7326 }, { "epoch": 0.8401559454191033, "grad_norm": 19.5, "learning_rate": 6.561663577792199e-07, "loss": 1.6575, "step": 7327 }, { "epoch": 0.840270611168444, "grad_norm": 16.75, "learning_rate": 6.55247054380913e-07, "loss": 1.7248, "step": 7328 }, { "epoch": 0.8403852769177846, "grad_norm": 14.5625, "learning_rate": 6.543283502584558e-07, "loss": 1.9068, "step": 7329 }, { "epoch": 0.8404999426671254, "grad_norm": 16.75, "learning_rate": 6.534102455385671e-07, "loss": 1.6799, "step": 7330 }, { "epoch": 0.840614608416466, "grad_norm": 13.4375, "learning_rate": 6.524927403478814e-07, "loss": 1.7128, "step": 7331 }, { "epoch": 0.8407292741658067, "grad_norm": 16.875, "learning_rate": 6.51575834812952e-07, "loss": 1.6309, "step": 7332 }, { "epoch": 0.8408439399151474, "grad_norm": 14.25, "learning_rate": 6.506595290602468e-07, "loss": 1.9735, "step": 7333 }, { "epoch": 0.840958605664488, "grad_norm": 16.875, "learning_rate": 6.497438232161524e-07, "loss": 1.3006, "step": 7334 }, { "epoch": 0.8410732714138287, "grad_norm": 14.9375, "learning_rate": 6.488287174069763e-07, "loss": 1.6492, "step": 7335 }, { "epoch": 0.8411879371631693, "grad_norm": 14.8125, "learning_rate": 6.479142117589371e-07, "loss": 1.8998, "step": 7336 }, { "epoch": 0.84130260291251, "grad_norm": 12.25, "learning_rate": 6.470003063981733e-07, "loss": 1.6028, "step": 7337 }, { "epoch": 0.8414172686618507, "grad_norm": 12.4375, "learning_rate": 6.460870014507431e-07, "loss": 1.6003, "step": 7338 }, { "epoch": 0.8415319344111913, "grad_norm": 12.0625, "learning_rate": 6.451742970426162e-07, "loss": 1.6327, "step": 7339 }, { "epoch": 0.841646600160532, "grad_norm": 18.75, "learning_rate": 6.442621932996851e-07, "loss": 1.8415, "step": 7340 }, { "epoch": 0.8417612659098728, "grad_norm": 11.8125, "learning_rate": 6.433506903477571e-07, "loss": 1.8096, "step": 7341 }, { "epoch": 0.8418759316592134, "grad_norm": 12.9375, "learning_rate": 6.424397883125555e-07, "loss": 1.7092, "step": 7342 }, { "epoch": 0.8419905974085541, "grad_norm": 16.75, "learning_rate": 6.415294873197226e-07, "loss": 1.5594, "step": 7343 }, { "epoch": 0.8421052631578947, "grad_norm": 11.9375, "learning_rate": 6.406197874948167e-07, "loss": 1.6918, "step": 7344 }, { "epoch": 0.8422199289072354, "grad_norm": 14.4375, "learning_rate": 6.397106889633136e-07, "loss": 1.8921, "step": 7345 }, { "epoch": 0.8423345946565761, "grad_norm": 14.0625, "learning_rate": 6.388021918506065e-07, "loss": 1.6538, "step": 7346 }, { "epoch": 0.8424492604059167, "grad_norm": 13.0625, "learning_rate": 6.378942962820062e-07, "loss": 1.7139, "step": 7347 }, { "epoch": 0.8425639261552574, "grad_norm": 14.5, "learning_rate": 6.369870023827374e-07, "loss": 1.6786, "step": 7348 }, { "epoch": 0.8426785919045982, "grad_norm": 14.8125, "learning_rate": 6.360803102779439e-07, "loss": 1.807, "step": 7349 }, { "epoch": 0.8427932576539388, "grad_norm": 13.25, "learning_rate": 6.351742200926897e-07, "loss": 1.5511, "step": 7350 }, { "epoch": 0.8429079234032795, "grad_norm": 14.4375, "learning_rate": 6.342687319519497e-07, "loss": 2.0791, "step": 7351 }, { "epoch": 0.8430225891526201, "grad_norm": 14.8125, "learning_rate": 6.333638459806197e-07, "loss": 1.8025, "step": 7352 }, { "epoch": 0.8431372549019608, "grad_norm": 16.875, "learning_rate": 6.324595623035124e-07, "loss": 2.2283, "step": 7353 }, { "epoch": 0.8432519206513015, "grad_norm": 16.5, "learning_rate": 6.315558810453537e-07, "loss": 1.8961, "step": 7354 }, { "epoch": 0.8433665864006421, "grad_norm": 11.25, "learning_rate": 6.306528023307923e-07, "loss": 1.8067, "step": 7355 }, { "epoch": 0.8434812521499828, "grad_norm": 15.875, "learning_rate": 6.297503262843901e-07, "loss": 2.0321, "step": 7356 }, { "epoch": 0.8435959178993234, "grad_norm": 14.125, "learning_rate": 6.288484530306249e-07, "loss": 1.9155, "step": 7357 }, { "epoch": 0.8437105836486641, "grad_norm": 15.6875, "learning_rate": 6.279471826938944e-07, "loss": 1.6852, "step": 7358 }, { "epoch": 0.8438252493980049, "grad_norm": 29.5, "learning_rate": 6.270465153985105e-07, "loss": 1.8566, "step": 7359 }, { "epoch": 0.8439399151473455, "grad_norm": 11.5625, "learning_rate": 6.261464512687038e-07, "loss": 1.7078, "step": 7360 }, { "epoch": 0.8440545808966862, "grad_norm": 14.4375, "learning_rate": 6.252469904286213e-07, "loss": 1.954, "step": 7361 }, { "epoch": 0.8441692466460269, "grad_norm": 12.0, "learning_rate": 6.243481330023266e-07, "loss": 1.4688, "step": 7362 }, { "epoch": 0.8442839123953675, "grad_norm": 33.0, "learning_rate": 6.234498791137988e-07, "loss": 1.9667, "step": 7363 }, { "epoch": 0.8443985781447082, "grad_norm": 12.5625, "learning_rate": 6.225522288869357e-07, "loss": 2.0317, "step": 7364 }, { "epoch": 0.8445132438940488, "grad_norm": 12.625, "learning_rate": 6.216551824455502e-07, "loss": 1.6987, "step": 7365 }, { "epoch": 0.8446279096433895, "grad_norm": 14.6875, "learning_rate": 6.207587399133741e-07, "loss": 1.738, "step": 7366 }, { "epoch": 0.8447425753927302, "grad_norm": 18.375, "learning_rate": 6.198629014140539e-07, "loss": 1.6648, "step": 7367 }, { "epoch": 0.8448572411420708, "grad_norm": 15.8125, "learning_rate": 6.189676670711542e-07, "loss": 1.7658, "step": 7368 }, { "epoch": 0.8449719068914116, "grad_norm": 15.8125, "learning_rate": 6.180730370081522e-07, "loss": 1.5393, "step": 7369 }, { "epoch": 0.8450865726407522, "grad_norm": 12.3125, "learning_rate": 6.171790113484505e-07, "loss": 1.7971, "step": 7370 }, { "epoch": 0.8452012383900929, "grad_norm": 16.0, "learning_rate": 6.162855902153586e-07, "loss": 1.9036, "step": 7371 }, { "epoch": 0.8453159041394336, "grad_norm": 16.0, "learning_rate": 6.153927737321086e-07, "loss": 1.7259, "step": 7372 }, { "epoch": 0.8454305698887742, "grad_norm": 15.5, "learning_rate": 6.145005620218469e-07, "loss": 2.4454, "step": 7373 }, { "epoch": 0.8455452356381149, "grad_norm": 12.125, "learning_rate": 6.13608955207638e-07, "loss": 1.8717, "step": 7374 }, { "epoch": 0.8456599013874556, "grad_norm": 16.5, "learning_rate": 6.127179534124606e-07, "loss": 1.4341, "step": 7375 }, { "epoch": 0.8457745671367962, "grad_norm": 15.375, "learning_rate": 6.118275567592141e-07, "loss": 1.5926, "step": 7376 }, { "epoch": 0.8458892328861369, "grad_norm": 16.25, "learning_rate": 6.109377653707088e-07, "loss": 1.5518, "step": 7377 }, { "epoch": 0.8460038986354775, "grad_norm": 11.9375, "learning_rate": 6.100485793696753e-07, "loss": 1.7411, "step": 7378 }, { "epoch": 0.8461185643848183, "grad_norm": 14.875, "learning_rate": 6.091599988787606e-07, "loss": 1.9504, "step": 7379 }, { "epoch": 0.846233230134159, "grad_norm": 13.9375, "learning_rate": 6.08272024020527e-07, "loss": 1.3825, "step": 7380 }, { "epoch": 0.8463478958834996, "grad_norm": 15.4375, "learning_rate": 6.073846549174534e-07, "loss": 1.5072, "step": 7381 }, { "epoch": 0.8464625616328403, "grad_norm": 16.5, "learning_rate": 6.064978916919373e-07, "loss": 1.6759, "step": 7382 }, { "epoch": 0.846577227382181, "grad_norm": 15.25, "learning_rate": 6.056117344662876e-07, "loss": 1.7355, "step": 7383 }, { "epoch": 0.8466918931315216, "grad_norm": 15.4375, "learning_rate": 6.047261833627338e-07, "loss": 1.7567, "step": 7384 }, { "epoch": 0.8468065588808623, "grad_norm": 16.0, "learning_rate": 6.038412385034237e-07, "loss": 1.8354, "step": 7385 }, { "epoch": 0.8469212246302029, "grad_norm": 17.125, "learning_rate": 6.029569000104151e-07, "loss": 2.0948, "step": 7386 }, { "epoch": 0.8470358903795436, "grad_norm": 17.0, "learning_rate": 6.020731680056869e-07, "loss": 1.845, "step": 7387 }, { "epoch": 0.8471505561288843, "grad_norm": 14.875, "learning_rate": 6.011900426111339e-07, "loss": 1.3761, "step": 7388 }, { "epoch": 0.847265221878225, "grad_norm": 13.75, "learning_rate": 6.00307523948564e-07, "loss": 2.0455, "step": 7389 }, { "epoch": 0.8473798876275657, "grad_norm": 12.5625, "learning_rate": 5.994256121397058e-07, "loss": 1.7674, "step": 7390 }, { "epoch": 0.8474945533769063, "grad_norm": 15.5625, "learning_rate": 5.985443073062036e-07, "loss": 1.7153, "step": 7391 }, { "epoch": 0.847609219126247, "grad_norm": 14.6875, "learning_rate": 5.976636095696131e-07, "loss": 1.7186, "step": 7392 }, { "epoch": 0.8477238848755877, "grad_norm": 12.5625, "learning_rate": 5.96783519051412e-07, "loss": 1.5469, "step": 7393 }, { "epoch": 0.8478385506249283, "grad_norm": 17.375, "learning_rate": 5.959040358729911e-07, "loss": 1.8276, "step": 7394 }, { "epoch": 0.847953216374269, "grad_norm": 16.5, "learning_rate": 5.95025160155659e-07, "loss": 1.5515, "step": 7395 }, { "epoch": 0.8480678821236097, "grad_norm": 10.125, "learning_rate": 5.941468920206395e-07, "loss": 1.441, "step": 7396 }, { "epoch": 0.8481825478729503, "grad_norm": 14.5625, "learning_rate": 5.932692315890742e-07, "loss": 1.6603, "step": 7397 }, { "epoch": 0.848297213622291, "grad_norm": 13.875, "learning_rate": 5.923921789820175e-07, "loss": 1.8631, "step": 7398 }, { "epoch": 0.8484118793716316, "grad_norm": 15.5625, "learning_rate": 5.91515734320442e-07, "loss": 1.9655, "step": 7399 }, { "epoch": 0.8485265451209724, "grad_norm": 13.8125, "learning_rate": 5.906398977252392e-07, "loss": 1.5185, "step": 7400 }, { "epoch": 0.8486412108703131, "grad_norm": 15.0625, "learning_rate": 5.897646693172121e-07, "loss": 1.7555, "step": 7401 }, { "epoch": 0.8487558766196537, "grad_norm": 11.5625, "learning_rate": 5.888900492170818e-07, "loss": 1.7078, "step": 7402 }, { "epoch": 0.8488705423689944, "grad_norm": 15.6875, "learning_rate": 5.880160375454869e-07, "loss": 1.5772, "step": 7403 }, { "epoch": 0.848985208118335, "grad_norm": 12.625, "learning_rate": 5.871426344229769e-07, "loss": 1.5327, "step": 7404 }, { "epoch": 0.8490998738676757, "grad_norm": 14.0625, "learning_rate": 5.862698399700256e-07, "loss": 1.5809, "step": 7405 }, { "epoch": 0.8492145396170164, "grad_norm": 24.625, "learning_rate": 5.853976543070173e-07, "loss": 1.7233, "step": 7406 }, { "epoch": 0.849329205366357, "grad_norm": 17.375, "learning_rate": 5.845260775542516e-07, "loss": 1.7531, "step": 7407 }, { "epoch": 0.8494438711156977, "grad_norm": 14.4375, "learning_rate": 5.836551098319471e-07, "loss": 1.5508, "step": 7408 }, { "epoch": 0.8495585368650385, "grad_norm": 16.375, "learning_rate": 5.827847512602364e-07, "loss": 1.6986, "step": 7409 }, { "epoch": 0.8496732026143791, "grad_norm": 14.125, "learning_rate": 5.819150019591701e-07, "loss": 1.8656, "step": 7410 }, { "epoch": 0.8497878683637198, "grad_norm": 14.5625, "learning_rate": 5.810458620487125e-07, "loss": 1.8513, "step": 7411 }, { "epoch": 0.8499025341130604, "grad_norm": 15.375, "learning_rate": 5.801773316487463e-07, "loss": 1.7326, "step": 7412 }, { "epoch": 0.8500171998624011, "grad_norm": 16.25, "learning_rate": 5.79309410879067e-07, "loss": 2.2074, "step": 7413 }, { "epoch": 0.8501318656117418, "grad_norm": 16.625, "learning_rate": 5.784420998593887e-07, "loss": 1.5767, "step": 7414 }, { "epoch": 0.8502465313610824, "grad_norm": 14.25, "learning_rate": 5.775753987093397e-07, "loss": 1.6403, "step": 7415 }, { "epoch": 0.8503611971104231, "grad_norm": 14.25, "learning_rate": 5.767093075484653e-07, "loss": 1.4494, "step": 7416 }, { "epoch": 0.8504758628597638, "grad_norm": 13.0, "learning_rate": 5.758438264962268e-07, "loss": 1.5329, "step": 7417 }, { "epoch": 0.8505905286091044, "grad_norm": 16.25, "learning_rate": 5.749789556720009e-07, "loss": 2.0409, "step": 7418 }, { "epoch": 0.8507051943584452, "grad_norm": 15.375, "learning_rate": 5.741146951950777e-07, "loss": 1.8381, "step": 7419 }, { "epoch": 0.8508198601077858, "grad_norm": 15.75, "learning_rate": 5.73251045184669e-07, "loss": 1.7425, "step": 7420 }, { "epoch": 0.8509345258571265, "grad_norm": 15.375, "learning_rate": 5.723880057598963e-07, "loss": 1.9957, "step": 7421 }, { "epoch": 0.8510491916064672, "grad_norm": 14.375, "learning_rate": 5.715255770398004e-07, "loss": 1.9215, "step": 7422 }, { "epoch": 0.8511638573558078, "grad_norm": 23.625, "learning_rate": 5.706637591433367e-07, "loss": 1.6019, "step": 7423 }, { "epoch": 0.8512785231051485, "grad_norm": 19.0, "learning_rate": 5.698025521893763e-07, "loss": 1.7685, "step": 7424 }, { "epoch": 0.8513931888544891, "grad_norm": 11.875, "learning_rate": 5.689419562967069e-07, "loss": 1.8828, "step": 7425 }, { "epoch": 0.8515078546038298, "grad_norm": 16.375, "learning_rate": 5.680819715840319e-07, "loss": 1.7656, "step": 7426 }, { "epoch": 0.8516225203531705, "grad_norm": 14.8125, "learning_rate": 5.672225981699675e-07, "loss": 1.4309, "step": 7427 }, { "epoch": 0.8517371861025111, "grad_norm": 12.8125, "learning_rate": 5.6636383617305e-07, "loss": 1.8541, "step": 7428 }, { "epoch": 0.8518518518518519, "grad_norm": 23.875, "learning_rate": 5.65505685711728e-07, "loss": 1.6094, "step": 7429 }, { "epoch": 0.8519665176011926, "grad_norm": 14.0625, "learning_rate": 5.646481469043675e-07, "loss": 1.6323, "step": 7430 }, { "epoch": 0.8520811833505332, "grad_norm": 14.875, "learning_rate": 5.637912198692497e-07, "loss": 1.9611, "step": 7431 }, { "epoch": 0.8521958490998739, "grad_norm": 20.5, "learning_rate": 5.629349047245724e-07, "loss": 2.0535, "step": 7432 }, { "epoch": 0.8523105148492145, "grad_norm": 18.625, "learning_rate": 5.620792015884458e-07, "loss": 1.8823, "step": 7433 }, { "epoch": 0.8524251805985552, "grad_norm": 15.6875, "learning_rate": 5.612241105788979e-07, "loss": 2.2602, "step": 7434 }, { "epoch": 0.8525398463478959, "grad_norm": 17.25, "learning_rate": 5.603696318138751e-07, "loss": 1.8474, "step": 7435 }, { "epoch": 0.8526545120972365, "grad_norm": 16.5, "learning_rate": 5.595157654112332e-07, "loss": 2.023, "step": 7436 }, { "epoch": 0.8527691778465772, "grad_norm": 13.5625, "learning_rate": 5.586625114887489e-07, "loss": 1.7628, "step": 7437 }, { "epoch": 0.8528838435959178, "grad_norm": 15.6875, "learning_rate": 5.578098701641121e-07, "loss": 2.0366, "step": 7438 }, { "epoch": 0.8529985093452586, "grad_norm": 13.5625, "learning_rate": 5.569578415549259e-07, "loss": 1.7156, "step": 7439 }, { "epoch": 0.8531131750945993, "grad_norm": 15.6875, "learning_rate": 5.561064257787141e-07, "loss": 1.7342, "step": 7440 }, { "epoch": 0.8532278408439399, "grad_norm": 15.8125, "learning_rate": 5.552556229529138e-07, "loss": 1.757, "step": 7441 }, { "epoch": 0.8533425065932806, "grad_norm": 11.5625, "learning_rate": 5.54405433194875e-07, "loss": 1.6062, "step": 7442 }, { "epoch": 0.8534571723426213, "grad_norm": 13.125, "learning_rate": 5.535558566218657e-07, "loss": 1.6336, "step": 7443 }, { "epoch": 0.8535718380919619, "grad_norm": 13.6875, "learning_rate": 5.527068933510687e-07, "loss": 1.8388, "step": 7444 }, { "epoch": 0.8536865038413026, "grad_norm": 17.75, "learning_rate": 5.518585434995832e-07, "loss": 1.7109, "step": 7445 }, { "epoch": 0.8538011695906432, "grad_norm": 18.125, "learning_rate": 5.510108071844223e-07, "loss": 1.7687, "step": 7446 }, { "epoch": 0.8539158353399839, "grad_norm": 13.375, "learning_rate": 5.501636845225161e-07, "loss": 1.6899, "step": 7447 }, { "epoch": 0.8540305010893247, "grad_norm": 10.8125, "learning_rate": 5.493171756307064e-07, "loss": 1.4893, "step": 7448 }, { "epoch": 0.8541451668386653, "grad_norm": 15.4375, "learning_rate": 5.484712806257542e-07, "loss": 1.5634, "step": 7449 }, { "epoch": 0.854259832588006, "grad_norm": 12.8125, "learning_rate": 5.476259996243371e-07, "loss": 1.6014, "step": 7450 }, { "epoch": 0.8543744983373467, "grad_norm": 12.625, "learning_rate": 5.467813327430421e-07, "loss": 1.5781, "step": 7451 }, { "epoch": 0.8544891640866873, "grad_norm": 13.625, "learning_rate": 5.459372800983758e-07, "loss": 1.5496, "step": 7452 }, { "epoch": 0.854603829836028, "grad_norm": 13.5625, "learning_rate": 5.450938418067614e-07, "loss": 1.4008, "step": 7453 }, { "epoch": 0.8547184955853686, "grad_norm": 11.4375, "learning_rate": 5.442510179845306e-07, "loss": 1.3497, "step": 7454 }, { "epoch": 0.8548331613347093, "grad_norm": 13.875, "learning_rate": 5.434088087479389e-07, "loss": 1.7529, "step": 7455 }, { "epoch": 0.85494782708405, "grad_norm": 13.5625, "learning_rate": 5.425672142131527e-07, "loss": 1.9628, "step": 7456 }, { "epoch": 0.8550624928333906, "grad_norm": 13.5625, "learning_rate": 5.417262344962515e-07, "loss": 1.6251, "step": 7457 }, { "epoch": 0.8551771585827314, "grad_norm": 13.4375, "learning_rate": 5.408858697132341e-07, "loss": 1.614, "step": 7458 }, { "epoch": 0.855291824332072, "grad_norm": 18.0, "learning_rate": 5.400461199800128e-07, "loss": 1.7665, "step": 7459 }, { "epoch": 0.8554064900814127, "grad_norm": 15.25, "learning_rate": 5.392069854124143e-07, "loss": 2.2699, "step": 7460 }, { "epoch": 0.8555211558307534, "grad_norm": 14.25, "learning_rate": 5.383684661261824e-07, "loss": 1.7066, "step": 7461 }, { "epoch": 0.855635821580094, "grad_norm": 14.75, "learning_rate": 5.375305622369742e-07, "loss": 1.6143, "step": 7462 }, { "epoch": 0.8557504873294347, "grad_norm": 20.75, "learning_rate": 5.36693273860362e-07, "loss": 1.8345, "step": 7463 }, { "epoch": 0.8558651530787754, "grad_norm": 19.0, "learning_rate": 5.358566011118343e-07, "loss": 2.0823, "step": 7464 }, { "epoch": 0.855979818828116, "grad_norm": 31.25, "learning_rate": 5.350205441067946e-07, "loss": 1.6561, "step": 7465 }, { "epoch": 0.8560944845774567, "grad_norm": 13.3125, "learning_rate": 5.341851029605599e-07, "loss": 1.5606, "step": 7466 }, { "epoch": 0.8562091503267973, "grad_norm": 14.125, "learning_rate": 5.333502777883643e-07, "loss": 1.9331, "step": 7467 }, { "epoch": 0.8563238160761381, "grad_norm": 14.3125, "learning_rate": 5.325160687053571e-07, "loss": 2.0768, "step": 7468 }, { "epoch": 0.8564384818254788, "grad_norm": 15.5625, "learning_rate": 5.316824758265982e-07, "loss": 1.4512, "step": 7469 }, { "epoch": 0.8565531475748194, "grad_norm": 10.625, "learning_rate": 5.308494992670693e-07, "loss": 1.5656, "step": 7470 }, { "epoch": 0.8566678133241601, "grad_norm": 15.4375, "learning_rate": 5.300171391416614e-07, "loss": 1.7837, "step": 7471 }, { "epoch": 0.8567824790735007, "grad_norm": 16.625, "learning_rate": 5.291853955651838e-07, "loss": 2.098, "step": 7472 }, { "epoch": 0.8568971448228414, "grad_norm": 12.875, "learning_rate": 5.283542686523596e-07, "loss": 1.7928, "step": 7473 }, { "epoch": 0.8570118105721821, "grad_norm": 15.25, "learning_rate": 5.275237585178261e-07, "loss": 1.8733, "step": 7474 }, { "epoch": 0.8571264763215227, "grad_norm": 11.8125, "learning_rate": 5.266938652761378e-07, "loss": 1.7719, "step": 7475 }, { "epoch": 0.8572411420708634, "grad_norm": 13.9375, "learning_rate": 5.258645890417619e-07, "loss": 2.1298, "step": 7476 }, { "epoch": 0.8573558078202042, "grad_norm": 15.4375, "learning_rate": 5.250359299290808e-07, "loss": 1.8462, "step": 7477 }, { "epoch": 0.8574704735695448, "grad_norm": 12.9375, "learning_rate": 5.242078880523921e-07, "loss": 1.8106, "step": 7478 }, { "epoch": 0.8575851393188855, "grad_norm": 15.3125, "learning_rate": 5.233804635259093e-07, "loss": 1.8612, "step": 7479 }, { "epoch": 0.8576998050682261, "grad_norm": 16.125, "learning_rate": 5.22553656463759e-07, "loss": 1.7468, "step": 7480 }, { "epoch": 0.8578144708175668, "grad_norm": 13.0625, "learning_rate": 5.217274669799843e-07, "loss": 1.5367, "step": 7481 }, { "epoch": 0.8579291365669075, "grad_norm": 16.125, "learning_rate": 5.20901895188543e-07, "loss": 1.5664, "step": 7482 }, { "epoch": 0.8580438023162481, "grad_norm": 18.5, "learning_rate": 5.200769412033046e-07, "loss": 1.4581, "step": 7483 }, { "epoch": 0.8581584680655888, "grad_norm": 16.625, "learning_rate": 5.192526051380559e-07, "loss": 1.7481, "step": 7484 }, { "epoch": 0.8582731338149295, "grad_norm": 14.5625, "learning_rate": 5.184288871065018e-07, "loss": 1.5086, "step": 7485 }, { "epoch": 0.8583877995642701, "grad_norm": 30.625, "learning_rate": 5.17605787222255e-07, "loss": 1.8359, "step": 7486 }, { "epoch": 0.8585024653136109, "grad_norm": 13.875, "learning_rate": 5.16783305598848e-07, "loss": 1.7432, "step": 7487 }, { "epoch": 0.8586171310629515, "grad_norm": 12.5, "learning_rate": 5.159614423497267e-07, "loss": 1.6588, "step": 7488 }, { "epoch": 0.8587317968122922, "grad_norm": 15.0625, "learning_rate": 5.151401975882492e-07, "loss": 1.8704, "step": 7489 }, { "epoch": 0.8588464625616329, "grad_norm": 16.125, "learning_rate": 5.143195714276927e-07, "loss": 1.7011, "step": 7490 }, { "epoch": 0.8589611283109735, "grad_norm": 36.75, "learning_rate": 5.134995639812479e-07, "loss": 1.7975, "step": 7491 }, { "epoch": 0.8590757940603142, "grad_norm": 17.0, "learning_rate": 5.12680175362017e-07, "loss": 1.4571, "step": 7492 }, { "epoch": 0.8591904598096548, "grad_norm": 14.5625, "learning_rate": 5.118614056830195e-07, "loss": 1.5935, "step": 7493 }, { "epoch": 0.8593051255589955, "grad_norm": 15.25, "learning_rate": 5.110432550571892e-07, "loss": 1.5423, "step": 7494 }, { "epoch": 0.8594197913083362, "grad_norm": 53.5, "learning_rate": 5.102257235973751e-07, "loss": 2.0227, "step": 7495 }, { "epoch": 0.8595344570576768, "grad_norm": 15.0, "learning_rate": 5.094088114163393e-07, "loss": 1.6329, "step": 7496 }, { "epoch": 0.8596491228070176, "grad_norm": 15.375, "learning_rate": 5.085925186267604e-07, "loss": 1.9627, "step": 7497 }, { "epoch": 0.8597637885563583, "grad_norm": 11.6875, "learning_rate": 5.077768453412285e-07, "loss": 1.5957, "step": 7498 }, { "epoch": 0.8598784543056989, "grad_norm": 12.375, "learning_rate": 5.069617916722502e-07, "loss": 1.7795, "step": 7499 }, { "epoch": 0.8599931200550396, "grad_norm": 29.25, "learning_rate": 5.061473577322496e-07, "loss": 1.651, "step": 7500 }, { "epoch": 0.8601077858043802, "grad_norm": 18.5, "learning_rate": 5.053335436335593e-07, "loss": 2.0596, "step": 7501 }, { "epoch": 0.8602224515537209, "grad_norm": 13.5, "learning_rate": 5.04520349488431e-07, "loss": 1.8007, "step": 7502 }, { "epoch": 0.8603371173030616, "grad_norm": 14.4375, "learning_rate": 5.037077754090298e-07, "loss": 1.8082, "step": 7503 }, { "epoch": 0.8604517830524022, "grad_norm": 14.4375, "learning_rate": 5.028958215074315e-07, "loss": 1.5344, "step": 7504 }, { "epoch": 0.8605664488017429, "grad_norm": 15.5, "learning_rate": 5.020844878956332e-07, "loss": 1.9141, "step": 7505 }, { "epoch": 0.8606811145510835, "grad_norm": 13.1875, "learning_rate": 5.012737746855429e-07, "loss": 1.6535, "step": 7506 }, { "epoch": 0.8607957803004243, "grad_norm": 13.25, "learning_rate": 5.004636819889802e-07, "loss": 1.4837, "step": 7507 }, { "epoch": 0.860910446049765, "grad_norm": 26.25, "learning_rate": 4.996542099176844e-07, "loss": 1.8349, "step": 7508 }, { "epoch": 0.8610251117991056, "grad_norm": 13.4375, "learning_rate": 4.988453585833053e-07, "loss": 1.8444, "step": 7509 }, { "epoch": 0.8611397775484463, "grad_norm": 19.125, "learning_rate": 4.980371280974095e-07, "loss": 1.7482, "step": 7510 }, { "epoch": 0.861254443297787, "grad_norm": 16.875, "learning_rate": 4.972295185714765e-07, "loss": 1.6187, "step": 7511 }, { "epoch": 0.8613691090471276, "grad_norm": 13.25, "learning_rate": 4.964225301169023e-07, "loss": 1.6249, "step": 7512 }, { "epoch": 0.8614837747964683, "grad_norm": 15.25, "learning_rate": 4.956161628449924e-07, "loss": 1.5221, "step": 7513 }, { "epoch": 0.8615984405458089, "grad_norm": 13.4375, "learning_rate": 4.948104168669721e-07, "loss": 1.5475, "step": 7514 }, { "epoch": 0.8617131062951496, "grad_norm": 17.5, "learning_rate": 4.940052922939775e-07, "loss": 1.7027, "step": 7515 }, { "epoch": 0.8618277720444903, "grad_norm": 15.6875, "learning_rate": 4.93200789237061e-07, "loss": 1.5804, "step": 7516 }, { "epoch": 0.861942437793831, "grad_norm": 16.25, "learning_rate": 4.923969078071882e-07, "loss": 1.9165, "step": 7517 }, { "epoch": 0.8620571035431717, "grad_norm": 13.9375, "learning_rate": 4.9159364811524e-07, "loss": 1.7664, "step": 7518 }, { "epoch": 0.8621717692925124, "grad_norm": 18.625, "learning_rate": 4.907910102720087e-07, "loss": 1.472, "step": 7519 }, { "epoch": 0.862286435041853, "grad_norm": 15.25, "learning_rate": 4.899889943882052e-07, "loss": 1.9275, "step": 7520 }, { "epoch": 0.8624011007911937, "grad_norm": 14.0625, "learning_rate": 4.891876005744511e-07, "loss": 1.6637, "step": 7521 }, { "epoch": 0.8625157665405343, "grad_norm": 14.4375, "learning_rate": 4.883868289412835e-07, "loss": 1.8778, "step": 7522 }, { "epoch": 0.862630432289875, "grad_norm": 18.5, "learning_rate": 4.875866795991535e-07, "loss": 1.6145, "step": 7523 }, { "epoch": 0.8627450980392157, "grad_norm": 32.75, "learning_rate": 4.867871526584262e-07, "loss": 1.6753, "step": 7524 }, { "epoch": 0.8628597637885563, "grad_norm": 17.25, "learning_rate": 4.85988248229382e-07, "loss": 1.52, "step": 7525 }, { "epoch": 0.862974429537897, "grad_norm": 13.5, "learning_rate": 4.851899664222154e-07, "loss": 1.9137, "step": 7526 }, { "epoch": 0.8630890952872377, "grad_norm": 14.6875, "learning_rate": 4.843923073470313e-07, "loss": 1.9219, "step": 7527 }, { "epoch": 0.8632037610365784, "grad_norm": 19.5, "learning_rate": 4.835952711138531e-07, "loss": 2.2564, "step": 7528 }, { "epoch": 0.8633184267859191, "grad_norm": 14.125, "learning_rate": 4.82798857832617e-07, "loss": 1.7453, "step": 7529 }, { "epoch": 0.8634330925352597, "grad_norm": 14.9375, "learning_rate": 4.820030676131727e-07, "loss": 1.9913, "step": 7530 }, { "epoch": 0.8635477582846004, "grad_norm": 30.125, "learning_rate": 4.812079005652842e-07, "loss": 1.9996, "step": 7531 }, { "epoch": 0.8636624240339411, "grad_norm": 15.25, "learning_rate": 4.804133567986308e-07, "loss": 1.8549, "step": 7532 }, { "epoch": 0.8637770897832817, "grad_norm": 13.9375, "learning_rate": 4.79619436422803e-07, "loss": 1.9366, "step": 7533 }, { "epoch": 0.8638917555326224, "grad_norm": 14.1875, "learning_rate": 4.788261395473065e-07, "loss": 1.7032, "step": 7534 }, { "epoch": 0.864006421281963, "grad_norm": 16.75, "learning_rate": 4.780334662815644e-07, "loss": 1.5725, "step": 7535 }, { "epoch": 0.8641210870313037, "grad_norm": 33.5, "learning_rate": 4.772414167349082e-07, "loss": 1.718, "step": 7536 }, { "epoch": 0.8642357527806445, "grad_norm": 14.4375, "learning_rate": 4.7644999101658705e-07, "loss": 1.6222, "step": 7537 }, { "epoch": 0.8643504185299851, "grad_norm": 14.4375, "learning_rate": 4.756591892357637e-07, "loss": 1.8662, "step": 7538 }, { "epoch": 0.8644650842793258, "grad_norm": 13.0, "learning_rate": 4.74869011501512e-07, "loss": 1.758, "step": 7539 }, { "epoch": 0.8645797500286664, "grad_norm": 11.25, "learning_rate": 4.740794579228236e-07, "loss": 1.5533, "step": 7540 }, { "epoch": 0.8646944157780071, "grad_norm": 12.6875, "learning_rate": 4.732905286086037e-07, "loss": 1.7357, "step": 7541 }, { "epoch": 0.8648090815273478, "grad_norm": 19.25, "learning_rate": 4.7250222366766683e-07, "loss": 1.6112, "step": 7542 }, { "epoch": 0.8649237472766884, "grad_norm": 14.125, "learning_rate": 4.7171454320874666e-07, "loss": 1.8239, "step": 7543 }, { "epoch": 0.8650384130260291, "grad_norm": 14.75, "learning_rate": 4.7092748734048844e-07, "loss": 1.9733, "step": 7544 }, { "epoch": 0.8651530787753698, "grad_norm": 13.1875, "learning_rate": 4.7014105617145087e-07, "loss": 1.672, "step": 7545 }, { "epoch": 0.8652677445247104, "grad_norm": 11.125, "learning_rate": 4.693552498101078e-07, "loss": 1.4246, "step": 7546 }, { "epoch": 0.8653824102740512, "grad_norm": 12.125, "learning_rate": 4.6857006836484695e-07, "loss": 1.5424, "step": 7547 }, { "epoch": 0.8654970760233918, "grad_norm": 14.1875, "learning_rate": 4.677855119439678e-07, "loss": 1.7292, "step": 7548 }, { "epoch": 0.8656117417727325, "grad_norm": 14.5625, "learning_rate": 4.6700158065568434e-07, "loss": 1.9702, "step": 7549 }, { "epoch": 0.8657264075220732, "grad_norm": 15.5625, "learning_rate": 4.662182746081273e-07, "loss": 1.4379, "step": 7550 }, { "epoch": 0.8658410732714138, "grad_norm": 15.3125, "learning_rate": 4.654355939093369e-07, "loss": 1.9743, "step": 7551 }, { "epoch": 0.8659557390207545, "grad_norm": 14.4375, "learning_rate": 4.6465353866727015e-07, "loss": 1.569, "step": 7552 }, { "epoch": 0.8660704047700952, "grad_norm": 27.875, "learning_rate": 4.6387210898979686e-07, "loss": 1.6596, "step": 7553 }, { "epoch": 0.8661850705194358, "grad_norm": 13.9375, "learning_rate": 4.6309130498469747e-07, "loss": 1.9474, "step": 7554 }, { "epoch": 0.8662997362687765, "grad_norm": 13.4375, "learning_rate": 4.623111267596725e-07, "loss": 1.7912, "step": 7555 }, { "epoch": 0.8664144020181171, "grad_norm": 15.3125, "learning_rate": 4.61531574422332e-07, "loss": 1.6084, "step": 7556 }, { "epoch": 0.8665290677674579, "grad_norm": 18.75, "learning_rate": 4.607526480801988e-07, "loss": 1.9542, "step": 7557 }, { "epoch": 0.8666437335167986, "grad_norm": 14.9375, "learning_rate": 4.59974347840712e-07, "loss": 1.9365, "step": 7558 }, { "epoch": 0.8667583992661392, "grad_norm": 17.75, "learning_rate": 4.59196673811223e-07, "loss": 1.6636, "step": 7559 }, { "epoch": 0.8668730650154799, "grad_norm": 12.9375, "learning_rate": 4.584196260989976e-07, "loss": 1.5667, "step": 7560 }, { "epoch": 0.8669877307648205, "grad_norm": 17.5, "learning_rate": 4.576432048112145e-07, "loss": 2.1365, "step": 7561 }, { "epoch": 0.8671023965141612, "grad_norm": 16.125, "learning_rate": 4.5686741005496637e-07, "loss": 1.7935, "step": 7562 }, { "epoch": 0.8672170622635019, "grad_norm": 16.25, "learning_rate": 4.5609224193725877e-07, "loss": 1.6597, "step": 7563 }, { "epoch": 0.8673317280128425, "grad_norm": 13.5, "learning_rate": 4.553177005650111e-07, "loss": 1.5995, "step": 7564 }, { "epoch": 0.8674463937621832, "grad_norm": 12.75, "learning_rate": 4.5454378604505746e-07, "loss": 1.4099, "step": 7565 }, { "epoch": 0.867561059511524, "grad_norm": 15.5, "learning_rate": 4.53770498484144e-07, "loss": 1.6638, "step": 7566 }, { "epoch": 0.8676757252608646, "grad_norm": 14.25, "learning_rate": 4.529978379889316e-07, "loss": 1.8953, "step": 7567 }, { "epoch": 0.8677903910102053, "grad_norm": 13.6875, "learning_rate": 4.522258046659944e-07, "loss": 1.5892, "step": 7568 }, { "epoch": 0.8679050567595459, "grad_norm": 14.5625, "learning_rate": 4.5145439862181726e-07, "loss": 1.8608, "step": 7569 }, { "epoch": 0.8680197225088866, "grad_norm": 13.875, "learning_rate": 4.50683619962804e-07, "loss": 1.4458, "step": 7570 }, { "epoch": 0.8681343882582273, "grad_norm": 12.25, "learning_rate": 4.499134687952672e-07, "loss": 1.8531, "step": 7571 }, { "epoch": 0.8682490540075679, "grad_norm": 37.5, "learning_rate": 4.4914394522543483e-07, "loss": 1.9561, "step": 7572 }, { "epoch": 0.8683637197569086, "grad_norm": 17.875, "learning_rate": 4.483750493594474e-07, "loss": 2.0421, "step": 7573 }, { "epoch": 0.8684783855062492, "grad_norm": 16.625, "learning_rate": 4.4760678130336077e-07, "loss": 1.6492, "step": 7574 }, { "epoch": 0.8685930512555899, "grad_norm": 16.75, "learning_rate": 4.468391411631418e-07, "loss": 1.9316, "step": 7575 }, { "epoch": 0.8687077170049307, "grad_norm": 16.125, "learning_rate": 4.460721290446729e-07, "loss": 1.3019, "step": 7576 }, { "epoch": 0.8688223827542713, "grad_norm": 14.3125, "learning_rate": 4.4530574505374734e-07, "loss": 1.6316, "step": 7577 }, { "epoch": 0.868937048503612, "grad_norm": 13.25, "learning_rate": 4.445399892960739e-07, "loss": 1.8147, "step": 7578 }, { "epoch": 0.8690517142529527, "grad_norm": 12.5, "learning_rate": 4.4377486187727415e-07, "loss": 1.6357, "step": 7579 }, { "epoch": 0.8691663800022933, "grad_norm": 67.0, "learning_rate": 4.430103629028826e-07, "loss": 2.0372, "step": 7580 }, { "epoch": 0.869281045751634, "grad_norm": 17.375, "learning_rate": 4.4224649247834705e-07, "loss": 1.8227, "step": 7581 }, { "epoch": 0.8693957115009746, "grad_norm": 13.375, "learning_rate": 4.4148325070903e-07, "loss": 1.3737, "step": 7582 }, { "epoch": 0.8695103772503153, "grad_norm": 16.0, "learning_rate": 4.407206377002049e-07, "loss": 1.9617, "step": 7583 }, { "epoch": 0.869625042999656, "grad_norm": 12.6875, "learning_rate": 4.399586535570588e-07, "loss": 1.4346, "step": 7584 }, { "epoch": 0.8697397087489966, "grad_norm": 33.25, "learning_rate": 4.3919729838469596e-07, "loss": 1.8765, "step": 7585 }, { "epoch": 0.8698543744983374, "grad_norm": 17.0, "learning_rate": 4.3843657228812843e-07, "loss": 2.023, "step": 7586 }, { "epoch": 0.8699690402476781, "grad_norm": 13.0625, "learning_rate": 4.3767647537228396e-07, "loss": 1.6459, "step": 7587 }, { "epoch": 0.8700837059970187, "grad_norm": 20.25, "learning_rate": 4.3691700774200485e-07, "loss": 1.7023, "step": 7588 }, { "epoch": 0.8701983717463594, "grad_norm": 12.875, "learning_rate": 4.361581695020428e-07, "loss": 1.5793, "step": 7589 }, { "epoch": 0.8703130374957, "grad_norm": 14.625, "learning_rate": 4.3539996075706747e-07, "loss": 1.9553, "step": 7590 }, { "epoch": 0.8704277032450407, "grad_norm": 12.875, "learning_rate": 4.3464238161165905e-07, "loss": 2.0246, "step": 7591 }, { "epoch": 0.8705423689943814, "grad_norm": 14.25, "learning_rate": 4.3388543217030956e-07, "loss": 1.8149, "step": 7592 }, { "epoch": 0.870657034743722, "grad_norm": 21.125, "learning_rate": 4.331291125374265e-07, "loss": 1.9729, "step": 7593 }, { "epoch": 0.8707717004930627, "grad_norm": 12.6875, "learning_rate": 4.3237342281732986e-07, "loss": 1.6431, "step": 7594 }, { "epoch": 0.8708863662424033, "grad_norm": 16.0, "learning_rate": 4.3161836311425287e-07, "loss": 2.0526, "step": 7595 }, { "epoch": 0.8710010319917441, "grad_norm": 12.1875, "learning_rate": 4.308639335323417e-07, "loss": 1.5358, "step": 7596 }, { "epoch": 0.8711156977410848, "grad_norm": 14.0625, "learning_rate": 4.3011013417565583e-07, "loss": 1.7252, "step": 7597 }, { "epoch": 0.8712303634904254, "grad_norm": 14.0, "learning_rate": 4.2935696514816605e-07, "loss": 1.7142, "step": 7598 }, { "epoch": 0.8713450292397661, "grad_norm": 14.0, "learning_rate": 4.2860442655375756e-07, "loss": 1.6085, "step": 7599 }, { "epoch": 0.8714596949891068, "grad_norm": 39.75, "learning_rate": 4.278525184962312e-07, "loss": 2.0269, "step": 7600 }, { "epoch": 0.8715743607384474, "grad_norm": 14.375, "learning_rate": 4.2710124107929627e-07, "loss": 1.806, "step": 7601 }, { "epoch": 0.8716890264877881, "grad_norm": 16.125, "learning_rate": 4.2635059440657765e-07, "loss": 1.5719, "step": 7602 }, { "epoch": 0.8718036922371287, "grad_norm": 14.25, "learning_rate": 4.256005785816142e-07, "loss": 1.6449, "step": 7603 }, { "epoch": 0.8719183579864694, "grad_norm": 12.6875, "learning_rate": 4.2485119370785255e-07, "loss": 1.595, "step": 7604 }, { "epoch": 0.8720330237358102, "grad_norm": 14.4375, "learning_rate": 4.2410243988866006e-07, "loss": 1.8373, "step": 7605 }, { "epoch": 0.8721476894851508, "grad_norm": 14.375, "learning_rate": 4.233543172273119e-07, "loss": 1.7904, "step": 7606 }, { "epoch": 0.8722623552344915, "grad_norm": 15.6875, "learning_rate": 4.226068258269955e-07, "loss": 1.4338, "step": 7607 }, { "epoch": 0.8723770209838321, "grad_norm": 20.875, "learning_rate": 4.2185996579081513e-07, "loss": 1.3652, "step": 7608 }, { "epoch": 0.8724916867331728, "grad_norm": 38.5, "learning_rate": 4.2111373722178493e-07, "loss": 1.9266, "step": 7609 }, { "epoch": 0.8726063524825135, "grad_norm": 15.3125, "learning_rate": 4.203681402228332e-07, "loss": 1.7614, "step": 7610 }, { "epoch": 0.8727210182318541, "grad_norm": 15.375, "learning_rate": 4.19623174896801e-07, "loss": 1.8404, "step": 7611 }, { "epoch": 0.8728356839811948, "grad_norm": 31.125, "learning_rate": 4.188788413464423e-07, "loss": 1.9648, "step": 7612 }, { "epoch": 0.8729503497305355, "grad_norm": 15.6875, "learning_rate": 4.181351396744227e-07, "loss": 1.74, "step": 7613 }, { "epoch": 0.8730650154798761, "grad_norm": 19.5, "learning_rate": 4.173920699833223e-07, "loss": 1.9512, "step": 7614 }, { "epoch": 0.8731796812292169, "grad_norm": 11.0, "learning_rate": 4.166496323756336e-07, "loss": 1.7706, "step": 7615 }, { "epoch": 0.8732943469785575, "grad_norm": 15.6875, "learning_rate": 4.159078269537614e-07, "loss": 1.761, "step": 7616 }, { "epoch": 0.8734090127278982, "grad_norm": 14.0625, "learning_rate": 4.1516665382002383e-07, "loss": 1.6637, "step": 7617 }, { "epoch": 0.8735236784772389, "grad_norm": 14.125, "learning_rate": 4.1442611307665195e-07, "loss": 1.9287, "step": 7618 }, { "epoch": 0.8736383442265795, "grad_norm": 12.8125, "learning_rate": 4.1368620482578723e-07, "loss": 1.5372, "step": 7619 }, { "epoch": 0.8737530099759202, "grad_norm": 14.75, "learning_rate": 4.129469291694893e-07, "loss": 1.5207, "step": 7620 }, { "epoch": 0.8738676757252609, "grad_norm": 16.875, "learning_rate": 4.122082862097243e-07, "loss": 2.0059, "step": 7621 }, { "epoch": 0.8739823414746015, "grad_norm": 13.6875, "learning_rate": 4.114702760483752e-07, "loss": 1.7114, "step": 7622 }, { "epoch": 0.8740970072239422, "grad_norm": 11.8125, "learning_rate": 4.107328987872361e-07, "loss": 1.9292, "step": 7623 }, { "epoch": 0.8742116729732828, "grad_norm": 17.125, "learning_rate": 4.09996154528014e-07, "loss": 1.4067, "step": 7624 }, { "epoch": 0.8743263387226236, "grad_norm": 14.875, "learning_rate": 4.0926004337232926e-07, "loss": 1.649, "step": 7625 }, { "epoch": 0.8744410044719643, "grad_norm": 18.875, "learning_rate": 4.0852456542171457e-07, "loss": 1.5934, "step": 7626 }, { "epoch": 0.8745556702213049, "grad_norm": 12.0625, "learning_rate": 4.077897207776138e-07, "loss": 1.4852, "step": 7627 }, { "epoch": 0.8746703359706456, "grad_norm": 14.625, "learning_rate": 4.0705550954138583e-07, "loss": 1.6567, "step": 7628 }, { "epoch": 0.8747850017199862, "grad_norm": 16.75, "learning_rate": 4.063219318143008e-07, "loss": 1.6747, "step": 7629 }, { "epoch": 0.8748996674693269, "grad_norm": 13.9375, "learning_rate": 4.0558898769754164e-07, "loss": 1.5983, "step": 7630 }, { "epoch": 0.8750143332186676, "grad_norm": 17.25, "learning_rate": 4.0485667729220414e-07, "loss": 1.9183, "step": 7631 }, { "epoch": 0.8751289989680082, "grad_norm": 13.125, "learning_rate": 4.041250006992975e-07, "loss": 1.6862, "step": 7632 }, { "epoch": 0.8752436647173489, "grad_norm": 13.625, "learning_rate": 4.0339395801974103e-07, "loss": 2.0771, "step": 7633 }, { "epoch": 0.8753583304666896, "grad_norm": 16.5, "learning_rate": 4.026635493543679e-07, "loss": 1.6274, "step": 7634 }, { "epoch": 0.8754729962160303, "grad_norm": 15.1875, "learning_rate": 4.0193377480392647e-07, "loss": 1.7877, "step": 7635 }, { "epoch": 0.875587661965371, "grad_norm": 20.5, "learning_rate": 4.0120463446907286e-07, "loss": 2.1031, "step": 7636 }, { "epoch": 0.8757023277147116, "grad_norm": 14.0, "learning_rate": 4.004761284503789e-07, "loss": 1.8658, "step": 7637 }, { "epoch": 0.8758169934640523, "grad_norm": 24.0, "learning_rate": 3.9974825684832906e-07, "loss": 2.2589, "step": 7638 }, { "epoch": 0.875931659213393, "grad_norm": 15.0, "learning_rate": 3.990210197633165e-07, "loss": 1.8111, "step": 7639 }, { "epoch": 0.8760463249627336, "grad_norm": 14.0, "learning_rate": 3.9829441729565265e-07, "loss": 2.1421, "step": 7640 }, { "epoch": 0.8761609907120743, "grad_norm": 16.0, "learning_rate": 3.975684495455584e-07, "loss": 2.0943, "step": 7641 }, { "epoch": 0.8762756564614149, "grad_norm": 16.125, "learning_rate": 3.968431166131648e-07, "loss": 2.0812, "step": 7642 }, { "epoch": 0.8763903222107556, "grad_norm": 29.625, "learning_rate": 3.961184185985195e-07, "loss": 2.2713, "step": 7643 }, { "epoch": 0.8765049879600963, "grad_norm": 16.375, "learning_rate": 3.9539435560158045e-07, "loss": 2.1314, "step": 7644 }, { "epoch": 0.876619653709437, "grad_norm": 14.0625, "learning_rate": 3.9467092772221825e-07, "loss": 1.6232, "step": 7645 }, { "epoch": 0.8767343194587777, "grad_norm": 15.3125, "learning_rate": 3.939481350602159e-07, "loss": 1.6974, "step": 7646 }, { "epoch": 0.8768489852081184, "grad_norm": 17.5, "learning_rate": 3.9322597771527025e-07, "loss": 1.7038, "step": 7647 }, { "epoch": 0.876963650957459, "grad_norm": 16.375, "learning_rate": 3.925044557869867e-07, "loss": 1.697, "step": 7648 }, { "epoch": 0.8770783167067997, "grad_norm": 14.1875, "learning_rate": 3.917835693748856e-07, "loss": 1.8041, "step": 7649 }, { "epoch": 0.8771929824561403, "grad_norm": 14.8125, "learning_rate": 3.910633185784024e-07, "loss": 1.8678, "step": 7650 }, { "epoch": 0.877307648205481, "grad_norm": 12.625, "learning_rate": 3.9034370349687935e-07, "loss": 1.3945, "step": 7651 }, { "epoch": 0.8774223139548217, "grad_norm": 11.8125, "learning_rate": 3.896247242295742e-07, "loss": 1.7931, "step": 7652 }, { "epoch": 0.8775369797041623, "grad_norm": 11.5625, "learning_rate": 3.889063808756577e-07, "loss": 1.569, "step": 7653 }, { "epoch": 0.877651645453503, "grad_norm": 15.3125, "learning_rate": 3.881886735342083e-07, "loss": 1.55, "step": 7654 }, { "epoch": 0.8777663112028438, "grad_norm": 17.125, "learning_rate": 3.8747160230422353e-07, "loss": 1.7464, "step": 7655 }, { "epoch": 0.8778809769521844, "grad_norm": 12.9375, "learning_rate": 3.8675516728460926e-07, "loss": 1.5346, "step": 7656 }, { "epoch": 0.8779956427015251, "grad_norm": 16.0, "learning_rate": 3.86039368574182e-07, "loss": 1.6056, "step": 7657 }, { "epoch": 0.8781103084508657, "grad_norm": 12.5, "learning_rate": 3.853242062716739e-07, "loss": 1.563, "step": 7658 }, { "epoch": 0.8782249742002064, "grad_norm": 25.125, "learning_rate": 3.846096804757282e-07, "loss": 1.9457, "step": 7659 }, { "epoch": 0.8783396399495471, "grad_norm": 14.9375, "learning_rate": 3.838957912848995e-07, "loss": 1.6241, "step": 7660 }, { "epoch": 0.8784543056988877, "grad_norm": 14.6875, "learning_rate": 3.831825387976551e-07, "loss": 1.4134, "step": 7661 }, { "epoch": 0.8785689714482284, "grad_norm": 12.5625, "learning_rate": 3.824699231123763e-07, "loss": 1.902, "step": 7662 }, { "epoch": 0.878683637197569, "grad_norm": 24.875, "learning_rate": 3.8175794432735225e-07, "loss": 1.7713, "step": 7663 }, { "epoch": 0.8787983029469097, "grad_norm": 14.5, "learning_rate": 3.810466025407883e-07, "loss": 1.5976, "step": 7664 }, { "epoch": 0.8789129686962505, "grad_norm": 17.125, "learning_rate": 3.8033589785080036e-07, "loss": 1.8417, "step": 7665 }, { "epoch": 0.8790276344455911, "grad_norm": 15.8125, "learning_rate": 3.796258303554162e-07, "loss": 1.6064, "step": 7666 }, { "epoch": 0.8791423001949318, "grad_norm": 14.375, "learning_rate": 3.789164001525769e-07, "loss": 1.8779, "step": 7667 }, { "epoch": 0.8792569659442725, "grad_norm": 14.875, "learning_rate": 3.782076073401347e-07, "loss": 1.8027, "step": 7668 }, { "epoch": 0.8793716316936131, "grad_norm": 17.75, "learning_rate": 3.7749945201585313e-07, "loss": 1.6037, "step": 7669 }, { "epoch": 0.8794862974429538, "grad_norm": 18.875, "learning_rate": 3.76791934277409e-07, "loss": 2.131, "step": 7670 }, { "epoch": 0.8796009631922944, "grad_norm": 13.25, "learning_rate": 3.7608505422239097e-07, "loss": 1.7429, "step": 7671 }, { "epoch": 0.8797156289416351, "grad_norm": 14.1875, "learning_rate": 3.753788119483004e-07, "loss": 1.6115, "step": 7672 }, { "epoch": 0.8798302946909758, "grad_norm": 34.25, "learning_rate": 3.7467320755254944e-07, "loss": 2.24, "step": 7673 }, { "epoch": 0.8799449604403164, "grad_norm": 13.0, "learning_rate": 3.73968241132463e-07, "loss": 1.7072, "step": 7674 }, { "epoch": 0.8800596261896572, "grad_norm": 18.75, "learning_rate": 3.732639127852761e-07, "loss": 1.7349, "step": 7675 }, { "epoch": 0.8801742919389978, "grad_norm": 15.375, "learning_rate": 3.7256022260814097e-07, "loss": 2.2083, "step": 7676 }, { "epoch": 0.8802889576883385, "grad_norm": 12.0625, "learning_rate": 3.7185717069811497e-07, "loss": 1.7413, "step": 7677 }, { "epoch": 0.8804036234376792, "grad_norm": 13.9375, "learning_rate": 3.7115475715217173e-07, "loss": 2.1476, "step": 7678 }, { "epoch": 0.8805182891870198, "grad_norm": 16.625, "learning_rate": 3.704529820671959e-07, "loss": 1.5992, "step": 7679 }, { "epoch": 0.8806329549363605, "grad_norm": 16.125, "learning_rate": 3.69751845539984e-07, "loss": 1.6242, "step": 7680 }, { "epoch": 0.8807476206857012, "grad_norm": 17.25, "learning_rate": 3.690513476672447e-07, "loss": 2.2802, "step": 7681 }, { "epoch": 0.8808622864350418, "grad_norm": 16.0, "learning_rate": 3.683514885455985e-07, "loss": 1.6696, "step": 7682 }, { "epoch": 0.8809769521843825, "grad_norm": 13.25, "learning_rate": 3.676522682715766e-07, "loss": 1.9122, "step": 7683 }, { "epoch": 0.8810916179337231, "grad_norm": 15.5, "learning_rate": 3.6695368694162346e-07, "loss": 1.6053, "step": 7684 }, { "epoch": 0.8812062836830639, "grad_norm": 15.1875, "learning_rate": 3.662557446520959e-07, "loss": 1.7391, "step": 7685 }, { "epoch": 0.8813209494324046, "grad_norm": 16.5, "learning_rate": 3.6555844149926076e-07, "loss": 1.5316, "step": 7686 }, { "epoch": 0.8814356151817452, "grad_norm": 14.125, "learning_rate": 3.6486177757929784e-07, "loss": 1.7669, "step": 7687 }, { "epoch": 0.8815502809310859, "grad_norm": 15.0625, "learning_rate": 3.6416575298830026e-07, "loss": 1.4859, "step": 7688 }, { "epoch": 0.8816649466804266, "grad_norm": 12.9375, "learning_rate": 3.6347036782226895e-07, "loss": 1.5783, "step": 7689 }, { "epoch": 0.8817796124297672, "grad_norm": 20.125, "learning_rate": 3.627756221771195e-07, "loss": 1.8493, "step": 7690 }, { "epoch": 0.8818942781791079, "grad_norm": 12.375, "learning_rate": 3.620815161486807e-07, "loss": 1.6091, "step": 7691 }, { "epoch": 0.8820089439284485, "grad_norm": 13.875, "learning_rate": 3.613880498326894e-07, "loss": 1.6444, "step": 7692 }, { "epoch": 0.8821236096777892, "grad_norm": 16.375, "learning_rate": 3.6069522332479677e-07, "loss": 2.006, "step": 7693 }, { "epoch": 0.88223827542713, "grad_norm": 13.75, "learning_rate": 3.6000303672056537e-07, "loss": 2.058, "step": 7694 }, { "epoch": 0.8823529411764706, "grad_norm": 11.5625, "learning_rate": 3.593114901154676e-07, "loss": 1.7831, "step": 7695 }, { "epoch": 0.8824676069258113, "grad_norm": 12.5625, "learning_rate": 3.586205836048906e-07, "loss": 1.921, "step": 7696 }, { "epoch": 0.8825822726751519, "grad_norm": 15.75, "learning_rate": 3.57930317284132e-07, "loss": 2.0887, "step": 7697 }, { "epoch": 0.8826969384244926, "grad_norm": 15.125, "learning_rate": 3.5724069124839946e-07, "loss": 1.9364, "step": 7698 }, { "epoch": 0.8828116041738333, "grad_norm": 19.125, "learning_rate": 3.565517055928147e-07, "loss": 1.9421, "step": 7699 }, { "epoch": 0.8829262699231739, "grad_norm": 12.5, "learning_rate": 3.558633604124101e-07, "loss": 1.6535, "step": 7700 }, { "epoch": 0.8830409356725146, "grad_norm": 18.0, "learning_rate": 3.5517565580212966e-07, "loss": 1.6945, "step": 7701 }, { "epoch": 0.8831556014218553, "grad_norm": 14.125, "learning_rate": 3.544885918568286e-07, "loss": 1.51, "step": 7702 }, { "epoch": 0.8832702671711959, "grad_norm": 17.25, "learning_rate": 3.5380216867127613e-07, "loss": 1.8888, "step": 7703 }, { "epoch": 0.8833849329205367, "grad_norm": 12.125, "learning_rate": 3.531163863401488e-07, "loss": 1.6853, "step": 7704 }, { "epoch": 0.8834995986698773, "grad_norm": 17.0, "learning_rate": 3.524312449580375e-07, "loss": 1.7945, "step": 7705 }, { "epoch": 0.883614264419218, "grad_norm": 10.9375, "learning_rate": 3.5174674461944727e-07, "loss": 1.632, "step": 7706 }, { "epoch": 0.8837289301685587, "grad_norm": 29.875, "learning_rate": 3.5106288541878866e-07, "loss": 1.9366, "step": 7707 }, { "epoch": 0.8838435959178993, "grad_norm": 13.375, "learning_rate": 3.5037966745038844e-07, "loss": 1.7585, "step": 7708 }, { "epoch": 0.88395826166724, "grad_norm": 13.5, "learning_rate": 3.4969709080848347e-07, "loss": 1.4031, "step": 7709 }, { "epoch": 0.8840729274165807, "grad_norm": 13.8125, "learning_rate": 3.490151555872212e-07, "loss": 1.9794, "step": 7710 }, { "epoch": 0.8841875931659213, "grad_norm": 17.25, "learning_rate": 3.4833386188066256e-07, "loss": 1.774, "step": 7711 }, { "epoch": 0.884302258915262, "grad_norm": 16.875, "learning_rate": 3.4765320978278006e-07, "loss": 1.4198, "step": 7712 }, { "epoch": 0.8844169246646026, "grad_norm": 17.5, "learning_rate": 3.469731993874542e-07, "loss": 1.7848, "step": 7713 }, { "epoch": 0.8845315904139434, "grad_norm": 17.0, "learning_rate": 3.462938307884811e-07, "loss": 2.0483, "step": 7714 }, { "epoch": 0.8846462561632841, "grad_norm": 16.375, "learning_rate": 3.456151040795663e-07, "loss": 2.1226, "step": 7715 }, { "epoch": 0.8847609219126247, "grad_norm": 14.0625, "learning_rate": 3.449370193543272e-07, "loss": 2.0883, "step": 7716 }, { "epoch": 0.8848755876619654, "grad_norm": 12.0625, "learning_rate": 3.4425957670629285e-07, "loss": 1.6966, "step": 7717 }, { "epoch": 0.884990253411306, "grad_norm": 15.75, "learning_rate": 3.43582776228904e-07, "loss": 1.9506, "step": 7718 }, { "epoch": 0.8851049191606467, "grad_norm": 13.75, "learning_rate": 3.429066180155111e-07, "loss": 1.8357, "step": 7719 }, { "epoch": 0.8852195849099874, "grad_norm": 12.3125, "learning_rate": 3.422311021593777e-07, "loss": 1.6435, "step": 7720 }, { "epoch": 0.885334250659328, "grad_norm": 14.0, "learning_rate": 3.415562287536789e-07, "loss": 1.8011, "step": 7721 }, { "epoch": 0.8854489164086687, "grad_norm": 11.0, "learning_rate": 3.408819978915001e-07, "loss": 1.7323, "step": 7722 }, { "epoch": 0.8855635821580095, "grad_norm": 12.625, "learning_rate": 3.4020840966583926e-07, "loss": 1.6967, "step": 7723 }, { "epoch": 0.8856782479073501, "grad_norm": 13.875, "learning_rate": 3.3953546416960526e-07, "loss": 1.7175, "step": 7724 }, { "epoch": 0.8857929136566908, "grad_norm": 14.375, "learning_rate": 3.388631614956156e-07, "loss": 1.8158, "step": 7725 }, { "epoch": 0.8859075794060314, "grad_norm": 16.5, "learning_rate": 3.38191501736605e-07, "loss": 1.8368, "step": 7726 }, { "epoch": 0.8860222451553721, "grad_norm": 15.0, "learning_rate": 3.3752048498521375e-07, "loss": 1.5439, "step": 7727 }, { "epoch": 0.8861369109047128, "grad_norm": 12.625, "learning_rate": 3.3685011133399725e-07, "loss": 1.6241, "step": 7728 }, { "epoch": 0.8862515766540534, "grad_norm": 13.125, "learning_rate": 3.361803808754194e-07, "loss": 1.9115, "step": 7729 }, { "epoch": 0.8863662424033941, "grad_norm": 11.9375, "learning_rate": 3.3551129370185785e-07, "loss": 1.7318, "step": 7730 }, { "epoch": 0.8864809081527347, "grad_norm": 11.4375, "learning_rate": 3.348428499056e-07, "loss": 1.6159, "step": 7731 }, { "epoch": 0.8865955739020754, "grad_norm": 15.1875, "learning_rate": 3.3417504957884594e-07, "loss": 1.5754, "step": 7732 }, { "epoch": 0.8867102396514162, "grad_norm": 15.875, "learning_rate": 3.3350789281370474e-07, "loss": 1.7309, "step": 7733 }, { "epoch": 0.8868249054007568, "grad_norm": 11.875, "learning_rate": 3.3284137970219734e-07, "loss": 1.7239, "step": 7734 }, { "epoch": 0.8869395711500975, "grad_norm": 18.625, "learning_rate": 3.321755103362584e-07, "loss": 2.2284, "step": 7735 }, { "epoch": 0.8870542368994382, "grad_norm": 13.8125, "learning_rate": 3.315102848077306e-07, "loss": 1.7142, "step": 7736 }, { "epoch": 0.8871689026487788, "grad_norm": 15.4375, "learning_rate": 3.3084570320836894e-07, "loss": 2.0374, "step": 7737 }, { "epoch": 0.8872835683981195, "grad_norm": 14.125, "learning_rate": 3.3018176562984174e-07, "loss": 1.7441, "step": 7738 }, { "epoch": 0.8873982341474601, "grad_norm": 16.75, "learning_rate": 3.2951847216372455e-07, "loss": 1.399, "step": 7739 }, { "epoch": 0.8875128998968008, "grad_norm": 14.375, "learning_rate": 3.288558229015054e-07, "loss": 1.6896, "step": 7740 }, { "epoch": 0.8876275656461415, "grad_norm": 15.5625, "learning_rate": 3.2819381793458725e-07, "loss": 1.9062, "step": 7741 }, { "epoch": 0.8877422313954821, "grad_norm": 33.0, "learning_rate": 3.275324573542782e-07, "loss": 1.9364, "step": 7742 }, { "epoch": 0.8878568971448229, "grad_norm": 37.5, "learning_rate": 3.268717412518013e-07, "loss": 1.6771, "step": 7743 }, { "epoch": 0.8879715628941636, "grad_norm": 12.9375, "learning_rate": 3.262116697182899e-07, "loss": 2.0154, "step": 7744 }, { "epoch": 0.8880862286435042, "grad_norm": 28.875, "learning_rate": 3.255522428447877e-07, "loss": 1.6716, "step": 7745 }, { "epoch": 0.8882008943928449, "grad_norm": 13.375, "learning_rate": 3.248934607222509e-07, "loss": 1.8085, "step": 7746 }, { "epoch": 0.8883155601421855, "grad_norm": 15.125, "learning_rate": 3.2423532344154687e-07, "loss": 1.671, "step": 7747 }, { "epoch": 0.8884302258915262, "grad_norm": 14.75, "learning_rate": 3.235778310934501e-07, "loss": 1.8004, "step": 7748 }, { "epoch": 0.8885448916408669, "grad_norm": 13.5625, "learning_rate": 3.2292098376865143e-07, "loss": 2.146, "step": 7749 }, { "epoch": 0.8886595573902075, "grad_norm": 17.5, "learning_rate": 3.2226478155774955e-07, "loss": 1.9544, "step": 7750 }, { "epoch": 0.8887742231395482, "grad_norm": 15.4375, "learning_rate": 3.216092245512553e-07, "loss": 1.783, "step": 7751 }, { "epoch": 0.8888888888888888, "grad_norm": 14.4375, "learning_rate": 3.2095431283959033e-07, "loss": 1.9101, "step": 7752 }, { "epoch": 0.8890035546382296, "grad_norm": 15.875, "learning_rate": 3.2030004651308786e-07, "loss": 1.342, "step": 7753 }, { "epoch": 0.8891182203875703, "grad_norm": 12.875, "learning_rate": 3.196464256619902e-07, "loss": 1.6666, "step": 7754 }, { "epoch": 0.8892328861369109, "grad_norm": 23.375, "learning_rate": 3.1899345037645135e-07, "loss": 1.9577, "step": 7755 }, { "epoch": 0.8893475518862516, "grad_norm": 15.25, "learning_rate": 3.1834112074653976e-07, "loss": 1.7096, "step": 7756 }, { "epoch": 0.8894622176355923, "grad_norm": 19.0, "learning_rate": 3.176894368622291e-07, "loss": 2.0567, "step": 7757 }, { "epoch": 0.8895768833849329, "grad_norm": 12.625, "learning_rate": 3.170383988134079e-07, "loss": 1.5789, "step": 7758 }, { "epoch": 0.8896915491342736, "grad_norm": 13.9375, "learning_rate": 3.16388006689875e-07, "loss": 1.8283, "step": 7759 }, { "epoch": 0.8898062148836142, "grad_norm": 13.9375, "learning_rate": 3.1573826058133696e-07, "loss": 1.554, "step": 7760 }, { "epoch": 0.8899208806329549, "grad_norm": 17.5, "learning_rate": 3.1508916057741645e-07, "loss": 1.6336, "step": 7761 }, { "epoch": 0.8900355463822957, "grad_norm": 11.625, "learning_rate": 3.144407067676447e-07, "loss": 1.7471, "step": 7762 }, { "epoch": 0.8901502121316363, "grad_norm": 18.375, "learning_rate": 3.137928992414618e-07, "loss": 1.6073, "step": 7763 }, { "epoch": 0.890264877880977, "grad_norm": 14.9375, "learning_rate": 3.1314573808822136e-07, "loss": 2.0282, "step": 7764 }, { "epoch": 0.8903795436303176, "grad_norm": 12.5, "learning_rate": 3.1249922339718684e-07, "loss": 1.5391, "step": 7765 }, { "epoch": 0.8904942093796583, "grad_norm": 14.375, "learning_rate": 3.1185335525753203e-07, "loss": 1.8654, "step": 7766 }, { "epoch": 0.890608875128999, "grad_norm": 13.4375, "learning_rate": 3.112081337583428e-07, "loss": 1.7855, "step": 7767 }, { "epoch": 0.8907235408783396, "grad_norm": 27.875, "learning_rate": 3.1056355898861634e-07, "loss": 2.1438, "step": 7768 }, { "epoch": 0.8908382066276803, "grad_norm": 13.0, "learning_rate": 3.0991963103725653e-07, "loss": 1.6736, "step": 7769 }, { "epoch": 0.890952872377021, "grad_norm": 15.0, "learning_rate": 3.092763499930834e-07, "loss": 1.6544, "step": 7770 }, { "epoch": 0.8910675381263616, "grad_norm": 14.25, "learning_rate": 3.086337159448238e-07, "loss": 1.8966, "step": 7771 }, { "epoch": 0.8911822038757024, "grad_norm": 12.25, "learning_rate": 3.0799172898111796e-07, "loss": 1.7513, "step": 7772 }, { "epoch": 0.891296869625043, "grad_norm": 14.875, "learning_rate": 3.0735038919051496e-07, "loss": 1.7798, "step": 7773 }, { "epoch": 0.8914115353743837, "grad_norm": 11.5625, "learning_rate": 3.0670969666147633e-07, "loss": 1.6338, "step": 7774 }, { "epoch": 0.8915262011237244, "grad_norm": 13.25, "learning_rate": 3.0606965148237186e-07, "loss": 1.3942, "step": 7775 }, { "epoch": 0.891640866873065, "grad_norm": 14.5625, "learning_rate": 3.054302537414855e-07, "loss": 1.5648, "step": 7776 }, { "epoch": 0.8917555326224057, "grad_norm": 15.6875, "learning_rate": 3.0479150352700883e-07, "loss": 1.7437, "step": 7777 }, { "epoch": 0.8918701983717464, "grad_norm": 14.8125, "learning_rate": 3.041534009270447e-07, "loss": 1.7156, "step": 7778 }, { "epoch": 0.891984864121087, "grad_norm": 13.5, "learning_rate": 3.035159460296083e-07, "loss": 1.3771, "step": 7779 }, { "epoch": 0.8920995298704277, "grad_norm": 16.125, "learning_rate": 3.028791389226243e-07, "loss": 1.9332, "step": 7780 }, { "epoch": 0.8922141956197683, "grad_norm": 12.375, "learning_rate": 3.0224297969392745e-07, "loss": 1.492, "step": 7781 }, { "epoch": 0.892328861369109, "grad_norm": 15.25, "learning_rate": 3.016074684312642e-07, "loss": 1.6255, "step": 7782 }, { "epoch": 0.8924435271184498, "grad_norm": 18.125, "learning_rate": 3.009726052222922e-07, "loss": 1.8382, "step": 7783 }, { "epoch": 0.8925581928677904, "grad_norm": 14.0, "learning_rate": 3.0033839015457646e-07, "loss": 1.6653, "step": 7784 }, { "epoch": 0.8926728586171311, "grad_norm": 14.5, "learning_rate": 2.9970482331559637e-07, "loss": 1.8656, "step": 7785 }, { "epoch": 0.8927875243664717, "grad_norm": 17.25, "learning_rate": 2.990719047927404e-07, "loss": 1.7898, "step": 7786 }, { "epoch": 0.8929021901158124, "grad_norm": 13.0625, "learning_rate": 2.9843963467330703e-07, "loss": 1.7488, "step": 7787 }, { "epoch": 0.8930168558651531, "grad_norm": 18.5, "learning_rate": 2.97808013044506e-07, "loss": 2.1577, "step": 7788 }, { "epoch": 0.8931315216144937, "grad_norm": 14.0625, "learning_rate": 2.971770399934587e-07, "loss": 1.6192, "step": 7789 }, { "epoch": 0.8932461873638344, "grad_norm": 14.0, "learning_rate": 2.9654671560719275e-07, "loss": 2.1646, "step": 7790 }, { "epoch": 0.8933608531131751, "grad_norm": 15.8125, "learning_rate": 2.959170399726535e-07, "loss": 2.2368, "step": 7791 }, { "epoch": 0.8934755188625157, "grad_norm": 13.375, "learning_rate": 2.952880131766894e-07, "loss": 1.4209, "step": 7792 }, { "epoch": 0.8935901846118565, "grad_norm": 17.5, "learning_rate": 2.9465963530606436e-07, "loss": 2.0424, "step": 7793 }, { "epoch": 0.8937048503611971, "grad_norm": 19.25, "learning_rate": 2.9403190644745063e-07, "loss": 2.043, "step": 7794 }, { "epoch": 0.8938195161105378, "grad_norm": 15.75, "learning_rate": 2.934048266874312e-07, "loss": 1.5531, "step": 7795 }, { "epoch": 0.8939341818598785, "grad_norm": 27.75, "learning_rate": 2.927783961125002e-07, "loss": 1.9465, "step": 7796 }, { "epoch": 0.8940488476092191, "grad_norm": 13.6875, "learning_rate": 2.9215261480906256e-07, "loss": 2.0642, "step": 7797 }, { "epoch": 0.8941635133585598, "grad_norm": 12.0625, "learning_rate": 2.915274828634307e-07, "loss": 2.0938, "step": 7798 }, { "epoch": 0.8942781791079004, "grad_norm": 12.6875, "learning_rate": 2.9090300036183127e-07, "loss": 1.6929, "step": 7799 }, { "epoch": 0.8943928448572411, "grad_norm": 13.875, "learning_rate": 2.902791673903993e-07, "loss": 1.5888, "step": 7800 }, { "epoch": 0.8945075106065818, "grad_norm": 14.0, "learning_rate": 2.8965598403518094e-07, "loss": 1.9698, "step": 7801 }, { "epoch": 0.8946221763559224, "grad_norm": 14.6875, "learning_rate": 2.8903345038213236e-07, "loss": 1.819, "step": 7802 }, { "epoch": 0.8947368421052632, "grad_norm": 16.375, "learning_rate": 2.8841156651712054e-07, "loss": 1.8022, "step": 7803 }, { "epoch": 0.8948515078546039, "grad_norm": 17.375, "learning_rate": 2.8779033252592127e-07, "loss": 1.9582, "step": 7804 }, { "epoch": 0.8949661736039445, "grad_norm": 14.1875, "learning_rate": 2.8716974849422264e-07, "loss": 1.907, "step": 7805 }, { "epoch": 0.8950808393532852, "grad_norm": 14.9375, "learning_rate": 2.865498145076234e-07, "loss": 1.6127, "step": 7806 }, { "epoch": 0.8951955051026258, "grad_norm": 17.875, "learning_rate": 2.859305306516297e-07, "loss": 1.5689, "step": 7807 }, { "epoch": 0.8953101708519665, "grad_norm": 16.375, "learning_rate": 2.853118970116614e-07, "loss": 1.3386, "step": 7808 }, { "epoch": 0.8954248366013072, "grad_norm": 12.75, "learning_rate": 2.8469391367304756e-07, "loss": 2.1909, "step": 7809 }, { "epoch": 0.8955395023506478, "grad_norm": 19.5, "learning_rate": 2.8407658072102496e-07, "loss": 2.2492, "step": 7810 }, { "epoch": 0.8956541680999885, "grad_norm": 16.375, "learning_rate": 2.8345989824074495e-07, "loss": 1.8365, "step": 7811 }, { "epoch": 0.8957688338493293, "grad_norm": 13.125, "learning_rate": 2.8284386631726726e-07, "loss": 1.7308, "step": 7812 }, { "epoch": 0.8958834995986699, "grad_norm": 15.6875, "learning_rate": 2.822284850355606e-07, "loss": 1.4328, "step": 7813 }, { "epoch": 0.8959981653480106, "grad_norm": 11.75, "learning_rate": 2.8161375448050475e-07, "loss": 1.8066, "step": 7814 }, { "epoch": 0.8961128310973512, "grad_norm": 17.125, "learning_rate": 2.809996747368915e-07, "loss": 1.7434, "step": 7815 }, { "epoch": 0.8962274968466919, "grad_norm": 15.1875, "learning_rate": 2.803862458894208e-07, "loss": 1.5682, "step": 7816 }, { "epoch": 0.8963421625960326, "grad_norm": 15.5625, "learning_rate": 2.7977346802270324e-07, "loss": 1.6, "step": 7817 }, { "epoch": 0.8964568283453732, "grad_norm": 12.25, "learning_rate": 2.791613412212607e-07, "loss": 1.7834, "step": 7818 }, { "epoch": 0.8965714940947139, "grad_norm": 13.0, "learning_rate": 2.7854986556952346e-07, "loss": 1.8595, "step": 7819 }, { "epoch": 0.8966861598440545, "grad_norm": 15.25, "learning_rate": 2.779390411518329e-07, "loss": 1.8299, "step": 7820 }, { "epoch": 0.8968008255933952, "grad_norm": 18.375, "learning_rate": 2.7732886805244106e-07, "loss": 2.2685, "step": 7821 }, { "epoch": 0.896915491342736, "grad_norm": 14.75, "learning_rate": 2.767193463555096e-07, "loss": 1.8456, "step": 7822 }, { "epoch": 0.8970301570920766, "grad_norm": 11.3125, "learning_rate": 2.7611047614511054e-07, "loss": 1.4385, "step": 7823 }, { "epoch": 0.8971448228414173, "grad_norm": 14.5, "learning_rate": 2.755022575052263e-07, "loss": 1.5672, "step": 7824 }, { "epoch": 0.897259488590758, "grad_norm": 13.5625, "learning_rate": 2.7489469051974695e-07, "loss": 1.3741, "step": 7825 }, { "epoch": 0.8973741543400986, "grad_norm": 15.9375, "learning_rate": 2.742877752724771e-07, "loss": 1.9901, "step": 7826 }, { "epoch": 0.8974888200894393, "grad_norm": 20.125, "learning_rate": 2.736815118471292e-07, "loss": 2.3368, "step": 7827 }, { "epoch": 0.8976034858387799, "grad_norm": 16.0, "learning_rate": 2.7307590032732425e-07, "loss": 1.9133, "step": 7828 }, { "epoch": 0.8977181515881206, "grad_norm": 17.375, "learning_rate": 2.724709407965953e-07, "loss": 2.1574, "step": 7829 }, { "epoch": 0.8978328173374613, "grad_norm": 13.875, "learning_rate": 2.7186663333838514e-07, "loss": 2.075, "step": 7830 }, { "epoch": 0.897947483086802, "grad_norm": 14.0, "learning_rate": 2.712629780360465e-07, "loss": 1.6459, "step": 7831 }, { "epoch": 0.8980621488361427, "grad_norm": 14.5, "learning_rate": 2.7065997497284225e-07, "loss": 1.819, "step": 7832 }, { "epoch": 0.8981768145854833, "grad_norm": 17.375, "learning_rate": 2.700576242319458e-07, "loss": 1.6222, "step": 7833 }, { "epoch": 0.898291480334824, "grad_norm": 36.5, "learning_rate": 2.6945592589643855e-07, "loss": 1.8821, "step": 7834 }, { "epoch": 0.8984061460841647, "grad_norm": 18.5, "learning_rate": 2.6885488004931346e-07, "loss": 1.7104, "step": 7835 }, { "epoch": 0.8985208118335053, "grad_norm": 15.0625, "learning_rate": 2.6825448677347433e-07, "loss": 1.8783, "step": 7836 }, { "epoch": 0.898635477582846, "grad_norm": 13.875, "learning_rate": 2.6765474615173314e-07, "loss": 1.6052, "step": 7837 }, { "epoch": 0.8987501433321867, "grad_norm": 25.25, "learning_rate": 2.6705565826681324e-07, "loss": 2.0737, "step": 7838 }, { "epoch": 0.8988648090815273, "grad_norm": 14.9375, "learning_rate": 2.6645722320134794e-07, "loss": 1.5462, "step": 7839 }, { "epoch": 0.898979474830868, "grad_norm": 16.0, "learning_rate": 2.658594410378773e-07, "loss": 1.7645, "step": 7840 }, { "epoch": 0.8990941405802086, "grad_norm": 15.3125, "learning_rate": 2.6526231185885763e-07, "loss": 1.6587, "step": 7841 }, { "epoch": 0.8992088063295494, "grad_norm": 14.9375, "learning_rate": 2.6466583574664915e-07, "loss": 1.5322, "step": 7842 }, { "epoch": 0.8993234720788901, "grad_norm": 13.0625, "learning_rate": 2.6407001278352496e-07, "loss": 1.7559, "step": 7843 }, { "epoch": 0.8994381378282307, "grad_norm": 16.625, "learning_rate": 2.6347484305166706e-07, "loss": 1.8346, "step": 7844 }, { "epoch": 0.8995528035775714, "grad_norm": 16.375, "learning_rate": 2.6288032663316867e-07, "loss": 1.8326, "step": 7845 }, { "epoch": 0.8996674693269121, "grad_norm": 13.5625, "learning_rate": 2.6228646361003083e-07, "loss": 1.833, "step": 7846 }, { "epoch": 0.8997821350762527, "grad_norm": 14.75, "learning_rate": 2.6169325406416757e-07, "loss": 1.7831, "step": 7847 }, { "epoch": 0.8998968008255934, "grad_norm": 12.875, "learning_rate": 2.6110069807739834e-07, "loss": 1.4719, "step": 7848 }, { "epoch": 0.900011466574934, "grad_norm": 30.0, "learning_rate": 2.605087957314567e-07, "loss": 1.7103, "step": 7849 }, { "epoch": 0.9001261323242747, "grad_norm": 19.125, "learning_rate": 2.599175471079834e-07, "loss": 1.9574, "step": 7850 }, { "epoch": 0.9002407980736155, "grad_norm": 12.25, "learning_rate": 2.5932695228853054e-07, "loss": 1.7622, "step": 7851 }, { "epoch": 0.9003554638229561, "grad_norm": 16.125, "learning_rate": 2.5873701135455953e-07, "loss": 1.7496, "step": 7852 }, { "epoch": 0.9004701295722968, "grad_norm": 16.5, "learning_rate": 2.581477243874414e-07, "loss": 2.0642, "step": 7853 }, { "epoch": 0.9005847953216374, "grad_norm": 15.5625, "learning_rate": 2.5755909146845556e-07, "loss": 1.5551, "step": 7854 }, { "epoch": 0.9006994610709781, "grad_norm": 17.5, "learning_rate": 2.569711126787938e-07, "loss": 1.7885, "step": 7855 }, { "epoch": 0.9008141268203188, "grad_norm": 17.25, "learning_rate": 2.5638378809955787e-07, "loss": 1.7572, "step": 7856 }, { "epoch": 0.9009287925696594, "grad_norm": 12.6875, "learning_rate": 2.557971178117563e-07, "loss": 1.527, "step": 7857 }, { "epoch": 0.9010434583190001, "grad_norm": 14.625, "learning_rate": 2.552111018963094e-07, "loss": 1.8552, "step": 7858 }, { "epoch": 0.9011581240683408, "grad_norm": 34.75, "learning_rate": 2.5462574043404795e-07, "loss": 2.1241, "step": 7859 }, { "epoch": 0.9012727898176814, "grad_norm": 19.375, "learning_rate": 2.5404103350570917e-07, "loss": 1.5072, "step": 7860 }, { "epoch": 0.9013874555670222, "grad_norm": 13.625, "learning_rate": 2.5345698119194407e-07, "loss": 1.331, "step": 7861 }, { "epoch": 0.9015021213163628, "grad_norm": 19.25, "learning_rate": 2.5287358357331207e-07, "loss": 1.564, "step": 7862 }, { "epoch": 0.9016167870657035, "grad_norm": 16.375, "learning_rate": 2.5229084073027985e-07, "loss": 1.8371, "step": 7863 }, { "epoch": 0.9017314528150442, "grad_norm": 17.375, "learning_rate": 2.5170875274322657e-07, "loss": 2.2243, "step": 7864 }, { "epoch": 0.9018461185643848, "grad_norm": 16.375, "learning_rate": 2.5112731969244077e-07, "loss": 1.6347, "step": 7865 }, { "epoch": 0.9019607843137255, "grad_norm": 12.1875, "learning_rate": 2.5054654165811874e-07, "loss": 1.7146, "step": 7866 }, { "epoch": 0.9020754500630661, "grad_norm": 13.6875, "learning_rate": 2.4996641872036874e-07, "loss": 1.9624, "step": 7867 }, { "epoch": 0.9021901158124068, "grad_norm": 20.125, "learning_rate": 2.4938695095920784e-07, "loss": 1.682, "step": 7868 }, { "epoch": 0.9023047815617475, "grad_norm": 29.125, "learning_rate": 2.4880813845456207e-07, "loss": 1.9617, "step": 7869 }, { "epoch": 0.9024194473110881, "grad_norm": 27.5, "learning_rate": 2.482299812862665e-07, "loss": 1.7651, "step": 7870 }, { "epoch": 0.9025341130604289, "grad_norm": 12.6875, "learning_rate": 2.47652479534069e-07, "loss": 1.7709, "step": 7871 }, { "epoch": 0.9026487788097696, "grad_norm": 14.5, "learning_rate": 2.470756332776242e-07, "loss": 1.6265, "step": 7872 }, { "epoch": 0.9027634445591102, "grad_norm": 16.625, "learning_rate": 2.4649944259649614e-07, "loss": 1.6926, "step": 7873 }, { "epoch": 0.9028781103084509, "grad_norm": 13.0625, "learning_rate": 2.459239075701608e-07, "loss": 1.8709, "step": 7874 }, { "epoch": 0.9029927760577915, "grad_norm": 15.3125, "learning_rate": 2.453490282780002e-07, "loss": 2.0482, "step": 7875 }, { "epoch": 0.9031074418071322, "grad_norm": 16.25, "learning_rate": 2.4477480479930984e-07, "loss": 1.7757, "step": 7876 }, { "epoch": 0.9032221075564729, "grad_norm": 16.5, "learning_rate": 2.4420123721329345e-07, "loss": 1.7806, "step": 7877 }, { "epoch": 0.9033367733058135, "grad_norm": 18.0, "learning_rate": 2.436283255990612e-07, "loss": 2.1659, "step": 7878 }, { "epoch": 0.9034514390551542, "grad_norm": 14.0625, "learning_rate": 2.4305607003563716e-07, "loss": 2.1197, "step": 7879 }, { "epoch": 0.903566104804495, "grad_norm": 11.75, "learning_rate": 2.42484470601953e-07, "loss": 1.5476, "step": 7880 }, { "epoch": 0.9036807705538356, "grad_norm": 13.625, "learning_rate": 2.4191352737684915e-07, "loss": 1.5961, "step": 7881 }, { "epoch": 0.9037954363031763, "grad_norm": 13.875, "learning_rate": 2.4134324043907696e-07, "loss": 1.6167, "step": 7882 }, { "epoch": 0.9039101020525169, "grad_norm": 17.75, "learning_rate": 2.407736098672969e-07, "loss": 1.7375, "step": 7883 }, { "epoch": 0.9040247678018576, "grad_norm": 11.625, "learning_rate": 2.402046357400778e-07, "loss": 1.5845, "step": 7884 }, { "epoch": 0.9041394335511983, "grad_norm": 15.25, "learning_rate": 2.396363181358996e-07, "loss": 1.8096, "step": 7885 }, { "epoch": 0.9042540993005389, "grad_norm": 13.1875, "learning_rate": 2.3906865713315017e-07, "loss": 1.7148, "step": 7886 }, { "epoch": 0.9043687650498796, "grad_norm": 15.4375, "learning_rate": 2.385016528101275e-07, "loss": 1.9211, "step": 7887 }, { "epoch": 0.9044834307992202, "grad_norm": 16.125, "learning_rate": 2.3793530524504005e-07, "loss": 1.8707, "step": 7888 }, { "epoch": 0.9045980965485609, "grad_norm": 14.625, "learning_rate": 2.3736961451600426e-07, "loss": 1.5621, "step": 7889 }, { "epoch": 0.9047127622979017, "grad_norm": 14.25, "learning_rate": 2.3680458070104496e-07, "loss": 1.6811, "step": 7890 }, { "epoch": 0.9048274280472423, "grad_norm": 15.8125, "learning_rate": 2.3624020387810031e-07, "loss": 1.6461, "step": 7891 }, { "epoch": 0.904942093796583, "grad_norm": 15.125, "learning_rate": 2.3567648412501365e-07, "loss": 1.9751, "step": 7892 }, { "epoch": 0.9050567595459237, "grad_norm": 16.5, "learning_rate": 2.3511342151953942e-07, "loss": 1.7703, "step": 7893 }, { "epoch": 0.9051714252952643, "grad_norm": 12.6875, "learning_rate": 2.345510161393416e-07, "loss": 1.7195, "step": 7894 }, { "epoch": 0.905286091044605, "grad_norm": 14.875, "learning_rate": 2.3398926806199374e-07, "loss": 1.4999, "step": 7895 }, { "epoch": 0.9054007567939456, "grad_norm": 12.625, "learning_rate": 2.3342817736497768e-07, "loss": 1.6374, "step": 7896 }, { "epoch": 0.9055154225432863, "grad_norm": 16.25, "learning_rate": 2.3286774412568602e-07, "loss": 1.6402, "step": 7897 }, { "epoch": 0.905630088292627, "grad_norm": 14.4375, "learning_rate": 2.3230796842141912e-07, "loss": 1.6802, "step": 7898 }, { "epoch": 0.9057447540419676, "grad_norm": 16.5, "learning_rate": 2.3174885032938687e-07, "loss": 2.0467, "step": 7899 }, { "epoch": 0.9058594197913084, "grad_norm": 12.4375, "learning_rate": 2.3119038992670985e-07, "loss": 1.5325, "step": 7900 }, { "epoch": 0.905974085540649, "grad_norm": 14.875, "learning_rate": 2.3063258729041637e-07, "loss": 1.9083, "step": 7901 }, { "epoch": 0.9060887512899897, "grad_norm": 15.375, "learning_rate": 2.300754424974455e-07, "loss": 2.2075, "step": 7902 }, { "epoch": 0.9062034170393304, "grad_norm": 20.25, "learning_rate": 2.2951895562464522e-07, "loss": 1.5334, "step": 7903 }, { "epoch": 0.906318082788671, "grad_norm": 13.3125, "learning_rate": 2.2896312674877018e-07, "loss": 1.6015, "step": 7904 }, { "epoch": 0.9064327485380117, "grad_norm": 23.0, "learning_rate": 2.284079559464869e-07, "loss": 2.2127, "step": 7905 }, { "epoch": 0.9065474142873524, "grad_norm": 15.875, "learning_rate": 2.2785344329437297e-07, "loss": 1.8621, "step": 7906 }, { "epoch": 0.906662080036693, "grad_norm": 13.875, "learning_rate": 2.2729958886891003e-07, "loss": 2.1442, "step": 7907 }, { "epoch": 0.9067767457860337, "grad_norm": 15.5, "learning_rate": 2.2674639274649358e-07, "loss": 1.8504, "step": 7908 }, { "epoch": 0.9068914115353743, "grad_norm": 15.8125, "learning_rate": 2.2619385500342595e-07, "loss": 1.5707, "step": 7909 }, { "epoch": 0.907006077284715, "grad_norm": 16.125, "learning_rate": 2.256419757159173e-07, "loss": 2.0148, "step": 7910 }, { "epoch": 0.9071207430340558, "grad_norm": 15.25, "learning_rate": 2.250907549600917e-07, "loss": 1.2368, "step": 7911 }, { "epoch": 0.9072354087833964, "grad_norm": 25.375, "learning_rate": 2.2454019281197892e-07, "loss": 2.0017, "step": 7912 }, { "epoch": 0.9073500745327371, "grad_norm": 17.0, "learning_rate": 2.2399028934751766e-07, "loss": 2.0109, "step": 7913 }, { "epoch": 0.9074647402820778, "grad_norm": 15.4375, "learning_rate": 2.2344104464255668e-07, "loss": 1.9362, "step": 7914 }, { "epoch": 0.9075794060314184, "grad_norm": 13.875, "learning_rate": 2.228924587728537e-07, "loss": 1.9118, "step": 7915 }, { "epoch": 0.9076940717807591, "grad_norm": 13.0625, "learning_rate": 2.2234453181407655e-07, "loss": 1.8597, "step": 7916 }, { "epoch": 0.9078087375300997, "grad_norm": 12.0625, "learning_rate": 2.217972638418009e-07, "loss": 1.5295, "step": 7917 }, { "epoch": 0.9079234032794404, "grad_norm": 13.8125, "learning_rate": 2.2125065493151243e-07, "loss": 1.933, "step": 7918 }, { "epoch": 0.9080380690287811, "grad_norm": 18.125, "learning_rate": 2.207047051586042e-07, "loss": 1.7979, "step": 7919 }, { "epoch": 0.9081527347781218, "grad_norm": 14.25, "learning_rate": 2.201594145983793e-07, "loss": 2.0279, "step": 7920 }, { "epoch": 0.9082674005274625, "grad_norm": 14.6875, "learning_rate": 2.1961478332605314e-07, "loss": 1.8515, "step": 7921 }, { "epoch": 0.9083820662768031, "grad_norm": 14.25, "learning_rate": 2.190708114167439e-07, "loss": 1.7882, "step": 7922 }, { "epoch": 0.9084967320261438, "grad_norm": 13.0, "learning_rate": 2.185274989454833e-07, "loss": 1.5203, "step": 7923 }, { "epoch": 0.9086113977754845, "grad_norm": 15.875, "learning_rate": 2.1798484598721248e-07, "loss": 1.9672, "step": 7924 }, { "epoch": 0.9087260635248251, "grad_norm": 18.75, "learning_rate": 2.1744285261677656e-07, "loss": 1.4228, "step": 7925 }, { "epoch": 0.9088407292741658, "grad_norm": 13.875, "learning_rate": 2.169015189089363e-07, "loss": 1.7731, "step": 7926 }, { "epoch": 0.9089553950235065, "grad_norm": 12.375, "learning_rate": 2.1636084493835809e-07, "loss": 1.4081, "step": 7927 }, { "epoch": 0.9090700607728471, "grad_norm": 14.75, "learning_rate": 2.1582083077961614e-07, "loss": 1.8716, "step": 7928 }, { "epoch": 0.9091847265221878, "grad_norm": 16.25, "learning_rate": 2.1528147650719532e-07, "loss": 1.9533, "step": 7929 }, { "epoch": 0.9092993922715285, "grad_norm": 24.625, "learning_rate": 2.147427821954906e-07, "loss": 1.5319, "step": 7930 }, { "epoch": 0.9094140580208692, "grad_norm": 14.4375, "learning_rate": 2.1420474791880307e-07, "loss": 1.9576, "step": 7931 }, { "epoch": 0.9095287237702099, "grad_norm": 16.625, "learning_rate": 2.136673737513456e-07, "loss": 1.9118, "step": 7932 }, { "epoch": 0.9096433895195505, "grad_norm": 13.875, "learning_rate": 2.1313065976723835e-07, "loss": 1.6339, "step": 7933 }, { "epoch": 0.9097580552688912, "grad_norm": 31.375, "learning_rate": 2.1259460604051042e-07, "loss": 1.7381, "step": 7934 }, { "epoch": 0.9098727210182318, "grad_norm": 26.75, "learning_rate": 2.1205921264509988e-07, "loss": 1.8871, "step": 7935 }, { "epoch": 0.9099873867675725, "grad_norm": 14.875, "learning_rate": 2.1152447965485435e-07, "loss": 1.7381, "step": 7936 }, { "epoch": 0.9101020525169132, "grad_norm": 16.5, "learning_rate": 2.1099040714353093e-07, "loss": 1.7603, "step": 7937 }, { "epoch": 0.9102167182662538, "grad_norm": 14.9375, "learning_rate": 2.1045699518479344e-07, "loss": 1.7388, "step": 7938 }, { "epoch": 0.9103313840155945, "grad_norm": 13.5625, "learning_rate": 2.09924243852217e-07, "loss": 1.5044, "step": 7939 }, { "epoch": 0.9104460497649353, "grad_norm": 16.125, "learning_rate": 2.093921532192833e-07, "loss": 1.9276, "step": 7940 }, { "epoch": 0.9105607155142759, "grad_norm": 12.6875, "learning_rate": 2.0886072335938534e-07, "loss": 1.7142, "step": 7941 }, { "epoch": 0.9106753812636166, "grad_norm": 16.375, "learning_rate": 2.083299543458228e-07, "loss": 2.1206, "step": 7942 }, { "epoch": 0.9107900470129572, "grad_norm": 12.5, "learning_rate": 2.077998462518055e-07, "loss": 1.7662, "step": 7943 }, { "epoch": 0.9109047127622979, "grad_norm": 12.625, "learning_rate": 2.072703991504521e-07, "loss": 1.7747, "step": 7944 }, { "epoch": 0.9110193785116386, "grad_norm": 12.1875, "learning_rate": 2.0674161311478923e-07, "loss": 1.9001, "step": 7945 }, { "epoch": 0.9111340442609792, "grad_norm": 12.5625, "learning_rate": 2.06213488217753e-07, "loss": 1.5857, "step": 7946 }, { "epoch": 0.9112487100103199, "grad_norm": 13.6875, "learning_rate": 2.0568602453218843e-07, "loss": 1.8982, "step": 7947 }, { "epoch": 0.9113633757596606, "grad_norm": 13.625, "learning_rate": 2.051592221308485e-07, "loss": 1.9087, "step": 7948 }, { "epoch": 0.9114780415090012, "grad_norm": 17.0, "learning_rate": 2.0463308108639612e-07, "loss": 1.8831, "step": 7949 }, { "epoch": 0.911592707258342, "grad_norm": 14.25, "learning_rate": 2.0410760147140163e-07, "loss": 1.6368, "step": 7950 }, { "epoch": 0.9117073730076826, "grad_norm": 16.125, "learning_rate": 2.035827833583459e-07, "loss": 2.0413, "step": 7951 }, { "epoch": 0.9118220387570233, "grad_norm": 22.625, "learning_rate": 2.0305862681961718e-07, "loss": 1.7678, "step": 7952 }, { "epoch": 0.911936704506364, "grad_norm": 13.625, "learning_rate": 2.0253513192751374e-07, "loss": 1.5208, "step": 7953 }, { "epoch": 0.9120513702557046, "grad_norm": 13.25, "learning_rate": 2.0201229875424e-07, "loss": 2.1939, "step": 7954 }, { "epoch": 0.9121660360050453, "grad_norm": 11.875, "learning_rate": 2.0149012737191054e-07, "loss": 1.6528, "step": 7955 }, { "epoch": 0.9122807017543859, "grad_norm": 14.5, "learning_rate": 2.0096861785255218e-07, "loss": 1.7145, "step": 7956 }, { "epoch": 0.9123953675037266, "grad_norm": 14.0625, "learning_rate": 2.0044777026809404e-07, "loss": 1.7404, "step": 7957 }, { "epoch": 0.9125100332530673, "grad_norm": 14.1875, "learning_rate": 1.9992758469037808e-07, "loss": 1.7393, "step": 7958 }, { "epoch": 0.912624699002408, "grad_norm": 14.125, "learning_rate": 1.994080611911553e-07, "loss": 2.0732, "step": 7959 }, { "epoch": 0.9127393647517487, "grad_norm": 18.25, "learning_rate": 1.988891998420811e-07, "loss": 2.3128, "step": 7960 }, { "epoch": 0.9128540305010894, "grad_norm": 14.625, "learning_rate": 1.9837100071472492e-07, "loss": 1.7313, "step": 7961 }, { "epoch": 0.91296869625043, "grad_norm": 14.0, "learning_rate": 1.9785346388056238e-07, "loss": 1.4565, "step": 7962 }, { "epoch": 0.9130833619997707, "grad_norm": 27.125, "learning_rate": 1.973365894109769e-07, "loss": 1.8423, "step": 7963 }, { "epoch": 0.9131980277491113, "grad_norm": 12.125, "learning_rate": 1.9682037737726202e-07, "loss": 1.7558, "step": 7964 }, { "epoch": 0.913312693498452, "grad_norm": 13.0625, "learning_rate": 1.9630482785061856e-07, "loss": 1.5248, "step": 7965 }, { "epoch": 0.9134273592477927, "grad_norm": 13.5, "learning_rate": 1.9578994090215796e-07, "loss": 1.6405, "step": 7966 }, { "epoch": 0.9135420249971333, "grad_norm": 13.5625, "learning_rate": 1.952757166028979e-07, "loss": 1.7361, "step": 7967 }, { "epoch": 0.913656690746474, "grad_norm": 15.375, "learning_rate": 1.9476215502376718e-07, "loss": 1.5241, "step": 7968 }, { "epoch": 0.9137713564958146, "grad_norm": 17.0, "learning_rate": 1.9424925623560077e-07, "loss": 2.2397, "step": 7969 }, { "epoch": 0.9138860222451554, "grad_norm": 16.375, "learning_rate": 1.937370203091421e-07, "loss": 2.012, "step": 7970 }, { "epoch": 0.9140006879944961, "grad_norm": 14.5625, "learning_rate": 1.9322544731504744e-07, "loss": 1.5331, "step": 7971 }, { "epoch": 0.9141153537438367, "grad_norm": 18.625, "learning_rate": 1.9271453732387647e-07, "loss": 2.0483, "step": 7972 }, { "epoch": 0.9142300194931774, "grad_norm": 11.9375, "learning_rate": 1.9220429040609945e-07, "loss": 1.5306, "step": 7973 }, { "epoch": 0.9143446852425181, "grad_norm": 12.0, "learning_rate": 1.9169470663209676e-07, "loss": 1.5575, "step": 7974 }, { "epoch": 0.9144593509918587, "grad_norm": 14.75, "learning_rate": 1.9118578607215333e-07, "loss": 1.642, "step": 7975 }, { "epoch": 0.9145740167411994, "grad_norm": 17.875, "learning_rate": 1.9067752879646684e-07, "loss": 2.1624, "step": 7976 }, { "epoch": 0.91468868249054, "grad_norm": 11.9375, "learning_rate": 1.901699348751418e-07, "loss": 1.5853, "step": 7977 }, { "epoch": 0.9148033482398807, "grad_norm": 13.0, "learning_rate": 1.896630043781894e-07, "loss": 1.7873, "step": 7978 }, { "epoch": 0.9149180139892215, "grad_norm": 14.75, "learning_rate": 1.891567373755332e-07, "loss": 2.1597, "step": 7979 }, { "epoch": 0.9150326797385621, "grad_norm": 16.375, "learning_rate": 1.8865113393700118e-07, "loss": 1.9722, "step": 7980 }, { "epoch": 0.9151473454879028, "grad_norm": 13.125, "learning_rate": 1.8814619413233316e-07, "loss": 1.7137, "step": 7981 }, { "epoch": 0.9152620112372435, "grad_norm": 13.5625, "learning_rate": 1.8764191803117505e-07, "loss": 1.8135, "step": 7982 }, { "epoch": 0.9153766769865841, "grad_norm": 14.875, "learning_rate": 1.8713830570308344e-07, "loss": 1.978, "step": 7983 }, { "epoch": 0.9154913427359248, "grad_norm": 14.6875, "learning_rate": 1.8663535721752002e-07, "loss": 2.0155, "step": 7984 }, { "epoch": 0.9156060084852654, "grad_norm": 16.125, "learning_rate": 1.8613307264385816e-07, "loss": 1.6498, "step": 7985 }, { "epoch": 0.9157206742346061, "grad_norm": 13.625, "learning_rate": 1.8563145205137856e-07, "loss": 1.6192, "step": 7986 }, { "epoch": 0.9158353399839468, "grad_norm": 19.125, "learning_rate": 1.8513049550926976e-07, "loss": 2.2138, "step": 7987 }, { "epoch": 0.9159500057332874, "grad_norm": 14.0, "learning_rate": 1.8463020308662926e-07, "loss": 1.6914, "step": 7988 }, { "epoch": 0.9160646714826282, "grad_norm": 14.625, "learning_rate": 1.8413057485246355e-07, "loss": 1.5389, "step": 7989 }, { "epoch": 0.9161793372319688, "grad_norm": 19.875, "learning_rate": 1.836316108756847e-07, "loss": 1.5264, "step": 7990 }, { "epoch": 0.9162940029813095, "grad_norm": 16.25, "learning_rate": 1.831333112251188e-07, "loss": 1.6442, "step": 7991 }, { "epoch": 0.9164086687306502, "grad_norm": 14.375, "learning_rate": 1.8263567596949362e-07, "loss": 1.7746, "step": 7992 }, { "epoch": 0.9165233344799908, "grad_norm": 20.0, "learning_rate": 1.821387051774498e-07, "loss": 2.0996, "step": 7993 }, { "epoch": 0.9166380002293315, "grad_norm": 20.0, "learning_rate": 1.8164239891753478e-07, "loss": 1.7182, "step": 7994 }, { "epoch": 0.9167526659786722, "grad_norm": 15.3125, "learning_rate": 1.811467572582043e-07, "loss": 1.6399, "step": 7995 }, { "epoch": 0.9168673317280128, "grad_norm": 17.125, "learning_rate": 1.8065178026782315e-07, "loss": 2.196, "step": 7996 }, { "epoch": 0.9169819974773535, "grad_norm": 17.875, "learning_rate": 1.801574680146645e-07, "loss": 1.5264, "step": 7997 }, { "epoch": 0.9170966632266941, "grad_norm": 19.375, "learning_rate": 1.7966382056690769e-07, "loss": 2.0183, "step": 7998 }, { "epoch": 0.9172113289760349, "grad_norm": 13.75, "learning_rate": 1.7917083799264323e-07, "loss": 1.9459, "step": 7999 }, { "epoch": 0.9173259947253756, "grad_norm": 17.625, "learning_rate": 1.786785203598679e-07, "loss": 1.7571, "step": 8000 }, { "epoch": 0.9174406604747162, "grad_norm": 16.125, "learning_rate": 1.7818686773648842e-07, "loss": 1.6949, "step": 8001 }, { "epoch": 0.9175553262240569, "grad_norm": 14.0, "learning_rate": 1.7769588019031837e-07, "loss": 1.5695, "step": 8002 }, { "epoch": 0.9176699919733975, "grad_norm": 13.9375, "learning_rate": 1.772055577890802e-07, "loss": 1.9438, "step": 8003 }, { "epoch": 0.9177846577227382, "grad_norm": 13.875, "learning_rate": 1.7671590060040434e-07, "loss": 2.0216, "step": 8004 }, { "epoch": 0.9178993234720789, "grad_norm": 14.1875, "learning_rate": 1.7622690869182945e-07, "loss": 1.9717, "step": 8005 }, { "epoch": 0.9180139892214195, "grad_norm": 16.375, "learning_rate": 1.7573858213080386e-07, "loss": 1.5144, "step": 8006 }, { "epoch": 0.9181286549707602, "grad_norm": 11.4375, "learning_rate": 1.7525092098468143e-07, "loss": 1.7895, "step": 8007 }, { "epoch": 0.918243320720101, "grad_norm": 13.75, "learning_rate": 1.7476392532072673e-07, "loss": 1.653, "step": 8008 }, { "epoch": 0.9183579864694416, "grad_norm": 14.625, "learning_rate": 1.7427759520611153e-07, "loss": 1.6562, "step": 8009 }, { "epoch": 0.9184726522187823, "grad_norm": 33.0, "learning_rate": 1.7379193070791388e-07, "loss": 2.0503, "step": 8010 }, { "epoch": 0.9185873179681229, "grad_norm": 15.625, "learning_rate": 1.7330693189312463e-07, "loss": 1.495, "step": 8011 }, { "epoch": 0.9187019837174636, "grad_norm": 17.25, "learning_rate": 1.7282259882863972e-07, "loss": 2.1224, "step": 8012 }, { "epoch": 0.9188166494668043, "grad_norm": 12.9375, "learning_rate": 1.723389315812618e-07, "loss": 1.7617, "step": 8013 }, { "epoch": 0.9189313152161449, "grad_norm": 12.8125, "learning_rate": 1.7185593021770474e-07, "loss": 1.7341, "step": 8014 }, { "epoch": 0.9190459809654856, "grad_norm": 14.0, "learning_rate": 1.713735948045897e-07, "loss": 1.5487, "step": 8015 }, { "epoch": 0.9191606467148263, "grad_norm": 14.625, "learning_rate": 1.7089192540844512e-07, "loss": 1.7311, "step": 8016 }, { "epoch": 0.9192753124641669, "grad_norm": 16.0, "learning_rate": 1.7041092209570785e-07, "loss": 1.7209, "step": 8017 }, { "epoch": 0.9193899782135077, "grad_norm": 14.625, "learning_rate": 1.6993058493272485e-07, "loss": 1.8215, "step": 8018 }, { "epoch": 0.9195046439628483, "grad_norm": 14.4375, "learning_rate": 1.6945091398574698e-07, "loss": 1.8749, "step": 8019 }, { "epoch": 0.919619309712189, "grad_norm": 15.4375, "learning_rate": 1.689719093209363e-07, "loss": 1.7797, "step": 8020 }, { "epoch": 0.9197339754615297, "grad_norm": 14.25, "learning_rate": 1.6849357100436447e-07, "loss": 1.984, "step": 8021 }, { "epoch": 0.9198486412108703, "grad_norm": 17.75, "learning_rate": 1.68015899102007e-07, "loss": 2.3068, "step": 8022 }, { "epoch": 0.919963306960211, "grad_norm": 12.0625, "learning_rate": 1.6753889367975007e-07, "loss": 1.7573, "step": 8023 }, { "epoch": 0.9200779727095516, "grad_norm": 14.625, "learning_rate": 1.670625548033883e-07, "loss": 1.3285, "step": 8024 }, { "epoch": 0.9201926384588923, "grad_norm": 13.0, "learning_rate": 1.6658688253862188e-07, "loss": 1.9462, "step": 8025 }, { "epoch": 0.920307304208233, "grad_norm": 15.4375, "learning_rate": 1.6611187695106222e-07, "loss": 1.761, "step": 8026 }, { "epoch": 0.9204219699575736, "grad_norm": 15.0625, "learning_rate": 1.6563753810622752e-07, "loss": 2.1504, "step": 8027 }, { "epoch": 0.9205366357069144, "grad_norm": 18.125, "learning_rate": 1.6516386606954204e-07, "loss": 1.7004, "step": 8028 }, { "epoch": 0.9206513014562551, "grad_norm": 31.875, "learning_rate": 1.6469086090634134e-07, "loss": 1.7731, "step": 8029 }, { "epoch": 0.9207659672055957, "grad_norm": 14.5625, "learning_rate": 1.6421852268186712e-07, "loss": 1.5433, "step": 8030 }, { "epoch": 0.9208806329549364, "grad_norm": 13.75, "learning_rate": 1.6374685146126945e-07, "loss": 1.6479, "step": 8031 }, { "epoch": 0.920995298704277, "grad_norm": 14.25, "learning_rate": 1.6327584730960578e-07, "loss": 1.7742, "step": 8032 }, { "epoch": 0.9211099644536177, "grad_norm": 11.9375, "learning_rate": 1.6280551029184354e-07, "loss": 1.696, "step": 8033 }, { "epoch": 0.9212246302029584, "grad_norm": 13.25, "learning_rate": 1.6233584047285477e-07, "loss": 1.6746, "step": 8034 }, { "epoch": 0.921339295952299, "grad_norm": 15.625, "learning_rate": 1.6186683791742318e-07, "loss": 1.4795, "step": 8035 }, { "epoch": 0.9214539617016397, "grad_norm": 15.625, "learning_rate": 1.6139850269023761e-07, "loss": 1.7983, "step": 8036 }, { "epoch": 0.9215686274509803, "grad_norm": 16.625, "learning_rate": 1.6093083485589633e-07, "loss": 1.8855, "step": 8037 }, { "epoch": 0.921683293200321, "grad_norm": 12.9375, "learning_rate": 1.604638344789061e-07, "loss": 1.8484, "step": 8038 }, { "epoch": 0.9217979589496618, "grad_norm": 14.375, "learning_rate": 1.5999750162367988e-07, "loss": 2.0297, "step": 8039 }, { "epoch": 0.9219126246990024, "grad_norm": 16.125, "learning_rate": 1.595318363545384e-07, "loss": 1.9807, "step": 8040 }, { "epoch": 0.9220272904483431, "grad_norm": 17.0, "learning_rate": 1.590668387357136e-07, "loss": 1.4929, "step": 8041 }, { "epoch": 0.9221419561976838, "grad_norm": 14.9375, "learning_rate": 1.5860250883134142e-07, "loss": 1.7644, "step": 8042 }, { "epoch": 0.9222566219470244, "grad_norm": 16.25, "learning_rate": 1.581388467054673e-07, "loss": 1.9432, "step": 8043 }, { "epoch": 0.9223712876963651, "grad_norm": 14.5, "learning_rate": 1.5767585242204564e-07, "loss": 1.8872, "step": 8044 }, { "epoch": 0.9224859534457057, "grad_norm": 14.625, "learning_rate": 1.5721352604493644e-07, "loss": 1.509, "step": 8045 }, { "epoch": 0.9226006191950464, "grad_norm": 13.3125, "learning_rate": 1.5675186763790985e-07, "loss": 1.6886, "step": 8046 }, { "epoch": 0.9227152849443871, "grad_norm": 13.8125, "learning_rate": 1.5629087726464264e-07, "loss": 1.7694, "step": 8047 }, { "epoch": 0.9228299506937278, "grad_norm": 16.25, "learning_rate": 1.5583055498871902e-07, "loss": 1.7768, "step": 8048 }, { "epoch": 0.9229446164430685, "grad_norm": 10.9375, "learning_rate": 1.5537090087363204e-07, "loss": 1.6386, "step": 8049 }, { "epoch": 0.9230592821924092, "grad_norm": 17.125, "learning_rate": 1.5491191498278214e-07, "loss": 2.0055, "step": 8050 }, { "epoch": 0.9231739479417498, "grad_norm": 14.9375, "learning_rate": 1.5445359737947807e-07, "loss": 1.646, "step": 8051 }, { "epoch": 0.9232886136910905, "grad_norm": 13.0, "learning_rate": 1.5399594812693543e-07, "loss": 1.68, "step": 8052 }, { "epoch": 0.9234032794404311, "grad_norm": 21.625, "learning_rate": 1.5353896728827922e-07, "loss": 1.7324, "step": 8053 }, { "epoch": 0.9235179451897718, "grad_norm": 16.875, "learning_rate": 1.5308265492654018e-07, "loss": 1.954, "step": 8054 }, { "epoch": 0.9236326109391125, "grad_norm": 14.9375, "learning_rate": 1.5262701110465737e-07, "loss": 1.3701, "step": 8055 }, { "epoch": 0.9237472766884531, "grad_norm": 17.875, "learning_rate": 1.5217203588548003e-07, "loss": 1.45, "step": 8056 }, { "epoch": 0.9238619424377938, "grad_norm": 14.125, "learning_rate": 1.5171772933176177e-07, "loss": 1.6219, "step": 8057 }, { "epoch": 0.9239766081871345, "grad_norm": 13.9375, "learning_rate": 1.5126409150616638e-07, "loss": 2.0411, "step": 8058 }, { "epoch": 0.9240912739364752, "grad_norm": 15.8125, "learning_rate": 1.5081112247126494e-07, "loss": 2.013, "step": 8059 }, { "epoch": 0.9242059396858159, "grad_norm": 16.25, "learning_rate": 1.503588222895336e-07, "loss": 1.7816, "step": 8060 }, { "epoch": 0.9243206054351565, "grad_norm": 17.125, "learning_rate": 1.4990719102336127e-07, "loss": 1.659, "step": 8061 }, { "epoch": 0.9244352711844972, "grad_norm": 14.25, "learning_rate": 1.4945622873504094e-07, "loss": 1.78, "step": 8062 }, { "epoch": 0.9245499369338379, "grad_norm": 16.75, "learning_rate": 1.4900593548677346e-07, "loss": 1.7673, "step": 8063 }, { "epoch": 0.9246646026831785, "grad_norm": 31.375, "learning_rate": 1.4855631134066907e-07, "loss": 1.7177, "step": 8064 }, { "epoch": 0.9247792684325192, "grad_norm": 14.125, "learning_rate": 1.481073563587443e-07, "loss": 1.8395, "step": 8065 }, { "epoch": 0.9248939341818598, "grad_norm": 20.25, "learning_rate": 1.4765907060292462e-07, "loss": 2.0543, "step": 8066 }, { "epoch": 0.9250085999312005, "grad_norm": 48.25, "learning_rate": 1.4721145413504168e-07, "loss": 2.1609, "step": 8067 }, { "epoch": 0.9251232656805413, "grad_norm": 12.9375, "learning_rate": 1.467645070168372e-07, "loss": 2.0873, "step": 8068 }, { "epoch": 0.9252379314298819, "grad_norm": 14.125, "learning_rate": 1.4631822930995687e-07, "loss": 1.8441, "step": 8069 }, { "epoch": 0.9253525971792226, "grad_norm": 16.625, "learning_rate": 1.4587262107595702e-07, "loss": 1.8324, "step": 8070 }, { "epoch": 0.9254672629285632, "grad_norm": 47.5, "learning_rate": 1.4542768237630178e-07, "loss": 1.826, "step": 8071 }, { "epoch": 0.9255819286779039, "grad_norm": 15.0, "learning_rate": 1.4498341327236044e-07, "loss": 1.7325, "step": 8072 }, { "epoch": 0.9256965944272446, "grad_norm": 22.5, "learning_rate": 1.4453981382541283e-07, "loss": 1.6838, "step": 8073 }, { "epoch": 0.9258112601765852, "grad_norm": 13.125, "learning_rate": 1.4409688409664447e-07, "loss": 1.4571, "step": 8074 }, { "epoch": 0.9259259259259259, "grad_norm": 15.625, "learning_rate": 1.4365462414714814e-07, "loss": 1.8811, "step": 8075 }, { "epoch": 0.9260405916752666, "grad_norm": 16.375, "learning_rate": 1.4321303403792674e-07, "loss": 2.303, "step": 8076 }, { "epoch": 0.9261552574246072, "grad_norm": 12.4375, "learning_rate": 1.4277211382988875e-07, "loss": 1.677, "step": 8077 }, { "epoch": 0.926269923173948, "grad_norm": 13.25, "learning_rate": 1.4233186358384998e-07, "loss": 2.2584, "step": 8078 }, { "epoch": 0.9263845889232886, "grad_norm": 12.5625, "learning_rate": 1.4189228336053519e-07, "loss": 1.5986, "step": 8079 }, { "epoch": 0.9264992546726293, "grad_norm": 14.5625, "learning_rate": 1.4145337322057584e-07, "loss": 2.0368, "step": 8080 }, { "epoch": 0.92661392042197, "grad_norm": 13.4375, "learning_rate": 1.4101513322451133e-07, "loss": 1.5175, "step": 8081 }, { "epoch": 0.9267285861713106, "grad_norm": 17.75, "learning_rate": 1.405775634327883e-07, "loss": 1.8683, "step": 8082 }, { "epoch": 0.9268432519206513, "grad_norm": 16.125, "learning_rate": 1.401406639057623e-07, "loss": 1.6429, "step": 8083 }, { "epoch": 0.926957917669992, "grad_norm": 16.75, "learning_rate": 1.3970443470369354e-07, "loss": 1.8486, "step": 8084 }, { "epoch": 0.9270725834193326, "grad_norm": 16.75, "learning_rate": 1.3926887588675276e-07, "loss": 1.657, "step": 8085 }, { "epoch": 0.9271872491686733, "grad_norm": 12.3125, "learning_rate": 1.388339875150163e-07, "loss": 2.0623, "step": 8086 }, { "epoch": 0.927301914918014, "grad_norm": 14.0, "learning_rate": 1.3839976964846903e-07, "loss": 1.6064, "step": 8087 }, { "epoch": 0.9274165806673547, "grad_norm": 15.875, "learning_rate": 1.3796622234700297e-07, "loss": 1.3991, "step": 8088 }, { "epoch": 0.9275312464166954, "grad_norm": 13.0625, "learning_rate": 1.375333456704181e-07, "loss": 1.8078, "step": 8089 }, { "epoch": 0.927645912166036, "grad_norm": 12.25, "learning_rate": 1.3710113967842054e-07, "loss": 1.892, "step": 8090 }, { "epoch": 0.9277605779153767, "grad_norm": 17.0, "learning_rate": 1.3666960443062648e-07, "loss": 1.8703, "step": 8091 }, { "epoch": 0.9278752436647173, "grad_norm": 14.0625, "learning_rate": 1.3623873998655612e-07, "loss": 1.8084, "step": 8092 }, { "epoch": 0.927989909414058, "grad_norm": 17.375, "learning_rate": 1.3580854640564023e-07, "loss": 2.0948, "step": 8093 }, { "epoch": 0.9281045751633987, "grad_norm": 16.5, "learning_rate": 1.3537902374721524e-07, "loss": 1.7088, "step": 8094 }, { "epoch": 0.9282192409127393, "grad_norm": 19.25, "learning_rate": 1.349501720705254e-07, "loss": 1.9312, "step": 8095 }, { "epoch": 0.92833390666208, "grad_norm": 12.375, "learning_rate": 1.3452199143472345e-07, "loss": 1.6589, "step": 8096 }, { "epoch": 0.9284485724114208, "grad_norm": 12.375, "learning_rate": 1.3409448189886875e-07, "loss": 1.6368, "step": 8097 }, { "epoch": 0.9285632381607614, "grad_norm": 14.375, "learning_rate": 1.3366764352192752e-07, "loss": 1.8738, "step": 8098 }, { "epoch": 0.9286779039101021, "grad_norm": 13.3125, "learning_rate": 1.3324147636277373e-07, "loss": 1.4972, "step": 8099 }, { "epoch": 0.9287925696594427, "grad_norm": 12.625, "learning_rate": 1.328159804801893e-07, "loss": 1.9279, "step": 8100 }, { "epoch": 0.9289072354087834, "grad_norm": 14.9375, "learning_rate": 1.323911559328639e-07, "loss": 1.8333, "step": 8101 }, { "epoch": 0.9290219011581241, "grad_norm": 13.4375, "learning_rate": 1.3196700277939346e-07, "loss": 1.6746, "step": 8102 }, { "epoch": 0.9291365669074647, "grad_norm": 15.875, "learning_rate": 1.315435210782817e-07, "loss": 2.0637, "step": 8103 }, { "epoch": 0.9292512326568054, "grad_norm": 14.25, "learning_rate": 1.3112071088793975e-07, "loss": 2.3586, "step": 8104 }, { "epoch": 0.929365898406146, "grad_norm": 13.5625, "learning_rate": 1.3069857226668648e-07, "loss": 1.5423, "step": 8105 }, { "epoch": 0.9294805641554867, "grad_norm": 12.75, "learning_rate": 1.3027710527274806e-07, "loss": 2.0134, "step": 8106 }, { "epoch": 0.9295952299048275, "grad_norm": 15.0625, "learning_rate": 1.2985630996425747e-07, "loss": 1.491, "step": 8107 }, { "epoch": 0.9297098956541681, "grad_norm": 19.125, "learning_rate": 1.2943618639925547e-07, "loss": 1.8038, "step": 8108 }, { "epoch": 0.9298245614035088, "grad_norm": 20.375, "learning_rate": 1.2901673463569075e-07, "loss": 1.9336, "step": 8109 }, { "epoch": 0.9299392271528495, "grad_norm": 16.75, "learning_rate": 1.2859795473141702e-07, "loss": 1.999, "step": 8110 }, { "epoch": 0.9300538929021901, "grad_norm": 13.125, "learning_rate": 1.2817984674419915e-07, "loss": 1.7513, "step": 8111 }, { "epoch": 0.9301685586515308, "grad_norm": 15.5, "learning_rate": 1.2776241073170603e-07, "loss": 1.5086, "step": 8112 }, { "epoch": 0.9302832244008714, "grad_norm": 16.125, "learning_rate": 1.2734564675151439e-07, "loss": 1.8857, "step": 8113 }, { "epoch": 0.9303978901502121, "grad_norm": 12.5, "learning_rate": 1.269295548611099e-07, "loss": 1.7038, "step": 8114 }, { "epoch": 0.9305125558995528, "grad_norm": 15.4375, "learning_rate": 1.265141351178839e-07, "loss": 1.5775, "step": 8115 }, { "epoch": 0.9306272216488934, "grad_norm": 13.5, "learning_rate": 1.2609938757913553e-07, "loss": 1.6842, "step": 8116 }, { "epoch": 0.9307418873982342, "grad_norm": 14.5, "learning_rate": 1.2568531230207238e-07, "loss": 1.7178, "step": 8117 }, { "epoch": 0.9308565531475749, "grad_norm": 13.125, "learning_rate": 1.2527190934380763e-07, "loss": 1.3409, "step": 8118 }, { "epoch": 0.9309712188969155, "grad_norm": 18.125, "learning_rate": 1.248591787613618e-07, "loss": 1.7998, "step": 8119 }, { "epoch": 0.9310858846462562, "grad_norm": 14.0625, "learning_rate": 1.244471206116632e-07, "loss": 2.095, "step": 8120 }, { "epoch": 0.9312005503955968, "grad_norm": 15.5625, "learning_rate": 1.2403573495154863e-07, "loss": 1.5119, "step": 8121 }, { "epoch": 0.9313152161449375, "grad_norm": 14.625, "learning_rate": 1.2362502183775939e-07, "loss": 1.6729, "step": 8122 }, { "epoch": 0.9314298818942782, "grad_norm": 33.75, "learning_rate": 1.2321498132694677e-07, "loss": 1.8015, "step": 8123 }, { "epoch": 0.9315445476436188, "grad_norm": 27.75, "learning_rate": 1.2280561347566778e-07, "loss": 2.1755, "step": 8124 }, { "epoch": 0.9316592133929595, "grad_norm": 15.6875, "learning_rate": 1.223969183403856e-07, "loss": 1.4493, "step": 8125 }, { "epoch": 0.9317738791423001, "grad_norm": 15.3125, "learning_rate": 1.2198889597747344e-07, "loss": 1.6007, "step": 8126 }, { "epoch": 0.9318885448916409, "grad_norm": 13.875, "learning_rate": 1.2158154644321018e-07, "loss": 1.6483, "step": 8127 }, { "epoch": 0.9320032106409816, "grad_norm": 29.875, "learning_rate": 1.211748697937809e-07, "loss": 1.871, "step": 8128 }, { "epoch": 0.9321178763903222, "grad_norm": 14.4375, "learning_rate": 1.207688660852796e-07, "loss": 1.596, "step": 8129 }, { "epoch": 0.9322325421396629, "grad_norm": 14.875, "learning_rate": 1.2036353537370648e-07, "loss": 1.9333, "step": 8130 }, { "epoch": 0.9323472078890036, "grad_norm": 16.625, "learning_rate": 1.199588777149696e-07, "loss": 2.0218, "step": 8131 }, { "epoch": 0.9324618736383442, "grad_norm": 13.125, "learning_rate": 1.1955489316488322e-07, "loss": 1.6303, "step": 8132 }, { "epoch": 0.9325765393876849, "grad_norm": 12.375, "learning_rate": 1.1915158177916996e-07, "loss": 1.7167, "step": 8133 }, { "epoch": 0.9326912051370255, "grad_norm": 13.3125, "learning_rate": 1.1874894361345868e-07, "loss": 1.5987, "step": 8134 }, { "epoch": 0.9328058708863662, "grad_norm": 14.125, "learning_rate": 1.1834697872328493e-07, "loss": 2.1778, "step": 8135 }, { "epoch": 0.932920536635707, "grad_norm": 14.125, "learning_rate": 1.1794568716409272e-07, "loss": 2.0892, "step": 8136 }, { "epoch": 0.9330352023850476, "grad_norm": 13.8125, "learning_rate": 1.1754506899123274e-07, "loss": 1.9907, "step": 8137 }, { "epoch": 0.9331498681343883, "grad_norm": 13.1875, "learning_rate": 1.1714512425996249e-07, "loss": 1.761, "step": 8138 }, { "epoch": 0.9332645338837289, "grad_norm": 14.6875, "learning_rate": 1.1674585302544728e-07, "loss": 2.1738, "step": 8139 }, { "epoch": 0.9333791996330696, "grad_norm": 11.9375, "learning_rate": 1.1634725534275693e-07, "loss": 1.4768, "step": 8140 }, { "epoch": 0.9334938653824103, "grad_norm": 13.6875, "learning_rate": 1.1594933126687303e-07, "loss": 1.6736, "step": 8141 }, { "epoch": 0.9336085311317509, "grad_norm": 14.8125, "learning_rate": 1.1555208085268e-07, "loss": 1.8164, "step": 8142 }, { "epoch": 0.9337231968810916, "grad_norm": 13.8125, "learning_rate": 1.1515550415497123e-07, "loss": 1.6771, "step": 8143 }, { "epoch": 0.9338378626304323, "grad_norm": 14.625, "learning_rate": 1.1475960122844743e-07, "loss": 1.6664, "step": 8144 }, { "epoch": 0.9339525283797729, "grad_norm": 16.25, "learning_rate": 1.1436437212771489e-07, "loss": 1.449, "step": 8145 }, { "epoch": 0.9340671941291137, "grad_norm": 13.875, "learning_rate": 1.1396981690728893e-07, "loss": 1.7123, "step": 8146 }, { "epoch": 0.9341818598784543, "grad_norm": 18.25, "learning_rate": 1.1357593562159097e-07, "loss": 2.1389, "step": 8147 }, { "epoch": 0.934296525627795, "grad_norm": 17.0, "learning_rate": 1.1318272832494814e-07, "loss": 1.6559, "step": 8148 }, { "epoch": 0.9344111913771357, "grad_norm": 13.1875, "learning_rate": 1.1279019507159705e-07, "loss": 1.8743, "step": 8149 }, { "epoch": 0.9345258571264763, "grad_norm": 14.0, "learning_rate": 1.1239833591567995e-07, "loss": 1.4511, "step": 8150 }, { "epoch": 0.934640522875817, "grad_norm": 13.125, "learning_rate": 1.1200715091124636e-07, "loss": 1.799, "step": 8151 }, { "epoch": 0.9347551886251577, "grad_norm": 15.5, "learning_rate": 1.1161664011225204e-07, "loss": 1.6188, "step": 8152 }, { "epoch": 0.9348698543744983, "grad_norm": 18.5, "learning_rate": 1.1122680357256221e-07, "loss": 1.6913, "step": 8153 }, { "epoch": 0.934984520123839, "grad_norm": 14.9375, "learning_rate": 1.1083764134594555e-07, "loss": 1.4061, "step": 8154 }, { "epoch": 0.9350991858731796, "grad_norm": 13.625, "learning_rate": 1.104491534860802e-07, "loss": 1.4617, "step": 8155 }, { "epoch": 0.9352138516225204, "grad_norm": 18.625, "learning_rate": 1.1006134004655112e-07, "loss": 1.8845, "step": 8156 }, { "epoch": 0.9353285173718611, "grad_norm": 27.625, "learning_rate": 1.0967420108084992e-07, "loss": 1.9655, "step": 8157 }, { "epoch": 0.9354431831212017, "grad_norm": 13.5, "learning_rate": 1.092877366423739e-07, "loss": 1.5383, "step": 8158 }, { "epoch": 0.9355578488705424, "grad_norm": 15.0625, "learning_rate": 1.0890194678442989e-07, "loss": 1.3996, "step": 8159 }, { "epoch": 0.935672514619883, "grad_norm": 18.375, "learning_rate": 1.0851683156022862e-07, "loss": 2.4259, "step": 8160 }, { "epoch": 0.9357871803692237, "grad_norm": 12.875, "learning_rate": 1.0813239102289041e-07, "loss": 2.0286, "step": 8161 }, { "epoch": 0.9359018461185644, "grad_norm": 18.0, "learning_rate": 1.0774862522544228e-07, "loss": 1.5939, "step": 8162 }, { "epoch": 0.936016511867905, "grad_norm": 17.0, "learning_rate": 1.0736553422081575e-07, "loss": 2.1974, "step": 8163 }, { "epoch": 0.9361311776172457, "grad_norm": 12.0625, "learning_rate": 1.0698311806185136e-07, "loss": 1.62, "step": 8164 }, { "epoch": 0.9362458433665864, "grad_norm": 17.25, "learning_rate": 1.0660137680129689e-07, "loss": 1.9784, "step": 8165 }, { "epoch": 0.936360509115927, "grad_norm": 14.0625, "learning_rate": 1.0622031049180581e-07, "loss": 1.8095, "step": 8166 }, { "epoch": 0.9364751748652678, "grad_norm": 14.1875, "learning_rate": 1.0583991918593883e-07, "loss": 1.7383, "step": 8167 }, { "epoch": 0.9365898406146084, "grad_norm": 14.4375, "learning_rate": 1.0546020293616454e-07, "loss": 1.6949, "step": 8168 }, { "epoch": 0.9367045063639491, "grad_norm": 16.75, "learning_rate": 1.0508116179485606e-07, "loss": 1.71, "step": 8169 }, { "epoch": 0.9368191721132898, "grad_norm": 13.0, "learning_rate": 1.0470279581429543e-07, "loss": 1.6338, "step": 8170 }, { "epoch": 0.9369338378626304, "grad_norm": 17.125, "learning_rate": 1.0432510504667203e-07, "loss": 1.8843, "step": 8171 }, { "epoch": 0.9370485036119711, "grad_norm": 17.5, "learning_rate": 1.0394808954408032e-07, "loss": 1.6249, "step": 8172 }, { "epoch": 0.9371631693613117, "grad_norm": 12.625, "learning_rate": 1.0357174935852199e-07, "loss": 1.8703, "step": 8173 }, { "epoch": 0.9372778351106524, "grad_norm": 17.75, "learning_rate": 1.0319608454190722e-07, "loss": 1.8329, "step": 8174 }, { "epoch": 0.9373925008599931, "grad_norm": 15.5, "learning_rate": 1.0282109514604954e-07, "loss": 1.5197, "step": 8175 }, { "epoch": 0.9375071666093338, "grad_norm": 15.125, "learning_rate": 1.024467812226737e-07, "loss": 1.5033, "step": 8176 }, { "epoch": 0.9376218323586745, "grad_norm": 11.4375, "learning_rate": 1.0207314282340952e-07, "loss": 1.5625, "step": 8177 }, { "epoch": 0.9377364981080152, "grad_norm": 16.75, "learning_rate": 1.0170017999979132e-07, "loss": 2.124, "step": 8178 }, { "epoch": 0.9378511638573558, "grad_norm": 20.0, "learning_rate": 1.0132789280326294e-07, "loss": 1.6104, "step": 8179 }, { "epoch": 0.9379658296066965, "grad_norm": 14.0625, "learning_rate": 1.0095628128517498e-07, "loss": 1.7638, "step": 8180 }, { "epoch": 0.9380804953560371, "grad_norm": 17.25, "learning_rate": 1.0058534549678368e-07, "loss": 2.0331, "step": 8181 }, { "epoch": 0.9381951611053778, "grad_norm": 15.6875, "learning_rate": 1.0021508548925308e-07, "loss": 1.7923, "step": 8182 }, { "epoch": 0.9383098268547185, "grad_norm": 14.4375, "learning_rate": 9.98455013136529e-08, "loss": 1.6159, "step": 8183 }, { "epoch": 0.9384244926040591, "grad_norm": 34.25, "learning_rate": 9.947659302096014e-08, "loss": 1.6987, "step": 8184 }, { "epoch": 0.9385391583533998, "grad_norm": 14.25, "learning_rate": 9.910836066205854e-08, "loss": 1.4172, "step": 8185 }, { "epoch": 0.9386538241027406, "grad_norm": 16.875, "learning_rate": 9.874080428773969e-08, "loss": 2.0751, "step": 8186 }, { "epoch": 0.9387684898520812, "grad_norm": 15.375, "learning_rate": 9.837392394870026e-08, "loss": 1.9192, "step": 8187 }, { "epoch": 0.9388831556014219, "grad_norm": 17.5, "learning_rate": 9.800771969554424e-08, "loss": 1.7014, "step": 8188 }, { "epoch": 0.9389978213507625, "grad_norm": 15.6875, "learning_rate": 9.764219157878341e-08, "loss": 1.42, "step": 8189 }, { "epoch": 0.9391124871001032, "grad_norm": 17.875, "learning_rate": 9.727733964883357e-08, "loss": 1.6845, "step": 8190 }, { "epoch": 0.9392271528494439, "grad_norm": 19.5, "learning_rate": 9.691316395602168e-08, "loss": 1.6631, "step": 8191 }, { "epoch": 0.9393418185987845, "grad_norm": 15.3125, "learning_rate": 9.654966455057701e-08, "loss": 1.9155, "step": 8192 }, { "epoch": 0.9394564843481252, "grad_norm": 11.6875, "learning_rate": 9.618684148263779e-08, "loss": 1.6659, "step": 8193 }, { "epoch": 0.9395711500974658, "grad_norm": 17.625, "learning_rate": 9.582469480224842e-08, "loss": 1.8888, "step": 8194 }, { "epoch": 0.9396858158468065, "grad_norm": 13.25, "learning_rate": 9.546322455936063e-08, "loss": 1.7822, "step": 8195 }, { "epoch": 0.9398004815961473, "grad_norm": 15.1875, "learning_rate": 9.510243080383175e-08, "loss": 1.5856, "step": 8196 }, { "epoch": 0.9399151473454879, "grad_norm": 13.9375, "learning_rate": 9.474231358542751e-08, "loss": 1.8681, "step": 8197 }, { "epoch": 0.9400298130948286, "grad_norm": 12.9375, "learning_rate": 9.438287295381765e-08, "loss": 1.4812, "step": 8198 }, { "epoch": 0.9401444788441693, "grad_norm": 13.8125, "learning_rate": 9.402410895858082e-08, "loss": 1.6455, "step": 8199 }, { "epoch": 0.9402591445935099, "grad_norm": 16.625, "learning_rate": 9.366602164920191e-08, "loss": 1.8775, "step": 8200 }, { "epoch": 0.9403738103428506, "grad_norm": 13.625, "learning_rate": 9.330861107507194e-08, "loss": 1.602, "step": 8201 }, { "epoch": 0.9404884760921912, "grad_norm": 15.5625, "learning_rate": 9.295187728548927e-08, "loss": 1.9846, "step": 8202 }, { "epoch": 0.9406031418415319, "grad_norm": 14.875, "learning_rate": 9.259582032965842e-08, "loss": 1.8031, "step": 8203 }, { "epoch": 0.9407178075908726, "grad_norm": 16.5, "learning_rate": 9.22404402566901e-08, "loss": 1.6689, "step": 8204 }, { "epoch": 0.9408324733402132, "grad_norm": 16.0, "learning_rate": 9.188573711560233e-08, "loss": 1.7927, "step": 8205 }, { "epoch": 0.940947139089554, "grad_norm": 13.6875, "learning_rate": 9.153171095532098e-08, "loss": 1.7385, "step": 8206 }, { "epoch": 0.9410618048388946, "grad_norm": 12.75, "learning_rate": 9.117836182467532e-08, "loss": 1.4048, "step": 8207 }, { "epoch": 0.9411764705882353, "grad_norm": 21.875, "learning_rate": 9.082568977240469e-08, "loss": 2.3729, "step": 8208 }, { "epoch": 0.941291136337576, "grad_norm": 12.5, "learning_rate": 9.04736948471524e-08, "loss": 1.5616, "step": 8209 }, { "epoch": 0.9414058020869166, "grad_norm": 13.625, "learning_rate": 9.012237709746963e-08, "loss": 1.8534, "step": 8210 }, { "epoch": 0.9415204678362573, "grad_norm": 14.3125, "learning_rate": 8.977173657181426e-08, "loss": 1.7127, "step": 8211 }, { "epoch": 0.941635133585598, "grad_norm": 14.5, "learning_rate": 8.942177331855095e-08, "loss": 1.6323, "step": 8212 }, { "epoch": 0.9417497993349386, "grad_norm": 12.0625, "learning_rate": 8.907248738594942e-08, "loss": 1.4972, "step": 8213 }, { "epoch": 0.9418644650842793, "grad_norm": 13.875, "learning_rate": 8.872387882218725e-08, "loss": 1.6161, "step": 8214 }, { "epoch": 0.94197913083362, "grad_norm": 13.3125, "learning_rate": 8.837594767534874e-08, "loss": 1.4192, "step": 8215 }, { "epoch": 0.9420937965829607, "grad_norm": 17.625, "learning_rate": 8.802869399342384e-08, "loss": 2.0239, "step": 8216 }, { "epoch": 0.9422084623323014, "grad_norm": 13.25, "learning_rate": 8.768211782431035e-08, "loss": 1.5638, "step": 8217 }, { "epoch": 0.942323128081642, "grad_norm": 17.25, "learning_rate": 8.73362192158117e-08, "loss": 2.0911, "step": 8218 }, { "epoch": 0.9424377938309827, "grad_norm": 14.0625, "learning_rate": 8.69909982156375e-08, "loss": 1.5379, "step": 8219 }, { "epoch": 0.9425524595803234, "grad_norm": 13.5, "learning_rate": 8.664645487140411e-08, "loss": 1.6593, "step": 8220 }, { "epoch": 0.942667125329664, "grad_norm": 11.875, "learning_rate": 8.630258923063572e-08, "loss": 1.4581, "step": 8221 }, { "epoch": 0.9427817910790047, "grad_norm": 15.5625, "learning_rate": 8.595940134076108e-08, "loss": 2.0333, "step": 8222 }, { "epoch": 0.9428964568283453, "grad_norm": 16.125, "learning_rate": 8.561689124911732e-08, "loss": 2.1848, "step": 8223 }, { "epoch": 0.943011122577686, "grad_norm": 13.6875, "learning_rate": 8.527505900294719e-08, "loss": 1.5438, "step": 8224 }, { "epoch": 0.9431257883270268, "grad_norm": 15.9375, "learning_rate": 8.493390464939854e-08, "loss": 1.811, "step": 8225 }, { "epoch": 0.9432404540763674, "grad_norm": 15.75, "learning_rate": 8.459342823552819e-08, "loss": 1.8788, "step": 8226 }, { "epoch": 0.9433551198257081, "grad_norm": 15.5, "learning_rate": 8.42536298082991e-08, "loss": 1.7872, "step": 8227 }, { "epoch": 0.9434697855750487, "grad_norm": 13.375, "learning_rate": 8.391450941457824e-08, "loss": 1.8775, "step": 8228 }, { "epoch": 0.9435844513243894, "grad_norm": 14.1875, "learning_rate": 8.357606710114207e-08, "loss": 1.8359, "step": 8229 }, { "epoch": 0.9436991170737301, "grad_norm": 14.3125, "learning_rate": 8.32383029146716e-08, "loss": 1.6839, "step": 8230 }, { "epoch": 0.9438137828230707, "grad_norm": 13.75, "learning_rate": 8.290121690175568e-08, "loss": 1.7002, "step": 8231 }, { "epoch": 0.9439284485724114, "grad_norm": 14.0625, "learning_rate": 8.25648091088882e-08, "loss": 1.9343, "step": 8232 }, { "epoch": 0.9440431143217521, "grad_norm": 14.875, "learning_rate": 8.222907958247095e-08, "loss": 1.8403, "step": 8233 }, { "epoch": 0.9441577800710927, "grad_norm": 11.8125, "learning_rate": 8.189402836881022e-08, "loss": 1.3444, "step": 8234 }, { "epoch": 0.9442724458204335, "grad_norm": 15.8125, "learning_rate": 8.15596555141207e-08, "loss": 1.8466, "step": 8235 }, { "epoch": 0.9443871115697741, "grad_norm": 13.1875, "learning_rate": 8.122596106452274e-08, "loss": 1.9053, "step": 8236 }, { "epoch": 0.9445017773191148, "grad_norm": 11.1875, "learning_rate": 8.089294506604339e-08, "loss": 1.9575, "step": 8237 }, { "epoch": 0.9446164430684555, "grad_norm": 13.1875, "learning_rate": 8.056060756461537e-08, "loss": 1.464, "step": 8238 }, { "epoch": 0.9447311088177961, "grad_norm": 17.125, "learning_rate": 8.022894860607866e-08, "loss": 1.5483, "step": 8239 }, { "epoch": 0.9448457745671368, "grad_norm": 15.125, "learning_rate": 7.98979682361789e-08, "loss": 1.7656, "step": 8240 }, { "epoch": 0.9449604403164775, "grad_norm": 12.0625, "learning_rate": 7.956766650056902e-08, "loss": 1.5595, "step": 8241 }, { "epoch": 0.9450751060658181, "grad_norm": 12.625, "learning_rate": 7.923804344480757e-08, "loss": 1.6003, "step": 8242 }, { "epoch": 0.9451897718151588, "grad_norm": 17.25, "learning_rate": 7.890909911435929e-08, "loss": 1.8739, "step": 8243 }, { "epoch": 0.9453044375644994, "grad_norm": 18.0, "learning_rate": 7.858083355459678e-08, "loss": 1.9228, "step": 8244 }, { "epoch": 0.9454191033138402, "grad_norm": 15.25, "learning_rate": 7.825324681079715e-08, "loss": 1.5403, "step": 8245 }, { "epoch": 0.9455337690631809, "grad_norm": 14.3125, "learning_rate": 7.792633892814482e-08, "loss": 1.4665, "step": 8246 }, { "epoch": 0.9456484348125215, "grad_norm": 15.0625, "learning_rate": 7.76001099517315e-08, "loss": 1.6458, "step": 8247 }, { "epoch": 0.9457631005618622, "grad_norm": 16.375, "learning_rate": 7.727455992655342e-08, "loss": 1.8613, "step": 8248 }, { "epoch": 0.9458777663112028, "grad_norm": 13.375, "learning_rate": 7.694968889751409e-08, "loss": 1.6225, "step": 8249 }, { "epoch": 0.9459924320605435, "grad_norm": 14.8125, "learning_rate": 7.662549690942267e-08, "loss": 1.7315, "step": 8250 }, { "epoch": 0.9461070978098842, "grad_norm": 15.5, "learning_rate": 7.630198400699673e-08, "loss": 1.9882, "step": 8251 }, { "epoch": 0.9462217635592248, "grad_norm": 14.875, "learning_rate": 7.597915023485725e-08, "loss": 1.9036, "step": 8252 }, { "epoch": 0.9463364293085655, "grad_norm": 14.75, "learning_rate": 7.565699563753415e-08, "loss": 1.5672, "step": 8253 }, { "epoch": 0.9464510950579063, "grad_norm": 12.5, "learning_rate": 7.533552025946189e-08, "loss": 1.6394, "step": 8254 }, { "epoch": 0.9465657608072469, "grad_norm": 14.9375, "learning_rate": 7.501472414498112e-08, "loss": 1.6514, "step": 8255 }, { "epoch": 0.9466804265565876, "grad_norm": 13.8125, "learning_rate": 7.469460733834145e-08, "loss": 1.6392, "step": 8256 }, { "epoch": 0.9467950923059282, "grad_norm": 15.125, "learning_rate": 7.43751698836953e-08, "loss": 1.6593, "step": 8257 }, { "epoch": 0.9469097580552689, "grad_norm": 12.1875, "learning_rate": 7.405641182510359e-08, "loss": 1.8517, "step": 8258 }, { "epoch": 0.9470244238046096, "grad_norm": 13.9375, "learning_rate": 7.373833320653334e-08, "loss": 1.8744, "step": 8259 }, { "epoch": 0.9471390895539502, "grad_norm": 14.4375, "learning_rate": 7.342093407185613e-08, "loss": 1.9508, "step": 8260 }, { "epoch": 0.9472537553032909, "grad_norm": 12.5, "learning_rate": 7.310421446485195e-08, "loss": 1.5175, "step": 8261 }, { "epoch": 0.9473684210526315, "grad_norm": 16.375, "learning_rate": 7.278817442920693e-08, "loss": 1.81, "step": 8262 }, { "epoch": 0.9474830868019722, "grad_norm": 13.5625, "learning_rate": 7.247281400851181e-08, "loss": 1.9547, "step": 8263 }, { "epoch": 0.947597752551313, "grad_norm": 12.1875, "learning_rate": 7.215813324626453e-08, "loss": 1.5628, "step": 8264 }, { "epoch": 0.9477124183006536, "grad_norm": 13.6875, "learning_rate": 7.184413218586928e-08, "loss": 1.8145, "step": 8265 }, { "epoch": 0.9478270840499943, "grad_norm": 17.625, "learning_rate": 7.153081087063751e-08, "loss": 2.0114, "step": 8266 }, { "epoch": 0.947941749799335, "grad_norm": 14.75, "learning_rate": 7.121816934378467e-08, "loss": 2.012, "step": 8267 }, { "epoch": 0.9480564155486756, "grad_norm": 18.5, "learning_rate": 7.09062076484346e-08, "loss": 2.1176, "step": 8268 }, { "epoch": 0.9481710812980163, "grad_norm": 13.6875, "learning_rate": 7.059492582761618e-08, "loss": 1.7889, "step": 8269 }, { "epoch": 0.9482857470473569, "grad_norm": 19.0, "learning_rate": 7.028432392426398e-08, "loss": 1.4442, "step": 8270 }, { "epoch": 0.9484004127966976, "grad_norm": 15.3125, "learning_rate": 6.997440198122152e-08, "loss": 2.2051, "step": 8271 }, { "epoch": 0.9485150785460383, "grad_norm": 14.0625, "learning_rate": 6.966516004123458e-08, "loss": 1.7101, "step": 8272 }, { "epoch": 0.9486297442953789, "grad_norm": 13.625, "learning_rate": 6.93565981469585e-08, "loss": 1.4283, "step": 8273 }, { "epoch": 0.9487444100447197, "grad_norm": 16.25, "learning_rate": 6.904871634095312e-08, "loss": 1.938, "step": 8274 }, { "epoch": 0.9488590757940604, "grad_norm": 26.5, "learning_rate": 6.87415146656839e-08, "loss": 1.8818, "step": 8275 }, { "epoch": 0.948973741543401, "grad_norm": 14.75, "learning_rate": 6.843499316352531e-08, "loss": 1.5976, "step": 8276 }, { "epoch": 0.9490884072927417, "grad_norm": 14.6875, "learning_rate": 6.812915187675517e-08, "loss": 1.5689, "step": 8277 }, { "epoch": 0.9492030730420823, "grad_norm": 17.25, "learning_rate": 6.782399084755864e-08, "loss": 1.651, "step": 8278 }, { "epoch": 0.949317738791423, "grad_norm": 14.4375, "learning_rate": 6.751951011802649e-08, "loss": 1.753, "step": 8279 }, { "epoch": 0.9494324045407637, "grad_norm": 13.1875, "learning_rate": 6.721570973015623e-08, "loss": 1.7797, "step": 8280 }, { "epoch": 0.9495470702901043, "grad_norm": 17.25, "learning_rate": 6.691258972585157e-08, "loss": 2.0396, "step": 8281 }, { "epoch": 0.949661736039445, "grad_norm": 14.0, "learning_rate": 6.661015014692185e-08, "loss": 1.8158, "step": 8282 }, { "epoch": 0.9497764017887856, "grad_norm": 17.75, "learning_rate": 6.630839103508313e-08, "loss": 1.8422, "step": 8283 }, { "epoch": 0.9498910675381264, "grad_norm": 15.8125, "learning_rate": 6.600731243195713e-08, "loss": 1.9506, "step": 8284 }, { "epoch": 0.9500057332874671, "grad_norm": 16.25, "learning_rate": 6.570691437907173e-08, "loss": 1.9744, "step": 8285 }, { "epoch": 0.9501203990368077, "grad_norm": 12.375, "learning_rate": 6.540719691786102e-08, "loss": 1.8839, "step": 8286 }, { "epoch": 0.9502350647861484, "grad_norm": 14.375, "learning_rate": 6.510816008966581e-08, "loss": 1.6708, "step": 8287 }, { "epoch": 0.9503497305354891, "grad_norm": 19.875, "learning_rate": 6.480980393573255e-08, "loss": 2.4268, "step": 8288 }, { "epoch": 0.9504643962848297, "grad_norm": 14.8125, "learning_rate": 6.451212849721389e-08, "loss": 1.512, "step": 8289 }, { "epoch": 0.9505790620341704, "grad_norm": 11.125, "learning_rate": 6.421513381516753e-08, "loss": 1.7912, "step": 8290 }, { "epoch": 0.950693727783511, "grad_norm": 14.4375, "learning_rate": 6.391881993055903e-08, "loss": 1.7824, "step": 8291 }, { "epoch": 0.9508083935328517, "grad_norm": 15.5, "learning_rate": 6.362318688425905e-08, "loss": 1.9146, "step": 8292 }, { "epoch": 0.9509230592821925, "grad_norm": 13.9375, "learning_rate": 6.332823471704441e-08, "loss": 1.7595, "step": 8293 }, { "epoch": 0.951037725031533, "grad_norm": 12.75, "learning_rate": 6.303396346959867e-08, "loss": 1.8088, "step": 8294 }, { "epoch": 0.9511523907808738, "grad_norm": 15.75, "learning_rate": 6.274037318250992e-08, "loss": 1.5574, "step": 8295 }, { "epoch": 0.9512670565302144, "grad_norm": 16.5, "learning_rate": 6.24474638962741e-08, "loss": 2.0306, "step": 8296 }, { "epoch": 0.9513817222795551, "grad_norm": 14.125, "learning_rate": 6.215523565129277e-08, "loss": 1.9253, "step": 8297 }, { "epoch": 0.9514963880288958, "grad_norm": 13.6875, "learning_rate": 6.18636884878726e-08, "loss": 1.8885, "step": 8298 }, { "epoch": 0.9516110537782364, "grad_norm": 11.9375, "learning_rate": 6.157282244622697e-08, "loss": 1.5525, "step": 8299 }, { "epoch": 0.9517257195275771, "grad_norm": 11.0625, "learning_rate": 6.128263756647601e-08, "loss": 1.4535, "step": 8300 }, { "epoch": 0.9518403852769178, "grad_norm": 13.0, "learning_rate": 6.099313388864381e-08, "loss": 1.5052, "step": 8301 }, { "epoch": 0.9519550510262584, "grad_norm": 13.25, "learning_rate": 6.070431145266342e-08, "loss": 1.6277, "step": 8302 }, { "epoch": 0.9520697167755992, "grad_norm": 14.9375, "learning_rate": 6.041617029837188e-08, "loss": 2.0713, "step": 8303 }, { "epoch": 0.9521843825249398, "grad_norm": 13.875, "learning_rate": 6.012871046551238e-08, "loss": 1.6997, "step": 8304 }, { "epoch": 0.9522990482742805, "grad_norm": 12.5625, "learning_rate": 5.984193199373433e-08, "loss": 1.4904, "step": 8305 }, { "epoch": 0.9524137140236212, "grad_norm": 16.5, "learning_rate": 5.955583492259442e-08, "loss": 1.9313, "step": 8306 }, { "epoch": 0.9525283797729618, "grad_norm": 13.25, "learning_rate": 5.9270419291553305e-08, "loss": 1.4475, "step": 8307 }, { "epoch": 0.9526430455223025, "grad_norm": 13.3125, "learning_rate": 5.898568513997893e-08, "loss": 1.692, "step": 8308 }, { "epoch": 0.9527577112716432, "grad_norm": 14.3125, "learning_rate": 5.8701632507145446e-08, "loss": 1.6977, "step": 8309 }, { "epoch": 0.9528723770209838, "grad_norm": 13.8125, "learning_rate": 5.841826143223095e-08, "loss": 1.755, "step": 8310 }, { "epoch": 0.9529870427703245, "grad_norm": 29.125, "learning_rate": 5.813557195432307e-08, "loss": 1.786, "step": 8311 }, { "epoch": 0.9531017085196651, "grad_norm": 13.1875, "learning_rate": 5.7853564112412273e-08, "loss": 1.6259, "step": 8312 }, { "epoch": 0.9532163742690059, "grad_norm": 15.75, "learning_rate": 5.7572237945396905e-08, "loss": 2.2585, "step": 8313 }, { "epoch": 0.9533310400183466, "grad_norm": 13.5, "learning_rate": 5.7291593492079245e-08, "loss": 1.3923, "step": 8314 }, { "epoch": 0.9534457057676872, "grad_norm": 12.9375, "learning_rate": 5.701163079117e-08, "loss": 1.7233, "step": 8315 }, { "epoch": 0.9535603715170279, "grad_norm": 12.4375, "learning_rate": 5.6732349881284396e-08, "loss": 1.7699, "step": 8316 }, { "epoch": 0.9536750372663685, "grad_norm": 18.5, "learning_rate": 5.645375080094439e-08, "loss": 1.6427, "step": 8317 }, { "epoch": 0.9537897030157092, "grad_norm": 25.25, "learning_rate": 5.6175833588576476e-08, "loss": 1.8228, "step": 8318 }, { "epoch": 0.9539043687650499, "grad_norm": 13.75, "learning_rate": 5.589859828251443e-08, "loss": 1.5995, "step": 8319 }, { "epoch": 0.9540190345143905, "grad_norm": 16.75, "learning_rate": 5.5622044920997096e-08, "loss": 1.7523, "step": 8320 }, { "epoch": 0.9541337002637312, "grad_norm": 16.0, "learning_rate": 5.53461735421712e-08, "loss": 1.8191, "step": 8321 }, { "epoch": 0.954248366013072, "grad_norm": 35.0, "learning_rate": 5.50709841840863e-08, "loss": 1.9855, "step": 8322 }, { "epoch": 0.9543630317624126, "grad_norm": 26.625, "learning_rate": 5.479647688470035e-08, "loss": 1.7416, "step": 8323 }, { "epoch": 0.9544776975117533, "grad_norm": 13.6875, "learning_rate": 5.4522651681876983e-08, "loss": 1.7801, "step": 8324 }, { "epoch": 0.9545923632610939, "grad_norm": 12.625, "learning_rate": 5.4249508613383185e-08, "loss": 1.616, "step": 8325 }, { "epoch": 0.9547070290104346, "grad_norm": 13.0625, "learning_rate": 5.397704771689549e-08, "loss": 1.5024, "step": 8326 }, { "epoch": 0.9548216947597753, "grad_norm": 17.5, "learning_rate": 5.370526902999496e-08, "loss": 1.7119, "step": 8327 }, { "epoch": 0.9549363605091159, "grad_norm": 13.75, "learning_rate": 5.3434172590167145e-08, "loss": 1.5399, "step": 8328 }, { "epoch": 0.9550510262584566, "grad_norm": 13.4375, "learning_rate": 5.316375843480492e-08, "loss": 1.5085, "step": 8329 }, { "epoch": 0.9551656920077972, "grad_norm": 13.875, "learning_rate": 5.289402660120735e-08, "loss": 1.7627, "step": 8330 }, { "epoch": 0.9552803577571379, "grad_norm": 13.0, "learning_rate": 5.262497712657799e-08, "loss": 1.7413, "step": 8331 }, { "epoch": 0.9553950235064786, "grad_norm": 15.625, "learning_rate": 5.2356610048027726e-08, "loss": 1.5359, "step": 8332 }, { "epoch": 0.9555096892558193, "grad_norm": 16.875, "learning_rate": 5.208892540257249e-08, "loss": 1.7542, "step": 8333 }, { "epoch": 0.95562435500516, "grad_norm": 17.125, "learning_rate": 5.182192322713442e-08, "loss": 2.0054, "step": 8334 }, { "epoch": 0.9557390207545007, "grad_norm": 13.25, "learning_rate": 5.155560355854072e-08, "loss": 1.549, "step": 8335 }, { "epoch": 0.9558536865038413, "grad_norm": 11.3125, "learning_rate": 5.128996643352591e-08, "loss": 1.883, "step": 8336 }, { "epoch": 0.955968352253182, "grad_norm": 13.75, "learning_rate": 5.102501188872899e-08, "loss": 1.9693, "step": 8337 }, { "epoch": 0.9560830180025226, "grad_norm": 14.0625, "learning_rate": 5.076073996069575e-08, "loss": 1.7238, "step": 8338 }, { "epoch": 0.9561976837518633, "grad_norm": 13.3125, "learning_rate": 5.049715068587757e-08, "loss": 1.6529, "step": 8339 }, { "epoch": 0.956312349501204, "grad_norm": 15.375, "learning_rate": 5.023424410063038e-08, "loss": 1.7472, "step": 8340 }, { "epoch": 0.9564270152505446, "grad_norm": 20.625, "learning_rate": 4.9972020241219054e-08, "loss": 2.2464, "step": 8341 }, { "epoch": 0.9565416809998853, "grad_norm": 13.5625, "learning_rate": 4.971047914381133e-08, "loss": 1.6153, "step": 8342 }, { "epoch": 0.9566563467492261, "grad_norm": 13.75, "learning_rate": 4.9449620844481685e-08, "loss": 1.6666, "step": 8343 }, { "epoch": 0.9567710124985667, "grad_norm": 11.5625, "learning_rate": 4.918944537921078e-08, "loss": 1.2697, "step": 8344 }, { "epoch": 0.9568856782479074, "grad_norm": 12.5, "learning_rate": 4.892995278388491e-08, "loss": 1.4042, "step": 8345 }, { "epoch": 0.957000343997248, "grad_norm": 20.0, "learning_rate": 4.867114309429599e-08, "loss": 1.7816, "step": 8346 }, { "epoch": 0.9571150097465887, "grad_norm": 14.125, "learning_rate": 4.8413016346142704e-08, "loss": 1.7091, "step": 8347 }, { "epoch": 0.9572296754959294, "grad_norm": 11.8125, "learning_rate": 4.8155572575027123e-08, "loss": 1.6379, "step": 8348 }, { "epoch": 0.95734434124527, "grad_norm": 10.9375, "learning_rate": 4.7898811816459723e-08, "loss": 1.557, "step": 8349 }, { "epoch": 0.9574590069946107, "grad_norm": 17.375, "learning_rate": 4.7642734105855514e-08, "loss": 1.9159, "step": 8350 }, { "epoch": 0.9575736727439513, "grad_norm": 13.3125, "learning_rate": 4.738733947853569e-08, "loss": 1.5018, "step": 8351 }, { "epoch": 0.957688338493292, "grad_norm": 15.375, "learning_rate": 4.7132627969727065e-08, "loss": 1.9649, "step": 8352 }, { "epoch": 0.9578030042426328, "grad_norm": 13.375, "learning_rate": 4.687859961456265e-08, "loss": 1.8062, "step": 8353 }, { "epoch": 0.9579176699919734, "grad_norm": 14.4375, "learning_rate": 4.6625254448079414e-08, "loss": 1.6477, "step": 8354 }, { "epoch": 0.9580323357413141, "grad_norm": 13.4375, "learning_rate": 4.6372592505222744e-08, "loss": 2.0887, "step": 8355 }, { "epoch": 0.9581470014906548, "grad_norm": 15.3125, "learning_rate": 4.612061382084254e-08, "loss": 1.7747, "step": 8356 }, { "epoch": 0.9582616672399954, "grad_norm": 12.5625, "learning_rate": 4.586931842969378e-08, "loss": 1.589, "step": 8357 }, { "epoch": 0.9583763329893361, "grad_norm": 13.125, "learning_rate": 4.561870636643817e-08, "loss": 1.9794, "step": 8358 }, { "epoch": 0.9584909987386767, "grad_norm": 11.6875, "learning_rate": 4.536877766564363e-08, "loss": 1.3366, "step": 8359 }, { "epoch": 0.9586056644880174, "grad_norm": 11.6875, "learning_rate": 4.511953236178146e-08, "loss": 1.9367, "step": 8360 }, { "epoch": 0.9587203302373581, "grad_norm": 12.0, "learning_rate": 4.487097048923139e-08, "loss": 1.6385, "step": 8361 }, { "epoch": 0.9588349959866987, "grad_norm": 16.625, "learning_rate": 4.462309208227822e-08, "loss": 1.9867, "step": 8362 }, { "epoch": 0.9589496617360395, "grad_norm": 12.9375, "learning_rate": 4.4375897175110705e-08, "loss": 1.5426, "step": 8363 }, { "epoch": 0.9590643274853801, "grad_norm": 13.5625, "learning_rate": 4.412938580182547e-08, "loss": 1.6106, "step": 8364 }, { "epoch": 0.9591789932347208, "grad_norm": 16.75, "learning_rate": 4.3883557996424187e-08, "loss": 2.2183, "step": 8365 }, { "epoch": 0.9592936589840615, "grad_norm": 24.375, "learning_rate": 4.363841379281364e-08, "loss": 1.6942, "step": 8366 }, { "epoch": 0.9594083247334021, "grad_norm": 13.125, "learning_rate": 4.339395322480733e-08, "loss": 1.5938, "step": 8367 }, { "epoch": 0.9595229904827428, "grad_norm": 14.125, "learning_rate": 4.315017632612384e-08, "loss": 2.183, "step": 8368 }, { "epoch": 0.9596376562320835, "grad_norm": 15.0625, "learning_rate": 4.2907083130387386e-08, "loss": 1.855, "step": 8369 }, { "epoch": 0.9597523219814241, "grad_norm": 15.5625, "learning_rate": 4.2664673671127256e-08, "loss": 2.1469, "step": 8370 }, { "epoch": 0.9598669877307648, "grad_norm": 14.625, "learning_rate": 4.242294798178059e-08, "loss": 1.648, "step": 8371 }, { "epoch": 0.9599816534801054, "grad_norm": 14.25, "learning_rate": 4.218190609568851e-08, "loss": 2.0272, "step": 8372 }, { "epoch": 0.9600963192294462, "grad_norm": 14.9375, "learning_rate": 4.194154804609718e-08, "loss": 1.8682, "step": 8373 }, { "epoch": 0.9602109849787869, "grad_norm": 15.1875, "learning_rate": 4.1701873866160646e-08, "loss": 2.044, "step": 8374 }, { "epoch": 0.9603256507281275, "grad_norm": 14.625, "learning_rate": 4.1462883588936355e-08, "loss": 1.5832, "step": 8375 }, { "epoch": 0.9604403164774682, "grad_norm": 14.8125, "learning_rate": 4.122457724738904e-08, "loss": 1.6987, "step": 8376 }, { "epoch": 0.9605549822268089, "grad_norm": 15.5, "learning_rate": 4.098695487438909e-08, "loss": 1.5827, "step": 8377 }, { "epoch": 0.9606696479761495, "grad_norm": 30.5, "learning_rate": 4.075001650271082e-08, "loss": 1.6202, "step": 8378 }, { "epoch": 0.9607843137254902, "grad_norm": 11.625, "learning_rate": 4.0513762165035884e-08, "loss": 1.4528, "step": 8379 }, { "epoch": 0.9608989794748308, "grad_norm": 17.5, "learning_rate": 4.027819189395099e-08, "loss": 1.8363, "step": 8380 }, { "epoch": 0.9610136452241715, "grad_norm": 15.3125, "learning_rate": 4.0043305721949036e-08, "loss": 1.8871, "step": 8381 }, { "epoch": 0.9611283109735123, "grad_norm": 12.3125, "learning_rate": 3.9809103681426875e-08, "loss": 1.5823, "step": 8382 }, { "epoch": 0.9612429767228529, "grad_norm": 13.375, "learning_rate": 3.957558580468979e-08, "loss": 1.7301, "step": 8383 }, { "epoch": 0.9613576424721936, "grad_norm": 14.3125, "learning_rate": 3.934275212394645e-08, "loss": 2.1133, "step": 8384 }, { "epoch": 0.9614723082215342, "grad_norm": 13.625, "learning_rate": 3.911060267131117e-08, "loss": 1.6569, "step": 8385 }, { "epoch": 0.9615869739708749, "grad_norm": 13.875, "learning_rate": 3.887913747880501e-08, "loss": 1.5455, "step": 8386 }, { "epoch": 0.9617016397202156, "grad_norm": 14.375, "learning_rate": 3.864835657835464e-08, "loss": 1.6316, "step": 8387 }, { "epoch": 0.9618163054695562, "grad_norm": 10.125, "learning_rate": 3.8418260001791273e-08, "loss": 1.6308, "step": 8388 }, { "epoch": 0.9619309712188969, "grad_norm": 18.125, "learning_rate": 3.818884778085285e-08, "loss": 1.9702, "step": 8389 }, { "epoch": 0.9620456369682376, "grad_norm": 13.0625, "learning_rate": 3.7960119947181853e-08, "loss": 1.7074, "step": 8390 }, { "epoch": 0.9621603027175782, "grad_norm": 16.5, "learning_rate": 3.7732076532327465e-08, "loss": 2.1731, "step": 8391 }, { "epoch": 0.962274968466919, "grad_norm": 28.125, "learning_rate": 3.7504717567743986e-08, "loss": 1.7551, "step": 8392 }, { "epoch": 0.9623896342162596, "grad_norm": 27.125, "learning_rate": 3.7278043084790217e-08, "loss": 1.7508, "step": 8393 }, { "epoch": 0.9625042999656003, "grad_norm": 16.5, "learning_rate": 3.705205311473281e-08, "loss": 1.7659, "step": 8394 }, { "epoch": 0.962618965714941, "grad_norm": 15.375, "learning_rate": 3.682674768874239e-08, "loss": 1.5544, "step": 8395 }, { "epoch": 0.9627336314642816, "grad_norm": 17.0, "learning_rate": 3.6602126837895215e-08, "loss": 1.9166, "step": 8396 }, { "epoch": 0.9628482972136223, "grad_norm": 17.125, "learning_rate": 3.637819059317371e-08, "loss": 2.1146, "step": 8397 }, { "epoch": 0.9629629629629629, "grad_norm": 19.5, "learning_rate": 3.61549389854654e-08, "loss": 1.8846, "step": 8398 }, { "epoch": 0.9630776287123036, "grad_norm": 13.75, "learning_rate": 3.593237204556399e-08, "loss": 1.5195, "step": 8399 }, { "epoch": 0.9631922944616443, "grad_norm": 15.375, "learning_rate": 3.5710489804167694e-08, "loss": 1.756, "step": 8400 }, { "epoch": 0.9633069602109849, "grad_norm": 18.5, "learning_rate": 3.548929229188203e-08, "loss": 1.9406, "step": 8401 }, { "epoch": 0.9634216259603257, "grad_norm": 17.625, "learning_rate": 3.526877953921592e-08, "loss": 1.9759, "step": 8402 }, { "epoch": 0.9635362917096664, "grad_norm": 10.4375, "learning_rate": 3.504895157658561e-08, "loss": 1.3208, "step": 8403 }, { "epoch": 0.963650957459007, "grad_norm": 15.0, "learning_rate": 3.482980843431127e-08, "loss": 2.0125, "step": 8404 }, { "epoch": 0.9637656232083477, "grad_norm": 12.75, "learning_rate": 3.46113501426204e-08, "loss": 1.4796, "step": 8405 }, { "epoch": 0.9638802889576883, "grad_norm": 14.375, "learning_rate": 3.4393576731645005e-08, "loss": 1.3069, "step": 8406 }, { "epoch": 0.963994954707029, "grad_norm": 15.5, "learning_rate": 3.4176488231422164e-08, "loss": 1.8192, "step": 8407 }, { "epoch": 0.9641096204563697, "grad_norm": 12.9375, "learning_rate": 3.396008467189627e-08, "loss": 1.8262, "step": 8408 }, { "epoch": 0.9642242862057103, "grad_norm": 18.0, "learning_rate": 3.3744366082915094e-08, "loss": 1.9044, "step": 8409 }, { "epoch": 0.964338951955051, "grad_norm": 11.375, "learning_rate": 3.352933249423262e-08, "loss": 1.5662, "step": 8410 }, { "epoch": 0.9644536177043918, "grad_norm": 14.625, "learning_rate": 3.3314983935509584e-08, "loss": 1.5097, "step": 8411 }, { "epoch": 0.9645682834537324, "grad_norm": 14.875, "learning_rate": 3.31013204363112e-08, "loss": 1.9711, "step": 8412 }, { "epoch": 0.9646829492030731, "grad_norm": 18.5, "learning_rate": 3.28883420261078e-08, "loss": 1.814, "step": 8413 }, { "epoch": 0.9647976149524137, "grad_norm": 12.9375, "learning_rate": 3.267604873427532e-08, "loss": 1.5601, "step": 8414 }, { "epoch": 0.9649122807017544, "grad_norm": 13.375, "learning_rate": 3.246444059009701e-08, "loss": 1.9476, "step": 8415 }, { "epoch": 0.9650269464510951, "grad_norm": 13.5, "learning_rate": 3.225351762275841e-08, "loss": 1.5224, "step": 8416 }, { "epoch": 0.9651416122004357, "grad_norm": 13.625, "learning_rate": 3.204327986135347e-08, "loss": 1.8261, "step": 8417 }, { "epoch": 0.9652562779497764, "grad_norm": 13.1875, "learning_rate": 3.183372733488066e-08, "loss": 1.9279, "step": 8418 }, { "epoch": 0.965370943699117, "grad_norm": 29.375, "learning_rate": 3.1624860072242966e-08, "loss": 1.8885, "step": 8419 }, { "epoch": 0.9654856094484577, "grad_norm": 18.25, "learning_rate": 3.1416678102249575e-08, "loss": 1.7219, "step": 8420 }, { "epoch": 0.9656002751977985, "grad_norm": 12.3125, "learning_rate": 3.12091814536164e-08, "loss": 1.7023, "step": 8421 }, { "epoch": 0.9657149409471391, "grad_norm": 13.4375, "learning_rate": 3.100237015496222e-08, "loss": 1.85, "step": 8422 }, { "epoch": 0.9658296066964798, "grad_norm": 16.125, "learning_rate": 3.0796244234813645e-08, "loss": 1.7195, "step": 8423 }, { "epoch": 0.9659442724458205, "grad_norm": 34.5, "learning_rate": 3.059080372160184e-08, "loss": 1.9784, "step": 8424 }, { "epoch": 0.9660589381951611, "grad_norm": 13.3125, "learning_rate": 3.038604864366246e-08, "loss": 1.5564, "step": 8425 }, { "epoch": 0.9661736039445018, "grad_norm": 27.25, "learning_rate": 3.0181979029238474e-08, "loss": 1.9623, "step": 8426 }, { "epoch": 0.9662882696938424, "grad_norm": 14.9375, "learning_rate": 2.9978594906477365e-08, "loss": 1.78, "step": 8427 }, { "epoch": 0.9664029354431831, "grad_norm": 15.3125, "learning_rate": 2.9775896303431694e-08, "loss": 1.6168, "step": 8428 }, { "epoch": 0.9665176011925238, "grad_norm": 12.5, "learning_rate": 2.957388324806021e-08, "loss": 1.7656, "step": 8429 }, { "epoch": 0.9666322669418644, "grad_norm": 14.6875, "learning_rate": 2.9372555768226175e-08, "loss": 1.8602, "step": 8430 }, { "epoch": 0.9667469326912052, "grad_norm": 13.5625, "learning_rate": 2.91719138916996e-08, "loss": 1.6831, "step": 8431 }, { "epoch": 0.9668615984405458, "grad_norm": 15.3125, "learning_rate": 2.8971957646155013e-08, "loss": 1.8141, "step": 8432 }, { "epoch": 0.9669762641898865, "grad_norm": 17.625, "learning_rate": 2.877268705917258e-08, "loss": 1.6464, "step": 8433 }, { "epoch": 0.9670909299392272, "grad_norm": 17.0, "learning_rate": 2.8574102158237528e-08, "loss": 1.6002, "step": 8434 }, { "epoch": 0.9672055956885678, "grad_norm": 14.5, "learning_rate": 2.837620297074073e-08, "loss": 1.4707, "step": 8435 }, { "epoch": 0.9673202614379085, "grad_norm": 14.75, "learning_rate": 2.817898952397924e-08, "loss": 1.7409, "step": 8436 }, { "epoch": 0.9674349271872492, "grad_norm": 14.75, "learning_rate": 2.798246184515463e-08, "loss": 1.8367, "step": 8437 }, { "epoch": 0.9675495929365898, "grad_norm": 16.375, "learning_rate": 2.7786619961374107e-08, "loss": 1.8049, "step": 8438 }, { "epoch": 0.9676642586859305, "grad_norm": 15.75, "learning_rate": 2.7591463899650506e-08, "loss": 1.6699, "step": 8439 }, { "epoch": 0.9677789244352711, "grad_norm": 16.0, "learning_rate": 2.7396993686901185e-08, "loss": 1.9852, "step": 8440 }, { "epoch": 0.9678935901846119, "grad_norm": 14.1875, "learning_rate": 2.7203209349950798e-08, "loss": 1.5535, "step": 8441 }, { "epoch": 0.9680082559339526, "grad_norm": 14.4375, "learning_rate": 2.701011091552741e-08, "loss": 1.711, "step": 8442 }, { "epoch": 0.9681229216832932, "grad_norm": 12.5625, "learning_rate": 2.681769841026527e-08, "loss": 1.5159, "step": 8443 }, { "epoch": 0.9682375874326339, "grad_norm": 13.75, "learning_rate": 2.6625971860704258e-08, "loss": 1.6536, "step": 8444 }, { "epoch": 0.9683522531819746, "grad_norm": 15.75, "learning_rate": 2.6434931293289333e-08, "loss": 1.8223, "step": 8445 }, { "epoch": 0.9684669189313152, "grad_norm": 13.75, "learning_rate": 2.624457673437053e-08, "loss": 1.853, "step": 8446 }, { "epoch": 0.9685815846806559, "grad_norm": 16.125, "learning_rate": 2.605490821020462e-08, "loss": 1.9513, "step": 8447 }, { "epoch": 0.9686962504299965, "grad_norm": 18.0, "learning_rate": 2.5865925746951237e-08, "loss": 1.9491, "step": 8448 }, { "epoch": 0.9688109161793372, "grad_norm": 10.875, "learning_rate": 2.5677629370678414e-08, "loss": 1.4875, "step": 8449 }, { "epoch": 0.968925581928678, "grad_norm": 13.9375, "learning_rate": 2.5490019107357043e-08, "loss": 1.8383, "step": 8450 }, { "epoch": 0.9690402476780186, "grad_norm": 32.75, "learning_rate": 2.530309498286476e-08, "loss": 2.0229, "step": 8451 }, { "epoch": 0.9691549134273593, "grad_norm": 12.6875, "learning_rate": 2.511685702298483e-08, "loss": 1.5761, "step": 8452 }, { "epoch": 0.9692695791766999, "grad_norm": 16.75, "learning_rate": 2.4931305253403927e-08, "loss": 2.056, "step": 8453 }, { "epoch": 0.9693842449260406, "grad_norm": 25.875, "learning_rate": 2.4746439699716017e-08, "loss": 2.2027, "step": 8454 }, { "epoch": 0.9694989106753813, "grad_norm": 14.4375, "learning_rate": 2.4562260387420157e-08, "loss": 1.6608, "step": 8455 }, { "epoch": 0.9696135764247219, "grad_norm": 13.75, "learning_rate": 2.4378767341919905e-08, "loss": 1.8363, "step": 8456 }, { "epoch": 0.9697282421740626, "grad_norm": 14.75, "learning_rate": 2.4195960588524468e-08, "loss": 1.5073, "step": 8457 }, { "epoch": 0.9698429079234033, "grad_norm": 15.25, "learning_rate": 2.4013840152448677e-08, "loss": 2.0584, "step": 8458 }, { "epoch": 0.9699575736727439, "grad_norm": 12.6875, "learning_rate": 2.383240605881354e-08, "loss": 1.5737, "step": 8459 }, { "epoch": 0.9700722394220846, "grad_norm": 14.625, "learning_rate": 2.3651658332642933e-08, "loss": 1.478, "step": 8460 }, { "epoch": 0.9701869051714253, "grad_norm": 16.125, "learning_rate": 2.347159699886803e-08, "loss": 1.7563, "step": 8461 }, { "epoch": 0.970301570920766, "grad_norm": 14.9375, "learning_rate": 2.3292222082325623e-08, "loss": 1.984, "step": 8462 }, { "epoch": 0.9704162366701067, "grad_norm": 13.8125, "learning_rate": 2.3113533607756477e-08, "loss": 1.4827, "step": 8463 }, { "epoch": 0.9705309024194473, "grad_norm": 12.5, "learning_rate": 2.2935531599806993e-08, "loss": 1.9, "step": 8464 }, { "epoch": 0.970645568168788, "grad_norm": 14.75, "learning_rate": 2.2758216083029194e-08, "loss": 1.7782, "step": 8465 }, { "epoch": 0.9707602339181286, "grad_norm": 13.9375, "learning_rate": 2.2581587081881296e-08, "loss": 1.48, "step": 8466 }, { "epoch": 0.9708748996674693, "grad_norm": 14.9375, "learning_rate": 2.240564462072492e-08, "loss": 2.0055, "step": 8467 }, { "epoch": 0.97098956541681, "grad_norm": 14.625, "learning_rate": 2.2230388723828434e-08, "loss": 1.5301, "step": 8468 }, { "epoch": 0.9711042311661506, "grad_norm": 14.75, "learning_rate": 2.2055819415364722e-08, "loss": 1.6008, "step": 8469 }, { "epoch": 0.9712188969154913, "grad_norm": 14.8125, "learning_rate": 2.1881936719411744e-08, "loss": 1.779, "step": 8470 }, { "epoch": 0.9713335626648321, "grad_norm": 15.125, "learning_rate": 2.1708740659955317e-08, "loss": 1.7957, "step": 8471 }, { "epoch": 0.9714482284141727, "grad_norm": 12.5, "learning_rate": 2.1536231260882445e-08, "loss": 1.6644, "step": 8472 }, { "epoch": 0.9715628941635134, "grad_norm": 15.9375, "learning_rate": 2.1364408545988537e-08, "loss": 1.9749, "step": 8473 }, { "epoch": 0.971677559912854, "grad_norm": 19.375, "learning_rate": 2.119327253897352e-08, "loss": 1.8092, "step": 8474 }, { "epoch": 0.9717922256621947, "grad_norm": 12.0625, "learning_rate": 2.1022823263441294e-08, "loss": 1.6061, "step": 8475 }, { "epoch": 0.9719068914115354, "grad_norm": 11.8125, "learning_rate": 2.08530607429025e-08, "loss": 1.38, "step": 8476 }, { "epoch": 0.972021557160876, "grad_norm": 15.5625, "learning_rate": 2.0683985000773398e-08, "loss": 2.0639, "step": 8477 }, { "epoch": 0.9721362229102167, "grad_norm": 17.0, "learning_rate": 2.0515596060373676e-08, "loss": 2.0112, "step": 8478 }, { "epoch": 0.9722508886595574, "grad_norm": 14.0625, "learning_rate": 2.034789394493031e-08, "loss": 1.935, "step": 8479 }, { "epoch": 0.972365554408898, "grad_norm": 14.125, "learning_rate": 2.0180878677573678e-08, "loss": 1.5959, "step": 8480 }, { "epoch": 0.9724802201582388, "grad_norm": 16.375, "learning_rate": 2.0014550281340918e-08, "loss": 2.0123, "step": 8481 }, { "epoch": 0.9725948859075794, "grad_norm": 14.0, "learning_rate": 1.9848908779174225e-08, "loss": 2.153, "step": 8482 }, { "epoch": 0.9727095516569201, "grad_norm": 13.625, "learning_rate": 1.968395419392033e-08, "loss": 1.6014, "step": 8483 }, { "epoch": 0.9728242174062608, "grad_norm": 15.5, "learning_rate": 1.9519686548331028e-08, "loss": 1.6656, "step": 8484 }, { "epoch": 0.9729388831556014, "grad_norm": 15.5625, "learning_rate": 1.9356105865064312e-08, "loss": 1.8004, "step": 8485 }, { "epoch": 0.9730535489049421, "grad_norm": 12.9375, "learning_rate": 1.919321216668324e-08, "loss": 1.7204, "step": 8486 }, { "epoch": 0.9731682146542827, "grad_norm": 14.8125, "learning_rate": 1.903100547565595e-08, "loss": 2.0723, "step": 8487 }, { "epoch": 0.9732828804036234, "grad_norm": 22.125, "learning_rate": 1.8869485814355105e-08, "loss": 1.6775, "step": 8488 }, { "epoch": 0.9733975461529641, "grad_norm": 13.6875, "learning_rate": 1.87086532050601e-08, "loss": 1.7645, "step": 8489 }, { "epoch": 0.9735122119023047, "grad_norm": 15.625, "learning_rate": 1.854850766995431e-08, "loss": 1.8554, "step": 8490 }, { "epoch": 0.9736268776516455, "grad_norm": 33.5, "learning_rate": 1.838904923112672e-08, "loss": 2.458, "step": 8491 }, { "epoch": 0.9737415434009862, "grad_norm": 13.6875, "learning_rate": 1.8230277910571414e-08, "loss": 1.3932, "step": 8492 }, { "epoch": 0.9738562091503268, "grad_norm": 14.4375, "learning_rate": 1.8072193730188094e-08, "loss": 1.8264, "step": 8493 }, { "epoch": 0.9739708748996675, "grad_norm": 13.3125, "learning_rate": 1.791479671178209e-08, "loss": 1.6348, "step": 8494 }, { "epoch": 0.9740855406490081, "grad_norm": 13.5, "learning_rate": 1.775808687706215e-08, "loss": 1.7277, "step": 8495 }, { "epoch": 0.9742002063983488, "grad_norm": 12.0, "learning_rate": 1.760206424764377e-08, "loss": 1.5907, "step": 8496 }, { "epoch": 0.9743148721476895, "grad_norm": 14.5, "learning_rate": 1.7446728845048055e-08, "loss": 1.6949, "step": 8497 }, { "epoch": 0.9744295378970301, "grad_norm": 16.5, "learning_rate": 1.7292080690699543e-08, "loss": 1.7347, "step": 8498 }, { "epoch": 0.9745442036463708, "grad_norm": 15.4375, "learning_rate": 1.7138119805929498e-08, "loss": 1.9534, "step": 8499 }, { "epoch": 0.9746588693957114, "grad_norm": 14.375, "learning_rate": 1.698484621197427e-08, "loss": 1.7034, "step": 8500 } ], "logging_steps": 1.0, "max_steps": 8721, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 7.636473198049812e+17, "train_batch_size": 4, "trial_name": null, "trial_params": null }