{ "best_metric": null, "best_model_checkpoint": null, "epoch": 199.75301204819277, "eval_steps": 500, "global_step": 20600, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00963855421686747, "grad_norm": 3.883495330810547, "learning_rate": 9.999514563106796e-05, "loss": 0.3982, "step": 1 }, { "epoch": 0.01927710843373494, "grad_norm": 5.607765197753906, "learning_rate": 9.999029126213592e-05, "loss": 0.5667, "step": 2 }, { "epoch": 0.02891566265060241, "grad_norm": 5.426010608673096, "learning_rate": 9.998543689320389e-05, "loss": 0.5459, "step": 3 }, { "epoch": 0.03855421686746988, "grad_norm": 5.313508987426758, "learning_rate": 9.998058252427185e-05, "loss": 0.502, "step": 4 }, { "epoch": 0.04819277108433735, "grad_norm": 4.864013195037842, "learning_rate": 9.997572815533981e-05, "loss": 0.4106, "step": 5 }, { "epoch": 0.05783132530120482, "grad_norm": 5.514575004577637, "learning_rate": 9.997087378640776e-05, "loss": 0.6682, "step": 6 }, { "epoch": 0.06746987951807229, "grad_norm": 3.787320613861084, "learning_rate": 9.996601941747574e-05, "loss": 0.392, "step": 7 }, { "epoch": 0.07710843373493977, "grad_norm": 4.317789554595947, "learning_rate": 9.99611650485437e-05, "loss": 0.3231, "step": 8 }, { "epoch": 0.08674698795180723, "grad_norm": 4.514036655426025, "learning_rate": 9.995631067961165e-05, "loss": 0.7775, "step": 9 }, { "epoch": 0.0963855421686747, "grad_norm": 5.2076616287231445, "learning_rate": 9.995145631067961e-05, "loss": 0.4871, "step": 10 }, { "epoch": 0.10602409638554217, "grad_norm": 4.678882122039795, "learning_rate": 9.994660194174758e-05, "loss": 0.4344, "step": 11 }, { "epoch": 0.11566265060240964, "grad_norm": 4.157716274261475, "learning_rate": 9.994174757281554e-05, "loss": 0.407, "step": 12 }, { "epoch": 0.12530120481927712, "grad_norm": 5.07155704498291, "learning_rate": 9.99368932038835e-05, "loss": 0.5023, "step": 13 }, { "epoch": 0.13493975903614458, "grad_norm": 6.698201656341553, "learning_rate": 9.993203883495145e-05, "loss": 0.5374, "step": 14 }, { "epoch": 0.14457831325301204, "grad_norm": 3.8603944778442383, "learning_rate": 9.992718446601943e-05, "loss": 0.3762, "step": 15 }, { "epoch": 0.15421686746987953, "grad_norm": 3.9224307537078857, "learning_rate": 9.992233009708738e-05, "loss": 0.3969, "step": 16 }, { "epoch": 0.163855421686747, "grad_norm": 5.238499164581299, "learning_rate": 9.991747572815534e-05, "loss": 0.4687, "step": 17 }, { "epoch": 0.17349397590361446, "grad_norm": 5.881870746612549, "learning_rate": 9.99126213592233e-05, "loss": 0.6128, "step": 18 }, { "epoch": 0.18313253012048192, "grad_norm": 7.116026878356934, "learning_rate": 9.990776699029126e-05, "loss": 0.5128, "step": 19 }, { "epoch": 0.1927710843373494, "grad_norm": 3.723599910736084, "learning_rate": 9.990291262135923e-05, "loss": 0.4186, "step": 20 }, { "epoch": 0.20240963855421687, "grad_norm": 2.8818421363830566, "learning_rate": 9.989805825242719e-05, "loss": 0.3187, "step": 21 }, { "epoch": 0.21204819277108433, "grad_norm": 4.825333118438721, "learning_rate": 9.989320388349514e-05, "loss": 0.3669, "step": 22 }, { "epoch": 0.2216867469879518, "grad_norm": 5.377885818481445, "learning_rate": 9.98883495145631e-05, "loss": 0.4165, "step": 23 }, { "epoch": 0.23132530120481928, "grad_norm": 7.343980312347412, "learning_rate": 9.988349514563107e-05, "loss": 0.606, "step": 24 }, { "epoch": 0.24096385542168675, "grad_norm": 5.265160083770752, "learning_rate": 9.987864077669903e-05, "loss": 0.433, "step": 25 }, { "epoch": 0.25060240963855424, "grad_norm": 4.292900562286377, "learning_rate": 9.9873786407767e-05, "loss": 0.4028, "step": 26 }, { "epoch": 0.26024096385542167, "grad_norm": 4.611701011657715, "learning_rate": 9.986893203883496e-05, "loss": 0.511, "step": 27 }, { "epoch": 0.26987951807228916, "grad_norm": 5.397536277770996, "learning_rate": 9.986407766990292e-05, "loss": 0.4194, "step": 28 }, { "epoch": 0.27951807228915665, "grad_norm": 4.468271732330322, "learning_rate": 9.985922330097089e-05, "loss": 0.654, "step": 29 }, { "epoch": 0.2891566265060241, "grad_norm": 2.85311222076416, "learning_rate": 9.985436893203885e-05, "loss": 0.3599, "step": 30 }, { "epoch": 0.2987951807228916, "grad_norm": 3.356994152069092, "learning_rate": 9.98495145631068e-05, "loss": 0.3652, "step": 31 }, { "epoch": 0.30843373493975906, "grad_norm": 4.449845314025879, "learning_rate": 9.984466019417476e-05, "loss": 0.4027, "step": 32 }, { "epoch": 0.3180722891566265, "grad_norm": 5.120013236999512, "learning_rate": 9.983980582524273e-05, "loss": 0.5381, "step": 33 }, { "epoch": 0.327710843373494, "grad_norm": 5.6530866622924805, "learning_rate": 9.983495145631069e-05, "loss": 0.3182, "step": 34 }, { "epoch": 0.3373493975903614, "grad_norm": 9.619132041931152, "learning_rate": 9.983009708737865e-05, "loss": 0.4018, "step": 35 }, { "epoch": 0.3469879518072289, "grad_norm": 4.274433612823486, "learning_rate": 9.98252427184466e-05, "loss": 0.372, "step": 36 }, { "epoch": 0.3566265060240964, "grad_norm": 10.34952163696289, "learning_rate": 9.982038834951458e-05, "loss": 0.3497, "step": 37 }, { "epoch": 0.36626506024096384, "grad_norm": 6.388422012329102, "learning_rate": 9.981553398058254e-05, "loss": 0.5507, "step": 38 }, { "epoch": 0.3759036144578313, "grad_norm": 4.177401542663574, "learning_rate": 9.981067961165049e-05, "loss": 0.356, "step": 39 }, { "epoch": 0.3855421686746988, "grad_norm": 5.476969242095947, "learning_rate": 9.980582524271845e-05, "loss": 0.4471, "step": 40 }, { "epoch": 0.39518072289156625, "grad_norm": 3.691624402999878, "learning_rate": 9.980097087378641e-05, "loss": 0.4384, "step": 41 }, { "epoch": 0.40481927710843374, "grad_norm": 4.587851524353027, "learning_rate": 9.979611650485438e-05, "loss": 0.3723, "step": 42 }, { "epoch": 0.41445783132530123, "grad_norm": 4.818039894104004, "learning_rate": 9.979126213592234e-05, "loss": 0.4318, "step": 43 }, { "epoch": 0.42409638554216866, "grad_norm": 3.3452248573303223, "learning_rate": 9.97864077669903e-05, "loss": 0.4139, "step": 44 }, { "epoch": 0.43373493975903615, "grad_norm": 2.9612808227539062, "learning_rate": 9.978155339805825e-05, "loss": 0.2394, "step": 45 }, { "epoch": 0.4433734939759036, "grad_norm": 5.043538570404053, "learning_rate": 9.977669902912622e-05, "loss": 0.4536, "step": 46 }, { "epoch": 0.4530120481927711, "grad_norm": 2.974832773208618, "learning_rate": 9.977184466019418e-05, "loss": 0.3886, "step": 47 }, { "epoch": 0.46265060240963857, "grad_norm": 4.45283842086792, "learning_rate": 9.976699029126214e-05, "loss": 0.5729, "step": 48 }, { "epoch": 0.472289156626506, "grad_norm": 5.013668060302734, "learning_rate": 9.97621359223301e-05, "loss": 0.3299, "step": 49 }, { "epoch": 0.4819277108433735, "grad_norm": 4.573915004730225, "learning_rate": 9.975728155339807e-05, "loss": 0.5541, "step": 50 }, { "epoch": 0.491566265060241, "grad_norm": 4.360326766967773, "learning_rate": 9.975242718446603e-05, "loss": 0.366, "step": 51 }, { "epoch": 0.5012048192771085, "grad_norm": 6.4832963943481445, "learning_rate": 9.974757281553398e-05, "loss": 0.4354, "step": 52 }, { "epoch": 0.5108433734939759, "grad_norm": 3.3396496772766113, "learning_rate": 9.974271844660194e-05, "loss": 0.257, "step": 53 }, { "epoch": 0.5204819277108433, "grad_norm": 6.501490592956543, "learning_rate": 9.973786407766991e-05, "loss": 0.5665, "step": 54 }, { "epoch": 0.5301204819277109, "grad_norm": 3.3325388431549072, "learning_rate": 9.973300970873787e-05, "loss": 0.3599, "step": 55 }, { "epoch": 0.5397590361445783, "grad_norm": 4.002194404602051, "learning_rate": 9.972815533980583e-05, "loss": 0.2691, "step": 56 }, { "epoch": 0.5493975903614458, "grad_norm": 5.255983829498291, "learning_rate": 9.972330097087379e-05, "loss": 0.4004, "step": 57 }, { "epoch": 0.5590361445783133, "grad_norm": 5.712917804718018, "learning_rate": 9.971844660194174e-05, "loss": 0.3543, "step": 58 }, { "epoch": 0.5686746987951807, "grad_norm": 3.3866543769836426, "learning_rate": 9.971359223300972e-05, "loss": 0.4485, "step": 59 }, { "epoch": 0.5783132530120482, "grad_norm": 4.196461200714111, "learning_rate": 9.970873786407767e-05, "loss": 0.3984, "step": 60 }, { "epoch": 0.5879518072289157, "grad_norm": 3.8618485927581787, "learning_rate": 9.970388349514563e-05, "loss": 0.3843, "step": 61 }, { "epoch": 0.5975903614457831, "grad_norm": 6.829723358154297, "learning_rate": 9.969902912621359e-05, "loss": 0.6552, "step": 62 }, { "epoch": 0.6072289156626506, "grad_norm": 4.419615745544434, "learning_rate": 9.969417475728156e-05, "loss": 0.3523, "step": 63 }, { "epoch": 0.6168674698795181, "grad_norm": 2.226353406906128, "learning_rate": 9.968932038834952e-05, "loss": 0.1973, "step": 64 }, { "epoch": 0.6265060240963856, "grad_norm": 4.696351528167725, "learning_rate": 9.968446601941748e-05, "loss": 0.3525, "step": 65 }, { "epoch": 0.636144578313253, "grad_norm": 3.438277244567871, "learning_rate": 9.967961165048543e-05, "loss": 0.3764, "step": 66 }, { "epoch": 0.6457831325301204, "grad_norm": 2.5591115951538086, "learning_rate": 9.96747572815534e-05, "loss": 0.2116, "step": 67 }, { "epoch": 0.655421686746988, "grad_norm": 4.576904773712158, "learning_rate": 9.966990291262136e-05, "loss": 0.4568, "step": 68 }, { "epoch": 0.6650602409638554, "grad_norm": 3.3983864784240723, "learning_rate": 9.966504854368932e-05, "loss": 0.4861, "step": 69 }, { "epoch": 0.6746987951807228, "grad_norm": 3.2781474590301514, "learning_rate": 9.966019417475728e-05, "loss": 0.4643, "step": 70 }, { "epoch": 0.6843373493975904, "grad_norm": 4.744735240936279, "learning_rate": 9.965533980582525e-05, "loss": 0.243, "step": 71 }, { "epoch": 0.6939759036144578, "grad_norm": 3.09490704536438, "learning_rate": 9.965048543689321e-05, "loss": 0.2916, "step": 72 }, { "epoch": 0.7036144578313253, "grad_norm": 4.593525409698486, "learning_rate": 9.964563106796117e-05, "loss": 0.53, "step": 73 }, { "epoch": 0.7132530120481928, "grad_norm": 2.6752147674560547, "learning_rate": 9.964077669902912e-05, "loss": 0.1456, "step": 74 }, { "epoch": 0.7228915662650602, "grad_norm": 3.4532177448272705, "learning_rate": 9.963592233009708e-05, "loss": 0.6119, "step": 75 }, { "epoch": 0.7325301204819277, "grad_norm": 4.644095420837402, "learning_rate": 9.963106796116507e-05, "loss": 0.3554, "step": 76 }, { "epoch": 0.7421686746987952, "grad_norm": 3.8956315517425537, "learning_rate": 9.962621359223302e-05, "loss": 0.4673, "step": 77 }, { "epoch": 0.7518072289156627, "grad_norm": 4.211650848388672, "learning_rate": 9.962135922330098e-05, "loss": 0.5125, "step": 78 }, { "epoch": 0.7614457831325301, "grad_norm": 5.862034320831299, "learning_rate": 9.961650485436894e-05, "loss": 0.3513, "step": 79 }, { "epoch": 0.7710843373493976, "grad_norm": 2.721202850341797, "learning_rate": 9.96116504854369e-05, "loss": 0.2133, "step": 80 }, { "epoch": 0.7807228915662651, "grad_norm": 3.8570311069488525, "learning_rate": 9.960679611650487e-05, "loss": 0.3927, "step": 81 }, { "epoch": 0.7903614457831325, "grad_norm": 3.575876235961914, "learning_rate": 9.960194174757283e-05, "loss": 0.3017, "step": 82 }, { "epoch": 0.8, "grad_norm": 6.375074863433838, "learning_rate": 9.959708737864078e-05, "loss": 0.2167, "step": 83 }, { "epoch": 0.8096385542168675, "grad_norm": 3.2719156742095947, "learning_rate": 9.959223300970874e-05, "loss": 0.4066, "step": 84 }, { "epoch": 0.8192771084337349, "grad_norm": 2.997389078140259, "learning_rate": 9.958737864077671e-05, "loss": 0.3016, "step": 85 }, { "epoch": 0.8289156626506025, "grad_norm": 3.0137360095977783, "learning_rate": 9.958252427184467e-05, "loss": 0.4143, "step": 86 }, { "epoch": 0.8385542168674699, "grad_norm": 7.212274074554443, "learning_rate": 9.957766990291263e-05, "loss": 1.0344, "step": 87 }, { "epoch": 0.8481927710843373, "grad_norm": 11.572142601013184, "learning_rate": 9.957281553398059e-05, "loss": 0.3464, "step": 88 }, { "epoch": 0.8578313253012049, "grad_norm": 2.302300214767456, "learning_rate": 9.956796116504856e-05, "loss": 0.1792, "step": 89 }, { "epoch": 0.8674698795180723, "grad_norm": 6.256125450134277, "learning_rate": 9.956310679611651e-05, "loss": 0.4526, "step": 90 }, { "epoch": 0.8771084337349397, "grad_norm": 6.348989963531494, "learning_rate": 9.955825242718447e-05, "loss": 0.3827, "step": 91 }, { "epoch": 0.8867469879518072, "grad_norm": 7.5619401931762695, "learning_rate": 9.955339805825243e-05, "loss": 0.5465, "step": 92 }, { "epoch": 0.8963855421686747, "grad_norm": 4.077050685882568, "learning_rate": 9.95485436893204e-05, "loss": 0.3531, "step": 93 }, { "epoch": 0.9060240963855422, "grad_norm": 5.459836483001709, "learning_rate": 9.954368932038836e-05, "loss": 0.7332, "step": 94 }, { "epoch": 0.9156626506024096, "grad_norm": 3.7656667232513428, "learning_rate": 9.953883495145632e-05, "loss": 0.308, "step": 95 }, { "epoch": 0.9253012048192771, "grad_norm": 4.956954479217529, "learning_rate": 9.953398058252427e-05, "loss": 0.4582, "step": 96 }, { "epoch": 0.9349397590361446, "grad_norm": 4.917057037353516, "learning_rate": 9.952912621359223e-05, "loss": 0.4959, "step": 97 }, { "epoch": 0.944578313253012, "grad_norm": 5.093995094299316, "learning_rate": 9.95242718446602e-05, "loss": 0.583, "step": 98 }, { "epoch": 0.9542168674698795, "grad_norm": 4.826685428619385, "learning_rate": 9.951941747572816e-05, "loss": 0.3877, "step": 99 }, { "epoch": 0.963855421686747, "grad_norm": 4.2527241706848145, "learning_rate": 9.951456310679612e-05, "loss": 0.2218, "step": 100 }, { "epoch": 0.9734939759036144, "grad_norm": 4.999881744384766, "learning_rate": 9.950970873786408e-05, "loss": 0.619, "step": 101 }, { "epoch": 0.983132530120482, "grad_norm": 7.628602027893066, "learning_rate": 9.950485436893205e-05, "loss": 0.3487, "step": 102 }, { "epoch": 0.9927710843373494, "grad_norm": 6.755055904388428, "learning_rate": 9.95e-05, "loss": 0.4397, "step": 103 }, { "epoch": 1.008433734939759, "grad_norm": 7.430731296539307, "learning_rate": 9.949514563106796e-05, "loss": 0.4581, "step": 104 }, { "epoch": 1.0180722891566265, "grad_norm": 3.703141689300537, "learning_rate": 9.949029126213592e-05, "loss": 0.2403, "step": 105 }, { "epoch": 1.027710843373494, "grad_norm": 2.5330729484558105, "learning_rate": 9.948543689320389e-05, "loss": 0.3297, "step": 106 }, { "epoch": 1.0373493975903614, "grad_norm": 14.256707191467285, "learning_rate": 9.948058252427185e-05, "loss": 0.4547, "step": 107 }, { "epoch": 1.046987951807229, "grad_norm": 11.471699714660645, "learning_rate": 9.947572815533981e-05, "loss": 0.3662, "step": 108 }, { "epoch": 1.0566265060240965, "grad_norm": 11.621235847473145, "learning_rate": 9.947087378640777e-05, "loss": 0.3364, "step": 109 }, { "epoch": 1.0662650602409638, "grad_norm": 9.583854675292969, "learning_rate": 9.946601941747574e-05, "loss": 0.3854, "step": 110 }, { "epoch": 1.0759036144578313, "grad_norm": 14.585342407226562, "learning_rate": 9.94611650485437e-05, "loss": 0.3982, "step": 111 }, { "epoch": 1.0855421686746989, "grad_norm": 1.7425644397735596, "learning_rate": 9.945631067961165e-05, "loss": 0.555, "step": 112 }, { "epoch": 1.0951807228915662, "grad_norm": 4.9239888191223145, "learning_rate": 9.945145631067961e-05, "loss": 0.3674, "step": 113 }, { "epoch": 1.1048192771084338, "grad_norm": 2.0184385776519775, "learning_rate": 9.944660194174757e-05, "loss": 0.4135, "step": 114 }, { "epoch": 1.1144578313253013, "grad_norm": 4.546851634979248, "learning_rate": 9.944174757281554e-05, "loss": 0.3578, "step": 115 }, { "epoch": 1.1240963855421686, "grad_norm": 6.238420486450195, "learning_rate": 9.94368932038835e-05, "loss": 0.4852, "step": 116 }, { "epoch": 1.1337349397590362, "grad_norm": 8.604491233825684, "learning_rate": 9.943203883495146e-05, "loss": 0.2144, "step": 117 }, { "epoch": 1.1433734939759037, "grad_norm": 20.710649490356445, "learning_rate": 9.942718446601941e-05, "loss": 0.2134, "step": 118 }, { "epoch": 1.153012048192771, "grad_norm": 3.5029444694519043, "learning_rate": 9.942233009708738e-05, "loss": 0.305, "step": 119 }, { "epoch": 1.1626506024096386, "grad_norm": 3.485933303833008, "learning_rate": 9.941747572815534e-05, "loss": 0.275, "step": 120 }, { "epoch": 1.1722891566265061, "grad_norm": 7.1451873779296875, "learning_rate": 9.94126213592233e-05, "loss": 0.4523, "step": 121 }, { "epoch": 1.1819277108433734, "grad_norm": 6.665964603424072, "learning_rate": 9.940776699029126e-05, "loss": 0.4313, "step": 122 }, { "epoch": 1.191566265060241, "grad_norm": 8.900416374206543, "learning_rate": 9.940291262135923e-05, "loss": 0.3081, "step": 123 }, { "epoch": 1.2012048192771085, "grad_norm": 13.466391563415527, "learning_rate": 9.939805825242719e-05, "loss": 0.3127, "step": 124 }, { "epoch": 1.2108433734939759, "grad_norm": 12.717873573303223, "learning_rate": 9.939320388349514e-05, "loss": 0.2696, "step": 125 }, { "epoch": 1.2204819277108434, "grad_norm": 9.354372024536133, "learning_rate": 9.93883495145631e-05, "loss": 0.3378, "step": 126 }, { "epoch": 1.230120481927711, "grad_norm": 4.800004005432129, "learning_rate": 9.938349514563107e-05, "loss": 0.3182, "step": 127 }, { "epoch": 1.2397590361445783, "grad_norm": 1.494340181350708, "learning_rate": 9.937864077669904e-05, "loss": 0.1858, "step": 128 }, { "epoch": 1.2493975903614458, "grad_norm": 2.613430976867676, "learning_rate": 9.9373786407767e-05, "loss": 0.29, "step": 129 }, { "epoch": 1.2590361445783134, "grad_norm": 7.588158130645752, "learning_rate": 9.936893203883496e-05, "loss": 0.2797, "step": 130 }, { "epoch": 1.2686746987951807, "grad_norm": 2.852006196975708, "learning_rate": 9.936407766990292e-05, "loss": 0.179, "step": 131 }, { "epoch": 1.2783132530120482, "grad_norm": 6.471401691436768, "learning_rate": 9.935922330097089e-05, "loss": 0.513, "step": 132 }, { "epoch": 1.2879518072289158, "grad_norm": 4.99775505065918, "learning_rate": 9.935436893203885e-05, "loss": 0.3733, "step": 133 }, { "epoch": 1.297590361445783, "grad_norm": 7.63673734664917, "learning_rate": 9.93495145631068e-05, "loss": 0.3892, "step": 134 }, { "epoch": 1.3072289156626506, "grad_norm": 12.84007740020752, "learning_rate": 9.934466019417476e-05, "loss": 0.4571, "step": 135 }, { "epoch": 1.3168674698795182, "grad_norm": 9.449222564697266, "learning_rate": 9.933980582524272e-05, "loss": 0.3872, "step": 136 }, { "epoch": 1.3265060240963855, "grad_norm": 8.361763000488281, "learning_rate": 9.933495145631069e-05, "loss": 0.5031, "step": 137 }, { "epoch": 1.336144578313253, "grad_norm": 18.75994873046875, "learning_rate": 9.933009708737865e-05, "loss": 0.5747, "step": 138 }, { "epoch": 1.3457831325301206, "grad_norm": 16.9029483795166, "learning_rate": 9.932524271844661e-05, "loss": 0.3096, "step": 139 }, { "epoch": 1.355421686746988, "grad_norm": 17.061098098754883, "learning_rate": 9.932038834951457e-05, "loss": 0.3278, "step": 140 }, { "epoch": 1.3650602409638555, "grad_norm": 5.571994304656982, "learning_rate": 9.931553398058254e-05, "loss": 0.463, "step": 141 }, { "epoch": 1.374698795180723, "grad_norm": 9.901453971862793, "learning_rate": 9.93106796116505e-05, "loss": 0.3786, "step": 142 }, { "epoch": 1.3843373493975903, "grad_norm": 3.8148891925811768, "learning_rate": 9.930582524271845e-05, "loss": 0.3111, "step": 143 }, { "epoch": 1.393975903614458, "grad_norm": 16.853355407714844, "learning_rate": 9.930097087378641e-05, "loss": 0.4104, "step": 144 }, { "epoch": 1.4036144578313254, "grad_norm": 4.489780902862549, "learning_rate": 9.929611650485438e-05, "loss": 0.2905, "step": 145 }, { "epoch": 1.4132530120481928, "grad_norm": 6.818675518035889, "learning_rate": 9.929126213592234e-05, "loss": 0.3973, "step": 146 }, { "epoch": 1.4228915662650603, "grad_norm": 16.261863708496094, "learning_rate": 9.92864077669903e-05, "loss": 0.4392, "step": 147 }, { "epoch": 1.4325301204819278, "grad_norm": 5.916988849639893, "learning_rate": 9.928155339805825e-05, "loss": 0.3014, "step": 148 }, { "epoch": 1.4421686746987952, "grad_norm": 13.718411445617676, "learning_rate": 9.927669902912621e-05, "loss": 0.4971, "step": 149 }, { "epoch": 1.4518072289156627, "grad_norm": 4.737636089324951, "learning_rate": 9.927184466019418e-05, "loss": 0.1806, "step": 150 }, { "epoch": 1.46144578313253, "grad_norm": 5.359301567077637, "learning_rate": 9.926699029126214e-05, "loss": 0.2651, "step": 151 }, { "epoch": 1.4710843373493976, "grad_norm": 12.544493675231934, "learning_rate": 9.92621359223301e-05, "loss": 0.4407, "step": 152 }, { "epoch": 1.4807228915662651, "grad_norm": 9.471081733703613, "learning_rate": 9.925728155339806e-05, "loss": 0.4255, "step": 153 }, { "epoch": 1.4903614457831325, "grad_norm": 10.421324729919434, "learning_rate": 9.925242718446603e-05, "loss": 0.3232, "step": 154 }, { "epoch": 1.5, "grad_norm": 2.2069265842437744, "learning_rate": 9.924757281553399e-05, "loss": 0.3474, "step": 155 }, { "epoch": 1.5096385542168673, "grad_norm": 5.023476600646973, "learning_rate": 9.924271844660194e-05, "loss": 0.2835, "step": 156 }, { "epoch": 1.519277108433735, "grad_norm": 3.485438346862793, "learning_rate": 9.92378640776699e-05, "loss": 0.417, "step": 157 }, { "epoch": 1.5289156626506024, "grad_norm": 2.2009072303771973, "learning_rate": 9.923300970873787e-05, "loss": 0.4476, "step": 158 }, { "epoch": 1.5385542168674697, "grad_norm": 3.3014838695526123, "learning_rate": 9.922815533980583e-05, "loss": 0.3925, "step": 159 }, { "epoch": 1.5481927710843375, "grad_norm": 4.839330196380615, "learning_rate": 9.922330097087379e-05, "loss": 0.3869, "step": 160 }, { "epoch": 1.5578313253012048, "grad_norm": 8.27634048461914, "learning_rate": 9.921844660194175e-05, "loss": 0.3928, "step": 161 }, { "epoch": 1.5674698795180722, "grad_norm": 16.446203231811523, "learning_rate": 9.921359223300972e-05, "loss": 0.5918, "step": 162 }, { "epoch": 1.57710843373494, "grad_norm": 4.974839210510254, "learning_rate": 9.920873786407767e-05, "loss": 0.372, "step": 163 }, { "epoch": 1.5867469879518072, "grad_norm": 9.715226173400879, "learning_rate": 9.920388349514563e-05, "loss": 0.3904, "step": 164 }, { "epoch": 1.5963855421686746, "grad_norm": 6.087960720062256, "learning_rate": 9.919902912621359e-05, "loss": 0.6289, "step": 165 }, { "epoch": 1.6060240963855423, "grad_norm": 3.8631999492645264, "learning_rate": 9.919417475728155e-05, "loss": 0.3785, "step": 166 }, { "epoch": 1.6156626506024097, "grad_norm": 6.3117523193359375, "learning_rate": 9.918932038834952e-05, "loss": 0.352, "step": 167 }, { "epoch": 1.625301204819277, "grad_norm": 12.935311317443848, "learning_rate": 9.918446601941748e-05, "loss": 0.303, "step": 168 }, { "epoch": 1.6349397590361445, "grad_norm": 11.414741516113281, "learning_rate": 9.917961165048544e-05, "loss": 0.4404, "step": 169 }, { "epoch": 1.644578313253012, "grad_norm": 5.09133768081665, "learning_rate": 9.917475728155339e-05, "loss": 0.3064, "step": 170 }, { "epoch": 1.6542168674698794, "grad_norm": 8.611676216125488, "learning_rate": 9.916990291262136e-05, "loss": 0.4883, "step": 171 }, { "epoch": 1.663855421686747, "grad_norm": 15.391538619995117, "learning_rate": 9.916504854368932e-05, "loss": 0.3433, "step": 172 }, { "epoch": 1.6734939759036145, "grad_norm": 11.154183387756348, "learning_rate": 9.916019417475728e-05, "loss": 0.658, "step": 173 }, { "epoch": 1.6831325301204818, "grad_norm": 3.634519338607788, "learning_rate": 9.915533980582524e-05, "loss": 0.3926, "step": 174 }, { "epoch": 1.6927710843373494, "grad_norm": 10.216323852539062, "learning_rate": 9.915048543689321e-05, "loss": 0.3094, "step": 175 }, { "epoch": 1.702409638554217, "grad_norm": 6.950342655181885, "learning_rate": 9.914563106796117e-05, "loss": 0.5867, "step": 176 }, { "epoch": 1.7120481927710842, "grad_norm": 8.928644180297852, "learning_rate": 9.914077669902914e-05, "loss": 0.4595, "step": 177 }, { "epoch": 1.7216867469879518, "grad_norm": 15.403507232666016, "learning_rate": 9.91359223300971e-05, "loss": 0.196, "step": 178 }, { "epoch": 1.7313253012048193, "grad_norm": 2.501918315887451, "learning_rate": 9.913106796116505e-05, "loss": 0.5005, "step": 179 }, { "epoch": 1.7409638554216866, "grad_norm": 6.640446662902832, "learning_rate": 9.912621359223302e-05, "loss": 0.4877, "step": 180 }, { "epoch": 1.7506024096385542, "grad_norm": 13.597162246704102, "learning_rate": 9.912135922330098e-05, "loss": 0.3758, "step": 181 }, { "epoch": 1.7602409638554217, "grad_norm": 11.171950340270996, "learning_rate": 9.911650485436894e-05, "loss": 0.4316, "step": 182 }, { "epoch": 1.769879518072289, "grad_norm": 2.251030683517456, "learning_rate": 9.91116504854369e-05, "loss": 0.4601, "step": 183 }, { "epoch": 1.7795180722891566, "grad_norm": 11.121167182922363, "learning_rate": 9.910679611650487e-05, "loss": 0.5347, "step": 184 }, { "epoch": 1.7891566265060241, "grad_norm": 16.896238327026367, "learning_rate": 9.910194174757283e-05, "loss": 0.5762, "step": 185 }, { "epoch": 1.7987951807228915, "grad_norm": 2.6431422233581543, "learning_rate": 9.909708737864078e-05, "loss": 0.1195, "step": 186 }, { "epoch": 1.808433734939759, "grad_norm": 15.916812896728516, "learning_rate": 9.909223300970874e-05, "loss": 0.3087, "step": 187 }, { "epoch": 1.8180722891566266, "grad_norm": 5.790849208831787, "learning_rate": 9.90873786407767e-05, "loss": 0.327, "step": 188 }, { "epoch": 1.8277108433734939, "grad_norm": 0.804267942905426, "learning_rate": 9.908252427184467e-05, "loss": 0.3266, "step": 189 }, { "epoch": 1.8373493975903614, "grad_norm": 4.111992835998535, "learning_rate": 9.907766990291263e-05, "loss": 0.524, "step": 190 }, { "epoch": 1.846987951807229, "grad_norm": 4.138038158416748, "learning_rate": 9.907281553398059e-05, "loss": 0.5034, "step": 191 }, { "epoch": 1.8566265060240963, "grad_norm": 13.590168952941895, "learning_rate": 9.906796116504854e-05, "loss": 0.4619, "step": 192 }, { "epoch": 1.8662650602409638, "grad_norm": 6.189218044281006, "learning_rate": 9.906310679611652e-05, "loss": 0.3435, "step": 193 }, { "epoch": 1.8759036144578314, "grad_norm": 9.813295364379883, "learning_rate": 9.905825242718447e-05, "loss": 0.3812, "step": 194 }, { "epoch": 1.8855421686746987, "grad_norm": 23.489343643188477, "learning_rate": 9.905339805825243e-05, "loss": 0.6257, "step": 195 }, { "epoch": 1.8951807228915662, "grad_norm": 17.796062469482422, "learning_rate": 9.904854368932039e-05, "loss": 0.4146, "step": 196 }, { "epoch": 1.9048192771084338, "grad_norm": 15.031364440917969, "learning_rate": 9.904368932038836e-05, "loss": 0.5335, "step": 197 }, { "epoch": 1.9144578313253011, "grad_norm": 13.860004425048828, "learning_rate": 9.903883495145632e-05, "loss": 0.3338, "step": 198 }, { "epoch": 1.9240963855421687, "grad_norm": 8.465548515319824, "learning_rate": 9.903398058252428e-05, "loss": 0.3642, "step": 199 }, { "epoch": 1.9337349397590362, "grad_norm": 9.409234046936035, "learning_rate": 9.902912621359223e-05, "loss": 0.4424, "step": 200 }, { "epoch": 1.9433734939759035, "grad_norm": 4.332907199859619, "learning_rate": 9.90242718446602e-05, "loss": 0.417, "step": 201 }, { "epoch": 1.953012048192771, "grad_norm": 20.38149642944336, "learning_rate": 9.901941747572816e-05, "loss": 0.4807, "step": 202 }, { "epoch": 1.9626506024096386, "grad_norm": 7.942718982696533, "learning_rate": 9.901456310679612e-05, "loss": 0.37, "step": 203 }, { "epoch": 1.972289156626506, "grad_norm": 10.420685768127441, "learning_rate": 9.900970873786408e-05, "loss": 0.3138, "step": 204 }, { "epoch": 1.9819277108433735, "grad_norm": 14.785261154174805, "learning_rate": 9.900485436893204e-05, "loss": 0.5137, "step": 205 }, { "epoch": 1.991566265060241, "grad_norm": 7.8974995613098145, "learning_rate": 9.900000000000001e-05, "loss": 0.5164, "step": 206 }, { "epoch": 2.0072289156626506, "grad_norm": 6.878636837005615, "learning_rate": 9.899514563106797e-05, "loss": 0.1964, "step": 207 }, { "epoch": 2.016867469879518, "grad_norm": 5.306710243225098, "learning_rate": 9.899029126213592e-05, "loss": 0.3327, "step": 208 }, { "epoch": 2.0265060240963857, "grad_norm": 22.76435661315918, "learning_rate": 9.898543689320388e-05, "loss": 0.5003, "step": 209 }, { "epoch": 2.036144578313253, "grad_norm": 10.484804153442383, "learning_rate": 9.898058252427185e-05, "loss": 0.3887, "step": 210 }, { "epoch": 2.0457831325301203, "grad_norm": 17.673585891723633, "learning_rate": 9.897572815533981e-05, "loss": 0.1955, "step": 211 }, { "epoch": 2.055421686746988, "grad_norm": 4.848171234130859, "learning_rate": 9.897087378640777e-05, "loss": 0.2828, "step": 212 }, { "epoch": 2.0650602409638554, "grad_norm": 4.728011608123779, "learning_rate": 9.896601941747573e-05, "loss": 0.2376, "step": 213 }, { "epoch": 2.0746987951807228, "grad_norm": 10.936849594116211, "learning_rate": 9.89611650485437e-05, "loss": 0.4679, "step": 214 }, { "epoch": 2.0843373493975905, "grad_norm": 10.92579460144043, "learning_rate": 9.895631067961165e-05, "loss": 0.399, "step": 215 }, { "epoch": 2.093975903614458, "grad_norm": 11.936409950256348, "learning_rate": 9.895145631067961e-05, "loss": 0.3897, "step": 216 }, { "epoch": 2.103614457831325, "grad_norm": 14.141229629516602, "learning_rate": 9.894660194174757e-05, "loss": 0.3101, "step": 217 }, { "epoch": 2.113253012048193, "grad_norm": 1.4583972692489624, "learning_rate": 9.894174757281554e-05, "loss": 0.3729, "step": 218 }, { "epoch": 2.1228915662650603, "grad_norm": 26.990882873535156, "learning_rate": 9.89368932038835e-05, "loss": 0.369, "step": 219 }, { "epoch": 2.1325301204819276, "grad_norm": 3.5186877250671387, "learning_rate": 9.893203883495146e-05, "loss": 0.3642, "step": 220 }, { "epoch": 2.1421686746987953, "grad_norm": 13.57888126373291, "learning_rate": 9.892718446601941e-05, "loss": 0.4176, "step": 221 }, { "epoch": 2.1518072289156627, "grad_norm": 11.960016250610352, "learning_rate": 9.892233009708737e-05, "loss": 0.281, "step": 222 }, { "epoch": 2.16144578313253, "grad_norm": 95.60669708251953, "learning_rate": 9.891747572815534e-05, "loss": 0.2473, "step": 223 }, { "epoch": 2.1710843373493978, "grad_norm": 170.03921508789062, "learning_rate": 9.89126213592233e-05, "loss": 0.3865, "step": 224 }, { "epoch": 2.180722891566265, "grad_norm": 4.594600200653076, "learning_rate": 9.890776699029126e-05, "loss": 0.3018, "step": 225 }, { "epoch": 2.1903614457831324, "grad_norm": 5.802493572235107, "learning_rate": 9.890291262135922e-05, "loss": 0.3344, "step": 226 }, { "epoch": 2.2, "grad_norm": 2.4812417030334473, "learning_rate": 9.889805825242719e-05, "loss": 0.3963, "step": 227 }, { "epoch": 2.2096385542168675, "grad_norm": 5.9767937660217285, "learning_rate": 9.889320388349516e-05, "loss": 0.1972, "step": 228 }, { "epoch": 2.219277108433735, "grad_norm": 16.531597137451172, "learning_rate": 9.888834951456312e-05, "loss": 0.4407, "step": 229 }, { "epoch": 2.2289156626506026, "grad_norm": 13.233558654785156, "learning_rate": 9.888349514563108e-05, "loss": 0.342, "step": 230 }, { "epoch": 2.23855421686747, "grad_norm": 7.297155380249023, "learning_rate": 9.887864077669903e-05, "loss": 0.8377, "step": 231 }, { "epoch": 2.2481927710843372, "grad_norm": 17.480396270751953, "learning_rate": 9.8873786407767e-05, "loss": 0.4511, "step": 232 }, { "epoch": 2.257831325301205, "grad_norm": 6.12341833114624, "learning_rate": 9.886893203883496e-05, "loss": 0.4099, "step": 233 }, { "epoch": 2.2674698795180723, "grad_norm": 12.639354705810547, "learning_rate": 9.886407766990292e-05, "loss": 0.6082, "step": 234 }, { "epoch": 2.2771084337349397, "grad_norm": 27.298809051513672, "learning_rate": 9.885922330097088e-05, "loss": 0.4357, "step": 235 }, { "epoch": 2.2867469879518074, "grad_norm": 17.494346618652344, "learning_rate": 9.885436893203885e-05, "loss": 0.2865, "step": 236 }, { "epoch": 2.2963855421686747, "grad_norm": 2.6573336124420166, "learning_rate": 9.88495145631068e-05, "loss": 0.361, "step": 237 }, { "epoch": 2.306024096385542, "grad_norm": 25.441619873046875, "learning_rate": 9.884466019417476e-05, "loss": 0.4023, "step": 238 }, { "epoch": 2.31566265060241, "grad_norm": 3.797698736190796, "learning_rate": 9.883980582524272e-05, "loss": 0.244, "step": 239 }, { "epoch": 2.325301204819277, "grad_norm": 42.9473991394043, "learning_rate": 9.88349514563107e-05, "loss": 0.4905, "step": 240 }, { "epoch": 2.3349397590361445, "grad_norm": 4.624748229980469, "learning_rate": 9.883009708737865e-05, "loss": 0.3059, "step": 241 }, { "epoch": 2.3445783132530122, "grad_norm": 9.89862060546875, "learning_rate": 9.882524271844661e-05, "loss": 0.3255, "step": 242 }, { "epoch": 2.3542168674698796, "grad_norm": 4.885334491729736, "learning_rate": 9.882038834951457e-05, "loss": 0.2182, "step": 243 }, { "epoch": 2.363855421686747, "grad_norm": 21.016395568847656, "learning_rate": 9.881553398058252e-05, "loss": 0.4267, "step": 244 }, { "epoch": 2.3734939759036147, "grad_norm": 16.577579498291016, "learning_rate": 9.88106796116505e-05, "loss": 0.3437, "step": 245 }, { "epoch": 2.383132530120482, "grad_norm": 15.13383674621582, "learning_rate": 9.880582524271845e-05, "loss": 0.2688, "step": 246 }, { "epoch": 2.3927710843373493, "grad_norm": 3.28494930267334, "learning_rate": 9.880097087378641e-05, "loss": 0.3788, "step": 247 }, { "epoch": 2.402409638554217, "grad_norm": 4.29710578918457, "learning_rate": 9.879611650485437e-05, "loss": 0.3107, "step": 248 }, { "epoch": 2.4120481927710844, "grad_norm": 4.441616058349609, "learning_rate": 9.879126213592234e-05, "loss": 0.5814, "step": 249 }, { "epoch": 2.4216867469879517, "grad_norm": 10.997577667236328, "learning_rate": 9.87864077669903e-05, "loss": 0.3012, "step": 250 }, { "epoch": 2.4313253012048195, "grad_norm": 7.565840721130371, "learning_rate": 9.878155339805826e-05, "loss": 0.2657, "step": 251 }, { "epoch": 2.440963855421687, "grad_norm": 12.89657211303711, "learning_rate": 9.877669902912621e-05, "loss": 0.4819, "step": 252 }, { "epoch": 2.450602409638554, "grad_norm": 22.417198181152344, "learning_rate": 9.877184466019418e-05, "loss": 0.5762, "step": 253 }, { "epoch": 2.460240963855422, "grad_norm": 2.254826068878174, "learning_rate": 9.876699029126214e-05, "loss": 0.4714, "step": 254 }, { "epoch": 2.4698795180722892, "grad_norm": 18.5875301361084, "learning_rate": 9.87621359223301e-05, "loss": 0.4355, "step": 255 }, { "epoch": 2.4795180722891565, "grad_norm": 6.165832996368408, "learning_rate": 9.875728155339806e-05, "loss": 0.5457, "step": 256 }, { "epoch": 2.4891566265060243, "grad_norm": 17.457088470458984, "learning_rate": 9.875242718446603e-05, "loss": 0.3453, "step": 257 }, { "epoch": 2.4987951807228916, "grad_norm": 13.856874465942383, "learning_rate": 9.874757281553399e-05, "loss": 0.5603, "step": 258 }, { "epoch": 2.508433734939759, "grad_norm": 5.642117500305176, "learning_rate": 9.874271844660194e-05, "loss": 0.4474, "step": 259 }, { "epoch": 2.5180722891566267, "grad_norm": 24.100811004638672, "learning_rate": 9.87378640776699e-05, "loss": 0.5442, "step": 260 }, { "epoch": 2.527710843373494, "grad_norm": 17.965471267700195, "learning_rate": 9.873300970873786e-05, "loss": 0.3377, "step": 261 }, { "epoch": 2.5373493975903614, "grad_norm": 3.2536072731018066, "learning_rate": 9.872815533980583e-05, "loss": 0.5581, "step": 262 }, { "epoch": 2.546987951807229, "grad_norm": 27.890121459960938, "learning_rate": 9.872330097087379e-05, "loss": 0.2462, "step": 263 }, { "epoch": 2.5566265060240965, "grad_norm": 10.826033592224121, "learning_rate": 9.871844660194175e-05, "loss": 0.3259, "step": 264 }, { "epoch": 2.566265060240964, "grad_norm": 3.6546010971069336, "learning_rate": 9.87135922330097e-05, "loss": 0.5901, "step": 265 }, { "epoch": 2.5759036144578316, "grad_norm": 19.892696380615234, "learning_rate": 9.870873786407768e-05, "loss": 0.4437, "step": 266 }, { "epoch": 2.585542168674699, "grad_norm": 11.988555908203125, "learning_rate": 9.870388349514563e-05, "loss": 0.3988, "step": 267 }, { "epoch": 2.595180722891566, "grad_norm": 4.279914855957031, "learning_rate": 9.869902912621359e-05, "loss": 0.5371, "step": 268 }, { "epoch": 2.604819277108434, "grad_norm": 8.743132591247559, "learning_rate": 9.869417475728155e-05, "loss": 0.3984, "step": 269 }, { "epoch": 2.6144578313253013, "grad_norm": 11.579974174499512, "learning_rate": 9.868932038834952e-05, "loss": 0.3211, "step": 270 }, { "epoch": 2.6240963855421686, "grad_norm": 16.063478469848633, "learning_rate": 9.868446601941748e-05, "loss": 0.5952, "step": 271 }, { "epoch": 2.6337349397590364, "grad_norm": 4.782046794891357, "learning_rate": 9.867961165048544e-05, "loss": 0.3432, "step": 272 }, { "epoch": 2.6433734939759037, "grad_norm": 30.110902786254883, "learning_rate": 9.86747572815534e-05, "loss": 0.3318, "step": 273 }, { "epoch": 2.653012048192771, "grad_norm": 1.944240927696228, "learning_rate": 9.866990291262137e-05, "loss": 0.1128, "step": 274 }, { "epoch": 2.662650602409639, "grad_norm": 2.2361717224121094, "learning_rate": 9.866504854368932e-05, "loss": 0.4804, "step": 275 }, { "epoch": 2.672289156626506, "grad_norm": 9.983328819274902, "learning_rate": 9.866019417475728e-05, "loss": 0.5448, "step": 276 }, { "epoch": 2.6819277108433734, "grad_norm": 5.701781272888184, "learning_rate": 9.865533980582524e-05, "loss": 0.3981, "step": 277 }, { "epoch": 2.691566265060241, "grad_norm": 27.209430694580078, "learning_rate": 9.865048543689321e-05, "loss": 0.5805, "step": 278 }, { "epoch": 2.7012048192771085, "grad_norm": 13.092561721801758, "learning_rate": 9.864563106796118e-05, "loss": 0.2604, "step": 279 }, { "epoch": 2.710843373493976, "grad_norm": 2.5776913166046143, "learning_rate": 9.864077669902914e-05, "loss": 0.477, "step": 280 }, { "epoch": 2.7204819277108436, "grad_norm": 6.065176963806152, "learning_rate": 9.86359223300971e-05, "loss": 0.2503, "step": 281 }, { "epoch": 2.730120481927711, "grad_norm": 3.190626621246338, "learning_rate": 9.863106796116505e-05, "loss": 0.2147, "step": 282 }, { "epoch": 2.7397590361445783, "grad_norm": 4.741774082183838, "learning_rate": 9.862621359223301e-05, "loss": 0.3074, "step": 283 }, { "epoch": 2.749397590361446, "grad_norm": 5.060323715209961, "learning_rate": 9.862135922330098e-05, "loss": 0.4318, "step": 284 }, { "epoch": 2.7590361445783134, "grad_norm": 11.733187675476074, "learning_rate": 9.861650485436894e-05, "loss": 0.497, "step": 285 }, { "epoch": 2.7686746987951807, "grad_norm": 12.384824752807617, "learning_rate": 9.86116504854369e-05, "loss": 0.3531, "step": 286 }, { "epoch": 2.7783132530120485, "grad_norm": 2.850745916366577, "learning_rate": 9.860679611650486e-05, "loss": 0.4023, "step": 287 }, { "epoch": 2.787951807228916, "grad_norm": 7.489830017089844, "learning_rate": 9.860194174757283e-05, "loss": 0.6008, "step": 288 }, { "epoch": 2.797590361445783, "grad_norm": 6.346521854400635, "learning_rate": 9.859708737864079e-05, "loss": 0.4823, "step": 289 }, { "epoch": 2.807228915662651, "grad_norm": 15.724376678466797, "learning_rate": 9.859223300970874e-05, "loss": 0.29, "step": 290 }, { "epoch": 2.816867469879518, "grad_norm": 12.846384048461914, "learning_rate": 9.85873786407767e-05, "loss": 0.4116, "step": 291 }, { "epoch": 2.8265060240963855, "grad_norm": 10.8715181350708, "learning_rate": 9.858252427184467e-05, "loss": 0.4754, "step": 292 }, { "epoch": 2.8361445783132533, "grad_norm": 16.69060516357422, "learning_rate": 9.857766990291263e-05, "loss": 0.4905, "step": 293 }, { "epoch": 2.8457831325301206, "grad_norm": 20.97838020324707, "learning_rate": 9.857281553398059e-05, "loss": 0.5451, "step": 294 }, { "epoch": 2.855421686746988, "grad_norm": 3.1126623153686523, "learning_rate": 9.856796116504855e-05, "loss": 0.3372, "step": 295 }, { "epoch": 2.8650602409638557, "grad_norm": 8.112845420837402, "learning_rate": 9.856310679611652e-05, "loss": 0.6201, "step": 296 }, { "epoch": 2.874698795180723, "grad_norm": 6.4273529052734375, "learning_rate": 9.855825242718448e-05, "loss": 0.517, "step": 297 }, { "epoch": 2.8843373493975903, "grad_norm": 21.647005081176758, "learning_rate": 9.855339805825243e-05, "loss": 0.3237, "step": 298 }, { "epoch": 2.8939759036144577, "grad_norm": 4.403151035308838, "learning_rate": 9.854854368932039e-05, "loss": 0.2729, "step": 299 }, { "epoch": 2.9036144578313254, "grad_norm": 3.295827627182007, "learning_rate": 9.854368932038835e-05, "loss": 0.2639, "step": 300 }, { "epoch": 2.9132530120481928, "grad_norm": 3.4065756797790527, "learning_rate": 9.853883495145632e-05, "loss": 0.3482, "step": 301 }, { "epoch": 2.92289156626506, "grad_norm": 3.218249797821045, "learning_rate": 9.853398058252428e-05, "loss": 0.453, "step": 302 }, { "epoch": 2.932530120481928, "grad_norm": 7.816202163696289, "learning_rate": 9.852912621359224e-05, "loss": 0.3446, "step": 303 }, { "epoch": 2.942168674698795, "grad_norm": 10.051136016845703, "learning_rate": 9.852427184466019e-05, "loss": 0.505, "step": 304 }, { "epoch": 2.9518072289156625, "grad_norm": 6.133807182312012, "learning_rate": 9.851941747572816e-05, "loss": 0.4106, "step": 305 }, { "epoch": 2.9614457831325303, "grad_norm": 17.848154067993164, "learning_rate": 9.851456310679612e-05, "loss": 0.4359, "step": 306 }, { "epoch": 2.9710843373493976, "grad_norm": 17.342662811279297, "learning_rate": 9.850970873786408e-05, "loss": 0.4404, "step": 307 }, { "epoch": 2.980722891566265, "grad_norm": 6.208151340484619, "learning_rate": 9.850485436893204e-05, "loss": 0.3672, "step": 308 }, { "epoch": 2.9903614457831327, "grad_norm": 25.10737419128418, "learning_rate": 9.850000000000001e-05, "loss": 0.3826, "step": 309 }, { "epoch": 3.0060240963855422, "grad_norm": 21.584033966064453, "learning_rate": 9.849514563106797e-05, "loss": 0.2967, "step": 310 }, { "epoch": 3.0156626506024096, "grad_norm": 4.147687911987305, "learning_rate": 9.849029126213592e-05, "loss": 0.3935, "step": 311 }, { "epoch": 3.025301204819277, "grad_norm": 2.6770193576812744, "learning_rate": 9.848543689320388e-05, "loss": 0.238, "step": 312 }, { "epoch": 3.0349397590361447, "grad_norm": 6.403909206390381, "learning_rate": 9.848058252427185e-05, "loss": 0.2805, "step": 313 }, { "epoch": 3.044578313253012, "grad_norm": 24.551992416381836, "learning_rate": 9.847572815533981e-05, "loss": 0.3035, "step": 314 }, { "epoch": 3.0542168674698793, "grad_norm": 4.704305171966553, "learning_rate": 9.847087378640777e-05, "loss": 0.2829, "step": 315 }, { "epoch": 3.063855421686747, "grad_norm": 17.196962356567383, "learning_rate": 9.846601941747573e-05, "loss": 0.2349, "step": 316 }, { "epoch": 3.0734939759036144, "grad_norm": 4.194557189941406, "learning_rate": 9.846116504854368e-05, "loss": 0.539, "step": 317 }, { "epoch": 3.0831325301204817, "grad_norm": 3.882075548171997, "learning_rate": 9.845631067961166e-05, "loss": 0.2978, "step": 318 }, { "epoch": 3.0927710843373495, "grad_norm": 6.577015399932861, "learning_rate": 9.845145631067961e-05, "loss": 0.3498, "step": 319 }, { "epoch": 3.102409638554217, "grad_norm": 15.55713939666748, "learning_rate": 9.844660194174757e-05, "loss": 0.4157, "step": 320 }, { "epoch": 3.112048192771084, "grad_norm": 4.560832977294922, "learning_rate": 9.844174757281553e-05, "loss": 0.2679, "step": 321 }, { "epoch": 3.121686746987952, "grad_norm": 2.364003896713257, "learning_rate": 9.84368932038835e-05, "loss": 0.3135, "step": 322 }, { "epoch": 3.1313253012048192, "grad_norm": 3.1542439460754395, "learning_rate": 9.843203883495146e-05, "loss": 0.3525, "step": 323 }, { "epoch": 3.1409638554216865, "grad_norm": 8.096260070800781, "learning_rate": 9.842718446601942e-05, "loss": 0.5209, "step": 324 }, { "epoch": 3.1506024096385543, "grad_norm": 2.3288137912750244, "learning_rate": 9.842233009708737e-05, "loss": 0.61, "step": 325 }, { "epoch": 3.1602409638554216, "grad_norm": 3.69069766998291, "learning_rate": 9.841747572815535e-05, "loss": 0.3042, "step": 326 }, { "epoch": 3.169879518072289, "grad_norm": 3.5665507316589355, "learning_rate": 9.84126213592233e-05, "loss": 0.3673, "step": 327 }, { "epoch": 3.1795180722891567, "grad_norm": 5.114205360412598, "learning_rate": 9.840776699029126e-05, "loss": 0.2643, "step": 328 }, { "epoch": 3.189156626506024, "grad_norm": 16.272281646728516, "learning_rate": 9.840291262135923e-05, "loss": 0.3686, "step": 329 }, { "epoch": 3.1987951807228914, "grad_norm": 0.8705376386642456, "learning_rate": 9.839805825242719e-05, "loss": 0.2834, "step": 330 }, { "epoch": 3.208433734939759, "grad_norm": 5.177937030792236, "learning_rate": 9.839320388349516e-05, "loss": 0.5481, "step": 331 }, { "epoch": 3.2180722891566265, "grad_norm": 5.056303977966309, "learning_rate": 9.838834951456312e-05, "loss": 0.3886, "step": 332 }, { "epoch": 3.227710843373494, "grad_norm": 10.20246696472168, "learning_rate": 9.838349514563108e-05, "loss": 0.2525, "step": 333 }, { "epoch": 3.2373493975903616, "grad_norm": 10.188186645507812, "learning_rate": 9.837864077669903e-05, "loss": 0.2084, "step": 334 }, { "epoch": 3.246987951807229, "grad_norm": 3.0801644325256348, "learning_rate": 9.8373786407767e-05, "loss": 0.2797, "step": 335 }, { "epoch": 3.256626506024096, "grad_norm": 5.198385715484619, "learning_rate": 9.836893203883496e-05, "loss": 0.5405, "step": 336 }, { "epoch": 3.266265060240964, "grad_norm": 10.135032653808594, "learning_rate": 9.836407766990292e-05, "loss": 0.409, "step": 337 }, { "epoch": 3.2759036144578313, "grad_norm": 10.360496520996094, "learning_rate": 9.835922330097088e-05, "loss": 0.365, "step": 338 }, { "epoch": 3.2855421686746986, "grad_norm": 45.94935989379883, "learning_rate": 9.835436893203884e-05, "loss": 0.4112, "step": 339 }, { "epoch": 3.2951807228915664, "grad_norm": 33.22359848022461, "learning_rate": 9.834951456310681e-05, "loss": 0.6038, "step": 340 }, { "epoch": 3.3048192771084337, "grad_norm": 14.917102813720703, "learning_rate": 9.834466019417477e-05, "loss": 0.5923, "step": 341 }, { "epoch": 3.314457831325301, "grad_norm": 20.41253089904785, "learning_rate": 9.833980582524272e-05, "loss": 0.4485, "step": 342 }, { "epoch": 3.324096385542169, "grad_norm": 6.451887130737305, "learning_rate": 9.833495145631068e-05, "loss": 0.3836, "step": 343 }, { "epoch": 3.333734939759036, "grad_norm": 11.612648963928223, "learning_rate": 9.833009708737865e-05, "loss": 0.2685, "step": 344 }, { "epoch": 3.3433734939759034, "grad_norm": 6.957334518432617, "learning_rate": 9.832524271844661e-05, "loss": 0.3717, "step": 345 }, { "epoch": 3.353012048192771, "grad_norm": 5.660593509674072, "learning_rate": 9.832038834951457e-05, "loss": 0.3139, "step": 346 }, { "epoch": 3.3626506024096385, "grad_norm": 4.2149834632873535, "learning_rate": 9.831553398058253e-05, "loss": 0.2433, "step": 347 }, { "epoch": 3.372289156626506, "grad_norm": 25.811721801757812, "learning_rate": 9.83106796116505e-05, "loss": 0.1895, "step": 348 }, { "epoch": 3.3819277108433736, "grad_norm": 3.7745108604431152, "learning_rate": 9.830582524271845e-05, "loss": 0.3437, "step": 349 }, { "epoch": 3.391566265060241, "grad_norm": 14.111804008483887, "learning_rate": 9.830097087378641e-05, "loss": 0.4596, "step": 350 }, { "epoch": 3.4012048192771083, "grad_norm": 27.494613647460938, "learning_rate": 9.829611650485437e-05, "loss": 0.453, "step": 351 }, { "epoch": 3.410843373493976, "grad_norm": 16.2795467376709, "learning_rate": 9.829126213592234e-05, "loss": 0.293, "step": 352 }, { "epoch": 3.4204819277108434, "grad_norm": 14.73820972442627, "learning_rate": 9.82864077669903e-05, "loss": 0.349, "step": 353 }, { "epoch": 3.4301204819277107, "grad_norm": 22.2900390625, "learning_rate": 9.828155339805826e-05, "loss": 0.3302, "step": 354 }, { "epoch": 3.4397590361445785, "grad_norm": 6.462294578552246, "learning_rate": 9.827669902912621e-05, "loss": 0.4262, "step": 355 }, { "epoch": 3.4493975903614458, "grad_norm": 9.658364295959473, "learning_rate": 9.827184466019417e-05, "loss": 0.5625, "step": 356 }, { "epoch": 3.459036144578313, "grad_norm": 5.045487880706787, "learning_rate": 9.826699029126214e-05, "loss": 0.3285, "step": 357 }, { "epoch": 3.468674698795181, "grad_norm": 3.600754499435425, "learning_rate": 9.82621359223301e-05, "loss": 0.4423, "step": 358 }, { "epoch": 3.478313253012048, "grad_norm": 4.518320083618164, "learning_rate": 9.825728155339806e-05, "loss": 0.4004, "step": 359 }, { "epoch": 3.4879518072289155, "grad_norm": 8.083192825317383, "learning_rate": 9.825242718446602e-05, "loss": 0.4032, "step": 360 }, { "epoch": 3.4975903614457833, "grad_norm": 9.587786674499512, "learning_rate": 9.824757281553399e-05, "loss": 0.424, "step": 361 }, { "epoch": 3.5072289156626506, "grad_norm": 4.2372822761535645, "learning_rate": 9.824271844660195e-05, "loss": 0.2306, "step": 362 }, { "epoch": 3.516867469879518, "grad_norm": 20.796327590942383, "learning_rate": 9.82378640776699e-05, "loss": 0.3794, "step": 363 }, { "epoch": 3.5265060240963857, "grad_norm": 2.6187992095947266, "learning_rate": 9.823300970873786e-05, "loss": 0.3404, "step": 364 }, { "epoch": 3.536144578313253, "grad_norm": 20.706758499145508, "learning_rate": 9.822815533980583e-05, "loss": 0.5544, "step": 365 }, { "epoch": 3.5457831325301203, "grad_norm": 5.865291595458984, "learning_rate": 9.822330097087379e-05, "loss": 0.3843, "step": 366 }, { "epoch": 3.555421686746988, "grad_norm": 7.424984455108643, "learning_rate": 9.821844660194175e-05, "loss": 0.2532, "step": 367 }, { "epoch": 3.5650602409638554, "grad_norm": 4.551677227020264, "learning_rate": 9.82135922330097e-05, "loss": 0.2238, "step": 368 }, { "epoch": 3.5746987951807228, "grad_norm": 15.146475791931152, "learning_rate": 9.820873786407768e-05, "loss": 0.5466, "step": 369 }, { "epoch": 3.5843373493975905, "grad_norm": 3.522535800933838, "learning_rate": 9.820388349514564e-05, "loss": 0.3243, "step": 370 }, { "epoch": 3.593975903614458, "grad_norm": 3.355036497116089, "learning_rate": 9.81990291262136e-05, "loss": 0.319, "step": 371 }, { "epoch": 3.603614457831325, "grad_norm": 7.781651020050049, "learning_rate": 9.819417475728155e-05, "loss": 0.4378, "step": 372 }, { "epoch": 3.613253012048193, "grad_norm": 17.717716217041016, "learning_rate": 9.818932038834951e-05, "loss": 0.4181, "step": 373 }, { "epoch": 3.6228915662650603, "grad_norm": 7.970596790313721, "learning_rate": 9.818446601941748e-05, "loss": 0.4801, "step": 374 }, { "epoch": 3.6325301204819276, "grad_norm": 3.317416191101074, "learning_rate": 9.817961165048544e-05, "loss": 0.3623, "step": 375 }, { "epoch": 3.6421686746987953, "grad_norm": 7.43623161315918, "learning_rate": 9.81747572815534e-05, "loss": 0.264, "step": 376 }, { "epoch": 3.6518072289156627, "grad_norm": 2.346316337585449, "learning_rate": 9.816990291262135e-05, "loss": 0.3079, "step": 377 }, { "epoch": 3.66144578313253, "grad_norm": 15.61959171295166, "learning_rate": 9.816504854368932e-05, "loss": 0.4136, "step": 378 }, { "epoch": 3.6710843373493978, "grad_norm": 3.4539170265197754, "learning_rate": 9.81601941747573e-05, "loss": 0.44, "step": 379 }, { "epoch": 3.680722891566265, "grad_norm": 9.218192100524902, "learning_rate": 9.815533980582525e-05, "loss": 0.4223, "step": 380 }, { "epoch": 3.6903614457831324, "grad_norm": 3.86674427986145, "learning_rate": 9.815048543689321e-05, "loss": 0.4025, "step": 381 }, { "epoch": 3.7, "grad_norm": 4.067527770996094, "learning_rate": 9.814563106796117e-05, "loss": 0.2721, "step": 382 }, { "epoch": 3.7096385542168675, "grad_norm": 12.99213981628418, "learning_rate": 9.814077669902914e-05, "loss": 0.3861, "step": 383 }, { "epoch": 3.719277108433735, "grad_norm": 15.263169288635254, "learning_rate": 9.81359223300971e-05, "loss": 0.2692, "step": 384 }, { "epoch": 3.7289156626506026, "grad_norm": 2.2644312381744385, "learning_rate": 9.813106796116506e-05, "loss": 0.4258, "step": 385 }, { "epoch": 3.73855421686747, "grad_norm": 14.439421653747559, "learning_rate": 9.812621359223301e-05, "loss": 0.4732, "step": 386 }, { "epoch": 3.7481927710843372, "grad_norm": 11.313796043395996, "learning_rate": 9.812135922330099e-05, "loss": 0.5839, "step": 387 }, { "epoch": 3.757831325301205, "grad_norm": 5.473857402801514, "learning_rate": 9.811650485436894e-05, "loss": 0.4415, "step": 388 }, { "epoch": 3.7674698795180723, "grad_norm": 4.994795799255371, "learning_rate": 9.81116504854369e-05, "loss": 0.4435, "step": 389 }, { "epoch": 3.7771084337349397, "grad_norm": 7.3361029624938965, "learning_rate": 9.810679611650486e-05, "loss": 0.4056, "step": 390 }, { "epoch": 3.7867469879518074, "grad_norm": 8.515096664428711, "learning_rate": 9.810194174757283e-05, "loss": 0.4878, "step": 391 }, { "epoch": 3.7963855421686747, "grad_norm": 15.866056442260742, "learning_rate": 9.809708737864079e-05, "loss": 0.4077, "step": 392 }, { "epoch": 3.806024096385542, "grad_norm": 12.353825569152832, "learning_rate": 9.809223300970875e-05, "loss": 0.3861, "step": 393 }, { "epoch": 3.81566265060241, "grad_norm": 1.8780932426452637, "learning_rate": 9.80873786407767e-05, "loss": 0.3572, "step": 394 }, { "epoch": 3.825301204819277, "grad_norm": 6.116151809692383, "learning_rate": 9.808252427184466e-05, "loss": 0.273, "step": 395 }, { "epoch": 3.8349397590361445, "grad_norm": 6.098775863647461, "learning_rate": 9.807766990291263e-05, "loss": 0.3803, "step": 396 }, { "epoch": 3.8445783132530122, "grad_norm": 13.579024314880371, "learning_rate": 9.807281553398059e-05, "loss": 0.357, "step": 397 }, { "epoch": 3.8542168674698796, "grad_norm": 36.76884841918945, "learning_rate": 9.806796116504855e-05, "loss": 0.1987, "step": 398 }, { "epoch": 3.863855421686747, "grad_norm": 10.541704177856445, "learning_rate": 9.80631067961165e-05, "loss": 0.3264, "step": 399 }, { "epoch": 3.8734939759036147, "grad_norm": 4.43033504486084, "learning_rate": 9.805825242718448e-05, "loss": 0.415, "step": 400 }, { "epoch": 3.883132530120482, "grad_norm": 3.02974009513855, "learning_rate": 9.805339805825243e-05, "loss": 0.4754, "step": 401 }, { "epoch": 3.8927710843373493, "grad_norm": 3.894240617752075, "learning_rate": 9.804854368932039e-05, "loss": 0.317, "step": 402 }, { "epoch": 3.902409638554217, "grad_norm": 6.8115010261535645, "learning_rate": 9.804368932038835e-05, "loss": 0.3305, "step": 403 }, { "epoch": 3.9120481927710844, "grad_norm": 6.58502721786499, "learning_rate": 9.803883495145632e-05, "loss": 0.5248, "step": 404 }, { "epoch": 3.9216867469879517, "grad_norm": 5.67113733291626, "learning_rate": 9.803398058252428e-05, "loss": 0.4866, "step": 405 }, { "epoch": 3.9313253012048195, "grad_norm": 23.920162200927734, "learning_rate": 9.802912621359224e-05, "loss": 0.5801, "step": 406 }, { "epoch": 3.940963855421687, "grad_norm": 2.9804444313049316, "learning_rate": 9.80242718446602e-05, "loss": 0.3695, "step": 407 }, { "epoch": 3.950602409638554, "grad_norm": 7.569250583648682, "learning_rate": 9.801941747572817e-05, "loss": 0.4071, "step": 408 }, { "epoch": 3.960240963855422, "grad_norm": 3.1722466945648193, "learning_rate": 9.801456310679612e-05, "loss": 0.3283, "step": 409 }, { "epoch": 3.9698795180722892, "grad_norm": 15.818419456481934, "learning_rate": 9.800970873786408e-05, "loss": 0.5281, "step": 410 }, { "epoch": 3.9795180722891565, "grad_norm": 6.693188667297363, "learning_rate": 9.800485436893204e-05, "loss": 0.2361, "step": 411 }, { "epoch": 3.9891566265060243, "grad_norm": 13.513023376464844, "learning_rate": 9.8e-05, "loss": 0.2469, "step": 412 }, { "epoch": 4.004819277108433, "grad_norm": 16.08788299560547, "learning_rate": 9.799514563106797e-05, "loss": 0.5318, "step": 413 }, { "epoch": 4.014457831325301, "grad_norm": 7.556865215301514, "learning_rate": 9.799029126213593e-05, "loss": 0.4318, "step": 414 }, { "epoch": 4.024096385542169, "grad_norm": 2.697539806365967, "learning_rate": 9.798543689320388e-05, "loss": 0.1511, "step": 415 }, { "epoch": 4.033734939759036, "grad_norm": 2.829149007797241, "learning_rate": 9.798058252427184e-05, "loss": 0.2263, "step": 416 }, { "epoch": 4.043373493975904, "grad_norm": 2.099283218383789, "learning_rate": 9.797572815533981e-05, "loss": 0.3768, "step": 417 }, { "epoch": 4.053012048192771, "grad_norm": 4.590118408203125, "learning_rate": 9.797087378640777e-05, "loss": 0.5811, "step": 418 }, { "epoch": 4.062650602409638, "grad_norm": 8.032024383544922, "learning_rate": 9.796601941747573e-05, "loss": 0.3949, "step": 419 }, { "epoch": 4.072289156626506, "grad_norm": 3.6267850399017334, "learning_rate": 9.796116504854369e-05, "loss": 0.3456, "step": 420 }, { "epoch": 4.081927710843374, "grad_norm": 29.117862701416016, "learning_rate": 9.795631067961166e-05, "loss": 0.4888, "step": 421 }, { "epoch": 4.091566265060241, "grad_norm": 17.16216278076172, "learning_rate": 9.795145631067962e-05, "loss": 0.3271, "step": 422 }, { "epoch": 4.1012048192771084, "grad_norm": 2.681671619415283, "learning_rate": 9.794660194174757e-05, "loss": 0.2027, "step": 423 }, { "epoch": 4.110843373493976, "grad_norm": 18.51841163635254, "learning_rate": 9.794174757281553e-05, "loss": 0.4805, "step": 424 }, { "epoch": 4.120481927710843, "grad_norm": 22.4951229095459, "learning_rate": 9.79368932038835e-05, "loss": 0.6594, "step": 425 }, { "epoch": 4.130120481927711, "grad_norm": 16.598773956298828, "learning_rate": 9.793203883495146e-05, "loss": 0.4198, "step": 426 }, { "epoch": 4.139759036144579, "grad_norm": 15.034675598144531, "learning_rate": 9.792718446601942e-05, "loss": 0.4282, "step": 427 }, { "epoch": 4.1493975903614455, "grad_norm": 6.4159321784973145, "learning_rate": 9.792233009708738e-05, "loss": 0.3168, "step": 428 }, { "epoch": 4.159036144578313, "grad_norm": 3.6176187992095947, "learning_rate": 9.791747572815533e-05, "loss": 0.2982, "step": 429 }, { "epoch": 4.168674698795181, "grad_norm": 21.37215805053711, "learning_rate": 9.791262135922332e-05, "loss": 0.4228, "step": 430 }, { "epoch": 4.178313253012048, "grad_norm": 8.743306159973145, "learning_rate": 9.790776699029128e-05, "loss": 0.3849, "step": 431 }, { "epoch": 4.187951807228916, "grad_norm": 4.8714118003845215, "learning_rate": 9.790291262135923e-05, "loss": 0.4176, "step": 432 }, { "epoch": 4.1975903614457835, "grad_norm": 9.536160469055176, "learning_rate": 9.789805825242719e-05, "loss": 0.5178, "step": 433 }, { "epoch": 4.20722891566265, "grad_norm": 2.8080973625183105, "learning_rate": 9.789320388349515e-05, "loss": 0.3479, "step": 434 }, { "epoch": 4.216867469879518, "grad_norm": 8.09260082244873, "learning_rate": 9.788834951456312e-05, "loss": 0.3212, "step": 435 }, { "epoch": 4.226506024096386, "grad_norm": 6.906187057495117, "learning_rate": 9.788349514563108e-05, "loss": 0.563, "step": 436 }, { "epoch": 4.236144578313253, "grad_norm": 3.930257558822632, "learning_rate": 9.787864077669904e-05, "loss": 0.3108, "step": 437 }, { "epoch": 4.2457831325301205, "grad_norm": 8.514981269836426, "learning_rate": 9.7873786407767e-05, "loss": 0.3783, "step": 438 }, { "epoch": 4.255421686746988, "grad_norm": 3.9780170917510986, "learning_rate": 9.786893203883496e-05, "loss": 0.4468, "step": 439 }, { "epoch": 4.265060240963855, "grad_norm": 3.4179248809814453, "learning_rate": 9.786407766990292e-05, "loss": 0.166, "step": 440 }, { "epoch": 4.274698795180723, "grad_norm": 16.926742553710938, "learning_rate": 9.785922330097088e-05, "loss": 0.3367, "step": 441 }, { "epoch": 4.284337349397591, "grad_norm": 8.009011268615723, "learning_rate": 9.785436893203884e-05, "loss": 0.3803, "step": 442 }, { "epoch": 4.293975903614458, "grad_norm": 6.151138782501221, "learning_rate": 9.784951456310681e-05, "loss": 0.5108, "step": 443 }, { "epoch": 4.303614457831325, "grad_norm": 1.8489304780960083, "learning_rate": 9.784466019417477e-05, "loss": 0.3334, "step": 444 }, { "epoch": 4.313253012048193, "grad_norm": 14.699933052062988, "learning_rate": 9.783980582524272e-05, "loss": 0.3766, "step": 445 }, { "epoch": 4.32289156626506, "grad_norm": 3.076385974884033, "learning_rate": 9.783495145631068e-05, "loss": 0.318, "step": 446 }, { "epoch": 4.332530120481928, "grad_norm": 9.761015892028809, "learning_rate": 9.783009708737864e-05, "loss": 0.6094, "step": 447 }, { "epoch": 4.3421686746987955, "grad_norm": 3.408341407775879, "learning_rate": 9.782524271844661e-05, "loss": 0.3132, "step": 448 }, { "epoch": 4.351807228915662, "grad_norm": 17.295690536499023, "learning_rate": 9.782038834951457e-05, "loss": 0.4108, "step": 449 }, { "epoch": 4.36144578313253, "grad_norm": 15.83061695098877, "learning_rate": 9.781553398058253e-05, "loss": 0.4261, "step": 450 }, { "epoch": 4.371084337349398, "grad_norm": 6.8014302253723145, "learning_rate": 9.781067961165048e-05, "loss": 0.3808, "step": 451 }, { "epoch": 4.380722891566265, "grad_norm": 3.952319622039795, "learning_rate": 9.780582524271846e-05, "loss": 0.2609, "step": 452 }, { "epoch": 4.390361445783133, "grad_norm": 5.225638389587402, "learning_rate": 9.780097087378641e-05, "loss": 0.3521, "step": 453 }, { "epoch": 4.4, "grad_norm": 19.15594482421875, "learning_rate": 9.779611650485437e-05, "loss": 0.3079, "step": 454 }, { "epoch": 4.409638554216867, "grad_norm": 11.074477195739746, "learning_rate": 9.779126213592233e-05, "loss": 0.4722, "step": 455 }, { "epoch": 4.419277108433735, "grad_norm": 5.34878396987915, "learning_rate": 9.77864077669903e-05, "loss": 0.3417, "step": 456 }, { "epoch": 4.428915662650603, "grad_norm": 3.628394603729248, "learning_rate": 9.778155339805826e-05, "loss": 0.2769, "step": 457 }, { "epoch": 4.43855421686747, "grad_norm": 4.62729549407959, "learning_rate": 9.777669902912622e-05, "loss": 0.4333, "step": 458 }, { "epoch": 4.448192771084337, "grad_norm": 3.5172879695892334, "learning_rate": 9.777184466019417e-05, "loss": 0.3523, "step": 459 }, { "epoch": 4.457831325301205, "grad_norm": 11.123435020446777, "learning_rate": 9.776699029126215e-05, "loss": 0.3553, "step": 460 }, { "epoch": 4.467469879518072, "grad_norm": 4.4058027267456055, "learning_rate": 9.77621359223301e-05, "loss": 0.2088, "step": 461 }, { "epoch": 4.47710843373494, "grad_norm": 23.35591697692871, "learning_rate": 9.775728155339806e-05, "loss": 0.5386, "step": 462 }, { "epoch": 4.486746987951808, "grad_norm": 11.836080551147461, "learning_rate": 9.775242718446602e-05, "loss": 0.3891, "step": 463 }, { "epoch": 4.4963855421686745, "grad_norm": 4.056002140045166, "learning_rate": 9.774757281553398e-05, "loss": 0.4364, "step": 464 }, { "epoch": 4.506024096385542, "grad_norm": 9.640485763549805, "learning_rate": 9.774271844660195e-05, "loss": 0.7388, "step": 465 }, { "epoch": 4.51566265060241, "grad_norm": 110.9031982421875, "learning_rate": 9.77378640776699e-05, "loss": 0.5235, "step": 466 }, { "epoch": 4.525301204819277, "grad_norm": 18.513212203979492, "learning_rate": 9.773300970873786e-05, "loss": 0.4734, "step": 467 }, { "epoch": 4.534939759036145, "grad_norm": 4.112464904785156, "learning_rate": 9.772815533980582e-05, "loss": 0.302, "step": 468 }, { "epoch": 4.544578313253012, "grad_norm": 11.905723571777344, "learning_rate": 9.772330097087379e-05, "loss": 0.5558, "step": 469 }, { "epoch": 4.554216867469879, "grad_norm": 2.9849166870117188, "learning_rate": 9.771844660194175e-05, "loss": 0.2403, "step": 470 }, { "epoch": 4.563855421686747, "grad_norm": 3.0800840854644775, "learning_rate": 9.771359223300971e-05, "loss": 0.549, "step": 471 }, { "epoch": 4.573493975903615, "grad_norm": 4.048053741455078, "learning_rate": 9.770873786407767e-05, "loss": 0.348, "step": 472 }, { "epoch": 4.583132530120482, "grad_norm": 5.64263391494751, "learning_rate": 9.770388349514564e-05, "loss": 0.398, "step": 473 }, { "epoch": 4.5927710843373495, "grad_norm": 4.909956932067871, "learning_rate": 9.76990291262136e-05, "loss": 0.2393, "step": 474 }, { "epoch": 4.602409638554217, "grad_norm": 10.933610916137695, "learning_rate": 9.769417475728155e-05, "loss": 0.4402, "step": 475 }, { "epoch": 4.612048192771084, "grad_norm": 11.007538795471191, "learning_rate": 9.768932038834951e-05, "loss": 0.5454, "step": 476 }, { "epoch": 4.621686746987952, "grad_norm": 2.3356399536132812, "learning_rate": 9.768446601941748e-05, "loss": 0.3883, "step": 477 }, { "epoch": 4.63132530120482, "grad_norm": 12.909774780273438, "learning_rate": 9.767961165048544e-05, "loss": 0.6427, "step": 478 }, { "epoch": 4.6409638554216865, "grad_norm": 3.423194169998169, "learning_rate": 9.76747572815534e-05, "loss": 0.4758, "step": 479 }, { "epoch": 4.650602409638554, "grad_norm": 10.095155715942383, "learning_rate": 9.766990291262137e-05, "loss": 0.4621, "step": 480 }, { "epoch": 4.660240963855422, "grad_norm": 4.085877895355225, "learning_rate": 9.766504854368933e-05, "loss": 0.261, "step": 481 }, { "epoch": 4.669879518072289, "grad_norm": 11.577909469604492, "learning_rate": 9.76601941747573e-05, "loss": 0.5028, "step": 482 }, { "epoch": 4.679518072289157, "grad_norm": 2.721404552459717, "learning_rate": 9.765533980582526e-05, "loss": 0.4458, "step": 483 }, { "epoch": 4.6891566265060245, "grad_norm": 6.817914962768555, "learning_rate": 9.765048543689321e-05, "loss": 0.374, "step": 484 }, { "epoch": 4.698795180722891, "grad_norm": 14.848785400390625, "learning_rate": 9.764563106796117e-05, "loss": 0.5241, "step": 485 }, { "epoch": 4.708433734939759, "grad_norm": 5.716707706451416, "learning_rate": 9.764077669902913e-05, "loss": 0.1949, "step": 486 }, { "epoch": 4.718072289156627, "grad_norm": 3.3421103954315186, "learning_rate": 9.76359223300971e-05, "loss": 0.382, "step": 487 }, { "epoch": 4.727710843373494, "grad_norm": 2.950603485107422, "learning_rate": 9.763106796116506e-05, "loss": 0.2498, "step": 488 }, { "epoch": 4.7373493975903616, "grad_norm": 1.9777179956436157, "learning_rate": 9.762621359223302e-05, "loss": 0.136, "step": 489 }, { "epoch": 4.746987951807229, "grad_norm": 11.578618049621582, "learning_rate": 9.762135922330097e-05, "loss": 0.2404, "step": 490 }, { "epoch": 4.756626506024096, "grad_norm": 8.707721710205078, "learning_rate": 9.761650485436894e-05, "loss": 0.3721, "step": 491 }, { "epoch": 4.766265060240964, "grad_norm": 6.816089153289795, "learning_rate": 9.76116504854369e-05, "loss": 0.6901, "step": 492 }, { "epoch": 4.775903614457832, "grad_norm": 6.138238430023193, "learning_rate": 9.760679611650486e-05, "loss": 0.3259, "step": 493 }, { "epoch": 4.785542168674699, "grad_norm": 2.7941300868988037, "learning_rate": 9.760194174757282e-05, "loss": 0.1739, "step": 494 }, { "epoch": 4.795180722891566, "grad_norm": 7.621683597564697, "learning_rate": 9.759708737864079e-05, "loss": 0.4377, "step": 495 }, { "epoch": 4.804819277108434, "grad_norm": 15.074524879455566, "learning_rate": 9.759223300970875e-05, "loss": 0.4997, "step": 496 }, { "epoch": 4.814457831325301, "grad_norm": 7.014825820922852, "learning_rate": 9.75873786407767e-05, "loss": 0.3981, "step": 497 }, { "epoch": 4.824096385542169, "grad_norm": 2.418645143508911, "learning_rate": 9.758252427184466e-05, "loss": 0.3909, "step": 498 }, { "epoch": 4.833734939759037, "grad_norm": 6.738986492156982, "learning_rate": 9.757766990291263e-05, "loss": 0.2388, "step": 499 }, { "epoch": 4.843373493975903, "grad_norm": 11.597596168518066, "learning_rate": 9.757281553398059e-05, "loss": 0.6102, "step": 500 }, { "epoch": 4.853012048192771, "grad_norm": 5.060056209564209, "learning_rate": 9.756796116504855e-05, "loss": 0.4347, "step": 501 }, { "epoch": 4.862650602409639, "grad_norm": 5.351769924163818, "learning_rate": 9.75631067961165e-05, "loss": 0.3648, "step": 502 }, { "epoch": 4.872289156626506, "grad_norm": 7.823605537414551, "learning_rate": 9.755825242718446e-05, "loss": 0.2829, "step": 503 }, { "epoch": 4.881927710843374, "grad_norm": 8.606855392456055, "learning_rate": 9.755339805825244e-05, "loss": 0.3277, "step": 504 }, { "epoch": 4.891566265060241, "grad_norm": 3.8559160232543945, "learning_rate": 9.75485436893204e-05, "loss": 0.29, "step": 505 }, { "epoch": 4.901204819277108, "grad_norm": 7.195594310760498, "learning_rate": 9.754368932038835e-05, "loss": 0.3727, "step": 506 }, { "epoch": 4.910843373493976, "grad_norm": 4.181766986846924, "learning_rate": 9.753883495145631e-05, "loss": 0.3753, "step": 507 }, { "epoch": 4.920481927710844, "grad_norm": 14.254661560058594, "learning_rate": 9.753398058252428e-05, "loss": 0.24, "step": 508 }, { "epoch": 4.930120481927711, "grad_norm": 3.6964826583862305, "learning_rate": 9.752912621359224e-05, "loss": 0.2396, "step": 509 }, { "epoch": 4.9397590361445785, "grad_norm": 13.593742370605469, "learning_rate": 9.75242718446602e-05, "loss": 0.4865, "step": 510 }, { "epoch": 4.949397590361446, "grad_norm": 9.74808406829834, "learning_rate": 9.751941747572815e-05, "loss": 0.1382, "step": 511 }, { "epoch": 4.959036144578313, "grad_norm": 8.12394905090332, "learning_rate": 9.751456310679612e-05, "loss": 0.2735, "step": 512 }, { "epoch": 4.968674698795181, "grad_norm": 1.4520745277404785, "learning_rate": 9.750970873786408e-05, "loss": 0.243, "step": 513 }, { "epoch": 4.978313253012049, "grad_norm": 7.542995929718018, "learning_rate": 9.750485436893204e-05, "loss": 0.5494, "step": 514 }, { "epoch": 4.9879518072289155, "grad_norm": 4.211097717285156, "learning_rate": 9.75e-05, "loss": 0.2908, "step": 515 }, { "epoch": 5.0036144578313255, "grad_norm": 6.431490421295166, "learning_rate": 9.749514563106797e-05, "loss": 0.2948, "step": 516 }, { "epoch": 5.013253012048192, "grad_norm": 5.521190643310547, "learning_rate": 9.749029126213593e-05, "loss": 0.3218, "step": 517 }, { "epoch": 5.02289156626506, "grad_norm": 7.382561206817627, "learning_rate": 9.748543689320389e-05, "loss": 0.417, "step": 518 }, { "epoch": 5.032530120481928, "grad_norm": 4.550668239593506, "learning_rate": 9.748058252427184e-05, "loss": 0.2511, "step": 519 }, { "epoch": 5.042168674698795, "grad_norm": 15.045186042785645, "learning_rate": 9.74757281553398e-05, "loss": 0.3375, "step": 520 }, { "epoch": 5.051807228915663, "grad_norm": 3.0288095474243164, "learning_rate": 9.747087378640777e-05, "loss": 0.3963, "step": 521 }, { "epoch": 5.06144578313253, "grad_norm": 11.14201545715332, "learning_rate": 9.746601941747573e-05, "loss": 0.4989, "step": 522 }, { "epoch": 5.071084337349397, "grad_norm": 5.496853828430176, "learning_rate": 9.746116504854369e-05, "loss": 0.591, "step": 523 }, { "epoch": 5.080722891566265, "grad_norm": 3.1127662658691406, "learning_rate": 9.745631067961165e-05, "loss": 0.3775, "step": 524 }, { "epoch": 5.090361445783133, "grad_norm": 2.386813163757324, "learning_rate": 9.745145631067962e-05, "loss": 0.2101, "step": 525 }, { "epoch": 5.1, "grad_norm": 7.020637512207031, "learning_rate": 9.744660194174757e-05, "loss": 0.3746, "step": 526 }, { "epoch": 5.109638554216867, "grad_norm": 12.287225723266602, "learning_rate": 9.744174757281553e-05, "loss": 0.3335, "step": 527 }, { "epoch": 5.119277108433735, "grad_norm": 2.9691338539123535, "learning_rate": 9.743689320388349e-05, "loss": 0.3329, "step": 528 }, { "epoch": 5.128915662650602, "grad_norm": 19.729228973388672, "learning_rate": 9.743203883495146e-05, "loss": 0.2058, "step": 529 }, { "epoch": 5.13855421686747, "grad_norm": 22.2899112701416, "learning_rate": 9.742718446601942e-05, "loss": 0.6783, "step": 530 }, { "epoch": 5.148192771084338, "grad_norm": 5.809994220733643, "learning_rate": 9.742233009708739e-05, "loss": 0.3466, "step": 531 }, { "epoch": 5.1578313253012045, "grad_norm": 3.520979166030884, "learning_rate": 9.741747572815535e-05, "loss": 0.4144, "step": 532 }, { "epoch": 5.167469879518072, "grad_norm": 3.2992398738861084, "learning_rate": 9.74126213592233e-05, "loss": 0.2977, "step": 533 }, { "epoch": 5.17710843373494, "grad_norm": 6.517961025238037, "learning_rate": 9.740776699029128e-05, "loss": 0.4895, "step": 534 }, { "epoch": 5.186746987951807, "grad_norm": 6.994864463806152, "learning_rate": 9.740291262135923e-05, "loss": 0.3131, "step": 535 }, { "epoch": 5.196385542168675, "grad_norm": 11.038994789123535, "learning_rate": 9.739805825242719e-05, "loss": 0.4415, "step": 536 }, { "epoch": 5.206024096385542, "grad_norm": 3.3362574577331543, "learning_rate": 9.739320388349515e-05, "loss": 0.2305, "step": 537 }, { "epoch": 5.215662650602409, "grad_norm": 4.746628761291504, "learning_rate": 9.738834951456312e-05, "loss": 0.4859, "step": 538 }, { "epoch": 5.225301204819277, "grad_norm": 4.624610424041748, "learning_rate": 9.738349514563108e-05, "loss": 0.3395, "step": 539 }, { "epoch": 5.234939759036145, "grad_norm": 2.5550737380981445, "learning_rate": 9.737864077669904e-05, "loss": 0.365, "step": 540 }, { "epoch": 5.244578313253012, "grad_norm": 4.390716075897217, "learning_rate": 9.7373786407767e-05, "loss": 0.2187, "step": 541 }, { "epoch": 5.2542168674698795, "grad_norm": 6.783937454223633, "learning_rate": 9.736893203883495e-05, "loss": 0.3163, "step": 542 }, { "epoch": 5.263855421686747, "grad_norm": 29.411609649658203, "learning_rate": 9.736407766990292e-05, "loss": 0.2451, "step": 543 }, { "epoch": 5.273493975903614, "grad_norm": 3.4033806324005127, "learning_rate": 9.735922330097088e-05, "loss": 0.3012, "step": 544 }, { "epoch": 5.283132530120482, "grad_norm": 11.801424980163574, "learning_rate": 9.735436893203884e-05, "loss": 0.3515, "step": 545 }, { "epoch": 5.29277108433735, "grad_norm": 9.667840957641602, "learning_rate": 9.73495145631068e-05, "loss": 0.4978, "step": 546 }, { "epoch": 5.3024096385542165, "grad_norm": 12.316267967224121, "learning_rate": 9.734466019417477e-05, "loss": 0.3727, "step": 547 }, { "epoch": 5.312048192771084, "grad_norm": 10.039875984191895, "learning_rate": 9.733980582524273e-05, "loss": 0.4108, "step": 548 }, { "epoch": 5.321686746987952, "grad_norm": 21.662303924560547, "learning_rate": 9.733495145631068e-05, "loss": 0.4796, "step": 549 }, { "epoch": 5.331325301204819, "grad_norm": 3.4687247276306152, "learning_rate": 9.733009708737864e-05, "loss": 0.2998, "step": 550 }, { "epoch": 5.340963855421687, "grad_norm": 13.638408660888672, "learning_rate": 9.732524271844661e-05, "loss": 0.2608, "step": 551 }, { "epoch": 5.3506024096385545, "grad_norm": 5.548796653747559, "learning_rate": 9.732038834951457e-05, "loss": 0.2726, "step": 552 }, { "epoch": 5.360240963855421, "grad_norm": 4.4398064613342285, "learning_rate": 9.731553398058253e-05, "loss": 0.2507, "step": 553 }, { "epoch": 5.369879518072289, "grad_norm": 6.214051723480225, "learning_rate": 9.731067961165049e-05, "loss": 0.286, "step": 554 }, { "epoch": 5.379518072289157, "grad_norm": 4.560336589813232, "learning_rate": 9.730582524271846e-05, "loss": 0.3495, "step": 555 }, { "epoch": 5.389156626506024, "grad_norm": 22.053131103515625, "learning_rate": 9.730097087378642e-05, "loss": 0.3226, "step": 556 }, { "epoch": 5.3987951807228916, "grad_norm": 4.4174323081970215, "learning_rate": 9.729611650485437e-05, "loss": 0.2073, "step": 557 }, { "epoch": 5.408433734939759, "grad_norm": 7.560439109802246, "learning_rate": 9.729126213592233e-05, "loss": 0.2456, "step": 558 }, { "epoch": 5.418072289156626, "grad_norm": 5.5289201736450195, "learning_rate": 9.728640776699029e-05, "loss": 0.29, "step": 559 }, { "epoch": 5.427710843373494, "grad_norm": 8.484601020812988, "learning_rate": 9.728155339805826e-05, "loss": 0.5718, "step": 560 }, { "epoch": 5.437349397590362, "grad_norm": 14.339473724365234, "learning_rate": 9.727669902912622e-05, "loss": 0.3111, "step": 561 }, { "epoch": 5.446987951807229, "grad_norm": 3.4083240032196045, "learning_rate": 9.727184466019418e-05, "loss": 0.3908, "step": 562 }, { "epoch": 5.456626506024096, "grad_norm": 2.2884581089019775, "learning_rate": 9.726699029126213e-05, "loss": 0.1626, "step": 563 }, { "epoch": 5.466265060240964, "grad_norm": 4.3083014488220215, "learning_rate": 9.72621359223301e-05, "loss": 0.3371, "step": 564 }, { "epoch": 5.475903614457831, "grad_norm": 3.0772769451141357, "learning_rate": 9.725728155339806e-05, "loss": 0.3099, "step": 565 }, { "epoch": 5.485542168674699, "grad_norm": 5.234801292419434, "learning_rate": 9.725242718446602e-05, "loss": 0.3764, "step": 566 }, { "epoch": 5.495180722891567, "grad_norm": 4.9415693283081055, "learning_rate": 9.724757281553398e-05, "loss": 0.3025, "step": 567 }, { "epoch": 5.504819277108433, "grad_norm": 7.8344902992248535, "learning_rate": 9.724271844660195e-05, "loss": 0.2002, "step": 568 }, { "epoch": 5.514457831325301, "grad_norm": 3.7597708702087402, "learning_rate": 9.723786407766991e-05, "loss": 0.3131, "step": 569 }, { "epoch": 5.524096385542169, "grad_norm": 2.984241485595703, "learning_rate": 9.723300970873786e-05, "loss": 0.4002, "step": 570 }, { "epoch": 5.533734939759036, "grad_norm": 9.253135681152344, "learning_rate": 9.722815533980582e-05, "loss": 0.2423, "step": 571 }, { "epoch": 5.543373493975904, "grad_norm": 5.082090854644775, "learning_rate": 9.72233009708738e-05, "loss": 0.4446, "step": 572 }, { "epoch": 5.553012048192771, "grad_norm": 9.18911075592041, "learning_rate": 9.721844660194175e-05, "loss": 0.3252, "step": 573 }, { "epoch": 5.562650602409638, "grad_norm": 14.098710060119629, "learning_rate": 9.721359223300971e-05, "loss": 0.4706, "step": 574 }, { "epoch": 5.572289156626506, "grad_norm": 12.182364463806152, "learning_rate": 9.720873786407767e-05, "loss": 0.5232, "step": 575 }, { "epoch": 5.581927710843374, "grad_norm": 3.344181537628174, "learning_rate": 9.720388349514562e-05, "loss": 0.1267, "step": 576 }, { "epoch": 5.591566265060241, "grad_norm": 5.444984436035156, "learning_rate": 9.71990291262136e-05, "loss": 0.533, "step": 577 }, { "epoch": 5.6012048192771084, "grad_norm": 10.007716178894043, "learning_rate": 9.719417475728155e-05, "loss": 0.3085, "step": 578 }, { "epoch": 5.610843373493976, "grad_norm": 4.4351487159729, "learning_rate": 9.718932038834951e-05, "loss": 0.2685, "step": 579 }, { "epoch": 5.620481927710843, "grad_norm": 3.3641762733459473, "learning_rate": 9.718446601941747e-05, "loss": 0.3622, "step": 580 }, { "epoch": 5.630120481927711, "grad_norm": 4.749971389770508, "learning_rate": 9.717961165048544e-05, "loss": 0.3498, "step": 581 }, { "epoch": 5.639759036144579, "grad_norm": 2.8244197368621826, "learning_rate": 9.717475728155341e-05, "loss": 0.3541, "step": 582 }, { "epoch": 5.6493975903614455, "grad_norm": 3.4756360054016113, "learning_rate": 9.716990291262137e-05, "loss": 0.4405, "step": 583 }, { "epoch": 5.659036144578313, "grad_norm": 5.044853210449219, "learning_rate": 9.716504854368933e-05, "loss": 0.6258, "step": 584 }, { "epoch": 5.668674698795181, "grad_norm": 5.278900623321533, "learning_rate": 9.716019417475729e-05, "loss": 0.3563, "step": 585 }, { "epoch": 5.678313253012048, "grad_norm": 10.800078392028809, "learning_rate": 9.715533980582526e-05, "loss": 0.4067, "step": 586 }, { "epoch": 5.687951807228916, "grad_norm": 6.552043437957764, "learning_rate": 9.715048543689321e-05, "loss": 0.6015, "step": 587 }, { "epoch": 5.6975903614457835, "grad_norm": 4.720762729644775, "learning_rate": 9.714563106796117e-05, "loss": 0.4954, "step": 588 }, { "epoch": 5.70722891566265, "grad_norm": 3.7900466918945312, "learning_rate": 9.714077669902913e-05, "loss": 0.2687, "step": 589 }, { "epoch": 5.716867469879518, "grad_norm": 7.368895053863525, "learning_rate": 9.71359223300971e-05, "loss": 0.5465, "step": 590 }, { "epoch": 5.726506024096386, "grad_norm": 3.2007830142974854, "learning_rate": 9.713106796116506e-05, "loss": 0.2703, "step": 591 }, { "epoch": 5.736144578313253, "grad_norm": 4.486144542694092, "learning_rate": 9.712621359223302e-05, "loss": 0.3606, "step": 592 }, { "epoch": 5.7457831325301205, "grad_norm": 3.4930150508880615, "learning_rate": 9.712135922330097e-05, "loss": 0.3354, "step": 593 }, { "epoch": 5.755421686746988, "grad_norm": 5.85201358795166, "learning_rate": 9.711650485436895e-05, "loss": 0.2814, "step": 594 }, { "epoch": 5.765060240963855, "grad_norm": 7.571313381195068, "learning_rate": 9.71116504854369e-05, "loss": 0.3754, "step": 595 }, { "epoch": 5.774698795180723, "grad_norm": 2.9975578784942627, "learning_rate": 9.710679611650486e-05, "loss": 0.3619, "step": 596 }, { "epoch": 5.784337349397591, "grad_norm": 1.9621005058288574, "learning_rate": 9.710194174757282e-05, "loss": 0.3714, "step": 597 }, { "epoch": 5.793975903614458, "grad_norm": 13.716328620910645, "learning_rate": 9.709708737864078e-05, "loss": 0.4961, "step": 598 }, { "epoch": 5.803614457831325, "grad_norm": 6.1292924880981445, "learning_rate": 9.709223300970875e-05, "loss": 0.2996, "step": 599 }, { "epoch": 5.813253012048193, "grad_norm": 5.713808536529541, "learning_rate": 9.70873786407767e-05, "loss": 0.5359, "step": 600 }, { "epoch": 5.82289156626506, "grad_norm": 6.7606401443481445, "learning_rate": 9.708252427184466e-05, "loss": 0.5288, "step": 601 }, { "epoch": 5.832530120481928, "grad_norm": 9.61403751373291, "learning_rate": 9.707766990291262e-05, "loss": 0.3285, "step": 602 }, { "epoch": 5.8421686746987955, "grad_norm": 8.50088119506836, "learning_rate": 9.707281553398059e-05, "loss": 0.3088, "step": 603 }, { "epoch": 5.851807228915662, "grad_norm": 2.8742969036102295, "learning_rate": 9.706796116504855e-05, "loss": 0.2879, "step": 604 }, { "epoch": 5.86144578313253, "grad_norm": 6.541842460632324, "learning_rate": 9.706310679611651e-05, "loss": 0.4835, "step": 605 }, { "epoch": 5.871084337349398, "grad_norm": 3.1671013832092285, "learning_rate": 9.705825242718447e-05, "loss": 0.4076, "step": 606 }, { "epoch": 5.880722891566265, "grad_norm": 6.507176876068115, "learning_rate": 9.705339805825244e-05, "loss": 0.4354, "step": 607 }, { "epoch": 5.890361445783133, "grad_norm": 3.9244186878204346, "learning_rate": 9.70485436893204e-05, "loss": 0.4522, "step": 608 }, { "epoch": 5.9, "grad_norm": 3.358614683151245, "learning_rate": 9.704368932038835e-05, "loss": 0.4078, "step": 609 }, { "epoch": 5.909638554216867, "grad_norm": 4.604615688323975, "learning_rate": 9.703883495145631e-05, "loss": 0.3683, "step": 610 }, { "epoch": 5.919277108433735, "grad_norm": 4.4669413566589355, "learning_rate": 9.703398058252428e-05, "loss": 0.4537, "step": 611 }, { "epoch": 5.928915662650603, "grad_norm": 4.283143520355225, "learning_rate": 9.702912621359224e-05, "loss": 0.3955, "step": 612 }, { "epoch": 5.93855421686747, "grad_norm": 5.0965142250061035, "learning_rate": 9.70242718446602e-05, "loss": 0.2693, "step": 613 }, { "epoch": 5.948192771084337, "grad_norm": 12.181734085083008, "learning_rate": 9.701941747572816e-05, "loss": 0.4427, "step": 614 }, { "epoch": 5.957831325301205, "grad_norm": 2.773944139480591, "learning_rate": 9.701456310679611e-05, "loss": 0.3196, "step": 615 }, { "epoch": 5.967469879518072, "grad_norm": 6.614069938659668, "learning_rate": 9.700970873786408e-05, "loss": 0.3691, "step": 616 }, { "epoch": 5.97710843373494, "grad_norm": 4.127647876739502, "learning_rate": 9.700485436893204e-05, "loss": 0.465, "step": 617 }, { "epoch": 5.986746987951808, "grad_norm": 8.419861793518066, "learning_rate": 9.7e-05, "loss": 0.4595, "step": 618 }, { "epoch": 6.002409638554217, "grad_norm": 5.838987827301025, "learning_rate": 9.699514563106796e-05, "loss": 0.5165, "step": 619 }, { "epoch": 6.0120481927710845, "grad_norm": 8.701430320739746, "learning_rate": 9.699029126213593e-05, "loss": 0.6928, "step": 620 }, { "epoch": 6.021686746987951, "grad_norm": 4.030494689941406, "learning_rate": 9.698543689320389e-05, "loss": 0.3621, "step": 621 }, { "epoch": 6.031325301204819, "grad_norm": 3.5283844470977783, "learning_rate": 9.698058252427184e-05, "loss": 0.3403, "step": 622 }, { "epoch": 6.040963855421687, "grad_norm": 17.598041534423828, "learning_rate": 9.69757281553398e-05, "loss": 0.4485, "step": 623 }, { "epoch": 6.050602409638554, "grad_norm": 12.916193008422852, "learning_rate": 9.697087378640777e-05, "loss": 0.6174, "step": 624 }, { "epoch": 6.0602409638554215, "grad_norm": 8.073314666748047, "learning_rate": 9.696601941747573e-05, "loss": 0.4074, "step": 625 }, { "epoch": 6.069879518072289, "grad_norm": 10.061405181884766, "learning_rate": 9.696116504854369e-05, "loss": 0.7998, "step": 626 }, { "epoch": 6.079518072289156, "grad_norm": 4.240304470062256, "learning_rate": 9.695631067961165e-05, "loss": 0.4825, "step": 627 }, { "epoch": 6.089156626506024, "grad_norm": 7.2652411460876465, "learning_rate": 9.695145631067962e-05, "loss": 0.4874, "step": 628 }, { "epoch": 6.098795180722892, "grad_norm": 15.362871170043945, "learning_rate": 9.694660194174758e-05, "loss": 0.612, "step": 629 }, { "epoch": 6.108433734939759, "grad_norm": 11.747915267944336, "learning_rate": 9.694174757281553e-05, "loss": 0.55, "step": 630 }, { "epoch": 6.118072289156626, "grad_norm": 3.444286823272705, "learning_rate": 9.693689320388349e-05, "loss": 0.3337, "step": 631 }, { "epoch": 6.127710843373494, "grad_norm": 3.7820537090301514, "learning_rate": 9.693203883495146e-05, "loss": 0.311, "step": 632 }, { "epoch": 6.137349397590361, "grad_norm": 4.31968879699707, "learning_rate": 9.692718446601943e-05, "loss": 0.3404, "step": 633 }, { "epoch": 6.146987951807229, "grad_norm": 3.1904773712158203, "learning_rate": 9.692233009708739e-05, "loss": 0.4707, "step": 634 }, { "epoch": 6.156626506024097, "grad_norm": 25.223587036132812, "learning_rate": 9.691747572815535e-05, "loss": 0.3499, "step": 635 }, { "epoch": 6.166265060240963, "grad_norm": 6.927448749542236, "learning_rate": 9.691262135922331e-05, "loss": 0.3315, "step": 636 }, { "epoch": 6.175903614457831, "grad_norm": 4.000790596008301, "learning_rate": 9.690776699029126e-05, "loss": 0.4355, "step": 637 }, { "epoch": 6.185542168674699, "grad_norm": 7.604026794433594, "learning_rate": 9.690291262135924e-05, "loss": 0.3347, "step": 638 }, { "epoch": 6.195180722891566, "grad_norm": 3.8774216175079346, "learning_rate": 9.68980582524272e-05, "loss": 0.1588, "step": 639 }, { "epoch": 6.204819277108434, "grad_norm": 29.128623962402344, "learning_rate": 9.689320388349515e-05, "loss": 0.6322, "step": 640 }, { "epoch": 6.214457831325301, "grad_norm": 8.396805763244629, "learning_rate": 9.688834951456311e-05, "loss": 0.383, "step": 641 }, { "epoch": 6.224096385542168, "grad_norm": 5.986485958099365, "learning_rate": 9.688349514563108e-05, "loss": 0.4035, "step": 642 }, { "epoch": 6.233734939759036, "grad_norm": 3.317140817642212, "learning_rate": 9.687864077669904e-05, "loss": 0.4854, "step": 643 }, { "epoch": 6.243373493975904, "grad_norm": 3.7621192932128906, "learning_rate": 9.6873786407767e-05, "loss": 0.3528, "step": 644 }, { "epoch": 6.253012048192771, "grad_norm": 3.6792593002319336, "learning_rate": 9.686893203883495e-05, "loss": 0.2329, "step": 645 }, { "epoch": 6.2626506024096384, "grad_norm": 4.294147491455078, "learning_rate": 9.686407766990293e-05, "loss": 0.3349, "step": 646 }, { "epoch": 6.272289156626506, "grad_norm": 15.853439331054688, "learning_rate": 9.685922330097088e-05, "loss": 0.3155, "step": 647 }, { "epoch": 6.281927710843373, "grad_norm": 2.7699801921844482, "learning_rate": 9.685436893203884e-05, "loss": 0.2744, "step": 648 }, { "epoch": 6.291566265060241, "grad_norm": 7.462607383728027, "learning_rate": 9.68495145631068e-05, "loss": 0.4067, "step": 649 }, { "epoch": 6.301204819277109, "grad_norm": 3.4757437705993652, "learning_rate": 9.684466019417477e-05, "loss": 0.2048, "step": 650 }, { "epoch": 6.3108433734939755, "grad_norm": 7.163461685180664, "learning_rate": 9.683980582524273e-05, "loss": 0.2032, "step": 651 }, { "epoch": 6.320481927710843, "grad_norm": 4.228833198547363, "learning_rate": 9.683495145631069e-05, "loss": 0.5136, "step": 652 }, { "epoch": 6.330120481927711, "grad_norm": 8.770180702209473, "learning_rate": 9.683009708737864e-05, "loss": 0.411, "step": 653 }, { "epoch": 6.339759036144578, "grad_norm": 4.789836406707764, "learning_rate": 9.68252427184466e-05, "loss": 0.3394, "step": 654 }, { "epoch": 6.349397590361446, "grad_norm": 7.417471885681152, "learning_rate": 9.682038834951457e-05, "loss": 0.5394, "step": 655 }, { "epoch": 6.3590361445783135, "grad_norm": 3.477372884750366, "learning_rate": 9.681553398058253e-05, "loss": 0.3498, "step": 656 }, { "epoch": 6.36867469879518, "grad_norm": 7.222618579864502, "learning_rate": 9.681067961165049e-05, "loss": 0.3629, "step": 657 }, { "epoch": 6.378313253012048, "grad_norm": 11.526228904724121, "learning_rate": 9.680582524271845e-05, "loss": 0.3157, "step": 658 }, { "epoch": 6.387951807228916, "grad_norm": 5.694446563720703, "learning_rate": 9.680097087378642e-05, "loss": 0.4878, "step": 659 }, { "epoch": 6.397590361445783, "grad_norm": 3.2963438034057617, "learning_rate": 9.679611650485437e-05, "loss": 0.2963, "step": 660 }, { "epoch": 6.4072289156626505, "grad_norm": 3.0829102993011475, "learning_rate": 9.679126213592233e-05, "loss": 0.2893, "step": 661 }, { "epoch": 6.416867469879518, "grad_norm": 5.114987850189209, "learning_rate": 9.678640776699029e-05, "loss": 0.5346, "step": 662 }, { "epoch": 6.426506024096385, "grad_norm": 1.8585237264633179, "learning_rate": 9.678155339805826e-05, "loss": 0.135, "step": 663 }, { "epoch": 6.436144578313253, "grad_norm": 1.9071263074874878, "learning_rate": 9.677669902912622e-05, "loss": 0.1609, "step": 664 }, { "epoch": 6.445783132530121, "grad_norm": 8.953192710876465, "learning_rate": 9.677184466019418e-05, "loss": 0.3816, "step": 665 }, { "epoch": 6.455421686746988, "grad_norm": 3.493570566177368, "learning_rate": 9.676699029126213e-05, "loss": 0.3377, "step": 666 }, { "epoch": 6.465060240963855, "grad_norm": 12.4081449508667, "learning_rate": 9.67621359223301e-05, "loss": 0.5904, "step": 667 }, { "epoch": 6.474698795180723, "grad_norm": 16.250198364257812, "learning_rate": 9.675728155339806e-05, "loss": 0.46, "step": 668 }, { "epoch": 6.48433734939759, "grad_norm": 6.161910533905029, "learning_rate": 9.675242718446602e-05, "loss": 0.5831, "step": 669 }, { "epoch": 6.493975903614458, "grad_norm": 12.501585960388184, "learning_rate": 9.674757281553398e-05, "loss": 0.4421, "step": 670 }, { "epoch": 6.5036144578313255, "grad_norm": 12.269323348999023, "learning_rate": 9.674271844660194e-05, "loss": 0.3764, "step": 671 }, { "epoch": 6.513253012048192, "grad_norm": 6.405348300933838, "learning_rate": 9.673786407766991e-05, "loss": 0.632, "step": 672 }, { "epoch": 6.52289156626506, "grad_norm": 14.515312194824219, "learning_rate": 9.673300970873787e-05, "loss": 0.4561, "step": 673 }, { "epoch": 6.532530120481928, "grad_norm": 3.2598206996917725, "learning_rate": 9.672815533980582e-05, "loss": 0.4059, "step": 674 }, { "epoch": 6.542168674698795, "grad_norm": 6.635264873504639, "learning_rate": 9.672330097087378e-05, "loss": 0.5577, "step": 675 }, { "epoch": 6.551807228915663, "grad_norm": 3.425060272216797, "learning_rate": 9.671844660194175e-05, "loss": 0.2471, "step": 676 }, { "epoch": 6.56144578313253, "grad_norm": 7.01643705368042, "learning_rate": 9.671359223300971e-05, "loss": 0.4699, "step": 677 }, { "epoch": 6.571084337349397, "grad_norm": 9.894701957702637, "learning_rate": 9.670873786407767e-05, "loss": 0.2956, "step": 678 }, { "epoch": 6.580722891566265, "grad_norm": 4.4224772453308105, "learning_rate": 9.670388349514563e-05, "loss": 0.3988, "step": 679 }, { "epoch": 6.590361445783133, "grad_norm": 2.9022045135498047, "learning_rate": 9.66990291262136e-05, "loss": 0.3129, "step": 680 }, { "epoch": 6.6, "grad_norm": 4.863816738128662, "learning_rate": 9.669417475728156e-05, "loss": 0.4078, "step": 681 }, { "epoch": 6.609638554216867, "grad_norm": 3.8550493717193604, "learning_rate": 9.668932038834953e-05, "loss": 0.352, "step": 682 }, { "epoch": 6.619277108433735, "grad_norm": 3.1150684356689453, "learning_rate": 9.668446601941748e-05, "loss": 0.2425, "step": 683 }, { "epoch": 6.628915662650602, "grad_norm": 9.734992027282715, "learning_rate": 9.667961165048544e-05, "loss": 0.522, "step": 684 }, { "epoch": 6.63855421686747, "grad_norm": 5.757567882537842, "learning_rate": 9.667475728155341e-05, "loss": 0.3286, "step": 685 }, { "epoch": 6.648192771084338, "grad_norm": 3.9384331703186035, "learning_rate": 9.666990291262137e-05, "loss": 0.3556, "step": 686 }, { "epoch": 6.6578313253012045, "grad_norm": 5.815218448638916, "learning_rate": 9.666504854368933e-05, "loss": 0.5679, "step": 687 }, { "epoch": 6.667469879518072, "grad_norm": 28.241580963134766, "learning_rate": 9.666019417475729e-05, "loss": 0.441, "step": 688 }, { "epoch": 6.67710843373494, "grad_norm": 7.464465618133545, "learning_rate": 9.665533980582526e-05, "loss": 0.4835, "step": 689 }, { "epoch": 6.686746987951807, "grad_norm": 6.2792205810546875, "learning_rate": 9.665048543689322e-05, "loss": 0.4996, "step": 690 }, { "epoch": 6.696385542168675, "grad_norm": 8.607760429382324, "learning_rate": 9.664563106796117e-05, "loss": 0.6143, "step": 691 }, { "epoch": 6.706024096385542, "grad_norm": 11.11461067199707, "learning_rate": 9.664077669902913e-05, "loss": 0.57, "step": 692 }, { "epoch": 6.715662650602409, "grad_norm": 2.3515470027923584, "learning_rate": 9.663592233009709e-05, "loss": 0.149, "step": 693 }, { "epoch": 6.725301204819277, "grad_norm": 4.765363693237305, "learning_rate": 9.663106796116506e-05, "loss": 0.4567, "step": 694 }, { "epoch": 6.734939759036145, "grad_norm": 8.933563232421875, "learning_rate": 9.662621359223302e-05, "loss": 0.4452, "step": 695 }, { "epoch": 6.744578313253012, "grad_norm": 8.791959762573242, "learning_rate": 9.662135922330098e-05, "loss": 0.3122, "step": 696 }, { "epoch": 6.7542168674698795, "grad_norm": 14.829087257385254, "learning_rate": 9.661650485436893e-05, "loss": 0.4847, "step": 697 }, { "epoch": 6.763855421686747, "grad_norm": 5.176752090454102, "learning_rate": 9.66116504854369e-05, "loss": 0.4558, "step": 698 }, { "epoch": 6.773493975903614, "grad_norm": 4.834455490112305, "learning_rate": 9.660679611650486e-05, "loss": 0.4348, "step": 699 }, { "epoch": 6.783132530120482, "grad_norm": 6.447503089904785, "learning_rate": 9.660194174757282e-05, "loss": 0.326, "step": 700 }, { "epoch": 6.79277108433735, "grad_norm": 11.233283996582031, "learning_rate": 9.659708737864078e-05, "loss": 0.3718, "step": 701 }, { "epoch": 6.8024096385542165, "grad_norm": 10.054475784301758, "learning_rate": 9.659223300970875e-05, "loss": 0.3082, "step": 702 }, { "epoch": 6.812048192771084, "grad_norm": 4.198498725891113, "learning_rate": 9.658737864077671e-05, "loss": 0.4103, "step": 703 }, { "epoch": 6.821686746987952, "grad_norm": 5.206996917724609, "learning_rate": 9.658252427184466e-05, "loss": 0.3389, "step": 704 }, { "epoch": 6.831325301204819, "grad_norm": 3.5955686569213867, "learning_rate": 9.657766990291262e-05, "loss": 0.3631, "step": 705 }, { "epoch": 6.840963855421687, "grad_norm": 9.99223804473877, "learning_rate": 9.65728155339806e-05, "loss": 0.2509, "step": 706 }, { "epoch": 6.8506024096385545, "grad_norm": 4.640846252441406, "learning_rate": 9.656796116504855e-05, "loss": 0.3202, "step": 707 }, { "epoch": 6.860240963855421, "grad_norm": 5.383673191070557, "learning_rate": 9.656310679611651e-05, "loss": 0.3087, "step": 708 }, { "epoch": 6.869879518072289, "grad_norm": 10.938912391662598, "learning_rate": 9.655825242718447e-05, "loss": 0.4941, "step": 709 }, { "epoch": 6.879518072289157, "grad_norm": 8.774730682373047, "learning_rate": 9.655339805825243e-05, "loss": 0.2547, "step": 710 }, { "epoch": 6.889156626506024, "grad_norm": 3.5676467418670654, "learning_rate": 9.65485436893204e-05, "loss": 0.4071, "step": 711 }, { "epoch": 6.8987951807228916, "grad_norm": 1.9039796590805054, "learning_rate": 9.654368932038835e-05, "loss": 0.1565, "step": 712 }, { "epoch": 6.908433734939759, "grad_norm": 8.64209270477295, "learning_rate": 9.653883495145631e-05, "loss": 0.5462, "step": 713 }, { "epoch": 6.918072289156626, "grad_norm": 5.304514408111572, "learning_rate": 9.653398058252427e-05, "loss": 0.3428, "step": 714 }, { "epoch": 6.927710843373494, "grad_norm": 4.340817928314209, "learning_rate": 9.652912621359224e-05, "loss": 0.4188, "step": 715 }, { "epoch": 6.937349397590362, "grad_norm": 2.5107250213623047, "learning_rate": 9.65242718446602e-05, "loss": 0.2695, "step": 716 }, { "epoch": 6.946987951807229, "grad_norm": 4.752511024475098, "learning_rate": 9.651941747572816e-05, "loss": 0.4283, "step": 717 }, { "epoch": 6.956626506024096, "grad_norm": 10.055572509765625, "learning_rate": 9.651456310679611e-05, "loss": 0.4968, "step": 718 }, { "epoch": 6.966265060240964, "grad_norm": 6.4650444984436035, "learning_rate": 9.650970873786409e-05, "loss": 0.5392, "step": 719 }, { "epoch": 6.975903614457831, "grad_norm": 5.570119857788086, "learning_rate": 9.650485436893204e-05, "loss": 0.2655, "step": 720 }, { "epoch": 6.985542168674699, "grad_norm": 6.629663944244385, "learning_rate": 9.65e-05, "loss": 0.1883, "step": 721 }, { "epoch": 7.001204819277109, "grad_norm": 4.807003498077393, "learning_rate": 9.649514563106796e-05, "loss": 0.2109, "step": 722 }, { "epoch": 7.010843373493976, "grad_norm": 9.003057479858398, "learning_rate": 9.649029126213593e-05, "loss": 0.3436, "step": 723 }, { "epoch": 7.0204819277108435, "grad_norm": 8.993918418884277, "learning_rate": 9.648543689320389e-05, "loss": 0.6094, "step": 724 }, { "epoch": 7.030120481927711, "grad_norm": 5.853751182556152, "learning_rate": 9.648058252427185e-05, "loss": 0.5131, "step": 725 }, { "epoch": 7.039759036144578, "grad_norm": 5.012195110321045, "learning_rate": 9.64757281553398e-05, "loss": 0.448, "step": 726 }, { "epoch": 7.049397590361446, "grad_norm": 3.6509947776794434, "learning_rate": 9.647087378640776e-05, "loss": 0.3494, "step": 727 }, { "epoch": 7.059036144578314, "grad_norm": 4.588027000427246, "learning_rate": 9.646601941747573e-05, "loss": 0.3857, "step": 728 }, { "epoch": 7.0686746987951805, "grad_norm": 4.667324066162109, "learning_rate": 9.646116504854369e-05, "loss": 0.3948, "step": 729 }, { "epoch": 7.078313253012048, "grad_norm": 5.1217241287231445, "learning_rate": 9.645631067961165e-05, "loss": 0.3862, "step": 730 }, { "epoch": 7.087951807228916, "grad_norm": 8.189798355102539, "learning_rate": 9.64514563106796e-05, "loss": 0.4271, "step": 731 }, { "epoch": 7.097590361445783, "grad_norm": 17.69722557067871, "learning_rate": 9.644660194174758e-05, "loss": 0.5089, "step": 732 }, { "epoch": 7.107228915662651, "grad_norm": 5.346834659576416, "learning_rate": 9.644174757281555e-05, "loss": 0.4627, "step": 733 }, { "epoch": 7.1168674698795185, "grad_norm": 4.509889125823975, "learning_rate": 9.64368932038835e-05, "loss": 0.3916, "step": 734 }, { "epoch": 7.126506024096385, "grad_norm": 4.796131610870361, "learning_rate": 9.643203883495146e-05, "loss": 0.3625, "step": 735 }, { "epoch": 7.136144578313253, "grad_norm": 9.690263748168945, "learning_rate": 9.642718446601942e-05, "loss": 0.6123, "step": 736 }, { "epoch": 7.145783132530121, "grad_norm": 6.031671524047852, "learning_rate": 9.642233009708739e-05, "loss": 0.2829, "step": 737 }, { "epoch": 7.155421686746988, "grad_norm": 6.707278251647949, "learning_rate": 9.641747572815535e-05, "loss": 0.4699, "step": 738 }, { "epoch": 7.1650602409638555, "grad_norm": 19.59365463256836, "learning_rate": 9.641262135922331e-05, "loss": 0.55, "step": 739 }, { "epoch": 7.174698795180723, "grad_norm": 3.241567373275757, "learning_rate": 9.640776699029127e-05, "loss": 0.2749, "step": 740 }, { "epoch": 7.18433734939759, "grad_norm": 4.702701091766357, "learning_rate": 9.640291262135924e-05, "loss": 0.4227, "step": 741 }, { "epoch": 7.193975903614458, "grad_norm": 10.722411155700684, "learning_rate": 9.63980582524272e-05, "loss": 0.2709, "step": 742 }, { "epoch": 7.203614457831326, "grad_norm": 5.884797096252441, "learning_rate": 9.639320388349515e-05, "loss": 0.3564, "step": 743 }, { "epoch": 7.213253012048193, "grad_norm": 4.261545181274414, "learning_rate": 9.638834951456311e-05, "loss": 0.513, "step": 744 }, { "epoch": 7.22289156626506, "grad_norm": 7.055700302124023, "learning_rate": 9.638349514563107e-05, "loss": 0.241, "step": 745 }, { "epoch": 7.232530120481928, "grad_norm": 6.914934158325195, "learning_rate": 9.637864077669904e-05, "loss": 0.4033, "step": 746 }, { "epoch": 7.242168674698795, "grad_norm": 3.929837465286255, "learning_rate": 9.6373786407767e-05, "loss": 0.2767, "step": 747 }, { "epoch": 7.251807228915663, "grad_norm": 3.2962918281555176, "learning_rate": 9.636893203883496e-05, "loss": 0.3163, "step": 748 }, { "epoch": 7.2614457831325305, "grad_norm": 2.476548433303833, "learning_rate": 9.636407766990291e-05, "loss": 0.3488, "step": 749 }, { "epoch": 7.271084337349397, "grad_norm": 5.7439422607421875, "learning_rate": 9.635922330097088e-05, "loss": 0.4226, "step": 750 }, { "epoch": 7.280722891566265, "grad_norm": 6.3358540534973145, "learning_rate": 9.635436893203884e-05, "loss": 0.4467, "step": 751 }, { "epoch": 7.290361445783133, "grad_norm": 5.50137186050415, "learning_rate": 9.63495145631068e-05, "loss": 0.4811, "step": 752 }, { "epoch": 7.3, "grad_norm": 6.158215045928955, "learning_rate": 9.634466019417476e-05, "loss": 0.5642, "step": 753 }, { "epoch": 7.309638554216868, "grad_norm": 5.337847709655762, "learning_rate": 9.633980582524273e-05, "loss": 0.2249, "step": 754 }, { "epoch": 7.3192771084337345, "grad_norm": 8.242106437683105, "learning_rate": 9.633495145631069e-05, "loss": 0.4688, "step": 755 }, { "epoch": 7.328915662650602, "grad_norm": 7.391642093658447, "learning_rate": 9.633009708737864e-05, "loss": 0.4196, "step": 756 }, { "epoch": 7.33855421686747, "grad_norm": 8.749537467956543, "learning_rate": 9.63252427184466e-05, "loss": 0.5925, "step": 757 }, { "epoch": 7.348192771084337, "grad_norm": 15.392102241516113, "learning_rate": 9.632038834951457e-05, "loss": 0.4398, "step": 758 }, { "epoch": 7.357831325301205, "grad_norm": 7.419500827789307, "learning_rate": 9.631553398058253e-05, "loss": 0.4579, "step": 759 }, { "epoch": 7.367469879518072, "grad_norm": 5.451685905456543, "learning_rate": 9.631067961165049e-05, "loss": 0.3536, "step": 760 }, { "epoch": 7.377108433734939, "grad_norm": 4.8872833251953125, "learning_rate": 9.630582524271845e-05, "loss": 0.6271, "step": 761 }, { "epoch": 7.386746987951807, "grad_norm": 3.1274280548095703, "learning_rate": 9.63009708737864e-05, "loss": 0.4068, "step": 762 }, { "epoch": 7.396385542168675, "grad_norm": 8.415763854980469, "learning_rate": 9.629611650485438e-05, "loss": 0.4278, "step": 763 }, { "epoch": 7.406024096385542, "grad_norm": 6.4552998542785645, "learning_rate": 9.629126213592233e-05, "loss": 0.6995, "step": 764 }, { "epoch": 7.4156626506024095, "grad_norm": 2.9876608848571777, "learning_rate": 9.628640776699029e-05, "loss": 0.2508, "step": 765 }, { "epoch": 7.425301204819277, "grad_norm": 6.730759143829346, "learning_rate": 9.628155339805825e-05, "loss": 0.2924, "step": 766 }, { "epoch": 7.434939759036144, "grad_norm": 11.147560119628906, "learning_rate": 9.627669902912622e-05, "loss": 0.5718, "step": 767 }, { "epoch": 7.444578313253012, "grad_norm": 12.048450469970703, "learning_rate": 9.627184466019418e-05, "loss": 0.376, "step": 768 }, { "epoch": 7.45421686746988, "grad_norm": 4.21327018737793, "learning_rate": 9.626699029126214e-05, "loss": 0.3615, "step": 769 }, { "epoch": 7.4638554216867465, "grad_norm": 15.504690170288086, "learning_rate": 9.62621359223301e-05, "loss": 0.4084, "step": 770 }, { "epoch": 7.473493975903614, "grad_norm": 2.7012202739715576, "learning_rate": 9.625728155339807e-05, "loss": 0.2207, "step": 771 }, { "epoch": 7.483132530120482, "grad_norm": 4.480315685272217, "learning_rate": 9.625242718446602e-05, "loss": 0.284, "step": 772 }, { "epoch": 7.492771084337349, "grad_norm": 11.162592887878418, "learning_rate": 9.624757281553398e-05, "loss": 0.387, "step": 773 }, { "epoch": 7.502409638554217, "grad_norm": 7.634221076965332, "learning_rate": 9.624271844660194e-05, "loss": 0.2801, "step": 774 }, { "epoch": 7.5120481927710845, "grad_norm": 4.160109996795654, "learning_rate": 9.623786407766991e-05, "loss": 0.6257, "step": 775 }, { "epoch": 7.521686746987951, "grad_norm": 6.77249813079834, "learning_rate": 9.623300970873787e-05, "loss": 0.1975, "step": 776 }, { "epoch": 7.531325301204819, "grad_norm": 4.138680458068848, "learning_rate": 9.622815533980583e-05, "loss": 0.3983, "step": 777 }, { "epoch": 7.540963855421687, "grad_norm": 4.438982009887695, "learning_rate": 9.622330097087378e-05, "loss": 0.4097, "step": 778 }, { "epoch": 7.550602409638554, "grad_norm": 23.509777069091797, "learning_rate": 9.621844660194174e-05, "loss": 0.3005, "step": 779 }, { "epoch": 7.5602409638554215, "grad_norm": 6.054694652557373, "learning_rate": 9.621359223300971e-05, "loss": 0.4804, "step": 780 }, { "epoch": 7.569879518072289, "grad_norm": 4.510021686553955, "learning_rate": 9.620873786407767e-05, "loss": 0.3944, "step": 781 }, { "epoch": 7.579518072289156, "grad_norm": 7.377563953399658, "learning_rate": 9.620388349514563e-05, "loss": 0.4043, "step": 782 }, { "epoch": 7.589156626506024, "grad_norm": 4.9431986808776855, "learning_rate": 9.61990291262136e-05, "loss": 0.4915, "step": 783 }, { "epoch": 7.598795180722892, "grad_norm": 7.974850654602051, "learning_rate": 9.619417475728156e-05, "loss": 0.4986, "step": 784 }, { "epoch": 7.608433734939759, "grad_norm": 5.247725963592529, "learning_rate": 9.618932038834953e-05, "loss": 0.5224, "step": 785 }, { "epoch": 7.618072289156626, "grad_norm": 8.340785026550293, "learning_rate": 9.618446601941749e-05, "loss": 0.4016, "step": 786 }, { "epoch": 7.627710843373494, "grad_norm": 4.208388328552246, "learning_rate": 9.617961165048544e-05, "loss": 0.335, "step": 787 }, { "epoch": 7.637349397590361, "grad_norm": 8.543506622314453, "learning_rate": 9.61747572815534e-05, "loss": 0.6419, "step": 788 }, { "epoch": 7.646987951807229, "grad_norm": 4.027740001678467, "learning_rate": 9.616990291262137e-05, "loss": 0.2894, "step": 789 }, { "epoch": 7.656626506024097, "grad_norm": 5.138880252838135, "learning_rate": 9.616504854368933e-05, "loss": 0.2215, "step": 790 }, { "epoch": 7.666265060240963, "grad_norm": 29.12859535217285, "learning_rate": 9.616019417475729e-05, "loss": 0.5487, "step": 791 }, { "epoch": 7.675903614457831, "grad_norm": 7.634925365447998, "learning_rate": 9.615533980582525e-05, "loss": 0.5394, "step": 792 }, { "epoch": 7.685542168674699, "grad_norm": 3.327205181121826, "learning_rate": 9.615048543689322e-05, "loss": 0.2183, "step": 793 }, { "epoch": 7.695180722891566, "grad_norm": 4.145867824554443, "learning_rate": 9.614563106796117e-05, "loss": 0.3846, "step": 794 }, { "epoch": 7.704819277108434, "grad_norm": 5.415701389312744, "learning_rate": 9.614077669902913e-05, "loss": 0.3597, "step": 795 }, { "epoch": 7.714457831325301, "grad_norm": 5.497639179229736, "learning_rate": 9.613592233009709e-05, "loss": 0.4793, "step": 796 }, { "epoch": 7.724096385542168, "grad_norm": 6.881802558898926, "learning_rate": 9.613106796116506e-05, "loss": 0.5345, "step": 797 }, { "epoch": 7.733734939759036, "grad_norm": 10.950495719909668, "learning_rate": 9.612621359223302e-05, "loss": 0.3819, "step": 798 }, { "epoch": 7.743373493975904, "grad_norm": 5.963572025299072, "learning_rate": 9.612135922330098e-05, "loss": 0.3861, "step": 799 }, { "epoch": 7.753012048192771, "grad_norm": 5.171678066253662, "learning_rate": 9.611650485436893e-05, "loss": 0.2369, "step": 800 }, { "epoch": 7.7626506024096384, "grad_norm": 5.9627604484558105, "learning_rate": 9.611165048543689e-05, "loss": 0.5896, "step": 801 }, { "epoch": 7.772289156626506, "grad_norm": 8.064327239990234, "learning_rate": 9.610679611650486e-05, "loss": 0.578, "step": 802 }, { "epoch": 7.781927710843373, "grad_norm": 5.890385150909424, "learning_rate": 9.610194174757282e-05, "loss": 0.3529, "step": 803 }, { "epoch": 7.791566265060241, "grad_norm": 5.393946647644043, "learning_rate": 9.609708737864078e-05, "loss": 0.189, "step": 804 }, { "epoch": 7.801204819277109, "grad_norm": 6.917865753173828, "learning_rate": 9.609223300970874e-05, "loss": 0.3246, "step": 805 }, { "epoch": 7.8108433734939755, "grad_norm": 6.79478645324707, "learning_rate": 9.608737864077671e-05, "loss": 0.3112, "step": 806 }, { "epoch": 7.820481927710843, "grad_norm": 7.082761764526367, "learning_rate": 9.608252427184467e-05, "loss": 0.3298, "step": 807 }, { "epoch": 7.830120481927711, "grad_norm": 2.761727809906006, "learning_rate": 9.607766990291262e-05, "loss": 0.231, "step": 808 }, { "epoch": 7.839759036144578, "grad_norm": 4.610779285430908, "learning_rate": 9.607281553398058e-05, "loss": 0.228, "step": 809 }, { "epoch": 7.849397590361446, "grad_norm": 3.1571786403656006, "learning_rate": 9.606796116504855e-05, "loss": 0.2017, "step": 810 }, { "epoch": 7.8590361445783135, "grad_norm": 4.171891689300537, "learning_rate": 9.606310679611651e-05, "loss": 0.1672, "step": 811 }, { "epoch": 7.86867469879518, "grad_norm": 8.522045135498047, "learning_rate": 9.605825242718447e-05, "loss": 0.5587, "step": 812 }, { "epoch": 7.878313253012048, "grad_norm": 5.456655979156494, "learning_rate": 9.605339805825243e-05, "loss": 0.1948, "step": 813 }, { "epoch": 7.887951807228916, "grad_norm": 3.319885015487671, "learning_rate": 9.60485436893204e-05, "loss": 0.3293, "step": 814 }, { "epoch": 7.897590361445783, "grad_norm": 3.5659282207489014, "learning_rate": 9.604368932038836e-05, "loss": 0.2099, "step": 815 }, { "epoch": 7.9072289156626505, "grad_norm": 11.019042015075684, "learning_rate": 9.603883495145631e-05, "loss": 0.6043, "step": 816 }, { "epoch": 7.916867469879518, "grad_norm": 3.4559497833251953, "learning_rate": 9.603398058252427e-05, "loss": 0.3007, "step": 817 }, { "epoch": 7.926506024096385, "grad_norm": 5.604502201080322, "learning_rate": 9.602912621359223e-05, "loss": 0.3385, "step": 818 }, { "epoch": 7.936144578313253, "grad_norm": 4.693770408630371, "learning_rate": 9.60242718446602e-05, "loss": 0.3897, "step": 819 }, { "epoch": 7.945783132530121, "grad_norm": 8.00755786895752, "learning_rate": 9.601941747572816e-05, "loss": 0.4954, "step": 820 }, { "epoch": 7.955421686746988, "grad_norm": 3.9966211318969727, "learning_rate": 9.601456310679612e-05, "loss": 0.2725, "step": 821 }, { "epoch": 7.965060240963855, "grad_norm": 6.793938159942627, "learning_rate": 9.600970873786407e-05, "loss": 0.602, "step": 822 }, { "epoch": 7.974698795180723, "grad_norm": 18.064476013183594, "learning_rate": 9.600485436893204e-05, "loss": 0.6575, "step": 823 }, { "epoch": 7.98433734939759, "grad_norm": 4.398540019989014, "learning_rate": 9.6e-05, "loss": 0.4861, "step": 824 }, { "epoch": 7.993975903614458, "grad_norm": 5.5235700607299805, "learning_rate": 9.599514563106796e-05, "loss": 0.3847, "step": 825 }, { "epoch": 8.009638554216867, "grad_norm": 8.431597709655762, "learning_rate": 9.599029126213592e-05, "loss": 0.5067, "step": 826 }, { "epoch": 8.019277108433736, "grad_norm": 5.619980335235596, "learning_rate": 9.598543689320389e-05, "loss": 0.5305, "step": 827 }, { "epoch": 8.028915662650602, "grad_norm": 14.176984786987305, "learning_rate": 9.598058252427185e-05, "loss": 0.386, "step": 828 }, { "epoch": 8.03855421686747, "grad_norm": 2.2295546531677246, "learning_rate": 9.59757281553398e-05, "loss": 0.1909, "step": 829 }, { "epoch": 8.048192771084338, "grad_norm": 3.9574522972106934, "learning_rate": 9.597087378640776e-05, "loss": 0.3124, "step": 830 }, { "epoch": 8.057831325301205, "grad_norm": 10.501607894897461, "learning_rate": 9.596601941747573e-05, "loss": 0.3922, "step": 831 }, { "epoch": 8.067469879518072, "grad_norm": 4.1614508628845215, "learning_rate": 9.596116504854369e-05, "loss": 0.2531, "step": 832 }, { "epoch": 8.07710843373494, "grad_norm": 8.508698463439941, "learning_rate": 9.595631067961165e-05, "loss": 0.4482, "step": 833 }, { "epoch": 8.086746987951807, "grad_norm": 5.893696308135986, "learning_rate": 9.595145631067962e-05, "loss": 0.2039, "step": 834 }, { "epoch": 8.096385542168674, "grad_norm": 3.974867582321167, "learning_rate": 9.594660194174758e-05, "loss": 0.1943, "step": 835 }, { "epoch": 8.106024096385543, "grad_norm": 6.8844122886657715, "learning_rate": 9.594174757281555e-05, "loss": 0.4509, "step": 836 }, { "epoch": 8.11566265060241, "grad_norm": 7.502753257751465, "learning_rate": 9.593689320388351e-05, "loss": 0.3833, "step": 837 }, { "epoch": 8.125301204819277, "grad_norm": 4.395756721496582, "learning_rate": 9.593203883495147e-05, "loss": 0.3924, "step": 838 }, { "epoch": 8.134939759036145, "grad_norm": 14.974113464355469, "learning_rate": 9.592718446601942e-05, "loss": 0.4458, "step": 839 }, { "epoch": 8.144578313253012, "grad_norm": 7.5383782386779785, "learning_rate": 9.592233009708738e-05, "loss": 0.5757, "step": 840 }, { "epoch": 8.154216867469879, "grad_norm": 8.029748916625977, "learning_rate": 9.591747572815535e-05, "loss": 0.3864, "step": 841 }, { "epoch": 8.163855421686748, "grad_norm": 4.352522850036621, "learning_rate": 9.591262135922331e-05, "loss": 0.4274, "step": 842 }, { "epoch": 8.173493975903614, "grad_norm": 6.003105163574219, "learning_rate": 9.590776699029127e-05, "loss": 0.234, "step": 843 }, { "epoch": 8.183132530120481, "grad_norm": 4.45254373550415, "learning_rate": 9.590291262135923e-05, "loss": 0.341, "step": 844 }, { "epoch": 8.19277108433735, "grad_norm": 2.46222186088562, "learning_rate": 9.58980582524272e-05, "loss": 0.1221, "step": 845 }, { "epoch": 8.202409638554217, "grad_norm": 16.1867618560791, "learning_rate": 9.589320388349515e-05, "loss": 0.782, "step": 846 }, { "epoch": 8.212048192771084, "grad_norm": 5.323999881744385, "learning_rate": 9.588834951456311e-05, "loss": 0.1952, "step": 847 }, { "epoch": 8.221686746987952, "grad_norm": 4.527608394622803, "learning_rate": 9.588349514563107e-05, "loss": 0.272, "step": 848 }, { "epoch": 8.23132530120482, "grad_norm": 6.195493698120117, "learning_rate": 9.587864077669904e-05, "loss": 0.3571, "step": 849 }, { "epoch": 8.240963855421686, "grad_norm": 6.394032001495361, "learning_rate": 9.5873786407767e-05, "loss": 0.5073, "step": 850 }, { "epoch": 8.250602409638555, "grad_norm": 3.6604530811309814, "learning_rate": 9.586893203883496e-05, "loss": 0.297, "step": 851 }, { "epoch": 8.260240963855422, "grad_norm": 3.134676933288574, "learning_rate": 9.586407766990291e-05, "loss": 0.2587, "step": 852 }, { "epoch": 8.269879518072289, "grad_norm": 9.267568588256836, "learning_rate": 9.585922330097089e-05, "loss": 0.2774, "step": 853 }, { "epoch": 8.279518072289157, "grad_norm": 4.405209541320801, "learning_rate": 9.585436893203884e-05, "loss": 0.2446, "step": 854 }, { "epoch": 8.289156626506024, "grad_norm": 6.455477714538574, "learning_rate": 9.58495145631068e-05, "loss": 0.3776, "step": 855 }, { "epoch": 8.298795180722891, "grad_norm": 4.959549903869629, "learning_rate": 9.584466019417476e-05, "loss": 0.3406, "step": 856 }, { "epoch": 8.30843373493976, "grad_norm": 8.275314331054688, "learning_rate": 9.583980582524272e-05, "loss": 0.5178, "step": 857 }, { "epoch": 8.318072289156627, "grad_norm": 5.288694858551025, "learning_rate": 9.583495145631069e-05, "loss": 0.3958, "step": 858 }, { "epoch": 8.327710843373493, "grad_norm": 2.4393229484558105, "learning_rate": 9.583009708737865e-05, "loss": 0.2104, "step": 859 }, { "epoch": 8.337349397590362, "grad_norm": 2.764439821243286, "learning_rate": 9.58252427184466e-05, "loss": 0.2361, "step": 860 }, { "epoch": 8.346987951807229, "grad_norm": 6.319993019104004, "learning_rate": 9.582038834951456e-05, "loss": 0.2685, "step": 861 }, { "epoch": 8.356626506024096, "grad_norm": 6.60000467300415, "learning_rate": 9.581553398058253e-05, "loss": 0.6034, "step": 862 }, { "epoch": 8.366265060240965, "grad_norm": 9.191675186157227, "learning_rate": 9.581067961165049e-05, "loss": 0.4558, "step": 863 }, { "epoch": 8.375903614457831, "grad_norm": 3.984930992126465, "learning_rate": 9.580582524271845e-05, "loss": 0.2383, "step": 864 }, { "epoch": 8.385542168674698, "grad_norm": 4.108213901519775, "learning_rate": 9.58009708737864e-05, "loss": 0.4179, "step": 865 }, { "epoch": 8.395180722891567, "grad_norm": 8.137394905090332, "learning_rate": 9.579611650485438e-05, "loss": 0.5296, "step": 866 }, { "epoch": 8.404819277108434, "grad_norm": 6.649435997009277, "learning_rate": 9.579126213592234e-05, "loss": 0.3432, "step": 867 }, { "epoch": 8.4144578313253, "grad_norm": 8.822879791259766, "learning_rate": 9.578640776699029e-05, "loss": 0.4052, "step": 868 }, { "epoch": 8.42409638554217, "grad_norm": 5.172307968139648, "learning_rate": 9.578155339805825e-05, "loss": 0.2552, "step": 869 }, { "epoch": 8.433734939759036, "grad_norm": 7.315627098083496, "learning_rate": 9.577669902912622e-05, "loss": 0.282, "step": 870 }, { "epoch": 8.443373493975903, "grad_norm": 6.523926734924316, "learning_rate": 9.577184466019418e-05, "loss": 0.2565, "step": 871 }, { "epoch": 8.453012048192772, "grad_norm": 9.507198333740234, "learning_rate": 9.576699029126214e-05, "loss": 0.4635, "step": 872 }, { "epoch": 8.462650602409639, "grad_norm": 8.053179740905762, "learning_rate": 9.57621359223301e-05, "loss": 0.3331, "step": 873 }, { "epoch": 8.472289156626506, "grad_norm": 6.262516975402832, "learning_rate": 9.575728155339805e-05, "loss": 0.3699, "step": 874 }, { "epoch": 8.481927710843374, "grad_norm": 3.44828724861145, "learning_rate": 9.575242718446602e-05, "loss": 0.3226, "step": 875 }, { "epoch": 8.491566265060241, "grad_norm": 6.028384208679199, "learning_rate": 9.574757281553398e-05, "loss": 0.4624, "step": 876 }, { "epoch": 8.501204819277108, "grad_norm": 7.1509013175964355, "learning_rate": 9.574271844660194e-05, "loss": 0.5553, "step": 877 }, { "epoch": 8.510843373493977, "grad_norm": 6.84628438949585, "learning_rate": 9.57378640776699e-05, "loss": 0.3552, "step": 878 }, { "epoch": 8.520481927710843, "grad_norm": 5.829349994659424, "learning_rate": 9.573300970873787e-05, "loss": 0.3953, "step": 879 }, { "epoch": 8.53012048192771, "grad_norm": 8.719776153564453, "learning_rate": 9.572815533980583e-05, "loss": 0.3344, "step": 880 }, { "epoch": 8.539759036144579, "grad_norm": 4.287723064422607, "learning_rate": 9.572330097087378e-05, "loss": 0.2309, "step": 881 }, { "epoch": 8.549397590361446, "grad_norm": 4.116950511932373, "learning_rate": 9.571844660194174e-05, "loss": 0.4048, "step": 882 }, { "epoch": 8.559036144578313, "grad_norm": 4.153770923614502, "learning_rate": 9.571359223300971e-05, "loss": 0.3319, "step": 883 }, { "epoch": 8.568674698795181, "grad_norm": 3.880678653717041, "learning_rate": 9.570873786407768e-05, "loss": 0.242, "step": 884 }, { "epoch": 8.578313253012048, "grad_norm": 10.701404571533203, "learning_rate": 9.570388349514564e-05, "loss": 0.2862, "step": 885 }, { "epoch": 8.587951807228915, "grad_norm": 5.59446382522583, "learning_rate": 9.56990291262136e-05, "loss": 0.2914, "step": 886 }, { "epoch": 8.597590361445784, "grad_norm": 4.238283634185791, "learning_rate": 9.569417475728156e-05, "loss": 0.256, "step": 887 }, { "epoch": 8.60722891566265, "grad_norm": 6.546535968780518, "learning_rate": 9.568932038834953e-05, "loss": 0.2986, "step": 888 }, { "epoch": 8.616867469879518, "grad_norm": 4.335517883300781, "learning_rate": 9.568446601941749e-05, "loss": 0.3791, "step": 889 }, { "epoch": 8.626506024096386, "grad_norm": 7.69927978515625, "learning_rate": 9.567961165048544e-05, "loss": 0.4016, "step": 890 }, { "epoch": 8.636144578313253, "grad_norm": 7.980609893798828, "learning_rate": 9.56747572815534e-05, "loss": 0.2789, "step": 891 }, { "epoch": 8.64578313253012, "grad_norm": 6.477508544921875, "learning_rate": 9.566990291262137e-05, "loss": 0.5016, "step": 892 }, { "epoch": 8.655421686746989, "grad_norm": 2.5606212615966797, "learning_rate": 9.566504854368933e-05, "loss": 0.127, "step": 893 }, { "epoch": 8.665060240963856, "grad_norm": 13.486055374145508, "learning_rate": 9.566019417475729e-05, "loss": 0.3238, "step": 894 }, { "epoch": 8.674698795180722, "grad_norm": 4.742214202880859, "learning_rate": 9.565533980582525e-05, "loss": 0.4959, "step": 895 }, { "epoch": 8.684337349397591, "grad_norm": 12.17177677154541, "learning_rate": 9.56504854368932e-05, "loss": 0.3137, "step": 896 }, { "epoch": 8.693975903614458, "grad_norm": 7.874961853027344, "learning_rate": 9.564563106796118e-05, "loss": 0.4677, "step": 897 }, { "epoch": 8.703614457831325, "grad_norm": 3.645742177963257, "learning_rate": 9.564077669902913e-05, "loss": 0.2798, "step": 898 }, { "epoch": 8.713253012048193, "grad_norm": 6.820101737976074, "learning_rate": 9.563592233009709e-05, "loss": 0.3727, "step": 899 }, { "epoch": 8.72289156626506, "grad_norm": 6.085211753845215, "learning_rate": 9.563106796116505e-05, "loss": 0.198, "step": 900 }, { "epoch": 8.732530120481927, "grad_norm": 6.188626289367676, "learning_rate": 9.562621359223302e-05, "loss": 0.453, "step": 901 }, { "epoch": 8.742168674698796, "grad_norm": 5.907408714294434, "learning_rate": 9.562135922330098e-05, "loss": 0.3598, "step": 902 }, { "epoch": 8.751807228915663, "grad_norm": 7.132232666015625, "learning_rate": 9.561650485436894e-05, "loss": 0.1602, "step": 903 }, { "epoch": 8.76144578313253, "grad_norm": 4.525591850280762, "learning_rate": 9.56116504854369e-05, "loss": 0.2623, "step": 904 }, { "epoch": 8.771084337349398, "grad_norm": 6.280615329742432, "learning_rate": 9.560679611650487e-05, "loss": 0.5075, "step": 905 }, { "epoch": 8.780722891566265, "grad_norm": 6.015519142150879, "learning_rate": 9.560194174757282e-05, "loss": 0.4705, "step": 906 }, { "epoch": 8.790361445783132, "grad_norm": 4.522576808929443, "learning_rate": 9.559708737864078e-05, "loss": 0.3178, "step": 907 }, { "epoch": 8.8, "grad_norm": 8.37850284576416, "learning_rate": 9.559223300970874e-05, "loss": 0.2284, "step": 908 }, { "epoch": 8.809638554216868, "grad_norm": 6.813063621520996, "learning_rate": 9.558737864077671e-05, "loss": 0.3216, "step": 909 }, { "epoch": 8.819277108433734, "grad_norm": 6.164306640625, "learning_rate": 9.558252427184467e-05, "loss": 0.3325, "step": 910 }, { "epoch": 8.828915662650603, "grad_norm": 6.686633586883545, "learning_rate": 9.557766990291263e-05, "loss": 0.4866, "step": 911 }, { "epoch": 8.83855421686747, "grad_norm": 4.340207099914551, "learning_rate": 9.557281553398058e-05, "loss": 0.349, "step": 912 }, { "epoch": 8.848192771084337, "grad_norm": 7.057284832000732, "learning_rate": 9.556796116504854e-05, "loss": 0.3246, "step": 913 }, { "epoch": 8.857831325301206, "grad_norm": 3.9535257816314697, "learning_rate": 9.556310679611651e-05, "loss": 0.2146, "step": 914 }, { "epoch": 8.867469879518072, "grad_norm": 4.591403007507324, "learning_rate": 9.555825242718447e-05, "loss": 0.3809, "step": 915 }, { "epoch": 8.87710843373494, "grad_norm": 5.763818740844727, "learning_rate": 9.555339805825243e-05, "loss": 0.4473, "step": 916 }, { "epoch": 8.886746987951808, "grad_norm": 5.436281204223633, "learning_rate": 9.554854368932039e-05, "loss": 0.3173, "step": 917 }, { "epoch": 8.896385542168675, "grad_norm": 2.75614857673645, "learning_rate": 9.554368932038836e-05, "loss": 0.2076, "step": 918 }, { "epoch": 8.906024096385542, "grad_norm": 4.113542079925537, "learning_rate": 9.553883495145631e-05, "loss": 0.3443, "step": 919 }, { "epoch": 8.91566265060241, "grad_norm": 5.693469047546387, "learning_rate": 9.553398058252427e-05, "loss": 0.4894, "step": 920 }, { "epoch": 8.925301204819277, "grad_norm": 3.452791690826416, "learning_rate": 9.552912621359223e-05, "loss": 0.3162, "step": 921 }, { "epoch": 8.934939759036144, "grad_norm": 3.6609370708465576, "learning_rate": 9.55242718446602e-05, "loss": 0.2498, "step": 922 }, { "epoch": 8.944578313253013, "grad_norm": 4.174489974975586, "learning_rate": 9.551941747572816e-05, "loss": 0.2475, "step": 923 }, { "epoch": 8.95421686746988, "grad_norm": 4.966146945953369, "learning_rate": 9.551456310679612e-05, "loss": 0.3147, "step": 924 }, { "epoch": 8.963855421686747, "grad_norm": 4.29378604888916, "learning_rate": 9.550970873786407e-05, "loss": 0.313, "step": 925 }, { "epoch": 8.973493975903615, "grad_norm": 7.167509078979492, "learning_rate": 9.550485436893205e-05, "loss": 0.5385, "step": 926 }, { "epoch": 8.983132530120482, "grad_norm": 4.268331527709961, "learning_rate": 9.55e-05, "loss": 0.4075, "step": 927 }, { "epoch": 8.992771084337349, "grad_norm": 10.635354042053223, "learning_rate": 9.549514563106796e-05, "loss": 0.2876, "step": 928 }, { "epoch": 9.008433734939759, "grad_norm": 7.979439735412598, "learning_rate": 9.549029126213592e-05, "loss": 0.2078, "step": 929 }, { "epoch": 9.018072289156626, "grad_norm": 3.7794876098632812, "learning_rate": 9.548543689320388e-05, "loss": 0.4421, "step": 930 }, { "epoch": 9.027710843373494, "grad_norm": 15.32957649230957, "learning_rate": 9.548058252427185e-05, "loss": 0.3354, "step": 931 }, { "epoch": 9.037349397590361, "grad_norm": 2.3296897411346436, "learning_rate": 9.54757281553398e-05, "loss": 0.2461, "step": 932 }, { "epoch": 9.046987951807228, "grad_norm": 12.953283309936523, "learning_rate": 9.547087378640776e-05, "loss": 0.515, "step": 933 }, { "epoch": 9.056626506024097, "grad_norm": 33.51260757446289, "learning_rate": 9.546601941747572e-05, "loss": 0.2169, "step": 934 }, { "epoch": 9.066265060240964, "grad_norm": 11.506074905395508, "learning_rate": 9.546116504854369e-05, "loss": 0.2704, "step": 935 }, { "epoch": 9.07590361445783, "grad_norm": 11.717639923095703, "learning_rate": 9.545631067961166e-05, "loss": 0.4196, "step": 936 }, { "epoch": 9.0855421686747, "grad_norm": 6.4125189781188965, "learning_rate": 9.545145631067962e-05, "loss": 0.3467, "step": 937 }, { "epoch": 9.095180722891566, "grad_norm": 12.349903106689453, "learning_rate": 9.544660194174758e-05, "loss": 0.2762, "step": 938 }, { "epoch": 9.104819277108433, "grad_norm": 7.61633825302124, "learning_rate": 9.544174757281554e-05, "loss": 0.5071, "step": 939 }, { "epoch": 9.114457831325302, "grad_norm": 10.173833847045898, "learning_rate": 9.543689320388351e-05, "loss": 0.2349, "step": 940 }, { "epoch": 9.124096385542169, "grad_norm": 23.164386749267578, "learning_rate": 9.543203883495147e-05, "loss": 0.5615, "step": 941 }, { "epoch": 9.133734939759035, "grad_norm": 9.151108741760254, "learning_rate": 9.542718446601942e-05, "loss": 0.4771, "step": 942 }, { "epoch": 9.143373493975904, "grad_norm": 33.896976470947266, "learning_rate": 9.542233009708738e-05, "loss": 0.4803, "step": 943 }, { "epoch": 9.153012048192771, "grad_norm": 23.88856315612793, "learning_rate": 9.541747572815535e-05, "loss": 0.4104, "step": 944 }, { "epoch": 9.162650602409638, "grad_norm": 17.708145141601562, "learning_rate": 9.541262135922331e-05, "loss": 0.3681, "step": 945 }, { "epoch": 9.172289156626507, "grad_norm": 3.293518304824829, "learning_rate": 9.540776699029127e-05, "loss": 0.1647, "step": 946 }, { "epoch": 9.181927710843373, "grad_norm": 44.920867919921875, "learning_rate": 9.540291262135923e-05, "loss": 0.3049, "step": 947 }, { "epoch": 9.19156626506024, "grad_norm": 3.0030977725982666, "learning_rate": 9.53980582524272e-05, "loss": 0.234, "step": 948 }, { "epoch": 9.201204819277109, "grad_norm": 8.501741409301758, "learning_rate": 9.539320388349516e-05, "loss": 0.188, "step": 949 }, { "epoch": 9.210843373493976, "grad_norm": 7.61237096786499, "learning_rate": 9.538834951456311e-05, "loss": 0.2812, "step": 950 }, { "epoch": 9.220481927710843, "grad_norm": 12.861624717712402, "learning_rate": 9.538349514563107e-05, "loss": 0.241, "step": 951 }, { "epoch": 9.230120481927711, "grad_norm": 3.4167985916137695, "learning_rate": 9.537864077669903e-05, "loss": 0.3686, "step": 952 }, { "epoch": 9.239759036144578, "grad_norm": 15.152127265930176, "learning_rate": 9.5373786407767e-05, "loss": 0.5103, "step": 953 }, { "epoch": 9.249397590361445, "grad_norm": 11.6535005569458, "learning_rate": 9.536893203883496e-05, "loss": 0.4635, "step": 954 }, { "epoch": 9.259036144578314, "grad_norm": 3.135021448135376, "learning_rate": 9.536407766990292e-05, "loss": 0.0787, "step": 955 }, { "epoch": 9.26867469879518, "grad_norm": 11.931025505065918, "learning_rate": 9.535922330097087e-05, "loss": 0.1995, "step": 956 }, { "epoch": 9.278313253012048, "grad_norm": 5.920558929443359, "learning_rate": 9.535436893203884e-05, "loss": 0.2702, "step": 957 }, { "epoch": 9.287951807228916, "grad_norm": 14.056199073791504, "learning_rate": 9.53495145631068e-05, "loss": 0.4392, "step": 958 }, { "epoch": 9.297590361445783, "grad_norm": 53.37441635131836, "learning_rate": 9.534466019417476e-05, "loss": 0.3453, "step": 959 }, { "epoch": 9.30722891566265, "grad_norm": 16.266544342041016, "learning_rate": 9.533980582524272e-05, "loss": 0.4896, "step": 960 }, { "epoch": 9.316867469879519, "grad_norm": 8.348306655883789, "learning_rate": 9.533495145631069e-05, "loss": 0.3247, "step": 961 }, { "epoch": 9.326506024096386, "grad_norm": 3.561495065689087, "learning_rate": 9.533009708737865e-05, "loss": 0.5294, "step": 962 }, { "epoch": 9.336144578313252, "grad_norm": 13.153687477111816, "learning_rate": 9.53252427184466e-05, "loss": 0.4894, "step": 963 }, { "epoch": 9.345783132530121, "grad_norm": 3.685177803039551, "learning_rate": 9.532038834951456e-05, "loss": 0.3749, "step": 964 }, { "epoch": 9.355421686746988, "grad_norm": 16.9092960357666, "learning_rate": 9.531553398058253e-05, "loss": 0.2428, "step": 965 }, { "epoch": 9.365060240963855, "grad_norm": 11.692341804504395, "learning_rate": 9.531067961165049e-05, "loss": 0.3553, "step": 966 }, { "epoch": 9.374698795180723, "grad_norm": 17.60066795349121, "learning_rate": 9.530582524271845e-05, "loss": 0.3999, "step": 967 }, { "epoch": 9.38433734939759, "grad_norm": 15.415523529052734, "learning_rate": 9.530097087378641e-05, "loss": 0.3592, "step": 968 }, { "epoch": 9.393975903614457, "grad_norm": 18.469871520996094, "learning_rate": 9.529611650485437e-05, "loss": 0.2615, "step": 969 }, { "epoch": 9.403614457831326, "grad_norm": 10.238862991333008, "learning_rate": 9.529126213592234e-05, "loss": 0.2208, "step": 970 }, { "epoch": 9.413253012048193, "grad_norm": 3.255254030227661, "learning_rate": 9.52864077669903e-05, "loss": 0.3067, "step": 971 }, { "epoch": 9.42289156626506, "grad_norm": 14.90932559967041, "learning_rate": 9.528155339805825e-05, "loss": 0.4363, "step": 972 }, { "epoch": 9.432530120481928, "grad_norm": 9.252453804016113, "learning_rate": 9.527669902912621e-05, "loss": 0.4449, "step": 973 }, { "epoch": 9.442168674698795, "grad_norm": 1.9764782190322876, "learning_rate": 9.527184466019418e-05, "loss": 0.1875, "step": 974 }, { "epoch": 9.451807228915662, "grad_norm": 7.760078430175781, "learning_rate": 9.526699029126214e-05, "loss": 0.309, "step": 975 }, { "epoch": 9.46144578313253, "grad_norm": 3.324559211730957, "learning_rate": 9.52621359223301e-05, "loss": 0.3291, "step": 976 }, { "epoch": 9.471084337349398, "grad_norm": 5.160024166107178, "learning_rate": 9.525728155339805e-05, "loss": 0.4379, "step": 977 }, { "epoch": 9.480722891566264, "grad_norm": 24.249513626098633, "learning_rate": 9.525242718446603e-05, "loss": 0.3001, "step": 978 }, { "epoch": 9.490361445783133, "grad_norm": 2.4967641830444336, "learning_rate": 9.524757281553398e-05, "loss": 0.4167, "step": 979 }, { "epoch": 9.5, "grad_norm": 8.144975662231445, "learning_rate": 9.524271844660194e-05, "loss": 0.5609, "step": 980 }, { "epoch": 9.509638554216867, "grad_norm": 3.4173836708068848, "learning_rate": 9.52378640776699e-05, "loss": 0.253, "step": 981 }, { "epoch": 9.519277108433736, "grad_norm": 9.164339065551758, "learning_rate": 9.523300970873787e-05, "loss": 0.3614, "step": 982 }, { "epoch": 9.528915662650602, "grad_norm": 20.31145668029785, "learning_rate": 9.522815533980583e-05, "loss": 0.4115, "step": 983 }, { "epoch": 9.53855421686747, "grad_norm": 13.304665565490723, "learning_rate": 9.522330097087379e-05, "loss": 0.4292, "step": 984 }, { "epoch": 9.548192771084338, "grad_norm": 22.90193748474121, "learning_rate": 9.521844660194176e-05, "loss": 0.2335, "step": 985 }, { "epoch": 9.557831325301205, "grad_norm": 6.815419673919678, "learning_rate": 9.521359223300971e-05, "loss": 0.3613, "step": 986 }, { "epoch": 9.567469879518072, "grad_norm": 17.64737319946289, "learning_rate": 9.520873786407769e-05, "loss": 0.4648, "step": 987 }, { "epoch": 9.57710843373494, "grad_norm": 2.7323696613311768, "learning_rate": 9.520388349514564e-05, "loss": 0.2249, "step": 988 }, { "epoch": 9.586746987951807, "grad_norm": 10.58283805847168, "learning_rate": 9.51990291262136e-05, "loss": 0.3417, "step": 989 }, { "epoch": 9.596385542168674, "grad_norm": 16.49382781982422, "learning_rate": 9.519417475728156e-05, "loss": 0.4734, "step": 990 }, { "epoch": 9.606024096385543, "grad_norm": 8.106828689575195, "learning_rate": 9.518932038834952e-05, "loss": 0.1922, "step": 991 }, { "epoch": 9.61566265060241, "grad_norm": 13.576820373535156, "learning_rate": 9.518446601941749e-05, "loss": 0.4994, "step": 992 }, { "epoch": 9.625301204819277, "grad_norm": 5.914340496063232, "learning_rate": 9.517961165048545e-05, "loss": 0.3214, "step": 993 }, { "epoch": 9.634939759036145, "grad_norm": 6.520223617553711, "learning_rate": 9.51747572815534e-05, "loss": 0.3547, "step": 994 }, { "epoch": 9.644578313253012, "grad_norm": 8.614643096923828, "learning_rate": 9.516990291262136e-05, "loss": 0.3389, "step": 995 }, { "epoch": 9.654216867469879, "grad_norm": 8.28887939453125, "learning_rate": 9.516504854368933e-05, "loss": 0.4748, "step": 996 }, { "epoch": 9.663855421686748, "grad_norm": 6.037021636962891, "learning_rate": 9.516019417475729e-05, "loss": 0.2778, "step": 997 }, { "epoch": 9.673493975903614, "grad_norm": 13.806314468383789, "learning_rate": 9.515533980582525e-05, "loss": 0.6265, "step": 998 }, { "epoch": 9.683132530120481, "grad_norm": 11.093732833862305, "learning_rate": 9.51504854368932e-05, "loss": 0.4734, "step": 999 }, { "epoch": 9.69277108433735, "grad_norm": 9.101907730102539, "learning_rate": 9.514563106796118e-05, "loss": 0.4745, "step": 1000 }, { "epoch": 9.702409638554217, "grad_norm": 16.201744079589844, "learning_rate": 9.514077669902914e-05, "loss": 0.2002, "step": 1001 }, { "epoch": 9.712048192771084, "grad_norm": 2.2347702980041504, "learning_rate": 9.513592233009709e-05, "loss": 0.5163, "step": 1002 }, { "epoch": 9.721686746987952, "grad_norm": 13.291701316833496, "learning_rate": 9.513106796116505e-05, "loss": 0.2493, "step": 1003 }, { "epoch": 9.73132530120482, "grad_norm": 10.74729061126709, "learning_rate": 9.512621359223302e-05, "loss": 0.2118, "step": 1004 }, { "epoch": 9.740963855421686, "grad_norm": 5.133732318878174, "learning_rate": 9.512135922330098e-05, "loss": 0.2528, "step": 1005 }, { "epoch": 9.750602409638555, "grad_norm": 8.883079528808594, "learning_rate": 9.511650485436894e-05, "loss": 0.2612, "step": 1006 }, { "epoch": 9.760240963855422, "grad_norm": 16.912948608398438, "learning_rate": 9.51116504854369e-05, "loss": 0.286, "step": 1007 }, { "epoch": 9.769879518072289, "grad_norm": 2.5654799938201904, "learning_rate": 9.510679611650485e-05, "loss": 0.2452, "step": 1008 }, { "epoch": 9.779518072289157, "grad_norm": 8.451665878295898, "learning_rate": 9.510194174757282e-05, "loss": 0.3327, "step": 1009 }, { "epoch": 9.789156626506024, "grad_norm": 6.127887725830078, "learning_rate": 9.509708737864078e-05, "loss": 0.1562, "step": 1010 }, { "epoch": 9.798795180722891, "grad_norm": 3.5389649868011475, "learning_rate": 9.509223300970874e-05, "loss": 0.1971, "step": 1011 }, { "epoch": 9.80843373493976, "grad_norm": 9.701542854309082, "learning_rate": 9.50873786407767e-05, "loss": 0.6627, "step": 1012 }, { "epoch": 9.818072289156627, "grad_norm": 11.72576904296875, "learning_rate": 9.508252427184467e-05, "loss": 0.3524, "step": 1013 }, { "epoch": 9.827710843373493, "grad_norm": 22.422971725463867, "learning_rate": 9.507766990291263e-05, "loss": 0.3205, "step": 1014 }, { "epoch": 9.837349397590362, "grad_norm": 4.493837356567383, "learning_rate": 9.507281553398058e-05, "loss": 0.4179, "step": 1015 }, { "epoch": 9.846987951807229, "grad_norm": 9.841538429260254, "learning_rate": 9.506796116504854e-05, "loss": 0.4543, "step": 1016 }, { "epoch": 9.856626506024096, "grad_norm": 11.978109359741211, "learning_rate": 9.506310679611651e-05, "loss": 0.5129, "step": 1017 }, { "epoch": 9.866265060240965, "grad_norm": 7.172471523284912, "learning_rate": 9.505825242718447e-05, "loss": 0.6335, "step": 1018 }, { "epoch": 9.875903614457831, "grad_norm": 2.8560235500335693, "learning_rate": 9.505339805825243e-05, "loss": 0.3924, "step": 1019 }, { "epoch": 9.885542168674698, "grad_norm": 48.07304382324219, "learning_rate": 9.504854368932039e-05, "loss": 0.5117, "step": 1020 }, { "epoch": 9.895180722891567, "grad_norm": 64.33214569091797, "learning_rate": 9.504368932038836e-05, "loss": 0.2729, "step": 1021 }, { "epoch": 9.904819277108434, "grad_norm": 15.568924903869629, "learning_rate": 9.503883495145632e-05, "loss": 0.4922, "step": 1022 }, { "epoch": 9.9144578313253, "grad_norm": 11.389730453491211, "learning_rate": 9.503398058252427e-05, "loss": 0.2401, "step": 1023 }, { "epoch": 9.92409638554217, "grad_norm": 8.367956161499023, "learning_rate": 9.502912621359223e-05, "loss": 0.3535, "step": 1024 }, { "epoch": 9.933734939759036, "grad_norm": 10.769963264465332, "learning_rate": 9.502427184466019e-05, "loss": 0.2976, "step": 1025 }, { "epoch": 9.943373493975903, "grad_norm": 3.716402769088745, "learning_rate": 9.501941747572816e-05, "loss": 0.2879, "step": 1026 }, { "epoch": 9.953012048192772, "grad_norm": 21.830402374267578, "learning_rate": 9.501456310679612e-05, "loss": 0.2722, "step": 1027 }, { "epoch": 9.962650602409639, "grad_norm": 14.56213092803955, "learning_rate": 9.500970873786408e-05, "loss": 0.6101, "step": 1028 }, { "epoch": 9.972289156626506, "grad_norm": 21.924989700317383, "learning_rate": 9.500485436893203e-05, "loss": 0.6417, "step": 1029 }, { "epoch": 9.981927710843374, "grad_norm": 3.6592490673065186, "learning_rate": 9.5e-05, "loss": 0.4355, "step": 1030 }, { "epoch": 9.991566265060241, "grad_norm": 19.190189361572266, "learning_rate": 9.499514563106796e-05, "loss": 0.5355, "step": 1031 }, { "epoch": 10.007228915662651, "grad_norm": 10.040400505065918, "learning_rate": 9.499029126213592e-05, "loss": 0.5576, "step": 1032 }, { "epoch": 10.016867469879518, "grad_norm": 5.667544841766357, "learning_rate": 9.498543689320388e-05, "loss": 0.43, "step": 1033 }, { "epoch": 10.026506024096385, "grad_norm": 12.370623588562012, "learning_rate": 9.498058252427185e-05, "loss": 0.3345, "step": 1034 }, { "epoch": 10.036144578313253, "grad_norm": 8.694356918334961, "learning_rate": 9.497572815533981e-05, "loss": 0.3589, "step": 1035 }, { "epoch": 10.04578313253012, "grad_norm": 32.16225814819336, "learning_rate": 9.497087378640778e-05, "loss": 0.4386, "step": 1036 }, { "epoch": 10.055421686746987, "grad_norm": 25.267925262451172, "learning_rate": 9.496601941747574e-05, "loss": 0.4419, "step": 1037 }, { "epoch": 10.065060240963856, "grad_norm": 6.537799835205078, "learning_rate": 9.49611650485437e-05, "loss": 0.3133, "step": 1038 }, { "epoch": 10.074698795180723, "grad_norm": 7.824986457824707, "learning_rate": 9.495631067961167e-05, "loss": 0.2514, "step": 1039 }, { "epoch": 10.08433734939759, "grad_norm": 4.506023406982422, "learning_rate": 9.495145631067962e-05, "loss": 0.5044, "step": 1040 }, { "epoch": 10.093975903614458, "grad_norm": 2.831132411956787, "learning_rate": 9.494660194174758e-05, "loss": 0.2046, "step": 1041 }, { "epoch": 10.103614457831325, "grad_norm": 27.669939041137695, "learning_rate": 9.494174757281554e-05, "loss": 0.4782, "step": 1042 }, { "epoch": 10.113253012048192, "grad_norm": 5.728501796722412, "learning_rate": 9.49368932038835e-05, "loss": 0.3415, "step": 1043 }, { "epoch": 10.12289156626506, "grad_norm": 15.909664154052734, "learning_rate": 9.493203883495147e-05, "loss": 0.3274, "step": 1044 }, { "epoch": 10.132530120481928, "grad_norm": 7.922830581665039, "learning_rate": 9.492718446601943e-05, "loss": 0.3086, "step": 1045 }, { "epoch": 10.142168674698794, "grad_norm": 8.775146484375, "learning_rate": 9.492233009708738e-05, "loss": 0.446, "step": 1046 }, { "epoch": 10.151807228915663, "grad_norm": 9.284746170043945, "learning_rate": 9.491747572815534e-05, "loss": 0.4281, "step": 1047 }, { "epoch": 10.16144578313253, "grad_norm": 19.984411239624023, "learning_rate": 9.491262135922331e-05, "loss": 0.2883, "step": 1048 }, { "epoch": 10.171084337349397, "grad_norm": 11.440886497497559, "learning_rate": 9.490776699029127e-05, "loss": 0.4109, "step": 1049 }, { "epoch": 10.180722891566266, "grad_norm": 2.8654439449310303, "learning_rate": 9.490291262135923e-05, "loss": 0.4606, "step": 1050 }, { "epoch": 10.190361445783132, "grad_norm": 8.872488975524902, "learning_rate": 9.489805825242719e-05, "loss": 0.3288, "step": 1051 }, { "epoch": 10.2, "grad_norm": 7.491482257843018, "learning_rate": 9.489320388349516e-05, "loss": 0.5068, "step": 1052 }, { "epoch": 10.209638554216868, "grad_norm": 7.990087032318115, "learning_rate": 9.488834951456311e-05, "loss": 0.3248, "step": 1053 }, { "epoch": 10.219277108433735, "grad_norm": 1.4819023609161377, "learning_rate": 9.488349514563107e-05, "loss": 0.222, "step": 1054 }, { "epoch": 10.228915662650602, "grad_norm": 12.95163631439209, "learning_rate": 9.487864077669903e-05, "loss": 0.3247, "step": 1055 }, { "epoch": 10.23855421686747, "grad_norm": 7.08576774597168, "learning_rate": 9.4873786407767e-05, "loss": 0.3543, "step": 1056 }, { "epoch": 10.248192771084337, "grad_norm": 15.44316291809082, "learning_rate": 9.486893203883496e-05, "loss": 0.4305, "step": 1057 }, { "epoch": 10.257831325301204, "grad_norm": 10.679027557373047, "learning_rate": 9.486407766990292e-05, "loss": 0.2945, "step": 1058 }, { "epoch": 10.267469879518073, "grad_norm": 6.489208221435547, "learning_rate": 9.485922330097088e-05, "loss": 0.4662, "step": 1059 }, { "epoch": 10.27710843373494, "grad_norm": 18.526269912719727, "learning_rate": 9.485436893203883e-05, "loss": 0.4317, "step": 1060 }, { "epoch": 10.286746987951807, "grad_norm": 3.189955472946167, "learning_rate": 9.48495145631068e-05, "loss": 0.2709, "step": 1061 }, { "epoch": 10.296385542168675, "grad_norm": 10.084278106689453, "learning_rate": 9.484466019417476e-05, "loss": 0.3651, "step": 1062 }, { "epoch": 10.306024096385542, "grad_norm": 5.338037967681885, "learning_rate": 9.483980582524272e-05, "loss": 0.2197, "step": 1063 }, { "epoch": 10.315662650602409, "grad_norm": 14.416592597961426, "learning_rate": 9.483495145631068e-05, "loss": 0.25, "step": 1064 }, { "epoch": 10.325301204819278, "grad_norm": 7.047606468200684, "learning_rate": 9.483009708737865e-05, "loss": 0.3057, "step": 1065 }, { "epoch": 10.334939759036144, "grad_norm": 8.511214256286621, "learning_rate": 9.48252427184466e-05, "loss": 0.3278, "step": 1066 }, { "epoch": 10.344578313253011, "grad_norm": 2.728191375732422, "learning_rate": 9.482038834951456e-05, "loss": 0.4158, "step": 1067 }, { "epoch": 10.35421686746988, "grad_norm": 16.185029983520508, "learning_rate": 9.481553398058252e-05, "loss": 0.4381, "step": 1068 }, { "epoch": 10.363855421686747, "grad_norm": 2.3805456161499023, "learning_rate": 9.48106796116505e-05, "loss": 0.2863, "step": 1069 }, { "epoch": 10.373493975903614, "grad_norm": 12.316679000854492, "learning_rate": 9.480582524271845e-05, "loss": 0.4991, "step": 1070 }, { "epoch": 10.383132530120482, "grad_norm": 7.693389892578125, "learning_rate": 9.480097087378641e-05, "loss": 0.173, "step": 1071 }, { "epoch": 10.39277108433735, "grad_norm": 25.200416564941406, "learning_rate": 9.479611650485437e-05, "loss": 0.3048, "step": 1072 }, { "epoch": 10.402409638554216, "grad_norm": 19.360721588134766, "learning_rate": 9.479126213592234e-05, "loss": 0.4744, "step": 1073 }, { "epoch": 10.412048192771085, "grad_norm": 4.1931891441345215, "learning_rate": 9.47864077669903e-05, "loss": 0.2712, "step": 1074 }, { "epoch": 10.421686746987952, "grad_norm": 3.6062424182891846, "learning_rate": 9.478155339805825e-05, "loss": 0.5521, "step": 1075 }, { "epoch": 10.431325301204819, "grad_norm": 10.143965721130371, "learning_rate": 9.477669902912621e-05, "loss": 0.5839, "step": 1076 }, { "epoch": 10.440963855421687, "grad_norm": 18.666362762451172, "learning_rate": 9.477184466019417e-05, "loss": 0.337, "step": 1077 }, { "epoch": 10.450602409638554, "grad_norm": 20.518930435180664, "learning_rate": 9.476699029126214e-05, "loss": 0.5296, "step": 1078 }, { "epoch": 10.460240963855421, "grad_norm": 15.532368659973145, "learning_rate": 9.47621359223301e-05, "loss": 0.2444, "step": 1079 }, { "epoch": 10.46987951807229, "grad_norm": 4.35224723815918, "learning_rate": 9.475728155339806e-05, "loss": 0.5003, "step": 1080 }, { "epoch": 10.479518072289157, "grad_norm": 4.736363410949707, "learning_rate": 9.475242718446601e-05, "loss": 0.379, "step": 1081 }, { "epoch": 10.489156626506023, "grad_norm": 19.39057731628418, "learning_rate": 9.474757281553398e-05, "loss": 0.3381, "step": 1082 }, { "epoch": 10.498795180722892, "grad_norm": 9.689014434814453, "learning_rate": 9.474271844660194e-05, "loss": 0.3828, "step": 1083 }, { "epoch": 10.508433734939759, "grad_norm": 11.691107749938965, "learning_rate": 9.47378640776699e-05, "loss": 0.4336, "step": 1084 }, { "epoch": 10.518072289156626, "grad_norm": 17.848012924194336, "learning_rate": 9.473300970873786e-05, "loss": 0.34, "step": 1085 }, { "epoch": 10.527710843373494, "grad_norm": 12.838643074035645, "learning_rate": 9.472815533980583e-05, "loss": 0.4401, "step": 1086 }, { "epoch": 10.537349397590361, "grad_norm": 22.676746368408203, "learning_rate": 9.47233009708738e-05, "loss": 0.3921, "step": 1087 }, { "epoch": 10.546987951807228, "grad_norm": 6.209812164306641, "learning_rate": 9.471844660194176e-05, "loss": 0.3107, "step": 1088 }, { "epoch": 10.556626506024097, "grad_norm": 18.091888427734375, "learning_rate": 9.471359223300972e-05, "loss": 0.429, "step": 1089 }, { "epoch": 10.566265060240964, "grad_norm": 2.2251713275909424, "learning_rate": 9.470873786407767e-05, "loss": 0.4012, "step": 1090 }, { "epoch": 10.57590361445783, "grad_norm": 18.102251052856445, "learning_rate": 9.470388349514565e-05, "loss": 0.4483, "step": 1091 }, { "epoch": 10.5855421686747, "grad_norm": 3.0563724040985107, "learning_rate": 9.46990291262136e-05, "loss": 0.2939, "step": 1092 }, { "epoch": 10.595180722891566, "grad_norm": 9.619827270507812, "learning_rate": 9.469417475728156e-05, "loss": 0.3631, "step": 1093 }, { "epoch": 10.604819277108433, "grad_norm": 15.747815132141113, "learning_rate": 9.468932038834952e-05, "loss": 0.2884, "step": 1094 }, { "epoch": 10.614457831325302, "grad_norm": 2.358610153198242, "learning_rate": 9.468446601941749e-05, "loss": 0.1893, "step": 1095 }, { "epoch": 10.624096385542169, "grad_norm": 4.085537433624268, "learning_rate": 9.467961165048545e-05, "loss": 0.4663, "step": 1096 }, { "epoch": 10.633734939759035, "grad_norm": 8.617310523986816, "learning_rate": 9.46747572815534e-05, "loss": 0.5424, "step": 1097 }, { "epoch": 10.643373493975904, "grad_norm": 14.695164680480957, "learning_rate": 9.466990291262136e-05, "loss": 0.4214, "step": 1098 }, { "epoch": 10.653012048192771, "grad_norm": 1.6092551946640015, "learning_rate": 9.466504854368932e-05, "loss": 0.3456, "step": 1099 }, { "epoch": 10.662650602409638, "grad_norm": 19.758892059326172, "learning_rate": 9.466019417475729e-05, "loss": 0.2515, "step": 1100 }, { "epoch": 10.672289156626507, "grad_norm": 15.715566635131836, "learning_rate": 9.465533980582525e-05, "loss": 0.375, "step": 1101 }, { "epoch": 10.681927710843373, "grad_norm": 7.141045093536377, "learning_rate": 9.465048543689321e-05, "loss": 0.4862, "step": 1102 }, { "epoch": 10.69156626506024, "grad_norm": 25.56489372253418, "learning_rate": 9.464563106796117e-05, "loss": 0.5232, "step": 1103 }, { "epoch": 10.701204819277109, "grad_norm": 9.890853881835938, "learning_rate": 9.464077669902914e-05, "loss": 0.4687, "step": 1104 }, { "epoch": 10.710843373493976, "grad_norm": 19.543123245239258, "learning_rate": 9.46359223300971e-05, "loss": 0.2011, "step": 1105 }, { "epoch": 10.720481927710843, "grad_norm": 6.735100746154785, "learning_rate": 9.463106796116505e-05, "loss": 0.2253, "step": 1106 }, { "epoch": 10.730120481927711, "grad_norm": 15.29575252532959, "learning_rate": 9.462621359223301e-05, "loss": 0.389, "step": 1107 }, { "epoch": 10.739759036144578, "grad_norm": 15.985859870910645, "learning_rate": 9.462135922330098e-05, "loss": 0.3561, "step": 1108 }, { "epoch": 10.749397590361445, "grad_norm": 22.890214920043945, "learning_rate": 9.461650485436894e-05, "loss": 0.4343, "step": 1109 }, { "epoch": 10.759036144578314, "grad_norm": 7.34912633895874, "learning_rate": 9.46116504854369e-05, "loss": 0.3928, "step": 1110 }, { "epoch": 10.76867469879518, "grad_norm": 11.022401809692383, "learning_rate": 9.460679611650485e-05, "loss": 0.2366, "step": 1111 }, { "epoch": 10.778313253012048, "grad_norm": 6.655566692352295, "learning_rate": 9.460194174757283e-05, "loss": 0.2703, "step": 1112 }, { "epoch": 10.787951807228916, "grad_norm": 14.001361846923828, "learning_rate": 9.459708737864078e-05, "loss": 0.3015, "step": 1113 }, { "epoch": 10.797590361445783, "grad_norm": 21.47203254699707, "learning_rate": 9.459223300970874e-05, "loss": 0.4856, "step": 1114 }, { "epoch": 10.80722891566265, "grad_norm": 2.257009983062744, "learning_rate": 9.45873786407767e-05, "loss": 0.1194, "step": 1115 }, { "epoch": 10.816867469879519, "grad_norm": 7.330010414123535, "learning_rate": 9.458252427184466e-05, "loss": 0.4243, "step": 1116 }, { "epoch": 10.826506024096386, "grad_norm": 3.5494229793548584, "learning_rate": 9.457766990291263e-05, "loss": 0.2662, "step": 1117 }, { "epoch": 10.836144578313252, "grad_norm": 31.855527877807617, "learning_rate": 9.457281553398059e-05, "loss": 0.4286, "step": 1118 }, { "epoch": 10.845783132530121, "grad_norm": 4.657224655151367, "learning_rate": 9.456796116504854e-05, "loss": 0.2777, "step": 1119 }, { "epoch": 10.855421686746988, "grad_norm": 6.560201644897461, "learning_rate": 9.45631067961165e-05, "loss": 0.2457, "step": 1120 }, { "epoch": 10.865060240963855, "grad_norm": 5.406611919403076, "learning_rate": 9.455825242718447e-05, "loss": 0.344, "step": 1121 }, { "epoch": 10.874698795180723, "grad_norm": 3.9546597003936768, "learning_rate": 9.455339805825243e-05, "loss": 0.3128, "step": 1122 }, { "epoch": 10.88433734939759, "grad_norm": 31.386327743530273, "learning_rate": 9.454854368932039e-05, "loss": 0.4126, "step": 1123 }, { "epoch": 10.893975903614457, "grad_norm": 14.564229011535645, "learning_rate": 9.454368932038835e-05, "loss": 0.3203, "step": 1124 }, { "epoch": 10.903614457831326, "grad_norm": 9.343304634094238, "learning_rate": 9.453883495145632e-05, "loss": 0.4957, "step": 1125 }, { "epoch": 10.913253012048193, "grad_norm": 19.55753517150879, "learning_rate": 9.453398058252428e-05, "loss": 0.5816, "step": 1126 }, { "epoch": 10.92289156626506, "grad_norm": 7.302728176116943, "learning_rate": 9.452912621359223e-05, "loss": 0.1751, "step": 1127 }, { "epoch": 10.932530120481928, "grad_norm": 11.926730155944824, "learning_rate": 9.452427184466019e-05, "loss": 0.3203, "step": 1128 }, { "epoch": 10.942168674698795, "grad_norm": 4.6094279289245605, "learning_rate": 9.451941747572816e-05, "loss": 0.3856, "step": 1129 }, { "epoch": 10.951807228915662, "grad_norm": 5.784184455871582, "learning_rate": 9.451456310679612e-05, "loss": 0.453, "step": 1130 }, { "epoch": 10.96144578313253, "grad_norm": 3.794538974761963, "learning_rate": 9.450970873786408e-05, "loss": 0.2729, "step": 1131 }, { "epoch": 10.971084337349398, "grad_norm": 12.080953598022461, "learning_rate": 9.450485436893204e-05, "loss": 0.2804, "step": 1132 }, { "epoch": 10.980722891566264, "grad_norm": 8.404790878295898, "learning_rate": 9.449999999999999e-05, "loss": 0.5715, "step": 1133 }, { "epoch": 10.990361445783133, "grad_norm": 21.766483306884766, "learning_rate": 9.449514563106796e-05, "loss": 0.335, "step": 1134 }, { "epoch": 11.006024096385541, "grad_norm": 5.744012832641602, "learning_rate": 9.449029126213592e-05, "loss": 0.4329, "step": 1135 }, { "epoch": 11.01566265060241, "grad_norm": 4.378734588623047, "learning_rate": 9.448543689320388e-05, "loss": 0.2965, "step": 1136 }, { "epoch": 11.025301204819277, "grad_norm": 15.626556396484375, "learning_rate": 9.448058252427185e-05, "loss": 0.7374, "step": 1137 }, { "epoch": 11.034939759036144, "grad_norm": 5.83518123626709, "learning_rate": 9.447572815533981e-05, "loss": 0.3256, "step": 1138 }, { "epoch": 11.044578313253012, "grad_norm": 24.504810333251953, "learning_rate": 9.447087378640778e-05, "loss": 0.2915, "step": 1139 }, { "epoch": 11.05421686746988, "grad_norm": 15.183797836303711, "learning_rate": 9.446601941747574e-05, "loss": 0.3525, "step": 1140 }, { "epoch": 11.063855421686746, "grad_norm": 4.3375749588012695, "learning_rate": 9.44611650485437e-05, "loss": 0.4032, "step": 1141 }, { "epoch": 11.073493975903615, "grad_norm": 12.138446807861328, "learning_rate": 9.445631067961165e-05, "loss": 0.4676, "step": 1142 }, { "epoch": 11.083132530120482, "grad_norm": 14.982931137084961, "learning_rate": 9.445145631067962e-05, "loss": 0.2277, "step": 1143 }, { "epoch": 11.092771084337349, "grad_norm": 9.667319297790527, "learning_rate": 9.444660194174758e-05, "loss": 0.246, "step": 1144 }, { "epoch": 11.102409638554217, "grad_norm": 7.368207931518555, "learning_rate": 9.444174757281554e-05, "loss": 0.3953, "step": 1145 }, { "epoch": 11.112048192771084, "grad_norm": 5.2928972244262695, "learning_rate": 9.44368932038835e-05, "loss": 0.1556, "step": 1146 }, { "epoch": 11.121686746987951, "grad_norm": 10.399062156677246, "learning_rate": 9.443203883495147e-05, "loss": 0.3021, "step": 1147 }, { "epoch": 11.13132530120482, "grad_norm": 10.52424144744873, "learning_rate": 9.442718446601943e-05, "loss": 0.4817, "step": 1148 }, { "epoch": 11.140963855421687, "grad_norm": 4.978135585784912, "learning_rate": 9.442233009708738e-05, "loss": 0.4621, "step": 1149 }, { "epoch": 11.150602409638553, "grad_norm": 11.406804084777832, "learning_rate": 9.441747572815534e-05, "loss": 0.4604, "step": 1150 }, { "epoch": 11.160240963855422, "grad_norm": 6.3644304275512695, "learning_rate": 9.441262135922331e-05, "loss": 0.3361, "step": 1151 }, { "epoch": 11.169879518072289, "grad_norm": 10.777481079101562, "learning_rate": 9.440776699029127e-05, "loss": 0.2869, "step": 1152 }, { "epoch": 11.179518072289156, "grad_norm": 5.800930023193359, "learning_rate": 9.440291262135923e-05, "loss": 0.1712, "step": 1153 }, { "epoch": 11.189156626506024, "grad_norm": 1.1745638847351074, "learning_rate": 9.439805825242719e-05, "loss": 0.2909, "step": 1154 }, { "epoch": 11.198795180722891, "grad_norm": 28.50299835205078, "learning_rate": 9.439320388349515e-05, "loss": 0.4394, "step": 1155 }, { "epoch": 11.208433734939758, "grad_norm": 27.473478317260742, "learning_rate": 9.438834951456312e-05, "loss": 0.3186, "step": 1156 }, { "epoch": 11.218072289156627, "grad_norm": 6.313675880432129, "learning_rate": 9.438349514563107e-05, "loss": 0.4178, "step": 1157 }, { "epoch": 11.227710843373494, "grad_norm": 7.832062721252441, "learning_rate": 9.437864077669903e-05, "loss": 0.5219, "step": 1158 }, { "epoch": 11.23734939759036, "grad_norm": 18.89845085144043, "learning_rate": 9.437378640776699e-05, "loss": 0.7173, "step": 1159 }, { "epoch": 11.24698795180723, "grad_norm": 9.031190872192383, "learning_rate": 9.436893203883496e-05, "loss": 0.5521, "step": 1160 }, { "epoch": 11.256626506024096, "grad_norm": 5.08243989944458, "learning_rate": 9.436407766990292e-05, "loss": 0.478, "step": 1161 }, { "epoch": 11.266265060240963, "grad_norm": 10.447175979614258, "learning_rate": 9.435922330097088e-05, "loss": 0.3973, "step": 1162 }, { "epoch": 11.275903614457832, "grad_norm": 5.599060535430908, "learning_rate": 9.435436893203883e-05, "loss": 0.3144, "step": 1163 }, { "epoch": 11.285542168674699, "grad_norm": 12.309122085571289, "learning_rate": 9.43495145631068e-05, "loss": 0.4372, "step": 1164 }, { "epoch": 11.295180722891565, "grad_norm": 7.778517246246338, "learning_rate": 9.434466019417476e-05, "loss": 0.299, "step": 1165 }, { "epoch": 11.304819277108434, "grad_norm": 11.024553298950195, "learning_rate": 9.433980582524272e-05, "loss": 0.3752, "step": 1166 }, { "epoch": 11.314457831325301, "grad_norm": 16.030969619750977, "learning_rate": 9.433495145631068e-05, "loss": 0.3927, "step": 1167 }, { "epoch": 11.324096385542168, "grad_norm": 3.1852493286132812, "learning_rate": 9.433009708737865e-05, "loss": 0.2259, "step": 1168 }, { "epoch": 11.333734939759037, "grad_norm": 25.44487190246582, "learning_rate": 9.432524271844661e-05, "loss": 0.4831, "step": 1169 }, { "epoch": 11.343373493975903, "grad_norm": 10.107372283935547, "learning_rate": 9.432038834951457e-05, "loss": 0.363, "step": 1170 }, { "epoch": 11.35301204819277, "grad_norm": 4.83876371383667, "learning_rate": 9.431553398058252e-05, "loss": 0.3271, "step": 1171 }, { "epoch": 11.362650602409639, "grad_norm": 4.263241767883301, "learning_rate": 9.431067961165048e-05, "loss": 0.1484, "step": 1172 }, { "epoch": 11.372289156626506, "grad_norm": 12.71500015258789, "learning_rate": 9.430582524271845e-05, "loss": 0.4628, "step": 1173 }, { "epoch": 11.381927710843373, "grad_norm": 2.4149274826049805, "learning_rate": 9.430097087378641e-05, "loss": 0.3692, "step": 1174 }, { "epoch": 11.391566265060241, "grad_norm": 2.1729819774627686, "learning_rate": 9.429611650485437e-05, "loss": 0.2216, "step": 1175 }, { "epoch": 11.401204819277108, "grad_norm": 9.090463638305664, "learning_rate": 9.429126213592233e-05, "loss": 0.3936, "step": 1176 }, { "epoch": 11.410843373493975, "grad_norm": 13.43541145324707, "learning_rate": 9.42864077669903e-05, "loss": 0.3515, "step": 1177 }, { "epoch": 11.420481927710844, "grad_norm": 3.6988446712493896, "learning_rate": 9.428155339805825e-05, "loss": 0.244, "step": 1178 }, { "epoch": 11.43012048192771, "grad_norm": 29.07266616821289, "learning_rate": 9.427669902912621e-05, "loss": 0.4161, "step": 1179 }, { "epoch": 11.439759036144578, "grad_norm": 30.750850677490234, "learning_rate": 9.427184466019417e-05, "loss": 0.4006, "step": 1180 }, { "epoch": 11.449397590361446, "grad_norm": 4.638474941253662, "learning_rate": 9.426699029126214e-05, "loss": 0.1778, "step": 1181 }, { "epoch": 11.459036144578313, "grad_norm": 6.326756954193115, "learning_rate": 9.42621359223301e-05, "loss": 0.4498, "step": 1182 }, { "epoch": 11.46867469879518, "grad_norm": 2.867757797241211, "learning_rate": 9.425728155339806e-05, "loss": 0.1445, "step": 1183 }, { "epoch": 11.478313253012049, "grad_norm": 1.9351630210876465, "learning_rate": 9.425242718446601e-05, "loss": 0.2488, "step": 1184 }, { "epoch": 11.487951807228916, "grad_norm": 3.3006675243377686, "learning_rate": 9.424757281553399e-05, "loss": 0.427, "step": 1185 }, { "epoch": 11.497590361445782, "grad_norm": 14.917038917541504, "learning_rate": 9.424271844660194e-05, "loss": 0.2807, "step": 1186 }, { "epoch": 11.507228915662651, "grad_norm": 12.562067985534668, "learning_rate": 9.423786407766992e-05, "loss": 0.3933, "step": 1187 }, { "epoch": 11.516867469879518, "grad_norm": 27.284088134765625, "learning_rate": 9.423300970873787e-05, "loss": 0.4954, "step": 1188 }, { "epoch": 11.526506024096385, "grad_norm": 9.157196044921875, "learning_rate": 9.422815533980583e-05, "loss": 0.6828, "step": 1189 }, { "epoch": 11.536144578313253, "grad_norm": 37.726539611816406, "learning_rate": 9.42233009708738e-05, "loss": 0.3065, "step": 1190 }, { "epoch": 11.54578313253012, "grad_norm": 4.501086235046387, "learning_rate": 9.421844660194176e-05, "loss": 0.4928, "step": 1191 }, { "epoch": 11.555421686746987, "grad_norm": 14.532305717468262, "learning_rate": 9.421359223300972e-05, "loss": 0.5234, "step": 1192 }, { "epoch": 11.565060240963856, "grad_norm": 24.87577247619629, "learning_rate": 9.420873786407768e-05, "loss": 0.4571, "step": 1193 }, { "epoch": 11.574698795180723, "grad_norm": 6.312738418579102, "learning_rate": 9.420388349514563e-05, "loss": 0.2264, "step": 1194 }, { "epoch": 11.58433734939759, "grad_norm": 9.380632400512695, "learning_rate": 9.41990291262136e-05, "loss": 0.272, "step": 1195 }, { "epoch": 11.593975903614458, "grad_norm": 4.956829071044922, "learning_rate": 9.419417475728156e-05, "loss": 0.2996, "step": 1196 }, { "epoch": 11.603614457831325, "grad_norm": 5.649827003479004, "learning_rate": 9.418932038834952e-05, "loss": 0.3484, "step": 1197 }, { "epoch": 11.613253012048192, "grad_norm": 16.479999542236328, "learning_rate": 9.418446601941748e-05, "loss": 0.6501, "step": 1198 }, { "epoch": 11.62289156626506, "grad_norm": 47.91411209106445, "learning_rate": 9.417961165048545e-05, "loss": 0.2021, "step": 1199 }, { "epoch": 11.632530120481928, "grad_norm": 13.795291900634766, "learning_rate": 9.417475728155341e-05, "loss": 0.5577, "step": 1200 }, { "epoch": 11.642168674698794, "grad_norm": 14.46544361114502, "learning_rate": 9.416990291262136e-05, "loss": 0.2834, "step": 1201 }, { "epoch": 11.651807228915663, "grad_norm": 4.656355857849121, "learning_rate": 9.416504854368932e-05, "loss": 0.2868, "step": 1202 }, { "epoch": 11.66144578313253, "grad_norm": 13.784445762634277, "learning_rate": 9.41601941747573e-05, "loss": 0.3648, "step": 1203 }, { "epoch": 11.671084337349397, "grad_norm": 2.37286114692688, "learning_rate": 9.415533980582525e-05, "loss": 0.2599, "step": 1204 }, { "epoch": 11.680722891566266, "grad_norm": 19.398046493530273, "learning_rate": 9.415048543689321e-05, "loss": 0.4259, "step": 1205 }, { "epoch": 11.690361445783132, "grad_norm": 8.759932518005371, "learning_rate": 9.414563106796117e-05, "loss": 0.6769, "step": 1206 }, { "epoch": 11.7, "grad_norm": 6.947553634643555, "learning_rate": 9.414077669902914e-05, "loss": 0.4047, "step": 1207 }, { "epoch": 11.709638554216868, "grad_norm": 6.137080669403076, "learning_rate": 9.41359223300971e-05, "loss": 0.348, "step": 1208 }, { "epoch": 11.719277108433735, "grad_norm": 19.019384384155273, "learning_rate": 9.413106796116505e-05, "loss": 0.297, "step": 1209 }, { "epoch": 11.728915662650602, "grad_norm": 9.781991958618164, "learning_rate": 9.412621359223301e-05, "loss": 0.3084, "step": 1210 }, { "epoch": 11.73855421686747, "grad_norm": 6.384054183959961, "learning_rate": 9.412135922330097e-05, "loss": 0.3353, "step": 1211 }, { "epoch": 11.748192771084337, "grad_norm": 15.789347648620605, "learning_rate": 9.411650485436894e-05, "loss": 0.4849, "step": 1212 }, { "epoch": 11.757831325301204, "grad_norm": 7.0925116539001465, "learning_rate": 9.41116504854369e-05, "loss": 0.5, "step": 1213 }, { "epoch": 11.767469879518073, "grad_norm": 7.8082170486450195, "learning_rate": 9.410679611650486e-05, "loss": 0.4921, "step": 1214 }, { "epoch": 11.77710843373494, "grad_norm": 1.9507871866226196, "learning_rate": 9.410194174757281e-05, "loss": 0.2521, "step": 1215 }, { "epoch": 11.786746987951807, "grad_norm": 1.9203262329101562, "learning_rate": 9.409708737864079e-05, "loss": 0.2227, "step": 1216 }, { "epoch": 11.796385542168675, "grad_norm": 4.82496452331543, "learning_rate": 9.409223300970874e-05, "loss": 0.3569, "step": 1217 }, { "epoch": 11.806024096385542, "grad_norm": 5.289453506469727, "learning_rate": 9.40873786407767e-05, "loss": 0.4595, "step": 1218 }, { "epoch": 11.815662650602409, "grad_norm": 16.47198486328125, "learning_rate": 9.408252427184466e-05, "loss": 0.319, "step": 1219 }, { "epoch": 11.825301204819278, "grad_norm": 5.464609146118164, "learning_rate": 9.407766990291263e-05, "loss": 0.4446, "step": 1220 }, { "epoch": 11.834939759036144, "grad_norm": 2.3145804405212402, "learning_rate": 9.407281553398059e-05, "loss": 0.2647, "step": 1221 }, { "epoch": 11.844578313253011, "grad_norm": 9.221813201904297, "learning_rate": 9.406796116504855e-05, "loss": 0.5505, "step": 1222 }, { "epoch": 11.85421686746988, "grad_norm": 15.935184478759766, "learning_rate": 9.40631067961165e-05, "loss": 0.4575, "step": 1223 }, { "epoch": 11.863855421686747, "grad_norm": 6.446454048156738, "learning_rate": 9.405825242718447e-05, "loss": 0.3682, "step": 1224 }, { "epoch": 11.873493975903614, "grad_norm": 8.394899368286133, "learning_rate": 9.405339805825243e-05, "loss": 0.5048, "step": 1225 }, { "epoch": 11.883132530120482, "grad_norm": 5.42984676361084, "learning_rate": 9.404854368932039e-05, "loss": 0.2546, "step": 1226 }, { "epoch": 11.89277108433735, "grad_norm": 12.85007381439209, "learning_rate": 9.404368932038835e-05, "loss": 0.5269, "step": 1227 }, { "epoch": 11.902409638554216, "grad_norm": 12.995246887207031, "learning_rate": 9.40388349514563e-05, "loss": 0.3748, "step": 1228 }, { "epoch": 11.912048192771085, "grad_norm": 6.245255947113037, "learning_rate": 9.403398058252428e-05, "loss": 0.5087, "step": 1229 }, { "epoch": 11.921686746987952, "grad_norm": 7.709681034088135, "learning_rate": 9.402912621359223e-05, "loss": 0.2668, "step": 1230 }, { "epoch": 11.931325301204819, "grad_norm": 10.259217262268066, "learning_rate": 9.402427184466019e-05, "loss": 0.4811, "step": 1231 }, { "epoch": 11.940963855421687, "grad_norm": 2.7288479804992676, "learning_rate": 9.401941747572815e-05, "loss": 0.2676, "step": 1232 }, { "epoch": 11.950602409638554, "grad_norm": 3.601778268814087, "learning_rate": 9.401456310679612e-05, "loss": 0.5205, "step": 1233 }, { "epoch": 11.960240963855421, "grad_norm": 17.75416374206543, "learning_rate": 9.400970873786408e-05, "loss": 0.442, "step": 1234 }, { "epoch": 11.96987951807229, "grad_norm": 3.9860520362854004, "learning_rate": 9.400485436893204e-05, "loss": 0.5629, "step": 1235 }, { "epoch": 11.979518072289157, "grad_norm": 10.688682556152344, "learning_rate": 9.4e-05, "loss": 0.7883, "step": 1236 }, { "epoch": 11.989156626506023, "grad_norm": 18.189403533935547, "learning_rate": 9.399514563106797e-05, "loss": 0.3352, "step": 1237 }, { "epoch": 12.004819277108433, "grad_norm": 6.055888652801514, "learning_rate": 9.399029126213594e-05, "loss": 0.3866, "step": 1238 }, { "epoch": 12.0144578313253, "grad_norm": 11.498871803283691, "learning_rate": 9.39854368932039e-05, "loss": 0.2459, "step": 1239 }, { "epoch": 12.024096385542169, "grad_norm": 3.1776044368743896, "learning_rate": 9.398058252427185e-05, "loss": 0.3669, "step": 1240 }, { "epoch": 12.033734939759036, "grad_norm": 4.01508092880249, "learning_rate": 9.397572815533981e-05, "loss": 0.2337, "step": 1241 }, { "epoch": 12.043373493975903, "grad_norm": 3.6897339820861816, "learning_rate": 9.397087378640778e-05, "loss": 0.2055, "step": 1242 }, { "epoch": 12.053012048192771, "grad_norm": 10.761018753051758, "learning_rate": 9.396601941747574e-05, "loss": 0.5527, "step": 1243 }, { "epoch": 12.062650602409638, "grad_norm": 5.749185562133789, "learning_rate": 9.39611650485437e-05, "loss": 0.399, "step": 1244 }, { "epoch": 12.072289156626505, "grad_norm": 8.599132537841797, "learning_rate": 9.395631067961165e-05, "loss": 0.3118, "step": 1245 }, { "epoch": 12.081927710843374, "grad_norm": 6.480676651000977, "learning_rate": 9.395145631067963e-05, "loss": 0.2758, "step": 1246 }, { "epoch": 12.09156626506024, "grad_norm": 13.757299423217773, "learning_rate": 9.394660194174758e-05, "loss": 0.3685, "step": 1247 }, { "epoch": 12.101204819277108, "grad_norm": 12.73832893371582, "learning_rate": 9.394174757281554e-05, "loss": 0.2446, "step": 1248 }, { "epoch": 12.110843373493976, "grad_norm": 7.840247631072998, "learning_rate": 9.39368932038835e-05, "loss": 0.5155, "step": 1249 }, { "epoch": 12.120481927710843, "grad_norm": 11.994718551635742, "learning_rate": 9.393203883495146e-05, "loss": 0.2431, "step": 1250 }, { "epoch": 12.13012048192771, "grad_norm": 11.541242599487305, "learning_rate": 9.392718446601943e-05, "loss": 0.4362, "step": 1251 }, { "epoch": 12.139759036144579, "grad_norm": 16.92323112487793, "learning_rate": 9.392233009708739e-05, "loss": 0.2876, "step": 1252 }, { "epoch": 12.149397590361446, "grad_norm": 6.078758716583252, "learning_rate": 9.391747572815534e-05, "loss": 0.4586, "step": 1253 }, { "epoch": 12.159036144578312, "grad_norm": 9.29987907409668, "learning_rate": 9.39126213592233e-05, "loss": 0.415, "step": 1254 }, { "epoch": 12.168674698795181, "grad_norm": 3.7802743911743164, "learning_rate": 9.390776699029127e-05, "loss": 0.3259, "step": 1255 }, { "epoch": 12.178313253012048, "grad_norm": 5.366605281829834, "learning_rate": 9.390291262135923e-05, "loss": 0.3653, "step": 1256 }, { "epoch": 12.187951807228915, "grad_norm": 4.1840901374816895, "learning_rate": 9.389805825242719e-05, "loss": 0.3563, "step": 1257 }, { "epoch": 12.197590361445783, "grad_norm": 4.649291515350342, "learning_rate": 9.389320388349515e-05, "loss": 0.5523, "step": 1258 }, { "epoch": 12.20722891566265, "grad_norm": 5.100485801696777, "learning_rate": 9.388834951456312e-05, "loss": 0.3571, "step": 1259 }, { "epoch": 12.216867469879517, "grad_norm": 14.123041152954102, "learning_rate": 9.388349514563108e-05, "loss": 0.4685, "step": 1260 }, { "epoch": 12.226506024096386, "grad_norm": 10.449247360229492, "learning_rate": 9.387864077669903e-05, "loss": 0.5813, "step": 1261 }, { "epoch": 12.236144578313253, "grad_norm": 3.1410346031188965, "learning_rate": 9.387378640776699e-05, "loss": 0.2836, "step": 1262 }, { "epoch": 12.24578313253012, "grad_norm": 14.193865776062012, "learning_rate": 9.386893203883496e-05, "loss": 0.5055, "step": 1263 }, { "epoch": 12.255421686746988, "grad_norm": 17.66584014892578, "learning_rate": 9.386407766990292e-05, "loss": 0.3433, "step": 1264 }, { "epoch": 12.265060240963855, "grad_norm": 8.754327774047852, "learning_rate": 9.385922330097088e-05, "loss": 0.3673, "step": 1265 }, { "epoch": 12.274698795180722, "grad_norm": 3.1368815898895264, "learning_rate": 9.385436893203884e-05, "loss": 0.279, "step": 1266 }, { "epoch": 12.28433734939759, "grad_norm": 5.792121410369873, "learning_rate": 9.38495145631068e-05, "loss": 0.3978, "step": 1267 }, { "epoch": 12.293975903614458, "grad_norm": 6.817379951477051, "learning_rate": 9.384466019417476e-05, "loss": 0.3857, "step": 1268 }, { "epoch": 12.303614457831324, "grad_norm": 5.009941101074219, "learning_rate": 9.383980582524272e-05, "loss": 0.3267, "step": 1269 }, { "epoch": 12.313253012048193, "grad_norm": 12.48479175567627, "learning_rate": 9.383495145631068e-05, "loss": 0.5132, "step": 1270 }, { "epoch": 12.32289156626506, "grad_norm": 2.3220138549804688, "learning_rate": 9.383009708737864e-05, "loss": 0.1851, "step": 1271 }, { "epoch": 12.332530120481927, "grad_norm": 5.546024322509766, "learning_rate": 9.382524271844661e-05, "loss": 0.5557, "step": 1272 }, { "epoch": 12.342168674698796, "grad_norm": 6.9567108154296875, "learning_rate": 9.382038834951457e-05, "loss": 0.4775, "step": 1273 }, { "epoch": 12.351807228915662, "grad_norm": 10.255085945129395, "learning_rate": 9.381553398058252e-05, "loss": 0.227, "step": 1274 }, { "epoch": 12.36144578313253, "grad_norm": 3.18597412109375, "learning_rate": 9.381067961165048e-05, "loss": 0.2491, "step": 1275 }, { "epoch": 12.371084337349398, "grad_norm": 11.86898136138916, "learning_rate": 9.380582524271845e-05, "loss": 0.4998, "step": 1276 }, { "epoch": 12.380722891566265, "grad_norm": 15.504060745239258, "learning_rate": 9.380097087378641e-05, "loss": 0.4358, "step": 1277 }, { "epoch": 12.390361445783132, "grad_norm": 19.54547119140625, "learning_rate": 9.379611650485437e-05, "loss": 0.1729, "step": 1278 }, { "epoch": 12.4, "grad_norm": 10.353401184082031, "learning_rate": 9.379126213592233e-05, "loss": 0.3478, "step": 1279 }, { "epoch": 12.409638554216867, "grad_norm": 11.437660217285156, "learning_rate": 9.37864077669903e-05, "loss": 0.683, "step": 1280 }, { "epoch": 12.419277108433734, "grad_norm": 4.321442127227783, "learning_rate": 9.378155339805826e-05, "loss": 0.2964, "step": 1281 }, { "epoch": 12.428915662650603, "grad_norm": 22.573335647583008, "learning_rate": 9.377669902912621e-05, "loss": 0.4616, "step": 1282 }, { "epoch": 12.43855421686747, "grad_norm": 28.008987426757812, "learning_rate": 9.377184466019417e-05, "loss": 0.4821, "step": 1283 }, { "epoch": 12.448192771084337, "grad_norm": 2.8142833709716797, "learning_rate": 9.376699029126213e-05, "loss": 0.3133, "step": 1284 }, { "epoch": 12.457831325301205, "grad_norm": 16.032344818115234, "learning_rate": 9.37621359223301e-05, "loss": 0.2278, "step": 1285 }, { "epoch": 12.467469879518072, "grad_norm": 5.009344577789307, "learning_rate": 9.375728155339806e-05, "loss": 0.1955, "step": 1286 }, { "epoch": 12.477108433734939, "grad_norm": 4.209919452667236, "learning_rate": 9.375242718446602e-05, "loss": 0.2844, "step": 1287 }, { "epoch": 12.486746987951808, "grad_norm": 4.806680679321289, "learning_rate": 9.374757281553399e-05, "loss": 0.4283, "step": 1288 }, { "epoch": 12.496385542168674, "grad_norm": 14.713456153869629, "learning_rate": 9.374271844660195e-05, "loss": 0.5711, "step": 1289 }, { "epoch": 12.506024096385541, "grad_norm": 6.662930965423584, "learning_rate": 9.373786407766992e-05, "loss": 0.3367, "step": 1290 }, { "epoch": 12.51566265060241, "grad_norm": 16.605873107910156, "learning_rate": 9.373300970873787e-05, "loss": 0.1899, "step": 1291 }, { "epoch": 12.525301204819277, "grad_norm": 6.677784442901611, "learning_rate": 9.372815533980583e-05, "loss": 0.5, "step": 1292 }, { "epoch": 12.534939759036144, "grad_norm": 4.99163293838501, "learning_rate": 9.372330097087379e-05, "loss": 0.3603, "step": 1293 }, { "epoch": 12.544578313253012, "grad_norm": 3.9569618701934814, "learning_rate": 9.371844660194176e-05, "loss": 0.3421, "step": 1294 }, { "epoch": 12.55421686746988, "grad_norm": 3.3518171310424805, "learning_rate": 9.371359223300972e-05, "loss": 0.2563, "step": 1295 }, { "epoch": 12.563855421686746, "grad_norm": 5.294371604919434, "learning_rate": 9.370873786407768e-05, "loss": 0.2859, "step": 1296 }, { "epoch": 12.573493975903615, "grad_norm": 6.049743175506592, "learning_rate": 9.370388349514563e-05, "loss": 0.2401, "step": 1297 }, { "epoch": 12.583132530120482, "grad_norm": 12.655534744262695, "learning_rate": 9.36990291262136e-05, "loss": 0.4711, "step": 1298 }, { "epoch": 12.592771084337349, "grad_norm": 3.591644763946533, "learning_rate": 9.369417475728156e-05, "loss": 0.3001, "step": 1299 }, { "epoch": 12.602409638554217, "grad_norm": 4.088390827178955, "learning_rate": 9.368932038834952e-05, "loss": 0.3755, "step": 1300 }, { "epoch": 12.612048192771084, "grad_norm": 4.803507328033447, "learning_rate": 9.368446601941748e-05, "loss": 0.4221, "step": 1301 }, { "epoch": 12.621686746987951, "grad_norm": 3.8316805362701416, "learning_rate": 9.367961165048545e-05, "loss": 0.4068, "step": 1302 }, { "epoch": 12.63132530120482, "grad_norm": 7.29962158203125, "learning_rate": 9.367475728155341e-05, "loss": 0.1637, "step": 1303 }, { "epoch": 12.640963855421687, "grad_norm": 5.984528064727783, "learning_rate": 9.366990291262137e-05, "loss": 0.2916, "step": 1304 }, { "epoch": 12.650602409638553, "grad_norm": 13.173746109008789, "learning_rate": 9.366504854368932e-05, "loss": 0.3988, "step": 1305 }, { "epoch": 12.660240963855422, "grad_norm": 7.061579704284668, "learning_rate": 9.366019417475728e-05, "loss": 0.1799, "step": 1306 }, { "epoch": 12.669879518072289, "grad_norm": 5.33928918838501, "learning_rate": 9.365533980582525e-05, "loss": 0.1672, "step": 1307 }, { "epoch": 12.679518072289156, "grad_norm": 6.33458948135376, "learning_rate": 9.365048543689321e-05, "loss": 0.409, "step": 1308 }, { "epoch": 12.689156626506024, "grad_norm": 7.878558158874512, "learning_rate": 9.364563106796117e-05, "loss": 0.2954, "step": 1309 }, { "epoch": 12.698795180722891, "grad_norm": 5.737586498260498, "learning_rate": 9.364077669902913e-05, "loss": 0.4048, "step": 1310 }, { "epoch": 12.708433734939758, "grad_norm": 3.413635015487671, "learning_rate": 9.36359223300971e-05, "loss": 0.3911, "step": 1311 }, { "epoch": 12.718072289156627, "grad_norm": 6.97313928604126, "learning_rate": 9.363106796116506e-05, "loss": 0.5424, "step": 1312 }, { "epoch": 12.727710843373494, "grad_norm": 3.046328067779541, "learning_rate": 9.362621359223301e-05, "loss": 0.3033, "step": 1313 }, { "epoch": 12.73734939759036, "grad_norm": 3.5816142559051514, "learning_rate": 9.362135922330097e-05, "loss": 0.2241, "step": 1314 }, { "epoch": 12.74698795180723, "grad_norm": 7.407858371734619, "learning_rate": 9.361650485436894e-05, "loss": 0.2572, "step": 1315 }, { "epoch": 12.756626506024096, "grad_norm": 2.3006644248962402, "learning_rate": 9.36116504854369e-05, "loss": 0.2575, "step": 1316 }, { "epoch": 12.766265060240963, "grad_norm": 19.019216537475586, "learning_rate": 9.360679611650486e-05, "loss": 0.2445, "step": 1317 }, { "epoch": 12.775903614457832, "grad_norm": 4.603786468505859, "learning_rate": 9.360194174757282e-05, "loss": 0.1624, "step": 1318 }, { "epoch": 12.785542168674699, "grad_norm": 7.780540466308594, "learning_rate": 9.359708737864079e-05, "loss": 0.5647, "step": 1319 }, { "epoch": 12.795180722891565, "grad_norm": 11.925387382507324, "learning_rate": 9.359223300970874e-05, "loss": 0.2732, "step": 1320 }, { "epoch": 12.804819277108434, "grad_norm": 4.6357421875, "learning_rate": 9.35873786407767e-05, "loss": 0.4015, "step": 1321 }, { "epoch": 12.814457831325301, "grad_norm": 20.222457885742188, "learning_rate": 9.358252427184466e-05, "loss": 0.3999, "step": 1322 }, { "epoch": 12.824096385542168, "grad_norm": 17.18768882751465, "learning_rate": 9.357766990291262e-05, "loss": 0.2449, "step": 1323 }, { "epoch": 12.833734939759037, "grad_norm": 13.713384628295898, "learning_rate": 9.357281553398059e-05, "loss": 0.4132, "step": 1324 }, { "epoch": 12.843373493975903, "grad_norm": 7.580006122589111, "learning_rate": 9.356796116504855e-05, "loss": 0.4442, "step": 1325 }, { "epoch": 12.85301204819277, "grad_norm": 4.768728733062744, "learning_rate": 9.35631067961165e-05, "loss": 0.3076, "step": 1326 }, { "epoch": 12.862650602409639, "grad_norm": 3.582719564437866, "learning_rate": 9.355825242718446e-05, "loss": 0.2873, "step": 1327 }, { "epoch": 12.872289156626506, "grad_norm": 7.023336887359619, "learning_rate": 9.355339805825243e-05, "loss": 0.4314, "step": 1328 }, { "epoch": 12.881927710843373, "grad_norm": 15.198623657226562, "learning_rate": 9.354854368932039e-05, "loss": 0.4674, "step": 1329 }, { "epoch": 12.891566265060241, "grad_norm": 13.564393043518066, "learning_rate": 9.354368932038835e-05, "loss": 0.4736, "step": 1330 }, { "epoch": 12.901204819277108, "grad_norm": 9.205514907836914, "learning_rate": 9.35388349514563e-05, "loss": 0.4623, "step": 1331 }, { "epoch": 12.910843373493975, "grad_norm": 7.120309829711914, "learning_rate": 9.353398058252428e-05, "loss": 0.3418, "step": 1332 }, { "epoch": 12.920481927710844, "grad_norm": 8.508384704589844, "learning_rate": 9.352912621359224e-05, "loss": 0.2143, "step": 1333 }, { "epoch": 12.93012048192771, "grad_norm": 6.850863456726074, "learning_rate": 9.35242718446602e-05, "loss": 0.4166, "step": 1334 }, { "epoch": 12.939759036144578, "grad_norm": 2.934960126876831, "learning_rate": 9.351941747572815e-05, "loss": 0.356, "step": 1335 }, { "epoch": 12.949397590361446, "grad_norm": 28.075666427612305, "learning_rate": 9.351456310679612e-05, "loss": 0.3904, "step": 1336 }, { "epoch": 12.959036144578313, "grad_norm": 16.521873474121094, "learning_rate": 9.350970873786408e-05, "loss": 0.3854, "step": 1337 }, { "epoch": 12.96867469879518, "grad_norm": 5.7674360275268555, "learning_rate": 9.350485436893205e-05, "loss": 0.3194, "step": 1338 }, { "epoch": 12.978313253012049, "grad_norm": 5.653547763824463, "learning_rate": 9.350000000000001e-05, "loss": 0.249, "step": 1339 }, { "epoch": 12.987951807228916, "grad_norm": 9.987544059753418, "learning_rate": 9.349514563106797e-05, "loss": 0.2508, "step": 1340 }, { "epoch": 13.003614457831326, "grad_norm": 6.048887729644775, "learning_rate": 9.349029126213592e-05, "loss": 0.4121, "step": 1341 }, { "epoch": 13.013253012048192, "grad_norm": 8.367447853088379, "learning_rate": 9.34854368932039e-05, "loss": 0.4616, "step": 1342 }, { "epoch": 13.022891566265061, "grad_norm": 17.97467803955078, "learning_rate": 9.348058252427185e-05, "loss": 0.6773, "step": 1343 }, { "epoch": 13.032530120481928, "grad_norm": 27.08138656616211, "learning_rate": 9.347572815533981e-05, "loss": 0.3343, "step": 1344 }, { "epoch": 13.042168674698795, "grad_norm": 18.754384994506836, "learning_rate": 9.347087378640777e-05, "loss": 0.2516, "step": 1345 }, { "epoch": 13.051807228915663, "grad_norm": 14.373141288757324, "learning_rate": 9.346601941747574e-05, "loss": 0.5075, "step": 1346 }, { "epoch": 13.06144578313253, "grad_norm": 4.881836414337158, "learning_rate": 9.34611650485437e-05, "loss": 0.3265, "step": 1347 }, { "epoch": 13.071084337349397, "grad_norm": 8.869173049926758, "learning_rate": 9.345631067961166e-05, "loss": 0.2905, "step": 1348 }, { "epoch": 13.080722891566266, "grad_norm": 4.058520793914795, "learning_rate": 9.345145631067961e-05, "loss": 0.2995, "step": 1349 }, { "epoch": 13.090361445783133, "grad_norm": 8.626431465148926, "learning_rate": 9.344660194174759e-05, "loss": 0.4207, "step": 1350 }, { "epoch": 13.1, "grad_norm": 4.392086982727051, "learning_rate": 9.344174757281554e-05, "loss": 0.5172, "step": 1351 }, { "epoch": 13.109638554216868, "grad_norm": 10.932074546813965, "learning_rate": 9.34368932038835e-05, "loss": 0.3865, "step": 1352 }, { "epoch": 13.119277108433735, "grad_norm": 17.936275482177734, "learning_rate": 9.343203883495146e-05, "loss": 0.4398, "step": 1353 }, { "epoch": 13.128915662650602, "grad_norm": 5.242444038391113, "learning_rate": 9.342718446601943e-05, "loss": 0.5044, "step": 1354 }, { "epoch": 13.13855421686747, "grad_norm": 9.249160766601562, "learning_rate": 9.342233009708739e-05, "loss": 0.3726, "step": 1355 }, { "epoch": 13.148192771084338, "grad_norm": 37.833309173583984, "learning_rate": 9.341747572815535e-05, "loss": 0.3527, "step": 1356 }, { "epoch": 13.157831325301204, "grad_norm": 13.841446876525879, "learning_rate": 9.34126213592233e-05, "loss": 0.4976, "step": 1357 }, { "epoch": 13.167469879518073, "grad_norm": 18.838228225708008, "learning_rate": 9.340776699029126e-05, "loss": 0.4684, "step": 1358 }, { "epoch": 13.17710843373494, "grad_norm": 11.048312187194824, "learning_rate": 9.340291262135923e-05, "loss": 0.2891, "step": 1359 }, { "epoch": 13.186746987951807, "grad_norm": 13.289352416992188, "learning_rate": 9.339805825242719e-05, "loss": 0.4821, "step": 1360 }, { "epoch": 13.196385542168676, "grad_norm": 4.33757209777832, "learning_rate": 9.339320388349515e-05, "loss": 0.1242, "step": 1361 }, { "epoch": 13.206024096385542, "grad_norm": 6.900155544281006, "learning_rate": 9.33883495145631e-05, "loss": 0.5455, "step": 1362 }, { "epoch": 13.21566265060241, "grad_norm": 3.9489893913269043, "learning_rate": 9.338349514563108e-05, "loss": 0.1994, "step": 1363 }, { "epoch": 13.225301204819278, "grad_norm": 6.5552778244018555, "learning_rate": 9.337864077669903e-05, "loss": 0.5124, "step": 1364 }, { "epoch": 13.234939759036145, "grad_norm": 10.840511322021484, "learning_rate": 9.337378640776699e-05, "loss": 0.5258, "step": 1365 }, { "epoch": 13.244578313253012, "grad_norm": 6.140920639038086, "learning_rate": 9.336893203883495e-05, "loss": 0.4899, "step": 1366 }, { "epoch": 13.25421686746988, "grad_norm": 4.573484420776367, "learning_rate": 9.336407766990292e-05, "loss": 0.4087, "step": 1367 }, { "epoch": 13.263855421686747, "grad_norm": 6.036597728729248, "learning_rate": 9.335922330097088e-05, "loss": 0.6238, "step": 1368 }, { "epoch": 13.273493975903614, "grad_norm": 6.305534362792969, "learning_rate": 9.335436893203884e-05, "loss": 0.3513, "step": 1369 }, { "epoch": 13.283132530120483, "grad_norm": 5.984772682189941, "learning_rate": 9.33495145631068e-05, "loss": 0.3474, "step": 1370 }, { "epoch": 13.29277108433735, "grad_norm": 12.28515911102295, "learning_rate": 9.334466019417477e-05, "loss": 0.4083, "step": 1371 }, { "epoch": 13.302409638554217, "grad_norm": 13.03615665435791, "learning_rate": 9.333980582524272e-05, "loss": 0.397, "step": 1372 }, { "epoch": 13.312048192771085, "grad_norm": 13.702859878540039, "learning_rate": 9.333495145631068e-05, "loss": 0.5026, "step": 1373 }, { "epoch": 13.321686746987952, "grad_norm": 4.798179626464844, "learning_rate": 9.333009708737864e-05, "loss": 0.2355, "step": 1374 }, { "epoch": 13.331325301204819, "grad_norm": 2.628031015396118, "learning_rate": 9.33252427184466e-05, "loss": 0.1906, "step": 1375 }, { "epoch": 13.340963855421688, "grad_norm": 20.4335880279541, "learning_rate": 9.332038834951457e-05, "loss": 0.5745, "step": 1376 }, { "epoch": 13.350602409638554, "grad_norm": 7.476676940917969, "learning_rate": 9.331553398058253e-05, "loss": 0.5134, "step": 1377 }, { "epoch": 13.360240963855421, "grad_norm": 19.246095657348633, "learning_rate": 9.331067961165048e-05, "loss": 0.4735, "step": 1378 }, { "epoch": 13.369879518072288, "grad_norm": 3.221208333969116, "learning_rate": 9.330582524271844e-05, "loss": 0.4662, "step": 1379 }, { "epoch": 13.379518072289157, "grad_norm": 6.332733154296875, "learning_rate": 9.330097087378641e-05, "loss": 0.5602, "step": 1380 }, { "epoch": 13.389156626506024, "grad_norm": 4.695082187652588, "learning_rate": 9.329611650485437e-05, "loss": 0.2409, "step": 1381 }, { "epoch": 13.398795180722892, "grad_norm": 2.477562665939331, "learning_rate": 9.329126213592233e-05, "loss": 0.2323, "step": 1382 }, { "epoch": 13.40843373493976, "grad_norm": 6.945754051208496, "learning_rate": 9.328640776699029e-05, "loss": 0.4056, "step": 1383 }, { "epoch": 13.418072289156626, "grad_norm": 14.457667350769043, "learning_rate": 9.328155339805826e-05, "loss": 0.3605, "step": 1384 }, { "epoch": 13.427710843373493, "grad_norm": 6.655060291290283, "learning_rate": 9.327669902912622e-05, "loss": 0.3901, "step": 1385 }, { "epoch": 13.437349397590362, "grad_norm": 5.354176998138428, "learning_rate": 9.327184466019417e-05, "loss": 0.4186, "step": 1386 }, { "epoch": 13.446987951807229, "grad_norm": 6.942165374755859, "learning_rate": 9.326699029126213e-05, "loss": 0.5545, "step": 1387 }, { "epoch": 13.456626506024097, "grad_norm": 11.345715522766113, "learning_rate": 9.32621359223301e-05, "loss": 0.3009, "step": 1388 }, { "epoch": 13.466265060240964, "grad_norm": 2.7492597103118896, "learning_rate": 9.325728155339807e-05, "loss": 0.1275, "step": 1389 }, { "epoch": 13.475903614457831, "grad_norm": 4.749251842498779, "learning_rate": 9.325242718446603e-05, "loss": 0.3526, "step": 1390 }, { "epoch": 13.485542168674698, "grad_norm": 13.447875022888184, "learning_rate": 9.324757281553399e-05, "loss": 0.3634, "step": 1391 }, { "epoch": 13.495180722891567, "grad_norm": 3.112013816833496, "learning_rate": 9.324271844660195e-05, "loss": 0.2163, "step": 1392 }, { "epoch": 13.504819277108433, "grad_norm": 8.050004959106445, "learning_rate": 9.323786407766992e-05, "loss": 0.4098, "step": 1393 }, { "epoch": 13.514457831325302, "grad_norm": 9.465788841247559, "learning_rate": 9.323300970873788e-05, "loss": 0.3827, "step": 1394 }, { "epoch": 13.524096385542169, "grad_norm": 11.553147315979004, "learning_rate": 9.322815533980583e-05, "loss": 0.3847, "step": 1395 }, { "epoch": 13.533734939759036, "grad_norm": 4.370859146118164, "learning_rate": 9.322330097087379e-05, "loss": 0.3105, "step": 1396 }, { "epoch": 13.543373493975903, "grad_norm": 13.98019790649414, "learning_rate": 9.321844660194175e-05, "loss": 0.4675, "step": 1397 }, { "epoch": 13.553012048192771, "grad_norm": 5.303406238555908, "learning_rate": 9.321359223300972e-05, "loss": 0.3601, "step": 1398 }, { "epoch": 13.562650602409638, "grad_norm": 4.781496524810791, "learning_rate": 9.320873786407768e-05, "loss": 0.2688, "step": 1399 }, { "epoch": 13.572289156626507, "grad_norm": 4.2262420654296875, "learning_rate": 9.320388349514564e-05, "loss": 0.3324, "step": 1400 }, { "epoch": 13.581927710843374, "grad_norm": 5.182602882385254, "learning_rate": 9.31990291262136e-05, "loss": 0.478, "step": 1401 }, { "epoch": 13.59156626506024, "grad_norm": 3.948228359222412, "learning_rate": 9.319417475728156e-05, "loss": 0.4144, "step": 1402 }, { "epoch": 13.601204819277108, "grad_norm": 40.303199768066406, "learning_rate": 9.318932038834952e-05, "loss": 0.4029, "step": 1403 }, { "epoch": 13.610843373493976, "grad_norm": 14.465543746948242, "learning_rate": 9.318446601941748e-05, "loss": 0.2245, "step": 1404 }, { "epoch": 13.620481927710843, "grad_norm": 4.108410358428955, "learning_rate": 9.317961165048544e-05, "loss": 0.311, "step": 1405 }, { "epoch": 13.630120481927712, "grad_norm": 11.583922386169434, "learning_rate": 9.317475728155341e-05, "loss": 0.2993, "step": 1406 }, { "epoch": 13.639759036144579, "grad_norm": 7.13770866394043, "learning_rate": 9.316990291262137e-05, "loss": 0.5647, "step": 1407 }, { "epoch": 13.649397590361446, "grad_norm": 16.396657943725586, "learning_rate": 9.316504854368933e-05, "loss": 0.2948, "step": 1408 }, { "epoch": 13.659036144578312, "grad_norm": 9.484367370605469, "learning_rate": 9.316019417475728e-05, "loss": 0.4006, "step": 1409 }, { "epoch": 13.668674698795181, "grad_norm": 8.152812957763672, "learning_rate": 9.315533980582525e-05, "loss": 0.5379, "step": 1410 }, { "epoch": 13.678313253012048, "grad_norm": 3.232314109802246, "learning_rate": 9.315048543689321e-05, "loss": 0.3414, "step": 1411 }, { "epoch": 13.687951807228917, "grad_norm": 2.942490339279175, "learning_rate": 9.314563106796117e-05, "loss": 0.3155, "step": 1412 }, { "epoch": 13.697590361445783, "grad_norm": 5.828101634979248, "learning_rate": 9.314077669902913e-05, "loss": 0.3858, "step": 1413 }, { "epoch": 13.70722891566265, "grad_norm": 3.7448432445526123, "learning_rate": 9.313592233009709e-05, "loss": 0.3452, "step": 1414 }, { "epoch": 13.716867469879517, "grad_norm": 4.372006893157959, "learning_rate": 9.313106796116506e-05, "loss": 0.2489, "step": 1415 }, { "epoch": 13.726506024096386, "grad_norm": 12.131375312805176, "learning_rate": 9.312621359223301e-05, "loss": 0.3128, "step": 1416 }, { "epoch": 13.736144578313253, "grad_norm": 11.173239707946777, "learning_rate": 9.312135922330097e-05, "loss": 0.333, "step": 1417 }, { "epoch": 13.745783132530121, "grad_norm": 12.094778060913086, "learning_rate": 9.311650485436893e-05, "loss": 0.3737, "step": 1418 }, { "epoch": 13.755421686746988, "grad_norm": 5.849565505981445, "learning_rate": 9.31116504854369e-05, "loss": 0.4329, "step": 1419 }, { "epoch": 13.765060240963855, "grad_norm": 19.06221580505371, "learning_rate": 9.310679611650486e-05, "loss": 0.7602, "step": 1420 }, { "epoch": 13.774698795180722, "grad_norm": 13.579615592956543, "learning_rate": 9.310194174757282e-05, "loss": 0.3453, "step": 1421 }, { "epoch": 13.78433734939759, "grad_norm": 8.7617826461792, "learning_rate": 9.309708737864077e-05, "loss": 0.3217, "step": 1422 }, { "epoch": 13.793975903614458, "grad_norm": 3.421010971069336, "learning_rate": 9.309223300970875e-05, "loss": 0.1925, "step": 1423 }, { "epoch": 13.803614457831326, "grad_norm": 6.122467517852783, "learning_rate": 9.30873786407767e-05, "loss": 0.2729, "step": 1424 }, { "epoch": 13.813253012048193, "grad_norm": 7.956698894500732, "learning_rate": 9.308252427184466e-05, "loss": 0.3731, "step": 1425 }, { "epoch": 13.82289156626506, "grad_norm": 4.389279842376709, "learning_rate": 9.307766990291262e-05, "loss": 0.196, "step": 1426 }, { "epoch": 13.832530120481927, "grad_norm": 11.515440940856934, "learning_rate": 9.307281553398059e-05, "loss": 0.3873, "step": 1427 }, { "epoch": 13.842168674698796, "grad_norm": 10.836444854736328, "learning_rate": 9.306796116504855e-05, "loss": 0.3497, "step": 1428 }, { "epoch": 13.851807228915662, "grad_norm": 7.224135875701904, "learning_rate": 9.30631067961165e-05, "loss": 0.5054, "step": 1429 }, { "epoch": 13.861445783132531, "grad_norm": 3.9216089248657227, "learning_rate": 9.305825242718446e-05, "loss": 0.3596, "step": 1430 }, { "epoch": 13.871084337349398, "grad_norm": 8.132318496704102, "learning_rate": 9.305339805825242e-05, "loss": 0.2719, "step": 1431 }, { "epoch": 13.880722891566265, "grad_norm": 16.069862365722656, "learning_rate": 9.304854368932039e-05, "loss": 0.3476, "step": 1432 }, { "epoch": 13.890361445783132, "grad_norm": 23.25846290588379, "learning_rate": 9.304368932038835e-05, "loss": 0.4707, "step": 1433 }, { "epoch": 13.9, "grad_norm": 5.438247203826904, "learning_rate": 9.303883495145631e-05, "loss": 0.2836, "step": 1434 }, { "epoch": 13.909638554216867, "grad_norm": 31.213356018066406, "learning_rate": 9.303398058252427e-05, "loss": 0.3196, "step": 1435 }, { "epoch": 13.919277108433734, "grad_norm": 34.594032287597656, "learning_rate": 9.302912621359224e-05, "loss": 0.3132, "step": 1436 }, { "epoch": 13.928915662650603, "grad_norm": 3.6323065757751465, "learning_rate": 9.30242718446602e-05, "loss": 0.2766, "step": 1437 }, { "epoch": 13.93855421686747, "grad_norm": 9.407398223876953, "learning_rate": 9.301941747572815e-05, "loss": 0.3246, "step": 1438 }, { "epoch": 13.948192771084337, "grad_norm": 35.27167510986328, "learning_rate": 9.301456310679612e-05, "loss": 0.3909, "step": 1439 }, { "epoch": 13.957831325301205, "grad_norm": 15.35251235961914, "learning_rate": 9.300970873786408e-05, "loss": 0.5112, "step": 1440 }, { "epoch": 13.967469879518072, "grad_norm": 16.826231002807617, "learning_rate": 9.300485436893205e-05, "loss": 0.3298, "step": 1441 }, { "epoch": 13.977108433734939, "grad_norm": 11.856441497802734, "learning_rate": 9.300000000000001e-05, "loss": 0.4733, "step": 1442 }, { "epoch": 13.986746987951808, "grad_norm": 11.250923156738281, "learning_rate": 9.299514563106797e-05, "loss": 0.5165, "step": 1443 }, { "epoch": 14.002409638554218, "grad_norm": 4.953524112701416, "learning_rate": 9.299029126213593e-05, "loss": 0.2081, "step": 1444 }, { "epoch": 14.012048192771084, "grad_norm": 5.832924842834473, "learning_rate": 9.29854368932039e-05, "loss": 0.3137, "step": 1445 }, { "epoch": 14.021686746987951, "grad_norm": 8.647130012512207, "learning_rate": 9.298058252427186e-05, "loss": 0.3816, "step": 1446 }, { "epoch": 14.03132530120482, "grad_norm": 41.06440734863281, "learning_rate": 9.297572815533981e-05, "loss": 0.7469, "step": 1447 }, { "epoch": 14.040963855421687, "grad_norm": 2.9899661540985107, "learning_rate": 9.297087378640777e-05, "loss": 0.2239, "step": 1448 }, { "epoch": 14.050602409638554, "grad_norm": 10.054325103759766, "learning_rate": 9.296601941747574e-05, "loss": 0.4759, "step": 1449 }, { "epoch": 14.060240963855422, "grad_norm": 7.502145290374756, "learning_rate": 9.29611650485437e-05, "loss": 0.4491, "step": 1450 }, { "epoch": 14.06987951807229, "grad_norm": 6.882865905761719, "learning_rate": 9.295631067961166e-05, "loss": 0.6423, "step": 1451 }, { "epoch": 14.079518072289156, "grad_norm": 7.0222883224487305, "learning_rate": 9.295145631067962e-05, "loss": 0.5843, "step": 1452 }, { "epoch": 14.089156626506025, "grad_norm": 10.987711906433105, "learning_rate": 9.294660194174757e-05, "loss": 0.2739, "step": 1453 }, { "epoch": 14.098795180722892, "grad_norm": 5.469206809997559, "learning_rate": 9.294174757281554e-05, "loss": 0.412, "step": 1454 }, { "epoch": 14.108433734939759, "grad_norm": 14.10290813446045, "learning_rate": 9.29368932038835e-05, "loss": 0.294, "step": 1455 }, { "epoch": 14.118072289156627, "grad_norm": 6.875463962554932, "learning_rate": 9.293203883495146e-05, "loss": 0.4112, "step": 1456 }, { "epoch": 14.127710843373494, "grad_norm": 5.343501091003418, "learning_rate": 9.292718446601942e-05, "loss": 0.3854, "step": 1457 }, { "epoch": 14.137349397590361, "grad_norm": 6.405155658721924, "learning_rate": 9.292233009708739e-05, "loss": 0.4877, "step": 1458 }, { "epoch": 14.14698795180723, "grad_norm": 4.308139324188232, "learning_rate": 9.291747572815535e-05, "loss": 0.2068, "step": 1459 }, { "epoch": 14.156626506024097, "grad_norm": 5.214849948883057, "learning_rate": 9.29126213592233e-05, "loss": 0.2615, "step": 1460 }, { "epoch": 14.166265060240963, "grad_norm": 19.297378540039062, "learning_rate": 9.290776699029126e-05, "loss": 0.5582, "step": 1461 }, { "epoch": 14.175903614457832, "grad_norm": 31.311044692993164, "learning_rate": 9.290291262135923e-05, "loss": 0.1109, "step": 1462 }, { "epoch": 14.185542168674699, "grad_norm": 10.226386070251465, "learning_rate": 9.289805825242719e-05, "loss": 0.3741, "step": 1463 }, { "epoch": 14.195180722891566, "grad_norm": 2.8055925369262695, "learning_rate": 9.289320388349515e-05, "loss": 0.2902, "step": 1464 }, { "epoch": 14.204819277108435, "grad_norm": 4.82480001449585, "learning_rate": 9.288834951456311e-05, "loss": 0.3202, "step": 1465 }, { "epoch": 14.214457831325301, "grad_norm": 7.632574081420898, "learning_rate": 9.288349514563108e-05, "loss": 0.5308, "step": 1466 }, { "epoch": 14.224096385542168, "grad_norm": 4.393543720245361, "learning_rate": 9.287864077669904e-05, "loss": 0.1752, "step": 1467 }, { "epoch": 14.233734939759037, "grad_norm": 8.522001266479492, "learning_rate": 9.2873786407767e-05, "loss": 0.2719, "step": 1468 }, { "epoch": 14.243373493975904, "grad_norm": 4.262711524963379, "learning_rate": 9.286893203883495e-05, "loss": 0.3873, "step": 1469 }, { "epoch": 14.25301204819277, "grad_norm": 5.792408466339111, "learning_rate": 9.286407766990291e-05, "loss": 0.4287, "step": 1470 }, { "epoch": 14.26265060240964, "grad_norm": 5.807891368865967, "learning_rate": 9.285922330097088e-05, "loss": 0.3174, "step": 1471 }, { "epoch": 14.272289156626506, "grad_norm": 13.885560035705566, "learning_rate": 9.285436893203884e-05, "loss": 0.2646, "step": 1472 }, { "epoch": 14.281927710843373, "grad_norm": 2.2262959480285645, "learning_rate": 9.28495145631068e-05, "loss": 0.2859, "step": 1473 }, { "epoch": 14.291566265060242, "grad_norm": 6.002269744873047, "learning_rate": 9.284466019417475e-05, "loss": 0.4235, "step": 1474 }, { "epoch": 14.301204819277109, "grad_norm": 5.378168106079102, "learning_rate": 9.283980582524273e-05, "loss": 0.5814, "step": 1475 }, { "epoch": 14.310843373493976, "grad_norm": 4.764664649963379, "learning_rate": 9.283495145631068e-05, "loss": 0.4518, "step": 1476 }, { "epoch": 14.320481927710844, "grad_norm": 3.793348789215088, "learning_rate": 9.283009708737864e-05, "loss": 0.1762, "step": 1477 }, { "epoch": 14.330120481927711, "grad_norm": 8.704983711242676, "learning_rate": 9.28252427184466e-05, "loss": 0.5287, "step": 1478 }, { "epoch": 14.339759036144578, "grad_norm": 8.10614013671875, "learning_rate": 9.282038834951457e-05, "loss": 0.5326, "step": 1479 }, { "epoch": 14.349397590361447, "grad_norm": 7.610810279846191, "learning_rate": 9.281553398058253e-05, "loss": 0.3396, "step": 1480 }, { "epoch": 14.359036144578313, "grad_norm": 6.553378582000732, "learning_rate": 9.281067961165049e-05, "loss": 0.3829, "step": 1481 }, { "epoch": 14.36867469879518, "grad_norm": 10.641654014587402, "learning_rate": 9.280582524271844e-05, "loss": 0.3206, "step": 1482 }, { "epoch": 14.378313253012049, "grad_norm": 15.987743377685547, "learning_rate": 9.280097087378641e-05, "loss": 0.2749, "step": 1483 }, { "epoch": 14.387951807228916, "grad_norm": 5.59743595123291, "learning_rate": 9.279611650485437e-05, "loss": 0.3033, "step": 1484 }, { "epoch": 14.397590361445783, "grad_norm": 17.60783576965332, "learning_rate": 9.279126213592233e-05, "loss": 0.5474, "step": 1485 }, { "epoch": 14.407228915662651, "grad_norm": 18.247018814086914, "learning_rate": 9.278640776699029e-05, "loss": 0.2472, "step": 1486 }, { "epoch": 14.416867469879518, "grad_norm": 12.43435287475586, "learning_rate": 9.278155339805825e-05, "loss": 0.4168, "step": 1487 }, { "epoch": 14.426506024096385, "grad_norm": 11.224918365478516, "learning_rate": 9.277669902912622e-05, "loss": 0.2723, "step": 1488 }, { "epoch": 14.436144578313254, "grad_norm": 5.971980571746826, "learning_rate": 9.277184466019417e-05, "loss": 0.2051, "step": 1489 }, { "epoch": 14.44578313253012, "grad_norm": 8.27527141571045, "learning_rate": 9.276699029126215e-05, "loss": 0.465, "step": 1490 }, { "epoch": 14.455421686746988, "grad_norm": 3.2338342666625977, "learning_rate": 9.27621359223301e-05, "loss": 0.1899, "step": 1491 }, { "epoch": 14.465060240963856, "grad_norm": 5.562441825866699, "learning_rate": 9.275728155339806e-05, "loss": 0.2794, "step": 1492 }, { "epoch": 14.474698795180723, "grad_norm": 6.7119975090026855, "learning_rate": 9.275242718446603e-05, "loss": 0.4451, "step": 1493 }, { "epoch": 14.48433734939759, "grad_norm": 3.6576380729675293, "learning_rate": 9.274757281553399e-05, "loss": 0.1446, "step": 1494 }, { "epoch": 14.493975903614459, "grad_norm": 7.164732456207275, "learning_rate": 9.274271844660195e-05, "loss": 0.2467, "step": 1495 }, { "epoch": 14.503614457831326, "grad_norm": 14.921198844909668, "learning_rate": 9.27378640776699e-05, "loss": 0.2313, "step": 1496 }, { "epoch": 14.513253012048192, "grad_norm": 14.738923072814941, "learning_rate": 9.273300970873788e-05, "loss": 0.4745, "step": 1497 }, { "epoch": 14.522891566265061, "grad_norm": 14.61955451965332, "learning_rate": 9.272815533980583e-05, "loss": 0.3162, "step": 1498 }, { "epoch": 14.532530120481928, "grad_norm": 4.791849136352539, "learning_rate": 9.272330097087379e-05, "loss": 0.4478, "step": 1499 }, { "epoch": 14.542168674698795, "grad_norm": 13.121952056884766, "learning_rate": 9.271844660194175e-05, "loss": 0.5372, "step": 1500 }, { "epoch": 14.551807228915663, "grad_norm": 10.747634887695312, "learning_rate": 9.271359223300972e-05, "loss": 0.5116, "step": 1501 }, { "epoch": 14.56144578313253, "grad_norm": 7.224635601043701, "learning_rate": 9.270873786407768e-05, "loss": 0.3762, "step": 1502 }, { "epoch": 14.571084337349397, "grad_norm": 7.758842468261719, "learning_rate": 9.270388349514564e-05, "loss": 0.4595, "step": 1503 }, { "epoch": 14.580722891566266, "grad_norm": 9.797385215759277, "learning_rate": 9.26990291262136e-05, "loss": 0.5919, "step": 1504 }, { "epoch": 14.590361445783133, "grad_norm": 4.554920673370361, "learning_rate": 9.269417475728157e-05, "loss": 0.2169, "step": 1505 }, { "epoch": 14.6, "grad_norm": 6.7516584396362305, "learning_rate": 9.268932038834952e-05, "loss": 0.3638, "step": 1506 }, { "epoch": 14.609638554216868, "grad_norm": 6.422217845916748, "learning_rate": 9.268446601941748e-05, "loss": 0.4081, "step": 1507 }, { "epoch": 14.619277108433735, "grad_norm": 7.548123836517334, "learning_rate": 9.267961165048544e-05, "loss": 0.3926, "step": 1508 }, { "epoch": 14.628915662650602, "grad_norm": 6.475811958312988, "learning_rate": 9.26747572815534e-05, "loss": 0.5296, "step": 1509 }, { "epoch": 14.638554216867469, "grad_norm": 3.5015716552734375, "learning_rate": 9.266990291262137e-05, "loss": 0.2812, "step": 1510 }, { "epoch": 14.648192771084338, "grad_norm": 5.013360500335693, "learning_rate": 9.266504854368933e-05, "loss": 0.363, "step": 1511 }, { "epoch": 14.657831325301204, "grad_norm": 5.024005889892578, "learning_rate": 9.266019417475728e-05, "loss": 0.3919, "step": 1512 }, { "epoch": 14.667469879518073, "grad_norm": 7.8624067306518555, "learning_rate": 9.265533980582524e-05, "loss": 0.3578, "step": 1513 }, { "epoch": 14.67710843373494, "grad_norm": 5.337090015411377, "learning_rate": 9.265048543689321e-05, "loss": 0.2945, "step": 1514 }, { "epoch": 14.686746987951807, "grad_norm": 4.094537258148193, "learning_rate": 9.264563106796117e-05, "loss": 0.4579, "step": 1515 }, { "epoch": 14.696385542168674, "grad_norm": 4.8127851486206055, "learning_rate": 9.264077669902913e-05, "loss": 0.3292, "step": 1516 }, { "epoch": 14.706024096385542, "grad_norm": 4.934291362762451, "learning_rate": 9.263592233009709e-05, "loss": 0.2629, "step": 1517 }, { "epoch": 14.71566265060241, "grad_norm": 6.6427459716796875, "learning_rate": 9.263106796116506e-05, "loss": 0.2146, "step": 1518 }, { "epoch": 14.725301204819278, "grad_norm": 10.713224411010742, "learning_rate": 9.262621359223302e-05, "loss": 0.5047, "step": 1519 }, { "epoch": 14.734939759036145, "grad_norm": 6.8613200187683105, "learning_rate": 9.262135922330097e-05, "loss": 0.3905, "step": 1520 }, { "epoch": 14.744578313253012, "grad_norm": 11.36563491821289, "learning_rate": 9.261650485436893e-05, "loss": 0.5894, "step": 1521 }, { "epoch": 14.754216867469879, "grad_norm": 8.87269115447998, "learning_rate": 9.26116504854369e-05, "loss": 0.137, "step": 1522 }, { "epoch": 14.763855421686747, "grad_norm": 29.6650447845459, "learning_rate": 9.260679611650486e-05, "loss": 0.3636, "step": 1523 }, { "epoch": 14.773493975903614, "grad_norm": 4.781250476837158, "learning_rate": 9.260194174757282e-05, "loss": 0.2777, "step": 1524 }, { "epoch": 14.783132530120483, "grad_norm": 7.804794788360596, "learning_rate": 9.259708737864078e-05, "loss": 0.2973, "step": 1525 }, { "epoch": 14.79277108433735, "grad_norm": 3.7492728233337402, "learning_rate": 9.259223300970873e-05, "loss": 0.2664, "step": 1526 }, { "epoch": 14.802409638554217, "grad_norm": 5.043986797332764, "learning_rate": 9.25873786407767e-05, "loss": 0.3334, "step": 1527 }, { "epoch": 14.812048192771083, "grad_norm": 6.466732978820801, "learning_rate": 9.258252427184466e-05, "loss": 0.4241, "step": 1528 }, { "epoch": 14.821686746987952, "grad_norm": 7.928110599517822, "learning_rate": 9.257766990291262e-05, "loss": 0.3403, "step": 1529 }, { "epoch": 14.831325301204819, "grad_norm": 5.035149574279785, "learning_rate": 9.257281553398058e-05, "loss": 0.2429, "step": 1530 }, { "epoch": 14.840963855421688, "grad_norm": 4.533519744873047, "learning_rate": 9.256796116504855e-05, "loss": 0.3094, "step": 1531 }, { "epoch": 14.850602409638554, "grad_norm": 5.712342739105225, "learning_rate": 9.256310679611651e-05, "loss": 0.339, "step": 1532 }, { "epoch": 14.860240963855421, "grad_norm": 4.508276462554932, "learning_rate": 9.255825242718446e-05, "loss": 0.2228, "step": 1533 }, { "epoch": 14.869879518072288, "grad_norm": 12.589387893676758, "learning_rate": 9.255339805825242e-05, "loss": 0.3678, "step": 1534 }, { "epoch": 14.879518072289157, "grad_norm": 4.4820876121521, "learning_rate": 9.25485436893204e-05, "loss": 0.4048, "step": 1535 }, { "epoch": 14.889156626506024, "grad_norm": 4.152180194854736, "learning_rate": 9.254368932038835e-05, "loss": 0.349, "step": 1536 }, { "epoch": 14.898795180722892, "grad_norm": 9.8953218460083, "learning_rate": 9.253883495145631e-05, "loss": 0.2752, "step": 1537 }, { "epoch": 14.90843373493976, "grad_norm": 1.9858856201171875, "learning_rate": 9.253398058252427e-05, "loss": 0.2933, "step": 1538 }, { "epoch": 14.918072289156626, "grad_norm": 10.235827445983887, "learning_rate": 9.252912621359224e-05, "loss": 0.2621, "step": 1539 }, { "epoch": 14.927710843373493, "grad_norm": 14.331528663635254, "learning_rate": 9.252427184466021e-05, "loss": 0.3469, "step": 1540 }, { "epoch": 14.937349397590362, "grad_norm": 16.442943572998047, "learning_rate": 9.251941747572817e-05, "loss": 0.8278, "step": 1541 }, { "epoch": 14.946987951807229, "grad_norm": 8.445901870727539, "learning_rate": 9.251456310679613e-05, "loss": 0.3646, "step": 1542 }, { "epoch": 14.956626506024097, "grad_norm": 8.137855529785156, "learning_rate": 9.250970873786408e-05, "loss": 0.446, "step": 1543 }, { "epoch": 14.966265060240964, "grad_norm": 3.9635872840881348, "learning_rate": 9.250485436893205e-05, "loss": 0.4426, "step": 1544 }, { "epoch": 14.975903614457831, "grad_norm": 8.932490348815918, "learning_rate": 9.250000000000001e-05, "loss": 0.5926, "step": 1545 }, { "epoch": 14.985542168674698, "grad_norm": 21.600534439086914, "learning_rate": 9.249514563106797e-05, "loss": 0.6436, "step": 1546 }, { "epoch": 15.001204819277108, "grad_norm": 9.61810302734375, "learning_rate": 9.249029126213593e-05, "loss": 0.4452, "step": 1547 }, { "epoch": 15.010843373493977, "grad_norm": 3.119255781173706, "learning_rate": 9.248543689320389e-05, "loss": 0.281, "step": 1548 }, { "epoch": 15.020481927710843, "grad_norm": 4.614575386047363, "learning_rate": 9.248058252427186e-05, "loss": 0.343, "step": 1549 }, { "epoch": 15.03012048192771, "grad_norm": 16.951297760009766, "learning_rate": 9.247572815533981e-05, "loss": 0.4835, "step": 1550 }, { "epoch": 15.039759036144579, "grad_norm": 6.40467643737793, "learning_rate": 9.247087378640777e-05, "loss": 0.2946, "step": 1551 }, { "epoch": 15.049397590361446, "grad_norm": 5.362975597381592, "learning_rate": 9.246601941747573e-05, "loss": 0.3059, "step": 1552 }, { "epoch": 15.059036144578313, "grad_norm": 4.575148105621338, "learning_rate": 9.24611650485437e-05, "loss": 0.1939, "step": 1553 }, { "epoch": 15.068674698795181, "grad_norm": 5.0897064208984375, "learning_rate": 9.245631067961166e-05, "loss": 0.2514, "step": 1554 }, { "epoch": 15.078313253012048, "grad_norm": 5.3925323486328125, "learning_rate": 9.245145631067962e-05, "loss": 0.3966, "step": 1555 }, { "epoch": 15.087951807228915, "grad_norm": 19.954631805419922, "learning_rate": 9.244660194174757e-05, "loss": 0.6574, "step": 1556 }, { "epoch": 15.097590361445784, "grad_norm": 7.612180709838867, "learning_rate": 9.244174757281555e-05, "loss": 0.3822, "step": 1557 }, { "epoch": 15.10722891566265, "grad_norm": 5.349355220794678, "learning_rate": 9.24368932038835e-05, "loss": 0.2921, "step": 1558 }, { "epoch": 15.116867469879518, "grad_norm": 5.827071666717529, "learning_rate": 9.243203883495146e-05, "loss": 0.3477, "step": 1559 }, { "epoch": 15.126506024096386, "grad_norm": 7.532491683959961, "learning_rate": 9.242718446601942e-05, "loss": 0.337, "step": 1560 }, { "epoch": 15.136144578313253, "grad_norm": 4.274010181427002, "learning_rate": 9.242233009708739e-05, "loss": 0.2905, "step": 1561 }, { "epoch": 15.14578313253012, "grad_norm": 4.096954345703125, "learning_rate": 9.241747572815535e-05, "loss": 0.2641, "step": 1562 }, { "epoch": 15.155421686746989, "grad_norm": 3.469541072845459, "learning_rate": 9.24126213592233e-05, "loss": 0.2425, "step": 1563 }, { "epoch": 15.165060240963856, "grad_norm": 4.285534858703613, "learning_rate": 9.240776699029126e-05, "loss": 0.2309, "step": 1564 }, { "epoch": 15.174698795180722, "grad_norm": 4.745375156402588, "learning_rate": 9.240291262135922e-05, "loss": 0.332, "step": 1565 }, { "epoch": 15.184337349397591, "grad_norm": 8.890633583068848, "learning_rate": 9.239805825242719e-05, "loss": 0.4461, "step": 1566 }, { "epoch": 15.193975903614458, "grad_norm": 3.476792573928833, "learning_rate": 9.239320388349515e-05, "loss": 0.1932, "step": 1567 }, { "epoch": 15.203614457831325, "grad_norm": 6.957151412963867, "learning_rate": 9.238834951456311e-05, "loss": 0.3302, "step": 1568 }, { "epoch": 15.213253012048193, "grad_norm": 7.93690299987793, "learning_rate": 9.238349514563107e-05, "loss": 0.3409, "step": 1569 }, { "epoch": 15.22289156626506, "grad_norm": 8.988520622253418, "learning_rate": 9.237864077669904e-05, "loss": 0.3152, "step": 1570 }, { "epoch": 15.232530120481927, "grad_norm": 7.014593601226807, "learning_rate": 9.2373786407767e-05, "loss": 0.1977, "step": 1571 }, { "epoch": 15.242168674698796, "grad_norm": 3.6575584411621094, "learning_rate": 9.236893203883495e-05, "loss": 0.2177, "step": 1572 }, { "epoch": 15.251807228915663, "grad_norm": 6.288464546203613, "learning_rate": 9.236407766990291e-05, "loss": 0.2034, "step": 1573 }, { "epoch": 15.26144578313253, "grad_norm": 10.316359519958496, "learning_rate": 9.235922330097088e-05, "loss": 0.5187, "step": 1574 }, { "epoch": 15.271084337349398, "grad_norm": 21.947919845581055, "learning_rate": 9.235436893203884e-05, "loss": 0.3678, "step": 1575 }, { "epoch": 15.280722891566265, "grad_norm": 6.784307479858398, "learning_rate": 9.23495145631068e-05, "loss": 0.5802, "step": 1576 }, { "epoch": 15.290361445783132, "grad_norm": 8.722726821899414, "learning_rate": 9.234466019417476e-05, "loss": 0.2135, "step": 1577 }, { "epoch": 15.3, "grad_norm": 8.54143238067627, "learning_rate": 9.233980582524273e-05, "loss": 0.5654, "step": 1578 }, { "epoch": 15.309638554216868, "grad_norm": 3.624420642852783, "learning_rate": 9.233495145631068e-05, "loss": 0.2716, "step": 1579 }, { "epoch": 15.319277108433734, "grad_norm": 4.342920780181885, "learning_rate": 9.233009708737864e-05, "loss": 0.2896, "step": 1580 }, { "epoch": 15.328915662650603, "grad_norm": 6.910046577453613, "learning_rate": 9.23252427184466e-05, "loss": 0.354, "step": 1581 }, { "epoch": 15.33855421686747, "grad_norm": 13.085598945617676, "learning_rate": 9.232038834951456e-05, "loss": 0.4162, "step": 1582 }, { "epoch": 15.348192771084337, "grad_norm": 13.557994842529297, "learning_rate": 9.231553398058253e-05, "loss": 0.3421, "step": 1583 }, { "epoch": 15.357831325301206, "grad_norm": 6.4669880867004395, "learning_rate": 9.231067961165049e-05, "loss": 0.4543, "step": 1584 }, { "epoch": 15.367469879518072, "grad_norm": 5.234957218170166, "learning_rate": 9.230582524271844e-05, "loss": 0.3188, "step": 1585 }, { "epoch": 15.37710843373494, "grad_norm": 4.702942371368408, "learning_rate": 9.23009708737864e-05, "loss": 0.3323, "step": 1586 }, { "epoch": 15.386746987951808, "grad_norm": 7.057127475738525, "learning_rate": 9.229611650485437e-05, "loss": 0.4914, "step": 1587 }, { "epoch": 15.396385542168675, "grad_norm": 10.77542495727539, "learning_rate": 9.229126213592233e-05, "loss": 0.2792, "step": 1588 }, { "epoch": 15.406024096385542, "grad_norm": 7.716588973999023, "learning_rate": 9.228640776699029e-05, "loss": 0.4589, "step": 1589 }, { "epoch": 15.41566265060241, "grad_norm": 8.130538940429688, "learning_rate": 9.228155339805825e-05, "loss": 0.3471, "step": 1590 }, { "epoch": 15.425301204819277, "grad_norm": 5.592360019683838, "learning_rate": 9.227669902912622e-05, "loss": 0.4435, "step": 1591 }, { "epoch": 15.434939759036144, "grad_norm": 8.914158821105957, "learning_rate": 9.227184466019419e-05, "loss": 0.5802, "step": 1592 }, { "epoch": 15.444578313253013, "grad_norm": 12.321364402770996, "learning_rate": 9.226699029126215e-05, "loss": 0.4182, "step": 1593 }, { "epoch": 15.45421686746988, "grad_norm": 5.9666361808776855, "learning_rate": 9.22621359223301e-05, "loss": 0.2595, "step": 1594 }, { "epoch": 15.463855421686747, "grad_norm": 3.471569776535034, "learning_rate": 9.225728155339806e-05, "loss": 0.2286, "step": 1595 }, { "epoch": 15.473493975903615, "grad_norm": 13.239213943481445, "learning_rate": 9.225242718446603e-05, "loss": 0.494, "step": 1596 }, { "epoch": 15.483132530120482, "grad_norm": 7.371853828430176, "learning_rate": 9.224757281553399e-05, "loss": 0.3324, "step": 1597 }, { "epoch": 15.492771084337349, "grad_norm": 4.169154167175293, "learning_rate": 9.224271844660195e-05, "loss": 0.2903, "step": 1598 }, { "epoch": 15.502409638554218, "grad_norm": 11.317599296569824, "learning_rate": 9.223786407766991e-05, "loss": 0.3701, "step": 1599 }, { "epoch": 15.512048192771084, "grad_norm": 13.254508018493652, "learning_rate": 9.223300970873788e-05, "loss": 0.4293, "step": 1600 }, { "epoch": 15.521686746987951, "grad_norm": 4.724128246307373, "learning_rate": 9.222815533980584e-05, "loss": 0.3587, "step": 1601 }, { "epoch": 15.53132530120482, "grad_norm": 7.291380882263184, "learning_rate": 9.22233009708738e-05, "loss": 0.2881, "step": 1602 }, { "epoch": 15.540963855421687, "grad_norm": 7.082474231719971, "learning_rate": 9.221844660194175e-05, "loss": 0.3562, "step": 1603 }, { "epoch": 15.550602409638554, "grad_norm": 5.224260330200195, "learning_rate": 9.221359223300971e-05, "loss": 0.3853, "step": 1604 }, { "epoch": 15.560240963855422, "grad_norm": 5.91819429397583, "learning_rate": 9.220873786407768e-05, "loss": 0.3235, "step": 1605 }, { "epoch": 15.56987951807229, "grad_norm": 4.0472235679626465, "learning_rate": 9.220388349514564e-05, "loss": 0.3073, "step": 1606 }, { "epoch": 15.579518072289156, "grad_norm": 9.398139953613281, "learning_rate": 9.21990291262136e-05, "loss": 0.2029, "step": 1607 }, { "epoch": 15.589156626506025, "grad_norm": 8.735954284667969, "learning_rate": 9.219417475728155e-05, "loss": 0.3331, "step": 1608 }, { "epoch": 15.598795180722892, "grad_norm": 5.2330098152160645, "learning_rate": 9.218932038834953e-05, "loss": 0.3004, "step": 1609 }, { "epoch": 15.608433734939759, "grad_norm": 3.2248306274414062, "learning_rate": 9.218446601941748e-05, "loss": 0.2168, "step": 1610 }, { "epoch": 15.618072289156627, "grad_norm": 5.428384780883789, "learning_rate": 9.217961165048544e-05, "loss": 0.3062, "step": 1611 }, { "epoch": 15.627710843373494, "grad_norm": 10.52034854888916, "learning_rate": 9.21747572815534e-05, "loss": 0.4313, "step": 1612 }, { "epoch": 15.637349397590361, "grad_norm": 7.306883335113525, "learning_rate": 9.216990291262137e-05, "loss": 0.3116, "step": 1613 }, { "epoch": 15.64698795180723, "grad_norm": 3.651287078857422, "learning_rate": 9.216504854368933e-05, "loss": 0.2305, "step": 1614 }, { "epoch": 15.656626506024097, "grad_norm": 15.880545616149902, "learning_rate": 9.216019417475729e-05, "loss": 0.4418, "step": 1615 }, { "epoch": 15.666265060240963, "grad_norm": 3.052318572998047, "learning_rate": 9.215533980582524e-05, "loss": 0.2946, "step": 1616 }, { "epoch": 15.675903614457832, "grad_norm": 9.543049812316895, "learning_rate": 9.215048543689321e-05, "loss": 0.3617, "step": 1617 }, { "epoch": 15.685542168674699, "grad_norm": 9.962630271911621, "learning_rate": 9.214563106796117e-05, "loss": 0.2829, "step": 1618 }, { "epoch": 15.695180722891566, "grad_norm": 9.684564590454102, "learning_rate": 9.214077669902913e-05, "loss": 0.2465, "step": 1619 }, { "epoch": 15.704819277108435, "grad_norm": 3.283167600631714, "learning_rate": 9.213592233009709e-05, "loss": 0.1878, "step": 1620 }, { "epoch": 15.714457831325301, "grad_norm": 7.049105644226074, "learning_rate": 9.213106796116505e-05, "loss": 0.4303, "step": 1621 }, { "epoch": 15.724096385542168, "grad_norm": 7.516069412231445, "learning_rate": 9.212621359223302e-05, "loss": 0.4685, "step": 1622 }, { "epoch": 15.733734939759037, "grad_norm": 7.902530193328857, "learning_rate": 9.212135922330097e-05, "loss": 0.349, "step": 1623 }, { "epoch": 15.743373493975904, "grad_norm": 3.0929043292999268, "learning_rate": 9.211650485436893e-05, "loss": 0.2122, "step": 1624 }, { "epoch": 15.75301204819277, "grad_norm": 6.276230812072754, "learning_rate": 9.211165048543689e-05, "loss": 0.2637, "step": 1625 }, { "epoch": 15.76265060240964, "grad_norm": 5.130654335021973, "learning_rate": 9.210679611650486e-05, "loss": 0.2348, "step": 1626 }, { "epoch": 15.772289156626506, "grad_norm": 5.027799606323242, "learning_rate": 9.210194174757282e-05, "loss": 0.2944, "step": 1627 }, { "epoch": 15.781927710843373, "grad_norm": 4.424139022827148, "learning_rate": 9.209708737864078e-05, "loss": 0.256, "step": 1628 }, { "epoch": 15.791566265060242, "grad_norm": 12.415945053100586, "learning_rate": 9.209223300970873e-05, "loss": 0.4215, "step": 1629 }, { "epoch": 15.801204819277109, "grad_norm": 5.060678958892822, "learning_rate": 9.20873786407767e-05, "loss": 0.1905, "step": 1630 }, { "epoch": 15.810843373493976, "grad_norm": 29.18994140625, "learning_rate": 9.208252427184466e-05, "loss": 0.2807, "step": 1631 }, { "epoch": 15.820481927710844, "grad_norm": 5.116558074951172, "learning_rate": 9.207766990291262e-05, "loss": 0.3819, "step": 1632 }, { "epoch": 15.830120481927711, "grad_norm": 4.7890729904174805, "learning_rate": 9.207281553398058e-05, "loss": 0.3864, "step": 1633 }, { "epoch": 15.839759036144578, "grad_norm": 8.054818153381348, "learning_rate": 9.206796116504855e-05, "loss": 0.4231, "step": 1634 }, { "epoch": 15.849397590361447, "grad_norm": 4.143812656402588, "learning_rate": 9.206310679611651e-05, "loss": 0.4434, "step": 1635 }, { "epoch": 15.859036144578313, "grad_norm": 4.982356071472168, "learning_rate": 9.205825242718447e-05, "loss": 0.1609, "step": 1636 }, { "epoch": 15.86867469879518, "grad_norm": 4.4528656005859375, "learning_rate": 9.205339805825242e-05, "loss": 0.2318, "step": 1637 }, { "epoch": 15.878313253012049, "grad_norm": 8.48991870880127, "learning_rate": 9.204854368932038e-05, "loss": 0.3649, "step": 1638 }, { "epoch": 15.887951807228916, "grad_norm": 5.296942234039307, "learning_rate": 9.204368932038835e-05, "loss": 0.3915, "step": 1639 }, { "epoch": 15.897590361445783, "grad_norm": 8.050596237182617, "learning_rate": 9.203883495145631e-05, "loss": 0.5495, "step": 1640 }, { "epoch": 15.907228915662651, "grad_norm": 7.035776138305664, "learning_rate": 9.203398058252428e-05, "loss": 0.3707, "step": 1641 }, { "epoch": 15.916867469879518, "grad_norm": 6.767737865447998, "learning_rate": 9.202912621359224e-05, "loss": 0.5796, "step": 1642 }, { "epoch": 15.926506024096385, "grad_norm": 6.260571002960205, "learning_rate": 9.20242718446602e-05, "loss": 0.3132, "step": 1643 }, { "epoch": 15.936144578313254, "grad_norm": 4.624007225036621, "learning_rate": 9.201941747572817e-05, "loss": 0.1722, "step": 1644 }, { "epoch": 15.94578313253012, "grad_norm": 4.396580696105957, "learning_rate": 9.201456310679613e-05, "loss": 0.3157, "step": 1645 }, { "epoch": 15.955421686746988, "grad_norm": 8.151201248168945, "learning_rate": 9.200970873786408e-05, "loss": 0.215, "step": 1646 }, { "epoch": 15.965060240963856, "grad_norm": 6.40098762512207, "learning_rate": 9.200485436893204e-05, "loss": 0.3255, "step": 1647 }, { "epoch": 15.974698795180723, "grad_norm": 5.153090000152588, "learning_rate": 9.200000000000001e-05, "loss": 0.3044, "step": 1648 }, { "epoch": 15.98433734939759, "grad_norm": 8.493891716003418, "learning_rate": 9.199514563106797e-05, "loss": 0.3363, "step": 1649 }, { "epoch": 15.993975903614459, "grad_norm": 4.6506123542785645, "learning_rate": 9.199029126213593e-05, "loss": 0.3697, "step": 1650 }, { "epoch": 16.009638554216867, "grad_norm": 5.462498188018799, "learning_rate": 9.198543689320389e-05, "loss": 0.3123, "step": 1651 }, { "epoch": 16.019277108433734, "grad_norm": 7.645285129547119, "learning_rate": 9.198058252427186e-05, "loss": 0.2318, "step": 1652 }, { "epoch": 16.0289156626506, "grad_norm": 5.7372941970825195, "learning_rate": 9.197572815533982e-05, "loss": 0.3462, "step": 1653 }, { "epoch": 16.03855421686747, "grad_norm": 3.322057008743286, "learning_rate": 9.197087378640777e-05, "loss": 0.2347, "step": 1654 }, { "epoch": 16.048192771084338, "grad_norm": 5.624446868896484, "learning_rate": 9.196601941747573e-05, "loss": 0.3283, "step": 1655 }, { "epoch": 16.057831325301205, "grad_norm": 10.312376976013184, "learning_rate": 9.196116504854369e-05, "loss": 0.282, "step": 1656 }, { "epoch": 16.06746987951807, "grad_norm": 4.830899238586426, "learning_rate": 9.195631067961166e-05, "loss": 0.5081, "step": 1657 }, { "epoch": 16.07710843373494, "grad_norm": 7.334006309509277, "learning_rate": 9.195145631067962e-05, "loss": 0.2416, "step": 1658 }, { "epoch": 16.086746987951805, "grad_norm": 6.953143119812012, "learning_rate": 9.194660194174758e-05, "loss": 0.4331, "step": 1659 }, { "epoch": 16.096385542168676, "grad_norm": 7.3238749504089355, "learning_rate": 9.194174757281553e-05, "loss": 0.3919, "step": 1660 }, { "epoch": 16.106024096385543, "grad_norm": 2.5695114135742188, "learning_rate": 9.19368932038835e-05, "loss": 0.1215, "step": 1661 }, { "epoch": 16.11566265060241, "grad_norm": 8.223772048950195, "learning_rate": 9.193203883495146e-05, "loss": 0.4999, "step": 1662 }, { "epoch": 16.125301204819277, "grad_norm": 5.9075117111206055, "learning_rate": 9.192718446601942e-05, "loss": 0.3055, "step": 1663 }, { "epoch": 16.134939759036143, "grad_norm": 6.687393665313721, "learning_rate": 9.192233009708738e-05, "loss": 0.4082, "step": 1664 }, { "epoch": 16.14457831325301, "grad_norm": 9.096417427062988, "learning_rate": 9.191747572815535e-05, "loss": 0.439, "step": 1665 }, { "epoch": 16.15421686746988, "grad_norm": 5.386663436889648, "learning_rate": 9.191262135922331e-05, "loss": 0.377, "step": 1666 }, { "epoch": 16.163855421686748, "grad_norm": 7.702605724334717, "learning_rate": 9.190776699029127e-05, "loss": 0.4625, "step": 1667 }, { "epoch": 16.173493975903614, "grad_norm": 6.250150680541992, "learning_rate": 9.190291262135922e-05, "loss": 0.4487, "step": 1668 }, { "epoch": 16.18313253012048, "grad_norm": 8.128484725952148, "learning_rate": 9.18980582524272e-05, "loss": 0.509, "step": 1669 }, { "epoch": 16.19277108433735, "grad_norm": 11.469080924987793, "learning_rate": 9.189320388349515e-05, "loss": 0.5061, "step": 1670 }, { "epoch": 16.202409638554215, "grad_norm": 4.6851396560668945, "learning_rate": 9.188834951456311e-05, "loss": 0.2588, "step": 1671 }, { "epoch": 16.212048192771086, "grad_norm": 8.054449081420898, "learning_rate": 9.188349514563107e-05, "loss": 0.5733, "step": 1672 }, { "epoch": 16.221686746987952, "grad_norm": 8.237906455993652, "learning_rate": 9.187864077669903e-05, "loss": 0.3693, "step": 1673 }, { "epoch": 16.23132530120482, "grad_norm": 20.55581283569336, "learning_rate": 9.1873786407767e-05, "loss": 0.3934, "step": 1674 }, { "epoch": 16.240963855421686, "grad_norm": 6.263428211212158, "learning_rate": 9.186893203883495e-05, "loss": 0.2096, "step": 1675 }, { "epoch": 16.250602409638553, "grad_norm": 10.260252952575684, "learning_rate": 9.186407766990291e-05, "loss": 0.4358, "step": 1676 }, { "epoch": 16.26024096385542, "grad_norm": 4.8274712562561035, "learning_rate": 9.185922330097087e-05, "loss": 0.3124, "step": 1677 }, { "epoch": 16.26987951807229, "grad_norm": 7.079626560211182, "learning_rate": 9.185436893203884e-05, "loss": 0.4284, "step": 1678 }, { "epoch": 16.279518072289157, "grad_norm": 4.897369861602783, "learning_rate": 9.18495145631068e-05, "loss": 0.3088, "step": 1679 }, { "epoch": 16.289156626506024, "grad_norm": 8.8363618850708, "learning_rate": 9.184466019417476e-05, "loss": 0.4002, "step": 1680 }, { "epoch": 16.29879518072289, "grad_norm": 5.527097225189209, "learning_rate": 9.183980582524271e-05, "loss": 0.356, "step": 1681 }, { "epoch": 16.308433734939758, "grad_norm": 3.486233711242676, "learning_rate": 9.183495145631069e-05, "loss": 0.2702, "step": 1682 }, { "epoch": 16.318072289156625, "grad_norm": 6.96115255355835, "learning_rate": 9.183009708737864e-05, "loss": 0.2642, "step": 1683 }, { "epoch": 16.327710843373495, "grad_norm": 4.332570552825928, "learning_rate": 9.18252427184466e-05, "loss": 0.2515, "step": 1684 }, { "epoch": 16.337349397590362, "grad_norm": 6.1505913734436035, "learning_rate": 9.182038834951456e-05, "loss": 0.5095, "step": 1685 }, { "epoch": 16.34698795180723, "grad_norm": 5.194479942321777, "learning_rate": 9.181553398058253e-05, "loss": 0.3303, "step": 1686 }, { "epoch": 16.356626506024096, "grad_norm": 5.008617877960205, "learning_rate": 9.181067961165049e-05, "loss": 0.2411, "step": 1687 }, { "epoch": 16.366265060240963, "grad_norm": 5.358800411224365, "learning_rate": 9.180582524271845e-05, "loss": 0.3849, "step": 1688 }, { "epoch": 16.37590361445783, "grad_norm": 9.695611953735352, "learning_rate": 9.18009708737864e-05, "loss": 0.3418, "step": 1689 }, { "epoch": 16.3855421686747, "grad_norm": 7.134877681732178, "learning_rate": 9.179611650485436e-05, "loss": 0.4273, "step": 1690 }, { "epoch": 16.395180722891567, "grad_norm": 6.154335021972656, "learning_rate": 9.179126213592233e-05, "loss": 0.2013, "step": 1691 }, { "epoch": 16.404819277108434, "grad_norm": 8.879928588867188, "learning_rate": 9.17864077669903e-05, "loss": 0.368, "step": 1692 }, { "epoch": 16.4144578313253, "grad_norm": 7.423298358917236, "learning_rate": 9.178155339805826e-05, "loss": 0.4837, "step": 1693 }, { "epoch": 16.424096385542168, "grad_norm": 3.6976327896118164, "learning_rate": 9.177669902912622e-05, "loss": 0.1487, "step": 1694 }, { "epoch": 16.433734939759034, "grad_norm": 4.770668029785156, "learning_rate": 9.177184466019418e-05, "loss": 0.3399, "step": 1695 }, { "epoch": 16.443373493975905, "grad_norm": 7.191640377044678, "learning_rate": 9.176699029126215e-05, "loss": 0.2689, "step": 1696 }, { "epoch": 16.45301204819277, "grad_norm": 4.964473724365234, "learning_rate": 9.17621359223301e-05, "loss": 0.5686, "step": 1697 }, { "epoch": 16.46265060240964, "grad_norm": 5.874170303344727, "learning_rate": 9.175728155339806e-05, "loss": 0.2434, "step": 1698 }, { "epoch": 16.472289156626506, "grad_norm": 4.536634922027588, "learning_rate": 9.175242718446602e-05, "loss": 0.2896, "step": 1699 }, { "epoch": 16.481927710843372, "grad_norm": 3.496643543243408, "learning_rate": 9.174757281553399e-05, "loss": 0.2465, "step": 1700 }, { "epoch": 16.49156626506024, "grad_norm": 4.980424880981445, "learning_rate": 9.174271844660195e-05, "loss": 0.203, "step": 1701 }, { "epoch": 16.50120481927711, "grad_norm": 10.216141700744629, "learning_rate": 9.173786407766991e-05, "loss": 0.5518, "step": 1702 }, { "epoch": 16.510843373493977, "grad_norm": 6.647271633148193, "learning_rate": 9.173300970873787e-05, "loss": 0.5191, "step": 1703 }, { "epoch": 16.520481927710843, "grad_norm": 3.986293315887451, "learning_rate": 9.172815533980584e-05, "loss": 0.2404, "step": 1704 }, { "epoch": 16.53012048192771, "grad_norm": 4.640357494354248, "learning_rate": 9.17233009708738e-05, "loss": 0.4323, "step": 1705 }, { "epoch": 16.539759036144577, "grad_norm": 5.518274784088135, "learning_rate": 9.171844660194175e-05, "loss": 0.2704, "step": 1706 }, { "epoch": 16.549397590361444, "grad_norm": 5.550935745239258, "learning_rate": 9.171359223300971e-05, "loss": 0.5125, "step": 1707 }, { "epoch": 16.559036144578315, "grad_norm": 6.473182678222656, "learning_rate": 9.170873786407768e-05, "loss": 0.5589, "step": 1708 }, { "epoch": 16.56867469879518, "grad_norm": 5.816985607147217, "learning_rate": 9.170388349514564e-05, "loss": 0.4933, "step": 1709 }, { "epoch": 16.57831325301205, "grad_norm": 4.932758331298828, "learning_rate": 9.16990291262136e-05, "loss": 0.3536, "step": 1710 }, { "epoch": 16.587951807228915, "grad_norm": 5.526294708251953, "learning_rate": 9.169417475728156e-05, "loss": 0.3851, "step": 1711 }, { "epoch": 16.597590361445782, "grad_norm": 9.560312271118164, "learning_rate": 9.168932038834951e-05, "loss": 0.4734, "step": 1712 }, { "epoch": 16.60722891566265, "grad_norm": 11.79024600982666, "learning_rate": 9.168446601941748e-05, "loss": 0.367, "step": 1713 }, { "epoch": 16.61686746987952, "grad_norm": 6.670504570007324, "learning_rate": 9.167961165048544e-05, "loss": 0.2983, "step": 1714 }, { "epoch": 16.626506024096386, "grad_norm": 6.678935527801514, "learning_rate": 9.16747572815534e-05, "loss": 0.3247, "step": 1715 }, { "epoch": 16.636144578313253, "grad_norm": 7.942539215087891, "learning_rate": 9.166990291262136e-05, "loss": 0.4295, "step": 1716 }, { "epoch": 16.64578313253012, "grad_norm": 6.608550548553467, "learning_rate": 9.166504854368933e-05, "loss": 0.3875, "step": 1717 }, { "epoch": 16.655421686746987, "grad_norm": 4.381741523742676, "learning_rate": 9.166019417475729e-05, "loss": 0.3623, "step": 1718 }, { "epoch": 16.665060240963854, "grad_norm": 9.40506362915039, "learning_rate": 9.165533980582524e-05, "loss": 0.4697, "step": 1719 }, { "epoch": 16.674698795180724, "grad_norm": 8.793902397155762, "learning_rate": 9.16504854368932e-05, "loss": 0.6468, "step": 1720 }, { "epoch": 16.68433734939759, "grad_norm": 7.4632182121276855, "learning_rate": 9.164563106796117e-05, "loss": 0.4648, "step": 1721 }, { "epoch": 16.693975903614458, "grad_norm": 8.954852104187012, "learning_rate": 9.164077669902913e-05, "loss": 0.5709, "step": 1722 }, { "epoch": 16.703614457831325, "grad_norm": 4.270956039428711, "learning_rate": 9.163592233009709e-05, "loss": 0.4378, "step": 1723 }, { "epoch": 16.71325301204819, "grad_norm": 6.30915641784668, "learning_rate": 9.163106796116505e-05, "loss": 0.4107, "step": 1724 }, { "epoch": 16.72289156626506, "grad_norm": 10.781926155090332, "learning_rate": 9.162621359223302e-05, "loss": 0.1976, "step": 1725 }, { "epoch": 16.73253012048193, "grad_norm": 5.5361504554748535, "learning_rate": 9.162135922330098e-05, "loss": 0.2665, "step": 1726 }, { "epoch": 16.742168674698796, "grad_norm": 7.219536304473877, "learning_rate": 9.161650485436893e-05, "loss": 0.4815, "step": 1727 }, { "epoch": 16.751807228915663, "grad_norm": 6.481014251708984, "learning_rate": 9.161165048543689e-05, "loss": 0.4543, "step": 1728 }, { "epoch": 16.76144578313253, "grad_norm": 4.0883073806762695, "learning_rate": 9.160679611650485e-05, "loss": 0.2033, "step": 1729 }, { "epoch": 16.771084337349397, "grad_norm": 7.168371677398682, "learning_rate": 9.160194174757282e-05, "loss": 0.3926, "step": 1730 }, { "epoch": 16.780722891566263, "grad_norm": 5.735024929046631, "learning_rate": 9.159708737864078e-05, "loss": 0.27, "step": 1731 }, { "epoch": 16.790361445783134, "grad_norm": 9.442459106445312, "learning_rate": 9.159223300970874e-05, "loss": 0.6061, "step": 1732 }, { "epoch": 16.8, "grad_norm": 4.096850395202637, "learning_rate": 9.15873786407767e-05, "loss": 0.2191, "step": 1733 }, { "epoch": 16.809638554216868, "grad_norm": 9.455533981323242, "learning_rate": 9.158252427184467e-05, "loss": 0.1966, "step": 1734 }, { "epoch": 16.819277108433734, "grad_norm": 8.573326110839844, "learning_rate": 9.157766990291262e-05, "loss": 0.4272, "step": 1735 }, { "epoch": 16.8289156626506, "grad_norm": 5.754106521606445, "learning_rate": 9.157281553398058e-05, "loss": 0.3295, "step": 1736 }, { "epoch": 16.83855421686747, "grad_norm": 9.518383026123047, "learning_rate": 9.156796116504854e-05, "loss": 0.3165, "step": 1737 }, { "epoch": 16.84819277108434, "grad_norm": 6.861412525177002, "learning_rate": 9.156310679611651e-05, "loss": 0.3475, "step": 1738 }, { "epoch": 16.857831325301206, "grad_norm": 6.245154857635498, "learning_rate": 9.155825242718447e-05, "loss": 0.3665, "step": 1739 }, { "epoch": 16.867469879518072, "grad_norm": 6.015488624572754, "learning_rate": 9.155339805825243e-05, "loss": 0.4426, "step": 1740 }, { "epoch": 16.87710843373494, "grad_norm": 4.97740364074707, "learning_rate": 9.154854368932038e-05, "loss": 0.3451, "step": 1741 }, { "epoch": 16.886746987951806, "grad_norm": 6.152440071105957, "learning_rate": 9.154368932038835e-05, "loss": 0.2757, "step": 1742 }, { "epoch": 16.896385542168673, "grad_norm": 11.317872047424316, "learning_rate": 9.153883495145633e-05, "loss": 0.303, "step": 1743 }, { "epoch": 16.906024096385543, "grad_norm": 6.498701095581055, "learning_rate": 9.153398058252428e-05, "loss": 0.5034, "step": 1744 }, { "epoch": 16.91566265060241, "grad_norm": 12.357467651367188, "learning_rate": 9.152912621359224e-05, "loss": 0.3073, "step": 1745 }, { "epoch": 16.925301204819277, "grad_norm": 6.438660621643066, "learning_rate": 9.15242718446602e-05, "loss": 0.3358, "step": 1746 }, { "epoch": 16.934939759036144, "grad_norm": 8.99237060546875, "learning_rate": 9.151941747572817e-05, "loss": 0.1816, "step": 1747 }, { "epoch": 16.94457831325301, "grad_norm": 7.025774955749512, "learning_rate": 9.151456310679613e-05, "loss": 0.3893, "step": 1748 }, { "epoch": 16.954216867469878, "grad_norm": 5.503346920013428, "learning_rate": 9.150970873786409e-05, "loss": 0.2762, "step": 1749 }, { "epoch": 16.96385542168675, "grad_norm": 7.412036418914795, "learning_rate": 9.150485436893204e-05, "loss": 0.2973, "step": 1750 }, { "epoch": 16.973493975903615, "grad_norm": 4.449939250946045, "learning_rate": 9.15e-05, "loss": 0.2189, "step": 1751 }, { "epoch": 16.983132530120482, "grad_norm": 5.258821487426758, "learning_rate": 9.149514563106797e-05, "loss": 0.3811, "step": 1752 }, { "epoch": 16.99277108433735, "grad_norm": 3.380094051361084, "learning_rate": 9.149029126213593e-05, "loss": 0.1714, "step": 1753 }, { "epoch": 17.00843373493976, "grad_norm": 16.458574295043945, "learning_rate": 9.148543689320389e-05, "loss": 0.3863, "step": 1754 }, { "epoch": 17.018072289156628, "grad_norm": 19.748685836791992, "learning_rate": 9.148058252427185e-05, "loss": 0.4179, "step": 1755 }, { "epoch": 17.027710843373494, "grad_norm": 9.114288330078125, "learning_rate": 9.147572815533982e-05, "loss": 0.4086, "step": 1756 }, { "epoch": 17.03734939759036, "grad_norm": 27.007736206054688, "learning_rate": 9.147087378640778e-05, "loss": 0.2917, "step": 1757 }, { "epoch": 17.04698795180723, "grad_norm": 15.069550514221191, "learning_rate": 9.146601941747573e-05, "loss": 0.7233, "step": 1758 }, { "epoch": 17.056626506024095, "grad_norm": 12.665718078613281, "learning_rate": 9.146116504854369e-05, "loss": 0.3031, "step": 1759 }, { "epoch": 17.066265060240966, "grad_norm": 13.290709495544434, "learning_rate": 9.145631067961166e-05, "loss": 0.3328, "step": 1760 }, { "epoch": 17.075903614457832, "grad_norm": 5.6854567527771, "learning_rate": 9.145145631067962e-05, "loss": 0.3756, "step": 1761 }, { "epoch": 17.0855421686747, "grad_norm": 29.93622398376465, "learning_rate": 9.144660194174758e-05, "loss": 0.5874, "step": 1762 }, { "epoch": 17.095180722891566, "grad_norm": 6.349117755889893, "learning_rate": 9.144174757281554e-05, "loss": 0.3644, "step": 1763 }, { "epoch": 17.104819277108433, "grad_norm": 5.813069820404053, "learning_rate": 9.14368932038835e-05, "loss": 0.2534, "step": 1764 }, { "epoch": 17.1144578313253, "grad_norm": 16.349294662475586, "learning_rate": 9.143203883495146e-05, "loss": 0.43, "step": 1765 }, { "epoch": 17.12409638554217, "grad_norm": 3.9526543617248535, "learning_rate": 9.142718446601942e-05, "loss": 0.4253, "step": 1766 }, { "epoch": 17.133734939759037, "grad_norm": 26.61086082458496, "learning_rate": 9.142233009708738e-05, "loss": 0.4237, "step": 1767 }, { "epoch": 17.143373493975904, "grad_norm": 7.712512969970703, "learning_rate": 9.141747572815534e-05, "loss": 0.2319, "step": 1768 }, { "epoch": 17.15301204819277, "grad_norm": 43.94197082519531, "learning_rate": 9.141262135922331e-05, "loss": 0.3731, "step": 1769 }, { "epoch": 17.162650602409638, "grad_norm": 13.25898265838623, "learning_rate": 9.140776699029127e-05, "loss": 0.3632, "step": 1770 }, { "epoch": 17.172289156626505, "grad_norm": 4.068243980407715, "learning_rate": 9.140291262135922e-05, "loss": 0.3795, "step": 1771 }, { "epoch": 17.181927710843375, "grad_norm": 21.277507781982422, "learning_rate": 9.139805825242718e-05, "loss": 0.3546, "step": 1772 }, { "epoch": 17.191566265060242, "grad_norm": 10.811474800109863, "learning_rate": 9.139320388349515e-05, "loss": 0.188, "step": 1773 }, { "epoch": 17.20120481927711, "grad_norm": 10.3988618850708, "learning_rate": 9.138834951456311e-05, "loss": 0.3119, "step": 1774 }, { "epoch": 17.210843373493976, "grad_norm": 7.200343608856201, "learning_rate": 9.138349514563107e-05, "loss": 0.3881, "step": 1775 }, { "epoch": 17.220481927710843, "grad_norm": 3.027552604675293, "learning_rate": 9.137864077669903e-05, "loss": 0.3209, "step": 1776 }, { "epoch": 17.23012048192771, "grad_norm": 1.9955863952636719, "learning_rate": 9.1373786407767e-05, "loss": 0.1626, "step": 1777 }, { "epoch": 17.23975903614458, "grad_norm": 12.239253044128418, "learning_rate": 9.136893203883496e-05, "loss": 0.4644, "step": 1778 }, { "epoch": 17.249397590361447, "grad_norm": 8.097572326660156, "learning_rate": 9.136407766990291e-05, "loss": 0.5881, "step": 1779 }, { "epoch": 17.259036144578314, "grad_norm": 7.779978275299072, "learning_rate": 9.135922330097087e-05, "loss": 0.3442, "step": 1780 }, { "epoch": 17.26867469879518, "grad_norm": 10.878144264221191, "learning_rate": 9.135436893203884e-05, "loss": 0.3797, "step": 1781 }, { "epoch": 17.278313253012048, "grad_norm": 19.178882598876953, "learning_rate": 9.13495145631068e-05, "loss": 0.3037, "step": 1782 }, { "epoch": 17.287951807228914, "grad_norm": 16.79674530029297, "learning_rate": 9.134466019417476e-05, "loss": 0.2521, "step": 1783 }, { "epoch": 17.297590361445785, "grad_norm": 3.985231876373291, "learning_rate": 9.133980582524272e-05, "loss": 0.3494, "step": 1784 }, { "epoch": 17.30722891566265, "grad_norm": 14.506234169006348, "learning_rate": 9.133495145631067e-05, "loss": 0.5588, "step": 1785 }, { "epoch": 17.31686746987952, "grad_norm": 17.994531631469727, "learning_rate": 9.133009708737864e-05, "loss": 0.5109, "step": 1786 }, { "epoch": 17.326506024096386, "grad_norm": 2.694624662399292, "learning_rate": 9.13252427184466e-05, "loss": 0.1521, "step": 1787 }, { "epoch": 17.336144578313252, "grad_norm": 12.63049030303955, "learning_rate": 9.132038834951456e-05, "loss": 0.2808, "step": 1788 }, { "epoch": 17.34578313253012, "grad_norm": 4.62308406829834, "learning_rate": 9.131553398058252e-05, "loss": 0.3093, "step": 1789 }, { "epoch": 17.355421686746986, "grad_norm": 11.51387882232666, "learning_rate": 9.131067961165049e-05, "loss": 0.3793, "step": 1790 }, { "epoch": 17.365060240963857, "grad_norm": 7.981752872467041, "learning_rate": 9.130582524271845e-05, "loss": 0.3804, "step": 1791 }, { "epoch": 17.374698795180723, "grad_norm": 4.777389049530029, "learning_rate": 9.13009708737864e-05, "loss": 0.3187, "step": 1792 }, { "epoch": 17.38433734939759, "grad_norm": 3.8667120933532715, "learning_rate": 9.129611650485438e-05, "loss": 0.48, "step": 1793 }, { "epoch": 17.393975903614457, "grad_norm": 5.996094226837158, "learning_rate": 9.129126213592233e-05, "loss": 0.1767, "step": 1794 }, { "epoch": 17.403614457831324, "grad_norm": 30.727758407592773, "learning_rate": 9.12864077669903e-05, "loss": 0.3512, "step": 1795 }, { "epoch": 17.413253012048195, "grad_norm": 29.8641300201416, "learning_rate": 9.128155339805826e-05, "loss": 0.3856, "step": 1796 }, { "epoch": 17.42289156626506, "grad_norm": 14.471880912780762, "learning_rate": 9.127669902912622e-05, "loss": 0.6311, "step": 1797 }, { "epoch": 17.43253012048193, "grad_norm": 27.500335693359375, "learning_rate": 9.127184466019418e-05, "loss": 0.3127, "step": 1798 }, { "epoch": 17.442168674698795, "grad_norm": 19.209325790405273, "learning_rate": 9.126699029126215e-05, "loss": 0.3858, "step": 1799 }, { "epoch": 17.451807228915662, "grad_norm": 9.984304428100586, "learning_rate": 9.126213592233011e-05, "loss": 0.6156, "step": 1800 }, { "epoch": 17.46144578313253, "grad_norm": 8.915853500366211, "learning_rate": 9.125728155339807e-05, "loss": 0.4848, "step": 1801 }, { "epoch": 17.471084337349396, "grad_norm": 26.15042495727539, "learning_rate": 9.125242718446602e-05, "loss": 0.4065, "step": 1802 }, { "epoch": 17.480722891566266, "grad_norm": 7.501951694488525, "learning_rate": 9.1247572815534e-05, "loss": 0.2656, "step": 1803 }, { "epoch": 17.490361445783133, "grad_norm": 20.7778263092041, "learning_rate": 9.124271844660195e-05, "loss": 0.2479, "step": 1804 }, { "epoch": 17.5, "grad_norm": 35.607059478759766, "learning_rate": 9.123786407766991e-05, "loss": 0.3405, "step": 1805 }, { "epoch": 17.509638554216867, "grad_norm": 11.551884651184082, "learning_rate": 9.123300970873787e-05, "loss": 0.3386, "step": 1806 }, { "epoch": 17.519277108433734, "grad_norm": 5.860926628112793, "learning_rate": 9.122815533980583e-05, "loss": 0.2686, "step": 1807 }, { "epoch": 17.528915662650604, "grad_norm": 2.737905263900757, "learning_rate": 9.12233009708738e-05, "loss": 0.392, "step": 1808 }, { "epoch": 17.53855421686747, "grad_norm": 9.187110900878906, "learning_rate": 9.121844660194175e-05, "loss": 0.2751, "step": 1809 }, { "epoch": 17.548192771084338, "grad_norm": 5.679970741271973, "learning_rate": 9.121359223300971e-05, "loss": 0.3385, "step": 1810 }, { "epoch": 17.557831325301205, "grad_norm": 7.9858808517456055, "learning_rate": 9.120873786407767e-05, "loss": 0.5136, "step": 1811 }, { "epoch": 17.56746987951807, "grad_norm": 2.584502696990967, "learning_rate": 9.120388349514564e-05, "loss": 0.1941, "step": 1812 }, { "epoch": 17.57710843373494, "grad_norm": 39.170963287353516, "learning_rate": 9.11990291262136e-05, "loss": 0.4179, "step": 1813 }, { "epoch": 17.586746987951805, "grad_norm": 24.13837432861328, "learning_rate": 9.119417475728156e-05, "loss": 0.4206, "step": 1814 }, { "epoch": 17.596385542168676, "grad_norm": 8.567994117736816, "learning_rate": 9.118932038834951e-05, "loss": 0.2043, "step": 1815 }, { "epoch": 17.606024096385543, "grad_norm": 12.35608196258545, "learning_rate": 9.118446601941749e-05, "loss": 0.3887, "step": 1816 }, { "epoch": 17.61566265060241, "grad_norm": 13.643354415893555, "learning_rate": 9.117961165048544e-05, "loss": 0.3822, "step": 1817 }, { "epoch": 17.625301204819277, "grad_norm": 46.201480865478516, "learning_rate": 9.11747572815534e-05, "loss": 0.3128, "step": 1818 }, { "epoch": 17.634939759036143, "grad_norm": 17.647132873535156, "learning_rate": 9.116990291262136e-05, "loss": 0.3395, "step": 1819 }, { "epoch": 17.644578313253014, "grad_norm": 4.240926265716553, "learning_rate": 9.116504854368933e-05, "loss": 0.1616, "step": 1820 }, { "epoch": 17.65421686746988, "grad_norm": 9.083270072937012, "learning_rate": 9.116019417475729e-05, "loss": 0.3386, "step": 1821 }, { "epoch": 17.663855421686748, "grad_norm": 14.101011276245117, "learning_rate": 9.115533980582525e-05, "loss": 0.4512, "step": 1822 }, { "epoch": 17.673493975903614, "grad_norm": 13.953588485717773, "learning_rate": 9.11504854368932e-05, "loss": 0.3045, "step": 1823 }, { "epoch": 17.68313253012048, "grad_norm": 7.748983860015869, "learning_rate": 9.114563106796116e-05, "loss": 0.3649, "step": 1824 }, { "epoch": 17.69277108433735, "grad_norm": 8.996646881103516, "learning_rate": 9.114077669902913e-05, "loss": 0.2518, "step": 1825 }, { "epoch": 17.702409638554215, "grad_norm": 19.015043258666992, "learning_rate": 9.113592233009709e-05, "loss": 0.5723, "step": 1826 }, { "epoch": 17.712048192771086, "grad_norm": 6.1306562423706055, "learning_rate": 9.113106796116505e-05, "loss": 0.2864, "step": 1827 }, { "epoch": 17.721686746987952, "grad_norm": 5.643147945404053, "learning_rate": 9.1126213592233e-05, "loss": 0.2746, "step": 1828 }, { "epoch": 17.73132530120482, "grad_norm": 10.467835426330566, "learning_rate": 9.112135922330098e-05, "loss": 0.3856, "step": 1829 }, { "epoch": 17.740963855421686, "grad_norm": 15.491721153259277, "learning_rate": 9.111650485436894e-05, "loss": 0.325, "step": 1830 }, { "epoch": 17.750602409638553, "grad_norm": 48.93743896484375, "learning_rate": 9.111165048543689e-05, "loss": 0.3387, "step": 1831 }, { "epoch": 17.760240963855424, "grad_norm": 11.964858055114746, "learning_rate": 9.110679611650485e-05, "loss": 0.6422, "step": 1832 }, { "epoch": 17.76987951807229, "grad_norm": 14.015859603881836, "learning_rate": 9.110194174757282e-05, "loss": 0.4091, "step": 1833 }, { "epoch": 17.779518072289157, "grad_norm": 3.969501256942749, "learning_rate": 9.109708737864078e-05, "loss": 0.1316, "step": 1834 }, { "epoch": 17.789156626506024, "grad_norm": 6.3397932052612305, "learning_rate": 9.109223300970874e-05, "loss": 0.1209, "step": 1835 }, { "epoch": 17.79879518072289, "grad_norm": 23.941513061523438, "learning_rate": 9.10873786407767e-05, "loss": 0.4545, "step": 1836 }, { "epoch": 17.808433734939758, "grad_norm": 1.672322154045105, "learning_rate": 9.108252427184467e-05, "loss": 0.3412, "step": 1837 }, { "epoch": 17.818072289156625, "grad_norm": 10.029732704162598, "learning_rate": 9.107766990291262e-05, "loss": 0.395, "step": 1838 }, { "epoch": 17.827710843373495, "grad_norm": 2.8784024715423584, "learning_rate": 9.107281553398058e-05, "loss": 0.4489, "step": 1839 }, { "epoch": 17.837349397590362, "grad_norm": 15.18856430053711, "learning_rate": 9.106796116504854e-05, "loss": 0.3835, "step": 1840 }, { "epoch": 17.84698795180723, "grad_norm": 24.669336318969727, "learning_rate": 9.10631067961165e-05, "loss": 0.29, "step": 1841 }, { "epoch": 17.856626506024096, "grad_norm": 8.78890323638916, "learning_rate": 9.105825242718447e-05, "loss": 0.5341, "step": 1842 }, { "epoch": 17.866265060240963, "grad_norm": 18.819622039794922, "learning_rate": 9.105339805825244e-05, "loss": 0.2072, "step": 1843 }, { "epoch": 17.87590361445783, "grad_norm": 21.283897399902344, "learning_rate": 9.10485436893204e-05, "loss": 0.4561, "step": 1844 }, { "epoch": 17.8855421686747, "grad_norm": 6.719386577606201, "learning_rate": 9.104368932038836e-05, "loss": 0.2931, "step": 1845 }, { "epoch": 17.895180722891567, "grad_norm": 3.4685840606689453, "learning_rate": 9.103883495145631e-05, "loss": 0.1296, "step": 1846 }, { "epoch": 17.904819277108434, "grad_norm": 26.054325103759766, "learning_rate": 9.103398058252429e-05, "loss": 0.2161, "step": 1847 }, { "epoch": 17.9144578313253, "grad_norm": 17.052818298339844, "learning_rate": 9.102912621359224e-05, "loss": 0.434, "step": 1848 }, { "epoch": 17.924096385542168, "grad_norm": 12.141855239868164, "learning_rate": 9.10242718446602e-05, "loss": 0.2146, "step": 1849 }, { "epoch": 17.933734939759034, "grad_norm": 34.0899772644043, "learning_rate": 9.101941747572816e-05, "loss": 0.4651, "step": 1850 }, { "epoch": 17.943373493975905, "grad_norm": 3.271799087524414, "learning_rate": 9.101456310679613e-05, "loss": 0.2982, "step": 1851 }, { "epoch": 17.95301204819277, "grad_norm": 8.938309669494629, "learning_rate": 9.100970873786409e-05, "loss": 0.3489, "step": 1852 }, { "epoch": 17.96265060240964, "grad_norm": 12.928946495056152, "learning_rate": 9.100485436893205e-05, "loss": 0.3894, "step": 1853 }, { "epoch": 17.972289156626506, "grad_norm": 9.481563568115234, "learning_rate": 9.1e-05, "loss": 0.3601, "step": 1854 }, { "epoch": 17.981927710843372, "grad_norm": 6.112974166870117, "learning_rate": 9.099514563106797e-05, "loss": 0.256, "step": 1855 }, { "epoch": 17.99156626506024, "grad_norm": 8.31019401550293, "learning_rate": 9.099029126213593e-05, "loss": 0.269, "step": 1856 }, { "epoch": 18.00722891566265, "grad_norm": 5.553253650665283, "learning_rate": 9.098543689320389e-05, "loss": 0.3395, "step": 1857 }, { "epoch": 18.016867469879518, "grad_norm": 14.393214225769043, "learning_rate": 9.098058252427185e-05, "loss": 0.4203, "step": 1858 }, { "epoch": 18.026506024096385, "grad_norm": 6.325401306152344, "learning_rate": 9.097572815533982e-05, "loss": 0.2488, "step": 1859 }, { "epoch": 18.03614457831325, "grad_norm": 22.933767318725586, "learning_rate": 9.097087378640778e-05, "loss": 0.3659, "step": 1860 }, { "epoch": 18.045783132530122, "grad_norm": 13.049522399902344, "learning_rate": 9.096601941747573e-05, "loss": 0.2861, "step": 1861 }, { "epoch": 18.05542168674699, "grad_norm": 9.371243476867676, "learning_rate": 9.096116504854369e-05, "loss": 0.2949, "step": 1862 }, { "epoch": 18.065060240963856, "grad_norm": 20.76903533935547, "learning_rate": 9.095631067961165e-05, "loss": 0.6277, "step": 1863 }, { "epoch": 18.074698795180723, "grad_norm": 7.206963539123535, "learning_rate": 9.095145631067962e-05, "loss": 0.4055, "step": 1864 }, { "epoch": 18.08433734939759, "grad_norm": 10.071020126342773, "learning_rate": 9.094660194174758e-05, "loss": 0.2779, "step": 1865 }, { "epoch": 18.093975903614457, "grad_norm": 8.644139289855957, "learning_rate": 9.094174757281554e-05, "loss": 0.3883, "step": 1866 }, { "epoch": 18.103614457831327, "grad_norm": 8.404914855957031, "learning_rate": 9.09368932038835e-05, "loss": 0.45, "step": 1867 }, { "epoch": 18.113253012048194, "grad_norm": 10.275274276733398, "learning_rate": 9.093203883495147e-05, "loss": 0.2849, "step": 1868 }, { "epoch": 18.12289156626506, "grad_norm": 12.168214797973633, "learning_rate": 9.092718446601942e-05, "loss": 0.4712, "step": 1869 }, { "epoch": 18.132530120481928, "grad_norm": 5.675642967224121, "learning_rate": 9.092233009708738e-05, "loss": 0.3111, "step": 1870 }, { "epoch": 18.142168674698794, "grad_norm": 9.66857624053955, "learning_rate": 9.091747572815534e-05, "loss": 0.3459, "step": 1871 }, { "epoch": 18.15180722891566, "grad_norm": 6.33837890625, "learning_rate": 9.091262135922331e-05, "loss": 0.2963, "step": 1872 }, { "epoch": 18.16144578313253, "grad_norm": 39.2539176940918, "learning_rate": 9.090776699029127e-05, "loss": 0.4131, "step": 1873 }, { "epoch": 18.1710843373494, "grad_norm": 2.0258982181549072, "learning_rate": 9.090291262135923e-05, "loss": 0.4199, "step": 1874 }, { "epoch": 18.180722891566266, "grad_norm": 6.382338523864746, "learning_rate": 9.089805825242718e-05, "loss": 0.3898, "step": 1875 }, { "epoch": 18.190361445783132, "grad_norm": 1.2065387964248657, "learning_rate": 9.089320388349515e-05, "loss": 0.3324, "step": 1876 }, { "epoch": 18.2, "grad_norm": 13.835840225219727, "learning_rate": 9.088834951456311e-05, "loss": 0.3927, "step": 1877 }, { "epoch": 18.209638554216866, "grad_norm": 6.520798683166504, "learning_rate": 9.088349514563107e-05, "loss": 0.3915, "step": 1878 }, { "epoch": 18.219277108433737, "grad_norm": 4.9430317878723145, "learning_rate": 9.087864077669903e-05, "loss": 0.4984, "step": 1879 }, { "epoch": 18.228915662650603, "grad_norm": 3.80684494972229, "learning_rate": 9.087378640776699e-05, "loss": 0.3194, "step": 1880 }, { "epoch": 18.23855421686747, "grad_norm": 7.060375213623047, "learning_rate": 9.086893203883496e-05, "loss": 0.4163, "step": 1881 }, { "epoch": 18.248192771084337, "grad_norm": 10.866677284240723, "learning_rate": 9.086407766990291e-05, "loss": 0.5033, "step": 1882 }, { "epoch": 18.257831325301204, "grad_norm": 46.77412033081055, "learning_rate": 9.085922330097087e-05, "loss": 0.5944, "step": 1883 }, { "epoch": 18.26746987951807, "grad_norm": 2.983064889907837, "learning_rate": 9.085436893203883e-05, "loss": 0.2176, "step": 1884 }, { "epoch": 18.27710843373494, "grad_norm": 11.164031982421875, "learning_rate": 9.08495145631068e-05, "loss": 0.4072, "step": 1885 }, { "epoch": 18.28674698795181, "grad_norm": 16.151260375976562, "learning_rate": 9.084466019417476e-05, "loss": 0.3656, "step": 1886 }, { "epoch": 18.296385542168675, "grad_norm": 20.2600040435791, "learning_rate": 9.083980582524272e-05, "loss": 0.4059, "step": 1887 }, { "epoch": 18.306024096385542, "grad_norm": 10.514196395874023, "learning_rate": 9.083495145631068e-05, "loss": 0.3394, "step": 1888 }, { "epoch": 18.31566265060241, "grad_norm": 10.537528991699219, "learning_rate": 9.083009708737865e-05, "loss": 0.3004, "step": 1889 }, { "epoch": 18.325301204819276, "grad_norm": 13.449474334716797, "learning_rate": 9.08252427184466e-05, "loss": 0.3875, "step": 1890 }, { "epoch": 18.334939759036146, "grad_norm": 4.013101577758789, "learning_rate": 9.082038834951456e-05, "loss": 0.2805, "step": 1891 }, { "epoch": 18.344578313253013, "grad_norm": 37.50716781616211, "learning_rate": 9.081553398058252e-05, "loss": 0.457, "step": 1892 }, { "epoch": 18.35421686746988, "grad_norm": 28.388357162475586, "learning_rate": 9.081067961165049e-05, "loss": 0.4566, "step": 1893 }, { "epoch": 18.363855421686747, "grad_norm": 2.9774982929229736, "learning_rate": 9.080582524271846e-05, "loss": 0.575, "step": 1894 }, { "epoch": 18.373493975903614, "grad_norm": 2.5393354892730713, "learning_rate": 9.080097087378642e-05, "loss": 0.3042, "step": 1895 }, { "epoch": 18.38313253012048, "grad_norm": 18.28884506225586, "learning_rate": 9.079611650485438e-05, "loss": 0.4862, "step": 1896 }, { "epoch": 18.39277108433735, "grad_norm": 5.787408828735352, "learning_rate": 9.079126213592234e-05, "loss": 0.506, "step": 1897 }, { "epoch": 18.402409638554218, "grad_norm": 14.960709571838379, "learning_rate": 9.078640776699031e-05, "loss": 0.2145, "step": 1898 }, { "epoch": 18.412048192771085, "grad_norm": 3.6574065685272217, "learning_rate": 9.078155339805826e-05, "loss": 0.4682, "step": 1899 }, { "epoch": 18.42168674698795, "grad_norm": 1.1072618961334229, "learning_rate": 9.077669902912622e-05, "loss": 0.1166, "step": 1900 }, { "epoch": 18.43132530120482, "grad_norm": 6.3251471519470215, "learning_rate": 9.077184466019418e-05, "loss": 0.2095, "step": 1901 }, { "epoch": 18.440963855421685, "grad_norm": 3.7630934715270996, "learning_rate": 9.076699029126214e-05, "loss": 0.3435, "step": 1902 }, { "epoch": 18.450602409638556, "grad_norm": 8.531841278076172, "learning_rate": 9.076213592233011e-05, "loss": 0.2526, "step": 1903 }, { "epoch": 18.460240963855423, "grad_norm": 21.853050231933594, "learning_rate": 9.075728155339807e-05, "loss": 0.3199, "step": 1904 }, { "epoch": 18.46987951807229, "grad_norm": 9.799336433410645, "learning_rate": 9.075242718446602e-05, "loss": 0.3571, "step": 1905 }, { "epoch": 18.479518072289157, "grad_norm": 1.8725082874298096, "learning_rate": 9.074757281553398e-05, "loss": 0.2828, "step": 1906 }, { "epoch": 18.489156626506023, "grad_norm": 17.509992599487305, "learning_rate": 9.074271844660195e-05, "loss": 0.404, "step": 1907 }, { "epoch": 18.49879518072289, "grad_norm": 6.285500526428223, "learning_rate": 9.073786407766991e-05, "loss": 0.4486, "step": 1908 }, { "epoch": 18.50843373493976, "grad_norm": 4.551450252532959, "learning_rate": 9.073300970873787e-05, "loss": 0.2775, "step": 1909 }, { "epoch": 18.518072289156628, "grad_norm": 2.975679636001587, "learning_rate": 9.072815533980583e-05, "loss": 0.2438, "step": 1910 }, { "epoch": 18.527710843373494, "grad_norm": 7.11352014541626, "learning_rate": 9.07233009708738e-05, "loss": 0.523, "step": 1911 }, { "epoch": 18.53734939759036, "grad_norm": 7.0740203857421875, "learning_rate": 9.071844660194176e-05, "loss": 0.3704, "step": 1912 }, { "epoch": 18.54698795180723, "grad_norm": 18.796663284301758, "learning_rate": 9.071359223300971e-05, "loss": 0.3185, "step": 1913 }, { "epoch": 18.556626506024095, "grad_norm": 33.88198471069336, "learning_rate": 9.070873786407767e-05, "loss": 0.5855, "step": 1914 }, { "epoch": 18.566265060240966, "grad_norm": 13.77143669128418, "learning_rate": 9.070388349514564e-05, "loss": 0.3843, "step": 1915 }, { "epoch": 18.575903614457832, "grad_norm": 13.71727466583252, "learning_rate": 9.06990291262136e-05, "loss": 0.5519, "step": 1916 }, { "epoch": 18.5855421686747, "grad_norm": 22.093868255615234, "learning_rate": 9.069417475728156e-05, "loss": 0.2695, "step": 1917 }, { "epoch": 18.595180722891566, "grad_norm": 4.121679782867432, "learning_rate": 9.068932038834952e-05, "loss": 0.3273, "step": 1918 }, { "epoch": 18.604819277108433, "grad_norm": 24.8153018951416, "learning_rate": 9.068446601941747e-05, "loss": 0.3171, "step": 1919 }, { "epoch": 18.6144578313253, "grad_norm": 41.17964172363281, "learning_rate": 9.067961165048545e-05, "loss": 0.2792, "step": 1920 }, { "epoch": 18.62409638554217, "grad_norm": 15.710625648498535, "learning_rate": 9.06747572815534e-05, "loss": 0.3418, "step": 1921 }, { "epoch": 18.633734939759037, "grad_norm": 23.154584884643555, "learning_rate": 9.066990291262136e-05, "loss": 0.7425, "step": 1922 }, { "epoch": 18.643373493975904, "grad_norm": 28.68702507019043, "learning_rate": 9.066504854368932e-05, "loss": 0.3451, "step": 1923 }, { "epoch": 18.65301204819277, "grad_norm": 28.581890106201172, "learning_rate": 9.066019417475729e-05, "loss": 0.2349, "step": 1924 }, { "epoch": 18.662650602409638, "grad_norm": 16.870763778686523, "learning_rate": 9.065533980582525e-05, "loss": 0.565, "step": 1925 }, { "epoch": 18.672289156626505, "grad_norm": 12.700529098510742, "learning_rate": 9.06504854368932e-05, "loss": 0.2898, "step": 1926 }, { "epoch": 18.681927710843375, "grad_norm": 18.58516502380371, "learning_rate": 9.064563106796116e-05, "loss": 0.328, "step": 1927 }, { "epoch": 18.691566265060242, "grad_norm": 7.8528547286987305, "learning_rate": 9.064077669902913e-05, "loss": 0.2021, "step": 1928 }, { "epoch": 18.70120481927711, "grad_norm": 21.841501235961914, "learning_rate": 9.063592233009709e-05, "loss": 0.3236, "step": 1929 }, { "epoch": 18.710843373493976, "grad_norm": 18.069580078125, "learning_rate": 9.063106796116505e-05, "loss": 0.3234, "step": 1930 }, { "epoch": 18.720481927710843, "grad_norm": 6.404432773590088, "learning_rate": 9.062621359223301e-05, "loss": 0.1946, "step": 1931 }, { "epoch": 18.73012048192771, "grad_norm": 23.552276611328125, "learning_rate": 9.062135922330098e-05, "loss": 0.5843, "step": 1932 }, { "epoch": 18.739759036144576, "grad_norm": 6.160479545593262, "learning_rate": 9.061650485436894e-05, "loss": 0.2558, "step": 1933 }, { "epoch": 18.749397590361447, "grad_norm": 4.1931939125061035, "learning_rate": 9.06116504854369e-05, "loss": 0.356, "step": 1934 }, { "epoch": 18.759036144578314, "grad_norm": 32.8124885559082, "learning_rate": 9.060679611650485e-05, "loss": 0.6644, "step": 1935 }, { "epoch": 18.76867469879518, "grad_norm": 2.180572986602783, "learning_rate": 9.060194174757281e-05, "loss": 0.3825, "step": 1936 }, { "epoch": 18.778313253012048, "grad_norm": 17.41013526916504, "learning_rate": 9.059708737864078e-05, "loss": 0.3782, "step": 1937 }, { "epoch": 18.787951807228914, "grad_norm": 69.6205062866211, "learning_rate": 9.059223300970874e-05, "loss": 0.452, "step": 1938 }, { "epoch": 18.797590361445785, "grad_norm": 5.008017539978027, "learning_rate": 9.05873786407767e-05, "loss": 0.3561, "step": 1939 }, { "epoch": 18.80722891566265, "grad_norm": 106.7082748413086, "learning_rate": 9.058252427184465e-05, "loss": 0.6522, "step": 1940 }, { "epoch": 18.81686746987952, "grad_norm": 5.262181282043457, "learning_rate": 9.057766990291263e-05, "loss": 0.2628, "step": 1941 }, { "epoch": 18.826506024096386, "grad_norm": 4.358756065368652, "learning_rate": 9.057281553398058e-05, "loss": 0.0669, "step": 1942 }, { "epoch": 18.836144578313252, "grad_norm": 7.816954612731934, "learning_rate": 9.056796116504854e-05, "loss": 0.4757, "step": 1943 }, { "epoch": 18.84578313253012, "grad_norm": 13.151540756225586, "learning_rate": 9.056310679611651e-05, "loss": 0.3077, "step": 1944 }, { "epoch": 18.855421686746986, "grad_norm": 27.509767532348633, "learning_rate": 9.055825242718447e-05, "loss": 0.3335, "step": 1945 }, { "epoch": 18.865060240963857, "grad_norm": 749.2785034179688, "learning_rate": 9.055339805825244e-05, "loss": 0.3976, "step": 1946 }, { "epoch": 18.874698795180723, "grad_norm": 7.60486364364624, "learning_rate": 9.05485436893204e-05, "loss": 0.157, "step": 1947 }, { "epoch": 18.88433734939759, "grad_norm": 6.721808910369873, "learning_rate": 9.054368932038836e-05, "loss": 0.2945, "step": 1948 }, { "epoch": 18.893975903614457, "grad_norm": 31.006450653076172, "learning_rate": 9.053883495145632e-05, "loss": 0.4307, "step": 1949 }, { "epoch": 18.903614457831324, "grad_norm": 5.3680572509765625, "learning_rate": 9.053398058252429e-05, "loss": 0.2525, "step": 1950 }, { "epoch": 18.913253012048195, "grad_norm": 21.249160766601562, "learning_rate": 9.052912621359224e-05, "loss": 0.3486, "step": 1951 }, { "epoch": 18.92289156626506, "grad_norm": 12.320511817932129, "learning_rate": 9.05242718446602e-05, "loss": 0.3021, "step": 1952 }, { "epoch": 18.93253012048193, "grad_norm": 20.454425811767578, "learning_rate": 9.051941747572816e-05, "loss": 0.4251, "step": 1953 }, { "epoch": 18.942168674698795, "grad_norm": 2.127168655395508, "learning_rate": 9.051456310679612e-05, "loss": 0.2793, "step": 1954 }, { "epoch": 18.951807228915662, "grad_norm": 6.932311534881592, "learning_rate": 9.050970873786409e-05, "loss": 0.2254, "step": 1955 }, { "epoch": 18.96144578313253, "grad_norm": 11.392114639282227, "learning_rate": 9.050485436893205e-05, "loss": 0.2795, "step": 1956 }, { "epoch": 18.971084337349396, "grad_norm": 6.458676338195801, "learning_rate": 9.05e-05, "loss": 0.2771, "step": 1957 }, { "epoch": 18.980722891566266, "grad_norm": 6.8888702392578125, "learning_rate": 9.049514563106796e-05, "loss": 0.2866, "step": 1958 }, { "epoch": 18.990361445783133, "grad_norm": 25.7474422454834, "learning_rate": 9.049029126213593e-05, "loss": 0.271, "step": 1959 }, { "epoch": 19.00602409638554, "grad_norm": 18.472780227661133, "learning_rate": 9.048543689320389e-05, "loss": 0.3828, "step": 1960 }, { "epoch": 19.01566265060241, "grad_norm": 8.302715301513672, "learning_rate": 9.048058252427185e-05, "loss": 0.5316, "step": 1961 }, { "epoch": 19.02530120481928, "grad_norm": 11.980663299560547, "learning_rate": 9.04757281553398e-05, "loss": 0.4566, "step": 1962 }, { "epoch": 19.034939759036146, "grad_norm": 1.617582082748413, "learning_rate": 9.047087378640778e-05, "loss": 0.1939, "step": 1963 }, { "epoch": 19.044578313253012, "grad_norm": 37.665584564208984, "learning_rate": 9.046601941747574e-05, "loss": 0.2976, "step": 1964 }, { "epoch": 19.05421686746988, "grad_norm": 40.813438415527344, "learning_rate": 9.04611650485437e-05, "loss": 0.5165, "step": 1965 }, { "epoch": 19.063855421686746, "grad_norm": 22.42719841003418, "learning_rate": 9.045631067961165e-05, "loss": 0.3491, "step": 1966 }, { "epoch": 19.073493975903613, "grad_norm": 7.657529354095459, "learning_rate": 9.045145631067962e-05, "loss": 0.3822, "step": 1967 }, { "epoch": 19.083132530120483, "grad_norm": 11.274700164794922, "learning_rate": 9.044660194174758e-05, "loss": 0.3889, "step": 1968 }, { "epoch": 19.09277108433735, "grad_norm": 23.001829147338867, "learning_rate": 9.044174757281554e-05, "loss": 0.4409, "step": 1969 }, { "epoch": 19.102409638554217, "grad_norm": 6.67434549331665, "learning_rate": 9.04368932038835e-05, "loss": 0.4214, "step": 1970 }, { "epoch": 19.112048192771084, "grad_norm": 15.106274604797363, "learning_rate": 9.043203883495145e-05, "loss": 0.2891, "step": 1971 }, { "epoch": 19.12168674698795, "grad_norm": 27.994667053222656, "learning_rate": 9.042718446601942e-05, "loss": 0.2828, "step": 1972 }, { "epoch": 19.131325301204818, "grad_norm": 22.340328216552734, "learning_rate": 9.042233009708738e-05, "loss": 0.1732, "step": 1973 }, { "epoch": 19.14096385542169, "grad_norm": 28.863956451416016, "learning_rate": 9.041747572815534e-05, "loss": 0.3714, "step": 1974 }, { "epoch": 19.150602409638555, "grad_norm": 3.518134593963623, "learning_rate": 9.04126213592233e-05, "loss": 0.2401, "step": 1975 }, { "epoch": 19.160240963855422, "grad_norm": 5.366899013519287, "learning_rate": 9.040776699029127e-05, "loss": 0.5159, "step": 1976 }, { "epoch": 19.16987951807229, "grad_norm": 143.35617065429688, "learning_rate": 9.040291262135923e-05, "loss": 0.4606, "step": 1977 }, { "epoch": 19.179518072289156, "grad_norm": 12.880592346191406, "learning_rate": 9.039805825242718e-05, "loss": 0.3043, "step": 1978 }, { "epoch": 19.189156626506023, "grad_norm": 26.5754451751709, "learning_rate": 9.039320388349514e-05, "loss": 0.4323, "step": 1979 }, { "epoch": 19.198795180722893, "grad_norm": 5.497648239135742, "learning_rate": 9.038834951456311e-05, "loss": 0.3875, "step": 1980 }, { "epoch": 19.20843373493976, "grad_norm": 23.419294357299805, "learning_rate": 9.038349514563107e-05, "loss": 0.4352, "step": 1981 }, { "epoch": 19.218072289156627, "grad_norm": 14.469138145446777, "learning_rate": 9.037864077669903e-05, "loss": 0.3827, "step": 1982 }, { "epoch": 19.227710843373494, "grad_norm": 6.852151393890381, "learning_rate": 9.037378640776699e-05, "loss": 0.3883, "step": 1983 }, { "epoch": 19.23734939759036, "grad_norm": 14.39210319519043, "learning_rate": 9.036893203883496e-05, "loss": 0.6146, "step": 1984 }, { "epoch": 19.246987951807228, "grad_norm": 6.397410869598389, "learning_rate": 9.036407766990292e-05, "loss": 0.2539, "step": 1985 }, { "epoch": 19.256626506024098, "grad_norm": 5.22564697265625, "learning_rate": 9.035922330097087e-05, "loss": 0.3862, "step": 1986 }, { "epoch": 19.266265060240965, "grad_norm": 14.262190818786621, "learning_rate": 9.035436893203883e-05, "loss": 0.2793, "step": 1987 }, { "epoch": 19.27590361445783, "grad_norm": 5.511660575866699, "learning_rate": 9.034951456310679e-05, "loss": 0.4268, "step": 1988 }, { "epoch": 19.2855421686747, "grad_norm": 6.0116286277771, "learning_rate": 9.034466019417476e-05, "loss": 0.3095, "step": 1989 }, { "epoch": 19.295180722891565, "grad_norm": 75.74551391601562, "learning_rate": 9.033980582524272e-05, "loss": 0.5076, "step": 1990 }, { "epoch": 19.304819277108432, "grad_norm": 5.390777111053467, "learning_rate": 9.033495145631068e-05, "loss": 0.2527, "step": 1991 }, { "epoch": 19.314457831325303, "grad_norm": 19.78178596496582, "learning_rate": 9.033009708737863e-05, "loss": 0.2911, "step": 1992 }, { "epoch": 19.32409638554217, "grad_norm": 15.074263572692871, "learning_rate": 9.03252427184466e-05, "loss": 0.4872, "step": 1993 }, { "epoch": 19.333734939759037, "grad_norm": 3.223311185836792, "learning_rate": 9.032038834951456e-05, "loss": 0.3167, "step": 1994 }, { "epoch": 19.343373493975903, "grad_norm": 25.175186157226562, "learning_rate": 9.031553398058253e-05, "loss": 0.4994, "step": 1995 }, { "epoch": 19.35301204819277, "grad_norm": 15.991721153259277, "learning_rate": 9.031067961165049e-05, "loss": 0.3914, "step": 1996 }, { "epoch": 19.362650602409637, "grad_norm": 29.90163230895996, "learning_rate": 9.030582524271845e-05, "loss": 0.4302, "step": 1997 }, { "epoch": 19.372289156626508, "grad_norm": 19.92572784423828, "learning_rate": 9.030097087378642e-05, "loss": 0.4308, "step": 1998 }, { "epoch": 19.381927710843375, "grad_norm": 31.25533676147461, "learning_rate": 9.029611650485438e-05, "loss": 0.5107, "step": 1999 }, { "epoch": 19.39156626506024, "grad_norm": 3.0810749530792236, "learning_rate": 9.029126213592234e-05, "loss": 0.1674, "step": 2000 }, { "epoch": 19.40120481927711, "grad_norm": 14.366427421569824, "learning_rate": 9.02864077669903e-05, "loss": 0.329, "step": 2001 }, { "epoch": 19.410843373493975, "grad_norm": 5.550936698913574, "learning_rate": 9.028155339805827e-05, "loss": 0.5086, "step": 2002 }, { "epoch": 19.420481927710842, "grad_norm": 23.085134506225586, "learning_rate": 9.027669902912622e-05, "loss": 0.3877, "step": 2003 }, { "epoch": 19.430120481927712, "grad_norm": 11.030743598937988, "learning_rate": 9.027184466019418e-05, "loss": 0.3401, "step": 2004 }, { "epoch": 19.43975903614458, "grad_norm": 19.53752326965332, "learning_rate": 9.026699029126214e-05, "loss": 0.4997, "step": 2005 }, { "epoch": 19.449397590361446, "grad_norm": 11.30378246307373, "learning_rate": 9.026213592233011e-05, "loss": 0.2654, "step": 2006 }, { "epoch": 19.459036144578313, "grad_norm": 28.366132736206055, "learning_rate": 9.025728155339807e-05, "loss": 0.4819, "step": 2007 }, { "epoch": 19.46867469879518, "grad_norm": 6.16321325302124, "learning_rate": 9.025242718446603e-05, "loss": 0.3545, "step": 2008 }, { "epoch": 19.478313253012047, "grad_norm": 9.346163749694824, "learning_rate": 9.024757281553398e-05, "loss": 0.2395, "step": 2009 }, { "epoch": 19.487951807228917, "grad_norm": 5.805694103240967, "learning_rate": 9.024271844660194e-05, "loss": 0.1134, "step": 2010 }, { "epoch": 19.497590361445784, "grad_norm": 3.2164974212646484, "learning_rate": 9.023786407766991e-05, "loss": 0.1286, "step": 2011 }, { "epoch": 19.50722891566265, "grad_norm": 3.163947343826294, "learning_rate": 9.023300970873787e-05, "loss": 0.2629, "step": 2012 }, { "epoch": 19.516867469879518, "grad_norm": 11.803749084472656, "learning_rate": 9.022815533980583e-05, "loss": 0.38, "step": 2013 }, { "epoch": 19.526506024096385, "grad_norm": 24.040760040283203, "learning_rate": 9.022330097087379e-05, "loss": 0.3806, "step": 2014 }, { "epoch": 19.53614457831325, "grad_norm": 5.5910139083862305, "learning_rate": 9.021844660194176e-05, "loss": 0.1435, "step": 2015 }, { "epoch": 19.545783132530122, "grad_norm": 4.220864772796631, "learning_rate": 9.021359223300972e-05, "loss": 0.3262, "step": 2016 }, { "epoch": 19.55542168674699, "grad_norm": 4.028323173522949, "learning_rate": 9.020873786407767e-05, "loss": 0.2912, "step": 2017 }, { "epoch": 19.565060240963856, "grad_norm": 10.990028381347656, "learning_rate": 9.020388349514563e-05, "loss": 0.3576, "step": 2018 }, { "epoch": 19.574698795180723, "grad_norm": 5.907679080963135, "learning_rate": 9.01990291262136e-05, "loss": 0.3235, "step": 2019 }, { "epoch": 19.58433734939759, "grad_norm": 4.594082832336426, "learning_rate": 9.019417475728156e-05, "loss": 0.1574, "step": 2020 }, { "epoch": 19.593975903614457, "grad_norm": 8.599654197692871, "learning_rate": 9.018932038834952e-05, "loss": 0.6652, "step": 2021 }, { "epoch": 19.603614457831327, "grad_norm": 8.49941349029541, "learning_rate": 9.018446601941748e-05, "loss": 0.378, "step": 2022 }, { "epoch": 19.613253012048194, "grad_norm": 16.5190486907959, "learning_rate": 9.017961165048545e-05, "loss": 0.3814, "step": 2023 }, { "epoch": 19.62289156626506, "grad_norm": 8.913263320922852, "learning_rate": 9.01747572815534e-05, "loss": 0.4697, "step": 2024 }, { "epoch": 19.632530120481928, "grad_norm": 6.766853332519531, "learning_rate": 9.016990291262136e-05, "loss": 0.7051, "step": 2025 }, { "epoch": 19.642168674698794, "grad_norm": 5.025972366333008, "learning_rate": 9.016504854368932e-05, "loss": 0.439, "step": 2026 }, { "epoch": 19.65180722891566, "grad_norm": 7.181129455566406, "learning_rate": 9.016019417475728e-05, "loss": 0.4974, "step": 2027 }, { "epoch": 19.66144578313253, "grad_norm": 18.580793380737305, "learning_rate": 9.015533980582525e-05, "loss": 0.3599, "step": 2028 }, { "epoch": 19.6710843373494, "grad_norm": 17.488819122314453, "learning_rate": 9.01504854368932e-05, "loss": 0.262, "step": 2029 }, { "epoch": 19.680722891566266, "grad_norm": 3.9519803524017334, "learning_rate": 9.014563106796116e-05, "loss": 0.3373, "step": 2030 }, { "epoch": 19.690361445783132, "grad_norm": 1.4464385509490967, "learning_rate": 9.014077669902912e-05, "loss": 0.0788, "step": 2031 }, { "epoch": 19.7, "grad_norm": 6.4420247077941895, "learning_rate": 9.01359223300971e-05, "loss": 0.2401, "step": 2032 }, { "epoch": 19.709638554216866, "grad_norm": 6.724270820617676, "learning_rate": 9.013106796116505e-05, "loss": 0.2325, "step": 2033 }, { "epoch": 19.719277108433737, "grad_norm": 18.577707290649414, "learning_rate": 9.012621359223301e-05, "loss": 0.3441, "step": 2034 }, { "epoch": 19.728915662650603, "grad_norm": 7.42727518081665, "learning_rate": 9.012135922330097e-05, "loss": 0.3215, "step": 2035 }, { "epoch": 19.73855421686747, "grad_norm": 39.625003814697266, "learning_rate": 9.011650485436894e-05, "loss": 0.3301, "step": 2036 }, { "epoch": 19.748192771084337, "grad_norm": 13.985305786132812, "learning_rate": 9.01116504854369e-05, "loss": 0.3046, "step": 2037 }, { "epoch": 19.757831325301204, "grad_norm": 5.898560523986816, "learning_rate": 9.010679611650485e-05, "loss": 0.5494, "step": 2038 }, { "epoch": 19.76746987951807, "grad_norm": 7.968297004699707, "learning_rate": 9.010194174757281e-05, "loss": 0.4292, "step": 2039 }, { "epoch": 19.77710843373494, "grad_norm": 8.065617561340332, "learning_rate": 9.009708737864078e-05, "loss": 0.3769, "step": 2040 }, { "epoch": 19.78674698795181, "grad_norm": 49.91721725463867, "learning_rate": 9.009223300970874e-05, "loss": 0.5272, "step": 2041 }, { "epoch": 19.796385542168675, "grad_norm": 4.002260208129883, "learning_rate": 9.00873786407767e-05, "loss": 0.4089, "step": 2042 }, { "epoch": 19.806024096385542, "grad_norm": 2.1274662017822266, "learning_rate": 9.008252427184466e-05, "loss": 0.1801, "step": 2043 }, { "epoch": 19.81566265060241, "grad_norm": 9.956897735595703, "learning_rate": 9.007766990291261e-05, "loss": 0.373, "step": 2044 }, { "epoch": 19.825301204819276, "grad_norm": 10.411615371704102, "learning_rate": 9.00728155339806e-05, "loss": 0.1868, "step": 2045 }, { "epoch": 19.834939759036146, "grad_norm": 23.378753662109375, "learning_rate": 9.006796116504856e-05, "loss": 0.4384, "step": 2046 }, { "epoch": 19.844578313253013, "grad_norm": 3.7979390621185303, "learning_rate": 9.006310679611651e-05, "loss": 0.1835, "step": 2047 }, { "epoch": 19.85421686746988, "grad_norm": 13.939805030822754, "learning_rate": 9.005825242718447e-05, "loss": 0.1341, "step": 2048 }, { "epoch": 19.863855421686747, "grad_norm": 12.882235527038574, "learning_rate": 9.005339805825243e-05, "loss": 0.5124, "step": 2049 }, { "epoch": 19.873493975903614, "grad_norm": 11.67038345336914, "learning_rate": 9.00485436893204e-05, "loss": 0.2942, "step": 2050 }, { "epoch": 19.88313253012048, "grad_norm": 28.39166259765625, "learning_rate": 9.004368932038836e-05, "loss": 0.605, "step": 2051 }, { "epoch": 19.89277108433735, "grad_norm": 10.78203010559082, "learning_rate": 9.003883495145632e-05, "loss": 0.3776, "step": 2052 }, { "epoch": 19.902409638554218, "grad_norm": 34.65802001953125, "learning_rate": 9.003398058252427e-05, "loss": 0.3788, "step": 2053 }, { "epoch": 19.912048192771085, "grad_norm": 7.280889987945557, "learning_rate": 9.002912621359225e-05, "loss": 0.4469, "step": 2054 }, { "epoch": 19.92168674698795, "grad_norm": 4.571901798248291, "learning_rate": 9.00242718446602e-05, "loss": 0.4164, "step": 2055 }, { "epoch": 19.93132530120482, "grad_norm": 8.124049186706543, "learning_rate": 9.001941747572816e-05, "loss": 0.2421, "step": 2056 }, { "epoch": 19.940963855421685, "grad_norm": 220.3855743408203, "learning_rate": 9.001456310679612e-05, "loss": 0.2864, "step": 2057 }, { "epoch": 19.950602409638556, "grad_norm": 12.107050895690918, "learning_rate": 9.000970873786409e-05, "loss": 0.508, "step": 2058 }, { "epoch": 19.960240963855423, "grad_norm": 2.2612123489379883, "learning_rate": 9.000485436893205e-05, "loss": 0.1257, "step": 2059 }, { "epoch": 19.96987951807229, "grad_norm": 22.08935546875, "learning_rate": 9e-05, "loss": 0.5618, "step": 2060 }, { "epoch": 19.979518072289157, "grad_norm": 14.696246147155762, "learning_rate": 8.999514563106796e-05, "loss": 0.6447, "step": 2061 }, { "epoch": 19.989156626506023, "grad_norm": 13.525457382202148, "learning_rate": 8.999029126213593e-05, "loss": 0.4408, "step": 2062 }, { "epoch": 20.004819277108435, "grad_norm": 6.313667297363281, "learning_rate": 8.998543689320389e-05, "loss": 0.2673, "step": 2063 }, { "epoch": 20.014457831325302, "grad_norm": 57.4716796875, "learning_rate": 8.998058252427185e-05, "loss": 0.4401, "step": 2064 }, { "epoch": 20.02409638554217, "grad_norm": 49.74363327026367, "learning_rate": 8.997572815533981e-05, "loss": 0.5437, "step": 2065 }, { "epoch": 20.033734939759036, "grad_norm": 2.707872152328491, "learning_rate": 8.997087378640777e-05, "loss": 0.1266, "step": 2066 }, { "epoch": 20.043373493975903, "grad_norm": 14.717106819152832, "learning_rate": 8.996601941747574e-05, "loss": 0.4459, "step": 2067 }, { "epoch": 20.05301204819277, "grad_norm": 5.237403869628906, "learning_rate": 8.99611650485437e-05, "loss": 0.2179, "step": 2068 }, { "epoch": 20.06265060240964, "grad_norm": 3.2765393257141113, "learning_rate": 8.995631067961165e-05, "loss": 0.2654, "step": 2069 }, { "epoch": 20.072289156626507, "grad_norm": 35.97175598144531, "learning_rate": 8.995145631067961e-05, "loss": 0.3329, "step": 2070 }, { "epoch": 20.081927710843374, "grad_norm": 7.446468830108643, "learning_rate": 8.994660194174758e-05, "loss": 0.3155, "step": 2071 }, { "epoch": 20.09156626506024, "grad_norm": 8.87435531616211, "learning_rate": 8.994174757281554e-05, "loss": 0.5776, "step": 2072 }, { "epoch": 20.101204819277108, "grad_norm": 15.281319618225098, "learning_rate": 8.99368932038835e-05, "loss": 0.3508, "step": 2073 }, { "epoch": 20.110843373493974, "grad_norm": 13.582765579223633, "learning_rate": 8.993203883495145e-05, "loss": 0.2705, "step": 2074 }, { "epoch": 20.120481927710845, "grad_norm": 34.4488410949707, "learning_rate": 8.992718446601943e-05, "loss": 0.3741, "step": 2075 }, { "epoch": 20.13012048192771, "grad_norm": 4.219476699829102, "learning_rate": 8.992233009708738e-05, "loss": 0.2117, "step": 2076 }, { "epoch": 20.13975903614458, "grad_norm": 11.815693855285645, "learning_rate": 8.991747572815534e-05, "loss": 0.5489, "step": 2077 }, { "epoch": 20.149397590361446, "grad_norm": 4.045429706573486, "learning_rate": 8.99126213592233e-05, "loss": 0.2143, "step": 2078 }, { "epoch": 20.159036144578312, "grad_norm": 17.151836395263672, "learning_rate": 8.990776699029127e-05, "loss": 0.4617, "step": 2079 }, { "epoch": 20.16867469879518, "grad_norm": 5.078327178955078, "learning_rate": 8.990291262135923e-05, "loss": 0.3471, "step": 2080 }, { "epoch": 20.17831325301205, "grad_norm": 5.681358814239502, "learning_rate": 8.989805825242719e-05, "loss": 0.4164, "step": 2081 }, { "epoch": 20.187951807228917, "grad_norm": 6.489326477050781, "learning_rate": 8.989320388349514e-05, "loss": 0.5032, "step": 2082 }, { "epoch": 20.197590361445783, "grad_norm": 3.9624381065368652, "learning_rate": 8.98883495145631e-05, "loss": 0.4798, "step": 2083 }, { "epoch": 20.20722891566265, "grad_norm": 4.472262859344482, "learning_rate": 8.988349514563107e-05, "loss": 0.2876, "step": 2084 }, { "epoch": 20.216867469879517, "grad_norm": 15.631071090698242, "learning_rate": 8.987864077669903e-05, "loss": 0.2867, "step": 2085 }, { "epoch": 20.226506024096384, "grad_norm": 6.713472843170166, "learning_rate": 8.987378640776699e-05, "loss": 0.319, "step": 2086 }, { "epoch": 20.236144578313255, "grad_norm": 13.252341270446777, "learning_rate": 8.986893203883495e-05, "loss": 0.411, "step": 2087 }, { "epoch": 20.24578313253012, "grad_norm": 8.282427787780762, "learning_rate": 8.986407766990292e-05, "loss": 0.4476, "step": 2088 }, { "epoch": 20.25542168674699, "grad_norm": 9.330663681030273, "learning_rate": 8.985922330097088e-05, "loss": 0.2457, "step": 2089 }, { "epoch": 20.265060240963855, "grad_norm": 20.354583740234375, "learning_rate": 8.985436893203883e-05, "loss": 0.353, "step": 2090 }, { "epoch": 20.274698795180722, "grad_norm": 16.470325469970703, "learning_rate": 8.984951456310679e-05, "loss": 0.3147, "step": 2091 }, { "epoch": 20.28433734939759, "grad_norm": 7.428584098815918, "learning_rate": 8.984466019417476e-05, "loss": 0.209, "step": 2092 }, { "epoch": 20.29397590361446, "grad_norm": 34.26204299926758, "learning_rate": 8.983980582524272e-05, "loss": 0.3284, "step": 2093 }, { "epoch": 20.303614457831326, "grad_norm": 16.338970184326172, "learning_rate": 8.983495145631068e-05, "loss": 0.4896, "step": 2094 }, { "epoch": 20.313253012048193, "grad_norm": 5.914295196533203, "learning_rate": 8.983009708737864e-05, "loss": 0.4449, "step": 2095 }, { "epoch": 20.32289156626506, "grad_norm": 5.473682403564453, "learning_rate": 8.982524271844661e-05, "loss": 0.4096, "step": 2096 }, { "epoch": 20.332530120481927, "grad_norm": 5.35717248916626, "learning_rate": 8.982038834951458e-05, "loss": 0.3922, "step": 2097 }, { "epoch": 20.342168674698794, "grad_norm": 11.921834945678711, "learning_rate": 8.981553398058254e-05, "loss": 0.2477, "step": 2098 }, { "epoch": 20.351807228915664, "grad_norm": 13.274727821350098, "learning_rate": 8.98106796116505e-05, "loss": 0.4534, "step": 2099 }, { "epoch": 20.36144578313253, "grad_norm": 4.9146623611450195, "learning_rate": 8.980582524271845e-05, "loss": 0.33, "step": 2100 }, { "epoch": 20.371084337349398, "grad_norm": 3.6917977333068848, "learning_rate": 8.980097087378642e-05, "loss": 0.2117, "step": 2101 }, { "epoch": 20.380722891566265, "grad_norm": 16.08298110961914, "learning_rate": 8.979611650485438e-05, "loss": 0.4564, "step": 2102 }, { "epoch": 20.39036144578313, "grad_norm": 4.126115798950195, "learning_rate": 8.979126213592234e-05, "loss": 0.2782, "step": 2103 }, { "epoch": 20.4, "grad_norm": 27.18509292602539, "learning_rate": 8.97864077669903e-05, "loss": 0.5302, "step": 2104 }, { "epoch": 20.40963855421687, "grad_norm": 3.1277759075164795, "learning_rate": 8.978155339805825e-05, "loss": 0.2374, "step": 2105 }, { "epoch": 20.419277108433736, "grad_norm": 12.805584907531738, "learning_rate": 8.977669902912623e-05, "loss": 0.3671, "step": 2106 }, { "epoch": 20.428915662650603, "grad_norm": 6.6775102615356445, "learning_rate": 8.977184466019418e-05, "loss": 0.1878, "step": 2107 }, { "epoch": 20.43855421686747, "grad_norm": 8.192072868347168, "learning_rate": 8.976699029126214e-05, "loss": 0.4062, "step": 2108 }, { "epoch": 20.448192771084337, "grad_norm": 10.663721084594727, "learning_rate": 8.97621359223301e-05, "loss": 0.1325, "step": 2109 }, { "epoch": 20.457831325301203, "grad_norm": 5.270077705383301, "learning_rate": 8.975728155339807e-05, "loss": 0.3945, "step": 2110 }, { "epoch": 20.467469879518074, "grad_norm": 3.0829622745513916, "learning_rate": 8.975242718446603e-05, "loss": 0.395, "step": 2111 }, { "epoch": 20.47710843373494, "grad_norm": 13.568376541137695, "learning_rate": 8.974757281553399e-05, "loss": 0.4273, "step": 2112 }, { "epoch": 20.486746987951808, "grad_norm": 11.87803840637207, "learning_rate": 8.974271844660194e-05, "loss": 0.4752, "step": 2113 }, { "epoch": 20.496385542168674, "grad_norm": 10.182350158691406, "learning_rate": 8.973786407766991e-05, "loss": 0.3748, "step": 2114 }, { "epoch": 20.50602409638554, "grad_norm": 19.21529769897461, "learning_rate": 8.973300970873787e-05, "loss": 0.3873, "step": 2115 }, { "epoch": 20.51566265060241, "grad_norm": 2.8993654251098633, "learning_rate": 8.972815533980583e-05, "loss": 0.2157, "step": 2116 }, { "epoch": 20.52530120481928, "grad_norm": 18.183183670043945, "learning_rate": 8.972330097087379e-05, "loss": 0.3571, "step": 2117 }, { "epoch": 20.534939759036146, "grad_norm": 10.602683067321777, "learning_rate": 8.971844660194176e-05, "loss": 0.2458, "step": 2118 }, { "epoch": 20.544578313253012, "grad_norm": 6.177969455718994, "learning_rate": 8.971359223300972e-05, "loss": 0.1321, "step": 2119 }, { "epoch": 20.55421686746988, "grad_norm": 16.6613826751709, "learning_rate": 8.970873786407767e-05, "loss": 0.2728, "step": 2120 }, { "epoch": 20.563855421686746, "grad_norm": 33.43659973144531, "learning_rate": 8.970388349514563e-05, "loss": 0.6706, "step": 2121 }, { "epoch": 20.573493975903613, "grad_norm": 13.2073974609375, "learning_rate": 8.969902912621359e-05, "loss": 0.4877, "step": 2122 }, { "epoch": 20.583132530120483, "grad_norm": 7.783329010009766, "learning_rate": 8.969417475728156e-05, "loss": 0.4537, "step": 2123 }, { "epoch": 20.59277108433735, "grad_norm": 22.0834903717041, "learning_rate": 8.968932038834952e-05, "loss": 0.4055, "step": 2124 }, { "epoch": 20.602409638554217, "grad_norm": 23.522777557373047, "learning_rate": 8.968446601941748e-05, "loss": 0.4189, "step": 2125 }, { "epoch": 20.612048192771084, "grad_norm": 2.235022783279419, "learning_rate": 8.967961165048543e-05, "loss": 0.1559, "step": 2126 }, { "epoch": 20.62168674698795, "grad_norm": 4.410442352294922, "learning_rate": 8.96747572815534e-05, "loss": 0.329, "step": 2127 }, { "epoch": 20.631325301204818, "grad_norm": 13.057551383972168, "learning_rate": 8.966990291262136e-05, "loss": 0.5143, "step": 2128 }, { "epoch": 20.64096385542169, "grad_norm": 35.374114990234375, "learning_rate": 8.966504854368932e-05, "loss": 0.6361, "step": 2129 }, { "epoch": 20.650602409638555, "grad_norm": 11.492179870605469, "learning_rate": 8.966019417475728e-05, "loss": 0.3624, "step": 2130 }, { "epoch": 20.660240963855422, "grad_norm": 42.179969787597656, "learning_rate": 8.965533980582525e-05, "loss": 0.8115, "step": 2131 }, { "epoch": 20.66987951807229, "grad_norm": 29.95408821105957, "learning_rate": 8.965048543689321e-05, "loss": 0.5923, "step": 2132 }, { "epoch": 20.679518072289156, "grad_norm": 12.665157318115234, "learning_rate": 8.964563106796117e-05, "loss": 0.6326, "step": 2133 }, { "epoch": 20.689156626506023, "grad_norm": 9.554508209228516, "learning_rate": 8.964077669902912e-05, "loss": 0.5554, "step": 2134 }, { "epoch": 20.698795180722893, "grad_norm": 8.493043899536133, "learning_rate": 8.96359223300971e-05, "loss": 0.4099, "step": 2135 }, { "epoch": 20.70843373493976, "grad_norm": 7.2256340980529785, "learning_rate": 8.963106796116505e-05, "loss": 0.2534, "step": 2136 }, { "epoch": 20.718072289156627, "grad_norm": 18.125925064086914, "learning_rate": 8.962621359223301e-05, "loss": 0.3869, "step": 2137 }, { "epoch": 20.727710843373494, "grad_norm": 19.75555419921875, "learning_rate": 8.962135922330097e-05, "loss": 0.3716, "step": 2138 }, { "epoch": 20.73734939759036, "grad_norm": 8.211896896362305, "learning_rate": 8.961650485436893e-05, "loss": 0.3678, "step": 2139 }, { "epoch": 20.746987951807228, "grad_norm": 10.074129104614258, "learning_rate": 8.96116504854369e-05, "loss": 0.6513, "step": 2140 }, { "epoch": 20.756626506024098, "grad_norm": 13.692898750305176, "learning_rate": 8.960679611650486e-05, "loss": 0.6182, "step": 2141 }, { "epoch": 20.766265060240965, "grad_norm": 4.157324314117432, "learning_rate": 8.960194174757281e-05, "loss": 0.2583, "step": 2142 }, { "epoch": 20.77590361445783, "grad_norm": 7.400536060333252, "learning_rate": 8.959708737864077e-05, "loss": 0.3967, "step": 2143 }, { "epoch": 20.7855421686747, "grad_norm": 8.679194450378418, "learning_rate": 8.959223300970874e-05, "loss": 0.3926, "step": 2144 }, { "epoch": 20.795180722891565, "grad_norm": 4.800831317901611, "learning_rate": 8.95873786407767e-05, "loss": 0.2955, "step": 2145 }, { "epoch": 20.804819277108432, "grad_norm": 1.8468519449234009, "learning_rate": 8.958252427184467e-05, "loss": 0.3872, "step": 2146 }, { "epoch": 20.814457831325303, "grad_norm": 5.510279178619385, "learning_rate": 8.957766990291263e-05, "loss": 0.3256, "step": 2147 }, { "epoch": 20.82409638554217, "grad_norm": 5.987325191497803, "learning_rate": 8.957281553398059e-05, "loss": 0.2461, "step": 2148 }, { "epoch": 20.833734939759037, "grad_norm": 3.479517698287964, "learning_rate": 8.956796116504856e-05, "loss": 0.2225, "step": 2149 }, { "epoch": 20.843373493975903, "grad_norm": 8.962663650512695, "learning_rate": 8.956310679611652e-05, "loss": 0.234, "step": 2150 }, { "epoch": 20.85301204819277, "grad_norm": 13.49216365814209, "learning_rate": 8.955825242718447e-05, "loss": 0.567, "step": 2151 }, { "epoch": 20.862650602409637, "grad_norm": 9.38683795928955, "learning_rate": 8.955339805825243e-05, "loss": 0.2522, "step": 2152 }, { "epoch": 20.872289156626508, "grad_norm": 16.17287826538086, "learning_rate": 8.95485436893204e-05, "loss": 0.5115, "step": 2153 }, { "epoch": 20.881927710843375, "grad_norm": 5.279826641082764, "learning_rate": 8.954368932038836e-05, "loss": 0.3347, "step": 2154 }, { "epoch": 20.89156626506024, "grad_norm": 21.29086685180664, "learning_rate": 8.953883495145632e-05, "loss": 0.1678, "step": 2155 }, { "epoch": 20.90120481927711, "grad_norm": 33.83600616455078, "learning_rate": 8.953398058252428e-05, "loss": 0.4382, "step": 2156 }, { "epoch": 20.910843373493975, "grad_norm": 25.58072280883789, "learning_rate": 8.952912621359225e-05, "loss": 0.2428, "step": 2157 }, { "epoch": 20.920481927710842, "grad_norm": 12.532552719116211, "learning_rate": 8.95242718446602e-05, "loss": 0.6308, "step": 2158 }, { "epoch": 20.930120481927712, "grad_norm": 4.72653341293335, "learning_rate": 8.951941747572816e-05, "loss": 0.4698, "step": 2159 }, { "epoch": 20.93975903614458, "grad_norm": 41.163902282714844, "learning_rate": 8.951456310679612e-05, "loss": 0.4512, "step": 2160 }, { "epoch": 20.949397590361446, "grad_norm": 7.927235126495361, "learning_rate": 8.950970873786408e-05, "loss": 0.385, "step": 2161 }, { "epoch": 20.959036144578313, "grad_norm": 9.06966495513916, "learning_rate": 8.950485436893205e-05, "loss": 0.3516, "step": 2162 }, { "epoch": 20.96867469879518, "grad_norm": 10.662839889526367, "learning_rate": 8.950000000000001e-05, "loss": 0.4908, "step": 2163 }, { "epoch": 20.978313253012047, "grad_norm": 4.485381126403809, "learning_rate": 8.949514563106796e-05, "loss": 0.2218, "step": 2164 }, { "epoch": 20.987951807228917, "grad_norm": 4.789339065551758, "learning_rate": 8.949029126213592e-05, "loss": 0.3733, "step": 2165 }, { "epoch": 21.003614457831326, "grad_norm": 7.369547367095947, "learning_rate": 8.94854368932039e-05, "loss": 0.3951, "step": 2166 }, { "epoch": 21.013253012048192, "grad_norm": 3.4512884616851807, "learning_rate": 8.948058252427185e-05, "loss": 0.2573, "step": 2167 }, { "epoch": 21.02289156626506, "grad_norm": 3.4308736324310303, "learning_rate": 8.947572815533981e-05, "loss": 0.2412, "step": 2168 }, { "epoch": 21.032530120481926, "grad_norm": 10.098041534423828, "learning_rate": 8.947087378640777e-05, "loss": 0.3049, "step": 2169 }, { "epoch": 21.042168674698797, "grad_norm": 9.963509559631348, "learning_rate": 8.946601941747574e-05, "loss": 0.275, "step": 2170 }, { "epoch": 21.051807228915663, "grad_norm": 11.161055564880371, "learning_rate": 8.94611650485437e-05, "loss": 0.3057, "step": 2171 }, { "epoch": 21.06144578313253, "grad_norm": 5.782242774963379, "learning_rate": 8.945631067961165e-05, "loss": 0.3176, "step": 2172 }, { "epoch": 21.071084337349397, "grad_norm": 7.302538871765137, "learning_rate": 8.945145631067961e-05, "loss": 0.1553, "step": 2173 }, { "epoch": 21.080722891566264, "grad_norm": 8.209023475646973, "learning_rate": 8.944660194174758e-05, "loss": 0.2805, "step": 2174 }, { "epoch": 21.09036144578313, "grad_norm": 15.56667709350586, "learning_rate": 8.944174757281554e-05, "loss": 0.3436, "step": 2175 }, { "epoch": 21.1, "grad_norm": 9.45315170288086, "learning_rate": 8.94368932038835e-05, "loss": 0.3594, "step": 2176 }, { "epoch": 21.10963855421687, "grad_norm": 3.2156314849853516, "learning_rate": 8.943203883495146e-05, "loss": 0.2686, "step": 2177 }, { "epoch": 21.119277108433735, "grad_norm": 7.794544219970703, "learning_rate": 8.942718446601941e-05, "loss": 0.3158, "step": 2178 }, { "epoch": 21.128915662650602, "grad_norm": 58.98945236206055, "learning_rate": 8.942233009708739e-05, "loss": 0.5345, "step": 2179 }, { "epoch": 21.13855421686747, "grad_norm": 5.813533782958984, "learning_rate": 8.941747572815534e-05, "loss": 0.1938, "step": 2180 }, { "epoch": 21.148192771084336, "grad_norm": 4.733443737030029, "learning_rate": 8.94126213592233e-05, "loss": 0.1925, "step": 2181 }, { "epoch": 21.157831325301206, "grad_norm": 6.682873249053955, "learning_rate": 8.940776699029126e-05, "loss": 0.2584, "step": 2182 }, { "epoch": 21.167469879518073, "grad_norm": 9.992301940917969, "learning_rate": 8.940291262135923e-05, "loss": 0.2758, "step": 2183 }, { "epoch": 21.17710843373494, "grad_norm": 9.687032699584961, "learning_rate": 8.939805825242719e-05, "loss": 0.2601, "step": 2184 }, { "epoch": 21.186746987951807, "grad_norm": 5.105895519256592, "learning_rate": 8.939320388349515e-05, "loss": 0.252, "step": 2185 }, { "epoch": 21.196385542168674, "grad_norm": 18.092496871948242, "learning_rate": 8.93883495145631e-05, "loss": 0.4009, "step": 2186 }, { "epoch": 21.20602409638554, "grad_norm": 4.396970272064209, "learning_rate": 8.938349514563107e-05, "loss": 0.2001, "step": 2187 }, { "epoch": 21.21566265060241, "grad_norm": 17.16271209716797, "learning_rate": 8.937864077669903e-05, "loss": 0.1916, "step": 2188 }, { "epoch": 21.225301204819278, "grad_norm": 3.1743574142456055, "learning_rate": 8.937378640776699e-05, "loss": 0.291, "step": 2189 }, { "epoch": 21.234939759036145, "grad_norm": 3.1832306385040283, "learning_rate": 8.936893203883495e-05, "loss": 0.2823, "step": 2190 }, { "epoch": 21.24457831325301, "grad_norm": 2.9928324222564697, "learning_rate": 8.936407766990292e-05, "loss": 0.2385, "step": 2191 }, { "epoch": 21.25421686746988, "grad_norm": 2.5955045223236084, "learning_rate": 8.935922330097088e-05, "loss": 0.115, "step": 2192 }, { "epoch": 21.263855421686745, "grad_norm": 3.2017085552215576, "learning_rate": 8.935436893203883e-05, "loss": 0.1046, "step": 2193 }, { "epoch": 21.273493975903616, "grad_norm": 7.880970001220703, "learning_rate": 8.934951456310679e-05, "loss": 0.3934, "step": 2194 }, { "epoch": 21.283132530120483, "grad_norm": 4.6827616691589355, "learning_rate": 8.934466019417475e-05, "loss": 0.3277, "step": 2195 }, { "epoch": 21.29277108433735, "grad_norm": 6.959837436676025, "learning_rate": 8.933980582524272e-05, "loss": 0.398, "step": 2196 }, { "epoch": 21.302409638554217, "grad_norm": 4.831371784210205, "learning_rate": 8.933495145631069e-05, "loss": 0.2212, "step": 2197 }, { "epoch": 21.312048192771083, "grad_norm": 8.524154663085938, "learning_rate": 8.933009708737865e-05, "loss": 0.4014, "step": 2198 }, { "epoch": 21.32168674698795, "grad_norm": 4.106417179107666, "learning_rate": 8.932524271844661e-05, "loss": 0.3689, "step": 2199 }, { "epoch": 21.33132530120482, "grad_norm": 5.393979072570801, "learning_rate": 8.932038834951457e-05, "loss": 0.3178, "step": 2200 }, { "epoch": 21.340963855421688, "grad_norm": 7.894619941711426, "learning_rate": 8.931553398058254e-05, "loss": 0.4339, "step": 2201 }, { "epoch": 21.350602409638554, "grad_norm": 8.788765907287598, "learning_rate": 8.93106796116505e-05, "loss": 0.3992, "step": 2202 }, { "epoch": 21.36024096385542, "grad_norm": 32.18729019165039, "learning_rate": 8.930582524271845e-05, "loss": 0.4806, "step": 2203 }, { "epoch": 21.36987951807229, "grad_norm": 5.528956413269043, "learning_rate": 8.930097087378641e-05, "loss": 0.4186, "step": 2204 }, { "epoch": 21.379518072289155, "grad_norm": 4.723628520965576, "learning_rate": 8.929611650485438e-05, "loss": 0.2912, "step": 2205 }, { "epoch": 21.389156626506026, "grad_norm": 9.421089172363281, "learning_rate": 8.929126213592234e-05, "loss": 0.309, "step": 2206 }, { "epoch": 21.398795180722892, "grad_norm": 12.350517272949219, "learning_rate": 8.92864077669903e-05, "loss": 0.432, "step": 2207 }, { "epoch": 21.40843373493976, "grad_norm": 5.303647518157959, "learning_rate": 8.928155339805826e-05, "loss": 0.5654, "step": 2208 }, { "epoch": 21.418072289156626, "grad_norm": 5.1570940017700195, "learning_rate": 8.927669902912623e-05, "loss": 0.3621, "step": 2209 }, { "epoch": 21.427710843373493, "grad_norm": 9.978538513183594, "learning_rate": 8.927184466019418e-05, "loss": 0.3502, "step": 2210 }, { "epoch": 21.43734939759036, "grad_norm": 5.228713035583496, "learning_rate": 8.926699029126214e-05, "loss": 0.38, "step": 2211 }, { "epoch": 21.44698795180723, "grad_norm": 19.514259338378906, "learning_rate": 8.92621359223301e-05, "loss": 0.6287, "step": 2212 }, { "epoch": 21.456626506024097, "grad_norm": 10.043380737304688, "learning_rate": 8.925728155339807e-05, "loss": 0.4937, "step": 2213 }, { "epoch": 21.466265060240964, "grad_norm": 6.514385223388672, "learning_rate": 8.925242718446603e-05, "loss": 0.3474, "step": 2214 }, { "epoch": 21.47590361445783, "grad_norm": 6.647812366485596, "learning_rate": 8.924757281553399e-05, "loss": 0.205, "step": 2215 }, { "epoch": 21.485542168674698, "grad_norm": 8.327917098999023, "learning_rate": 8.924271844660194e-05, "loss": 0.2832, "step": 2216 }, { "epoch": 21.495180722891565, "grad_norm": 4.976629734039307, "learning_rate": 8.92378640776699e-05, "loss": 0.5847, "step": 2217 }, { "epoch": 21.504819277108435, "grad_norm": 3.9394454956054688, "learning_rate": 8.923300970873787e-05, "loss": 0.1427, "step": 2218 }, { "epoch": 21.514457831325302, "grad_norm": 5.867509365081787, "learning_rate": 8.922815533980583e-05, "loss": 0.1778, "step": 2219 }, { "epoch": 21.52409638554217, "grad_norm": 5.7492780685424805, "learning_rate": 8.922330097087379e-05, "loss": 0.3979, "step": 2220 }, { "epoch": 21.533734939759036, "grad_norm": 2.8899099826812744, "learning_rate": 8.921844660194175e-05, "loss": 0.1972, "step": 2221 }, { "epoch": 21.543373493975903, "grad_norm": 22.50688362121582, "learning_rate": 8.921359223300972e-05, "loss": 0.5119, "step": 2222 }, { "epoch": 21.55301204819277, "grad_norm": 10.788880348205566, "learning_rate": 8.920873786407768e-05, "loss": 0.1955, "step": 2223 }, { "epoch": 21.56265060240964, "grad_norm": 18.277280807495117, "learning_rate": 8.920388349514563e-05, "loss": 0.2986, "step": 2224 }, { "epoch": 21.572289156626507, "grad_norm": 11.814428329467773, "learning_rate": 8.919902912621359e-05, "loss": 0.4075, "step": 2225 }, { "epoch": 21.581927710843374, "grad_norm": 13.266114234924316, "learning_rate": 8.919417475728156e-05, "loss": 0.6683, "step": 2226 }, { "epoch": 21.59156626506024, "grad_norm": 14.55667781829834, "learning_rate": 8.918932038834952e-05, "loss": 0.4796, "step": 2227 }, { "epoch": 21.601204819277108, "grad_norm": 6.38918399810791, "learning_rate": 8.918446601941748e-05, "loss": 0.4027, "step": 2228 }, { "epoch": 21.610843373493974, "grad_norm": 12.806866645812988, "learning_rate": 8.917961165048544e-05, "loss": 0.4499, "step": 2229 }, { "epoch": 21.620481927710845, "grad_norm": 7.101380825042725, "learning_rate": 8.917475728155341e-05, "loss": 0.3116, "step": 2230 }, { "epoch": 21.63012048192771, "grad_norm": 5.717641830444336, "learning_rate": 8.916990291262136e-05, "loss": 0.1562, "step": 2231 }, { "epoch": 21.63975903614458, "grad_norm": 24.720659255981445, "learning_rate": 8.916504854368932e-05, "loss": 0.2937, "step": 2232 }, { "epoch": 21.649397590361446, "grad_norm": 33.33662033081055, "learning_rate": 8.916019417475728e-05, "loss": 0.3483, "step": 2233 }, { "epoch": 21.659036144578312, "grad_norm": 4.026053428649902, "learning_rate": 8.915533980582524e-05, "loss": 0.3662, "step": 2234 }, { "epoch": 21.66867469879518, "grad_norm": 18.116111755371094, "learning_rate": 8.915048543689321e-05, "loss": 0.2703, "step": 2235 }, { "epoch": 21.67831325301205, "grad_norm": 8.309191703796387, "learning_rate": 8.914563106796117e-05, "loss": 0.453, "step": 2236 }, { "epoch": 21.687951807228917, "grad_norm": 9.078965187072754, "learning_rate": 8.914077669902913e-05, "loss": 0.4172, "step": 2237 }, { "epoch": 21.697590361445783, "grad_norm": 8.935972213745117, "learning_rate": 8.913592233009708e-05, "loss": 0.3513, "step": 2238 }, { "epoch": 21.70722891566265, "grad_norm": 4.645467281341553, "learning_rate": 8.913106796116505e-05, "loss": 0.1753, "step": 2239 }, { "epoch": 21.716867469879517, "grad_norm": 13.592841148376465, "learning_rate": 8.912621359223301e-05, "loss": 0.1697, "step": 2240 }, { "epoch": 21.726506024096384, "grad_norm": 10.69394588470459, "learning_rate": 8.912135922330097e-05, "loss": 0.2842, "step": 2241 }, { "epoch": 21.736144578313255, "grad_norm": 5.775249481201172, "learning_rate": 8.911650485436893e-05, "loss": 0.4383, "step": 2242 }, { "epoch": 21.74578313253012, "grad_norm": 69.12234497070312, "learning_rate": 8.91116504854369e-05, "loss": 0.3264, "step": 2243 }, { "epoch": 21.75542168674699, "grad_norm": 10.044442176818848, "learning_rate": 8.910679611650486e-05, "loss": 0.4047, "step": 2244 }, { "epoch": 21.765060240963855, "grad_norm": 4.251047611236572, "learning_rate": 8.910194174757281e-05, "loss": 0.261, "step": 2245 }, { "epoch": 21.774698795180722, "grad_norm": 6.1335883140563965, "learning_rate": 8.909708737864077e-05, "loss": 0.2207, "step": 2246 }, { "epoch": 21.78433734939759, "grad_norm": 18.331199645996094, "learning_rate": 8.909223300970874e-05, "loss": 0.2695, "step": 2247 }, { "epoch": 21.79397590361446, "grad_norm": 2.672563076019287, "learning_rate": 8.908737864077671e-05, "loss": 0.2102, "step": 2248 }, { "epoch": 21.803614457831326, "grad_norm": 8.34067440032959, "learning_rate": 8.908252427184467e-05, "loss": 0.5791, "step": 2249 }, { "epoch": 21.813253012048193, "grad_norm": 24.728717803955078, "learning_rate": 8.907766990291263e-05, "loss": 0.2953, "step": 2250 }, { "epoch": 21.82289156626506, "grad_norm": 5.944009304046631, "learning_rate": 8.907281553398059e-05, "loss": 0.4072, "step": 2251 }, { "epoch": 21.832530120481927, "grad_norm": 6.052402019500732, "learning_rate": 8.906796116504855e-05, "loss": 0.3828, "step": 2252 }, { "epoch": 21.842168674698794, "grad_norm": 2.8070995807647705, "learning_rate": 8.906310679611652e-05, "loss": 0.2066, "step": 2253 }, { "epoch": 21.851807228915664, "grad_norm": 43.89365768432617, "learning_rate": 8.905825242718447e-05, "loss": 0.4229, "step": 2254 }, { "epoch": 21.86144578313253, "grad_norm": 20.200777053833008, "learning_rate": 8.905339805825243e-05, "loss": 0.4277, "step": 2255 }, { "epoch": 21.871084337349398, "grad_norm": 8.241342544555664, "learning_rate": 8.904854368932039e-05, "loss": 0.3024, "step": 2256 }, { "epoch": 21.880722891566265, "grad_norm": 7.126091957092285, "learning_rate": 8.904368932038836e-05, "loss": 0.1705, "step": 2257 }, { "epoch": 21.89036144578313, "grad_norm": 6.851171493530273, "learning_rate": 8.903883495145632e-05, "loss": 0.27, "step": 2258 }, { "epoch": 21.9, "grad_norm": 40.99858474731445, "learning_rate": 8.903398058252428e-05, "loss": 0.4416, "step": 2259 }, { "epoch": 21.90963855421687, "grad_norm": 5.371386528015137, "learning_rate": 8.902912621359223e-05, "loss": 0.3777, "step": 2260 }, { "epoch": 21.919277108433736, "grad_norm": 4.695680618286133, "learning_rate": 8.90242718446602e-05, "loss": 0.4232, "step": 2261 }, { "epoch": 21.928915662650603, "grad_norm": 24.651968002319336, "learning_rate": 8.901941747572816e-05, "loss": 0.305, "step": 2262 }, { "epoch": 21.93855421686747, "grad_norm": 3.8853671550750732, "learning_rate": 8.901456310679612e-05, "loss": 0.5102, "step": 2263 }, { "epoch": 21.948192771084337, "grad_norm": 3.9605679512023926, "learning_rate": 8.900970873786408e-05, "loss": 0.1941, "step": 2264 }, { "epoch": 21.957831325301203, "grad_norm": 4.55346155166626, "learning_rate": 8.900485436893205e-05, "loss": 0.2539, "step": 2265 }, { "epoch": 21.967469879518074, "grad_norm": 49.152217864990234, "learning_rate": 8.900000000000001e-05, "loss": 0.4005, "step": 2266 }, { "epoch": 21.97710843373494, "grad_norm": 16.92533302307129, "learning_rate": 8.899514563106797e-05, "loss": 0.517, "step": 2267 }, { "epoch": 21.986746987951808, "grad_norm": 70.50060272216797, "learning_rate": 8.899029126213592e-05, "loss": 0.3672, "step": 2268 }, { "epoch": 22.002409638554216, "grad_norm": 9.411640167236328, "learning_rate": 8.898543689320388e-05, "loss": 0.3732, "step": 2269 }, { "epoch": 22.012048192771083, "grad_norm": 5.9638142585754395, "learning_rate": 8.898058252427185e-05, "loss": 0.1742, "step": 2270 }, { "epoch": 22.021686746987953, "grad_norm": 9.944381713867188, "learning_rate": 8.897572815533981e-05, "loss": 0.3888, "step": 2271 }, { "epoch": 22.03132530120482, "grad_norm": 6.661789894104004, "learning_rate": 8.897087378640777e-05, "loss": 0.3869, "step": 2272 }, { "epoch": 22.040963855421687, "grad_norm": 9.188225746154785, "learning_rate": 8.896601941747573e-05, "loss": 0.3758, "step": 2273 }, { "epoch": 22.050602409638554, "grad_norm": 7.673594951629639, "learning_rate": 8.89611650485437e-05, "loss": 0.4006, "step": 2274 }, { "epoch": 22.06024096385542, "grad_norm": 13.952990531921387, "learning_rate": 8.895631067961166e-05, "loss": 0.4339, "step": 2275 }, { "epoch": 22.069879518072288, "grad_norm": 5.736727714538574, "learning_rate": 8.895145631067961e-05, "loss": 0.3256, "step": 2276 }, { "epoch": 22.079518072289158, "grad_norm": 13.934999465942383, "learning_rate": 8.894660194174757e-05, "loss": 0.3737, "step": 2277 }, { "epoch": 22.089156626506025, "grad_norm": 26.79831886291504, "learning_rate": 8.894174757281554e-05, "loss": 0.2459, "step": 2278 }, { "epoch": 22.09879518072289, "grad_norm": 11.087240219116211, "learning_rate": 8.89368932038835e-05, "loss": 0.3214, "step": 2279 }, { "epoch": 22.10843373493976, "grad_norm": 5.920054912567139, "learning_rate": 8.893203883495146e-05, "loss": 0.1604, "step": 2280 }, { "epoch": 22.118072289156625, "grad_norm": 4.4026408195495605, "learning_rate": 8.892718446601942e-05, "loss": 0.2555, "step": 2281 }, { "epoch": 22.127710843373492, "grad_norm": 2.8953864574432373, "learning_rate": 8.892233009708739e-05, "loss": 0.0992, "step": 2282 }, { "epoch": 22.137349397590363, "grad_norm": 5.4221930503845215, "learning_rate": 8.891747572815534e-05, "loss": 0.3199, "step": 2283 }, { "epoch": 22.14698795180723, "grad_norm": 5.971908092498779, "learning_rate": 8.89126213592233e-05, "loss": 0.1795, "step": 2284 }, { "epoch": 22.156626506024097, "grad_norm": 10.78135871887207, "learning_rate": 8.890776699029126e-05, "loss": 0.3039, "step": 2285 }, { "epoch": 22.166265060240963, "grad_norm": 9.793379783630371, "learning_rate": 8.890291262135922e-05, "loss": 0.3186, "step": 2286 }, { "epoch": 22.17590361445783, "grad_norm": 7.433806419372559, "learning_rate": 8.889805825242719e-05, "loss": 0.284, "step": 2287 }, { "epoch": 22.185542168674697, "grad_norm": 9.388676643371582, "learning_rate": 8.889320388349515e-05, "loss": 0.6106, "step": 2288 }, { "epoch": 22.195180722891568, "grad_norm": 8.30748462677002, "learning_rate": 8.88883495145631e-05, "loss": 0.2489, "step": 2289 }, { "epoch": 22.204819277108435, "grad_norm": 2.490084171295166, "learning_rate": 8.888349514563106e-05, "loss": 0.3218, "step": 2290 }, { "epoch": 22.2144578313253, "grad_norm": 3.6807234287261963, "learning_rate": 8.887864077669903e-05, "loss": 0.221, "step": 2291 }, { "epoch": 22.22409638554217, "grad_norm": 7.12135124206543, "learning_rate": 8.887378640776699e-05, "loss": 0.5173, "step": 2292 }, { "epoch": 22.233734939759035, "grad_norm": 5.5195231437683105, "learning_rate": 8.886893203883495e-05, "loss": 0.4115, "step": 2293 }, { "epoch": 22.243373493975902, "grad_norm": 3.960705518722534, "learning_rate": 8.886407766990291e-05, "loss": 0.4101, "step": 2294 }, { "epoch": 22.253012048192772, "grad_norm": 2.1653120517730713, "learning_rate": 8.885922330097088e-05, "loss": 0.0662, "step": 2295 }, { "epoch": 22.26265060240964, "grad_norm": 7.167950630187988, "learning_rate": 8.885436893203884e-05, "loss": 0.3288, "step": 2296 }, { "epoch": 22.272289156626506, "grad_norm": 7.159619331359863, "learning_rate": 8.88495145631068e-05, "loss": 0.5226, "step": 2297 }, { "epoch": 22.281927710843373, "grad_norm": 7.723639011383057, "learning_rate": 8.884466019417477e-05, "loss": 0.3133, "step": 2298 }, { "epoch": 22.29156626506024, "grad_norm": 5.092340469360352, "learning_rate": 8.883980582524272e-05, "loss": 0.3447, "step": 2299 }, { "epoch": 22.301204819277107, "grad_norm": 5.925654888153076, "learning_rate": 8.88349514563107e-05, "loss": 0.1438, "step": 2300 }, { "epoch": 22.310843373493977, "grad_norm": 2.9508299827575684, "learning_rate": 8.883009708737865e-05, "loss": 0.3234, "step": 2301 }, { "epoch": 22.320481927710844, "grad_norm": 9.699308395385742, "learning_rate": 8.882524271844661e-05, "loss": 0.5409, "step": 2302 }, { "epoch": 22.33012048192771, "grad_norm": 5.175774574279785, "learning_rate": 8.882038834951457e-05, "loss": 0.5124, "step": 2303 }, { "epoch": 22.339759036144578, "grad_norm": 12.118961334228516, "learning_rate": 8.881553398058254e-05, "loss": 0.4335, "step": 2304 }, { "epoch": 22.349397590361445, "grad_norm": 9.111946105957031, "learning_rate": 8.88106796116505e-05, "loss": 0.1574, "step": 2305 }, { "epoch": 22.35903614457831, "grad_norm": 12.111213684082031, "learning_rate": 8.880582524271845e-05, "loss": 0.3866, "step": 2306 }, { "epoch": 22.368674698795182, "grad_norm": 5.433626174926758, "learning_rate": 8.880097087378641e-05, "loss": 0.3809, "step": 2307 }, { "epoch": 22.37831325301205, "grad_norm": 17.172056198120117, "learning_rate": 8.879611650485437e-05, "loss": 0.2237, "step": 2308 }, { "epoch": 22.387951807228916, "grad_norm": 4.179693222045898, "learning_rate": 8.879126213592234e-05, "loss": 0.2608, "step": 2309 }, { "epoch": 22.397590361445783, "grad_norm": 5.491171836853027, "learning_rate": 8.87864077669903e-05, "loss": 0.3804, "step": 2310 }, { "epoch": 22.40722891566265, "grad_norm": 15.430594444274902, "learning_rate": 8.878155339805826e-05, "loss": 0.3902, "step": 2311 }, { "epoch": 22.416867469879517, "grad_norm": 11.711423873901367, "learning_rate": 8.877669902912621e-05, "loss": 0.2108, "step": 2312 }, { "epoch": 22.426506024096387, "grad_norm": 5.917972087860107, "learning_rate": 8.877184466019419e-05, "loss": 0.3221, "step": 2313 }, { "epoch": 22.436144578313254, "grad_norm": 10.444304466247559, "learning_rate": 8.876699029126214e-05, "loss": 0.3828, "step": 2314 }, { "epoch": 22.44578313253012, "grad_norm": 5.4181976318359375, "learning_rate": 8.87621359223301e-05, "loss": 0.335, "step": 2315 }, { "epoch": 22.455421686746988, "grad_norm": 13.203408241271973, "learning_rate": 8.875728155339806e-05, "loss": 0.3757, "step": 2316 }, { "epoch": 22.465060240963854, "grad_norm": 6.349584102630615, "learning_rate": 8.875242718446603e-05, "loss": 0.1602, "step": 2317 }, { "epoch": 22.47469879518072, "grad_norm": 8.600847244262695, "learning_rate": 8.874757281553399e-05, "loss": 0.2564, "step": 2318 }, { "epoch": 22.48433734939759, "grad_norm": 9.4829683303833, "learning_rate": 8.874271844660195e-05, "loss": 0.6118, "step": 2319 }, { "epoch": 22.49397590361446, "grad_norm": 6.168498992919922, "learning_rate": 8.87378640776699e-05, "loss": 0.4104, "step": 2320 }, { "epoch": 22.503614457831326, "grad_norm": 23.096345901489258, "learning_rate": 8.873300970873787e-05, "loss": 0.5419, "step": 2321 }, { "epoch": 22.513253012048192, "grad_norm": 9.068732261657715, "learning_rate": 8.872815533980583e-05, "loss": 0.5227, "step": 2322 }, { "epoch": 22.52289156626506, "grad_norm": 4.735579490661621, "learning_rate": 8.872330097087379e-05, "loss": 0.2868, "step": 2323 }, { "epoch": 22.532530120481926, "grad_norm": 13.671121597290039, "learning_rate": 8.871844660194175e-05, "loss": 0.7295, "step": 2324 }, { "epoch": 22.542168674698797, "grad_norm": 8.085368156433105, "learning_rate": 8.87135922330097e-05, "loss": 0.3306, "step": 2325 }, { "epoch": 22.551807228915663, "grad_norm": 16.58965301513672, "learning_rate": 8.870873786407768e-05, "loss": 0.2256, "step": 2326 }, { "epoch": 22.56144578313253, "grad_norm": 9.91273021697998, "learning_rate": 8.870388349514563e-05, "loss": 0.4303, "step": 2327 }, { "epoch": 22.571084337349397, "grad_norm": 17.645578384399414, "learning_rate": 8.869902912621359e-05, "loss": 0.4738, "step": 2328 }, { "epoch": 22.580722891566264, "grad_norm": 6.704995155334473, "learning_rate": 8.869417475728155e-05, "loss": 0.183, "step": 2329 }, { "epoch": 22.59036144578313, "grad_norm": 5.7616095542907715, "learning_rate": 8.868932038834952e-05, "loss": 0.2478, "step": 2330 }, { "epoch": 22.6, "grad_norm": 17.930498123168945, "learning_rate": 8.868446601941748e-05, "loss": 0.3774, "step": 2331 }, { "epoch": 22.60963855421687, "grad_norm": 10.873723983764648, "learning_rate": 8.867961165048544e-05, "loss": 0.3746, "step": 2332 }, { "epoch": 22.619277108433735, "grad_norm": 21.151397705078125, "learning_rate": 8.86747572815534e-05, "loss": 0.5141, "step": 2333 }, { "epoch": 22.628915662650602, "grad_norm": 8.936529159545898, "learning_rate": 8.866990291262137e-05, "loss": 0.3422, "step": 2334 }, { "epoch": 22.63855421686747, "grad_norm": 30.713390350341797, "learning_rate": 8.866504854368932e-05, "loss": 0.3733, "step": 2335 }, { "epoch": 22.648192771084336, "grad_norm": 6.689733982086182, "learning_rate": 8.866019417475728e-05, "loss": 0.2188, "step": 2336 }, { "epoch": 22.657831325301206, "grad_norm": 5.703059196472168, "learning_rate": 8.865533980582524e-05, "loss": 0.2181, "step": 2337 }, { "epoch": 22.667469879518073, "grad_norm": 5.3546953201293945, "learning_rate": 8.865048543689321e-05, "loss": 0.3058, "step": 2338 }, { "epoch": 22.67710843373494, "grad_norm": 7.112314224243164, "learning_rate": 8.864563106796117e-05, "loss": 0.2733, "step": 2339 }, { "epoch": 22.686746987951807, "grad_norm": 6.224013328552246, "learning_rate": 8.864077669902913e-05, "loss": 0.383, "step": 2340 }, { "epoch": 22.696385542168674, "grad_norm": 8.542447090148926, "learning_rate": 8.863592233009708e-05, "loss": 0.341, "step": 2341 }, { "epoch": 22.70602409638554, "grad_norm": 19.56521987915039, "learning_rate": 8.863106796116504e-05, "loss": 0.3389, "step": 2342 }, { "epoch": 22.71566265060241, "grad_norm": 32.94246292114258, "learning_rate": 8.862621359223301e-05, "loss": 0.3996, "step": 2343 }, { "epoch": 22.725301204819278, "grad_norm": 14.823163032531738, "learning_rate": 8.862135922330097e-05, "loss": 0.3136, "step": 2344 }, { "epoch": 22.734939759036145, "grad_norm": 19.600584030151367, "learning_rate": 8.861650485436893e-05, "loss": 0.2409, "step": 2345 }, { "epoch": 22.74457831325301, "grad_norm": 7.905206203460693, "learning_rate": 8.861165048543689e-05, "loss": 0.4585, "step": 2346 }, { "epoch": 22.75421686746988, "grad_norm": 10.400909423828125, "learning_rate": 8.860679611650486e-05, "loss": 0.3756, "step": 2347 }, { "epoch": 22.763855421686745, "grad_norm": 24.103391647338867, "learning_rate": 8.860194174757283e-05, "loss": 0.4726, "step": 2348 }, { "epoch": 22.773493975903616, "grad_norm": 5.229384422302246, "learning_rate": 8.859708737864079e-05, "loss": 0.3104, "step": 2349 }, { "epoch": 22.783132530120483, "grad_norm": 10.89517593383789, "learning_rate": 8.859223300970874e-05, "loss": 0.6508, "step": 2350 }, { "epoch": 22.79277108433735, "grad_norm": 11.541398048400879, "learning_rate": 8.85873786407767e-05, "loss": 0.317, "step": 2351 }, { "epoch": 22.802409638554217, "grad_norm": 10.844027519226074, "learning_rate": 8.858252427184467e-05, "loss": 0.3822, "step": 2352 }, { "epoch": 22.812048192771083, "grad_norm": 6.221378326416016, "learning_rate": 8.857766990291263e-05, "loss": 0.535, "step": 2353 }, { "epoch": 22.82168674698795, "grad_norm": 4.544120788574219, "learning_rate": 8.857281553398059e-05, "loss": 0.3173, "step": 2354 }, { "epoch": 22.83132530120482, "grad_norm": 6.770412445068359, "learning_rate": 8.856796116504855e-05, "loss": 0.3166, "step": 2355 }, { "epoch": 22.840963855421688, "grad_norm": 3.9131217002868652, "learning_rate": 8.856310679611652e-05, "loss": 0.3504, "step": 2356 }, { "epoch": 22.850602409638554, "grad_norm": 2.9571902751922607, "learning_rate": 8.855825242718448e-05, "loss": 0.2144, "step": 2357 }, { "epoch": 22.86024096385542, "grad_norm": 9.245591163635254, "learning_rate": 8.855339805825243e-05, "loss": 0.2327, "step": 2358 }, { "epoch": 22.86987951807229, "grad_norm": 7.118414878845215, "learning_rate": 8.854854368932039e-05, "loss": 0.3281, "step": 2359 }, { "epoch": 22.879518072289155, "grad_norm": 40.724788665771484, "learning_rate": 8.854368932038836e-05, "loss": 0.251, "step": 2360 }, { "epoch": 22.889156626506026, "grad_norm": 3.2315735816955566, "learning_rate": 8.853883495145632e-05, "loss": 0.1619, "step": 2361 }, { "epoch": 22.898795180722892, "grad_norm": 21.445344924926758, "learning_rate": 8.853398058252428e-05, "loss": 0.3078, "step": 2362 }, { "epoch": 22.90843373493976, "grad_norm": 11.86123275756836, "learning_rate": 8.852912621359224e-05, "loss": 0.2419, "step": 2363 }, { "epoch": 22.918072289156626, "grad_norm": 16.692970275878906, "learning_rate": 8.85242718446602e-05, "loss": 0.6674, "step": 2364 }, { "epoch": 22.927710843373493, "grad_norm": 10.146451950073242, "learning_rate": 8.851941747572817e-05, "loss": 0.2616, "step": 2365 }, { "epoch": 22.93734939759036, "grad_norm": 26.01495933532715, "learning_rate": 8.851456310679612e-05, "loss": 0.2067, "step": 2366 }, { "epoch": 22.94698795180723, "grad_norm": 5.297593116760254, "learning_rate": 8.850970873786408e-05, "loss": 0.4464, "step": 2367 }, { "epoch": 22.956626506024097, "grad_norm": 8.902032852172852, "learning_rate": 8.850485436893204e-05, "loss": 0.2905, "step": 2368 }, { "epoch": 22.966265060240964, "grad_norm": 18.66410255432129, "learning_rate": 8.850000000000001e-05, "loss": 0.5941, "step": 2369 }, { "epoch": 22.97590361445783, "grad_norm": 16.110923767089844, "learning_rate": 8.849514563106797e-05, "loss": 0.6222, "step": 2370 }, { "epoch": 22.985542168674698, "grad_norm": 6.884692192077637, "learning_rate": 8.849029126213593e-05, "loss": 0.3035, "step": 2371 }, { "epoch": 23.00120481927711, "grad_norm": 9.338570594787598, "learning_rate": 8.848543689320388e-05, "loss": 0.2694, "step": 2372 }, { "epoch": 23.010843373493977, "grad_norm": 3.6689558029174805, "learning_rate": 8.848058252427185e-05, "loss": 0.2062, "step": 2373 }, { "epoch": 23.020481927710843, "grad_norm": 9.398517608642578, "learning_rate": 8.847572815533981e-05, "loss": 0.3904, "step": 2374 }, { "epoch": 23.03012048192771, "grad_norm": 4.809319972991943, "learning_rate": 8.847087378640777e-05, "loss": 0.187, "step": 2375 }, { "epoch": 23.039759036144577, "grad_norm": 6.42117977142334, "learning_rate": 8.846601941747573e-05, "loss": 0.378, "step": 2376 }, { "epoch": 23.049397590361444, "grad_norm": 10.445579528808594, "learning_rate": 8.84611650485437e-05, "loss": 0.7037, "step": 2377 }, { "epoch": 23.059036144578315, "grad_norm": 10.950546264648438, "learning_rate": 8.845631067961166e-05, "loss": 0.3571, "step": 2378 }, { "epoch": 23.06867469879518, "grad_norm": 6.176374912261963, "learning_rate": 8.845145631067961e-05, "loss": 0.274, "step": 2379 }, { "epoch": 23.07831325301205, "grad_norm": 6.812893390655518, "learning_rate": 8.844660194174757e-05, "loss": 0.4682, "step": 2380 }, { "epoch": 23.087951807228915, "grad_norm": 12.182948112487793, "learning_rate": 8.844174757281553e-05, "loss": 0.4056, "step": 2381 }, { "epoch": 23.097590361445782, "grad_norm": 7.55920934677124, "learning_rate": 8.84368932038835e-05, "loss": 0.3697, "step": 2382 }, { "epoch": 23.10722891566265, "grad_norm": 11.229445457458496, "learning_rate": 8.843203883495146e-05, "loss": 0.3422, "step": 2383 }, { "epoch": 23.11686746987952, "grad_norm": 11.865169525146484, "learning_rate": 8.842718446601942e-05, "loss": 0.3814, "step": 2384 }, { "epoch": 23.126506024096386, "grad_norm": 6.694626808166504, "learning_rate": 8.842233009708737e-05, "loss": 0.3451, "step": 2385 }, { "epoch": 23.136144578313253, "grad_norm": 7.442386627197266, "learning_rate": 8.841747572815535e-05, "loss": 0.5363, "step": 2386 }, { "epoch": 23.14578313253012, "grad_norm": 8.670738220214844, "learning_rate": 8.84126213592233e-05, "loss": 0.4594, "step": 2387 }, { "epoch": 23.155421686746987, "grad_norm": 7.363620758056641, "learning_rate": 8.840776699029126e-05, "loss": 0.3353, "step": 2388 }, { "epoch": 23.165060240963854, "grad_norm": 10.106439590454102, "learning_rate": 8.840291262135922e-05, "loss": 0.3538, "step": 2389 }, { "epoch": 23.174698795180724, "grad_norm": 8.438190460205078, "learning_rate": 8.839805825242719e-05, "loss": 0.2509, "step": 2390 }, { "epoch": 23.18433734939759, "grad_norm": 15.062203407287598, "learning_rate": 8.839320388349515e-05, "loss": 0.4237, "step": 2391 }, { "epoch": 23.193975903614458, "grad_norm": 14.69808292388916, "learning_rate": 8.83883495145631e-05, "loss": 0.2582, "step": 2392 }, { "epoch": 23.203614457831325, "grad_norm": 4.9254374504089355, "learning_rate": 8.838349514563106e-05, "loss": 0.2224, "step": 2393 }, { "epoch": 23.21325301204819, "grad_norm": 2.8890750408172607, "learning_rate": 8.837864077669904e-05, "loss": 0.129, "step": 2394 }, { "epoch": 23.22289156626506, "grad_norm": 7.423701286315918, "learning_rate": 8.837378640776699e-05, "loss": 0.4033, "step": 2395 }, { "epoch": 23.23253012048193, "grad_norm": 5.030467510223389, "learning_rate": 8.836893203883495e-05, "loss": 0.1838, "step": 2396 }, { "epoch": 23.242168674698796, "grad_norm": 8.800880432128906, "learning_rate": 8.836407766990291e-05, "loss": 0.398, "step": 2397 }, { "epoch": 23.251807228915663, "grad_norm": 4.649676322937012, "learning_rate": 8.835922330097087e-05, "loss": 0.339, "step": 2398 }, { "epoch": 23.26144578313253, "grad_norm": 8.585814476013184, "learning_rate": 8.835436893203885e-05, "loss": 0.4038, "step": 2399 }, { "epoch": 23.271084337349397, "grad_norm": 9.01313591003418, "learning_rate": 8.834951456310681e-05, "loss": 0.3132, "step": 2400 }, { "epoch": 23.280722891566263, "grad_norm": 15.724675178527832, "learning_rate": 8.834466019417477e-05, "loss": 0.3675, "step": 2401 }, { "epoch": 23.290361445783134, "grad_norm": 5.222177028656006, "learning_rate": 8.833980582524272e-05, "loss": 0.2304, "step": 2402 }, { "epoch": 23.3, "grad_norm": 10.188176155090332, "learning_rate": 8.833495145631068e-05, "loss": 0.2581, "step": 2403 }, { "epoch": 23.309638554216868, "grad_norm": 50.505279541015625, "learning_rate": 8.833009708737865e-05, "loss": 0.4739, "step": 2404 }, { "epoch": 23.319277108433734, "grad_norm": 22.67613410949707, "learning_rate": 8.832524271844661e-05, "loss": 0.789, "step": 2405 }, { "epoch": 23.3289156626506, "grad_norm": 5.652853965759277, "learning_rate": 8.832038834951457e-05, "loss": 0.3546, "step": 2406 }, { "epoch": 23.33855421686747, "grad_norm": 19.82429313659668, "learning_rate": 8.831553398058253e-05, "loss": 0.393, "step": 2407 }, { "epoch": 23.34819277108434, "grad_norm": 9.993599891662598, "learning_rate": 8.83106796116505e-05, "loss": 0.8376, "step": 2408 }, { "epoch": 23.357831325301206, "grad_norm": 16.18553924560547, "learning_rate": 8.830582524271846e-05, "loss": 0.4411, "step": 2409 }, { "epoch": 23.367469879518072, "grad_norm": 16.628480911254883, "learning_rate": 8.830097087378641e-05, "loss": 0.6528, "step": 2410 }, { "epoch": 23.37710843373494, "grad_norm": 8.783699989318848, "learning_rate": 8.829611650485437e-05, "loss": 0.3765, "step": 2411 }, { "epoch": 23.386746987951806, "grad_norm": 8.253890991210938, "learning_rate": 8.829126213592234e-05, "loss": 0.4763, "step": 2412 }, { "epoch": 23.396385542168673, "grad_norm": 10.883673667907715, "learning_rate": 8.82864077669903e-05, "loss": 0.2573, "step": 2413 }, { "epoch": 23.406024096385543, "grad_norm": 11.242916107177734, "learning_rate": 8.828155339805826e-05, "loss": 0.2947, "step": 2414 }, { "epoch": 23.41566265060241, "grad_norm": 4.7782111167907715, "learning_rate": 8.827669902912622e-05, "loss": 0.2514, "step": 2415 }, { "epoch": 23.425301204819277, "grad_norm": 7.600667953491211, "learning_rate": 8.827184466019419e-05, "loss": 0.3766, "step": 2416 }, { "epoch": 23.434939759036144, "grad_norm": 10.797026634216309, "learning_rate": 8.826699029126214e-05, "loss": 0.3269, "step": 2417 }, { "epoch": 23.44457831325301, "grad_norm": 7.3057451248168945, "learning_rate": 8.82621359223301e-05, "loss": 0.3022, "step": 2418 }, { "epoch": 23.454216867469878, "grad_norm": 24.309101104736328, "learning_rate": 8.825728155339806e-05, "loss": 0.4429, "step": 2419 }, { "epoch": 23.46385542168675, "grad_norm": 3.391212224960327, "learning_rate": 8.825242718446602e-05, "loss": 0.2056, "step": 2420 }, { "epoch": 23.473493975903615, "grad_norm": 11.969708442687988, "learning_rate": 8.824757281553399e-05, "loss": 0.2677, "step": 2421 }, { "epoch": 23.483132530120482, "grad_norm": 21.958181381225586, "learning_rate": 8.824271844660195e-05, "loss": 0.3427, "step": 2422 }, { "epoch": 23.49277108433735, "grad_norm": 5.496153354644775, "learning_rate": 8.82378640776699e-05, "loss": 0.2417, "step": 2423 }, { "epoch": 23.502409638554216, "grad_norm": 5.812340259552002, "learning_rate": 8.823300970873786e-05, "loss": 0.2159, "step": 2424 }, { "epoch": 23.512048192771083, "grad_norm": 4.958152770996094, "learning_rate": 8.822815533980583e-05, "loss": 0.175, "step": 2425 }, { "epoch": 23.521686746987953, "grad_norm": 26.579345703125, "learning_rate": 8.822330097087379e-05, "loss": 0.3156, "step": 2426 }, { "epoch": 23.53132530120482, "grad_norm": 15.123591423034668, "learning_rate": 8.821844660194175e-05, "loss": 0.4551, "step": 2427 }, { "epoch": 23.540963855421687, "grad_norm": 5.475025653839111, "learning_rate": 8.821359223300971e-05, "loss": 0.3228, "step": 2428 }, { "epoch": 23.550602409638554, "grad_norm": 13.381158828735352, "learning_rate": 8.820873786407768e-05, "loss": 0.3956, "step": 2429 }, { "epoch": 23.56024096385542, "grad_norm": 12.051216125488281, "learning_rate": 8.820388349514564e-05, "loss": 0.4405, "step": 2430 }, { "epoch": 23.569879518072288, "grad_norm": 18.99673080444336, "learning_rate": 8.81990291262136e-05, "loss": 0.4256, "step": 2431 }, { "epoch": 23.579518072289158, "grad_norm": 25.322582244873047, "learning_rate": 8.819417475728155e-05, "loss": 0.472, "step": 2432 }, { "epoch": 23.589156626506025, "grad_norm": 5.698084354400635, "learning_rate": 8.818932038834952e-05, "loss": 0.1798, "step": 2433 }, { "epoch": 23.59879518072289, "grad_norm": 5.542449474334717, "learning_rate": 8.818446601941748e-05, "loss": 0.223, "step": 2434 }, { "epoch": 23.60843373493976, "grad_norm": 19.386343002319336, "learning_rate": 8.817961165048544e-05, "loss": 0.6179, "step": 2435 }, { "epoch": 23.618072289156625, "grad_norm": 15.177533149719238, "learning_rate": 8.81747572815534e-05, "loss": 0.2086, "step": 2436 }, { "epoch": 23.627710843373492, "grad_norm": 18.957439422607422, "learning_rate": 8.816990291262135e-05, "loss": 0.2166, "step": 2437 }, { "epoch": 23.637349397590363, "grad_norm": 7.625329971313477, "learning_rate": 8.816504854368933e-05, "loss": 0.3106, "step": 2438 }, { "epoch": 23.64698795180723, "grad_norm": 10.079035758972168, "learning_rate": 8.816019417475728e-05, "loss": 0.1829, "step": 2439 }, { "epoch": 23.656626506024097, "grad_norm": 12.052436828613281, "learning_rate": 8.815533980582524e-05, "loss": 0.286, "step": 2440 }, { "epoch": 23.666265060240963, "grad_norm": 37.90221405029297, "learning_rate": 8.81504854368932e-05, "loss": 0.576, "step": 2441 }, { "epoch": 23.67590361445783, "grad_norm": 5.054784774780273, "learning_rate": 8.814563106796117e-05, "loss": 0.349, "step": 2442 }, { "epoch": 23.685542168674697, "grad_norm": 10.88890266418457, "learning_rate": 8.814077669902913e-05, "loss": 0.5186, "step": 2443 }, { "epoch": 23.695180722891568, "grad_norm": 12.370150566101074, "learning_rate": 8.813592233009709e-05, "loss": 0.3378, "step": 2444 }, { "epoch": 23.704819277108435, "grad_norm": 9.756922721862793, "learning_rate": 8.813106796116504e-05, "loss": 0.2015, "step": 2445 }, { "epoch": 23.7144578313253, "grad_norm": 8.217723846435547, "learning_rate": 8.812621359223301e-05, "loss": 0.2525, "step": 2446 }, { "epoch": 23.72409638554217, "grad_norm": 20.703338623046875, "learning_rate": 8.812135922330097e-05, "loss": 0.2868, "step": 2447 }, { "epoch": 23.733734939759035, "grad_norm": 4.175050735473633, "learning_rate": 8.811650485436893e-05, "loss": 0.2305, "step": 2448 }, { "epoch": 23.743373493975902, "grad_norm": 3.531217098236084, "learning_rate": 8.81116504854369e-05, "loss": 0.1466, "step": 2449 }, { "epoch": 23.753012048192772, "grad_norm": 3.6006453037261963, "learning_rate": 8.810679611650486e-05, "loss": 0.201, "step": 2450 }, { "epoch": 23.76265060240964, "grad_norm": 8.171937942504883, "learning_rate": 8.810194174757283e-05, "loss": 0.4473, "step": 2451 }, { "epoch": 23.772289156626506, "grad_norm": 8.267675399780273, "learning_rate": 8.809708737864079e-05, "loss": 0.3576, "step": 2452 }, { "epoch": 23.781927710843373, "grad_norm": 9.75684928894043, "learning_rate": 8.809223300970875e-05, "loss": 0.3207, "step": 2453 }, { "epoch": 23.79156626506024, "grad_norm": 17.090871810913086, "learning_rate": 8.80873786407767e-05, "loss": 0.478, "step": 2454 }, { "epoch": 23.801204819277107, "grad_norm": 11.409150123596191, "learning_rate": 8.808252427184468e-05, "loss": 0.3141, "step": 2455 }, { "epoch": 23.810843373493977, "grad_norm": 8.033745765686035, "learning_rate": 8.807766990291263e-05, "loss": 0.4501, "step": 2456 }, { "epoch": 23.820481927710844, "grad_norm": 11.341726303100586, "learning_rate": 8.807281553398059e-05, "loss": 0.2498, "step": 2457 }, { "epoch": 23.83012048192771, "grad_norm": 5.298240661621094, "learning_rate": 8.806796116504855e-05, "loss": 0.1787, "step": 2458 }, { "epoch": 23.839759036144578, "grad_norm": 18.371496200561523, "learning_rate": 8.80631067961165e-05, "loss": 0.2331, "step": 2459 }, { "epoch": 23.849397590361445, "grad_norm": 14.533903121948242, "learning_rate": 8.805825242718448e-05, "loss": 0.2541, "step": 2460 }, { "epoch": 23.85903614457831, "grad_norm": 21.966995239257812, "learning_rate": 8.805339805825244e-05, "loss": 0.2874, "step": 2461 }, { "epoch": 23.868674698795182, "grad_norm": 4.344491958618164, "learning_rate": 8.804854368932039e-05, "loss": 0.2862, "step": 2462 }, { "epoch": 23.87831325301205, "grad_norm": 16.518966674804688, "learning_rate": 8.804368932038835e-05, "loss": 0.2374, "step": 2463 }, { "epoch": 23.887951807228916, "grad_norm": 26.704919815063477, "learning_rate": 8.803883495145632e-05, "loss": 0.4599, "step": 2464 }, { "epoch": 23.897590361445783, "grad_norm": 4.222492218017578, "learning_rate": 8.803398058252428e-05, "loss": 0.2156, "step": 2465 }, { "epoch": 23.90722891566265, "grad_norm": 4.272043704986572, "learning_rate": 8.802912621359224e-05, "loss": 0.3431, "step": 2466 }, { "epoch": 23.916867469879517, "grad_norm": 7.7692999839782715, "learning_rate": 8.80242718446602e-05, "loss": 0.2819, "step": 2467 }, { "epoch": 23.926506024096387, "grad_norm": 6.725187301635742, "learning_rate": 8.801941747572817e-05, "loss": 0.319, "step": 2468 }, { "epoch": 23.936144578313254, "grad_norm": 16.336647033691406, "learning_rate": 8.801456310679612e-05, "loss": 0.4632, "step": 2469 }, { "epoch": 23.94578313253012, "grad_norm": 3.2318532466888428, "learning_rate": 8.800970873786408e-05, "loss": 0.1466, "step": 2470 }, { "epoch": 23.955421686746988, "grad_norm": 7.0436787605285645, "learning_rate": 8.800485436893204e-05, "loss": 0.3513, "step": 2471 }, { "epoch": 23.965060240963854, "grad_norm": 5.374818325042725, "learning_rate": 8.800000000000001e-05, "loss": 0.2692, "step": 2472 }, { "epoch": 23.97469879518072, "grad_norm": 19.128652572631836, "learning_rate": 8.799514563106797e-05, "loss": 0.3919, "step": 2473 }, { "epoch": 23.98433734939759, "grad_norm": 15.220536231994629, "learning_rate": 8.799029126213593e-05, "loss": 0.3972, "step": 2474 }, { "epoch": 23.99397590361446, "grad_norm": 10.130197525024414, "learning_rate": 8.798543689320388e-05, "loss": 0.4931, "step": 2475 }, { "epoch": 24.009638554216867, "grad_norm": 7.574868679046631, "learning_rate": 8.798058252427184e-05, "loss": 0.2705, "step": 2476 }, { "epoch": 24.019277108433734, "grad_norm": 12.12050724029541, "learning_rate": 8.797572815533981e-05, "loss": 0.3859, "step": 2477 }, { "epoch": 24.0289156626506, "grad_norm": 6.212700843811035, "learning_rate": 8.797087378640777e-05, "loss": 0.3983, "step": 2478 }, { "epoch": 24.03855421686747, "grad_norm": 12.477849960327148, "learning_rate": 8.796601941747573e-05, "loss": 0.4236, "step": 2479 }, { "epoch": 24.048192771084338, "grad_norm": 4.971896648406982, "learning_rate": 8.796116504854369e-05, "loss": 0.1741, "step": 2480 }, { "epoch": 24.057831325301205, "grad_norm": 8.887489318847656, "learning_rate": 8.795631067961166e-05, "loss": 0.3358, "step": 2481 }, { "epoch": 24.06746987951807, "grad_norm": 8.410202980041504, "learning_rate": 8.795145631067962e-05, "loss": 0.2551, "step": 2482 }, { "epoch": 24.07710843373494, "grad_norm": 7.5903167724609375, "learning_rate": 8.794660194174757e-05, "loss": 0.168, "step": 2483 }, { "epoch": 24.086746987951805, "grad_norm": 6.678199291229248, "learning_rate": 8.794174757281553e-05, "loss": 0.2764, "step": 2484 }, { "epoch": 24.096385542168676, "grad_norm": 37.77317810058594, "learning_rate": 8.79368932038835e-05, "loss": 0.4839, "step": 2485 }, { "epoch": 24.106024096385543, "grad_norm": 7.907444953918457, "learning_rate": 8.793203883495146e-05, "loss": 0.463, "step": 2486 }, { "epoch": 24.11566265060241, "grad_norm": 10.522565841674805, "learning_rate": 8.792718446601942e-05, "loss": 0.3313, "step": 2487 }, { "epoch": 24.125301204819277, "grad_norm": 8.188711166381836, "learning_rate": 8.792233009708738e-05, "loss": 0.3307, "step": 2488 }, { "epoch": 24.134939759036143, "grad_norm": 8.380213737487793, "learning_rate": 8.791747572815535e-05, "loss": 0.2051, "step": 2489 }, { "epoch": 24.14457831325301, "grad_norm": 11.567877769470215, "learning_rate": 8.79126213592233e-05, "loss": 0.5899, "step": 2490 }, { "epoch": 24.15421686746988, "grad_norm": 7.796013355255127, "learning_rate": 8.790776699029126e-05, "loss": 0.4988, "step": 2491 }, { "epoch": 24.163855421686748, "grad_norm": 8.984952926635742, "learning_rate": 8.790291262135922e-05, "loss": 0.4394, "step": 2492 }, { "epoch": 24.173493975903614, "grad_norm": 29.538759231567383, "learning_rate": 8.789805825242718e-05, "loss": 0.4708, "step": 2493 }, { "epoch": 24.18313253012048, "grad_norm": 4.936911106109619, "learning_rate": 8.789320388349515e-05, "loss": 0.2965, "step": 2494 }, { "epoch": 24.19277108433735, "grad_norm": 2.400874614715576, "learning_rate": 8.788834951456311e-05, "loss": 0.1656, "step": 2495 }, { "epoch": 24.202409638554215, "grad_norm": 80.97468566894531, "learning_rate": 8.788349514563107e-05, "loss": 0.273, "step": 2496 }, { "epoch": 24.212048192771086, "grad_norm": 11.852928161621094, "learning_rate": 8.787864077669902e-05, "loss": 0.4748, "step": 2497 }, { "epoch": 24.221686746987952, "grad_norm": 9.244354248046875, "learning_rate": 8.7873786407767e-05, "loss": 0.395, "step": 2498 }, { "epoch": 24.23132530120482, "grad_norm": 6.950978755950928, "learning_rate": 8.786893203883495e-05, "loss": 0.4616, "step": 2499 }, { "epoch": 24.240963855421686, "grad_norm": 13.602249145507812, "learning_rate": 8.786407766990292e-05, "loss": 0.6348, "step": 2500 }, { "epoch": 24.250602409638553, "grad_norm": 5.607892036437988, "learning_rate": 8.785922330097088e-05, "loss": 0.1736, "step": 2501 }, { "epoch": 24.26024096385542, "grad_norm": 6.52144718170166, "learning_rate": 8.785436893203884e-05, "loss": 0.4092, "step": 2502 }, { "epoch": 24.26987951807229, "grad_norm": 17.93943977355957, "learning_rate": 8.784951456310681e-05, "loss": 0.4766, "step": 2503 }, { "epoch": 24.279518072289157, "grad_norm": 24.070423126220703, "learning_rate": 8.784466019417477e-05, "loss": 0.3149, "step": 2504 }, { "epoch": 24.289156626506024, "grad_norm": 7.990084648132324, "learning_rate": 8.783980582524273e-05, "loss": 0.365, "step": 2505 }, { "epoch": 24.29879518072289, "grad_norm": 9.116954803466797, "learning_rate": 8.783495145631068e-05, "loss": 0.2938, "step": 2506 }, { "epoch": 24.308433734939758, "grad_norm": 5.4895429611206055, "learning_rate": 8.783009708737865e-05, "loss": 0.327, "step": 2507 }, { "epoch": 24.318072289156625, "grad_norm": 22.537752151489258, "learning_rate": 8.782524271844661e-05, "loss": 0.3569, "step": 2508 }, { "epoch": 24.327710843373495, "grad_norm": 31.00585174560547, "learning_rate": 8.782038834951457e-05, "loss": 0.2946, "step": 2509 }, { "epoch": 24.337349397590362, "grad_norm": 5.246865749359131, "learning_rate": 8.781553398058253e-05, "loss": 0.2763, "step": 2510 }, { "epoch": 24.34698795180723, "grad_norm": 5.656888008117676, "learning_rate": 8.78106796116505e-05, "loss": 0.2315, "step": 2511 }, { "epoch": 24.356626506024096, "grad_norm": 10.73576831817627, "learning_rate": 8.780582524271846e-05, "loss": 0.4754, "step": 2512 }, { "epoch": 24.366265060240963, "grad_norm": 11.857471466064453, "learning_rate": 8.780097087378641e-05, "loss": 0.4194, "step": 2513 }, { "epoch": 24.37590361445783, "grad_norm": 11.627461433410645, "learning_rate": 8.779611650485437e-05, "loss": 0.4155, "step": 2514 }, { "epoch": 24.3855421686747, "grad_norm": 8.87336540222168, "learning_rate": 8.779126213592233e-05, "loss": 0.2952, "step": 2515 }, { "epoch": 24.395180722891567, "grad_norm": 5.224743843078613, "learning_rate": 8.77864077669903e-05, "loss": 0.177, "step": 2516 }, { "epoch": 24.404819277108434, "grad_norm": 11.618436813354492, "learning_rate": 8.778155339805826e-05, "loss": 0.449, "step": 2517 }, { "epoch": 24.4144578313253, "grad_norm": 15.73175048828125, "learning_rate": 8.777669902912622e-05, "loss": 0.448, "step": 2518 }, { "epoch": 24.424096385542168, "grad_norm": 11.760454177856445, "learning_rate": 8.777184466019417e-05, "loss": 0.4259, "step": 2519 }, { "epoch": 24.433734939759034, "grad_norm": 3.3888182640075684, "learning_rate": 8.776699029126215e-05, "loss": 0.1256, "step": 2520 }, { "epoch": 24.443373493975905, "grad_norm": 16.604084014892578, "learning_rate": 8.77621359223301e-05, "loss": 0.3024, "step": 2521 }, { "epoch": 24.45301204819277, "grad_norm": 2.8638696670532227, "learning_rate": 8.775728155339806e-05, "loss": 0.2454, "step": 2522 }, { "epoch": 24.46265060240964, "grad_norm": 9.017637252807617, "learning_rate": 8.775242718446602e-05, "loss": 0.4361, "step": 2523 }, { "epoch": 24.472289156626506, "grad_norm": 6.397702217102051, "learning_rate": 8.774757281553399e-05, "loss": 0.5411, "step": 2524 }, { "epoch": 24.481927710843372, "grad_norm": 9.359827041625977, "learning_rate": 8.774271844660195e-05, "loss": 0.4137, "step": 2525 }, { "epoch": 24.49156626506024, "grad_norm": 7.203212261199951, "learning_rate": 8.77378640776699e-05, "loss": 0.5953, "step": 2526 }, { "epoch": 24.50120481927711, "grad_norm": 6.754775047302246, "learning_rate": 8.773300970873786e-05, "loss": 0.2481, "step": 2527 }, { "epoch": 24.510843373493977, "grad_norm": 4.916004180908203, "learning_rate": 8.772815533980584e-05, "loss": 0.3271, "step": 2528 }, { "epoch": 24.520481927710843, "grad_norm": 7.858489513397217, "learning_rate": 8.772330097087379e-05, "loss": 0.538, "step": 2529 }, { "epoch": 24.53012048192771, "grad_norm": 10.870919227600098, "learning_rate": 8.771844660194175e-05, "loss": 0.3752, "step": 2530 }, { "epoch": 24.539759036144577, "grad_norm": 7.402121067047119, "learning_rate": 8.771359223300971e-05, "loss": 0.5621, "step": 2531 }, { "epoch": 24.549397590361444, "grad_norm": 4.264886856079102, "learning_rate": 8.770873786407767e-05, "loss": 0.2261, "step": 2532 }, { "epoch": 24.559036144578315, "grad_norm": 6.986367225646973, "learning_rate": 8.770388349514564e-05, "loss": 0.3664, "step": 2533 }, { "epoch": 24.56867469879518, "grad_norm": 4.308691501617432, "learning_rate": 8.76990291262136e-05, "loss": 0.3459, "step": 2534 }, { "epoch": 24.57831325301205, "grad_norm": 5.466145038604736, "learning_rate": 8.769417475728155e-05, "loss": 0.2708, "step": 2535 }, { "epoch": 24.587951807228915, "grad_norm": 7.744576930999756, "learning_rate": 8.768932038834951e-05, "loss": 0.379, "step": 2536 }, { "epoch": 24.597590361445782, "grad_norm": 6.966259956359863, "learning_rate": 8.768446601941748e-05, "loss": 0.3652, "step": 2537 }, { "epoch": 24.60722891566265, "grad_norm": 3.5537617206573486, "learning_rate": 8.767961165048544e-05, "loss": 0.1534, "step": 2538 }, { "epoch": 24.61686746987952, "grad_norm": 4.398351192474365, "learning_rate": 8.76747572815534e-05, "loss": 0.2205, "step": 2539 }, { "epoch": 24.626506024096386, "grad_norm": 9.67757511138916, "learning_rate": 8.766990291262136e-05, "loss": 0.361, "step": 2540 }, { "epoch": 24.636144578313253, "grad_norm": 12.020818710327148, "learning_rate": 8.766504854368933e-05, "loss": 0.5368, "step": 2541 }, { "epoch": 24.64578313253012, "grad_norm": 6.368501663208008, "learning_rate": 8.766019417475728e-05, "loss": 0.2953, "step": 2542 }, { "epoch": 24.655421686746987, "grad_norm": 10.435840606689453, "learning_rate": 8.765533980582524e-05, "loss": 0.2876, "step": 2543 }, { "epoch": 24.665060240963854, "grad_norm": 5.764291763305664, "learning_rate": 8.76504854368932e-05, "loss": 0.3987, "step": 2544 }, { "epoch": 24.674698795180724, "grad_norm": 5.799643516540527, "learning_rate": 8.764563106796117e-05, "loss": 0.166, "step": 2545 }, { "epoch": 24.68433734939759, "grad_norm": 4.533388614654541, "learning_rate": 8.764077669902913e-05, "loss": 0.3917, "step": 2546 }, { "epoch": 24.693975903614458, "grad_norm": 7.980547904968262, "learning_rate": 8.763592233009709e-05, "loss": 0.3448, "step": 2547 }, { "epoch": 24.703614457831325, "grad_norm": 8.106343269348145, "learning_rate": 8.763106796116504e-05, "loss": 0.2636, "step": 2548 }, { "epoch": 24.71325301204819, "grad_norm": 9.05624008178711, "learning_rate": 8.7626213592233e-05, "loss": 0.7019, "step": 2549 }, { "epoch": 24.72289156626506, "grad_norm": 6.527524471282959, "learning_rate": 8.762135922330097e-05, "loss": 0.3235, "step": 2550 }, { "epoch": 24.73253012048193, "grad_norm": 7.09128999710083, "learning_rate": 8.761650485436895e-05, "loss": 0.3879, "step": 2551 }, { "epoch": 24.742168674698796, "grad_norm": 6.967550754547119, "learning_rate": 8.76116504854369e-05, "loss": 0.344, "step": 2552 }, { "epoch": 24.751807228915663, "grad_norm": 6.681129455566406, "learning_rate": 8.760679611650486e-05, "loss": 0.3374, "step": 2553 }, { "epoch": 24.76144578313253, "grad_norm": 6.192886829376221, "learning_rate": 8.760194174757282e-05, "loss": 0.3916, "step": 2554 }, { "epoch": 24.771084337349397, "grad_norm": 5.846513271331787, "learning_rate": 8.759708737864079e-05, "loss": 0.2451, "step": 2555 }, { "epoch": 24.780722891566263, "grad_norm": 6.4917378425598145, "learning_rate": 8.759223300970875e-05, "loss": 0.2565, "step": 2556 }, { "epoch": 24.790361445783134, "grad_norm": 5.824562072753906, "learning_rate": 8.75873786407767e-05, "loss": 0.1575, "step": 2557 }, { "epoch": 24.8, "grad_norm": 3.574695587158203, "learning_rate": 8.758252427184466e-05, "loss": 0.285, "step": 2558 }, { "epoch": 24.809638554216868, "grad_norm": 5.774848937988281, "learning_rate": 8.757766990291263e-05, "loss": 0.3313, "step": 2559 }, { "epoch": 24.819277108433734, "grad_norm": 27.635116577148438, "learning_rate": 8.757281553398059e-05, "loss": 0.4225, "step": 2560 }, { "epoch": 24.8289156626506, "grad_norm": 11.56636905670166, "learning_rate": 8.756796116504855e-05, "loss": 0.4574, "step": 2561 }, { "epoch": 24.83855421686747, "grad_norm": 10.465139389038086, "learning_rate": 8.756310679611651e-05, "loss": 0.3708, "step": 2562 }, { "epoch": 24.84819277108434, "grad_norm": 7.147889614105225, "learning_rate": 8.755825242718448e-05, "loss": 0.4125, "step": 2563 }, { "epoch": 24.857831325301206, "grad_norm": 7.756060600280762, "learning_rate": 8.755339805825244e-05, "loss": 0.649, "step": 2564 }, { "epoch": 24.867469879518072, "grad_norm": 12.842037200927734, "learning_rate": 8.75485436893204e-05, "loss": 0.4858, "step": 2565 }, { "epoch": 24.87710843373494, "grad_norm": 10.310794830322266, "learning_rate": 8.754368932038835e-05, "loss": 0.4039, "step": 2566 }, { "epoch": 24.886746987951806, "grad_norm": 8.412572860717773, "learning_rate": 8.753883495145631e-05, "loss": 0.2021, "step": 2567 }, { "epoch": 24.896385542168673, "grad_norm": 9.954697608947754, "learning_rate": 8.753398058252428e-05, "loss": 0.3327, "step": 2568 }, { "epoch": 24.906024096385543, "grad_norm": 7.425406455993652, "learning_rate": 8.752912621359224e-05, "loss": 0.3291, "step": 2569 }, { "epoch": 24.91566265060241, "grad_norm": 5.145172595977783, "learning_rate": 8.75242718446602e-05, "loss": 0.5486, "step": 2570 }, { "epoch": 24.925301204819277, "grad_norm": 7.744809627532959, "learning_rate": 8.751941747572815e-05, "loss": 0.3145, "step": 2571 }, { "epoch": 24.934939759036144, "grad_norm": 8.741817474365234, "learning_rate": 8.751456310679613e-05, "loss": 0.4625, "step": 2572 }, { "epoch": 24.94457831325301, "grad_norm": 8.192500114440918, "learning_rate": 8.750970873786408e-05, "loss": 0.2358, "step": 2573 }, { "epoch": 24.954216867469878, "grad_norm": 5.898991107940674, "learning_rate": 8.750485436893204e-05, "loss": 0.393, "step": 2574 }, { "epoch": 24.96385542168675, "grad_norm": 6.124575614929199, "learning_rate": 8.75e-05, "loss": 0.1432, "step": 2575 }, { "epoch": 24.973493975903615, "grad_norm": 20.44960594177246, "learning_rate": 8.749514563106797e-05, "loss": 0.2698, "step": 2576 }, { "epoch": 24.983132530120482, "grad_norm": 6.107243061065674, "learning_rate": 8.749029126213593e-05, "loss": 0.2352, "step": 2577 }, { "epoch": 24.99277108433735, "grad_norm": 10.626370429992676, "learning_rate": 8.748543689320389e-05, "loss": 0.601, "step": 2578 }, { "epoch": 25.00843373493976, "grad_norm": 4.3554768562316895, "learning_rate": 8.748058252427184e-05, "loss": 0.2137, "step": 2579 }, { "epoch": 25.018072289156628, "grad_norm": 32.01624298095703, "learning_rate": 8.747572815533982e-05, "loss": 0.295, "step": 2580 }, { "epoch": 25.027710843373494, "grad_norm": 20.479446411132812, "learning_rate": 8.747087378640777e-05, "loss": 0.4541, "step": 2581 }, { "epoch": 25.03734939759036, "grad_norm": 6.841614246368408, "learning_rate": 8.746601941747573e-05, "loss": 0.3593, "step": 2582 }, { "epoch": 25.04698795180723, "grad_norm": 33.522361755371094, "learning_rate": 8.746116504854369e-05, "loss": 0.3027, "step": 2583 }, { "epoch": 25.056626506024095, "grad_norm": 2.4572205543518066, "learning_rate": 8.745631067961165e-05, "loss": 0.1877, "step": 2584 }, { "epoch": 25.066265060240966, "grad_norm": 1.5255359411239624, "learning_rate": 8.745145631067962e-05, "loss": 0.2601, "step": 2585 }, { "epoch": 25.075903614457832, "grad_norm": 31.003602981567383, "learning_rate": 8.744660194174758e-05, "loss": 0.434, "step": 2586 }, { "epoch": 25.0855421686747, "grad_norm": 9.357420921325684, "learning_rate": 8.744174757281553e-05, "loss": 0.2731, "step": 2587 }, { "epoch": 25.095180722891566, "grad_norm": 42.63052749633789, "learning_rate": 8.743689320388349e-05, "loss": 0.3764, "step": 2588 }, { "epoch": 25.104819277108433, "grad_norm": 21.932579040527344, "learning_rate": 8.743203883495146e-05, "loss": 0.3289, "step": 2589 }, { "epoch": 25.1144578313253, "grad_norm": 12.68036937713623, "learning_rate": 8.742718446601942e-05, "loss": 0.3571, "step": 2590 }, { "epoch": 25.12409638554217, "grad_norm": 10.316309928894043, "learning_rate": 8.742233009708738e-05, "loss": 0.4527, "step": 2591 }, { "epoch": 25.133734939759037, "grad_norm": 6.519512176513672, "learning_rate": 8.741747572815534e-05, "loss": 0.4259, "step": 2592 }, { "epoch": 25.143373493975904, "grad_norm": 51.47377395629883, "learning_rate": 8.74126213592233e-05, "loss": 0.3802, "step": 2593 }, { "epoch": 25.15301204819277, "grad_norm": 19.63567543029785, "learning_rate": 8.740776699029126e-05, "loss": 0.2862, "step": 2594 }, { "epoch": 25.162650602409638, "grad_norm": 4.4662699699401855, "learning_rate": 8.740291262135922e-05, "loss": 0.3794, "step": 2595 }, { "epoch": 25.172289156626505, "grad_norm": 48.03007507324219, "learning_rate": 8.739805825242718e-05, "loss": 0.346, "step": 2596 }, { "epoch": 25.181927710843375, "grad_norm": 6.70247220993042, "learning_rate": 8.739320388349515e-05, "loss": 0.1817, "step": 2597 }, { "epoch": 25.191566265060242, "grad_norm": 30.052778244018555, "learning_rate": 8.738834951456311e-05, "loss": 0.3521, "step": 2598 }, { "epoch": 25.20120481927711, "grad_norm": 4.511881351470947, "learning_rate": 8.738349514563107e-05, "loss": 0.1596, "step": 2599 }, { "epoch": 25.210843373493976, "grad_norm": 10.132587432861328, "learning_rate": 8.737864077669902e-05, "loss": 0.2278, "step": 2600 }, { "epoch": 25.220481927710843, "grad_norm": 13.826497077941895, "learning_rate": 8.7373786407767e-05, "loss": 0.225, "step": 2601 }, { "epoch": 25.23012048192771, "grad_norm": 1.5125885009765625, "learning_rate": 8.736893203883497e-05, "loss": 0.2628, "step": 2602 }, { "epoch": 25.23975903614458, "grad_norm": 16.332334518432617, "learning_rate": 8.736407766990292e-05, "loss": 0.6302, "step": 2603 }, { "epoch": 25.249397590361447, "grad_norm": 26.600452423095703, "learning_rate": 8.735922330097088e-05, "loss": 0.4069, "step": 2604 }, { "epoch": 25.259036144578314, "grad_norm": 49.7502326965332, "learning_rate": 8.735436893203884e-05, "loss": 0.3286, "step": 2605 }, { "epoch": 25.26867469879518, "grad_norm": 15.290031433105469, "learning_rate": 8.73495145631068e-05, "loss": 0.5938, "step": 2606 }, { "epoch": 25.278313253012048, "grad_norm": 14.829207420349121, "learning_rate": 8.734466019417477e-05, "loss": 0.2838, "step": 2607 }, { "epoch": 25.287951807228914, "grad_norm": 28.321247100830078, "learning_rate": 8.733980582524273e-05, "loss": 0.2779, "step": 2608 }, { "epoch": 25.297590361445785, "grad_norm": 0.7608890533447266, "learning_rate": 8.733495145631068e-05, "loss": 0.3227, "step": 2609 }, { "epoch": 25.30722891566265, "grad_norm": 1.8827495574951172, "learning_rate": 8.733009708737864e-05, "loss": 0.0789, "step": 2610 }, { "epoch": 25.31686746987952, "grad_norm": 28.40556526184082, "learning_rate": 8.732524271844661e-05, "loss": 0.2156, "step": 2611 }, { "epoch": 25.326506024096386, "grad_norm": 5.663457870483398, "learning_rate": 8.732038834951457e-05, "loss": 0.2635, "step": 2612 }, { "epoch": 25.336144578313252, "grad_norm": 4.741570472717285, "learning_rate": 8.731553398058253e-05, "loss": 0.2783, "step": 2613 }, { "epoch": 25.34578313253012, "grad_norm": 1.7705707550048828, "learning_rate": 8.731067961165049e-05, "loss": 0.2196, "step": 2614 }, { "epoch": 25.355421686746986, "grad_norm": 21.791126251220703, "learning_rate": 8.730582524271846e-05, "loss": 0.4371, "step": 2615 }, { "epoch": 25.365060240963857, "grad_norm": 2.3404295444488525, "learning_rate": 8.730097087378642e-05, "loss": 0.2891, "step": 2616 }, { "epoch": 25.374698795180723, "grad_norm": 11.804431915283203, "learning_rate": 8.729611650485437e-05, "loss": 0.3322, "step": 2617 }, { "epoch": 25.38433734939759, "grad_norm": 21.588850021362305, "learning_rate": 8.729126213592233e-05, "loss": 0.3963, "step": 2618 }, { "epoch": 25.393975903614457, "grad_norm": 4.951204776763916, "learning_rate": 8.72864077669903e-05, "loss": 0.4497, "step": 2619 }, { "epoch": 25.403614457831324, "grad_norm": 15.560357093811035, "learning_rate": 8.728155339805826e-05, "loss": 0.387, "step": 2620 }, { "epoch": 25.413253012048195, "grad_norm": 31.390806198120117, "learning_rate": 8.727669902912622e-05, "loss": 0.3555, "step": 2621 }, { "epoch": 25.42289156626506, "grad_norm": 5.358893394470215, "learning_rate": 8.727184466019418e-05, "loss": 0.3523, "step": 2622 }, { "epoch": 25.43253012048193, "grad_norm": 17.3353214263916, "learning_rate": 8.726699029126213e-05, "loss": 0.3102, "step": 2623 }, { "epoch": 25.442168674698795, "grad_norm": 25.74901008605957, "learning_rate": 8.72621359223301e-05, "loss": 0.3614, "step": 2624 }, { "epoch": 25.451807228915662, "grad_norm": 7.469470024108887, "learning_rate": 8.725728155339806e-05, "loss": 0.3401, "step": 2625 }, { "epoch": 25.46144578313253, "grad_norm": 37.44545364379883, "learning_rate": 8.725242718446602e-05, "loss": 0.5246, "step": 2626 }, { "epoch": 25.471084337349396, "grad_norm": 20.30995750427246, "learning_rate": 8.724757281553398e-05, "loss": 0.4284, "step": 2627 }, { "epoch": 25.480722891566266, "grad_norm": 30.916126251220703, "learning_rate": 8.724271844660195e-05, "loss": 0.441, "step": 2628 }, { "epoch": 25.490361445783133, "grad_norm": 65.37963104248047, "learning_rate": 8.723786407766991e-05, "loss": 0.5137, "step": 2629 }, { "epoch": 25.5, "grad_norm": 10.805392265319824, "learning_rate": 8.723300970873787e-05, "loss": 0.4406, "step": 2630 }, { "epoch": 25.509638554216867, "grad_norm": 8.223143577575684, "learning_rate": 8.722815533980582e-05, "loss": 0.2571, "step": 2631 }, { "epoch": 25.519277108433734, "grad_norm": 20.975963592529297, "learning_rate": 8.72233009708738e-05, "loss": 0.265, "step": 2632 }, { "epoch": 25.528915662650604, "grad_norm": 25.320505142211914, "learning_rate": 8.721844660194175e-05, "loss": 0.5324, "step": 2633 }, { "epoch": 25.53855421686747, "grad_norm": 11.523727416992188, "learning_rate": 8.721359223300971e-05, "loss": 0.3455, "step": 2634 }, { "epoch": 25.548192771084338, "grad_norm": 25.05559539794922, "learning_rate": 8.720873786407767e-05, "loss": 0.2931, "step": 2635 }, { "epoch": 25.557831325301205, "grad_norm": 31.656810760498047, "learning_rate": 8.720388349514564e-05, "loss": 0.4335, "step": 2636 }, { "epoch": 25.56746987951807, "grad_norm": 10.734986305236816, "learning_rate": 8.71990291262136e-05, "loss": 0.3302, "step": 2637 }, { "epoch": 25.57710843373494, "grad_norm": 4.215928077697754, "learning_rate": 8.719417475728155e-05, "loss": 0.2581, "step": 2638 }, { "epoch": 25.586746987951805, "grad_norm": 15.636452674865723, "learning_rate": 8.718932038834951e-05, "loss": 0.4367, "step": 2639 }, { "epoch": 25.596385542168676, "grad_norm": 22.90837860107422, "learning_rate": 8.718446601941747e-05, "loss": 0.4959, "step": 2640 }, { "epoch": 25.606024096385543, "grad_norm": 28.82097625732422, "learning_rate": 8.717961165048544e-05, "loss": 0.5437, "step": 2641 }, { "epoch": 25.61566265060241, "grad_norm": 9.382704734802246, "learning_rate": 8.71747572815534e-05, "loss": 0.478, "step": 2642 }, { "epoch": 25.625301204819277, "grad_norm": 18.756980895996094, "learning_rate": 8.716990291262136e-05, "loss": 0.3678, "step": 2643 }, { "epoch": 25.634939759036143, "grad_norm": 36.897377014160156, "learning_rate": 8.716504854368931e-05, "loss": 0.6141, "step": 2644 }, { "epoch": 25.644578313253014, "grad_norm": 11.841390609741211, "learning_rate": 8.716019417475729e-05, "loss": 0.1986, "step": 2645 }, { "epoch": 25.65421686746988, "grad_norm": 19.75030517578125, "learning_rate": 8.715533980582524e-05, "loss": 0.381, "step": 2646 }, { "epoch": 25.663855421686748, "grad_norm": 2.8368470668792725, "learning_rate": 8.71504854368932e-05, "loss": 0.1198, "step": 2647 }, { "epoch": 25.673493975903614, "grad_norm": 10.377007484436035, "learning_rate": 8.714563106796116e-05, "loss": 0.3713, "step": 2648 }, { "epoch": 25.68313253012048, "grad_norm": 21.53451156616211, "learning_rate": 8.714077669902913e-05, "loss": 0.3317, "step": 2649 }, { "epoch": 25.69277108433735, "grad_norm": 6.906198978424072, "learning_rate": 8.713592233009709e-05, "loss": 0.4077, "step": 2650 }, { "epoch": 25.702409638554215, "grad_norm": 23.73684310913086, "learning_rate": 8.713106796116506e-05, "loss": 0.2581, "step": 2651 }, { "epoch": 25.712048192771086, "grad_norm": 5.958120346069336, "learning_rate": 8.712621359223302e-05, "loss": 0.2359, "step": 2652 }, { "epoch": 25.721686746987952, "grad_norm": 6.8374152183532715, "learning_rate": 8.712135922330098e-05, "loss": 0.4047, "step": 2653 }, { "epoch": 25.73132530120482, "grad_norm": 16.292781829833984, "learning_rate": 8.711650485436895e-05, "loss": 0.2334, "step": 2654 }, { "epoch": 25.740963855421686, "grad_norm": 25.662919998168945, "learning_rate": 8.71116504854369e-05, "loss": 0.2704, "step": 2655 }, { "epoch": 25.750602409638553, "grad_norm": 3.171391248703003, "learning_rate": 8.710679611650486e-05, "loss": 0.3001, "step": 2656 }, { "epoch": 25.760240963855424, "grad_norm": 21.493976593017578, "learning_rate": 8.710194174757282e-05, "loss": 0.3269, "step": 2657 }, { "epoch": 25.76987951807229, "grad_norm": 26.321170806884766, "learning_rate": 8.709708737864079e-05, "loss": 0.3321, "step": 2658 }, { "epoch": 25.779518072289157, "grad_norm": 89.84083557128906, "learning_rate": 8.709223300970875e-05, "loss": 0.5376, "step": 2659 }, { "epoch": 25.789156626506024, "grad_norm": 12.911752700805664, "learning_rate": 8.70873786407767e-05, "loss": 0.1782, "step": 2660 }, { "epoch": 25.79879518072289, "grad_norm": 24.782026290893555, "learning_rate": 8.708252427184466e-05, "loss": 0.4272, "step": 2661 }, { "epoch": 25.808433734939758, "grad_norm": 5.1897430419921875, "learning_rate": 8.707766990291262e-05, "loss": 0.4704, "step": 2662 }, { "epoch": 25.818072289156625, "grad_norm": 32.30056381225586, "learning_rate": 8.70728155339806e-05, "loss": 0.3807, "step": 2663 }, { "epoch": 25.827710843373495, "grad_norm": 8.552945137023926, "learning_rate": 8.706796116504855e-05, "loss": 0.376, "step": 2664 }, { "epoch": 25.837349397590362, "grad_norm": 4.782879829406738, "learning_rate": 8.706310679611651e-05, "loss": 0.3439, "step": 2665 }, { "epoch": 25.84698795180723, "grad_norm": 20.169267654418945, "learning_rate": 8.705825242718447e-05, "loss": 0.2341, "step": 2666 }, { "epoch": 25.856626506024096, "grad_norm": 6.449965000152588, "learning_rate": 8.705339805825244e-05, "loss": 0.2245, "step": 2667 }, { "epoch": 25.866265060240963, "grad_norm": 28.14720344543457, "learning_rate": 8.70485436893204e-05, "loss": 0.2073, "step": 2668 }, { "epoch": 25.87590361445783, "grad_norm": 2.475722312927246, "learning_rate": 8.704368932038835e-05, "loss": 0.1733, "step": 2669 }, { "epoch": 25.8855421686747, "grad_norm": 17.316591262817383, "learning_rate": 8.703883495145631e-05, "loss": 0.2114, "step": 2670 }, { "epoch": 25.895180722891567, "grad_norm": 26.96521759033203, "learning_rate": 8.703398058252428e-05, "loss": 0.3405, "step": 2671 }, { "epoch": 25.904819277108434, "grad_norm": 19.87383270263672, "learning_rate": 8.702912621359224e-05, "loss": 0.6302, "step": 2672 }, { "epoch": 25.9144578313253, "grad_norm": 7.018223285675049, "learning_rate": 8.70242718446602e-05, "loss": 0.4234, "step": 2673 }, { "epoch": 25.924096385542168, "grad_norm": 17.16767120361328, "learning_rate": 8.701941747572816e-05, "loss": 0.2121, "step": 2674 }, { "epoch": 25.933734939759034, "grad_norm": 39.581539154052734, "learning_rate": 8.701456310679613e-05, "loss": 0.5075, "step": 2675 }, { "epoch": 25.943373493975905, "grad_norm": 8.619064331054688, "learning_rate": 8.700970873786408e-05, "loss": 0.407, "step": 2676 }, { "epoch": 25.95301204819277, "grad_norm": 7.200294494628906, "learning_rate": 8.700485436893204e-05, "loss": 0.2692, "step": 2677 }, { "epoch": 25.96265060240964, "grad_norm": 6.644932746887207, "learning_rate": 8.7e-05, "loss": 0.4116, "step": 2678 }, { "epoch": 25.972289156626506, "grad_norm": 4.891565799713135, "learning_rate": 8.699514563106796e-05, "loss": 0.2515, "step": 2679 }, { "epoch": 25.981927710843372, "grad_norm": 28.044130325317383, "learning_rate": 8.699029126213593e-05, "loss": 0.2872, "step": 2680 }, { "epoch": 25.99156626506024, "grad_norm": 16.394561767578125, "learning_rate": 8.698543689320389e-05, "loss": 0.2464, "step": 2681 }, { "epoch": 26.00722891566265, "grad_norm": 10.349292755126953, "learning_rate": 8.698058252427185e-05, "loss": 0.4712, "step": 2682 }, { "epoch": 26.016867469879518, "grad_norm": 3.3079185485839844, "learning_rate": 8.69757281553398e-05, "loss": 0.4945, "step": 2683 }, { "epoch": 26.026506024096385, "grad_norm": 10.621663093566895, "learning_rate": 8.697087378640777e-05, "loss": 0.4357, "step": 2684 }, { "epoch": 26.03614457831325, "grad_norm": 4.3516716957092285, "learning_rate": 8.696601941747573e-05, "loss": 0.2147, "step": 2685 }, { "epoch": 26.045783132530122, "grad_norm": 6.249943733215332, "learning_rate": 8.696116504854369e-05, "loss": 0.534, "step": 2686 }, { "epoch": 26.05542168674699, "grad_norm": 12.525423049926758, "learning_rate": 8.695631067961165e-05, "loss": 0.5934, "step": 2687 }, { "epoch": 26.065060240963856, "grad_norm": 5.415981292724609, "learning_rate": 8.695145631067962e-05, "loss": 0.2821, "step": 2688 }, { "epoch": 26.074698795180723, "grad_norm": 5.253728866577148, "learning_rate": 8.694660194174758e-05, "loss": 0.2089, "step": 2689 }, { "epoch": 26.08433734939759, "grad_norm": 25.673919677734375, "learning_rate": 8.694174757281553e-05, "loss": 0.3347, "step": 2690 }, { "epoch": 26.093975903614457, "grad_norm": 7.240187168121338, "learning_rate": 8.693689320388349e-05, "loss": 0.3135, "step": 2691 }, { "epoch": 26.103614457831327, "grad_norm": 19.776472091674805, "learning_rate": 8.693203883495146e-05, "loss": 0.4498, "step": 2692 }, { "epoch": 26.113253012048194, "grad_norm": 17.527542114257812, "learning_rate": 8.692718446601942e-05, "loss": 0.4114, "step": 2693 }, { "epoch": 26.12289156626506, "grad_norm": 18.281450271606445, "learning_rate": 8.692233009708738e-05, "loss": 0.3652, "step": 2694 }, { "epoch": 26.132530120481928, "grad_norm": 5.322199821472168, "learning_rate": 8.691747572815534e-05, "loss": 0.303, "step": 2695 }, { "epoch": 26.142168674698794, "grad_norm": 54.17495346069336, "learning_rate": 8.69126213592233e-05, "loss": 0.3914, "step": 2696 }, { "epoch": 26.15180722891566, "grad_norm": 4.615414619445801, "learning_rate": 8.690776699029127e-05, "loss": 0.3519, "step": 2697 }, { "epoch": 26.16144578313253, "grad_norm": 9.957255363464355, "learning_rate": 8.690291262135922e-05, "loss": 0.3816, "step": 2698 }, { "epoch": 26.1710843373494, "grad_norm": 3.211622714996338, "learning_rate": 8.689805825242718e-05, "loss": 0.3597, "step": 2699 }, { "epoch": 26.180722891566266, "grad_norm": 19.089141845703125, "learning_rate": 8.689320388349514e-05, "loss": 0.4502, "step": 2700 }, { "epoch": 26.190361445783132, "grad_norm": 3.095336437225342, "learning_rate": 8.688834951456311e-05, "loss": 0.1913, "step": 2701 }, { "epoch": 26.2, "grad_norm": 45.24293899536133, "learning_rate": 8.688349514563108e-05, "loss": 0.4729, "step": 2702 }, { "epoch": 26.209638554216866, "grad_norm": 5.985403537750244, "learning_rate": 8.687864077669904e-05, "loss": 0.435, "step": 2703 }, { "epoch": 26.219277108433737, "grad_norm": 4.920817852020264, "learning_rate": 8.6873786407767e-05, "loss": 0.2576, "step": 2704 }, { "epoch": 26.228915662650603, "grad_norm": 20.255672454833984, "learning_rate": 8.686893203883495e-05, "loss": 0.3161, "step": 2705 }, { "epoch": 26.23855421686747, "grad_norm": 5.484216690063477, "learning_rate": 8.686407766990293e-05, "loss": 0.3783, "step": 2706 }, { "epoch": 26.248192771084337, "grad_norm": 15.27100658416748, "learning_rate": 8.685922330097088e-05, "loss": 0.4481, "step": 2707 }, { "epoch": 26.257831325301204, "grad_norm": 7.567043304443359, "learning_rate": 8.685436893203884e-05, "loss": 0.506, "step": 2708 }, { "epoch": 26.26746987951807, "grad_norm": 8.784266471862793, "learning_rate": 8.68495145631068e-05, "loss": 0.3405, "step": 2709 }, { "epoch": 26.27710843373494, "grad_norm": 7.303818702697754, "learning_rate": 8.684466019417477e-05, "loss": 0.3619, "step": 2710 }, { "epoch": 26.28674698795181, "grad_norm": 6.871300220489502, "learning_rate": 8.683980582524273e-05, "loss": 0.253, "step": 2711 }, { "epoch": 26.296385542168675, "grad_norm": 47.96216583251953, "learning_rate": 8.683495145631069e-05, "loss": 0.4178, "step": 2712 }, { "epoch": 26.306024096385542, "grad_norm": 6.228971004486084, "learning_rate": 8.683009708737864e-05, "loss": 0.214, "step": 2713 }, { "epoch": 26.31566265060241, "grad_norm": 18.44927215576172, "learning_rate": 8.682524271844662e-05, "loss": 0.3073, "step": 2714 }, { "epoch": 26.325301204819276, "grad_norm": 12.38174819946289, "learning_rate": 8.682038834951457e-05, "loss": 0.339, "step": 2715 }, { "epoch": 26.334939759036146, "grad_norm": 6.870080471038818, "learning_rate": 8.681553398058253e-05, "loss": 0.3656, "step": 2716 }, { "epoch": 26.344578313253013, "grad_norm": 2.7304935455322266, "learning_rate": 8.681067961165049e-05, "loss": 0.1822, "step": 2717 }, { "epoch": 26.35421686746988, "grad_norm": 27.59406089782715, "learning_rate": 8.680582524271845e-05, "loss": 0.4028, "step": 2718 }, { "epoch": 26.363855421686747, "grad_norm": 42.9849967956543, "learning_rate": 8.680097087378642e-05, "loss": 0.6519, "step": 2719 }, { "epoch": 26.373493975903614, "grad_norm": 20.677715301513672, "learning_rate": 8.679611650485438e-05, "loss": 0.2961, "step": 2720 }, { "epoch": 26.38313253012048, "grad_norm": 10.224438667297363, "learning_rate": 8.679126213592233e-05, "loss": 0.3001, "step": 2721 }, { "epoch": 26.39277108433735, "grad_norm": 11.224401473999023, "learning_rate": 8.678640776699029e-05, "loss": 0.2667, "step": 2722 }, { "epoch": 26.402409638554218, "grad_norm": 72.58869171142578, "learning_rate": 8.678155339805826e-05, "loss": 0.3378, "step": 2723 }, { "epoch": 26.412048192771085, "grad_norm": 17.492326736450195, "learning_rate": 8.677669902912622e-05, "loss": 0.3803, "step": 2724 }, { "epoch": 26.42168674698795, "grad_norm": 14.15921401977539, "learning_rate": 8.677184466019418e-05, "loss": 0.3038, "step": 2725 }, { "epoch": 26.43132530120482, "grad_norm": 5.767305374145508, "learning_rate": 8.676699029126214e-05, "loss": 0.1984, "step": 2726 }, { "epoch": 26.440963855421685, "grad_norm": 16.426198959350586, "learning_rate": 8.676213592233011e-05, "loss": 0.3913, "step": 2727 }, { "epoch": 26.450602409638556, "grad_norm": 9.85080623626709, "learning_rate": 8.675728155339806e-05, "loss": 0.2701, "step": 2728 }, { "epoch": 26.460240963855423, "grad_norm": 17.387407302856445, "learning_rate": 8.675242718446602e-05, "loss": 0.3322, "step": 2729 }, { "epoch": 26.46987951807229, "grad_norm": 12.492700576782227, "learning_rate": 8.674757281553398e-05, "loss": 0.2767, "step": 2730 }, { "epoch": 26.479518072289157, "grad_norm": 4.436362266540527, "learning_rate": 8.674271844660195e-05, "loss": 0.6466, "step": 2731 }, { "epoch": 26.489156626506023, "grad_norm": 20.261178970336914, "learning_rate": 8.673786407766991e-05, "loss": 0.59, "step": 2732 }, { "epoch": 26.49879518072289, "grad_norm": 9.230469703674316, "learning_rate": 8.673300970873787e-05, "loss": 0.3014, "step": 2733 }, { "epoch": 26.50843373493976, "grad_norm": 6.934201717376709, "learning_rate": 8.672815533980582e-05, "loss": 0.2571, "step": 2734 }, { "epoch": 26.518072289156628, "grad_norm": 4.341486930847168, "learning_rate": 8.672330097087378e-05, "loss": 0.3298, "step": 2735 }, { "epoch": 26.527710843373494, "grad_norm": 9.404462814331055, "learning_rate": 8.671844660194175e-05, "loss": 0.5226, "step": 2736 }, { "epoch": 26.53734939759036, "grad_norm": 9.212148666381836, "learning_rate": 8.671359223300971e-05, "loss": 0.1745, "step": 2737 }, { "epoch": 26.54698795180723, "grad_norm": 4.394214630126953, "learning_rate": 8.670873786407767e-05, "loss": 0.3513, "step": 2738 }, { "epoch": 26.556626506024095, "grad_norm": 3.5557045936584473, "learning_rate": 8.670388349514563e-05, "loss": 0.216, "step": 2739 }, { "epoch": 26.566265060240966, "grad_norm": 4.409439563751221, "learning_rate": 8.66990291262136e-05, "loss": 0.2571, "step": 2740 }, { "epoch": 26.575903614457832, "grad_norm": 10.649624824523926, "learning_rate": 8.669417475728156e-05, "loss": 0.2407, "step": 2741 }, { "epoch": 26.5855421686747, "grad_norm": 13.031817436218262, "learning_rate": 8.668932038834951e-05, "loss": 0.3861, "step": 2742 }, { "epoch": 26.595180722891566, "grad_norm": 11.286710739135742, "learning_rate": 8.668446601941747e-05, "loss": 0.3854, "step": 2743 }, { "epoch": 26.604819277108433, "grad_norm": 3.033686399459839, "learning_rate": 8.667961165048544e-05, "loss": 0.2828, "step": 2744 }, { "epoch": 26.6144578313253, "grad_norm": 7.866096019744873, "learning_rate": 8.66747572815534e-05, "loss": 0.6019, "step": 2745 }, { "epoch": 26.62409638554217, "grad_norm": 22.120153427124023, "learning_rate": 8.666990291262136e-05, "loss": 0.3526, "step": 2746 }, { "epoch": 26.633734939759037, "grad_norm": 5.668933868408203, "learning_rate": 8.666504854368932e-05, "loss": 0.3402, "step": 2747 }, { "epoch": 26.643373493975904, "grad_norm": 15.106249809265137, "learning_rate": 8.666019417475729e-05, "loss": 0.4325, "step": 2748 }, { "epoch": 26.65301204819277, "grad_norm": 5.442444324493408, "learning_rate": 8.665533980582525e-05, "loss": 0.2634, "step": 2749 }, { "epoch": 26.662650602409638, "grad_norm": 18.587539672851562, "learning_rate": 8.66504854368932e-05, "loss": 0.4064, "step": 2750 }, { "epoch": 26.672289156626505, "grad_norm": 6.040508270263672, "learning_rate": 8.664563106796116e-05, "loss": 0.4123, "step": 2751 }, { "epoch": 26.681927710843375, "grad_norm": 77.74638366699219, "learning_rate": 8.664077669902913e-05, "loss": 0.2535, "step": 2752 }, { "epoch": 26.691566265060242, "grad_norm": 6.617156505584717, "learning_rate": 8.66359223300971e-05, "loss": 0.3521, "step": 2753 }, { "epoch": 26.70120481927711, "grad_norm": 10.694194793701172, "learning_rate": 8.663106796116506e-05, "loss": 0.4147, "step": 2754 }, { "epoch": 26.710843373493976, "grad_norm": 29.077180862426758, "learning_rate": 8.662621359223302e-05, "loss": 0.4748, "step": 2755 }, { "epoch": 26.720481927710843, "grad_norm": 15.324470520019531, "learning_rate": 8.662135922330098e-05, "loss": 0.7125, "step": 2756 }, { "epoch": 26.73012048192771, "grad_norm": 30.380037307739258, "learning_rate": 8.661650485436893e-05, "loss": 0.3037, "step": 2757 }, { "epoch": 26.739759036144576, "grad_norm": 10.448144912719727, "learning_rate": 8.66116504854369e-05, "loss": 0.6388, "step": 2758 }, { "epoch": 26.749397590361447, "grad_norm": 14.643841743469238, "learning_rate": 8.660679611650486e-05, "loss": 0.4211, "step": 2759 }, { "epoch": 26.759036144578314, "grad_norm": 11.942283630371094, "learning_rate": 8.660194174757282e-05, "loss": 0.4509, "step": 2760 }, { "epoch": 26.76867469879518, "grad_norm": 2.4624991416931152, "learning_rate": 8.659708737864078e-05, "loss": 0.2372, "step": 2761 }, { "epoch": 26.778313253012048, "grad_norm": 11.333992004394531, "learning_rate": 8.659223300970875e-05, "loss": 0.3368, "step": 2762 }, { "epoch": 26.787951807228914, "grad_norm": 39.93413162231445, "learning_rate": 8.658737864077671e-05, "loss": 0.4305, "step": 2763 }, { "epoch": 26.797590361445785, "grad_norm": 13.1572904586792, "learning_rate": 8.658252427184467e-05, "loss": 0.2214, "step": 2764 }, { "epoch": 26.80722891566265, "grad_norm": 11.939718246459961, "learning_rate": 8.657766990291262e-05, "loss": 0.5229, "step": 2765 }, { "epoch": 26.81686746987952, "grad_norm": 4.938444137573242, "learning_rate": 8.65728155339806e-05, "loss": 0.4758, "step": 2766 }, { "epoch": 26.826506024096386, "grad_norm": 26.910888671875, "learning_rate": 8.656796116504855e-05, "loss": 0.3507, "step": 2767 }, { "epoch": 26.836144578313252, "grad_norm": 4.403737545013428, "learning_rate": 8.656310679611651e-05, "loss": 0.4508, "step": 2768 }, { "epoch": 26.84578313253012, "grad_norm": 24.604572296142578, "learning_rate": 8.655825242718447e-05, "loss": 0.2328, "step": 2769 }, { "epoch": 26.855421686746986, "grad_norm": 12.50339126586914, "learning_rate": 8.655339805825244e-05, "loss": 0.4993, "step": 2770 }, { "epoch": 26.865060240963857, "grad_norm": 6.462963104248047, "learning_rate": 8.65485436893204e-05, "loss": 0.3468, "step": 2771 }, { "epoch": 26.874698795180723, "grad_norm": 12.36984634399414, "learning_rate": 8.654368932038835e-05, "loss": 0.3072, "step": 2772 }, { "epoch": 26.88433734939759, "grad_norm": 21.35106658935547, "learning_rate": 8.653883495145631e-05, "loss": 0.3183, "step": 2773 }, { "epoch": 26.893975903614457, "grad_norm": 4.569802284240723, "learning_rate": 8.653398058252427e-05, "loss": 0.3408, "step": 2774 }, { "epoch": 26.903614457831324, "grad_norm": 10.695528984069824, "learning_rate": 8.652912621359224e-05, "loss": 0.3548, "step": 2775 }, { "epoch": 26.913253012048195, "grad_norm": 27.60000991821289, "learning_rate": 8.65242718446602e-05, "loss": 0.3854, "step": 2776 }, { "epoch": 26.92289156626506, "grad_norm": 12.099767684936523, "learning_rate": 8.651941747572816e-05, "loss": 0.2711, "step": 2777 }, { "epoch": 26.93253012048193, "grad_norm": 6.549857139587402, "learning_rate": 8.651456310679612e-05, "loss": 0.1675, "step": 2778 }, { "epoch": 26.942168674698795, "grad_norm": 52.0595817565918, "learning_rate": 8.650970873786409e-05, "loss": 0.3963, "step": 2779 }, { "epoch": 26.951807228915662, "grad_norm": 22.129573822021484, "learning_rate": 8.650485436893204e-05, "loss": 0.3643, "step": 2780 }, { "epoch": 26.96144578313253, "grad_norm": 5.2629780769348145, "learning_rate": 8.65e-05, "loss": 0.3502, "step": 2781 }, { "epoch": 26.971084337349396, "grad_norm": 7.590737342834473, "learning_rate": 8.649514563106796e-05, "loss": 0.2133, "step": 2782 }, { "epoch": 26.980722891566266, "grad_norm": 20.116849899291992, "learning_rate": 8.649029126213593e-05, "loss": 0.2812, "step": 2783 }, { "epoch": 26.990361445783133, "grad_norm": 7.869743347167969, "learning_rate": 8.648543689320389e-05, "loss": 0.3565, "step": 2784 }, { "epoch": 27.00602409638554, "grad_norm": 20.406387329101562, "learning_rate": 8.648058252427185e-05, "loss": 0.5697, "step": 2785 }, { "epoch": 27.01566265060241, "grad_norm": 7.788662433624268, "learning_rate": 8.64757281553398e-05, "loss": 0.4305, "step": 2786 }, { "epoch": 27.02530120481928, "grad_norm": 10.321410179138184, "learning_rate": 8.647087378640778e-05, "loss": 0.4336, "step": 2787 }, { "epoch": 27.034939759036146, "grad_norm": 6.047469139099121, "learning_rate": 8.646601941747573e-05, "loss": 0.2236, "step": 2788 }, { "epoch": 27.044578313253012, "grad_norm": 15.372793197631836, "learning_rate": 8.646116504854369e-05, "loss": 0.3456, "step": 2789 }, { "epoch": 27.05421686746988, "grad_norm": 5.882079124450684, "learning_rate": 8.645631067961165e-05, "loss": 0.3987, "step": 2790 }, { "epoch": 27.063855421686746, "grad_norm": 2.574841022491455, "learning_rate": 8.64514563106796e-05, "loss": 0.2502, "step": 2791 }, { "epoch": 27.073493975903613, "grad_norm": 10.712586402893066, "learning_rate": 8.644660194174758e-05, "loss": 0.5661, "step": 2792 }, { "epoch": 27.083132530120483, "grad_norm": 8.380182266235352, "learning_rate": 8.644174757281554e-05, "loss": 0.2317, "step": 2793 }, { "epoch": 27.09277108433735, "grad_norm": 12.586956024169922, "learning_rate": 8.64368932038835e-05, "loss": 0.313, "step": 2794 }, { "epoch": 27.102409638554217, "grad_norm": 3.5677218437194824, "learning_rate": 8.643203883495145e-05, "loss": 0.197, "step": 2795 }, { "epoch": 27.112048192771084, "grad_norm": 9.612369537353516, "learning_rate": 8.642718446601942e-05, "loss": 0.2569, "step": 2796 }, { "epoch": 27.12168674698795, "grad_norm": 17.264667510986328, "learning_rate": 8.642233009708738e-05, "loss": 0.4052, "step": 2797 }, { "epoch": 27.131325301204818, "grad_norm": 12.776878356933594, "learning_rate": 8.641747572815534e-05, "loss": 0.4303, "step": 2798 }, { "epoch": 27.14096385542169, "grad_norm": 28.569555282592773, "learning_rate": 8.64126213592233e-05, "loss": 0.3618, "step": 2799 }, { "epoch": 27.150602409638555, "grad_norm": 10.942296028137207, "learning_rate": 8.640776699029127e-05, "loss": 0.2832, "step": 2800 }, { "epoch": 27.160240963855422, "grad_norm": 15.205896377563477, "learning_rate": 8.640291262135922e-05, "loss": 0.1329, "step": 2801 }, { "epoch": 27.16987951807229, "grad_norm": 6.857314109802246, "learning_rate": 8.639805825242718e-05, "loss": 0.3054, "step": 2802 }, { "epoch": 27.179518072289156, "grad_norm": 5.797438621520996, "learning_rate": 8.639320388349515e-05, "loss": 0.3228, "step": 2803 }, { "epoch": 27.189156626506023, "grad_norm": 13.960712432861328, "learning_rate": 8.638834951456311e-05, "loss": 0.2106, "step": 2804 }, { "epoch": 27.198795180722893, "grad_norm": 9.559858322143555, "learning_rate": 8.638349514563108e-05, "loss": 0.2948, "step": 2805 }, { "epoch": 27.20843373493976, "grad_norm": 2.6710619926452637, "learning_rate": 8.637864077669904e-05, "loss": 0.2448, "step": 2806 }, { "epoch": 27.218072289156627, "grad_norm": 3.886125326156616, "learning_rate": 8.6373786407767e-05, "loss": 0.3342, "step": 2807 }, { "epoch": 27.227710843373494, "grad_norm": 30.19447135925293, "learning_rate": 8.636893203883496e-05, "loss": 0.5637, "step": 2808 }, { "epoch": 27.23734939759036, "grad_norm": 3.9394047260284424, "learning_rate": 8.636407766990293e-05, "loss": 0.348, "step": 2809 }, { "epoch": 27.246987951807228, "grad_norm": 55.2707633972168, "learning_rate": 8.635922330097089e-05, "loss": 0.5086, "step": 2810 }, { "epoch": 27.256626506024098, "grad_norm": 16.90347671508789, "learning_rate": 8.635436893203884e-05, "loss": 0.3435, "step": 2811 }, { "epoch": 27.266265060240965, "grad_norm": 19.778629302978516, "learning_rate": 8.63495145631068e-05, "loss": 0.3345, "step": 2812 }, { "epoch": 27.27590361445783, "grad_norm": 16.896642684936523, "learning_rate": 8.634466019417476e-05, "loss": 0.3143, "step": 2813 }, { "epoch": 27.2855421686747, "grad_norm": 9.113320350646973, "learning_rate": 8.633980582524273e-05, "loss": 0.2756, "step": 2814 }, { "epoch": 27.295180722891565, "grad_norm": 11.0147705078125, "learning_rate": 8.633495145631069e-05, "loss": 0.3368, "step": 2815 }, { "epoch": 27.304819277108432, "grad_norm": 20.427433013916016, "learning_rate": 8.633009708737865e-05, "loss": 0.3328, "step": 2816 }, { "epoch": 27.314457831325303, "grad_norm": 14.765094757080078, "learning_rate": 8.63252427184466e-05, "loss": 0.2065, "step": 2817 }, { "epoch": 27.32409638554217, "grad_norm": 34.33006286621094, "learning_rate": 8.632038834951457e-05, "loss": 0.4137, "step": 2818 }, { "epoch": 27.333734939759037, "grad_norm": 8.345902442932129, "learning_rate": 8.631553398058253e-05, "loss": 0.4351, "step": 2819 }, { "epoch": 27.343373493975903, "grad_norm": 4.929204940795898, "learning_rate": 8.631067961165049e-05, "loss": 0.2842, "step": 2820 }, { "epoch": 27.35301204819277, "grad_norm": 3.9907519817352295, "learning_rate": 8.630582524271845e-05, "loss": 0.223, "step": 2821 }, { "epoch": 27.362650602409637, "grad_norm": 3.9935998916625977, "learning_rate": 8.630097087378642e-05, "loss": 0.2325, "step": 2822 }, { "epoch": 27.372289156626508, "grad_norm": 7.124224662780762, "learning_rate": 8.629611650485438e-05, "loss": 0.1841, "step": 2823 }, { "epoch": 27.381927710843375, "grad_norm": 24.0301456451416, "learning_rate": 8.629126213592233e-05, "loss": 0.1627, "step": 2824 }, { "epoch": 27.39156626506024, "grad_norm": 4.829151153564453, "learning_rate": 8.628640776699029e-05, "loss": 0.2513, "step": 2825 }, { "epoch": 27.40120481927711, "grad_norm": 7.26607608795166, "learning_rate": 8.628155339805826e-05, "loss": 0.3915, "step": 2826 }, { "epoch": 27.410843373493975, "grad_norm": 7.954512119293213, "learning_rate": 8.627669902912622e-05, "loss": 0.429, "step": 2827 }, { "epoch": 27.420481927710842, "grad_norm": 11.731574058532715, "learning_rate": 8.627184466019418e-05, "loss": 0.4364, "step": 2828 }, { "epoch": 27.430120481927712, "grad_norm": 22.015949249267578, "learning_rate": 8.626699029126214e-05, "loss": 0.3653, "step": 2829 }, { "epoch": 27.43975903614458, "grad_norm": 9.933445930480957, "learning_rate": 8.62621359223301e-05, "loss": 0.5847, "step": 2830 }, { "epoch": 27.449397590361446, "grad_norm": 3.0563886165618896, "learning_rate": 8.625728155339807e-05, "loss": 0.1753, "step": 2831 }, { "epoch": 27.459036144578313, "grad_norm": 4.6682658195495605, "learning_rate": 8.625242718446602e-05, "loss": 0.3622, "step": 2832 }, { "epoch": 27.46867469879518, "grad_norm": 52.90237808227539, "learning_rate": 8.624757281553398e-05, "loss": 0.6009, "step": 2833 }, { "epoch": 27.478313253012047, "grad_norm": 8.161681175231934, "learning_rate": 8.624271844660194e-05, "loss": 0.1505, "step": 2834 }, { "epoch": 27.487951807228917, "grad_norm": 17.49845314025879, "learning_rate": 8.623786407766991e-05, "loss": 0.4582, "step": 2835 }, { "epoch": 27.497590361445784, "grad_norm": 3.939586639404297, "learning_rate": 8.623300970873787e-05, "loss": 0.2937, "step": 2836 }, { "epoch": 27.50722891566265, "grad_norm": 6.029016971588135, "learning_rate": 8.622815533980583e-05, "loss": 0.3955, "step": 2837 }, { "epoch": 27.516867469879518, "grad_norm": 5.0929036140441895, "learning_rate": 8.622330097087378e-05, "loss": 0.3093, "step": 2838 }, { "epoch": 27.526506024096385, "grad_norm": 7.777419567108154, "learning_rate": 8.621844660194176e-05, "loss": 0.271, "step": 2839 }, { "epoch": 27.53614457831325, "grad_norm": 27.363086700439453, "learning_rate": 8.621359223300971e-05, "loss": 0.4292, "step": 2840 }, { "epoch": 27.545783132530122, "grad_norm": 6.935954570770264, "learning_rate": 8.620873786407767e-05, "loss": 0.28, "step": 2841 }, { "epoch": 27.55542168674699, "grad_norm": 14.086926460266113, "learning_rate": 8.620388349514563e-05, "loss": 0.631, "step": 2842 }, { "epoch": 27.565060240963856, "grad_norm": 1.4322339296340942, "learning_rate": 8.619902912621359e-05, "loss": 0.1568, "step": 2843 }, { "epoch": 27.574698795180723, "grad_norm": 19.142501831054688, "learning_rate": 8.619417475728156e-05, "loss": 0.3125, "step": 2844 }, { "epoch": 27.58433734939759, "grad_norm": 8.415404319763184, "learning_rate": 8.618932038834952e-05, "loss": 0.1303, "step": 2845 }, { "epoch": 27.593975903614457, "grad_norm": 12.108388900756836, "learning_rate": 8.618446601941747e-05, "loss": 0.4389, "step": 2846 }, { "epoch": 27.603614457831327, "grad_norm": 12.301694869995117, "learning_rate": 8.617961165048543e-05, "loss": 0.2797, "step": 2847 }, { "epoch": 27.613253012048194, "grad_norm": 15.611754417419434, "learning_rate": 8.61747572815534e-05, "loss": 0.3638, "step": 2848 }, { "epoch": 27.62289156626506, "grad_norm": 11.776840209960938, "learning_rate": 8.616990291262136e-05, "loss": 0.5579, "step": 2849 }, { "epoch": 27.632530120481928, "grad_norm": 12.593303680419922, "learning_rate": 8.616504854368932e-05, "loss": 0.3036, "step": 2850 }, { "epoch": 27.642168674698794, "grad_norm": 18.902782440185547, "learning_rate": 8.616019417475728e-05, "loss": 0.4814, "step": 2851 }, { "epoch": 27.65180722891566, "grad_norm": 23.853729248046875, "learning_rate": 8.615533980582525e-05, "loss": 0.5175, "step": 2852 }, { "epoch": 27.66144578313253, "grad_norm": 5.227821350097656, "learning_rate": 8.615048543689322e-05, "loss": 0.2962, "step": 2853 }, { "epoch": 27.6710843373494, "grad_norm": 16.23773193359375, "learning_rate": 8.614563106796118e-05, "loss": 0.4656, "step": 2854 }, { "epoch": 27.680722891566266, "grad_norm": 28.030059814453125, "learning_rate": 8.614077669902913e-05, "loss": 0.5006, "step": 2855 }, { "epoch": 27.690361445783132, "grad_norm": 23.352182388305664, "learning_rate": 8.613592233009709e-05, "loss": 0.2299, "step": 2856 }, { "epoch": 27.7, "grad_norm": 8.857762336730957, "learning_rate": 8.613106796116506e-05, "loss": 0.3987, "step": 2857 }, { "epoch": 27.709638554216866, "grad_norm": 4.9631195068359375, "learning_rate": 8.612621359223302e-05, "loss": 0.3952, "step": 2858 }, { "epoch": 27.719277108433737, "grad_norm": 3.0207300186157227, "learning_rate": 8.612135922330098e-05, "loss": 0.3223, "step": 2859 }, { "epoch": 27.728915662650603, "grad_norm": 2.8613853454589844, "learning_rate": 8.611650485436894e-05, "loss": 0.2194, "step": 2860 }, { "epoch": 27.73855421686747, "grad_norm": 2.392639636993408, "learning_rate": 8.611165048543691e-05, "loss": 0.2811, "step": 2861 }, { "epoch": 27.748192771084337, "grad_norm": 16.669513702392578, "learning_rate": 8.610679611650486e-05, "loss": 0.2557, "step": 2862 }, { "epoch": 27.757831325301204, "grad_norm": 9.46247386932373, "learning_rate": 8.610194174757282e-05, "loss": 0.2322, "step": 2863 }, { "epoch": 27.76746987951807, "grad_norm": 15.99009895324707, "learning_rate": 8.609708737864078e-05, "loss": 0.2595, "step": 2864 }, { "epoch": 27.77710843373494, "grad_norm": 7.769036293029785, "learning_rate": 8.609223300970874e-05, "loss": 0.3093, "step": 2865 }, { "epoch": 27.78674698795181, "grad_norm": 63.769378662109375, "learning_rate": 8.608737864077671e-05, "loss": 0.5704, "step": 2866 }, { "epoch": 27.796385542168675, "grad_norm": 16.838245391845703, "learning_rate": 8.608252427184467e-05, "loss": 0.3829, "step": 2867 }, { "epoch": 27.806024096385542, "grad_norm": 14.420866966247559, "learning_rate": 8.607766990291262e-05, "loss": 0.2857, "step": 2868 }, { "epoch": 27.81566265060241, "grad_norm": 21.701997756958008, "learning_rate": 8.607281553398058e-05, "loss": 0.3953, "step": 2869 }, { "epoch": 27.825301204819276, "grad_norm": 15.979990005493164, "learning_rate": 8.606796116504855e-05, "loss": 0.4335, "step": 2870 }, { "epoch": 27.834939759036146, "grad_norm": 10.505325317382812, "learning_rate": 8.606310679611651e-05, "loss": 0.4044, "step": 2871 }, { "epoch": 27.844578313253013, "grad_norm": 4.555185794830322, "learning_rate": 8.605825242718447e-05, "loss": 0.2202, "step": 2872 }, { "epoch": 27.85421686746988, "grad_norm": 10.597148895263672, "learning_rate": 8.605339805825243e-05, "loss": 0.6074, "step": 2873 }, { "epoch": 27.863855421686747, "grad_norm": 17.31587791442871, "learning_rate": 8.60485436893204e-05, "loss": 0.4726, "step": 2874 }, { "epoch": 27.873493975903614, "grad_norm": 24.93529510498047, "learning_rate": 8.604368932038836e-05, "loss": 0.4486, "step": 2875 }, { "epoch": 27.88313253012048, "grad_norm": 10.2748384475708, "learning_rate": 8.603883495145631e-05, "loss": 0.3961, "step": 2876 }, { "epoch": 27.89277108433735, "grad_norm": 9.691417694091797, "learning_rate": 8.603398058252427e-05, "loss": 0.2876, "step": 2877 }, { "epoch": 27.902409638554218, "grad_norm": 2.782005548477173, "learning_rate": 8.602912621359224e-05, "loss": 0.1963, "step": 2878 }, { "epoch": 27.912048192771085, "grad_norm": 6.8931169509887695, "learning_rate": 8.60242718446602e-05, "loss": 0.1281, "step": 2879 }, { "epoch": 27.92168674698795, "grad_norm": 4.155185699462891, "learning_rate": 8.601941747572816e-05, "loss": 0.2971, "step": 2880 }, { "epoch": 27.93132530120482, "grad_norm": 8.240036010742188, "learning_rate": 8.601456310679612e-05, "loss": 0.3163, "step": 2881 }, { "epoch": 27.940963855421685, "grad_norm": 7.104522228240967, "learning_rate": 8.600970873786407e-05, "loss": 0.3777, "step": 2882 }, { "epoch": 27.950602409638556, "grad_norm": 10.49619197845459, "learning_rate": 8.600485436893205e-05, "loss": 0.3706, "step": 2883 }, { "epoch": 27.960240963855423, "grad_norm": 14.344712257385254, "learning_rate": 8.6e-05, "loss": 0.365, "step": 2884 }, { "epoch": 27.96987951807229, "grad_norm": 15.016012191772461, "learning_rate": 8.599514563106796e-05, "loss": 0.3012, "step": 2885 }, { "epoch": 27.979518072289157, "grad_norm": 5.277706623077393, "learning_rate": 8.599029126213592e-05, "loss": 0.3836, "step": 2886 }, { "epoch": 27.989156626506023, "grad_norm": 7.363961696624756, "learning_rate": 8.598543689320389e-05, "loss": 0.3934, "step": 2887 }, { "epoch": 28.004819277108435, "grad_norm": 10.905040740966797, "learning_rate": 8.598058252427185e-05, "loss": 0.3847, "step": 2888 }, { "epoch": 28.014457831325302, "grad_norm": 7.2057576179504395, "learning_rate": 8.59757281553398e-05, "loss": 0.2437, "step": 2889 }, { "epoch": 28.02409638554217, "grad_norm": 22.072303771972656, "learning_rate": 8.597087378640776e-05, "loss": 0.2772, "step": 2890 }, { "epoch": 28.033734939759036, "grad_norm": 23.5151309967041, "learning_rate": 8.596601941747573e-05, "loss": 0.3663, "step": 2891 }, { "epoch": 28.043373493975903, "grad_norm": 14.47026252746582, "learning_rate": 8.596116504854369e-05, "loss": 0.6183, "step": 2892 }, { "epoch": 28.05301204819277, "grad_norm": 3.450645923614502, "learning_rate": 8.595631067961165e-05, "loss": 0.3228, "step": 2893 }, { "epoch": 28.06265060240964, "grad_norm": 6.464987277984619, "learning_rate": 8.595145631067961e-05, "loss": 0.2191, "step": 2894 }, { "epoch": 28.072289156626507, "grad_norm": 4.975175857543945, "learning_rate": 8.594660194174758e-05, "loss": 0.4701, "step": 2895 }, { "epoch": 28.081927710843374, "grad_norm": 8.185086250305176, "learning_rate": 8.594174757281554e-05, "loss": 0.6048, "step": 2896 }, { "epoch": 28.09156626506024, "grad_norm": 12.119087219238281, "learning_rate": 8.59368932038835e-05, "loss": 0.4446, "step": 2897 }, { "epoch": 28.101204819277108, "grad_norm": 19.452978134155273, "learning_rate": 8.593203883495145e-05, "loss": 0.3486, "step": 2898 }, { "epoch": 28.110843373493974, "grad_norm": 8.564955711364746, "learning_rate": 8.592718446601941e-05, "loss": 0.2678, "step": 2899 }, { "epoch": 28.120481927710845, "grad_norm": 25.03343391418457, "learning_rate": 8.592233009708738e-05, "loss": 0.3877, "step": 2900 }, { "epoch": 28.13012048192771, "grad_norm": 64.899658203125, "learning_rate": 8.591747572815534e-05, "loss": 0.3291, "step": 2901 }, { "epoch": 28.13975903614458, "grad_norm": 3.175647258758545, "learning_rate": 8.59126213592233e-05, "loss": 0.1862, "step": 2902 }, { "epoch": 28.149397590361446, "grad_norm": 46.21463394165039, "learning_rate": 8.590776699029125e-05, "loss": 0.3419, "step": 2903 }, { "epoch": 28.159036144578312, "grad_norm": 11.553979873657227, "learning_rate": 8.590291262135923e-05, "loss": 0.421, "step": 2904 }, { "epoch": 28.16867469879518, "grad_norm": 4.514410972595215, "learning_rate": 8.58980582524272e-05, "loss": 0.1516, "step": 2905 }, { "epoch": 28.17831325301205, "grad_norm": 7.930041313171387, "learning_rate": 8.589320388349516e-05, "loss": 0.5077, "step": 2906 }, { "epoch": 28.187951807228917, "grad_norm": 5.916649341583252, "learning_rate": 8.588834951456311e-05, "loss": 0.4513, "step": 2907 }, { "epoch": 28.197590361445783, "grad_norm": 8.695697784423828, "learning_rate": 8.588349514563107e-05, "loss": 0.4369, "step": 2908 }, { "epoch": 28.20722891566265, "grad_norm": 4.273078918457031, "learning_rate": 8.587864077669904e-05, "loss": 0.315, "step": 2909 }, { "epoch": 28.216867469879517, "grad_norm": 9.097066879272461, "learning_rate": 8.5873786407767e-05, "loss": 0.2999, "step": 2910 }, { "epoch": 28.226506024096384, "grad_norm": 6.651758193969727, "learning_rate": 8.586893203883496e-05, "loss": 0.1555, "step": 2911 }, { "epoch": 28.236144578313255, "grad_norm": 3.058601140975952, "learning_rate": 8.586407766990292e-05, "loss": 0.1798, "step": 2912 }, { "epoch": 28.24578313253012, "grad_norm": 3.9574921131134033, "learning_rate": 8.585922330097089e-05, "loss": 0.363, "step": 2913 }, { "epoch": 28.25542168674699, "grad_norm": 11.274078369140625, "learning_rate": 8.585436893203884e-05, "loss": 0.4224, "step": 2914 }, { "epoch": 28.265060240963855, "grad_norm": 14.63796615600586, "learning_rate": 8.58495145631068e-05, "loss": 0.3786, "step": 2915 }, { "epoch": 28.274698795180722, "grad_norm": 6.005289077758789, "learning_rate": 8.584466019417476e-05, "loss": 0.3454, "step": 2916 }, { "epoch": 28.28433734939759, "grad_norm": 8.698692321777344, "learning_rate": 8.583980582524273e-05, "loss": 0.4661, "step": 2917 }, { "epoch": 28.29397590361446, "grad_norm": 17.316452026367188, "learning_rate": 8.583495145631069e-05, "loss": 0.2661, "step": 2918 }, { "epoch": 28.303614457831326, "grad_norm": 26.727895736694336, "learning_rate": 8.583009708737865e-05, "loss": 0.3245, "step": 2919 }, { "epoch": 28.313253012048193, "grad_norm": 6.808135032653809, "learning_rate": 8.58252427184466e-05, "loss": 0.512, "step": 2920 }, { "epoch": 28.32289156626506, "grad_norm": 14.730539321899414, "learning_rate": 8.582038834951456e-05, "loss": 0.5134, "step": 2921 }, { "epoch": 28.332530120481927, "grad_norm": 27.456445693969727, "learning_rate": 8.581553398058253e-05, "loss": 0.3959, "step": 2922 }, { "epoch": 28.342168674698794, "grad_norm": 5.546400547027588, "learning_rate": 8.581067961165049e-05, "loss": 0.3752, "step": 2923 }, { "epoch": 28.351807228915664, "grad_norm": 6.427655220031738, "learning_rate": 8.580582524271845e-05, "loss": 0.4953, "step": 2924 }, { "epoch": 28.36144578313253, "grad_norm": 20.344276428222656, "learning_rate": 8.580097087378641e-05, "loss": 0.3683, "step": 2925 }, { "epoch": 28.371084337349398, "grad_norm": 4.481100082397461, "learning_rate": 8.579611650485438e-05, "loss": 0.3407, "step": 2926 }, { "epoch": 28.380722891566265, "grad_norm": 6.600910663604736, "learning_rate": 8.579126213592234e-05, "loss": 0.2017, "step": 2927 }, { "epoch": 28.39036144578313, "grad_norm": 13.50104808807373, "learning_rate": 8.57864077669903e-05, "loss": 0.5541, "step": 2928 }, { "epoch": 28.4, "grad_norm": 12.858631134033203, "learning_rate": 8.578155339805825e-05, "loss": 0.3544, "step": 2929 }, { "epoch": 28.40963855421687, "grad_norm": 8.25964641571045, "learning_rate": 8.577669902912622e-05, "loss": 0.3663, "step": 2930 }, { "epoch": 28.419277108433736, "grad_norm": 15.049766540527344, "learning_rate": 8.577184466019418e-05, "loss": 0.3087, "step": 2931 }, { "epoch": 28.428915662650603, "grad_norm": 9.599963188171387, "learning_rate": 8.576699029126214e-05, "loss": 0.4191, "step": 2932 }, { "epoch": 28.43855421686747, "grad_norm": 10.358292579650879, "learning_rate": 8.57621359223301e-05, "loss": 0.3072, "step": 2933 }, { "epoch": 28.448192771084337, "grad_norm": 14.062864303588867, "learning_rate": 8.575728155339807e-05, "loss": 0.4306, "step": 2934 }, { "epoch": 28.457831325301203, "grad_norm": 15.20287036895752, "learning_rate": 8.575242718446603e-05, "loss": 0.4153, "step": 2935 }, { "epoch": 28.467469879518074, "grad_norm": 5.9701924324035645, "learning_rate": 8.574757281553398e-05, "loss": 0.2702, "step": 2936 }, { "epoch": 28.47710843373494, "grad_norm": 8.931381225585938, "learning_rate": 8.574271844660194e-05, "loss": 0.45, "step": 2937 }, { "epoch": 28.486746987951808, "grad_norm": 13.579507827758789, "learning_rate": 8.57378640776699e-05, "loss": 0.4779, "step": 2938 }, { "epoch": 28.496385542168674, "grad_norm": 3.5002095699310303, "learning_rate": 8.573300970873787e-05, "loss": 0.2504, "step": 2939 }, { "epoch": 28.50602409638554, "grad_norm": 9.169312477111816, "learning_rate": 8.572815533980583e-05, "loss": 0.2798, "step": 2940 }, { "epoch": 28.51566265060241, "grad_norm": 10.638721466064453, "learning_rate": 8.572330097087379e-05, "loss": 0.3214, "step": 2941 }, { "epoch": 28.52530120481928, "grad_norm": 3.814138650894165, "learning_rate": 8.571844660194174e-05, "loss": 0.2214, "step": 2942 }, { "epoch": 28.534939759036146, "grad_norm": 7.422600269317627, "learning_rate": 8.571359223300971e-05, "loss": 0.354, "step": 2943 }, { "epoch": 28.544578313253012, "grad_norm": 15.272479057312012, "learning_rate": 8.570873786407767e-05, "loss": 0.372, "step": 2944 }, { "epoch": 28.55421686746988, "grad_norm": 11.283559799194336, "learning_rate": 8.570388349514563e-05, "loss": 0.3979, "step": 2945 }, { "epoch": 28.563855421686746, "grad_norm": 48.7728271484375, "learning_rate": 8.569902912621359e-05, "loss": 0.3506, "step": 2946 }, { "epoch": 28.573493975903613, "grad_norm": 17.062244415283203, "learning_rate": 8.569417475728156e-05, "loss": 0.267, "step": 2947 }, { "epoch": 28.583132530120483, "grad_norm": 7.213787078857422, "learning_rate": 8.568932038834952e-05, "loss": 0.3048, "step": 2948 }, { "epoch": 28.59277108433735, "grad_norm": 7.894152641296387, "learning_rate": 8.568446601941747e-05, "loss": 0.3679, "step": 2949 }, { "epoch": 28.602409638554217, "grad_norm": 14.474533081054688, "learning_rate": 8.567961165048543e-05, "loss": 0.3437, "step": 2950 }, { "epoch": 28.612048192771084, "grad_norm": 9.343036651611328, "learning_rate": 8.56747572815534e-05, "loss": 0.3263, "step": 2951 }, { "epoch": 28.62168674698795, "grad_norm": 13.417638778686523, "learning_rate": 8.566990291262136e-05, "loss": 0.4043, "step": 2952 }, { "epoch": 28.631325301204818, "grad_norm": 4.0008931159973145, "learning_rate": 8.566504854368932e-05, "loss": 0.3418, "step": 2953 }, { "epoch": 28.64096385542169, "grad_norm": 11.675044059753418, "learning_rate": 8.566019417475729e-05, "loss": 0.3569, "step": 2954 }, { "epoch": 28.650602409638555, "grad_norm": 8.922859191894531, "learning_rate": 8.565533980582525e-05, "loss": 0.2707, "step": 2955 }, { "epoch": 28.660240963855422, "grad_norm": 16.440942764282227, "learning_rate": 8.565048543689322e-05, "loss": 0.4667, "step": 2956 }, { "epoch": 28.66987951807229, "grad_norm": 7.144199371337891, "learning_rate": 8.564563106796118e-05, "loss": 0.2415, "step": 2957 }, { "epoch": 28.679518072289156, "grad_norm": 5.612502098083496, "learning_rate": 8.564077669902913e-05, "loss": 0.3226, "step": 2958 }, { "epoch": 28.689156626506023, "grad_norm": 19.61693572998047, "learning_rate": 8.563592233009709e-05, "loss": 0.4663, "step": 2959 }, { "epoch": 28.698795180722893, "grad_norm": 13.395596504211426, "learning_rate": 8.563106796116505e-05, "loss": 0.284, "step": 2960 }, { "epoch": 28.70843373493976, "grad_norm": 3.5202789306640625, "learning_rate": 8.562621359223302e-05, "loss": 0.2349, "step": 2961 }, { "epoch": 28.718072289156627, "grad_norm": 12.998739242553711, "learning_rate": 8.562135922330098e-05, "loss": 0.5235, "step": 2962 }, { "epoch": 28.727710843373494, "grad_norm": 15.850679397583008, "learning_rate": 8.561650485436894e-05, "loss": 0.4784, "step": 2963 }, { "epoch": 28.73734939759036, "grad_norm": 19.151254653930664, "learning_rate": 8.56116504854369e-05, "loss": 0.5245, "step": 2964 }, { "epoch": 28.746987951807228, "grad_norm": 9.348943710327148, "learning_rate": 8.560679611650487e-05, "loss": 0.1754, "step": 2965 }, { "epoch": 28.756626506024098, "grad_norm": 35.660987854003906, "learning_rate": 8.560194174757282e-05, "loss": 0.39, "step": 2966 }, { "epoch": 28.766265060240965, "grad_norm": 12.261303901672363, "learning_rate": 8.559708737864078e-05, "loss": 0.3021, "step": 2967 }, { "epoch": 28.77590361445783, "grad_norm": 11.636183738708496, "learning_rate": 8.559223300970874e-05, "loss": 0.2827, "step": 2968 }, { "epoch": 28.7855421686747, "grad_norm": 6.058892250061035, "learning_rate": 8.558737864077671e-05, "loss": 0.3436, "step": 2969 }, { "epoch": 28.795180722891565, "grad_norm": 10.139787673950195, "learning_rate": 8.558252427184467e-05, "loss": 0.2261, "step": 2970 }, { "epoch": 28.804819277108432, "grad_norm": 3.5248608589172363, "learning_rate": 8.557766990291263e-05, "loss": 0.1811, "step": 2971 }, { "epoch": 28.814457831325303, "grad_norm": 4.070901870727539, "learning_rate": 8.557281553398058e-05, "loss": 0.1966, "step": 2972 }, { "epoch": 28.82409638554217, "grad_norm": 18.580781936645508, "learning_rate": 8.556796116504856e-05, "loss": 0.4204, "step": 2973 }, { "epoch": 28.833734939759037, "grad_norm": 35.9090461730957, "learning_rate": 8.556310679611651e-05, "loss": 0.3418, "step": 2974 }, { "epoch": 28.843373493975903, "grad_norm": 13.18864631652832, "learning_rate": 8.555825242718447e-05, "loss": 0.5969, "step": 2975 }, { "epoch": 28.85301204819277, "grad_norm": 6.4267802238464355, "learning_rate": 8.555339805825243e-05, "loss": 0.3065, "step": 2976 }, { "epoch": 28.862650602409637, "grad_norm": 9.215806007385254, "learning_rate": 8.554854368932039e-05, "loss": 0.359, "step": 2977 }, { "epoch": 28.872289156626508, "grad_norm": 9.167598724365234, "learning_rate": 8.554368932038836e-05, "loss": 0.4642, "step": 2978 }, { "epoch": 28.881927710843375, "grad_norm": 8.446127891540527, "learning_rate": 8.553883495145632e-05, "loss": 0.4355, "step": 2979 }, { "epoch": 28.89156626506024, "grad_norm": 18.242685317993164, "learning_rate": 8.553398058252427e-05, "loss": 0.4034, "step": 2980 }, { "epoch": 28.90120481927711, "grad_norm": 14.284258842468262, "learning_rate": 8.552912621359223e-05, "loss": 0.2418, "step": 2981 }, { "epoch": 28.910843373493975, "grad_norm": 18.112340927124023, "learning_rate": 8.55242718446602e-05, "loss": 0.3953, "step": 2982 }, { "epoch": 28.920481927710842, "grad_norm": 11.242133140563965, "learning_rate": 8.551941747572816e-05, "loss": 0.2622, "step": 2983 }, { "epoch": 28.930120481927712, "grad_norm": 8.591757774353027, "learning_rate": 8.551456310679612e-05, "loss": 0.2734, "step": 2984 }, { "epoch": 28.93975903614458, "grad_norm": 2.900252342224121, "learning_rate": 8.550970873786408e-05, "loss": 0.2819, "step": 2985 }, { "epoch": 28.949397590361446, "grad_norm": 7.803743839263916, "learning_rate": 8.550485436893205e-05, "loss": 0.3184, "step": 2986 }, { "epoch": 28.959036144578313, "grad_norm": 11.234395027160645, "learning_rate": 8.55e-05, "loss": 0.2417, "step": 2987 }, { "epoch": 28.96867469879518, "grad_norm": 5.834792613983154, "learning_rate": 8.549514563106796e-05, "loss": 0.2734, "step": 2988 }, { "epoch": 28.978313253012047, "grad_norm": 43.462127685546875, "learning_rate": 8.549029126213592e-05, "loss": 0.2246, "step": 2989 }, { "epoch": 28.987951807228917, "grad_norm": 2.637827157974243, "learning_rate": 8.548543689320389e-05, "loss": 0.2287, "step": 2990 }, { "epoch": 29.003614457831326, "grad_norm": 3.178344249725342, "learning_rate": 8.548058252427185e-05, "loss": 0.2071, "step": 2991 }, { "epoch": 29.013253012048192, "grad_norm": 23.80452537536621, "learning_rate": 8.547572815533981e-05, "loss": 0.4311, "step": 2992 }, { "epoch": 29.02289156626506, "grad_norm": 8.007723808288574, "learning_rate": 8.547087378640776e-05, "loss": 0.3812, "step": 2993 }, { "epoch": 29.032530120481926, "grad_norm": 9.923715591430664, "learning_rate": 8.546601941747572e-05, "loss": 0.3087, "step": 2994 }, { "epoch": 29.042168674698797, "grad_norm": 10.874634742736816, "learning_rate": 8.54611650485437e-05, "loss": 0.4471, "step": 2995 }, { "epoch": 29.051807228915663, "grad_norm": 15.939070701599121, "learning_rate": 8.545631067961165e-05, "loss": 0.5492, "step": 2996 }, { "epoch": 29.06144578313253, "grad_norm": 7.2166852951049805, "learning_rate": 8.545145631067961e-05, "loss": 0.414, "step": 2997 }, { "epoch": 29.071084337349397, "grad_norm": 7.5202813148498535, "learning_rate": 8.544660194174757e-05, "loss": 0.3606, "step": 2998 }, { "epoch": 29.080722891566264, "grad_norm": 9.761240005493164, "learning_rate": 8.544174757281554e-05, "loss": 0.4337, "step": 2999 }, { "epoch": 29.09036144578313, "grad_norm": 11.99882984161377, "learning_rate": 8.54368932038835e-05, "loss": 0.4207, "step": 3000 }, { "epoch": 29.1, "grad_norm": 25.32556915283203, "learning_rate": 8.543203883495145e-05, "loss": 0.2212, "step": 3001 }, { "epoch": 29.10963855421687, "grad_norm": 6.292162895202637, "learning_rate": 8.542718446601941e-05, "loss": 0.2876, "step": 3002 }, { "epoch": 29.119277108433735, "grad_norm": 4.94800329208374, "learning_rate": 8.542233009708738e-05, "loss": 0.2513, "step": 3003 }, { "epoch": 29.128915662650602, "grad_norm": 10.216076850891113, "learning_rate": 8.541747572815534e-05, "loss": 0.2647, "step": 3004 }, { "epoch": 29.13855421686747, "grad_norm": 10.397260665893555, "learning_rate": 8.541262135922331e-05, "loss": 0.5453, "step": 3005 }, { "epoch": 29.148192771084336, "grad_norm": 6.268781661987305, "learning_rate": 8.540776699029127e-05, "loss": 0.4071, "step": 3006 }, { "epoch": 29.157831325301206, "grad_norm": 6.945646286010742, "learning_rate": 8.540291262135923e-05, "loss": 0.3255, "step": 3007 }, { "epoch": 29.167469879518073, "grad_norm": 4.759183406829834, "learning_rate": 8.53980582524272e-05, "loss": 0.4222, "step": 3008 }, { "epoch": 29.17710843373494, "grad_norm": 6.664562225341797, "learning_rate": 8.539320388349516e-05, "loss": 0.1608, "step": 3009 }, { "epoch": 29.186746987951807, "grad_norm": 6.169371604919434, "learning_rate": 8.538834951456311e-05, "loss": 0.2646, "step": 3010 }, { "epoch": 29.196385542168674, "grad_norm": 15.404926300048828, "learning_rate": 8.538349514563107e-05, "loss": 0.2267, "step": 3011 }, { "epoch": 29.20602409638554, "grad_norm": 6.040475845336914, "learning_rate": 8.537864077669904e-05, "loss": 0.2475, "step": 3012 }, { "epoch": 29.21566265060241, "grad_norm": 14.314696311950684, "learning_rate": 8.5373786407767e-05, "loss": 0.5123, "step": 3013 }, { "epoch": 29.225301204819278, "grad_norm": 20.88374900817871, "learning_rate": 8.536893203883496e-05, "loss": 0.3646, "step": 3014 }, { "epoch": 29.234939759036145, "grad_norm": 7.520414352416992, "learning_rate": 8.536407766990292e-05, "loss": 0.4913, "step": 3015 }, { "epoch": 29.24457831325301, "grad_norm": 11.695136070251465, "learning_rate": 8.535922330097087e-05, "loss": 0.468, "step": 3016 }, { "epoch": 29.25421686746988, "grad_norm": 7.176530361175537, "learning_rate": 8.535436893203885e-05, "loss": 0.3293, "step": 3017 }, { "epoch": 29.263855421686745, "grad_norm": 6.520160675048828, "learning_rate": 8.53495145631068e-05, "loss": 0.4217, "step": 3018 }, { "epoch": 29.273493975903616, "grad_norm": 3.2545807361602783, "learning_rate": 8.534466019417476e-05, "loss": 0.3089, "step": 3019 }, { "epoch": 29.283132530120483, "grad_norm": 5.769707202911377, "learning_rate": 8.533980582524272e-05, "loss": 0.3082, "step": 3020 }, { "epoch": 29.29277108433735, "grad_norm": 12.225700378417969, "learning_rate": 8.533495145631069e-05, "loss": 0.3799, "step": 3021 }, { "epoch": 29.302409638554217, "grad_norm": 7.259023666381836, "learning_rate": 8.533009708737865e-05, "loss": 0.2698, "step": 3022 }, { "epoch": 29.312048192771083, "grad_norm": 5.7515387535095215, "learning_rate": 8.53252427184466e-05, "loss": 0.3535, "step": 3023 }, { "epoch": 29.32168674698795, "grad_norm": 2.978097677230835, "learning_rate": 8.532038834951456e-05, "loss": 0.1994, "step": 3024 }, { "epoch": 29.33132530120482, "grad_norm": 4.878520965576172, "learning_rate": 8.531553398058254e-05, "loss": 0.4018, "step": 3025 }, { "epoch": 29.340963855421688, "grad_norm": 6.466853141784668, "learning_rate": 8.531067961165049e-05, "loss": 0.2258, "step": 3026 }, { "epoch": 29.350602409638554, "grad_norm": 7.976125240325928, "learning_rate": 8.530582524271845e-05, "loss": 0.2288, "step": 3027 }, { "epoch": 29.36024096385542, "grad_norm": 4.406592845916748, "learning_rate": 8.530097087378641e-05, "loss": 0.296, "step": 3028 }, { "epoch": 29.36987951807229, "grad_norm": 11.641940116882324, "learning_rate": 8.529611650485438e-05, "loss": 0.3653, "step": 3029 }, { "epoch": 29.379518072289155, "grad_norm": 24.223018646240234, "learning_rate": 8.529126213592234e-05, "loss": 0.1808, "step": 3030 }, { "epoch": 29.389156626506026, "grad_norm": 14.747148513793945, "learning_rate": 8.52864077669903e-05, "loss": 0.3923, "step": 3031 }, { "epoch": 29.398795180722892, "grad_norm": 12.363137245178223, "learning_rate": 8.528155339805825e-05, "loss": 0.3319, "step": 3032 }, { "epoch": 29.40843373493976, "grad_norm": 13.089143753051758, "learning_rate": 8.527669902912621e-05, "loss": 0.3028, "step": 3033 }, { "epoch": 29.418072289156626, "grad_norm": 16.569236755371094, "learning_rate": 8.527184466019418e-05, "loss": 0.354, "step": 3034 }, { "epoch": 29.427710843373493, "grad_norm": 21.62982177734375, "learning_rate": 8.526699029126214e-05, "loss": 0.4375, "step": 3035 }, { "epoch": 29.43734939759036, "grad_norm": 4.689580917358398, "learning_rate": 8.52621359223301e-05, "loss": 0.3691, "step": 3036 }, { "epoch": 29.44698795180723, "grad_norm": 12.390045166015625, "learning_rate": 8.525728155339806e-05, "loss": 0.273, "step": 3037 }, { "epoch": 29.456626506024097, "grad_norm": 11.610526084899902, "learning_rate": 8.525242718446603e-05, "loss": 0.4203, "step": 3038 }, { "epoch": 29.466265060240964, "grad_norm": 2.9698431491851807, "learning_rate": 8.524757281553398e-05, "loss": 0.2689, "step": 3039 }, { "epoch": 29.47590361445783, "grad_norm": 15.335389137268066, "learning_rate": 8.524271844660194e-05, "loss": 0.3754, "step": 3040 }, { "epoch": 29.485542168674698, "grad_norm": 8.651880264282227, "learning_rate": 8.52378640776699e-05, "loss": 0.3649, "step": 3041 }, { "epoch": 29.495180722891565, "grad_norm": 10.671987533569336, "learning_rate": 8.523300970873787e-05, "loss": 0.5608, "step": 3042 }, { "epoch": 29.504819277108435, "grad_norm": 11.553400993347168, "learning_rate": 8.522815533980583e-05, "loss": 0.2585, "step": 3043 }, { "epoch": 29.514457831325302, "grad_norm": 20.341794967651367, "learning_rate": 8.522330097087379e-05, "loss": 0.2688, "step": 3044 }, { "epoch": 29.52409638554217, "grad_norm": 17.084880828857422, "learning_rate": 8.521844660194174e-05, "loss": 0.4624, "step": 3045 }, { "epoch": 29.533734939759036, "grad_norm": 6.467948913574219, "learning_rate": 8.521359223300972e-05, "loss": 0.493, "step": 3046 }, { "epoch": 29.543373493975903, "grad_norm": 7.170800685882568, "learning_rate": 8.520873786407767e-05, "loss": 0.4441, "step": 3047 }, { "epoch": 29.55301204819277, "grad_norm": 4.5928826332092285, "learning_rate": 8.520388349514563e-05, "loss": 0.2423, "step": 3048 }, { "epoch": 29.56265060240964, "grad_norm": 5.052304744720459, "learning_rate": 8.519902912621359e-05, "loss": 0.2844, "step": 3049 }, { "epoch": 29.572289156626507, "grad_norm": 9.112431526184082, "learning_rate": 8.519417475728155e-05, "loss": 0.2591, "step": 3050 }, { "epoch": 29.581927710843374, "grad_norm": 7.325663089752197, "learning_rate": 8.518932038834952e-05, "loss": 0.3867, "step": 3051 }, { "epoch": 29.59156626506024, "grad_norm": 13.697641372680664, "learning_rate": 8.518446601941748e-05, "loss": 0.3326, "step": 3052 }, { "epoch": 29.601204819277108, "grad_norm": 5.184481620788574, "learning_rate": 8.517961165048543e-05, "loss": 0.3418, "step": 3053 }, { "epoch": 29.610843373493974, "grad_norm": 6.497147560119629, "learning_rate": 8.517475728155339e-05, "loss": 0.2027, "step": 3054 }, { "epoch": 29.620481927710845, "grad_norm": 14.572168350219727, "learning_rate": 8.516990291262136e-05, "loss": 0.4221, "step": 3055 }, { "epoch": 29.63012048192771, "grad_norm": 11.454270362854004, "learning_rate": 8.516504854368933e-05, "loss": 0.3819, "step": 3056 }, { "epoch": 29.63975903614458, "grad_norm": 35.656089782714844, "learning_rate": 8.516019417475729e-05, "loss": 0.4239, "step": 3057 }, { "epoch": 29.649397590361446, "grad_norm": 15.826101303100586, "learning_rate": 8.515533980582525e-05, "loss": 0.4019, "step": 3058 }, { "epoch": 29.659036144578312, "grad_norm": 8.982465744018555, "learning_rate": 8.515048543689321e-05, "loss": 0.4683, "step": 3059 }, { "epoch": 29.66867469879518, "grad_norm": 7.667304039001465, "learning_rate": 8.514563106796118e-05, "loss": 0.3982, "step": 3060 }, { "epoch": 29.67831325301205, "grad_norm": 4.439653396606445, "learning_rate": 8.514077669902914e-05, "loss": 0.3474, "step": 3061 }, { "epoch": 29.687951807228917, "grad_norm": 35.12008285522461, "learning_rate": 8.51359223300971e-05, "loss": 0.2303, "step": 3062 }, { "epoch": 29.697590361445783, "grad_norm": 11.975335121154785, "learning_rate": 8.513106796116505e-05, "loss": 0.3251, "step": 3063 }, { "epoch": 29.70722891566265, "grad_norm": 11.455517768859863, "learning_rate": 8.512621359223302e-05, "loss": 0.3089, "step": 3064 }, { "epoch": 29.716867469879517, "grad_norm": 6.263235569000244, "learning_rate": 8.512135922330098e-05, "loss": 0.524, "step": 3065 }, { "epoch": 29.726506024096384, "grad_norm": 4.845265865325928, "learning_rate": 8.511650485436894e-05, "loss": 0.2614, "step": 3066 }, { "epoch": 29.736144578313255, "grad_norm": 28.736276626586914, "learning_rate": 8.51116504854369e-05, "loss": 0.2811, "step": 3067 }, { "epoch": 29.74578313253012, "grad_norm": 3.1134912967681885, "learning_rate": 8.510679611650487e-05, "loss": 0.2256, "step": 3068 }, { "epoch": 29.75542168674699, "grad_norm": 4.489585876464844, "learning_rate": 8.510194174757283e-05, "loss": 0.3127, "step": 3069 }, { "epoch": 29.765060240963855, "grad_norm": 1.9391590356826782, "learning_rate": 8.509708737864078e-05, "loss": 0.2659, "step": 3070 }, { "epoch": 29.774698795180722, "grad_norm": 8.189793586730957, "learning_rate": 8.509223300970874e-05, "loss": 0.2702, "step": 3071 }, { "epoch": 29.78433734939759, "grad_norm": 13.022055625915527, "learning_rate": 8.50873786407767e-05, "loss": 0.4377, "step": 3072 }, { "epoch": 29.79397590361446, "grad_norm": 6.643534183502197, "learning_rate": 8.508252427184467e-05, "loss": 0.3893, "step": 3073 }, { "epoch": 29.803614457831326, "grad_norm": 17.62685203552246, "learning_rate": 8.507766990291263e-05, "loss": 0.2026, "step": 3074 }, { "epoch": 29.813253012048193, "grad_norm": 4.672639846801758, "learning_rate": 8.507281553398059e-05, "loss": 0.4723, "step": 3075 }, { "epoch": 29.82289156626506, "grad_norm": 11.100628852844238, "learning_rate": 8.506796116504854e-05, "loss": 0.3009, "step": 3076 }, { "epoch": 29.832530120481927, "grad_norm": 3.58146595954895, "learning_rate": 8.506310679611651e-05, "loss": 0.1717, "step": 3077 }, { "epoch": 29.842168674698794, "grad_norm": 7.049989700317383, "learning_rate": 8.505825242718447e-05, "loss": 0.4922, "step": 3078 }, { "epoch": 29.851807228915664, "grad_norm": 7.387608051300049, "learning_rate": 8.505339805825243e-05, "loss": 0.336, "step": 3079 }, { "epoch": 29.86144578313253, "grad_norm": 2.340620517730713, "learning_rate": 8.504854368932039e-05, "loss": 0.1364, "step": 3080 }, { "epoch": 29.871084337349398, "grad_norm": 6.653561115264893, "learning_rate": 8.504368932038836e-05, "loss": 0.3593, "step": 3081 }, { "epoch": 29.880722891566265, "grad_norm": 7.153271675109863, "learning_rate": 8.503883495145632e-05, "loss": 0.2923, "step": 3082 }, { "epoch": 29.89036144578313, "grad_norm": 17.8656063079834, "learning_rate": 8.503398058252427e-05, "loss": 0.4709, "step": 3083 }, { "epoch": 29.9, "grad_norm": 10.390910148620605, "learning_rate": 8.502912621359223e-05, "loss": 0.5064, "step": 3084 }, { "epoch": 29.90963855421687, "grad_norm": 5.768812656402588, "learning_rate": 8.50242718446602e-05, "loss": 0.4392, "step": 3085 }, { "epoch": 29.919277108433736, "grad_norm": 10.674601554870605, "learning_rate": 8.501941747572816e-05, "loss": 0.5186, "step": 3086 }, { "epoch": 29.928915662650603, "grad_norm": 27.4941463470459, "learning_rate": 8.501456310679612e-05, "loss": 0.3646, "step": 3087 }, { "epoch": 29.93855421686747, "grad_norm": 14.10010051727295, "learning_rate": 8.500970873786408e-05, "loss": 0.6199, "step": 3088 }, { "epoch": 29.948192771084337, "grad_norm": 11.148367881774902, "learning_rate": 8.500485436893203e-05, "loss": 0.4664, "step": 3089 }, { "epoch": 29.957831325301203, "grad_norm": 14.22563648223877, "learning_rate": 8.5e-05, "loss": 0.4225, "step": 3090 }, { "epoch": 29.967469879518074, "grad_norm": 9.640792846679688, "learning_rate": 8.499514563106796e-05, "loss": 0.262, "step": 3091 }, { "epoch": 29.97710843373494, "grad_norm": 17.84135627746582, "learning_rate": 8.499029126213592e-05, "loss": 0.2716, "step": 3092 }, { "epoch": 29.986746987951808, "grad_norm": 23.23920440673828, "learning_rate": 8.498543689320388e-05, "loss": 0.3759, "step": 3093 }, { "epoch": 30.002409638554216, "grad_norm": 21.50931167602539, "learning_rate": 8.498058252427185e-05, "loss": 0.2964, "step": 3094 }, { "epoch": 30.012048192771083, "grad_norm": 7.500164031982422, "learning_rate": 8.497572815533981e-05, "loss": 0.3826, "step": 3095 }, { "epoch": 30.021686746987953, "grad_norm": 19.680469512939453, "learning_rate": 8.497087378640777e-05, "loss": 0.3171, "step": 3096 }, { "epoch": 30.03132530120482, "grad_norm": 5.270541191101074, "learning_rate": 8.496601941747572e-05, "loss": 0.3431, "step": 3097 }, { "epoch": 30.040963855421687, "grad_norm": 8.84352970123291, "learning_rate": 8.49611650485437e-05, "loss": 0.3669, "step": 3098 }, { "epoch": 30.050602409638554, "grad_norm": 4.614744663238525, "learning_rate": 8.495631067961165e-05, "loss": 0.3754, "step": 3099 }, { "epoch": 30.06024096385542, "grad_norm": 6.8042144775390625, "learning_rate": 8.495145631067961e-05, "loss": 0.3869, "step": 3100 }, { "epoch": 30.069879518072288, "grad_norm": 10.287026405334473, "learning_rate": 8.494660194174757e-05, "loss": 0.4672, "step": 3101 }, { "epoch": 30.079518072289158, "grad_norm": 11.524218559265137, "learning_rate": 8.494174757281554e-05, "loss": 0.4029, "step": 3102 }, { "epoch": 30.089156626506025, "grad_norm": 9.07363224029541, "learning_rate": 8.49368932038835e-05, "loss": 0.1463, "step": 3103 }, { "epoch": 30.09879518072289, "grad_norm": 14.096338272094727, "learning_rate": 8.493203883495146e-05, "loss": 0.2699, "step": 3104 }, { "epoch": 30.10843373493976, "grad_norm": 7.807704925537109, "learning_rate": 8.492718446601941e-05, "loss": 0.3529, "step": 3105 }, { "epoch": 30.118072289156625, "grad_norm": 9.455755233764648, "learning_rate": 8.492233009708738e-05, "loss": 0.4581, "step": 3106 }, { "epoch": 30.127710843373492, "grad_norm": 4.330613136291504, "learning_rate": 8.491747572815536e-05, "loss": 0.4233, "step": 3107 }, { "epoch": 30.137349397590363, "grad_norm": 6.24312686920166, "learning_rate": 8.491262135922331e-05, "loss": 0.1829, "step": 3108 }, { "epoch": 30.14698795180723, "grad_norm": 10.799115180969238, "learning_rate": 8.490776699029127e-05, "loss": 0.4005, "step": 3109 }, { "epoch": 30.156626506024097, "grad_norm": 6.553548812866211, "learning_rate": 8.490291262135923e-05, "loss": 0.3833, "step": 3110 }, { "epoch": 30.166265060240963, "grad_norm": 10.69524097442627, "learning_rate": 8.489805825242719e-05, "loss": 0.3882, "step": 3111 }, { "epoch": 30.17590361445783, "grad_norm": 7.8175177574157715, "learning_rate": 8.489320388349516e-05, "loss": 0.3663, "step": 3112 }, { "epoch": 30.185542168674697, "grad_norm": 14.863716125488281, "learning_rate": 8.488834951456312e-05, "loss": 0.2946, "step": 3113 }, { "epoch": 30.195180722891568, "grad_norm": 6.084674835205078, "learning_rate": 8.488349514563107e-05, "loss": 0.2761, "step": 3114 }, { "epoch": 30.204819277108435, "grad_norm": 4.576884746551514, "learning_rate": 8.487864077669903e-05, "loss": 0.3092, "step": 3115 }, { "epoch": 30.2144578313253, "grad_norm": 6.067967891693115, "learning_rate": 8.4873786407767e-05, "loss": 0.333, "step": 3116 }, { "epoch": 30.22409638554217, "grad_norm": 8.16746711730957, "learning_rate": 8.486893203883496e-05, "loss": 0.2105, "step": 3117 }, { "epoch": 30.233734939759035, "grad_norm": 6.151790142059326, "learning_rate": 8.486407766990292e-05, "loss": 0.4123, "step": 3118 }, { "epoch": 30.243373493975902, "grad_norm": 12.384239196777344, "learning_rate": 8.485922330097088e-05, "loss": 0.4551, "step": 3119 }, { "epoch": 30.253012048192772, "grad_norm": 11.887920379638672, "learning_rate": 8.485436893203885e-05, "loss": 0.3049, "step": 3120 }, { "epoch": 30.26265060240964, "grad_norm": 4.68925666809082, "learning_rate": 8.48495145631068e-05, "loss": 0.252, "step": 3121 }, { "epoch": 30.272289156626506, "grad_norm": 9.144904136657715, "learning_rate": 8.484466019417476e-05, "loss": 0.391, "step": 3122 }, { "epoch": 30.281927710843373, "grad_norm": 15.108392715454102, "learning_rate": 8.483980582524272e-05, "loss": 0.3226, "step": 3123 }, { "epoch": 30.29156626506024, "grad_norm": 5.019140243530273, "learning_rate": 8.483495145631069e-05, "loss": 0.2753, "step": 3124 }, { "epoch": 30.301204819277107, "grad_norm": 20.573806762695312, "learning_rate": 8.483009708737865e-05, "loss": 0.354, "step": 3125 }, { "epoch": 30.310843373493977, "grad_norm": 5.071398735046387, "learning_rate": 8.482524271844661e-05, "loss": 0.202, "step": 3126 }, { "epoch": 30.320481927710844, "grad_norm": 18.709705352783203, "learning_rate": 8.482038834951457e-05, "loss": 0.4823, "step": 3127 }, { "epoch": 30.33012048192771, "grad_norm": 24.05435562133789, "learning_rate": 8.481553398058252e-05, "loss": 0.2929, "step": 3128 }, { "epoch": 30.339759036144578, "grad_norm": 22.132516860961914, "learning_rate": 8.48106796116505e-05, "loss": 0.4166, "step": 3129 }, { "epoch": 30.349397590361445, "grad_norm": 5.765704154968262, "learning_rate": 8.480582524271845e-05, "loss": 0.4036, "step": 3130 }, { "epoch": 30.35903614457831, "grad_norm": 16.236255645751953, "learning_rate": 8.480097087378641e-05, "loss": 0.2576, "step": 3131 }, { "epoch": 30.368674698795182, "grad_norm": 6.971453666687012, "learning_rate": 8.479611650485437e-05, "loss": 0.4333, "step": 3132 }, { "epoch": 30.37831325301205, "grad_norm": 4.204174518585205, "learning_rate": 8.479126213592234e-05, "loss": 0.3013, "step": 3133 }, { "epoch": 30.387951807228916, "grad_norm": 7.243431568145752, "learning_rate": 8.47864077669903e-05, "loss": 0.2637, "step": 3134 }, { "epoch": 30.397590361445783, "grad_norm": 3.421721935272217, "learning_rate": 8.478155339805825e-05, "loss": 0.215, "step": 3135 }, { "epoch": 30.40722891566265, "grad_norm": 7.447136402130127, "learning_rate": 8.477669902912621e-05, "loss": 0.1857, "step": 3136 }, { "epoch": 30.416867469879517, "grad_norm": 10.388875961303711, "learning_rate": 8.477184466019418e-05, "loss": 0.3346, "step": 3137 }, { "epoch": 30.426506024096387, "grad_norm": 13.941240310668945, "learning_rate": 8.476699029126214e-05, "loss": 0.2796, "step": 3138 }, { "epoch": 30.436144578313254, "grad_norm": 6.210986137390137, "learning_rate": 8.47621359223301e-05, "loss": 0.3425, "step": 3139 }, { "epoch": 30.44578313253012, "grad_norm": 6.484493255615234, "learning_rate": 8.475728155339806e-05, "loss": 0.4229, "step": 3140 }, { "epoch": 30.455421686746988, "grad_norm": 6.2796311378479, "learning_rate": 8.475242718446601e-05, "loss": 0.2182, "step": 3141 }, { "epoch": 30.465060240963854, "grad_norm": 10.52352237701416, "learning_rate": 8.474757281553399e-05, "loss": 0.3124, "step": 3142 }, { "epoch": 30.47469879518072, "grad_norm": 10.007392883300781, "learning_rate": 8.474271844660194e-05, "loss": 0.4261, "step": 3143 }, { "epoch": 30.48433734939759, "grad_norm": 6.57614803314209, "learning_rate": 8.47378640776699e-05, "loss": 0.2727, "step": 3144 }, { "epoch": 30.49397590361446, "grad_norm": 5.66255521774292, "learning_rate": 8.473300970873786e-05, "loss": 0.3882, "step": 3145 }, { "epoch": 30.503614457831326, "grad_norm": 7.634706497192383, "learning_rate": 8.472815533980583e-05, "loss": 0.4014, "step": 3146 }, { "epoch": 30.513253012048192, "grad_norm": 20.75388526916504, "learning_rate": 8.472330097087379e-05, "loss": 0.5041, "step": 3147 }, { "epoch": 30.52289156626506, "grad_norm": 8.83803653717041, "learning_rate": 8.471844660194175e-05, "loss": 0.2543, "step": 3148 }, { "epoch": 30.532530120481926, "grad_norm": 10.217299461364746, "learning_rate": 8.47135922330097e-05, "loss": 0.3909, "step": 3149 }, { "epoch": 30.542168674698797, "grad_norm": 8.734871864318848, "learning_rate": 8.470873786407767e-05, "loss": 0.2174, "step": 3150 }, { "epoch": 30.551807228915663, "grad_norm": 13.518681526184082, "learning_rate": 8.470388349514563e-05, "loss": 0.2646, "step": 3151 }, { "epoch": 30.56144578313253, "grad_norm": 23.441972732543945, "learning_rate": 8.469902912621359e-05, "loss": 0.343, "step": 3152 }, { "epoch": 30.571084337349397, "grad_norm": 5.370384693145752, "learning_rate": 8.469417475728155e-05, "loss": 0.3591, "step": 3153 }, { "epoch": 30.580722891566264, "grad_norm": 11.24913501739502, "learning_rate": 8.468932038834952e-05, "loss": 0.2796, "step": 3154 }, { "epoch": 30.59036144578313, "grad_norm": 4.860820293426514, "learning_rate": 8.468446601941748e-05, "loss": 0.2412, "step": 3155 }, { "epoch": 30.6, "grad_norm": 6.086719512939453, "learning_rate": 8.467961165048545e-05, "loss": 0.1559, "step": 3156 }, { "epoch": 30.60963855421687, "grad_norm": 4.613293647766113, "learning_rate": 8.46747572815534e-05, "loss": 0.3105, "step": 3157 }, { "epoch": 30.619277108433735, "grad_norm": 6.289015293121338, "learning_rate": 8.466990291262136e-05, "loss": 0.4034, "step": 3158 }, { "epoch": 30.628915662650602, "grad_norm": 5.361303329467773, "learning_rate": 8.466504854368934e-05, "loss": 0.4657, "step": 3159 }, { "epoch": 30.63855421686747, "grad_norm": 7.905696868896484, "learning_rate": 8.466019417475729e-05, "loss": 0.6334, "step": 3160 }, { "epoch": 30.648192771084336, "grad_norm": 14.323894500732422, "learning_rate": 8.465533980582525e-05, "loss": 0.4982, "step": 3161 }, { "epoch": 30.657831325301206, "grad_norm": 10.828669548034668, "learning_rate": 8.465048543689321e-05, "loss": 0.4213, "step": 3162 }, { "epoch": 30.667469879518073, "grad_norm": 8.670848846435547, "learning_rate": 8.464563106796117e-05, "loss": 0.4487, "step": 3163 }, { "epoch": 30.67710843373494, "grad_norm": 8.917008399963379, "learning_rate": 8.464077669902914e-05, "loss": 0.4363, "step": 3164 }, { "epoch": 30.686746987951807, "grad_norm": 4.688762664794922, "learning_rate": 8.46359223300971e-05, "loss": 0.3048, "step": 3165 }, { "epoch": 30.696385542168674, "grad_norm": 16.541271209716797, "learning_rate": 8.463106796116505e-05, "loss": 0.6555, "step": 3166 }, { "epoch": 30.70602409638554, "grad_norm": 6.477208137512207, "learning_rate": 8.462621359223301e-05, "loss": 0.137, "step": 3167 }, { "epoch": 30.71566265060241, "grad_norm": 11.821768760681152, "learning_rate": 8.462135922330098e-05, "loss": 0.5018, "step": 3168 }, { "epoch": 30.725301204819278, "grad_norm": 7.930075168609619, "learning_rate": 8.461650485436894e-05, "loss": 0.3982, "step": 3169 }, { "epoch": 30.734939759036145, "grad_norm": 8.984326362609863, "learning_rate": 8.46116504854369e-05, "loss": 0.2854, "step": 3170 }, { "epoch": 30.74457831325301, "grad_norm": 5.2116241455078125, "learning_rate": 8.460679611650486e-05, "loss": 0.2418, "step": 3171 }, { "epoch": 30.75421686746988, "grad_norm": 11.705560684204102, "learning_rate": 8.460194174757283e-05, "loss": 0.4239, "step": 3172 }, { "epoch": 30.763855421686745, "grad_norm": 9.212186813354492, "learning_rate": 8.459708737864078e-05, "loss": 0.3593, "step": 3173 }, { "epoch": 30.773493975903616, "grad_norm": 14.446362495422363, "learning_rate": 8.459223300970874e-05, "loss": 0.2961, "step": 3174 }, { "epoch": 30.783132530120483, "grad_norm": 6.870103359222412, "learning_rate": 8.45873786407767e-05, "loss": 0.3341, "step": 3175 }, { "epoch": 30.79277108433735, "grad_norm": 3.910998582839966, "learning_rate": 8.458252427184467e-05, "loss": 0.1215, "step": 3176 }, { "epoch": 30.802409638554217, "grad_norm": 4.796658039093018, "learning_rate": 8.457766990291263e-05, "loss": 0.19, "step": 3177 }, { "epoch": 30.812048192771083, "grad_norm": 22.75408363342285, "learning_rate": 8.457281553398059e-05, "loss": 0.24, "step": 3178 }, { "epoch": 30.82168674698795, "grad_norm": 18.516342163085938, "learning_rate": 8.456796116504854e-05, "loss": 0.3727, "step": 3179 }, { "epoch": 30.83132530120482, "grad_norm": 4.659321308135986, "learning_rate": 8.45631067961165e-05, "loss": 0.2366, "step": 3180 }, { "epoch": 30.840963855421688, "grad_norm": 5.3813090324401855, "learning_rate": 8.455825242718447e-05, "loss": 0.3781, "step": 3181 }, { "epoch": 30.850602409638554, "grad_norm": 16.757755279541016, "learning_rate": 8.455339805825243e-05, "loss": 0.6238, "step": 3182 }, { "epoch": 30.86024096385542, "grad_norm": 8.950188636779785, "learning_rate": 8.454854368932039e-05, "loss": 0.3504, "step": 3183 }, { "epoch": 30.86987951807229, "grad_norm": 5.642615795135498, "learning_rate": 8.454368932038835e-05, "loss": 0.2704, "step": 3184 }, { "epoch": 30.879518072289155, "grad_norm": 18.29671287536621, "learning_rate": 8.453883495145632e-05, "loss": 0.3774, "step": 3185 }, { "epoch": 30.889156626506026, "grad_norm": 37.7179069519043, "learning_rate": 8.453398058252428e-05, "loss": 0.4663, "step": 3186 }, { "epoch": 30.898795180722892, "grad_norm": 9.889169692993164, "learning_rate": 8.452912621359223e-05, "loss": 0.2509, "step": 3187 }, { "epoch": 30.90843373493976, "grad_norm": 7.004913806915283, "learning_rate": 8.452427184466019e-05, "loss": 0.4207, "step": 3188 }, { "epoch": 30.918072289156626, "grad_norm": 7.3625807762146, "learning_rate": 8.451941747572816e-05, "loss": 0.332, "step": 3189 }, { "epoch": 30.927710843373493, "grad_norm": 21.969953536987305, "learning_rate": 8.451456310679612e-05, "loss": 0.4672, "step": 3190 }, { "epoch": 30.93734939759036, "grad_norm": 8.469635009765625, "learning_rate": 8.450970873786408e-05, "loss": 0.4434, "step": 3191 }, { "epoch": 30.94698795180723, "grad_norm": 4.653131008148193, "learning_rate": 8.450485436893204e-05, "loss": 0.2358, "step": 3192 }, { "epoch": 30.956626506024097, "grad_norm": 12.849065780639648, "learning_rate": 8.450000000000001e-05, "loss": 0.412, "step": 3193 }, { "epoch": 30.966265060240964, "grad_norm": 4.536248207092285, "learning_rate": 8.449514563106797e-05, "loss": 0.2766, "step": 3194 }, { "epoch": 30.97590361445783, "grad_norm": 7.010354042053223, "learning_rate": 8.449029126213592e-05, "loss": 0.3974, "step": 3195 }, { "epoch": 30.985542168674698, "grad_norm": 6.215043067932129, "learning_rate": 8.448543689320388e-05, "loss": 0.1798, "step": 3196 }, { "epoch": 31.00120481927711, "grad_norm": 10.779264450073242, "learning_rate": 8.448058252427184e-05, "loss": 0.3505, "step": 3197 }, { "epoch": 31.010843373493977, "grad_norm": 23.369613647460938, "learning_rate": 8.447572815533981e-05, "loss": 0.4554, "step": 3198 }, { "epoch": 31.020481927710843, "grad_norm": 9.945026397705078, "learning_rate": 8.447087378640777e-05, "loss": 0.2816, "step": 3199 }, { "epoch": 31.03012048192771, "grad_norm": 17.52831268310547, "learning_rate": 8.446601941747573e-05, "loss": 0.3841, "step": 3200 }, { "epoch": 31.039759036144577, "grad_norm": 7.987008094787598, "learning_rate": 8.446116504854368e-05, "loss": 0.4252, "step": 3201 }, { "epoch": 31.049397590361444, "grad_norm": 5.797526836395264, "learning_rate": 8.445631067961165e-05, "loss": 0.2483, "step": 3202 }, { "epoch": 31.059036144578315, "grad_norm": 36.45345687866211, "learning_rate": 8.445145631067961e-05, "loss": 0.3812, "step": 3203 }, { "epoch": 31.06867469879518, "grad_norm": 3.8947784900665283, "learning_rate": 8.444660194174757e-05, "loss": 0.2543, "step": 3204 }, { "epoch": 31.07831325301205, "grad_norm": 5.931452751159668, "learning_rate": 8.444174757281553e-05, "loss": 0.163, "step": 3205 }, { "epoch": 31.087951807228915, "grad_norm": 13.735323905944824, "learning_rate": 8.44368932038835e-05, "loss": 0.3808, "step": 3206 }, { "epoch": 31.097590361445782, "grad_norm": 9.618124961853027, "learning_rate": 8.443203883495147e-05, "loss": 0.5078, "step": 3207 }, { "epoch": 31.10722891566265, "grad_norm": 7.891617774963379, "learning_rate": 8.442718446601943e-05, "loss": 0.378, "step": 3208 }, { "epoch": 31.11686746987952, "grad_norm": 15.27062702178955, "learning_rate": 8.442233009708739e-05, "loss": 0.246, "step": 3209 }, { "epoch": 31.126506024096386, "grad_norm": 5.679168701171875, "learning_rate": 8.441747572815534e-05, "loss": 0.2888, "step": 3210 }, { "epoch": 31.136144578313253, "grad_norm": 37.40653610229492, "learning_rate": 8.441262135922331e-05, "loss": 0.3978, "step": 3211 }, { "epoch": 31.14578313253012, "grad_norm": 9.816060066223145, "learning_rate": 8.440776699029127e-05, "loss": 0.3665, "step": 3212 }, { "epoch": 31.155421686746987, "grad_norm": 4.341335296630859, "learning_rate": 8.440291262135923e-05, "loss": 0.2804, "step": 3213 }, { "epoch": 31.165060240963854, "grad_norm": 8.69811725616455, "learning_rate": 8.439805825242719e-05, "loss": 0.3165, "step": 3214 }, { "epoch": 31.174698795180724, "grad_norm": 8.00328540802002, "learning_rate": 8.439320388349516e-05, "loss": 0.3354, "step": 3215 }, { "epoch": 31.18433734939759, "grad_norm": 3.9932398796081543, "learning_rate": 8.438834951456312e-05, "loss": 0.1749, "step": 3216 }, { "epoch": 31.193975903614458, "grad_norm": 12.167257308959961, "learning_rate": 8.438349514563108e-05, "loss": 0.3998, "step": 3217 }, { "epoch": 31.203614457831325, "grad_norm": 12.551100730895996, "learning_rate": 8.437864077669903e-05, "loss": 0.2929, "step": 3218 }, { "epoch": 31.21325301204819, "grad_norm": 8.774446487426758, "learning_rate": 8.437378640776699e-05, "loss": 0.2174, "step": 3219 }, { "epoch": 31.22289156626506, "grad_norm": 5.697792053222656, "learning_rate": 8.436893203883496e-05, "loss": 0.3152, "step": 3220 }, { "epoch": 31.23253012048193, "grad_norm": 27.653207778930664, "learning_rate": 8.436407766990292e-05, "loss": 0.485, "step": 3221 }, { "epoch": 31.242168674698796, "grad_norm": 11.533764839172363, "learning_rate": 8.435922330097088e-05, "loss": 0.3238, "step": 3222 }, { "epoch": 31.251807228915663, "grad_norm": 12.126066207885742, "learning_rate": 8.435436893203884e-05, "loss": 0.2592, "step": 3223 }, { "epoch": 31.26144578313253, "grad_norm": 21.18977928161621, "learning_rate": 8.43495145631068e-05, "loss": 0.3654, "step": 3224 }, { "epoch": 31.271084337349397, "grad_norm": 7.643187522888184, "learning_rate": 8.434466019417476e-05, "loss": 0.3398, "step": 3225 }, { "epoch": 31.280722891566263, "grad_norm": 12.963462829589844, "learning_rate": 8.433980582524272e-05, "loss": 0.272, "step": 3226 }, { "epoch": 31.290361445783134, "grad_norm": 5.640630722045898, "learning_rate": 8.433495145631068e-05, "loss": 0.3363, "step": 3227 }, { "epoch": 31.3, "grad_norm": 3.27242374420166, "learning_rate": 8.433009708737865e-05, "loss": 0.2226, "step": 3228 }, { "epoch": 31.309638554216868, "grad_norm": 14.159363746643066, "learning_rate": 8.432524271844661e-05, "loss": 0.2147, "step": 3229 }, { "epoch": 31.319277108433734, "grad_norm": 32.55813980102539, "learning_rate": 8.432038834951457e-05, "loss": 0.2836, "step": 3230 }, { "epoch": 31.3289156626506, "grad_norm": 10.975354194641113, "learning_rate": 8.431553398058252e-05, "loss": 0.7918, "step": 3231 }, { "epoch": 31.33855421686747, "grad_norm": 13.256075859069824, "learning_rate": 8.43106796116505e-05, "loss": 0.2898, "step": 3232 }, { "epoch": 31.34819277108434, "grad_norm": 12.180657386779785, "learning_rate": 8.430582524271845e-05, "loss": 0.386, "step": 3233 }, { "epoch": 31.357831325301206, "grad_norm": 4.177336692810059, "learning_rate": 8.430097087378641e-05, "loss": 0.3038, "step": 3234 }, { "epoch": 31.367469879518072, "grad_norm": 36.04138946533203, "learning_rate": 8.429611650485437e-05, "loss": 0.7937, "step": 3235 }, { "epoch": 31.37710843373494, "grad_norm": 10.834895133972168, "learning_rate": 8.429126213592233e-05, "loss": 0.2413, "step": 3236 }, { "epoch": 31.386746987951806, "grad_norm": 14.02003002166748, "learning_rate": 8.42864077669903e-05, "loss": 0.4764, "step": 3237 }, { "epoch": 31.396385542168673, "grad_norm": 15.565476417541504, "learning_rate": 8.428155339805826e-05, "loss": 0.4941, "step": 3238 }, { "epoch": 31.406024096385543, "grad_norm": 16.33933448791504, "learning_rate": 8.427669902912621e-05, "loss": 0.2787, "step": 3239 }, { "epoch": 31.41566265060241, "grad_norm": 4.4709906578063965, "learning_rate": 8.427184466019417e-05, "loss": 0.2068, "step": 3240 }, { "epoch": 31.425301204819277, "grad_norm": 8.416055679321289, "learning_rate": 8.426699029126214e-05, "loss": 0.2705, "step": 3241 }, { "epoch": 31.434939759036144, "grad_norm": 6.966923713684082, "learning_rate": 8.42621359223301e-05, "loss": 0.1867, "step": 3242 }, { "epoch": 31.44457831325301, "grad_norm": 6.188630104064941, "learning_rate": 8.425728155339806e-05, "loss": 0.3019, "step": 3243 }, { "epoch": 31.454216867469878, "grad_norm": 17.356618881225586, "learning_rate": 8.425242718446602e-05, "loss": 0.4416, "step": 3244 }, { "epoch": 31.46385542168675, "grad_norm": 5.857809543609619, "learning_rate": 8.424757281553399e-05, "loss": 0.2, "step": 3245 }, { "epoch": 31.473493975903615, "grad_norm": 16.80712127685547, "learning_rate": 8.424271844660194e-05, "loss": 0.5234, "step": 3246 }, { "epoch": 31.483132530120482, "grad_norm": 11.445703506469727, "learning_rate": 8.42378640776699e-05, "loss": 0.3782, "step": 3247 }, { "epoch": 31.49277108433735, "grad_norm": 11.23994255065918, "learning_rate": 8.423300970873786e-05, "loss": 0.5005, "step": 3248 }, { "epoch": 31.502409638554216, "grad_norm": 8.880938529968262, "learning_rate": 8.422815533980583e-05, "loss": 0.3008, "step": 3249 }, { "epoch": 31.512048192771083, "grad_norm": 4.22500467300415, "learning_rate": 8.422330097087379e-05, "loss": 0.3157, "step": 3250 }, { "epoch": 31.521686746987953, "grad_norm": 11.315045356750488, "learning_rate": 8.421844660194175e-05, "loss": 0.4888, "step": 3251 }, { "epoch": 31.53132530120482, "grad_norm": 6.430030822753906, "learning_rate": 8.42135922330097e-05, "loss": 0.4513, "step": 3252 }, { "epoch": 31.540963855421687, "grad_norm": 12.470541000366211, "learning_rate": 8.420873786407766e-05, "loss": 0.189, "step": 3253 }, { "epoch": 31.550602409638554, "grad_norm": 44.923240661621094, "learning_rate": 8.420388349514563e-05, "loss": 0.4432, "step": 3254 }, { "epoch": 31.56024096385542, "grad_norm": 10.474471092224121, "learning_rate": 8.419902912621359e-05, "loss": 0.2579, "step": 3255 }, { "epoch": 31.569879518072288, "grad_norm": 9.99461555480957, "learning_rate": 8.419417475728155e-05, "loss": 0.3068, "step": 3256 }, { "epoch": 31.579518072289158, "grad_norm": 15.396228790283203, "learning_rate": 8.418932038834952e-05, "loss": 0.3281, "step": 3257 }, { "epoch": 31.589156626506025, "grad_norm": 9.822854995727539, "learning_rate": 8.418446601941748e-05, "loss": 0.2933, "step": 3258 }, { "epoch": 31.59879518072289, "grad_norm": 5.361859321594238, "learning_rate": 8.417961165048545e-05, "loss": 0.3183, "step": 3259 }, { "epoch": 31.60843373493976, "grad_norm": 7.46078634262085, "learning_rate": 8.417475728155341e-05, "loss": 0.3161, "step": 3260 }, { "epoch": 31.618072289156625, "grad_norm": 4.941464424133301, "learning_rate": 8.416990291262137e-05, "loss": 0.1086, "step": 3261 }, { "epoch": 31.627710843373492, "grad_norm": 12.675556182861328, "learning_rate": 8.416504854368932e-05, "loss": 0.554, "step": 3262 }, { "epoch": 31.637349397590363, "grad_norm": 4.040639877319336, "learning_rate": 8.41601941747573e-05, "loss": 0.2876, "step": 3263 }, { "epoch": 31.64698795180723, "grad_norm": 8.406182289123535, "learning_rate": 8.415533980582525e-05, "loss": 0.2773, "step": 3264 }, { "epoch": 31.656626506024097, "grad_norm": 9.305062294006348, "learning_rate": 8.415048543689321e-05, "loss": 0.4775, "step": 3265 }, { "epoch": 31.666265060240963, "grad_norm": 5.431548595428467, "learning_rate": 8.414563106796117e-05, "loss": 0.4787, "step": 3266 }, { "epoch": 31.67590361445783, "grad_norm": 3.4261422157287598, "learning_rate": 8.414077669902914e-05, "loss": 0.2473, "step": 3267 }, { "epoch": 31.685542168674697, "grad_norm": 12.126899719238281, "learning_rate": 8.41359223300971e-05, "loss": 0.3823, "step": 3268 }, { "epoch": 31.695180722891568, "grad_norm": 6.96376895904541, "learning_rate": 8.413106796116505e-05, "loss": 0.4104, "step": 3269 }, { "epoch": 31.704819277108435, "grad_norm": 3.6781952381134033, "learning_rate": 8.412621359223301e-05, "loss": 0.1753, "step": 3270 }, { "epoch": 31.7144578313253, "grad_norm": 8.127955436706543, "learning_rate": 8.412135922330098e-05, "loss": 0.5103, "step": 3271 }, { "epoch": 31.72409638554217, "grad_norm": 5.707311153411865, "learning_rate": 8.411650485436894e-05, "loss": 0.4179, "step": 3272 }, { "epoch": 31.733734939759035, "grad_norm": 4.575580596923828, "learning_rate": 8.41116504854369e-05, "loss": 0.3149, "step": 3273 }, { "epoch": 31.743373493975902, "grad_norm": 5.7741923332214355, "learning_rate": 8.410679611650486e-05, "loss": 0.5222, "step": 3274 }, { "epoch": 31.753012048192772, "grad_norm": 9.91787052154541, "learning_rate": 8.410194174757281e-05, "loss": 0.3457, "step": 3275 }, { "epoch": 31.76265060240964, "grad_norm": 5.601621627807617, "learning_rate": 8.409708737864079e-05, "loss": 0.3206, "step": 3276 }, { "epoch": 31.772289156626506, "grad_norm": 4.969506740570068, "learning_rate": 8.409223300970874e-05, "loss": 0.2586, "step": 3277 }, { "epoch": 31.781927710843373, "grad_norm": 14.460419654846191, "learning_rate": 8.40873786407767e-05, "loss": 0.2489, "step": 3278 }, { "epoch": 31.79156626506024, "grad_norm": 7.3963518142700195, "learning_rate": 8.408252427184466e-05, "loss": 0.2452, "step": 3279 }, { "epoch": 31.801204819277107, "grad_norm": 5.160113334655762, "learning_rate": 8.407766990291263e-05, "loss": 0.3745, "step": 3280 }, { "epoch": 31.810843373493977, "grad_norm": 6.117737293243408, "learning_rate": 8.407281553398059e-05, "loss": 0.3383, "step": 3281 }, { "epoch": 31.820481927710844, "grad_norm": 65.23914337158203, "learning_rate": 8.406796116504855e-05, "loss": 0.2422, "step": 3282 }, { "epoch": 31.83012048192771, "grad_norm": 9.273151397705078, "learning_rate": 8.40631067961165e-05, "loss": 0.475, "step": 3283 }, { "epoch": 31.839759036144578, "grad_norm": 7.755721092224121, "learning_rate": 8.405825242718448e-05, "loss": 0.5577, "step": 3284 }, { "epoch": 31.849397590361445, "grad_norm": 4.447584629058838, "learning_rate": 8.405339805825243e-05, "loss": 0.1932, "step": 3285 }, { "epoch": 31.85903614457831, "grad_norm": 13.088644981384277, "learning_rate": 8.404854368932039e-05, "loss": 0.4085, "step": 3286 }, { "epoch": 31.868674698795182, "grad_norm": 4.254502773284912, "learning_rate": 8.404368932038835e-05, "loss": 0.1502, "step": 3287 }, { "epoch": 31.87831325301205, "grad_norm": 8.989267349243164, "learning_rate": 8.403883495145632e-05, "loss": 0.2469, "step": 3288 }, { "epoch": 31.887951807228916, "grad_norm": 4.4627885818481445, "learning_rate": 8.403398058252428e-05, "loss": 0.2565, "step": 3289 }, { "epoch": 31.897590361445783, "grad_norm": 6.514201641082764, "learning_rate": 8.402912621359224e-05, "loss": 0.4845, "step": 3290 }, { "epoch": 31.90722891566265, "grad_norm": 36.016265869140625, "learning_rate": 8.402427184466019e-05, "loss": 0.3041, "step": 3291 }, { "epoch": 31.916867469879517, "grad_norm": 19.2015380859375, "learning_rate": 8.401941747572815e-05, "loss": 0.516, "step": 3292 }, { "epoch": 31.926506024096387, "grad_norm": 6.616302013397217, "learning_rate": 8.401456310679612e-05, "loss": 0.1653, "step": 3293 }, { "epoch": 31.936144578313254, "grad_norm": 17.38219451904297, "learning_rate": 8.400970873786408e-05, "loss": 0.3621, "step": 3294 }, { "epoch": 31.94578313253012, "grad_norm": 12.820596694946289, "learning_rate": 8.400485436893204e-05, "loss": 0.2474, "step": 3295 }, { "epoch": 31.955421686746988, "grad_norm": 7.68613338470459, "learning_rate": 8.4e-05, "loss": 0.162, "step": 3296 }, { "epoch": 31.965060240963854, "grad_norm": 8.506052017211914, "learning_rate": 8.399514563106797e-05, "loss": 0.4173, "step": 3297 }, { "epoch": 31.97469879518072, "grad_norm": 5.9227070808410645, "learning_rate": 8.399029126213592e-05, "loss": 0.4294, "step": 3298 }, { "epoch": 31.98433734939759, "grad_norm": 3.36710786819458, "learning_rate": 8.398543689320388e-05, "loss": 0.3557, "step": 3299 }, { "epoch": 31.99397590361446, "grad_norm": 6.295999050140381, "learning_rate": 8.398058252427184e-05, "loss": 0.2105, "step": 3300 }, { "epoch": 32.00963855421687, "grad_norm": 6.771306037902832, "learning_rate": 8.397572815533981e-05, "loss": 0.3144, "step": 3301 }, { "epoch": 32.019277108433734, "grad_norm": 5.884930610656738, "learning_rate": 8.397087378640777e-05, "loss": 0.115, "step": 3302 }, { "epoch": 32.028915662650604, "grad_norm": 15.133679389953613, "learning_rate": 8.396601941747573e-05, "loss": 0.5763, "step": 3303 }, { "epoch": 32.03855421686747, "grad_norm": 5.046120643615723, "learning_rate": 8.396116504854368e-05, "loss": 0.3225, "step": 3304 }, { "epoch": 32.04819277108434, "grad_norm": 4.669327735900879, "learning_rate": 8.395631067961166e-05, "loss": 0.3551, "step": 3305 }, { "epoch": 32.0578313253012, "grad_norm": 5.844996452331543, "learning_rate": 8.395145631067961e-05, "loss": 0.2945, "step": 3306 }, { "epoch": 32.06746987951807, "grad_norm": 6.014493465423584, "learning_rate": 8.394660194174757e-05, "loss": 0.2955, "step": 3307 }, { "epoch": 32.07710843373494, "grad_norm": 5.416995048522949, "learning_rate": 8.394174757281554e-05, "loss": 0.227, "step": 3308 }, { "epoch": 32.086746987951805, "grad_norm": 6.3639068603515625, "learning_rate": 8.39368932038835e-05, "loss": 0.3416, "step": 3309 }, { "epoch": 32.096385542168676, "grad_norm": 8.266217231750488, "learning_rate": 8.393203883495147e-05, "loss": 0.3869, "step": 3310 }, { "epoch": 32.10602409638554, "grad_norm": 10.764619827270508, "learning_rate": 8.392718446601943e-05, "loss": 0.2291, "step": 3311 }, { "epoch": 32.11566265060241, "grad_norm": 18.782154083251953, "learning_rate": 8.392233009708739e-05, "loss": 0.3608, "step": 3312 }, { "epoch": 32.12530120481928, "grad_norm": 12.994498252868652, "learning_rate": 8.391747572815534e-05, "loss": 0.5255, "step": 3313 }, { "epoch": 32.13493975903614, "grad_norm": 7.333801746368408, "learning_rate": 8.39126213592233e-05, "loss": 0.2482, "step": 3314 }, { "epoch": 32.144578313253014, "grad_norm": 7.520168781280518, "learning_rate": 8.390776699029127e-05, "loss": 0.3067, "step": 3315 }, { "epoch": 32.15421686746988, "grad_norm": 7.055604934692383, "learning_rate": 8.390291262135923e-05, "loss": 0.2983, "step": 3316 }, { "epoch": 32.16385542168675, "grad_norm": 56.76035690307617, "learning_rate": 8.389805825242719e-05, "loss": 0.3352, "step": 3317 }, { "epoch": 32.17349397590361, "grad_norm": 11.129471778869629, "learning_rate": 8.389320388349515e-05, "loss": 0.3002, "step": 3318 }, { "epoch": 32.18313253012048, "grad_norm": 5.441463947296143, "learning_rate": 8.388834951456312e-05, "loss": 0.2846, "step": 3319 }, { "epoch": 32.19277108433735, "grad_norm": 37.32143020629883, "learning_rate": 8.388349514563108e-05, "loss": 0.501, "step": 3320 }, { "epoch": 32.202409638554215, "grad_norm": 7.138887882232666, "learning_rate": 8.387864077669903e-05, "loss": 0.3097, "step": 3321 }, { "epoch": 32.212048192771086, "grad_norm": 45.85586929321289, "learning_rate": 8.387378640776699e-05, "loss": 0.4304, "step": 3322 }, { "epoch": 32.22168674698795, "grad_norm": 24.445331573486328, "learning_rate": 8.386893203883496e-05, "loss": 0.2076, "step": 3323 }, { "epoch": 32.23132530120482, "grad_norm": 8.789253234863281, "learning_rate": 8.386407766990292e-05, "loss": 0.3364, "step": 3324 }, { "epoch": 32.24096385542169, "grad_norm": 5.368090629577637, "learning_rate": 8.385922330097088e-05, "loss": 0.3521, "step": 3325 }, { "epoch": 32.25060240963855, "grad_norm": 20.092763900756836, "learning_rate": 8.385436893203884e-05, "loss": 0.452, "step": 3326 }, { "epoch": 32.26024096385542, "grad_norm": 18.567913055419922, "learning_rate": 8.384951456310681e-05, "loss": 0.4194, "step": 3327 }, { "epoch": 32.26987951807229, "grad_norm": 10.393457412719727, "learning_rate": 8.384466019417477e-05, "loss": 0.5839, "step": 3328 }, { "epoch": 32.27951807228916, "grad_norm": 9.64769172668457, "learning_rate": 8.383980582524272e-05, "loss": 0.2712, "step": 3329 }, { "epoch": 32.28915662650602, "grad_norm": 5.335543155670166, "learning_rate": 8.383495145631068e-05, "loss": 0.1968, "step": 3330 }, { "epoch": 32.29879518072289, "grad_norm": 4.752079010009766, "learning_rate": 8.383009708737864e-05, "loss": 0.238, "step": 3331 }, { "epoch": 32.30843373493976, "grad_norm": 6.49189567565918, "learning_rate": 8.382524271844661e-05, "loss": 0.3262, "step": 3332 }, { "epoch": 32.318072289156625, "grad_norm": 12.05092716217041, "learning_rate": 8.382038834951457e-05, "loss": 0.2062, "step": 3333 }, { "epoch": 32.327710843373495, "grad_norm": 14.734648704528809, "learning_rate": 8.381553398058253e-05, "loss": 0.3929, "step": 3334 }, { "epoch": 32.33734939759036, "grad_norm": 18.902488708496094, "learning_rate": 8.381067961165048e-05, "loss": 0.2552, "step": 3335 }, { "epoch": 32.34698795180723, "grad_norm": 6.075798511505127, "learning_rate": 8.380582524271845e-05, "loss": 0.2173, "step": 3336 }, { "epoch": 32.3566265060241, "grad_norm": 9.644990921020508, "learning_rate": 8.380097087378641e-05, "loss": 0.3829, "step": 3337 }, { "epoch": 32.36626506024096, "grad_norm": 8.840384483337402, "learning_rate": 8.379611650485437e-05, "loss": 0.3434, "step": 3338 }, { "epoch": 32.37590361445783, "grad_norm": 10.719453811645508, "learning_rate": 8.379126213592233e-05, "loss": 0.4437, "step": 3339 }, { "epoch": 32.3855421686747, "grad_norm": 10.687088966369629, "learning_rate": 8.37864077669903e-05, "loss": 0.3448, "step": 3340 }, { "epoch": 32.39518072289157, "grad_norm": 4.765726089477539, "learning_rate": 8.378155339805826e-05, "loss": 0.1989, "step": 3341 }, { "epoch": 32.40481927710843, "grad_norm": 10.666707992553711, "learning_rate": 8.377669902912621e-05, "loss": 0.464, "step": 3342 }, { "epoch": 32.4144578313253, "grad_norm": 12.071531295776367, "learning_rate": 8.377184466019417e-05, "loss": 0.3411, "step": 3343 }, { "epoch": 32.42409638554217, "grad_norm": 3.3640060424804688, "learning_rate": 8.376699029126214e-05, "loss": 0.1507, "step": 3344 }, { "epoch": 32.433734939759034, "grad_norm": 7.688880920410156, "learning_rate": 8.37621359223301e-05, "loss": 0.1589, "step": 3345 }, { "epoch": 32.443373493975905, "grad_norm": 7.303642749786377, "learning_rate": 8.375728155339806e-05, "loss": 0.377, "step": 3346 }, { "epoch": 32.45301204819277, "grad_norm": 7.747814655303955, "learning_rate": 8.375242718446602e-05, "loss": 0.4659, "step": 3347 }, { "epoch": 32.46265060240964, "grad_norm": 17.706634521484375, "learning_rate": 8.374757281553397e-05, "loss": 0.7175, "step": 3348 }, { "epoch": 32.47228915662651, "grad_norm": 21.996280670166016, "learning_rate": 8.374271844660195e-05, "loss": 0.2712, "step": 3349 }, { "epoch": 32.48192771084337, "grad_norm": 12.518506050109863, "learning_rate": 8.37378640776699e-05, "loss": 0.3784, "step": 3350 }, { "epoch": 32.49156626506024, "grad_norm": 18.065099716186523, "learning_rate": 8.373300970873786e-05, "loss": 0.4022, "step": 3351 }, { "epoch": 32.501204819277106, "grad_norm": 9.654213905334473, "learning_rate": 8.372815533980582e-05, "loss": 0.3877, "step": 3352 }, { "epoch": 32.51084337349398, "grad_norm": 5.528099536895752, "learning_rate": 8.372330097087379e-05, "loss": 0.2608, "step": 3353 }, { "epoch": 32.52048192771084, "grad_norm": 14.565610885620117, "learning_rate": 8.371844660194175e-05, "loss": 0.3477, "step": 3354 }, { "epoch": 32.53012048192771, "grad_norm": 4.732031345367432, "learning_rate": 8.37135922330097e-05, "loss": 0.2661, "step": 3355 }, { "epoch": 32.53975903614458, "grad_norm": 9.591453552246094, "learning_rate": 8.370873786407766e-05, "loss": 0.4451, "step": 3356 }, { "epoch": 32.549397590361444, "grad_norm": 13.367165565490723, "learning_rate": 8.370388349514564e-05, "loss": 0.2844, "step": 3357 }, { "epoch": 32.559036144578315, "grad_norm": 7.286766052246094, "learning_rate": 8.36990291262136e-05, "loss": 0.346, "step": 3358 }, { "epoch": 32.56867469879518, "grad_norm": 7.974287986755371, "learning_rate": 8.369417475728156e-05, "loss": 0.3229, "step": 3359 }, { "epoch": 32.57831325301205, "grad_norm": 7.1707329750061035, "learning_rate": 8.368932038834952e-05, "loss": 0.3863, "step": 3360 }, { "epoch": 32.58795180722892, "grad_norm": 18.816274642944336, "learning_rate": 8.368446601941748e-05, "loss": 0.3831, "step": 3361 }, { "epoch": 32.59759036144578, "grad_norm": 17.559593200683594, "learning_rate": 8.367961165048545e-05, "loss": 0.244, "step": 3362 }, { "epoch": 32.60722891566265, "grad_norm": 4.399951457977295, "learning_rate": 8.367475728155341e-05, "loss": 0.2032, "step": 3363 }, { "epoch": 32.616867469879516, "grad_norm": 10.194437980651855, "learning_rate": 8.366990291262137e-05, "loss": 0.313, "step": 3364 }, { "epoch": 32.626506024096386, "grad_norm": 8.268061637878418, "learning_rate": 8.366504854368932e-05, "loss": 0.37, "step": 3365 }, { "epoch": 32.63614457831325, "grad_norm": 5.180586338043213, "learning_rate": 8.36601941747573e-05, "loss": 0.2049, "step": 3366 }, { "epoch": 32.64578313253012, "grad_norm": 10.994796752929688, "learning_rate": 8.365533980582525e-05, "loss": 0.7179, "step": 3367 }, { "epoch": 32.65542168674699, "grad_norm": 7.681466579437256, "learning_rate": 8.365048543689321e-05, "loss": 0.1398, "step": 3368 }, { "epoch": 32.665060240963854, "grad_norm": 6.383668899536133, "learning_rate": 8.364563106796117e-05, "loss": 0.3981, "step": 3369 }, { "epoch": 32.674698795180724, "grad_norm": 5.262146949768066, "learning_rate": 8.364077669902913e-05, "loss": 0.1576, "step": 3370 }, { "epoch": 32.68433734939759, "grad_norm": 8.565531730651855, "learning_rate": 8.36359223300971e-05, "loss": 0.5132, "step": 3371 }, { "epoch": 32.69397590361446, "grad_norm": 12.202187538146973, "learning_rate": 8.363106796116506e-05, "loss": 0.3814, "step": 3372 }, { "epoch": 32.70361445783133, "grad_norm": 9.590249061584473, "learning_rate": 8.362621359223301e-05, "loss": 0.4621, "step": 3373 }, { "epoch": 32.71325301204819, "grad_norm": 10.347330093383789, "learning_rate": 8.362135922330097e-05, "loss": 0.4741, "step": 3374 }, { "epoch": 32.72289156626506, "grad_norm": 4.485520839691162, "learning_rate": 8.361650485436894e-05, "loss": 0.2511, "step": 3375 }, { "epoch": 32.732530120481925, "grad_norm": 5.70994234085083, "learning_rate": 8.36116504854369e-05, "loss": 0.1392, "step": 3376 }, { "epoch": 32.742168674698796, "grad_norm": 4.036057472229004, "learning_rate": 8.360679611650486e-05, "loss": 0.2161, "step": 3377 }, { "epoch": 32.75180722891566, "grad_norm": 7.225859642028809, "learning_rate": 8.360194174757282e-05, "loss": 0.2529, "step": 3378 }, { "epoch": 32.76144578313253, "grad_norm": 9.137582778930664, "learning_rate": 8.359708737864079e-05, "loss": 0.4933, "step": 3379 }, { "epoch": 32.7710843373494, "grad_norm": 6.673788547515869, "learning_rate": 8.359223300970875e-05, "loss": 0.2909, "step": 3380 }, { "epoch": 32.78072289156626, "grad_norm": 4.890994548797607, "learning_rate": 8.35873786407767e-05, "loss": 0.1713, "step": 3381 }, { "epoch": 32.790361445783134, "grad_norm": 7.616218090057373, "learning_rate": 8.358252427184466e-05, "loss": 0.3039, "step": 3382 }, { "epoch": 32.8, "grad_norm": 8.401321411132812, "learning_rate": 8.357766990291263e-05, "loss": 0.1858, "step": 3383 }, { "epoch": 32.80963855421687, "grad_norm": 6.480374813079834, "learning_rate": 8.357281553398059e-05, "loss": 0.2553, "step": 3384 }, { "epoch": 32.81927710843374, "grad_norm": 6.416882514953613, "learning_rate": 8.356796116504855e-05, "loss": 0.2671, "step": 3385 }, { "epoch": 32.8289156626506, "grad_norm": 18.971887588500977, "learning_rate": 8.35631067961165e-05, "loss": 0.3897, "step": 3386 }, { "epoch": 32.83855421686747, "grad_norm": 6.110680103302002, "learning_rate": 8.355825242718446e-05, "loss": 0.175, "step": 3387 }, { "epoch": 32.848192771084335, "grad_norm": 2.764166831970215, "learning_rate": 8.355339805825243e-05, "loss": 0.2098, "step": 3388 }, { "epoch": 32.857831325301206, "grad_norm": 9.086139678955078, "learning_rate": 8.354854368932039e-05, "loss": 0.3732, "step": 3389 }, { "epoch": 32.86746987951807, "grad_norm": 5.355843544006348, "learning_rate": 8.354368932038835e-05, "loss": 0.3615, "step": 3390 }, { "epoch": 32.87710843373494, "grad_norm": 5.246227741241455, "learning_rate": 8.353883495145631e-05, "loss": 0.4011, "step": 3391 }, { "epoch": 32.88674698795181, "grad_norm": 8.971302032470703, "learning_rate": 8.353398058252428e-05, "loss": 0.3005, "step": 3392 }, { "epoch": 32.89638554216867, "grad_norm": 14.277738571166992, "learning_rate": 8.352912621359224e-05, "loss": 0.3157, "step": 3393 }, { "epoch": 32.90602409638554, "grad_norm": 7.831887722015381, "learning_rate": 8.35242718446602e-05, "loss": 0.4768, "step": 3394 }, { "epoch": 32.91566265060241, "grad_norm": 10.141061782836914, "learning_rate": 8.351941747572815e-05, "loss": 0.3649, "step": 3395 }, { "epoch": 32.92530120481928, "grad_norm": 6.074619770050049, "learning_rate": 8.351456310679612e-05, "loss": 0.3591, "step": 3396 }, { "epoch": 32.93493975903615, "grad_norm": 5.904169082641602, "learning_rate": 8.350970873786408e-05, "loss": 0.3586, "step": 3397 }, { "epoch": 32.94457831325301, "grad_norm": 7.205076217651367, "learning_rate": 8.350485436893204e-05, "loss": 0.4813, "step": 3398 }, { "epoch": 32.95421686746988, "grad_norm": 4.500741004943848, "learning_rate": 8.35e-05, "loss": 0.2415, "step": 3399 }, { "epoch": 32.963855421686745, "grad_norm": 3.6359474658966064, "learning_rate": 8.349514563106797e-05, "loss": 0.1673, "step": 3400 }, { "epoch": 32.973493975903615, "grad_norm": 8.097408294677734, "learning_rate": 8.349029126213593e-05, "loss": 0.2119, "step": 3401 }, { "epoch": 32.98313253012048, "grad_norm": 3.758024215698242, "learning_rate": 8.348543689320388e-05, "loss": 0.2766, "step": 3402 }, { "epoch": 32.99277108433735, "grad_norm": 5.415801525115967, "learning_rate": 8.348058252427184e-05, "loss": 0.4099, "step": 3403 }, { "epoch": 33.00843373493976, "grad_norm": 14.394957542419434, "learning_rate": 8.34757281553398e-05, "loss": 0.2387, "step": 3404 }, { "epoch": 33.01807228915663, "grad_norm": 7.430574893951416, "learning_rate": 8.347087378640777e-05, "loss": 0.4398, "step": 3405 }, { "epoch": 33.02771084337349, "grad_norm": 20.88284683227539, "learning_rate": 8.346601941747573e-05, "loss": 0.4016, "step": 3406 }, { "epoch": 33.03734939759036, "grad_norm": 3.913435697555542, "learning_rate": 8.346116504854369e-05, "loss": 0.309, "step": 3407 }, { "epoch": 33.04698795180723, "grad_norm": 4.9575114250183105, "learning_rate": 8.345631067961164e-05, "loss": 0.5305, "step": 3408 }, { "epoch": 33.056626506024095, "grad_norm": 44.305423736572266, "learning_rate": 8.345145631067961e-05, "loss": 0.3361, "step": 3409 }, { "epoch": 33.066265060240966, "grad_norm": 8.151001930236816, "learning_rate": 8.344660194174759e-05, "loss": 0.2286, "step": 3410 }, { "epoch": 33.07590361445783, "grad_norm": 4.192224025726318, "learning_rate": 8.344174757281554e-05, "loss": 0.4445, "step": 3411 }, { "epoch": 33.0855421686747, "grad_norm": 1.7146950960159302, "learning_rate": 8.34368932038835e-05, "loss": 0.2092, "step": 3412 }, { "epoch": 33.09518072289157, "grad_norm": 14.653775215148926, "learning_rate": 8.343203883495146e-05, "loss": 0.3992, "step": 3413 }, { "epoch": 33.10481927710843, "grad_norm": 4.747289180755615, "learning_rate": 8.342718446601943e-05, "loss": 0.3358, "step": 3414 }, { "epoch": 33.1144578313253, "grad_norm": 14.501933097839355, "learning_rate": 8.342233009708739e-05, "loss": 0.2638, "step": 3415 }, { "epoch": 33.12409638554217, "grad_norm": 6.1873393058776855, "learning_rate": 8.341747572815535e-05, "loss": 0.4136, "step": 3416 }, { "epoch": 33.13373493975904, "grad_norm": 5.4371538162231445, "learning_rate": 8.34126213592233e-05, "loss": 0.2514, "step": 3417 }, { "epoch": 33.1433734939759, "grad_norm": 15.387518882751465, "learning_rate": 8.340776699029128e-05, "loss": 0.3593, "step": 3418 }, { "epoch": 33.15301204819277, "grad_norm": 17.815156936645508, "learning_rate": 8.340291262135923e-05, "loss": 0.4381, "step": 3419 }, { "epoch": 33.16265060240964, "grad_norm": 22.59377098083496, "learning_rate": 8.339805825242719e-05, "loss": 0.3577, "step": 3420 }, { "epoch": 33.172289156626505, "grad_norm": 1.9740952253341675, "learning_rate": 8.339320388349515e-05, "loss": 0.2164, "step": 3421 }, { "epoch": 33.181927710843375, "grad_norm": 22.380889892578125, "learning_rate": 8.338834951456312e-05, "loss": 0.267, "step": 3422 }, { "epoch": 33.19156626506024, "grad_norm": 6.287942886352539, "learning_rate": 8.338349514563108e-05, "loss": 0.5393, "step": 3423 }, { "epoch": 33.20120481927711, "grad_norm": 57.69078063964844, "learning_rate": 8.337864077669904e-05, "loss": 0.5016, "step": 3424 }, { "epoch": 33.21084337349398, "grad_norm": 23.27435874938965, "learning_rate": 8.3373786407767e-05, "loss": 0.2832, "step": 3425 }, { "epoch": 33.22048192771084, "grad_norm": 19.175912857055664, "learning_rate": 8.336893203883495e-05, "loss": 0.1729, "step": 3426 }, { "epoch": 33.23012048192771, "grad_norm": 10.671881675720215, "learning_rate": 8.336407766990292e-05, "loss": 0.3495, "step": 3427 }, { "epoch": 33.23975903614458, "grad_norm": 16.61208724975586, "learning_rate": 8.335922330097088e-05, "loss": 0.5112, "step": 3428 }, { "epoch": 33.24939759036145, "grad_norm": 25.391948699951172, "learning_rate": 8.335436893203884e-05, "loss": 0.2957, "step": 3429 }, { "epoch": 33.25903614457831, "grad_norm": 20.124937057495117, "learning_rate": 8.33495145631068e-05, "loss": 0.2794, "step": 3430 }, { "epoch": 33.26867469879518, "grad_norm": 2.974222421646118, "learning_rate": 8.334466019417477e-05, "loss": 0.1118, "step": 3431 }, { "epoch": 33.27831325301205, "grad_norm": 25.354019165039062, "learning_rate": 8.333980582524272e-05, "loss": 0.2668, "step": 3432 }, { "epoch": 33.287951807228914, "grad_norm": 3.462249755859375, "learning_rate": 8.333495145631068e-05, "loss": 0.2331, "step": 3433 }, { "epoch": 33.297590361445785, "grad_norm": 7.333792209625244, "learning_rate": 8.333009708737864e-05, "loss": 0.4069, "step": 3434 }, { "epoch": 33.30722891566265, "grad_norm": 19.556991577148438, "learning_rate": 8.332524271844661e-05, "loss": 0.4491, "step": 3435 }, { "epoch": 33.31686746987952, "grad_norm": 1.378514051437378, "learning_rate": 8.332038834951457e-05, "loss": 0.2912, "step": 3436 }, { "epoch": 33.32650602409639, "grad_norm": 7.092177867889404, "learning_rate": 8.331553398058253e-05, "loss": 0.209, "step": 3437 }, { "epoch": 33.33614457831325, "grad_norm": 20.15179443359375, "learning_rate": 8.331067961165048e-05, "loss": 0.5625, "step": 3438 }, { "epoch": 33.34578313253012, "grad_norm": 7.733619689941406, "learning_rate": 8.330582524271844e-05, "loss": 0.3746, "step": 3439 }, { "epoch": 33.355421686746986, "grad_norm": 16.086942672729492, "learning_rate": 8.330097087378641e-05, "loss": 0.2607, "step": 3440 }, { "epoch": 33.36506024096386, "grad_norm": 2.935986280441284, "learning_rate": 8.329611650485437e-05, "loss": 0.1751, "step": 3441 }, { "epoch": 33.37469879518072, "grad_norm": 3.818817615509033, "learning_rate": 8.329126213592233e-05, "loss": 0.1829, "step": 3442 }, { "epoch": 33.38433734939759, "grad_norm": 32.412349700927734, "learning_rate": 8.328640776699029e-05, "loss": 0.2059, "step": 3443 }, { "epoch": 33.39397590361446, "grad_norm": 13.305150032043457, "learning_rate": 8.328155339805826e-05, "loss": 0.4659, "step": 3444 }, { "epoch": 33.403614457831324, "grad_norm": 8.327999114990234, "learning_rate": 8.327669902912622e-05, "loss": 0.5331, "step": 3445 }, { "epoch": 33.413253012048195, "grad_norm": 4.726358890533447, "learning_rate": 8.327184466019417e-05, "loss": 0.2812, "step": 3446 }, { "epoch": 33.42289156626506, "grad_norm": 14.344779968261719, "learning_rate": 8.326699029126213e-05, "loss": 0.329, "step": 3447 }, { "epoch": 33.43253012048193, "grad_norm": 20.088560104370117, "learning_rate": 8.32621359223301e-05, "loss": 0.4668, "step": 3448 }, { "epoch": 33.44216867469879, "grad_norm": 29.1879940032959, "learning_rate": 8.325728155339806e-05, "loss": 0.3042, "step": 3449 }, { "epoch": 33.45180722891566, "grad_norm": 9.924168586730957, "learning_rate": 8.325242718446602e-05, "loss": 0.3417, "step": 3450 }, { "epoch": 33.46144578313253, "grad_norm": 22.221296310424805, "learning_rate": 8.324757281553398e-05, "loss": 0.1814, "step": 3451 }, { "epoch": 33.471084337349396, "grad_norm": 14.974710464477539, "learning_rate": 8.324271844660195e-05, "loss": 0.3674, "step": 3452 }, { "epoch": 33.480722891566266, "grad_norm": 5.925137996673584, "learning_rate": 8.32378640776699e-05, "loss": 0.3373, "step": 3453 }, { "epoch": 33.49036144578313, "grad_norm": 7.422428131103516, "learning_rate": 8.323300970873786e-05, "loss": 0.1981, "step": 3454 }, { "epoch": 33.5, "grad_norm": 5.144602298736572, "learning_rate": 8.322815533980582e-05, "loss": 0.3278, "step": 3455 }, { "epoch": 33.50963855421687, "grad_norm": 7.527886867523193, "learning_rate": 8.322330097087378e-05, "loss": 0.408, "step": 3456 }, { "epoch": 33.519277108433734, "grad_norm": 2.3941924571990967, "learning_rate": 8.321844660194175e-05, "loss": 0.3441, "step": 3457 }, { "epoch": 33.528915662650604, "grad_norm": 7.155960559844971, "learning_rate": 8.321359223300971e-05, "loss": 0.4581, "step": 3458 }, { "epoch": 33.53855421686747, "grad_norm": 11.816633224487305, "learning_rate": 8.320873786407768e-05, "loss": 0.2598, "step": 3459 }, { "epoch": 33.54819277108434, "grad_norm": 34.96955108642578, "learning_rate": 8.320388349514564e-05, "loss": 0.5653, "step": 3460 }, { "epoch": 33.55783132530121, "grad_norm": 9.633713722229004, "learning_rate": 8.31990291262136e-05, "loss": 0.2472, "step": 3461 }, { "epoch": 33.56746987951807, "grad_norm": 12.55183219909668, "learning_rate": 8.319417475728157e-05, "loss": 0.2042, "step": 3462 }, { "epoch": 33.57710843373494, "grad_norm": 14.970256805419922, "learning_rate": 8.318932038834952e-05, "loss": 0.3202, "step": 3463 }, { "epoch": 33.586746987951805, "grad_norm": 8.290794372558594, "learning_rate": 8.318446601941748e-05, "loss": 0.3909, "step": 3464 }, { "epoch": 33.596385542168676, "grad_norm": 3.847588062286377, "learning_rate": 8.317961165048544e-05, "loss": 0.3917, "step": 3465 }, { "epoch": 33.60602409638554, "grad_norm": 9.455058097839355, "learning_rate": 8.317475728155341e-05, "loss": 0.2174, "step": 3466 }, { "epoch": 33.61566265060241, "grad_norm": 26.27269744873047, "learning_rate": 8.316990291262137e-05, "loss": 0.3684, "step": 3467 }, { "epoch": 33.62530120481928, "grad_norm": 18.545488357543945, "learning_rate": 8.316504854368933e-05, "loss": 0.3338, "step": 3468 }, { "epoch": 33.63493975903614, "grad_norm": 15.199871063232422, "learning_rate": 8.316019417475728e-05, "loss": 0.2431, "step": 3469 }, { "epoch": 33.644578313253014, "grad_norm": 12.563154220581055, "learning_rate": 8.315533980582526e-05, "loss": 0.2646, "step": 3470 }, { "epoch": 33.65421686746988, "grad_norm": 11.79407787322998, "learning_rate": 8.315048543689321e-05, "loss": 0.3923, "step": 3471 }, { "epoch": 33.66385542168675, "grad_norm": 5.9661149978637695, "learning_rate": 8.314563106796117e-05, "loss": 0.2727, "step": 3472 }, { "epoch": 33.67349397590361, "grad_norm": 12.438655853271484, "learning_rate": 8.314077669902913e-05, "loss": 0.2752, "step": 3473 }, { "epoch": 33.68313253012048, "grad_norm": 26.904884338378906, "learning_rate": 8.31359223300971e-05, "loss": 0.4168, "step": 3474 }, { "epoch": 33.69277108433735, "grad_norm": 23.735504150390625, "learning_rate": 8.313106796116506e-05, "loss": 0.2424, "step": 3475 }, { "epoch": 33.702409638554215, "grad_norm": 17.382768630981445, "learning_rate": 8.312621359223302e-05, "loss": 0.4415, "step": 3476 }, { "epoch": 33.712048192771086, "grad_norm": 19.13584327697754, "learning_rate": 8.312135922330097e-05, "loss": 0.3502, "step": 3477 }, { "epoch": 33.72168674698795, "grad_norm": 21.0825252532959, "learning_rate": 8.311650485436893e-05, "loss": 0.3354, "step": 3478 }, { "epoch": 33.73132530120482, "grad_norm": 2.5214040279388428, "learning_rate": 8.31116504854369e-05, "loss": 0.2809, "step": 3479 }, { "epoch": 33.74096385542169, "grad_norm": 33.40382766723633, "learning_rate": 8.310679611650486e-05, "loss": 0.5068, "step": 3480 }, { "epoch": 33.75060240963855, "grad_norm": 19.668132781982422, "learning_rate": 8.310194174757282e-05, "loss": 0.2382, "step": 3481 }, { "epoch": 33.76024096385542, "grad_norm": 3.9800775051116943, "learning_rate": 8.309708737864078e-05, "loss": 0.3652, "step": 3482 }, { "epoch": 33.76987951807229, "grad_norm": 23.785306930541992, "learning_rate": 8.309223300970875e-05, "loss": 0.4389, "step": 3483 }, { "epoch": 33.77951807228916, "grad_norm": 1.7568644285202026, "learning_rate": 8.30873786407767e-05, "loss": 0.1518, "step": 3484 }, { "epoch": 33.78915662650603, "grad_norm": 25.108482360839844, "learning_rate": 8.308252427184466e-05, "loss": 0.4649, "step": 3485 }, { "epoch": 33.79879518072289, "grad_norm": 27.744306564331055, "learning_rate": 8.307766990291262e-05, "loss": 0.6528, "step": 3486 }, { "epoch": 33.80843373493976, "grad_norm": 28.777307510375977, "learning_rate": 8.307281553398059e-05, "loss": 0.3033, "step": 3487 }, { "epoch": 33.818072289156625, "grad_norm": 8.623684883117676, "learning_rate": 8.306796116504855e-05, "loss": 0.2785, "step": 3488 }, { "epoch": 33.827710843373495, "grad_norm": 5.540765285491943, "learning_rate": 8.30631067961165e-05, "loss": 0.3507, "step": 3489 }, { "epoch": 33.83734939759036, "grad_norm": 17.630596160888672, "learning_rate": 8.305825242718446e-05, "loss": 0.3751, "step": 3490 }, { "epoch": 33.84698795180723, "grad_norm": 23.420228958129883, "learning_rate": 8.305339805825244e-05, "loss": 0.4537, "step": 3491 }, { "epoch": 33.8566265060241, "grad_norm": 9.21094036102295, "learning_rate": 8.30485436893204e-05, "loss": 0.2398, "step": 3492 }, { "epoch": 33.86626506024096, "grad_norm": 19.524076461791992, "learning_rate": 8.304368932038835e-05, "loss": 0.2801, "step": 3493 }, { "epoch": 33.87590361445783, "grad_norm": 6.276140213012695, "learning_rate": 8.303883495145631e-05, "loss": 0.3011, "step": 3494 }, { "epoch": 33.8855421686747, "grad_norm": 3.9532358646392822, "learning_rate": 8.303398058252427e-05, "loss": 0.3424, "step": 3495 }, { "epoch": 33.89518072289157, "grad_norm": 6.040084362030029, "learning_rate": 8.302912621359224e-05, "loss": 0.3926, "step": 3496 }, { "epoch": 33.90481927710843, "grad_norm": 2.012232780456543, "learning_rate": 8.30242718446602e-05, "loss": 0.284, "step": 3497 }, { "epoch": 33.9144578313253, "grad_norm": 8.332056045532227, "learning_rate": 8.301941747572815e-05, "loss": 0.276, "step": 3498 }, { "epoch": 33.92409638554217, "grad_norm": 23.66985321044922, "learning_rate": 8.301456310679611e-05, "loss": 0.2989, "step": 3499 }, { "epoch": 33.933734939759034, "grad_norm": 12.885994911193848, "learning_rate": 8.300970873786408e-05, "loss": 0.1896, "step": 3500 }, { "epoch": 33.943373493975905, "grad_norm": 8.251449584960938, "learning_rate": 8.300485436893204e-05, "loss": 0.2715, "step": 3501 }, { "epoch": 33.95301204819277, "grad_norm": 21.28561019897461, "learning_rate": 8.3e-05, "loss": 0.4414, "step": 3502 }, { "epoch": 33.96265060240964, "grad_norm": 5.479055881500244, "learning_rate": 8.299514563106796e-05, "loss": 0.241, "step": 3503 }, { "epoch": 33.97228915662651, "grad_norm": 25.05200958251953, "learning_rate": 8.299029126213593e-05, "loss": 0.1556, "step": 3504 }, { "epoch": 33.98192771084337, "grad_norm": 2.681151866912842, "learning_rate": 8.298543689320388e-05, "loss": 0.4159, "step": 3505 }, { "epoch": 33.99156626506024, "grad_norm": 19.966899871826172, "learning_rate": 8.298058252427184e-05, "loss": 0.2377, "step": 3506 }, { "epoch": 34.00722891566265, "grad_norm": 56.667972564697266, "learning_rate": 8.29757281553398e-05, "loss": 0.3695, "step": 3507 }, { "epoch": 34.01686746987952, "grad_norm": 25.97195053100586, "learning_rate": 8.297087378640777e-05, "loss": 0.4731, "step": 3508 }, { "epoch": 34.026506024096385, "grad_norm": 6.23313570022583, "learning_rate": 8.296601941747573e-05, "loss": 0.3173, "step": 3509 }, { "epoch": 34.036144578313255, "grad_norm": 4.041944980621338, "learning_rate": 8.29611650485437e-05, "loss": 0.2174, "step": 3510 }, { "epoch": 34.04578313253012, "grad_norm": 4.889497756958008, "learning_rate": 8.295631067961166e-05, "loss": 0.2976, "step": 3511 }, { "epoch": 34.05542168674699, "grad_norm": 10.642328262329102, "learning_rate": 8.295145631067962e-05, "loss": 0.3275, "step": 3512 }, { "epoch": 34.06506024096385, "grad_norm": 8.475943565368652, "learning_rate": 8.294660194174759e-05, "loss": 0.473, "step": 3513 }, { "epoch": 34.07469879518072, "grad_norm": 14.84055233001709, "learning_rate": 8.294174757281555e-05, "loss": 0.2435, "step": 3514 }, { "epoch": 34.08433734939759, "grad_norm": 2.079157829284668, "learning_rate": 8.29368932038835e-05, "loss": 0.2124, "step": 3515 }, { "epoch": 34.09397590361446, "grad_norm": 14.239971160888672, "learning_rate": 8.293203883495146e-05, "loss": 0.1963, "step": 3516 }, { "epoch": 34.10361445783133, "grad_norm": 25.32750701904297, "learning_rate": 8.292718446601942e-05, "loss": 0.2598, "step": 3517 }, { "epoch": 34.11325301204819, "grad_norm": 2.2504193782806396, "learning_rate": 8.292233009708739e-05, "loss": 0.2389, "step": 3518 }, { "epoch": 34.12289156626506, "grad_norm": 5.886616230010986, "learning_rate": 8.291747572815535e-05, "loss": 0.1134, "step": 3519 }, { "epoch": 34.13253012048193, "grad_norm": 7.1720662117004395, "learning_rate": 8.29126213592233e-05, "loss": 0.3178, "step": 3520 }, { "epoch": 34.142168674698794, "grad_norm": 36.79134750366211, "learning_rate": 8.290776699029126e-05, "loss": 0.3753, "step": 3521 }, { "epoch": 34.151807228915665, "grad_norm": 7.098700046539307, "learning_rate": 8.290291262135923e-05, "loss": 0.3885, "step": 3522 }, { "epoch": 34.16144578313253, "grad_norm": 16.91274642944336, "learning_rate": 8.289805825242719e-05, "loss": 0.4481, "step": 3523 }, { "epoch": 34.1710843373494, "grad_norm": 34.54959487915039, "learning_rate": 8.289320388349515e-05, "loss": 0.313, "step": 3524 }, { "epoch": 34.18072289156626, "grad_norm": 8.167377471923828, "learning_rate": 8.288834951456311e-05, "loss": 0.3491, "step": 3525 }, { "epoch": 34.19036144578313, "grad_norm": 13.113202095031738, "learning_rate": 8.288349514563108e-05, "loss": 0.3186, "step": 3526 }, { "epoch": 34.2, "grad_norm": 11.16904067993164, "learning_rate": 8.287864077669904e-05, "loss": 0.3135, "step": 3527 }, { "epoch": 34.209638554216866, "grad_norm": 6.388293266296387, "learning_rate": 8.2873786407767e-05, "loss": 0.2785, "step": 3528 }, { "epoch": 34.21927710843374, "grad_norm": 2.49979829788208, "learning_rate": 8.286893203883495e-05, "loss": 0.3443, "step": 3529 }, { "epoch": 34.2289156626506, "grad_norm": 5.697763442993164, "learning_rate": 8.286407766990292e-05, "loss": 0.2565, "step": 3530 }, { "epoch": 34.23855421686747, "grad_norm": 4.179762363433838, "learning_rate": 8.285922330097088e-05, "loss": 0.0978, "step": 3531 }, { "epoch": 34.24819277108434, "grad_norm": 25.728979110717773, "learning_rate": 8.285436893203884e-05, "loss": 0.3613, "step": 3532 }, { "epoch": 34.257831325301204, "grad_norm": 17.708547592163086, "learning_rate": 8.28495145631068e-05, "loss": 0.5065, "step": 3533 }, { "epoch": 34.267469879518075, "grad_norm": 36.49027633666992, "learning_rate": 8.284466019417475e-05, "loss": 0.1837, "step": 3534 }, { "epoch": 34.27710843373494, "grad_norm": 12.102835655212402, "learning_rate": 8.283980582524273e-05, "loss": 0.3045, "step": 3535 }, { "epoch": 34.28674698795181, "grad_norm": 5.591787815093994, "learning_rate": 8.283495145631068e-05, "loss": 0.3901, "step": 3536 }, { "epoch": 34.29638554216867, "grad_norm": 23.43857765197754, "learning_rate": 8.283009708737864e-05, "loss": 0.4095, "step": 3537 }, { "epoch": 34.30602409638554, "grad_norm": 10.887813568115234, "learning_rate": 8.28252427184466e-05, "loss": 0.4119, "step": 3538 }, { "epoch": 34.31566265060241, "grad_norm": 6.179720401763916, "learning_rate": 8.282038834951457e-05, "loss": 0.2129, "step": 3539 }, { "epoch": 34.325301204819276, "grad_norm": 45.1407356262207, "learning_rate": 8.281553398058253e-05, "loss": 0.4018, "step": 3540 }, { "epoch": 34.334939759036146, "grad_norm": 3.8742265701293945, "learning_rate": 8.281067961165049e-05, "loss": 0.2367, "step": 3541 }, { "epoch": 34.34457831325301, "grad_norm": 11.373680114746094, "learning_rate": 8.280582524271844e-05, "loss": 0.4844, "step": 3542 }, { "epoch": 34.35421686746988, "grad_norm": 22.087690353393555, "learning_rate": 8.280097087378642e-05, "loss": 0.2905, "step": 3543 }, { "epoch": 34.36385542168675, "grad_norm": 4.468733787536621, "learning_rate": 8.279611650485437e-05, "loss": 0.2288, "step": 3544 }, { "epoch": 34.373493975903614, "grad_norm": 10.415626525878906, "learning_rate": 8.279126213592233e-05, "loss": 0.1934, "step": 3545 }, { "epoch": 34.383132530120484, "grad_norm": 15.002968788146973, "learning_rate": 8.278640776699029e-05, "loss": 0.4491, "step": 3546 }, { "epoch": 34.39277108433735, "grad_norm": 6.046592712402344, "learning_rate": 8.278155339805826e-05, "loss": 0.2671, "step": 3547 }, { "epoch": 34.40240963855422, "grad_norm": 10.723692893981934, "learning_rate": 8.277669902912622e-05, "loss": 0.3495, "step": 3548 }, { "epoch": 34.41204819277108, "grad_norm": 7.56532621383667, "learning_rate": 8.277184466019418e-05, "loss": 0.2327, "step": 3549 }, { "epoch": 34.42168674698795, "grad_norm": 5.92044734954834, "learning_rate": 8.276699029126213e-05, "loss": 0.2815, "step": 3550 }, { "epoch": 34.43132530120482, "grad_norm": 13.390686988830566, "learning_rate": 8.276213592233009e-05, "loss": 0.2305, "step": 3551 }, { "epoch": 34.440963855421685, "grad_norm": 29.305992126464844, "learning_rate": 8.275728155339806e-05, "loss": 0.2481, "step": 3552 }, { "epoch": 34.450602409638556, "grad_norm": 10.355457305908203, "learning_rate": 8.275242718446602e-05, "loss": 0.4274, "step": 3553 }, { "epoch": 34.46024096385542, "grad_norm": 26.843238830566406, "learning_rate": 8.274757281553398e-05, "loss": 0.4453, "step": 3554 }, { "epoch": 34.46987951807229, "grad_norm": 4.2548675537109375, "learning_rate": 8.274271844660194e-05, "loss": 0.2443, "step": 3555 }, { "epoch": 34.47951807228916, "grad_norm": 20.00388526916504, "learning_rate": 8.273786407766991e-05, "loss": 0.6052, "step": 3556 }, { "epoch": 34.48915662650602, "grad_norm": 13.759217262268066, "learning_rate": 8.273300970873786e-05, "loss": 0.5629, "step": 3557 }, { "epoch": 34.498795180722894, "grad_norm": 32.125709533691406, "learning_rate": 8.272815533980582e-05, "loss": 0.3556, "step": 3558 }, { "epoch": 34.50843373493976, "grad_norm": 12.757682800292969, "learning_rate": 8.272330097087378e-05, "loss": 0.281, "step": 3559 }, { "epoch": 34.51807228915663, "grad_norm": 19.840852737426758, "learning_rate": 8.271844660194175e-05, "loss": 0.336, "step": 3560 }, { "epoch": 34.52771084337349, "grad_norm": 6.786267280578613, "learning_rate": 8.271359223300972e-05, "loss": 0.324, "step": 3561 }, { "epoch": 34.53734939759036, "grad_norm": 10.756643295288086, "learning_rate": 8.270873786407768e-05, "loss": 0.2677, "step": 3562 }, { "epoch": 34.54698795180723, "grad_norm": 7.528720378875732, "learning_rate": 8.270388349514564e-05, "loss": 0.3986, "step": 3563 }, { "epoch": 34.556626506024095, "grad_norm": 15.508282661437988, "learning_rate": 8.26990291262136e-05, "loss": 0.461, "step": 3564 }, { "epoch": 34.566265060240966, "grad_norm": 12.95091438293457, "learning_rate": 8.269417475728157e-05, "loss": 0.33, "step": 3565 }, { "epoch": 34.57590361445783, "grad_norm": 11.840478897094727, "learning_rate": 8.268932038834953e-05, "loss": 0.3662, "step": 3566 }, { "epoch": 34.5855421686747, "grad_norm": 15.157370567321777, "learning_rate": 8.268446601941748e-05, "loss": 0.2103, "step": 3567 }, { "epoch": 34.59518072289157, "grad_norm": 15.316061973571777, "learning_rate": 8.267961165048544e-05, "loss": 0.3613, "step": 3568 }, { "epoch": 34.60481927710843, "grad_norm": 14.879199981689453, "learning_rate": 8.267475728155341e-05, "loss": 0.4141, "step": 3569 }, { "epoch": 34.6144578313253, "grad_norm": 11.551783561706543, "learning_rate": 8.266990291262137e-05, "loss": 0.2034, "step": 3570 }, { "epoch": 34.62409638554217, "grad_norm": 15.20410442352295, "learning_rate": 8.266504854368933e-05, "loss": 0.4146, "step": 3571 }, { "epoch": 34.63373493975904, "grad_norm": 6.653562068939209, "learning_rate": 8.266019417475729e-05, "loss": 0.3212, "step": 3572 }, { "epoch": 34.6433734939759, "grad_norm": 13.820991516113281, "learning_rate": 8.265533980582524e-05, "loss": 0.3904, "step": 3573 }, { "epoch": 34.65301204819277, "grad_norm": 5.014748573303223, "learning_rate": 8.265048543689321e-05, "loss": 0.3037, "step": 3574 }, { "epoch": 34.66265060240964, "grad_norm": 14.768170356750488, "learning_rate": 8.264563106796117e-05, "loss": 0.2421, "step": 3575 }, { "epoch": 34.672289156626505, "grad_norm": 10.59442138671875, "learning_rate": 8.264077669902913e-05, "loss": 0.26, "step": 3576 }, { "epoch": 34.681927710843375, "grad_norm": 12.930530548095703, "learning_rate": 8.263592233009709e-05, "loss": 0.2249, "step": 3577 }, { "epoch": 34.69156626506024, "grad_norm": 7.063936710357666, "learning_rate": 8.263106796116506e-05, "loss": 0.2245, "step": 3578 }, { "epoch": 34.70120481927711, "grad_norm": 3.2606921195983887, "learning_rate": 8.262621359223302e-05, "loss": 0.222, "step": 3579 }, { "epoch": 34.71084337349397, "grad_norm": 7.494451999664307, "learning_rate": 8.262135922330097e-05, "loss": 0.2439, "step": 3580 }, { "epoch": 34.72048192771084, "grad_norm": 28.574039459228516, "learning_rate": 8.261650485436893e-05, "loss": 0.639, "step": 3581 }, { "epoch": 34.73012048192771, "grad_norm": 28.728397369384766, "learning_rate": 8.26116504854369e-05, "loss": 0.3234, "step": 3582 }, { "epoch": 34.73975903614458, "grad_norm": 4.3399152755737305, "learning_rate": 8.260679611650486e-05, "loss": 0.1988, "step": 3583 }, { "epoch": 34.74939759036145, "grad_norm": 16.528114318847656, "learning_rate": 8.260194174757282e-05, "loss": 0.5445, "step": 3584 }, { "epoch": 34.75903614457831, "grad_norm": 10.328648567199707, "learning_rate": 8.259708737864078e-05, "loss": 0.3303, "step": 3585 }, { "epoch": 34.76867469879518, "grad_norm": 6.769685745239258, "learning_rate": 8.259223300970875e-05, "loss": 0.5374, "step": 3586 }, { "epoch": 34.77831325301205, "grad_norm": 67.15789794921875, "learning_rate": 8.25873786407767e-05, "loss": 0.3624, "step": 3587 }, { "epoch": 34.787951807228914, "grad_norm": 25.54003143310547, "learning_rate": 8.258252427184466e-05, "loss": 0.3416, "step": 3588 }, { "epoch": 34.797590361445785, "grad_norm": 2.307351589202881, "learning_rate": 8.257766990291262e-05, "loss": 0.2946, "step": 3589 }, { "epoch": 34.80722891566265, "grad_norm": 9.619743347167969, "learning_rate": 8.257281553398058e-05, "loss": 0.3022, "step": 3590 }, { "epoch": 34.81686746987952, "grad_norm": 7.396059989929199, "learning_rate": 8.256796116504855e-05, "loss": 0.2474, "step": 3591 }, { "epoch": 34.82650602409639, "grad_norm": 9.141266822814941, "learning_rate": 8.256310679611651e-05, "loss": 0.2698, "step": 3592 }, { "epoch": 34.83614457831325, "grad_norm": 5.058606147766113, "learning_rate": 8.255825242718447e-05, "loss": 0.223, "step": 3593 }, { "epoch": 34.84578313253012, "grad_norm": 12.86490249633789, "learning_rate": 8.255339805825242e-05, "loss": 0.4033, "step": 3594 }, { "epoch": 34.855421686746986, "grad_norm": 15.124314308166504, "learning_rate": 8.25485436893204e-05, "loss": 0.3821, "step": 3595 }, { "epoch": 34.86506024096386, "grad_norm": 7.328298568725586, "learning_rate": 8.254368932038835e-05, "loss": 0.2122, "step": 3596 }, { "epoch": 34.87469879518072, "grad_norm": 31.788814544677734, "learning_rate": 8.253883495145631e-05, "loss": 0.4541, "step": 3597 }, { "epoch": 34.88433734939759, "grad_norm": 16.550630569458008, "learning_rate": 8.253398058252427e-05, "loss": 0.258, "step": 3598 }, { "epoch": 34.89397590361446, "grad_norm": 19.58555793762207, "learning_rate": 8.252912621359224e-05, "loss": 0.4209, "step": 3599 }, { "epoch": 34.903614457831324, "grad_norm": 13.247913360595703, "learning_rate": 8.25242718446602e-05, "loss": 0.3427, "step": 3600 }, { "epoch": 34.913253012048195, "grad_norm": 9.622272491455078, "learning_rate": 8.251941747572815e-05, "loss": 0.4344, "step": 3601 }, { "epoch": 34.92289156626506, "grad_norm": 14.35428237915039, "learning_rate": 8.251456310679611e-05, "loss": 0.2343, "step": 3602 }, { "epoch": 34.93253012048193, "grad_norm": 36.10122299194336, "learning_rate": 8.250970873786408e-05, "loss": 0.3074, "step": 3603 }, { "epoch": 34.94216867469879, "grad_norm": 1.9433635473251343, "learning_rate": 8.250485436893204e-05, "loss": 0.2195, "step": 3604 }, { "epoch": 34.95180722891566, "grad_norm": 31.51957893371582, "learning_rate": 8.25e-05, "loss": 0.5095, "step": 3605 }, { "epoch": 34.96144578313253, "grad_norm": 10.85533332824707, "learning_rate": 8.249514563106796e-05, "loss": 0.3376, "step": 3606 }, { "epoch": 34.971084337349396, "grad_norm": 11.975939750671387, "learning_rate": 8.249029126213592e-05, "loss": 0.4463, "step": 3607 }, { "epoch": 34.980722891566266, "grad_norm": 4.8512043952941895, "learning_rate": 8.248543689320389e-05, "loss": 0.2156, "step": 3608 }, { "epoch": 34.99036144578313, "grad_norm": 11.816681861877441, "learning_rate": 8.248058252427184e-05, "loss": 0.1957, "step": 3609 }, { "epoch": 35.006024096385545, "grad_norm": 11.114798545837402, "learning_rate": 8.24757281553398e-05, "loss": 0.2754, "step": 3610 }, { "epoch": 35.01566265060241, "grad_norm": 18.120695114135742, "learning_rate": 8.247087378640777e-05, "loss": 0.4653, "step": 3611 }, { "epoch": 35.02530120481928, "grad_norm": 11.807130813598633, "learning_rate": 8.246601941747573e-05, "loss": 0.1033, "step": 3612 }, { "epoch": 35.03493975903614, "grad_norm": 5.448370933532715, "learning_rate": 8.24611650485437e-05, "loss": 0.3047, "step": 3613 }, { "epoch": 35.04457831325301, "grad_norm": 41.648372650146484, "learning_rate": 8.245631067961166e-05, "loss": 0.4206, "step": 3614 }, { "epoch": 35.05421686746988, "grad_norm": 6.248712539672852, "learning_rate": 8.245145631067962e-05, "loss": 0.4928, "step": 3615 }, { "epoch": 35.063855421686746, "grad_norm": 12.14569091796875, "learning_rate": 8.244660194174758e-05, "loss": 0.3534, "step": 3616 }, { "epoch": 35.07349397590362, "grad_norm": 7.727815628051758, "learning_rate": 8.244174757281555e-05, "loss": 0.2966, "step": 3617 }, { "epoch": 35.08313253012048, "grad_norm": 3.2107880115509033, "learning_rate": 8.24368932038835e-05, "loss": 0.1668, "step": 3618 }, { "epoch": 35.09277108433735, "grad_norm": 11.864227294921875, "learning_rate": 8.243203883495146e-05, "loss": 0.2779, "step": 3619 }, { "epoch": 35.102409638554214, "grad_norm": 32.029624938964844, "learning_rate": 8.242718446601942e-05, "loss": 0.4679, "step": 3620 }, { "epoch": 35.112048192771084, "grad_norm": 7.245349884033203, "learning_rate": 8.242233009708739e-05, "loss": 0.2924, "step": 3621 }, { "epoch": 35.121686746987955, "grad_norm": 54.079063415527344, "learning_rate": 8.241747572815535e-05, "loss": 0.3941, "step": 3622 }, { "epoch": 35.13132530120482, "grad_norm": 4.2106194496154785, "learning_rate": 8.241262135922331e-05, "loss": 0.2772, "step": 3623 }, { "epoch": 35.14096385542169, "grad_norm": 10.7633056640625, "learning_rate": 8.240776699029126e-05, "loss": 0.3186, "step": 3624 }, { "epoch": 35.15060240963855, "grad_norm": 8.576643943786621, "learning_rate": 8.240291262135924e-05, "loss": 0.2749, "step": 3625 }, { "epoch": 35.16024096385542, "grad_norm": 4.846395015716553, "learning_rate": 8.23980582524272e-05, "loss": 0.3183, "step": 3626 }, { "epoch": 35.16987951807229, "grad_norm": 18.947998046875, "learning_rate": 8.239320388349515e-05, "loss": 0.1359, "step": 3627 }, { "epoch": 35.179518072289156, "grad_norm": 21.252674102783203, "learning_rate": 8.238834951456311e-05, "loss": 0.5468, "step": 3628 }, { "epoch": 35.189156626506026, "grad_norm": 5.854010581970215, "learning_rate": 8.238349514563107e-05, "loss": 0.4401, "step": 3629 }, { "epoch": 35.19879518072289, "grad_norm": 5.815518856048584, "learning_rate": 8.237864077669904e-05, "loss": 0.5055, "step": 3630 }, { "epoch": 35.20843373493976, "grad_norm": 3.7849369049072266, "learning_rate": 8.2373786407767e-05, "loss": 0.3366, "step": 3631 }, { "epoch": 35.21807228915662, "grad_norm": 11.41663646697998, "learning_rate": 8.236893203883495e-05, "loss": 0.2049, "step": 3632 }, { "epoch": 35.227710843373494, "grad_norm": 11.716797828674316, "learning_rate": 8.236407766990291e-05, "loss": 0.2177, "step": 3633 }, { "epoch": 35.237349397590364, "grad_norm": 14.937954902648926, "learning_rate": 8.235922330097088e-05, "loss": 0.1509, "step": 3634 }, { "epoch": 35.24698795180723, "grad_norm": 14.352350234985352, "learning_rate": 8.235436893203884e-05, "loss": 0.3914, "step": 3635 }, { "epoch": 35.2566265060241, "grad_norm": 7.82935905456543, "learning_rate": 8.23495145631068e-05, "loss": 0.3923, "step": 3636 }, { "epoch": 35.26626506024096, "grad_norm": 17.804738998413086, "learning_rate": 8.234466019417476e-05, "loss": 0.486, "step": 3637 }, { "epoch": 35.27590361445783, "grad_norm": 11.626181602478027, "learning_rate": 8.233980582524273e-05, "loss": 0.1768, "step": 3638 }, { "epoch": 35.2855421686747, "grad_norm": 5.49812126159668, "learning_rate": 8.233495145631069e-05, "loss": 0.3175, "step": 3639 }, { "epoch": 35.295180722891565, "grad_norm": 5.241804599761963, "learning_rate": 8.233009708737864e-05, "loss": 0.2911, "step": 3640 }, { "epoch": 35.304819277108436, "grad_norm": 4.539982318878174, "learning_rate": 8.23252427184466e-05, "loss": 0.2115, "step": 3641 }, { "epoch": 35.3144578313253, "grad_norm": 6.775118350982666, "learning_rate": 8.232038834951457e-05, "loss": 0.4008, "step": 3642 }, { "epoch": 35.32409638554217, "grad_norm": 15.812118530273438, "learning_rate": 8.231553398058253e-05, "loss": 0.55, "step": 3643 }, { "epoch": 35.33373493975903, "grad_norm": 9.42573070526123, "learning_rate": 8.231067961165049e-05, "loss": 0.5373, "step": 3644 }, { "epoch": 35.3433734939759, "grad_norm": 10.99870777130127, "learning_rate": 8.230582524271845e-05, "loss": 0.3206, "step": 3645 }, { "epoch": 35.353012048192774, "grad_norm": 1.6969120502471924, "learning_rate": 8.23009708737864e-05, "loss": 0.1595, "step": 3646 }, { "epoch": 35.36265060240964, "grad_norm": 7.729437828063965, "learning_rate": 8.229611650485437e-05, "loss": 0.6238, "step": 3647 }, { "epoch": 35.37228915662651, "grad_norm": 20.44061851501465, "learning_rate": 8.229126213592233e-05, "loss": 0.1263, "step": 3648 }, { "epoch": 35.38192771084337, "grad_norm": 16.409456253051758, "learning_rate": 8.228640776699029e-05, "loss": 0.4584, "step": 3649 }, { "epoch": 35.39156626506024, "grad_norm": 4.560053825378418, "learning_rate": 8.228155339805825e-05, "loss": 0.3665, "step": 3650 }, { "epoch": 35.40120481927711, "grad_norm": 11.84373664855957, "learning_rate": 8.227669902912622e-05, "loss": 0.4193, "step": 3651 }, { "epoch": 35.410843373493975, "grad_norm": 2.768983840942383, "learning_rate": 8.227184466019418e-05, "loss": 0.3377, "step": 3652 }, { "epoch": 35.420481927710846, "grad_norm": 1.8228681087493896, "learning_rate": 8.226699029126213e-05, "loss": 0.2884, "step": 3653 }, { "epoch": 35.43012048192771, "grad_norm": 6.108584403991699, "learning_rate": 8.226213592233009e-05, "loss": 0.383, "step": 3654 }, { "epoch": 35.43975903614458, "grad_norm": 6.983123302459717, "learning_rate": 8.225728155339806e-05, "loss": 0.329, "step": 3655 }, { "epoch": 35.44939759036144, "grad_norm": 2.3021438121795654, "learning_rate": 8.225242718446602e-05, "loss": 0.2667, "step": 3656 }, { "epoch": 35.45903614457831, "grad_norm": 3.198138952255249, "learning_rate": 8.224757281553398e-05, "loss": 0.1742, "step": 3657 }, { "epoch": 35.46867469879518, "grad_norm": 36.37656784057617, "learning_rate": 8.224271844660194e-05, "loss": 0.3838, "step": 3658 }, { "epoch": 35.47831325301205, "grad_norm": 16.271575927734375, "learning_rate": 8.223786407766991e-05, "loss": 0.3938, "step": 3659 }, { "epoch": 35.48795180722892, "grad_norm": 10.628961563110352, "learning_rate": 8.223300970873787e-05, "loss": 0.4826, "step": 3660 }, { "epoch": 35.49759036144578, "grad_norm": 30.7509822845459, "learning_rate": 8.222815533980584e-05, "loss": 0.4321, "step": 3661 }, { "epoch": 35.50722891566265, "grad_norm": 14.89616870880127, "learning_rate": 8.22233009708738e-05, "loss": 0.5969, "step": 3662 }, { "epoch": 35.51686746987952, "grad_norm": 8.238472938537598, "learning_rate": 8.221844660194175e-05, "loss": 0.2447, "step": 3663 }, { "epoch": 35.526506024096385, "grad_norm": 13.290135383605957, "learning_rate": 8.221359223300972e-05, "loss": 0.5383, "step": 3664 }, { "epoch": 35.536144578313255, "grad_norm": 8.048593521118164, "learning_rate": 8.220873786407768e-05, "loss": 0.4244, "step": 3665 }, { "epoch": 35.54578313253012, "grad_norm": 6.191959381103516, "learning_rate": 8.220388349514564e-05, "loss": 0.1769, "step": 3666 }, { "epoch": 35.55542168674699, "grad_norm": 9.056086540222168, "learning_rate": 8.21990291262136e-05, "loss": 0.4177, "step": 3667 }, { "epoch": 35.56506024096385, "grad_norm": 11.962039947509766, "learning_rate": 8.219417475728156e-05, "loss": 0.2471, "step": 3668 }, { "epoch": 35.57469879518072, "grad_norm": 18.670867919921875, "learning_rate": 8.218932038834953e-05, "loss": 0.5128, "step": 3669 }, { "epoch": 35.58433734939759, "grad_norm": 17.08327293395996, "learning_rate": 8.218446601941748e-05, "loss": 0.6135, "step": 3670 }, { "epoch": 35.59397590361446, "grad_norm": 24.13373374938965, "learning_rate": 8.217961165048544e-05, "loss": 0.409, "step": 3671 }, { "epoch": 35.60361445783133, "grad_norm": 5.973228454589844, "learning_rate": 8.21747572815534e-05, "loss": 0.1768, "step": 3672 }, { "epoch": 35.61325301204819, "grad_norm": 33.59287643432617, "learning_rate": 8.216990291262137e-05, "loss": 0.2988, "step": 3673 }, { "epoch": 35.62289156626506, "grad_norm": 7.641523361206055, "learning_rate": 8.216504854368933e-05, "loss": 0.3408, "step": 3674 }, { "epoch": 35.63253012048193, "grad_norm": 4.887516498565674, "learning_rate": 8.216019417475729e-05, "loss": 0.1191, "step": 3675 }, { "epoch": 35.642168674698794, "grad_norm": 6.72687292098999, "learning_rate": 8.215533980582524e-05, "loss": 0.3484, "step": 3676 }, { "epoch": 35.651807228915665, "grad_norm": 17.90115737915039, "learning_rate": 8.215048543689322e-05, "loss": 0.4506, "step": 3677 }, { "epoch": 35.66144578313253, "grad_norm": 9.942901611328125, "learning_rate": 8.214563106796117e-05, "loss": 0.3789, "step": 3678 }, { "epoch": 35.6710843373494, "grad_norm": 14.865127563476562, "learning_rate": 8.214077669902913e-05, "loss": 0.3974, "step": 3679 }, { "epoch": 35.68072289156626, "grad_norm": 4.080019474029541, "learning_rate": 8.213592233009709e-05, "loss": 0.1088, "step": 3680 }, { "epoch": 35.69036144578313, "grad_norm": 15.887353897094727, "learning_rate": 8.213106796116506e-05, "loss": 0.3269, "step": 3681 }, { "epoch": 35.7, "grad_norm": 4.949869632720947, "learning_rate": 8.212621359223302e-05, "loss": 0.2284, "step": 3682 }, { "epoch": 35.709638554216866, "grad_norm": 9.088884353637695, "learning_rate": 8.212135922330098e-05, "loss": 0.2659, "step": 3683 }, { "epoch": 35.71927710843374, "grad_norm": 19.335996627807617, "learning_rate": 8.211650485436893e-05, "loss": 0.3249, "step": 3684 }, { "epoch": 35.7289156626506, "grad_norm": 10.142772674560547, "learning_rate": 8.211165048543689e-05, "loss": 0.4134, "step": 3685 }, { "epoch": 35.73855421686747, "grad_norm": 24.415300369262695, "learning_rate": 8.210679611650486e-05, "loss": 0.4625, "step": 3686 }, { "epoch": 35.74819277108434, "grad_norm": 14.774507522583008, "learning_rate": 8.210194174757282e-05, "loss": 0.3775, "step": 3687 }, { "epoch": 35.757831325301204, "grad_norm": 31.664796829223633, "learning_rate": 8.209708737864078e-05, "loss": 0.4352, "step": 3688 }, { "epoch": 35.767469879518075, "grad_norm": 10.273613929748535, "learning_rate": 8.209223300970874e-05, "loss": 0.214, "step": 3689 }, { "epoch": 35.77710843373494, "grad_norm": 6.174567222595215, "learning_rate": 8.208737864077671e-05, "loss": 0.1753, "step": 3690 }, { "epoch": 35.78674698795181, "grad_norm": 25.453113555908203, "learning_rate": 8.208252427184466e-05, "loss": 0.4329, "step": 3691 }, { "epoch": 35.79638554216867, "grad_norm": 14.615409851074219, "learning_rate": 8.207766990291262e-05, "loss": 0.3962, "step": 3692 }, { "epoch": 35.80602409638554, "grad_norm": 8.651169776916504, "learning_rate": 8.207281553398058e-05, "loss": 0.541, "step": 3693 }, { "epoch": 35.81566265060241, "grad_norm": 22.099624633789062, "learning_rate": 8.206796116504855e-05, "loss": 0.1625, "step": 3694 }, { "epoch": 35.825301204819276, "grad_norm": 9.391587257385254, "learning_rate": 8.206310679611651e-05, "loss": 0.3386, "step": 3695 }, { "epoch": 35.834939759036146, "grad_norm": 6.049586772918701, "learning_rate": 8.205825242718447e-05, "loss": 0.2742, "step": 3696 }, { "epoch": 35.84457831325301, "grad_norm": 22.420536041259766, "learning_rate": 8.205339805825242e-05, "loss": 0.342, "step": 3697 }, { "epoch": 35.85421686746988, "grad_norm": 24.89956283569336, "learning_rate": 8.20485436893204e-05, "loss": 0.3809, "step": 3698 }, { "epoch": 35.86385542168675, "grad_norm": 9.617170333862305, "learning_rate": 8.204368932038835e-05, "loss": 0.3119, "step": 3699 }, { "epoch": 35.873493975903614, "grad_norm": 11.631295204162598, "learning_rate": 8.203883495145631e-05, "loss": 0.2381, "step": 3700 }, { "epoch": 35.883132530120484, "grad_norm": 3.2917978763580322, "learning_rate": 8.203398058252427e-05, "loss": 0.3184, "step": 3701 }, { "epoch": 35.89277108433735, "grad_norm": 2.42942476272583, "learning_rate": 8.202912621359223e-05, "loss": 0.2474, "step": 3702 }, { "epoch": 35.90240963855422, "grad_norm": 10.044413566589355, "learning_rate": 8.20242718446602e-05, "loss": 0.2914, "step": 3703 }, { "epoch": 35.91204819277108, "grad_norm": 20.89301300048828, "learning_rate": 8.201941747572816e-05, "loss": 0.3992, "step": 3704 }, { "epoch": 35.92168674698795, "grad_norm": 27.591211318969727, "learning_rate": 8.201456310679611e-05, "loss": 0.3241, "step": 3705 }, { "epoch": 35.93132530120482, "grad_norm": 22.55430030822754, "learning_rate": 8.200970873786407e-05, "loss": 0.339, "step": 3706 }, { "epoch": 35.940963855421685, "grad_norm": 10.619304656982422, "learning_rate": 8.200485436893204e-05, "loss": 0.394, "step": 3707 }, { "epoch": 35.950602409638556, "grad_norm": 12.438398361206055, "learning_rate": 8.2e-05, "loss": 0.2608, "step": 3708 }, { "epoch": 35.96024096385542, "grad_norm": 6.0824055671691895, "learning_rate": 8.199514563106796e-05, "loss": 0.2477, "step": 3709 }, { "epoch": 35.96987951807229, "grad_norm": 6.654388904571533, "learning_rate": 8.199029126213592e-05, "loss": 0.1242, "step": 3710 }, { "epoch": 35.97951807228916, "grad_norm": 14.809720039367676, "learning_rate": 8.198543689320389e-05, "loss": 0.4275, "step": 3711 }, { "epoch": 35.98915662650602, "grad_norm": 23.52486228942871, "learning_rate": 8.198058252427186e-05, "loss": 0.4159, "step": 3712 }, { "epoch": 36.00481927710843, "grad_norm": 2.798367500305176, "learning_rate": 8.197572815533982e-05, "loss": 0.3739, "step": 3713 }, { "epoch": 36.0144578313253, "grad_norm": 10.016836166381836, "learning_rate": 8.197087378640777e-05, "loss": 0.3172, "step": 3714 }, { "epoch": 36.024096385542165, "grad_norm": 5.075667858123779, "learning_rate": 8.196601941747573e-05, "loss": 0.2012, "step": 3715 }, { "epoch": 36.033734939759036, "grad_norm": 12.841075897216797, "learning_rate": 8.19611650485437e-05, "loss": 0.4686, "step": 3716 }, { "epoch": 36.043373493975906, "grad_norm": 12.097411155700684, "learning_rate": 8.195631067961166e-05, "loss": 0.3171, "step": 3717 }, { "epoch": 36.05301204819277, "grad_norm": 16.389474868774414, "learning_rate": 8.195145631067962e-05, "loss": 0.2529, "step": 3718 }, { "epoch": 36.06265060240964, "grad_norm": 25.744754791259766, "learning_rate": 8.194660194174758e-05, "loss": 0.1771, "step": 3719 }, { "epoch": 36.0722891566265, "grad_norm": 5.08447265625, "learning_rate": 8.194174757281555e-05, "loss": 0.1784, "step": 3720 }, { "epoch": 36.081927710843374, "grad_norm": 6.640146255493164, "learning_rate": 8.19368932038835e-05, "loss": 0.4637, "step": 3721 }, { "epoch": 36.091566265060244, "grad_norm": 2.6853983402252197, "learning_rate": 8.193203883495146e-05, "loss": 0.2688, "step": 3722 }, { "epoch": 36.10120481927711, "grad_norm": 31.966400146484375, "learning_rate": 8.192718446601942e-05, "loss": 0.3465, "step": 3723 }, { "epoch": 36.11084337349398, "grad_norm": 4.629513740539551, "learning_rate": 8.192233009708738e-05, "loss": 0.3366, "step": 3724 }, { "epoch": 36.12048192771084, "grad_norm": 17.5805606842041, "learning_rate": 8.191747572815535e-05, "loss": 0.3493, "step": 3725 }, { "epoch": 36.13012048192771, "grad_norm": 20.559513092041016, "learning_rate": 8.191262135922331e-05, "loss": 0.2459, "step": 3726 }, { "epoch": 36.139759036144575, "grad_norm": 2.9306297302246094, "learning_rate": 8.190776699029127e-05, "loss": 0.1961, "step": 3727 }, { "epoch": 36.149397590361446, "grad_norm": 13.099404335021973, "learning_rate": 8.190291262135922e-05, "loss": 0.4843, "step": 3728 }, { "epoch": 36.159036144578316, "grad_norm": 32.0052375793457, "learning_rate": 8.18980582524272e-05, "loss": 0.4324, "step": 3729 }, { "epoch": 36.16867469879518, "grad_norm": 9.793212890625, "learning_rate": 8.189320388349515e-05, "loss": 0.2321, "step": 3730 }, { "epoch": 36.17831325301205, "grad_norm": 33.85744094848633, "learning_rate": 8.188834951456311e-05, "loss": 0.7894, "step": 3731 }, { "epoch": 36.18795180722891, "grad_norm": 5.57317590713501, "learning_rate": 8.188349514563107e-05, "loss": 0.4931, "step": 3732 }, { "epoch": 36.19759036144578, "grad_norm": 4.314291000366211, "learning_rate": 8.187864077669904e-05, "loss": 0.1826, "step": 3733 }, { "epoch": 36.207228915662654, "grad_norm": 15.276759147644043, "learning_rate": 8.1873786407767e-05, "loss": 0.3231, "step": 3734 }, { "epoch": 36.21686746987952, "grad_norm": 6.622117042541504, "learning_rate": 8.186893203883496e-05, "loss": 0.2984, "step": 3735 }, { "epoch": 36.22650602409639, "grad_norm": 6.563251495361328, "learning_rate": 8.186407766990291e-05, "loss": 0.2579, "step": 3736 }, { "epoch": 36.23614457831325, "grad_norm": 9.688237190246582, "learning_rate": 8.185922330097087e-05, "loss": 0.2383, "step": 3737 }, { "epoch": 36.24578313253012, "grad_norm": 12.92557144165039, "learning_rate": 8.185436893203884e-05, "loss": 0.6746, "step": 3738 }, { "epoch": 36.255421686746985, "grad_norm": 8.211503028869629, "learning_rate": 8.18495145631068e-05, "loss": 0.448, "step": 3739 }, { "epoch": 36.265060240963855, "grad_norm": 11.95729923248291, "learning_rate": 8.184466019417476e-05, "loss": 0.2264, "step": 3740 }, { "epoch": 36.274698795180726, "grad_norm": 13.431320190429688, "learning_rate": 8.183980582524272e-05, "loss": 0.5068, "step": 3741 }, { "epoch": 36.28433734939759, "grad_norm": 9.196406364440918, "learning_rate": 8.183495145631069e-05, "loss": 0.3037, "step": 3742 }, { "epoch": 36.29397590361446, "grad_norm": 4.091110706329346, "learning_rate": 8.183009708737864e-05, "loss": 0.2046, "step": 3743 }, { "epoch": 36.30361445783132, "grad_norm": 4.904778480529785, "learning_rate": 8.18252427184466e-05, "loss": 0.5436, "step": 3744 }, { "epoch": 36.31325301204819, "grad_norm": 12.532116889953613, "learning_rate": 8.182038834951456e-05, "loss": 0.3281, "step": 3745 }, { "epoch": 36.32289156626506, "grad_norm": 19.5455265045166, "learning_rate": 8.181553398058253e-05, "loss": 0.339, "step": 3746 }, { "epoch": 36.33253012048193, "grad_norm": 14.0618314743042, "learning_rate": 8.181067961165049e-05, "loss": 0.2856, "step": 3747 }, { "epoch": 36.3421686746988, "grad_norm": 4.58949089050293, "learning_rate": 8.180582524271845e-05, "loss": 0.2194, "step": 3748 }, { "epoch": 36.35180722891566, "grad_norm": 12.2847261428833, "learning_rate": 8.18009708737864e-05, "loss": 0.4005, "step": 3749 }, { "epoch": 36.36144578313253, "grad_norm": 11.49608325958252, "learning_rate": 8.179611650485438e-05, "loss": 0.3275, "step": 3750 }, { "epoch": 36.371084337349394, "grad_norm": 4.630154609680176, "learning_rate": 8.179126213592233e-05, "loss": 0.2312, "step": 3751 }, { "epoch": 36.380722891566265, "grad_norm": 10.700677871704102, "learning_rate": 8.178640776699029e-05, "loss": 0.2696, "step": 3752 }, { "epoch": 36.390361445783135, "grad_norm": 10.144579887390137, "learning_rate": 8.178155339805825e-05, "loss": 0.4448, "step": 3753 }, { "epoch": 36.4, "grad_norm": 24.313671112060547, "learning_rate": 8.177669902912621e-05, "loss": 0.2701, "step": 3754 }, { "epoch": 36.40963855421687, "grad_norm": 19.6468505859375, "learning_rate": 8.177184466019418e-05, "loss": 0.3729, "step": 3755 }, { "epoch": 36.41927710843373, "grad_norm": 16.56260871887207, "learning_rate": 8.176699029126214e-05, "loss": 0.2514, "step": 3756 }, { "epoch": 36.4289156626506, "grad_norm": 8.296265602111816, "learning_rate": 8.17621359223301e-05, "loss": 0.3277, "step": 3757 }, { "epoch": 36.43855421686747, "grad_norm": 10.962532997131348, "learning_rate": 8.175728155339805e-05, "loss": 0.3542, "step": 3758 }, { "epoch": 36.44819277108434, "grad_norm": 17.008386611938477, "learning_rate": 8.175242718446602e-05, "loss": 0.224, "step": 3759 }, { "epoch": 36.45783132530121, "grad_norm": 19.602535247802734, "learning_rate": 8.174757281553398e-05, "loss": 0.3251, "step": 3760 }, { "epoch": 36.46746987951807, "grad_norm": 18.187488555908203, "learning_rate": 8.174271844660194e-05, "loss": 0.4401, "step": 3761 }, { "epoch": 36.47710843373494, "grad_norm": 1.7896546125411987, "learning_rate": 8.173786407766991e-05, "loss": 0.1169, "step": 3762 }, { "epoch": 36.486746987951804, "grad_norm": 7.512508392333984, "learning_rate": 8.173300970873787e-05, "loss": 0.228, "step": 3763 }, { "epoch": 36.496385542168674, "grad_norm": 19.43288230895996, "learning_rate": 8.172815533980584e-05, "loss": 0.3397, "step": 3764 }, { "epoch": 36.506024096385545, "grad_norm": 3.79841685295105, "learning_rate": 8.17233009708738e-05, "loss": 0.0903, "step": 3765 }, { "epoch": 36.51566265060241, "grad_norm": 5.243253231048584, "learning_rate": 8.171844660194175e-05, "loss": 0.408, "step": 3766 }, { "epoch": 36.52530120481928, "grad_norm": 6.752379894256592, "learning_rate": 8.171359223300971e-05, "loss": 0.3397, "step": 3767 }, { "epoch": 36.53493975903614, "grad_norm": 11.285721778869629, "learning_rate": 8.170873786407768e-05, "loss": 0.2595, "step": 3768 }, { "epoch": 36.54457831325301, "grad_norm": 11.864870071411133, "learning_rate": 8.170388349514564e-05, "loss": 0.3192, "step": 3769 }, { "epoch": 36.55421686746988, "grad_norm": 5.393152713775635, "learning_rate": 8.16990291262136e-05, "loss": 0.372, "step": 3770 }, { "epoch": 36.563855421686746, "grad_norm": 4.6825995445251465, "learning_rate": 8.169417475728156e-05, "loss": 0.2739, "step": 3771 }, { "epoch": 36.57349397590362, "grad_norm": 18.838598251342773, "learning_rate": 8.168932038834953e-05, "loss": 0.7504, "step": 3772 }, { "epoch": 36.58313253012048, "grad_norm": 7.639337062835693, "learning_rate": 8.168446601941749e-05, "loss": 0.3222, "step": 3773 }, { "epoch": 36.59277108433735, "grad_norm": 14.507381439208984, "learning_rate": 8.167961165048544e-05, "loss": 0.3581, "step": 3774 }, { "epoch": 36.602409638554214, "grad_norm": 14.539176940917969, "learning_rate": 8.16747572815534e-05, "loss": 0.5108, "step": 3775 }, { "epoch": 36.612048192771084, "grad_norm": 13.171611785888672, "learning_rate": 8.166990291262136e-05, "loss": 0.609, "step": 3776 }, { "epoch": 36.621686746987955, "grad_norm": 21.658416748046875, "learning_rate": 8.166504854368933e-05, "loss": 0.3026, "step": 3777 }, { "epoch": 36.63132530120482, "grad_norm": 6.364629745483398, "learning_rate": 8.166019417475729e-05, "loss": 0.2764, "step": 3778 }, { "epoch": 36.64096385542169, "grad_norm": 12.024164199829102, "learning_rate": 8.165533980582525e-05, "loss": 0.4719, "step": 3779 }, { "epoch": 36.65060240963855, "grad_norm": 6.873937129974365, "learning_rate": 8.16504854368932e-05, "loss": 0.1311, "step": 3780 }, { "epoch": 36.66024096385542, "grad_norm": 21.63239860534668, "learning_rate": 8.164563106796117e-05, "loss": 0.248, "step": 3781 }, { "epoch": 36.66987951807229, "grad_norm": 3.9394805431365967, "learning_rate": 8.164077669902913e-05, "loss": 0.1698, "step": 3782 }, { "epoch": 36.679518072289156, "grad_norm": 7.559859752655029, "learning_rate": 8.163592233009709e-05, "loss": 0.3255, "step": 3783 }, { "epoch": 36.689156626506026, "grad_norm": 4.579981327056885, "learning_rate": 8.163106796116505e-05, "loss": 0.4181, "step": 3784 }, { "epoch": 36.69879518072289, "grad_norm": 29.415666580200195, "learning_rate": 8.162621359223302e-05, "loss": 0.4837, "step": 3785 }, { "epoch": 36.70843373493976, "grad_norm": 12.544160842895508, "learning_rate": 8.162135922330098e-05, "loss": 0.271, "step": 3786 }, { "epoch": 36.71807228915662, "grad_norm": 14.876622200012207, "learning_rate": 8.161650485436893e-05, "loss": 0.3728, "step": 3787 }, { "epoch": 36.727710843373494, "grad_norm": 4.769433975219727, "learning_rate": 8.161165048543689e-05, "loss": 0.3158, "step": 3788 }, { "epoch": 36.737349397590364, "grad_norm": 14.918008804321289, "learning_rate": 8.160679611650486e-05, "loss": 0.4538, "step": 3789 }, { "epoch": 36.74698795180723, "grad_norm": 11.511811256408691, "learning_rate": 8.160194174757282e-05, "loss": 0.3868, "step": 3790 }, { "epoch": 36.7566265060241, "grad_norm": 9.56380558013916, "learning_rate": 8.159708737864078e-05, "loss": 0.2551, "step": 3791 }, { "epoch": 36.76626506024096, "grad_norm": 31.416833877563477, "learning_rate": 8.159223300970874e-05, "loss": 0.2176, "step": 3792 }, { "epoch": 36.77590361445783, "grad_norm": 8.093338966369629, "learning_rate": 8.15873786407767e-05, "loss": 0.3234, "step": 3793 }, { "epoch": 36.7855421686747, "grad_norm": 13.867096900939941, "learning_rate": 8.158252427184467e-05, "loss": 0.1913, "step": 3794 }, { "epoch": 36.795180722891565, "grad_norm": 6.378793716430664, "learning_rate": 8.157766990291262e-05, "loss": 0.233, "step": 3795 }, { "epoch": 36.804819277108436, "grad_norm": 15.184615135192871, "learning_rate": 8.157281553398058e-05, "loss": 0.2294, "step": 3796 }, { "epoch": 36.8144578313253, "grad_norm": 6.019163131713867, "learning_rate": 8.156796116504854e-05, "loss": 0.3482, "step": 3797 }, { "epoch": 36.82409638554217, "grad_norm": 2.105684995651245, "learning_rate": 8.156310679611651e-05, "loss": 0.2756, "step": 3798 }, { "epoch": 36.83373493975903, "grad_norm": 12.788158416748047, "learning_rate": 8.155825242718447e-05, "loss": 0.477, "step": 3799 }, { "epoch": 36.8433734939759, "grad_norm": 14.246427536010742, "learning_rate": 8.155339805825243e-05, "loss": 0.4106, "step": 3800 }, { "epoch": 36.853012048192774, "grad_norm": 76.9118423461914, "learning_rate": 8.154854368932038e-05, "loss": 0.1803, "step": 3801 }, { "epoch": 36.86265060240964, "grad_norm": 5.985220909118652, "learning_rate": 8.154368932038836e-05, "loss": 0.2729, "step": 3802 }, { "epoch": 36.87228915662651, "grad_norm": 5.579717636108398, "learning_rate": 8.153883495145631e-05, "loss": 0.4068, "step": 3803 }, { "epoch": 36.88192771084337, "grad_norm": 9.035822868347168, "learning_rate": 8.153398058252427e-05, "loss": 0.4903, "step": 3804 }, { "epoch": 36.89156626506024, "grad_norm": 14.879098892211914, "learning_rate": 8.152912621359223e-05, "loss": 0.4114, "step": 3805 }, { "epoch": 36.90120481927711, "grad_norm": 50.74674606323242, "learning_rate": 8.15242718446602e-05, "loss": 0.3095, "step": 3806 }, { "epoch": 36.910843373493975, "grad_norm": 33.40483474731445, "learning_rate": 8.151941747572816e-05, "loss": 0.5368, "step": 3807 }, { "epoch": 36.920481927710846, "grad_norm": 21.722787857055664, "learning_rate": 8.151456310679612e-05, "loss": 0.2417, "step": 3808 }, { "epoch": 36.93012048192771, "grad_norm": 10.666155815124512, "learning_rate": 8.150970873786407e-05, "loss": 0.5038, "step": 3809 }, { "epoch": 36.93975903614458, "grad_norm": 5.924186706542969, "learning_rate": 8.150485436893203e-05, "loss": 0.4015, "step": 3810 }, { "epoch": 36.94939759036144, "grad_norm": 18.525665283203125, "learning_rate": 8.15e-05, "loss": 0.4294, "step": 3811 }, { "epoch": 36.95903614457831, "grad_norm": 3.885098934173584, "learning_rate": 8.149514563106796e-05, "loss": 0.2112, "step": 3812 }, { "epoch": 36.96867469879518, "grad_norm": 12.040545463562012, "learning_rate": 8.149029126213593e-05, "loss": 0.312, "step": 3813 }, { "epoch": 36.97831325301205, "grad_norm": 10.903711318969727, "learning_rate": 8.148543689320389e-05, "loss": 0.4346, "step": 3814 }, { "epoch": 36.98795180722892, "grad_norm": 52.69401550292969, "learning_rate": 8.148058252427185e-05, "loss": 0.343, "step": 3815 }, { "epoch": 37.003614457831326, "grad_norm": 8.415332794189453, "learning_rate": 8.147572815533982e-05, "loss": 0.3344, "step": 3816 }, { "epoch": 37.013253012048196, "grad_norm": 44.99927520751953, "learning_rate": 8.147087378640778e-05, "loss": 0.4847, "step": 3817 }, { "epoch": 37.02289156626506, "grad_norm": 5.42719030380249, "learning_rate": 8.146601941747573e-05, "loss": 0.243, "step": 3818 }, { "epoch": 37.03253012048193, "grad_norm": 4.120513916015625, "learning_rate": 8.146116504854369e-05, "loss": 0.3535, "step": 3819 }, { "epoch": 37.04216867469879, "grad_norm": 15.924821853637695, "learning_rate": 8.145631067961166e-05, "loss": 0.5238, "step": 3820 }, { "epoch": 37.05180722891566, "grad_norm": 5.300693035125732, "learning_rate": 8.145145631067962e-05, "loss": 0.3103, "step": 3821 }, { "epoch": 37.06144578313253, "grad_norm": 26.921693801879883, "learning_rate": 8.144660194174758e-05, "loss": 0.2898, "step": 3822 }, { "epoch": 37.0710843373494, "grad_norm": 47.1968879699707, "learning_rate": 8.144174757281554e-05, "loss": 0.5042, "step": 3823 }, { "epoch": 37.08072289156627, "grad_norm": 20.213937759399414, "learning_rate": 8.143689320388351e-05, "loss": 0.5225, "step": 3824 }, { "epoch": 37.09036144578313, "grad_norm": 9.633237838745117, "learning_rate": 8.143203883495147e-05, "loss": 0.5083, "step": 3825 }, { "epoch": 37.1, "grad_norm": 8.698235511779785, "learning_rate": 8.142718446601942e-05, "loss": 0.3549, "step": 3826 }, { "epoch": 37.109638554216865, "grad_norm": 10.254057884216309, "learning_rate": 8.142233009708738e-05, "loss": 0.4741, "step": 3827 }, { "epoch": 37.119277108433735, "grad_norm": 13.359136581420898, "learning_rate": 8.141747572815535e-05, "loss": 0.2445, "step": 3828 }, { "epoch": 37.128915662650606, "grad_norm": 11.557134628295898, "learning_rate": 8.141262135922331e-05, "loss": 0.2803, "step": 3829 }, { "epoch": 37.13855421686747, "grad_norm": 3.7647106647491455, "learning_rate": 8.140776699029127e-05, "loss": 0.1685, "step": 3830 }, { "epoch": 37.14819277108434, "grad_norm": 16.883703231811523, "learning_rate": 8.140291262135923e-05, "loss": 0.1626, "step": 3831 }, { "epoch": 37.1578313253012, "grad_norm": 7.349201679229736, "learning_rate": 8.139805825242718e-05, "loss": 0.3495, "step": 3832 }, { "epoch": 37.16746987951807, "grad_norm": 4.075558185577393, "learning_rate": 8.139320388349515e-05, "loss": 0.4284, "step": 3833 }, { "epoch": 37.17710843373494, "grad_norm": 39.16014099121094, "learning_rate": 8.138834951456311e-05, "loss": 0.2527, "step": 3834 }, { "epoch": 37.18674698795181, "grad_norm": 51.4917106628418, "learning_rate": 8.138349514563107e-05, "loss": 0.3416, "step": 3835 }, { "epoch": 37.19638554216868, "grad_norm": 8.316917419433594, "learning_rate": 8.137864077669903e-05, "loss": 0.342, "step": 3836 }, { "epoch": 37.20602409638554, "grad_norm": 5.684871673583984, "learning_rate": 8.1373786407767e-05, "loss": 0.2293, "step": 3837 }, { "epoch": 37.21566265060241, "grad_norm": 18.86922264099121, "learning_rate": 8.136893203883496e-05, "loss": 0.3819, "step": 3838 }, { "epoch": 37.225301204819274, "grad_norm": 10.288963317871094, "learning_rate": 8.136407766990291e-05, "loss": 0.4579, "step": 3839 }, { "epoch": 37.234939759036145, "grad_norm": 5.185425281524658, "learning_rate": 8.135922330097087e-05, "loss": 0.1988, "step": 3840 }, { "epoch": 37.244578313253015, "grad_norm": 9.912493705749512, "learning_rate": 8.135436893203884e-05, "loss": 0.2611, "step": 3841 }, { "epoch": 37.25421686746988, "grad_norm": 6.748018264770508, "learning_rate": 8.13495145631068e-05, "loss": 0.1705, "step": 3842 }, { "epoch": 37.26385542168675, "grad_norm": 5.352945327758789, "learning_rate": 8.134466019417476e-05, "loss": 0.2301, "step": 3843 }, { "epoch": 37.27349397590361, "grad_norm": 13.213763236999512, "learning_rate": 8.133980582524272e-05, "loss": 0.5069, "step": 3844 }, { "epoch": 37.28313253012048, "grad_norm": 9.687968254089355, "learning_rate": 8.133495145631069e-05, "loss": 0.3741, "step": 3845 }, { "epoch": 37.292771084337346, "grad_norm": 6.757462024688721, "learning_rate": 8.133009708737865e-05, "loss": 0.3615, "step": 3846 }, { "epoch": 37.30240963855422, "grad_norm": 14.761463165283203, "learning_rate": 8.13252427184466e-05, "loss": 0.4125, "step": 3847 }, { "epoch": 37.31204819277109, "grad_norm": 25.298242568969727, "learning_rate": 8.132038834951456e-05, "loss": 0.6993, "step": 3848 }, { "epoch": 37.32168674698795, "grad_norm": 9.814125061035156, "learning_rate": 8.131553398058252e-05, "loss": 0.2729, "step": 3849 }, { "epoch": 37.33132530120482, "grad_norm": 10.421504974365234, "learning_rate": 8.131067961165049e-05, "loss": 0.2618, "step": 3850 }, { "epoch": 37.340963855421684, "grad_norm": 15.942606925964355, "learning_rate": 8.130582524271845e-05, "loss": 0.2692, "step": 3851 }, { "epoch": 37.350602409638554, "grad_norm": 40.85688781738281, "learning_rate": 8.13009708737864e-05, "loss": 0.3512, "step": 3852 }, { "epoch": 37.360240963855425, "grad_norm": 6.324808120727539, "learning_rate": 8.129611650485436e-05, "loss": 0.2953, "step": 3853 }, { "epoch": 37.36987951807229, "grad_norm": 5.931598663330078, "learning_rate": 8.129126213592234e-05, "loss": 0.2807, "step": 3854 }, { "epoch": 37.37951807228916, "grad_norm": 7.002747058868408, "learning_rate": 8.128640776699029e-05, "loss": 0.2419, "step": 3855 }, { "epoch": 37.38915662650602, "grad_norm": 19.149471282958984, "learning_rate": 8.128155339805825e-05, "loss": 0.1762, "step": 3856 }, { "epoch": 37.39879518072289, "grad_norm": 4.391636371612549, "learning_rate": 8.127669902912621e-05, "loss": 0.0919, "step": 3857 }, { "epoch": 37.408433734939756, "grad_norm": 9.583170890808105, "learning_rate": 8.127184466019418e-05, "loss": 0.4504, "step": 3858 }, { "epoch": 37.418072289156626, "grad_norm": 7.1490864753723145, "learning_rate": 8.126699029126214e-05, "loss": 0.3013, "step": 3859 }, { "epoch": 37.4277108433735, "grad_norm": 27.50499725341797, "learning_rate": 8.12621359223301e-05, "loss": 0.4011, "step": 3860 }, { "epoch": 37.43734939759036, "grad_norm": 5.801513195037842, "learning_rate": 8.125728155339805e-05, "loss": 0.258, "step": 3861 }, { "epoch": 37.44698795180723, "grad_norm": 26.931711196899414, "learning_rate": 8.125242718446602e-05, "loss": 0.4336, "step": 3862 }, { "epoch": 37.456626506024094, "grad_norm": 12.394509315490723, "learning_rate": 8.1247572815534e-05, "loss": 0.4056, "step": 3863 }, { "epoch": 37.466265060240964, "grad_norm": 8.967199325561523, "learning_rate": 8.124271844660195e-05, "loss": 0.3286, "step": 3864 }, { "epoch": 37.475903614457835, "grad_norm": 7.060080051422119, "learning_rate": 8.123786407766991e-05, "loss": 0.3261, "step": 3865 }, { "epoch": 37.4855421686747, "grad_norm": 31.864648818969727, "learning_rate": 8.123300970873787e-05, "loss": 0.2879, "step": 3866 }, { "epoch": 37.49518072289157, "grad_norm": 7.830379962921143, "learning_rate": 8.122815533980584e-05, "loss": 0.4209, "step": 3867 }, { "epoch": 37.50481927710843, "grad_norm": 17.550498962402344, "learning_rate": 8.12233009708738e-05, "loss": 0.4138, "step": 3868 }, { "epoch": 37.5144578313253, "grad_norm": 5.101518154144287, "learning_rate": 8.121844660194176e-05, "loss": 0.3109, "step": 3869 }, { "epoch": 37.524096385542165, "grad_norm": 5.2498555183410645, "learning_rate": 8.121359223300971e-05, "loss": 0.4182, "step": 3870 }, { "epoch": 37.533734939759036, "grad_norm": 17.454345703125, "learning_rate": 8.120873786407767e-05, "loss": 0.3941, "step": 3871 }, { "epoch": 37.543373493975906, "grad_norm": 7.935922622680664, "learning_rate": 8.120388349514564e-05, "loss": 0.2574, "step": 3872 }, { "epoch": 37.55301204819277, "grad_norm": 3.9284090995788574, "learning_rate": 8.11990291262136e-05, "loss": 0.1242, "step": 3873 }, { "epoch": 37.56265060240964, "grad_norm": 6.877845287322998, "learning_rate": 8.119417475728156e-05, "loss": 0.348, "step": 3874 }, { "epoch": 37.5722891566265, "grad_norm": 9.15022087097168, "learning_rate": 8.118932038834952e-05, "loss": 0.3234, "step": 3875 }, { "epoch": 37.581927710843374, "grad_norm": 11.9785795211792, "learning_rate": 8.118446601941749e-05, "loss": 0.3478, "step": 3876 }, { "epoch": 37.591566265060244, "grad_norm": 5.055971622467041, "learning_rate": 8.117961165048544e-05, "loss": 0.2664, "step": 3877 }, { "epoch": 37.60120481927711, "grad_norm": 19.572906494140625, "learning_rate": 8.11747572815534e-05, "loss": 0.4074, "step": 3878 }, { "epoch": 37.61084337349398, "grad_norm": 8.384222030639648, "learning_rate": 8.116990291262136e-05, "loss": 0.2127, "step": 3879 }, { "epoch": 37.62048192771084, "grad_norm": 17.292327880859375, "learning_rate": 8.116504854368933e-05, "loss": 0.3558, "step": 3880 }, { "epoch": 37.63012048192771, "grad_norm": 8.607071876525879, "learning_rate": 8.116019417475729e-05, "loss": 0.1581, "step": 3881 }, { "epoch": 37.639759036144575, "grad_norm": 9.184027671813965, "learning_rate": 8.115533980582525e-05, "loss": 0.4175, "step": 3882 }, { "epoch": 37.649397590361446, "grad_norm": 3.252326011657715, "learning_rate": 8.11504854368932e-05, "loss": 0.2084, "step": 3883 }, { "epoch": 37.659036144578316, "grad_norm": 5.447022914886475, "learning_rate": 8.114563106796118e-05, "loss": 0.2736, "step": 3884 }, { "epoch": 37.66867469879518, "grad_norm": 6.661081314086914, "learning_rate": 8.114077669902913e-05, "loss": 0.4706, "step": 3885 }, { "epoch": 37.67831325301205, "grad_norm": 5.475314140319824, "learning_rate": 8.113592233009709e-05, "loss": 0.3553, "step": 3886 }, { "epoch": 37.68795180722891, "grad_norm": 37.32486343383789, "learning_rate": 8.113106796116505e-05, "loss": 0.3588, "step": 3887 }, { "epoch": 37.69759036144578, "grad_norm": 12.98508358001709, "learning_rate": 8.112621359223301e-05, "loss": 0.2378, "step": 3888 }, { "epoch": 37.707228915662654, "grad_norm": 18.50927734375, "learning_rate": 8.112135922330098e-05, "loss": 0.3082, "step": 3889 }, { "epoch": 37.71686746987952, "grad_norm": 10.518655776977539, "learning_rate": 8.111650485436894e-05, "loss": 0.2141, "step": 3890 }, { "epoch": 37.72650602409639, "grad_norm": 8.712448120117188, "learning_rate": 8.11116504854369e-05, "loss": 0.2987, "step": 3891 }, { "epoch": 37.73614457831325, "grad_norm": 24.333648681640625, "learning_rate": 8.110679611650485e-05, "loss": 0.2339, "step": 3892 }, { "epoch": 37.74578313253012, "grad_norm": 9.336552619934082, "learning_rate": 8.110194174757282e-05, "loss": 0.4244, "step": 3893 }, { "epoch": 37.755421686746985, "grad_norm": 6.140659809112549, "learning_rate": 8.109708737864078e-05, "loss": 0.1764, "step": 3894 }, { "epoch": 37.765060240963855, "grad_norm": 23.414775848388672, "learning_rate": 8.109223300970874e-05, "loss": 0.1562, "step": 3895 }, { "epoch": 37.774698795180726, "grad_norm": 5.330310821533203, "learning_rate": 8.10873786407767e-05, "loss": 0.3004, "step": 3896 }, { "epoch": 37.78433734939759, "grad_norm": 6.624322891235352, "learning_rate": 8.108252427184467e-05, "loss": 0.4095, "step": 3897 }, { "epoch": 37.79397590361446, "grad_norm": 22.373767852783203, "learning_rate": 8.107766990291263e-05, "loss": 0.4154, "step": 3898 }, { "epoch": 37.80361445783132, "grad_norm": 7.758865833282471, "learning_rate": 8.107281553398058e-05, "loss": 0.3925, "step": 3899 }, { "epoch": 37.81325301204819, "grad_norm": 4.150254249572754, "learning_rate": 8.106796116504854e-05, "loss": 0.1296, "step": 3900 }, { "epoch": 37.82289156626506, "grad_norm": 9.071002960205078, "learning_rate": 8.106310679611651e-05, "loss": 0.4992, "step": 3901 }, { "epoch": 37.83253012048193, "grad_norm": 17.057144165039062, "learning_rate": 8.105825242718447e-05, "loss": 0.4771, "step": 3902 }, { "epoch": 37.8421686746988, "grad_norm": 37.0244255065918, "learning_rate": 8.105339805825243e-05, "loss": 0.5271, "step": 3903 }, { "epoch": 37.85180722891566, "grad_norm": 15.651301383972168, "learning_rate": 8.104854368932039e-05, "loss": 0.3846, "step": 3904 }, { "epoch": 37.86144578313253, "grad_norm": 6.662036895751953, "learning_rate": 8.104368932038834e-05, "loss": 0.4791, "step": 3905 }, { "epoch": 37.871084337349394, "grad_norm": 8.74703311920166, "learning_rate": 8.103883495145631e-05, "loss": 0.2789, "step": 3906 }, { "epoch": 37.880722891566265, "grad_norm": 22.52289581298828, "learning_rate": 8.103398058252427e-05, "loss": 0.395, "step": 3907 }, { "epoch": 37.890361445783135, "grad_norm": 15.006443977355957, "learning_rate": 8.102912621359223e-05, "loss": 0.3462, "step": 3908 }, { "epoch": 37.9, "grad_norm": 18.643461227416992, "learning_rate": 8.102427184466019e-05, "loss": 0.207, "step": 3909 }, { "epoch": 37.90963855421687, "grad_norm": 5.891425132751465, "learning_rate": 8.101941747572816e-05, "loss": 0.3673, "step": 3910 }, { "epoch": 37.91927710843373, "grad_norm": 9.132133483886719, "learning_rate": 8.101456310679612e-05, "loss": 0.2048, "step": 3911 }, { "epoch": 37.9289156626506, "grad_norm": 11.270477294921875, "learning_rate": 8.100970873786407e-05, "loss": 0.5568, "step": 3912 }, { "epoch": 37.93855421686747, "grad_norm": 9.238426208496094, "learning_rate": 8.100485436893205e-05, "loss": 0.3278, "step": 3913 }, { "epoch": 37.94819277108434, "grad_norm": 12.941641807556152, "learning_rate": 8.1e-05, "loss": 0.2816, "step": 3914 }, { "epoch": 37.95783132530121, "grad_norm": 16.09032440185547, "learning_rate": 8.099514563106798e-05, "loss": 0.4573, "step": 3915 }, { "epoch": 37.96746987951807, "grad_norm": 14.644217491149902, "learning_rate": 8.099029126213593e-05, "loss": 0.4867, "step": 3916 }, { "epoch": 37.97710843373494, "grad_norm": 8.886375427246094, "learning_rate": 8.098543689320389e-05, "loss": 0.2145, "step": 3917 }, { "epoch": 37.986746987951804, "grad_norm": 8.041826248168945, "learning_rate": 8.098058252427185e-05, "loss": 0.5593, "step": 3918 }, { "epoch": 38.00240963855422, "grad_norm": 7.62418270111084, "learning_rate": 8.097572815533982e-05, "loss": 0.3597, "step": 3919 }, { "epoch": 38.01204819277108, "grad_norm": 21.13396453857422, "learning_rate": 8.097087378640778e-05, "loss": 0.2468, "step": 3920 }, { "epoch": 38.02168674698795, "grad_norm": 7.972153186798096, "learning_rate": 8.096601941747574e-05, "loss": 0.4146, "step": 3921 }, { "epoch": 38.03132530120482, "grad_norm": 4.108306884765625, "learning_rate": 8.096116504854369e-05, "loss": 0.2632, "step": 3922 }, { "epoch": 38.04096385542169, "grad_norm": 13.150860786437988, "learning_rate": 8.095631067961166e-05, "loss": 0.3399, "step": 3923 }, { "epoch": 38.05060240963856, "grad_norm": 9.863807678222656, "learning_rate": 8.095145631067962e-05, "loss": 0.3414, "step": 3924 }, { "epoch": 38.06024096385542, "grad_norm": 6.725522994995117, "learning_rate": 8.094660194174758e-05, "loss": 0.4718, "step": 3925 }, { "epoch": 38.06987951807229, "grad_norm": 12.942422866821289, "learning_rate": 8.094174757281554e-05, "loss": 0.2781, "step": 3926 }, { "epoch": 38.079518072289154, "grad_norm": 5.710423946380615, "learning_rate": 8.09368932038835e-05, "loss": 0.2559, "step": 3927 }, { "epoch": 38.089156626506025, "grad_norm": 11.344510078430176, "learning_rate": 8.093203883495147e-05, "loss": 0.2826, "step": 3928 }, { "epoch": 38.09879518072289, "grad_norm": 4.150822162628174, "learning_rate": 8.092718446601942e-05, "loss": 0.3415, "step": 3929 }, { "epoch": 38.10843373493976, "grad_norm": 8.09303092956543, "learning_rate": 8.092233009708738e-05, "loss": 0.4391, "step": 3930 }, { "epoch": 38.11807228915663, "grad_norm": 13.146254539489746, "learning_rate": 8.091747572815534e-05, "loss": 0.181, "step": 3931 }, { "epoch": 38.12771084337349, "grad_norm": 4.507794380187988, "learning_rate": 8.091262135922331e-05, "loss": 0.3125, "step": 3932 }, { "epoch": 38.13734939759036, "grad_norm": 7.166329860687256, "learning_rate": 8.090776699029127e-05, "loss": 0.292, "step": 3933 }, { "epoch": 38.146987951807226, "grad_norm": 5.279947280883789, "learning_rate": 8.090291262135923e-05, "loss": 0.446, "step": 3934 }, { "epoch": 38.1566265060241, "grad_norm": 3.061060667037964, "learning_rate": 8.089805825242718e-05, "loss": 0.2407, "step": 3935 }, { "epoch": 38.16626506024097, "grad_norm": 4.999633312225342, "learning_rate": 8.089320388349516e-05, "loss": 0.3125, "step": 3936 }, { "epoch": 38.17590361445783, "grad_norm": 18.110027313232422, "learning_rate": 8.088834951456311e-05, "loss": 0.1888, "step": 3937 }, { "epoch": 38.1855421686747, "grad_norm": 10.521879196166992, "learning_rate": 8.088349514563107e-05, "loss": 0.356, "step": 3938 }, { "epoch": 38.195180722891564, "grad_norm": 6.466208457946777, "learning_rate": 8.087864077669903e-05, "loss": 0.2824, "step": 3939 }, { "epoch": 38.204819277108435, "grad_norm": 5.672955513000488, "learning_rate": 8.0873786407767e-05, "loss": 0.3016, "step": 3940 }, { "epoch": 38.2144578313253, "grad_norm": 11.274888038635254, "learning_rate": 8.086893203883496e-05, "loss": 0.266, "step": 3941 }, { "epoch": 38.22409638554217, "grad_norm": 13.446539878845215, "learning_rate": 8.086407766990292e-05, "loss": 0.3376, "step": 3942 }, { "epoch": 38.23373493975904, "grad_norm": 6.704113006591797, "learning_rate": 8.085922330097087e-05, "loss": 0.3448, "step": 3943 }, { "epoch": 38.2433734939759, "grad_norm": 5.719286918640137, "learning_rate": 8.085436893203883e-05, "loss": 0.2465, "step": 3944 }, { "epoch": 38.25301204819277, "grad_norm": 24.486997604370117, "learning_rate": 8.08495145631068e-05, "loss": 0.342, "step": 3945 }, { "epoch": 38.262650602409636, "grad_norm": 7.481293678283691, "learning_rate": 8.084466019417476e-05, "loss": 0.1969, "step": 3946 }, { "epoch": 38.272289156626506, "grad_norm": 3.553082227706909, "learning_rate": 8.083980582524272e-05, "loss": 0.1442, "step": 3947 }, { "epoch": 38.28192771084338, "grad_norm": 11.239554405212402, "learning_rate": 8.083495145631068e-05, "loss": 0.3784, "step": 3948 }, { "epoch": 38.29156626506024, "grad_norm": 4.893836975097656, "learning_rate": 8.083009708737865e-05, "loss": 0.316, "step": 3949 }, { "epoch": 38.30120481927711, "grad_norm": 13.56302261352539, "learning_rate": 8.08252427184466e-05, "loss": 0.4457, "step": 3950 }, { "epoch": 38.310843373493974, "grad_norm": 5.371337413787842, "learning_rate": 8.082038834951456e-05, "loss": 0.3949, "step": 3951 }, { "epoch": 38.320481927710844, "grad_norm": 4.506810665130615, "learning_rate": 8.081553398058252e-05, "loss": 0.245, "step": 3952 }, { "epoch": 38.33012048192771, "grad_norm": 6.346377849578857, "learning_rate": 8.081067961165049e-05, "loss": 0.1656, "step": 3953 }, { "epoch": 38.33975903614458, "grad_norm": 15.949641227722168, "learning_rate": 8.080582524271845e-05, "loss": 0.4859, "step": 3954 }, { "epoch": 38.34939759036145, "grad_norm": 9.77806282043457, "learning_rate": 8.080097087378641e-05, "loss": 0.2319, "step": 3955 }, { "epoch": 38.35903614457831, "grad_norm": 6.240684509277344, "learning_rate": 8.079611650485437e-05, "loss": 0.2986, "step": 3956 }, { "epoch": 38.36867469879518, "grad_norm": 10.721518516540527, "learning_rate": 8.079126213592234e-05, "loss": 0.4465, "step": 3957 }, { "epoch": 38.378313253012045, "grad_norm": 11.387398719787598, "learning_rate": 8.07864077669903e-05, "loss": 0.2814, "step": 3958 }, { "epoch": 38.387951807228916, "grad_norm": 4.880520820617676, "learning_rate": 8.078155339805825e-05, "loss": 0.4168, "step": 3959 }, { "epoch": 38.397590361445786, "grad_norm": 12.651737213134766, "learning_rate": 8.077669902912621e-05, "loss": 0.4312, "step": 3960 }, { "epoch": 38.40722891566265, "grad_norm": 10.835700988769531, "learning_rate": 8.077184466019417e-05, "loss": 0.4539, "step": 3961 }, { "epoch": 38.41686746987952, "grad_norm": 27.823179244995117, "learning_rate": 8.076699029126214e-05, "loss": 0.2965, "step": 3962 }, { "epoch": 38.42650602409638, "grad_norm": 3.4255006313323975, "learning_rate": 8.07621359223301e-05, "loss": 0.2431, "step": 3963 }, { "epoch": 38.436144578313254, "grad_norm": 17.393993377685547, "learning_rate": 8.075728155339807e-05, "loss": 0.3814, "step": 3964 }, { "epoch": 38.44578313253012, "grad_norm": 6.139851093292236, "learning_rate": 8.075242718446603e-05, "loss": 0.4217, "step": 3965 }, { "epoch": 38.45542168674699, "grad_norm": 3.982860803604126, "learning_rate": 8.074757281553398e-05, "loss": 0.2036, "step": 3966 }, { "epoch": 38.46506024096386, "grad_norm": 31.059064865112305, "learning_rate": 8.074271844660195e-05, "loss": 0.2855, "step": 3967 }, { "epoch": 38.47469879518072, "grad_norm": 3.9341537952423096, "learning_rate": 8.073786407766991e-05, "loss": 0.2431, "step": 3968 }, { "epoch": 38.48433734939759, "grad_norm": 8.847949028015137, "learning_rate": 8.073300970873787e-05, "loss": 0.2323, "step": 3969 }, { "epoch": 38.493975903614455, "grad_norm": 6.069029331207275, "learning_rate": 8.072815533980583e-05, "loss": 0.2598, "step": 3970 }, { "epoch": 38.503614457831326, "grad_norm": 9.903377532958984, "learning_rate": 8.07233009708738e-05, "loss": 0.4022, "step": 3971 }, { "epoch": 38.513253012048196, "grad_norm": 7.314094543457031, "learning_rate": 8.071844660194176e-05, "loss": 0.5566, "step": 3972 }, { "epoch": 38.52289156626506, "grad_norm": 7.448326587677002, "learning_rate": 8.071359223300971e-05, "loss": 0.3865, "step": 3973 }, { "epoch": 38.53253012048193, "grad_norm": 7.196884632110596, "learning_rate": 8.070873786407767e-05, "loss": 0.2946, "step": 3974 }, { "epoch": 38.54216867469879, "grad_norm": 6.12889289855957, "learning_rate": 8.070388349514564e-05, "loss": 0.3961, "step": 3975 }, { "epoch": 38.55180722891566, "grad_norm": 7.678639888763428, "learning_rate": 8.06990291262136e-05, "loss": 0.5359, "step": 3976 }, { "epoch": 38.56144578313253, "grad_norm": 14.29857063293457, "learning_rate": 8.069417475728156e-05, "loss": 0.5647, "step": 3977 }, { "epoch": 38.5710843373494, "grad_norm": 7.072974681854248, "learning_rate": 8.068932038834952e-05, "loss": 0.3814, "step": 3978 }, { "epoch": 38.58072289156627, "grad_norm": 4.421789169311523, "learning_rate": 8.068446601941749e-05, "loss": 0.4071, "step": 3979 }, { "epoch": 38.59036144578313, "grad_norm": 8.434439659118652, "learning_rate": 8.067961165048545e-05, "loss": 0.3894, "step": 3980 }, { "epoch": 38.6, "grad_norm": 5.739391326904297, "learning_rate": 8.06747572815534e-05, "loss": 0.2889, "step": 3981 }, { "epoch": 38.609638554216865, "grad_norm": 7.64754056930542, "learning_rate": 8.066990291262136e-05, "loss": 0.4089, "step": 3982 }, { "epoch": 38.619277108433735, "grad_norm": 5.940700531005859, "learning_rate": 8.066504854368932e-05, "loss": 0.4027, "step": 3983 }, { "epoch": 38.628915662650606, "grad_norm": 24.654550552368164, "learning_rate": 8.066019417475729e-05, "loss": 0.3457, "step": 3984 }, { "epoch": 38.63855421686747, "grad_norm": 7.616548538208008, "learning_rate": 8.065533980582525e-05, "loss": 0.3286, "step": 3985 }, { "epoch": 38.64819277108434, "grad_norm": 11.089035987854004, "learning_rate": 8.06504854368932e-05, "loss": 0.2282, "step": 3986 }, { "epoch": 38.6578313253012, "grad_norm": 13.277202606201172, "learning_rate": 8.064563106796116e-05, "loss": 0.4409, "step": 3987 }, { "epoch": 38.66746987951807, "grad_norm": 8.076203346252441, "learning_rate": 8.064077669902914e-05, "loss": 0.313, "step": 3988 }, { "epoch": 38.67710843373494, "grad_norm": 5.489664077758789, "learning_rate": 8.063592233009709e-05, "loss": 0.3002, "step": 3989 }, { "epoch": 38.68674698795181, "grad_norm": 7.292996406555176, "learning_rate": 8.063106796116505e-05, "loss": 0.403, "step": 3990 }, { "epoch": 38.69638554216868, "grad_norm": 29.901304244995117, "learning_rate": 8.062621359223301e-05, "loss": 0.2521, "step": 3991 }, { "epoch": 38.70602409638554, "grad_norm": 6.946818828582764, "learning_rate": 8.062135922330098e-05, "loss": 0.3229, "step": 3992 }, { "epoch": 38.71566265060241, "grad_norm": 9.189666748046875, "learning_rate": 8.061650485436894e-05, "loss": 0.5781, "step": 3993 }, { "epoch": 38.725301204819274, "grad_norm": 8.512144088745117, "learning_rate": 8.06116504854369e-05, "loss": 0.1991, "step": 3994 }, { "epoch": 38.734939759036145, "grad_norm": 12.731412887573242, "learning_rate": 8.060679611650485e-05, "loss": 0.1675, "step": 3995 }, { "epoch": 38.744578313253015, "grad_norm": 3.2871227264404297, "learning_rate": 8.060194174757282e-05, "loss": 0.1278, "step": 3996 }, { "epoch": 38.75421686746988, "grad_norm": 10.834494590759277, "learning_rate": 8.059708737864078e-05, "loss": 0.3249, "step": 3997 }, { "epoch": 38.76385542168675, "grad_norm": 3.8304367065429688, "learning_rate": 8.059223300970874e-05, "loss": 0.1849, "step": 3998 }, { "epoch": 38.77349397590361, "grad_norm": 6.116730213165283, "learning_rate": 8.05873786407767e-05, "loss": 0.3675, "step": 3999 }, { "epoch": 38.78313253012048, "grad_norm": 5.970306873321533, "learning_rate": 8.058252427184466e-05, "loss": 0.3647, "step": 4000 }, { "epoch": 38.792771084337346, "grad_norm": 3.7795701026916504, "learning_rate": 8.057766990291263e-05, "loss": 0.094, "step": 4001 }, { "epoch": 38.80240963855422, "grad_norm": 346.2492370605469, "learning_rate": 8.057281553398058e-05, "loss": 0.509, "step": 4002 }, { "epoch": 38.81204819277109, "grad_norm": 10.831836700439453, "learning_rate": 8.056796116504854e-05, "loss": 0.3364, "step": 4003 }, { "epoch": 38.82168674698795, "grad_norm": 4.828188419342041, "learning_rate": 8.05631067961165e-05, "loss": 0.2257, "step": 4004 }, { "epoch": 38.83132530120482, "grad_norm": 4.37513542175293, "learning_rate": 8.055825242718447e-05, "loss": 0.2701, "step": 4005 }, { "epoch": 38.840963855421684, "grad_norm": 13.021529197692871, "learning_rate": 8.055339805825243e-05, "loss": 0.2379, "step": 4006 }, { "epoch": 38.850602409638554, "grad_norm": 13.112868309020996, "learning_rate": 8.054854368932039e-05, "loss": 0.2278, "step": 4007 }, { "epoch": 38.860240963855425, "grad_norm": 5.133359432220459, "learning_rate": 8.054368932038834e-05, "loss": 0.2857, "step": 4008 }, { "epoch": 38.86987951807229, "grad_norm": 8.093008041381836, "learning_rate": 8.053883495145632e-05, "loss": 0.4572, "step": 4009 }, { "epoch": 38.87951807228916, "grad_norm": 10.617939949035645, "learning_rate": 8.053398058252427e-05, "loss": 0.3493, "step": 4010 }, { "epoch": 38.88915662650602, "grad_norm": 3.087597608566284, "learning_rate": 8.052912621359223e-05, "loss": 0.2002, "step": 4011 }, { "epoch": 38.89879518072289, "grad_norm": 7.8604021072387695, "learning_rate": 8.052427184466019e-05, "loss": 0.3912, "step": 4012 }, { "epoch": 38.908433734939756, "grad_norm": 11.092252731323242, "learning_rate": 8.051941747572816e-05, "loss": 0.2057, "step": 4013 }, { "epoch": 38.918072289156626, "grad_norm": 7.254083633422852, "learning_rate": 8.051456310679613e-05, "loss": 0.354, "step": 4014 }, { "epoch": 38.9277108433735, "grad_norm": 4.51936149597168, "learning_rate": 8.050970873786409e-05, "loss": 0.1864, "step": 4015 }, { "epoch": 38.93734939759036, "grad_norm": 8.151479721069336, "learning_rate": 8.050485436893205e-05, "loss": 0.3135, "step": 4016 }, { "epoch": 38.94698795180723, "grad_norm": 9.742079734802246, "learning_rate": 8.05e-05, "loss": 0.3724, "step": 4017 }, { "epoch": 38.956626506024094, "grad_norm": 10.242034912109375, "learning_rate": 8.049514563106798e-05, "loss": 0.2867, "step": 4018 }, { "epoch": 38.966265060240964, "grad_norm": 8.48585033416748, "learning_rate": 8.049029126213593e-05, "loss": 0.3114, "step": 4019 }, { "epoch": 38.975903614457835, "grad_norm": 12.933178901672363, "learning_rate": 8.048543689320389e-05, "loss": 0.2326, "step": 4020 }, { "epoch": 38.9855421686747, "grad_norm": 5.7837982177734375, "learning_rate": 8.048058252427185e-05, "loss": 0.1985, "step": 4021 }, { "epoch": 39.001204819277106, "grad_norm": 10.481948852539062, "learning_rate": 8.047572815533981e-05, "loss": 0.6237, "step": 4022 }, { "epoch": 39.01084337349398, "grad_norm": 7.791828632354736, "learning_rate": 8.047087378640778e-05, "loss": 0.4192, "step": 4023 }, { "epoch": 39.02048192771084, "grad_norm": 8.64647102355957, "learning_rate": 8.046601941747574e-05, "loss": 0.4311, "step": 4024 }, { "epoch": 39.03012048192771, "grad_norm": 6.441244125366211, "learning_rate": 8.04611650485437e-05, "loss": 0.3651, "step": 4025 }, { "epoch": 39.03975903614458, "grad_norm": 6.47193717956543, "learning_rate": 8.045631067961165e-05, "loss": 0.3208, "step": 4026 }, { "epoch": 39.049397590361444, "grad_norm": 8.246183395385742, "learning_rate": 8.045145631067962e-05, "loss": 0.3131, "step": 4027 }, { "epoch": 39.059036144578315, "grad_norm": 9.260546684265137, "learning_rate": 8.044660194174758e-05, "loss": 0.2987, "step": 4028 }, { "epoch": 39.06867469879518, "grad_norm": 4.706030368804932, "learning_rate": 8.044174757281554e-05, "loss": 0.2687, "step": 4029 }, { "epoch": 39.07831325301205, "grad_norm": 8.61475944519043, "learning_rate": 8.04368932038835e-05, "loss": 0.4341, "step": 4030 }, { "epoch": 39.08795180722892, "grad_norm": 12.946808815002441, "learning_rate": 8.043203883495147e-05, "loss": 0.2942, "step": 4031 }, { "epoch": 39.09759036144578, "grad_norm": 5.5704169273376465, "learning_rate": 8.042718446601943e-05, "loss": 0.3968, "step": 4032 }, { "epoch": 39.10722891566265, "grad_norm": 15.890684127807617, "learning_rate": 8.042233009708738e-05, "loss": 0.3093, "step": 4033 }, { "epoch": 39.116867469879516, "grad_norm": 3.638728141784668, "learning_rate": 8.041747572815534e-05, "loss": 0.1589, "step": 4034 }, { "epoch": 39.126506024096386, "grad_norm": 7.3190107345581055, "learning_rate": 8.04126213592233e-05, "loss": 0.2131, "step": 4035 }, { "epoch": 39.13614457831325, "grad_norm": 3.8857736587524414, "learning_rate": 8.040776699029127e-05, "loss": 0.3298, "step": 4036 }, { "epoch": 39.14578313253012, "grad_norm": 5.37900447845459, "learning_rate": 8.040291262135923e-05, "loss": 0.3512, "step": 4037 }, { "epoch": 39.15542168674699, "grad_norm": 12.759865760803223, "learning_rate": 8.039805825242719e-05, "loss": 0.1944, "step": 4038 }, { "epoch": 39.165060240963854, "grad_norm": 11.135438919067383, "learning_rate": 8.039320388349514e-05, "loss": 0.2779, "step": 4039 }, { "epoch": 39.174698795180724, "grad_norm": 9.52955150604248, "learning_rate": 8.038834951456311e-05, "loss": 0.5295, "step": 4040 }, { "epoch": 39.18433734939759, "grad_norm": 9.008644104003906, "learning_rate": 8.038349514563107e-05, "loss": 0.2891, "step": 4041 }, { "epoch": 39.19397590361446, "grad_norm": 10.141979217529297, "learning_rate": 8.037864077669903e-05, "loss": 0.4096, "step": 4042 }, { "epoch": 39.20361445783133, "grad_norm": 11.025555610656738, "learning_rate": 8.037378640776699e-05, "loss": 0.3537, "step": 4043 }, { "epoch": 39.21325301204819, "grad_norm": 8.728321075439453, "learning_rate": 8.036893203883496e-05, "loss": 0.358, "step": 4044 }, { "epoch": 39.22289156626506, "grad_norm": 20.033342361450195, "learning_rate": 8.036407766990292e-05, "loss": 0.2902, "step": 4045 }, { "epoch": 39.232530120481925, "grad_norm": 6.531880855560303, "learning_rate": 8.035922330097087e-05, "loss": 0.3204, "step": 4046 }, { "epoch": 39.242168674698796, "grad_norm": 21.578550338745117, "learning_rate": 8.035436893203883e-05, "loss": 0.2911, "step": 4047 }, { "epoch": 39.25180722891566, "grad_norm": 8.941112518310547, "learning_rate": 8.03495145631068e-05, "loss": 0.4161, "step": 4048 }, { "epoch": 39.26144578313253, "grad_norm": 6.526405334472656, "learning_rate": 8.034466019417476e-05, "loss": 0.2683, "step": 4049 }, { "epoch": 39.2710843373494, "grad_norm": 10.813644409179688, "learning_rate": 8.033980582524272e-05, "loss": 0.4006, "step": 4050 }, { "epoch": 39.28072289156626, "grad_norm": 21.883161544799805, "learning_rate": 8.033495145631068e-05, "loss": 0.2958, "step": 4051 }, { "epoch": 39.290361445783134, "grad_norm": 25.660255432128906, "learning_rate": 8.033009708737864e-05, "loss": 0.2855, "step": 4052 }, { "epoch": 39.3, "grad_norm": 8.173921585083008, "learning_rate": 8.03252427184466e-05, "loss": 0.4399, "step": 4053 }, { "epoch": 39.30963855421687, "grad_norm": 6.260617256164551, "learning_rate": 8.032038834951456e-05, "loss": 0.4168, "step": 4054 }, { "epoch": 39.31927710843374, "grad_norm": 9.688565254211426, "learning_rate": 8.031553398058252e-05, "loss": 0.3493, "step": 4055 }, { "epoch": 39.3289156626506, "grad_norm": 7.310895919799805, "learning_rate": 8.031067961165048e-05, "loss": 0.2829, "step": 4056 }, { "epoch": 39.33855421686747, "grad_norm": 8.586281776428223, "learning_rate": 8.030582524271845e-05, "loss": 0.4397, "step": 4057 }, { "epoch": 39.348192771084335, "grad_norm": 10.26656723022461, "learning_rate": 8.030097087378641e-05, "loss": 0.2883, "step": 4058 }, { "epoch": 39.357831325301206, "grad_norm": 8.039036750793457, "learning_rate": 8.029611650485437e-05, "loss": 0.4445, "step": 4059 }, { "epoch": 39.36746987951807, "grad_norm": 12.902581214904785, "learning_rate": 8.029126213592232e-05, "loss": 0.3559, "step": 4060 }, { "epoch": 39.37710843373494, "grad_norm": 3.608762741088867, "learning_rate": 8.02864077669903e-05, "loss": 0.2097, "step": 4061 }, { "epoch": 39.38674698795181, "grad_norm": 6.361729621887207, "learning_rate": 8.028155339805825e-05, "loss": 0.4275, "step": 4062 }, { "epoch": 39.39638554216867, "grad_norm": 3.273109197616577, "learning_rate": 8.027669902912621e-05, "loss": 0.2024, "step": 4063 }, { "epoch": 39.40602409638554, "grad_norm": 15.479167938232422, "learning_rate": 8.027184466019417e-05, "loss": 0.299, "step": 4064 }, { "epoch": 39.41566265060241, "grad_norm": 7.594846725463867, "learning_rate": 8.026699029126214e-05, "loss": 0.26, "step": 4065 }, { "epoch": 39.42530120481928, "grad_norm": 7.5038981437683105, "learning_rate": 8.026213592233011e-05, "loss": 0.3027, "step": 4066 }, { "epoch": 39.43493975903615, "grad_norm": 6.8467326164245605, "learning_rate": 8.025728155339807e-05, "loss": 0.2638, "step": 4067 }, { "epoch": 39.44457831325301, "grad_norm": 6.066018104553223, "learning_rate": 8.025242718446603e-05, "loss": 0.2736, "step": 4068 }, { "epoch": 39.45421686746988, "grad_norm": 6.860747814178467, "learning_rate": 8.024757281553398e-05, "loss": 0.2694, "step": 4069 }, { "epoch": 39.463855421686745, "grad_norm": 7.745129585266113, "learning_rate": 8.024271844660196e-05, "loss": 0.228, "step": 4070 }, { "epoch": 39.473493975903615, "grad_norm": 9.145703315734863, "learning_rate": 8.023786407766991e-05, "loss": 0.4593, "step": 4071 }, { "epoch": 39.48313253012048, "grad_norm": 5.115575313568115, "learning_rate": 8.023300970873787e-05, "loss": 0.2671, "step": 4072 }, { "epoch": 39.49277108433735, "grad_norm": 13.48025131225586, "learning_rate": 8.022815533980583e-05, "loss": 0.3965, "step": 4073 }, { "epoch": 39.50240963855422, "grad_norm": 7.1075968742370605, "learning_rate": 8.022330097087379e-05, "loss": 0.1481, "step": 4074 }, { "epoch": 39.51204819277108, "grad_norm": 6.68690824508667, "learning_rate": 8.021844660194176e-05, "loss": 0.418, "step": 4075 }, { "epoch": 39.52168674698795, "grad_norm": 5.960308074951172, "learning_rate": 8.021359223300972e-05, "loss": 0.2696, "step": 4076 }, { "epoch": 39.53132530120482, "grad_norm": 3.2916040420532227, "learning_rate": 8.020873786407767e-05, "loss": 0.2137, "step": 4077 }, { "epoch": 39.54096385542169, "grad_norm": 5.732546329498291, "learning_rate": 8.020388349514563e-05, "loss": 0.3392, "step": 4078 }, { "epoch": 39.55060240963856, "grad_norm": 5.114494323730469, "learning_rate": 8.01990291262136e-05, "loss": 0.2173, "step": 4079 }, { "epoch": 39.56024096385542, "grad_norm": 8.732937812805176, "learning_rate": 8.019417475728156e-05, "loss": 0.3078, "step": 4080 }, { "epoch": 39.56987951807229, "grad_norm": 5.804482936859131, "learning_rate": 8.018932038834952e-05, "loss": 0.2559, "step": 4081 }, { "epoch": 39.579518072289154, "grad_norm": 5.19765567779541, "learning_rate": 8.018446601941748e-05, "loss": 0.2543, "step": 4082 }, { "epoch": 39.589156626506025, "grad_norm": 22.505523681640625, "learning_rate": 8.017961165048545e-05, "loss": 0.2769, "step": 4083 }, { "epoch": 39.59879518072289, "grad_norm": 5.370514869689941, "learning_rate": 8.01747572815534e-05, "loss": 0.4434, "step": 4084 }, { "epoch": 39.60843373493976, "grad_norm": 7.919530391693115, "learning_rate": 8.016990291262136e-05, "loss": 0.381, "step": 4085 }, { "epoch": 39.61807228915663, "grad_norm": 29.90304946899414, "learning_rate": 8.016504854368932e-05, "loss": 0.6466, "step": 4086 }, { "epoch": 39.62771084337349, "grad_norm": 11.911151885986328, "learning_rate": 8.016019417475729e-05, "loss": 0.388, "step": 4087 }, { "epoch": 39.63734939759036, "grad_norm": 10.840731620788574, "learning_rate": 8.015533980582525e-05, "loss": 0.2269, "step": 4088 }, { "epoch": 39.646987951807226, "grad_norm": 6.5389323234558105, "learning_rate": 8.015048543689321e-05, "loss": 0.2631, "step": 4089 }, { "epoch": 39.6566265060241, "grad_norm": 10.183120727539062, "learning_rate": 8.014563106796117e-05, "loss": 0.3507, "step": 4090 }, { "epoch": 39.66626506024097, "grad_norm": 7.7039594650268555, "learning_rate": 8.014077669902912e-05, "loss": 0.3756, "step": 4091 }, { "epoch": 39.67590361445783, "grad_norm": 6.482124328613281, "learning_rate": 8.01359223300971e-05, "loss": 0.2582, "step": 4092 }, { "epoch": 39.6855421686747, "grad_norm": 6.649734020233154, "learning_rate": 8.013106796116505e-05, "loss": 0.3624, "step": 4093 }, { "epoch": 39.695180722891564, "grad_norm": 10.990633010864258, "learning_rate": 8.012621359223301e-05, "loss": 0.316, "step": 4094 }, { "epoch": 39.704819277108435, "grad_norm": 5.26398229598999, "learning_rate": 8.012135922330097e-05, "loss": 0.1793, "step": 4095 }, { "epoch": 39.7144578313253, "grad_norm": 5.075741767883301, "learning_rate": 8.011650485436894e-05, "loss": 0.4338, "step": 4096 }, { "epoch": 39.72409638554217, "grad_norm": 4.123610019683838, "learning_rate": 8.01116504854369e-05, "loss": 0.1735, "step": 4097 }, { "epoch": 39.73373493975904, "grad_norm": 9.886674880981445, "learning_rate": 8.010679611650485e-05, "loss": 0.3834, "step": 4098 }, { "epoch": 39.7433734939759, "grad_norm": 12.181526184082031, "learning_rate": 8.010194174757281e-05, "loss": 0.4476, "step": 4099 }, { "epoch": 39.75301204819277, "grad_norm": 5.001387596130371, "learning_rate": 8.009708737864078e-05, "loss": 0.3539, "step": 4100 }, { "epoch": 39.762650602409636, "grad_norm": 9.939635276794434, "learning_rate": 8.009223300970874e-05, "loss": 0.235, "step": 4101 }, { "epoch": 39.772289156626506, "grad_norm": 12.954181671142578, "learning_rate": 8.00873786407767e-05, "loss": 0.4061, "step": 4102 }, { "epoch": 39.78192771084338, "grad_norm": 6.677494049072266, "learning_rate": 8.008252427184466e-05, "loss": 0.3255, "step": 4103 }, { "epoch": 39.79156626506024, "grad_norm": 16.085229873657227, "learning_rate": 8.007766990291263e-05, "loss": 0.3619, "step": 4104 }, { "epoch": 39.80120481927711, "grad_norm": 10.206055641174316, "learning_rate": 8.007281553398059e-05, "loss": 0.3746, "step": 4105 }, { "epoch": 39.810843373493974, "grad_norm": 4.636328220367432, "learning_rate": 8.006796116504854e-05, "loss": 0.2152, "step": 4106 }, { "epoch": 39.820481927710844, "grad_norm": 6.400730133056641, "learning_rate": 8.00631067961165e-05, "loss": 0.2917, "step": 4107 }, { "epoch": 39.83012048192771, "grad_norm": 6.659059524536133, "learning_rate": 8.005825242718446e-05, "loss": 0.331, "step": 4108 }, { "epoch": 39.83975903614458, "grad_norm": 10.500040054321289, "learning_rate": 8.005339805825243e-05, "loss": 0.3378, "step": 4109 }, { "epoch": 39.84939759036145, "grad_norm": 5.915572643280029, "learning_rate": 8.004854368932039e-05, "loss": 0.3373, "step": 4110 }, { "epoch": 39.85903614457831, "grad_norm": 4.957621097564697, "learning_rate": 8.004368932038835e-05, "loss": 0.2462, "step": 4111 }, { "epoch": 39.86867469879518, "grad_norm": 6.929385185241699, "learning_rate": 8.00388349514563e-05, "loss": 0.5315, "step": 4112 }, { "epoch": 39.878313253012045, "grad_norm": 7.225522041320801, "learning_rate": 8.003398058252428e-05, "loss": 0.2209, "step": 4113 }, { "epoch": 39.887951807228916, "grad_norm": 8.79845142364502, "learning_rate": 8.002912621359223e-05, "loss": 0.3212, "step": 4114 }, { "epoch": 39.897590361445786, "grad_norm": 6.0581955909729, "learning_rate": 8.00242718446602e-05, "loss": 0.3591, "step": 4115 }, { "epoch": 39.90722891566265, "grad_norm": 6.3124775886535645, "learning_rate": 8.001941747572816e-05, "loss": 0.4047, "step": 4116 }, { "epoch": 39.91686746987952, "grad_norm": 10.112924575805664, "learning_rate": 8.001456310679612e-05, "loss": 0.3123, "step": 4117 }, { "epoch": 39.92650602409638, "grad_norm": 4.060420989990234, "learning_rate": 8.000970873786409e-05, "loss": 0.236, "step": 4118 }, { "epoch": 39.936144578313254, "grad_norm": 24.50531768798828, "learning_rate": 8.000485436893205e-05, "loss": 0.4106, "step": 4119 }, { "epoch": 39.94578313253012, "grad_norm": 10.527265548706055, "learning_rate": 8e-05, "loss": 0.3615, "step": 4120 }, { "epoch": 39.95542168674699, "grad_norm": 29.684038162231445, "learning_rate": 7.999514563106796e-05, "loss": 0.364, "step": 4121 }, { "epoch": 39.96506024096386, "grad_norm": 14.673189163208008, "learning_rate": 7.999029126213594e-05, "loss": 0.5183, "step": 4122 }, { "epoch": 39.97469879518072, "grad_norm": 8.854025840759277, "learning_rate": 7.99854368932039e-05, "loss": 0.4891, "step": 4123 }, { "epoch": 39.98433734939759, "grad_norm": 6.656195640563965, "learning_rate": 7.998058252427185e-05, "loss": 0.3794, "step": 4124 }, { "epoch": 39.993975903614455, "grad_norm": 6.848031520843506, "learning_rate": 7.997572815533981e-05, "loss": 0.3321, "step": 4125 }, { "epoch": 40.00963855421687, "grad_norm": 9.54970932006836, "learning_rate": 7.997087378640778e-05, "loss": 0.5553, "step": 4126 }, { "epoch": 40.019277108433734, "grad_norm": 9.19192123413086, "learning_rate": 7.996601941747574e-05, "loss": 0.2242, "step": 4127 }, { "epoch": 40.028915662650604, "grad_norm": 9.495926856994629, "learning_rate": 7.99611650485437e-05, "loss": 0.3409, "step": 4128 }, { "epoch": 40.03855421686747, "grad_norm": 5.317587375640869, "learning_rate": 7.995631067961165e-05, "loss": 0.2455, "step": 4129 }, { "epoch": 40.04819277108434, "grad_norm": 8.498479843139648, "learning_rate": 7.995145631067961e-05, "loss": 0.3058, "step": 4130 }, { "epoch": 40.0578313253012, "grad_norm": 2.7024364471435547, "learning_rate": 7.994660194174758e-05, "loss": 0.1072, "step": 4131 }, { "epoch": 40.06746987951807, "grad_norm": 6.68545389175415, "learning_rate": 7.994174757281554e-05, "loss": 0.2324, "step": 4132 }, { "epoch": 40.07710843373494, "grad_norm": 20.6758975982666, "learning_rate": 7.99368932038835e-05, "loss": 0.1095, "step": 4133 }, { "epoch": 40.086746987951805, "grad_norm": 9.56200885772705, "learning_rate": 7.993203883495146e-05, "loss": 0.4368, "step": 4134 }, { "epoch": 40.096385542168676, "grad_norm": 14.13261890411377, "learning_rate": 7.992718446601943e-05, "loss": 0.4041, "step": 4135 }, { "epoch": 40.10602409638554, "grad_norm": 5.622531890869141, "learning_rate": 7.992233009708738e-05, "loss": 0.3563, "step": 4136 }, { "epoch": 40.11566265060241, "grad_norm": 13.296247482299805, "learning_rate": 7.991747572815534e-05, "loss": 0.2382, "step": 4137 }, { "epoch": 40.12530120481928, "grad_norm": 11.465093612670898, "learning_rate": 7.99126213592233e-05, "loss": 0.2774, "step": 4138 }, { "epoch": 40.13493975903614, "grad_norm": 20.511083602905273, "learning_rate": 7.990776699029127e-05, "loss": 0.3478, "step": 4139 }, { "epoch": 40.144578313253014, "grad_norm": 5.4416680335998535, "learning_rate": 7.990291262135923e-05, "loss": 0.1192, "step": 4140 }, { "epoch": 40.15421686746988, "grad_norm": 15.457120895385742, "learning_rate": 7.989805825242719e-05, "loss": 0.3278, "step": 4141 }, { "epoch": 40.16385542168675, "grad_norm": 16.102581024169922, "learning_rate": 7.989320388349514e-05, "loss": 0.3283, "step": 4142 }, { "epoch": 40.17349397590361, "grad_norm": 11.864717483520508, "learning_rate": 7.988834951456312e-05, "loss": 0.5119, "step": 4143 }, { "epoch": 40.18313253012048, "grad_norm": 25.16305923461914, "learning_rate": 7.988349514563107e-05, "loss": 0.507, "step": 4144 }, { "epoch": 40.19277108433735, "grad_norm": 7.946400165557861, "learning_rate": 7.987864077669903e-05, "loss": 0.1812, "step": 4145 }, { "epoch": 40.202409638554215, "grad_norm": 17.494169235229492, "learning_rate": 7.987378640776699e-05, "loss": 0.574, "step": 4146 }, { "epoch": 40.212048192771086, "grad_norm": 8.53831672668457, "learning_rate": 7.986893203883495e-05, "loss": 0.2863, "step": 4147 }, { "epoch": 40.22168674698795, "grad_norm": 4.083259105682373, "learning_rate": 7.986407766990292e-05, "loss": 0.1214, "step": 4148 }, { "epoch": 40.23132530120482, "grad_norm": 5.100589275360107, "learning_rate": 7.985922330097088e-05, "loss": 0.1957, "step": 4149 }, { "epoch": 40.24096385542169, "grad_norm": 11.03868579864502, "learning_rate": 7.985436893203883e-05, "loss": 0.2224, "step": 4150 }, { "epoch": 40.25060240963855, "grad_norm": 9.881985664367676, "learning_rate": 7.984951456310679e-05, "loss": 0.2541, "step": 4151 }, { "epoch": 40.26024096385542, "grad_norm": 11.189480781555176, "learning_rate": 7.984466019417476e-05, "loss": 0.6283, "step": 4152 }, { "epoch": 40.26987951807229, "grad_norm": 12.873272895812988, "learning_rate": 7.983980582524272e-05, "loss": 0.4208, "step": 4153 }, { "epoch": 40.27951807228916, "grad_norm": 10.403754234313965, "learning_rate": 7.983495145631068e-05, "loss": 0.4431, "step": 4154 }, { "epoch": 40.28915662650602, "grad_norm": 8.211030960083008, "learning_rate": 7.983009708737864e-05, "loss": 0.5307, "step": 4155 }, { "epoch": 40.29879518072289, "grad_norm": 10.01301097869873, "learning_rate": 7.982524271844661e-05, "loss": 0.5553, "step": 4156 }, { "epoch": 40.30843373493976, "grad_norm": 9.947439193725586, "learning_rate": 7.982038834951457e-05, "loss": 0.4312, "step": 4157 }, { "epoch": 40.318072289156625, "grad_norm": 5.156205177307129, "learning_rate": 7.981553398058252e-05, "loss": 0.1997, "step": 4158 }, { "epoch": 40.327710843373495, "grad_norm": 8.50404167175293, "learning_rate": 7.981067961165048e-05, "loss": 0.3242, "step": 4159 }, { "epoch": 40.33734939759036, "grad_norm": 14.337474822998047, "learning_rate": 7.980582524271845e-05, "loss": 0.323, "step": 4160 }, { "epoch": 40.34698795180723, "grad_norm": 2.3395397663116455, "learning_rate": 7.980097087378641e-05, "loss": 0.0883, "step": 4161 }, { "epoch": 40.3566265060241, "grad_norm": 22.946989059448242, "learning_rate": 7.979611650485437e-05, "loss": 0.3971, "step": 4162 }, { "epoch": 40.36626506024096, "grad_norm": 8.104813575744629, "learning_rate": 7.979126213592233e-05, "loss": 0.3214, "step": 4163 }, { "epoch": 40.37590361445783, "grad_norm": 7.97607421875, "learning_rate": 7.978640776699028e-05, "loss": 0.2447, "step": 4164 }, { "epoch": 40.3855421686747, "grad_norm": 7.9766998291015625, "learning_rate": 7.978155339805825e-05, "loss": 0.317, "step": 4165 }, { "epoch": 40.39518072289157, "grad_norm": 7.797740459442139, "learning_rate": 7.977669902912623e-05, "loss": 0.2977, "step": 4166 }, { "epoch": 40.40481927710843, "grad_norm": 7.681482791900635, "learning_rate": 7.977184466019418e-05, "loss": 0.1388, "step": 4167 }, { "epoch": 40.4144578313253, "grad_norm": 4.631894111633301, "learning_rate": 7.976699029126214e-05, "loss": 0.2493, "step": 4168 }, { "epoch": 40.42409638554217, "grad_norm": 6.1631669998168945, "learning_rate": 7.97621359223301e-05, "loss": 0.362, "step": 4169 }, { "epoch": 40.433734939759034, "grad_norm": 8.877609252929688, "learning_rate": 7.975728155339807e-05, "loss": 0.2514, "step": 4170 }, { "epoch": 40.443373493975905, "grad_norm": 9.7259521484375, "learning_rate": 7.975242718446603e-05, "loss": 0.3407, "step": 4171 }, { "epoch": 40.45301204819277, "grad_norm": 9.544964790344238, "learning_rate": 7.974757281553399e-05, "loss": 0.3603, "step": 4172 }, { "epoch": 40.46265060240964, "grad_norm": 5.985980987548828, "learning_rate": 7.974271844660194e-05, "loss": 0.3763, "step": 4173 }, { "epoch": 40.47228915662651, "grad_norm": 9.371065139770508, "learning_rate": 7.973786407766992e-05, "loss": 0.3181, "step": 4174 }, { "epoch": 40.48192771084337, "grad_norm": 8.269669532775879, "learning_rate": 7.973300970873787e-05, "loss": 0.3227, "step": 4175 }, { "epoch": 40.49156626506024, "grad_norm": 7.792767524719238, "learning_rate": 7.972815533980583e-05, "loss": 0.4553, "step": 4176 }, { "epoch": 40.501204819277106, "grad_norm": 6.849043369293213, "learning_rate": 7.972330097087379e-05, "loss": 0.2682, "step": 4177 }, { "epoch": 40.51084337349398, "grad_norm": 6.607674598693848, "learning_rate": 7.971844660194176e-05, "loss": 0.1848, "step": 4178 }, { "epoch": 40.52048192771084, "grad_norm": 9.397296905517578, "learning_rate": 7.971359223300972e-05, "loss": 0.2703, "step": 4179 }, { "epoch": 40.53012048192771, "grad_norm": 7.600361347198486, "learning_rate": 7.970873786407768e-05, "loss": 0.3121, "step": 4180 }, { "epoch": 40.53975903614458, "grad_norm": 9.062417030334473, "learning_rate": 7.970388349514563e-05, "loss": 0.4241, "step": 4181 }, { "epoch": 40.549397590361444, "grad_norm": 8.254925727844238, "learning_rate": 7.96990291262136e-05, "loss": 0.378, "step": 4182 }, { "epoch": 40.559036144578315, "grad_norm": 4.672442436218262, "learning_rate": 7.969417475728156e-05, "loss": 0.2836, "step": 4183 }, { "epoch": 40.56867469879518, "grad_norm": 10.848471641540527, "learning_rate": 7.968932038834952e-05, "loss": 0.5032, "step": 4184 }, { "epoch": 40.57831325301205, "grad_norm": 24.394996643066406, "learning_rate": 7.968446601941748e-05, "loss": 0.5071, "step": 4185 }, { "epoch": 40.58795180722892, "grad_norm": 8.5025634765625, "learning_rate": 7.967961165048544e-05, "loss": 0.3566, "step": 4186 }, { "epoch": 40.59759036144578, "grad_norm": 11.782153129577637, "learning_rate": 7.96747572815534e-05, "loss": 0.1677, "step": 4187 }, { "epoch": 40.60722891566265, "grad_norm": 3.8303260803222656, "learning_rate": 7.966990291262136e-05, "loss": 0.2239, "step": 4188 }, { "epoch": 40.616867469879516, "grad_norm": 7.309577941894531, "learning_rate": 7.966504854368932e-05, "loss": 0.4375, "step": 4189 }, { "epoch": 40.626506024096386, "grad_norm": 23.849409103393555, "learning_rate": 7.966019417475728e-05, "loss": 0.357, "step": 4190 }, { "epoch": 40.63614457831325, "grad_norm": 6.621732711791992, "learning_rate": 7.965533980582525e-05, "loss": 0.5186, "step": 4191 }, { "epoch": 40.64578313253012, "grad_norm": 13.870115280151367, "learning_rate": 7.965048543689321e-05, "loss": 0.3028, "step": 4192 }, { "epoch": 40.65542168674699, "grad_norm": 8.583327293395996, "learning_rate": 7.964563106796117e-05, "loss": 0.229, "step": 4193 }, { "epoch": 40.665060240963854, "grad_norm": 4.473170280456543, "learning_rate": 7.964077669902912e-05, "loss": 0.3211, "step": 4194 }, { "epoch": 40.674698795180724, "grad_norm": 8.378066062927246, "learning_rate": 7.96359223300971e-05, "loss": 0.2225, "step": 4195 }, { "epoch": 40.68433734939759, "grad_norm": 5.55860710144043, "learning_rate": 7.963106796116505e-05, "loss": 0.2975, "step": 4196 }, { "epoch": 40.69397590361446, "grad_norm": 12.428603172302246, "learning_rate": 7.962621359223301e-05, "loss": 0.5236, "step": 4197 }, { "epoch": 40.70361445783133, "grad_norm": 6.894827842712402, "learning_rate": 7.962135922330097e-05, "loss": 0.4495, "step": 4198 }, { "epoch": 40.71325301204819, "grad_norm": 9.440921783447266, "learning_rate": 7.961650485436894e-05, "loss": 0.203, "step": 4199 }, { "epoch": 40.72289156626506, "grad_norm": 7.967469215393066, "learning_rate": 7.96116504854369e-05, "loss": 0.319, "step": 4200 }, { "epoch": 40.732530120481925, "grad_norm": 4.094396114349365, "learning_rate": 7.960679611650486e-05, "loss": 0.1619, "step": 4201 }, { "epoch": 40.742168674698796, "grad_norm": 8.086089134216309, "learning_rate": 7.960194174757281e-05, "loss": 0.4076, "step": 4202 }, { "epoch": 40.75180722891566, "grad_norm": 4.242579936981201, "learning_rate": 7.959708737864077e-05, "loss": 0.1997, "step": 4203 }, { "epoch": 40.76144578313253, "grad_norm": 4.749634742736816, "learning_rate": 7.959223300970874e-05, "loss": 0.2471, "step": 4204 }, { "epoch": 40.7710843373494, "grad_norm": 11.244935035705566, "learning_rate": 7.95873786407767e-05, "loss": 0.369, "step": 4205 }, { "epoch": 40.78072289156626, "grad_norm": 22.346696853637695, "learning_rate": 7.958252427184466e-05, "loss": 0.4519, "step": 4206 }, { "epoch": 40.790361445783134, "grad_norm": 6.815518379211426, "learning_rate": 7.957766990291262e-05, "loss": 0.1937, "step": 4207 }, { "epoch": 40.8, "grad_norm": 7.36038875579834, "learning_rate": 7.957281553398059e-05, "loss": 0.2998, "step": 4208 }, { "epoch": 40.80963855421687, "grad_norm": 6.936501979827881, "learning_rate": 7.956796116504855e-05, "loss": 0.4839, "step": 4209 }, { "epoch": 40.81927710843374, "grad_norm": 7.367212295532227, "learning_rate": 7.95631067961165e-05, "loss": 0.2856, "step": 4210 }, { "epoch": 40.8289156626506, "grad_norm": 6.1016764640808105, "learning_rate": 7.955825242718446e-05, "loss": 0.3082, "step": 4211 }, { "epoch": 40.83855421686747, "grad_norm": 5.3812031745910645, "learning_rate": 7.955339805825243e-05, "loss": 0.2628, "step": 4212 }, { "epoch": 40.848192771084335, "grad_norm": 6.534830570220947, "learning_rate": 7.954854368932039e-05, "loss": 0.2881, "step": 4213 }, { "epoch": 40.857831325301206, "grad_norm": 7.396784782409668, "learning_rate": 7.954368932038835e-05, "loss": 0.2654, "step": 4214 }, { "epoch": 40.86746987951807, "grad_norm": 12.614002227783203, "learning_rate": 7.95388349514563e-05, "loss": 0.478, "step": 4215 }, { "epoch": 40.87710843373494, "grad_norm": 7.910267353057861, "learning_rate": 7.953398058252428e-05, "loss": 0.217, "step": 4216 }, { "epoch": 40.88674698795181, "grad_norm": 7.917011737823486, "learning_rate": 7.952912621359225e-05, "loss": 0.2068, "step": 4217 }, { "epoch": 40.89638554216867, "grad_norm": 9.244901657104492, "learning_rate": 7.95242718446602e-05, "loss": 0.2209, "step": 4218 }, { "epoch": 40.90602409638554, "grad_norm": 6.159719944000244, "learning_rate": 7.951941747572816e-05, "loss": 0.2169, "step": 4219 }, { "epoch": 40.91566265060241, "grad_norm": 6.625805377960205, "learning_rate": 7.951456310679612e-05, "loss": 0.2535, "step": 4220 }, { "epoch": 40.92530120481928, "grad_norm": 22.05204963684082, "learning_rate": 7.950970873786409e-05, "loss": 0.3457, "step": 4221 }, { "epoch": 40.93493975903615, "grad_norm": 7.8204026222229, "learning_rate": 7.950485436893205e-05, "loss": 0.1887, "step": 4222 }, { "epoch": 40.94457831325301, "grad_norm": 7.934009552001953, "learning_rate": 7.950000000000001e-05, "loss": 0.2567, "step": 4223 }, { "epoch": 40.95421686746988, "grad_norm": 12.116974830627441, "learning_rate": 7.949514563106797e-05, "loss": 0.613, "step": 4224 }, { "epoch": 40.963855421686745, "grad_norm": 6.331194877624512, "learning_rate": 7.949029126213592e-05, "loss": 0.2829, "step": 4225 }, { "epoch": 40.973493975903615, "grad_norm": 3.4131922721862793, "learning_rate": 7.94854368932039e-05, "loss": 0.1805, "step": 4226 }, { "epoch": 40.98313253012048, "grad_norm": 7.39122200012207, "learning_rate": 7.948058252427185e-05, "loss": 0.2432, "step": 4227 }, { "epoch": 40.99277108433735, "grad_norm": 10.291929244995117, "learning_rate": 7.947572815533981e-05, "loss": 0.1757, "step": 4228 }, { "epoch": 41.00843373493976, "grad_norm": 15.371843338012695, "learning_rate": 7.947087378640777e-05, "loss": 0.4693, "step": 4229 }, { "epoch": 41.01807228915663, "grad_norm": 21.269350051879883, "learning_rate": 7.946601941747574e-05, "loss": 0.5163, "step": 4230 }, { "epoch": 41.02771084337349, "grad_norm": 8.059710502624512, "learning_rate": 7.94611650485437e-05, "loss": 0.1951, "step": 4231 }, { "epoch": 41.03734939759036, "grad_norm": 2.33252215385437, "learning_rate": 7.945631067961165e-05, "loss": 0.2456, "step": 4232 }, { "epoch": 41.04698795180723, "grad_norm": 16.17742919921875, "learning_rate": 7.945145631067961e-05, "loss": 0.4768, "step": 4233 }, { "epoch": 41.056626506024095, "grad_norm": 1.9827156066894531, "learning_rate": 7.944660194174758e-05, "loss": 0.1626, "step": 4234 }, { "epoch": 41.066265060240966, "grad_norm": 8.53732681274414, "learning_rate": 7.944174757281554e-05, "loss": 0.3264, "step": 4235 }, { "epoch": 41.07590361445783, "grad_norm": 9.096097946166992, "learning_rate": 7.94368932038835e-05, "loss": 0.1776, "step": 4236 }, { "epoch": 41.0855421686747, "grad_norm": 3.362032413482666, "learning_rate": 7.943203883495146e-05, "loss": 0.2637, "step": 4237 }, { "epoch": 41.09518072289157, "grad_norm": 16.319190979003906, "learning_rate": 7.942718446601943e-05, "loss": 0.378, "step": 4238 }, { "epoch": 41.10481927710843, "grad_norm": 5.994530200958252, "learning_rate": 7.942233009708739e-05, "loss": 0.1621, "step": 4239 }, { "epoch": 41.1144578313253, "grad_norm": 32.34727096557617, "learning_rate": 7.941747572815534e-05, "loss": 0.3126, "step": 4240 }, { "epoch": 41.12409638554217, "grad_norm": 16.95684051513672, "learning_rate": 7.94126213592233e-05, "loss": 0.4491, "step": 4241 }, { "epoch": 41.13373493975904, "grad_norm": 6.454630374908447, "learning_rate": 7.940776699029126e-05, "loss": 0.2652, "step": 4242 }, { "epoch": 41.1433734939759, "grad_norm": 5.816340446472168, "learning_rate": 7.940291262135923e-05, "loss": 0.4058, "step": 4243 }, { "epoch": 41.15301204819277, "grad_norm": 5.722314357757568, "learning_rate": 7.939805825242719e-05, "loss": 0.2063, "step": 4244 }, { "epoch": 41.16265060240964, "grad_norm": 5.970911026000977, "learning_rate": 7.939320388349515e-05, "loss": 0.1156, "step": 4245 }, { "epoch": 41.172289156626505, "grad_norm": 19.08677864074707, "learning_rate": 7.93883495145631e-05, "loss": 0.3477, "step": 4246 }, { "epoch": 41.181927710843375, "grad_norm": 14.366501808166504, "learning_rate": 7.938349514563108e-05, "loss": 0.3057, "step": 4247 }, { "epoch": 41.19156626506024, "grad_norm": 2.014803647994995, "learning_rate": 7.937864077669903e-05, "loss": 0.266, "step": 4248 }, { "epoch": 41.20120481927711, "grad_norm": 10.422992706298828, "learning_rate": 7.937378640776699e-05, "loss": 0.4586, "step": 4249 }, { "epoch": 41.21084337349398, "grad_norm": 9.070619583129883, "learning_rate": 7.936893203883495e-05, "loss": 0.2375, "step": 4250 }, { "epoch": 41.22048192771084, "grad_norm": 25.933198928833008, "learning_rate": 7.936407766990292e-05, "loss": 0.2086, "step": 4251 }, { "epoch": 41.23012048192771, "grad_norm": 15.443026542663574, "learning_rate": 7.935922330097088e-05, "loss": 0.4617, "step": 4252 }, { "epoch": 41.23975903614458, "grad_norm": 10.949661254882812, "learning_rate": 7.935436893203884e-05, "loss": 0.3137, "step": 4253 }, { "epoch": 41.24939759036145, "grad_norm": 17.73229217529297, "learning_rate": 7.93495145631068e-05, "loss": 0.2811, "step": 4254 }, { "epoch": 41.25903614457831, "grad_norm": 26.41978645324707, "learning_rate": 7.934466019417476e-05, "loss": 0.274, "step": 4255 }, { "epoch": 41.26867469879518, "grad_norm": 8.952932357788086, "learning_rate": 7.933980582524272e-05, "loss": 0.2667, "step": 4256 }, { "epoch": 41.27831325301205, "grad_norm": 30.165388107299805, "learning_rate": 7.933495145631068e-05, "loss": 0.3778, "step": 4257 }, { "epoch": 41.287951807228914, "grad_norm": 14.694620132446289, "learning_rate": 7.933009708737864e-05, "loss": 0.254, "step": 4258 }, { "epoch": 41.297590361445785, "grad_norm": 16.93602752685547, "learning_rate": 7.93252427184466e-05, "loss": 0.4596, "step": 4259 }, { "epoch": 41.30722891566265, "grad_norm": 8.606695175170898, "learning_rate": 7.932038834951457e-05, "loss": 0.286, "step": 4260 }, { "epoch": 41.31686746987952, "grad_norm": 7.999302387237549, "learning_rate": 7.931553398058252e-05, "loss": 0.34, "step": 4261 }, { "epoch": 41.32650602409639, "grad_norm": 20.666275024414062, "learning_rate": 7.931067961165048e-05, "loss": 0.2534, "step": 4262 }, { "epoch": 41.33614457831325, "grad_norm": 12.305364608764648, "learning_rate": 7.930582524271844e-05, "loss": 0.3442, "step": 4263 }, { "epoch": 41.34578313253012, "grad_norm": 20.789060592651367, "learning_rate": 7.930097087378641e-05, "loss": 0.2379, "step": 4264 }, { "epoch": 41.355421686746986, "grad_norm": 2.812117576599121, "learning_rate": 7.929611650485437e-05, "loss": 0.3031, "step": 4265 }, { "epoch": 41.36506024096386, "grad_norm": 26.173952102661133, "learning_rate": 7.929126213592233e-05, "loss": 0.3559, "step": 4266 }, { "epoch": 41.37469879518072, "grad_norm": 28.454137802124023, "learning_rate": 7.92864077669903e-05, "loss": 0.5561, "step": 4267 }, { "epoch": 41.38433734939759, "grad_norm": 14.492112159729004, "learning_rate": 7.928155339805826e-05, "loss": 0.2967, "step": 4268 }, { "epoch": 41.39397590361446, "grad_norm": 4.11317253112793, "learning_rate": 7.927669902912623e-05, "loss": 0.3207, "step": 4269 }, { "epoch": 41.403614457831324, "grad_norm": 3.70601487159729, "learning_rate": 7.927184466019419e-05, "loss": 0.3463, "step": 4270 }, { "epoch": 41.413253012048195, "grad_norm": 33.51222229003906, "learning_rate": 7.926699029126214e-05, "loss": 0.2842, "step": 4271 }, { "epoch": 41.42289156626506, "grad_norm": 27.64628028869629, "learning_rate": 7.92621359223301e-05, "loss": 0.4209, "step": 4272 }, { "epoch": 41.43253012048193, "grad_norm": 19.165485382080078, "learning_rate": 7.925728155339807e-05, "loss": 0.2722, "step": 4273 }, { "epoch": 41.44216867469879, "grad_norm": 24.2318115234375, "learning_rate": 7.925242718446603e-05, "loss": 0.2624, "step": 4274 }, { "epoch": 41.45180722891566, "grad_norm": 24.277660369873047, "learning_rate": 7.924757281553399e-05, "loss": 0.3255, "step": 4275 }, { "epoch": 41.46144578313253, "grad_norm": 32.554603576660156, "learning_rate": 7.924271844660195e-05, "loss": 0.3422, "step": 4276 }, { "epoch": 41.471084337349396, "grad_norm": 18.175901412963867, "learning_rate": 7.923786407766992e-05, "loss": 0.4417, "step": 4277 }, { "epoch": 41.480722891566266, "grad_norm": 40.71525955200195, "learning_rate": 7.923300970873787e-05, "loss": 0.3227, "step": 4278 }, { "epoch": 41.49036144578313, "grad_norm": 13.06006145477295, "learning_rate": 7.922815533980583e-05, "loss": 0.2867, "step": 4279 }, { "epoch": 41.5, "grad_norm": 11.338165283203125, "learning_rate": 7.922330097087379e-05, "loss": 0.2926, "step": 4280 }, { "epoch": 41.50963855421687, "grad_norm": 47.14375305175781, "learning_rate": 7.921844660194175e-05, "loss": 0.4835, "step": 4281 }, { "epoch": 41.519277108433734, "grad_norm": 3.8065733909606934, "learning_rate": 7.921359223300972e-05, "loss": 0.3514, "step": 4282 }, { "epoch": 41.528915662650604, "grad_norm": 4.494954586029053, "learning_rate": 7.920873786407768e-05, "loss": 0.3994, "step": 4283 }, { "epoch": 41.53855421686747, "grad_norm": 2.398707866668701, "learning_rate": 7.920388349514563e-05, "loss": 0.2892, "step": 4284 }, { "epoch": 41.54819277108434, "grad_norm": 1.9431506395339966, "learning_rate": 7.919902912621359e-05, "loss": 0.2942, "step": 4285 }, { "epoch": 41.55783132530121, "grad_norm": 7.239547252655029, "learning_rate": 7.919417475728156e-05, "loss": 0.3429, "step": 4286 }, { "epoch": 41.56746987951807, "grad_norm": 7.295486927032471, "learning_rate": 7.918932038834952e-05, "loss": 0.2955, "step": 4287 }, { "epoch": 41.57710843373494, "grad_norm": 18.468955993652344, "learning_rate": 7.918446601941748e-05, "loss": 0.3261, "step": 4288 }, { "epoch": 41.586746987951805, "grad_norm": 8.079458236694336, "learning_rate": 7.917961165048544e-05, "loss": 0.3237, "step": 4289 }, { "epoch": 41.596385542168676, "grad_norm": 34.70955276489258, "learning_rate": 7.917475728155341e-05, "loss": 0.311, "step": 4290 }, { "epoch": 41.60602409638554, "grad_norm": 46.03289031982422, "learning_rate": 7.916990291262137e-05, "loss": 0.3186, "step": 4291 }, { "epoch": 41.61566265060241, "grad_norm": 21.17268943786621, "learning_rate": 7.916504854368932e-05, "loss": 0.5231, "step": 4292 }, { "epoch": 41.62530120481928, "grad_norm": 30.256181716918945, "learning_rate": 7.916019417475728e-05, "loss": 0.3352, "step": 4293 }, { "epoch": 41.63493975903614, "grad_norm": 39.75346374511719, "learning_rate": 7.915533980582525e-05, "loss": 0.6556, "step": 4294 }, { "epoch": 41.644578313253014, "grad_norm": 19.019622802734375, "learning_rate": 7.915048543689321e-05, "loss": 0.2691, "step": 4295 }, { "epoch": 41.65421686746988, "grad_norm": 2.5070486068725586, "learning_rate": 7.914563106796117e-05, "loss": 0.1846, "step": 4296 }, { "epoch": 41.66385542168675, "grad_norm": 6.577592849731445, "learning_rate": 7.914077669902913e-05, "loss": 0.5042, "step": 4297 }, { "epoch": 41.67349397590361, "grad_norm": 12.170208930969238, "learning_rate": 7.913592233009708e-05, "loss": 0.3124, "step": 4298 }, { "epoch": 41.68313253012048, "grad_norm": 7.564280033111572, "learning_rate": 7.913106796116506e-05, "loss": 0.2122, "step": 4299 }, { "epoch": 41.69277108433735, "grad_norm": 10.422323226928711, "learning_rate": 7.912621359223301e-05, "loss": 0.2336, "step": 4300 }, { "epoch": 41.702409638554215, "grad_norm": 13.031074523925781, "learning_rate": 7.912135922330097e-05, "loss": 0.5544, "step": 4301 }, { "epoch": 41.712048192771086, "grad_norm": 6.174999237060547, "learning_rate": 7.911650485436893e-05, "loss": 0.3997, "step": 4302 }, { "epoch": 41.72168674698795, "grad_norm": 16.26624298095703, "learning_rate": 7.91116504854369e-05, "loss": 0.3602, "step": 4303 }, { "epoch": 41.73132530120482, "grad_norm": 9.976166725158691, "learning_rate": 7.910679611650486e-05, "loss": 0.3387, "step": 4304 }, { "epoch": 41.74096385542169, "grad_norm": 9.739113807678223, "learning_rate": 7.910194174757282e-05, "loss": 0.3155, "step": 4305 }, { "epoch": 41.75060240963855, "grad_norm": 81.55187225341797, "learning_rate": 7.909708737864077e-05, "loss": 0.4, "step": 4306 }, { "epoch": 41.76024096385542, "grad_norm": 9.527631759643555, "learning_rate": 7.909223300970874e-05, "loss": 0.3239, "step": 4307 }, { "epoch": 41.76987951807229, "grad_norm": 20.700963973999023, "learning_rate": 7.90873786407767e-05, "loss": 0.1535, "step": 4308 }, { "epoch": 41.77951807228916, "grad_norm": 14.20887565612793, "learning_rate": 7.908252427184466e-05, "loss": 0.3955, "step": 4309 }, { "epoch": 41.78915662650603, "grad_norm": 47.190303802490234, "learning_rate": 7.907766990291262e-05, "loss": 0.3186, "step": 4310 }, { "epoch": 41.79879518072289, "grad_norm": 9.670117378234863, "learning_rate": 7.907281553398059e-05, "loss": 0.1674, "step": 4311 }, { "epoch": 41.80843373493976, "grad_norm": 8.113764762878418, "learning_rate": 7.906796116504855e-05, "loss": 0.4282, "step": 4312 }, { "epoch": 41.818072289156625, "grad_norm": 4.40543794631958, "learning_rate": 7.90631067961165e-05, "loss": 0.2318, "step": 4313 }, { "epoch": 41.827710843373495, "grad_norm": 9.669439315795898, "learning_rate": 7.905825242718446e-05, "loss": 0.3526, "step": 4314 }, { "epoch": 41.83734939759036, "grad_norm": 13.430965423583984, "learning_rate": 7.905339805825242e-05, "loss": 0.5563, "step": 4315 }, { "epoch": 41.84698795180723, "grad_norm": 22.6013240814209, "learning_rate": 7.904854368932039e-05, "loss": 0.2461, "step": 4316 }, { "epoch": 41.8566265060241, "grad_norm": 7.748489856719971, "learning_rate": 7.904368932038836e-05, "loss": 0.2136, "step": 4317 }, { "epoch": 41.86626506024096, "grad_norm": 13.117732048034668, "learning_rate": 7.903883495145632e-05, "loss": 0.2662, "step": 4318 }, { "epoch": 41.87590361445783, "grad_norm": 6.868534564971924, "learning_rate": 7.903398058252428e-05, "loss": 0.356, "step": 4319 }, { "epoch": 41.8855421686747, "grad_norm": 9.53190803527832, "learning_rate": 7.902912621359224e-05, "loss": 0.5325, "step": 4320 }, { "epoch": 41.89518072289157, "grad_norm": 11.085938453674316, "learning_rate": 7.902427184466021e-05, "loss": 0.2993, "step": 4321 }, { "epoch": 41.90481927710843, "grad_norm": 14.867979049682617, "learning_rate": 7.901941747572816e-05, "loss": 0.293, "step": 4322 }, { "epoch": 41.9144578313253, "grad_norm": 21.537832260131836, "learning_rate": 7.901456310679612e-05, "loss": 0.39, "step": 4323 }, { "epoch": 41.92409638554217, "grad_norm": 6.190297603607178, "learning_rate": 7.900970873786408e-05, "loss": 0.1918, "step": 4324 }, { "epoch": 41.933734939759034, "grad_norm": 3.869019031524658, "learning_rate": 7.900485436893205e-05, "loss": 0.3723, "step": 4325 }, { "epoch": 41.943373493975905, "grad_norm": 16.818180084228516, "learning_rate": 7.900000000000001e-05, "loss": 0.4625, "step": 4326 }, { "epoch": 41.95301204819277, "grad_norm": 7.612067222595215, "learning_rate": 7.899514563106797e-05, "loss": 0.338, "step": 4327 }, { "epoch": 41.96265060240964, "grad_norm": 13.563863754272461, "learning_rate": 7.899029126213592e-05, "loss": 0.3297, "step": 4328 }, { "epoch": 41.97228915662651, "grad_norm": 6.411911487579346, "learning_rate": 7.89854368932039e-05, "loss": 0.2711, "step": 4329 }, { "epoch": 41.98192771084337, "grad_norm": 7.195040702819824, "learning_rate": 7.898058252427185e-05, "loss": 0.4244, "step": 4330 }, { "epoch": 41.99156626506024, "grad_norm": 6.407606601715088, "learning_rate": 7.897572815533981e-05, "loss": 0.2603, "step": 4331 }, { "epoch": 42.00722891566265, "grad_norm": 1.904571294784546, "learning_rate": 7.897087378640777e-05, "loss": 0.2091, "step": 4332 }, { "epoch": 42.01686746987952, "grad_norm": 33.283416748046875, "learning_rate": 7.896601941747573e-05, "loss": 0.4397, "step": 4333 }, { "epoch": 42.026506024096385, "grad_norm": 14.30436897277832, "learning_rate": 7.89611650485437e-05, "loss": 0.5404, "step": 4334 }, { "epoch": 42.036144578313255, "grad_norm": 9.268548011779785, "learning_rate": 7.895631067961166e-05, "loss": 0.2574, "step": 4335 }, { "epoch": 42.04578313253012, "grad_norm": 6.6942901611328125, "learning_rate": 7.895145631067961e-05, "loss": 0.171, "step": 4336 }, { "epoch": 42.05542168674699, "grad_norm": 21.456607818603516, "learning_rate": 7.894660194174757e-05, "loss": 0.5492, "step": 4337 }, { "epoch": 42.06506024096385, "grad_norm": 25.767915725708008, "learning_rate": 7.894174757281554e-05, "loss": 0.2871, "step": 4338 }, { "epoch": 42.07469879518072, "grad_norm": 11.5248384475708, "learning_rate": 7.89368932038835e-05, "loss": 0.3246, "step": 4339 }, { "epoch": 42.08433734939759, "grad_norm": 6.448894023895264, "learning_rate": 7.893203883495146e-05, "loss": 0.2795, "step": 4340 }, { "epoch": 42.09397590361446, "grad_norm": 11.648497581481934, "learning_rate": 7.892718446601942e-05, "loss": 0.3607, "step": 4341 }, { "epoch": 42.10361445783133, "grad_norm": 2.4234201908111572, "learning_rate": 7.892233009708739e-05, "loss": 0.1873, "step": 4342 }, { "epoch": 42.11325301204819, "grad_norm": 7.081000328063965, "learning_rate": 7.891747572815535e-05, "loss": 0.2338, "step": 4343 }, { "epoch": 42.12289156626506, "grad_norm": 2.6857659816741943, "learning_rate": 7.89126213592233e-05, "loss": 0.3405, "step": 4344 }, { "epoch": 42.13253012048193, "grad_norm": 5.888179302215576, "learning_rate": 7.890776699029126e-05, "loss": 0.2814, "step": 4345 }, { "epoch": 42.142168674698794, "grad_norm": 10.959869384765625, "learning_rate": 7.890291262135923e-05, "loss": 0.4804, "step": 4346 }, { "epoch": 42.151807228915665, "grad_norm": 3.1350746154785156, "learning_rate": 7.889805825242719e-05, "loss": 0.258, "step": 4347 }, { "epoch": 42.16144578313253, "grad_norm": 6.47755765914917, "learning_rate": 7.889320388349515e-05, "loss": 0.321, "step": 4348 }, { "epoch": 42.1710843373494, "grad_norm": 8.035584449768066, "learning_rate": 7.88883495145631e-05, "loss": 0.2437, "step": 4349 }, { "epoch": 42.18072289156626, "grad_norm": 47.1693000793457, "learning_rate": 7.888349514563106e-05, "loss": 0.6268, "step": 4350 }, { "epoch": 42.19036144578313, "grad_norm": 14.045479774475098, "learning_rate": 7.887864077669903e-05, "loss": 0.323, "step": 4351 }, { "epoch": 42.2, "grad_norm": 3.6725616455078125, "learning_rate": 7.887378640776699e-05, "loss": 0.1612, "step": 4352 }, { "epoch": 42.209638554216866, "grad_norm": 26.03337860107422, "learning_rate": 7.886893203883495e-05, "loss": 0.3556, "step": 4353 }, { "epoch": 42.21927710843374, "grad_norm": 10.942137718200684, "learning_rate": 7.886407766990291e-05, "loss": 0.2923, "step": 4354 }, { "epoch": 42.2289156626506, "grad_norm": 14.167069435119629, "learning_rate": 7.885922330097088e-05, "loss": 0.1439, "step": 4355 }, { "epoch": 42.23855421686747, "grad_norm": 12.325176239013672, "learning_rate": 7.885436893203884e-05, "loss": 0.4318, "step": 4356 }, { "epoch": 42.24819277108434, "grad_norm": 4.191765785217285, "learning_rate": 7.88495145631068e-05, "loss": 0.3737, "step": 4357 }, { "epoch": 42.257831325301204, "grad_norm": 5.371860504150391, "learning_rate": 7.884466019417475e-05, "loss": 0.2198, "step": 4358 }, { "epoch": 42.267469879518075, "grad_norm": 30.0861759185791, "learning_rate": 7.883980582524272e-05, "loss": 0.3051, "step": 4359 }, { "epoch": 42.27710843373494, "grad_norm": 13.918561935424805, "learning_rate": 7.883495145631068e-05, "loss": 0.271, "step": 4360 }, { "epoch": 42.28674698795181, "grad_norm": 18.41411781311035, "learning_rate": 7.883009708737864e-05, "loss": 0.3589, "step": 4361 }, { "epoch": 42.29638554216867, "grad_norm": 13.65330696105957, "learning_rate": 7.88252427184466e-05, "loss": 0.3399, "step": 4362 }, { "epoch": 42.30602409638554, "grad_norm": 6.18696403503418, "learning_rate": 7.882038834951457e-05, "loss": 0.3662, "step": 4363 }, { "epoch": 42.31566265060241, "grad_norm": 10.926609992980957, "learning_rate": 7.881553398058253e-05, "loss": 0.3106, "step": 4364 }, { "epoch": 42.325301204819276, "grad_norm": 20.9941349029541, "learning_rate": 7.881067961165048e-05, "loss": 0.3858, "step": 4365 }, { "epoch": 42.334939759036146, "grad_norm": 5.194899082183838, "learning_rate": 7.880582524271844e-05, "loss": 0.2031, "step": 4366 }, { "epoch": 42.34457831325301, "grad_norm": 24.321992874145508, "learning_rate": 7.88009708737864e-05, "loss": 0.4668, "step": 4367 }, { "epoch": 42.35421686746988, "grad_norm": 19.234582901000977, "learning_rate": 7.879611650485438e-05, "loss": 0.3039, "step": 4368 }, { "epoch": 42.36385542168675, "grad_norm": 2.660968780517578, "learning_rate": 7.879126213592234e-05, "loss": 0.2918, "step": 4369 }, { "epoch": 42.373493975903614, "grad_norm": 22.844497680664062, "learning_rate": 7.87864077669903e-05, "loss": 0.4998, "step": 4370 }, { "epoch": 42.383132530120484, "grad_norm": 9.406044960021973, "learning_rate": 7.878155339805826e-05, "loss": 0.3088, "step": 4371 }, { "epoch": 42.39277108433735, "grad_norm": 13.015501976013184, "learning_rate": 7.877669902912622e-05, "loss": 0.2065, "step": 4372 }, { "epoch": 42.40240963855422, "grad_norm": 17.995574951171875, "learning_rate": 7.877184466019419e-05, "loss": 0.4224, "step": 4373 }, { "epoch": 42.41204819277108, "grad_norm": 2.605130910873413, "learning_rate": 7.876699029126214e-05, "loss": 0.2044, "step": 4374 }, { "epoch": 42.42168674698795, "grad_norm": 10.658329963684082, "learning_rate": 7.87621359223301e-05, "loss": 0.2384, "step": 4375 }, { "epoch": 42.43132530120482, "grad_norm": 7.206455707550049, "learning_rate": 7.875728155339806e-05, "loss": 0.4204, "step": 4376 }, { "epoch": 42.440963855421685, "grad_norm": 7.271116733551025, "learning_rate": 7.875242718446603e-05, "loss": 0.2247, "step": 4377 }, { "epoch": 42.450602409638556, "grad_norm": 7.292287826538086, "learning_rate": 7.874757281553399e-05, "loss": 0.3712, "step": 4378 }, { "epoch": 42.46024096385542, "grad_norm": 12.260629653930664, "learning_rate": 7.874271844660195e-05, "loss": 0.2404, "step": 4379 }, { "epoch": 42.46987951807229, "grad_norm": 20.018598556518555, "learning_rate": 7.87378640776699e-05, "loss": 0.3165, "step": 4380 }, { "epoch": 42.47951807228916, "grad_norm": 7.806912899017334, "learning_rate": 7.873300970873788e-05, "loss": 0.2956, "step": 4381 }, { "epoch": 42.48915662650602, "grad_norm": 24.39199447631836, "learning_rate": 7.872815533980583e-05, "loss": 0.3585, "step": 4382 }, { "epoch": 42.498795180722894, "grad_norm": 23.68526268005371, "learning_rate": 7.872330097087379e-05, "loss": 0.3272, "step": 4383 }, { "epoch": 42.50843373493976, "grad_norm": 5.6618266105651855, "learning_rate": 7.871844660194175e-05, "loss": 0.3751, "step": 4384 }, { "epoch": 42.51807228915663, "grad_norm": 19.724010467529297, "learning_rate": 7.871359223300972e-05, "loss": 0.3687, "step": 4385 }, { "epoch": 42.52771084337349, "grad_norm": 25.29085922241211, "learning_rate": 7.870873786407768e-05, "loss": 0.675, "step": 4386 }, { "epoch": 42.53734939759036, "grad_norm": 9.784281730651855, "learning_rate": 7.870388349514564e-05, "loss": 0.3193, "step": 4387 }, { "epoch": 42.54698795180723, "grad_norm": 13.443415641784668, "learning_rate": 7.86990291262136e-05, "loss": 0.5795, "step": 4388 }, { "epoch": 42.556626506024095, "grad_norm": 15.690094947814941, "learning_rate": 7.869417475728155e-05, "loss": 0.3763, "step": 4389 }, { "epoch": 42.566265060240966, "grad_norm": 38.17797088623047, "learning_rate": 7.868932038834952e-05, "loss": 0.2238, "step": 4390 }, { "epoch": 42.57590361445783, "grad_norm": 5.717434883117676, "learning_rate": 7.868446601941748e-05, "loss": 0.375, "step": 4391 }, { "epoch": 42.5855421686747, "grad_norm": 13.148069381713867, "learning_rate": 7.867961165048544e-05, "loss": 0.4264, "step": 4392 }, { "epoch": 42.59518072289157, "grad_norm": 21.344100952148438, "learning_rate": 7.86747572815534e-05, "loss": 0.2786, "step": 4393 }, { "epoch": 42.60481927710843, "grad_norm": 7.004610061645508, "learning_rate": 7.866990291262137e-05, "loss": 0.2338, "step": 4394 }, { "epoch": 42.6144578313253, "grad_norm": 30.183210372924805, "learning_rate": 7.866504854368933e-05, "loss": 0.2926, "step": 4395 }, { "epoch": 42.62409638554217, "grad_norm": 9.565695762634277, "learning_rate": 7.866019417475728e-05, "loss": 0.5803, "step": 4396 }, { "epoch": 42.63373493975904, "grad_norm": 12.455824851989746, "learning_rate": 7.865533980582524e-05, "loss": 0.319, "step": 4397 }, { "epoch": 42.6433734939759, "grad_norm": 7.364126205444336, "learning_rate": 7.865048543689321e-05, "loss": 0.4019, "step": 4398 }, { "epoch": 42.65301204819277, "grad_norm": 10.754877090454102, "learning_rate": 7.864563106796117e-05, "loss": 0.1946, "step": 4399 }, { "epoch": 42.66265060240964, "grad_norm": 7.5091633796691895, "learning_rate": 7.864077669902913e-05, "loss": 0.2146, "step": 4400 }, { "epoch": 42.672289156626505, "grad_norm": 8.784540176391602, "learning_rate": 7.863592233009709e-05, "loss": 0.323, "step": 4401 }, { "epoch": 42.681927710843375, "grad_norm": 12.247650146484375, "learning_rate": 7.863106796116506e-05, "loss": 0.3659, "step": 4402 }, { "epoch": 42.69156626506024, "grad_norm": 2.976785182952881, "learning_rate": 7.862621359223301e-05, "loss": 0.4201, "step": 4403 }, { "epoch": 42.70120481927711, "grad_norm": 7.3646440505981445, "learning_rate": 7.862135922330097e-05, "loss": 0.2393, "step": 4404 }, { "epoch": 42.71084337349397, "grad_norm": 8.5873384475708, "learning_rate": 7.861650485436893e-05, "loss": 0.2575, "step": 4405 }, { "epoch": 42.72048192771084, "grad_norm": 27.687580108642578, "learning_rate": 7.861165048543689e-05, "loss": 0.5369, "step": 4406 }, { "epoch": 42.73012048192771, "grad_norm": 8.498696327209473, "learning_rate": 7.860679611650486e-05, "loss": 0.4538, "step": 4407 }, { "epoch": 42.73975903614458, "grad_norm": 15.597735404968262, "learning_rate": 7.860194174757282e-05, "loss": 0.241, "step": 4408 }, { "epoch": 42.74939759036145, "grad_norm": 32.57813262939453, "learning_rate": 7.859708737864077e-05, "loss": 0.4405, "step": 4409 }, { "epoch": 42.75903614457831, "grad_norm": 26.31399154663086, "learning_rate": 7.859223300970873e-05, "loss": 0.2615, "step": 4410 }, { "epoch": 42.76867469879518, "grad_norm": 7.755858421325684, "learning_rate": 7.85873786407767e-05, "loss": 0.2475, "step": 4411 }, { "epoch": 42.77831325301205, "grad_norm": 13.61223030090332, "learning_rate": 7.858252427184466e-05, "loss": 0.3894, "step": 4412 }, { "epoch": 42.787951807228914, "grad_norm": 6.328771591186523, "learning_rate": 7.857766990291262e-05, "loss": 0.4079, "step": 4413 }, { "epoch": 42.797590361445785, "grad_norm": 19.31540298461914, "learning_rate": 7.857281553398058e-05, "loss": 0.6004, "step": 4414 }, { "epoch": 42.80722891566265, "grad_norm": 6.927850246429443, "learning_rate": 7.856796116504855e-05, "loss": 0.2881, "step": 4415 }, { "epoch": 42.81686746987952, "grad_norm": 3.326998710632324, "learning_rate": 7.85631067961165e-05, "loss": 0.3019, "step": 4416 }, { "epoch": 42.82650602409639, "grad_norm": 16.574827194213867, "learning_rate": 7.855825242718446e-05, "loss": 0.385, "step": 4417 }, { "epoch": 42.83614457831325, "grad_norm": 5.399365425109863, "learning_rate": 7.855339805825243e-05, "loss": 0.3085, "step": 4418 }, { "epoch": 42.84578313253012, "grad_norm": 6.792282581329346, "learning_rate": 7.854854368932039e-05, "loss": 0.3062, "step": 4419 }, { "epoch": 42.855421686746986, "grad_norm": 6.3826141357421875, "learning_rate": 7.854368932038836e-05, "loss": 0.2494, "step": 4420 }, { "epoch": 42.86506024096386, "grad_norm": 21.791948318481445, "learning_rate": 7.853883495145632e-05, "loss": 0.5598, "step": 4421 }, { "epoch": 42.87469879518072, "grad_norm": 25.42186737060547, "learning_rate": 7.853398058252428e-05, "loss": 0.3724, "step": 4422 }, { "epoch": 42.88433734939759, "grad_norm": 5.3072381019592285, "learning_rate": 7.852912621359224e-05, "loss": 0.3166, "step": 4423 }, { "epoch": 42.89397590361446, "grad_norm": 68.40231323242188, "learning_rate": 7.852427184466021e-05, "loss": 0.1817, "step": 4424 }, { "epoch": 42.903614457831324, "grad_norm": 6.484898090362549, "learning_rate": 7.851941747572817e-05, "loss": 0.3527, "step": 4425 }, { "epoch": 42.913253012048195, "grad_norm": 2.755032777786255, "learning_rate": 7.851456310679612e-05, "loss": 0.261, "step": 4426 }, { "epoch": 42.92289156626506, "grad_norm": 19.307954788208008, "learning_rate": 7.850970873786408e-05, "loss": 0.2839, "step": 4427 }, { "epoch": 42.93253012048193, "grad_norm": 6.711308002471924, "learning_rate": 7.850485436893204e-05, "loss": 0.3093, "step": 4428 }, { "epoch": 42.94216867469879, "grad_norm": 30.866928100585938, "learning_rate": 7.850000000000001e-05, "loss": 0.4841, "step": 4429 }, { "epoch": 42.95180722891566, "grad_norm": 11.699029922485352, "learning_rate": 7.849514563106797e-05, "loss": 0.5169, "step": 4430 }, { "epoch": 42.96144578313253, "grad_norm": 23.947067260742188, "learning_rate": 7.849029126213593e-05, "loss": 0.2389, "step": 4431 }, { "epoch": 42.971084337349396, "grad_norm": 6.792580604553223, "learning_rate": 7.848543689320388e-05, "loss": 0.369, "step": 4432 }, { "epoch": 42.980722891566266, "grad_norm": 3.1628565788269043, "learning_rate": 7.848058252427186e-05, "loss": 0.1394, "step": 4433 }, { "epoch": 42.99036144578313, "grad_norm": 12.106389045715332, "learning_rate": 7.847572815533981e-05, "loss": 0.6861, "step": 4434 }, { "epoch": 43.006024096385545, "grad_norm": 10.854767799377441, "learning_rate": 7.847087378640777e-05, "loss": 0.528, "step": 4435 }, { "epoch": 43.01566265060241, "grad_norm": 24.236225128173828, "learning_rate": 7.846601941747573e-05, "loss": 0.3446, "step": 4436 }, { "epoch": 43.02530120481928, "grad_norm": 13.430481910705566, "learning_rate": 7.84611650485437e-05, "loss": 0.3455, "step": 4437 }, { "epoch": 43.03493975903614, "grad_norm": 4.708858013153076, "learning_rate": 7.845631067961166e-05, "loss": 0.1993, "step": 4438 }, { "epoch": 43.04457831325301, "grad_norm": 10.82209300994873, "learning_rate": 7.845145631067962e-05, "loss": 0.3744, "step": 4439 }, { "epoch": 43.05421686746988, "grad_norm": 13.50937557220459, "learning_rate": 7.844660194174757e-05, "loss": 0.2439, "step": 4440 }, { "epoch": 43.063855421686746, "grad_norm": 2.495741605758667, "learning_rate": 7.844174757281554e-05, "loss": 0.2962, "step": 4441 }, { "epoch": 43.07349397590362, "grad_norm": 32.24626159667969, "learning_rate": 7.84368932038835e-05, "loss": 0.2612, "step": 4442 }, { "epoch": 43.08313253012048, "grad_norm": 4.695351600646973, "learning_rate": 7.843203883495146e-05, "loss": 0.4437, "step": 4443 }, { "epoch": 43.09277108433735, "grad_norm": 8.962446212768555, "learning_rate": 7.842718446601942e-05, "loss": 0.2272, "step": 4444 }, { "epoch": 43.102409638554214, "grad_norm": 18.257862091064453, "learning_rate": 7.842233009708738e-05, "loss": 0.1947, "step": 4445 }, { "epoch": 43.112048192771084, "grad_norm": 14.030902862548828, "learning_rate": 7.841747572815535e-05, "loss": 0.3677, "step": 4446 }, { "epoch": 43.121686746987955, "grad_norm": 27.006120681762695, "learning_rate": 7.84126213592233e-05, "loss": 0.4396, "step": 4447 }, { "epoch": 43.13132530120482, "grad_norm": 28.48261070251465, "learning_rate": 7.840776699029126e-05, "loss": 0.5421, "step": 4448 }, { "epoch": 43.14096385542169, "grad_norm": 21.723270416259766, "learning_rate": 7.840291262135922e-05, "loss": 0.4953, "step": 4449 }, { "epoch": 43.15060240963855, "grad_norm": 24.018512725830078, "learning_rate": 7.839805825242719e-05, "loss": 0.3482, "step": 4450 }, { "epoch": 43.16024096385542, "grad_norm": 22.605390548706055, "learning_rate": 7.839320388349515e-05, "loss": 0.3037, "step": 4451 }, { "epoch": 43.16987951807229, "grad_norm": 5.130201816558838, "learning_rate": 7.838834951456311e-05, "loss": 0.2179, "step": 4452 }, { "epoch": 43.179518072289156, "grad_norm": 3.309171438217163, "learning_rate": 7.838349514563106e-05, "loss": 0.2127, "step": 4453 }, { "epoch": 43.189156626506026, "grad_norm": 4.739785671234131, "learning_rate": 7.837864077669904e-05, "loss": 0.1933, "step": 4454 }, { "epoch": 43.19879518072289, "grad_norm": 8.773955345153809, "learning_rate": 7.8373786407767e-05, "loss": 0.3052, "step": 4455 }, { "epoch": 43.20843373493976, "grad_norm": 30.50819969177246, "learning_rate": 7.836893203883495e-05, "loss": 0.4997, "step": 4456 }, { "epoch": 43.21807228915662, "grad_norm": 4.317107200622559, "learning_rate": 7.836407766990291e-05, "loss": 0.4226, "step": 4457 }, { "epoch": 43.227710843373494, "grad_norm": 82.28577423095703, "learning_rate": 7.835922330097088e-05, "loss": 0.2566, "step": 4458 }, { "epoch": 43.237349397590364, "grad_norm": 32.80439376831055, "learning_rate": 7.835436893203884e-05, "loss": 0.1453, "step": 4459 }, { "epoch": 43.24698795180723, "grad_norm": 13.377714157104492, "learning_rate": 7.83495145631068e-05, "loss": 0.3807, "step": 4460 }, { "epoch": 43.2566265060241, "grad_norm": 10.527002334594727, "learning_rate": 7.834466019417475e-05, "loss": 0.3961, "step": 4461 }, { "epoch": 43.26626506024096, "grad_norm": 10.033833503723145, "learning_rate": 7.833980582524271e-05, "loss": 0.3773, "step": 4462 }, { "epoch": 43.27590361445783, "grad_norm": 7.08451509475708, "learning_rate": 7.833495145631068e-05, "loss": 0.1869, "step": 4463 }, { "epoch": 43.2855421686747, "grad_norm": 13.071532249450684, "learning_rate": 7.833009708737864e-05, "loss": 0.2665, "step": 4464 }, { "epoch": 43.295180722891565, "grad_norm": 8.676480293273926, "learning_rate": 7.83252427184466e-05, "loss": 0.3615, "step": 4465 }, { "epoch": 43.304819277108436, "grad_norm": 20.43577003479004, "learning_rate": 7.832038834951456e-05, "loss": 0.5097, "step": 4466 }, { "epoch": 43.3144578313253, "grad_norm": 34.07159423828125, "learning_rate": 7.831553398058253e-05, "loss": 0.5609, "step": 4467 }, { "epoch": 43.32409638554217, "grad_norm": 4.842332363128662, "learning_rate": 7.831067961165049e-05, "loss": 0.2979, "step": 4468 }, { "epoch": 43.33373493975903, "grad_norm": 10.389739036560059, "learning_rate": 7.830582524271846e-05, "loss": 0.4698, "step": 4469 }, { "epoch": 43.3433734939759, "grad_norm": 13.815397262573242, "learning_rate": 7.830097087378641e-05, "loss": 0.2117, "step": 4470 }, { "epoch": 43.353012048192774, "grad_norm": 9.114405632019043, "learning_rate": 7.829611650485437e-05, "loss": 0.1874, "step": 4471 }, { "epoch": 43.36265060240964, "grad_norm": 8.257594108581543, "learning_rate": 7.829126213592234e-05, "loss": 0.3624, "step": 4472 }, { "epoch": 43.37228915662651, "grad_norm": 14.089897155761719, "learning_rate": 7.82864077669903e-05, "loss": 0.3998, "step": 4473 }, { "epoch": 43.38192771084337, "grad_norm": 6.197944164276123, "learning_rate": 7.828155339805826e-05, "loss": 0.2758, "step": 4474 }, { "epoch": 43.39156626506024, "grad_norm": 26.588346481323242, "learning_rate": 7.827669902912622e-05, "loss": 0.4176, "step": 4475 }, { "epoch": 43.40120481927711, "grad_norm": 41.317020416259766, "learning_rate": 7.827184466019419e-05, "loss": 0.5326, "step": 4476 }, { "epoch": 43.410843373493975, "grad_norm": 2.7339439392089844, "learning_rate": 7.826699029126215e-05, "loss": 0.2273, "step": 4477 }, { "epoch": 43.420481927710846, "grad_norm": 52.69807815551758, "learning_rate": 7.82621359223301e-05, "loss": 0.3785, "step": 4478 }, { "epoch": 43.43012048192771, "grad_norm": 27.403318405151367, "learning_rate": 7.825728155339806e-05, "loss": 0.2664, "step": 4479 }, { "epoch": 43.43975903614458, "grad_norm": 6.582303047180176, "learning_rate": 7.825242718446603e-05, "loss": 0.4169, "step": 4480 }, { "epoch": 43.44939759036144, "grad_norm": 7.1570329666137695, "learning_rate": 7.824757281553399e-05, "loss": 0.3712, "step": 4481 }, { "epoch": 43.45903614457831, "grad_norm": 1.1530169248580933, "learning_rate": 7.824271844660195e-05, "loss": 0.2655, "step": 4482 }, { "epoch": 43.46867469879518, "grad_norm": 7.155737400054932, "learning_rate": 7.82378640776699e-05, "loss": 0.2697, "step": 4483 }, { "epoch": 43.47831325301205, "grad_norm": 17.85148811340332, "learning_rate": 7.823300970873786e-05, "loss": 0.3553, "step": 4484 }, { "epoch": 43.48795180722892, "grad_norm": 23.22091293334961, "learning_rate": 7.822815533980583e-05, "loss": 0.3483, "step": 4485 }, { "epoch": 43.49759036144578, "grad_norm": 4.993752956390381, "learning_rate": 7.822330097087379e-05, "loss": 0.4543, "step": 4486 }, { "epoch": 43.50722891566265, "grad_norm": 7.203827857971191, "learning_rate": 7.821844660194175e-05, "loss": 0.381, "step": 4487 }, { "epoch": 43.51686746987952, "grad_norm": 34.89204025268555, "learning_rate": 7.821359223300971e-05, "loss": 0.5751, "step": 4488 }, { "epoch": 43.526506024096385, "grad_norm": 6.181936740875244, "learning_rate": 7.820873786407768e-05, "loss": 0.2147, "step": 4489 }, { "epoch": 43.536144578313255, "grad_norm": 10.032718658447266, "learning_rate": 7.820388349514564e-05, "loss": 0.332, "step": 4490 }, { "epoch": 43.54578313253012, "grad_norm": 4.373081684112549, "learning_rate": 7.81990291262136e-05, "loss": 0.363, "step": 4491 }, { "epoch": 43.55542168674699, "grad_norm": 6.2738189697265625, "learning_rate": 7.819417475728155e-05, "loss": 0.2202, "step": 4492 }, { "epoch": 43.56506024096385, "grad_norm": 14.751449584960938, "learning_rate": 7.818932038834952e-05, "loss": 0.3898, "step": 4493 }, { "epoch": 43.57469879518072, "grad_norm": 9.062698364257812, "learning_rate": 7.818446601941748e-05, "loss": 0.2862, "step": 4494 }, { "epoch": 43.58433734939759, "grad_norm": 20.981367111206055, "learning_rate": 7.817961165048544e-05, "loss": 0.4003, "step": 4495 }, { "epoch": 43.59397590361446, "grad_norm": 3.0316457748413086, "learning_rate": 7.81747572815534e-05, "loss": 0.2451, "step": 4496 }, { "epoch": 43.60361445783133, "grad_norm": 13.28626537322998, "learning_rate": 7.816990291262137e-05, "loss": 0.2958, "step": 4497 }, { "epoch": 43.61325301204819, "grad_norm": 4.0340495109558105, "learning_rate": 7.816504854368933e-05, "loss": 0.4705, "step": 4498 }, { "epoch": 43.62289156626506, "grad_norm": 3.640990972518921, "learning_rate": 7.816019417475728e-05, "loss": 0.3038, "step": 4499 }, { "epoch": 43.63253012048193, "grad_norm": 6.989196300506592, "learning_rate": 7.815533980582524e-05, "loss": 0.2053, "step": 4500 }, { "epoch": 43.642168674698794, "grad_norm": 23.28146743774414, "learning_rate": 7.81504854368932e-05, "loss": 0.1534, "step": 4501 }, { "epoch": 43.651807228915665, "grad_norm": 30.499034881591797, "learning_rate": 7.814563106796117e-05, "loss": 0.4935, "step": 4502 }, { "epoch": 43.66144578313253, "grad_norm": 15.548855781555176, "learning_rate": 7.814077669902913e-05, "loss": 0.3218, "step": 4503 }, { "epoch": 43.6710843373494, "grad_norm": 7.864441394805908, "learning_rate": 7.813592233009709e-05, "loss": 0.5232, "step": 4504 }, { "epoch": 43.68072289156626, "grad_norm": 10.358319282531738, "learning_rate": 7.813106796116504e-05, "loss": 0.2582, "step": 4505 }, { "epoch": 43.69036144578313, "grad_norm": 14.841388702392578, "learning_rate": 7.812621359223302e-05, "loss": 0.7057, "step": 4506 }, { "epoch": 43.7, "grad_norm": 6.2376298904418945, "learning_rate": 7.812135922330097e-05, "loss": 0.2543, "step": 4507 }, { "epoch": 43.709638554216866, "grad_norm": 10.432683944702148, "learning_rate": 7.811650485436893e-05, "loss": 0.2896, "step": 4508 }, { "epoch": 43.71927710843374, "grad_norm": 35.56601333618164, "learning_rate": 7.811165048543689e-05, "loss": 0.3924, "step": 4509 }, { "epoch": 43.7289156626506, "grad_norm": 12.88729476928711, "learning_rate": 7.810679611650486e-05, "loss": 0.2969, "step": 4510 }, { "epoch": 43.73855421686747, "grad_norm": 16.77302360534668, "learning_rate": 7.810194174757282e-05, "loss": 0.3963, "step": 4511 }, { "epoch": 43.74819277108434, "grad_norm": 5.285420894622803, "learning_rate": 7.809708737864078e-05, "loss": 0.2672, "step": 4512 }, { "epoch": 43.757831325301204, "grad_norm": 7.802116870880127, "learning_rate": 7.809223300970873e-05, "loss": 0.2225, "step": 4513 }, { "epoch": 43.767469879518075, "grad_norm": 2.7946059703826904, "learning_rate": 7.80873786407767e-05, "loss": 0.2143, "step": 4514 }, { "epoch": 43.77710843373494, "grad_norm": 5.598272800445557, "learning_rate": 7.808252427184466e-05, "loss": 0.3488, "step": 4515 }, { "epoch": 43.78674698795181, "grad_norm": 6.3809285163879395, "learning_rate": 7.807766990291262e-05, "loss": 0.3511, "step": 4516 }, { "epoch": 43.79638554216867, "grad_norm": 3.517470359802246, "learning_rate": 7.807281553398058e-05, "loss": 0.3964, "step": 4517 }, { "epoch": 43.80602409638554, "grad_norm": 3.689328908920288, "learning_rate": 7.806796116504854e-05, "loss": 0.2015, "step": 4518 }, { "epoch": 43.81566265060241, "grad_norm": 10.2308931350708, "learning_rate": 7.806310679611652e-05, "loss": 0.2592, "step": 4519 }, { "epoch": 43.825301204819276, "grad_norm": 13.564299583435059, "learning_rate": 7.805825242718448e-05, "loss": 0.3929, "step": 4520 }, { "epoch": 43.834939759036146, "grad_norm": 13.269978523254395, "learning_rate": 7.805339805825244e-05, "loss": 0.2958, "step": 4521 }, { "epoch": 43.84457831325301, "grad_norm": 13.169804573059082, "learning_rate": 7.80485436893204e-05, "loss": 0.3376, "step": 4522 }, { "epoch": 43.85421686746988, "grad_norm": 9.446765899658203, "learning_rate": 7.804368932038835e-05, "loss": 0.431, "step": 4523 }, { "epoch": 43.86385542168675, "grad_norm": 13.705148696899414, "learning_rate": 7.803883495145632e-05, "loss": 0.3173, "step": 4524 }, { "epoch": 43.873493975903614, "grad_norm": 16.0914249420166, "learning_rate": 7.803398058252428e-05, "loss": 0.2722, "step": 4525 }, { "epoch": 43.883132530120484, "grad_norm": 15.652511596679688, "learning_rate": 7.802912621359224e-05, "loss": 0.3538, "step": 4526 }, { "epoch": 43.89277108433735, "grad_norm": 23.32219886779785, "learning_rate": 7.80242718446602e-05, "loss": 0.3743, "step": 4527 }, { "epoch": 43.90240963855422, "grad_norm": 12.347796440124512, "learning_rate": 7.801941747572817e-05, "loss": 0.3753, "step": 4528 }, { "epoch": 43.91204819277108, "grad_norm": 9.173962593078613, "learning_rate": 7.801456310679613e-05, "loss": 0.6237, "step": 4529 }, { "epoch": 43.92168674698795, "grad_norm": 31.608747482299805, "learning_rate": 7.800970873786408e-05, "loss": 0.3188, "step": 4530 }, { "epoch": 43.93132530120482, "grad_norm": 38.19337844848633, "learning_rate": 7.800485436893204e-05, "loss": 0.561, "step": 4531 }, { "epoch": 43.940963855421685, "grad_norm": 20.001598358154297, "learning_rate": 7.800000000000001e-05, "loss": 0.4341, "step": 4532 }, { "epoch": 43.950602409638556, "grad_norm": 5.4322333335876465, "learning_rate": 7.799514563106797e-05, "loss": 0.307, "step": 4533 }, { "epoch": 43.96024096385542, "grad_norm": 10.896688461303711, "learning_rate": 7.799029126213593e-05, "loss": 0.4248, "step": 4534 }, { "epoch": 43.96987951807229, "grad_norm": 22.167036056518555, "learning_rate": 7.798543689320389e-05, "loss": 0.2875, "step": 4535 }, { "epoch": 43.97951807228916, "grad_norm": 17.364986419677734, "learning_rate": 7.798058252427186e-05, "loss": 0.237, "step": 4536 }, { "epoch": 43.98915662650602, "grad_norm": 18.47341537475586, "learning_rate": 7.797572815533981e-05, "loss": 0.3077, "step": 4537 }, { "epoch": 44.00481927710843, "grad_norm": 4.411258220672607, "learning_rate": 7.797087378640777e-05, "loss": 0.2216, "step": 4538 }, { "epoch": 44.0144578313253, "grad_norm": 4.054451942443848, "learning_rate": 7.796601941747573e-05, "loss": 0.1997, "step": 4539 }, { "epoch": 44.024096385542165, "grad_norm": 41.84880447387695, "learning_rate": 7.796116504854369e-05, "loss": 0.379, "step": 4540 }, { "epoch": 44.033734939759036, "grad_norm": 24.516292572021484, "learning_rate": 7.795631067961166e-05, "loss": 0.2756, "step": 4541 }, { "epoch": 44.043373493975906, "grad_norm": 7.7370405197143555, "learning_rate": 7.795145631067962e-05, "loss": 0.2928, "step": 4542 }, { "epoch": 44.05301204819277, "grad_norm": 3.6078548431396484, "learning_rate": 7.794660194174757e-05, "loss": 0.2469, "step": 4543 }, { "epoch": 44.06265060240964, "grad_norm": 4.252645015716553, "learning_rate": 7.794174757281553e-05, "loss": 0.1168, "step": 4544 }, { "epoch": 44.0722891566265, "grad_norm": 14.45711612701416, "learning_rate": 7.79368932038835e-05, "loss": 0.4789, "step": 4545 }, { "epoch": 44.081927710843374, "grad_norm": 9.790465354919434, "learning_rate": 7.793203883495146e-05, "loss": 0.3326, "step": 4546 }, { "epoch": 44.091566265060244, "grad_norm": 23.44277572631836, "learning_rate": 7.792718446601942e-05, "loss": 0.6439, "step": 4547 }, { "epoch": 44.10120481927711, "grad_norm": 11.377091407775879, "learning_rate": 7.792233009708738e-05, "loss": 0.2817, "step": 4548 }, { "epoch": 44.11084337349398, "grad_norm": 13.652661323547363, "learning_rate": 7.791747572815535e-05, "loss": 0.4028, "step": 4549 }, { "epoch": 44.12048192771084, "grad_norm": 7.710618019104004, "learning_rate": 7.79126213592233e-05, "loss": 0.1546, "step": 4550 }, { "epoch": 44.13012048192771, "grad_norm": 17.731346130371094, "learning_rate": 7.790776699029126e-05, "loss": 0.3891, "step": 4551 }, { "epoch": 44.139759036144575, "grad_norm": 5.20151948928833, "learning_rate": 7.790291262135922e-05, "loss": 0.1474, "step": 4552 }, { "epoch": 44.149397590361446, "grad_norm": 28.378509521484375, "learning_rate": 7.789805825242719e-05, "loss": 0.4405, "step": 4553 }, { "epoch": 44.159036144578316, "grad_norm": 7.344108581542969, "learning_rate": 7.789320388349515e-05, "loss": 0.2921, "step": 4554 }, { "epoch": 44.16867469879518, "grad_norm": 15.550301551818848, "learning_rate": 7.788834951456311e-05, "loss": 0.1499, "step": 4555 }, { "epoch": 44.17831325301205, "grad_norm": 9.920475959777832, "learning_rate": 7.788349514563107e-05, "loss": 0.4456, "step": 4556 }, { "epoch": 44.18795180722891, "grad_norm": 6.480612754821777, "learning_rate": 7.787864077669902e-05, "loss": 0.4415, "step": 4557 }, { "epoch": 44.19759036144578, "grad_norm": 5.353660583496094, "learning_rate": 7.7873786407767e-05, "loss": 0.4719, "step": 4558 }, { "epoch": 44.207228915662654, "grad_norm": 4.030913829803467, "learning_rate": 7.786893203883495e-05, "loss": 0.4294, "step": 4559 }, { "epoch": 44.21686746987952, "grad_norm": 13.821046829223633, "learning_rate": 7.786407766990291e-05, "loss": 0.4025, "step": 4560 }, { "epoch": 44.22650602409639, "grad_norm": 13.318949699401855, "learning_rate": 7.785922330097087e-05, "loss": 0.357, "step": 4561 }, { "epoch": 44.23614457831325, "grad_norm": 10.001596450805664, "learning_rate": 7.785436893203884e-05, "loss": 0.2115, "step": 4562 }, { "epoch": 44.24578313253012, "grad_norm": 5.343698978424072, "learning_rate": 7.78495145631068e-05, "loss": 0.264, "step": 4563 }, { "epoch": 44.255421686746985, "grad_norm": 9.620122909545898, "learning_rate": 7.784466019417476e-05, "loss": 0.3788, "step": 4564 }, { "epoch": 44.265060240963855, "grad_norm": 15.800562858581543, "learning_rate": 7.783980582524271e-05, "loss": 0.1658, "step": 4565 }, { "epoch": 44.274698795180726, "grad_norm": 6.620005130767822, "learning_rate": 7.783495145631068e-05, "loss": 0.406, "step": 4566 }, { "epoch": 44.28433734939759, "grad_norm": 18.818748474121094, "learning_rate": 7.783009708737864e-05, "loss": 0.3581, "step": 4567 }, { "epoch": 44.29397590361446, "grad_norm": 17.13374137878418, "learning_rate": 7.78252427184466e-05, "loss": 0.4572, "step": 4568 }, { "epoch": 44.30361445783132, "grad_norm": 8.118782997131348, "learning_rate": 7.782038834951456e-05, "loss": 0.2858, "step": 4569 }, { "epoch": 44.31325301204819, "grad_norm": 7.000833511352539, "learning_rate": 7.781553398058253e-05, "loss": 0.2993, "step": 4570 }, { "epoch": 44.32289156626506, "grad_norm": 5.78543758392334, "learning_rate": 7.78106796116505e-05, "loss": 0.415, "step": 4571 }, { "epoch": 44.33253012048193, "grad_norm": 8.757676124572754, "learning_rate": 7.780582524271846e-05, "loss": 0.3099, "step": 4572 }, { "epoch": 44.3421686746988, "grad_norm": 8.134239196777344, "learning_rate": 7.780097087378642e-05, "loss": 0.4476, "step": 4573 }, { "epoch": 44.35180722891566, "grad_norm": 6.50494909286499, "learning_rate": 7.779611650485437e-05, "loss": 0.3871, "step": 4574 }, { "epoch": 44.36144578313253, "grad_norm": 7.789238452911377, "learning_rate": 7.779126213592234e-05, "loss": 0.3814, "step": 4575 }, { "epoch": 44.371084337349394, "grad_norm": 14.148582458496094, "learning_rate": 7.77864077669903e-05, "loss": 0.278, "step": 4576 }, { "epoch": 44.380722891566265, "grad_norm": 54.83828353881836, "learning_rate": 7.778155339805826e-05, "loss": 0.2455, "step": 4577 }, { "epoch": 44.390361445783135, "grad_norm": 9.09475040435791, "learning_rate": 7.777669902912622e-05, "loss": 0.2585, "step": 4578 }, { "epoch": 44.4, "grad_norm": 14.2112398147583, "learning_rate": 7.777184466019418e-05, "loss": 0.3509, "step": 4579 }, { "epoch": 44.40963855421687, "grad_norm": 24.949892044067383, "learning_rate": 7.776699029126215e-05, "loss": 0.4761, "step": 4580 }, { "epoch": 44.41927710843373, "grad_norm": 8.094832420349121, "learning_rate": 7.77621359223301e-05, "loss": 0.3557, "step": 4581 }, { "epoch": 44.4289156626506, "grad_norm": 2.670300006866455, "learning_rate": 7.775728155339806e-05, "loss": 0.4657, "step": 4582 }, { "epoch": 44.43855421686747, "grad_norm": 43.05268478393555, "learning_rate": 7.775242718446602e-05, "loss": 0.1555, "step": 4583 }, { "epoch": 44.44819277108434, "grad_norm": 22.1136531829834, "learning_rate": 7.774757281553399e-05, "loss": 0.4545, "step": 4584 }, { "epoch": 44.45783132530121, "grad_norm": 4.684062957763672, "learning_rate": 7.774271844660195e-05, "loss": 0.2532, "step": 4585 }, { "epoch": 44.46746987951807, "grad_norm": 3.281196355819702, "learning_rate": 7.773786407766991e-05, "loss": 0.3205, "step": 4586 }, { "epoch": 44.47710843373494, "grad_norm": 2.500913381576538, "learning_rate": 7.773300970873786e-05, "loss": 0.4691, "step": 4587 }, { "epoch": 44.486746987951804, "grad_norm": 6.245790004730225, "learning_rate": 7.772815533980584e-05, "loss": 0.1358, "step": 4588 }, { "epoch": 44.496385542168674, "grad_norm": 6.4294633865356445, "learning_rate": 7.77233009708738e-05, "loss": 0.2039, "step": 4589 }, { "epoch": 44.506024096385545, "grad_norm": 6.244689464569092, "learning_rate": 7.771844660194175e-05, "loss": 0.2586, "step": 4590 }, { "epoch": 44.51566265060241, "grad_norm": 5.590883255004883, "learning_rate": 7.771359223300971e-05, "loss": 0.3325, "step": 4591 }, { "epoch": 44.52530120481928, "grad_norm": 4.3979997634887695, "learning_rate": 7.770873786407768e-05, "loss": 0.2291, "step": 4592 }, { "epoch": 44.53493975903614, "grad_norm": 11.911394119262695, "learning_rate": 7.770388349514564e-05, "loss": 0.3712, "step": 4593 }, { "epoch": 44.54457831325301, "grad_norm": 14.595010757446289, "learning_rate": 7.76990291262136e-05, "loss": 0.2108, "step": 4594 }, { "epoch": 44.55421686746988, "grad_norm": 7.801776885986328, "learning_rate": 7.769417475728155e-05, "loss": 0.2574, "step": 4595 }, { "epoch": 44.563855421686746, "grad_norm": 18.1627140045166, "learning_rate": 7.768932038834951e-05, "loss": 0.4454, "step": 4596 }, { "epoch": 44.57349397590362, "grad_norm": 3.089900493621826, "learning_rate": 7.768446601941748e-05, "loss": 0.2757, "step": 4597 }, { "epoch": 44.58313253012048, "grad_norm": 5.412187099456787, "learning_rate": 7.767961165048544e-05, "loss": 0.2696, "step": 4598 }, { "epoch": 44.59277108433735, "grad_norm": 6.4366230964660645, "learning_rate": 7.76747572815534e-05, "loss": 0.2588, "step": 4599 }, { "epoch": 44.602409638554214, "grad_norm": 21.81267547607422, "learning_rate": 7.766990291262136e-05, "loss": 0.3443, "step": 4600 }, { "epoch": 44.612048192771084, "grad_norm": 4.057292938232422, "learning_rate": 7.766504854368933e-05, "loss": 0.3613, "step": 4601 }, { "epoch": 44.621686746987955, "grad_norm": 4.942378997802734, "learning_rate": 7.766019417475729e-05, "loss": 0.463, "step": 4602 }, { "epoch": 44.63132530120482, "grad_norm": 6.498761177062988, "learning_rate": 7.765533980582524e-05, "loss": 0.2752, "step": 4603 }, { "epoch": 44.64096385542169, "grad_norm": 8.75404167175293, "learning_rate": 7.76504854368932e-05, "loss": 0.3005, "step": 4604 }, { "epoch": 44.65060240963855, "grad_norm": 3.243192434310913, "learning_rate": 7.764563106796117e-05, "loss": 0.1808, "step": 4605 }, { "epoch": 44.66024096385542, "grad_norm": 13.108922958374023, "learning_rate": 7.764077669902913e-05, "loss": 0.2317, "step": 4606 }, { "epoch": 44.66987951807229, "grad_norm": 28.06548500061035, "learning_rate": 7.763592233009709e-05, "loss": 0.3513, "step": 4607 }, { "epoch": 44.679518072289156, "grad_norm": 7.108211994171143, "learning_rate": 7.763106796116505e-05, "loss": 0.2862, "step": 4608 }, { "epoch": 44.689156626506026, "grad_norm": 14.857620239257812, "learning_rate": 7.762621359223302e-05, "loss": 0.4044, "step": 4609 }, { "epoch": 44.69879518072289, "grad_norm": 22.287179946899414, "learning_rate": 7.762135922330097e-05, "loss": 0.4882, "step": 4610 }, { "epoch": 44.70843373493976, "grad_norm": 10.877161026000977, "learning_rate": 7.761650485436893e-05, "loss": 0.2772, "step": 4611 }, { "epoch": 44.71807228915662, "grad_norm": 12.203481674194336, "learning_rate": 7.761165048543689e-05, "loss": 0.5534, "step": 4612 }, { "epoch": 44.727710843373494, "grad_norm": 29.3207950592041, "learning_rate": 7.760679611650485e-05, "loss": 0.3219, "step": 4613 }, { "epoch": 44.737349397590364, "grad_norm": 18.072046279907227, "learning_rate": 7.760194174757282e-05, "loss": 0.3655, "step": 4614 }, { "epoch": 44.74698795180723, "grad_norm": 21.4361629486084, "learning_rate": 7.759708737864078e-05, "loss": 0.3281, "step": 4615 }, { "epoch": 44.7566265060241, "grad_norm": 5.046407222747803, "learning_rate": 7.759223300970873e-05, "loss": 0.2742, "step": 4616 }, { "epoch": 44.76626506024096, "grad_norm": 20.545211791992188, "learning_rate": 7.758737864077669e-05, "loss": 0.3188, "step": 4617 }, { "epoch": 44.77590361445783, "grad_norm": 9.697029113769531, "learning_rate": 7.758252427184466e-05, "loss": 0.2558, "step": 4618 }, { "epoch": 44.7855421686747, "grad_norm": 8.236445426940918, "learning_rate": 7.757766990291262e-05, "loss": 0.1253, "step": 4619 }, { "epoch": 44.795180722891565, "grad_norm": 38.7866325378418, "learning_rate": 7.757281553398059e-05, "loss": 0.5251, "step": 4620 }, { "epoch": 44.804819277108436, "grad_norm": 6.23081636428833, "learning_rate": 7.756796116504855e-05, "loss": 0.2192, "step": 4621 }, { "epoch": 44.8144578313253, "grad_norm": 5.594736576080322, "learning_rate": 7.756310679611651e-05, "loss": 0.4514, "step": 4622 }, { "epoch": 44.82409638554217, "grad_norm": 16.24919319152832, "learning_rate": 7.755825242718448e-05, "loss": 0.4587, "step": 4623 }, { "epoch": 44.83373493975903, "grad_norm": 8.230205535888672, "learning_rate": 7.755339805825244e-05, "loss": 0.3533, "step": 4624 }, { "epoch": 44.8433734939759, "grad_norm": 10.193048477172852, "learning_rate": 7.75485436893204e-05, "loss": 0.3496, "step": 4625 }, { "epoch": 44.853012048192774, "grad_norm": 4.538084983825684, "learning_rate": 7.754368932038835e-05, "loss": 0.248, "step": 4626 }, { "epoch": 44.86265060240964, "grad_norm": 9.910898208618164, "learning_rate": 7.753883495145632e-05, "loss": 0.4587, "step": 4627 }, { "epoch": 44.87228915662651, "grad_norm": 5.540097713470459, "learning_rate": 7.753398058252428e-05, "loss": 0.226, "step": 4628 }, { "epoch": 44.88192771084337, "grad_norm": 35.35204315185547, "learning_rate": 7.752912621359224e-05, "loss": 0.2553, "step": 4629 }, { "epoch": 44.89156626506024, "grad_norm": 8.228525161743164, "learning_rate": 7.75242718446602e-05, "loss": 0.4014, "step": 4630 }, { "epoch": 44.90120481927711, "grad_norm": 8.635978698730469, "learning_rate": 7.751941747572816e-05, "loss": 0.448, "step": 4631 }, { "epoch": 44.910843373493975, "grad_norm": 11.161723136901855, "learning_rate": 7.751456310679613e-05, "loss": 0.4167, "step": 4632 }, { "epoch": 44.920481927710846, "grad_norm": 5.237980842590332, "learning_rate": 7.750970873786408e-05, "loss": 0.3123, "step": 4633 }, { "epoch": 44.93012048192771, "grad_norm": 5.503510475158691, "learning_rate": 7.750485436893204e-05, "loss": 0.3886, "step": 4634 }, { "epoch": 44.93975903614458, "grad_norm": 21.127119064331055, "learning_rate": 7.75e-05, "loss": 0.3696, "step": 4635 }, { "epoch": 44.94939759036144, "grad_norm": 7.685153961181641, "learning_rate": 7.749514563106797e-05, "loss": 0.1336, "step": 4636 }, { "epoch": 44.95903614457831, "grad_norm": 17.573951721191406, "learning_rate": 7.749029126213593e-05, "loss": 0.4743, "step": 4637 }, { "epoch": 44.96867469879518, "grad_norm": 4.928545951843262, "learning_rate": 7.748543689320389e-05, "loss": 0.3848, "step": 4638 }, { "epoch": 44.97831325301205, "grad_norm": 5.047237396240234, "learning_rate": 7.748058252427184e-05, "loss": 0.3651, "step": 4639 }, { "epoch": 44.98795180722892, "grad_norm": 28.81960678100586, "learning_rate": 7.747572815533982e-05, "loss": 0.3673, "step": 4640 }, { "epoch": 45.003614457831326, "grad_norm": 10.046671867370605, "learning_rate": 7.747087378640777e-05, "loss": 0.3563, "step": 4641 }, { "epoch": 45.013253012048196, "grad_norm": 15.415156364440918, "learning_rate": 7.746601941747573e-05, "loss": 0.4793, "step": 4642 }, { "epoch": 45.02289156626506, "grad_norm": 9.558704376220703, "learning_rate": 7.746116504854369e-05, "loss": 0.3351, "step": 4643 }, { "epoch": 45.03253012048193, "grad_norm": 3.7863194942474365, "learning_rate": 7.745631067961166e-05, "loss": 0.2752, "step": 4644 }, { "epoch": 45.04216867469879, "grad_norm": 5.472381114959717, "learning_rate": 7.745145631067962e-05, "loss": 0.3452, "step": 4645 }, { "epoch": 45.05180722891566, "grad_norm": 8.03809642791748, "learning_rate": 7.744660194174758e-05, "loss": 0.3124, "step": 4646 }, { "epoch": 45.06144578313253, "grad_norm": 17.505142211914062, "learning_rate": 7.744174757281553e-05, "loss": 0.3931, "step": 4647 }, { "epoch": 45.0710843373494, "grad_norm": 4.384190082550049, "learning_rate": 7.743689320388349e-05, "loss": 0.3427, "step": 4648 }, { "epoch": 45.08072289156627, "grad_norm": 7.163971900939941, "learning_rate": 7.743203883495146e-05, "loss": 0.3912, "step": 4649 }, { "epoch": 45.09036144578313, "grad_norm": 4.9158148765563965, "learning_rate": 7.742718446601942e-05, "loss": 0.3616, "step": 4650 }, { "epoch": 45.1, "grad_norm": 9.951620101928711, "learning_rate": 7.742233009708738e-05, "loss": 0.3227, "step": 4651 }, { "epoch": 45.109638554216865, "grad_norm": 26.395503997802734, "learning_rate": 7.741747572815534e-05, "loss": 0.4023, "step": 4652 }, { "epoch": 45.119277108433735, "grad_norm": 4.840840816497803, "learning_rate": 7.741262135922331e-05, "loss": 0.2719, "step": 4653 }, { "epoch": 45.128915662650606, "grad_norm": 12.584598541259766, "learning_rate": 7.740776699029127e-05, "loss": 0.3844, "step": 4654 }, { "epoch": 45.13855421686747, "grad_norm": 59.11317825317383, "learning_rate": 7.740291262135922e-05, "loss": 0.2734, "step": 4655 }, { "epoch": 45.14819277108434, "grad_norm": 11.40229606628418, "learning_rate": 7.739805825242718e-05, "loss": 0.2746, "step": 4656 }, { "epoch": 45.1578313253012, "grad_norm": 7.337160587310791, "learning_rate": 7.739320388349515e-05, "loss": 0.409, "step": 4657 }, { "epoch": 45.16746987951807, "grad_norm": 6.160838603973389, "learning_rate": 7.738834951456311e-05, "loss": 0.1609, "step": 4658 }, { "epoch": 45.17710843373494, "grad_norm": 6.44001579284668, "learning_rate": 7.738349514563107e-05, "loss": 0.3692, "step": 4659 }, { "epoch": 45.18674698795181, "grad_norm": 14.004420280456543, "learning_rate": 7.737864077669903e-05, "loss": 0.3771, "step": 4660 }, { "epoch": 45.19638554216868, "grad_norm": 23.24720001220703, "learning_rate": 7.7373786407767e-05, "loss": 0.2417, "step": 4661 }, { "epoch": 45.20602409638554, "grad_norm": 10.391663551330566, "learning_rate": 7.736893203883495e-05, "loss": 0.3105, "step": 4662 }, { "epoch": 45.21566265060241, "grad_norm": 6.146125793457031, "learning_rate": 7.736407766990291e-05, "loss": 0.225, "step": 4663 }, { "epoch": 45.225301204819274, "grad_norm": 9.994850158691406, "learning_rate": 7.735922330097087e-05, "loss": 0.3239, "step": 4664 }, { "epoch": 45.234939759036145, "grad_norm": 20.957462310791016, "learning_rate": 7.735436893203883e-05, "loss": 0.1434, "step": 4665 }, { "epoch": 45.244578313253015, "grad_norm": 9.925419807434082, "learning_rate": 7.73495145631068e-05, "loss": 0.4376, "step": 4666 }, { "epoch": 45.25421686746988, "grad_norm": 9.5278959274292, "learning_rate": 7.734466019417476e-05, "loss": 0.286, "step": 4667 }, { "epoch": 45.26385542168675, "grad_norm": 5.136429309844971, "learning_rate": 7.733980582524271e-05, "loss": 0.2469, "step": 4668 }, { "epoch": 45.27349397590361, "grad_norm": 18.654834747314453, "learning_rate": 7.733495145631067e-05, "loss": 0.3703, "step": 4669 }, { "epoch": 45.28313253012048, "grad_norm": 19.296545028686523, "learning_rate": 7.733009708737864e-05, "loss": 0.36, "step": 4670 }, { "epoch": 45.292771084337346, "grad_norm": 6.963071346282959, "learning_rate": 7.732524271844661e-05, "loss": 0.226, "step": 4671 }, { "epoch": 45.30240963855422, "grad_norm": 12.040593147277832, "learning_rate": 7.732038834951457e-05, "loss": 0.3922, "step": 4672 }, { "epoch": 45.31204819277109, "grad_norm": 9.186380386352539, "learning_rate": 7.731553398058253e-05, "loss": 0.3106, "step": 4673 }, { "epoch": 45.32168674698795, "grad_norm": 3.6376779079437256, "learning_rate": 7.731067961165049e-05, "loss": 0.144, "step": 4674 }, { "epoch": 45.33132530120482, "grad_norm": 5.644275665283203, "learning_rate": 7.730582524271846e-05, "loss": 0.3874, "step": 4675 }, { "epoch": 45.340963855421684, "grad_norm": 20.997846603393555, "learning_rate": 7.730097087378642e-05, "loss": 0.3843, "step": 4676 }, { "epoch": 45.350602409638554, "grad_norm": 9.731096267700195, "learning_rate": 7.729611650485437e-05, "loss": 0.3638, "step": 4677 }, { "epoch": 45.360240963855425, "grad_norm": 8.424317359924316, "learning_rate": 7.729126213592233e-05, "loss": 0.3538, "step": 4678 }, { "epoch": 45.36987951807229, "grad_norm": 4.119542121887207, "learning_rate": 7.72864077669903e-05, "loss": 0.1176, "step": 4679 }, { "epoch": 45.37951807228916, "grad_norm": 55.229373931884766, "learning_rate": 7.728155339805826e-05, "loss": 0.2279, "step": 4680 }, { "epoch": 45.38915662650602, "grad_norm": 13.308351516723633, "learning_rate": 7.727669902912622e-05, "loss": 0.2786, "step": 4681 }, { "epoch": 45.39879518072289, "grad_norm": 8.420271873474121, "learning_rate": 7.727184466019418e-05, "loss": 0.3651, "step": 4682 }, { "epoch": 45.408433734939756, "grad_norm": 7.089565277099609, "learning_rate": 7.726699029126215e-05, "loss": 0.414, "step": 4683 }, { "epoch": 45.418072289156626, "grad_norm": 18.228418350219727, "learning_rate": 7.72621359223301e-05, "loss": 0.4795, "step": 4684 }, { "epoch": 45.4277108433735, "grad_norm": 4.908432960510254, "learning_rate": 7.725728155339806e-05, "loss": 0.2004, "step": 4685 }, { "epoch": 45.43734939759036, "grad_norm": 4.711953639984131, "learning_rate": 7.725242718446602e-05, "loss": 0.2048, "step": 4686 }, { "epoch": 45.44698795180723, "grad_norm": 12.990696907043457, "learning_rate": 7.724757281553398e-05, "loss": 0.2968, "step": 4687 }, { "epoch": 45.456626506024094, "grad_norm": 6.943695545196533, "learning_rate": 7.724271844660195e-05, "loss": 0.0961, "step": 4688 }, { "epoch": 45.466265060240964, "grad_norm": 10.33127212524414, "learning_rate": 7.723786407766991e-05, "loss": 0.2804, "step": 4689 }, { "epoch": 45.475903614457835, "grad_norm": 9.055458068847656, "learning_rate": 7.723300970873787e-05, "loss": 0.2192, "step": 4690 }, { "epoch": 45.4855421686747, "grad_norm": 10.031561851501465, "learning_rate": 7.722815533980582e-05, "loss": 0.5372, "step": 4691 }, { "epoch": 45.49518072289157, "grad_norm": 25.263132095336914, "learning_rate": 7.72233009708738e-05, "loss": 0.3293, "step": 4692 }, { "epoch": 45.50481927710843, "grad_norm": 15.959466934204102, "learning_rate": 7.721844660194175e-05, "loss": 0.4155, "step": 4693 }, { "epoch": 45.5144578313253, "grad_norm": 25.487329483032227, "learning_rate": 7.721359223300971e-05, "loss": 0.3362, "step": 4694 }, { "epoch": 45.524096385542165, "grad_norm": 10.231945991516113, "learning_rate": 7.720873786407767e-05, "loss": 0.2677, "step": 4695 }, { "epoch": 45.533734939759036, "grad_norm": 21.76970100402832, "learning_rate": 7.720388349514564e-05, "loss": 0.2602, "step": 4696 }, { "epoch": 45.543373493975906, "grad_norm": 3.874420642852783, "learning_rate": 7.71990291262136e-05, "loss": 0.298, "step": 4697 }, { "epoch": 45.55301204819277, "grad_norm": 4.686958312988281, "learning_rate": 7.719417475728156e-05, "loss": 0.2077, "step": 4698 }, { "epoch": 45.56265060240964, "grad_norm": 34.12770462036133, "learning_rate": 7.718932038834951e-05, "loss": 0.4068, "step": 4699 }, { "epoch": 45.5722891566265, "grad_norm": 15.528741836547852, "learning_rate": 7.718446601941748e-05, "loss": 0.4106, "step": 4700 }, { "epoch": 45.581927710843374, "grad_norm": 8.841434478759766, "learning_rate": 7.717961165048544e-05, "loss": 0.3279, "step": 4701 }, { "epoch": 45.591566265060244, "grad_norm": 10.839820861816406, "learning_rate": 7.71747572815534e-05, "loss": 0.2081, "step": 4702 }, { "epoch": 45.60120481927711, "grad_norm": 5.06611442565918, "learning_rate": 7.716990291262136e-05, "loss": 0.2918, "step": 4703 }, { "epoch": 45.61084337349398, "grad_norm": 20.112398147583008, "learning_rate": 7.716504854368932e-05, "loss": 0.4377, "step": 4704 }, { "epoch": 45.62048192771084, "grad_norm": 15.030588150024414, "learning_rate": 7.716019417475729e-05, "loss": 0.333, "step": 4705 }, { "epoch": 45.63012048192771, "grad_norm": 2.6236157417297363, "learning_rate": 7.715533980582524e-05, "loss": 0.1338, "step": 4706 }, { "epoch": 45.639759036144575, "grad_norm": 35.16373825073242, "learning_rate": 7.71504854368932e-05, "loss": 0.4447, "step": 4707 }, { "epoch": 45.649397590361446, "grad_norm": 4.803242206573486, "learning_rate": 7.714563106796116e-05, "loss": 0.2714, "step": 4708 }, { "epoch": 45.659036144578316, "grad_norm": 9.574975967407227, "learning_rate": 7.714077669902913e-05, "loss": 0.2529, "step": 4709 }, { "epoch": 45.66867469879518, "grad_norm": 4.2943315505981445, "learning_rate": 7.713592233009709e-05, "loss": 0.2128, "step": 4710 }, { "epoch": 45.67831325301205, "grad_norm": 6.715332984924316, "learning_rate": 7.713106796116505e-05, "loss": 0.2956, "step": 4711 }, { "epoch": 45.68795180722891, "grad_norm": 9.307597160339355, "learning_rate": 7.7126213592233e-05, "loss": 0.366, "step": 4712 }, { "epoch": 45.69759036144578, "grad_norm": 9.974461555480957, "learning_rate": 7.712135922330098e-05, "loss": 0.5059, "step": 4713 }, { "epoch": 45.707228915662654, "grad_norm": 26.38043212890625, "learning_rate": 7.711650485436893e-05, "loss": 0.4102, "step": 4714 }, { "epoch": 45.71686746987952, "grad_norm": 5.663285732269287, "learning_rate": 7.711165048543689e-05, "loss": 0.3932, "step": 4715 }, { "epoch": 45.72650602409639, "grad_norm": 7.896357536315918, "learning_rate": 7.710679611650485e-05, "loss": 0.2116, "step": 4716 }, { "epoch": 45.73614457831325, "grad_norm": 24.00981330871582, "learning_rate": 7.710194174757282e-05, "loss": 0.3034, "step": 4717 }, { "epoch": 45.74578313253012, "grad_norm": 12.70981216430664, "learning_rate": 7.709708737864078e-05, "loss": 0.3079, "step": 4718 }, { "epoch": 45.755421686746985, "grad_norm": 3.9624316692352295, "learning_rate": 7.709223300970874e-05, "loss": 0.2818, "step": 4719 }, { "epoch": 45.765060240963855, "grad_norm": 7.192288398742676, "learning_rate": 7.70873786407767e-05, "loss": 0.3445, "step": 4720 }, { "epoch": 45.774698795180726, "grad_norm": 7.647115230560303, "learning_rate": 7.708252427184467e-05, "loss": 0.2999, "step": 4721 }, { "epoch": 45.78433734939759, "grad_norm": 23.373615264892578, "learning_rate": 7.707766990291264e-05, "loss": 0.3257, "step": 4722 }, { "epoch": 45.79397590361446, "grad_norm": 14.084808349609375, "learning_rate": 7.70728155339806e-05, "loss": 0.1336, "step": 4723 }, { "epoch": 45.80361445783132, "grad_norm": 4.618678092956543, "learning_rate": 7.706796116504855e-05, "loss": 0.2224, "step": 4724 }, { "epoch": 45.81325301204819, "grad_norm": 5.343328952789307, "learning_rate": 7.706310679611651e-05, "loss": 0.2122, "step": 4725 }, { "epoch": 45.82289156626506, "grad_norm": 3.39003849029541, "learning_rate": 7.705825242718447e-05, "loss": 0.1473, "step": 4726 }, { "epoch": 45.83253012048193, "grad_norm": 3.2744078636169434, "learning_rate": 7.705339805825244e-05, "loss": 0.2073, "step": 4727 }, { "epoch": 45.8421686746988, "grad_norm": 6.98894739151001, "learning_rate": 7.70485436893204e-05, "loss": 0.2161, "step": 4728 }, { "epoch": 45.85180722891566, "grad_norm": 6.222350597381592, "learning_rate": 7.704368932038835e-05, "loss": 0.2768, "step": 4729 }, { "epoch": 45.86144578313253, "grad_norm": 22.514015197753906, "learning_rate": 7.703883495145631e-05, "loss": 0.2804, "step": 4730 }, { "epoch": 45.871084337349394, "grad_norm": 5.242111682891846, "learning_rate": 7.703398058252428e-05, "loss": 0.4019, "step": 4731 }, { "epoch": 45.880722891566265, "grad_norm": 23.532787322998047, "learning_rate": 7.702912621359224e-05, "loss": 0.2412, "step": 4732 }, { "epoch": 45.890361445783135, "grad_norm": 21.279281616210938, "learning_rate": 7.70242718446602e-05, "loss": 0.4695, "step": 4733 }, { "epoch": 45.9, "grad_norm": 10.35426139831543, "learning_rate": 7.701941747572816e-05, "loss": 0.2377, "step": 4734 }, { "epoch": 45.90963855421687, "grad_norm": 4.410060405731201, "learning_rate": 7.701456310679613e-05, "loss": 0.2753, "step": 4735 }, { "epoch": 45.91927710843373, "grad_norm": 15.348751068115234, "learning_rate": 7.700970873786409e-05, "loss": 0.3686, "step": 4736 }, { "epoch": 45.9289156626506, "grad_norm": 7.368531227111816, "learning_rate": 7.700485436893204e-05, "loss": 0.1783, "step": 4737 }, { "epoch": 45.93855421686747, "grad_norm": 5.072981357574463, "learning_rate": 7.7e-05, "loss": 0.2221, "step": 4738 }, { "epoch": 45.94819277108434, "grad_norm": 19.229568481445312, "learning_rate": 7.699514563106797e-05, "loss": 0.5631, "step": 4739 }, { "epoch": 45.95783132530121, "grad_norm": 7.714302062988281, "learning_rate": 7.699029126213593e-05, "loss": 0.1948, "step": 4740 }, { "epoch": 45.96746987951807, "grad_norm": 10.798008918762207, "learning_rate": 7.698543689320389e-05, "loss": 0.7247, "step": 4741 }, { "epoch": 45.97710843373494, "grad_norm": 11.576087951660156, "learning_rate": 7.698058252427185e-05, "loss": 0.272, "step": 4742 }, { "epoch": 45.986746987951804, "grad_norm": 11.548660278320312, "learning_rate": 7.69757281553398e-05, "loss": 0.2699, "step": 4743 }, { "epoch": 46.00240963855422, "grad_norm": 4.577508449554443, "learning_rate": 7.697087378640778e-05, "loss": 0.2275, "step": 4744 }, { "epoch": 46.01204819277108, "grad_norm": 12.689103126525879, "learning_rate": 7.696601941747573e-05, "loss": 0.4229, "step": 4745 }, { "epoch": 46.02168674698795, "grad_norm": 9.12517261505127, "learning_rate": 7.696116504854369e-05, "loss": 0.3993, "step": 4746 }, { "epoch": 46.03132530120482, "grad_norm": 9.250137329101562, "learning_rate": 7.695631067961165e-05, "loss": 0.3119, "step": 4747 }, { "epoch": 46.04096385542169, "grad_norm": 10.160589218139648, "learning_rate": 7.695145631067962e-05, "loss": 0.3491, "step": 4748 }, { "epoch": 46.05060240963856, "grad_norm": 19.502687454223633, "learning_rate": 7.694660194174758e-05, "loss": 0.4556, "step": 4749 }, { "epoch": 46.06024096385542, "grad_norm": 7.487708568572998, "learning_rate": 7.694174757281554e-05, "loss": 0.2698, "step": 4750 }, { "epoch": 46.06987951807229, "grad_norm": 9.598434448242188, "learning_rate": 7.693689320388349e-05, "loss": 0.2573, "step": 4751 }, { "epoch": 46.079518072289154, "grad_norm": 6.737821102142334, "learning_rate": 7.693203883495146e-05, "loss": 0.1445, "step": 4752 }, { "epoch": 46.089156626506025, "grad_norm": 8.229142189025879, "learning_rate": 7.692718446601942e-05, "loss": 0.4257, "step": 4753 }, { "epoch": 46.09879518072289, "grad_norm": 4.536818981170654, "learning_rate": 7.692233009708738e-05, "loss": 0.1632, "step": 4754 }, { "epoch": 46.10843373493976, "grad_norm": 7.324138641357422, "learning_rate": 7.691747572815534e-05, "loss": 0.3128, "step": 4755 }, { "epoch": 46.11807228915663, "grad_norm": 7.248870849609375, "learning_rate": 7.691262135922331e-05, "loss": 0.2962, "step": 4756 }, { "epoch": 46.12771084337349, "grad_norm": 12.329904556274414, "learning_rate": 7.690776699029127e-05, "loss": 0.3454, "step": 4757 }, { "epoch": 46.13734939759036, "grad_norm": 14.670171737670898, "learning_rate": 7.690291262135922e-05, "loss": 0.3417, "step": 4758 }, { "epoch": 46.146987951807226, "grad_norm": 3.087315797805786, "learning_rate": 7.689805825242718e-05, "loss": 0.1509, "step": 4759 }, { "epoch": 46.1566265060241, "grad_norm": 30.574129104614258, "learning_rate": 7.689320388349514e-05, "loss": 0.2532, "step": 4760 }, { "epoch": 46.16626506024097, "grad_norm": 4.286678314208984, "learning_rate": 7.688834951456311e-05, "loss": 0.3924, "step": 4761 }, { "epoch": 46.17590361445783, "grad_norm": 1.9077366590499878, "learning_rate": 7.688349514563107e-05, "loss": 0.0978, "step": 4762 }, { "epoch": 46.1855421686747, "grad_norm": 7.155038356781006, "learning_rate": 7.687864077669903e-05, "loss": 0.3062, "step": 4763 }, { "epoch": 46.195180722891564, "grad_norm": 59.65201950073242, "learning_rate": 7.687378640776698e-05, "loss": 0.3881, "step": 4764 }, { "epoch": 46.204819277108435, "grad_norm": 38.925785064697266, "learning_rate": 7.686893203883496e-05, "loss": 0.3567, "step": 4765 }, { "epoch": 46.2144578313253, "grad_norm": 7.297222137451172, "learning_rate": 7.686407766990291e-05, "loss": 0.2727, "step": 4766 }, { "epoch": 46.22409638554217, "grad_norm": 8.600992202758789, "learning_rate": 7.685922330097087e-05, "loss": 0.3363, "step": 4767 }, { "epoch": 46.23373493975904, "grad_norm": 5.184842586517334, "learning_rate": 7.685436893203883e-05, "loss": 0.2683, "step": 4768 }, { "epoch": 46.2433734939759, "grad_norm": 6.2600884437561035, "learning_rate": 7.68495145631068e-05, "loss": 0.2324, "step": 4769 }, { "epoch": 46.25301204819277, "grad_norm": 11.915632247924805, "learning_rate": 7.684466019417476e-05, "loss": 0.3481, "step": 4770 }, { "epoch": 46.262650602409636, "grad_norm": 17.24550437927246, "learning_rate": 7.683980582524272e-05, "loss": 0.3992, "step": 4771 }, { "epoch": 46.272289156626506, "grad_norm": 6.754281997680664, "learning_rate": 7.683495145631069e-05, "loss": 0.5396, "step": 4772 }, { "epoch": 46.28192771084338, "grad_norm": 6.379152774810791, "learning_rate": 7.683009708737864e-05, "loss": 0.195, "step": 4773 }, { "epoch": 46.29156626506024, "grad_norm": 36.022483825683594, "learning_rate": 7.682524271844662e-05, "loss": 0.6224, "step": 4774 }, { "epoch": 46.30120481927711, "grad_norm": 3.9837684631347656, "learning_rate": 7.682038834951457e-05, "loss": 0.1331, "step": 4775 }, { "epoch": 46.310843373493974, "grad_norm": 7.944559574127197, "learning_rate": 7.681553398058253e-05, "loss": 0.3752, "step": 4776 }, { "epoch": 46.320481927710844, "grad_norm": 7.867758274078369, "learning_rate": 7.681067961165049e-05, "loss": 0.5008, "step": 4777 }, { "epoch": 46.33012048192771, "grad_norm": 10.763291358947754, "learning_rate": 7.680582524271846e-05, "loss": 0.3443, "step": 4778 }, { "epoch": 46.33975903614458, "grad_norm": 7.469329357147217, "learning_rate": 7.680097087378642e-05, "loss": 0.3375, "step": 4779 }, { "epoch": 46.34939759036145, "grad_norm": 12.597834587097168, "learning_rate": 7.679611650485438e-05, "loss": 0.2792, "step": 4780 }, { "epoch": 46.35903614457831, "grad_norm": 9.120831489562988, "learning_rate": 7.679126213592233e-05, "loss": 0.4297, "step": 4781 }, { "epoch": 46.36867469879518, "grad_norm": 4.568827152252197, "learning_rate": 7.678640776699029e-05, "loss": 0.2321, "step": 4782 }, { "epoch": 46.378313253012045, "grad_norm": 25.75, "learning_rate": 7.678155339805826e-05, "loss": 0.4139, "step": 4783 }, { "epoch": 46.387951807228916, "grad_norm": 3.608969211578369, "learning_rate": 7.677669902912622e-05, "loss": 0.2364, "step": 4784 }, { "epoch": 46.397590361445786, "grad_norm": 7.300573348999023, "learning_rate": 7.677184466019418e-05, "loss": 0.286, "step": 4785 }, { "epoch": 46.40722891566265, "grad_norm": 15.061266899108887, "learning_rate": 7.676699029126214e-05, "loss": 0.3398, "step": 4786 }, { "epoch": 46.41686746987952, "grad_norm": 4.927909851074219, "learning_rate": 7.676213592233011e-05, "loss": 0.3915, "step": 4787 }, { "epoch": 46.42650602409638, "grad_norm": 23.791282653808594, "learning_rate": 7.675728155339807e-05, "loss": 0.5665, "step": 4788 }, { "epoch": 46.436144578313254, "grad_norm": 13.254396438598633, "learning_rate": 7.675242718446602e-05, "loss": 0.2148, "step": 4789 }, { "epoch": 46.44578313253012, "grad_norm": 27.7601261138916, "learning_rate": 7.674757281553398e-05, "loss": 0.298, "step": 4790 }, { "epoch": 46.45542168674699, "grad_norm": 4.158872604370117, "learning_rate": 7.674271844660195e-05, "loss": 0.2119, "step": 4791 }, { "epoch": 46.46506024096386, "grad_norm": 7.344633102416992, "learning_rate": 7.673786407766991e-05, "loss": 0.4466, "step": 4792 }, { "epoch": 46.47469879518072, "grad_norm": 19.462099075317383, "learning_rate": 7.673300970873787e-05, "loss": 0.2833, "step": 4793 }, { "epoch": 46.48433734939759, "grad_norm": 11.011359214782715, "learning_rate": 7.672815533980583e-05, "loss": 0.3598, "step": 4794 }, { "epoch": 46.493975903614455, "grad_norm": 8.78458309173584, "learning_rate": 7.67233009708738e-05, "loss": 0.3716, "step": 4795 }, { "epoch": 46.503614457831326, "grad_norm": 6.721675872802734, "learning_rate": 7.671844660194175e-05, "loss": 0.418, "step": 4796 }, { "epoch": 46.513253012048196, "grad_norm": 21.57415199279785, "learning_rate": 7.671359223300971e-05, "loss": 0.1955, "step": 4797 }, { "epoch": 46.52289156626506, "grad_norm": 8.277522087097168, "learning_rate": 7.670873786407767e-05, "loss": 0.3472, "step": 4798 }, { "epoch": 46.53253012048193, "grad_norm": 9.826462745666504, "learning_rate": 7.670388349514563e-05, "loss": 0.3181, "step": 4799 }, { "epoch": 46.54216867469879, "grad_norm": 11.60020637512207, "learning_rate": 7.66990291262136e-05, "loss": 0.3229, "step": 4800 }, { "epoch": 46.55180722891566, "grad_norm": 26.433042526245117, "learning_rate": 7.669417475728156e-05, "loss": 0.5572, "step": 4801 }, { "epoch": 46.56144578313253, "grad_norm": 9.456788063049316, "learning_rate": 7.668932038834951e-05, "loss": 0.3834, "step": 4802 }, { "epoch": 46.5710843373494, "grad_norm": 10.601426124572754, "learning_rate": 7.668446601941747e-05, "loss": 0.2631, "step": 4803 }, { "epoch": 46.58072289156627, "grad_norm": 18.407302856445312, "learning_rate": 7.667961165048544e-05, "loss": 0.3654, "step": 4804 }, { "epoch": 46.59036144578313, "grad_norm": 18.07807159423828, "learning_rate": 7.66747572815534e-05, "loss": 0.2194, "step": 4805 }, { "epoch": 46.6, "grad_norm": 6.161979675292969, "learning_rate": 7.666990291262136e-05, "loss": 0.3113, "step": 4806 }, { "epoch": 46.609638554216865, "grad_norm": 5.1578474044799805, "learning_rate": 7.666504854368932e-05, "loss": 0.1764, "step": 4807 }, { "epoch": 46.619277108433735, "grad_norm": 9.109139442443848, "learning_rate": 7.666019417475729e-05, "loss": 0.2097, "step": 4808 }, { "epoch": 46.628915662650606, "grad_norm": 7.363020420074463, "learning_rate": 7.665533980582525e-05, "loss": 0.3518, "step": 4809 }, { "epoch": 46.63855421686747, "grad_norm": 11.07642936706543, "learning_rate": 7.66504854368932e-05, "loss": 0.4857, "step": 4810 }, { "epoch": 46.64819277108434, "grad_norm": 6.357198238372803, "learning_rate": 7.664563106796116e-05, "loss": 0.1665, "step": 4811 }, { "epoch": 46.6578313253012, "grad_norm": 6.517282962799072, "learning_rate": 7.664077669902913e-05, "loss": 0.297, "step": 4812 }, { "epoch": 46.66746987951807, "grad_norm": 7.882625579833984, "learning_rate": 7.663592233009709e-05, "loss": 0.4205, "step": 4813 }, { "epoch": 46.67710843373494, "grad_norm": 14.59648323059082, "learning_rate": 7.663106796116505e-05, "loss": 0.2896, "step": 4814 }, { "epoch": 46.68674698795181, "grad_norm": 1.865039587020874, "learning_rate": 7.6626213592233e-05, "loss": 0.1425, "step": 4815 }, { "epoch": 46.69638554216868, "grad_norm": 6.165207386016846, "learning_rate": 7.662135922330096e-05, "loss": 0.3328, "step": 4816 }, { "epoch": 46.70602409638554, "grad_norm": 9.228484153747559, "learning_rate": 7.661650485436894e-05, "loss": 0.2766, "step": 4817 }, { "epoch": 46.71566265060241, "grad_norm": 9.435720443725586, "learning_rate": 7.661165048543689e-05, "loss": 0.312, "step": 4818 }, { "epoch": 46.725301204819274, "grad_norm": 25.65481948852539, "learning_rate": 7.660679611650485e-05, "loss": 0.3829, "step": 4819 }, { "epoch": 46.734939759036145, "grad_norm": 9.094706535339355, "learning_rate": 7.660194174757281e-05, "loss": 0.2374, "step": 4820 }, { "epoch": 46.744578313253015, "grad_norm": 11.083735466003418, "learning_rate": 7.659708737864078e-05, "loss": 0.6111, "step": 4821 }, { "epoch": 46.75421686746988, "grad_norm": 25.46224594116211, "learning_rate": 7.659223300970875e-05, "loss": 0.3576, "step": 4822 }, { "epoch": 46.76385542168675, "grad_norm": 12.306243896484375, "learning_rate": 7.658737864077671e-05, "loss": 0.2464, "step": 4823 }, { "epoch": 46.77349397590361, "grad_norm": 8.049965858459473, "learning_rate": 7.658252427184467e-05, "loss": 0.4683, "step": 4824 }, { "epoch": 46.78313253012048, "grad_norm": 68.15637969970703, "learning_rate": 7.657766990291262e-05, "loss": 0.4174, "step": 4825 }, { "epoch": 46.792771084337346, "grad_norm": 7.259721755981445, "learning_rate": 7.65728155339806e-05, "loss": 0.2211, "step": 4826 }, { "epoch": 46.80240963855422, "grad_norm": 4.4344706535339355, "learning_rate": 7.656796116504855e-05, "loss": 0.2571, "step": 4827 }, { "epoch": 46.81204819277109, "grad_norm": 41.946693420410156, "learning_rate": 7.656310679611651e-05, "loss": 0.3124, "step": 4828 }, { "epoch": 46.82168674698795, "grad_norm": 8.648445129394531, "learning_rate": 7.655825242718447e-05, "loss": 0.4738, "step": 4829 }, { "epoch": 46.83132530120482, "grad_norm": 10.158480644226074, "learning_rate": 7.655339805825244e-05, "loss": 0.3547, "step": 4830 }, { "epoch": 46.840963855421684, "grad_norm": 11.898297309875488, "learning_rate": 7.65485436893204e-05, "loss": 0.3807, "step": 4831 }, { "epoch": 46.850602409638554, "grad_norm": 11.433245658874512, "learning_rate": 7.654368932038836e-05, "loss": 0.2522, "step": 4832 }, { "epoch": 46.860240963855425, "grad_norm": 32.09307861328125, "learning_rate": 7.653883495145631e-05, "loss": 0.2102, "step": 4833 }, { "epoch": 46.86987951807229, "grad_norm": 36.58565902709961, "learning_rate": 7.653398058252428e-05, "loss": 0.3367, "step": 4834 }, { "epoch": 46.87951807228916, "grad_norm": 6.164538860321045, "learning_rate": 7.652912621359224e-05, "loss": 0.3502, "step": 4835 }, { "epoch": 46.88915662650602, "grad_norm": 18.9925479888916, "learning_rate": 7.65242718446602e-05, "loss": 0.3441, "step": 4836 }, { "epoch": 46.89879518072289, "grad_norm": 5.909270286560059, "learning_rate": 7.651941747572816e-05, "loss": 0.23, "step": 4837 }, { "epoch": 46.908433734939756, "grad_norm": 23.33896827697754, "learning_rate": 7.651456310679612e-05, "loss": 0.2543, "step": 4838 }, { "epoch": 46.918072289156626, "grad_norm": 12.247896194458008, "learning_rate": 7.650970873786409e-05, "loss": 0.5032, "step": 4839 }, { "epoch": 46.9277108433735, "grad_norm": 6.465395927429199, "learning_rate": 7.650485436893205e-05, "loss": 0.458, "step": 4840 }, { "epoch": 46.93734939759036, "grad_norm": 8.54661750793457, "learning_rate": 7.65e-05, "loss": 0.2577, "step": 4841 }, { "epoch": 46.94698795180723, "grad_norm": 5.2553300857543945, "learning_rate": 7.649514563106796e-05, "loss": 0.2111, "step": 4842 }, { "epoch": 46.956626506024094, "grad_norm": 27.606605529785156, "learning_rate": 7.649029126213593e-05, "loss": 0.3668, "step": 4843 }, { "epoch": 46.966265060240964, "grad_norm": 9.528993606567383, "learning_rate": 7.648543689320389e-05, "loss": 0.4714, "step": 4844 }, { "epoch": 46.975903614457835, "grad_norm": 8.437125205993652, "learning_rate": 7.648058252427185e-05, "loss": 0.4012, "step": 4845 }, { "epoch": 46.9855421686747, "grad_norm": 12.24316120147705, "learning_rate": 7.64757281553398e-05, "loss": 0.4344, "step": 4846 }, { "epoch": 47.001204819277106, "grad_norm": 6.511977672576904, "learning_rate": 7.647087378640778e-05, "loss": 0.3831, "step": 4847 }, { "epoch": 47.01084337349398, "grad_norm": 7.194410800933838, "learning_rate": 7.646601941747573e-05, "loss": 0.4045, "step": 4848 }, { "epoch": 47.02048192771084, "grad_norm": 13.231522560119629, "learning_rate": 7.646116504854369e-05, "loss": 0.1037, "step": 4849 }, { "epoch": 47.03012048192771, "grad_norm": 4.764149188995361, "learning_rate": 7.645631067961165e-05, "loss": 0.2737, "step": 4850 }, { "epoch": 47.03975903614458, "grad_norm": 18.0467472076416, "learning_rate": 7.645145631067962e-05, "loss": 0.3807, "step": 4851 }, { "epoch": 47.049397590361444, "grad_norm": 8.700337409973145, "learning_rate": 7.644660194174758e-05, "loss": 0.2831, "step": 4852 }, { "epoch": 47.059036144578315, "grad_norm": 25.067338943481445, "learning_rate": 7.644174757281554e-05, "loss": 0.2542, "step": 4853 }, { "epoch": 47.06867469879518, "grad_norm": 3.752527952194214, "learning_rate": 7.64368932038835e-05, "loss": 0.1871, "step": 4854 }, { "epoch": 47.07831325301205, "grad_norm": 12.858832359313965, "learning_rate": 7.643203883495145e-05, "loss": 0.4084, "step": 4855 }, { "epoch": 47.08795180722892, "grad_norm": 12.499919891357422, "learning_rate": 7.642718446601942e-05, "loss": 0.425, "step": 4856 }, { "epoch": 47.09759036144578, "grad_norm": 8.523588180541992, "learning_rate": 7.642233009708738e-05, "loss": 0.3232, "step": 4857 }, { "epoch": 47.10722891566265, "grad_norm": 13.493267059326172, "learning_rate": 7.641747572815534e-05, "loss": 0.2871, "step": 4858 }, { "epoch": 47.116867469879516, "grad_norm": 13.37389850616455, "learning_rate": 7.64126213592233e-05, "loss": 0.4963, "step": 4859 }, { "epoch": 47.126506024096386, "grad_norm": 8.171547889709473, "learning_rate": 7.640776699029127e-05, "loss": 0.3647, "step": 4860 }, { "epoch": 47.13614457831325, "grad_norm": 32.36699676513672, "learning_rate": 7.640291262135923e-05, "loss": 0.5692, "step": 4861 }, { "epoch": 47.14578313253012, "grad_norm": 5.357143878936768, "learning_rate": 7.639805825242718e-05, "loss": 0.2889, "step": 4862 }, { "epoch": 47.15542168674699, "grad_norm": 17.166868209838867, "learning_rate": 7.639320388349514e-05, "loss": 0.2222, "step": 4863 }, { "epoch": 47.165060240963854, "grad_norm": 12.811493873596191, "learning_rate": 7.638834951456311e-05, "loss": 0.2204, "step": 4864 }, { "epoch": 47.174698795180724, "grad_norm": 11.96480655670166, "learning_rate": 7.638349514563107e-05, "loss": 0.3134, "step": 4865 }, { "epoch": 47.18433734939759, "grad_norm": 20.397624969482422, "learning_rate": 7.637864077669903e-05, "loss": 0.4025, "step": 4866 }, { "epoch": 47.19397590361446, "grad_norm": 7.323365211486816, "learning_rate": 7.637378640776699e-05, "loss": 0.207, "step": 4867 }, { "epoch": 47.20361445783133, "grad_norm": 13.608860969543457, "learning_rate": 7.636893203883496e-05, "loss": 0.2634, "step": 4868 }, { "epoch": 47.21325301204819, "grad_norm": 20.585092544555664, "learning_rate": 7.636407766990291e-05, "loss": 0.3635, "step": 4869 }, { "epoch": 47.22289156626506, "grad_norm": 11.753555297851562, "learning_rate": 7.635922330097087e-05, "loss": 0.3484, "step": 4870 }, { "epoch": 47.232530120481925, "grad_norm": 12.979461669921875, "learning_rate": 7.635436893203883e-05, "loss": 0.2298, "step": 4871 }, { "epoch": 47.242168674698796, "grad_norm": 6.672684192657471, "learning_rate": 7.634951456310679e-05, "loss": 0.4589, "step": 4872 }, { "epoch": 47.25180722891566, "grad_norm": 9.209179878234863, "learning_rate": 7.634466019417477e-05, "loss": 0.377, "step": 4873 }, { "epoch": 47.26144578313253, "grad_norm": 9.058496475219727, "learning_rate": 7.633980582524273e-05, "loss": 0.3664, "step": 4874 }, { "epoch": 47.2710843373494, "grad_norm": 6.123219966888428, "learning_rate": 7.633495145631069e-05, "loss": 0.2028, "step": 4875 }, { "epoch": 47.28072289156626, "grad_norm": 4.920437335968018, "learning_rate": 7.633009708737865e-05, "loss": 0.1299, "step": 4876 }, { "epoch": 47.290361445783134, "grad_norm": 11.162537574768066, "learning_rate": 7.63252427184466e-05, "loss": 0.3966, "step": 4877 }, { "epoch": 47.3, "grad_norm": 6.685653209686279, "learning_rate": 7.632038834951458e-05, "loss": 0.3009, "step": 4878 }, { "epoch": 47.30963855421687, "grad_norm": 6.037079811096191, "learning_rate": 7.631553398058253e-05, "loss": 0.2895, "step": 4879 }, { "epoch": 47.31927710843374, "grad_norm": 6.885039806365967, "learning_rate": 7.631067961165049e-05, "loss": 0.2186, "step": 4880 }, { "epoch": 47.3289156626506, "grad_norm": 6.5388593673706055, "learning_rate": 7.630582524271845e-05, "loss": 0.2859, "step": 4881 }, { "epoch": 47.33855421686747, "grad_norm": 7.864712238311768, "learning_rate": 7.630097087378642e-05, "loss": 0.1815, "step": 4882 }, { "epoch": 47.348192771084335, "grad_norm": 10.855761528015137, "learning_rate": 7.629611650485438e-05, "loss": 0.3908, "step": 4883 }, { "epoch": 47.357831325301206, "grad_norm": 26.317039489746094, "learning_rate": 7.629126213592234e-05, "loss": 0.3455, "step": 4884 }, { "epoch": 47.36746987951807, "grad_norm": 5.172886848449707, "learning_rate": 7.628640776699029e-05, "loss": 0.2488, "step": 4885 }, { "epoch": 47.37710843373494, "grad_norm": 6.776750564575195, "learning_rate": 7.628155339805826e-05, "loss": 0.2186, "step": 4886 }, { "epoch": 47.38674698795181, "grad_norm": 6.066345691680908, "learning_rate": 7.627669902912622e-05, "loss": 0.5294, "step": 4887 }, { "epoch": 47.39638554216867, "grad_norm": 15.221108436584473, "learning_rate": 7.627184466019418e-05, "loss": 0.4959, "step": 4888 }, { "epoch": 47.40602409638554, "grad_norm": 14.185855865478516, "learning_rate": 7.626699029126214e-05, "loss": 0.2284, "step": 4889 }, { "epoch": 47.41566265060241, "grad_norm": 13.02571964263916, "learning_rate": 7.626213592233011e-05, "loss": 0.3251, "step": 4890 }, { "epoch": 47.42530120481928, "grad_norm": 9.570011138916016, "learning_rate": 7.625728155339807e-05, "loss": 0.5118, "step": 4891 }, { "epoch": 47.43493975903615, "grad_norm": 5.698034286499023, "learning_rate": 7.625242718446602e-05, "loss": 0.2421, "step": 4892 }, { "epoch": 47.44457831325301, "grad_norm": 8.313006401062012, "learning_rate": 7.624757281553398e-05, "loss": 0.3412, "step": 4893 }, { "epoch": 47.45421686746988, "grad_norm": 3.504427194595337, "learning_rate": 7.624271844660194e-05, "loss": 0.2212, "step": 4894 }, { "epoch": 47.463855421686745, "grad_norm": 8.452329635620117, "learning_rate": 7.623786407766991e-05, "loss": 0.3651, "step": 4895 }, { "epoch": 47.473493975903615, "grad_norm": 5.922992706298828, "learning_rate": 7.623300970873787e-05, "loss": 0.4093, "step": 4896 }, { "epoch": 47.48313253012048, "grad_norm": 9.58006763458252, "learning_rate": 7.622815533980583e-05, "loss": 0.4638, "step": 4897 }, { "epoch": 47.49277108433735, "grad_norm": 25.375272750854492, "learning_rate": 7.622330097087378e-05, "loss": 0.2776, "step": 4898 }, { "epoch": 47.50240963855422, "grad_norm": 10.94299030303955, "learning_rate": 7.621844660194176e-05, "loss": 0.2113, "step": 4899 }, { "epoch": 47.51204819277108, "grad_norm": 6.961006164550781, "learning_rate": 7.621359223300971e-05, "loss": 0.3504, "step": 4900 }, { "epoch": 47.52168674698795, "grad_norm": 7.676585674285889, "learning_rate": 7.620873786407767e-05, "loss": 0.3528, "step": 4901 }, { "epoch": 47.53132530120482, "grad_norm": 7.165614604949951, "learning_rate": 7.620388349514563e-05, "loss": 0.1573, "step": 4902 }, { "epoch": 47.54096385542169, "grad_norm": 4.781308174133301, "learning_rate": 7.61990291262136e-05, "loss": 0.1695, "step": 4903 }, { "epoch": 47.55060240963856, "grad_norm": 3.669848918914795, "learning_rate": 7.619417475728156e-05, "loss": 0.1277, "step": 4904 }, { "epoch": 47.56024096385542, "grad_norm": 6.788479804992676, "learning_rate": 7.618932038834952e-05, "loss": 0.179, "step": 4905 }, { "epoch": 47.56987951807229, "grad_norm": 7.790316581726074, "learning_rate": 7.618446601941747e-05, "loss": 0.2533, "step": 4906 }, { "epoch": 47.579518072289154, "grad_norm": 7.235328197479248, "learning_rate": 7.617961165048545e-05, "loss": 0.3174, "step": 4907 }, { "epoch": 47.589156626506025, "grad_norm": 5.274322032928467, "learning_rate": 7.61747572815534e-05, "loss": 0.1887, "step": 4908 }, { "epoch": 47.59879518072289, "grad_norm": 10.291972160339355, "learning_rate": 7.616990291262136e-05, "loss": 0.3052, "step": 4909 }, { "epoch": 47.60843373493976, "grad_norm": 5.139192581176758, "learning_rate": 7.616504854368932e-05, "loss": 0.2819, "step": 4910 }, { "epoch": 47.61807228915663, "grad_norm": 8.981620788574219, "learning_rate": 7.616019417475728e-05, "loss": 0.323, "step": 4911 }, { "epoch": 47.62771084337349, "grad_norm": 9.367114067077637, "learning_rate": 7.615533980582525e-05, "loss": 0.4214, "step": 4912 }, { "epoch": 47.63734939759036, "grad_norm": 4.099111080169678, "learning_rate": 7.61504854368932e-05, "loss": 0.1683, "step": 4913 }, { "epoch": 47.646987951807226, "grad_norm": 5.013329029083252, "learning_rate": 7.614563106796116e-05, "loss": 0.3034, "step": 4914 }, { "epoch": 47.6566265060241, "grad_norm": 8.356827735900879, "learning_rate": 7.614077669902912e-05, "loss": 0.3161, "step": 4915 }, { "epoch": 47.66626506024097, "grad_norm": 11.531270027160645, "learning_rate": 7.613592233009709e-05, "loss": 0.4529, "step": 4916 }, { "epoch": 47.67590361445783, "grad_norm": 4.927653789520264, "learning_rate": 7.613106796116505e-05, "loss": 0.3589, "step": 4917 }, { "epoch": 47.6855421686747, "grad_norm": 26.113658905029297, "learning_rate": 7.612621359223301e-05, "loss": 0.2568, "step": 4918 }, { "epoch": 47.695180722891564, "grad_norm": 3.984858751296997, "learning_rate": 7.612135922330097e-05, "loss": 0.1503, "step": 4919 }, { "epoch": 47.704819277108435, "grad_norm": 5.809622764587402, "learning_rate": 7.611650485436894e-05, "loss": 0.1639, "step": 4920 }, { "epoch": 47.7144578313253, "grad_norm": 6.335975646972656, "learning_rate": 7.61116504854369e-05, "loss": 0.2277, "step": 4921 }, { "epoch": 47.72409638554217, "grad_norm": 5.999996662139893, "learning_rate": 7.610679611650485e-05, "loss": 0.2602, "step": 4922 }, { "epoch": 47.73373493975904, "grad_norm": 9.414304733276367, "learning_rate": 7.610194174757282e-05, "loss": 0.3716, "step": 4923 }, { "epoch": 47.7433734939759, "grad_norm": 5.866245269775391, "learning_rate": 7.609708737864078e-05, "loss": 0.3885, "step": 4924 }, { "epoch": 47.75301204819277, "grad_norm": 6.492210865020752, "learning_rate": 7.609223300970875e-05, "loss": 0.335, "step": 4925 }, { "epoch": 47.762650602409636, "grad_norm": 6.64755392074585, "learning_rate": 7.608737864077671e-05, "loss": 0.3357, "step": 4926 }, { "epoch": 47.772289156626506, "grad_norm": 4.171189785003662, "learning_rate": 7.608252427184467e-05, "loss": 0.128, "step": 4927 }, { "epoch": 47.78192771084338, "grad_norm": 8.140571594238281, "learning_rate": 7.607766990291263e-05, "loss": 0.5001, "step": 4928 }, { "epoch": 47.79156626506024, "grad_norm": 8.532171249389648, "learning_rate": 7.607281553398058e-05, "loss": 0.4104, "step": 4929 }, { "epoch": 47.80120481927711, "grad_norm": 10.22121524810791, "learning_rate": 7.606796116504855e-05, "loss": 0.2995, "step": 4930 }, { "epoch": 47.810843373493974, "grad_norm": 11.126678466796875, "learning_rate": 7.606310679611651e-05, "loss": 0.3617, "step": 4931 }, { "epoch": 47.820481927710844, "grad_norm": 5.446324825286865, "learning_rate": 7.605825242718447e-05, "loss": 0.4669, "step": 4932 }, { "epoch": 47.83012048192771, "grad_norm": 24.726797103881836, "learning_rate": 7.605339805825243e-05, "loss": 0.5446, "step": 4933 }, { "epoch": 47.83975903614458, "grad_norm": 25.06206703186035, "learning_rate": 7.60485436893204e-05, "loss": 0.4952, "step": 4934 }, { "epoch": 47.84939759036145, "grad_norm": 13.043954849243164, "learning_rate": 7.604368932038836e-05, "loss": 0.4769, "step": 4935 }, { "epoch": 47.85903614457831, "grad_norm": 9.800690650939941, "learning_rate": 7.603883495145632e-05, "loss": 0.6255, "step": 4936 }, { "epoch": 47.86867469879518, "grad_norm": 8.64914608001709, "learning_rate": 7.603398058252427e-05, "loss": 0.5243, "step": 4937 }, { "epoch": 47.878313253012045, "grad_norm": 6.075289726257324, "learning_rate": 7.602912621359224e-05, "loss": 0.1885, "step": 4938 }, { "epoch": 47.887951807228916, "grad_norm": 16.015710830688477, "learning_rate": 7.60242718446602e-05, "loss": 0.2884, "step": 4939 }, { "epoch": 47.897590361445786, "grad_norm": 16.531681060791016, "learning_rate": 7.601941747572816e-05, "loss": 0.2349, "step": 4940 }, { "epoch": 47.90722891566265, "grad_norm": 5.721368312835693, "learning_rate": 7.601456310679612e-05, "loss": 0.2175, "step": 4941 }, { "epoch": 47.91686746987952, "grad_norm": 8.220489501953125, "learning_rate": 7.600970873786409e-05, "loss": 0.3052, "step": 4942 }, { "epoch": 47.92650602409638, "grad_norm": 9.852360725402832, "learning_rate": 7.600485436893205e-05, "loss": 0.3313, "step": 4943 }, { "epoch": 47.936144578313254, "grad_norm": 9.699403762817383, "learning_rate": 7.6e-05, "loss": 0.4282, "step": 4944 }, { "epoch": 47.94578313253012, "grad_norm": 4.672789096832275, "learning_rate": 7.599514563106796e-05, "loss": 0.1883, "step": 4945 }, { "epoch": 47.95542168674699, "grad_norm": 8.826748847961426, "learning_rate": 7.599029126213592e-05, "loss": 0.4103, "step": 4946 }, { "epoch": 47.96506024096386, "grad_norm": 12.00413990020752, "learning_rate": 7.598543689320389e-05, "loss": 0.3022, "step": 4947 }, { "epoch": 47.97469879518072, "grad_norm": 18.16322898864746, "learning_rate": 7.598058252427185e-05, "loss": 0.1586, "step": 4948 }, { "epoch": 47.98433734939759, "grad_norm": 13.812177658081055, "learning_rate": 7.59757281553398e-05, "loss": 0.3861, "step": 4949 }, { "epoch": 47.993975903614455, "grad_norm": 20.733291625976562, "learning_rate": 7.597087378640776e-05, "loss": 0.3932, "step": 4950 }, { "epoch": 48.00963855421687, "grad_norm": 9.9173583984375, "learning_rate": 7.596601941747574e-05, "loss": 0.5236, "step": 4951 }, { "epoch": 48.019277108433734, "grad_norm": 4.960967063903809, "learning_rate": 7.59611650485437e-05, "loss": 0.1506, "step": 4952 }, { "epoch": 48.028915662650604, "grad_norm": 4.25662899017334, "learning_rate": 7.595631067961165e-05, "loss": 0.1599, "step": 4953 }, { "epoch": 48.03855421686747, "grad_norm": 7.020098686218262, "learning_rate": 7.595145631067961e-05, "loss": 0.31, "step": 4954 }, { "epoch": 48.04819277108434, "grad_norm": 10.336992263793945, "learning_rate": 7.594660194174758e-05, "loss": 0.4554, "step": 4955 }, { "epoch": 48.0578313253012, "grad_norm": 14.860848426818848, "learning_rate": 7.594174757281554e-05, "loss": 0.3927, "step": 4956 }, { "epoch": 48.06746987951807, "grad_norm": 8.128817558288574, "learning_rate": 7.59368932038835e-05, "loss": 0.1134, "step": 4957 }, { "epoch": 48.07710843373494, "grad_norm": 8.414308547973633, "learning_rate": 7.593203883495145e-05, "loss": 0.2618, "step": 4958 }, { "epoch": 48.086746987951805, "grad_norm": 11.423940658569336, "learning_rate": 7.592718446601942e-05, "loss": 0.4627, "step": 4959 }, { "epoch": 48.096385542168676, "grad_norm": 10.576011657714844, "learning_rate": 7.592233009708738e-05, "loss": 0.2156, "step": 4960 }, { "epoch": 48.10602409638554, "grad_norm": 7.70115327835083, "learning_rate": 7.591747572815534e-05, "loss": 0.2113, "step": 4961 }, { "epoch": 48.11566265060241, "grad_norm": 7.331251621246338, "learning_rate": 7.59126213592233e-05, "loss": 0.2184, "step": 4962 }, { "epoch": 48.12530120481928, "grad_norm": 17.661563873291016, "learning_rate": 7.590776699029126e-05, "loss": 0.5808, "step": 4963 }, { "epoch": 48.13493975903614, "grad_norm": 6.262159824371338, "learning_rate": 7.590291262135923e-05, "loss": 0.2098, "step": 4964 }, { "epoch": 48.144578313253014, "grad_norm": 17.499208450317383, "learning_rate": 7.589805825242718e-05, "loss": 0.3611, "step": 4965 }, { "epoch": 48.15421686746988, "grad_norm": 7.450377464294434, "learning_rate": 7.589320388349514e-05, "loss": 0.2899, "step": 4966 }, { "epoch": 48.16385542168675, "grad_norm": 4.277986526489258, "learning_rate": 7.58883495145631e-05, "loss": 0.3006, "step": 4967 }, { "epoch": 48.17349397590361, "grad_norm": 14.242559432983398, "learning_rate": 7.588349514563107e-05, "loss": 0.2814, "step": 4968 }, { "epoch": 48.18313253012048, "grad_norm": 11.477522850036621, "learning_rate": 7.587864077669903e-05, "loss": 0.1774, "step": 4969 }, { "epoch": 48.19277108433735, "grad_norm": 12.057822227478027, "learning_rate": 7.587378640776699e-05, "loss": 0.4397, "step": 4970 }, { "epoch": 48.202409638554215, "grad_norm": 6.787048816680908, "learning_rate": 7.586893203883494e-05, "loss": 0.2628, "step": 4971 }, { "epoch": 48.212048192771086, "grad_norm": 19.511451721191406, "learning_rate": 7.586407766990292e-05, "loss": 0.3408, "step": 4972 }, { "epoch": 48.22168674698795, "grad_norm": 17.018218994140625, "learning_rate": 7.585922330097087e-05, "loss": 0.3993, "step": 4973 }, { "epoch": 48.23132530120482, "grad_norm": 6.828583717346191, "learning_rate": 7.585436893203885e-05, "loss": 0.3749, "step": 4974 }, { "epoch": 48.24096385542169, "grad_norm": 5.38796854019165, "learning_rate": 7.58495145631068e-05, "loss": 0.2819, "step": 4975 }, { "epoch": 48.25060240963855, "grad_norm": 10.302882194519043, "learning_rate": 7.584466019417476e-05, "loss": 0.4683, "step": 4976 }, { "epoch": 48.26024096385542, "grad_norm": 12.566476821899414, "learning_rate": 7.583980582524273e-05, "loss": 0.4344, "step": 4977 }, { "epoch": 48.26987951807229, "grad_norm": 17.108957290649414, "learning_rate": 7.583495145631069e-05, "loss": 0.4161, "step": 4978 }, { "epoch": 48.27951807228916, "grad_norm": 8.325989723205566, "learning_rate": 7.583009708737865e-05, "loss": 0.3043, "step": 4979 }, { "epoch": 48.28915662650602, "grad_norm": 8.03777027130127, "learning_rate": 7.58252427184466e-05, "loss": 0.352, "step": 4980 }, { "epoch": 48.29879518072289, "grad_norm": 6.1020355224609375, "learning_rate": 7.582038834951458e-05, "loss": 0.1863, "step": 4981 }, { "epoch": 48.30843373493976, "grad_norm": 9.04576301574707, "learning_rate": 7.581553398058253e-05, "loss": 0.3914, "step": 4982 }, { "epoch": 48.318072289156625, "grad_norm": 10.165420532226562, "learning_rate": 7.581067961165049e-05, "loss": 0.3712, "step": 4983 }, { "epoch": 48.327710843373495, "grad_norm": 6.838231563568115, "learning_rate": 7.580582524271845e-05, "loss": 0.3706, "step": 4984 }, { "epoch": 48.33734939759036, "grad_norm": 8.19084358215332, "learning_rate": 7.580097087378641e-05, "loss": 0.2733, "step": 4985 }, { "epoch": 48.34698795180723, "grad_norm": 7.320016860961914, "learning_rate": 7.579611650485438e-05, "loss": 0.2261, "step": 4986 }, { "epoch": 48.3566265060241, "grad_norm": 11.467435836791992, "learning_rate": 7.579126213592234e-05, "loss": 0.7099, "step": 4987 }, { "epoch": 48.36626506024096, "grad_norm": 7.621162414550781, "learning_rate": 7.57864077669903e-05, "loss": 0.2376, "step": 4988 }, { "epoch": 48.37590361445783, "grad_norm": 7.612390041351318, "learning_rate": 7.578155339805825e-05, "loss": 0.547, "step": 4989 }, { "epoch": 48.3855421686747, "grad_norm": 8.663681030273438, "learning_rate": 7.577669902912622e-05, "loss": 0.2827, "step": 4990 }, { "epoch": 48.39518072289157, "grad_norm": 16.780099868774414, "learning_rate": 7.577184466019418e-05, "loss": 0.5171, "step": 4991 }, { "epoch": 48.40481927710843, "grad_norm": 5.926018238067627, "learning_rate": 7.576699029126214e-05, "loss": 0.3364, "step": 4992 }, { "epoch": 48.4144578313253, "grad_norm": 5.679141998291016, "learning_rate": 7.57621359223301e-05, "loss": 0.313, "step": 4993 }, { "epoch": 48.42409638554217, "grad_norm": 24.960739135742188, "learning_rate": 7.575728155339807e-05, "loss": 0.4337, "step": 4994 }, { "epoch": 48.433734939759034, "grad_norm": 4.563435077667236, "learning_rate": 7.575242718446603e-05, "loss": 0.2722, "step": 4995 }, { "epoch": 48.443373493975905, "grad_norm": 5.482316970825195, "learning_rate": 7.574757281553398e-05, "loss": 0.3125, "step": 4996 }, { "epoch": 48.45301204819277, "grad_norm": 6.081084251403809, "learning_rate": 7.574271844660194e-05, "loss": 0.3536, "step": 4997 }, { "epoch": 48.46265060240964, "grad_norm": 7.877464771270752, "learning_rate": 7.573786407766991e-05, "loss": 0.4319, "step": 4998 }, { "epoch": 48.47228915662651, "grad_norm": 7.460249423980713, "learning_rate": 7.573300970873787e-05, "loss": 0.4128, "step": 4999 }, { "epoch": 48.48192771084337, "grad_norm": 26.38889503479004, "learning_rate": 7.572815533980583e-05, "loss": 0.6486, "step": 5000 }, { "epoch": 48.49156626506024, "grad_norm": 5.914973258972168, "learning_rate": 7.572330097087379e-05, "loss": 0.2777, "step": 5001 }, { "epoch": 48.501204819277106, "grad_norm": 14.397209167480469, "learning_rate": 7.571844660194174e-05, "loss": 0.3138, "step": 5002 }, { "epoch": 48.51084337349398, "grad_norm": 8.157575607299805, "learning_rate": 7.571359223300972e-05, "loss": 0.2015, "step": 5003 }, { "epoch": 48.52048192771084, "grad_norm": 23.657684326171875, "learning_rate": 7.570873786407767e-05, "loss": 0.3272, "step": 5004 }, { "epoch": 48.53012048192771, "grad_norm": 56.14508819580078, "learning_rate": 7.570388349514563e-05, "loss": 0.3693, "step": 5005 }, { "epoch": 48.53975903614458, "grad_norm": 9.274670600891113, "learning_rate": 7.569902912621359e-05, "loss": 0.2352, "step": 5006 }, { "epoch": 48.549397590361444, "grad_norm": 8.943968772888184, "learning_rate": 7.569417475728156e-05, "loss": 0.3671, "step": 5007 }, { "epoch": 48.559036144578315, "grad_norm": 24.33614730834961, "learning_rate": 7.568932038834952e-05, "loss": 0.3393, "step": 5008 }, { "epoch": 48.56867469879518, "grad_norm": 10.622578620910645, "learning_rate": 7.568446601941748e-05, "loss": 0.4663, "step": 5009 }, { "epoch": 48.57831325301205, "grad_norm": 11.23092269897461, "learning_rate": 7.567961165048543e-05, "loss": 0.2989, "step": 5010 }, { "epoch": 48.58795180722892, "grad_norm": 10.941829681396484, "learning_rate": 7.56747572815534e-05, "loss": 0.2028, "step": 5011 }, { "epoch": 48.59759036144578, "grad_norm": 12.078180313110352, "learning_rate": 7.566990291262136e-05, "loss": 0.2947, "step": 5012 }, { "epoch": 48.60722891566265, "grad_norm": 13.008063316345215, "learning_rate": 7.566504854368932e-05, "loss": 0.266, "step": 5013 }, { "epoch": 48.616867469879516, "grad_norm": 8.436932563781738, "learning_rate": 7.566019417475728e-05, "loss": 0.4203, "step": 5014 }, { "epoch": 48.626506024096386, "grad_norm": 9.864270210266113, "learning_rate": 7.565533980582525e-05, "loss": 0.3107, "step": 5015 }, { "epoch": 48.63614457831325, "grad_norm": 11.598262786865234, "learning_rate": 7.56504854368932e-05, "loss": 0.4457, "step": 5016 }, { "epoch": 48.64578313253012, "grad_norm": 6.436229228973389, "learning_rate": 7.564563106796116e-05, "loss": 0.2423, "step": 5017 }, { "epoch": 48.65542168674699, "grad_norm": 6.455815315246582, "learning_rate": 7.564077669902912e-05, "loss": 0.5049, "step": 5018 }, { "epoch": 48.665060240963854, "grad_norm": 7.718199729919434, "learning_rate": 7.563592233009708e-05, "loss": 0.2189, "step": 5019 }, { "epoch": 48.674698795180724, "grad_norm": 18.15264320373535, "learning_rate": 7.563106796116505e-05, "loss": 0.3287, "step": 5020 }, { "epoch": 48.68433734939759, "grad_norm": 10.997127532958984, "learning_rate": 7.562621359223301e-05, "loss": 0.2068, "step": 5021 }, { "epoch": 48.69397590361446, "grad_norm": 8.263200759887695, "learning_rate": 7.562135922330097e-05, "loss": 0.28, "step": 5022 }, { "epoch": 48.70361445783133, "grad_norm": 10.86503791809082, "learning_rate": 7.561650485436892e-05, "loss": 0.3457, "step": 5023 }, { "epoch": 48.71325301204819, "grad_norm": 5.807268142700195, "learning_rate": 7.56116504854369e-05, "loss": 0.1347, "step": 5024 }, { "epoch": 48.72289156626506, "grad_norm": 10.040094375610352, "learning_rate": 7.560679611650487e-05, "loss": 0.3538, "step": 5025 }, { "epoch": 48.732530120481925, "grad_norm": 5.0554399490356445, "learning_rate": 7.560194174757282e-05, "loss": 0.2884, "step": 5026 }, { "epoch": 48.742168674698796, "grad_norm": 24.253250122070312, "learning_rate": 7.559708737864078e-05, "loss": 0.6199, "step": 5027 }, { "epoch": 48.75180722891566, "grad_norm": 10.266837120056152, "learning_rate": 7.559223300970874e-05, "loss": 0.2882, "step": 5028 }, { "epoch": 48.76144578313253, "grad_norm": 7.624542236328125, "learning_rate": 7.558737864077671e-05, "loss": 0.4329, "step": 5029 }, { "epoch": 48.7710843373494, "grad_norm": 17.47589111328125, "learning_rate": 7.558252427184467e-05, "loss": 0.4484, "step": 5030 }, { "epoch": 48.78072289156626, "grad_norm": 11.268594741821289, "learning_rate": 7.557766990291263e-05, "loss": 0.3934, "step": 5031 }, { "epoch": 48.790361445783134, "grad_norm": 24.045927047729492, "learning_rate": 7.557281553398059e-05, "loss": 0.3068, "step": 5032 }, { "epoch": 48.8, "grad_norm": 21.333736419677734, "learning_rate": 7.556796116504856e-05, "loss": 0.2442, "step": 5033 }, { "epoch": 48.80963855421687, "grad_norm": 3.786625862121582, "learning_rate": 7.556310679611651e-05, "loss": 0.2298, "step": 5034 }, { "epoch": 48.81927710843374, "grad_norm": 21.367189407348633, "learning_rate": 7.555825242718447e-05, "loss": 0.4953, "step": 5035 }, { "epoch": 48.8289156626506, "grad_norm": 4.956246852874756, "learning_rate": 7.555339805825243e-05, "loss": 0.251, "step": 5036 }, { "epoch": 48.83855421686747, "grad_norm": 5.545447826385498, "learning_rate": 7.55485436893204e-05, "loss": 0.2473, "step": 5037 }, { "epoch": 48.848192771084335, "grad_norm": 4.2946014404296875, "learning_rate": 7.554368932038836e-05, "loss": 0.2155, "step": 5038 }, { "epoch": 48.857831325301206, "grad_norm": 5.154994964599609, "learning_rate": 7.553883495145632e-05, "loss": 0.1867, "step": 5039 }, { "epoch": 48.86746987951807, "grad_norm": 6.501673698425293, "learning_rate": 7.553398058252427e-05, "loss": 0.2288, "step": 5040 }, { "epoch": 48.87710843373494, "grad_norm": 8.005448341369629, "learning_rate": 7.552912621359223e-05, "loss": 0.265, "step": 5041 }, { "epoch": 48.88674698795181, "grad_norm": 7.029966354370117, "learning_rate": 7.55242718446602e-05, "loss": 0.4382, "step": 5042 }, { "epoch": 48.89638554216867, "grad_norm": 12.746597290039062, "learning_rate": 7.551941747572816e-05, "loss": 0.4575, "step": 5043 }, { "epoch": 48.90602409638554, "grad_norm": 6.068904876708984, "learning_rate": 7.551456310679612e-05, "loss": 0.1589, "step": 5044 }, { "epoch": 48.91566265060241, "grad_norm": 8.277040481567383, "learning_rate": 7.550970873786408e-05, "loss": 0.3113, "step": 5045 }, { "epoch": 48.92530120481928, "grad_norm": 6.074435710906982, "learning_rate": 7.550485436893205e-05, "loss": 0.343, "step": 5046 }, { "epoch": 48.93493975903615, "grad_norm": 6.0567145347595215, "learning_rate": 7.55e-05, "loss": 0.2345, "step": 5047 }, { "epoch": 48.94457831325301, "grad_norm": 7.202183723449707, "learning_rate": 7.549514563106796e-05, "loss": 0.1265, "step": 5048 }, { "epoch": 48.95421686746988, "grad_norm": 20.079347610473633, "learning_rate": 7.549029126213592e-05, "loss": 0.3843, "step": 5049 }, { "epoch": 48.963855421686745, "grad_norm": 12.981492042541504, "learning_rate": 7.548543689320389e-05, "loss": 0.1942, "step": 5050 }, { "epoch": 48.973493975903615, "grad_norm": 7.867464065551758, "learning_rate": 7.548058252427185e-05, "loss": 0.2265, "step": 5051 }, { "epoch": 48.98313253012048, "grad_norm": 7.11667013168335, "learning_rate": 7.547572815533981e-05, "loss": 0.4179, "step": 5052 }, { "epoch": 48.99277108433735, "grad_norm": 8.109174728393555, "learning_rate": 7.547087378640777e-05, "loss": 0.1991, "step": 5053 }, { "epoch": 49.00843373493976, "grad_norm": 5.601223945617676, "learning_rate": 7.546601941747574e-05, "loss": 0.2817, "step": 5054 }, { "epoch": 49.01807228915663, "grad_norm": 3.803972005844116, "learning_rate": 7.54611650485437e-05, "loss": 0.113, "step": 5055 }, { "epoch": 49.02771084337349, "grad_norm": 50.34025955200195, "learning_rate": 7.545631067961165e-05, "loss": 0.1709, "step": 5056 }, { "epoch": 49.03734939759036, "grad_norm": 3.646616220474243, "learning_rate": 7.545145631067961e-05, "loss": 0.2125, "step": 5057 }, { "epoch": 49.04698795180723, "grad_norm": 13.129755973815918, "learning_rate": 7.544660194174757e-05, "loss": 0.443, "step": 5058 }, { "epoch": 49.056626506024095, "grad_norm": 94.06072235107422, "learning_rate": 7.544174757281554e-05, "loss": 0.5193, "step": 5059 }, { "epoch": 49.066265060240966, "grad_norm": 14.267996788024902, "learning_rate": 7.54368932038835e-05, "loss": 0.2844, "step": 5060 }, { "epoch": 49.07590361445783, "grad_norm": 23.921533584594727, "learning_rate": 7.543203883495145e-05, "loss": 0.3409, "step": 5061 }, { "epoch": 49.0855421686747, "grad_norm": 26.892732620239258, "learning_rate": 7.542718446601941e-05, "loss": 0.4205, "step": 5062 }, { "epoch": 49.09518072289157, "grad_norm": 25.52305793762207, "learning_rate": 7.542233009708738e-05, "loss": 0.4448, "step": 5063 }, { "epoch": 49.10481927710843, "grad_norm": 22.1208553314209, "learning_rate": 7.541747572815534e-05, "loss": 0.2435, "step": 5064 }, { "epoch": 49.1144578313253, "grad_norm": 18.534412384033203, "learning_rate": 7.54126213592233e-05, "loss": 0.1891, "step": 5065 }, { "epoch": 49.12409638554217, "grad_norm": 4.399944305419922, "learning_rate": 7.540776699029126e-05, "loss": 0.3636, "step": 5066 }, { "epoch": 49.13373493975904, "grad_norm": 8.942761421203613, "learning_rate": 7.540291262135923e-05, "loss": 0.2961, "step": 5067 }, { "epoch": 49.1433734939759, "grad_norm": 5.392336368560791, "learning_rate": 7.539805825242719e-05, "loss": 0.3747, "step": 5068 }, { "epoch": 49.15301204819277, "grad_norm": 20.338876724243164, "learning_rate": 7.539320388349514e-05, "loss": 0.4032, "step": 5069 }, { "epoch": 49.16265060240964, "grad_norm": 1.5982916355133057, "learning_rate": 7.53883495145631e-05, "loss": 0.1307, "step": 5070 }, { "epoch": 49.172289156626505, "grad_norm": 19.756254196166992, "learning_rate": 7.538349514563107e-05, "loss": 0.5289, "step": 5071 }, { "epoch": 49.181927710843375, "grad_norm": 11.569816589355469, "learning_rate": 7.537864077669903e-05, "loss": 0.2204, "step": 5072 }, { "epoch": 49.19156626506024, "grad_norm": 20.551048278808594, "learning_rate": 7.537378640776699e-05, "loss": 0.2211, "step": 5073 }, { "epoch": 49.20120481927711, "grad_norm": 59.00621795654297, "learning_rate": 7.536893203883495e-05, "loss": 0.3421, "step": 5074 }, { "epoch": 49.21084337349398, "grad_norm": 49.967628479003906, "learning_rate": 7.536407766990292e-05, "loss": 0.2893, "step": 5075 }, { "epoch": 49.22048192771084, "grad_norm": 5.522302627563477, "learning_rate": 7.535922330097089e-05, "loss": 0.2397, "step": 5076 }, { "epoch": 49.23012048192771, "grad_norm": 6.279780864715576, "learning_rate": 7.535436893203885e-05, "loss": 0.4973, "step": 5077 }, { "epoch": 49.23975903614458, "grad_norm": 14.02269172668457, "learning_rate": 7.53495145631068e-05, "loss": 0.3896, "step": 5078 }, { "epoch": 49.24939759036145, "grad_norm": 46.73685073852539, "learning_rate": 7.534466019417476e-05, "loss": 0.4158, "step": 5079 }, { "epoch": 49.25903614457831, "grad_norm": 7.129205703735352, "learning_rate": 7.533980582524272e-05, "loss": 0.3434, "step": 5080 }, { "epoch": 49.26867469879518, "grad_norm": 3.6438586711883545, "learning_rate": 7.533495145631069e-05, "loss": 0.428, "step": 5081 }, { "epoch": 49.27831325301205, "grad_norm": 2.0456998348236084, "learning_rate": 7.533009708737865e-05, "loss": 0.1333, "step": 5082 }, { "epoch": 49.287951807228914, "grad_norm": 25.06875228881836, "learning_rate": 7.532524271844661e-05, "loss": 0.3246, "step": 5083 }, { "epoch": 49.297590361445785, "grad_norm": 5.900460243225098, "learning_rate": 7.532038834951456e-05, "loss": 0.3721, "step": 5084 }, { "epoch": 49.30722891566265, "grad_norm": 49.1645393371582, "learning_rate": 7.531553398058254e-05, "loss": 0.4617, "step": 5085 }, { "epoch": 49.31686746987952, "grad_norm": 15.043237686157227, "learning_rate": 7.53106796116505e-05, "loss": 0.42, "step": 5086 }, { "epoch": 49.32650602409639, "grad_norm": 8.952193260192871, "learning_rate": 7.530582524271845e-05, "loss": 0.2721, "step": 5087 }, { "epoch": 49.33614457831325, "grad_norm": 17.979013442993164, "learning_rate": 7.530097087378641e-05, "loss": 0.3365, "step": 5088 }, { "epoch": 49.34578313253012, "grad_norm": 23.41096305847168, "learning_rate": 7.529611650485438e-05, "loss": 0.4132, "step": 5089 }, { "epoch": 49.355421686746986, "grad_norm": 25.46953773498535, "learning_rate": 7.529126213592234e-05, "loss": 0.4742, "step": 5090 }, { "epoch": 49.36506024096386, "grad_norm": 16.628414154052734, "learning_rate": 7.52864077669903e-05, "loss": 0.3595, "step": 5091 }, { "epoch": 49.37469879518072, "grad_norm": 37.41392135620117, "learning_rate": 7.528155339805825e-05, "loss": 0.2338, "step": 5092 }, { "epoch": 49.38433734939759, "grad_norm": 3.859917163848877, "learning_rate": 7.527669902912623e-05, "loss": 0.4309, "step": 5093 }, { "epoch": 49.39397590361446, "grad_norm": 3.6250174045562744, "learning_rate": 7.527184466019418e-05, "loss": 0.2524, "step": 5094 }, { "epoch": 49.403614457831324, "grad_norm": 6.629832744598389, "learning_rate": 7.526699029126214e-05, "loss": 0.3983, "step": 5095 }, { "epoch": 49.413253012048195, "grad_norm": 47.41777038574219, "learning_rate": 7.52621359223301e-05, "loss": 0.3843, "step": 5096 }, { "epoch": 49.42289156626506, "grad_norm": 39.21114730834961, "learning_rate": 7.525728155339806e-05, "loss": 0.3617, "step": 5097 }, { "epoch": 49.43253012048193, "grad_norm": 8.403002738952637, "learning_rate": 7.525242718446603e-05, "loss": 0.4038, "step": 5098 }, { "epoch": 49.44216867469879, "grad_norm": 71.91637420654297, "learning_rate": 7.524757281553399e-05, "loss": 0.5298, "step": 5099 }, { "epoch": 49.45180722891566, "grad_norm": 10.288312911987305, "learning_rate": 7.524271844660194e-05, "loss": 0.1887, "step": 5100 }, { "epoch": 49.46144578313253, "grad_norm": 18.763566970825195, "learning_rate": 7.52378640776699e-05, "loss": 0.2036, "step": 5101 }, { "epoch": 49.471084337349396, "grad_norm": 1.3244184255599976, "learning_rate": 7.523300970873787e-05, "loss": 0.2479, "step": 5102 }, { "epoch": 49.480722891566266, "grad_norm": 4.036909580230713, "learning_rate": 7.522815533980583e-05, "loss": 0.3261, "step": 5103 }, { "epoch": 49.49036144578313, "grad_norm": 10.507697105407715, "learning_rate": 7.522330097087379e-05, "loss": 0.2287, "step": 5104 }, { "epoch": 49.5, "grad_norm": 6.35495662689209, "learning_rate": 7.521844660194175e-05, "loss": 0.387, "step": 5105 }, { "epoch": 49.50963855421687, "grad_norm": 26.62969207763672, "learning_rate": 7.521359223300972e-05, "loss": 0.4947, "step": 5106 }, { "epoch": 49.519277108433734, "grad_norm": 5.300750255584717, "learning_rate": 7.520873786407767e-05, "loss": 0.1111, "step": 5107 }, { "epoch": 49.528915662650604, "grad_norm": 36.03458023071289, "learning_rate": 7.520388349514563e-05, "loss": 0.3133, "step": 5108 }, { "epoch": 49.53855421686747, "grad_norm": 59.519500732421875, "learning_rate": 7.519902912621359e-05, "loss": 0.3331, "step": 5109 }, { "epoch": 49.54819277108434, "grad_norm": 19.66541290283203, "learning_rate": 7.519417475728156e-05, "loss": 0.2996, "step": 5110 }, { "epoch": 49.55783132530121, "grad_norm": 43.7995491027832, "learning_rate": 7.518932038834952e-05, "loss": 0.5299, "step": 5111 }, { "epoch": 49.56746987951807, "grad_norm": 54.439453125, "learning_rate": 7.518446601941748e-05, "loss": 0.6687, "step": 5112 }, { "epoch": 49.57710843373494, "grad_norm": 25.67877197265625, "learning_rate": 7.517961165048543e-05, "loss": 0.3681, "step": 5113 }, { "epoch": 49.586746987951805, "grad_norm": 10.21973705291748, "learning_rate": 7.517475728155339e-05, "loss": 0.2422, "step": 5114 }, { "epoch": 49.596385542168676, "grad_norm": 5.445949077606201, "learning_rate": 7.516990291262136e-05, "loss": 0.2684, "step": 5115 }, { "epoch": 49.60602409638554, "grad_norm": 13.109206199645996, "learning_rate": 7.516504854368932e-05, "loss": 0.2465, "step": 5116 }, { "epoch": 49.61566265060241, "grad_norm": 12.484163284301758, "learning_rate": 7.516019417475728e-05, "loss": 0.2234, "step": 5117 }, { "epoch": 49.62530120481928, "grad_norm": 21.031280517578125, "learning_rate": 7.515533980582524e-05, "loss": 0.2, "step": 5118 }, { "epoch": 49.63493975903614, "grad_norm": 12.177804946899414, "learning_rate": 7.515048543689321e-05, "loss": 0.2979, "step": 5119 }, { "epoch": 49.644578313253014, "grad_norm": 2.938734292984009, "learning_rate": 7.514563106796117e-05, "loss": 0.3252, "step": 5120 }, { "epoch": 49.65421686746988, "grad_norm": 2.365003824234009, "learning_rate": 7.514077669902912e-05, "loss": 0.4091, "step": 5121 }, { "epoch": 49.66385542168675, "grad_norm": 0.8750266432762146, "learning_rate": 7.513592233009708e-05, "loss": 0.1783, "step": 5122 }, { "epoch": 49.67349397590361, "grad_norm": 16.395036697387695, "learning_rate": 7.513106796116505e-05, "loss": 0.3553, "step": 5123 }, { "epoch": 49.68313253012048, "grad_norm": 7.050961494445801, "learning_rate": 7.512621359223301e-05, "loss": 0.1854, "step": 5124 }, { "epoch": 49.69277108433735, "grad_norm": 14.335088729858398, "learning_rate": 7.512135922330098e-05, "loss": 0.3523, "step": 5125 }, { "epoch": 49.702409638554215, "grad_norm": 28.2888126373291, "learning_rate": 7.511650485436894e-05, "loss": 0.4079, "step": 5126 }, { "epoch": 49.712048192771086, "grad_norm": 14.16639518737793, "learning_rate": 7.51116504854369e-05, "loss": 0.3135, "step": 5127 }, { "epoch": 49.72168674698795, "grad_norm": 15.495466232299805, "learning_rate": 7.510679611650487e-05, "loss": 0.5838, "step": 5128 }, { "epoch": 49.73132530120482, "grad_norm": 29.25273895263672, "learning_rate": 7.510194174757283e-05, "loss": 0.2903, "step": 5129 }, { "epoch": 49.74096385542169, "grad_norm": 61.18498992919922, "learning_rate": 7.509708737864078e-05, "loss": 0.3684, "step": 5130 }, { "epoch": 49.75060240963855, "grad_norm": 14.894485473632812, "learning_rate": 7.509223300970874e-05, "loss": 0.3039, "step": 5131 }, { "epoch": 49.76024096385542, "grad_norm": 17.56907081604004, "learning_rate": 7.508737864077671e-05, "loss": 0.2799, "step": 5132 }, { "epoch": 49.76987951807229, "grad_norm": 10.219970703125, "learning_rate": 7.508252427184467e-05, "loss": 0.4189, "step": 5133 }, { "epoch": 49.77951807228916, "grad_norm": 47.9036865234375, "learning_rate": 7.507766990291263e-05, "loss": 0.5693, "step": 5134 }, { "epoch": 49.78915662650603, "grad_norm": 10.488384246826172, "learning_rate": 7.507281553398059e-05, "loss": 0.444, "step": 5135 }, { "epoch": 49.79879518072289, "grad_norm": 18.872312545776367, "learning_rate": 7.506796116504854e-05, "loss": 0.4811, "step": 5136 }, { "epoch": 49.80843373493976, "grad_norm": 35.12120819091797, "learning_rate": 7.506310679611652e-05, "loss": 0.4388, "step": 5137 }, { "epoch": 49.818072289156625, "grad_norm": 5.6900739669799805, "learning_rate": 7.505825242718447e-05, "loss": 0.26, "step": 5138 }, { "epoch": 49.827710843373495, "grad_norm": 15.351757049560547, "learning_rate": 7.505339805825243e-05, "loss": 0.2175, "step": 5139 }, { "epoch": 49.83734939759036, "grad_norm": 24.504220962524414, "learning_rate": 7.504854368932039e-05, "loss": 0.1459, "step": 5140 }, { "epoch": 49.84698795180723, "grad_norm": 16.25591468811035, "learning_rate": 7.504368932038836e-05, "loss": 0.3737, "step": 5141 }, { "epoch": 49.8566265060241, "grad_norm": 5.752495765686035, "learning_rate": 7.503883495145632e-05, "loss": 0.2285, "step": 5142 }, { "epoch": 49.86626506024096, "grad_norm": 17.121185302734375, "learning_rate": 7.503398058252428e-05, "loss": 0.3887, "step": 5143 }, { "epoch": 49.87590361445783, "grad_norm": 17.796327590942383, "learning_rate": 7.502912621359223e-05, "loss": 0.3441, "step": 5144 }, { "epoch": 49.8855421686747, "grad_norm": 23.801677703857422, "learning_rate": 7.50242718446602e-05, "loss": 0.3439, "step": 5145 }, { "epoch": 49.89518072289157, "grad_norm": 3.3046844005584717, "learning_rate": 7.501941747572816e-05, "loss": 0.4251, "step": 5146 }, { "epoch": 49.90481927710843, "grad_norm": 5.827195644378662, "learning_rate": 7.501456310679612e-05, "loss": 0.2516, "step": 5147 }, { "epoch": 49.9144578313253, "grad_norm": 31.04408836364746, "learning_rate": 7.500970873786408e-05, "loss": 0.288, "step": 5148 }, { "epoch": 49.92409638554217, "grad_norm": 1.9879337549209595, "learning_rate": 7.500485436893205e-05, "loss": 0.3023, "step": 5149 }, { "epoch": 49.933734939759034, "grad_norm": 6.768683910369873, "learning_rate": 7.500000000000001e-05, "loss": 0.3375, "step": 5150 }, { "epoch": 49.943373493975905, "grad_norm": 11.484048843383789, "learning_rate": 7.499514563106796e-05, "loss": 0.3175, "step": 5151 }, { "epoch": 49.95301204819277, "grad_norm": 16.953989028930664, "learning_rate": 7.499029126213592e-05, "loss": 0.3063, "step": 5152 }, { "epoch": 49.96265060240964, "grad_norm": 54.40238952636719, "learning_rate": 7.498543689320388e-05, "loss": 0.3654, "step": 5153 }, { "epoch": 49.97228915662651, "grad_norm": 18.726594924926758, "learning_rate": 7.498058252427185e-05, "loss": 0.3088, "step": 5154 }, { "epoch": 49.98192771084337, "grad_norm": 16.228464126586914, "learning_rate": 7.497572815533981e-05, "loss": 0.157, "step": 5155 }, { "epoch": 49.99156626506024, "grad_norm": 20.731800079345703, "learning_rate": 7.497087378640777e-05, "loss": 0.567, "step": 5156 }, { "epoch": 50.00722891566265, "grad_norm": 14.113591194152832, "learning_rate": 7.496601941747572e-05, "loss": 0.3999, "step": 5157 }, { "epoch": 50.01686746987952, "grad_norm": 6.927577018737793, "learning_rate": 7.49611650485437e-05, "loss": 0.3869, "step": 5158 }, { "epoch": 50.026506024096385, "grad_norm": 21.99730682373047, "learning_rate": 7.495631067961165e-05, "loss": 0.3059, "step": 5159 }, { "epoch": 50.036144578313255, "grad_norm": 21.530609130859375, "learning_rate": 7.495145631067961e-05, "loss": 0.5012, "step": 5160 }, { "epoch": 50.04578313253012, "grad_norm": 5.850213527679443, "learning_rate": 7.494660194174757e-05, "loss": 0.4351, "step": 5161 }, { "epoch": 50.05542168674699, "grad_norm": 7.249020576477051, "learning_rate": 7.494174757281554e-05, "loss": 0.4405, "step": 5162 }, { "epoch": 50.06506024096385, "grad_norm": 10.174236297607422, "learning_rate": 7.49368932038835e-05, "loss": 0.2996, "step": 5163 }, { "epoch": 50.07469879518072, "grad_norm": 7.843321800231934, "learning_rate": 7.493203883495146e-05, "loss": 0.4344, "step": 5164 }, { "epoch": 50.08433734939759, "grad_norm": 18.523332595825195, "learning_rate": 7.492718446601941e-05, "loss": 0.3057, "step": 5165 }, { "epoch": 50.09397590361446, "grad_norm": 23.232656478881836, "learning_rate": 7.492233009708739e-05, "loss": 0.378, "step": 5166 }, { "epoch": 50.10361445783133, "grad_norm": 4.980871200561523, "learning_rate": 7.491747572815534e-05, "loss": 0.3403, "step": 5167 }, { "epoch": 50.11325301204819, "grad_norm": 15.712224006652832, "learning_rate": 7.49126213592233e-05, "loss": 0.2524, "step": 5168 }, { "epoch": 50.12289156626506, "grad_norm": 38.69635772705078, "learning_rate": 7.490776699029126e-05, "loss": 0.498, "step": 5169 }, { "epoch": 50.13253012048193, "grad_norm": 6.862545490264893, "learning_rate": 7.490291262135922e-05, "loss": 0.4431, "step": 5170 }, { "epoch": 50.142168674698794, "grad_norm": 25.68387794494629, "learning_rate": 7.489805825242719e-05, "loss": 0.3687, "step": 5171 }, { "epoch": 50.151807228915665, "grad_norm": 17.92797088623047, "learning_rate": 7.489320388349515e-05, "loss": 0.3648, "step": 5172 }, { "epoch": 50.16144578313253, "grad_norm": 6.988628387451172, "learning_rate": 7.48883495145631e-05, "loss": 0.5125, "step": 5173 }, { "epoch": 50.1710843373494, "grad_norm": 18.46043586730957, "learning_rate": 7.488349514563106e-05, "loss": 0.2367, "step": 5174 }, { "epoch": 50.18072289156626, "grad_norm": 8.474717140197754, "learning_rate": 7.487864077669903e-05, "loss": 0.3779, "step": 5175 }, { "epoch": 50.19036144578313, "grad_norm": 15.48946475982666, "learning_rate": 7.4873786407767e-05, "loss": 0.4448, "step": 5176 }, { "epoch": 50.2, "grad_norm": 9.004589080810547, "learning_rate": 7.486893203883496e-05, "loss": 0.3869, "step": 5177 }, { "epoch": 50.209638554216866, "grad_norm": 8.80899715423584, "learning_rate": 7.486407766990292e-05, "loss": 0.3258, "step": 5178 }, { "epoch": 50.21927710843374, "grad_norm": 5.577889919281006, "learning_rate": 7.485922330097088e-05, "loss": 0.3616, "step": 5179 }, { "epoch": 50.2289156626506, "grad_norm": 8.719786643981934, "learning_rate": 7.485436893203885e-05, "loss": 0.2143, "step": 5180 }, { "epoch": 50.23855421686747, "grad_norm": 82.28839111328125, "learning_rate": 7.48495145631068e-05, "loss": 0.4933, "step": 5181 }, { "epoch": 50.24819277108434, "grad_norm": 17.502046585083008, "learning_rate": 7.484466019417476e-05, "loss": 0.4147, "step": 5182 }, { "epoch": 50.257831325301204, "grad_norm": 16.847070693969727, "learning_rate": 7.483980582524272e-05, "loss": 0.4936, "step": 5183 }, { "epoch": 50.267469879518075, "grad_norm": 3.8430798053741455, "learning_rate": 7.483495145631069e-05, "loss": 0.1957, "step": 5184 }, { "epoch": 50.27710843373494, "grad_norm": 43.979698181152344, "learning_rate": 7.483009708737865e-05, "loss": 0.3173, "step": 5185 }, { "epoch": 50.28674698795181, "grad_norm": 9.730890274047852, "learning_rate": 7.482524271844661e-05, "loss": 0.2394, "step": 5186 }, { "epoch": 50.29638554216867, "grad_norm": 56.6920051574707, "learning_rate": 7.482038834951457e-05, "loss": 0.2642, "step": 5187 }, { "epoch": 50.30602409638554, "grad_norm": 62.732444763183594, "learning_rate": 7.481553398058254e-05, "loss": 0.4196, "step": 5188 }, { "epoch": 50.31566265060241, "grad_norm": 19.031734466552734, "learning_rate": 7.48106796116505e-05, "loss": 0.3537, "step": 5189 }, { "epoch": 50.325301204819276, "grad_norm": 5.396056652069092, "learning_rate": 7.480582524271845e-05, "loss": 0.2212, "step": 5190 }, { "epoch": 50.334939759036146, "grad_norm": 22.0457706451416, "learning_rate": 7.480097087378641e-05, "loss": 0.4036, "step": 5191 }, { "epoch": 50.34457831325301, "grad_norm": 14.047567367553711, "learning_rate": 7.479611650485437e-05, "loss": 0.5067, "step": 5192 }, { "epoch": 50.35421686746988, "grad_norm": 25.56201934814453, "learning_rate": 7.479126213592234e-05, "loss": 0.5358, "step": 5193 }, { "epoch": 50.36385542168675, "grad_norm": 35.62563705444336, "learning_rate": 7.47864077669903e-05, "loss": 0.4605, "step": 5194 }, { "epoch": 50.373493975903614, "grad_norm": 3.6699352264404297, "learning_rate": 7.478155339805826e-05, "loss": 0.3544, "step": 5195 }, { "epoch": 50.383132530120484, "grad_norm": 30.11778450012207, "learning_rate": 7.477669902912621e-05, "loss": 0.3193, "step": 5196 }, { "epoch": 50.39277108433735, "grad_norm": 95.06318664550781, "learning_rate": 7.477184466019418e-05, "loss": 0.4921, "step": 5197 }, { "epoch": 50.40240963855422, "grad_norm": 43.98596954345703, "learning_rate": 7.476699029126214e-05, "loss": 0.479, "step": 5198 }, { "epoch": 50.41204819277108, "grad_norm": 14.852743148803711, "learning_rate": 7.47621359223301e-05, "loss": 0.3563, "step": 5199 }, { "epoch": 50.42168674698795, "grad_norm": 22.15607261657715, "learning_rate": 7.475728155339806e-05, "loss": 0.3082, "step": 5200 }, { "epoch": 50.43132530120482, "grad_norm": 21.791379928588867, "learning_rate": 7.475242718446603e-05, "loss": 0.208, "step": 5201 }, { "epoch": 50.440963855421685, "grad_norm": 50.32787322998047, "learning_rate": 7.474757281553399e-05, "loss": 0.3343, "step": 5202 }, { "epoch": 50.450602409638556, "grad_norm": 15.419114112854004, "learning_rate": 7.474271844660194e-05, "loss": 0.2844, "step": 5203 }, { "epoch": 50.46024096385542, "grad_norm": 26.769012451171875, "learning_rate": 7.47378640776699e-05, "loss": 0.526, "step": 5204 }, { "epoch": 50.46987951807229, "grad_norm": 11.072179794311523, "learning_rate": 7.473300970873787e-05, "loss": 0.3866, "step": 5205 }, { "epoch": 50.47951807228916, "grad_norm": 5.472208499908447, "learning_rate": 7.472815533980583e-05, "loss": 0.3625, "step": 5206 }, { "epoch": 50.48915662650602, "grad_norm": 37.91725540161133, "learning_rate": 7.472330097087379e-05, "loss": 0.5274, "step": 5207 }, { "epoch": 50.498795180722894, "grad_norm": 2.523550271987915, "learning_rate": 7.471844660194175e-05, "loss": 0.4019, "step": 5208 }, { "epoch": 50.50843373493976, "grad_norm": 9.717705726623535, "learning_rate": 7.47135922330097e-05, "loss": 0.4426, "step": 5209 }, { "epoch": 50.51807228915663, "grad_norm": 15.720003128051758, "learning_rate": 7.470873786407768e-05, "loss": 0.4943, "step": 5210 }, { "epoch": 50.52771084337349, "grad_norm": 102.92877960205078, "learning_rate": 7.470388349514563e-05, "loss": 0.4015, "step": 5211 }, { "epoch": 50.53734939759036, "grad_norm": 10.440110206604004, "learning_rate": 7.469902912621359e-05, "loss": 0.3607, "step": 5212 }, { "epoch": 50.54698795180723, "grad_norm": 2.3808863162994385, "learning_rate": 7.469417475728155e-05, "loss": 0.1633, "step": 5213 }, { "epoch": 50.556626506024095, "grad_norm": 32.68952178955078, "learning_rate": 7.468932038834952e-05, "loss": 0.4474, "step": 5214 }, { "epoch": 50.566265060240966, "grad_norm": 25.84304428100586, "learning_rate": 7.468446601941748e-05, "loss": 0.2595, "step": 5215 }, { "epoch": 50.57590361445783, "grad_norm": 22.65744972229004, "learning_rate": 7.467961165048544e-05, "loss": 0.4331, "step": 5216 }, { "epoch": 50.5855421686747, "grad_norm": 4.8591814041137695, "learning_rate": 7.46747572815534e-05, "loss": 0.5153, "step": 5217 }, { "epoch": 50.59518072289157, "grad_norm": 11.683565139770508, "learning_rate": 7.466990291262136e-05, "loss": 0.1623, "step": 5218 }, { "epoch": 50.60481927710843, "grad_norm": 9.72556209564209, "learning_rate": 7.466504854368932e-05, "loss": 0.4066, "step": 5219 }, { "epoch": 50.6144578313253, "grad_norm": 4.254095554351807, "learning_rate": 7.466019417475728e-05, "loss": 0.4372, "step": 5220 }, { "epoch": 50.62409638554217, "grad_norm": 12.180986404418945, "learning_rate": 7.465533980582524e-05, "loss": 0.402, "step": 5221 }, { "epoch": 50.63373493975904, "grad_norm": 21.917695999145508, "learning_rate": 7.465048543689321e-05, "loss": 0.2101, "step": 5222 }, { "epoch": 50.6433734939759, "grad_norm": 4.22808313369751, "learning_rate": 7.464563106796117e-05, "loss": 0.3011, "step": 5223 }, { "epoch": 50.65301204819277, "grad_norm": 27.3529109954834, "learning_rate": 7.464077669902912e-05, "loss": 0.4049, "step": 5224 }, { "epoch": 50.66265060240964, "grad_norm": 7.411851406097412, "learning_rate": 7.463592233009708e-05, "loss": 0.308, "step": 5225 }, { "epoch": 50.672289156626505, "grad_norm": 12.894536972045898, "learning_rate": 7.463106796116505e-05, "loss": 0.4054, "step": 5226 }, { "epoch": 50.681927710843375, "grad_norm": 11.215344429016113, "learning_rate": 7.462621359223301e-05, "loss": 0.2281, "step": 5227 }, { "epoch": 50.69156626506024, "grad_norm": 17.135541915893555, "learning_rate": 7.462135922330098e-05, "loss": 0.1566, "step": 5228 }, { "epoch": 50.70120481927711, "grad_norm": 27.171709060668945, "learning_rate": 7.461650485436894e-05, "loss": 0.2605, "step": 5229 }, { "epoch": 50.71084337349397, "grad_norm": 11.591547012329102, "learning_rate": 7.46116504854369e-05, "loss": 0.2078, "step": 5230 }, { "epoch": 50.72048192771084, "grad_norm": 9.638972282409668, "learning_rate": 7.460679611650486e-05, "loss": 0.2985, "step": 5231 }, { "epoch": 50.73012048192771, "grad_norm": 1.784536600112915, "learning_rate": 7.460194174757283e-05, "loss": 0.1425, "step": 5232 }, { "epoch": 50.73975903614458, "grad_norm": 18.409832000732422, "learning_rate": 7.459708737864079e-05, "loss": 0.2667, "step": 5233 }, { "epoch": 50.74939759036145, "grad_norm": 29.17789077758789, "learning_rate": 7.459223300970874e-05, "loss": 0.536, "step": 5234 }, { "epoch": 50.75903614457831, "grad_norm": 2.9893293380737305, "learning_rate": 7.45873786407767e-05, "loss": 0.3012, "step": 5235 }, { "epoch": 50.76867469879518, "grad_norm": 37.74193572998047, "learning_rate": 7.458252427184467e-05, "loss": 0.3857, "step": 5236 }, { "epoch": 50.77831325301205, "grad_norm": 13.084625244140625, "learning_rate": 7.457766990291263e-05, "loss": 0.3656, "step": 5237 }, { "epoch": 50.787951807228914, "grad_norm": 26.249404907226562, "learning_rate": 7.457281553398059e-05, "loss": 0.4122, "step": 5238 }, { "epoch": 50.797590361445785, "grad_norm": 30.74405860900879, "learning_rate": 7.456796116504855e-05, "loss": 0.3268, "step": 5239 }, { "epoch": 50.80722891566265, "grad_norm": 8.967023849487305, "learning_rate": 7.456310679611652e-05, "loss": 0.3965, "step": 5240 }, { "epoch": 50.81686746987952, "grad_norm": 8.458280563354492, "learning_rate": 7.455825242718447e-05, "loss": 0.3839, "step": 5241 }, { "epoch": 50.82650602409639, "grad_norm": 5.97642183303833, "learning_rate": 7.455339805825243e-05, "loss": 0.3906, "step": 5242 }, { "epoch": 50.83614457831325, "grad_norm": 2.769089698791504, "learning_rate": 7.454854368932039e-05, "loss": 0.2878, "step": 5243 }, { "epoch": 50.84578313253012, "grad_norm": 166.94973754882812, "learning_rate": 7.454368932038835e-05, "loss": 0.6053, "step": 5244 }, { "epoch": 50.855421686746986, "grad_norm": 16.320688247680664, "learning_rate": 7.453883495145632e-05, "loss": 0.3662, "step": 5245 }, { "epoch": 50.86506024096386, "grad_norm": 200.76882934570312, "learning_rate": 7.453398058252428e-05, "loss": 0.3153, "step": 5246 }, { "epoch": 50.87469879518072, "grad_norm": 94.53239440917969, "learning_rate": 7.452912621359223e-05, "loss": 0.4942, "step": 5247 }, { "epoch": 50.88433734939759, "grad_norm": 81.24492645263672, "learning_rate": 7.452427184466019e-05, "loss": 0.3227, "step": 5248 }, { "epoch": 50.89397590361446, "grad_norm": 10.818285942077637, "learning_rate": 7.451941747572816e-05, "loss": 0.6116, "step": 5249 }, { "epoch": 50.903614457831324, "grad_norm": 26.056795120239258, "learning_rate": 7.451456310679612e-05, "loss": 0.4212, "step": 5250 }, { "epoch": 50.913253012048195, "grad_norm": 5.1854329109191895, "learning_rate": 7.450970873786408e-05, "loss": 0.3056, "step": 5251 }, { "epoch": 50.92289156626506, "grad_norm": 28.38758087158203, "learning_rate": 7.450485436893204e-05, "loss": 0.3057, "step": 5252 }, { "epoch": 50.93253012048193, "grad_norm": 38.39419174194336, "learning_rate": 7.450000000000001e-05, "loss": 0.2716, "step": 5253 }, { "epoch": 50.94216867469879, "grad_norm": 56.20404052734375, "learning_rate": 7.449514563106797e-05, "loss": 0.3361, "step": 5254 }, { "epoch": 50.95180722891566, "grad_norm": 12.22274398803711, "learning_rate": 7.449029126213592e-05, "loss": 0.2544, "step": 5255 }, { "epoch": 50.96144578313253, "grad_norm": 36.33196258544922, "learning_rate": 7.448543689320388e-05, "loss": 0.2361, "step": 5256 }, { "epoch": 50.971084337349396, "grad_norm": 6.618801593780518, "learning_rate": 7.448058252427185e-05, "loss": 0.4259, "step": 5257 }, { "epoch": 50.980722891566266, "grad_norm": 50.66907501220703, "learning_rate": 7.447572815533981e-05, "loss": 0.4689, "step": 5258 }, { "epoch": 50.99036144578313, "grad_norm": 14.809078216552734, "learning_rate": 7.447087378640777e-05, "loss": 0.2875, "step": 5259 }, { "epoch": 51.006024096385545, "grad_norm": 24.641828536987305, "learning_rate": 7.446601941747573e-05, "loss": 0.4508, "step": 5260 }, { "epoch": 51.01566265060241, "grad_norm": 3.4296560287475586, "learning_rate": 7.446116504854368e-05, "loss": 0.2121, "step": 5261 }, { "epoch": 51.02530120481928, "grad_norm": 29.45868492126465, "learning_rate": 7.445631067961166e-05, "loss": 0.2976, "step": 5262 }, { "epoch": 51.03493975903614, "grad_norm": 5.255611896514893, "learning_rate": 7.445145631067961e-05, "loss": 0.3131, "step": 5263 }, { "epoch": 51.04457831325301, "grad_norm": 12.620471954345703, "learning_rate": 7.444660194174757e-05, "loss": 0.4368, "step": 5264 }, { "epoch": 51.05421686746988, "grad_norm": 30.653779983520508, "learning_rate": 7.444174757281553e-05, "loss": 0.3457, "step": 5265 }, { "epoch": 51.063855421686746, "grad_norm": 7.97125244140625, "learning_rate": 7.44368932038835e-05, "loss": 0.2435, "step": 5266 }, { "epoch": 51.07349397590362, "grad_norm": 20.896406173706055, "learning_rate": 7.443203883495146e-05, "loss": 0.3017, "step": 5267 }, { "epoch": 51.08313253012048, "grad_norm": 38.93675231933594, "learning_rate": 7.442718446601942e-05, "loss": 0.0971, "step": 5268 }, { "epoch": 51.09277108433735, "grad_norm": 28.739452362060547, "learning_rate": 7.442233009708737e-05, "loss": 0.2869, "step": 5269 }, { "epoch": 51.102409638554214, "grad_norm": 23.703237533569336, "learning_rate": 7.441747572815534e-05, "loss": 0.4403, "step": 5270 }, { "epoch": 51.112048192771084, "grad_norm": 31.948652267456055, "learning_rate": 7.44126213592233e-05, "loss": 0.2096, "step": 5271 }, { "epoch": 51.121686746987955, "grad_norm": 8.26285457611084, "learning_rate": 7.440776699029126e-05, "loss": 0.2779, "step": 5272 }, { "epoch": 51.13132530120482, "grad_norm": 17.485586166381836, "learning_rate": 7.440291262135922e-05, "loss": 0.3399, "step": 5273 }, { "epoch": 51.14096385542169, "grad_norm": 148.92288208007812, "learning_rate": 7.439805825242719e-05, "loss": 0.2286, "step": 5274 }, { "epoch": 51.15060240963855, "grad_norm": 14.014787673950195, "learning_rate": 7.439320388349515e-05, "loss": 0.2263, "step": 5275 }, { "epoch": 51.16024096385542, "grad_norm": 5.282577991485596, "learning_rate": 7.43883495145631e-05, "loss": 0.2348, "step": 5276 }, { "epoch": 51.16987951807229, "grad_norm": 46.94712829589844, "learning_rate": 7.438349514563108e-05, "loss": 0.4299, "step": 5277 }, { "epoch": 51.179518072289156, "grad_norm": 8.122282028198242, "learning_rate": 7.437864077669903e-05, "loss": 0.1958, "step": 5278 }, { "epoch": 51.189156626506026, "grad_norm": 6.20375394821167, "learning_rate": 7.4373786407767e-05, "loss": 0.3204, "step": 5279 }, { "epoch": 51.19879518072289, "grad_norm": 59.222450256347656, "learning_rate": 7.436893203883496e-05, "loss": 0.2073, "step": 5280 }, { "epoch": 51.20843373493976, "grad_norm": 13.835104942321777, "learning_rate": 7.436407766990292e-05, "loss": 0.3172, "step": 5281 }, { "epoch": 51.21807228915662, "grad_norm": 5.154575347900391, "learning_rate": 7.435922330097088e-05, "loss": 0.315, "step": 5282 }, { "epoch": 51.227710843373494, "grad_norm": 6.145111560821533, "learning_rate": 7.435436893203884e-05, "loss": 0.3816, "step": 5283 }, { "epoch": 51.237349397590364, "grad_norm": 39.832725524902344, "learning_rate": 7.434951456310681e-05, "loss": 0.2916, "step": 5284 }, { "epoch": 51.24698795180723, "grad_norm": 10.643549919128418, "learning_rate": 7.434466019417477e-05, "loss": 0.4879, "step": 5285 }, { "epoch": 51.2566265060241, "grad_norm": 47.1585578918457, "learning_rate": 7.433980582524272e-05, "loss": 0.4469, "step": 5286 }, { "epoch": 51.26626506024096, "grad_norm": 23.641071319580078, "learning_rate": 7.433495145631068e-05, "loss": 0.3286, "step": 5287 }, { "epoch": 51.27590361445783, "grad_norm": 7.7532196044921875, "learning_rate": 7.433009708737865e-05, "loss": 0.395, "step": 5288 }, { "epoch": 51.2855421686747, "grad_norm": 11.657002449035645, "learning_rate": 7.432524271844661e-05, "loss": 0.2607, "step": 5289 }, { "epoch": 51.295180722891565, "grad_norm": 56.086273193359375, "learning_rate": 7.432038834951457e-05, "loss": 0.4706, "step": 5290 }, { "epoch": 51.304819277108436, "grad_norm": 15.930200576782227, "learning_rate": 7.431553398058253e-05, "loss": 0.3069, "step": 5291 }, { "epoch": 51.3144578313253, "grad_norm": 3.1287877559661865, "learning_rate": 7.43106796116505e-05, "loss": 0.3332, "step": 5292 }, { "epoch": 51.32409638554217, "grad_norm": 14.977863311767578, "learning_rate": 7.430582524271845e-05, "loss": 0.3694, "step": 5293 }, { "epoch": 51.33373493975903, "grad_norm": 39.28998565673828, "learning_rate": 7.430097087378641e-05, "loss": 0.3926, "step": 5294 }, { "epoch": 51.3433734939759, "grad_norm": 2.5513999462127686, "learning_rate": 7.429611650485437e-05, "loss": 0.3523, "step": 5295 }, { "epoch": 51.353012048192774, "grad_norm": 23.255159378051758, "learning_rate": 7.429126213592234e-05, "loss": 0.284, "step": 5296 }, { "epoch": 51.36265060240964, "grad_norm": 3.0801169872283936, "learning_rate": 7.42864077669903e-05, "loss": 0.201, "step": 5297 }, { "epoch": 51.37228915662651, "grad_norm": 6.547496318817139, "learning_rate": 7.428155339805826e-05, "loss": 0.3189, "step": 5298 }, { "epoch": 51.38192771084337, "grad_norm": 32.55241775512695, "learning_rate": 7.427669902912621e-05, "loss": 0.2628, "step": 5299 }, { "epoch": 51.39156626506024, "grad_norm": 13.393462181091309, "learning_rate": 7.427184466019417e-05, "loss": 0.1593, "step": 5300 }, { "epoch": 51.40120481927711, "grad_norm": 3.7954134941101074, "learning_rate": 7.426699029126214e-05, "loss": 0.1781, "step": 5301 }, { "epoch": 51.410843373493975, "grad_norm": 7.173990249633789, "learning_rate": 7.42621359223301e-05, "loss": 0.1938, "step": 5302 }, { "epoch": 51.420481927710846, "grad_norm": 11.607857704162598, "learning_rate": 7.425728155339806e-05, "loss": 0.3758, "step": 5303 }, { "epoch": 51.43012048192771, "grad_norm": 27.725841522216797, "learning_rate": 7.425242718446602e-05, "loss": 0.322, "step": 5304 }, { "epoch": 51.43975903614458, "grad_norm": 4.230848789215088, "learning_rate": 7.424757281553399e-05, "loss": 0.24, "step": 5305 }, { "epoch": 51.44939759036144, "grad_norm": 8.570454597473145, "learning_rate": 7.424271844660195e-05, "loss": 0.385, "step": 5306 }, { "epoch": 51.45903614457831, "grad_norm": 6.649279594421387, "learning_rate": 7.42378640776699e-05, "loss": 0.3486, "step": 5307 }, { "epoch": 51.46867469879518, "grad_norm": 82.74542236328125, "learning_rate": 7.423300970873786e-05, "loss": 0.3117, "step": 5308 }, { "epoch": 51.47831325301205, "grad_norm": 22.28921127319336, "learning_rate": 7.422815533980583e-05, "loss": 0.3366, "step": 5309 }, { "epoch": 51.48795180722892, "grad_norm": 7.908844470977783, "learning_rate": 7.422330097087379e-05, "loss": 0.2571, "step": 5310 }, { "epoch": 51.49759036144578, "grad_norm": 6.350122928619385, "learning_rate": 7.421844660194175e-05, "loss": 0.4585, "step": 5311 }, { "epoch": 51.50722891566265, "grad_norm": 6.363731384277344, "learning_rate": 7.42135922330097e-05, "loss": 0.475, "step": 5312 }, { "epoch": 51.51686746987952, "grad_norm": 3.8622212409973145, "learning_rate": 7.420873786407768e-05, "loss": 0.2239, "step": 5313 }, { "epoch": 51.526506024096385, "grad_norm": 22.664581298828125, "learning_rate": 7.420388349514563e-05, "loss": 0.2826, "step": 5314 }, { "epoch": 51.536144578313255, "grad_norm": 13.50506591796875, "learning_rate": 7.419902912621359e-05, "loss": 0.3087, "step": 5315 }, { "epoch": 51.54578313253012, "grad_norm": 50.68806457519531, "learning_rate": 7.419417475728155e-05, "loss": 0.4357, "step": 5316 }, { "epoch": 51.55542168674699, "grad_norm": 9.207572937011719, "learning_rate": 7.418932038834951e-05, "loss": 0.6034, "step": 5317 }, { "epoch": 51.56506024096385, "grad_norm": 9.58560848236084, "learning_rate": 7.418446601941748e-05, "loss": 0.3549, "step": 5318 }, { "epoch": 51.57469879518072, "grad_norm": 11.441438674926758, "learning_rate": 7.417961165048544e-05, "loss": 0.3203, "step": 5319 }, { "epoch": 51.58433734939759, "grad_norm": 31.264074325561523, "learning_rate": 7.41747572815534e-05, "loss": 0.425, "step": 5320 }, { "epoch": 51.59397590361446, "grad_norm": 9.53303050994873, "learning_rate": 7.416990291262135e-05, "loss": 0.4313, "step": 5321 }, { "epoch": 51.60361445783133, "grad_norm": 21.06154441833496, "learning_rate": 7.416504854368932e-05, "loss": 0.3888, "step": 5322 }, { "epoch": 51.61325301204819, "grad_norm": 22.862024307250977, "learning_rate": 7.416019417475728e-05, "loss": 0.2223, "step": 5323 }, { "epoch": 51.62289156626506, "grad_norm": 21.429485321044922, "learning_rate": 7.415533980582524e-05, "loss": 0.3418, "step": 5324 }, { "epoch": 51.63253012048193, "grad_norm": 40.249671936035156, "learning_rate": 7.41504854368932e-05, "loss": 0.6164, "step": 5325 }, { "epoch": 51.642168674698794, "grad_norm": 6.883729457855225, "learning_rate": 7.414563106796117e-05, "loss": 0.0825, "step": 5326 }, { "epoch": 51.651807228915665, "grad_norm": 38.08344268798828, "learning_rate": 7.414077669902914e-05, "loss": 0.5318, "step": 5327 }, { "epoch": 51.66144578313253, "grad_norm": 20.15545082092285, "learning_rate": 7.41359223300971e-05, "loss": 0.2755, "step": 5328 }, { "epoch": 51.6710843373494, "grad_norm": 33.78176498413086, "learning_rate": 7.413106796116506e-05, "loss": 0.2472, "step": 5329 }, { "epoch": 51.68072289156626, "grad_norm": 23.527305603027344, "learning_rate": 7.412621359223301e-05, "loss": 0.3508, "step": 5330 }, { "epoch": 51.69036144578313, "grad_norm": 5.274231433868408, "learning_rate": 7.412135922330098e-05, "loss": 0.1964, "step": 5331 }, { "epoch": 51.7, "grad_norm": 79.74822998046875, "learning_rate": 7.411650485436894e-05, "loss": 0.5829, "step": 5332 }, { "epoch": 51.709638554216866, "grad_norm": 8.357091903686523, "learning_rate": 7.41116504854369e-05, "loss": 0.3756, "step": 5333 }, { "epoch": 51.71927710843374, "grad_norm": 4.665406703948975, "learning_rate": 7.410679611650486e-05, "loss": 0.1989, "step": 5334 }, { "epoch": 51.7289156626506, "grad_norm": 7.441854000091553, "learning_rate": 7.410194174757283e-05, "loss": 0.3686, "step": 5335 }, { "epoch": 51.73855421686747, "grad_norm": 36.25518798828125, "learning_rate": 7.409708737864079e-05, "loss": 0.3018, "step": 5336 }, { "epoch": 51.74819277108434, "grad_norm": 40.90165710449219, "learning_rate": 7.409223300970874e-05, "loss": 0.1719, "step": 5337 }, { "epoch": 51.757831325301204, "grad_norm": 39.52376937866211, "learning_rate": 7.40873786407767e-05, "loss": 0.5624, "step": 5338 }, { "epoch": 51.767469879518075, "grad_norm": 8.353097915649414, "learning_rate": 7.408252427184466e-05, "loss": 0.1564, "step": 5339 }, { "epoch": 51.77710843373494, "grad_norm": 6.76699686050415, "learning_rate": 7.407766990291263e-05, "loss": 0.3737, "step": 5340 }, { "epoch": 51.78674698795181, "grad_norm": 45.03110885620117, "learning_rate": 7.407281553398059e-05, "loss": 0.4235, "step": 5341 }, { "epoch": 51.79638554216867, "grad_norm": 47.88637161254883, "learning_rate": 7.406796116504855e-05, "loss": 0.2398, "step": 5342 }, { "epoch": 51.80602409638554, "grad_norm": 6.03065299987793, "learning_rate": 7.40631067961165e-05, "loss": 0.2542, "step": 5343 }, { "epoch": 51.81566265060241, "grad_norm": 42.640708923339844, "learning_rate": 7.405825242718448e-05, "loss": 0.1005, "step": 5344 }, { "epoch": 51.825301204819276, "grad_norm": 44.25596237182617, "learning_rate": 7.405339805825243e-05, "loss": 0.3312, "step": 5345 }, { "epoch": 51.834939759036146, "grad_norm": 11.154715538024902, "learning_rate": 7.404854368932039e-05, "loss": 0.3647, "step": 5346 }, { "epoch": 51.84457831325301, "grad_norm": 31.223899841308594, "learning_rate": 7.404368932038835e-05, "loss": 0.1378, "step": 5347 }, { "epoch": 51.85421686746988, "grad_norm": 6.98850679397583, "learning_rate": 7.403883495145632e-05, "loss": 0.3918, "step": 5348 }, { "epoch": 51.86385542168675, "grad_norm": 3.3281800746917725, "learning_rate": 7.403398058252428e-05, "loss": 0.1116, "step": 5349 }, { "epoch": 51.873493975903614, "grad_norm": 117.4404067993164, "learning_rate": 7.402912621359224e-05, "loss": 0.436, "step": 5350 }, { "epoch": 51.883132530120484, "grad_norm": 5.89901876449585, "learning_rate": 7.40242718446602e-05, "loss": 0.3444, "step": 5351 }, { "epoch": 51.89277108433735, "grad_norm": 6.486180782318115, "learning_rate": 7.401941747572817e-05, "loss": 0.4284, "step": 5352 }, { "epoch": 51.90240963855422, "grad_norm": 7.822134971618652, "learning_rate": 7.401456310679612e-05, "loss": 0.1734, "step": 5353 }, { "epoch": 51.91204819277108, "grad_norm": 34.348167419433594, "learning_rate": 7.400970873786408e-05, "loss": 0.3509, "step": 5354 }, { "epoch": 51.92168674698795, "grad_norm": 10.183485984802246, "learning_rate": 7.400485436893204e-05, "loss": 0.3815, "step": 5355 }, { "epoch": 51.93132530120482, "grad_norm": 30.247966766357422, "learning_rate": 7.4e-05, "loss": 0.2669, "step": 5356 }, { "epoch": 51.940963855421685, "grad_norm": 6.300027847290039, "learning_rate": 7.399514563106797e-05, "loss": 0.3267, "step": 5357 }, { "epoch": 51.950602409638556, "grad_norm": 74.91439056396484, "learning_rate": 7.399029126213593e-05, "loss": 0.5369, "step": 5358 }, { "epoch": 51.96024096385542, "grad_norm": 40.0877571105957, "learning_rate": 7.398543689320388e-05, "loss": 0.4976, "step": 5359 }, { "epoch": 51.96987951807229, "grad_norm": 6.36478328704834, "learning_rate": 7.398058252427184e-05, "loss": 0.5247, "step": 5360 }, { "epoch": 51.97951807228916, "grad_norm": 20.63363265991211, "learning_rate": 7.397572815533981e-05, "loss": 0.5616, "step": 5361 }, { "epoch": 51.98915662650602, "grad_norm": 10.4453706741333, "learning_rate": 7.397087378640777e-05, "loss": 0.278, "step": 5362 }, { "epoch": 52.00481927710843, "grad_norm": 12.269205093383789, "learning_rate": 7.396601941747573e-05, "loss": 0.3217, "step": 5363 }, { "epoch": 52.0144578313253, "grad_norm": 42.54534149169922, "learning_rate": 7.396116504854369e-05, "loss": 0.2862, "step": 5364 }, { "epoch": 52.024096385542165, "grad_norm": 9.284284591674805, "learning_rate": 7.395631067961166e-05, "loss": 0.3175, "step": 5365 }, { "epoch": 52.033734939759036, "grad_norm": 24.78369140625, "learning_rate": 7.395145631067961e-05, "loss": 0.3252, "step": 5366 }, { "epoch": 52.043373493975906, "grad_norm": 16.43492317199707, "learning_rate": 7.394660194174757e-05, "loss": 0.3271, "step": 5367 }, { "epoch": 52.05301204819277, "grad_norm": 8.628532409667969, "learning_rate": 7.394174757281553e-05, "loss": 0.331, "step": 5368 }, { "epoch": 52.06265060240964, "grad_norm": 7.418787479400635, "learning_rate": 7.39368932038835e-05, "loss": 0.1803, "step": 5369 }, { "epoch": 52.0722891566265, "grad_norm": 10.44642162322998, "learning_rate": 7.393203883495146e-05, "loss": 0.4759, "step": 5370 }, { "epoch": 52.081927710843374, "grad_norm": 4.402227878570557, "learning_rate": 7.392718446601942e-05, "loss": 0.2446, "step": 5371 }, { "epoch": 52.091566265060244, "grad_norm": 5.468954086303711, "learning_rate": 7.392233009708737e-05, "loss": 0.2274, "step": 5372 }, { "epoch": 52.10120481927711, "grad_norm": 21.091217041015625, "learning_rate": 7.391747572815533e-05, "loss": 0.2547, "step": 5373 }, { "epoch": 52.11084337349398, "grad_norm": 37.66365432739258, "learning_rate": 7.39126213592233e-05, "loss": 0.3098, "step": 5374 }, { "epoch": 52.12048192771084, "grad_norm": 10.238794326782227, "learning_rate": 7.390776699029126e-05, "loss": 0.519, "step": 5375 }, { "epoch": 52.13012048192771, "grad_norm": 54.13196563720703, "learning_rate": 7.390291262135922e-05, "loss": 0.4839, "step": 5376 }, { "epoch": 52.139759036144575, "grad_norm": 12.35814094543457, "learning_rate": 7.389805825242718e-05, "loss": 0.3444, "step": 5377 }, { "epoch": 52.149397590361446, "grad_norm": 11.045419692993164, "learning_rate": 7.389320388349515e-05, "loss": 0.4779, "step": 5378 }, { "epoch": 52.159036144578316, "grad_norm": 30.945486068725586, "learning_rate": 7.388834951456312e-05, "loss": 0.3609, "step": 5379 }, { "epoch": 52.16867469879518, "grad_norm": 26.777708053588867, "learning_rate": 7.388349514563108e-05, "loss": 0.318, "step": 5380 }, { "epoch": 52.17831325301205, "grad_norm": 11.2344970703125, "learning_rate": 7.387864077669904e-05, "loss": 0.2449, "step": 5381 }, { "epoch": 52.18795180722891, "grad_norm": 6.925745487213135, "learning_rate": 7.387378640776699e-05, "loss": 0.1181, "step": 5382 }, { "epoch": 52.19759036144578, "grad_norm": 28.607166290283203, "learning_rate": 7.386893203883496e-05, "loss": 0.3006, "step": 5383 }, { "epoch": 52.207228915662654, "grad_norm": 23.273670196533203, "learning_rate": 7.386407766990292e-05, "loss": 0.3756, "step": 5384 }, { "epoch": 52.21686746987952, "grad_norm": 26.59632682800293, "learning_rate": 7.385922330097088e-05, "loss": 0.3873, "step": 5385 }, { "epoch": 52.22650602409639, "grad_norm": 10.94918155670166, "learning_rate": 7.385436893203884e-05, "loss": 0.5158, "step": 5386 }, { "epoch": 52.23614457831325, "grad_norm": 1.7908010482788086, "learning_rate": 7.384951456310681e-05, "loss": 0.121, "step": 5387 }, { "epoch": 52.24578313253012, "grad_norm": 9.669342994689941, "learning_rate": 7.384466019417477e-05, "loss": 0.3571, "step": 5388 }, { "epoch": 52.255421686746985, "grad_norm": 2.4785850048065186, "learning_rate": 7.383980582524272e-05, "loss": 0.0916, "step": 5389 }, { "epoch": 52.265060240963855, "grad_norm": 14.218269348144531, "learning_rate": 7.383495145631068e-05, "loss": 0.3775, "step": 5390 }, { "epoch": 52.274698795180726, "grad_norm": 16.31375503540039, "learning_rate": 7.383009708737865e-05, "loss": 0.2352, "step": 5391 }, { "epoch": 52.28433734939759, "grad_norm": 19.389968872070312, "learning_rate": 7.382524271844661e-05, "loss": 0.2363, "step": 5392 }, { "epoch": 52.29397590361446, "grad_norm": 5.711338043212891, "learning_rate": 7.382038834951457e-05, "loss": 0.3525, "step": 5393 }, { "epoch": 52.30361445783132, "grad_norm": 14.490965843200684, "learning_rate": 7.381553398058253e-05, "loss": 0.4091, "step": 5394 }, { "epoch": 52.31325301204819, "grad_norm": 18.409500122070312, "learning_rate": 7.381067961165048e-05, "loss": 0.2098, "step": 5395 }, { "epoch": 52.32289156626506, "grad_norm": 14.491890907287598, "learning_rate": 7.380582524271846e-05, "loss": 0.3053, "step": 5396 }, { "epoch": 52.33253012048193, "grad_norm": 20.205759048461914, "learning_rate": 7.380097087378641e-05, "loss": 0.2543, "step": 5397 }, { "epoch": 52.3421686746988, "grad_norm": 3.467949628829956, "learning_rate": 7.379611650485437e-05, "loss": 0.1939, "step": 5398 }, { "epoch": 52.35180722891566, "grad_norm": 20.316129684448242, "learning_rate": 7.379126213592233e-05, "loss": 0.2309, "step": 5399 }, { "epoch": 52.36144578313253, "grad_norm": 11.939044952392578, "learning_rate": 7.37864077669903e-05, "loss": 0.4064, "step": 5400 }, { "epoch": 52.371084337349394, "grad_norm": 6.571177959442139, "learning_rate": 7.378155339805826e-05, "loss": 0.2728, "step": 5401 }, { "epoch": 52.380722891566265, "grad_norm": 5.2621355056762695, "learning_rate": 7.377669902912622e-05, "loss": 0.2638, "step": 5402 }, { "epoch": 52.390361445783135, "grad_norm": 7.427818775177002, "learning_rate": 7.377184466019417e-05, "loss": 0.2867, "step": 5403 }, { "epoch": 52.4, "grad_norm": 8.258096694946289, "learning_rate": 7.376699029126214e-05, "loss": 0.2876, "step": 5404 }, { "epoch": 52.40963855421687, "grad_norm": 14.22062873840332, "learning_rate": 7.37621359223301e-05, "loss": 0.3561, "step": 5405 }, { "epoch": 52.41927710843373, "grad_norm": 6.875739097595215, "learning_rate": 7.375728155339806e-05, "loss": 0.329, "step": 5406 }, { "epoch": 52.4289156626506, "grad_norm": 14.065689086914062, "learning_rate": 7.375242718446602e-05, "loss": 0.2468, "step": 5407 }, { "epoch": 52.43855421686747, "grad_norm": 29.652578353881836, "learning_rate": 7.374757281553399e-05, "loss": 0.3446, "step": 5408 }, { "epoch": 52.44819277108434, "grad_norm": 4.134801864624023, "learning_rate": 7.374271844660195e-05, "loss": 0.1796, "step": 5409 }, { "epoch": 52.45783132530121, "grad_norm": 20.327226638793945, "learning_rate": 7.37378640776699e-05, "loss": 0.4989, "step": 5410 }, { "epoch": 52.46746987951807, "grad_norm": 18.70574378967285, "learning_rate": 7.373300970873786e-05, "loss": 0.1209, "step": 5411 }, { "epoch": 52.47710843373494, "grad_norm": 19.711193084716797, "learning_rate": 7.372815533980582e-05, "loss": 0.4539, "step": 5412 }, { "epoch": 52.486746987951804, "grad_norm": 13.511402130126953, "learning_rate": 7.372330097087379e-05, "loss": 0.0975, "step": 5413 }, { "epoch": 52.496385542168674, "grad_norm": 67.76435852050781, "learning_rate": 7.371844660194175e-05, "loss": 0.3681, "step": 5414 }, { "epoch": 52.506024096385545, "grad_norm": 5.353764533996582, "learning_rate": 7.371359223300971e-05, "loss": 0.3876, "step": 5415 }, { "epoch": 52.51566265060241, "grad_norm": 10.810968399047852, "learning_rate": 7.370873786407766e-05, "loss": 0.3776, "step": 5416 }, { "epoch": 52.52530120481928, "grad_norm": 15.90303897857666, "learning_rate": 7.370388349514564e-05, "loss": 0.5514, "step": 5417 }, { "epoch": 52.53493975903614, "grad_norm": 4.065897464752197, "learning_rate": 7.36990291262136e-05, "loss": 0.1852, "step": 5418 }, { "epoch": 52.54457831325301, "grad_norm": 77.82865905761719, "learning_rate": 7.369417475728155e-05, "loss": 0.5184, "step": 5419 }, { "epoch": 52.55421686746988, "grad_norm": 15.693861961364746, "learning_rate": 7.368932038834951e-05, "loss": 0.4754, "step": 5420 }, { "epoch": 52.563855421686746, "grad_norm": 20.083045959472656, "learning_rate": 7.368446601941748e-05, "loss": 0.3428, "step": 5421 }, { "epoch": 52.57349397590362, "grad_norm": 23.700374603271484, "learning_rate": 7.367961165048544e-05, "loss": 0.1979, "step": 5422 }, { "epoch": 52.58313253012048, "grad_norm": 5.498600482940674, "learning_rate": 7.36747572815534e-05, "loss": 0.3563, "step": 5423 }, { "epoch": 52.59277108433735, "grad_norm": 18.55752182006836, "learning_rate": 7.366990291262135e-05, "loss": 0.2106, "step": 5424 }, { "epoch": 52.602409638554214, "grad_norm": 4.520443916320801, "learning_rate": 7.366504854368933e-05, "loss": 0.3459, "step": 5425 }, { "epoch": 52.612048192771084, "grad_norm": 8.079276084899902, "learning_rate": 7.366019417475728e-05, "loss": 0.3202, "step": 5426 }, { "epoch": 52.621686746987955, "grad_norm": 7.4281392097473145, "learning_rate": 7.365533980582524e-05, "loss": 0.4119, "step": 5427 }, { "epoch": 52.63132530120482, "grad_norm": 31.908784866333008, "learning_rate": 7.365048543689321e-05, "loss": 0.5374, "step": 5428 }, { "epoch": 52.64096385542169, "grad_norm": 83.97752380371094, "learning_rate": 7.364563106796117e-05, "loss": 0.2629, "step": 5429 }, { "epoch": 52.65060240963855, "grad_norm": 8.08786392211914, "learning_rate": 7.364077669902914e-05, "loss": 0.2478, "step": 5430 }, { "epoch": 52.66024096385542, "grad_norm": 50.98091506958008, "learning_rate": 7.36359223300971e-05, "loss": 0.1883, "step": 5431 }, { "epoch": 52.66987951807229, "grad_norm": 7.548646926879883, "learning_rate": 7.363106796116506e-05, "loss": 0.3301, "step": 5432 }, { "epoch": 52.679518072289156, "grad_norm": 25.288467407226562, "learning_rate": 7.362621359223301e-05, "loss": 0.2881, "step": 5433 }, { "epoch": 52.689156626506026, "grad_norm": 26.82941436767578, "learning_rate": 7.362135922330097e-05, "loss": 0.3771, "step": 5434 }, { "epoch": 52.69879518072289, "grad_norm": 12.312726974487305, "learning_rate": 7.361650485436894e-05, "loss": 0.1864, "step": 5435 }, { "epoch": 52.70843373493976, "grad_norm": 39.726985931396484, "learning_rate": 7.36116504854369e-05, "loss": 0.3099, "step": 5436 }, { "epoch": 52.71807228915662, "grad_norm": 12.96284294128418, "learning_rate": 7.360679611650486e-05, "loss": 0.1257, "step": 5437 }, { "epoch": 52.727710843373494, "grad_norm": 9.540044784545898, "learning_rate": 7.360194174757282e-05, "loss": 0.4321, "step": 5438 }, { "epoch": 52.737349397590364, "grad_norm": 8.619926452636719, "learning_rate": 7.359708737864079e-05, "loss": 0.3836, "step": 5439 }, { "epoch": 52.74698795180723, "grad_norm": 11.165061950683594, "learning_rate": 7.359223300970875e-05, "loss": 0.2387, "step": 5440 }, { "epoch": 52.7566265060241, "grad_norm": 15.535923957824707, "learning_rate": 7.35873786407767e-05, "loss": 0.3512, "step": 5441 }, { "epoch": 52.76626506024096, "grad_norm": 64.84275817871094, "learning_rate": 7.358252427184466e-05, "loss": 0.45, "step": 5442 }, { "epoch": 52.77590361445783, "grad_norm": 11.77036190032959, "learning_rate": 7.357766990291263e-05, "loss": 0.4342, "step": 5443 }, { "epoch": 52.7855421686747, "grad_norm": 25.764923095703125, "learning_rate": 7.357281553398059e-05, "loss": 0.225, "step": 5444 }, { "epoch": 52.795180722891565, "grad_norm": 28.092212677001953, "learning_rate": 7.356796116504855e-05, "loss": 0.2191, "step": 5445 }, { "epoch": 52.804819277108436, "grad_norm": 128.91461181640625, "learning_rate": 7.35631067961165e-05, "loss": 0.4024, "step": 5446 }, { "epoch": 52.8144578313253, "grad_norm": 47.43532180786133, "learning_rate": 7.355825242718448e-05, "loss": 0.3106, "step": 5447 }, { "epoch": 52.82409638554217, "grad_norm": 6.992377281188965, "learning_rate": 7.355339805825244e-05, "loss": 0.2112, "step": 5448 }, { "epoch": 52.83373493975903, "grad_norm": 9.413662910461426, "learning_rate": 7.354854368932039e-05, "loss": 0.2422, "step": 5449 }, { "epoch": 52.8433734939759, "grad_norm": 13.775390625, "learning_rate": 7.354368932038835e-05, "loss": 0.4003, "step": 5450 }, { "epoch": 52.853012048192774, "grad_norm": 22.51371192932129, "learning_rate": 7.353883495145631e-05, "loss": 0.445, "step": 5451 }, { "epoch": 52.86265060240964, "grad_norm": 6.657142639160156, "learning_rate": 7.353398058252428e-05, "loss": 0.2643, "step": 5452 }, { "epoch": 52.87228915662651, "grad_norm": 8.586576461791992, "learning_rate": 7.352912621359224e-05, "loss": 0.3672, "step": 5453 }, { "epoch": 52.88192771084337, "grad_norm": 41.17131423950195, "learning_rate": 7.35242718446602e-05, "loss": 0.414, "step": 5454 }, { "epoch": 52.89156626506024, "grad_norm": 8.64588737487793, "learning_rate": 7.351941747572815e-05, "loss": 0.2555, "step": 5455 }, { "epoch": 52.90120481927711, "grad_norm": 6.160858154296875, "learning_rate": 7.351456310679612e-05, "loss": 0.2433, "step": 5456 }, { "epoch": 52.910843373493975, "grad_norm": 28.59868621826172, "learning_rate": 7.350970873786408e-05, "loss": 0.3479, "step": 5457 }, { "epoch": 52.920481927710846, "grad_norm": 10.92858600616455, "learning_rate": 7.350485436893204e-05, "loss": 0.4587, "step": 5458 }, { "epoch": 52.93012048192771, "grad_norm": 4.798215866088867, "learning_rate": 7.35e-05, "loss": 0.3693, "step": 5459 }, { "epoch": 52.93975903614458, "grad_norm": 10.672432899475098, "learning_rate": 7.349514563106797e-05, "loss": 0.3987, "step": 5460 }, { "epoch": 52.94939759036144, "grad_norm": 19.211565017700195, "learning_rate": 7.349029126213593e-05, "loss": 0.5225, "step": 5461 }, { "epoch": 52.95903614457831, "grad_norm": 6.022468566894531, "learning_rate": 7.348543689320388e-05, "loss": 0.2255, "step": 5462 }, { "epoch": 52.96867469879518, "grad_norm": 4.641674995422363, "learning_rate": 7.348058252427184e-05, "loss": 0.1072, "step": 5463 }, { "epoch": 52.97831325301205, "grad_norm": 22.526498794555664, "learning_rate": 7.347572815533981e-05, "loss": 0.2166, "step": 5464 }, { "epoch": 52.98795180722892, "grad_norm": 23.271629333496094, "learning_rate": 7.347087378640777e-05, "loss": 0.3564, "step": 5465 }, { "epoch": 53.003614457831326, "grad_norm": 2.477247953414917, "learning_rate": 7.346601941747573e-05, "loss": 0.1616, "step": 5466 }, { "epoch": 53.013253012048196, "grad_norm": 7.369833469390869, "learning_rate": 7.346116504854369e-05, "loss": 0.2704, "step": 5467 }, { "epoch": 53.02289156626506, "grad_norm": 8.472530364990234, "learning_rate": 7.345631067961164e-05, "loss": 0.3201, "step": 5468 }, { "epoch": 53.03253012048193, "grad_norm": 3.3533809185028076, "learning_rate": 7.345145631067962e-05, "loss": 0.2192, "step": 5469 }, { "epoch": 53.04216867469879, "grad_norm": 7.289004325866699, "learning_rate": 7.344660194174757e-05, "loss": 0.323, "step": 5470 }, { "epoch": 53.05180722891566, "grad_norm": 7.680224895477295, "learning_rate": 7.344174757281553e-05, "loss": 0.4153, "step": 5471 }, { "epoch": 53.06144578313253, "grad_norm": 16.014127731323242, "learning_rate": 7.343689320388349e-05, "loss": 0.4009, "step": 5472 }, { "epoch": 53.0710843373494, "grad_norm": 7.980946063995361, "learning_rate": 7.343203883495146e-05, "loss": 0.3888, "step": 5473 }, { "epoch": 53.08072289156627, "grad_norm": 7.8207926750183105, "learning_rate": 7.342718446601942e-05, "loss": 0.1982, "step": 5474 }, { "epoch": 53.09036144578313, "grad_norm": 8.1029634475708, "learning_rate": 7.342233009708738e-05, "loss": 0.3334, "step": 5475 }, { "epoch": 53.1, "grad_norm": 7.958981513977051, "learning_rate": 7.341747572815533e-05, "loss": 0.393, "step": 5476 }, { "epoch": 53.109638554216865, "grad_norm": 4.552763938903809, "learning_rate": 7.34126213592233e-05, "loss": 0.2731, "step": 5477 }, { "epoch": 53.119277108433735, "grad_norm": 17.3854923248291, "learning_rate": 7.340776699029126e-05, "loss": 0.2755, "step": 5478 }, { "epoch": 53.128915662650606, "grad_norm": 18.9511661529541, "learning_rate": 7.340291262135923e-05, "loss": 0.2157, "step": 5479 }, { "epoch": 53.13855421686747, "grad_norm": 5.555345058441162, "learning_rate": 7.339805825242719e-05, "loss": 0.2771, "step": 5480 }, { "epoch": 53.14819277108434, "grad_norm": 26.3504695892334, "learning_rate": 7.339320388349515e-05, "loss": 0.281, "step": 5481 }, { "epoch": 53.1578313253012, "grad_norm": 9.791186332702637, "learning_rate": 7.338834951456312e-05, "loss": 0.2656, "step": 5482 }, { "epoch": 53.16746987951807, "grad_norm": 9.079300880432129, "learning_rate": 7.338349514563108e-05, "loss": 0.3383, "step": 5483 }, { "epoch": 53.17710843373494, "grad_norm": 4.140090465545654, "learning_rate": 7.337864077669904e-05, "loss": 0.4246, "step": 5484 }, { "epoch": 53.18674698795181, "grad_norm": 28.63450813293457, "learning_rate": 7.3373786407767e-05, "loss": 0.4442, "step": 5485 }, { "epoch": 53.19638554216868, "grad_norm": 4.34373140335083, "learning_rate": 7.336893203883497e-05, "loss": 0.2694, "step": 5486 }, { "epoch": 53.20602409638554, "grad_norm": 4.98762321472168, "learning_rate": 7.336407766990292e-05, "loss": 0.2475, "step": 5487 }, { "epoch": 53.21566265060241, "grad_norm": 10.601836204528809, "learning_rate": 7.335922330097088e-05, "loss": 0.1088, "step": 5488 }, { "epoch": 53.225301204819274, "grad_norm": 3.9358673095703125, "learning_rate": 7.335436893203884e-05, "loss": 0.3791, "step": 5489 }, { "epoch": 53.234939759036145, "grad_norm": 12.421573638916016, "learning_rate": 7.33495145631068e-05, "loss": 0.2398, "step": 5490 }, { "epoch": 53.244578313253015, "grad_norm": 15.819973945617676, "learning_rate": 7.334466019417477e-05, "loss": 0.2847, "step": 5491 }, { "epoch": 53.25421686746988, "grad_norm": 34.30818557739258, "learning_rate": 7.333980582524273e-05, "loss": 0.4825, "step": 5492 }, { "epoch": 53.26385542168675, "grad_norm": 11.600661277770996, "learning_rate": 7.333495145631068e-05, "loss": 0.2508, "step": 5493 }, { "epoch": 53.27349397590361, "grad_norm": 14.113556861877441, "learning_rate": 7.333009708737864e-05, "loss": 0.5214, "step": 5494 }, { "epoch": 53.28313253012048, "grad_norm": 5.753725528717041, "learning_rate": 7.332524271844661e-05, "loss": 0.2827, "step": 5495 }, { "epoch": 53.292771084337346, "grad_norm": 10.628528594970703, "learning_rate": 7.332038834951457e-05, "loss": 0.5535, "step": 5496 }, { "epoch": 53.30240963855422, "grad_norm": 9.063275337219238, "learning_rate": 7.331553398058253e-05, "loss": 0.319, "step": 5497 }, { "epoch": 53.31204819277109, "grad_norm": 28.071455001831055, "learning_rate": 7.331067961165049e-05, "loss": 0.429, "step": 5498 }, { "epoch": 53.32168674698795, "grad_norm": 7.099904537200928, "learning_rate": 7.330582524271846e-05, "loss": 0.3138, "step": 5499 }, { "epoch": 53.33132530120482, "grad_norm": 36.05520248413086, "learning_rate": 7.330097087378641e-05, "loss": 0.445, "step": 5500 }, { "epoch": 53.340963855421684, "grad_norm": 8.482436180114746, "learning_rate": 7.329611650485437e-05, "loss": 0.3796, "step": 5501 }, { "epoch": 53.350602409638554, "grad_norm": 17.443605422973633, "learning_rate": 7.329126213592233e-05, "loss": 0.2909, "step": 5502 }, { "epoch": 53.360240963855425, "grad_norm": 3.8508076667785645, "learning_rate": 7.32864077669903e-05, "loss": 0.3938, "step": 5503 }, { "epoch": 53.36987951807229, "grad_norm": 16.814048767089844, "learning_rate": 7.328155339805826e-05, "loss": 0.3694, "step": 5504 }, { "epoch": 53.37951807228916, "grad_norm": 7.5858941078186035, "learning_rate": 7.327669902912622e-05, "loss": 0.3689, "step": 5505 }, { "epoch": 53.38915662650602, "grad_norm": 8.078968048095703, "learning_rate": 7.327184466019417e-05, "loss": 0.4956, "step": 5506 }, { "epoch": 53.39879518072289, "grad_norm": 26.823650360107422, "learning_rate": 7.326699029126213e-05, "loss": 0.367, "step": 5507 }, { "epoch": 53.408433734939756, "grad_norm": 43.319366455078125, "learning_rate": 7.32621359223301e-05, "loss": 0.3322, "step": 5508 }, { "epoch": 53.418072289156626, "grad_norm": 8.341794967651367, "learning_rate": 7.325728155339806e-05, "loss": 0.1822, "step": 5509 }, { "epoch": 53.4277108433735, "grad_norm": 7.072680473327637, "learning_rate": 7.325242718446602e-05, "loss": 0.1883, "step": 5510 }, { "epoch": 53.43734939759036, "grad_norm": 11.598054885864258, "learning_rate": 7.324757281553398e-05, "loss": 0.4071, "step": 5511 }, { "epoch": 53.44698795180723, "grad_norm": 12.422612190246582, "learning_rate": 7.324271844660195e-05, "loss": 0.46, "step": 5512 }, { "epoch": 53.456626506024094, "grad_norm": 7.613259792327881, "learning_rate": 7.32378640776699e-05, "loss": 0.2893, "step": 5513 }, { "epoch": 53.466265060240964, "grad_norm": 7.33617639541626, "learning_rate": 7.323300970873786e-05, "loss": 0.3303, "step": 5514 }, { "epoch": 53.475903614457835, "grad_norm": 63.62054443359375, "learning_rate": 7.322815533980582e-05, "loss": 0.4552, "step": 5515 }, { "epoch": 53.4855421686747, "grad_norm": 3.817491292953491, "learning_rate": 7.322330097087379e-05, "loss": 0.1698, "step": 5516 }, { "epoch": 53.49518072289157, "grad_norm": 4.376890659332275, "learning_rate": 7.321844660194175e-05, "loss": 0.2458, "step": 5517 }, { "epoch": 53.50481927710843, "grad_norm": 4.6040802001953125, "learning_rate": 7.321359223300971e-05, "loss": 0.162, "step": 5518 }, { "epoch": 53.5144578313253, "grad_norm": 10.731978416442871, "learning_rate": 7.320873786407767e-05, "loss": 0.2447, "step": 5519 }, { "epoch": 53.524096385542165, "grad_norm": 9.652530670166016, "learning_rate": 7.320388349514564e-05, "loss": 0.409, "step": 5520 }, { "epoch": 53.533734939759036, "grad_norm": 13.385933876037598, "learning_rate": 7.31990291262136e-05, "loss": 0.4281, "step": 5521 }, { "epoch": 53.543373493975906, "grad_norm": 8.001100540161133, "learning_rate": 7.319417475728155e-05, "loss": 0.2858, "step": 5522 }, { "epoch": 53.55301204819277, "grad_norm": 10.149064064025879, "learning_rate": 7.318932038834951e-05, "loss": 0.1595, "step": 5523 }, { "epoch": 53.56265060240964, "grad_norm": 13.486555099487305, "learning_rate": 7.318446601941747e-05, "loss": 0.4073, "step": 5524 }, { "epoch": 53.5722891566265, "grad_norm": 3.7000765800476074, "learning_rate": 7.317961165048544e-05, "loss": 0.3257, "step": 5525 }, { "epoch": 53.581927710843374, "grad_norm": 50.448787689208984, "learning_rate": 7.31747572815534e-05, "loss": 0.5182, "step": 5526 }, { "epoch": 53.591566265060244, "grad_norm": 7.891375541687012, "learning_rate": 7.316990291262136e-05, "loss": 0.178, "step": 5527 }, { "epoch": 53.60120481927711, "grad_norm": 2.6309001445770264, "learning_rate": 7.316504854368931e-05, "loss": 0.3126, "step": 5528 }, { "epoch": 53.61084337349398, "grad_norm": 3.322702646255493, "learning_rate": 7.316019417475728e-05, "loss": 0.115, "step": 5529 }, { "epoch": 53.62048192771084, "grad_norm": 6.114804744720459, "learning_rate": 7.315533980582526e-05, "loss": 0.3804, "step": 5530 }, { "epoch": 53.63012048192771, "grad_norm": 2.1470947265625, "learning_rate": 7.315048543689321e-05, "loss": 0.2535, "step": 5531 }, { "epoch": 53.639759036144575, "grad_norm": 18.160364151000977, "learning_rate": 7.314563106796117e-05, "loss": 0.4808, "step": 5532 }, { "epoch": 53.649397590361446, "grad_norm": 6.833508014678955, "learning_rate": 7.314077669902913e-05, "loss": 0.2116, "step": 5533 }, { "epoch": 53.659036144578316, "grad_norm": 8.988892555236816, "learning_rate": 7.31359223300971e-05, "loss": 0.3864, "step": 5534 }, { "epoch": 53.66867469879518, "grad_norm": 49.22731018066406, "learning_rate": 7.313106796116506e-05, "loss": 0.3147, "step": 5535 }, { "epoch": 53.67831325301205, "grad_norm": 5.922362804412842, "learning_rate": 7.312621359223302e-05, "loss": 0.5252, "step": 5536 }, { "epoch": 53.68795180722891, "grad_norm": 11.658049583435059, "learning_rate": 7.312135922330097e-05, "loss": 0.1915, "step": 5537 }, { "epoch": 53.69759036144578, "grad_norm": 65.0096206665039, "learning_rate": 7.311650485436895e-05, "loss": 0.2535, "step": 5538 }, { "epoch": 53.707228915662654, "grad_norm": 12.201210021972656, "learning_rate": 7.31116504854369e-05, "loss": 0.3238, "step": 5539 }, { "epoch": 53.71686746987952, "grad_norm": 49.78684997558594, "learning_rate": 7.310679611650486e-05, "loss": 0.3874, "step": 5540 }, { "epoch": 53.72650602409639, "grad_norm": 18.656299591064453, "learning_rate": 7.310194174757282e-05, "loss": 0.5974, "step": 5541 }, { "epoch": 53.73614457831325, "grad_norm": 25.44793701171875, "learning_rate": 7.309708737864078e-05, "loss": 0.3549, "step": 5542 }, { "epoch": 53.74578313253012, "grad_norm": 6.190340042114258, "learning_rate": 7.309223300970875e-05, "loss": 0.3453, "step": 5543 }, { "epoch": 53.755421686746985, "grad_norm": 68.41165161132812, "learning_rate": 7.30873786407767e-05, "loss": 0.2982, "step": 5544 }, { "epoch": 53.765060240963855, "grad_norm": 15.960406303405762, "learning_rate": 7.308252427184466e-05, "loss": 0.2691, "step": 5545 }, { "epoch": 53.774698795180726, "grad_norm": 5.283202171325684, "learning_rate": 7.307766990291262e-05, "loss": 0.4372, "step": 5546 }, { "epoch": 53.78433734939759, "grad_norm": 3.43932843208313, "learning_rate": 7.307281553398059e-05, "loss": 0.2526, "step": 5547 }, { "epoch": 53.79397590361446, "grad_norm": 15.74207592010498, "learning_rate": 7.306796116504855e-05, "loss": 0.2819, "step": 5548 }, { "epoch": 53.80361445783132, "grad_norm": 5.463065147399902, "learning_rate": 7.306310679611651e-05, "loss": 0.1783, "step": 5549 }, { "epoch": 53.81325301204819, "grad_norm": 5.404597759246826, "learning_rate": 7.305825242718447e-05, "loss": 0.2778, "step": 5550 }, { "epoch": 53.82289156626506, "grad_norm": 27.09171485900879, "learning_rate": 7.305339805825244e-05, "loss": 0.5365, "step": 5551 }, { "epoch": 53.83253012048193, "grad_norm": 14.52691650390625, "learning_rate": 7.30485436893204e-05, "loss": 0.2075, "step": 5552 }, { "epoch": 53.8421686746988, "grad_norm": 9.353397369384766, "learning_rate": 7.304368932038835e-05, "loss": 0.3092, "step": 5553 }, { "epoch": 53.85180722891566, "grad_norm": 6.555534839630127, "learning_rate": 7.303883495145631e-05, "loss": 0.2916, "step": 5554 }, { "epoch": 53.86144578313253, "grad_norm": 7.223914623260498, "learning_rate": 7.303398058252428e-05, "loss": 0.2575, "step": 5555 }, { "epoch": 53.871084337349394, "grad_norm": 14.904142379760742, "learning_rate": 7.302912621359224e-05, "loss": 0.2604, "step": 5556 }, { "epoch": 53.880722891566265, "grad_norm": 12.828025817871094, "learning_rate": 7.30242718446602e-05, "loss": 0.3346, "step": 5557 }, { "epoch": 53.890361445783135, "grad_norm": 9.355344772338867, "learning_rate": 7.301941747572815e-05, "loss": 0.321, "step": 5558 }, { "epoch": 53.9, "grad_norm": 4.851235866546631, "learning_rate": 7.301456310679611e-05, "loss": 0.1544, "step": 5559 }, { "epoch": 53.90963855421687, "grad_norm": 22.40256690979004, "learning_rate": 7.300970873786408e-05, "loss": 0.4237, "step": 5560 }, { "epoch": 53.91927710843373, "grad_norm": 9.024208068847656, "learning_rate": 7.300485436893204e-05, "loss": 0.2496, "step": 5561 }, { "epoch": 53.9289156626506, "grad_norm": 6.739879131317139, "learning_rate": 7.3e-05, "loss": 0.3665, "step": 5562 }, { "epoch": 53.93855421686747, "grad_norm": 27.005889892578125, "learning_rate": 7.299514563106796e-05, "loss": 0.2473, "step": 5563 }, { "epoch": 53.94819277108434, "grad_norm": 12.445219039916992, "learning_rate": 7.299029126213593e-05, "loss": 0.3607, "step": 5564 }, { "epoch": 53.95783132530121, "grad_norm": 5.833708763122559, "learning_rate": 7.298543689320389e-05, "loss": 0.1671, "step": 5565 }, { "epoch": 53.96746987951807, "grad_norm": 6.164813041687012, "learning_rate": 7.298058252427184e-05, "loss": 0.2437, "step": 5566 }, { "epoch": 53.97710843373494, "grad_norm": 23.046171188354492, "learning_rate": 7.29757281553398e-05, "loss": 0.3552, "step": 5567 }, { "epoch": 53.986746987951804, "grad_norm": 16.235824584960938, "learning_rate": 7.297087378640777e-05, "loss": 0.5262, "step": 5568 }, { "epoch": 54.00240963855422, "grad_norm": 4.870986461639404, "learning_rate": 7.296601941747573e-05, "loss": 0.2208, "step": 5569 }, { "epoch": 54.01204819277108, "grad_norm": 8.469711303710938, "learning_rate": 7.296116504854369e-05, "loss": 0.3677, "step": 5570 }, { "epoch": 54.02168674698795, "grad_norm": 6.951744079589844, "learning_rate": 7.295631067961165e-05, "loss": 0.481, "step": 5571 }, { "epoch": 54.03132530120482, "grad_norm": 9.102513313293457, "learning_rate": 7.295145631067962e-05, "loss": 0.4108, "step": 5572 }, { "epoch": 54.04096385542169, "grad_norm": 7.313479423522949, "learning_rate": 7.294660194174758e-05, "loss": 0.5959, "step": 5573 }, { "epoch": 54.05060240963856, "grad_norm": 18.693086624145508, "learning_rate": 7.294174757281553e-05, "loss": 0.183, "step": 5574 }, { "epoch": 54.06024096385542, "grad_norm": 8.469559669494629, "learning_rate": 7.293689320388349e-05, "loss": 0.446, "step": 5575 }, { "epoch": 54.06987951807229, "grad_norm": 7.727418422698975, "learning_rate": 7.293203883495145e-05, "loss": 0.2153, "step": 5576 }, { "epoch": 54.079518072289154, "grad_norm": 17.884410858154297, "learning_rate": 7.292718446601942e-05, "loss": 0.3274, "step": 5577 }, { "epoch": 54.089156626506025, "grad_norm": 8.507340431213379, "learning_rate": 7.292233009708738e-05, "loss": 0.2362, "step": 5578 }, { "epoch": 54.09879518072289, "grad_norm": 7.916188716888428, "learning_rate": 7.291747572815534e-05, "loss": 0.1929, "step": 5579 }, { "epoch": 54.10843373493976, "grad_norm": 10.701672554016113, "learning_rate": 7.29126213592233e-05, "loss": 0.2723, "step": 5580 }, { "epoch": 54.11807228915663, "grad_norm": 8.108180046081543, "learning_rate": 7.290776699029126e-05, "loss": 0.3158, "step": 5581 }, { "epoch": 54.12771084337349, "grad_norm": 14.705476760864258, "learning_rate": 7.290291262135924e-05, "loss": 0.293, "step": 5582 }, { "epoch": 54.13734939759036, "grad_norm": 9.370555877685547, "learning_rate": 7.28980582524272e-05, "loss": 0.3856, "step": 5583 }, { "epoch": 54.146987951807226, "grad_norm": 13.250414848327637, "learning_rate": 7.289320388349515e-05, "loss": 0.4765, "step": 5584 }, { "epoch": 54.1566265060241, "grad_norm": 11.847804069519043, "learning_rate": 7.288834951456311e-05, "loss": 0.3106, "step": 5585 }, { "epoch": 54.16626506024097, "grad_norm": 15.176905632019043, "learning_rate": 7.288349514563108e-05, "loss": 0.3317, "step": 5586 }, { "epoch": 54.17590361445783, "grad_norm": 11.429770469665527, "learning_rate": 7.287864077669904e-05, "loss": 0.3734, "step": 5587 }, { "epoch": 54.1855421686747, "grad_norm": 6.57033634185791, "learning_rate": 7.2873786407767e-05, "loss": 0.0972, "step": 5588 }, { "epoch": 54.195180722891564, "grad_norm": 18.42082977294922, "learning_rate": 7.286893203883495e-05, "loss": 0.3784, "step": 5589 }, { "epoch": 54.204819277108435, "grad_norm": 7.812590599060059, "learning_rate": 7.286407766990292e-05, "loss": 0.353, "step": 5590 }, { "epoch": 54.2144578313253, "grad_norm": 4.180182456970215, "learning_rate": 7.285922330097088e-05, "loss": 0.3086, "step": 5591 }, { "epoch": 54.22409638554217, "grad_norm": 8.228464126586914, "learning_rate": 7.285436893203884e-05, "loss": 0.2483, "step": 5592 }, { "epoch": 54.23373493975904, "grad_norm": 7.442071437835693, "learning_rate": 7.28495145631068e-05, "loss": 0.2277, "step": 5593 }, { "epoch": 54.2433734939759, "grad_norm": 95.82807159423828, "learning_rate": 7.284466019417477e-05, "loss": 0.3543, "step": 5594 }, { "epoch": 54.25301204819277, "grad_norm": 6.0535430908203125, "learning_rate": 7.283980582524273e-05, "loss": 0.335, "step": 5595 }, { "epoch": 54.262650602409636, "grad_norm": 11.134868621826172, "learning_rate": 7.283495145631068e-05, "loss": 0.2707, "step": 5596 }, { "epoch": 54.272289156626506, "grad_norm": 16.56490707397461, "learning_rate": 7.283009708737864e-05, "loss": 0.4395, "step": 5597 }, { "epoch": 54.28192771084338, "grad_norm": 2.270195722579956, "learning_rate": 7.28252427184466e-05, "loss": 0.2474, "step": 5598 }, { "epoch": 54.29156626506024, "grad_norm": 22.0529842376709, "learning_rate": 7.282038834951457e-05, "loss": 0.2245, "step": 5599 }, { "epoch": 54.30120481927711, "grad_norm": 9.030118942260742, "learning_rate": 7.281553398058253e-05, "loss": 0.194, "step": 5600 }, { "epoch": 54.310843373493974, "grad_norm": 10.32287883758545, "learning_rate": 7.281067961165049e-05, "loss": 0.3694, "step": 5601 }, { "epoch": 54.320481927710844, "grad_norm": 28.371118545532227, "learning_rate": 7.280582524271844e-05, "loss": 0.3725, "step": 5602 }, { "epoch": 54.33012048192771, "grad_norm": 25.077646255493164, "learning_rate": 7.280097087378642e-05, "loss": 0.5632, "step": 5603 }, { "epoch": 54.33975903614458, "grad_norm": 8.943123817443848, "learning_rate": 7.279611650485437e-05, "loss": 0.4801, "step": 5604 }, { "epoch": 54.34939759036145, "grad_norm": 8.852677345275879, "learning_rate": 7.279126213592233e-05, "loss": 0.2609, "step": 5605 }, { "epoch": 54.35903614457831, "grad_norm": 10.10508918762207, "learning_rate": 7.278640776699029e-05, "loss": 0.2182, "step": 5606 }, { "epoch": 54.36867469879518, "grad_norm": 5.957070827484131, "learning_rate": 7.278155339805826e-05, "loss": 0.1376, "step": 5607 }, { "epoch": 54.378313253012045, "grad_norm": 3.643643856048584, "learning_rate": 7.277669902912622e-05, "loss": 0.2573, "step": 5608 }, { "epoch": 54.387951807228916, "grad_norm": 5.951881408691406, "learning_rate": 7.277184466019418e-05, "loss": 0.227, "step": 5609 }, { "epoch": 54.397590361445786, "grad_norm": 69.95732879638672, "learning_rate": 7.276699029126213e-05, "loss": 0.4508, "step": 5610 }, { "epoch": 54.40722891566265, "grad_norm": 3.788789987564087, "learning_rate": 7.27621359223301e-05, "loss": 0.1411, "step": 5611 }, { "epoch": 54.41686746987952, "grad_norm": 3.248851776123047, "learning_rate": 7.275728155339806e-05, "loss": 0.1777, "step": 5612 }, { "epoch": 54.42650602409638, "grad_norm": 12.842695236206055, "learning_rate": 7.275242718446602e-05, "loss": 0.2222, "step": 5613 }, { "epoch": 54.436144578313254, "grad_norm": 12.017558097839355, "learning_rate": 7.274757281553398e-05, "loss": 0.29, "step": 5614 }, { "epoch": 54.44578313253012, "grad_norm": 5.9389753341674805, "learning_rate": 7.274271844660194e-05, "loss": 0.3546, "step": 5615 }, { "epoch": 54.45542168674699, "grad_norm": 6.121877193450928, "learning_rate": 7.273786407766991e-05, "loss": 0.3164, "step": 5616 }, { "epoch": 54.46506024096386, "grad_norm": 13.666754722595215, "learning_rate": 7.273300970873787e-05, "loss": 0.2293, "step": 5617 }, { "epoch": 54.47469879518072, "grad_norm": 9.815033912658691, "learning_rate": 7.272815533980582e-05, "loss": 0.3185, "step": 5618 }, { "epoch": 54.48433734939759, "grad_norm": 4.800584316253662, "learning_rate": 7.272330097087378e-05, "loss": 0.392, "step": 5619 }, { "epoch": 54.493975903614455, "grad_norm": 3.436988353729248, "learning_rate": 7.271844660194175e-05, "loss": 0.2049, "step": 5620 }, { "epoch": 54.503614457831326, "grad_norm": 5.144093036651611, "learning_rate": 7.271359223300971e-05, "loss": 0.4462, "step": 5621 }, { "epoch": 54.513253012048196, "grad_norm": 8.125924110412598, "learning_rate": 7.270873786407767e-05, "loss": 0.4543, "step": 5622 }, { "epoch": 54.52289156626506, "grad_norm": 9.322978973388672, "learning_rate": 7.270388349514563e-05, "loss": 0.466, "step": 5623 }, { "epoch": 54.53253012048193, "grad_norm": 6.627340316772461, "learning_rate": 7.26990291262136e-05, "loss": 0.3097, "step": 5624 }, { "epoch": 54.54216867469879, "grad_norm": 14.150897026062012, "learning_rate": 7.269417475728155e-05, "loss": 0.3651, "step": 5625 }, { "epoch": 54.55180722891566, "grad_norm": 15.735712051391602, "learning_rate": 7.268932038834951e-05, "loss": 0.259, "step": 5626 }, { "epoch": 54.56144578313253, "grad_norm": 5.447292804718018, "learning_rate": 7.268446601941747e-05, "loss": 0.4135, "step": 5627 }, { "epoch": 54.5710843373494, "grad_norm": 3.9625322818756104, "learning_rate": 7.267961165048544e-05, "loss": 0.0854, "step": 5628 }, { "epoch": 54.58072289156627, "grad_norm": 12.018198013305664, "learning_rate": 7.26747572815534e-05, "loss": 0.3721, "step": 5629 }, { "epoch": 54.59036144578313, "grad_norm": 10.545872688293457, "learning_rate": 7.266990291262137e-05, "loss": 0.5496, "step": 5630 }, { "epoch": 54.6, "grad_norm": 5.354023456573486, "learning_rate": 7.266504854368933e-05, "loss": 0.4393, "step": 5631 }, { "epoch": 54.609638554216865, "grad_norm": 14.995448112487793, "learning_rate": 7.266019417475729e-05, "loss": 0.3175, "step": 5632 }, { "epoch": 54.619277108433735, "grad_norm": 11.000618934631348, "learning_rate": 7.265533980582526e-05, "loss": 0.3721, "step": 5633 }, { "epoch": 54.628915662650606, "grad_norm": 8.071256637573242, "learning_rate": 7.265048543689322e-05, "loss": 0.2828, "step": 5634 }, { "epoch": 54.63855421686747, "grad_norm": 15.461649894714355, "learning_rate": 7.264563106796117e-05, "loss": 0.3272, "step": 5635 }, { "epoch": 54.64819277108434, "grad_norm": 6.455082893371582, "learning_rate": 7.264077669902913e-05, "loss": 0.2064, "step": 5636 }, { "epoch": 54.6578313253012, "grad_norm": 26.985485076904297, "learning_rate": 7.263592233009709e-05, "loss": 0.4972, "step": 5637 }, { "epoch": 54.66746987951807, "grad_norm": 14.84084415435791, "learning_rate": 7.263106796116506e-05, "loss": 0.4202, "step": 5638 }, { "epoch": 54.67710843373494, "grad_norm": 5.1501359939575195, "learning_rate": 7.262621359223302e-05, "loss": 0.1614, "step": 5639 }, { "epoch": 54.68674698795181, "grad_norm": 5.859149932861328, "learning_rate": 7.262135922330098e-05, "loss": 0.2181, "step": 5640 }, { "epoch": 54.69638554216868, "grad_norm": 5.323098659515381, "learning_rate": 7.261650485436893e-05, "loss": 0.2971, "step": 5641 }, { "epoch": 54.70602409638554, "grad_norm": 34.04116439819336, "learning_rate": 7.26116504854369e-05, "loss": 0.2515, "step": 5642 }, { "epoch": 54.71566265060241, "grad_norm": 4.023012638092041, "learning_rate": 7.260679611650486e-05, "loss": 0.3267, "step": 5643 }, { "epoch": 54.725301204819274, "grad_norm": 15.855236053466797, "learning_rate": 7.260194174757282e-05, "loss": 0.5824, "step": 5644 }, { "epoch": 54.734939759036145, "grad_norm": 11.096109390258789, "learning_rate": 7.259708737864078e-05, "loss": 0.41, "step": 5645 }, { "epoch": 54.744578313253015, "grad_norm": 18.552406311035156, "learning_rate": 7.259223300970875e-05, "loss": 0.3099, "step": 5646 }, { "epoch": 54.75421686746988, "grad_norm": 11.097726821899414, "learning_rate": 7.25873786407767e-05, "loss": 0.4486, "step": 5647 }, { "epoch": 54.76385542168675, "grad_norm": 6.477497577667236, "learning_rate": 7.258252427184466e-05, "loss": 0.2209, "step": 5648 }, { "epoch": 54.77349397590361, "grad_norm": 5.002108097076416, "learning_rate": 7.257766990291262e-05, "loss": 0.3717, "step": 5649 }, { "epoch": 54.78313253012048, "grad_norm": 8.313592910766602, "learning_rate": 7.25728155339806e-05, "loss": 0.3946, "step": 5650 }, { "epoch": 54.792771084337346, "grad_norm": 20.570926666259766, "learning_rate": 7.256796116504855e-05, "loss": 0.4514, "step": 5651 }, { "epoch": 54.80240963855422, "grad_norm": 5.512195587158203, "learning_rate": 7.256310679611651e-05, "loss": 0.2786, "step": 5652 }, { "epoch": 54.81204819277109, "grad_norm": 8.476048469543457, "learning_rate": 7.255825242718447e-05, "loss": 0.319, "step": 5653 }, { "epoch": 54.82168674698795, "grad_norm": 14.92805004119873, "learning_rate": 7.255339805825242e-05, "loss": 0.5393, "step": 5654 }, { "epoch": 54.83132530120482, "grad_norm": 6.349785327911377, "learning_rate": 7.25485436893204e-05, "loss": 0.2691, "step": 5655 }, { "epoch": 54.840963855421684, "grad_norm": 5.208187103271484, "learning_rate": 7.254368932038835e-05, "loss": 0.314, "step": 5656 }, { "epoch": 54.850602409638554, "grad_norm": 15.854289054870605, "learning_rate": 7.253883495145631e-05, "loss": 0.3734, "step": 5657 }, { "epoch": 54.860240963855425, "grad_norm": 7.965912818908691, "learning_rate": 7.253398058252427e-05, "loss": 0.2447, "step": 5658 }, { "epoch": 54.86987951807229, "grad_norm": 9.107571601867676, "learning_rate": 7.252912621359224e-05, "loss": 0.4368, "step": 5659 }, { "epoch": 54.87951807228916, "grad_norm": 7.5124735832214355, "learning_rate": 7.25242718446602e-05, "loss": 0.1943, "step": 5660 }, { "epoch": 54.88915662650602, "grad_norm": 4.42144775390625, "learning_rate": 7.251941747572816e-05, "loss": 0.181, "step": 5661 }, { "epoch": 54.89879518072289, "grad_norm": 6.992580413818359, "learning_rate": 7.251456310679611e-05, "loss": 0.2642, "step": 5662 }, { "epoch": 54.908433734939756, "grad_norm": 15.27769660949707, "learning_rate": 7.250970873786408e-05, "loss": 0.4165, "step": 5663 }, { "epoch": 54.918072289156626, "grad_norm": 5.624616622924805, "learning_rate": 7.250485436893204e-05, "loss": 0.2591, "step": 5664 }, { "epoch": 54.9277108433735, "grad_norm": 8.28520393371582, "learning_rate": 7.25e-05, "loss": 0.4112, "step": 5665 }, { "epoch": 54.93734939759036, "grad_norm": 7.963480472564697, "learning_rate": 7.249514563106796e-05, "loss": 0.2588, "step": 5666 }, { "epoch": 54.94698795180723, "grad_norm": 5.467836856842041, "learning_rate": 7.249029126213593e-05, "loss": 0.2088, "step": 5667 }, { "epoch": 54.956626506024094, "grad_norm": 14.026128768920898, "learning_rate": 7.248543689320389e-05, "loss": 0.729, "step": 5668 }, { "epoch": 54.966265060240964, "grad_norm": 5.541787147521973, "learning_rate": 7.248058252427185e-05, "loss": 0.226, "step": 5669 }, { "epoch": 54.975903614457835, "grad_norm": 23.418025970458984, "learning_rate": 7.24757281553398e-05, "loss": 0.4597, "step": 5670 }, { "epoch": 54.9855421686747, "grad_norm": 5.047794342041016, "learning_rate": 7.247087378640776e-05, "loss": 0.296, "step": 5671 }, { "epoch": 55.001204819277106, "grad_norm": 8.444014549255371, "learning_rate": 7.246601941747573e-05, "loss": 0.3519, "step": 5672 }, { "epoch": 55.01084337349398, "grad_norm": 18.949207305908203, "learning_rate": 7.246116504854369e-05, "loss": 0.4135, "step": 5673 }, { "epoch": 55.02048192771084, "grad_norm": 13.267598152160645, "learning_rate": 7.245631067961165e-05, "loss": 0.318, "step": 5674 }, { "epoch": 55.03012048192771, "grad_norm": 9.930100440979004, "learning_rate": 7.24514563106796e-05, "loss": 0.5596, "step": 5675 }, { "epoch": 55.03975903614458, "grad_norm": 8.99979019165039, "learning_rate": 7.244660194174758e-05, "loss": 0.3909, "step": 5676 }, { "epoch": 55.049397590361444, "grad_norm": 4.407816410064697, "learning_rate": 7.244174757281553e-05, "loss": 0.2297, "step": 5677 }, { "epoch": 55.059036144578315, "grad_norm": 9.424246788024902, "learning_rate": 7.243689320388349e-05, "loss": 0.2954, "step": 5678 }, { "epoch": 55.06867469879518, "grad_norm": 5.923033237457275, "learning_rate": 7.243203883495145e-05, "loss": 0.2741, "step": 5679 }, { "epoch": 55.07831325301205, "grad_norm": 5.160238742828369, "learning_rate": 7.242718446601942e-05, "loss": 0.2087, "step": 5680 }, { "epoch": 55.08795180722892, "grad_norm": 8.625714302062988, "learning_rate": 7.242233009708739e-05, "loss": 0.2419, "step": 5681 }, { "epoch": 55.09759036144578, "grad_norm": 13.035065650939941, "learning_rate": 7.241747572815535e-05, "loss": 0.5289, "step": 5682 }, { "epoch": 55.10722891566265, "grad_norm": 7.618344783782959, "learning_rate": 7.241262135922331e-05, "loss": 0.2561, "step": 5683 }, { "epoch": 55.116867469879516, "grad_norm": 6.791806697845459, "learning_rate": 7.240776699029127e-05, "loss": 0.1938, "step": 5684 }, { "epoch": 55.126506024096386, "grad_norm": 12.335515022277832, "learning_rate": 7.240291262135924e-05, "loss": 0.5543, "step": 5685 }, { "epoch": 55.13614457831325, "grad_norm": 8.849443435668945, "learning_rate": 7.23980582524272e-05, "loss": 0.4188, "step": 5686 }, { "epoch": 55.14578313253012, "grad_norm": 5.650587558746338, "learning_rate": 7.239320388349515e-05, "loss": 0.242, "step": 5687 }, { "epoch": 55.15542168674699, "grad_norm": 4.404826641082764, "learning_rate": 7.238834951456311e-05, "loss": 0.2159, "step": 5688 }, { "epoch": 55.165060240963854, "grad_norm": 11.011306762695312, "learning_rate": 7.238349514563108e-05, "loss": 0.3726, "step": 5689 }, { "epoch": 55.174698795180724, "grad_norm": 5.942828178405762, "learning_rate": 7.237864077669904e-05, "loss": 0.1921, "step": 5690 }, { "epoch": 55.18433734939759, "grad_norm": 48.98286056518555, "learning_rate": 7.2373786407767e-05, "loss": 0.3239, "step": 5691 }, { "epoch": 55.19397590361446, "grad_norm": 8.754592895507812, "learning_rate": 7.236893203883495e-05, "loss": 0.323, "step": 5692 }, { "epoch": 55.20361445783133, "grad_norm": 5.138117790222168, "learning_rate": 7.236407766990291e-05, "loss": 0.1986, "step": 5693 }, { "epoch": 55.21325301204819, "grad_norm": 25.83629608154297, "learning_rate": 7.235922330097088e-05, "loss": 0.2718, "step": 5694 }, { "epoch": 55.22289156626506, "grad_norm": 6.870785713195801, "learning_rate": 7.235436893203884e-05, "loss": 0.2471, "step": 5695 }, { "epoch": 55.232530120481925, "grad_norm": 7.729568958282471, "learning_rate": 7.23495145631068e-05, "loss": 0.2781, "step": 5696 }, { "epoch": 55.242168674698796, "grad_norm": 17.522008895874023, "learning_rate": 7.234466019417476e-05, "loss": 0.2769, "step": 5697 }, { "epoch": 55.25180722891566, "grad_norm": 20.337005615234375, "learning_rate": 7.233980582524273e-05, "loss": 0.3572, "step": 5698 }, { "epoch": 55.26144578313253, "grad_norm": 10.650087356567383, "learning_rate": 7.233495145631069e-05, "loss": 0.4413, "step": 5699 }, { "epoch": 55.2710843373494, "grad_norm": 15.481714248657227, "learning_rate": 7.233009708737864e-05, "loss": 0.214, "step": 5700 }, { "epoch": 55.28072289156626, "grad_norm": 7.198193073272705, "learning_rate": 7.23252427184466e-05, "loss": 0.2157, "step": 5701 }, { "epoch": 55.290361445783134, "grad_norm": 4.254854679107666, "learning_rate": 7.232038834951457e-05, "loss": 0.21, "step": 5702 }, { "epoch": 55.3, "grad_norm": 9.653910636901855, "learning_rate": 7.231553398058253e-05, "loss": 0.3071, "step": 5703 }, { "epoch": 55.30963855421687, "grad_norm": 9.058385848999023, "learning_rate": 7.231067961165049e-05, "loss": 0.3779, "step": 5704 }, { "epoch": 55.31927710843374, "grad_norm": 10.086677551269531, "learning_rate": 7.230582524271845e-05, "loss": 0.273, "step": 5705 }, { "epoch": 55.3289156626506, "grad_norm": 10.887735366821289, "learning_rate": 7.230097087378642e-05, "loss": 0.1922, "step": 5706 }, { "epoch": 55.33855421686747, "grad_norm": 29.314558029174805, "learning_rate": 7.229611650485438e-05, "loss": 0.2943, "step": 5707 }, { "epoch": 55.348192771084335, "grad_norm": 21.261531829833984, "learning_rate": 7.229126213592233e-05, "loss": 0.4263, "step": 5708 }, { "epoch": 55.357831325301206, "grad_norm": 8.37771224975586, "learning_rate": 7.228640776699029e-05, "loss": 0.2564, "step": 5709 }, { "epoch": 55.36746987951807, "grad_norm": 11.175371170043945, "learning_rate": 7.228155339805825e-05, "loss": 0.4269, "step": 5710 }, { "epoch": 55.37710843373494, "grad_norm": 5.651427745819092, "learning_rate": 7.227669902912622e-05, "loss": 0.2784, "step": 5711 }, { "epoch": 55.38674698795181, "grad_norm": 67.91177368164062, "learning_rate": 7.227184466019418e-05, "loss": 0.3328, "step": 5712 }, { "epoch": 55.39638554216867, "grad_norm": 13.349127769470215, "learning_rate": 7.226699029126214e-05, "loss": 0.5334, "step": 5713 }, { "epoch": 55.40602409638554, "grad_norm": 10.051610946655273, "learning_rate": 7.226213592233009e-05, "loss": 0.3756, "step": 5714 }, { "epoch": 55.41566265060241, "grad_norm": 7.93691349029541, "learning_rate": 7.225728155339806e-05, "loss": 0.2133, "step": 5715 }, { "epoch": 55.42530120481928, "grad_norm": 10.08812141418457, "learning_rate": 7.225242718446602e-05, "loss": 0.3729, "step": 5716 }, { "epoch": 55.43493975903615, "grad_norm": 8.875655174255371, "learning_rate": 7.224757281553398e-05, "loss": 0.2873, "step": 5717 }, { "epoch": 55.44457831325301, "grad_norm": 13.021071434020996, "learning_rate": 7.224271844660194e-05, "loss": 0.3344, "step": 5718 }, { "epoch": 55.45421686746988, "grad_norm": 13.063701629638672, "learning_rate": 7.223786407766991e-05, "loss": 0.494, "step": 5719 }, { "epoch": 55.463855421686745, "grad_norm": 5.736455917358398, "learning_rate": 7.223300970873787e-05, "loss": 0.2266, "step": 5720 }, { "epoch": 55.473493975903615, "grad_norm": 70.1322021484375, "learning_rate": 7.222815533980582e-05, "loss": 0.1852, "step": 5721 }, { "epoch": 55.48313253012048, "grad_norm": 11.496011734008789, "learning_rate": 7.222330097087378e-05, "loss": 0.5977, "step": 5722 }, { "epoch": 55.49277108433735, "grad_norm": 5.850667476654053, "learning_rate": 7.221844660194175e-05, "loss": 0.1902, "step": 5723 }, { "epoch": 55.50240963855422, "grad_norm": 7.703179359436035, "learning_rate": 7.221359223300971e-05, "loss": 0.2707, "step": 5724 }, { "epoch": 55.51204819277108, "grad_norm": 4.20508337020874, "learning_rate": 7.220873786407767e-05, "loss": 0.2411, "step": 5725 }, { "epoch": 55.52168674698795, "grad_norm": 9.620675086975098, "learning_rate": 7.220388349514563e-05, "loss": 0.2471, "step": 5726 }, { "epoch": 55.53132530120482, "grad_norm": 16.751026153564453, "learning_rate": 7.219902912621358e-05, "loss": 0.4051, "step": 5727 }, { "epoch": 55.54096385542169, "grad_norm": 11.814072608947754, "learning_rate": 7.219417475728156e-05, "loss": 0.2105, "step": 5728 }, { "epoch": 55.55060240963856, "grad_norm": 8.963741302490234, "learning_rate": 7.218932038834951e-05, "loss": 0.5323, "step": 5729 }, { "epoch": 55.56024096385542, "grad_norm": 22.5251522064209, "learning_rate": 7.218446601941747e-05, "loss": 0.5632, "step": 5730 }, { "epoch": 55.56987951807229, "grad_norm": 16.01742935180664, "learning_rate": 7.217961165048544e-05, "loss": 0.5171, "step": 5731 }, { "epoch": 55.579518072289154, "grad_norm": 6.521199703216553, "learning_rate": 7.21747572815534e-05, "loss": 0.3269, "step": 5732 }, { "epoch": 55.589156626506025, "grad_norm": 8.779431343078613, "learning_rate": 7.216990291262137e-05, "loss": 0.25, "step": 5733 }, { "epoch": 55.59879518072289, "grad_norm": 6.6170334815979, "learning_rate": 7.216504854368933e-05, "loss": 0.312, "step": 5734 }, { "epoch": 55.60843373493976, "grad_norm": 8.927961349487305, "learning_rate": 7.216019417475729e-05, "loss": 0.2786, "step": 5735 }, { "epoch": 55.61807228915663, "grad_norm": 85.0300064086914, "learning_rate": 7.215533980582525e-05, "loss": 0.2906, "step": 5736 }, { "epoch": 55.62771084337349, "grad_norm": 12.61474895477295, "learning_rate": 7.215048543689322e-05, "loss": 0.3033, "step": 5737 }, { "epoch": 55.63734939759036, "grad_norm": 14.566374778747559, "learning_rate": 7.214563106796117e-05, "loss": 0.2841, "step": 5738 }, { "epoch": 55.646987951807226, "grad_norm": 10.412740707397461, "learning_rate": 7.214077669902913e-05, "loss": 0.5054, "step": 5739 }, { "epoch": 55.6566265060241, "grad_norm": 5.357616424560547, "learning_rate": 7.213592233009709e-05, "loss": 0.2899, "step": 5740 }, { "epoch": 55.66626506024097, "grad_norm": 18.636985778808594, "learning_rate": 7.213106796116506e-05, "loss": 0.1882, "step": 5741 }, { "epoch": 55.67590361445783, "grad_norm": 5.711742877960205, "learning_rate": 7.212621359223302e-05, "loss": 0.3134, "step": 5742 }, { "epoch": 55.6855421686747, "grad_norm": 7.079282760620117, "learning_rate": 7.212135922330098e-05, "loss": 0.2103, "step": 5743 }, { "epoch": 55.695180722891564, "grad_norm": 18.331422805786133, "learning_rate": 7.211650485436893e-05, "loss": 0.4252, "step": 5744 }, { "epoch": 55.704819277108435, "grad_norm": 8.691535949707031, "learning_rate": 7.21116504854369e-05, "loss": 0.2964, "step": 5745 }, { "epoch": 55.7144578313253, "grad_norm": 9.143135070800781, "learning_rate": 7.210679611650486e-05, "loss": 0.2707, "step": 5746 }, { "epoch": 55.72409638554217, "grad_norm": 12.6238374710083, "learning_rate": 7.210194174757282e-05, "loss": 0.1922, "step": 5747 }, { "epoch": 55.73373493975904, "grad_norm": 7.133221626281738, "learning_rate": 7.209708737864078e-05, "loss": 0.4294, "step": 5748 }, { "epoch": 55.7433734939759, "grad_norm": 6.821170330047607, "learning_rate": 7.209223300970874e-05, "loss": 0.3423, "step": 5749 }, { "epoch": 55.75301204819277, "grad_norm": 7.252663612365723, "learning_rate": 7.208737864077671e-05, "loss": 0.3791, "step": 5750 }, { "epoch": 55.762650602409636, "grad_norm": 5.684885025024414, "learning_rate": 7.208252427184467e-05, "loss": 0.3455, "step": 5751 }, { "epoch": 55.772289156626506, "grad_norm": 5.718132495880127, "learning_rate": 7.207766990291262e-05, "loss": 0.2751, "step": 5752 }, { "epoch": 55.78192771084338, "grad_norm": 5.488896369934082, "learning_rate": 7.207281553398058e-05, "loss": 0.1928, "step": 5753 }, { "epoch": 55.79156626506024, "grad_norm": 23.190305709838867, "learning_rate": 7.206796116504855e-05, "loss": 0.2814, "step": 5754 }, { "epoch": 55.80120481927711, "grad_norm": 56.54740905761719, "learning_rate": 7.206310679611651e-05, "loss": 0.4589, "step": 5755 }, { "epoch": 55.810843373493974, "grad_norm": 9.143678665161133, "learning_rate": 7.205825242718447e-05, "loss": 0.3287, "step": 5756 }, { "epoch": 55.820481927710844, "grad_norm": 5.447449684143066, "learning_rate": 7.205339805825243e-05, "loss": 0.1467, "step": 5757 }, { "epoch": 55.83012048192771, "grad_norm": 8.876921653747559, "learning_rate": 7.20485436893204e-05, "loss": 0.3514, "step": 5758 }, { "epoch": 55.83975903614458, "grad_norm": 14.212651252746582, "learning_rate": 7.204368932038835e-05, "loss": 0.2993, "step": 5759 }, { "epoch": 55.84939759036145, "grad_norm": 7.205406188964844, "learning_rate": 7.203883495145631e-05, "loss": 0.3877, "step": 5760 }, { "epoch": 55.85903614457831, "grad_norm": 9.453951835632324, "learning_rate": 7.203398058252427e-05, "loss": 0.4108, "step": 5761 }, { "epoch": 55.86867469879518, "grad_norm": 153.16612243652344, "learning_rate": 7.202912621359224e-05, "loss": 0.1347, "step": 5762 }, { "epoch": 55.878313253012045, "grad_norm": 456.8193359375, "learning_rate": 7.20242718446602e-05, "loss": 0.3742, "step": 5763 }, { "epoch": 55.887951807228916, "grad_norm": 20.455955505371094, "learning_rate": 7.201941747572816e-05, "loss": 0.4873, "step": 5764 }, { "epoch": 55.897590361445786, "grad_norm": 18.76081657409668, "learning_rate": 7.201456310679611e-05, "loss": 0.1877, "step": 5765 }, { "epoch": 55.90722891566265, "grad_norm": 5.962982177734375, "learning_rate": 7.200970873786407e-05, "loss": 0.3975, "step": 5766 }, { "epoch": 55.91686746987952, "grad_norm": 2.6511032581329346, "learning_rate": 7.200485436893204e-05, "loss": 0.0933, "step": 5767 }, { "epoch": 55.92650602409638, "grad_norm": 9.029298782348633, "learning_rate": 7.2e-05, "loss": 0.2384, "step": 5768 }, { "epoch": 55.936144578313254, "grad_norm": 9.03533935546875, "learning_rate": 7.199514563106796e-05, "loss": 0.21, "step": 5769 }, { "epoch": 55.94578313253012, "grad_norm": 13.263437271118164, "learning_rate": 7.199029126213592e-05, "loss": 0.402, "step": 5770 }, { "epoch": 55.95542168674699, "grad_norm": 43.86027145385742, "learning_rate": 7.198543689320389e-05, "loss": 0.2906, "step": 5771 }, { "epoch": 55.96506024096386, "grad_norm": 11.031147003173828, "learning_rate": 7.198058252427185e-05, "loss": 0.3955, "step": 5772 }, { "epoch": 55.97469879518072, "grad_norm": 6.7954020500183105, "learning_rate": 7.19757281553398e-05, "loss": 0.3139, "step": 5773 }, { "epoch": 55.98433734939759, "grad_norm": 11.190664291381836, "learning_rate": 7.197087378640776e-05, "loss": 0.3776, "step": 5774 }, { "epoch": 55.993975903614455, "grad_norm": 12.050446510314941, "learning_rate": 7.196601941747573e-05, "loss": 0.4243, "step": 5775 }, { "epoch": 56.00963855421687, "grad_norm": 4.672457218170166, "learning_rate": 7.196116504854369e-05, "loss": 0.2484, "step": 5776 }, { "epoch": 56.019277108433734, "grad_norm": 6.613987922668457, "learning_rate": 7.195631067961165e-05, "loss": 0.4131, "step": 5777 }, { "epoch": 56.028915662650604, "grad_norm": 12.888992309570312, "learning_rate": 7.19514563106796e-05, "loss": 0.5827, "step": 5778 }, { "epoch": 56.03855421686747, "grad_norm": 9.964414596557617, "learning_rate": 7.194660194174758e-05, "loss": 0.3054, "step": 5779 }, { "epoch": 56.04819277108434, "grad_norm": 5.401790142059326, "learning_rate": 7.194174757281554e-05, "loss": 0.3205, "step": 5780 }, { "epoch": 56.0578313253012, "grad_norm": 8.015822410583496, "learning_rate": 7.19368932038835e-05, "loss": 0.3128, "step": 5781 }, { "epoch": 56.06746987951807, "grad_norm": 5.464121341705322, "learning_rate": 7.193203883495146e-05, "loss": 0.1508, "step": 5782 }, { "epoch": 56.07710843373494, "grad_norm": 9.539134979248047, "learning_rate": 7.192718446601942e-05, "loss": 0.3023, "step": 5783 }, { "epoch": 56.086746987951805, "grad_norm": 14.245201110839844, "learning_rate": 7.19223300970874e-05, "loss": 0.3953, "step": 5784 }, { "epoch": 56.096385542168676, "grad_norm": 8.921121597290039, "learning_rate": 7.191747572815535e-05, "loss": 0.1336, "step": 5785 }, { "epoch": 56.10602409638554, "grad_norm": 7.35275936126709, "learning_rate": 7.191262135922331e-05, "loss": 0.2704, "step": 5786 }, { "epoch": 56.11566265060241, "grad_norm": 6.875333309173584, "learning_rate": 7.190776699029127e-05, "loss": 0.161, "step": 5787 }, { "epoch": 56.12530120481928, "grad_norm": 18.704736709594727, "learning_rate": 7.190291262135922e-05, "loss": 0.4223, "step": 5788 }, { "epoch": 56.13493975903614, "grad_norm": 11.376660346984863, "learning_rate": 7.18980582524272e-05, "loss": 0.2643, "step": 5789 }, { "epoch": 56.144578313253014, "grad_norm": 10.322937965393066, "learning_rate": 7.189320388349515e-05, "loss": 0.3213, "step": 5790 }, { "epoch": 56.15421686746988, "grad_norm": 7.22531795501709, "learning_rate": 7.188834951456311e-05, "loss": 0.4315, "step": 5791 }, { "epoch": 56.16385542168675, "grad_norm": 7.114387512207031, "learning_rate": 7.188349514563107e-05, "loss": 0.1981, "step": 5792 }, { "epoch": 56.17349397590361, "grad_norm": 9.694056510925293, "learning_rate": 7.187864077669904e-05, "loss": 0.2316, "step": 5793 }, { "epoch": 56.18313253012048, "grad_norm": 9.089702606201172, "learning_rate": 7.1873786407767e-05, "loss": 0.4903, "step": 5794 }, { "epoch": 56.19277108433735, "grad_norm": 7.789444446563721, "learning_rate": 7.186893203883496e-05, "loss": 0.3421, "step": 5795 }, { "epoch": 56.202409638554215, "grad_norm": 5.066715717315674, "learning_rate": 7.186407766990291e-05, "loss": 0.2183, "step": 5796 }, { "epoch": 56.212048192771086, "grad_norm": 5.810678958892822, "learning_rate": 7.185922330097089e-05, "loss": 0.3333, "step": 5797 }, { "epoch": 56.22168674698795, "grad_norm": 13.804726600646973, "learning_rate": 7.185436893203884e-05, "loss": 0.2908, "step": 5798 }, { "epoch": 56.23132530120482, "grad_norm": 13.568094253540039, "learning_rate": 7.18495145631068e-05, "loss": 0.2424, "step": 5799 }, { "epoch": 56.24096385542169, "grad_norm": 11.362260818481445, "learning_rate": 7.184466019417476e-05, "loss": 0.3212, "step": 5800 }, { "epoch": 56.25060240963855, "grad_norm": 7.392895698547363, "learning_rate": 7.183980582524273e-05, "loss": 0.4492, "step": 5801 }, { "epoch": 56.26024096385542, "grad_norm": 9.773407936096191, "learning_rate": 7.183495145631069e-05, "loss": 0.257, "step": 5802 }, { "epoch": 56.26987951807229, "grad_norm": 6.854273319244385, "learning_rate": 7.183009708737865e-05, "loss": 0.2154, "step": 5803 }, { "epoch": 56.27951807228916, "grad_norm": 11.74950122833252, "learning_rate": 7.18252427184466e-05, "loss": 0.418, "step": 5804 }, { "epoch": 56.28915662650602, "grad_norm": 6.252941131591797, "learning_rate": 7.182038834951456e-05, "loss": 0.269, "step": 5805 }, { "epoch": 56.29879518072289, "grad_norm": 8.65678596496582, "learning_rate": 7.181553398058253e-05, "loss": 0.3051, "step": 5806 }, { "epoch": 56.30843373493976, "grad_norm": 10.292756080627441, "learning_rate": 7.181067961165049e-05, "loss": 0.2774, "step": 5807 }, { "epoch": 56.318072289156625, "grad_norm": 11.609627723693848, "learning_rate": 7.180582524271845e-05, "loss": 0.231, "step": 5808 }, { "epoch": 56.327710843373495, "grad_norm": 11.074420928955078, "learning_rate": 7.18009708737864e-05, "loss": 0.3206, "step": 5809 }, { "epoch": 56.33734939759036, "grad_norm": 15.014052391052246, "learning_rate": 7.179611650485438e-05, "loss": 0.3289, "step": 5810 }, { "epoch": 56.34698795180723, "grad_norm": 13.809057235717773, "learning_rate": 7.179126213592233e-05, "loss": 0.3656, "step": 5811 }, { "epoch": 56.3566265060241, "grad_norm": 5.957476615905762, "learning_rate": 7.178640776699029e-05, "loss": 0.3038, "step": 5812 }, { "epoch": 56.36626506024096, "grad_norm": 12.19198226928711, "learning_rate": 7.178155339805825e-05, "loss": 0.3033, "step": 5813 }, { "epoch": 56.37590361445783, "grad_norm": 12.64076042175293, "learning_rate": 7.177669902912622e-05, "loss": 0.517, "step": 5814 }, { "epoch": 56.3855421686747, "grad_norm": 7.1970696449279785, "learning_rate": 7.177184466019418e-05, "loss": 0.1661, "step": 5815 }, { "epoch": 56.39518072289157, "grad_norm": 9.189763069152832, "learning_rate": 7.176699029126214e-05, "loss": 0.3607, "step": 5816 }, { "epoch": 56.40481927710843, "grad_norm": 3.8362653255462646, "learning_rate": 7.17621359223301e-05, "loss": 0.1634, "step": 5817 }, { "epoch": 56.4144578313253, "grad_norm": 10.8461332321167, "learning_rate": 7.175728155339807e-05, "loss": 0.4437, "step": 5818 }, { "epoch": 56.42409638554217, "grad_norm": 8.594403266906738, "learning_rate": 7.175242718446602e-05, "loss": 0.2751, "step": 5819 }, { "epoch": 56.433734939759034, "grad_norm": 7.384901523590088, "learning_rate": 7.174757281553398e-05, "loss": 0.2299, "step": 5820 }, { "epoch": 56.443373493975905, "grad_norm": 8.885339736938477, "learning_rate": 7.174271844660194e-05, "loss": 0.4613, "step": 5821 }, { "epoch": 56.45301204819277, "grad_norm": 9.210165023803711, "learning_rate": 7.17378640776699e-05, "loss": 0.5643, "step": 5822 }, { "epoch": 56.46265060240964, "grad_norm": 7.744014739990234, "learning_rate": 7.173300970873787e-05, "loss": 0.4147, "step": 5823 }, { "epoch": 56.47228915662651, "grad_norm": 6.621869087219238, "learning_rate": 7.172815533980583e-05, "loss": 0.2839, "step": 5824 }, { "epoch": 56.48192771084337, "grad_norm": 10.292949676513672, "learning_rate": 7.172330097087378e-05, "loss": 0.4541, "step": 5825 }, { "epoch": 56.49156626506024, "grad_norm": 8.602326393127441, "learning_rate": 7.171844660194174e-05, "loss": 0.2284, "step": 5826 }, { "epoch": 56.501204819277106, "grad_norm": 5.292244911193848, "learning_rate": 7.171359223300971e-05, "loss": 0.3277, "step": 5827 }, { "epoch": 56.51084337349398, "grad_norm": 9.743860244750977, "learning_rate": 7.170873786407767e-05, "loss": 0.2815, "step": 5828 }, { "epoch": 56.52048192771084, "grad_norm": 5.778561115264893, "learning_rate": 7.170388349514563e-05, "loss": 0.2264, "step": 5829 }, { "epoch": 56.53012048192771, "grad_norm": 10.028334617614746, "learning_rate": 7.169902912621359e-05, "loss": 0.5483, "step": 5830 }, { "epoch": 56.53975903614458, "grad_norm": 4.870185375213623, "learning_rate": 7.169417475728156e-05, "loss": 0.1792, "step": 5831 }, { "epoch": 56.549397590361444, "grad_norm": 7.319553852081299, "learning_rate": 7.168932038834953e-05, "loss": 0.4249, "step": 5832 }, { "epoch": 56.559036144578315, "grad_norm": 12.327747344970703, "learning_rate": 7.168446601941749e-05, "loss": 0.2309, "step": 5833 }, { "epoch": 56.56867469879518, "grad_norm": 7.45686674118042, "learning_rate": 7.167961165048544e-05, "loss": 0.2274, "step": 5834 }, { "epoch": 56.57831325301205, "grad_norm": 7.159858226776123, "learning_rate": 7.16747572815534e-05, "loss": 0.2331, "step": 5835 }, { "epoch": 56.58795180722892, "grad_norm": 23.59441375732422, "learning_rate": 7.166990291262137e-05, "loss": 0.4388, "step": 5836 }, { "epoch": 56.59759036144578, "grad_norm": 11.072267532348633, "learning_rate": 7.166504854368933e-05, "loss": 0.4606, "step": 5837 }, { "epoch": 56.60722891566265, "grad_norm": 4.519339084625244, "learning_rate": 7.166019417475729e-05, "loss": 0.0864, "step": 5838 }, { "epoch": 56.616867469879516, "grad_norm": 5.460470676422119, "learning_rate": 7.165533980582525e-05, "loss": 0.2264, "step": 5839 }, { "epoch": 56.626506024096386, "grad_norm": 12.047262191772461, "learning_rate": 7.16504854368932e-05, "loss": 0.2602, "step": 5840 }, { "epoch": 56.63614457831325, "grad_norm": 10.550187110900879, "learning_rate": 7.164563106796118e-05, "loss": 0.2923, "step": 5841 }, { "epoch": 56.64578313253012, "grad_norm": 9.54238510131836, "learning_rate": 7.164077669902913e-05, "loss": 0.3461, "step": 5842 }, { "epoch": 56.65542168674699, "grad_norm": 9.640588760375977, "learning_rate": 7.163592233009709e-05, "loss": 0.319, "step": 5843 }, { "epoch": 56.665060240963854, "grad_norm": 7.077660083770752, "learning_rate": 7.163106796116505e-05, "loss": 0.3005, "step": 5844 }, { "epoch": 56.674698795180724, "grad_norm": 7.656529903411865, "learning_rate": 7.162621359223302e-05, "loss": 0.1793, "step": 5845 }, { "epoch": 56.68433734939759, "grad_norm": 7.45171594619751, "learning_rate": 7.162135922330098e-05, "loss": 0.3105, "step": 5846 }, { "epoch": 56.69397590361446, "grad_norm": 8.894721984863281, "learning_rate": 7.161650485436894e-05, "loss": 0.2301, "step": 5847 }, { "epoch": 56.70361445783133, "grad_norm": 9.324665069580078, "learning_rate": 7.16116504854369e-05, "loss": 0.4218, "step": 5848 }, { "epoch": 56.71325301204819, "grad_norm": 6.046548366546631, "learning_rate": 7.160679611650486e-05, "loss": 0.2446, "step": 5849 }, { "epoch": 56.72289156626506, "grad_norm": 9.617477416992188, "learning_rate": 7.160194174757282e-05, "loss": 0.4019, "step": 5850 }, { "epoch": 56.732530120481925, "grad_norm": 7.17501163482666, "learning_rate": 7.159708737864078e-05, "loss": 0.2516, "step": 5851 }, { "epoch": 56.742168674698796, "grad_norm": 12.406990051269531, "learning_rate": 7.159223300970874e-05, "loss": 0.6859, "step": 5852 }, { "epoch": 56.75180722891566, "grad_norm": 6.755099773406982, "learning_rate": 7.158737864077671e-05, "loss": 0.2935, "step": 5853 }, { "epoch": 56.76144578313253, "grad_norm": 7.481417179107666, "learning_rate": 7.158252427184467e-05, "loss": 0.4838, "step": 5854 }, { "epoch": 56.7710843373494, "grad_norm": 7.238933086395264, "learning_rate": 7.157766990291262e-05, "loss": 0.2793, "step": 5855 }, { "epoch": 56.78072289156626, "grad_norm": 7.226253509521484, "learning_rate": 7.157281553398058e-05, "loss": 0.1862, "step": 5856 }, { "epoch": 56.790361445783134, "grad_norm": 8.896430969238281, "learning_rate": 7.156796116504854e-05, "loss": 0.2241, "step": 5857 }, { "epoch": 56.8, "grad_norm": 9.288312911987305, "learning_rate": 7.156310679611651e-05, "loss": 0.1878, "step": 5858 }, { "epoch": 56.80963855421687, "grad_norm": 5.8184380531311035, "learning_rate": 7.155825242718447e-05, "loss": 0.2225, "step": 5859 }, { "epoch": 56.81927710843374, "grad_norm": 9.257943153381348, "learning_rate": 7.155339805825243e-05, "loss": 0.3587, "step": 5860 }, { "epoch": 56.8289156626506, "grad_norm": 6.604793071746826, "learning_rate": 7.154854368932038e-05, "loss": 0.1504, "step": 5861 }, { "epoch": 56.83855421686747, "grad_norm": 4.454575538635254, "learning_rate": 7.154368932038836e-05, "loss": 0.2838, "step": 5862 }, { "epoch": 56.848192771084335, "grad_norm": 3.9078705310821533, "learning_rate": 7.153883495145631e-05, "loss": 0.1392, "step": 5863 }, { "epoch": 56.857831325301206, "grad_norm": 12.218320846557617, "learning_rate": 7.153398058252427e-05, "loss": 0.2442, "step": 5864 }, { "epoch": 56.86746987951807, "grad_norm": 10.40406608581543, "learning_rate": 7.152912621359223e-05, "loss": 0.219, "step": 5865 }, { "epoch": 56.87710843373494, "grad_norm": 5.830802917480469, "learning_rate": 7.15242718446602e-05, "loss": 0.1852, "step": 5866 }, { "epoch": 56.88674698795181, "grad_norm": 5.726907730102539, "learning_rate": 7.151941747572816e-05, "loss": 0.2548, "step": 5867 }, { "epoch": 56.89638554216867, "grad_norm": 9.034757614135742, "learning_rate": 7.151456310679612e-05, "loss": 0.3311, "step": 5868 }, { "epoch": 56.90602409638554, "grad_norm": 7.533843040466309, "learning_rate": 7.150970873786407e-05, "loss": 0.4986, "step": 5869 }, { "epoch": 56.91566265060241, "grad_norm": 4.425135135650635, "learning_rate": 7.150485436893205e-05, "loss": 0.2091, "step": 5870 }, { "epoch": 56.92530120481928, "grad_norm": 5.864797115325928, "learning_rate": 7.15e-05, "loss": 0.2489, "step": 5871 }, { "epoch": 56.93493975903615, "grad_norm": 17.941951751708984, "learning_rate": 7.149514563106796e-05, "loss": 0.2647, "step": 5872 }, { "epoch": 56.94457831325301, "grad_norm": 12.345484733581543, "learning_rate": 7.149029126213592e-05, "loss": 0.3493, "step": 5873 }, { "epoch": 56.95421686746988, "grad_norm": 9.171649932861328, "learning_rate": 7.148543689320388e-05, "loss": 0.4696, "step": 5874 }, { "epoch": 56.963855421686745, "grad_norm": 16.928218841552734, "learning_rate": 7.148058252427185e-05, "loss": 0.614, "step": 5875 }, { "epoch": 56.973493975903615, "grad_norm": 5.2492241859436035, "learning_rate": 7.14757281553398e-05, "loss": 0.1889, "step": 5876 }, { "epoch": 56.98313253012048, "grad_norm": 11.205419540405273, "learning_rate": 7.147087378640776e-05, "loss": 0.4413, "step": 5877 }, { "epoch": 56.99277108433735, "grad_norm": 7.071478843688965, "learning_rate": 7.146601941747572e-05, "loss": 0.3589, "step": 5878 }, { "epoch": 57.00843373493976, "grad_norm": 35.0014762878418, "learning_rate": 7.146116504854369e-05, "loss": 0.4847, "step": 5879 }, { "epoch": 57.01807228915663, "grad_norm": 12.199165344238281, "learning_rate": 7.145631067961165e-05, "loss": 0.3146, "step": 5880 }, { "epoch": 57.02771084337349, "grad_norm": 10.15896987915039, "learning_rate": 7.145145631067961e-05, "loss": 0.294, "step": 5881 }, { "epoch": 57.03734939759036, "grad_norm": 19.968046188354492, "learning_rate": 7.144660194174757e-05, "loss": 0.2111, "step": 5882 }, { "epoch": 57.04698795180723, "grad_norm": 24.73168182373047, "learning_rate": 7.144174757281554e-05, "loss": 0.3159, "step": 5883 }, { "epoch": 57.056626506024095, "grad_norm": 5.519067764282227, "learning_rate": 7.143689320388351e-05, "loss": 0.5859, "step": 5884 }, { "epoch": 57.066265060240966, "grad_norm": 18.656497955322266, "learning_rate": 7.143203883495147e-05, "loss": 0.1583, "step": 5885 }, { "epoch": 57.07590361445783, "grad_norm": 10.34675407409668, "learning_rate": 7.142718446601942e-05, "loss": 0.2449, "step": 5886 }, { "epoch": 57.0855421686747, "grad_norm": 22.1093807220459, "learning_rate": 7.142233009708738e-05, "loss": 0.26, "step": 5887 }, { "epoch": 57.09518072289157, "grad_norm": 31.642772674560547, "learning_rate": 7.141747572815535e-05, "loss": 0.5104, "step": 5888 }, { "epoch": 57.10481927710843, "grad_norm": 65.09416198730469, "learning_rate": 7.141262135922331e-05, "loss": 0.2366, "step": 5889 }, { "epoch": 57.1144578313253, "grad_norm": 6.645605087280273, "learning_rate": 7.140776699029127e-05, "loss": 0.2896, "step": 5890 }, { "epoch": 57.12409638554217, "grad_norm": 6.370213508605957, "learning_rate": 7.140291262135923e-05, "loss": 0.4092, "step": 5891 }, { "epoch": 57.13373493975904, "grad_norm": 23.345962524414062, "learning_rate": 7.13980582524272e-05, "loss": 0.2759, "step": 5892 }, { "epoch": 57.1433734939759, "grad_norm": 4.080952167510986, "learning_rate": 7.139320388349516e-05, "loss": 0.2193, "step": 5893 }, { "epoch": 57.15301204819277, "grad_norm": 23.500873565673828, "learning_rate": 7.138834951456311e-05, "loss": 0.3193, "step": 5894 }, { "epoch": 57.16265060240964, "grad_norm": 14.565120697021484, "learning_rate": 7.138349514563107e-05, "loss": 0.4249, "step": 5895 }, { "epoch": 57.172289156626505, "grad_norm": 7.905410289764404, "learning_rate": 7.137864077669903e-05, "loss": 0.2595, "step": 5896 }, { "epoch": 57.181927710843375, "grad_norm": 44.893524169921875, "learning_rate": 7.1373786407767e-05, "loss": 0.3205, "step": 5897 }, { "epoch": 57.19156626506024, "grad_norm": 19.742748260498047, "learning_rate": 7.136893203883496e-05, "loss": 0.221, "step": 5898 }, { "epoch": 57.20120481927711, "grad_norm": 2.715893030166626, "learning_rate": 7.136407766990292e-05, "loss": 0.3588, "step": 5899 }, { "epoch": 57.21084337349398, "grad_norm": 108.95024108886719, "learning_rate": 7.135922330097087e-05, "loss": 0.2044, "step": 5900 }, { "epoch": 57.22048192771084, "grad_norm": 3.413957118988037, "learning_rate": 7.135436893203884e-05, "loss": 0.2053, "step": 5901 }, { "epoch": 57.23012048192771, "grad_norm": 25.783634185791016, "learning_rate": 7.13495145631068e-05, "loss": 0.307, "step": 5902 }, { "epoch": 57.23975903614458, "grad_norm": 28.90869903564453, "learning_rate": 7.134466019417476e-05, "loss": 0.3054, "step": 5903 }, { "epoch": 57.24939759036145, "grad_norm": 18.959680557250977, "learning_rate": 7.133980582524272e-05, "loss": 0.384, "step": 5904 }, { "epoch": 57.25903614457831, "grad_norm": 2.5676076412200928, "learning_rate": 7.133495145631069e-05, "loss": 0.208, "step": 5905 }, { "epoch": 57.26867469879518, "grad_norm": 4.6831746101379395, "learning_rate": 7.133009708737865e-05, "loss": 0.2058, "step": 5906 }, { "epoch": 57.27831325301205, "grad_norm": 2.3476669788360596, "learning_rate": 7.13252427184466e-05, "loss": 0.2518, "step": 5907 }, { "epoch": 57.287951807228914, "grad_norm": 48.1645622253418, "learning_rate": 7.132038834951456e-05, "loss": 0.2615, "step": 5908 }, { "epoch": 57.297590361445785, "grad_norm": 13.551375389099121, "learning_rate": 7.131553398058253e-05, "loss": 0.096, "step": 5909 }, { "epoch": 57.30722891566265, "grad_norm": 21.068397521972656, "learning_rate": 7.131067961165049e-05, "loss": 0.2674, "step": 5910 }, { "epoch": 57.31686746987952, "grad_norm": 3.6894710063934326, "learning_rate": 7.130582524271845e-05, "loss": 0.3538, "step": 5911 }, { "epoch": 57.32650602409639, "grad_norm": 31.208751678466797, "learning_rate": 7.130097087378641e-05, "loss": 0.3065, "step": 5912 }, { "epoch": 57.33614457831325, "grad_norm": 28.668811798095703, "learning_rate": 7.129611650485436e-05, "loss": 0.1947, "step": 5913 }, { "epoch": 57.34578313253012, "grad_norm": 33.27718734741211, "learning_rate": 7.129126213592234e-05, "loss": 0.2367, "step": 5914 }, { "epoch": 57.355421686746986, "grad_norm": 2.2028210163116455, "learning_rate": 7.12864077669903e-05, "loss": 0.1763, "step": 5915 }, { "epoch": 57.36506024096386, "grad_norm": 2.8801727294921875, "learning_rate": 7.128155339805825e-05, "loss": 0.5282, "step": 5916 }, { "epoch": 57.37469879518072, "grad_norm": 13.316731452941895, "learning_rate": 7.127669902912621e-05, "loss": 0.2279, "step": 5917 }, { "epoch": 57.38433734939759, "grad_norm": 18.15968894958496, "learning_rate": 7.127184466019418e-05, "loss": 0.3043, "step": 5918 }, { "epoch": 57.39397590361446, "grad_norm": 366.03759765625, "learning_rate": 7.126699029126214e-05, "loss": 0.4113, "step": 5919 }, { "epoch": 57.403614457831324, "grad_norm": 3.885953903198242, "learning_rate": 7.12621359223301e-05, "loss": 0.2191, "step": 5920 }, { "epoch": 57.413253012048195, "grad_norm": 33.265953063964844, "learning_rate": 7.125728155339805e-05, "loss": 0.2775, "step": 5921 }, { "epoch": 57.42289156626506, "grad_norm": 22.363534927368164, "learning_rate": 7.125242718446603e-05, "loss": 0.4469, "step": 5922 }, { "epoch": 57.43253012048193, "grad_norm": 8.310577392578125, "learning_rate": 7.124757281553398e-05, "loss": 0.1949, "step": 5923 }, { "epoch": 57.44216867469879, "grad_norm": 83.1961898803711, "learning_rate": 7.124271844660194e-05, "loss": 0.4258, "step": 5924 }, { "epoch": 57.45180722891566, "grad_norm": 37.539730072021484, "learning_rate": 7.12378640776699e-05, "loss": 0.4069, "step": 5925 }, { "epoch": 57.46144578313253, "grad_norm": 26.40621566772461, "learning_rate": 7.123300970873787e-05, "loss": 0.382, "step": 5926 }, { "epoch": 57.471084337349396, "grad_norm": 25.279598236083984, "learning_rate": 7.122815533980583e-05, "loss": 0.3352, "step": 5927 }, { "epoch": 57.480722891566266, "grad_norm": 28.034622192382812, "learning_rate": 7.122330097087379e-05, "loss": 0.2555, "step": 5928 }, { "epoch": 57.49036144578313, "grad_norm": 29.1401424407959, "learning_rate": 7.121844660194174e-05, "loss": 0.2262, "step": 5929 }, { "epoch": 57.5, "grad_norm": 15.42707347869873, "learning_rate": 7.12135922330097e-05, "loss": 0.5392, "step": 5930 }, { "epoch": 57.50963855421687, "grad_norm": 99.32242584228516, "learning_rate": 7.120873786407767e-05, "loss": 0.2301, "step": 5931 }, { "epoch": 57.519277108433734, "grad_norm": 9.842738151550293, "learning_rate": 7.120388349514563e-05, "loss": 0.263, "step": 5932 }, { "epoch": 57.528915662650604, "grad_norm": 35.93917465209961, "learning_rate": 7.11990291262136e-05, "loss": 0.3076, "step": 5933 }, { "epoch": 57.53855421686747, "grad_norm": 25.178178787231445, "learning_rate": 7.119417475728156e-05, "loss": 0.3614, "step": 5934 }, { "epoch": 57.54819277108434, "grad_norm": 3.282830238342285, "learning_rate": 7.118932038834952e-05, "loss": 0.4229, "step": 5935 }, { "epoch": 57.55783132530121, "grad_norm": 4.155424118041992, "learning_rate": 7.118446601941749e-05, "loss": 0.1651, "step": 5936 }, { "epoch": 57.56746987951807, "grad_norm": 23.1149845123291, "learning_rate": 7.117961165048545e-05, "loss": 0.3844, "step": 5937 }, { "epoch": 57.57710843373494, "grad_norm": 15.912914276123047, "learning_rate": 7.11747572815534e-05, "loss": 0.3591, "step": 5938 }, { "epoch": 57.586746987951805, "grad_norm": 2.7474968433380127, "learning_rate": 7.116990291262136e-05, "loss": 0.2241, "step": 5939 }, { "epoch": 57.596385542168676, "grad_norm": 24.816951751708984, "learning_rate": 7.116504854368933e-05, "loss": 0.4684, "step": 5940 }, { "epoch": 57.60602409638554, "grad_norm": 24.677165985107422, "learning_rate": 7.116019417475729e-05, "loss": 0.2167, "step": 5941 }, { "epoch": 57.61566265060241, "grad_norm": 12.69338607788086, "learning_rate": 7.115533980582525e-05, "loss": 0.4329, "step": 5942 }, { "epoch": 57.62530120481928, "grad_norm": 47.633094787597656, "learning_rate": 7.11504854368932e-05, "loss": 0.281, "step": 5943 }, { "epoch": 57.63493975903614, "grad_norm": 14.884298324584961, "learning_rate": 7.114563106796118e-05, "loss": 0.2837, "step": 5944 }, { "epoch": 57.644578313253014, "grad_norm": 66.37792205810547, "learning_rate": 7.114077669902913e-05, "loss": 0.3808, "step": 5945 }, { "epoch": 57.65421686746988, "grad_norm": 14.393485069274902, "learning_rate": 7.113592233009709e-05, "loss": 0.3974, "step": 5946 }, { "epoch": 57.66385542168675, "grad_norm": 34.26787567138672, "learning_rate": 7.113106796116505e-05, "loss": 0.3437, "step": 5947 }, { "epoch": 57.67349397590361, "grad_norm": 6.152584552764893, "learning_rate": 7.112621359223302e-05, "loss": 0.4124, "step": 5948 }, { "epoch": 57.68313253012048, "grad_norm": 15.192813873291016, "learning_rate": 7.112135922330098e-05, "loss": 0.3615, "step": 5949 }, { "epoch": 57.69277108433735, "grad_norm": 10.177088737487793, "learning_rate": 7.111650485436894e-05, "loss": 0.3128, "step": 5950 }, { "epoch": 57.702409638554215, "grad_norm": 11.486393928527832, "learning_rate": 7.11116504854369e-05, "loss": 0.2859, "step": 5951 }, { "epoch": 57.712048192771086, "grad_norm": 1.4642585515975952, "learning_rate": 7.110679611650485e-05, "loss": 0.3002, "step": 5952 }, { "epoch": 57.72168674698795, "grad_norm": 11.883005142211914, "learning_rate": 7.110194174757282e-05, "loss": 0.3564, "step": 5953 }, { "epoch": 57.73132530120482, "grad_norm": 14.111127853393555, "learning_rate": 7.109708737864078e-05, "loss": 0.4219, "step": 5954 }, { "epoch": 57.74096385542169, "grad_norm": 2.0021142959594727, "learning_rate": 7.109223300970874e-05, "loss": 0.1824, "step": 5955 }, { "epoch": 57.75060240963855, "grad_norm": 46.199642181396484, "learning_rate": 7.10873786407767e-05, "loss": 0.2427, "step": 5956 }, { "epoch": 57.76024096385542, "grad_norm": 10.259454727172852, "learning_rate": 7.108252427184467e-05, "loss": 0.2041, "step": 5957 }, { "epoch": 57.76987951807229, "grad_norm": 61.653846740722656, "learning_rate": 7.107766990291263e-05, "loss": 0.4671, "step": 5958 }, { "epoch": 57.77951807228916, "grad_norm": 12.274060249328613, "learning_rate": 7.107281553398058e-05, "loss": 0.3381, "step": 5959 }, { "epoch": 57.78915662650603, "grad_norm": 76.5816650390625, "learning_rate": 7.106796116504854e-05, "loss": 0.3664, "step": 5960 }, { "epoch": 57.79879518072289, "grad_norm": 4.601001262664795, "learning_rate": 7.106310679611651e-05, "loss": 0.3052, "step": 5961 }, { "epoch": 57.80843373493976, "grad_norm": 3.45884370803833, "learning_rate": 7.105825242718447e-05, "loss": 0.2587, "step": 5962 }, { "epoch": 57.818072289156625, "grad_norm": 31.64818000793457, "learning_rate": 7.105339805825243e-05, "loss": 0.5442, "step": 5963 }, { "epoch": 57.827710843373495, "grad_norm": 57.13949203491211, "learning_rate": 7.104854368932039e-05, "loss": 0.1901, "step": 5964 }, { "epoch": 57.83734939759036, "grad_norm": 8.51352596282959, "learning_rate": 7.104368932038836e-05, "loss": 0.2717, "step": 5965 }, { "epoch": 57.84698795180723, "grad_norm": 20.861482620239258, "learning_rate": 7.103883495145632e-05, "loss": 0.2986, "step": 5966 }, { "epoch": 57.8566265060241, "grad_norm": 42.378501892089844, "learning_rate": 7.103398058252427e-05, "loss": 0.3191, "step": 5967 }, { "epoch": 57.86626506024096, "grad_norm": 81.85310363769531, "learning_rate": 7.102912621359223e-05, "loss": 0.2664, "step": 5968 }, { "epoch": 57.87590361445783, "grad_norm": 24.744544982910156, "learning_rate": 7.102427184466019e-05, "loss": 0.4738, "step": 5969 }, { "epoch": 57.8855421686747, "grad_norm": 4.036928653717041, "learning_rate": 7.101941747572816e-05, "loss": 0.3231, "step": 5970 }, { "epoch": 57.89518072289157, "grad_norm": 7.954691410064697, "learning_rate": 7.101456310679612e-05, "loss": 0.1931, "step": 5971 }, { "epoch": 57.90481927710843, "grad_norm": 72.40713500976562, "learning_rate": 7.100970873786408e-05, "loss": 0.2257, "step": 5972 }, { "epoch": 57.9144578313253, "grad_norm": 10.289909362792969, "learning_rate": 7.100485436893203e-05, "loss": 0.2511, "step": 5973 }, { "epoch": 57.92409638554217, "grad_norm": 4.5219221115112305, "learning_rate": 7.1e-05, "loss": 0.3511, "step": 5974 }, { "epoch": 57.933734939759034, "grad_norm": 2.202993392944336, "learning_rate": 7.099514563106796e-05, "loss": 0.4291, "step": 5975 }, { "epoch": 57.943373493975905, "grad_norm": 24.943252563476562, "learning_rate": 7.099029126213592e-05, "loss": 0.3503, "step": 5976 }, { "epoch": 57.95301204819277, "grad_norm": 118.09846496582031, "learning_rate": 7.098543689320388e-05, "loss": 0.2483, "step": 5977 }, { "epoch": 57.96265060240964, "grad_norm": 47.56529235839844, "learning_rate": 7.098058252427185e-05, "loss": 0.3816, "step": 5978 }, { "epoch": 57.97228915662651, "grad_norm": 11.98597526550293, "learning_rate": 7.097572815533981e-05, "loss": 0.2276, "step": 5979 }, { "epoch": 57.98192771084337, "grad_norm": 11.118152618408203, "learning_rate": 7.097087378640776e-05, "loss": 0.2909, "step": 5980 }, { "epoch": 57.99156626506024, "grad_norm": 15.218603134155273, "learning_rate": 7.096601941747572e-05, "loss": 0.3071, "step": 5981 }, { "epoch": 58.00722891566265, "grad_norm": 4.117258548736572, "learning_rate": 7.09611650485437e-05, "loss": 0.1689, "step": 5982 }, { "epoch": 58.01686746987952, "grad_norm": 32.744937896728516, "learning_rate": 7.095631067961165e-05, "loss": 0.1767, "step": 5983 }, { "epoch": 58.026506024096385, "grad_norm": 11.70785140991211, "learning_rate": 7.095145631067962e-05, "loss": 0.1908, "step": 5984 }, { "epoch": 58.036144578313255, "grad_norm": 8.060894966125488, "learning_rate": 7.094660194174758e-05, "loss": 0.2788, "step": 5985 }, { "epoch": 58.04578313253012, "grad_norm": 12.986445426940918, "learning_rate": 7.094174757281554e-05, "loss": 0.2706, "step": 5986 }, { "epoch": 58.05542168674699, "grad_norm": 4.905580997467041, "learning_rate": 7.093689320388351e-05, "loss": 0.2729, "step": 5987 }, { "epoch": 58.06506024096385, "grad_norm": 6.758049488067627, "learning_rate": 7.093203883495147e-05, "loss": 0.2138, "step": 5988 }, { "epoch": 58.07469879518072, "grad_norm": 3.7222702503204346, "learning_rate": 7.092718446601943e-05, "loss": 0.1819, "step": 5989 }, { "epoch": 58.08433734939759, "grad_norm": 23.7821102142334, "learning_rate": 7.092233009708738e-05, "loss": 0.3071, "step": 5990 }, { "epoch": 58.09397590361446, "grad_norm": 7.77810525894165, "learning_rate": 7.091747572815534e-05, "loss": 0.1738, "step": 5991 }, { "epoch": 58.10361445783133, "grad_norm": 12.457103729248047, "learning_rate": 7.091262135922331e-05, "loss": 0.4792, "step": 5992 }, { "epoch": 58.11325301204819, "grad_norm": 11.137775421142578, "learning_rate": 7.090776699029127e-05, "loss": 0.2151, "step": 5993 }, { "epoch": 58.12289156626506, "grad_norm": 9.158154487609863, "learning_rate": 7.090291262135923e-05, "loss": 0.3157, "step": 5994 }, { "epoch": 58.13253012048193, "grad_norm": 7.4003376960754395, "learning_rate": 7.089805825242719e-05, "loss": 0.3828, "step": 5995 }, { "epoch": 58.142168674698794, "grad_norm": 10.50894546508789, "learning_rate": 7.089320388349516e-05, "loss": 0.2403, "step": 5996 }, { "epoch": 58.151807228915665, "grad_norm": 22.28265953063965, "learning_rate": 7.088834951456311e-05, "loss": 0.3557, "step": 5997 }, { "epoch": 58.16144578313253, "grad_norm": 5.217268466949463, "learning_rate": 7.088349514563107e-05, "loss": 0.3941, "step": 5998 }, { "epoch": 58.1710843373494, "grad_norm": 14.396159172058105, "learning_rate": 7.087864077669903e-05, "loss": 0.221, "step": 5999 }, { "epoch": 58.18072289156626, "grad_norm": 6.872837543487549, "learning_rate": 7.0873786407767e-05, "loss": 0.265, "step": 6000 }, { "epoch": 58.19036144578313, "grad_norm": 6.8125481605529785, "learning_rate": 7.086893203883496e-05, "loss": 0.2037, "step": 6001 }, { "epoch": 58.2, "grad_norm": 7.974712371826172, "learning_rate": 7.086407766990292e-05, "loss": 0.2962, "step": 6002 }, { "epoch": 58.209638554216866, "grad_norm": 20.337661743164062, "learning_rate": 7.085922330097087e-05, "loss": 0.3963, "step": 6003 }, { "epoch": 58.21927710843374, "grad_norm": 4.441798210144043, "learning_rate": 7.085436893203885e-05, "loss": 0.2353, "step": 6004 }, { "epoch": 58.2289156626506, "grad_norm": 18.371173858642578, "learning_rate": 7.08495145631068e-05, "loss": 0.3621, "step": 6005 }, { "epoch": 58.23855421686747, "grad_norm": 33.346778869628906, "learning_rate": 7.084466019417476e-05, "loss": 0.3313, "step": 6006 }, { "epoch": 58.24819277108434, "grad_norm": 8.904777526855469, "learning_rate": 7.083980582524272e-05, "loss": 0.2218, "step": 6007 }, { "epoch": 58.257831325301204, "grad_norm": 15.274462699890137, "learning_rate": 7.083495145631068e-05, "loss": 0.2459, "step": 6008 }, { "epoch": 58.267469879518075, "grad_norm": 21.803539276123047, "learning_rate": 7.083009708737865e-05, "loss": 0.3724, "step": 6009 }, { "epoch": 58.27710843373494, "grad_norm": 1.7218002080917358, "learning_rate": 7.08252427184466e-05, "loss": 0.3184, "step": 6010 }, { "epoch": 58.28674698795181, "grad_norm": 4.798499584197998, "learning_rate": 7.082038834951456e-05, "loss": 0.4593, "step": 6011 }, { "epoch": 58.29638554216867, "grad_norm": 15.586691856384277, "learning_rate": 7.081553398058252e-05, "loss": 0.2081, "step": 6012 }, { "epoch": 58.30602409638554, "grad_norm": 5.368218898773193, "learning_rate": 7.081067961165049e-05, "loss": 0.3639, "step": 6013 }, { "epoch": 58.31566265060241, "grad_norm": 17.63429832458496, "learning_rate": 7.080582524271845e-05, "loss": 0.3782, "step": 6014 }, { "epoch": 58.325301204819276, "grad_norm": 10.313745498657227, "learning_rate": 7.080097087378641e-05, "loss": 0.1856, "step": 6015 }, { "epoch": 58.334939759036146, "grad_norm": 10.429176330566406, "learning_rate": 7.079611650485437e-05, "loss": 0.2821, "step": 6016 }, { "epoch": 58.34457831325301, "grad_norm": 4.900957107543945, "learning_rate": 7.079126213592234e-05, "loss": 0.3814, "step": 6017 }, { "epoch": 58.35421686746988, "grad_norm": 32.809024810791016, "learning_rate": 7.07864077669903e-05, "loss": 0.2063, "step": 6018 }, { "epoch": 58.36385542168675, "grad_norm": 8.110899925231934, "learning_rate": 7.078155339805825e-05, "loss": 0.2915, "step": 6019 }, { "epoch": 58.373493975903614, "grad_norm": 5.200425148010254, "learning_rate": 7.077669902912621e-05, "loss": 0.2048, "step": 6020 }, { "epoch": 58.383132530120484, "grad_norm": 15.125365257263184, "learning_rate": 7.077184466019418e-05, "loss": 0.3735, "step": 6021 }, { "epoch": 58.39277108433735, "grad_norm": 36.91615295410156, "learning_rate": 7.076699029126214e-05, "loss": 0.3521, "step": 6022 }, { "epoch": 58.40240963855422, "grad_norm": 7.6955132484436035, "learning_rate": 7.07621359223301e-05, "loss": 0.3023, "step": 6023 }, { "epoch": 58.41204819277108, "grad_norm": 8.902777671813965, "learning_rate": 7.075728155339806e-05, "loss": 0.3479, "step": 6024 }, { "epoch": 58.42168674698795, "grad_norm": 4.098647117614746, "learning_rate": 7.075242718446601e-05, "loss": 0.4184, "step": 6025 }, { "epoch": 58.43132530120482, "grad_norm": 12.329976081848145, "learning_rate": 7.074757281553398e-05, "loss": 0.3185, "step": 6026 }, { "epoch": 58.440963855421685, "grad_norm": 59.36407470703125, "learning_rate": 7.074271844660194e-05, "loss": 0.144, "step": 6027 }, { "epoch": 58.450602409638556, "grad_norm": 14.707812309265137, "learning_rate": 7.07378640776699e-05, "loss": 0.2134, "step": 6028 }, { "epoch": 58.46024096385542, "grad_norm": 30.280986785888672, "learning_rate": 7.073300970873786e-05, "loss": 0.3322, "step": 6029 }, { "epoch": 58.46987951807229, "grad_norm": 3.9153389930725098, "learning_rate": 7.072815533980583e-05, "loss": 0.2217, "step": 6030 }, { "epoch": 58.47951807228916, "grad_norm": 20.02893829345703, "learning_rate": 7.072330097087379e-05, "loss": 0.3328, "step": 6031 }, { "epoch": 58.48915662650602, "grad_norm": 11.7337646484375, "learning_rate": 7.071844660194174e-05, "loss": 0.3594, "step": 6032 }, { "epoch": 58.498795180722894, "grad_norm": 19.667871475219727, "learning_rate": 7.07135922330097e-05, "loss": 0.3994, "step": 6033 }, { "epoch": 58.50843373493976, "grad_norm": 48.69419860839844, "learning_rate": 7.070873786407767e-05, "loss": 0.2332, "step": 6034 }, { "epoch": 58.51807228915663, "grad_norm": 44.134952545166016, "learning_rate": 7.070388349514564e-05, "loss": 0.4169, "step": 6035 }, { "epoch": 58.52771084337349, "grad_norm": 12.781841278076172, "learning_rate": 7.06990291262136e-05, "loss": 0.3306, "step": 6036 }, { "epoch": 58.53734939759036, "grad_norm": 103.7414779663086, "learning_rate": 7.069417475728156e-05, "loss": 0.2319, "step": 6037 }, { "epoch": 58.54698795180723, "grad_norm": 10.857986450195312, "learning_rate": 7.068932038834952e-05, "loss": 0.6378, "step": 6038 }, { "epoch": 58.556626506024095, "grad_norm": 63.51497268676758, "learning_rate": 7.068446601941749e-05, "loss": 0.4384, "step": 6039 }, { "epoch": 58.566265060240966, "grad_norm": 15.370553016662598, "learning_rate": 7.067961165048545e-05, "loss": 0.3895, "step": 6040 }, { "epoch": 58.57590361445783, "grad_norm": 4.1522417068481445, "learning_rate": 7.06747572815534e-05, "loss": 0.1674, "step": 6041 }, { "epoch": 58.5855421686747, "grad_norm": 4.102275848388672, "learning_rate": 7.066990291262136e-05, "loss": 0.0786, "step": 6042 }, { "epoch": 58.59518072289157, "grad_norm": 20.856788635253906, "learning_rate": 7.066504854368933e-05, "loss": 0.248, "step": 6043 }, { "epoch": 58.60481927710843, "grad_norm": 30.84920310974121, "learning_rate": 7.066019417475729e-05, "loss": 0.2215, "step": 6044 }, { "epoch": 58.6144578313253, "grad_norm": 9.357306480407715, "learning_rate": 7.065533980582525e-05, "loss": 0.2293, "step": 6045 }, { "epoch": 58.62409638554217, "grad_norm": 29.898426055908203, "learning_rate": 7.065048543689321e-05, "loss": 0.2926, "step": 6046 }, { "epoch": 58.63373493975904, "grad_norm": 46.72187042236328, "learning_rate": 7.064563106796116e-05, "loss": 0.2115, "step": 6047 }, { "epoch": 58.6433734939759, "grad_norm": 3.705967903137207, "learning_rate": 7.064077669902914e-05, "loss": 0.1731, "step": 6048 }, { "epoch": 58.65301204819277, "grad_norm": 4.428170204162598, "learning_rate": 7.06359223300971e-05, "loss": 0.1845, "step": 6049 }, { "epoch": 58.66265060240964, "grad_norm": 4.430201530456543, "learning_rate": 7.063106796116505e-05, "loss": 0.2773, "step": 6050 }, { "epoch": 58.672289156626505, "grad_norm": 3.7726547718048096, "learning_rate": 7.062621359223301e-05, "loss": 0.2792, "step": 6051 }, { "epoch": 58.681927710843375, "grad_norm": 20.317337036132812, "learning_rate": 7.062135922330098e-05, "loss": 0.5034, "step": 6052 }, { "epoch": 58.69156626506024, "grad_norm": 49.419761657714844, "learning_rate": 7.061650485436894e-05, "loss": 0.5778, "step": 6053 }, { "epoch": 58.70120481927711, "grad_norm": 5.917753219604492, "learning_rate": 7.06116504854369e-05, "loss": 0.2683, "step": 6054 }, { "epoch": 58.71084337349397, "grad_norm": 25.627952575683594, "learning_rate": 7.060679611650485e-05, "loss": 0.3089, "step": 6055 }, { "epoch": 58.72048192771084, "grad_norm": 12.737092018127441, "learning_rate": 7.060194174757283e-05, "loss": 0.2677, "step": 6056 }, { "epoch": 58.73012048192771, "grad_norm": 16.305194854736328, "learning_rate": 7.059708737864078e-05, "loss": 0.3072, "step": 6057 }, { "epoch": 58.73975903614458, "grad_norm": 42.423465728759766, "learning_rate": 7.059223300970874e-05, "loss": 0.3581, "step": 6058 }, { "epoch": 58.74939759036145, "grad_norm": 19.84027862548828, "learning_rate": 7.05873786407767e-05, "loss": 0.1809, "step": 6059 }, { "epoch": 58.75903614457831, "grad_norm": 4.43892240524292, "learning_rate": 7.058252427184467e-05, "loss": 0.2103, "step": 6060 }, { "epoch": 58.76867469879518, "grad_norm": 23.835411071777344, "learning_rate": 7.057766990291263e-05, "loss": 0.3074, "step": 6061 }, { "epoch": 58.77831325301205, "grad_norm": 4.687929630279541, "learning_rate": 7.057281553398059e-05, "loss": 0.2369, "step": 6062 }, { "epoch": 58.787951807228914, "grad_norm": 47.8568115234375, "learning_rate": 7.056796116504854e-05, "loss": 0.3719, "step": 6063 }, { "epoch": 58.797590361445785, "grad_norm": 16.354656219482422, "learning_rate": 7.05631067961165e-05, "loss": 0.2372, "step": 6064 }, { "epoch": 58.80722891566265, "grad_norm": 99.37674713134766, "learning_rate": 7.055825242718447e-05, "loss": 0.3652, "step": 6065 }, { "epoch": 58.81686746987952, "grad_norm": 4.707158088684082, "learning_rate": 7.055339805825243e-05, "loss": 0.1274, "step": 6066 }, { "epoch": 58.82650602409639, "grad_norm": 21.322256088256836, "learning_rate": 7.054854368932039e-05, "loss": 0.3948, "step": 6067 }, { "epoch": 58.83614457831325, "grad_norm": 6.514432430267334, "learning_rate": 7.054368932038835e-05, "loss": 0.2687, "step": 6068 }, { "epoch": 58.84578313253012, "grad_norm": 9.12800121307373, "learning_rate": 7.053883495145632e-05, "loss": 0.2367, "step": 6069 }, { "epoch": 58.855421686746986, "grad_norm": 1.6558384895324707, "learning_rate": 7.053398058252427e-05, "loss": 0.114, "step": 6070 }, { "epoch": 58.86506024096386, "grad_norm": 15.102873802185059, "learning_rate": 7.052912621359223e-05, "loss": 0.3324, "step": 6071 }, { "epoch": 58.87469879518072, "grad_norm": 7.894067764282227, "learning_rate": 7.052427184466019e-05, "loss": 0.2112, "step": 6072 }, { "epoch": 58.88433734939759, "grad_norm": 13.744731903076172, "learning_rate": 7.051941747572816e-05, "loss": 0.446, "step": 6073 }, { "epoch": 58.89397590361446, "grad_norm": 23.285259246826172, "learning_rate": 7.051456310679612e-05, "loss": 0.5711, "step": 6074 }, { "epoch": 58.903614457831324, "grad_norm": 10.27879810333252, "learning_rate": 7.050970873786408e-05, "loss": 0.3959, "step": 6075 }, { "epoch": 58.913253012048195, "grad_norm": 56.6749267578125, "learning_rate": 7.050485436893203e-05, "loss": 0.4134, "step": 6076 }, { "epoch": 58.92289156626506, "grad_norm": 17.76399803161621, "learning_rate": 7.05e-05, "loss": 0.2955, "step": 6077 }, { "epoch": 58.93253012048193, "grad_norm": 5.995838642120361, "learning_rate": 7.049514563106796e-05, "loss": 0.3037, "step": 6078 }, { "epoch": 58.94216867469879, "grad_norm": 45.201900482177734, "learning_rate": 7.049029126213592e-05, "loss": 0.2636, "step": 6079 }, { "epoch": 58.95180722891566, "grad_norm": 17.94347381591797, "learning_rate": 7.048543689320388e-05, "loss": 0.2287, "step": 6080 }, { "epoch": 58.96144578313253, "grad_norm": 17.564350128173828, "learning_rate": 7.048058252427184e-05, "loss": 0.1621, "step": 6081 }, { "epoch": 58.971084337349396, "grad_norm": 5.1823577880859375, "learning_rate": 7.047572815533981e-05, "loss": 0.329, "step": 6082 }, { "epoch": 58.980722891566266, "grad_norm": 9.806084632873535, "learning_rate": 7.047087378640777e-05, "loss": 0.3494, "step": 6083 }, { "epoch": 58.99036144578313, "grad_norm": 8.042252540588379, "learning_rate": 7.046601941747572e-05, "loss": 0.2847, "step": 6084 }, { "epoch": 59.006024096385545, "grad_norm": 4.665970802307129, "learning_rate": 7.04611650485437e-05, "loss": 0.1334, "step": 6085 }, { "epoch": 59.01566265060241, "grad_norm": 13.456045150756836, "learning_rate": 7.045631067961165e-05, "loss": 0.5102, "step": 6086 }, { "epoch": 59.02530120481928, "grad_norm": 24.973146438598633, "learning_rate": 7.045145631067962e-05, "loss": 0.2335, "step": 6087 }, { "epoch": 59.03493975903614, "grad_norm": 20.539506912231445, "learning_rate": 7.044660194174758e-05, "loss": 0.4454, "step": 6088 }, { "epoch": 59.04457831325301, "grad_norm": 5.866421699523926, "learning_rate": 7.044174757281554e-05, "loss": 0.4624, "step": 6089 }, { "epoch": 59.05421686746988, "grad_norm": 6.847850322723389, "learning_rate": 7.04368932038835e-05, "loss": 0.3769, "step": 6090 }, { "epoch": 59.063855421686746, "grad_norm": 5.505662441253662, "learning_rate": 7.043203883495147e-05, "loss": 0.2885, "step": 6091 }, { "epoch": 59.07349397590362, "grad_norm": 10.910268783569336, "learning_rate": 7.042718446601943e-05, "loss": 0.1604, "step": 6092 }, { "epoch": 59.08313253012048, "grad_norm": 17.65015983581543, "learning_rate": 7.042233009708738e-05, "loss": 0.2623, "step": 6093 }, { "epoch": 59.09277108433735, "grad_norm": 28.374683380126953, "learning_rate": 7.041747572815534e-05, "loss": 0.3392, "step": 6094 }, { "epoch": 59.102409638554214, "grad_norm": 5.0863423347473145, "learning_rate": 7.041262135922331e-05, "loss": 0.2408, "step": 6095 }, { "epoch": 59.112048192771084, "grad_norm": 6.947498321533203, "learning_rate": 7.040776699029127e-05, "loss": 0.4075, "step": 6096 }, { "epoch": 59.121686746987955, "grad_norm": 6.2796711921691895, "learning_rate": 7.040291262135923e-05, "loss": 0.1739, "step": 6097 }, { "epoch": 59.13132530120482, "grad_norm": 3.864281415939331, "learning_rate": 7.039805825242719e-05, "loss": 0.2423, "step": 6098 }, { "epoch": 59.14096385542169, "grad_norm": 2.866774082183838, "learning_rate": 7.039320388349516e-05, "loss": 0.302, "step": 6099 }, { "epoch": 59.15060240963855, "grad_norm": 42.58252716064453, "learning_rate": 7.038834951456312e-05, "loss": 0.4699, "step": 6100 }, { "epoch": 59.16024096385542, "grad_norm": 33.92342758178711, "learning_rate": 7.038349514563107e-05, "loss": 0.377, "step": 6101 }, { "epoch": 59.16987951807229, "grad_norm": 9.565282821655273, "learning_rate": 7.037864077669903e-05, "loss": 0.1424, "step": 6102 }, { "epoch": 59.179518072289156, "grad_norm": 20.098363876342773, "learning_rate": 7.037378640776699e-05, "loss": 0.2007, "step": 6103 }, { "epoch": 59.189156626506026, "grad_norm": 5.998055458068848, "learning_rate": 7.036893203883496e-05, "loss": 0.1975, "step": 6104 }, { "epoch": 59.19879518072289, "grad_norm": 5.959310531616211, "learning_rate": 7.036407766990292e-05, "loss": 0.2138, "step": 6105 }, { "epoch": 59.20843373493976, "grad_norm": 3.3035120964050293, "learning_rate": 7.035922330097088e-05, "loss": 0.1718, "step": 6106 }, { "epoch": 59.21807228915662, "grad_norm": 4.046144962310791, "learning_rate": 7.035436893203883e-05, "loss": 0.3157, "step": 6107 }, { "epoch": 59.227710843373494, "grad_norm": 1.8491371870040894, "learning_rate": 7.03495145631068e-05, "loss": 0.2532, "step": 6108 }, { "epoch": 59.237349397590364, "grad_norm": 38.1419563293457, "learning_rate": 7.034466019417476e-05, "loss": 0.4803, "step": 6109 }, { "epoch": 59.24698795180723, "grad_norm": 35.26784896850586, "learning_rate": 7.033980582524272e-05, "loss": 0.4363, "step": 6110 }, { "epoch": 59.2566265060241, "grad_norm": 12.915849685668945, "learning_rate": 7.033495145631068e-05, "loss": 0.2495, "step": 6111 }, { "epoch": 59.26626506024096, "grad_norm": 21.461748123168945, "learning_rate": 7.033009708737865e-05, "loss": 0.3598, "step": 6112 }, { "epoch": 59.27590361445783, "grad_norm": 4.177813529968262, "learning_rate": 7.032524271844661e-05, "loss": 0.2247, "step": 6113 }, { "epoch": 59.2855421686747, "grad_norm": 25.30834197998047, "learning_rate": 7.032038834951457e-05, "loss": 0.4619, "step": 6114 }, { "epoch": 59.295180722891565, "grad_norm": 15.198158264160156, "learning_rate": 7.031553398058252e-05, "loss": 0.3397, "step": 6115 }, { "epoch": 59.304819277108436, "grad_norm": 29.295507431030273, "learning_rate": 7.03106796116505e-05, "loss": 0.4064, "step": 6116 }, { "epoch": 59.3144578313253, "grad_norm": 28.156124114990234, "learning_rate": 7.030582524271845e-05, "loss": 0.3196, "step": 6117 }, { "epoch": 59.32409638554217, "grad_norm": 5.710334300994873, "learning_rate": 7.030097087378641e-05, "loss": 0.3079, "step": 6118 }, { "epoch": 59.33373493975903, "grad_norm": 9.613948822021484, "learning_rate": 7.029611650485437e-05, "loss": 0.3625, "step": 6119 }, { "epoch": 59.3433734939759, "grad_norm": 7.151418685913086, "learning_rate": 7.029126213592233e-05, "loss": 0.3275, "step": 6120 }, { "epoch": 59.353012048192774, "grad_norm": 48.058624267578125, "learning_rate": 7.02864077669903e-05, "loss": 0.5627, "step": 6121 }, { "epoch": 59.36265060240964, "grad_norm": 10.370744705200195, "learning_rate": 7.028155339805825e-05, "loss": 0.2252, "step": 6122 }, { "epoch": 59.37228915662651, "grad_norm": 4.85703182220459, "learning_rate": 7.027669902912621e-05, "loss": 0.2516, "step": 6123 }, { "epoch": 59.38192771084337, "grad_norm": 19.405193328857422, "learning_rate": 7.027184466019417e-05, "loss": 0.2986, "step": 6124 }, { "epoch": 59.39156626506024, "grad_norm": 26.960342407226562, "learning_rate": 7.026699029126214e-05, "loss": 0.4326, "step": 6125 }, { "epoch": 59.40120481927711, "grad_norm": 1.4128098487854004, "learning_rate": 7.02621359223301e-05, "loss": 0.0568, "step": 6126 }, { "epoch": 59.410843373493975, "grad_norm": 7.0770721435546875, "learning_rate": 7.025728155339806e-05, "loss": 0.0856, "step": 6127 }, { "epoch": 59.420481927710846, "grad_norm": 30.670547485351562, "learning_rate": 7.025242718446601e-05, "loss": 0.3114, "step": 6128 }, { "epoch": 59.43012048192771, "grad_norm": 2.579728126525879, "learning_rate": 7.024757281553399e-05, "loss": 0.295, "step": 6129 }, { "epoch": 59.43975903614458, "grad_norm": 26.886856079101562, "learning_rate": 7.024271844660194e-05, "loss": 0.4589, "step": 6130 }, { "epoch": 59.44939759036144, "grad_norm": 62.191307067871094, "learning_rate": 7.02378640776699e-05, "loss": 0.2969, "step": 6131 }, { "epoch": 59.45903614457831, "grad_norm": 12.190589904785156, "learning_rate": 7.023300970873786e-05, "loss": 0.2351, "step": 6132 }, { "epoch": 59.46867469879518, "grad_norm": 4.9674973487854, "learning_rate": 7.022815533980582e-05, "loss": 0.3258, "step": 6133 }, { "epoch": 59.47831325301205, "grad_norm": 2.9563510417938232, "learning_rate": 7.022330097087379e-05, "loss": 0.3102, "step": 6134 }, { "epoch": 59.48795180722892, "grad_norm": 3.4315335750579834, "learning_rate": 7.021844660194176e-05, "loss": 0.408, "step": 6135 }, { "epoch": 59.49759036144578, "grad_norm": 9.33852767944336, "learning_rate": 7.021359223300972e-05, "loss": 0.3773, "step": 6136 }, { "epoch": 59.50722891566265, "grad_norm": 22.463884353637695, "learning_rate": 7.020873786407767e-05, "loss": 0.3947, "step": 6137 }, { "epoch": 59.51686746987952, "grad_norm": 4.014305114746094, "learning_rate": 7.020388349514563e-05, "loss": 0.2561, "step": 6138 }, { "epoch": 59.526506024096385, "grad_norm": 9.649206161499023, "learning_rate": 7.01990291262136e-05, "loss": 0.2349, "step": 6139 }, { "epoch": 59.536144578313255, "grad_norm": 7.3253173828125, "learning_rate": 7.019417475728156e-05, "loss": 0.2577, "step": 6140 }, { "epoch": 59.54578313253012, "grad_norm": 11.750785827636719, "learning_rate": 7.018932038834952e-05, "loss": 0.4215, "step": 6141 }, { "epoch": 59.55542168674699, "grad_norm": 20.22617530822754, "learning_rate": 7.018446601941748e-05, "loss": 0.3769, "step": 6142 }, { "epoch": 59.56506024096385, "grad_norm": 18.979562759399414, "learning_rate": 7.017961165048545e-05, "loss": 0.5168, "step": 6143 }, { "epoch": 59.57469879518072, "grad_norm": 5.194829940795898, "learning_rate": 7.01747572815534e-05, "loss": 0.2328, "step": 6144 }, { "epoch": 59.58433734939759, "grad_norm": 4.148833751678467, "learning_rate": 7.016990291262136e-05, "loss": 0.164, "step": 6145 }, { "epoch": 59.59397590361446, "grad_norm": 12.066973686218262, "learning_rate": 7.016504854368932e-05, "loss": 0.3309, "step": 6146 }, { "epoch": 59.60361445783133, "grad_norm": 33.60124969482422, "learning_rate": 7.016019417475729e-05, "loss": 0.5126, "step": 6147 }, { "epoch": 59.61325301204819, "grad_norm": 30.318824768066406, "learning_rate": 7.015533980582525e-05, "loss": 0.3783, "step": 6148 }, { "epoch": 59.62289156626506, "grad_norm": 16.086240768432617, "learning_rate": 7.015048543689321e-05, "loss": 0.2469, "step": 6149 }, { "epoch": 59.63253012048193, "grad_norm": 15.623514175415039, "learning_rate": 7.014563106796117e-05, "loss": 0.2372, "step": 6150 }, { "epoch": 59.642168674698794, "grad_norm": 108.2080307006836, "learning_rate": 7.014077669902914e-05, "loss": 0.2141, "step": 6151 }, { "epoch": 59.651807228915665, "grad_norm": 46.17551040649414, "learning_rate": 7.01359223300971e-05, "loss": 0.4171, "step": 6152 }, { "epoch": 59.66144578313253, "grad_norm": 27.42816162109375, "learning_rate": 7.013106796116505e-05, "loss": 0.3469, "step": 6153 }, { "epoch": 59.6710843373494, "grad_norm": 16.5891170501709, "learning_rate": 7.012621359223301e-05, "loss": 0.2895, "step": 6154 }, { "epoch": 59.68072289156626, "grad_norm": 11.866456985473633, "learning_rate": 7.012135922330097e-05, "loss": 0.2977, "step": 6155 }, { "epoch": 59.69036144578313, "grad_norm": 49.88699722290039, "learning_rate": 7.011650485436894e-05, "loss": 0.3689, "step": 6156 }, { "epoch": 59.7, "grad_norm": 20.290353775024414, "learning_rate": 7.01116504854369e-05, "loss": 0.4415, "step": 6157 }, { "epoch": 59.709638554216866, "grad_norm": 22.532913208007812, "learning_rate": 7.010679611650486e-05, "loss": 0.3952, "step": 6158 }, { "epoch": 59.71927710843374, "grad_norm": 11.924129486083984, "learning_rate": 7.010194174757281e-05, "loss": 0.2286, "step": 6159 }, { "epoch": 59.7289156626506, "grad_norm": 7.29273796081543, "learning_rate": 7.009708737864078e-05, "loss": 0.3998, "step": 6160 }, { "epoch": 59.73855421686747, "grad_norm": 37.19719696044922, "learning_rate": 7.009223300970874e-05, "loss": 0.3897, "step": 6161 }, { "epoch": 59.74819277108434, "grad_norm": 23.291948318481445, "learning_rate": 7.00873786407767e-05, "loss": 0.405, "step": 6162 }, { "epoch": 59.757831325301204, "grad_norm": 70.21282958984375, "learning_rate": 7.008252427184466e-05, "loss": 0.3995, "step": 6163 }, { "epoch": 59.767469879518075, "grad_norm": 5.090480327606201, "learning_rate": 7.007766990291263e-05, "loss": 0.2089, "step": 6164 }, { "epoch": 59.77710843373494, "grad_norm": 18.158424377441406, "learning_rate": 7.007281553398059e-05, "loss": 0.3339, "step": 6165 }, { "epoch": 59.78674698795181, "grad_norm": 194.77932739257812, "learning_rate": 7.006796116504854e-05, "loss": 0.1999, "step": 6166 }, { "epoch": 59.79638554216867, "grad_norm": 22.45768165588379, "learning_rate": 7.00631067961165e-05, "loss": 0.2916, "step": 6167 }, { "epoch": 59.80602409638554, "grad_norm": 4.8734869956970215, "learning_rate": 7.005825242718447e-05, "loss": 0.1476, "step": 6168 }, { "epoch": 59.81566265060241, "grad_norm": 7.266288757324219, "learning_rate": 7.005339805825243e-05, "loss": 0.3244, "step": 6169 }, { "epoch": 59.825301204819276, "grad_norm": 46.44111251831055, "learning_rate": 7.004854368932039e-05, "loss": 0.4737, "step": 6170 }, { "epoch": 59.834939759036146, "grad_norm": 11.169783592224121, "learning_rate": 7.004368932038835e-05, "loss": 0.234, "step": 6171 }, { "epoch": 59.84457831325301, "grad_norm": 8.694070816040039, "learning_rate": 7.00388349514563e-05, "loss": 0.2577, "step": 6172 }, { "epoch": 59.85421686746988, "grad_norm": 16.51235008239746, "learning_rate": 7.003398058252428e-05, "loss": 0.1558, "step": 6173 }, { "epoch": 59.86385542168675, "grad_norm": 18.406661987304688, "learning_rate": 7.002912621359223e-05, "loss": 0.2402, "step": 6174 }, { "epoch": 59.873493975903614, "grad_norm": 4.25161600112915, "learning_rate": 7.002427184466019e-05, "loss": 0.1644, "step": 6175 }, { "epoch": 59.883132530120484, "grad_norm": 8.451951026916504, "learning_rate": 7.001941747572815e-05, "loss": 0.4348, "step": 6176 }, { "epoch": 59.89277108433735, "grad_norm": 5.172547340393066, "learning_rate": 7.001456310679612e-05, "loss": 0.3798, "step": 6177 }, { "epoch": 59.90240963855422, "grad_norm": 15.161669731140137, "learning_rate": 7.000970873786408e-05, "loss": 0.5282, "step": 6178 }, { "epoch": 59.91204819277108, "grad_norm": 22.24619483947754, "learning_rate": 7.000485436893204e-05, "loss": 0.2454, "step": 6179 }, { "epoch": 59.92168674698795, "grad_norm": 8.749283790588379, "learning_rate": 7e-05, "loss": 0.1873, "step": 6180 }, { "epoch": 59.93132530120482, "grad_norm": 23.391441345214844, "learning_rate": 6.999514563106797e-05, "loss": 0.2377, "step": 6181 }, { "epoch": 59.940963855421685, "grad_norm": 16.9244327545166, "learning_rate": 6.999029126213592e-05, "loss": 0.4685, "step": 6182 }, { "epoch": 59.950602409638556, "grad_norm": 2.479670763015747, "learning_rate": 6.998543689320388e-05, "loss": 0.5529, "step": 6183 }, { "epoch": 59.96024096385542, "grad_norm": 15.427210807800293, "learning_rate": 6.998058252427184e-05, "loss": 0.3712, "step": 6184 }, { "epoch": 59.96987951807229, "grad_norm": 4.416212558746338, "learning_rate": 6.997572815533981e-05, "loss": 0.4243, "step": 6185 }, { "epoch": 59.97951807228916, "grad_norm": 5.4264044761657715, "learning_rate": 6.997087378640778e-05, "loss": 0.2872, "step": 6186 }, { "epoch": 59.98915662650602, "grad_norm": 7.519083499908447, "learning_rate": 6.996601941747574e-05, "loss": 0.2787, "step": 6187 }, { "epoch": 60.00481927710843, "grad_norm": 18.86211395263672, "learning_rate": 6.99611650485437e-05, "loss": 0.2043, "step": 6188 }, { "epoch": 60.0144578313253, "grad_norm": 14.453343391418457, "learning_rate": 6.995631067961165e-05, "loss": 0.6386, "step": 6189 }, { "epoch": 60.024096385542165, "grad_norm": 78.510986328125, "learning_rate": 6.995145631067963e-05, "loss": 0.5603, "step": 6190 }, { "epoch": 60.033734939759036, "grad_norm": 22.418336868286133, "learning_rate": 6.994660194174758e-05, "loss": 0.1161, "step": 6191 }, { "epoch": 60.043373493975906, "grad_norm": 12.576719284057617, "learning_rate": 6.994174757281554e-05, "loss": 0.5376, "step": 6192 }, { "epoch": 60.05301204819277, "grad_norm": 10.230596542358398, "learning_rate": 6.99368932038835e-05, "loss": 0.3742, "step": 6193 }, { "epoch": 60.06265060240964, "grad_norm": 6.437158584594727, "learning_rate": 6.993203883495146e-05, "loss": 0.302, "step": 6194 }, { "epoch": 60.0722891566265, "grad_norm": 8.033187866210938, "learning_rate": 6.992718446601943e-05, "loss": 0.5675, "step": 6195 }, { "epoch": 60.081927710843374, "grad_norm": 12.111137390136719, "learning_rate": 6.992233009708739e-05, "loss": 0.3672, "step": 6196 }, { "epoch": 60.091566265060244, "grad_norm": 36.75690841674805, "learning_rate": 6.991747572815534e-05, "loss": 0.3991, "step": 6197 }, { "epoch": 60.10120481927711, "grad_norm": 8.771151542663574, "learning_rate": 6.99126213592233e-05, "loss": 0.5134, "step": 6198 }, { "epoch": 60.11084337349398, "grad_norm": 7.148996353149414, "learning_rate": 6.990776699029127e-05, "loss": 0.2823, "step": 6199 }, { "epoch": 60.12048192771084, "grad_norm": 7.120368003845215, "learning_rate": 6.990291262135923e-05, "loss": 0.3399, "step": 6200 }, { "epoch": 60.13012048192771, "grad_norm": 15.05333423614502, "learning_rate": 6.989805825242719e-05, "loss": 0.3381, "step": 6201 }, { "epoch": 60.139759036144575, "grad_norm": 17.315885543823242, "learning_rate": 6.989320388349515e-05, "loss": 0.2898, "step": 6202 }, { "epoch": 60.149397590361446, "grad_norm": 24.581066131591797, "learning_rate": 6.988834951456312e-05, "loss": 0.5869, "step": 6203 }, { "epoch": 60.159036144578316, "grad_norm": 8.24342155456543, "learning_rate": 6.988349514563107e-05, "loss": 0.2584, "step": 6204 }, { "epoch": 60.16867469879518, "grad_norm": 6.443493366241455, "learning_rate": 6.987864077669903e-05, "loss": 0.2403, "step": 6205 }, { "epoch": 60.17831325301205, "grad_norm": 23.032270431518555, "learning_rate": 6.987378640776699e-05, "loss": 0.1147, "step": 6206 }, { "epoch": 60.18795180722891, "grad_norm": 30.159448623657227, "learning_rate": 6.986893203883496e-05, "loss": 0.4214, "step": 6207 }, { "epoch": 60.19759036144578, "grad_norm": 9.703255653381348, "learning_rate": 6.986407766990292e-05, "loss": 0.2332, "step": 6208 }, { "epoch": 60.207228915662654, "grad_norm": 10.56840991973877, "learning_rate": 6.985922330097088e-05, "loss": 0.2613, "step": 6209 }, { "epoch": 60.21686746987952, "grad_norm": 5.914285182952881, "learning_rate": 6.985436893203884e-05, "loss": 0.2999, "step": 6210 }, { "epoch": 60.22650602409639, "grad_norm": 42.622100830078125, "learning_rate": 6.984951456310679e-05, "loss": 0.3872, "step": 6211 }, { "epoch": 60.23614457831325, "grad_norm": 10.675061225891113, "learning_rate": 6.984466019417476e-05, "loss": 0.1449, "step": 6212 }, { "epoch": 60.24578313253012, "grad_norm": 26.08148765563965, "learning_rate": 6.983980582524272e-05, "loss": 0.6932, "step": 6213 }, { "epoch": 60.255421686746985, "grad_norm": 16.26009178161621, "learning_rate": 6.983495145631068e-05, "loss": 0.2149, "step": 6214 }, { "epoch": 60.265060240963855, "grad_norm": 26.16333770751953, "learning_rate": 6.983009708737864e-05, "loss": 0.2692, "step": 6215 }, { "epoch": 60.274698795180726, "grad_norm": 8.567550659179688, "learning_rate": 6.982524271844661e-05, "loss": 0.3799, "step": 6216 }, { "epoch": 60.28433734939759, "grad_norm": 118.56143951416016, "learning_rate": 6.982038834951457e-05, "loss": 0.365, "step": 6217 }, { "epoch": 60.29397590361446, "grad_norm": 2.545334815979004, "learning_rate": 6.981553398058252e-05, "loss": 0.1881, "step": 6218 }, { "epoch": 60.30361445783132, "grad_norm": 11.771743774414062, "learning_rate": 6.981067961165048e-05, "loss": 0.3153, "step": 6219 }, { "epoch": 60.31325301204819, "grad_norm": 15.490499496459961, "learning_rate": 6.980582524271845e-05, "loss": 0.2359, "step": 6220 }, { "epoch": 60.32289156626506, "grad_norm": 15.517301559448242, "learning_rate": 6.980097087378641e-05, "loss": 0.2375, "step": 6221 }, { "epoch": 60.33253012048193, "grad_norm": 6.228183269500732, "learning_rate": 6.979611650485437e-05, "loss": 0.3287, "step": 6222 }, { "epoch": 60.3421686746988, "grad_norm": 18.158451080322266, "learning_rate": 6.979126213592233e-05, "loss": 0.3862, "step": 6223 }, { "epoch": 60.35180722891566, "grad_norm": 5.280929088592529, "learning_rate": 6.97864077669903e-05, "loss": 0.3287, "step": 6224 }, { "epoch": 60.36144578313253, "grad_norm": 10.992741584777832, "learning_rate": 6.978155339805826e-05, "loss": 0.3552, "step": 6225 }, { "epoch": 60.371084337349394, "grad_norm": 146.40142822265625, "learning_rate": 6.977669902912621e-05, "loss": 0.1929, "step": 6226 }, { "epoch": 60.380722891566265, "grad_norm": 36.16669464111328, "learning_rate": 6.977184466019417e-05, "loss": 0.4657, "step": 6227 }, { "epoch": 60.390361445783135, "grad_norm": 7.34702730178833, "learning_rate": 6.976699029126213e-05, "loss": 0.3619, "step": 6228 }, { "epoch": 60.4, "grad_norm": 25.086950302124023, "learning_rate": 6.97621359223301e-05, "loss": 0.2681, "step": 6229 }, { "epoch": 60.40963855421687, "grad_norm": 28.114601135253906, "learning_rate": 6.975728155339806e-05, "loss": 0.2399, "step": 6230 }, { "epoch": 60.41927710843373, "grad_norm": 18.712995529174805, "learning_rate": 6.975242718446602e-05, "loss": 0.393, "step": 6231 }, { "epoch": 60.4289156626506, "grad_norm": 4.941356658935547, "learning_rate": 6.974757281553397e-05, "loss": 0.2249, "step": 6232 }, { "epoch": 60.43855421686747, "grad_norm": 25.636865615844727, "learning_rate": 6.974271844660194e-05, "loss": 0.4929, "step": 6233 }, { "epoch": 60.44819277108434, "grad_norm": 7.448472499847412, "learning_rate": 6.97378640776699e-05, "loss": 0.2755, "step": 6234 }, { "epoch": 60.45783132530121, "grad_norm": 5.393628120422363, "learning_rate": 6.973300970873786e-05, "loss": 0.1666, "step": 6235 }, { "epoch": 60.46746987951807, "grad_norm": 16.40706443786621, "learning_rate": 6.972815533980583e-05, "loss": 0.2266, "step": 6236 }, { "epoch": 60.47710843373494, "grad_norm": 33.789215087890625, "learning_rate": 6.972330097087379e-05, "loss": 0.4907, "step": 6237 }, { "epoch": 60.486746987951804, "grad_norm": 4.059194564819336, "learning_rate": 6.971844660194176e-05, "loss": 0.1373, "step": 6238 }, { "epoch": 60.496385542168674, "grad_norm": 2.9227757453918457, "learning_rate": 6.971359223300972e-05, "loss": 0.1412, "step": 6239 }, { "epoch": 60.506024096385545, "grad_norm": 6.165503025054932, "learning_rate": 6.970873786407768e-05, "loss": 0.2984, "step": 6240 }, { "epoch": 60.51566265060241, "grad_norm": 18.571027755737305, "learning_rate": 6.970388349514563e-05, "loss": 0.2612, "step": 6241 }, { "epoch": 60.52530120481928, "grad_norm": 17.541851043701172, "learning_rate": 6.96990291262136e-05, "loss": 0.3062, "step": 6242 }, { "epoch": 60.53493975903614, "grad_norm": 4.076896667480469, "learning_rate": 6.969417475728156e-05, "loss": 0.3055, "step": 6243 }, { "epoch": 60.54457831325301, "grad_norm": 17.235288619995117, "learning_rate": 6.968932038834952e-05, "loss": 0.354, "step": 6244 }, { "epoch": 60.55421686746988, "grad_norm": 28.6440486907959, "learning_rate": 6.968446601941748e-05, "loss": 0.4252, "step": 6245 }, { "epoch": 60.563855421686746, "grad_norm": 4.513937950134277, "learning_rate": 6.967961165048545e-05, "loss": 0.3498, "step": 6246 }, { "epoch": 60.57349397590362, "grad_norm": 16.93352699279785, "learning_rate": 6.967475728155341e-05, "loss": 0.2947, "step": 6247 }, { "epoch": 60.58313253012048, "grad_norm": 7.879796981811523, "learning_rate": 6.966990291262137e-05, "loss": 0.1049, "step": 6248 }, { "epoch": 60.59277108433735, "grad_norm": 11.126198768615723, "learning_rate": 6.966504854368932e-05, "loss": 0.3148, "step": 6249 }, { "epoch": 60.602409638554214, "grad_norm": 15.716898918151855, "learning_rate": 6.966019417475728e-05, "loss": 0.3029, "step": 6250 }, { "epoch": 60.612048192771084, "grad_norm": 5.1114583015441895, "learning_rate": 6.965533980582525e-05, "loss": 0.3273, "step": 6251 }, { "epoch": 60.621686746987955, "grad_norm": 5.314015865325928, "learning_rate": 6.965048543689321e-05, "loss": 0.3391, "step": 6252 }, { "epoch": 60.63132530120482, "grad_norm": 212.64657592773438, "learning_rate": 6.964563106796117e-05, "loss": 0.4047, "step": 6253 }, { "epoch": 60.64096385542169, "grad_norm": 15.157876968383789, "learning_rate": 6.964077669902913e-05, "loss": 0.3742, "step": 6254 }, { "epoch": 60.65060240963855, "grad_norm": 29.250572204589844, "learning_rate": 6.96359223300971e-05, "loss": 0.3304, "step": 6255 }, { "epoch": 60.66024096385542, "grad_norm": 5.983762264251709, "learning_rate": 6.963106796116505e-05, "loss": 0.2045, "step": 6256 }, { "epoch": 60.66987951807229, "grad_norm": 218.52145385742188, "learning_rate": 6.962621359223301e-05, "loss": 0.3798, "step": 6257 }, { "epoch": 60.679518072289156, "grad_norm": 6.959513187408447, "learning_rate": 6.962135922330097e-05, "loss": 0.2602, "step": 6258 }, { "epoch": 60.689156626506026, "grad_norm": 11.793533325195312, "learning_rate": 6.961650485436894e-05, "loss": 0.2972, "step": 6259 }, { "epoch": 60.69879518072289, "grad_norm": 4.939290523529053, "learning_rate": 6.96116504854369e-05, "loss": 0.2317, "step": 6260 }, { "epoch": 60.70843373493976, "grad_norm": 1.9601110219955444, "learning_rate": 6.960679611650486e-05, "loss": 0.0656, "step": 6261 }, { "epoch": 60.71807228915662, "grad_norm": 32.994049072265625, "learning_rate": 6.960194174757281e-05, "loss": 0.2184, "step": 6262 }, { "epoch": 60.727710843373494, "grad_norm": 25.587770462036133, "learning_rate": 6.959708737864079e-05, "loss": 0.1748, "step": 6263 }, { "epoch": 60.737349397590364, "grad_norm": 29.706851959228516, "learning_rate": 6.959223300970874e-05, "loss": 0.5463, "step": 6264 }, { "epoch": 60.74698795180723, "grad_norm": 16.002277374267578, "learning_rate": 6.95873786407767e-05, "loss": 0.1618, "step": 6265 }, { "epoch": 60.7566265060241, "grad_norm": 3.9799275398254395, "learning_rate": 6.958252427184466e-05, "loss": 0.2335, "step": 6266 }, { "epoch": 60.76626506024096, "grad_norm": 15.75780963897705, "learning_rate": 6.957766990291262e-05, "loss": 0.4663, "step": 6267 }, { "epoch": 60.77590361445783, "grad_norm": 7.435276031494141, "learning_rate": 6.957281553398059e-05, "loss": 0.279, "step": 6268 }, { "epoch": 60.7855421686747, "grad_norm": 6.259965419769287, "learning_rate": 6.956796116504855e-05, "loss": 0.3283, "step": 6269 }, { "epoch": 60.795180722891565, "grad_norm": 18.59534454345703, "learning_rate": 6.95631067961165e-05, "loss": 0.3617, "step": 6270 }, { "epoch": 60.804819277108436, "grad_norm": 12.082884788513184, "learning_rate": 6.955825242718446e-05, "loss": 0.5503, "step": 6271 }, { "epoch": 60.8144578313253, "grad_norm": 8.317916870117188, "learning_rate": 6.955339805825243e-05, "loss": 0.417, "step": 6272 }, { "epoch": 60.82409638554217, "grad_norm": 33.291526794433594, "learning_rate": 6.954854368932039e-05, "loss": 0.4349, "step": 6273 }, { "epoch": 60.83373493975903, "grad_norm": 14.824965476989746, "learning_rate": 6.954368932038835e-05, "loss": 0.24, "step": 6274 }, { "epoch": 60.8433734939759, "grad_norm": 9.43129825592041, "learning_rate": 6.95388349514563e-05, "loss": 0.3654, "step": 6275 }, { "epoch": 60.853012048192774, "grad_norm": 5.525013446807861, "learning_rate": 6.953398058252428e-05, "loss": 0.3079, "step": 6276 }, { "epoch": 60.86265060240964, "grad_norm": 15.526979446411133, "learning_rate": 6.952912621359224e-05, "loss": 0.2955, "step": 6277 }, { "epoch": 60.87228915662651, "grad_norm": 104.87100219726562, "learning_rate": 6.952427184466019e-05, "loss": 0.2001, "step": 6278 }, { "epoch": 60.88192771084337, "grad_norm": 15.199545860290527, "learning_rate": 6.951941747572815e-05, "loss": 0.1683, "step": 6279 }, { "epoch": 60.89156626506024, "grad_norm": 18.860265731811523, "learning_rate": 6.951456310679612e-05, "loss": 0.4078, "step": 6280 }, { "epoch": 60.90120481927711, "grad_norm": 32.93169021606445, "learning_rate": 6.950970873786408e-05, "loss": 0.4986, "step": 6281 }, { "epoch": 60.910843373493975, "grad_norm": 48.51258850097656, "learning_rate": 6.950485436893204e-05, "loss": 0.4011, "step": 6282 }, { "epoch": 60.920481927710846, "grad_norm": 36.912513732910156, "learning_rate": 6.95e-05, "loss": 0.3266, "step": 6283 }, { "epoch": 60.93012048192771, "grad_norm": 9.38900375366211, "learning_rate": 6.949514563106795e-05, "loss": 0.2717, "step": 6284 }, { "epoch": 60.93975903614458, "grad_norm": 8.758744239807129, "learning_rate": 6.949029126213592e-05, "loss": 0.5479, "step": 6285 }, { "epoch": 60.94939759036144, "grad_norm": 3.591907501220703, "learning_rate": 6.948543689320388e-05, "loss": 0.1868, "step": 6286 }, { "epoch": 60.95903614457831, "grad_norm": 12.783344268798828, "learning_rate": 6.948058252427185e-05, "loss": 0.2722, "step": 6287 }, { "epoch": 60.96867469879518, "grad_norm": 23.460664749145508, "learning_rate": 6.947572815533981e-05, "loss": 0.4397, "step": 6288 }, { "epoch": 60.97831325301205, "grad_norm": 8.19221305847168, "learning_rate": 6.947087378640777e-05, "loss": 0.1968, "step": 6289 }, { "epoch": 60.98795180722892, "grad_norm": 6.766294956207275, "learning_rate": 6.946601941747574e-05, "loss": 0.4511, "step": 6290 }, { "epoch": 61.003614457831326, "grad_norm": 5.037155628204346, "learning_rate": 6.94611650485437e-05, "loss": 0.1651, "step": 6291 }, { "epoch": 61.013253012048196, "grad_norm": 6.099465370178223, "learning_rate": 6.945631067961166e-05, "loss": 0.2823, "step": 6292 }, { "epoch": 61.02289156626506, "grad_norm": 33.06064224243164, "learning_rate": 6.945145631067961e-05, "loss": 0.3813, "step": 6293 }, { "epoch": 61.03253012048193, "grad_norm": 16.060771942138672, "learning_rate": 6.944660194174758e-05, "loss": 0.6958, "step": 6294 }, { "epoch": 61.04216867469879, "grad_norm": 7.5338521003723145, "learning_rate": 6.944174757281554e-05, "loss": 0.2894, "step": 6295 }, { "epoch": 61.05180722891566, "grad_norm": 11.570768356323242, "learning_rate": 6.94368932038835e-05, "loss": 0.2547, "step": 6296 }, { "epoch": 61.06144578313253, "grad_norm": 11.143168449401855, "learning_rate": 6.943203883495146e-05, "loss": 0.2007, "step": 6297 }, { "epoch": 61.0710843373494, "grad_norm": 24.728147506713867, "learning_rate": 6.942718446601943e-05, "loss": 0.4501, "step": 6298 }, { "epoch": 61.08072289156627, "grad_norm": 8.43761920928955, "learning_rate": 6.942233009708739e-05, "loss": 0.3784, "step": 6299 }, { "epoch": 61.09036144578313, "grad_norm": 5.830540657043457, "learning_rate": 6.941747572815534e-05, "loss": 0.2512, "step": 6300 }, { "epoch": 61.1, "grad_norm": 8.898932456970215, "learning_rate": 6.94126213592233e-05, "loss": 0.4357, "step": 6301 }, { "epoch": 61.109638554216865, "grad_norm": 15.165218353271484, "learning_rate": 6.940776699029127e-05, "loss": 0.3633, "step": 6302 }, { "epoch": 61.119277108433735, "grad_norm": 126.0708999633789, "learning_rate": 6.940291262135923e-05, "loss": 0.3695, "step": 6303 }, { "epoch": 61.128915662650606, "grad_norm": 10.471776962280273, "learning_rate": 6.939805825242719e-05, "loss": 0.6086, "step": 6304 }, { "epoch": 61.13855421686747, "grad_norm": 8.173190116882324, "learning_rate": 6.939320388349515e-05, "loss": 0.3193, "step": 6305 }, { "epoch": 61.14819277108434, "grad_norm": 3.9174885749816895, "learning_rate": 6.93883495145631e-05, "loss": 0.2384, "step": 6306 }, { "epoch": 61.1578313253012, "grad_norm": 10.375699043273926, "learning_rate": 6.938349514563108e-05, "loss": 0.2063, "step": 6307 }, { "epoch": 61.16746987951807, "grad_norm": 3.961442470550537, "learning_rate": 6.937864077669903e-05, "loss": 0.1213, "step": 6308 }, { "epoch": 61.17710843373494, "grad_norm": 36.552188873291016, "learning_rate": 6.937378640776699e-05, "loss": 0.2173, "step": 6309 }, { "epoch": 61.18674698795181, "grad_norm": 13.128689765930176, "learning_rate": 6.936893203883495e-05, "loss": 0.5922, "step": 6310 }, { "epoch": 61.19638554216868, "grad_norm": 4.007899761199951, "learning_rate": 6.936407766990292e-05, "loss": 0.2568, "step": 6311 }, { "epoch": 61.20602409638554, "grad_norm": 1.340217113494873, "learning_rate": 6.935922330097088e-05, "loss": 0.1579, "step": 6312 }, { "epoch": 61.21566265060241, "grad_norm": 42.17866897583008, "learning_rate": 6.935436893203884e-05, "loss": 0.5411, "step": 6313 }, { "epoch": 61.225301204819274, "grad_norm": 13.52660083770752, "learning_rate": 6.93495145631068e-05, "loss": 0.3305, "step": 6314 }, { "epoch": 61.234939759036145, "grad_norm": 14.152567863464355, "learning_rate": 6.934466019417477e-05, "loss": 0.2966, "step": 6315 }, { "epoch": 61.244578313253015, "grad_norm": 16.483245849609375, "learning_rate": 6.933980582524272e-05, "loss": 0.3372, "step": 6316 }, { "epoch": 61.25421686746988, "grad_norm": 27.50861167907715, "learning_rate": 6.933495145631068e-05, "loss": 0.3044, "step": 6317 }, { "epoch": 61.26385542168675, "grad_norm": 15.813180923461914, "learning_rate": 6.933009708737864e-05, "loss": 0.5452, "step": 6318 }, { "epoch": 61.27349397590361, "grad_norm": 8.911149978637695, "learning_rate": 6.932524271844661e-05, "loss": 0.4304, "step": 6319 }, { "epoch": 61.28313253012048, "grad_norm": 10.664557456970215, "learning_rate": 6.932038834951457e-05, "loss": 0.3627, "step": 6320 }, { "epoch": 61.292771084337346, "grad_norm": 46.09815216064453, "learning_rate": 6.931553398058253e-05, "loss": 0.3382, "step": 6321 }, { "epoch": 61.30240963855422, "grad_norm": 5.970519542694092, "learning_rate": 6.931067961165048e-05, "loss": 0.2355, "step": 6322 }, { "epoch": 61.31204819277109, "grad_norm": 9.042041778564453, "learning_rate": 6.930582524271844e-05, "loss": 0.3342, "step": 6323 }, { "epoch": 61.32168674698795, "grad_norm": 9.99138355255127, "learning_rate": 6.930097087378641e-05, "loss": 0.3992, "step": 6324 }, { "epoch": 61.33132530120482, "grad_norm": 5.9511895179748535, "learning_rate": 6.929611650485437e-05, "loss": 0.276, "step": 6325 }, { "epoch": 61.340963855421684, "grad_norm": 28.321733474731445, "learning_rate": 6.929126213592233e-05, "loss": 0.3755, "step": 6326 }, { "epoch": 61.350602409638554, "grad_norm": 6.989144802093506, "learning_rate": 6.928640776699029e-05, "loss": 0.2462, "step": 6327 }, { "epoch": 61.360240963855425, "grad_norm": 12.865959167480469, "learning_rate": 6.928155339805826e-05, "loss": 0.3739, "step": 6328 }, { "epoch": 61.36987951807229, "grad_norm": 23.630483627319336, "learning_rate": 6.927669902912621e-05, "loss": 0.3088, "step": 6329 }, { "epoch": 61.37951807228916, "grad_norm": 32.45876693725586, "learning_rate": 6.927184466019417e-05, "loss": 0.4829, "step": 6330 }, { "epoch": 61.38915662650602, "grad_norm": 7.842516899108887, "learning_rate": 6.926699029126213e-05, "loss": 0.2232, "step": 6331 }, { "epoch": 61.39879518072289, "grad_norm": 9.218047142028809, "learning_rate": 6.92621359223301e-05, "loss": 0.1886, "step": 6332 }, { "epoch": 61.408433734939756, "grad_norm": 10.352782249450684, "learning_rate": 6.925728155339806e-05, "loss": 0.2749, "step": 6333 }, { "epoch": 61.418072289156626, "grad_norm": 16.173614501953125, "learning_rate": 6.925242718446602e-05, "loss": 0.3501, "step": 6334 }, { "epoch": 61.4277108433735, "grad_norm": 5.093896865844727, "learning_rate": 6.924757281553397e-05, "loss": 0.2022, "step": 6335 }, { "epoch": 61.43734939759036, "grad_norm": 34.7584228515625, "learning_rate": 6.924271844660195e-05, "loss": 0.4202, "step": 6336 }, { "epoch": 61.44698795180723, "grad_norm": 10.770505905151367, "learning_rate": 6.923786407766992e-05, "loss": 0.3911, "step": 6337 }, { "epoch": 61.456626506024094, "grad_norm": 4.2232465744018555, "learning_rate": 6.923300970873788e-05, "loss": 0.219, "step": 6338 }, { "epoch": 61.466265060240964, "grad_norm": 22.11460304260254, "learning_rate": 6.922815533980583e-05, "loss": 0.2796, "step": 6339 }, { "epoch": 61.475903614457835, "grad_norm": 4.662457466125488, "learning_rate": 6.922330097087379e-05, "loss": 0.3358, "step": 6340 }, { "epoch": 61.4855421686747, "grad_norm": 11.733428001403809, "learning_rate": 6.921844660194176e-05, "loss": 0.3686, "step": 6341 }, { "epoch": 61.49518072289157, "grad_norm": 9.048224449157715, "learning_rate": 6.921359223300972e-05, "loss": 0.2834, "step": 6342 }, { "epoch": 61.50481927710843, "grad_norm": 18.718284606933594, "learning_rate": 6.920873786407768e-05, "loss": 0.3719, "step": 6343 }, { "epoch": 61.5144578313253, "grad_norm": 17.446975708007812, "learning_rate": 6.920388349514564e-05, "loss": 0.4526, "step": 6344 }, { "epoch": 61.524096385542165, "grad_norm": 20.701284408569336, "learning_rate": 6.919902912621359e-05, "loss": 0.4355, "step": 6345 }, { "epoch": 61.533734939759036, "grad_norm": 24.05842399597168, "learning_rate": 6.919417475728156e-05, "loss": 0.2845, "step": 6346 }, { "epoch": 61.543373493975906, "grad_norm": 8.5890531539917, "learning_rate": 6.918932038834952e-05, "loss": 0.1793, "step": 6347 }, { "epoch": 61.55301204819277, "grad_norm": 7.2247138023376465, "learning_rate": 6.918446601941748e-05, "loss": 0.139, "step": 6348 }, { "epoch": 61.56265060240964, "grad_norm": 27.109642028808594, "learning_rate": 6.917961165048544e-05, "loss": 0.3049, "step": 6349 }, { "epoch": 61.5722891566265, "grad_norm": 11.94164752960205, "learning_rate": 6.917475728155341e-05, "loss": 0.3162, "step": 6350 }, { "epoch": 61.581927710843374, "grad_norm": 25.810476303100586, "learning_rate": 6.916990291262137e-05, "loss": 0.3588, "step": 6351 }, { "epoch": 61.591566265060244, "grad_norm": 7.691949367523193, "learning_rate": 6.916504854368932e-05, "loss": 0.3763, "step": 6352 }, { "epoch": 61.60120481927711, "grad_norm": 41.37137222290039, "learning_rate": 6.916019417475728e-05, "loss": 0.2357, "step": 6353 }, { "epoch": 61.61084337349398, "grad_norm": 14.74811840057373, "learning_rate": 6.915533980582525e-05, "loss": 0.4555, "step": 6354 }, { "epoch": 61.62048192771084, "grad_norm": 8.34361457824707, "learning_rate": 6.915048543689321e-05, "loss": 0.2258, "step": 6355 }, { "epoch": 61.63012048192771, "grad_norm": 15.320693016052246, "learning_rate": 6.914563106796117e-05, "loss": 0.3496, "step": 6356 }, { "epoch": 61.639759036144575, "grad_norm": 9.311901092529297, "learning_rate": 6.914077669902913e-05, "loss": 0.1893, "step": 6357 }, { "epoch": 61.649397590361446, "grad_norm": 10.551115036010742, "learning_rate": 6.91359223300971e-05, "loss": 0.3189, "step": 6358 }, { "epoch": 61.659036144578316, "grad_norm": 7.311534881591797, "learning_rate": 6.913106796116506e-05, "loss": 0.3388, "step": 6359 }, { "epoch": 61.66867469879518, "grad_norm": 5.953284740447998, "learning_rate": 6.912621359223301e-05, "loss": 0.409, "step": 6360 }, { "epoch": 61.67831325301205, "grad_norm": 34.77224349975586, "learning_rate": 6.912135922330097e-05, "loss": 0.4528, "step": 6361 }, { "epoch": 61.68795180722891, "grad_norm": 9.582858085632324, "learning_rate": 6.911650485436893e-05, "loss": 0.2575, "step": 6362 }, { "epoch": 61.69759036144578, "grad_norm": 13.558136940002441, "learning_rate": 6.91116504854369e-05, "loss": 0.2377, "step": 6363 }, { "epoch": 61.707228915662654, "grad_norm": 6.043269634246826, "learning_rate": 6.910679611650486e-05, "loss": 0.231, "step": 6364 }, { "epoch": 61.71686746987952, "grad_norm": 8.829724311828613, "learning_rate": 6.910194174757282e-05, "loss": 0.2741, "step": 6365 }, { "epoch": 61.72650602409639, "grad_norm": 26.17068099975586, "learning_rate": 6.909708737864077e-05, "loss": 0.3079, "step": 6366 }, { "epoch": 61.73614457831325, "grad_norm": 9.622418403625488, "learning_rate": 6.909223300970875e-05, "loss": 0.2523, "step": 6367 }, { "epoch": 61.74578313253012, "grad_norm": 13.48233699798584, "learning_rate": 6.90873786407767e-05, "loss": 0.3606, "step": 6368 }, { "epoch": 61.755421686746985, "grad_norm": 4.90410852432251, "learning_rate": 6.908252427184466e-05, "loss": 0.305, "step": 6369 }, { "epoch": 61.765060240963855, "grad_norm": 6.545161247253418, "learning_rate": 6.907766990291262e-05, "loss": 0.2621, "step": 6370 }, { "epoch": 61.774698795180726, "grad_norm": 11.502851486206055, "learning_rate": 6.907281553398059e-05, "loss": 0.448, "step": 6371 }, { "epoch": 61.78433734939759, "grad_norm": 5.076004981994629, "learning_rate": 6.906796116504855e-05, "loss": 0.2148, "step": 6372 }, { "epoch": 61.79397590361446, "grad_norm": 84.9070816040039, "learning_rate": 6.90631067961165e-05, "loss": 0.2132, "step": 6373 }, { "epoch": 61.80361445783132, "grad_norm": 39.902122497558594, "learning_rate": 6.905825242718446e-05, "loss": 0.4352, "step": 6374 }, { "epoch": 61.81325301204819, "grad_norm": 19.455978393554688, "learning_rate": 6.905339805825243e-05, "loss": 0.3958, "step": 6375 }, { "epoch": 61.82289156626506, "grad_norm": 6.285456657409668, "learning_rate": 6.904854368932039e-05, "loss": 0.4138, "step": 6376 }, { "epoch": 61.83253012048193, "grad_norm": 13.010750770568848, "learning_rate": 6.904368932038835e-05, "loss": 0.309, "step": 6377 }, { "epoch": 61.8421686746988, "grad_norm": 3.6684865951538086, "learning_rate": 6.903883495145631e-05, "loss": 0.0908, "step": 6378 }, { "epoch": 61.85180722891566, "grad_norm": 5.060720443725586, "learning_rate": 6.903398058252427e-05, "loss": 0.1727, "step": 6379 }, { "epoch": 61.86144578313253, "grad_norm": 31.730682373046875, "learning_rate": 6.902912621359224e-05, "loss": 0.413, "step": 6380 }, { "epoch": 61.871084337349394, "grad_norm": 8.20959186553955, "learning_rate": 6.90242718446602e-05, "loss": 0.3245, "step": 6381 }, { "epoch": 61.880722891566265, "grad_norm": 6.152855396270752, "learning_rate": 6.901941747572815e-05, "loss": 0.3303, "step": 6382 }, { "epoch": 61.890361445783135, "grad_norm": 4.587690353393555, "learning_rate": 6.901456310679611e-05, "loss": 0.1523, "step": 6383 }, { "epoch": 61.9, "grad_norm": 9.258966445922852, "learning_rate": 6.900970873786408e-05, "loss": 0.4599, "step": 6384 }, { "epoch": 61.90963855421687, "grad_norm": 7.526381969451904, "learning_rate": 6.900485436893204e-05, "loss": 0.3691, "step": 6385 }, { "epoch": 61.91927710843373, "grad_norm": 45.06354904174805, "learning_rate": 6.9e-05, "loss": 0.2446, "step": 6386 }, { "epoch": 61.9289156626506, "grad_norm": 7.573967933654785, "learning_rate": 6.899514563106795e-05, "loss": 0.2871, "step": 6387 }, { "epoch": 61.93855421686747, "grad_norm": 12.915884971618652, "learning_rate": 6.899029126213593e-05, "loss": 0.2166, "step": 6388 }, { "epoch": 61.94819277108434, "grad_norm": 6.555178642272949, "learning_rate": 6.89854368932039e-05, "loss": 0.1919, "step": 6389 }, { "epoch": 61.95783132530121, "grad_norm": 10.842819213867188, "learning_rate": 6.898058252427185e-05, "loss": 0.4854, "step": 6390 }, { "epoch": 61.96746987951807, "grad_norm": 13.70837116241455, "learning_rate": 6.897572815533981e-05, "loss": 0.3334, "step": 6391 }, { "epoch": 61.97710843373494, "grad_norm": 17.682964324951172, "learning_rate": 6.897087378640777e-05, "loss": 0.328, "step": 6392 }, { "epoch": 61.986746987951804, "grad_norm": 7.460265159606934, "learning_rate": 6.896601941747574e-05, "loss": 0.3369, "step": 6393 }, { "epoch": 62.00240963855422, "grad_norm": 9.434895515441895, "learning_rate": 6.89611650485437e-05, "loss": 0.4037, "step": 6394 }, { "epoch": 62.01204819277108, "grad_norm": 8.573987007141113, "learning_rate": 6.895631067961166e-05, "loss": 0.309, "step": 6395 }, { "epoch": 62.02168674698795, "grad_norm": 6.93269681930542, "learning_rate": 6.895145631067961e-05, "loss": 0.3133, "step": 6396 }, { "epoch": 62.03132530120482, "grad_norm": 12.27397346496582, "learning_rate": 6.894660194174759e-05, "loss": 0.3984, "step": 6397 }, { "epoch": 62.04096385542169, "grad_norm": 6.394061088562012, "learning_rate": 6.894174757281554e-05, "loss": 0.2571, "step": 6398 }, { "epoch": 62.05060240963856, "grad_norm": 18.338642120361328, "learning_rate": 6.89368932038835e-05, "loss": 0.1981, "step": 6399 }, { "epoch": 62.06024096385542, "grad_norm": 5.940738201141357, "learning_rate": 6.893203883495146e-05, "loss": 0.3581, "step": 6400 }, { "epoch": 62.06987951807229, "grad_norm": 13.897854804992676, "learning_rate": 6.892718446601942e-05, "loss": 0.4318, "step": 6401 }, { "epoch": 62.079518072289154, "grad_norm": 6.437197685241699, "learning_rate": 6.892233009708739e-05, "loss": 0.1446, "step": 6402 }, { "epoch": 62.089156626506025, "grad_norm": 6.754671096801758, "learning_rate": 6.891747572815535e-05, "loss": 0.2396, "step": 6403 }, { "epoch": 62.09879518072289, "grad_norm": 8.502546310424805, "learning_rate": 6.89126213592233e-05, "loss": 0.2964, "step": 6404 }, { "epoch": 62.10843373493976, "grad_norm": 7.9955973625183105, "learning_rate": 6.890776699029126e-05, "loss": 0.2401, "step": 6405 }, { "epoch": 62.11807228915663, "grad_norm": 10.818915367126465, "learning_rate": 6.890291262135923e-05, "loss": 0.3997, "step": 6406 }, { "epoch": 62.12771084337349, "grad_norm": 27.916015625, "learning_rate": 6.889805825242719e-05, "loss": 0.1859, "step": 6407 }, { "epoch": 62.13734939759036, "grad_norm": 17.752944946289062, "learning_rate": 6.889320388349515e-05, "loss": 0.2712, "step": 6408 }, { "epoch": 62.146987951807226, "grad_norm": 35.90187072753906, "learning_rate": 6.88883495145631e-05, "loss": 0.3498, "step": 6409 }, { "epoch": 62.1566265060241, "grad_norm": 12.181302070617676, "learning_rate": 6.888349514563108e-05, "loss": 0.3559, "step": 6410 }, { "epoch": 62.16626506024097, "grad_norm": 11.81771183013916, "learning_rate": 6.887864077669904e-05, "loss": 0.252, "step": 6411 }, { "epoch": 62.17590361445783, "grad_norm": 9.795564651489258, "learning_rate": 6.8873786407767e-05, "loss": 0.2326, "step": 6412 }, { "epoch": 62.1855421686747, "grad_norm": 7.736691474914551, "learning_rate": 6.886893203883495e-05, "loss": 0.3579, "step": 6413 }, { "epoch": 62.195180722891564, "grad_norm": 5.11856746673584, "learning_rate": 6.886407766990292e-05, "loss": 0.1267, "step": 6414 }, { "epoch": 62.204819277108435, "grad_norm": 8.973358154296875, "learning_rate": 6.885922330097088e-05, "loss": 0.3733, "step": 6415 }, { "epoch": 62.2144578313253, "grad_norm": 8.466896057128906, "learning_rate": 6.885436893203884e-05, "loss": 0.3624, "step": 6416 }, { "epoch": 62.22409638554217, "grad_norm": 12.480368614196777, "learning_rate": 6.88495145631068e-05, "loss": 0.3408, "step": 6417 }, { "epoch": 62.23373493975904, "grad_norm": 42.71347427368164, "learning_rate": 6.884466019417475e-05, "loss": 0.3337, "step": 6418 }, { "epoch": 62.2433734939759, "grad_norm": 16.374372482299805, "learning_rate": 6.883980582524272e-05, "loss": 0.212, "step": 6419 }, { "epoch": 62.25301204819277, "grad_norm": 6.64813756942749, "learning_rate": 6.883495145631068e-05, "loss": 0.1408, "step": 6420 }, { "epoch": 62.262650602409636, "grad_norm": 5.048003673553467, "learning_rate": 6.883009708737864e-05, "loss": 0.3004, "step": 6421 }, { "epoch": 62.272289156626506, "grad_norm": 19.788461685180664, "learning_rate": 6.88252427184466e-05, "loss": 0.5376, "step": 6422 }, { "epoch": 62.28192771084338, "grad_norm": 8.105327606201172, "learning_rate": 6.882038834951457e-05, "loss": 0.4102, "step": 6423 }, { "epoch": 62.29156626506024, "grad_norm": 6.821069240570068, "learning_rate": 6.881553398058253e-05, "loss": 0.3429, "step": 6424 }, { "epoch": 62.30120481927711, "grad_norm": 6.648042678833008, "learning_rate": 6.881067961165048e-05, "loss": 0.3387, "step": 6425 }, { "epoch": 62.310843373493974, "grad_norm": 11.314849853515625, "learning_rate": 6.880582524271844e-05, "loss": 0.3419, "step": 6426 }, { "epoch": 62.320481927710844, "grad_norm": 7.435220718383789, "learning_rate": 6.880097087378641e-05, "loss": 0.228, "step": 6427 }, { "epoch": 62.33012048192771, "grad_norm": 11.045910835266113, "learning_rate": 6.879611650485437e-05, "loss": 0.2802, "step": 6428 }, { "epoch": 62.33975903614458, "grad_norm": 10.954133987426758, "learning_rate": 6.879126213592233e-05, "loss": 0.2917, "step": 6429 }, { "epoch": 62.34939759036145, "grad_norm": 7.564442157745361, "learning_rate": 6.878640776699029e-05, "loss": 0.3, "step": 6430 }, { "epoch": 62.35903614457831, "grad_norm": 15.530710220336914, "learning_rate": 6.878155339805824e-05, "loss": 0.3964, "step": 6431 }, { "epoch": 62.36867469879518, "grad_norm": 4.31205940246582, "learning_rate": 6.877669902912622e-05, "loss": 0.1859, "step": 6432 }, { "epoch": 62.378313253012045, "grad_norm": 10.045417785644531, "learning_rate": 6.877184466019417e-05, "loss": 0.3103, "step": 6433 }, { "epoch": 62.387951807228916, "grad_norm": 14.565006256103516, "learning_rate": 6.876699029126213e-05, "loss": 0.172, "step": 6434 }, { "epoch": 62.397590361445786, "grad_norm": 11.876593589782715, "learning_rate": 6.876213592233009e-05, "loss": 0.2727, "step": 6435 }, { "epoch": 62.40722891566265, "grad_norm": 6.311703681945801, "learning_rate": 6.875728155339806e-05, "loss": 0.429, "step": 6436 }, { "epoch": 62.41686746987952, "grad_norm": 23.402719497680664, "learning_rate": 6.875242718446602e-05, "loss": 0.4803, "step": 6437 }, { "epoch": 62.42650602409638, "grad_norm": 9.724924087524414, "learning_rate": 6.874757281553399e-05, "loss": 0.3426, "step": 6438 }, { "epoch": 62.436144578313254, "grad_norm": 6.375317573547363, "learning_rate": 6.874271844660195e-05, "loss": 0.4372, "step": 6439 }, { "epoch": 62.44578313253012, "grad_norm": 4.387547969818115, "learning_rate": 6.87378640776699e-05, "loss": 0.3004, "step": 6440 }, { "epoch": 62.45542168674699, "grad_norm": 7.670004844665527, "learning_rate": 6.873300970873788e-05, "loss": 0.3271, "step": 6441 }, { "epoch": 62.46506024096386, "grad_norm": 20.48999786376953, "learning_rate": 6.872815533980583e-05, "loss": 0.5715, "step": 6442 }, { "epoch": 62.47469879518072, "grad_norm": 18.74738121032715, "learning_rate": 6.872330097087379e-05, "loss": 0.2837, "step": 6443 }, { "epoch": 62.48433734939759, "grad_norm": 6.385318756103516, "learning_rate": 6.871844660194175e-05, "loss": 0.1595, "step": 6444 }, { "epoch": 62.493975903614455, "grad_norm": 8.635098457336426, "learning_rate": 6.871359223300972e-05, "loss": 0.1477, "step": 6445 }, { "epoch": 62.503614457831326, "grad_norm": 10.484916687011719, "learning_rate": 6.870873786407768e-05, "loss": 0.2934, "step": 6446 }, { "epoch": 62.513253012048196, "grad_norm": 35.54679870605469, "learning_rate": 6.870388349514564e-05, "loss": 0.471, "step": 6447 }, { "epoch": 62.52289156626506, "grad_norm": 7.026519298553467, "learning_rate": 6.86990291262136e-05, "loss": 0.4854, "step": 6448 }, { "epoch": 62.53253012048193, "grad_norm": 12.246330261230469, "learning_rate": 6.869417475728157e-05, "loss": 0.3536, "step": 6449 }, { "epoch": 62.54216867469879, "grad_norm": 11.250908851623535, "learning_rate": 6.868932038834952e-05, "loss": 0.2857, "step": 6450 }, { "epoch": 62.55180722891566, "grad_norm": 18.446670532226562, "learning_rate": 6.868446601941748e-05, "loss": 0.345, "step": 6451 }, { "epoch": 62.56144578313253, "grad_norm": 10.021488189697266, "learning_rate": 6.867961165048544e-05, "loss": 0.3006, "step": 6452 }, { "epoch": 62.5710843373494, "grad_norm": 31.60624885559082, "learning_rate": 6.86747572815534e-05, "loss": 0.223, "step": 6453 }, { "epoch": 62.58072289156627, "grad_norm": 14.449596405029297, "learning_rate": 6.866990291262137e-05, "loss": 0.3091, "step": 6454 }, { "epoch": 62.59036144578313, "grad_norm": 3.9738943576812744, "learning_rate": 6.866504854368933e-05, "loss": 0.1833, "step": 6455 }, { "epoch": 62.6, "grad_norm": 15.949701309204102, "learning_rate": 6.866019417475728e-05, "loss": 0.4307, "step": 6456 }, { "epoch": 62.609638554216865, "grad_norm": 10.974160194396973, "learning_rate": 6.865533980582524e-05, "loss": 0.3346, "step": 6457 }, { "epoch": 62.619277108433735, "grad_norm": 6.424174785614014, "learning_rate": 6.865048543689321e-05, "loss": 0.2575, "step": 6458 }, { "epoch": 62.628915662650606, "grad_norm": 19.276094436645508, "learning_rate": 6.864563106796117e-05, "loss": 0.2633, "step": 6459 }, { "epoch": 62.63855421686747, "grad_norm": 9.54018783569336, "learning_rate": 6.864077669902913e-05, "loss": 0.4067, "step": 6460 }, { "epoch": 62.64819277108434, "grad_norm": 5.108623504638672, "learning_rate": 6.863592233009709e-05, "loss": 0.3534, "step": 6461 }, { "epoch": 62.6578313253012, "grad_norm": 19.736764907836914, "learning_rate": 6.863106796116506e-05, "loss": 0.3721, "step": 6462 }, { "epoch": 62.66746987951807, "grad_norm": 19.167329788208008, "learning_rate": 6.862621359223302e-05, "loss": 0.2377, "step": 6463 }, { "epoch": 62.67710843373494, "grad_norm": 7.252110481262207, "learning_rate": 6.862135922330097e-05, "loss": 0.2592, "step": 6464 }, { "epoch": 62.68674698795181, "grad_norm": 17.22828483581543, "learning_rate": 6.861650485436893e-05, "loss": 0.2356, "step": 6465 }, { "epoch": 62.69638554216868, "grad_norm": 8.46289348602295, "learning_rate": 6.86116504854369e-05, "loss": 0.2071, "step": 6466 }, { "epoch": 62.70602409638554, "grad_norm": 20.60588836669922, "learning_rate": 6.860679611650486e-05, "loss": 0.2551, "step": 6467 }, { "epoch": 62.71566265060241, "grad_norm": 6.114762306213379, "learning_rate": 6.860194174757282e-05, "loss": 0.1829, "step": 6468 }, { "epoch": 62.725301204819274, "grad_norm": 9.368194580078125, "learning_rate": 6.859708737864078e-05, "loss": 0.3792, "step": 6469 }, { "epoch": 62.734939759036145, "grad_norm": 8.968371391296387, "learning_rate": 6.859223300970873e-05, "loss": 0.3058, "step": 6470 }, { "epoch": 62.744578313253015, "grad_norm": 11.366025924682617, "learning_rate": 6.85873786407767e-05, "loss": 0.5793, "step": 6471 }, { "epoch": 62.75421686746988, "grad_norm": 24.56887435913086, "learning_rate": 6.858252427184466e-05, "loss": 0.2924, "step": 6472 }, { "epoch": 62.76385542168675, "grad_norm": 80.3097152709961, "learning_rate": 6.857766990291262e-05, "loss": 0.358, "step": 6473 }, { "epoch": 62.77349397590361, "grad_norm": 7.158097743988037, "learning_rate": 6.857281553398058e-05, "loss": 0.222, "step": 6474 }, { "epoch": 62.78313253012048, "grad_norm": 19.331308364868164, "learning_rate": 6.856796116504855e-05, "loss": 0.3474, "step": 6475 }, { "epoch": 62.792771084337346, "grad_norm": 92.49017333984375, "learning_rate": 6.85631067961165e-05, "loss": 0.4492, "step": 6476 }, { "epoch": 62.80240963855422, "grad_norm": 6.801537990570068, "learning_rate": 6.855825242718446e-05, "loss": 0.2092, "step": 6477 }, { "epoch": 62.81204819277109, "grad_norm": 16.636873245239258, "learning_rate": 6.855339805825242e-05, "loss": 0.4596, "step": 6478 }, { "epoch": 62.82168674698795, "grad_norm": 20.340736389160156, "learning_rate": 6.85485436893204e-05, "loss": 0.3992, "step": 6479 }, { "epoch": 62.83132530120482, "grad_norm": 8.477664947509766, "learning_rate": 6.854368932038835e-05, "loss": 0.2321, "step": 6480 }, { "epoch": 62.840963855421684, "grad_norm": 20.380632400512695, "learning_rate": 6.853883495145631e-05, "loss": 0.5298, "step": 6481 }, { "epoch": 62.850602409638554, "grad_norm": 9.530606269836426, "learning_rate": 6.853398058252427e-05, "loss": 0.2192, "step": 6482 }, { "epoch": 62.860240963855425, "grad_norm": 9.410666465759277, "learning_rate": 6.852912621359224e-05, "loss": 0.2117, "step": 6483 }, { "epoch": 62.86987951807229, "grad_norm": 22.531875610351562, "learning_rate": 6.85242718446602e-05, "loss": 0.3189, "step": 6484 }, { "epoch": 62.87951807228916, "grad_norm": 17.163921356201172, "learning_rate": 6.851941747572815e-05, "loss": 0.331, "step": 6485 }, { "epoch": 62.88915662650602, "grad_norm": 8.26671028137207, "learning_rate": 6.851456310679611e-05, "loss": 0.2516, "step": 6486 }, { "epoch": 62.89879518072289, "grad_norm": 5.151186466217041, "learning_rate": 6.850970873786407e-05, "loss": 0.26, "step": 6487 }, { "epoch": 62.908433734939756, "grad_norm": 12.449390411376953, "learning_rate": 6.850485436893205e-05, "loss": 0.2058, "step": 6488 }, { "epoch": 62.918072289156626, "grad_norm": 6.776327610015869, "learning_rate": 6.850000000000001e-05, "loss": 0.2821, "step": 6489 }, { "epoch": 62.9277108433735, "grad_norm": 16.989912033081055, "learning_rate": 6.849514563106797e-05, "loss": 0.4629, "step": 6490 }, { "epoch": 62.93734939759036, "grad_norm": 12.660614967346191, "learning_rate": 6.849029126213593e-05, "loss": 0.2923, "step": 6491 }, { "epoch": 62.94698795180723, "grad_norm": 74.27203369140625, "learning_rate": 6.848543689320388e-05, "loss": 0.3029, "step": 6492 }, { "epoch": 62.956626506024094, "grad_norm": 9.997055053710938, "learning_rate": 6.848058252427186e-05, "loss": 0.2296, "step": 6493 }, { "epoch": 62.966265060240964, "grad_norm": 60.254249572753906, "learning_rate": 6.847572815533981e-05, "loss": 0.1587, "step": 6494 }, { "epoch": 62.975903614457835, "grad_norm": 6.587606430053711, "learning_rate": 6.847087378640777e-05, "loss": 0.143, "step": 6495 }, { "epoch": 62.9855421686747, "grad_norm": 5.999547481536865, "learning_rate": 6.846601941747573e-05, "loss": 0.3603, "step": 6496 }, { "epoch": 63.001204819277106, "grad_norm": 18.328760147094727, "learning_rate": 6.84611650485437e-05, "loss": 0.2259, "step": 6497 }, { "epoch": 63.01084337349398, "grad_norm": 9.275300979614258, "learning_rate": 6.845631067961166e-05, "loss": 0.3316, "step": 6498 }, { "epoch": 63.02048192771084, "grad_norm": 14.018143653869629, "learning_rate": 6.845145631067962e-05, "loss": 0.4184, "step": 6499 }, { "epoch": 63.03012048192771, "grad_norm": 11.030092239379883, "learning_rate": 6.844660194174757e-05, "loss": 0.2217, "step": 6500 }, { "epoch": 63.03975903614458, "grad_norm": 8.231854438781738, "learning_rate": 6.844174757281555e-05, "loss": 0.2338, "step": 6501 }, { "epoch": 63.049397590361444, "grad_norm": 12.193792343139648, "learning_rate": 6.84368932038835e-05, "loss": 0.5138, "step": 6502 }, { "epoch": 63.059036144578315, "grad_norm": 3.2835001945495605, "learning_rate": 6.843203883495146e-05, "loss": 0.1544, "step": 6503 }, { "epoch": 63.06867469879518, "grad_norm": 13.756373405456543, "learning_rate": 6.842718446601942e-05, "loss": 0.2548, "step": 6504 }, { "epoch": 63.07831325301205, "grad_norm": 10.076456069946289, "learning_rate": 6.842233009708739e-05, "loss": 0.2193, "step": 6505 }, { "epoch": 63.08795180722892, "grad_norm": 32.76690673828125, "learning_rate": 6.841747572815535e-05, "loss": 0.4422, "step": 6506 }, { "epoch": 63.09759036144578, "grad_norm": 13.778871536254883, "learning_rate": 6.84126213592233e-05, "loss": 0.4065, "step": 6507 }, { "epoch": 63.10722891566265, "grad_norm": 6.891286849975586, "learning_rate": 6.840776699029126e-05, "loss": 0.2206, "step": 6508 }, { "epoch": 63.116867469879516, "grad_norm": 41.69014358520508, "learning_rate": 6.840291262135922e-05, "loss": 0.3763, "step": 6509 }, { "epoch": 63.126506024096386, "grad_norm": 6.72434139251709, "learning_rate": 6.839805825242719e-05, "loss": 0.3107, "step": 6510 }, { "epoch": 63.13614457831325, "grad_norm": 11.950814247131348, "learning_rate": 6.839320388349515e-05, "loss": 0.1602, "step": 6511 }, { "epoch": 63.14578313253012, "grad_norm": 92.93323516845703, "learning_rate": 6.838834951456311e-05, "loss": 0.3132, "step": 6512 }, { "epoch": 63.15542168674699, "grad_norm": 11.96870231628418, "learning_rate": 6.838349514563107e-05, "loss": 0.3322, "step": 6513 }, { "epoch": 63.165060240963854, "grad_norm": 14.801607131958008, "learning_rate": 6.837864077669904e-05, "loss": 0.2962, "step": 6514 }, { "epoch": 63.174698795180724, "grad_norm": 17.071117401123047, "learning_rate": 6.8373786407767e-05, "loss": 0.362, "step": 6515 }, { "epoch": 63.18433734939759, "grad_norm": 16.317119598388672, "learning_rate": 6.836893203883495e-05, "loss": 0.2671, "step": 6516 }, { "epoch": 63.19397590361446, "grad_norm": 6.606090545654297, "learning_rate": 6.836407766990291e-05, "loss": 0.2332, "step": 6517 }, { "epoch": 63.20361445783133, "grad_norm": 7.269511699676514, "learning_rate": 6.835922330097088e-05, "loss": 0.2657, "step": 6518 }, { "epoch": 63.21325301204819, "grad_norm": 7.112913131713867, "learning_rate": 6.835436893203884e-05, "loss": 0.4088, "step": 6519 }, { "epoch": 63.22289156626506, "grad_norm": 13.119656562805176, "learning_rate": 6.83495145631068e-05, "loss": 0.265, "step": 6520 }, { "epoch": 63.232530120481925, "grad_norm": 11.714738845825195, "learning_rate": 6.834466019417475e-05, "loss": 0.3729, "step": 6521 }, { "epoch": 63.242168674698796, "grad_norm": 9.834344863891602, "learning_rate": 6.833980582524273e-05, "loss": 0.4556, "step": 6522 }, { "epoch": 63.25180722891566, "grad_norm": 11.574919700622559, "learning_rate": 6.833495145631068e-05, "loss": 0.1485, "step": 6523 }, { "epoch": 63.26144578313253, "grad_norm": 27.294536590576172, "learning_rate": 6.833009708737864e-05, "loss": 0.3039, "step": 6524 }, { "epoch": 63.2710843373494, "grad_norm": 42.96181106567383, "learning_rate": 6.83252427184466e-05, "loss": 0.4003, "step": 6525 }, { "epoch": 63.28072289156626, "grad_norm": 51.96223068237305, "learning_rate": 6.832038834951456e-05, "loss": 0.1867, "step": 6526 }, { "epoch": 63.290361445783134, "grad_norm": 13.146376609802246, "learning_rate": 6.831553398058253e-05, "loss": 0.3269, "step": 6527 }, { "epoch": 63.3, "grad_norm": 17.72578239440918, "learning_rate": 6.831067961165049e-05, "loss": 0.3297, "step": 6528 }, { "epoch": 63.30963855421687, "grad_norm": 75.84606170654297, "learning_rate": 6.830582524271844e-05, "loss": 0.3005, "step": 6529 }, { "epoch": 63.31927710843374, "grad_norm": 11.28494930267334, "learning_rate": 6.83009708737864e-05, "loss": 0.2797, "step": 6530 }, { "epoch": 63.3289156626506, "grad_norm": 11.631440162658691, "learning_rate": 6.829611650485437e-05, "loss": 0.3246, "step": 6531 }, { "epoch": 63.33855421686747, "grad_norm": 6.986293792724609, "learning_rate": 6.829126213592233e-05, "loss": 0.1696, "step": 6532 }, { "epoch": 63.348192771084335, "grad_norm": 12.094707489013672, "learning_rate": 6.828640776699029e-05, "loss": 0.2185, "step": 6533 }, { "epoch": 63.357831325301206, "grad_norm": 17.49024200439453, "learning_rate": 6.828155339805825e-05, "loss": 0.5067, "step": 6534 }, { "epoch": 63.36746987951807, "grad_norm": 25.335298538208008, "learning_rate": 6.827669902912622e-05, "loss": 0.2719, "step": 6535 }, { "epoch": 63.37710843373494, "grad_norm": 3.8017992973327637, "learning_rate": 6.827184466019418e-05, "loss": 0.1254, "step": 6536 }, { "epoch": 63.38674698795181, "grad_norm": 7.698104381561279, "learning_rate": 6.826699029126213e-05, "loss": 0.1644, "step": 6537 }, { "epoch": 63.39638554216867, "grad_norm": 24.540807723999023, "learning_rate": 6.826213592233009e-05, "loss": 0.1988, "step": 6538 }, { "epoch": 63.40602409638554, "grad_norm": 11.105274200439453, "learning_rate": 6.825728155339806e-05, "loss": 0.2051, "step": 6539 }, { "epoch": 63.41566265060241, "grad_norm": 9.205570220947266, "learning_rate": 6.825242718446603e-05, "loss": 0.2294, "step": 6540 }, { "epoch": 63.42530120481928, "grad_norm": 8.36434555053711, "learning_rate": 6.824757281553399e-05, "loss": 0.2996, "step": 6541 }, { "epoch": 63.43493975903615, "grad_norm": 7.536224842071533, "learning_rate": 6.824271844660195e-05, "loss": 0.2768, "step": 6542 }, { "epoch": 63.44457831325301, "grad_norm": 5.133694171905518, "learning_rate": 6.82378640776699e-05, "loss": 0.1721, "step": 6543 }, { "epoch": 63.45421686746988, "grad_norm": 38.27645492553711, "learning_rate": 6.823300970873788e-05, "loss": 0.3332, "step": 6544 }, { "epoch": 63.463855421686745, "grad_norm": 12.0804443359375, "learning_rate": 6.822815533980584e-05, "loss": 0.4834, "step": 6545 }, { "epoch": 63.473493975903615, "grad_norm": 29.64126205444336, "learning_rate": 6.82233009708738e-05, "loss": 0.4615, "step": 6546 }, { "epoch": 63.48313253012048, "grad_norm": 6.831211090087891, "learning_rate": 6.821844660194175e-05, "loss": 0.1841, "step": 6547 }, { "epoch": 63.49277108433735, "grad_norm": 11.772541999816895, "learning_rate": 6.821359223300971e-05, "loss": 0.1922, "step": 6548 }, { "epoch": 63.50240963855422, "grad_norm": 43.602603912353516, "learning_rate": 6.820873786407768e-05, "loss": 0.1969, "step": 6549 }, { "epoch": 63.51204819277108, "grad_norm": 13.717942237854004, "learning_rate": 6.820388349514564e-05, "loss": 0.3422, "step": 6550 }, { "epoch": 63.52168674698795, "grad_norm": 16.730253219604492, "learning_rate": 6.81990291262136e-05, "loss": 0.3497, "step": 6551 }, { "epoch": 63.53132530120482, "grad_norm": 6.309300899505615, "learning_rate": 6.819417475728155e-05, "loss": 0.2242, "step": 6552 }, { "epoch": 63.54096385542169, "grad_norm": 18.385787963867188, "learning_rate": 6.818932038834952e-05, "loss": 0.3716, "step": 6553 }, { "epoch": 63.55060240963856, "grad_norm": 10.04916763305664, "learning_rate": 6.818446601941748e-05, "loss": 0.3941, "step": 6554 }, { "epoch": 63.56024096385542, "grad_norm": 11.65825366973877, "learning_rate": 6.817961165048544e-05, "loss": 0.2541, "step": 6555 }, { "epoch": 63.56987951807229, "grad_norm": 52.544921875, "learning_rate": 6.81747572815534e-05, "loss": 0.5079, "step": 6556 }, { "epoch": 63.579518072289154, "grad_norm": 9.912969589233398, "learning_rate": 6.816990291262137e-05, "loss": 0.1889, "step": 6557 }, { "epoch": 63.589156626506025, "grad_norm": 14.021163940429688, "learning_rate": 6.816504854368933e-05, "loss": 0.2256, "step": 6558 }, { "epoch": 63.59879518072289, "grad_norm": 9.563776016235352, "learning_rate": 6.816019417475729e-05, "loss": 0.3121, "step": 6559 }, { "epoch": 63.60843373493976, "grad_norm": 5.431805610656738, "learning_rate": 6.815533980582524e-05, "loss": 0.2959, "step": 6560 }, { "epoch": 63.61807228915663, "grad_norm": 13.52214241027832, "learning_rate": 6.815048543689321e-05, "loss": 0.5025, "step": 6561 }, { "epoch": 63.62771084337349, "grad_norm": 21.583330154418945, "learning_rate": 6.814563106796117e-05, "loss": 0.3109, "step": 6562 }, { "epoch": 63.63734939759036, "grad_norm": 8.785005569458008, "learning_rate": 6.814077669902913e-05, "loss": 0.3558, "step": 6563 }, { "epoch": 63.646987951807226, "grad_norm": 21.5073184967041, "learning_rate": 6.813592233009709e-05, "loss": 0.3581, "step": 6564 }, { "epoch": 63.6566265060241, "grad_norm": 9.861814498901367, "learning_rate": 6.813106796116505e-05, "loss": 0.3493, "step": 6565 }, { "epoch": 63.66626506024097, "grad_norm": 22.509761810302734, "learning_rate": 6.812621359223302e-05, "loss": 0.3184, "step": 6566 }, { "epoch": 63.67590361445783, "grad_norm": 13.472326278686523, "learning_rate": 6.812135922330097e-05, "loss": 0.3573, "step": 6567 }, { "epoch": 63.6855421686747, "grad_norm": 17.533132553100586, "learning_rate": 6.811650485436893e-05, "loss": 0.3228, "step": 6568 }, { "epoch": 63.695180722891564, "grad_norm": 7.453120708465576, "learning_rate": 6.811165048543689e-05, "loss": 0.2488, "step": 6569 }, { "epoch": 63.704819277108435, "grad_norm": 27.86464500427246, "learning_rate": 6.810679611650486e-05, "loss": 0.4567, "step": 6570 }, { "epoch": 63.7144578313253, "grad_norm": 8.279170989990234, "learning_rate": 6.810194174757282e-05, "loss": 0.4926, "step": 6571 }, { "epoch": 63.72409638554217, "grad_norm": 21.50187110900879, "learning_rate": 6.809708737864078e-05, "loss": 0.1321, "step": 6572 }, { "epoch": 63.73373493975904, "grad_norm": 11.119619369506836, "learning_rate": 6.809223300970873e-05, "loss": 0.4192, "step": 6573 }, { "epoch": 63.7433734939759, "grad_norm": 5.176772594451904, "learning_rate": 6.80873786407767e-05, "loss": 0.2482, "step": 6574 }, { "epoch": 63.75301204819277, "grad_norm": 10.259053230285645, "learning_rate": 6.808252427184466e-05, "loss": 0.4598, "step": 6575 }, { "epoch": 63.762650602409636, "grad_norm": 12.109048843383789, "learning_rate": 6.807766990291262e-05, "loss": 0.3127, "step": 6576 }, { "epoch": 63.772289156626506, "grad_norm": 19.103668212890625, "learning_rate": 6.807281553398058e-05, "loss": 0.3176, "step": 6577 }, { "epoch": 63.78192771084338, "grad_norm": 41.02485275268555, "learning_rate": 6.806796116504855e-05, "loss": 0.3422, "step": 6578 }, { "epoch": 63.79156626506024, "grad_norm": 6.0682525634765625, "learning_rate": 6.806310679611651e-05, "loss": 0.2611, "step": 6579 }, { "epoch": 63.80120481927711, "grad_norm": 9.456031799316406, "learning_rate": 6.805825242718447e-05, "loss": 0.4434, "step": 6580 }, { "epoch": 63.810843373493974, "grad_norm": 13.804224014282227, "learning_rate": 6.805339805825242e-05, "loss": 0.4458, "step": 6581 }, { "epoch": 63.820481927710844, "grad_norm": 10.544405937194824, "learning_rate": 6.804854368932038e-05, "loss": 0.4162, "step": 6582 }, { "epoch": 63.83012048192771, "grad_norm": 29.260412216186523, "learning_rate": 6.804368932038835e-05, "loss": 0.2923, "step": 6583 }, { "epoch": 63.83975903614458, "grad_norm": 8.63221263885498, "learning_rate": 6.803883495145631e-05, "loss": 0.3153, "step": 6584 }, { "epoch": 63.84939759036145, "grad_norm": 14.757081985473633, "learning_rate": 6.803398058252427e-05, "loss": 0.259, "step": 6585 }, { "epoch": 63.85903614457831, "grad_norm": 15.442715644836426, "learning_rate": 6.802912621359223e-05, "loss": 0.2445, "step": 6586 }, { "epoch": 63.86867469879518, "grad_norm": 8.280701637268066, "learning_rate": 6.80242718446602e-05, "loss": 0.4502, "step": 6587 }, { "epoch": 63.878313253012045, "grad_norm": 4.262746810913086, "learning_rate": 6.801941747572815e-05, "loss": 0.1731, "step": 6588 }, { "epoch": 63.887951807228916, "grad_norm": 6.901000022888184, "learning_rate": 6.801456310679613e-05, "loss": 0.2311, "step": 6589 }, { "epoch": 63.897590361445786, "grad_norm": 29.3737735748291, "learning_rate": 6.800970873786408e-05, "loss": 0.4179, "step": 6590 }, { "epoch": 63.90722891566265, "grad_norm": 20.23407554626465, "learning_rate": 6.800485436893204e-05, "loss": 0.317, "step": 6591 }, { "epoch": 63.91686746987952, "grad_norm": 5.964539527893066, "learning_rate": 6.800000000000001e-05, "loss": 0.2511, "step": 6592 }, { "epoch": 63.92650602409638, "grad_norm": 128.07113647460938, "learning_rate": 6.799514563106797e-05, "loss": 0.3353, "step": 6593 }, { "epoch": 63.936144578313254, "grad_norm": 11.40372371673584, "learning_rate": 6.799029126213593e-05, "loss": 0.285, "step": 6594 }, { "epoch": 63.94578313253012, "grad_norm": 6.818202972412109, "learning_rate": 6.798543689320389e-05, "loss": 0.2084, "step": 6595 }, { "epoch": 63.95542168674699, "grad_norm": 3.798379421234131, "learning_rate": 6.798058252427186e-05, "loss": 0.1732, "step": 6596 }, { "epoch": 63.96506024096386, "grad_norm": 7.033954620361328, "learning_rate": 6.797572815533982e-05, "loss": 0.353, "step": 6597 }, { "epoch": 63.97469879518072, "grad_norm": 23.67427635192871, "learning_rate": 6.797087378640777e-05, "loss": 0.313, "step": 6598 }, { "epoch": 63.98433734939759, "grad_norm": 19.553096771240234, "learning_rate": 6.796601941747573e-05, "loss": 0.3089, "step": 6599 }, { "epoch": 63.993975903614455, "grad_norm": 21.687435150146484, "learning_rate": 6.79611650485437e-05, "loss": 0.3461, "step": 6600 }, { "epoch": 64.00963855421686, "grad_norm": 30.39788246154785, "learning_rate": 6.795631067961166e-05, "loss": 0.1855, "step": 6601 }, { "epoch": 64.01927710843374, "grad_norm": 9.083596229553223, "learning_rate": 6.795145631067962e-05, "loss": 0.2233, "step": 6602 }, { "epoch": 64.0289156626506, "grad_norm": 8.84108829498291, "learning_rate": 6.794660194174758e-05, "loss": 0.0986, "step": 6603 }, { "epoch": 64.03855421686747, "grad_norm": 8.503985404968262, "learning_rate": 6.794174757281553e-05, "loss": 0.2893, "step": 6604 }, { "epoch": 64.04819277108433, "grad_norm": 15.122008323669434, "learning_rate": 6.79368932038835e-05, "loss": 0.5266, "step": 6605 }, { "epoch": 64.05783132530121, "grad_norm": 10.823512077331543, "learning_rate": 6.793203883495146e-05, "loss": 0.489, "step": 6606 }, { "epoch": 64.06746987951807, "grad_norm": 9.155203819274902, "learning_rate": 6.792718446601942e-05, "loss": 0.4195, "step": 6607 }, { "epoch": 64.07710843373494, "grad_norm": 10.085539817810059, "learning_rate": 6.792233009708738e-05, "loss": 0.3562, "step": 6608 }, { "epoch": 64.08674698795181, "grad_norm": 19.845531463623047, "learning_rate": 6.791747572815535e-05, "loss": 0.4095, "step": 6609 }, { "epoch": 64.09638554216868, "grad_norm": 13.049212455749512, "learning_rate": 6.791262135922331e-05, "loss": 0.2934, "step": 6610 }, { "epoch": 64.10602409638554, "grad_norm": 5.7338643074035645, "learning_rate": 6.790776699029126e-05, "loss": 0.2608, "step": 6611 }, { "epoch": 64.1156626506024, "grad_norm": 10.213212013244629, "learning_rate": 6.790291262135922e-05, "loss": 0.2535, "step": 6612 }, { "epoch": 64.12530120481928, "grad_norm": 9.549551010131836, "learning_rate": 6.78980582524272e-05, "loss": 0.3425, "step": 6613 }, { "epoch": 64.13493975903614, "grad_norm": 8.62022876739502, "learning_rate": 6.789320388349515e-05, "loss": 0.2361, "step": 6614 }, { "epoch": 64.144578313253, "grad_norm": 8.224886894226074, "learning_rate": 6.788834951456311e-05, "loss": 0.2592, "step": 6615 }, { "epoch": 64.15421686746988, "grad_norm": 9.748292922973633, "learning_rate": 6.788349514563107e-05, "loss": 0.5216, "step": 6616 }, { "epoch": 64.16385542168675, "grad_norm": 7.545963764190674, "learning_rate": 6.787864077669904e-05, "loss": 0.2385, "step": 6617 }, { "epoch": 64.17349397590361, "grad_norm": 14.474712371826172, "learning_rate": 6.7873786407767e-05, "loss": 0.2552, "step": 6618 }, { "epoch": 64.18313253012049, "grad_norm": 27.862123489379883, "learning_rate": 6.786893203883495e-05, "loss": 0.5082, "step": 6619 }, { "epoch": 64.19277108433735, "grad_norm": 22.04400062561035, "learning_rate": 6.786407766990291e-05, "loss": 0.3369, "step": 6620 }, { "epoch": 64.20240963855422, "grad_norm": 51.749534606933594, "learning_rate": 6.785922330097087e-05, "loss": 0.5877, "step": 6621 }, { "epoch": 64.21204819277108, "grad_norm": 15.513418197631836, "learning_rate": 6.785436893203884e-05, "loss": 0.6258, "step": 6622 }, { "epoch": 64.22168674698796, "grad_norm": 4.118950843811035, "learning_rate": 6.78495145631068e-05, "loss": 0.1254, "step": 6623 }, { "epoch": 64.23132530120482, "grad_norm": 9.804727554321289, "learning_rate": 6.784466019417476e-05, "loss": 0.3955, "step": 6624 }, { "epoch": 64.24096385542168, "grad_norm": 15.981637001037598, "learning_rate": 6.783980582524271e-05, "loss": 0.3404, "step": 6625 }, { "epoch": 64.25060240963856, "grad_norm": 10.968348503112793, "learning_rate": 6.783495145631069e-05, "loss": 0.6454, "step": 6626 }, { "epoch": 64.26024096385542, "grad_norm": 10.217589378356934, "learning_rate": 6.783009708737864e-05, "loss": 0.4794, "step": 6627 }, { "epoch": 64.26987951807229, "grad_norm": 11.381455421447754, "learning_rate": 6.78252427184466e-05, "loss": 0.2971, "step": 6628 }, { "epoch": 64.27951807228915, "grad_norm": 11.959698677062988, "learning_rate": 6.782038834951456e-05, "loss": 0.2699, "step": 6629 }, { "epoch": 64.28915662650603, "grad_norm": 24.653282165527344, "learning_rate": 6.781553398058253e-05, "loss": 0.4566, "step": 6630 }, { "epoch": 64.29879518072289, "grad_norm": 7.989689826965332, "learning_rate": 6.781067961165049e-05, "loss": 0.2774, "step": 6631 }, { "epoch": 64.30843373493975, "grad_norm": 7.421775817871094, "learning_rate": 6.780582524271845e-05, "loss": 0.1319, "step": 6632 }, { "epoch": 64.31807228915663, "grad_norm": 21.14030647277832, "learning_rate": 6.78009708737864e-05, "loss": 0.268, "step": 6633 }, { "epoch": 64.3277108433735, "grad_norm": 13.76484203338623, "learning_rate": 6.779611650485437e-05, "loss": 0.5109, "step": 6634 }, { "epoch": 64.33734939759036, "grad_norm": 16.499319076538086, "learning_rate": 6.779126213592233e-05, "loss": 0.4887, "step": 6635 }, { "epoch": 64.34698795180722, "grad_norm": 6.855433464050293, "learning_rate": 6.778640776699029e-05, "loss": 0.1854, "step": 6636 }, { "epoch": 64.3566265060241, "grad_norm": 9.769676208496094, "learning_rate": 6.778155339805825e-05, "loss": 0.3764, "step": 6637 }, { "epoch": 64.36626506024096, "grad_norm": 6.734720230102539, "learning_rate": 6.77766990291262e-05, "loss": 0.1383, "step": 6638 }, { "epoch": 64.37590361445783, "grad_norm": 15.59817123413086, "learning_rate": 6.777184466019418e-05, "loss": 0.4984, "step": 6639 }, { "epoch": 64.3855421686747, "grad_norm": 5.852312088012695, "learning_rate": 6.776699029126215e-05, "loss": 0.3294, "step": 6640 }, { "epoch": 64.39518072289157, "grad_norm": 15.583005905151367, "learning_rate": 6.77621359223301e-05, "loss": 0.4019, "step": 6641 }, { "epoch": 64.40481927710843, "grad_norm": 8.678729057312012, "learning_rate": 6.775728155339806e-05, "loss": 0.2841, "step": 6642 }, { "epoch": 64.41445783132531, "grad_norm": 21.439847946166992, "learning_rate": 6.775242718446602e-05, "loss": 0.3684, "step": 6643 }, { "epoch": 64.42409638554217, "grad_norm": 7.720766067504883, "learning_rate": 6.774757281553399e-05, "loss": 0.2996, "step": 6644 }, { "epoch": 64.43373493975903, "grad_norm": 7.125592231750488, "learning_rate": 6.774271844660195e-05, "loss": 0.4145, "step": 6645 }, { "epoch": 64.4433734939759, "grad_norm": 7.205031394958496, "learning_rate": 6.773786407766991e-05, "loss": 0.2048, "step": 6646 }, { "epoch": 64.45301204819278, "grad_norm": 6.899515151977539, "learning_rate": 6.773300970873787e-05, "loss": 0.3472, "step": 6647 }, { "epoch": 64.46265060240964, "grad_norm": 10.64297866821289, "learning_rate": 6.772815533980584e-05, "loss": 0.2225, "step": 6648 }, { "epoch": 64.4722891566265, "grad_norm": 10.481184005737305, "learning_rate": 6.77233009708738e-05, "loss": 0.2725, "step": 6649 }, { "epoch": 64.48192771084338, "grad_norm": 9.914949417114258, "learning_rate": 6.771844660194175e-05, "loss": 0.2865, "step": 6650 }, { "epoch": 64.49156626506024, "grad_norm": 28.952199935913086, "learning_rate": 6.771359223300971e-05, "loss": 0.5354, "step": 6651 }, { "epoch": 64.5012048192771, "grad_norm": 16.13902473449707, "learning_rate": 6.770873786407768e-05, "loss": 0.5189, "step": 6652 }, { "epoch": 64.51084337349397, "grad_norm": 15.583375930786133, "learning_rate": 6.770388349514564e-05, "loss": 0.239, "step": 6653 }, { "epoch": 64.52048192771085, "grad_norm": 9.85528564453125, "learning_rate": 6.76990291262136e-05, "loss": 0.2915, "step": 6654 }, { "epoch": 64.53012048192771, "grad_norm": 16.200708389282227, "learning_rate": 6.769417475728156e-05, "loss": 0.2226, "step": 6655 }, { "epoch": 64.53975903614457, "grad_norm": 7.187249183654785, "learning_rate": 6.768932038834953e-05, "loss": 0.1883, "step": 6656 }, { "epoch": 64.54939759036145, "grad_norm": 7.2717604637146, "learning_rate": 6.768446601941748e-05, "loss": 0.4586, "step": 6657 }, { "epoch": 64.55903614457831, "grad_norm": 5.229335308074951, "learning_rate": 6.767961165048544e-05, "loss": 0.2811, "step": 6658 }, { "epoch": 64.56867469879518, "grad_norm": 20.980224609375, "learning_rate": 6.76747572815534e-05, "loss": 0.2432, "step": 6659 }, { "epoch": 64.57831325301204, "grad_norm": 8.96668529510498, "learning_rate": 6.766990291262136e-05, "loss": 0.2795, "step": 6660 }, { "epoch": 64.58795180722892, "grad_norm": 10.551125526428223, "learning_rate": 6.766504854368933e-05, "loss": 0.2159, "step": 6661 }, { "epoch": 64.59759036144578, "grad_norm": 9.594823837280273, "learning_rate": 6.766019417475729e-05, "loss": 0.2514, "step": 6662 }, { "epoch": 64.60722891566265, "grad_norm": 24.13766098022461, "learning_rate": 6.765533980582524e-05, "loss": 0.4695, "step": 6663 }, { "epoch": 64.61686746987952, "grad_norm": 10.075274467468262, "learning_rate": 6.76504854368932e-05, "loss": 0.4258, "step": 6664 }, { "epoch": 64.62650602409639, "grad_norm": 4.816701889038086, "learning_rate": 6.764563106796117e-05, "loss": 0.1992, "step": 6665 }, { "epoch": 64.63614457831325, "grad_norm": 6.2585906982421875, "learning_rate": 6.764077669902913e-05, "loss": 0.39, "step": 6666 }, { "epoch": 64.64578313253013, "grad_norm": 6.472599029541016, "learning_rate": 6.763592233009709e-05, "loss": 0.2018, "step": 6667 }, { "epoch": 64.65542168674699, "grad_norm": 10.50671672821045, "learning_rate": 6.763106796116505e-05, "loss": 0.3601, "step": 6668 }, { "epoch": 64.66506024096385, "grad_norm": 10.882716178894043, "learning_rate": 6.762621359223302e-05, "loss": 0.3026, "step": 6669 }, { "epoch": 64.67469879518072, "grad_norm": 12.925894737243652, "learning_rate": 6.762135922330098e-05, "loss": 0.3236, "step": 6670 }, { "epoch": 64.6843373493976, "grad_norm": 13.81333065032959, "learning_rate": 6.761650485436893e-05, "loss": 0.2445, "step": 6671 }, { "epoch": 64.69397590361446, "grad_norm": 20.92262840270996, "learning_rate": 6.761165048543689e-05, "loss": 0.308, "step": 6672 }, { "epoch": 64.70361445783132, "grad_norm": 11.153858184814453, "learning_rate": 6.760679611650486e-05, "loss": 0.3202, "step": 6673 }, { "epoch": 64.7132530120482, "grad_norm": 12.782146453857422, "learning_rate": 6.760194174757282e-05, "loss": 0.3293, "step": 6674 }, { "epoch": 64.72289156626506, "grad_norm": 9.498788833618164, "learning_rate": 6.759708737864078e-05, "loss": 0.282, "step": 6675 }, { "epoch": 64.73253012048193, "grad_norm": 12.638035774230957, "learning_rate": 6.759223300970874e-05, "loss": 0.4861, "step": 6676 }, { "epoch": 64.74216867469879, "grad_norm": 5.346834182739258, "learning_rate": 6.75873786407767e-05, "loss": 0.1857, "step": 6677 }, { "epoch": 64.75180722891567, "grad_norm": 9.161766052246094, "learning_rate": 6.758252427184466e-05, "loss": 0.1465, "step": 6678 }, { "epoch": 64.76144578313253, "grad_norm": 4.915650367736816, "learning_rate": 6.757766990291262e-05, "loss": 0.1523, "step": 6679 }, { "epoch": 64.7710843373494, "grad_norm": 11.989215850830078, "learning_rate": 6.757281553398058e-05, "loss": 0.2255, "step": 6680 }, { "epoch": 64.78072289156627, "grad_norm": 9.083693504333496, "learning_rate": 6.756796116504854e-05, "loss": 0.3183, "step": 6681 }, { "epoch": 64.79036144578313, "grad_norm": 8.482139587402344, "learning_rate": 6.756310679611651e-05, "loss": 0.293, "step": 6682 }, { "epoch": 64.8, "grad_norm": 7.112489223480225, "learning_rate": 6.755825242718447e-05, "loss": 0.2459, "step": 6683 }, { "epoch": 64.80963855421686, "grad_norm": 14.884392738342285, "learning_rate": 6.755339805825242e-05, "loss": 0.4068, "step": 6684 }, { "epoch": 64.81927710843374, "grad_norm": 6.096707820892334, "learning_rate": 6.754854368932038e-05, "loss": 0.1967, "step": 6685 }, { "epoch": 64.8289156626506, "grad_norm": 3.9081695079803467, "learning_rate": 6.754368932038835e-05, "loss": 0.1584, "step": 6686 }, { "epoch": 64.83855421686746, "grad_norm": 6.120231628417969, "learning_rate": 6.753883495145631e-05, "loss": 0.3109, "step": 6687 }, { "epoch": 64.84819277108434, "grad_norm": 18.08721351623535, "learning_rate": 6.753398058252427e-05, "loss": 0.4122, "step": 6688 }, { "epoch": 64.8578313253012, "grad_norm": 14.620268821716309, "learning_rate": 6.752912621359223e-05, "loss": 0.2443, "step": 6689 }, { "epoch": 64.86746987951807, "grad_norm": 3.6893582344055176, "learning_rate": 6.75242718446602e-05, "loss": 0.1369, "step": 6690 }, { "epoch": 64.87710843373495, "grad_norm": 14.484678268432617, "learning_rate": 6.751941747572817e-05, "loss": 0.5801, "step": 6691 }, { "epoch": 64.88674698795181, "grad_norm": 8.833699226379395, "learning_rate": 6.751456310679613e-05, "loss": 0.377, "step": 6692 }, { "epoch": 64.89638554216867, "grad_norm": 10.308846473693848, "learning_rate": 6.750970873786409e-05, "loss": 0.2109, "step": 6693 }, { "epoch": 64.90602409638554, "grad_norm": 23.20980453491211, "learning_rate": 6.750485436893204e-05, "loss": 0.4592, "step": 6694 }, { "epoch": 64.91566265060241, "grad_norm": 6.481101989746094, "learning_rate": 6.750000000000001e-05, "loss": 0.2559, "step": 6695 }, { "epoch": 64.92530120481928, "grad_norm": 7.852914810180664, "learning_rate": 6.749514563106797e-05, "loss": 0.1919, "step": 6696 }, { "epoch": 64.93493975903614, "grad_norm": 4.723352909088135, "learning_rate": 6.749029126213593e-05, "loss": 0.2339, "step": 6697 }, { "epoch": 64.94457831325302, "grad_norm": 10.93952751159668, "learning_rate": 6.748543689320389e-05, "loss": 0.0747, "step": 6698 }, { "epoch": 64.95421686746988, "grad_norm": 7.6336212158203125, "learning_rate": 6.748058252427185e-05, "loss": 0.1783, "step": 6699 }, { "epoch": 64.96385542168674, "grad_norm": 9.947287559509277, "learning_rate": 6.747572815533982e-05, "loss": 0.3005, "step": 6700 }, { "epoch": 64.97349397590361, "grad_norm": 19.318958282470703, "learning_rate": 6.747087378640777e-05, "loss": 0.3542, "step": 6701 }, { "epoch": 64.98313253012049, "grad_norm": 13.126338958740234, "learning_rate": 6.746601941747573e-05, "loss": 0.2464, "step": 6702 }, { "epoch": 64.99277108433735, "grad_norm": 20.439340591430664, "learning_rate": 6.746116504854369e-05, "loss": 0.5137, "step": 6703 }, { "epoch": 65.00843373493976, "grad_norm": 34.6386604309082, "learning_rate": 6.745631067961166e-05, "loss": 0.4831, "step": 6704 }, { "epoch": 65.01807228915662, "grad_norm": 1.7786166667938232, "learning_rate": 6.745145631067962e-05, "loss": 0.3429, "step": 6705 }, { "epoch": 65.0277108433735, "grad_norm": 3.265942096710205, "learning_rate": 6.744660194174758e-05, "loss": 0.2837, "step": 6706 }, { "epoch": 65.03734939759036, "grad_norm": 29.244482040405273, "learning_rate": 6.744174757281553e-05, "loss": 0.4278, "step": 6707 }, { "epoch": 65.04698795180722, "grad_norm": 32.6114387512207, "learning_rate": 6.74368932038835e-05, "loss": 0.2245, "step": 6708 }, { "epoch": 65.0566265060241, "grad_norm": 2.0974786281585693, "learning_rate": 6.743203883495146e-05, "loss": 0.2709, "step": 6709 }, { "epoch": 65.06626506024097, "grad_norm": 30.59119987487793, "learning_rate": 6.742718446601942e-05, "loss": 0.5478, "step": 6710 }, { "epoch": 65.07590361445783, "grad_norm": 13.907892227172852, "learning_rate": 6.742233009708738e-05, "loss": 0.2651, "step": 6711 }, { "epoch": 65.08554216867469, "grad_norm": 14.559274673461914, "learning_rate": 6.741747572815535e-05, "loss": 0.39, "step": 6712 }, { "epoch": 65.09518072289157, "grad_norm": 10.892230033874512, "learning_rate": 6.741262135922331e-05, "loss": 0.2421, "step": 6713 }, { "epoch": 65.10481927710843, "grad_norm": 40.69132995605469, "learning_rate": 6.740776699029127e-05, "loss": 0.2406, "step": 6714 }, { "epoch": 65.1144578313253, "grad_norm": 10.093097686767578, "learning_rate": 6.740291262135922e-05, "loss": 0.1987, "step": 6715 }, { "epoch": 65.12409638554217, "grad_norm": 81.79721069335938, "learning_rate": 6.739805825242718e-05, "loss": 0.2961, "step": 6716 }, { "epoch": 65.13373493975904, "grad_norm": 7.429294109344482, "learning_rate": 6.739320388349515e-05, "loss": 0.2901, "step": 6717 }, { "epoch": 65.1433734939759, "grad_norm": 9.073250770568848, "learning_rate": 6.738834951456311e-05, "loss": 0.22, "step": 6718 }, { "epoch": 65.15301204819278, "grad_norm": 16.22707176208496, "learning_rate": 6.738349514563107e-05, "loss": 0.3257, "step": 6719 }, { "epoch": 65.16265060240964, "grad_norm": 50.135040283203125, "learning_rate": 6.737864077669903e-05, "loss": 0.2457, "step": 6720 }, { "epoch": 65.1722891566265, "grad_norm": 20.371580123901367, "learning_rate": 6.7373786407767e-05, "loss": 0.492, "step": 6721 }, { "epoch": 65.18192771084337, "grad_norm": 7.4820966720581055, "learning_rate": 6.736893203883496e-05, "loss": 0.1112, "step": 6722 }, { "epoch": 65.19156626506025, "grad_norm": 4.952035903930664, "learning_rate": 6.736407766990291e-05, "loss": 0.2127, "step": 6723 }, { "epoch": 65.20120481927711, "grad_norm": 10.750466346740723, "learning_rate": 6.735922330097087e-05, "loss": 0.2721, "step": 6724 }, { "epoch": 65.21084337349397, "grad_norm": 13.067193031311035, "learning_rate": 6.735436893203884e-05, "loss": 0.2743, "step": 6725 }, { "epoch": 65.22048192771085, "grad_norm": 25.9171085357666, "learning_rate": 6.73495145631068e-05, "loss": 0.5795, "step": 6726 }, { "epoch": 65.23012048192771, "grad_norm": 27.778413772583008, "learning_rate": 6.734466019417476e-05, "loss": 0.3033, "step": 6727 }, { "epoch": 65.23975903614458, "grad_norm": 29.9765567779541, "learning_rate": 6.733980582524272e-05, "loss": 0.298, "step": 6728 }, { "epoch": 65.24939759036144, "grad_norm": 8.11925983428955, "learning_rate": 6.733495145631067e-05, "loss": 0.3656, "step": 6729 }, { "epoch": 65.25903614457832, "grad_norm": 14.246257781982422, "learning_rate": 6.733009708737864e-05, "loss": 0.3973, "step": 6730 }, { "epoch": 65.26867469879518, "grad_norm": 11.720973014831543, "learning_rate": 6.73252427184466e-05, "loss": 0.2208, "step": 6731 }, { "epoch": 65.27831325301204, "grad_norm": 48.22745895385742, "learning_rate": 6.732038834951456e-05, "loss": 0.4788, "step": 6732 }, { "epoch": 65.28795180722892, "grad_norm": 27.277536392211914, "learning_rate": 6.731553398058252e-05, "loss": 0.4556, "step": 6733 }, { "epoch": 65.29759036144578, "grad_norm": 14.486797332763672, "learning_rate": 6.731067961165049e-05, "loss": 0.3826, "step": 6734 }, { "epoch": 65.30722891566265, "grad_norm": 15.380736351013184, "learning_rate": 6.730582524271845e-05, "loss": 0.3541, "step": 6735 }, { "epoch": 65.31686746987951, "grad_norm": 7.36184024810791, "learning_rate": 6.73009708737864e-05, "loss": 0.2064, "step": 6736 }, { "epoch": 65.32650602409639, "grad_norm": 15.536354064941406, "learning_rate": 6.729611650485436e-05, "loss": 0.2336, "step": 6737 }, { "epoch": 65.33614457831325, "grad_norm": 2.431804656982422, "learning_rate": 6.729126213592233e-05, "loss": 0.3449, "step": 6738 }, { "epoch": 65.34578313253012, "grad_norm": 65.62654113769531, "learning_rate": 6.728640776699029e-05, "loss": 0.4447, "step": 6739 }, { "epoch": 65.355421686747, "grad_norm": 7.643687725067139, "learning_rate": 6.728155339805825e-05, "loss": 0.1588, "step": 6740 }, { "epoch": 65.36506024096386, "grad_norm": 6.860755920410156, "learning_rate": 6.727669902912622e-05, "loss": 0.5033, "step": 6741 }, { "epoch": 65.37469879518072, "grad_norm": 10.889143943786621, "learning_rate": 6.727184466019418e-05, "loss": 0.2425, "step": 6742 }, { "epoch": 65.38433734939758, "grad_norm": 37.06499481201172, "learning_rate": 6.726699029126215e-05, "loss": 0.2638, "step": 6743 }, { "epoch": 65.39397590361446, "grad_norm": 5.228896617889404, "learning_rate": 6.726213592233011e-05, "loss": 0.1184, "step": 6744 }, { "epoch": 65.40361445783132, "grad_norm": 8.17452621459961, "learning_rate": 6.725728155339806e-05, "loss": 0.2532, "step": 6745 }, { "epoch": 65.41325301204819, "grad_norm": 31.31730079650879, "learning_rate": 6.725242718446602e-05, "loss": 0.3033, "step": 6746 }, { "epoch": 65.42289156626506, "grad_norm": 21.358415603637695, "learning_rate": 6.7247572815534e-05, "loss": 0.2979, "step": 6747 }, { "epoch": 65.43253012048193, "grad_norm": 22.408750534057617, "learning_rate": 6.724271844660195e-05, "loss": 0.3567, "step": 6748 }, { "epoch": 65.44216867469879, "grad_norm": 26.376901626586914, "learning_rate": 6.723786407766991e-05, "loss": 0.4534, "step": 6749 }, { "epoch": 65.45180722891567, "grad_norm": 8.98001480102539, "learning_rate": 6.723300970873787e-05, "loss": 0.3241, "step": 6750 }, { "epoch": 65.46144578313253, "grad_norm": 1.9523696899414062, "learning_rate": 6.722815533980583e-05, "loss": 0.1396, "step": 6751 }, { "epoch": 65.4710843373494, "grad_norm": 13.782907485961914, "learning_rate": 6.72233009708738e-05, "loss": 0.4547, "step": 6752 }, { "epoch": 65.48072289156626, "grad_norm": 5.92196798324585, "learning_rate": 6.721844660194175e-05, "loss": 0.2314, "step": 6753 }, { "epoch": 65.49036144578314, "grad_norm": 18.25611686706543, "learning_rate": 6.721359223300971e-05, "loss": 0.2903, "step": 6754 }, { "epoch": 65.5, "grad_norm": 28.531898498535156, "learning_rate": 6.720873786407767e-05, "loss": 0.319, "step": 6755 }, { "epoch": 65.50963855421686, "grad_norm": 21.111448287963867, "learning_rate": 6.720388349514564e-05, "loss": 0.493, "step": 6756 }, { "epoch": 65.51927710843374, "grad_norm": 22.794017791748047, "learning_rate": 6.71990291262136e-05, "loss": 0.1625, "step": 6757 }, { "epoch": 65.5289156626506, "grad_norm": 7.1161723136901855, "learning_rate": 6.719417475728156e-05, "loss": 0.2508, "step": 6758 }, { "epoch": 65.53855421686747, "grad_norm": 6.881239891052246, "learning_rate": 6.718932038834951e-05, "loss": 0.3456, "step": 6759 }, { "epoch": 65.54819277108433, "grad_norm": 14.819785118103027, "learning_rate": 6.718446601941749e-05, "loss": 0.2637, "step": 6760 }, { "epoch": 65.55783132530121, "grad_norm": 34.53084182739258, "learning_rate": 6.717961165048544e-05, "loss": 0.3815, "step": 6761 }, { "epoch": 65.56746987951807, "grad_norm": 5.081968784332275, "learning_rate": 6.71747572815534e-05, "loss": 0.3213, "step": 6762 }, { "epoch": 65.57710843373494, "grad_norm": 8.368021965026855, "learning_rate": 6.716990291262136e-05, "loss": 0.3115, "step": 6763 }, { "epoch": 65.58674698795181, "grad_norm": 5.981809139251709, "learning_rate": 6.716504854368933e-05, "loss": 0.3471, "step": 6764 }, { "epoch": 65.59638554216868, "grad_norm": 8.531255722045898, "learning_rate": 6.716019417475729e-05, "loss": 0.5099, "step": 6765 }, { "epoch": 65.60602409638554, "grad_norm": 15.77514362335205, "learning_rate": 6.715533980582525e-05, "loss": 0.2766, "step": 6766 }, { "epoch": 65.61566265060242, "grad_norm": 13.29734992980957, "learning_rate": 6.71504854368932e-05, "loss": 0.369, "step": 6767 }, { "epoch": 65.62530120481928, "grad_norm": 1.1521291732788086, "learning_rate": 6.714563106796116e-05, "loss": 0.1278, "step": 6768 }, { "epoch": 65.63493975903614, "grad_norm": 6.187132358551025, "learning_rate": 6.714077669902913e-05, "loss": 0.3504, "step": 6769 }, { "epoch": 65.644578313253, "grad_norm": 1.7071126699447632, "learning_rate": 6.713592233009709e-05, "loss": 0.3064, "step": 6770 }, { "epoch": 65.65421686746988, "grad_norm": 3.8540291786193848, "learning_rate": 6.713106796116505e-05, "loss": 0.1761, "step": 6771 }, { "epoch": 65.66385542168675, "grad_norm": 41.79216384887695, "learning_rate": 6.7126213592233e-05, "loss": 0.2765, "step": 6772 }, { "epoch": 65.67349397590361, "grad_norm": 2.884713649749756, "learning_rate": 6.712135922330098e-05, "loss": 0.2942, "step": 6773 }, { "epoch": 65.68313253012049, "grad_norm": 10.390591621398926, "learning_rate": 6.711650485436893e-05, "loss": 0.2371, "step": 6774 }, { "epoch": 65.69277108433735, "grad_norm": 17.886817932128906, "learning_rate": 6.711165048543689e-05, "loss": 0.4048, "step": 6775 }, { "epoch": 65.70240963855422, "grad_norm": 20.59205436706543, "learning_rate": 6.710679611650485e-05, "loss": 0.3715, "step": 6776 }, { "epoch": 65.71204819277108, "grad_norm": 10.499261856079102, "learning_rate": 6.710194174757282e-05, "loss": 0.3365, "step": 6777 }, { "epoch": 65.72168674698796, "grad_norm": 14.705559730529785, "learning_rate": 6.709708737864078e-05, "loss": 0.2165, "step": 6778 }, { "epoch": 65.73132530120482, "grad_norm": 17.61156463623047, "learning_rate": 6.709223300970874e-05, "loss": 0.2821, "step": 6779 }, { "epoch": 65.74096385542168, "grad_norm": 17.764175415039062, "learning_rate": 6.70873786407767e-05, "loss": 0.315, "step": 6780 }, { "epoch": 65.75060240963856, "grad_norm": 5.623595237731934, "learning_rate": 6.708252427184467e-05, "loss": 0.1831, "step": 6781 }, { "epoch": 65.76024096385542, "grad_norm": 5.110619068145752, "learning_rate": 6.707766990291262e-05, "loss": 0.2844, "step": 6782 }, { "epoch": 65.76987951807229, "grad_norm": 9.655028343200684, "learning_rate": 6.707281553398058e-05, "loss": 0.3273, "step": 6783 }, { "epoch": 65.77951807228915, "grad_norm": 33.42417907714844, "learning_rate": 6.706796116504854e-05, "loss": 0.2327, "step": 6784 }, { "epoch": 65.78915662650603, "grad_norm": 32.30169677734375, "learning_rate": 6.70631067961165e-05, "loss": 0.3126, "step": 6785 }, { "epoch": 65.79879518072289, "grad_norm": 2.6114838123321533, "learning_rate": 6.705825242718447e-05, "loss": 0.1867, "step": 6786 }, { "epoch": 65.80843373493975, "grad_norm": 3.4021615982055664, "learning_rate": 6.705339805825243e-05, "loss": 0.273, "step": 6787 }, { "epoch": 65.81807228915663, "grad_norm": 5.410560607910156, "learning_rate": 6.704854368932038e-05, "loss": 0.1929, "step": 6788 }, { "epoch": 65.8277108433735, "grad_norm": 8.691788673400879, "learning_rate": 6.704368932038834e-05, "loss": 0.1009, "step": 6789 }, { "epoch": 65.83734939759036, "grad_norm": 5.847733974456787, "learning_rate": 6.703883495145631e-05, "loss": 0.1117, "step": 6790 }, { "epoch": 65.84698795180722, "grad_norm": 3.7002973556518555, "learning_rate": 6.703398058252428e-05, "loss": 0.3395, "step": 6791 }, { "epoch": 65.8566265060241, "grad_norm": 14.820626258850098, "learning_rate": 6.702912621359224e-05, "loss": 0.2399, "step": 6792 }, { "epoch": 65.86626506024096, "grad_norm": 4.034241199493408, "learning_rate": 6.70242718446602e-05, "loss": 0.3887, "step": 6793 }, { "epoch": 65.87590361445783, "grad_norm": 2.980968713760376, "learning_rate": 6.701941747572816e-05, "loss": 0.2438, "step": 6794 }, { "epoch": 65.8855421686747, "grad_norm": 11.394244194030762, "learning_rate": 6.701456310679613e-05, "loss": 0.3899, "step": 6795 }, { "epoch": 65.89518072289157, "grad_norm": 7.474178314208984, "learning_rate": 6.700970873786409e-05, "loss": 0.3125, "step": 6796 }, { "epoch": 65.90481927710843, "grad_norm": 5.150025844573975, "learning_rate": 6.700485436893204e-05, "loss": 0.4399, "step": 6797 }, { "epoch": 65.91445783132531, "grad_norm": 20.850555419921875, "learning_rate": 6.7e-05, "loss": 0.2189, "step": 6798 }, { "epoch": 65.92409638554217, "grad_norm": 48.65440368652344, "learning_rate": 6.699514563106797e-05, "loss": 0.6589, "step": 6799 }, { "epoch": 65.93373493975903, "grad_norm": 14.380315780639648, "learning_rate": 6.699029126213593e-05, "loss": 0.1611, "step": 6800 }, { "epoch": 65.9433734939759, "grad_norm": 23.71790313720703, "learning_rate": 6.698543689320389e-05, "loss": 0.2302, "step": 6801 }, { "epoch": 65.95301204819278, "grad_norm": 4.295592308044434, "learning_rate": 6.698058252427185e-05, "loss": 0.1582, "step": 6802 }, { "epoch": 65.96265060240964, "grad_norm": 4.244384765625, "learning_rate": 6.697572815533982e-05, "loss": 0.2418, "step": 6803 }, { "epoch": 65.9722891566265, "grad_norm": 12.098833084106445, "learning_rate": 6.697087378640778e-05, "loss": 0.3528, "step": 6804 }, { "epoch": 65.98192771084338, "grad_norm": 6.388802528381348, "learning_rate": 6.696601941747573e-05, "loss": 0.1481, "step": 6805 }, { "epoch": 65.99156626506024, "grad_norm": 6.298279285430908, "learning_rate": 6.696116504854369e-05, "loss": 0.2168, "step": 6806 }, { "epoch": 66.00722891566265, "grad_norm": 10.275309562683105, "learning_rate": 6.695631067961165e-05, "loss": 0.231, "step": 6807 }, { "epoch": 66.01686746987951, "grad_norm": 19.73981285095215, "learning_rate": 6.695145631067962e-05, "loss": 0.2787, "step": 6808 }, { "epoch": 66.02650602409639, "grad_norm": 47.467777252197266, "learning_rate": 6.694660194174758e-05, "loss": 0.1936, "step": 6809 }, { "epoch": 66.03614457831326, "grad_norm": 4.198078155517578, "learning_rate": 6.694174757281554e-05, "loss": 0.1911, "step": 6810 }, { "epoch": 66.04578313253012, "grad_norm": 15.561351776123047, "learning_rate": 6.69368932038835e-05, "loss": 0.4972, "step": 6811 }, { "epoch": 66.05542168674698, "grad_norm": 24.915813446044922, "learning_rate": 6.693203883495147e-05, "loss": 0.3589, "step": 6812 }, { "epoch": 66.06506024096386, "grad_norm": 7.011115550994873, "learning_rate": 6.692718446601942e-05, "loss": 0.3148, "step": 6813 }, { "epoch": 66.07469879518072, "grad_norm": 6.772200584411621, "learning_rate": 6.692233009708738e-05, "loss": 0.1499, "step": 6814 }, { "epoch": 66.08433734939759, "grad_norm": 4.220663547515869, "learning_rate": 6.691747572815534e-05, "loss": 0.1718, "step": 6815 }, { "epoch": 66.09397590361446, "grad_norm": 14.249181747436523, "learning_rate": 6.691262135922331e-05, "loss": 0.3149, "step": 6816 }, { "epoch": 66.10361445783133, "grad_norm": 9.50501537322998, "learning_rate": 6.690776699029127e-05, "loss": 0.1665, "step": 6817 }, { "epoch": 66.11325301204819, "grad_norm": 27.256900787353516, "learning_rate": 6.690291262135923e-05, "loss": 0.4023, "step": 6818 }, { "epoch": 66.12289156626505, "grad_norm": 17.672595977783203, "learning_rate": 6.689805825242718e-05, "loss": 0.3616, "step": 6819 }, { "epoch": 66.13253012048193, "grad_norm": 15.460383415222168, "learning_rate": 6.689320388349515e-05, "loss": 0.1843, "step": 6820 }, { "epoch": 66.1421686746988, "grad_norm": 14.843539237976074, "learning_rate": 6.688834951456311e-05, "loss": 0.3005, "step": 6821 }, { "epoch": 66.15180722891566, "grad_norm": 4.322976589202881, "learning_rate": 6.688349514563107e-05, "loss": 0.4463, "step": 6822 }, { "epoch": 66.16144578313254, "grad_norm": 11.614998817443848, "learning_rate": 6.687864077669903e-05, "loss": 0.3097, "step": 6823 }, { "epoch": 66.1710843373494, "grad_norm": 8.998124122619629, "learning_rate": 6.687378640776699e-05, "loss": 0.2869, "step": 6824 }, { "epoch": 66.18072289156626, "grad_norm": 7.685115814208984, "learning_rate": 6.686893203883496e-05, "loss": 0.4349, "step": 6825 }, { "epoch": 66.19036144578314, "grad_norm": 8.633055686950684, "learning_rate": 6.686407766990291e-05, "loss": 0.2315, "step": 6826 }, { "epoch": 66.2, "grad_norm": 5.919215679168701, "learning_rate": 6.685922330097087e-05, "loss": 0.1857, "step": 6827 }, { "epoch": 66.20963855421687, "grad_norm": 4.966214656829834, "learning_rate": 6.685436893203883e-05, "loss": 0.167, "step": 6828 }, { "epoch": 66.21927710843373, "grad_norm": 12.35540771484375, "learning_rate": 6.68495145631068e-05, "loss": 0.4759, "step": 6829 }, { "epoch": 66.2289156626506, "grad_norm": 9.928396224975586, "learning_rate": 6.684466019417476e-05, "loss": 0.2983, "step": 6830 }, { "epoch": 66.23855421686747, "grad_norm": 14.919381141662598, "learning_rate": 6.683980582524272e-05, "loss": 0.343, "step": 6831 }, { "epoch": 66.24819277108433, "grad_norm": 20.139665603637695, "learning_rate": 6.683495145631067e-05, "loss": 0.4327, "step": 6832 }, { "epoch": 66.25783132530121, "grad_norm": 29.315977096557617, "learning_rate": 6.683009708737865e-05, "loss": 0.2381, "step": 6833 }, { "epoch": 66.26746987951807, "grad_norm": 7.488012790679932, "learning_rate": 6.68252427184466e-05, "loss": 0.3346, "step": 6834 }, { "epoch": 66.27710843373494, "grad_norm": 4.024205684661865, "learning_rate": 6.682038834951456e-05, "loss": 0.2484, "step": 6835 }, { "epoch": 66.2867469879518, "grad_norm": 27.332263946533203, "learning_rate": 6.681553398058252e-05, "loss": 0.2185, "step": 6836 }, { "epoch": 66.29638554216868, "grad_norm": 4.176853179931641, "learning_rate": 6.681067961165049e-05, "loss": 0.5463, "step": 6837 }, { "epoch": 66.30602409638554, "grad_norm": 0.8683127760887146, "learning_rate": 6.680582524271845e-05, "loss": 0.3562, "step": 6838 }, { "epoch": 66.3156626506024, "grad_norm": 7.868536472320557, "learning_rate": 6.68009708737864e-05, "loss": 0.476, "step": 6839 }, { "epoch": 66.32530120481928, "grad_norm": 20.47072982788086, "learning_rate": 6.679611650485436e-05, "loss": 0.2721, "step": 6840 }, { "epoch": 66.33493975903615, "grad_norm": 7.247819423675537, "learning_rate": 6.679126213592232e-05, "loss": 0.3107, "step": 6841 }, { "epoch": 66.34457831325301, "grad_norm": 3.970270872116089, "learning_rate": 6.67864077669903e-05, "loss": 0.1831, "step": 6842 }, { "epoch": 66.35421686746987, "grad_norm": 7.544982433319092, "learning_rate": 6.678155339805826e-05, "loss": 0.2994, "step": 6843 }, { "epoch": 66.36385542168675, "grad_norm": 9.243553161621094, "learning_rate": 6.677669902912622e-05, "loss": 0.3438, "step": 6844 }, { "epoch": 66.37349397590361, "grad_norm": 20.522079467773438, "learning_rate": 6.677184466019418e-05, "loss": 0.2486, "step": 6845 }, { "epoch": 66.38313253012048, "grad_norm": 21.25244140625, "learning_rate": 6.676699029126214e-05, "loss": 0.2486, "step": 6846 }, { "epoch": 66.39277108433735, "grad_norm": 35.20885467529297, "learning_rate": 6.676213592233011e-05, "loss": 0.1806, "step": 6847 }, { "epoch": 66.40240963855422, "grad_norm": 77.81859588623047, "learning_rate": 6.675728155339807e-05, "loss": 0.6433, "step": 6848 }, { "epoch": 66.41204819277108, "grad_norm": 6.868830680847168, "learning_rate": 6.675242718446602e-05, "loss": 0.2211, "step": 6849 }, { "epoch": 66.42168674698796, "grad_norm": 17.35057830810547, "learning_rate": 6.674757281553398e-05, "loss": 0.4695, "step": 6850 }, { "epoch": 66.43132530120482, "grad_norm": 13.605826377868652, "learning_rate": 6.674271844660195e-05, "loss": 0.1446, "step": 6851 }, { "epoch": 66.44096385542169, "grad_norm": 39.8603630065918, "learning_rate": 6.673786407766991e-05, "loss": 0.3387, "step": 6852 }, { "epoch": 66.45060240963855, "grad_norm": 9.176685333251953, "learning_rate": 6.673300970873787e-05, "loss": 0.4238, "step": 6853 }, { "epoch": 66.46024096385543, "grad_norm": 26.13434600830078, "learning_rate": 6.672815533980583e-05, "loss": 0.4516, "step": 6854 }, { "epoch": 66.46987951807229, "grad_norm": 5.427332401275635, "learning_rate": 6.67233009708738e-05, "loss": 0.2057, "step": 6855 }, { "epoch": 66.47951807228915, "grad_norm": 10.139290809631348, "learning_rate": 6.671844660194176e-05, "loss": 0.3157, "step": 6856 }, { "epoch": 66.48915662650603, "grad_norm": 6.34350061416626, "learning_rate": 6.671359223300971e-05, "loss": 0.4388, "step": 6857 }, { "epoch": 66.4987951807229, "grad_norm": 39.61480712890625, "learning_rate": 6.670873786407767e-05, "loss": 0.3252, "step": 6858 }, { "epoch": 66.50843373493976, "grad_norm": 43.455665588378906, "learning_rate": 6.670388349514564e-05, "loss": 0.3394, "step": 6859 }, { "epoch": 66.51807228915662, "grad_norm": 22.59040069580078, "learning_rate": 6.66990291262136e-05, "loss": 0.2952, "step": 6860 }, { "epoch": 66.5277108433735, "grad_norm": 19.33199691772461, "learning_rate": 6.669417475728156e-05, "loss": 0.3657, "step": 6861 }, { "epoch": 66.53734939759036, "grad_norm": 1.5650545358657837, "learning_rate": 6.668932038834952e-05, "loss": 0.238, "step": 6862 }, { "epoch": 66.54698795180722, "grad_norm": 31.2493953704834, "learning_rate": 6.668446601941747e-05, "loss": 0.421, "step": 6863 }, { "epoch": 66.5566265060241, "grad_norm": 68.62315368652344, "learning_rate": 6.667961165048544e-05, "loss": 0.3423, "step": 6864 }, { "epoch": 66.56626506024097, "grad_norm": 16.051542282104492, "learning_rate": 6.66747572815534e-05, "loss": 0.2381, "step": 6865 }, { "epoch": 66.57590361445783, "grad_norm": 27.237756729125977, "learning_rate": 6.666990291262136e-05, "loss": 0.3707, "step": 6866 }, { "epoch": 66.58554216867469, "grad_norm": 16.8570613861084, "learning_rate": 6.666504854368932e-05, "loss": 0.3477, "step": 6867 }, { "epoch": 66.59518072289157, "grad_norm": 23.62989044189453, "learning_rate": 6.666019417475729e-05, "loss": 0.3991, "step": 6868 }, { "epoch": 66.60481927710843, "grad_norm": 0.9256627559661865, "learning_rate": 6.665533980582525e-05, "loss": 0.2644, "step": 6869 }, { "epoch": 66.6144578313253, "grad_norm": 15.670389175415039, "learning_rate": 6.66504854368932e-05, "loss": 0.3545, "step": 6870 }, { "epoch": 66.62409638554217, "grad_norm": 16.516809463500977, "learning_rate": 6.664563106796116e-05, "loss": 0.3599, "step": 6871 }, { "epoch": 66.63373493975904, "grad_norm": 10.025965690612793, "learning_rate": 6.664077669902913e-05, "loss": 0.3068, "step": 6872 }, { "epoch": 66.6433734939759, "grad_norm": 7.950473308563232, "learning_rate": 6.663592233009709e-05, "loss": 0.3032, "step": 6873 }, { "epoch": 66.65301204819278, "grad_norm": 9.55843734741211, "learning_rate": 6.663106796116505e-05, "loss": 0.3299, "step": 6874 }, { "epoch": 66.66265060240964, "grad_norm": 13.082752227783203, "learning_rate": 6.662621359223301e-05, "loss": 0.5145, "step": 6875 }, { "epoch": 66.6722891566265, "grad_norm": 13.250744819641113, "learning_rate": 6.662135922330098e-05, "loss": 0.369, "step": 6876 }, { "epoch": 66.68192771084337, "grad_norm": 14.482141494750977, "learning_rate": 6.661650485436894e-05, "loss": 0.236, "step": 6877 }, { "epoch": 66.69156626506025, "grad_norm": 4.092867851257324, "learning_rate": 6.66116504854369e-05, "loss": 0.3851, "step": 6878 }, { "epoch": 66.70120481927711, "grad_norm": 16.320545196533203, "learning_rate": 6.660679611650485e-05, "loss": 0.4071, "step": 6879 }, { "epoch": 66.71084337349397, "grad_norm": 5.359086036682129, "learning_rate": 6.660194174757281e-05, "loss": 0.2431, "step": 6880 }, { "epoch": 66.72048192771085, "grad_norm": 17.647945404052734, "learning_rate": 6.659708737864078e-05, "loss": 0.3516, "step": 6881 }, { "epoch": 66.73012048192771, "grad_norm": 20.399169921875, "learning_rate": 6.659223300970874e-05, "loss": 0.3136, "step": 6882 }, { "epoch": 66.73975903614458, "grad_norm": 6.042880535125732, "learning_rate": 6.65873786407767e-05, "loss": 0.1603, "step": 6883 }, { "epoch": 66.74939759036144, "grad_norm": 9.163426399230957, "learning_rate": 6.658252427184465e-05, "loss": 0.2416, "step": 6884 }, { "epoch": 66.75903614457832, "grad_norm": 32.64604187011719, "learning_rate": 6.657766990291263e-05, "loss": 0.2166, "step": 6885 }, { "epoch": 66.76867469879518, "grad_norm": 27.181529998779297, "learning_rate": 6.657281553398058e-05, "loss": 0.3909, "step": 6886 }, { "epoch": 66.77831325301204, "grad_norm": 9.097960472106934, "learning_rate": 6.656796116504854e-05, "loss": 0.2724, "step": 6887 }, { "epoch": 66.78795180722892, "grad_norm": 31.110193252563477, "learning_rate": 6.65631067961165e-05, "loss": 0.261, "step": 6888 }, { "epoch": 66.79759036144578, "grad_norm": 7.298301696777344, "learning_rate": 6.655825242718447e-05, "loss": 0.1996, "step": 6889 }, { "epoch": 66.80722891566265, "grad_norm": 16.969091415405273, "learning_rate": 6.655339805825243e-05, "loss": 0.3116, "step": 6890 }, { "epoch": 66.81686746987951, "grad_norm": 19.45387840270996, "learning_rate": 6.654854368932039e-05, "loss": 0.3346, "step": 6891 }, { "epoch": 66.82650602409639, "grad_norm": 10.382437705993652, "learning_rate": 6.654368932038836e-05, "loss": 0.2718, "step": 6892 }, { "epoch": 66.83614457831325, "grad_norm": 14.909063339233398, "learning_rate": 6.653883495145631e-05, "loss": 0.2642, "step": 6893 }, { "epoch": 66.84578313253012, "grad_norm": 4.442719459533691, "learning_rate": 6.653398058252429e-05, "loss": 0.3495, "step": 6894 }, { "epoch": 66.855421686747, "grad_norm": 23.52539825439453, "learning_rate": 6.652912621359224e-05, "loss": 0.433, "step": 6895 }, { "epoch": 66.86506024096386, "grad_norm": 4.58492374420166, "learning_rate": 6.65242718446602e-05, "loss": 0.39, "step": 6896 }, { "epoch": 66.87469879518072, "grad_norm": 5.760512828826904, "learning_rate": 6.651941747572816e-05, "loss": 0.281, "step": 6897 }, { "epoch": 66.88433734939758, "grad_norm": 22.907546997070312, "learning_rate": 6.651456310679613e-05, "loss": 0.2443, "step": 6898 }, { "epoch": 66.89397590361446, "grad_norm": 25.25837516784668, "learning_rate": 6.650970873786409e-05, "loss": 0.2667, "step": 6899 }, { "epoch": 66.90361445783132, "grad_norm": 16.23615837097168, "learning_rate": 6.650485436893205e-05, "loss": 0.328, "step": 6900 }, { "epoch": 66.91325301204819, "grad_norm": 6.353790283203125, "learning_rate": 6.65e-05, "loss": 0.3233, "step": 6901 }, { "epoch": 66.92289156626506, "grad_norm": 9.502952575683594, "learning_rate": 6.649514563106796e-05, "loss": 0.3574, "step": 6902 }, { "epoch": 66.93253012048193, "grad_norm": 12.621936798095703, "learning_rate": 6.649029126213593e-05, "loss": 0.1661, "step": 6903 }, { "epoch": 66.94216867469879, "grad_norm": 3.4734232425689697, "learning_rate": 6.648543689320389e-05, "loss": 0.2439, "step": 6904 }, { "epoch": 66.95180722891567, "grad_norm": 14.819450378417969, "learning_rate": 6.648058252427185e-05, "loss": 0.1943, "step": 6905 }, { "epoch": 66.96144578313253, "grad_norm": 5.62116003036499, "learning_rate": 6.64757281553398e-05, "loss": 0.3838, "step": 6906 }, { "epoch": 66.9710843373494, "grad_norm": 5.387557506561279, "learning_rate": 6.647087378640778e-05, "loss": 0.3241, "step": 6907 }, { "epoch": 66.98072289156626, "grad_norm": 27.78126335144043, "learning_rate": 6.646601941747574e-05, "loss": 0.4467, "step": 6908 }, { "epoch": 66.99036144578314, "grad_norm": 7.371760368347168, "learning_rate": 6.646116504854369e-05, "loss": 0.4043, "step": 6909 }, { "epoch": 67.00602409638554, "grad_norm": 2.0224149227142334, "learning_rate": 6.645631067961165e-05, "loss": 0.2516, "step": 6910 }, { "epoch": 67.01566265060241, "grad_norm": 2.9906623363494873, "learning_rate": 6.645145631067962e-05, "loss": 0.1263, "step": 6911 }, { "epoch": 67.02530120481927, "grad_norm": 12.836463928222656, "learning_rate": 6.644660194174758e-05, "loss": 0.2565, "step": 6912 }, { "epoch": 67.03493975903615, "grad_norm": 23.006961822509766, "learning_rate": 6.644174757281554e-05, "loss": 0.3249, "step": 6913 }, { "epoch": 67.04457831325301, "grad_norm": 9.004240989685059, "learning_rate": 6.64368932038835e-05, "loss": 0.2944, "step": 6914 }, { "epoch": 67.05421686746988, "grad_norm": 10.682459831237793, "learning_rate": 6.643203883495147e-05, "loss": 0.4059, "step": 6915 }, { "epoch": 67.06385542168675, "grad_norm": 7.20647668838501, "learning_rate": 6.642718446601942e-05, "loss": 0.3308, "step": 6916 }, { "epoch": 67.07349397590362, "grad_norm": 76.17681884765625, "learning_rate": 6.642233009708738e-05, "loss": 0.3259, "step": 6917 }, { "epoch": 67.08313253012048, "grad_norm": 14.72319507598877, "learning_rate": 6.641747572815534e-05, "loss": 0.3016, "step": 6918 }, { "epoch": 67.09277108433734, "grad_norm": 11.526976585388184, "learning_rate": 6.64126213592233e-05, "loss": 0.4823, "step": 6919 }, { "epoch": 67.10240963855422, "grad_norm": 14.73157024383545, "learning_rate": 6.640776699029127e-05, "loss": 0.489, "step": 6920 }, { "epoch": 67.11204819277108, "grad_norm": 2.2957651615142822, "learning_rate": 6.640291262135923e-05, "loss": 0.2275, "step": 6921 }, { "epoch": 67.12168674698795, "grad_norm": 5.350086688995361, "learning_rate": 6.639805825242718e-05, "loss": 0.3964, "step": 6922 }, { "epoch": 67.13132530120482, "grad_norm": 17.27556800842285, "learning_rate": 6.639320388349514e-05, "loss": 0.5064, "step": 6923 }, { "epoch": 67.14096385542169, "grad_norm": 20.186580657958984, "learning_rate": 6.638834951456311e-05, "loss": 0.4604, "step": 6924 }, { "epoch": 67.15060240963855, "grad_norm": 4.2861504554748535, "learning_rate": 6.638349514563107e-05, "loss": 0.2427, "step": 6925 }, { "epoch": 67.16024096385541, "grad_norm": 16.441457748413086, "learning_rate": 6.637864077669903e-05, "loss": 0.288, "step": 6926 }, { "epoch": 67.16987951807229, "grad_norm": 4.717936038970947, "learning_rate": 6.637378640776699e-05, "loss": 0.2959, "step": 6927 }, { "epoch": 67.17951807228916, "grad_norm": 6.64107084274292, "learning_rate": 6.636893203883496e-05, "loss": 0.2422, "step": 6928 }, { "epoch": 67.18915662650602, "grad_norm": 15.340292930603027, "learning_rate": 6.636407766990292e-05, "loss": 0.4308, "step": 6929 }, { "epoch": 67.1987951807229, "grad_norm": 24.458778381347656, "learning_rate": 6.635922330097087e-05, "loss": 0.2317, "step": 6930 }, { "epoch": 67.20843373493976, "grad_norm": 7.241652011871338, "learning_rate": 6.635436893203883e-05, "loss": 0.2048, "step": 6931 }, { "epoch": 67.21807228915662, "grad_norm": 11.004559516906738, "learning_rate": 6.63495145631068e-05, "loss": 0.2741, "step": 6932 }, { "epoch": 67.2277108433735, "grad_norm": 7.391747951507568, "learning_rate": 6.634466019417476e-05, "loss": 0.2938, "step": 6933 }, { "epoch": 67.23734939759036, "grad_norm": 24.637548446655273, "learning_rate": 6.633980582524272e-05, "loss": 0.4097, "step": 6934 }, { "epoch": 67.24698795180723, "grad_norm": 8.332649230957031, "learning_rate": 6.633495145631068e-05, "loss": 0.3105, "step": 6935 }, { "epoch": 67.25662650602409, "grad_norm": 4.05476188659668, "learning_rate": 6.633009708737863e-05, "loss": 0.2643, "step": 6936 }, { "epoch": 67.26626506024097, "grad_norm": 12.405923843383789, "learning_rate": 6.63252427184466e-05, "loss": 0.2271, "step": 6937 }, { "epoch": 67.27590361445783, "grad_norm": 22.410276412963867, "learning_rate": 6.632038834951456e-05, "loss": 0.372, "step": 6938 }, { "epoch": 67.2855421686747, "grad_norm": 5.12653112411499, "learning_rate": 6.631553398058252e-05, "loss": 0.2616, "step": 6939 }, { "epoch": 67.29518072289157, "grad_norm": 3.2737627029418945, "learning_rate": 6.631067961165048e-05, "loss": 0.1879, "step": 6940 }, { "epoch": 67.30481927710844, "grad_norm": 15.846311569213867, "learning_rate": 6.630582524271845e-05, "loss": 0.3105, "step": 6941 }, { "epoch": 67.3144578313253, "grad_norm": 24.27496910095215, "learning_rate": 6.630097087378641e-05, "loss": 0.2024, "step": 6942 }, { "epoch": 67.32409638554216, "grad_norm": 25.943466186523438, "learning_rate": 6.629611650485438e-05, "loss": 0.3925, "step": 6943 }, { "epoch": 67.33373493975904, "grad_norm": 27.482751846313477, "learning_rate": 6.629126213592234e-05, "loss": 0.4324, "step": 6944 }, { "epoch": 67.3433734939759, "grad_norm": 10.63064956665039, "learning_rate": 6.62864077669903e-05, "loss": 0.4094, "step": 6945 }, { "epoch": 67.35301204819277, "grad_norm": 19.025602340698242, "learning_rate": 6.628155339805827e-05, "loss": 0.2957, "step": 6946 }, { "epoch": 67.36265060240964, "grad_norm": 12.388166427612305, "learning_rate": 6.627669902912622e-05, "loss": 0.3227, "step": 6947 }, { "epoch": 67.37228915662651, "grad_norm": 3.596177816390991, "learning_rate": 6.627184466019418e-05, "loss": 0.2358, "step": 6948 }, { "epoch": 67.38192771084337, "grad_norm": 20.72718620300293, "learning_rate": 6.626699029126214e-05, "loss": 0.2477, "step": 6949 }, { "epoch": 67.39156626506023, "grad_norm": 22.32515525817871, "learning_rate": 6.626213592233011e-05, "loss": 0.2769, "step": 6950 }, { "epoch": 67.40120481927711, "grad_norm": 12.892574310302734, "learning_rate": 6.625728155339807e-05, "loss": 0.3525, "step": 6951 }, { "epoch": 67.41084337349398, "grad_norm": 1.842398762702942, "learning_rate": 6.625242718446603e-05, "loss": 0.1439, "step": 6952 }, { "epoch": 67.42048192771084, "grad_norm": 15.87617015838623, "learning_rate": 6.624757281553398e-05, "loss": 0.2274, "step": 6953 }, { "epoch": 67.43012048192772, "grad_norm": 22.072221755981445, "learning_rate": 6.624271844660195e-05, "loss": 0.3389, "step": 6954 }, { "epoch": 67.43975903614458, "grad_norm": 9.513219833374023, "learning_rate": 6.623786407766991e-05, "loss": 0.2096, "step": 6955 }, { "epoch": 67.44939759036144, "grad_norm": 19.55085563659668, "learning_rate": 6.623300970873787e-05, "loss": 0.4185, "step": 6956 }, { "epoch": 67.45903614457832, "grad_norm": 2.6015572547912598, "learning_rate": 6.622815533980583e-05, "loss": 0.2823, "step": 6957 }, { "epoch": 67.46867469879518, "grad_norm": 11.00049114227295, "learning_rate": 6.622330097087379e-05, "loss": 0.3998, "step": 6958 }, { "epoch": 67.47831325301205, "grad_norm": 10.328651428222656, "learning_rate": 6.621844660194176e-05, "loss": 0.262, "step": 6959 }, { "epoch": 67.48795180722891, "grad_norm": 9.593841552734375, "learning_rate": 6.621359223300971e-05, "loss": 0.1086, "step": 6960 }, { "epoch": 67.49759036144579, "grad_norm": 4.75465202331543, "learning_rate": 6.620873786407767e-05, "loss": 0.2696, "step": 6961 }, { "epoch": 67.50722891566265, "grad_norm": 6.704061508178711, "learning_rate": 6.620388349514563e-05, "loss": 0.1734, "step": 6962 }, { "epoch": 67.51686746987951, "grad_norm": 43.330970764160156, "learning_rate": 6.61990291262136e-05, "loss": 0.4194, "step": 6963 }, { "epoch": 67.52650602409639, "grad_norm": 8.849410057067871, "learning_rate": 6.619417475728156e-05, "loss": 0.1409, "step": 6964 }, { "epoch": 67.53614457831326, "grad_norm": 33.05470275878906, "learning_rate": 6.618932038834952e-05, "loss": 0.4812, "step": 6965 }, { "epoch": 67.54578313253012, "grad_norm": 10.912384986877441, "learning_rate": 6.618446601941747e-05, "loss": 0.5373, "step": 6966 }, { "epoch": 67.55542168674698, "grad_norm": 11.469688415527344, "learning_rate": 6.617961165048545e-05, "loss": 0.2698, "step": 6967 }, { "epoch": 67.56506024096386, "grad_norm": 8.762679100036621, "learning_rate": 6.61747572815534e-05, "loss": 0.3047, "step": 6968 }, { "epoch": 67.57469879518072, "grad_norm": 7.443022727966309, "learning_rate": 6.616990291262136e-05, "loss": 0.5266, "step": 6969 }, { "epoch": 67.58433734939759, "grad_norm": 5.080967903137207, "learning_rate": 6.616504854368932e-05, "loss": 0.2664, "step": 6970 }, { "epoch": 67.59397590361446, "grad_norm": 10.985661506652832, "learning_rate": 6.616019417475729e-05, "loss": 0.3834, "step": 6971 }, { "epoch": 67.60361445783133, "grad_norm": 4.479037761688232, "learning_rate": 6.615533980582525e-05, "loss": 0.2912, "step": 6972 }, { "epoch": 67.61325301204819, "grad_norm": 3.6992383003234863, "learning_rate": 6.61504854368932e-05, "loss": 0.2132, "step": 6973 }, { "epoch": 67.62289156626505, "grad_norm": 26.032169342041016, "learning_rate": 6.614563106796116e-05, "loss": 0.2217, "step": 6974 }, { "epoch": 67.63253012048193, "grad_norm": 31.26017189025879, "learning_rate": 6.614077669902912e-05, "loss": 0.615, "step": 6975 }, { "epoch": 67.6421686746988, "grad_norm": 20.11307144165039, "learning_rate": 6.613592233009709e-05, "loss": 0.3072, "step": 6976 }, { "epoch": 67.65180722891566, "grad_norm": 6.32933235168457, "learning_rate": 6.613106796116505e-05, "loss": 0.2927, "step": 6977 }, { "epoch": 67.66144578313254, "grad_norm": 8.185161590576172, "learning_rate": 6.612621359223301e-05, "loss": 0.3824, "step": 6978 }, { "epoch": 67.6710843373494, "grad_norm": 6.3541035652160645, "learning_rate": 6.612135922330097e-05, "loss": 0.3757, "step": 6979 }, { "epoch": 67.68072289156626, "grad_norm": 10.390789031982422, "learning_rate": 6.611650485436894e-05, "loss": 0.1077, "step": 6980 }, { "epoch": 67.69036144578314, "grad_norm": 9.786429405212402, "learning_rate": 6.61116504854369e-05, "loss": 0.3163, "step": 6981 }, { "epoch": 67.7, "grad_norm": 4.867819309234619, "learning_rate": 6.610679611650485e-05, "loss": 0.1954, "step": 6982 }, { "epoch": 67.70963855421687, "grad_norm": 33.35776901245117, "learning_rate": 6.610194174757281e-05, "loss": 0.4139, "step": 6983 }, { "epoch": 67.71927710843373, "grad_norm": 18.81363868713379, "learning_rate": 6.609708737864078e-05, "loss": 0.2338, "step": 6984 }, { "epoch": 67.7289156626506, "grad_norm": 13.136401176452637, "learning_rate": 6.609223300970874e-05, "loss": 0.2985, "step": 6985 }, { "epoch": 67.73855421686747, "grad_norm": 13.617165565490723, "learning_rate": 6.60873786407767e-05, "loss": 0.2848, "step": 6986 }, { "epoch": 67.74819277108433, "grad_norm": 24.46392822265625, "learning_rate": 6.608252427184466e-05, "loss": 0.2115, "step": 6987 }, { "epoch": 67.75783132530121, "grad_norm": 9.384690284729004, "learning_rate": 6.607766990291263e-05, "loss": 0.4379, "step": 6988 }, { "epoch": 67.76746987951807, "grad_norm": 6.95485258102417, "learning_rate": 6.607281553398058e-05, "loss": 0.2567, "step": 6989 }, { "epoch": 67.77710843373494, "grad_norm": 9.485891342163086, "learning_rate": 6.606796116504854e-05, "loss": 0.3934, "step": 6990 }, { "epoch": 67.7867469879518, "grad_norm": 18.373289108276367, "learning_rate": 6.60631067961165e-05, "loss": 0.2996, "step": 6991 }, { "epoch": 67.79638554216868, "grad_norm": 22.268898010253906, "learning_rate": 6.605825242718446e-05, "loss": 0.2525, "step": 6992 }, { "epoch": 67.80602409638554, "grad_norm": 16.658239364624023, "learning_rate": 6.605339805825244e-05, "loss": 0.2097, "step": 6993 }, { "epoch": 67.8156626506024, "grad_norm": 7.816825866699219, "learning_rate": 6.60485436893204e-05, "loss": 0.2339, "step": 6994 }, { "epoch": 67.82530120481928, "grad_norm": 27.972305297851562, "learning_rate": 6.604368932038836e-05, "loss": 0.1909, "step": 6995 }, { "epoch": 67.83493975903615, "grad_norm": 33.8155403137207, "learning_rate": 6.603883495145632e-05, "loss": 0.3047, "step": 6996 }, { "epoch": 67.84457831325301, "grad_norm": 2.5950565338134766, "learning_rate": 6.603398058252427e-05, "loss": 0.3436, "step": 6997 }, { "epoch": 67.85421686746987, "grad_norm": 4.109153747558594, "learning_rate": 6.602912621359224e-05, "loss": 0.3878, "step": 6998 }, { "epoch": 67.86385542168675, "grad_norm": 6.832097053527832, "learning_rate": 6.60242718446602e-05, "loss": 0.2956, "step": 6999 }, { "epoch": 67.87349397590361, "grad_norm": 9.255756378173828, "learning_rate": 6.601941747572816e-05, "loss": 0.23, "step": 7000 }, { "epoch": 67.88313253012048, "grad_norm": 7.841081619262695, "learning_rate": 6.601456310679612e-05, "loss": 0.3773, "step": 7001 }, { "epoch": 67.89277108433735, "grad_norm": 14.316874504089355, "learning_rate": 6.600970873786409e-05, "loss": 0.4345, "step": 7002 }, { "epoch": 67.90240963855422, "grad_norm": 9.451342582702637, "learning_rate": 6.600485436893205e-05, "loss": 0.5018, "step": 7003 }, { "epoch": 67.91204819277108, "grad_norm": 17.57292366027832, "learning_rate": 6.6e-05, "loss": 0.3531, "step": 7004 }, { "epoch": 67.92168674698796, "grad_norm": 3.090430498123169, "learning_rate": 6.599514563106796e-05, "loss": 0.2081, "step": 7005 }, { "epoch": 67.93132530120482, "grad_norm": 16.564340591430664, "learning_rate": 6.599029126213593e-05, "loss": 0.3568, "step": 7006 }, { "epoch": 67.94096385542169, "grad_norm": 5.842123031616211, "learning_rate": 6.598543689320389e-05, "loss": 0.2579, "step": 7007 }, { "epoch": 67.95060240963855, "grad_norm": 9.48395824432373, "learning_rate": 6.598058252427185e-05, "loss": 0.2801, "step": 7008 }, { "epoch": 67.96024096385543, "grad_norm": 28.02885627746582, "learning_rate": 6.597572815533981e-05, "loss": 0.3056, "step": 7009 }, { "epoch": 67.96987951807229, "grad_norm": 10.501373291015625, "learning_rate": 6.597087378640778e-05, "loss": 0.2528, "step": 7010 }, { "epoch": 67.97951807228915, "grad_norm": 8.950587272644043, "learning_rate": 6.596601941747574e-05, "loss": 0.2113, "step": 7011 }, { "epoch": 67.98915662650603, "grad_norm": 13.64896011352539, "learning_rate": 6.59611650485437e-05, "loss": 0.1992, "step": 7012 }, { "epoch": 68.00481927710844, "grad_norm": 3.045367956161499, "learning_rate": 6.595631067961165e-05, "loss": 0.1203, "step": 7013 }, { "epoch": 68.0144578313253, "grad_norm": 11.943473815917969, "learning_rate": 6.595145631067961e-05, "loss": 0.1273, "step": 7014 }, { "epoch": 68.02409638554217, "grad_norm": 9.439373016357422, "learning_rate": 6.594660194174758e-05, "loss": 0.1961, "step": 7015 }, { "epoch": 68.03373493975904, "grad_norm": 19.08601951599121, "learning_rate": 6.594174757281554e-05, "loss": 0.2016, "step": 7016 }, { "epoch": 68.0433734939759, "grad_norm": 5.262884616851807, "learning_rate": 6.59368932038835e-05, "loss": 0.1174, "step": 7017 }, { "epoch": 68.05301204819277, "grad_norm": 6.69572114944458, "learning_rate": 6.593203883495145e-05, "loss": 0.3097, "step": 7018 }, { "epoch": 68.06265060240963, "grad_norm": 7.155301094055176, "learning_rate": 6.592718446601943e-05, "loss": 0.2841, "step": 7019 }, { "epoch": 68.07228915662651, "grad_norm": 5.708433628082275, "learning_rate": 6.592233009708738e-05, "loss": 0.1677, "step": 7020 }, { "epoch": 68.08192771084337, "grad_norm": 16.59807777404785, "learning_rate": 6.591747572815534e-05, "loss": 0.4096, "step": 7021 }, { "epoch": 68.09156626506024, "grad_norm": 4.650554180145264, "learning_rate": 6.59126213592233e-05, "loss": 0.2552, "step": 7022 }, { "epoch": 68.10120481927711, "grad_norm": 18.19145965576172, "learning_rate": 6.590776699029127e-05, "loss": 0.2839, "step": 7023 }, { "epoch": 68.11084337349398, "grad_norm": 10.207982063293457, "learning_rate": 6.590291262135923e-05, "loss": 0.1965, "step": 7024 }, { "epoch": 68.12048192771084, "grad_norm": 4.428188323974609, "learning_rate": 6.589805825242719e-05, "loss": 0.1996, "step": 7025 }, { "epoch": 68.1301204819277, "grad_norm": 8.128618240356445, "learning_rate": 6.589320388349514e-05, "loss": 0.3234, "step": 7026 }, { "epoch": 68.13975903614458, "grad_norm": 7.907955646514893, "learning_rate": 6.58883495145631e-05, "loss": 0.2863, "step": 7027 }, { "epoch": 68.14939759036145, "grad_norm": 3.6006295680999756, "learning_rate": 6.588349514563107e-05, "loss": 0.2143, "step": 7028 }, { "epoch": 68.15903614457831, "grad_norm": 11.591547012329102, "learning_rate": 6.587864077669903e-05, "loss": 0.4322, "step": 7029 }, { "epoch": 68.16867469879519, "grad_norm": 7.1129560470581055, "learning_rate": 6.587378640776699e-05, "loss": 0.4785, "step": 7030 }, { "epoch": 68.17831325301205, "grad_norm": 7.35779333114624, "learning_rate": 6.586893203883495e-05, "loss": 0.5185, "step": 7031 }, { "epoch": 68.18795180722891, "grad_norm": 4.335166931152344, "learning_rate": 6.586407766990292e-05, "loss": 0.2609, "step": 7032 }, { "epoch": 68.19759036144578, "grad_norm": 10.56141471862793, "learning_rate": 6.585922330097087e-05, "loss": 0.3105, "step": 7033 }, { "epoch": 68.20722891566265, "grad_norm": 6.786253929138184, "learning_rate": 6.585436893203883e-05, "loss": 0.272, "step": 7034 }, { "epoch": 68.21686746987952, "grad_norm": 24.865663528442383, "learning_rate": 6.584951456310679e-05, "loss": 0.2864, "step": 7035 }, { "epoch": 68.22650602409638, "grad_norm": 6.350722312927246, "learning_rate": 6.584466019417476e-05, "loss": 0.299, "step": 7036 }, { "epoch": 68.23614457831326, "grad_norm": 25.330617904663086, "learning_rate": 6.583980582524272e-05, "loss": 0.2073, "step": 7037 }, { "epoch": 68.24578313253012, "grad_norm": 5.666957855224609, "learning_rate": 6.583495145631068e-05, "loss": 0.199, "step": 7038 }, { "epoch": 68.25542168674698, "grad_norm": 3.745028018951416, "learning_rate": 6.583009708737863e-05, "loss": 0.115, "step": 7039 }, { "epoch": 68.26506024096386, "grad_norm": 5.930240154266357, "learning_rate": 6.58252427184466e-05, "loss": 0.3227, "step": 7040 }, { "epoch": 68.27469879518073, "grad_norm": 29.205196380615234, "learning_rate": 6.582038834951456e-05, "loss": 0.2585, "step": 7041 }, { "epoch": 68.28433734939759, "grad_norm": 3.4721412658691406, "learning_rate": 6.581553398058252e-05, "loss": 0.4129, "step": 7042 }, { "epoch": 68.29397590361445, "grad_norm": 9.516021728515625, "learning_rate": 6.581067961165048e-05, "loss": 0.3432, "step": 7043 }, { "epoch": 68.30361445783133, "grad_norm": 12.691375732421875, "learning_rate": 6.580582524271845e-05, "loss": 0.3893, "step": 7044 }, { "epoch": 68.3132530120482, "grad_norm": 12.163009643554688, "learning_rate": 6.580097087378642e-05, "loss": 0.2893, "step": 7045 }, { "epoch": 68.32289156626506, "grad_norm": 18.936534881591797, "learning_rate": 6.579611650485438e-05, "loss": 0.3924, "step": 7046 }, { "epoch": 68.33253012048193, "grad_norm": 15.468582153320312, "learning_rate": 6.579126213592234e-05, "loss": 0.2879, "step": 7047 }, { "epoch": 68.3421686746988, "grad_norm": 5.646069049835205, "learning_rate": 6.57864077669903e-05, "loss": 0.2673, "step": 7048 }, { "epoch": 68.35180722891566, "grad_norm": 9.90992259979248, "learning_rate": 6.578155339805825e-05, "loss": 0.3043, "step": 7049 }, { "epoch": 68.36144578313252, "grad_norm": 22.045421600341797, "learning_rate": 6.577669902912622e-05, "loss": 0.2428, "step": 7050 }, { "epoch": 68.3710843373494, "grad_norm": 7.470091819763184, "learning_rate": 6.577184466019418e-05, "loss": 0.3902, "step": 7051 }, { "epoch": 68.38072289156626, "grad_norm": 3.852588415145874, "learning_rate": 6.576699029126214e-05, "loss": 0.1755, "step": 7052 }, { "epoch": 68.39036144578313, "grad_norm": 30.413433074951172, "learning_rate": 6.57621359223301e-05, "loss": 0.2675, "step": 7053 }, { "epoch": 68.4, "grad_norm": 2.5544660091400146, "learning_rate": 6.575728155339807e-05, "loss": 0.258, "step": 7054 }, { "epoch": 68.40963855421687, "grad_norm": 9.119236946105957, "learning_rate": 6.575242718446603e-05, "loss": 0.2398, "step": 7055 }, { "epoch": 68.41927710843373, "grad_norm": 10.091704368591309, "learning_rate": 6.574757281553398e-05, "loss": 0.4534, "step": 7056 }, { "epoch": 68.4289156626506, "grad_norm": 8.3302640914917, "learning_rate": 6.574271844660194e-05, "loss": 0.1527, "step": 7057 }, { "epoch": 68.43855421686747, "grad_norm": 14.33139419555664, "learning_rate": 6.573786407766991e-05, "loss": 0.1816, "step": 7058 }, { "epoch": 68.44819277108434, "grad_norm": 9.688758850097656, "learning_rate": 6.573300970873787e-05, "loss": 0.3958, "step": 7059 }, { "epoch": 68.4578313253012, "grad_norm": 8.322739601135254, "learning_rate": 6.572815533980583e-05, "loss": 0.2041, "step": 7060 }, { "epoch": 68.46746987951808, "grad_norm": 10.896190643310547, "learning_rate": 6.572330097087379e-05, "loss": 0.2362, "step": 7061 }, { "epoch": 68.47710843373494, "grad_norm": 19.712162017822266, "learning_rate": 6.571844660194176e-05, "loss": 0.1518, "step": 7062 }, { "epoch": 68.4867469879518, "grad_norm": 15.322548866271973, "learning_rate": 6.571359223300972e-05, "loss": 0.3088, "step": 7063 }, { "epoch": 68.49638554216868, "grad_norm": 8.353448867797852, "learning_rate": 6.570873786407767e-05, "loss": 0.2284, "step": 7064 }, { "epoch": 68.50602409638554, "grad_norm": 24.15607261657715, "learning_rate": 6.570388349514563e-05, "loss": 0.3746, "step": 7065 }, { "epoch": 68.51566265060241, "grad_norm": 18.94684410095215, "learning_rate": 6.569902912621359e-05, "loss": 0.454, "step": 7066 }, { "epoch": 68.52530120481927, "grad_norm": 11.644722938537598, "learning_rate": 6.569417475728156e-05, "loss": 0.1901, "step": 7067 }, { "epoch": 68.53493975903615, "grad_norm": 11.079106330871582, "learning_rate": 6.568932038834952e-05, "loss": 0.2641, "step": 7068 }, { "epoch": 68.54457831325301, "grad_norm": 9.247814178466797, "learning_rate": 6.568446601941748e-05, "loss": 0.3955, "step": 7069 }, { "epoch": 68.55421686746988, "grad_norm": 18.46755599975586, "learning_rate": 6.567961165048543e-05, "loss": 0.2554, "step": 7070 }, { "epoch": 68.56385542168675, "grad_norm": 12.58596420288086, "learning_rate": 6.56747572815534e-05, "loss": 0.3868, "step": 7071 }, { "epoch": 68.57349397590362, "grad_norm": 19.43853187561035, "learning_rate": 6.566990291262136e-05, "loss": 0.2416, "step": 7072 }, { "epoch": 68.58313253012048, "grad_norm": 6.1478047370910645, "learning_rate": 6.566504854368932e-05, "loss": 0.1426, "step": 7073 }, { "epoch": 68.59277108433734, "grad_norm": 10.442956924438477, "learning_rate": 6.566019417475728e-05, "loss": 0.3916, "step": 7074 }, { "epoch": 68.60240963855422, "grad_norm": 3.930123805999756, "learning_rate": 6.565533980582525e-05, "loss": 0.1643, "step": 7075 }, { "epoch": 68.61204819277108, "grad_norm": 22.96587371826172, "learning_rate": 6.565048543689321e-05, "loss": 0.3408, "step": 7076 }, { "epoch": 68.62168674698795, "grad_norm": 15.246909141540527, "learning_rate": 6.564563106796117e-05, "loss": 0.2897, "step": 7077 }, { "epoch": 68.63132530120482, "grad_norm": 7.154242038726807, "learning_rate": 6.564077669902912e-05, "loss": 0.4245, "step": 7078 }, { "epoch": 68.64096385542169, "grad_norm": 26.592910766601562, "learning_rate": 6.56359223300971e-05, "loss": 0.304, "step": 7079 }, { "epoch": 68.65060240963855, "grad_norm": 25.273422241210938, "learning_rate": 6.563106796116505e-05, "loss": 0.3714, "step": 7080 }, { "epoch": 68.66024096385541, "grad_norm": 30.248945236206055, "learning_rate": 6.562621359223301e-05, "loss": 0.3922, "step": 7081 }, { "epoch": 68.66987951807229, "grad_norm": 27.663265228271484, "learning_rate": 6.562135922330097e-05, "loss": 0.4516, "step": 7082 }, { "epoch": 68.67951807228916, "grad_norm": 7.751377105712891, "learning_rate": 6.561650485436893e-05, "loss": 0.2367, "step": 7083 }, { "epoch": 68.68915662650602, "grad_norm": 6.612028121948242, "learning_rate": 6.56116504854369e-05, "loss": 0.3117, "step": 7084 }, { "epoch": 68.6987951807229, "grad_norm": 5.454747676849365, "learning_rate": 6.560679611650485e-05, "loss": 0.4302, "step": 7085 }, { "epoch": 68.70843373493976, "grad_norm": 14.928614616394043, "learning_rate": 6.560194174757281e-05, "loss": 0.4118, "step": 7086 }, { "epoch": 68.71807228915662, "grad_norm": 10.80131721496582, "learning_rate": 6.559708737864077e-05, "loss": 0.3488, "step": 7087 }, { "epoch": 68.7277108433735, "grad_norm": 15.705007553100586, "learning_rate": 6.559223300970874e-05, "loss": 0.2135, "step": 7088 }, { "epoch": 68.73734939759036, "grad_norm": 2.917572498321533, "learning_rate": 6.55873786407767e-05, "loss": 0.1905, "step": 7089 }, { "epoch": 68.74698795180723, "grad_norm": 18.517377853393555, "learning_rate": 6.558252427184466e-05, "loss": 0.3934, "step": 7090 }, { "epoch": 68.75662650602409, "grad_norm": 8.08890438079834, "learning_rate": 6.557766990291261e-05, "loss": 0.1631, "step": 7091 }, { "epoch": 68.76626506024097, "grad_norm": 1.8895989656448364, "learning_rate": 6.557281553398059e-05, "loss": 0.1462, "step": 7092 }, { "epoch": 68.77590361445783, "grad_norm": 11.438352584838867, "learning_rate": 6.556796116504854e-05, "loss": 0.3636, "step": 7093 }, { "epoch": 68.7855421686747, "grad_norm": 37.686527252197266, "learning_rate": 6.556310679611651e-05, "loss": 0.3205, "step": 7094 }, { "epoch": 68.79518072289157, "grad_norm": 11.26423454284668, "learning_rate": 6.555825242718447e-05, "loss": 0.3696, "step": 7095 }, { "epoch": 68.80481927710844, "grad_norm": 6.737812042236328, "learning_rate": 6.555339805825243e-05, "loss": 0.3205, "step": 7096 }, { "epoch": 68.8144578313253, "grad_norm": 66.14535522460938, "learning_rate": 6.55485436893204e-05, "loss": 0.2699, "step": 7097 }, { "epoch": 68.82409638554216, "grad_norm": 9.942907333374023, "learning_rate": 6.554368932038836e-05, "loss": 0.2158, "step": 7098 }, { "epoch": 68.83373493975904, "grad_norm": 7.305004119873047, "learning_rate": 6.553883495145632e-05, "loss": 0.3077, "step": 7099 }, { "epoch": 68.8433734939759, "grad_norm": 17.117849349975586, "learning_rate": 6.553398058252428e-05, "loss": 0.1545, "step": 7100 }, { "epoch": 68.85301204819277, "grad_norm": 11.14328670501709, "learning_rate": 6.552912621359225e-05, "loss": 0.3212, "step": 7101 }, { "epoch": 68.86265060240964, "grad_norm": 23.451400756835938, "learning_rate": 6.55242718446602e-05, "loss": 0.3016, "step": 7102 }, { "epoch": 68.87228915662651, "grad_norm": 16.366710662841797, "learning_rate": 6.551941747572816e-05, "loss": 0.476, "step": 7103 }, { "epoch": 68.88192771084337, "grad_norm": 6.627814292907715, "learning_rate": 6.551456310679612e-05, "loss": 0.08, "step": 7104 }, { "epoch": 68.89156626506023, "grad_norm": 10.154376983642578, "learning_rate": 6.550970873786408e-05, "loss": 0.3014, "step": 7105 }, { "epoch": 68.90120481927711, "grad_norm": 8.263618469238281, "learning_rate": 6.550485436893205e-05, "loss": 0.2703, "step": 7106 }, { "epoch": 68.91084337349398, "grad_norm": 10.53853988647461, "learning_rate": 6.55e-05, "loss": 0.5888, "step": 7107 }, { "epoch": 68.92048192771084, "grad_norm": 10.755415916442871, "learning_rate": 6.549514563106796e-05, "loss": 0.2962, "step": 7108 }, { "epoch": 68.93012048192772, "grad_norm": 9.18686294555664, "learning_rate": 6.549029126213592e-05, "loss": 0.1849, "step": 7109 }, { "epoch": 68.93975903614458, "grad_norm": 30.756105422973633, "learning_rate": 6.54854368932039e-05, "loss": 0.5639, "step": 7110 }, { "epoch": 68.94939759036144, "grad_norm": 6.271714687347412, "learning_rate": 6.548058252427185e-05, "loss": 0.2513, "step": 7111 }, { "epoch": 68.95903614457832, "grad_norm": 25.439945220947266, "learning_rate": 6.547572815533981e-05, "loss": 0.3924, "step": 7112 }, { "epoch": 68.96867469879518, "grad_norm": 15.752678871154785, "learning_rate": 6.547087378640777e-05, "loss": 0.4221, "step": 7113 }, { "epoch": 68.97831325301205, "grad_norm": 14.244562149047852, "learning_rate": 6.546601941747574e-05, "loss": 0.3819, "step": 7114 }, { "epoch": 68.98795180722891, "grad_norm": 4.422541618347168, "learning_rate": 6.54611650485437e-05, "loss": 0.2247, "step": 7115 }, { "epoch": 69.00361445783132, "grad_norm": 10.642311096191406, "learning_rate": 6.545631067961165e-05, "loss": 0.5087, "step": 7116 }, { "epoch": 69.0132530120482, "grad_norm": 24.499814987182617, "learning_rate": 6.545145631067961e-05, "loss": 0.4322, "step": 7117 }, { "epoch": 69.02289156626506, "grad_norm": 24.839113235473633, "learning_rate": 6.544660194174758e-05, "loss": 0.5177, "step": 7118 }, { "epoch": 69.03253012048192, "grad_norm": 15.045598983764648, "learning_rate": 6.544174757281554e-05, "loss": 0.2919, "step": 7119 }, { "epoch": 69.0421686746988, "grad_norm": 17.752588272094727, "learning_rate": 6.54368932038835e-05, "loss": 0.2245, "step": 7120 }, { "epoch": 69.05180722891566, "grad_norm": 22.09603500366211, "learning_rate": 6.543203883495146e-05, "loss": 0.3453, "step": 7121 }, { "epoch": 69.06144578313253, "grad_norm": 48.225521087646484, "learning_rate": 6.542718446601941e-05, "loss": 0.5551, "step": 7122 }, { "epoch": 69.0710843373494, "grad_norm": 20.653337478637695, "learning_rate": 6.542233009708738e-05, "loss": 0.3268, "step": 7123 }, { "epoch": 69.08072289156627, "grad_norm": 8.759187698364258, "learning_rate": 6.541747572815534e-05, "loss": 0.3119, "step": 7124 }, { "epoch": 69.09036144578313, "grad_norm": 9.116815567016602, "learning_rate": 6.54126213592233e-05, "loss": 0.2813, "step": 7125 }, { "epoch": 69.1, "grad_norm": 10.715038299560547, "learning_rate": 6.540776699029126e-05, "loss": 0.2171, "step": 7126 }, { "epoch": 69.10963855421687, "grad_norm": 46.82402801513672, "learning_rate": 6.540291262135923e-05, "loss": 0.3459, "step": 7127 }, { "epoch": 69.11927710843374, "grad_norm": 10.274482727050781, "learning_rate": 6.539805825242719e-05, "loss": 0.2251, "step": 7128 }, { "epoch": 69.1289156626506, "grad_norm": 13.459163665771484, "learning_rate": 6.539320388349514e-05, "loss": 0.1736, "step": 7129 }, { "epoch": 69.13855421686748, "grad_norm": 3.9024651050567627, "learning_rate": 6.53883495145631e-05, "loss": 0.112, "step": 7130 }, { "epoch": 69.14819277108434, "grad_norm": 4.838319778442383, "learning_rate": 6.538349514563107e-05, "loss": 0.3752, "step": 7131 }, { "epoch": 69.1578313253012, "grad_norm": 10.849996566772461, "learning_rate": 6.537864077669903e-05, "loss": 0.3889, "step": 7132 }, { "epoch": 69.16746987951807, "grad_norm": 13.432497024536133, "learning_rate": 6.537378640776699e-05, "loss": 0.3448, "step": 7133 }, { "epoch": 69.17710843373494, "grad_norm": 17.067712783813477, "learning_rate": 6.536893203883495e-05, "loss": 0.2059, "step": 7134 }, { "epoch": 69.1867469879518, "grad_norm": 49.70146179199219, "learning_rate": 6.536407766990292e-05, "loss": 0.194, "step": 7135 }, { "epoch": 69.19638554216867, "grad_norm": 9.59664535522461, "learning_rate": 6.535922330097088e-05, "loss": 0.3227, "step": 7136 }, { "epoch": 69.20602409638555, "grad_norm": 6.347351551055908, "learning_rate": 6.535436893203883e-05, "loss": 0.4357, "step": 7137 }, { "epoch": 69.21566265060241, "grad_norm": 5.233338832855225, "learning_rate": 6.534951456310679e-05, "loss": 0.2099, "step": 7138 }, { "epoch": 69.22530120481927, "grad_norm": 8.584495544433594, "learning_rate": 6.534466019417475e-05, "loss": 0.4027, "step": 7139 }, { "epoch": 69.23493975903614, "grad_norm": 6.266733169555664, "learning_rate": 6.533980582524272e-05, "loss": 0.3397, "step": 7140 }, { "epoch": 69.24457831325302, "grad_norm": 3.9011571407318115, "learning_rate": 6.533495145631068e-05, "loss": 0.2342, "step": 7141 }, { "epoch": 69.25421686746988, "grad_norm": 23.987579345703125, "learning_rate": 6.533009708737864e-05, "loss": 0.5645, "step": 7142 }, { "epoch": 69.26385542168674, "grad_norm": 44.89698791503906, "learning_rate": 6.53252427184466e-05, "loss": 0.3856, "step": 7143 }, { "epoch": 69.27349397590362, "grad_norm": 14.284536361694336, "learning_rate": 6.532038834951457e-05, "loss": 0.3125, "step": 7144 }, { "epoch": 69.28313253012048, "grad_norm": 45.834476470947266, "learning_rate": 6.531553398058254e-05, "loss": 0.3724, "step": 7145 }, { "epoch": 69.29277108433735, "grad_norm": 7.305481910705566, "learning_rate": 6.53106796116505e-05, "loss": 0.3365, "step": 7146 }, { "epoch": 69.30240963855422, "grad_norm": 7.308683395385742, "learning_rate": 6.530582524271845e-05, "loss": 0.3178, "step": 7147 }, { "epoch": 69.31204819277109, "grad_norm": 20.22273826599121, "learning_rate": 6.530097087378641e-05, "loss": 0.2506, "step": 7148 }, { "epoch": 69.32168674698795, "grad_norm": 11.130992889404297, "learning_rate": 6.529611650485438e-05, "loss": 0.3111, "step": 7149 }, { "epoch": 69.33132530120481, "grad_norm": 17.61437225341797, "learning_rate": 6.529126213592234e-05, "loss": 0.5604, "step": 7150 }, { "epoch": 69.34096385542169, "grad_norm": 6.039733409881592, "learning_rate": 6.52864077669903e-05, "loss": 0.1999, "step": 7151 }, { "epoch": 69.35060240963855, "grad_norm": 7.305150985717773, "learning_rate": 6.528155339805825e-05, "loss": 0.315, "step": 7152 }, { "epoch": 69.36024096385542, "grad_norm": 9.3922758102417, "learning_rate": 6.527669902912623e-05, "loss": 0.2595, "step": 7153 }, { "epoch": 69.3698795180723, "grad_norm": 6.7861223220825195, "learning_rate": 6.527184466019418e-05, "loss": 0.4689, "step": 7154 }, { "epoch": 69.37951807228916, "grad_norm": 10.756797790527344, "learning_rate": 6.526699029126214e-05, "loss": 0.2321, "step": 7155 }, { "epoch": 69.38915662650602, "grad_norm": 6.296529293060303, "learning_rate": 6.52621359223301e-05, "loss": 0.3326, "step": 7156 }, { "epoch": 69.39879518072289, "grad_norm": 7.473015308380127, "learning_rate": 6.525728155339807e-05, "loss": 0.3076, "step": 7157 }, { "epoch": 69.40843373493976, "grad_norm": 5.612462520599365, "learning_rate": 6.525242718446603e-05, "loss": 0.1765, "step": 7158 }, { "epoch": 69.41807228915663, "grad_norm": 9.487133026123047, "learning_rate": 6.524757281553399e-05, "loss": 0.2965, "step": 7159 }, { "epoch": 69.42771084337349, "grad_norm": 9.686266899108887, "learning_rate": 6.524271844660194e-05, "loss": 0.4236, "step": 7160 }, { "epoch": 69.43734939759037, "grad_norm": 45.42413330078125, "learning_rate": 6.52378640776699e-05, "loss": 0.2248, "step": 7161 }, { "epoch": 69.44698795180723, "grad_norm": 22.058195114135742, "learning_rate": 6.523300970873787e-05, "loss": 0.3015, "step": 7162 }, { "epoch": 69.4566265060241, "grad_norm": 7.606264591217041, "learning_rate": 6.522815533980583e-05, "loss": 0.4537, "step": 7163 }, { "epoch": 69.46626506024096, "grad_norm": 31.056781768798828, "learning_rate": 6.522330097087379e-05, "loss": 0.2692, "step": 7164 }, { "epoch": 69.47590361445783, "grad_norm": 18.740591049194336, "learning_rate": 6.521844660194175e-05, "loss": 0.4417, "step": 7165 }, { "epoch": 69.4855421686747, "grad_norm": 17.38400650024414, "learning_rate": 6.521359223300972e-05, "loss": 0.4385, "step": 7166 }, { "epoch": 69.49518072289156, "grad_norm": 8.47407341003418, "learning_rate": 6.520873786407768e-05, "loss": 0.3123, "step": 7167 }, { "epoch": 69.50481927710844, "grad_norm": 14.791160583496094, "learning_rate": 6.520388349514563e-05, "loss": 0.3822, "step": 7168 }, { "epoch": 69.5144578313253, "grad_norm": 22.134199142456055, "learning_rate": 6.519902912621359e-05, "loss": 0.4514, "step": 7169 }, { "epoch": 69.52409638554217, "grad_norm": 5.925281524658203, "learning_rate": 6.519417475728156e-05, "loss": 0.2965, "step": 7170 }, { "epoch": 69.53373493975904, "grad_norm": 8.37100887298584, "learning_rate": 6.518932038834952e-05, "loss": 0.2975, "step": 7171 }, { "epoch": 69.5433734939759, "grad_norm": 6.15388822555542, "learning_rate": 6.518446601941748e-05, "loss": 0.1784, "step": 7172 }, { "epoch": 69.55301204819277, "grad_norm": 7.258922576904297, "learning_rate": 6.517961165048544e-05, "loss": 0.2091, "step": 7173 }, { "epoch": 69.56265060240963, "grad_norm": 9.255908012390137, "learning_rate": 6.51747572815534e-05, "loss": 0.3397, "step": 7174 }, { "epoch": 69.57228915662651, "grad_norm": 8.913677215576172, "learning_rate": 6.516990291262136e-05, "loss": 0.1714, "step": 7175 }, { "epoch": 69.58192771084337, "grad_norm": 7.010499000549316, "learning_rate": 6.516504854368932e-05, "loss": 0.3922, "step": 7176 }, { "epoch": 69.59156626506024, "grad_norm": 8.560977935791016, "learning_rate": 6.516019417475728e-05, "loss": 0.1938, "step": 7177 }, { "epoch": 69.60120481927711, "grad_norm": 10.461222648620605, "learning_rate": 6.515533980582524e-05, "loss": 0.5003, "step": 7178 }, { "epoch": 69.61084337349398, "grad_norm": 267.7281188964844, "learning_rate": 6.515048543689321e-05, "loss": 0.4539, "step": 7179 }, { "epoch": 69.62048192771084, "grad_norm": 6.955995082855225, "learning_rate": 6.514563106796117e-05, "loss": 0.3715, "step": 7180 }, { "epoch": 69.6301204819277, "grad_norm": 7.497957229614258, "learning_rate": 6.514077669902912e-05, "loss": 0.3153, "step": 7181 }, { "epoch": 69.63975903614458, "grad_norm": 18.521066665649414, "learning_rate": 6.513592233009708e-05, "loss": 0.4374, "step": 7182 }, { "epoch": 69.64939759036145, "grad_norm": 10.886887550354004, "learning_rate": 6.513106796116505e-05, "loss": 0.4666, "step": 7183 }, { "epoch": 69.65903614457831, "grad_norm": 11.49543285369873, "learning_rate": 6.512621359223301e-05, "loss": 0.3459, "step": 7184 }, { "epoch": 69.66867469879519, "grad_norm": 8.504899024963379, "learning_rate": 6.512135922330097e-05, "loss": 0.2931, "step": 7185 }, { "epoch": 69.67831325301205, "grad_norm": 34.74405288696289, "learning_rate": 6.511650485436893e-05, "loss": 0.3978, "step": 7186 }, { "epoch": 69.68795180722891, "grad_norm": 29.007070541381836, "learning_rate": 6.51116504854369e-05, "loss": 0.3835, "step": 7187 }, { "epoch": 69.69759036144578, "grad_norm": 17.578245162963867, "learning_rate": 6.510679611650486e-05, "loss": 0.4415, "step": 7188 }, { "epoch": 69.70722891566265, "grad_norm": 15.236289978027344, "learning_rate": 6.510194174757281e-05, "loss": 0.5136, "step": 7189 }, { "epoch": 69.71686746987952, "grad_norm": 9.13182258605957, "learning_rate": 6.509708737864077e-05, "loss": 0.2121, "step": 7190 }, { "epoch": 69.72650602409638, "grad_norm": 31.407747268676758, "learning_rate": 6.509223300970874e-05, "loss": 0.3608, "step": 7191 }, { "epoch": 69.73614457831326, "grad_norm": 14.146490097045898, "learning_rate": 6.50873786407767e-05, "loss": 0.484, "step": 7192 }, { "epoch": 69.74578313253012, "grad_norm": 13.16893196105957, "learning_rate": 6.508252427184466e-05, "loss": 0.3717, "step": 7193 }, { "epoch": 69.75542168674698, "grad_norm": 4.011922836303711, "learning_rate": 6.507766990291262e-05, "loss": 0.1808, "step": 7194 }, { "epoch": 69.76506024096386, "grad_norm": 7.41953182220459, "learning_rate": 6.507281553398059e-05, "loss": 0.2998, "step": 7195 }, { "epoch": 69.77469879518073, "grad_norm": 3.3289685249328613, "learning_rate": 6.506796116504856e-05, "loss": 0.1086, "step": 7196 }, { "epoch": 69.78433734939759, "grad_norm": 7.984865188598633, "learning_rate": 6.506310679611652e-05, "loss": 0.3056, "step": 7197 }, { "epoch": 69.79397590361445, "grad_norm": 20.13418960571289, "learning_rate": 6.505825242718447e-05, "loss": 0.4292, "step": 7198 }, { "epoch": 69.80361445783133, "grad_norm": 7.130465984344482, "learning_rate": 6.505339805825243e-05, "loss": 0.1697, "step": 7199 }, { "epoch": 69.8132530120482, "grad_norm": 11.135109901428223, "learning_rate": 6.504854368932039e-05, "loss": 0.1745, "step": 7200 }, { "epoch": 69.82289156626506, "grad_norm": 4.777661323547363, "learning_rate": 6.504368932038836e-05, "loss": 0.1642, "step": 7201 }, { "epoch": 69.83253012048193, "grad_norm": 22.22011947631836, "learning_rate": 6.503883495145632e-05, "loss": 0.3596, "step": 7202 }, { "epoch": 69.8421686746988, "grad_norm": 9.09559154510498, "learning_rate": 6.503398058252428e-05, "loss": 0.4331, "step": 7203 }, { "epoch": 69.85180722891566, "grad_norm": 6.793276309967041, "learning_rate": 6.502912621359223e-05, "loss": 0.2071, "step": 7204 }, { "epoch": 69.86144578313252, "grad_norm": 12.622445106506348, "learning_rate": 6.50242718446602e-05, "loss": 0.2598, "step": 7205 }, { "epoch": 69.8710843373494, "grad_norm": 11.797663688659668, "learning_rate": 6.501941747572816e-05, "loss": 0.2018, "step": 7206 }, { "epoch": 69.88072289156626, "grad_norm": 8.169713973999023, "learning_rate": 6.501456310679612e-05, "loss": 0.3216, "step": 7207 }, { "epoch": 69.89036144578313, "grad_norm": 10.272798538208008, "learning_rate": 6.500970873786408e-05, "loss": 0.2596, "step": 7208 }, { "epoch": 69.9, "grad_norm": 8.378496170043945, "learning_rate": 6.500485436893205e-05, "loss": 0.0814, "step": 7209 }, { "epoch": 69.90963855421687, "grad_norm": 12.845830917358398, "learning_rate": 6.500000000000001e-05, "loss": 0.1829, "step": 7210 }, { "epoch": 69.91927710843373, "grad_norm": 19.380571365356445, "learning_rate": 6.499514563106797e-05, "loss": 0.4742, "step": 7211 }, { "epoch": 69.9289156626506, "grad_norm": 16.32282829284668, "learning_rate": 6.499029126213592e-05, "loss": 0.2962, "step": 7212 }, { "epoch": 69.93855421686747, "grad_norm": 4.069680690765381, "learning_rate": 6.49854368932039e-05, "loss": 0.2644, "step": 7213 }, { "epoch": 69.94819277108434, "grad_norm": 9.152461051940918, "learning_rate": 6.498058252427185e-05, "loss": 0.18, "step": 7214 }, { "epoch": 69.9578313253012, "grad_norm": 5.410151958465576, "learning_rate": 6.497572815533981e-05, "loss": 0.2362, "step": 7215 }, { "epoch": 69.96746987951808, "grad_norm": 16.195697784423828, "learning_rate": 6.497087378640777e-05, "loss": 0.238, "step": 7216 }, { "epoch": 69.97710843373494, "grad_norm": 8.768218994140625, "learning_rate": 6.496601941747573e-05, "loss": 0.1273, "step": 7217 }, { "epoch": 69.9867469879518, "grad_norm": 9.065863609313965, "learning_rate": 6.49611650485437e-05, "loss": 0.1972, "step": 7218 }, { "epoch": 70.00240963855421, "grad_norm": 4.080842018127441, "learning_rate": 6.495631067961165e-05, "loss": 0.1181, "step": 7219 }, { "epoch": 70.01204819277109, "grad_norm": 16.245275497436523, "learning_rate": 6.495145631067961e-05, "loss": 0.2853, "step": 7220 }, { "epoch": 70.02168674698795, "grad_norm": 6.773704528808594, "learning_rate": 6.494660194174757e-05, "loss": 0.2714, "step": 7221 }, { "epoch": 70.03132530120482, "grad_norm": 49.664100646972656, "learning_rate": 6.494174757281554e-05, "loss": 0.1971, "step": 7222 }, { "epoch": 70.04096385542168, "grad_norm": 8.032182693481445, "learning_rate": 6.49368932038835e-05, "loss": 0.229, "step": 7223 }, { "epoch": 70.05060240963856, "grad_norm": 33.464332580566406, "learning_rate": 6.493203883495146e-05, "loss": 0.3672, "step": 7224 }, { "epoch": 70.06024096385542, "grad_norm": 6.405123233795166, "learning_rate": 6.492718446601941e-05, "loss": 0.2219, "step": 7225 }, { "epoch": 70.06987951807228, "grad_norm": 15.448386192321777, "learning_rate": 6.492233009708739e-05, "loss": 0.2113, "step": 7226 }, { "epoch": 70.07951807228916, "grad_norm": 3.376163959503174, "learning_rate": 6.491747572815534e-05, "loss": 0.2098, "step": 7227 }, { "epoch": 70.08915662650602, "grad_norm": 30.6037540435791, "learning_rate": 6.49126213592233e-05, "loss": 0.318, "step": 7228 }, { "epoch": 70.09879518072289, "grad_norm": 9.735767364501953, "learning_rate": 6.490776699029126e-05, "loss": 0.125, "step": 7229 }, { "epoch": 70.10843373493977, "grad_norm": 10.558013916015625, "learning_rate": 6.490291262135923e-05, "loss": 0.446, "step": 7230 }, { "epoch": 70.11807228915663, "grad_norm": 13.540790557861328, "learning_rate": 6.489805825242719e-05, "loss": 0.3775, "step": 7231 }, { "epoch": 70.12771084337349, "grad_norm": 12.920905113220215, "learning_rate": 6.489320388349515e-05, "loss": 0.5354, "step": 7232 }, { "epoch": 70.13734939759036, "grad_norm": 22.426921844482422, "learning_rate": 6.48883495145631e-05, "loss": 0.2593, "step": 7233 }, { "epoch": 70.14698795180723, "grad_norm": 27.14362907409668, "learning_rate": 6.488349514563106e-05, "loss": 0.4239, "step": 7234 }, { "epoch": 70.1566265060241, "grad_norm": 2.176504135131836, "learning_rate": 6.487864077669903e-05, "loss": 0.2298, "step": 7235 }, { "epoch": 70.16626506024096, "grad_norm": 7.181814670562744, "learning_rate": 6.487378640776699e-05, "loss": 0.2281, "step": 7236 }, { "epoch": 70.17590361445784, "grad_norm": 4.9585113525390625, "learning_rate": 6.486893203883495e-05, "loss": 0.1334, "step": 7237 }, { "epoch": 70.1855421686747, "grad_norm": 12.166047096252441, "learning_rate": 6.48640776699029e-05, "loss": 0.1645, "step": 7238 }, { "epoch": 70.19518072289156, "grad_norm": 13.430472373962402, "learning_rate": 6.485922330097088e-05, "loss": 0.5065, "step": 7239 }, { "epoch": 70.20481927710843, "grad_norm": 11.380019187927246, "learning_rate": 6.485436893203884e-05, "loss": 0.4061, "step": 7240 }, { "epoch": 70.2144578313253, "grad_norm": 2.5540478229522705, "learning_rate": 6.484951456310679e-05, "loss": 0.1921, "step": 7241 }, { "epoch": 70.22409638554217, "grad_norm": 7.36912202835083, "learning_rate": 6.484466019417475e-05, "loss": 0.4499, "step": 7242 }, { "epoch": 70.23373493975903, "grad_norm": 3.864319086074829, "learning_rate": 6.483980582524272e-05, "loss": 0.2228, "step": 7243 }, { "epoch": 70.24337349397591, "grad_norm": 7.218557357788086, "learning_rate": 6.483495145631068e-05, "loss": 0.4752, "step": 7244 }, { "epoch": 70.25301204819277, "grad_norm": 8.082222938537598, "learning_rate": 6.483009708737864e-05, "loss": 0.0955, "step": 7245 }, { "epoch": 70.26265060240964, "grad_norm": 6.252008438110352, "learning_rate": 6.482524271844661e-05, "loss": 0.256, "step": 7246 }, { "epoch": 70.2722891566265, "grad_norm": 10.328380584716797, "learning_rate": 6.482038834951457e-05, "loss": 0.3045, "step": 7247 }, { "epoch": 70.28192771084338, "grad_norm": 10.518881797790527, "learning_rate": 6.481553398058254e-05, "loss": 0.2264, "step": 7248 }, { "epoch": 70.29156626506024, "grad_norm": 7.62623405456543, "learning_rate": 6.48106796116505e-05, "loss": 0.2285, "step": 7249 }, { "epoch": 70.3012048192771, "grad_norm": 4.782765865325928, "learning_rate": 6.480582524271845e-05, "loss": 0.2104, "step": 7250 }, { "epoch": 70.31084337349398, "grad_norm": 6.906754493713379, "learning_rate": 6.480097087378641e-05, "loss": 0.1792, "step": 7251 }, { "epoch": 70.32048192771084, "grad_norm": 7.099820613861084, "learning_rate": 6.479611650485438e-05, "loss": 0.4758, "step": 7252 }, { "epoch": 70.33012048192771, "grad_norm": 8.645214080810547, "learning_rate": 6.479126213592234e-05, "loss": 0.4047, "step": 7253 }, { "epoch": 70.33975903614459, "grad_norm": 6.722445011138916, "learning_rate": 6.47864077669903e-05, "loss": 0.2445, "step": 7254 }, { "epoch": 70.34939759036145, "grad_norm": 6.66391658782959, "learning_rate": 6.478155339805826e-05, "loss": 0.3072, "step": 7255 }, { "epoch": 70.35903614457831, "grad_norm": 5.578937530517578, "learning_rate": 6.477669902912621e-05, "loss": 0.2169, "step": 7256 }, { "epoch": 70.36867469879518, "grad_norm": 17.282609939575195, "learning_rate": 6.477184466019419e-05, "loss": 0.3085, "step": 7257 }, { "epoch": 70.37831325301205, "grad_norm": 12.414284706115723, "learning_rate": 6.476699029126214e-05, "loss": 0.3077, "step": 7258 }, { "epoch": 70.38795180722892, "grad_norm": 21.306760787963867, "learning_rate": 6.47621359223301e-05, "loss": 0.4661, "step": 7259 }, { "epoch": 70.39759036144578, "grad_norm": 6.859759330749512, "learning_rate": 6.475728155339806e-05, "loss": 0.3058, "step": 7260 }, { "epoch": 70.40722891566266, "grad_norm": 3.031981945037842, "learning_rate": 6.475242718446603e-05, "loss": 0.1919, "step": 7261 }, { "epoch": 70.41686746987952, "grad_norm": 5.115494728088379, "learning_rate": 6.474757281553399e-05, "loss": 0.2828, "step": 7262 }, { "epoch": 70.42650602409638, "grad_norm": 16.321819305419922, "learning_rate": 6.474271844660195e-05, "loss": 0.3923, "step": 7263 }, { "epoch": 70.43614457831325, "grad_norm": 9.435128211975098, "learning_rate": 6.47378640776699e-05, "loss": 0.242, "step": 7264 }, { "epoch": 70.44578313253012, "grad_norm": 9.367366790771484, "learning_rate": 6.473300970873787e-05, "loss": 0.3818, "step": 7265 }, { "epoch": 70.45542168674699, "grad_norm": 9.524446487426758, "learning_rate": 6.472815533980583e-05, "loss": 0.3068, "step": 7266 }, { "epoch": 70.46506024096385, "grad_norm": 5.649787902832031, "learning_rate": 6.472330097087379e-05, "loss": 0.1342, "step": 7267 }, { "epoch": 70.47469879518073, "grad_norm": 9.604188919067383, "learning_rate": 6.471844660194175e-05, "loss": 0.37, "step": 7268 }, { "epoch": 70.48433734939759, "grad_norm": 9.028570175170898, "learning_rate": 6.471359223300972e-05, "loss": 0.2587, "step": 7269 }, { "epoch": 70.49397590361446, "grad_norm": 9.891403198242188, "learning_rate": 6.470873786407768e-05, "loss": 0.3383, "step": 7270 }, { "epoch": 70.50361445783132, "grad_norm": 36.89849853515625, "learning_rate": 6.470388349514563e-05, "loss": 0.511, "step": 7271 }, { "epoch": 70.5132530120482, "grad_norm": 4.385432243347168, "learning_rate": 6.469902912621359e-05, "loss": 0.2439, "step": 7272 }, { "epoch": 70.52289156626506, "grad_norm": 47.67609405517578, "learning_rate": 6.469417475728155e-05, "loss": 0.3099, "step": 7273 }, { "epoch": 70.53253012048192, "grad_norm": 6.702789306640625, "learning_rate": 6.468932038834952e-05, "loss": 0.1996, "step": 7274 }, { "epoch": 70.5421686746988, "grad_norm": 13.597227096557617, "learning_rate": 6.468446601941748e-05, "loss": 0.4945, "step": 7275 }, { "epoch": 70.55180722891566, "grad_norm": 11.669273376464844, "learning_rate": 6.467961165048544e-05, "loss": 0.2145, "step": 7276 }, { "epoch": 70.56144578313253, "grad_norm": 10.924080848693848, "learning_rate": 6.46747572815534e-05, "loss": 0.2754, "step": 7277 }, { "epoch": 70.5710843373494, "grad_norm": 19.83393096923828, "learning_rate": 6.466990291262137e-05, "loss": 0.4053, "step": 7278 }, { "epoch": 70.58072289156627, "grad_norm": 61.83207702636719, "learning_rate": 6.466504854368932e-05, "loss": 0.3344, "step": 7279 }, { "epoch": 70.59036144578313, "grad_norm": 31.04407501220703, "learning_rate": 6.466019417475728e-05, "loss": 0.2593, "step": 7280 }, { "epoch": 70.6, "grad_norm": 14.401809692382812, "learning_rate": 6.465533980582524e-05, "loss": 0.2812, "step": 7281 }, { "epoch": 70.60963855421687, "grad_norm": 3.6399619579315186, "learning_rate": 6.465048543689321e-05, "loss": 0.1465, "step": 7282 }, { "epoch": 70.61927710843374, "grad_norm": 9.240497589111328, "learning_rate": 6.464563106796117e-05, "loss": 0.2995, "step": 7283 }, { "epoch": 70.6289156626506, "grad_norm": 8.631453514099121, "learning_rate": 6.464077669902913e-05, "loss": 0.2437, "step": 7284 }, { "epoch": 70.63855421686748, "grad_norm": 10.906907081604004, "learning_rate": 6.463592233009708e-05, "loss": 0.2365, "step": 7285 }, { "epoch": 70.64819277108434, "grad_norm": 14.266180992126465, "learning_rate": 6.463106796116505e-05, "loss": 0.1654, "step": 7286 }, { "epoch": 70.6578313253012, "grad_norm": 4.173593997955322, "learning_rate": 6.462621359223301e-05, "loss": 0.1558, "step": 7287 }, { "epoch": 70.66746987951807, "grad_norm": 6.7250142097473145, "learning_rate": 6.462135922330097e-05, "loss": 0.233, "step": 7288 }, { "epoch": 70.67710843373494, "grad_norm": 6.727275848388672, "learning_rate": 6.461650485436893e-05, "loss": 0.2727, "step": 7289 }, { "epoch": 70.6867469879518, "grad_norm": 7.127630233764648, "learning_rate": 6.461165048543689e-05, "loss": 0.4021, "step": 7290 }, { "epoch": 70.69638554216867, "grad_norm": 7.019252300262451, "learning_rate": 6.460679611650486e-05, "loss": 0.2783, "step": 7291 }, { "epoch": 70.70602409638555, "grad_norm": 32.813941955566406, "learning_rate": 6.460194174757282e-05, "loss": 0.4622, "step": 7292 }, { "epoch": 70.71566265060241, "grad_norm": 6.108610153198242, "learning_rate": 6.459708737864077e-05, "loss": 0.161, "step": 7293 }, { "epoch": 70.72530120481927, "grad_norm": 5.826447010040283, "learning_rate": 6.459223300970873e-05, "loss": 0.1641, "step": 7294 }, { "epoch": 70.73493975903614, "grad_norm": 9.170880317687988, "learning_rate": 6.45873786407767e-05, "loss": 0.3207, "step": 7295 }, { "epoch": 70.74457831325302, "grad_norm": 7.974483489990234, "learning_rate": 6.458252427184467e-05, "loss": 0.2173, "step": 7296 }, { "epoch": 70.75421686746988, "grad_norm": 4.973574638366699, "learning_rate": 6.457766990291263e-05, "loss": 0.2777, "step": 7297 }, { "epoch": 70.76385542168674, "grad_norm": 30.059185028076172, "learning_rate": 6.457281553398059e-05, "loss": 0.4221, "step": 7298 }, { "epoch": 70.77349397590362, "grad_norm": 23.534570693969727, "learning_rate": 6.456796116504855e-05, "loss": 0.2092, "step": 7299 }, { "epoch": 70.78313253012048, "grad_norm": 7.544641971588135, "learning_rate": 6.456310679611652e-05, "loss": 0.2145, "step": 7300 }, { "epoch": 70.79277108433735, "grad_norm": 7.999875068664551, "learning_rate": 6.455825242718448e-05, "loss": 0.2097, "step": 7301 }, { "epoch": 70.80240963855422, "grad_norm": 10.062063217163086, "learning_rate": 6.455339805825243e-05, "loss": 0.3131, "step": 7302 }, { "epoch": 70.81204819277109, "grad_norm": 10.911698341369629, "learning_rate": 6.454854368932039e-05, "loss": 0.3072, "step": 7303 }, { "epoch": 70.82168674698795, "grad_norm": 11.023191452026367, "learning_rate": 6.454368932038836e-05, "loss": 0.4074, "step": 7304 }, { "epoch": 70.83132530120481, "grad_norm": 14.378754615783691, "learning_rate": 6.453883495145632e-05, "loss": 0.2016, "step": 7305 }, { "epoch": 70.84096385542169, "grad_norm": 11.316359519958496, "learning_rate": 6.453398058252428e-05, "loss": 0.2265, "step": 7306 }, { "epoch": 70.85060240963855, "grad_norm": 7.164603233337402, "learning_rate": 6.452912621359224e-05, "loss": 0.2879, "step": 7307 }, { "epoch": 70.86024096385542, "grad_norm": 4.391999244689941, "learning_rate": 6.452427184466021e-05, "loss": 0.1551, "step": 7308 }, { "epoch": 70.8698795180723, "grad_norm": 10.127483367919922, "learning_rate": 6.451941747572816e-05, "loss": 0.385, "step": 7309 }, { "epoch": 70.87951807228916, "grad_norm": 4.423041820526123, "learning_rate": 6.451456310679612e-05, "loss": 0.2486, "step": 7310 }, { "epoch": 70.88915662650602, "grad_norm": 8.918227195739746, "learning_rate": 6.450970873786408e-05, "loss": 0.3802, "step": 7311 }, { "epoch": 70.89879518072289, "grad_norm": 6.891106605529785, "learning_rate": 6.450485436893204e-05, "loss": 0.1725, "step": 7312 }, { "epoch": 70.90843373493976, "grad_norm": 16.015365600585938, "learning_rate": 6.450000000000001e-05, "loss": 0.3996, "step": 7313 }, { "epoch": 70.91807228915663, "grad_norm": 5.061431884765625, "learning_rate": 6.449514563106797e-05, "loss": 0.3524, "step": 7314 }, { "epoch": 70.92771084337349, "grad_norm": 5.98171854019165, "learning_rate": 6.449029126213592e-05, "loss": 0.3621, "step": 7315 }, { "epoch": 70.93734939759037, "grad_norm": 18.39668846130371, "learning_rate": 6.448543689320388e-05, "loss": 0.4275, "step": 7316 }, { "epoch": 70.94698795180723, "grad_norm": 14.77922248840332, "learning_rate": 6.448058252427185e-05, "loss": 0.2462, "step": 7317 }, { "epoch": 70.9566265060241, "grad_norm": 9.469416618347168, "learning_rate": 6.447572815533981e-05, "loss": 0.4219, "step": 7318 }, { "epoch": 70.96626506024096, "grad_norm": 12.911130905151367, "learning_rate": 6.447087378640777e-05, "loss": 0.3552, "step": 7319 }, { "epoch": 70.97590361445783, "grad_norm": 8.943942070007324, "learning_rate": 6.446601941747573e-05, "loss": 0.2685, "step": 7320 }, { "epoch": 70.9855421686747, "grad_norm": 10.141770362854004, "learning_rate": 6.44611650485437e-05, "loss": 0.2497, "step": 7321 }, { "epoch": 71.0012048192771, "grad_norm": 9.685593605041504, "learning_rate": 6.445631067961166e-05, "loss": 0.3434, "step": 7322 }, { "epoch": 71.01084337349397, "grad_norm": 10.981783866882324, "learning_rate": 6.445145631067961e-05, "loss": 0.1912, "step": 7323 }, { "epoch": 71.02048192771085, "grad_norm": 23.5967960357666, "learning_rate": 6.444660194174757e-05, "loss": 0.2983, "step": 7324 }, { "epoch": 71.03012048192771, "grad_norm": 17.013845443725586, "learning_rate": 6.444174757281553e-05, "loss": 0.3463, "step": 7325 }, { "epoch": 71.03975903614457, "grad_norm": 8.412620544433594, "learning_rate": 6.44368932038835e-05, "loss": 0.3448, "step": 7326 }, { "epoch": 71.04939759036145, "grad_norm": 7.814188003540039, "learning_rate": 6.443203883495146e-05, "loss": 0.2319, "step": 7327 }, { "epoch": 71.05903614457831, "grad_norm": 12.306654930114746, "learning_rate": 6.442718446601942e-05, "loss": 0.1993, "step": 7328 }, { "epoch": 71.06867469879518, "grad_norm": 11.081880569458008, "learning_rate": 6.442233009708737e-05, "loss": 0.3424, "step": 7329 }, { "epoch": 71.07831325301204, "grad_norm": 9.314651489257812, "learning_rate": 6.441747572815535e-05, "loss": 0.2432, "step": 7330 }, { "epoch": 71.08795180722892, "grad_norm": 9.278436660766602, "learning_rate": 6.44126213592233e-05, "loss": 0.3073, "step": 7331 }, { "epoch": 71.09759036144578, "grad_norm": 7.685369491577148, "learning_rate": 6.440776699029126e-05, "loss": 0.3558, "step": 7332 }, { "epoch": 71.10722891566265, "grad_norm": 8.097548484802246, "learning_rate": 6.440291262135922e-05, "loss": 0.4553, "step": 7333 }, { "epoch": 71.11686746987952, "grad_norm": 7.888114929199219, "learning_rate": 6.439805825242719e-05, "loss": 0.3429, "step": 7334 }, { "epoch": 71.12650602409639, "grad_norm": 13.474534034729004, "learning_rate": 6.439320388349515e-05, "loss": 0.3798, "step": 7335 }, { "epoch": 71.13614457831325, "grad_norm": 11.221755027770996, "learning_rate": 6.43883495145631e-05, "loss": 0.3253, "step": 7336 }, { "epoch": 71.14578313253013, "grad_norm": 11.313956260681152, "learning_rate": 6.438349514563106e-05, "loss": 0.3309, "step": 7337 }, { "epoch": 71.15542168674699, "grad_norm": 16.39841651916504, "learning_rate": 6.437864077669903e-05, "loss": 0.4355, "step": 7338 }, { "epoch": 71.16506024096385, "grad_norm": 8.551618576049805, "learning_rate": 6.437378640776699e-05, "loss": 0.3586, "step": 7339 }, { "epoch": 71.17469879518072, "grad_norm": 13.702754020690918, "learning_rate": 6.436893203883495e-05, "loss": 0.4378, "step": 7340 }, { "epoch": 71.1843373493976, "grad_norm": 10.046923637390137, "learning_rate": 6.436407766990291e-05, "loss": 0.3587, "step": 7341 }, { "epoch": 71.19397590361446, "grad_norm": 14.188945770263672, "learning_rate": 6.435922330097087e-05, "loss": 0.3458, "step": 7342 }, { "epoch": 71.20361445783132, "grad_norm": 11.285877227783203, "learning_rate": 6.435436893203884e-05, "loss": 0.33, "step": 7343 }, { "epoch": 71.2132530120482, "grad_norm": 7.3962082862854, "learning_rate": 6.43495145631068e-05, "loss": 0.2635, "step": 7344 }, { "epoch": 71.22289156626506, "grad_norm": 6.701144695281982, "learning_rate": 6.434466019417475e-05, "loss": 0.2644, "step": 7345 }, { "epoch": 71.23253012048193, "grad_norm": 10.811965942382812, "learning_rate": 6.433980582524271e-05, "loss": 0.2265, "step": 7346 }, { "epoch": 71.24216867469879, "grad_norm": 7.624083042144775, "learning_rate": 6.433495145631068e-05, "loss": 0.3485, "step": 7347 }, { "epoch": 71.25180722891567, "grad_norm": 7.824960708618164, "learning_rate": 6.433009708737865e-05, "loss": 0.4691, "step": 7348 }, { "epoch": 71.26144578313253, "grad_norm": 5.680123329162598, "learning_rate": 6.432524271844661e-05, "loss": 0.2884, "step": 7349 }, { "epoch": 71.2710843373494, "grad_norm": 6.210968494415283, "learning_rate": 6.432038834951457e-05, "loss": 0.177, "step": 7350 }, { "epoch": 71.28072289156627, "grad_norm": 14.918134689331055, "learning_rate": 6.431553398058253e-05, "loss": 0.4463, "step": 7351 }, { "epoch": 71.29036144578313, "grad_norm": 6.098474979400635, "learning_rate": 6.43106796116505e-05, "loss": 0.2546, "step": 7352 }, { "epoch": 71.3, "grad_norm": 13.264848709106445, "learning_rate": 6.430582524271846e-05, "loss": 0.42, "step": 7353 }, { "epoch": 71.30963855421686, "grad_norm": 6.253969192504883, "learning_rate": 6.430097087378641e-05, "loss": 0.1742, "step": 7354 }, { "epoch": 71.31927710843374, "grad_norm": 4.963611602783203, "learning_rate": 6.429611650485437e-05, "loss": 0.2672, "step": 7355 }, { "epoch": 71.3289156626506, "grad_norm": 11.778057098388672, "learning_rate": 6.429126213592234e-05, "loss": 0.3778, "step": 7356 }, { "epoch": 71.33855421686746, "grad_norm": 8.487443923950195, "learning_rate": 6.42864077669903e-05, "loss": 0.2417, "step": 7357 }, { "epoch": 71.34819277108434, "grad_norm": 6.854743480682373, "learning_rate": 6.428155339805826e-05, "loss": 0.314, "step": 7358 }, { "epoch": 71.3578313253012, "grad_norm": 10.919300079345703, "learning_rate": 6.427669902912622e-05, "loss": 0.44, "step": 7359 }, { "epoch": 71.36746987951807, "grad_norm": 9.480720520019531, "learning_rate": 6.427184466019419e-05, "loss": 0.4088, "step": 7360 }, { "epoch": 71.37710843373495, "grad_norm": 11.179807662963867, "learning_rate": 6.426699029126214e-05, "loss": 0.3129, "step": 7361 }, { "epoch": 71.38674698795181, "grad_norm": 16.568256378173828, "learning_rate": 6.42621359223301e-05, "loss": 0.3501, "step": 7362 }, { "epoch": 71.39638554216867, "grad_norm": 8.735665321350098, "learning_rate": 6.425728155339806e-05, "loss": 0.2859, "step": 7363 }, { "epoch": 71.40602409638554, "grad_norm": 5.909602165222168, "learning_rate": 6.425242718446602e-05, "loss": 0.1714, "step": 7364 }, { "epoch": 71.41566265060241, "grad_norm": 6.085857391357422, "learning_rate": 6.424757281553399e-05, "loss": 0.1916, "step": 7365 }, { "epoch": 71.42530120481928, "grad_norm": 9.752533912658691, "learning_rate": 6.424271844660195e-05, "loss": 0.3774, "step": 7366 }, { "epoch": 71.43493975903614, "grad_norm": 11.767082214355469, "learning_rate": 6.42378640776699e-05, "loss": 0.5339, "step": 7367 }, { "epoch": 71.44457831325302, "grad_norm": 7.920121192932129, "learning_rate": 6.423300970873786e-05, "loss": 0.2577, "step": 7368 }, { "epoch": 71.45421686746988, "grad_norm": 10.772014617919922, "learning_rate": 6.422815533980583e-05, "loss": 0.4684, "step": 7369 }, { "epoch": 71.46385542168674, "grad_norm": 9.115220069885254, "learning_rate": 6.422330097087379e-05, "loss": 0.3308, "step": 7370 }, { "epoch": 71.47349397590361, "grad_norm": 7.4478583335876465, "learning_rate": 6.421844660194175e-05, "loss": 0.2636, "step": 7371 }, { "epoch": 71.48313253012049, "grad_norm": 9.397438049316406, "learning_rate": 6.42135922330097e-05, "loss": 0.3626, "step": 7372 }, { "epoch": 71.49277108433735, "grad_norm": 4.075189113616943, "learning_rate": 6.420873786407768e-05, "loss": 0.1627, "step": 7373 }, { "epoch": 71.50240963855421, "grad_norm": 8.16367244720459, "learning_rate": 6.420388349514564e-05, "loss": 0.3966, "step": 7374 }, { "epoch": 71.51204819277109, "grad_norm": 6.360706329345703, "learning_rate": 6.41990291262136e-05, "loss": 0.1132, "step": 7375 }, { "epoch": 71.52168674698795, "grad_norm": 5.592076778411865, "learning_rate": 6.419417475728155e-05, "loss": 0.3443, "step": 7376 }, { "epoch": 71.53132530120482, "grad_norm": 7.184415340423584, "learning_rate": 6.418932038834952e-05, "loss": 0.3541, "step": 7377 }, { "epoch": 71.54096385542168, "grad_norm": 12.132803916931152, "learning_rate": 6.418446601941748e-05, "loss": 0.261, "step": 7378 }, { "epoch": 71.55060240963856, "grad_norm": 14.554036140441895, "learning_rate": 6.417961165048544e-05, "loss": 0.4333, "step": 7379 }, { "epoch": 71.56024096385542, "grad_norm": 8.93655776977539, "learning_rate": 6.41747572815534e-05, "loss": 0.2802, "step": 7380 }, { "epoch": 71.56987951807228, "grad_norm": 23.014780044555664, "learning_rate": 6.416990291262135e-05, "loss": 0.4299, "step": 7381 }, { "epoch": 71.57951807228916, "grad_norm": 8.18402099609375, "learning_rate": 6.416504854368932e-05, "loss": 0.2632, "step": 7382 }, { "epoch": 71.58915662650602, "grad_norm": 13.584380149841309, "learning_rate": 6.416019417475728e-05, "loss": 0.3467, "step": 7383 }, { "epoch": 71.59879518072289, "grad_norm": 5.107515335083008, "learning_rate": 6.415533980582524e-05, "loss": 0.1655, "step": 7384 }, { "epoch": 71.60843373493977, "grad_norm": 2.209731101989746, "learning_rate": 6.41504854368932e-05, "loss": 0.1614, "step": 7385 }, { "epoch": 71.61807228915663, "grad_norm": 18.432634353637695, "learning_rate": 6.414563106796117e-05, "loss": 0.5609, "step": 7386 }, { "epoch": 71.62771084337349, "grad_norm": 14.594766616821289, "learning_rate": 6.414077669902913e-05, "loss": 0.6029, "step": 7387 }, { "epoch": 71.63734939759036, "grad_norm": 11.60970401763916, "learning_rate": 6.413592233009709e-05, "loss": 0.2014, "step": 7388 }, { "epoch": 71.64698795180723, "grad_norm": 7.536970138549805, "learning_rate": 6.413106796116504e-05, "loss": 0.37, "step": 7389 }, { "epoch": 71.6566265060241, "grad_norm": 7.912495136260986, "learning_rate": 6.412621359223301e-05, "loss": 0.2308, "step": 7390 }, { "epoch": 71.66626506024096, "grad_norm": 11.040520668029785, "learning_rate": 6.412135922330097e-05, "loss": 0.2891, "step": 7391 }, { "epoch": 71.67590361445784, "grad_norm": 6.821792125701904, "learning_rate": 6.411650485436893e-05, "loss": 0.2943, "step": 7392 }, { "epoch": 71.6855421686747, "grad_norm": 14.027216911315918, "learning_rate": 6.411165048543689e-05, "loss": 0.5167, "step": 7393 }, { "epoch": 71.69518072289156, "grad_norm": 32.68931198120117, "learning_rate": 6.410679611650486e-05, "loss": 0.3108, "step": 7394 }, { "epoch": 71.70481927710843, "grad_norm": 7.392988204956055, "learning_rate": 6.410194174757282e-05, "loss": 0.2579, "step": 7395 }, { "epoch": 71.7144578313253, "grad_norm": 4.885997295379639, "learning_rate": 6.409708737864077e-05, "loss": 0.3413, "step": 7396 }, { "epoch": 71.72409638554217, "grad_norm": 4.510356903076172, "learning_rate": 6.409223300970875e-05, "loss": 0.3217, "step": 7397 }, { "epoch": 71.73373493975903, "grad_norm": 10.073853492736816, "learning_rate": 6.40873786407767e-05, "loss": 0.2465, "step": 7398 }, { "epoch": 71.74337349397591, "grad_norm": 8.656076431274414, "learning_rate": 6.408252427184467e-05, "loss": 0.371, "step": 7399 }, { "epoch": 71.75301204819277, "grad_norm": 18.240530014038086, "learning_rate": 6.407766990291263e-05, "loss": 0.6131, "step": 7400 }, { "epoch": 71.76265060240964, "grad_norm": 7.330814838409424, "learning_rate": 6.407281553398059e-05, "loss": 0.2725, "step": 7401 }, { "epoch": 71.7722891566265, "grad_norm": 15.61225700378418, "learning_rate": 6.406796116504855e-05, "loss": 0.4477, "step": 7402 }, { "epoch": 71.78192771084338, "grad_norm": 13.134319305419922, "learning_rate": 6.40631067961165e-05, "loss": 0.333, "step": 7403 }, { "epoch": 71.79156626506024, "grad_norm": 2.846884250640869, "learning_rate": 6.405825242718448e-05, "loss": 0.1731, "step": 7404 }, { "epoch": 71.8012048192771, "grad_norm": 7.352818489074707, "learning_rate": 6.405339805825243e-05, "loss": 0.263, "step": 7405 }, { "epoch": 71.81084337349398, "grad_norm": 5.731307506561279, "learning_rate": 6.404854368932039e-05, "loss": 0.1901, "step": 7406 }, { "epoch": 71.82048192771084, "grad_norm": 6.536317825317383, "learning_rate": 6.404368932038835e-05, "loss": 0.2436, "step": 7407 }, { "epoch": 71.83012048192771, "grad_norm": 8.845735549926758, "learning_rate": 6.403883495145632e-05, "loss": 0.4674, "step": 7408 }, { "epoch": 71.83975903614459, "grad_norm": 15.16864013671875, "learning_rate": 6.403398058252428e-05, "loss": 0.1965, "step": 7409 }, { "epoch": 71.84939759036145, "grad_norm": 10.977246284484863, "learning_rate": 6.402912621359224e-05, "loss": 0.2382, "step": 7410 }, { "epoch": 71.85903614457831, "grad_norm": 19.843936920166016, "learning_rate": 6.40242718446602e-05, "loss": 0.2782, "step": 7411 }, { "epoch": 71.86867469879518, "grad_norm": 7.058949947357178, "learning_rate": 6.401941747572817e-05, "loss": 0.2574, "step": 7412 }, { "epoch": 71.87831325301205, "grad_norm": 6.607901096343994, "learning_rate": 6.401456310679612e-05, "loss": 0.2454, "step": 7413 }, { "epoch": 71.88795180722892, "grad_norm": 10.672398567199707, "learning_rate": 6.400970873786408e-05, "loss": 0.3965, "step": 7414 }, { "epoch": 71.89759036144578, "grad_norm": 7.068012714385986, "learning_rate": 6.400485436893204e-05, "loss": 0.3306, "step": 7415 }, { "epoch": 71.90722891566266, "grad_norm": 6.255394458770752, "learning_rate": 6.400000000000001e-05, "loss": 0.1968, "step": 7416 }, { "epoch": 71.91686746987952, "grad_norm": 9.392524719238281, "learning_rate": 6.399514563106797e-05, "loss": 0.2039, "step": 7417 }, { "epoch": 71.92650602409638, "grad_norm": 7.6816630363464355, "learning_rate": 6.399029126213593e-05, "loss": 0.3502, "step": 7418 }, { "epoch": 71.93614457831325, "grad_norm": 7.854898452758789, "learning_rate": 6.398543689320388e-05, "loss": 0.3314, "step": 7419 }, { "epoch": 71.94578313253012, "grad_norm": 5.4778642654418945, "learning_rate": 6.398058252427184e-05, "loss": 0.2424, "step": 7420 }, { "epoch": 71.95542168674699, "grad_norm": 6.212159633636475, "learning_rate": 6.397572815533981e-05, "loss": 0.2518, "step": 7421 }, { "epoch": 71.96506024096385, "grad_norm": 11.387024879455566, "learning_rate": 6.397087378640777e-05, "loss": 0.5054, "step": 7422 }, { "epoch": 71.97469879518073, "grad_norm": 10.959242820739746, "learning_rate": 6.396601941747573e-05, "loss": 0.228, "step": 7423 }, { "epoch": 71.98433734939759, "grad_norm": 5.937744140625, "learning_rate": 6.396116504854369e-05, "loss": 0.275, "step": 7424 }, { "epoch": 71.99397590361446, "grad_norm": 8.763184547424316, "learning_rate": 6.395631067961166e-05, "loss": 0.3139, "step": 7425 }, { "epoch": 72.00963855421686, "grad_norm": 10.502995491027832, "learning_rate": 6.395145631067962e-05, "loss": 0.2832, "step": 7426 }, { "epoch": 72.01927710843374, "grad_norm": 9.109021186828613, "learning_rate": 6.394660194174757e-05, "loss": 0.3398, "step": 7427 }, { "epoch": 72.0289156626506, "grad_norm": 9.298200607299805, "learning_rate": 6.394174757281553e-05, "loss": 0.4073, "step": 7428 }, { "epoch": 72.03855421686747, "grad_norm": 4.940126895904541, "learning_rate": 6.39368932038835e-05, "loss": 0.2939, "step": 7429 }, { "epoch": 72.04819277108433, "grad_norm": 33.354736328125, "learning_rate": 6.393203883495146e-05, "loss": 0.3529, "step": 7430 }, { "epoch": 72.05783132530121, "grad_norm": 7.1235032081604, "learning_rate": 6.392718446601942e-05, "loss": 0.1373, "step": 7431 }, { "epoch": 72.06746987951807, "grad_norm": 19.986825942993164, "learning_rate": 6.392233009708738e-05, "loss": 0.3828, "step": 7432 }, { "epoch": 72.07710843373494, "grad_norm": 7.842347145080566, "learning_rate": 6.391747572815535e-05, "loss": 0.272, "step": 7433 }, { "epoch": 72.08674698795181, "grad_norm": 8.19542407989502, "learning_rate": 6.39126213592233e-05, "loss": 0.194, "step": 7434 }, { "epoch": 72.09638554216868, "grad_norm": 14.760595321655273, "learning_rate": 6.390776699029126e-05, "loss": 0.4206, "step": 7435 }, { "epoch": 72.10602409638554, "grad_norm": 7.4459381103515625, "learning_rate": 6.390291262135922e-05, "loss": 0.2664, "step": 7436 }, { "epoch": 72.1156626506024, "grad_norm": 9.3551025390625, "learning_rate": 6.389805825242718e-05, "loss": 0.3224, "step": 7437 }, { "epoch": 72.12530120481928, "grad_norm": 13.025769233703613, "learning_rate": 6.389320388349515e-05, "loss": 0.5675, "step": 7438 }, { "epoch": 72.13493975903614, "grad_norm": 8.927226066589355, "learning_rate": 6.388834951456311e-05, "loss": 0.2201, "step": 7439 }, { "epoch": 72.144578313253, "grad_norm": 14.392306327819824, "learning_rate": 6.388349514563106e-05, "loss": 0.3769, "step": 7440 }, { "epoch": 72.15421686746988, "grad_norm": 12.9969482421875, "learning_rate": 6.387864077669902e-05, "loss": 0.3109, "step": 7441 }, { "epoch": 72.16385542168675, "grad_norm": 7.523914337158203, "learning_rate": 6.3873786407767e-05, "loss": 0.4129, "step": 7442 }, { "epoch": 72.17349397590361, "grad_norm": 8.156290054321289, "learning_rate": 6.386893203883495e-05, "loss": 0.3264, "step": 7443 }, { "epoch": 72.18313253012049, "grad_norm": 10.181373596191406, "learning_rate": 6.386407766990291e-05, "loss": 0.3815, "step": 7444 }, { "epoch": 72.19277108433735, "grad_norm": 7.248602867126465, "learning_rate": 6.385922330097087e-05, "loss": 0.3181, "step": 7445 }, { "epoch": 72.20240963855422, "grad_norm": 8.669292449951172, "learning_rate": 6.385436893203884e-05, "loss": 0.3713, "step": 7446 }, { "epoch": 72.21204819277108, "grad_norm": 8.289119720458984, "learning_rate": 6.38495145631068e-05, "loss": 0.367, "step": 7447 }, { "epoch": 72.22168674698796, "grad_norm": 13.702099800109863, "learning_rate": 6.384466019417477e-05, "loss": 0.3266, "step": 7448 }, { "epoch": 72.23132530120482, "grad_norm": 8.920489311218262, "learning_rate": 6.383980582524273e-05, "loss": 0.4325, "step": 7449 }, { "epoch": 72.24096385542168, "grad_norm": 10.141921997070312, "learning_rate": 6.383495145631068e-05, "loss": 0.2556, "step": 7450 }, { "epoch": 72.25060240963856, "grad_norm": 4.249011993408203, "learning_rate": 6.383009708737865e-05, "loss": 0.2212, "step": 7451 }, { "epoch": 72.26024096385542, "grad_norm": 7.9086432456970215, "learning_rate": 6.382524271844661e-05, "loss": 0.3553, "step": 7452 }, { "epoch": 72.26987951807229, "grad_norm": 10.792892456054688, "learning_rate": 6.382038834951457e-05, "loss": 0.3235, "step": 7453 }, { "epoch": 72.27951807228915, "grad_norm": 7.770871639251709, "learning_rate": 6.381553398058253e-05, "loss": 0.3829, "step": 7454 }, { "epoch": 72.28915662650603, "grad_norm": 10.25910472869873, "learning_rate": 6.38106796116505e-05, "loss": 0.4014, "step": 7455 }, { "epoch": 72.29879518072289, "grad_norm": 10.237753868103027, "learning_rate": 6.380582524271846e-05, "loss": 0.2416, "step": 7456 }, { "epoch": 72.30843373493975, "grad_norm": 10.591200828552246, "learning_rate": 6.380097087378641e-05, "loss": 0.3328, "step": 7457 }, { "epoch": 72.31807228915663, "grad_norm": 14.221961975097656, "learning_rate": 6.379611650485437e-05, "loss": 0.6309, "step": 7458 }, { "epoch": 72.3277108433735, "grad_norm": 7.002182960510254, "learning_rate": 6.379126213592233e-05, "loss": 0.3622, "step": 7459 }, { "epoch": 72.33734939759036, "grad_norm": 9.799590110778809, "learning_rate": 6.37864077669903e-05, "loss": 0.1316, "step": 7460 }, { "epoch": 72.34698795180722, "grad_norm": 5.650688648223877, "learning_rate": 6.378155339805826e-05, "loss": 0.3466, "step": 7461 }, { "epoch": 72.3566265060241, "grad_norm": 21.353042602539062, "learning_rate": 6.377669902912622e-05, "loss": 0.3972, "step": 7462 }, { "epoch": 72.36626506024096, "grad_norm": 3.943783760070801, "learning_rate": 6.377184466019417e-05, "loss": 0.1332, "step": 7463 }, { "epoch": 72.37590361445783, "grad_norm": 6.0886054039001465, "learning_rate": 6.376699029126215e-05, "loss": 0.1752, "step": 7464 }, { "epoch": 72.3855421686747, "grad_norm": 14.409316062927246, "learning_rate": 6.37621359223301e-05, "loss": 0.4261, "step": 7465 }, { "epoch": 72.39518072289157, "grad_norm": 5.746764659881592, "learning_rate": 6.375728155339806e-05, "loss": 0.2079, "step": 7466 }, { "epoch": 72.40481927710843, "grad_norm": 17.438756942749023, "learning_rate": 6.375242718446602e-05, "loss": 0.4211, "step": 7467 }, { "epoch": 72.41445783132531, "grad_norm": 15.481274604797363, "learning_rate": 6.374757281553399e-05, "loss": 0.3904, "step": 7468 }, { "epoch": 72.42409638554217, "grad_norm": 5.701845169067383, "learning_rate": 6.374271844660195e-05, "loss": 0.4121, "step": 7469 }, { "epoch": 72.43373493975903, "grad_norm": 5.199228286743164, "learning_rate": 6.37378640776699e-05, "loss": 0.1857, "step": 7470 }, { "epoch": 72.4433734939759, "grad_norm": 8.989158630371094, "learning_rate": 6.373300970873786e-05, "loss": 0.2919, "step": 7471 }, { "epoch": 72.45301204819278, "grad_norm": 5.984496593475342, "learning_rate": 6.372815533980583e-05, "loss": 0.2177, "step": 7472 }, { "epoch": 72.46265060240964, "grad_norm": 10.83534049987793, "learning_rate": 6.372330097087379e-05, "loss": 0.3387, "step": 7473 }, { "epoch": 72.4722891566265, "grad_norm": 8.412758827209473, "learning_rate": 6.371844660194175e-05, "loss": 0.298, "step": 7474 }, { "epoch": 72.48192771084338, "grad_norm": 7.920607089996338, "learning_rate": 6.371359223300971e-05, "loss": 0.2707, "step": 7475 }, { "epoch": 72.49156626506024, "grad_norm": 8.795783996582031, "learning_rate": 6.370873786407767e-05, "loss": 0.4304, "step": 7476 }, { "epoch": 72.5012048192771, "grad_norm": 8.780950546264648, "learning_rate": 6.370388349514564e-05, "loss": 0.298, "step": 7477 }, { "epoch": 72.51084337349397, "grad_norm": 10.307188987731934, "learning_rate": 6.36990291262136e-05, "loss": 0.3322, "step": 7478 }, { "epoch": 72.52048192771085, "grad_norm": 4.163301944732666, "learning_rate": 6.369417475728155e-05, "loss": 0.2133, "step": 7479 }, { "epoch": 72.53012048192771, "grad_norm": 11.665488243103027, "learning_rate": 6.368932038834951e-05, "loss": 0.2272, "step": 7480 }, { "epoch": 72.53975903614457, "grad_norm": 5.082964897155762, "learning_rate": 6.368446601941748e-05, "loss": 0.2168, "step": 7481 }, { "epoch": 72.54939759036145, "grad_norm": 6.0027947425842285, "learning_rate": 6.367961165048544e-05, "loss": 0.3325, "step": 7482 }, { "epoch": 72.55903614457831, "grad_norm": 6.799987316131592, "learning_rate": 6.36747572815534e-05, "loss": 0.2083, "step": 7483 }, { "epoch": 72.56867469879518, "grad_norm": 10.122349739074707, "learning_rate": 6.366990291262136e-05, "loss": 0.3489, "step": 7484 }, { "epoch": 72.57831325301204, "grad_norm": 8.736347198486328, "learning_rate": 6.366504854368933e-05, "loss": 0.3747, "step": 7485 }, { "epoch": 72.58795180722892, "grad_norm": 4.469943523406982, "learning_rate": 6.366019417475728e-05, "loss": 0.2021, "step": 7486 }, { "epoch": 72.59759036144578, "grad_norm": 13.354405403137207, "learning_rate": 6.365533980582524e-05, "loss": 0.2754, "step": 7487 }, { "epoch": 72.60722891566265, "grad_norm": 6.765976905822754, "learning_rate": 6.36504854368932e-05, "loss": 0.414, "step": 7488 }, { "epoch": 72.61686746987952, "grad_norm": 7.559991836547852, "learning_rate": 6.364563106796117e-05, "loss": 0.3047, "step": 7489 }, { "epoch": 72.62650602409639, "grad_norm": 11.071096420288086, "learning_rate": 6.364077669902913e-05, "loss": 0.2408, "step": 7490 }, { "epoch": 72.63614457831325, "grad_norm": 7.879492282867432, "learning_rate": 6.363592233009709e-05, "loss": 0.4042, "step": 7491 }, { "epoch": 72.64578313253013, "grad_norm": 6.717682361602783, "learning_rate": 6.363106796116504e-05, "loss": 0.462, "step": 7492 }, { "epoch": 72.65542168674699, "grad_norm": 10.557415008544922, "learning_rate": 6.3626213592233e-05, "loss": 0.3258, "step": 7493 }, { "epoch": 72.66506024096385, "grad_norm": 5.300591468811035, "learning_rate": 6.362135922330097e-05, "loss": 0.2517, "step": 7494 }, { "epoch": 72.67469879518072, "grad_norm": 14.405998229980469, "learning_rate": 6.361650485436893e-05, "loss": 0.4009, "step": 7495 }, { "epoch": 72.6843373493976, "grad_norm": 8.432844161987305, "learning_rate": 6.361165048543689e-05, "loss": 0.1011, "step": 7496 }, { "epoch": 72.69397590361446, "grad_norm": 12.0535249710083, "learning_rate": 6.360679611650485e-05, "loss": 0.2568, "step": 7497 }, { "epoch": 72.70361445783132, "grad_norm": 11.044111251831055, "learning_rate": 6.360194174757282e-05, "loss": 0.3508, "step": 7498 }, { "epoch": 72.7132530120482, "grad_norm": 22.753665924072266, "learning_rate": 6.359708737864079e-05, "loss": 0.1444, "step": 7499 }, { "epoch": 72.72289156626506, "grad_norm": 9.736742973327637, "learning_rate": 6.359223300970875e-05, "loss": 0.3818, "step": 7500 }, { "epoch": 72.73253012048193, "grad_norm": 6.808864116668701, "learning_rate": 6.35873786407767e-05, "loss": 0.351, "step": 7501 }, { "epoch": 72.74216867469879, "grad_norm": 11.95871353149414, "learning_rate": 6.358252427184466e-05, "loss": 0.3095, "step": 7502 }, { "epoch": 72.75180722891567, "grad_norm": 6.713091850280762, "learning_rate": 6.357766990291263e-05, "loss": 0.203, "step": 7503 }, { "epoch": 72.76144578313253, "grad_norm": 5.771105766296387, "learning_rate": 6.357281553398059e-05, "loss": 0.2769, "step": 7504 }, { "epoch": 72.7710843373494, "grad_norm": 7.603653430938721, "learning_rate": 6.356796116504855e-05, "loss": 0.1892, "step": 7505 }, { "epoch": 72.78072289156627, "grad_norm": 6.591773986816406, "learning_rate": 6.356310679611651e-05, "loss": 0.2681, "step": 7506 }, { "epoch": 72.79036144578313, "grad_norm": 3.816061496734619, "learning_rate": 6.355825242718448e-05, "loss": 0.0803, "step": 7507 }, { "epoch": 72.8, "grad_norm": 12.114439964294434, "learning_rate": 6.355339805825244e-05, "loss": 0.1649, "step": 7508 }, { "epoch": 72.80963855421686, "grad_norm": 13.849995613098145, "learning_rate": 6.35485436893204e-05, "loss": 0.4317, "step": 7509 }, { "epoch": 72.81927710843374, "grad_norm": 14.141149520874023, "learning_rate": 6.354368932038835e-05, "loss": 0.3542, "step": 7510 }, { "epoch": 72.8289156626506, "grad_norm": 14.167909622192383, "learning_rate": 6.353883495145632e-05, "loss": 0.2884, "step": 7511 }, { "epoch": 72.83855421686746, "grad_norm": 9.079463005065918, "learning_rate": 6.353398058252428e-05, "loss": 0.175, "step": 7512 }, { "epoch": 72.84819277108434, "grad_norm": 7.480349063873291, "learning_rate": 6.352912621359224e-05, "loss": 0.3542, "step": 7513 }, { "epoch": 72.8578313253012, "grad_norm": 6.052505016326904, "learning_rate": 6.35242718446602e-05, "loss": 0.3106, "step": 7514 }, { "epoch": 72.86746987951807, "grad_norm": 7.749671459197998, "learning_rate": 6.351941747572815e-05, "loss": 0.1624, "step": 7515 }, { "epoch": 72.87710843373495, "grad_norm": 4.073670387268066, "learning_rate": 6.351456310679613e-05, "loss": 0.1726, "step": 7516 }, { "epoch": 72.88674698795181, "grad_norm": 5.487112045288086, "learning_rate": 6.350970873786408e-05, "loss": 0.2376, "step": 7517 }, { "epoch": 72.89638554216867, "grad_norm": 7.683554172515869, "learning_rate": 6.350485436893204e-05, "loss": 0.2923, "step": 7518 }, { "epoch": 72.90602409638554, "grad_norm": 11.007654190063477, "learning_rate": 6.35e-05, "loss": 0.4729, "step": 7519 }, { "epoch": 72.91566265060241, "grad_norm": 11.950287818908691, "learning_rate": 6.349514563106797e-05, "loss": 0.453, "step": 7520 }, { "epoch": 72.92530120481928, "grad_norm": 8.515439987182617, "learning_rate": 6.349029126213593e-05, "loss": 0.2308, "step": 7521 }, { "epoch": 72.93493975903614, "grad_norm": 7.257815837860107, "learning_rate": 6.348543689320389e-05, "loss": 0.4219, "step": 7522 }, { "epoch": 72.94457831325302, "grad_norm": 6.605511665344238, "learning_rate": 6.348058252427184e-05, "loss": 0.2163, "step": 7523 }, { "epoch": 72.95421686746988, "grad_norm": 9.616215705871582, "learning_rate": 6.347572815533981e-05, "loss": 0.3259, "step": 7524 }, { "epoch": 72.96385542168674, "grad_norm": 7.674861431121826, "learning_rate": 6.347087378640777e-05, "loss": 0.1778, "step": 7525 }, { "epoch": 72.97349397590361, "grad_norm": 11.418124198913574, "learning_rate": 6.346601941747573e-05, "loss": 0.3728, "step": 7526 }, { "epoch": 72.98313253012049, "grad_norm": 3.1472365856170654, "learning_rate": 6.346116504854369e-05, "loss": 0.0947, "step": 7527 }, { "epoch": 72.99277108433735, "grad_norm": 12.054003715515137, "learning_rate": 6.345631067961166e-05, "loss": 0.4893, "step": 7528 }, { "epoch": 73.00843373493976, "grad_norm": 4.5426788330078125, "learning_rate": 6.345145631067962e-05, "loss": 0.3368, "step": 7529 }, { "epoch": 73.01807228915662, "grad_norm": 8.796182632446289, "learning_rate": 6.344660194174757e-05, "loss": 0.218, "step": 7530 }, { "epoch": 73.0277108433735, "grad_norm": 101.5186767578125, "learning_rate": 6.344174757281553e-05, "loss": 0.2787, "step": 7531 }, { "epoch": 73.03734939759036, "grad_norm": 26.251239776611328, "learning_rate": 6.343689320388349e-05, "loss": 0.3157, "step": 7532 }, { "epoch": 73.04698795180722, "grad_norm": 32.08222198486328, "learning_rate": 6.343203883495146e-05, "loss": 0.3901, "step": 7533 }, { "epoch": 73.0566265060241, "grad_norm": 15.891142845153809, "learning_rate": 6.342718446601942e-05, "loss": 0.3131, "step": 7534 }, { "epoch": 73.06626506024097, "grad_norm": 7.318151473999023, "learning_rate": 6.342233009708738e-05, "loss": 0.4403, "step": 7535 }, { "epoch": 73.07590361445783, "grad_norm": 38.157066345214844, "learning_rate": 6.341747572815533e-05, "loss": 0.3025, "step": 7536 }, { "epoch": 73.08554216867469, "grad_norm": 2.903658628463745, "learning_rate": 6.34126213592233e-05, "loss": 0.3742, "step": 7537 }, { "epoch": 73.09518072289157, "grad_norm": 11.911237716674805, "learning_rate": 6.340776699029126e-05, "loss": 0.2511, "step": 7538 }, { "epoch": 73.10481927710843, "grad_norm": 22.097959518432617, "learning_rate": 6.340291262135922e-05, "loss": 0.4407, "step": 7539 }, { "epoch": 73.1144578313253, "grad_norm": 35.85429382324219, "learning_rate": 6.339805825242718e-05, "loss": 0.3109, "step": 7540 }, { "epoch": 73.12409638554217, "grad_norm": 40.10539627075195, "learning_rate": 6.339320388349515e-05, "loss": 0.2225, "step": 7541 }, { "epoch": 73.13373493975904, "grad_norm": 29.00745964050293, "learning_rate": 6.338834951456311e-05, "loss": 0.4061, "step": 7542 }, { "epoch": 73.1433734939759, "grad_norm": 30.259580612182617, "learning_rate": 6.338349514563107e-05, "loss": 0.3583, "step": 7543 }, { "epoch": 73.15301204819278, "grad_norm": 4.957788467407227, "learning_rate": 6.337864077669902e-05, "loss": 0.3749, "step": 7544 }, { "epoch": 73.16265060240964, "grad_norm": 21.522071838378906, "learning_rate": 6.3373786407767e-05, "loss": 0.3948, "step": 7545 }, { "epoch": 73.1722891566265, "grad_norm": 24.741580963134766, "learning_rate": 6.336893203883495e-05, "loss": 0.6508, "step": 7546 }, { "epoch": 73.18192771084337, "grad_norm": 11.945276260375977, "learning_rate": 6.336407766990291e-05, "loss": 0.3471, "step": 7547 }, { "epoch": 73.19156626506025, "grad_norm": 18.37032699584961, "learning_rate": 6.335922330097087e-05, "loss": 0.3767, "step": 7548 }, { "epoch": 73.20120481927711, "grad_norm": 15.145791053771973, "learning_rate": 6.335436893203884e-05, "loss": 0.2415, "step": 7549 }, { "epoch": 73.21084337349397, "grad_norm": 27.26062774658203, "learning_rate": 6.334951456310681e-05, "loss": 0.192, "step": 7550 }, { "epoch": 73.22048192771085, "grad_norm": 19.054702758789062, "learning_rate": 6.334466019417477e-05, "loss": 0.2599, "step": 7551 }, { "epoch": 73.23012048192771, "grad_norm": 15.401594161987305, "learning_rate": 6.333980582524273e-05, "loss": 0.215, "step": 7552 }, { "epoch": 73.23975903614458, "grad_norm": 8.456254959106445, "learning_rate": 6.333495145631068e-05, "loss": 0.1406, "step": 7553 }, { "epoch": 73.24939759036144, "grad_norm": 5.5859551429748535, "learning_rate": 6.333009708737864e-05, "loss": 0.2296, "step": 7554 }, { "epoch": 73.25903614457832, "grad_norm": 22.448286056518555, "learning_rate": 6.332524271844661e-05, "loss": 0.4772, "step": 7555 }, { "epoch": 73.26867469879518, "grad_norm": 62.86838912963867, "learning_rate": 6.332038834951457e-05, "loss": 0.2762, "step": 7556 }, { "epoch": 73.27831325301204, "grad_norm": 2.9751601219177246, "learning_rate": 6.331553398058253e-05, "loss": 0.1829, "step": 7557 }, { "epoch": 73.28795180722892, "grad_norm": 22.474369049072266, "learning_rate": 6.331067961165049e-05, "loss": 0.4148, "step": 7558 }, { "epoch": 73.29759036144578, "grad_norm": 30.99778938293457, "learning_rate": 6.330582524271846e-05, "loss": 0.4255, "step": 7559 }, { "epoch": 73.30722891566265, "grad_norm": 10.248549461364746, "learning_rate": 6.330097087378642e-05, "loss": 0.3979, "step": 7560 }, { "epoch": 73.31686746987951, "grad_norm": 4.331265926361084, "learning_rate": 6.329611650485437e-05, "loss": 0.3537, "step": 7561 }, { "epoch": 73.32650602409639, "grad_norm": 1.2604910135269165, "learning_rate": 6.329126213592233e-05, "loss": 0.4168, "step": 7562 }, { "epoch": 73.33614457831325, "grad_norm": 42.444480895996094, "learning_rate": 6.32864077669903e-05, "loss": 0.3549, "step": 7563 }, { "epoch": 73.34578313253012, "grad_norm": 12.57491397857666, "learning_rate": 6.328155339805826e-05, "loss": 0.3049, "step": 7564 }, { "epoch": 73.355421686747, "grad_norm": 16.30405616760254, "learning_rate": 6.327669902912622e-05, "loss": 0.312, "step": 7565 }, { "epoch": 73.36506024096386, "grad_norm": 36.26911544799805, "learning_rate": 6.327184466019418e-05, "loss": 0.3466, "step": 7566 }, { "epoch": 73.37469879518072, "grad_norm": 16.309728622436523, "learning_rate": 6.326699029126215e-05, "loss": 0.215, "step": 7567 }, { "epoch": 73.38433734939758, "grad_norm": 23.40841293334961, "learning_rate": 6.32621359223301e-05, "loss": 0.2613, "step": 7568 }, { "epoch": 73.39397590361446, "grad_norm": 28.22943115234375, "learning_rate": 6.325728155339806e-05, "loss": 0.7679, "step": 7569 }, { "epoch": 73.40361445783132, "grad_norm": 8.758723258972168, "learning_rate": 6.325242718446602e-05, "loss": 0.1774, "step": 7570 }, { "epoch": 73.41325301204819, "grad_norm": 31.170257568359375, "learning_rate": 6.324757281553398e-05, "loss": 0.284, "step": 7571 }, { "epoch": 73.42289156626506, "grad_norm": 16.841402053833008, "learning_rate": 6.324271844660195e-05, "loss": 0.2971, "step": 7572 }, { "epoch": 73.43253012048193, "grad_norm": 20.938705444335938, "learning_rate": 6.323786407766991e-05, "loss": 0.3523, "step": 7573 }, { "epoch": 73.44216867469879, "grad_norm": 18.7911319732666, "learning_rate": 6.323300970873786e-05, "loss": 0.3942, "step": 7574 }, { "epoch": 73.45180722891567, "grad_norm": 6.4134087562561035, "learning_rate": 6.322815533980582e-05, "loss": 0.3777, "step": 7575 }, { "epoch": 73.46144578313253, "grad_norm": 4.8862528800964355, "learning_rate": 6.32233009708738e-05, "loss": 0.1876, "step": 7576 }, { "epoch": 73.4710843373494, "grad_norm": 33.46821212768555, "learning_rate": 6.321844660194175e-05, "loss": 0.1949, "step": 7577 }, { "epoch": 73.48072289156626, "grad_norm": 11.437289237976074, "learning_rate": 6.321359223300971e-05, "loss": 0.2743, "step": 7578 }, { "epoch": 73.49036144578314, "grad_norm": 9.860628128051758, "learning_rate": 6.320873786407767e-05, "loss": 0.3388, "step": 7579 }, { "epoch": 73.5, "grad_norm": 16.38128089904785, "learning_rate": 6.320388349514564e-05, "loss": 0.5525, "step": 7580 }, { "epoch": 73.50963855421686, "grad_norm": 29.646888732910156, "learning_rate": 6.31990291262136e-05, "loss": 0.4961, "step": 7581 }, { "epoch": 73.51927710843374, "grad_norm": 9.456090927124023, "learning_rate": 6.319417475728155e-05, "loss": 0.388, "step": 7582 }, { "epoch": 73.5289156626506, "grad_norm": 9.487298965454102, "learning_rate": 6.318932038834951e-05, "loss": 0.3367, "step": 7583 }, { "epoch": 73.53855421686747, "grad_norm": 13.3428316116333, "learning_rate": 6.318446601941748e-05, "loss": 0.4948, "step": 7584 }, { "epoch": 73.54819277108433, "grad_norm": 15.558694839477539, "learning_rate": 6.317961165048544e-05, "loss": 0.1568, "step": 7585 }, { "epoch": 73.55783132530121, "grad_norm": 16.410799026489258, "learning_rate": 6.31747572815534e-05, "loss": 0.18, "step": 7586 }, { "epoch": 73.56746987951807, "grad_norm": 20.43592071533203, "learning_rate": 6.316990291262136e-05, "loss": 0.285, "step": 7587 }, { "epoch": 73.57710843373494, "grad_norm": 14.65477466583252, "learning_rate": 6.316504854368931e-05, "loss": 0.3727, "step": 7588 }, { "epoch": 73.58674698795181, "grad_norm": 14.445898056030273, "learning_rate": 6.316019417475729e-05, "loss": 0.2357, "step": 7589 }, { "epoch": 73.59638554216868, "grad_norm": 29.750473022460938, "learning_rate": 6.315533980582524e-05, "loss": 0.2872, "step": 7590 }, { "epoch": 73.60602409638554, "grad_norm": 51.471771240234375, "learning_rate": 6.31504854368932e-05, "loss": 0.1653, "step": 7591 }, { "epoch": 73.61566265060242, "grad_norm": 22.293603897094727, "learning_rate": 6.314563106796116e-05, "loss": 0.2633, "step": 7592 }, { "epoch": 73.62530120481928, "grad_norm": 6.193510055541992, "learning_rate": 6.314077669902913e-05, "loss": 0.3167, "step": 7593 }, { "epoch": 73.63493975903614, "grad_norm": 13.237101554870605, "learning_rate": 6.313592233009709e-05, "loss": 0.2289, "step": 7594 }, { "epoch": 73.644578313253, "grad_norm": 25.590843200683594, "learning_rate": 6.313106796116505e-05, "loss": 0.198, "step": 7595 }, { "epoch": 73.65421686746988, "grad_norm": 28.23094367980957, "learning_rate": 6.3126213592233e-05, "loss": 0.4124, "step": 7596 }, { "epoch": 73.66385542168675, "grad_norm": 45.22690200805664, "learning_rate": 6.312135922330097e-05, "loss": 0.2582, "step": 7597 }, { "epoch": 73.67349397590361, "grad_norm": 10.734393119812012, "learning_rate": 6.311650485436893e-05, "loss": 0.2542, "step": 7598 }, { "epoch": 73.68313253012049, "grad_norm": 27.068321228027344, "learning_rate": 6.31116504854369e-05, "loss": 0.3742, "step": 7599 }, { "epoch": 73.69277108433735, "grad_norm": 0.8725348711013794, "learning_rate": 6.310679611650486e-05, "loss": 0.1471, "step": 7600 }, { "epoch": 73.70240963855422, "grad_norm": 14.313016891479492, "learning_rate": 6.310194174757282e-05, "loss": 0.2059, "step": 7601 }, { "epoch": 73.71204819277108, "grad_norm": 2.8761465549468994, "learning_rate": 6.309708737864079e-05, "loss": 0.2494, "step": 7602 }, { "epoch": 73.72168674698796, "grad_norm": 3.255800247192383, "learning_rate": 6.309223300970875e-05, "loss": 0.2555, "step": 7603 }, { "epoch": 73.73132530120482, "grad_norm": 2.2095746994018555, "learning_rate": 6.30873786407767e-05, "loss": 0.3299, "step": 7604 }, { "epoch": 73.74096385542168, "grad_norm": 12.723995208740234, "learning_rate": 6.308252427184466e-05, "loss": 0.3017, "step": 7605 }, { "epoch": 73.75060240963856, "grad_norm": 21.0331974029541, "learning_rate": 6.307766990291262e-05, "loss": 0.2957, "step": 7606 }, { "epoch": 73.76024096385542, "grad_norm": 37.310218811035156, "learning_rate": 6.307281553398059e-05, "loss": 0.4201, "step": 7607 }, { "epoch": 73.76987951807229, "grad_norm": 7.7636590003967285, "learning_rate": 6.306796116504855e-05, "loss": 0.265, "step": 7608 }, { "epoch": 73.77951807228915, "grad_norm": 3.4533751010894775, "learning_rate": 6.306310679611651e-05, "loss": 0.3488, "step": 7609 }, { "epoch": 73.78915662650603, "grad_norm": 10.871837615966797, "learning_rate": 6.305825242718447e-05, "loss": 0.3476, "step": 7610 }, { "epoch": 73.79879518072289, "grad_norm": 14.562366485595703, "learning_rate": 6.305339805825244e-05, "loss": 0.2714, "step": 7611 }, { "epoch": 73.80843373493975, "grad_norm": 9.995960235595703, "learning_rate": 6.30485436893204e-05, "loss": 0.4156, "step": 7612 }, { "epoch": 73.81807228915663, "grad_norm": 2.228496789932251, "learning_rate": 6.304368932038835e-05, "loss": 0.2303, "step": 7613 }, { "epoch": 73.8277108433735, "grad_norm": 41.3006591796875, "learning_rate": 6.303883495145631e-05, "loss": 0.3302, "step": 7614 }, { "epoch": 73.83734939759036, "grad_norm": 24.578651428222656, "learning_rate": 6.303398058252428e-05, "loss": 0.3095, "step": 7615 }, { "epoch": 73.84698795180722, "grad_norm": 37.26167297363281, "learning_rate": 6.302912621359224e-05, "loss": 0.2243, "step": 7616 }, { "epoch": 73.8566265060241, "grad_norm": 5.251718521118164, "learning_rate": 6.30242718446602e-05, "loss": 0.1344, "step": 7617 }, { "epoch": 73.86626506024096, "grad_norm": 17.216150283813477, "learning_rate": 6.301941747572816e-05, "loss": 0.36, "step": 7618 }, { "epoch": 73.87590361445783, "grad_norm": 1.4991302490234375, "learning_rate": 6.301456310679613e-05, "loss": 0.2615, "step": 7619 }, { "epoch": 73.8855421686747, "grad_norm": 13.276453018188477, "learning_rate": 6.300970873786408e-05, "loss": 0.3167, "step": 7620 }, { "epoch": 73.89518072289157, "grad_norm": 71.28157806396484, "learning_rate": 6.300485436893204e-05, "loss": 0.1913, "step": 7621 }, { "epoch": 73.90481927710843, "grad_norm": 2.829674005508423, "learning_rate": 6.3e-05, "loss": 0.4667, "step": 7622 }, { "epoch": 73.91445783132531, "grad_norm": 14.245857238769531, "learning_rate": 6.299514563106796e-05, "loss": 0.5245, "step": 7623 }, { "epoch": 73.92409638554217, "grad_norm": 52.61729431152344, "learning_rate": 6.299029126213593e-05, "loss": 0.2256, "step": 7624 }, { "epoch": 73.93373493975903, "grad_norm": 10.35590934753418, "learning_rate": 6.298543689320389e-05, "loss": 0.2699, "step": 7625 }, { "epoch": 73.9433734939759, "grad_norm": 3.7863779067993164, "learning_rate": 6.298058252427184e-05, "loss": 0.1231, "step": 7626 }, { "epoch": 73.95301204819278, "grad_norm": 10.823022842407227, "learning_rate": 6.29757281553398e-05, "loss": 0.2208, "step": 7627 }, { "epoch": 73.96265060240964, "grad_norm": 2.594604969024658, "learning_rate": 6.297087378640777e-05, "loss": 0.4786, "step": 7628 }, { "epoch": 73.9722891566265, "grad_norm": 12.882964134216309, "learning_rate": 6.296601941747573e-05, "loss": 0.231, "step": 7629 }, { "epoch": 73.98192771084338, "grad_norm": 21.747098922729492, "learning_rate": 6.296116504854369e-05, "loss": 0.4426, "step": 7630 }, { "epoch": 73.99156626506024, "grad_norm": 4.7779645919799805, "learning_rate": 6.295631067961165e-05, "loss": 0.2851, "step": 7631 }, { "epoch": 74.00722891566265, "grad_norm": 24.875568389892578, "learning_rate": 6.295145631067962e-05, "loss": 0.1972, "step": 7632 }, { "epoch": 74.01686746987951, "grad_norm": 14.120491027832031, "learning_rate": 6.294660194174758e-05, "loss": 0.2504, "step": 7633 }, { "epoch": 74.02650602409639, "grad_norm": 45.83795928955078, "learning_rate": 6.294174757281553e-05, "loss": 0.3459, "step": 7634 }, { "epoch": 74.03614457831326, "grad_norm": 24.61786651611328, "learning_rate": 6.293689320388349e-05, "loss": 0.1808, "step": 7635 }, { "epoch": 74.04578313253012, "grad_norm": 2.575287103652954, "learning_rate": 6.293203883495146e-05, "loss": 0.2808, "step": 7636 }, { "epoch": 74.05542168674698, "grad_norm": 11.047318458557129, "learning_rate": 6.292718446601942e-05, "loss": 0.3893, "step": 7637 }, { "epoch": 74.06506024096386, "grad_norm": 21.409990310668945, "learning_rate": 6.292233009708738e-05, "loss": 0.5034, "step": 7638 }, { "epoch": 74.07469879518072, "grad_norm": 6.108425140380859, "learning_rate": 6.291747572815534e-05, "loss": 0.2563, "step": 7639 }, { "epoch": 74.08433734939759, "grad_norm": 7.496042728424072, "learning_rate": 6.29126213592233e-05, "loss": 0.2435, "step": 7640 }, { "epoch": 74.09397590361446, "grad_norm": 11.549663543701172, "learning_rate": 6.290776699029127e-05, "loss": 0.234, "step": 7641 }, { "epoch": 74.10361445783133, "grad_norm": 27.66507339477539, "learning_rate": 6.290291262135922e-05, "loss": 0.3935, "step": 7642 }, { "epoch": 74.11325301204819, "grad_norm": 13.5260648727417, "learning_rate": 6.289805825242718e-05, "loss": 0.1964, "step": 7643 }, { "epoch": 74.12289156626505, "grad_norm": 19.86374282836914, "learning_rate": 6.289320388349514e-05, "loss": 0.2507, "step": 7644 }, { "epoch": 74.13253012048193, "grad_norm": 11.513385772705078, "learning_rate": 6.288834951456311e-05, "loss": 0.2005, "step": 7645 }, { "epoch": 74.1421686746988, "grad_norm": 13.483284950256348, "learning_rate": 6.288349514563107e-05, "loss": 0.3452, "step": 7646 }, { "epoch": 74.15180722891566, "grad_norm": 11.16487979888916, "learning_rate": 6.287864077669903e-05, "loss": 0.4237, "step": 7647 }, { "epoch": 74.16144578313254, "grad_norm": 2.613826036453247, "learning_rate": 6.287378640776698e-05, "loss": 0.1445, "step": 7648 }, { "epoch": 74.1710843373494, "grad_norm": 15.01909351348877, "learning_rate": 6.286893203883495e-05, "loss": 0.1293, "step": 7649 }, { "epoch": 74.18072289156626, "grad_norm": 15.11666202545166, "learning_rate": 6.286407766990293e-05, "loss": 0.4927, "step": 7650 }, { "epoch": 74.19036144578314, "grad_norm": 24.159936904907227, "learning_rate": 6.285922330097088e-05, "loss": 0.2002, "step": 7651 }, { "epoch": 74.2, "grad_norm": 4.654689788818359, "learning_rate": 6.285436893203884e-05, "loss": 0.4172, "step": 7652 }, { "epoch": 74.20963855421687, "grad_norm": 10.614511489868164, "learning_rate": 6.28495145631068e-05, "loss": 0.2389, "step": 7653 }, { "epoch": 74.21927710843373, "grad_norm": 14.731425285339355, "learning_rate": 6.284466019417477e-05, "loss": 0.4705, "step": 7654 }, { "epoch": 74.2289156626506, "grad_norm": 21.121936798095703, "learning_rate": 6.283980582524273e-05, "loss": 0.2794, "step": 7655 }, { "epoch": 74.23855421686747, "grad_norm": 7.387024879455566, "learning_rate": 6.283495145631069e-05, "loss": 0.2185, "step": 7656 }, { "epoch": 74.24819277108433, "grad_norm": 20.30878257751465, "learning_rate": 6.283009708737864e-05, "loss": 0.3102, "step": 7657 }, { "epoch": 74.25783132530121, "grad_norm": 10.836851119995117, "learning_rate": 6.282524271844661e-05, "loss": 0.2951, "step": 7658 }, { "epoch": 74.26746987951807, "grad_norm": 35.84870147705078, "learning_rate": 6.282038834951457e-05, "loss": 0.4004, "step": 7659 }, { "epoch": 74.27710843373494, "grad_norm": 18.996625900268555, "learning_rate": 6.281553398058253e-05, "loss": 0.4791, "step": 7660 }, { "epoch": 74.2867469879518, "grad_norm": 13.559199333190918, "learning_rate": 6.281067961165049e-05, "loss": 0.4903, "step": 7661 }, { "epoch": 74.29638554216868, "grad_norm": 10.905721664428711, "learning_rate": 6.280582524271845e-05, "loss": 0.2837, "step": 7662 }, { "epoch": 74.30602409638554, "grad_norm": 12.462872505187988, "learning_rate": 6.280097087378642e-05, "loss": 0.25, "step": 7663 }, { "epoch": 74.3156626506024, "grad_norm": 9.557612419128418, "learning_rate": 6.279611650485437e-05, "loss": 0.1867, "step": 7664 }, { "epoch": 74.32530120481928, "grad_norm": 8.256611824035645, "learning_rate": 6.279126213592233e-05, "loss": 0.1925, "step": 7665 }, { "epoch": 74.33493975903615, "grad_norm": 6.925449371337891, "learning_rate": 6.278640776699029e-05, "loss": 0.2171, "step": 7666 }, { "epoch": 74.34457831325301, "grad_norm": 11.559043884277344, "learning_rate": 6.278155339805826e-05, "loss": 0.2624, "step": 7667 }, { "epoch": 74.35421686746987, "grad_norm": 24.088102340698242, "learning_rate": 6.277669902912622e-05, "loss": 0.5199, "step": 7668 }, { "epoch": 74.36385542168675, "grad_norm": 79.17728424072266, "learning_rate": 6.277184466019418e-05, "loss": 0.1739, "step": 7669 }, { "epoch": 74.37349397590361, "grad_norm": 21.688705444335938, "learning_rate": 6.276699029126213e-05, "loss": 0.4277, "step": 7670 }, { "epoch": 74.38313253012048, "grad_norm": 10.55517578125, "learning_rate": 6.27621359223301e-05, "loss": 0.2946, "step": 7671 }, { "epoch": 74.39277108433735, "grad_norm": 13.87467098236084, "learning_rate": 6.275728155339806e-05, "loss": 0.3228, "step": 7672 }, { "epoch": 74.40240963855422, "grad_norm": 24.371463775634766, "learning_rate": 6.275242718446602e-05, "loss": 0.2943, "step": 7673 }, { "epoch": 74.41204819277108, "grad_norm": 2.0696756839752197, "learning_rate": 6.274757281553398e-05, "loss": 0.2233, "step": 7674 }, { "epoch": 74.42168674698796, "grad_norm": 8.082834243774414, "learning_rate": 6.274271844660195e-05, "loss": 0.2513, "step": 7675 }, { "epoch": 74.43132530120482, "grad_norm": 56.73862838745117, "learning_rate": 6.273786407766991e-05, "loss": 0.2908, "step": 7676 }, { "epoch": 74.44096385542169, "grad_norm": 7.583291053771973, "learning_rate": 6.273300970873787e-05, "loss": 0.257, "step": 7677 }, { "epoch": 74.45060240963855, "grad_norm": 26.376201629638672, "learning_rate": 6.272815533980582e-05, "loss": 0.3845, "step": 7678 }, { "epoch": 74.46024096385543, "grad_norm": 15.338433265686035, "learning_rate": 6.272330097087378e-05, "loss": 0.4939, "step": 7679 }, { "epoch": 74.46987951807229, "grad_norm": 13.964509963989258, "learning_rate": 6.271844660194175e-05, "loss": 0.484, "step": 7680 }, { "epoch": 74.47951807228915, "grad_norm": 4.41622257232666, "learning_rate": 6.271359223300971e-05, "loss": 0.3523, "step": 7681 }, { "epoch": 74.48915662650603, "grad_norm": 1.7041834592819214, "learning_rate": 6.270873786407767e-05, "loss": 0.1533, "step": 7682 }, { "epoch": 74.4987951807229, "grad_norm": 6.491270065307617, "learning_rate": 6.270388349514563e-05, "loss": 0.2212, "step": 7683 }, { "epoch": 74.50843373493976, "grad_norm": 12.078897476196289, "learning_rate": 6.26990291262136e-05, "loss": 0.2541, "step": 7684 }, { "epoch": 74.51807228915662, "grad_norm": 65.4902572631836, "learning_rate": 6.269417475728156e-05, "loss": 0.3537, "step": 7685 }, { "epoch": 74.5277108433735, "grad_norm": 0.9952031970024109, "learning_rate": 6.268932038834951e-05, "loss": 0.2972, "step": 7686 }, { "epoch": 74.53734939759036, "grad_norm": 57.706974029541016, "learning_rate": 6.268446601941747e-05, "loss": 0.3491, "step": 7687 }, { "epoch": 74.54698795180722, "grad_norm": 13.061381340026855, "learning_rate": 6.267961165048544e-05, "loss": 0.3949, "step": 7688 }, { "epoch": 74.5566265060241, "grad_norm": 43.187400817871094, "learning_rate": 6.26747572815534e-05, "loss": 0.3452, "step": 7689 }, { "epoch": 74.56626506024097, "grad_norm": 2.8720781803131104, "learning_rate": 6.266990291262136e-05, "loss": 0.3516, "step": 7690 }, { "epoch": 74.57590361445783, "grad_norm": 5.665189743041992, "learning_rate": 6.266504854368932e-05, "loss": 0.3498, "step": 7691 }, { "epoch": 74.58554216867469, "grad_norm": 7.563119888305664, "learning_rate": 6.266019417475729e-05, "loss": 0.2045, "step": 7692 }, { "epoch": 74.59518072289157, "grad_norm": 9.978095054626465, "learning_rate": 6.265533980582524e-05, "loss": 0.2259, "step": 7693 }, { "epoch": 74.60481927710843, "grad_norm": 14.539473533630371, "learning_rate": 6.26504854368932e-05, "loss": 0.3806, "step": 7694 }, { "epoch": 74.6144578313253, "grad_norm": 8.272656440734863, "learning_rate": 6.264563106796116e-05, "loss": 0.4887, "step": 7695 }, { "epoch": 74.62409638554217, "grad_norm": 5.197551250457764, "learning_rate": 6.264077669902912e-05, "loss": 0.2876, "step": 7696 }, { "epoch": 74.63373493975904, "grad_norm": 16.170000076293945, "learning_rate": 6.263592233009709e-05, "loss": 0.3364, "step": 7697 }, { "epoch": 74.6433734939759, "grad_norm": 4.3618903160095215, "learning_rate": 6.263106796116505e-05, "loss": 0.2517, "step": 7698 }, { "epoch": 74.65301204819278, "grad_norm": 113.35206604003906, "learning_rate": 6.2626213592233e-05, "loss": 0.2922, "step": 7699 }, { "epoch": 74.66265060240964, "grad_norm": 69.160400390625, "learning_rate": 6.262135922330098e-05, "loss": 0.164, "step": 7700 }, { "epoch": 74.6722891566265, "grad_norm": 13.458179473876953, "learning_rate": 6.261650485436893e-05, "loss": 0.3878, "step": 7701 }, { "epoch": 74.68192771084337, "grad_norm": 30.609439849853516, "learning_rate": 6.26116504854369e-05, "loss": 0.2123, "step": 7702 }, { "epoch": 74.69156626506025, "grad_norm": 7.219809055328369, "learning_rate": 6.260679611650486e-05, "loss": 0.3453, "step": 7703 }, { "epoch": 74.70120481927711, "grad_norm": 4.733546257019043, "learning_rate": 6.260194174757282e-05, "loss": 0.2838, "step": 7704 }, { "epoch": 74.71084337349397, "grad_norm": 3.2554125785827637, "learning_rate": 6.259708737864078e-05, "loss": 0.0918, "step": 7705 }, { "epoch": 74.72048192771085, "grad_norm": 8.205339431762695, "learning_rate": 6.259223300970875e-05, "loss": 0.3971, "step": 7706 }, { "epoch": 74.73012048192771, "grad_norm": 11.113314628601074, "learning_rate": 6.258737864077671e-05, "loss": 0.3096, "step": 7707 }, { "epoch": 74.73975903614458, "grad_norm": 7.816232681274414, "learning_rate": 6.258252427184467e-05, "loss": 0.2035, "step": 7708 }, { "epoch": 74.74939759036144, "grad_norm": 8.339568138122559, "learning_rate": 6.257766990291262e-05, "loss": 0.4025, "step": 7709 }, { "epoch": 74.75903614457832, "grad_norm": 6.301371097564697, "learning_rate": 6.25728155339806e-05, "loss": 0.3262, "step": 7710 }, { "epoch": 74.76867469879518, "grad_norm": 5.923267364501953, "learning_rate": 6.256796116504855e-05, "loss": 0.198, "step": 7711 }, { "epoch": 74.77831325301204, "grad_norm": 24.422788619995117, "learning_rate": 6.256310679611651e-05, "loss": 0.2224, "step": 7712 }, { "epoch": 74.78795180722892, "grad_norm": 16.016233444213867, "learning_rate": 6.255825242718447e-05, "loss": 0.3745, "step": 7713 }, { "epoch": 74.79759036144578, "grad_norm": 12.530221939086914, "learning_rate": 6.255339805825244e-05, "loss": 0.3612, "step": 7714 }, { "epoch": 74.80722891566265, "grad_norm": 2.983701229095459, "learning_rate": 6.25485436893204e-05, "loss": 0.0934, "step": 7715 }, { "epoch": 74.81686746987951, "grad_norm": 33.80431365966797, "learning_rate": 6.254368932038835e-05, "loss": 0.302, "step": 7716 }, { "epoch": 74.82650602409639, "grad_norm": 2.4999327659606934, "learning_rate": 6.253883495145631e-05, "loss": 0.1129, "step": 7717 }, { "epoch": 74.83614457831325, "grad_norm": 6.104265213012695, "learning_rate": 6.253398058252427e-05, "loss": 0.2298, "step": 7718 }, { "epoch": 74.84578313253012, "grad_norm": 9.92497444152832, "learning_rate": 6.252912621359224e-05, "loss": 0.2097, "step": 7719 }, { "epoch": 74.855421686747, "grad_norm": 2.756629467010498, "learning_rate": 6.25242718446602e-05, "loss": 0.2078, "step": 7720 }, { "epoch": 74.86506024096386, "grad_norm": 11.527539253234863, "learning_rate": 6.251941747572816e-05, "loss": 0.3251, "step": 7721 }, { "epoch": 74.87469879518072, "grad_norm": 4.021393299102783, "learning_rate": 6.251456310679611e-05, "loss": 0.1249, "step": 7722 }, { "epoch": 74.88433734939758, "grad_norm": 15.83170223236084, "learning_rate": 6.250970873786409e-05, "loss": 0.3165, "step": 7723 }, { "epoch": 74.89397590361446, "grad_norm": 61.90603256225586, "learning_rate": 6.250485436893204e-05, "loss": 0.3108, "step": 7724 }, { "epoch": 74.90361445783132, "grad_norm": 6.10499906539917, "learning_rate": 6.25e-05, "loss": 0.2819, "step": 7725 }, { "epoch": 74.91325301204819, "grad_norm": 21.759601593017578, "learning_rate": 6.249514563106796e-05, "loss": 0.3046, "step": 7726 }, { "epoch": 74.92289156626506, "grad_norm": 11.30009937286377, "learning_rate": 6.249029126213593e-05, "loss": 0.5599, "step": 7727 }, { "epoch": 74.93253012048193, "grad_norm": 5.143789291381836, "learning_rate": 6.248543689320389e-05, "loss": 0.2781, "step": 7728 }, { "epoch": 74.94216867469879, "grad_norm": 7.396533966064453, "learning_rate": 6.248058252427185e-05, "loss": 0.1044, "step": 7729 }, { "epoch": 74.95180722891567, "grad_norm": 13.883987426757812, "learning_rate": 6.24757281553398e-05, "loss": 0.298, "step": 7730 }, { "epoch": 74.96144578313253, "grad_norm": 6.3558669090271, "learning_rate": 6.247087378640777e-05, "loss": 0.301, "step": 7731 }, { "epoch": 74.9710843373494, "grad_norm": 15.44318962097168, "learning_rate": 6.246601941747573e-05, "loss": 0.3905, "step": 7732 }, { "epoch": 74.98072289156626, "grad_norm": 8.214715957641602, "learning_rate": 6.246116504854369e-05, "loss": 0.0981, "step": 7733 }, { "epoch": 74.99036144578314, "grad_norm": 3.1946909427642822, "learning_rate": 6.245631067961165e-05, "loss": 0.2618, "step": 7734 }, { "epoch": 75.00602409638554, "grad_norm": 28.708833694458008, "learning_rate": 6.24514563106796e-05, "loss": 0.3893, "step": 7735 }, { "epoch": 75.01566265060241, "grad_norm": 8.433294296264648, "learning_rate": 6.244660194174758e-05, "loss": 0.439, "step": 7736 }, { "epoch": 75.02530120481927, "grad_norm": 41.13554382324219, "learning_rate": 6.244174757281554e-05, "loss": 0.2934, "step": 7737 }, { "epoch": 75.03493975903615, "grad_norm": 27.968685150146484, "learning_rate": 6.243689320388349e-05, "loss": 0.2681, "step": 7738 }, { "epoch": 75.04457831325301, "grad_norm": 4.587564468383789, "learning_rate": 6.243203883495145e-05, "loss": 0.2116, "step": 7739 }, { "epoch": 75.05421686746988, "grad_norm": 20.899377822875977, "learning_rate": 6.242718446601942e-05, "loss": 0.4843, "step": 7740 }, { "epoch": 75.06385542168675, "grad_norm": 15.23525333404541, "learning_rate": 6.242233009708738e-05, "loss": 0.4499, "step": 7741 }, { "epoch": 75.07349397590362, "grad_norm": 156.9478302001953, "learning_rate": 6.241747572815534e-05, "loss": 0.233, "step": 7742 }, { "epoch": 75.08313253012048, "grad_norm": 21.579238891601562, "learning_rate": 6.24126213592233e-05, "loss": 0.2422, "step": 7743 }, { "epoch": 75.09277108433734, "grad_norm": 3.894019842147827, "learning_rate": 6.240776699029127e-05, "loss": 0.2041, "step": 7744 }, { "epoch": 75.10240963855422, "grad_norm": 27.90090560913086, "learning_rate": 6.240291262135922e-05, "loss": 0.266, "step": 7745 }, { "epoch": 75.11204819277108, "grad_norm": 30.146928787231445, "learning_rate": 6.239805825242718e-05, "loss": 0.3682, "step": 7746 }, { "epoch": 75.12168674698795, "grad_norm": 34.63692092895508, "learning_rate": 6.239320388349514e-05, "loss": 0.2702, "step": 7747 }, { "epoch": 75.13132530120482, "grad_norm": 4.324772357940674, "learning_rate": 6.238834951456311e-05, "loss": 0.2817, "step": 7748 }, { "epoch": 75.14096385542169, "grad_norm": 22.21502113342285, "learning_rate": 6.238349514563107e-05, "loss": 0.2542, "step": 7749 }, { "epoch": 75.15060240963855, "grad_norm": 10.823016166687012, "learning_rate": 6.237864077669903e-05, "loss": 0.3405, "step": 7750 }, { "epoch": 75.16024096385541, "grad_norm": 12.50327205657959, "learning_rate": 6.2373786407767e-05, "loss": 0.3308, "step": 7751 }, { "epoch": 75.16987951807229, "grad_norm": 7.749337673187256, "learning_rate": 6.236893203883496e-05, "loss": 0.2575, "step": 7752 }, { "epoch": 75.17951807228916, "grad_norm": 24.931215286254883, "learning_rate": 6.236407766990293e-05, "loss": 0.7244, "step": 7753 }, { "epoch": 75.18915662650602, "grad_norm": 4.5623321533203125, "learning_rate": 6.235922330097088e-05, "loss": 0.2506, "step": 7754 }, { "epoch": 75.1987951807229, "grad_norm": 54.13434600830078, "learning_rate": 6.235436893203884e-05, "loss": 0.4781, "step": 7755 }, { "epoch": 75.20843373493976, "grad_norm": 4.6174421310424805, "learning_rate": 6.23495145631068e-05, "loss": 0.2846, "step": 7756 }, { "epoch": 75.21807228915662, "grad_norm": 2.501255989074707, "learning_rate": 6.234466019417476e-05, "loss": 0.1941, "step": 7757 }, { "epoch": 75.2277108433735, "grad_norm": 5.192638397216797, "learning_rate": 6.233980582524273e-05, "loss": 0.2511, "step": 7758 }, { "epoch": 75.23734939759036, "grad_norm": 13.030994415283203, "learning_rate": 6.233495145631069e-05, "loss": 0.3534, "step": 7759 }, { "epoch": 75.24698795180723, "grad_norm": 10.36681842803955, "learning_rate": 6.233009708737864e-05, "loss": 0.2638, "step": 7760 }, { "epoch": 75.25662650602409, "grad_norm": 13.433392524719238, "learning_rate": 6.23252427184466e-05, "loss": 0.2687, "step": 7761 }, { "epoch": 75.26626506024097, "grad_norm": 6.678405284881592, "learning_rate": 6.232038834951457e-05, "loss": 0.3954, "step": 7762 }, { "epoch": 75.27590361445783, "grad_norm": 9.081711769104004, "learning_rate": 6.231553398058253e-05, "loss": 0.2677, "step": 7763 }, { "epoch": 75.2855421686747, "grad_norm": 22.760635375976562, "learning_rate": 6.231067961165049e-05, "loss": 0.2587, "step": 7764 }, { "epoch": 75.29518072289157, "grad_norm": 6.728288650512695, "learning_rate": 6.230582524271845e-05, "loss": 0.2902, "step": 7765 }, { "epoch": 75.30481927710844, "grad_norm": 7.261501312255859, "learning_rate": 6.230097087378642e-05, "loss": 0.4307, "step": 7766 }, { "epoch": 75.3144578313253, "grad_norm": 13.741621017456055, "learning_rate": 6.229611650485438e-05, "loss": 0.3326, "step": 7767 }, { "epoch": 75.32409638554216, "grad_norm": 18.41766357421875, "learning_rate": 6.229126213592233e-05, "loss": 0.5143, "step": 7768 }, { "epoch": 75.33373493975904, "grad_norm": 5.676761150360107, "learning_rate": 6.228640776699029e-05, "loss": 0.2788, "step": 7769 }, { "epoch": 75.3433734939759, "grad_norm": 12.367594718933105, "learning_rate": 6.228155339805826e-05, "loss": 0.4434, "step": 7770 }, { "epoch": 75.35301204819277, "grad_norm": 3.184445381164551, "learning_rate": 6.227669902912622e-05, "loss": 0.1534, "step": 7771 }, { "epoch": 75.36265060240964, "grad_norm": 4.39246129989624, "learning_rate": 6.227184466019418e-05, "loss": 0.2764, "step": 7772 }, { "epoch": 75.37228915662651, "grad_norm": 4.763950347900391, "learning_rate": 6.226699029126214e-05, "loss": 0.3113, "step": 7773 }, { "epoch": 75.38192771084337, "grad_norm": 10.468350410461426, "learning_rate": 6.22621359223301e-05, "loss": 0.3904, "step": 7774 }, { "epoch": 75.39156626506023, "grad_norm": 14.826666831970215, "learning_rate": 6.225728155339807e-05, "loss": 0.2147, "step": 7775 }, { "epoch": 75.40120481927711, "grad_norm": 23.92942237854004, "learning_rate": 6.225242718446602e-05, "loss": 0.3957, "step": 7776 }, { "epoch": 75.41084337349398, "grad_norm": 2.6916840076446533, "learning_rate": 6.224757281553398e-05, "loss": 0.1403, "step": 7777 }, { "epoch": 75.42048192771084, "grad_norm": 10.634319305419922, "learning_rate": 6.224271844660194e-05, "loss": 0.1916, "step": 7778 }, { "epoch": 75.43012048192772, "grad_norm": 11.503338813781738, "learning_rate": 6.223786407766991e-05, "loss": 0.3663, "step": 7779 }, { "epoch": 75.43975903614458, "grad_norm": 13.564749717712402, "learning_rate": 6.223300970873787e-05, "loss": 0.2352, "step": 7780 }, { "epoch": 75.44939759036144, "grad_norm": 7.126152992248535, "learning_rate": 6.222815533980583e-05, "loss": 0.2679, "step": 7781 }, { "epoch": 75.45903614457832, "grad_norm": 42.14423370361328, "learning_rate": 6.222330097087378e-05, "loss": 0.1669, "step": 7782 }, { "epoch": 75.46867469879518, "grad_norm": 16.970401763916016, "learning_rate": 6.221844660194175e-05, "loss": 0.5946, "step": 7783 }, { "epoch": 75.47831325301205, "grad_norm": 12.72386360168457, "learning_rate": 6.221359223300971e-05, "loss": 0.1616, "step": 7784 }, { "epoch": 75.48795180722891, "grad_norm": 3.295565605163574, "learning_rate": 6.220873786407767e-05, "loss": 0.1892, "step": 7785 }, { "epoch": 75.49759036144579, "grad_norm": 4.907328128814697, "learning_rate": 6.220388349514563e-05, "loss": 0.2915, "step": 7786 }, { "epoch": 75.50722891566265, "grad_norm": 12.445056915283203, "learning_rate": 6.21990291262136e-05, "loss": 0.26, "step": 7787 }, { "epoch": 75.51686746987951, "grad_norm": 17.094627380371094, "learning_rate": 6.219417475728156e-05, "loss": 0.2412, "step": 7788 }, { "epoch": 75.52650602409639, "grad_norm": 12.18015193939209, "learning_rate": 6.218932038834951e-05, "loss": 0.2761, "step": 7789 }, { "epoch": 75.53614457831326, "grad_norm": 2.3912429809570312, "learning_rate": 6.218446601941747e-05, "loss": 0.2566, "step": 7790 }, { "epoch": 75.54578313253012, "grad_norm": 23.163166046142578, "learning_rate": 6.217961165048543e-05, "loss": 0.367, "step": 7791 }, { "epoch": 75.55542168674698, "grad_norm": 8.371794700622559, "learning_rate": 6.21747572815534e-05, "loss": 0.2, "step": 7792 }, { "epoch": 75.56506024096386, "grad_norm": 18.914705276489258, "learning_rate": 6.216990291262136e-05, "loss": 0.1733, "step": 7793 }, { "epoch": 75.57469879518072, "grad_norm": 27.913860321044922, "learning_rate": 6.216504854368932e-05, "loss": 0.3523, "step": 7794 }, { "epoch": 75.58433734939759, "grad_norm": 9.909224510192871, "learning_rate": 6.216019417475727e-05, "loss": 0.2102, "step": 7795 }, { "epoch": 75.59397590361446, "grad_norm": 7.771755218505859, "learning_rate": 6.215533980582525e-05, "loss": 0.3304, "step": 7796 }, { "epoch": 75.60361445783133, "grad_norm": 27.253543853759766, "learning_rate": 6.21504854368932e-05, "loss": 0.3426, "step": 7797 }, { "epoch": 75.61325301204819, "grad_norm": 17.71184539794922, "learning_rate": 6.214563106796116e-05, "loss": 0.3378, "step": 7798 }, { "epoch": 75.62289156626505, "grad_norm": 11.897692680358887, "learning_rate": 6.214077669902912e-05, "loss": 0.4301, "step": 7799 }, { "epoch": 75.63253012048193, "grad_norm": 6.67252779006958, "learning_rate": 6.213592233009709e-05, "loss": 0.3399, "step": 7800 }, { "epoch": 75.6421686746988, "grad_norm": 13.430048942565918, "learning_rate": 6.213106796116506e-05, "loss": 0.2638, "step": 7801 }, { "epoch": 75.65180722891566, "grad_norm": 10.359349250793457, "learning_rate": 6.212621359223302e-05, "loss": 0.3294, "step": 7802 }, { "epoch": 75.66144578313254, "grad_norm": 3.920853614807129, "learning_rate": 6.212135922330098e-05, "loss": 0.3259, "step": 7803 }, { "epoch": 75.6710843373494, "grad_norm": 14.979547500610352, "learning_rate": 6.211650485436894e-05, "loss": 0.613, "step": 7804 }, { "epoch": 75.68072289156626, "grad_norm": 7.755629539489746, "learning_rate": 6.21116504854369e-05, "loss": 0.251, "step": 7805 }, { "epoch": 75.69036144578314, "grad_norm": 4.350035667419434, "learning_rate": 6.210679611650486e-05, "loss": 0.3415, "step": 7806 }, { "epoch": 75.7, "grad_norm": 27.813467025756836, "learning_rate": 6.210194174757282e-05, "loss": 0.1549, "step": 7807 }, { "epoch": 75.70963855421687, "grad_norm": 11.004875183105469, "learning_rate": 6.209708737864078e-05, "loss": 0.3439, "step": 7808 }, { "epoch": 75.71927710843373, "grad_norm": 2.4994242191314697, "learning_rate": 6.209223300970875e-05, "loss": 0.3146, "step": 7809 }, { "epoch": 75.7289156626506, "grad_norm": 25.426023483276367, "learning_rate": 6.208737864077671e-05, "loss": 0.1289, "step": 7810 }, { "epoch": 75.73855421686747, "grad_norm": 6.561537265777588, "learning_rate": 6.208252427184467e-05, "loss": 0.3274, "step": 7811 }, { "epoch": 75.74819277108433, "grad_norm": 22.826736450195312, "learning_rate": 6.207766990291262e-05, "loss": 0.1723, "step": 7812 }, { "epoch": 75.75783132530121, "grad_norm": 22.85336685180664, "learning_rate": 6.207281553398058e-05, "loss": 0.2569, "step": 7813 }, { "epoch": 75.76746987951807, "grad_norm": 14.790824890136719, "learning_rate": 6.206796116504855e-05, "loss": 0.4327, "step": 7814 }, { "epoch": 75.77710843373494, "grad_norm": 6.064547061920166, "learning_rate": 6.206310679611651e-05, "loss": 0.3327, "step": 7815 }, { "epoch": 75.7867469879518, "grad_norm": 5.407444000244141, "learning_rate": 6.205825242718447e-05, "loss": 0.1741, "step": 7816 }, { "epoch": 75.79638554216868, "grad_norm": 8.97878360748291, "learning_rate": 6.205339805825243e-05, "loss": 0.3158, "step": 7817 }, { "epoch": 75.80602409638554, "grad_norm": 14.949835777282715, "learning_rate": 6.20485436893204e-05, "loss": 0.4796, "step": 7818 }, { "epoch": 75.8156626506024, "grad_norm": 14.669405937194824, "learning_rate": 6.204368932038836e-05, "loss": 0.2995, "step": 7819 }, { "epoch": 75.82530120481928, "grad_norm": 10.82479190826416, "learning_rate": 6.203883495145631e-05, "loss": 0.1849, "step": 7820 }, { "epoch": 75.83493975903615, "grad_norm": 16.37142562866211, "learning_rate": 6.203398058252427e-05, "loss": 0.3494, "step": 7821 }, { "epoch": 75.84457831325301, "grad_norm": 22.101587295532227, "learning_rate": 6.202912621359224e-05, "loss": 0.3827, "step": 7822 }, { "epoch": 75.85421686746987, "grad_norm": 7.659794330596924, "learning_rate": 6.20242718446602e-05, "loss": 0.2731, "step": 7823 }, { "epoch": 75.86385542168675, "grad_norm": 5.555226802825928, "learning_rate": 6.201941747572816e-05, "loss": 0.2159, "step": 7824 }, { "epoch": 75.87349397590361, "grad_norm": 13.616129875183105, "learning_rate": 6.201456310679612e-05, "loss": 0.3702, "step": 7825 }, { "epoch": 75.88313253012048, "grad_norm": 26.98531723022461, "learning_rate": 6.200970873786409e-05, "loss": 0.2568, "step": 7826 }, { "epoch": 75.89277108433735, "grad_norm": 10.707069396972656, "learning_rate": 6.200485436893204e-05, "loss": 0.3582, "step": 7827 }, { "epoch": 75.90240963855422, "grad_norm": 19.172199249267578, "learning_rate": 6.2e-05, "loss": 0.1994, "step": 7828 }, { "epoch": 75.91204819277108, "grad_norm": 3.7607767581939697, "learning_rate": 6.199514563106796e-05, "loss": 0.1228, "step": 7829 }, { "epoch": 75.92168674698796, "grad_norm": 1.5623345375061035, "learning_rate": 6.199029126213592e-05, "loss": 0.2482, "step": 7830 }, { "epoch": 75.93132530120482, "grad_norm": 22.022960662841797, "learning_rate": 6.198543689320389e-05, "loss": 0.3368, "step": 7831 }, { "epoch": 75.94096385542169, "grad_norm": 4.501099586486816, "learning_rate": 6.198058252427185e-05, "loss": 0.1468, "step": 7832 }, { "epoch": 75.95060240963855, "grad_norm": 11.920125007629395, "learning_rate": 6.19757281553398e-05, "loss": 0.339, "step": 7833 }, { "epoch": 75.96024096385543, "grad_norm": 31.834186553955078, "learning_rate": 6.197087378640776e-05, "loss": 0.4254, "step": 7834 }, { "epoch": 75.96987951807229, "grad_norm": 6.413938999176025, "learning_rate": 6.196601941747573e-05, "loss": 0.2538, "step": 7835 }, { "epoch": 75.97951807228915, "grad_norm": 23.069612503051758, "learning_rate": 6.196116504854369e-05, "loss": 0.3462, "step": 7836 }, { "epoch": 75.98915662650603, "grad_norm": 30.82686996459961, "learning_rate": 6.195631067961165e-05, "loss": 0.3904, "step": 7837 }, { "epoch": 76.00481927710844, "grad_norm": 10.20435905456543, "learning_rate": 6.195145631067961e-05, "loss": 0.3298, "step": 7838 }, { "epoch": 76.0144578313253, "grad_norm": 9.638174057006836, "learning_rate": 6.194660194174758e-05, "loss": 0.5658, "step": 7839 }, { "epoch": 76.02409638554217, "grad_norm": 2.782881021499634, "learning_rate": 6.194174757281554e-05, "loss": 0.2462, "step": 7840 }, { "epoch": 76.03373493975904, "grad_norm": 26.561498641967773, "learning_rate": 6.19368932038835e-05, "loss": 0.4295, "step": 7841 }, { "epoch": 76.0433734939759, "grad_norm": 7.53177547454834, "learning_rate": 6.193203883495145e-05, "loss": 0.3519, "step": 7842 }, { "epoch": 76.05301204819277, "grad_norm": 26.81083869934082, "learning_rate": 6.192718446601942e-05, "loss": 0.3802, "step": 7843 }, { "epoch": 76.06265060240963, "grad_norm": 9.879721641540527, "learning_rate": 6.192233009708738e-05, "loss": 0.4172, "step": 7844 }, { "epoch": 76.07228915662651, "grad_norm": 22.975566864013672, "learning_rate": 6.191747572815534e-05, "loss": 0.4307, "step": 7845 }, { "epoch": 76.08192771084337, "grad_norm": 3.941720724105835, "learning_rate": 6.19126213592233e-05, "loss": 0.1803, "step": 7846 }, { "epoch": 76.09156626506024, "grad_norm": 3.8162875175476074, "learning_rate": 6.190776699029125e-05, "loss": 0.2535, "step": 7847 }, { "epoch": 76.10120481927711, "grad_norm": 16.34026527404785, "learning_rate": 6.190291262135923e-05, "loss": 0.4053, "step": 7848 }, { "epoch": 76.11084337349398, "grad_norm": 6.806787014007568, "learning_rate": 6.189805825242718e-05, "loss": 0.3387, "step": 7849 }, { "epoch": 76.12048192771084, "grad_norm": 20.4476261138916, "learning_rate": 6.189320388349514e-05, "loss": 0.4609, "step": 7850 }, { "epoch": 76.1301204819277, "grad_norm": 1.963713526725769, "learning_rate": 6.18883495145631e-05, "loss": 0.1462, "step": 7851 }, { "epoch": 76.13975903614458, "grad_norm": 26.22223472595215, "learning_rate": 6.188349514563107e-05, "loss": 0.2947, "step": 7852 }, { "epoch": 76.14939759036145, "grad_norm": 12.747044563293457, "learning_rate": 6.187864077669904e-05, "loss": 0.4123, "step": 7853 }, { "epoch": 76.15903614457831, "grad_norm": 10.887367248535156, "learning_rate": 6.1873786407767e-05, "loss": 0.1871, "step": 7854 }, { "epoch": 76.16867469879519, "grad_norm": 11.576189041137695, "learning_rate": 6.186893203883496e-05, "loss": 0.3042, "step": 7855 }, { "epoch": 76.17831325301205, "grad_norm": 12.790206909179688, "learning_rate": 6.186407766990291e-05, "loss": 0.3099, "step": 7856 }, { "epoch": 76.18795180722891, "grad_norm": 9.334723472595215, "learning_rate": 6.185922330097089e-05, "loss": 0.2189, "step": 7857 }, { "epoch": 76.19759036144578, "grad_norm": 5.323615550994873, "learning_rate": 6.185436893203884e-05, "loss": 0.2041, "step": 7858 }, { "epoch": 76.20722891566265, "grad_norm": 5.410968780517578, "learning_rate": 6.18495145631068e-05, "loss": 0.2028, "step": 7859 }, { "epoch": 76.21686746987952, "grad_norm": 6.311045169830322, "learning_rate": 6.184466019417476e-05, "loss": 0.3175, "step": 7860 }, { "epoch": 76.22650602409638, "grad_norm": 21.26617431640625, "learning_rate": 6.183980582524273e-05, "loss": 0.3095, "step": 7861 }, { "epoch": 76.23614457831326, "grad_norm": 12.659114837646484, "learning_rate": 6.183495145631069e-05, "loss": 0.2711, "step": 7862 }, { "epoch": 76.24578313253012, "grad_norm": 11.873754501342773, "learning_rate": 6.183009708737865e-05, "loss": 0.4499, "step": 7863 }, { "epoch": 76.25542168674698, "grad_norm": 3.8055472373962402, "learning_rate": 6.18252427184466e-05, "loss": 0.2534, "step": 7864 }, { "epoch": 76.26506024096386, "grad_norm": 2.447486400604248, "learning_rate": 6.182038834951458e-05, "loss": 0.1181, "step": 7865 }, { "epoch": 76.27469879518073, "grad_norm": 16.911602020263672, "learning_rate": 6.181553398058253e-05, "loss": 0.1878, "step": 7866 }, { "epoch": 76.28433734939759, "grad_norm": 168.58157348632812, "learning_rate": 6.181067961165049e-05, "loss": 0.1468, "step": 7867 }, { "epoch": 76.29397590361445, "grad_norm": 3.446259021759033, "learning_rate": 6.180582524271845e-05, "loss": 0.2438, "step": 7868 }, { "epoch": 76.30361445783133, "grad_norm": 17.584972381591797, "learning_rate": 6.18009708737864e-05, "loss": 0.6245, "step": 7869 }, { "epoch": 76.3132530120482, "grad_norm": 4.477282524108887, "learning_rate": 6.179611650485438e-05, "loss": 0.1344, "step": 7870 }, { "epoch": 76.32289156626506, "grad_norm": 5.956451892852783, "learning_rate": 6.179126213592234e-05, "loss": 0.2677, "step": 7871 }, { "epoch": 76.33253012048193, "grad_norm": 13.106675148010254, "learning_rate": 6.178640776699029e-05, "loss": 0.1812, "step": 7872 }, { "epoch": 76.3421686746988, "grad_norm": 28.239578247070312, "learning_rate": 6.178155339805825e-05, "loss": 0.4392, "step": 7873 }, { "epoch": 76.35180722891566, "grad_norm": 5.142886638641357, "learning_rate": 6.177669902912622e-05, "loss": 0.2074, "step": 7874 }, { "epoch": 76.36144578313252, "grad_norm": 12.046384811401367, "learning_rate": 6.177184466019418e-05, "loss": 0.2236, "step": 7875 }, { "epoch": 76.3710843373494, "grad_norm": 16.230493545532227, "learning_rate": 6.176699029126214e-05, "loss": 0.429, "step": 7876 }, { "epoch": 76.38072289156626, "grad_norm": 11.152230262756348, "learning_rate": 6.17621359223301e-05, "loss": 0.1977, "step": 7877 }, { "epoch": 76.39036144578313, "grad_norm": 3.8433852195739746, "learning_rate": 6.175728155339807e-05, "loss": 0.1357, "step": 7878 }, { "epoch": 76.4, "grad_norm": 9.632987976074219, "learning_rate": 6.175242718446602e-05, "loss": 0.3739, "step": 7879 }, { "epoch": 76.40963855421687, "grad_norm": 10.331855773925781, "learning_rate": 6.174757281553398e-05, "loss": 0.4864, "step": 7880 }, { "epoch": 76.41927710843373, "grad_norm": 9.184535026550293, "learning_rate": 6.174271844660194e-05, "loss": 0.2384, "step": 7881 }, { "epoch": 76.4289156626506, "grad_norm": 62.31989288330078, "learning_rate": 6.173786407766991e-05, "loss": 0.3739, "step": 7882 }, { "epoch": 76.43855421686747, "grad_norm": 15.98682975769043, "learning_rate": 6.173300970873787e-05, "loss": 0.3978, "step": 7883 }, { "epoch": 76.44819277108434, "grad_norm": 23.217885971069336, "learning_rate": 6.172815533980583e-05, "loss": 0.485, "step": 7884 }, { "epoch": 76.4578313253012, "grad_norm": 12.345030784606934, "learning_rate": 6.172330097087378e-05, "loss": 0.2013, "step": 7885 }, { "epoch": 76.46746987951808, "grad_norm": 7.41091251373291, "learning_rate": 6.171844660194174e-05, "loss": 0.2925, "step": 7886 }, { "epoch": 76.47710843373494, "grad_norm": 25.852731704711914, "learning_rate": 6.171359223300971e-05, "loss": 0.3997, "step": 7887 }, { "epoch": 76.4867469879518, "grad_norm": 6.296528339385986, "learning_rate": 6.170873786407767e-05, "loss": 0.4202, "step": 7888 }, { "epoch": 76.49638554216868, "grad_norm": 11.259102821350098, "learning_rate": 6.170388349514563e-05, "loss": 0.2661, "step": 7889 }, { "epoch": 76.50602409638554, "grad_norm": 16.922460556030273, "learning_rate": 6.169902912621359e-05, "loss": 0.2627, "step": 7890 }, { "epoch": 76.51566265060241, "grad_norm": 17.005962371826172, "learning_rate": 6.169417475728156e-05, "loss": 0.3002, "step": 7891 }, { "epoch": 76.52530120481927, "grad_norm": 9.36590576171875, "learning_rate": 6.168932038834952e-05, "loss": 0.1961, "step": 7892 }, { "epoch": 76.53493975903615, "grad_norm": 29.282629013061523, "learning_rate": 6.168446601941747e-05, "loss": 0.3688, "step": 7893 }, { "epoch": 76.54457831325301, "grad_norm": 13.700407981872559, "learning_rate": 6.167961165048543e-05, "loss": 0.2753, "step": 7894 }, { "epoch": 76.55421686746988, "grad_norm": 3.9915316104888916, "learning_rate": 6.16747572815534e-05, "loss": 0.18, "step": 7895 }, { "epoch": 76.56385542168675, "grad_norm": 7.645717144012451, "learning_rate": 6.166990291262136e-05, "loss": 0.2526, "step": 7896 }, { "epoch": 76.57349397590362, "grad_norm": 8.802512168884277, "learning_rate": 6.166504854368932e-05, "loss": 0.2169, "step": 7897 }, { "epoch": 76.58313253012048, "grad_norm": 12.673576354980469, "learning_rate": 6.166019417475728e-05, "loss": 0.1769, "step": 7898 }, { "epoch": 76.59277108433734, "grad_norm": 8.501120567321777, "learning_rate": 6.165533980582525e-05, "loss": 0.1528, "step": 7899 }, { "epoch": 76.60240963855422, "grad_norm": 14.141979217529297, "learning_rate": 6.16504854368932e-05, "loss": 0.2458, "step": 7900 }, { "epoch": 76.61204819277108, "grad_norm": 30.6793155670166, "learning_rate": 6.164563106796116e-05, "loss": 0.1677, "step": 7901 }, { "epoch": 76.62168674698795, "grad_norm": 1.961316466331482, "learning_rate": 6.164077669902913e-05, "loss": 0.0865, "step": 7902 }, { "epoch": 76.63132530120482, "grad_norm": 6.02357292175293, "learning_rate": 6.163592233009709e-05, "loss": 0.4831, "step": 7903 }, { "epoch": 76.64096385542169, "grad_norm": 12.866796493530273, "learning_rate": 6.163106796116505e-05, "loss": 0.2818, "step": 7904 }, { "epoch": 76.65060240963855, "grad_norm": 13.449037551879883, "learning_rate": 6.162621359223302e-05, "loss": 0.5439, "step": 7905 }, { "epoch": 76.66024096385541, "grad_norm": 8.882939338684082, "learning_rate": 6.162135922330098e-05, "loss": 0.3038, "step": 7906 }, { "epoch": 76.66987951807229, "grad_norm": 9.144289016723633, "learning_rate": 6.161650485436894e-05, "loss": 0.3387, "step": 7907 }, { "epoch": 76.67951807228916, "grad_norm": 27.856843948364258, "learning_rate": 6.16116504854369e-05, "loss": 0.3604, "step": 7908 }, { "epoch": 76.68915662650602, "grad_norm": 5.352812767028809, "learning_rate": 6.160679611650487e-05, "loss": 0.3002, "step": 7909 }, { "epoch": 76.6987951807229, "grad_norm": 21.190845489501953, "learning_rate": 6.160194174757282e-05, "loss": 0.2645, "step": 7910 }, { "epoch": 76.70843373493976, "grad_norm": 5.918322563171387, "learning_rate": 6.159708737864078e-05, "loss": 0.1532, "step": 7911 }, { "epoch": 76.71807228915662, "grad_norm": 30.565547943115234, "learning_rate": 6.159223300970874e-05, "loss": 0.4011, "step": 7912 }, { "epoch": 76.7277108433735, "grad_norm": 18.696399688720703, "learning_rate": 6.158737864077671e-05, "loss": 0.3826, "step": 7913 }, { "epoch": 76.73734939759036, "grad_norm": 9.77871322631836, "learning_rate": 6.158252427184467e-05, "loss": 0.3298, "step": 7914 }, { "epoch": 76.74698795180723, "grad_norm": 28.185880661010742, "learning_rate": 6.157766990291263e-05, "loss": 0.4889, "step": 7915 }, { "epoch": 76.75662650602409, "grad_norm": 7.563174724578857, "learning_rate": 6.157281553398058e-05, "loss": 0.3776, "step": 7916 }, { "epoch": 76.76626506024097, "grad_norm": 8.649179458618164, "learning_rate": 6.156796116504855e-05, "loss": 0.2452, "step": 7917 }, { "epoch": 76.77590361445783, "grad_norm": 4.1661481857299805, "learning_rate": 6.156310679611651e-05, "loss": 0.1243, "step": 7918 }, { "epoch": 76.7855421686747, "grad_norm": 12.135919570922852, "learning_rate": 6.155825242718447e-05, "loss": 0.2846, "step": 7919 }, { "epoch": 76.79518072289157, "grad_norm": 14.86008071899414, "learning_rate": 6.155339805825243e-05, "loss": 0.2755, "step": 7920 }, { "epoch": 76.80481927710844, "grad_norm": 7.372538089752197, "learning_rate": 6.154854368932039e-05, "loss": 0.2951, "step": 7921 }, { "epoch": 76.8144578313253, "grad_norm": 11.023324012756348, "learning_rate": 6.154368932038836e-05, "loss": 0.2731, "step": 7922 }, { "epoch": 76.82409638554216, "grad_norm": 6.319502353668213, "learning_rate": 6.153883495145631e-05, "loss": 0.3563, "step": 7923 }, { "epoch": 76.83373493975904, "grad_norm": 26.319734573364258, "learning_rate": 6.153398058252427e-05, "loss": 0.4781, "step": 7924 }, { "epoch": 76.8433734939759, "grad_norm": 6.871543884277344, "learning_rate": 6.152912621359223e-05, "loss": 0.336, "step": 7925 }, { "epoch": 76.85301204819277, "grad_norm": 6.842790126800537, "learning_rate": 6.15242718446602e-05, "loss": 0.3302, "step": 7926 }, { "epoch": 76.86265060240964, "grad_norm": 12.891127586364746, "learning_rate": 6.151941747572816e-05, "loss": 0.3792, "step": 7927 }, { "epoch": 76.87228915662651, "grad_norm": 28.265111923217773, "learning_rate": 6.151456310679612e-05, "loss": 0.1228, "step": 7928 }, { "epoch": 76.88192771084337, "grad_norm": 10.233856201171875, "learning_rate": 6.150970873786408e-05, "loss": 0.2783, "step": 7929 }, { "epoch": 76.89156626506023, "grad_norm": 18.259048461914062, "learning_rate": 6.150485436893205e-05, "loss": 0.3556, "step": 7930 }, { "epoch": 76.90120481927711, "grad_norm": 4.745781421661377, "learning_rate": 6.15e-05, "loss": 0.2483, "step": 7931 }, { "epoch": 76.91084337349398, "grad_norm": 8.02773380279541, "learning_rate": 6.149514563106796e-05, "loss": 0.1564, "step": 7932 }, { "epoch": 76.92048192771084, "grad_norm": 4.517307281494141, "learning_rate": 6.149029126213592e-05, "loss": 0.1586, "step": 7933 }, { "epoch": 76.93012048192772, "grad_norm": 15.183832168579102, "learning_rate": 6.148543689320389e-05, "loss": 0.3694, "step": 7934 }, { "epoch": 76.93975903614458, "grad_norm": 38.127235412597656, "learning_rate": 6.148058252427185e-05, "loss": 0.1737, "step": 7935 }, { "epoch": 76.94939759036144, "grad_norm": 27.556434631347656, "learning_rate": 6.14757281553398e-05, "loss": 0.3328, "step": 7936 }, { "epoch": 76.95903614457832, "grad_norm": 4.256343841552734, "learning_rate": 6.147087378640776e-05, "loss": 0.1958, "step": 7937 }, { "epoch": 76.96867469879518, "grad_norm": 5.214723587036133, "learning_rate": 6.146601941747572e-05, "loss": 0.3109, "step": 7938 }, { "epoch": 76.97831325301205, "grad_norm": 17.687219619750977, "learning_rate": 6.14611650485437e-05, "loss": 0.2438, "step": 7939 }, { "epoch": 76.98795180722891, "grad_norm": 16.867908477783203, "learning_rate": 6.145631067961165e-05, "loss": 0.365, "step": 7940 }, { "epoch": 77.00361445783132, "grad_norm": 18.684778213500977, "learning_rate": 6.145145631067961e-05, "loss": 0.2565, "step": 7941 }, { "epoch": 77.0132530120482, "grad_norm": 4.545130252838135, "learning_rate": 6.144660194174757e-05, "loss": 0.1995, "step": 7942 }, { "epoch": 77.02289156626506, "grad_norm": 4.41927433013916, "learning_rate": 6.144174757281554e-05, "loss": 0.183, "step": 7943 }, { "epoch": 77.03253012048192, "grad_norm": 10.8616943359375, "learning_rate": 6.14368932038835e-05, "loss": 0.4114, "step": 7944 }, { "epoch": 77.0421686746988, "grad_norm": 16.671985626220703, "learning_rate": 6.143203883495145e-05, "loss": 0.375, "step": 7945 }, { "epoch": 77.05180722891566, "grad_norm": 10.466780662536621, "learning_rate": 6.142718446601941e-05, "loss": 0.4342, "step": 7946 }, { "epoch": 77.06144578313253, "grad_norm": 6.505780220031738, "learning_rate": 6.142233009708738e-05, "loss": 0.3394, "step": 7947 }, { "epoch": 77.0710843373494, "grad_norm": 7.867213249206543, "learning_rate": 6.141747572815534e-05, "loss": 0.1475, "step": 7948 }, { "epoch": 77.08072289156627, "grad_norm": 4.781960487365723, "learning_rate": 6.14126213592233e-05, "loss": 0.2026, "step": 7949 }, { "epoch": 77.09036144578313, "grad_norm": 32.032169342041016, "learning_rate": 6.140776699029126e-05, "loss": 0.3327, "step": 7950 }, { "epoch": 77.1, "grad_norm": 8.797717094421387, "learning_rate": 6.140291262135923e-05, "loss": 0.221, "step": 7951 }, { "epoch": 77.10963855421687, "grad_norm": 19.244482040405273, "learning_rate": 6.139805825242718e-05, "loss": 0.3516, "step": 7952 }, { "epoch": 77.11927710843374, "grad_norm": 6.41421365737915, "learning_rate": 6.139320388349516e-05, "loss": 0.3298, "step": 7953 }, { "epoch": 77.1289156626506, "grad_norm": 31.623218536376953, "learning_rate": 6.138834951456311e-05, "loss": 0.4321, "step": 7954 }, { "epoch": 77.13855421686748, "grad_norm": 8.255788803100586, "learning_rate": 6.138349514563107e-05, "loss": 0.3381, "step": 7955 }, { "epoch": 77.14819277108434, "grad_norm": 14.276327133178711, "learning_rate": 6.137864077669904e-05, "loss": 0.3339, "step": 7956 }, { "epoch": 77.1578313253012, "grad_norm": 33.28899383544922, "learning_rate": 6.1373786407767e-05, "loss": 0.4192, "step": 7957 }, { "epoch": 77.16746987951807, "grad_norm": 13.462610244750977, "learning_rate": 6.136893203883496e-05, "loss": 0.3134, "step": 7958 }, { "epoch": 77.17710843373494, "grad_norm": 14.171197891235352, "learning_rate": 6.136407766990292e-05, "loss": 0.3393, "step": 7959 }, { "epoch": 77.1867469879518, "grad_norm": 4.370251655578613, "learning_rate": 6.135922330097087e-05, "loss": 0.1668, "step": 7960 }, { "epoch": 77.19638554216867, "grad_norm": 1.1883342266082764, "learning_rate": 6.135436893203885e-05, "loss": 0.0453, "step": 7961 }, { "epoch": 77.20602409638555, "grad_norm": 8.182515144348145, "learning_rate": 6.13495145631068e-05, "loss": 0.1794, "step": 7962 }, { "epoch": 77.21566265060241, "grad_norm": 12.362141609191895, "learning_rate": 6.134466019417476e-05, "loss": 0.4242, "step": 7963 }, { "epoch": 77.22530120481927, "grad_norm": 16.888837814331055, "learning_rate": 6.133980582524272e-05, "loss": 0.4161, "step": 7964 }, { "epoch": 77.23493975903614, "grad_norm": 4.985570430755615, "learning_rate": 6.133495145631069e-05, "loss": 0.2898, "step": 7965 }, { "epoch": 77.24457831325302, "grad_norm": 28.995393753051758, "learning_rate": 6.133009708737865e-05, "loss": 0.2644, "step": 7966 }, { "epoch": 77.25421686746988, "grad_norm": 13.021946907043457, "learning_rate": 6.13252427184466e-05, "loss": 0.6004, "step": 7967 }, { "epoch": 77.26385542168674, "grad_norm": 8.133496284484863, "learning_rate": 6.132038834951456e-05, "loss": 0.2415, "step": 7968 }, { "epoch": 77.27349397590362, "grad_norm": 8.771163940429688, "learning_rate": 6.131553398058253e-05, "loss": 0.3714, "step": 7969 }, { "epoch": 77.28313253012048, "grad_norm": 7.6585693359375, "learning_rate": 6.131067961165049e-05, "loss": 0.2252, "step": 7970 }, { "epoch": 77.29277108433735, "grad_norm": 15.56506061553955, "learning_rate": 6.130582524271845e-05, "loss": 0.2806, "step": 7971 }, { "epoch": 77.30240963855422, "grad_norm": 14.072212219238281, "learning_rate": 6.130097087378641e-05, "loss": 0.3998, "step": 7972 }, { "epoch": 77.31204819277109, "grad_norm": 17.402788162231445, "learning_rate": 6.129611650485438e-05, "loss": 0.3787, "step": 7973 }, { "epoch": 77.32168674698795, "grad_norm": 11.101677894592285, "learning_rate": 6.129126213592234e-05, "loss": 0.4758, "step": 7974 }, { "epoch": 77.33132530120481, "grad_norm": 8.321344375610352, "learning_rate": 6.12864077669903e-05, "loss": 0.3916, "step": 7975 }, { "epoch": 77.34096385542169, "grad_norm": 8.328547477722168, "learning_rate": 6.128155339805825e-05, "loss": 0.4207, "step": 7976 }, { "epoch": 77.35060240963855, "grad_norm": 9.684812545776367, "learning_rate": 6.127669902912621e-05, "loss": 0.3892, "step": 7977 }, { "epoch": 77.36024096385542, "grad_norm": 9.081186294555664, "learning_rate": 6.127184466019418e-05, "loss": 0.2948, "step": 7978 }, { "epoch": 77.3698795180723, "grad_norm": 30.02557945251465, "learning_rate": 6.126699029126214e-05, "loss": 0.3738, "step": 7979 }, { "epoch": 77.37951807228916, "grad_norm": 17.371673583984375, "learning_rate": 6.12621359223301e-05, "loss": 0.2775, "step": 7980 }, { "epoch": 77.38915662650602, "grad_norm": 5.7056779861450195, "learning_rate": 6.125728155339805e-05, "loss": 0.1824, "step": 7981 }, { "epoch": 77.39879518072289, "grad_norm": 10.220314979553223, "learning_rate": 6.125242718446603e-05, "loss": 0.3853, "step": 7982 }, { "epoch": 77.40843373493976, "grad_norm": 6.301544666290283, "learning_rate": 6.124757281553398e-05, "loss": 0.2269, "step": 7983 }, { "epoch": 77.41807228915663, "grad_norm": 7.266261100769043, "learning_rate": 6.124271844660194e-05, "loss": 0.318, "step": 7984 }, { "epoch": 77.42771084337349, "grad_norm": 7.6300201416015625, "learning_rate": 6.12378640776699e-05, "loss": 0.3607, "step": 7985 }, { "epoch": 77.43734939759037, "grad_norm": 6.416317462921143, "learning_rate": 6.123300970873787e-05, "loss": 0.311, "step": 7986 }, { "epoch": 77.44698795180723, "grad_norm": 11.800814628601074, "learning_rate": 6.122815533980583e-05, "loss": 0.3769, "step": 7987 }, { "epoch": 77.4566265060241, "grad_norm": 14.609283447265625, "learning_rate": 6.122330097087379e-05, "loss": 0.2303, "step": 7988 }, { "epoch": 77.46626506024096, "grad_norm": 38.72018814086914, "learning_rate": 6.121844660194174e-05, "loss": 0.225, "step": 7989 }, { "epoch": 77.47590361445783, "grad_norm": 9.023536682128906, "learning_rate": 6.121359223300972e-05, "loss": 0.3815, "step": 7990 }, { "epoch": 77.4855421686747, "grad_norm": 25.06102752685547, "learning_rate": 6.120873786407767e-05, "loss": 0.4002, "step": 7991 }, { "epoch": 77.49518072289156, "grad_norm": 10.329904556274414, "learning_rate": 6.120388349514563e-05, "loss": 0.2609, "step": 7992 }, { "epoch": 77.50481927710844, "grad_norm": 25.86661720275879, "learning_rate": 6.119902912621359e-05, "loss": 0.4413, "step": 7993 }, { "epoch": 77.5144578313253, "grad_norm": 5.770347595214844, "learning_rate": 6.119417475728155e-05, "loss": 0.1475, "step": 7994 }, { "epoch": 77.52409638554217, "grad_norm": 20.453094482421875, "learning_rate": 6.118932038834952e-05, "loss": 0.5253, "step": 7995 }, { "epoch": 77.53373493975904, "grad_norm": 9.010120391845703, "learning_rate": 6.118446601941748e-05, "loss": 0.393, "step": 7996 }, { "epoch": 77.5433734939759, "grad_norm": 6.20469856262207, "learning_rate": 6.117961165048543e-05, "loss": 0.1101, "step": 7997 }, { "epoch": 77.55301204819277, "grad_norm": 8.237406730651855, "learning_rate": 6.117475728155339e-05, "loss": 0.3601, "step": 7998 }, { "epoch": 77.56265060240963, "grad_norm": 3.9544613361358643, "learning_rate": 6.116990291262136e-05, "loss": 0.1931, "step": 7999 }, { "epoch": 77.57228915662651, "grad_norm": 12.042486190795898, "learning_rate": 6.116504854368932e-05, "loss": 0.3547, "step": 8000 }, { "epoch": 77.58192771084337, "grad_norm": 8.183585166931152, "learning_rate": 6.116019417475728e-05, "loss": 0.3229, "step": 8001 }, { "epoch": 77.59156626506024, "grad_norm": 7.998006343841553, "learning_rate": 6.115533980582524e-05, "loss": 0.1828, "step": 8002 }, { "epoch": 77.60120481927711, "grad_norm": 14.184162139892578, "learning_rate": 6.11504854368932e-05, "loss": 0.3792, "step": 8003 }, { "epoch": 77.61084337349398, "grad_norm": 10.878597259521484, "learning_rate": 6.114563106796118e-05, "loss": 0.3155, "step": 8004 }, { "epoch": 77.62048192771084, "grad_norm": 5.628983497619629, "learning_rate": 6.114077669902914e-05, "loss": 0.2127, "step": 8005 }, { "epoch": 77.6301204819277, "grad_norm": 2.5313167572021484, "learning_rate": 6.11359223300971e-05, "loss": 0.1442, "step": 8006 }, { "epoch": 77.63975903614458, "grad_norm": 37.383888244628906, "learning_rate": 6.113106796116505e-05, "loss": 0.3616, "step": 8007 }, { "epoch": 77.64939759036145, "grad_norm": 9.230957984924316, "learning_rate": 6.112621359223302e-05, "loss": 0.4334, "step": 8008 }, { "epoch": 77.65903614457831, "grad_norm": 6.345993995666504, "learning_rate": 6.112135922330098e-05, "loss": 0.2529, "step": 8009 }, { "epoch": 77.66867469879519, "grad_norm": 5.07950496673584, "learning_rate": 6.111650485436894e-05, "loss": 0.1866, "step": 8010 }, { "epoch": 77.67831325301205, "grad_norm": 10.372097969055176, "learning_rate": 6.11116504854369e-05, "loss": 0.2818, "step": 8011 }, { "epoch": 77.68795180722891, "grad_norm": 7.998571872711182, "learning_rate": 6.110679611650487e-05, "loss": 0.4306, "step": 8012 }, { "epoch": 77.69759036144578, "grad_norm": 14.818253517150879, "learning_rate": 6.110194174757282e-05, "loss": 0.4106, "step": 8013 }, { "epoch": 77.70722891566265, "grad_norm": 5.929659366607666, "learning_rate": 6.109708737864078e-05, "loss": 0.1799, "step": 8014 }, { "epoch": 77.71686746987952, "grad_norm": 18.57018280029297, "learning_rate": 6.109223300970874e-05, "loss": 0.5021, "step": 8015 }, { "epoch": 77.72650602409638, "grad_norm": 14.591517448425293, "learning_rate": 6.10873786407767e-05, "loss": 0.2384, "step": 8016 }, { "epoch": 77.73614457831326, "grad_norm": 9.075658798217773, "learning_rate": 6.108252427184467e-05, "loss": 0.5937, "step": 8017 }, { "epoch": 77.74578313253012, "grad_norm": 14.352015495300293, "learning_rate": 6.107766990291263e-05, "loss": 0.244, "step": 8018 }, { "epoch": 77.75542168674698, "grad_norm": 12.334345817565918, "learning_rate": 6.107281553398058e-05, "loss": 0.2939, "step": 8019 }, { "epoch": 77.76506024096386, "grad_norm": 10.16297721862793, "learning_rate": 6.106796116504854e-05, "loss": 0.179, "step": 8020 }, { "epoch": 77.77469879518073, "grad_norm": 6.244452953338623, "learning_rate": 6.106310679611651e-05, "loss": 0.2001, "step": 8021 }, { "epoch": 77.78433734939759, "grad_norm": 9.392842292785645, "learning_rate": 6.105825242718447e-05, "loss": 0.4369, "step": 8022 }, { "epoch": 77.79397590361445, "grad_norm": 38.24441909790039, "learning_rate": 6.105339805825243e-05, "loss": 0.3848, "step": 8023 }, { "epoch": 77.80361445783133, "grad_norm": 5.313321590423584, "learning_rate": 6.104854368932039e-05, "loss": 0.1879, "step": 8024 }, { "epoch": 77.8132530120482, "grad_norm": 3.639146327972412, "learning_rate": 6.104368932038836e-05, "loss": 0.2887, "step": 8025 }, { "epoch": 77.82289156626506, "grad_norm": 26.406391143798828, "learning_rate": 6.103883495145632e-05, "loss": 0.3076, "step": 8026 }, { "epoch": 77.83253012048193, "grad_norm": 11.78233814239502, "learning_rate": 6.1033980582524274e-05, "loss": 0.5692, "step": 8027 }, { "epoch": 77.8421686746988, "grad_norm": 6.701729774475098, "learning_rate": 6.102912621359224e-05, "loss": 0.2231, "step": 8028 }, { "epoch": 77.85180722891566, "grad_norm": 10.957231521606445, "learning_rate": 6.1024271844660196e-05, "loss": 0.3976, "step": 8029 }, { "epoch": 77.86144578313252, "grad_norm": 9.857926368713379, "learning_rate": 6.101941747572816e-05, "loss": 0.2318, "step": 8030 }, { "epoch": 77.8710843373494, "grad_norm": 15.026426315307617, "learning_rate": 6.101456310679612e-05, "loss": 0.3296, "step": 8031 }, { "epoch": 77.88072289156626, "grad_norm": 7.2064714431762695, "learning_rate": 6.1009708737864076e-05, "loss": 0.4899, "step": 8032 }, { "epoch": 77.89036144578313, "grad_norm": 6.448047637939453, "learning_rate": 6.100485436893204e-05, "loss": 0.2015, "step": 8033 }, { "epoch": 77.9, "grad_norm": 19.043031692504883, "learning_rate": 6.1e-05, "loss": 0.4144, "step": 8034 }, { "epoch": 77.90963855421687, "grad_norm": 25.80837631225586, "learning_rate": 6.099514563106796e-05, "loss": 0.563, "step": 8035 }, { "epoch": 77.91927710843373, "grad_norm": 7.536501407623291, "learning_rate": 6.099029126213592e-05, "loss": 0.2473, "step": 8036 }, { "epoch": 77.9289156626506, "grad_norm": 12.312479972839355, "learning_rate": 6.0985436893203886e-05, "loss": 0.3293, "step": 8037 }, { "epoch": 77.93855421686747, "grad_norm": 6.87277364730835, "learning_rate": 6.098058252427184e-05, "loss": 0.3717, "step": 8038 }, { "epoch": 77.94819277108434, "grad_norm": 5.1503729820251465, "learning_rate": 6.097572815533981e-05, "loss": 0.09, "step": 8039 }, { "epoch": 77.9578313253012, "grad_norm": 13.454160690307617, "learning_rate": 6.0970873786407766e-05, "loss": 0.4574, "step": 8040 }, { "epoch": 77.96746987951808, "grad_norm": 8.208318710327148, "learning_rate": 6.096601941747573e-05, "loss": 0.3252, "step": 8041 }, { "epoch": 77.97710843373494, "grad_norm": 7.386109352111816, "learning_rate": 6.096116504854369e-05, "loss": 0.2941, "step": 8042 }, { "epoch": 77.9867469879518, "grad_norm": 35.537784576416016, "learning_rate": 6.095631067961165e-05, "loss": 0.3393, "step": 8043 }, { "epoch": 78.00240963855421, "grad_norm": 13.224456787109375, "learning_rate": 6.095145631067961e-05, "loss": 0.4691, "step": 8044 }, { "epoch": 78.01204819277109, "grad_norm": 7.192540168762207, "learning_rate": 6.0946601941747575e-05, "loss": 0.4861, "step": 8045 }, { "epoch": 78.02168674698795, "grad_norm": 7.699985504150391, "learning_rate": 6.094174757281553e-05, "loss": 0.2134, "step": 8046 }, { "epoch": 78.03132530120482, "grad_norm": 14.313530921936035, "learning_rate": 6.09368932038835e-05, "loss": 0.3509, "step": 8047 }, { "epoch": 78.04096385542168, "grad_norm": 8.117380142211914, "learning_rate": 6.0932038834951455e-05, "loss": 0.2034, "step": 8048 }, { "epoch": 78.05060240963856, "grad_norm": 9.463349342346191, "learning_rate": 6.092718446601941e-05, "loss": 0.2811, "step": 8049 }, { "epoch": 78.06024096385542, "grad_norm": 39.06373596191406, "learning_rate": 6.092233009708738e-05, "loss": 0.2852, "step": 8050 }, { "epoch": 78.06987951807228, "grad_norm": 6.838380336761475, "learning_rate": 6.0917475728155335e-05, "loss": 0.3031, "step": 8051 }, { "epoch": 78.07951807228916, "grad_norm": 5.933814525604248, "learning_rate": 6.09126213592233e-05, "loss": 0.2053, "step": 8052 }, { "epoch": 78.08915662650602, "grad_norm": 5.818180561065674, "learning_rate": 6.090776699029126e-05, "loss": 0.307, "step": 8053 }, { "epoch": 78.09879518072289, "grad_norm": 10.10960578918457, "learning_rate": 6.090291262135923e-05, "loss": 0.2857, "step": 8054 }, { "epoch": 78.10843373493977, "grad_norm": 8.790959358215332, "learning_rate": 6.089805825242719e-05, "loss": 0.5015, "step": 8055 }, { "epoch": 78.11807228915663, "grad_norm": 10.314340591430664, "learning_rate": 6.089320388349515e-05, "loss": 0.3113, "step": 8056 }, { "epoch": 78.12771084337349, "grad_norm": 9.607742309570312, "learning_rate": 6.0888349514563115e-05, "loss": 0.5174, "step": 8057 }, { "epoch": 78.13734939759036, "grad_norm": 6.088291168212891, "learning_rate": 6.088349514563107e-05, "loss": 0.1879, "step": 8058 }, { "epoch": 78.14698795180723, "grad_norm": 7.06526517868042, "learning_rate": 6.087864077669904e-05, "loss": 0.3641, "step": 8059 }, { "epoch": 78.1566265060241, "grad_norm": 57.831275939941406, "learning_rate": 6.0873786407766995e-05, "loss": 0.1919, "step": 8060 }, { "epoch": 78.16626506024096, "grad_norm": 6.611273765563965, "learning_rate": 6.086893203883496e-05, "loss": 0.2613, "step": 8061 }, { "epoch": 78.17590361445784, "grad_norm": 8.987013816833496, "learning_rate": 6.086407766990292e-05, "loss": 0.3289, "step": 8062 }, { "epoch": 78.1855421686747, "grad_norm": 13.059560775756836, "learning_rate": 6.085922330097088e-05, "loss": 0.4979, "step": 8063 }, { "epoch": 78.19518072289156, "grad_norm": 16.73979377746582, "learning_rate": 6.085436893203884e-05, "loss": 0.3209, "step": 8064 }, { "epoch": 78.20481927710843, "grad_norm": 7.310500144958496, "learning_rate": 6.0849514563106804e-05, "loss": 0.2481, "step": 8065 }, { "epoch": 78.2144578313253, "grad_norm": 7.565945625305176, "learning_rate": 6.084466019417476e-05, "loss": 0.3369, "step": 8066 }, { "epoch": 78.22409638554217, "grad_norm": 8.896239280700684, "learning_rate": 6.083980582524273e-05, "loss": 0.6368, "step": 8067 }, { "epoch": 78.23373493975903, "grad_norm": 7.526298999786377, "learning_rate": 6.0834951456310684e-05, "loss": 0.2633, "step": 8068 }, { "epoch": 78.24337349397591, "grad_norm": 21.32575035095215, "learning_rate": 6.083009708737865e-05, "loss": 0.4603, "step": 8069 }, { "epoch": 78.25301204819277, "grad_norm": 32.441287994384766, "learning_rate": 6.082524271844661e-05, "loss": 0.2754, "step": 8070 }, { "epoch": 78.26265060240964, "grad_norm": 18.11207389831543, "learning_rate": 6.0820388349514565e-05, "loss": 0.295, "step": 8071 }, { "epoch": 78.2722891566265, "grad_norm": 23.616846084594727, "learning_rate": 6.081553398058253e-05, "loss": 0.504, "step": 8072 }, { "epoch": 78.28192771084338, "grad_norm": 8.364171981811523, "learning_rate": 6.081067961165049e-05, "loss": 0.2869, "step": 8073 }, { "epoch": 78.29156626506024, "grad_norm": 6.691619873046875, "learning_rate": 6.080582524271845e-05, "loss": 0.2995, "step": 8074 }, { "epoch": 78.3012048192771, "grad_norm": 7.370383262634277, "learning_rate": 6.080097087378641e-05, "loss": 0.2156, "step": 8075 }, { "epoch": 78.31084337349398, "grad_norm": 7.0226640701293945, "learning_rate": 6.0796116504854374e-05, "loss": 0.3083, "step": 8076 }, { "epoch": 78.32048192771084, "grad_norm": 8.337292671203613, "learning_rate": 6.079126213592233e-05, "loss": 0.0795, "step": 8077 }, { "epoch": 78.33012048192771, "grad_norm": 16.782190322875977, "learning_rate": 6.0786407766990296e-05, "loss": 0.4177, "step": 8078 }, { "epoch": 78.33975903614459, "grad_norm": 17.97268295288086, "learning_rate": 6.0781553398058254e-05, "loss": 0.3373, "step": 8079 }, { "epoch": 78.34939759036145, "grad_norm": 8.442547798156738, "learning_rate": 6.077669902912622e-05, "loss": 0.2345, "step": 8080 }, { "epoch": 78.35903614457831, "grad_norm": 12.170958518981934, "learning_rate": 6.0771844660194176e-05, "loss": 0.5168, "step": 8081 }, { "epoch": 78.36867469879518, "grad_norm": 7.473206043243408, "learning_rate": 6.076699029126214e-05, "loss": 0.1367, "step": 8082 }, { "epoch": 78.37831325301205, "grad_norm": 39.30195617675781, "learning_rate": 6.07621359223301e-05, "loss": 0.4049, "step": 8083 }, { "epoch": 78.38795180722892, "grad_norm": 17.1811466217041, "learning_rate": 6.075728155339806e-05, "loss": 0.2267, "step": 8084 }, { "epoch": 78.39759036144578, "grad_norm": 4.743587970733643, "learning_rate": 6.075242718446602e-05, "loss": 0.0946, "step": 8085 }, { "epoch": 78.40722891566266, "grad_norm": 13.595513343811035, "learning_rate": 6.0747572815533985e-05, "loss": 0.3612, "step": 8086 }, { "epoch": 78.41686746987952, "grad_norm": 5.09011173248291, "learning_rate": 6.074271844660194e-05, "loss": 0.2695, "step": 8087 }, { "epoch": 78.42650602409638, "grad_norm": 7.925656318664551, "learning_rate": 6.07378640776699e-05, "loss": 0.633, "step": 8088 }, { "epoch": 78.43614457831325, "grad_norm": 41.30249786376953, "learning_rate": 6.0733009708737865e-05, "loss": 0.2963, "step": 8089 }, { "epoch": 78.44578313253012, "grad_norm": 10.305331230163574, "learning_rate": 6.072815533980582e-05, "loss": 0.2518, "step": 8090 }, { "epoch": 78.45542168674699, "grad_norm": 11.003633499145508, "learning_rate": 6.072330097087379e-05, "loss": 0.2781, "step": 8091 }, { "epoch": 78.46506024096385, "grad_norm": 2.798888921737671, "learning_rate": 6.0718446601941745e-05, "loss": 0.1661, "step": 8092 }, { "epoch": 78.47469879518073, "grad_norm": 7.790566921234131, "learning_rate": 6.071359223300971e-05, "loss": 0.3266, "step": 8093 }, { "epoch": 78.48433734939759, "grad_norm": 8.93099308013916, "learning_rate": 6.070873786407767e-05, "loss": 0.3485, "step": 8094 }, { "epoch": 78.49397590361446, "grad_norm": 16.438203811645508, "learning_rate": 6.070388349514563e-05, "loss": 0.5992, "step": 8095 }, { "epoch": 78.50361445783132, "grad_norm": 19.177091598510742, "learning_rate": 6.069902912621359e-05, "loss": 0.2775, "step": 8096 }, { "epoch": 78.5132530120482, "grad_norm": 26.324520111083984, "learning_rate": 6.0694174757281554e-05, "loss": 0.3651, "step": 8097 }, { "epoch": 78.52289156626506, "grad_norm": 13.438678741455078, "learning_rate": 6.068932038834951e-05, "loss": 0.2956, "step": 8098 }, { "epoch": 78.53253012048192, "grad_norm": 5.694369792938232, "learning_rate": 6.0684466019417477e-05, "loss": 0.2735, "step": 8099 }, { "epoch": 78.5421686746988, "grad_norm": 12.186934471130371, "learning_rate": 6.0679611650485434e-05, "loss": 0.3491, "step": 8100 }, { "epoch": 78.55180722891566, "grad_norm": 9.660648345947266, "learning_rate": 6.06747572815534e-05, "loss": 0.2763, "step": 8101 }, { "epoch": 78.56144578313253, "grad_norm": 14.120985984802246, "learning_rate": 6.066990291262136e-05, "loss": 0.3422, "step": 8102 }, { "epoch": 78.5710843373494, "grad_norm": 14.867323875427246, "learning_rate": 6.066504854368932e-05, "loss": 0.1873, "step": 8103 }, { "epoch": 78.58072289156627, "grad_norm": 8.524511337280273, "learning_rate": 6.066019417475729e-05, "loss": 0.3118, "step": 8104 }, { "epoch": 78.59036144578313, "grad_norm": 4.607027530670166, "learning_rate": 6.065533980582525e-05, "loss": 0.3525, "step": 8105 }, { "epoch": 78.6, "grad_norm": 7.977153778076172, "learning_rate": 6.0650485436893215e-05, "loss": 0.3875, "step": 8106 }, { "epoch": 78.60963855421687, "grad_norm": 7.52069091796875, "learning_rate": 6.064563106796117e-05, "loss": 0.2727, "step": 8107 }, { "epoch": 78.61927710843374, "grad_norm": 15.210343360900879, "learning_rate": 6.064077669902914e-05, "loss": 0.3747, "step": 8108 }, { "epoch": 78.6289156626506, "grad_norm": 7.789963722229004, "learning_rate": 6.0635922330097095e-05, "loss": 0.456, "step": 8109 }, { "epoch": 78.63855421686748, "grad_norm": 11.170942306518555, "learning_rate": 6.063106796116505e-05, "loss": 0.2736, "step": 8110 }, { "epoch": 78.64819277108434, "grad_norm": 6.227780342102051, "learning_rate": 6.062621359223302e-05, "loss": 0.3831, "step": 8111 }, { "epoch": 78.6578313253012, "grad_norm": 12.149619102478027, "learning_rate": 6.0621359223300975e-05, "loss": 0.1772, "step": 8112 }, { "epoch": 78.66746987951807, "grad_norm": 6.787862777709961, "learning_rate": 6.061650485436894e-05, "loss": 0.2756, "step": 8113 }, { "epoch": 78.67710843373494, "grad_norm": 7.062891006469727, "learning_rate": 6.06116504854369e-05, "loss": 0.2431, "step": 8114 }, { "epoch": 78.6867469879518, "grad_norm": 17.90228271484375, "learning_rate": 6.060679611650486e-05, "loss": 0.2112, "step": 8115 }, { "epoch": 78.69638554216867, "grad_norm": 3.692312479019165, "learning_rate": 6.060194174757282e-05, "loss": 0.1469, "step": 8116 }, { "epoch": 78.70602409638555, "grad_norm": 20.736196517944336, "learning_rate": 6.0597087378640784e-05, "loss": 0.2449, "step": 8117 }, { "epoch": 78.71566265060241, "grad_norm": 6.7666707038879395, "learning_rate": 6.059223300970874e-05, "loss": 0.2606, "step": 8118 }, { "epoch": 78.72530120481927, "grad_norm": 9.956622123718262, "learning_rate": 6.0587378640776706e-05, "loss": 0.1803, "step": 8119 }, { "epoch": 78.73493975903614, "grad_norm": 10.333930015563965, "learning_rate": 6.0582524271844664e-05, "loss": 0.4384, "step": 8120 }, { "epoch": 78.74457831325302, "grad_norm": 12.753165245056152, "learning_rate": 6.057766990291263e-05, "loss": 0.3487, "step": 8121 }, { "epoch": 78.75421686746988, "grad_norm": 4.092166423797607, "learning_rate": 6.0572815533980586e-05, "loss": 0.2251, "step": 8122 }, { "epoch": 78.76385542168674, "grad_norm": 19.988330841064453, "learning_rate": 6.056796116504855e-05, "loss": 0.3191, "step": 8123 }, { "epoch": 78.77349397590362, "grad_norm": 9.778780937194824, "learning_rate": 6.056310679611651e-05, "loss": 0.3282, "step": 8124 }, { "epoch": 78.78313253012048, "grad_norm": 6.904632091522217, "learning_rate": 6.055825242718447e-05, "loss": 0.282, "step": 8125 }, { "epoch": 78.79277108433735, "grad_norm": 6.653721809387207, "learning_rate": 6.055339805825243e-05, "loss": 0.2304, "step": 8126 }, { "epoch": 78.80240963855422, "grad_norm": 6.918817520141602, "learning_rate": 6.054854368932039e-05, "loss": 0.2276, "step": 8127 }, { "epoch": 78.81204819277109, "grad_norm": 28.146310806274414, "learning_rate": 6.054368932038835e-05, "loss": 0.3278, "step": 8128 }, { "epoch": 78.82168674698795, "grad_norm": 3.7951271533966064, "learning_rate": 6.053883495145631e-05, "loss": 0.0789, "step": 8129 }, { "epoch": 78.83132530120481, "grad_norm": 11.29300594329834, "learning_rate": 6.0533980582524275e-05, "loss": 0.2582, "step": 8130 }, { "epoch": 78.84096385542169, "grad_norm": 31.690956115722656, "learning_rate": 6.052912621359223e-05, "loss": 0.3898, "step": 8131 }, { "epoch": 78.85060240963855, "grad_norm": 7.865332126617432, "learning_rate": 6.05242718446602e-05, "loss": 0.3033, "step": 8132 }, { "epoch": 78.86024096385542, "grad_norm": 6.6068572998046875, "learning_rate": 6.0519417475728156e-05, "loss": 0.1521, "step": 8133 }, { "epoch": 78.8698795180723, "grad_norm": 16.421228408813477, "learning_rate": 6.051456310679612e-05, "loss": 0.1495, "step": 8134 }, { "epoch": 78.87951807228916, "grad_norm": 6.191936492919922, "learning_rate": 6.050970873786408e-05, "loss": 0.3149, "step": 8135 }, { "epoch": 78.88915662650602, "grad_norm": 17.25916290283203, "learning_rate": 6.050485436893204e-05, "loss": 0.2991, "step": 8136 }, { "epoch": 78.89879518072289, "grad_norm": 10.076903343200684, "learning_rate": 6.05e-05, "loss": 0.4083, "step": 8137 }, { "epoch": 78.90843373493976, "grad_norm": 11.344497680664062, "learning_rate": 6.0495145631067965e-05, "loss": 0.3046, "step": 8138 }, { "epoch": 78.91807228915663, "grad_norm": 17.881986618041992, "learning_rate": 6.049029126213592e-05, "loss": 0.3973, "step": 8139 }, { "epoch": 78.92771084337349, "grad_norm": 8.26010513305664, "learning_rate": 6.048543689320389e-05, "loss": 0.332, "step": 8140 }, { "epoch": 78.93734939759037, "grad_norm": 5.626612186431885, "learning_rate": 6.0480582524271845e-05, "loss": 0.1768, "step": 8141 }, { "epoch": 78.94698795180723, "grad_norm": 4.045575141906738, "learning_rate": 6.04757281553398e-05, "loss": 0.2508, "step": 8142 }, { "epoch": 78.9566265060241, "grad_norm": 3.853586435317993, "learning_rate": 6.047087378640777e-05, "loss": 0.1462, "step": 8143 }, { "epoch": 78.96626506024096, "grad_norm": 10.469511032104492, "learning_rate": 6.0466019417475725e-05, "loss": 0.2875, "step": 8144 }, { "epoch": 78.97590361445783, "grad_norm": 11.46811580657959, "learning_rate": 6.046116504854369e-05, "loss": 0.1722, "step": 8145 }, { "epoch": 78.9855421686747, "grad_norm": 7.373281478881836, "learning_rate": 6.045631067961165e-05, "loss": 0.3296, "step": 8146 }, { "epoch": 79.0012048192771, "grad_norm": 8.157715797424316, "learning_rate": 6.045145631067961e-05, "loss": 0.4402, "step": 8147 }, { "epoch": 79.01084337349397, "grad_norm": 5.288763046264648, "learning_rate": 6.044660194174757e-05, "loss": 0.238, "step": 8148 }, { "epoch": 79.02048192771085, "grad_norm": 14.374700546264648, "learning_rate": 6.0441747572815534e-05, "loss": 0.3708, "step": 8149 }, { "epoch": 79.03012048192771, "grad_norm": 18.36875343322754, "learning_rate": 6.043689320388349e-05, "loss": 0.2944, "step": 8150 }, { "epoch": 79.03975903614457, "grad_norm": 10.833747863769531, "learning_rate": 6.0432038834951456e-05, "loss": 0.4577, "step": 8151 }, { "epoch": 79.04939759036145, "grad_norm": 63.94399642944336, "learning_rate": 6.0427184466019414e-05, "loss": 0.3771, "step": 8152 }, { "epoch": 79.05903614457831, "grad_norm": 11.361528396606445, "learning_rate": 6.042233009708738e-05, "loss": 0.4762, "step": 8153 }, { "epoch": 79.06867469879518, "grad_norm": 6.390401840209961, "learning_rate": 6.0417475728155336e-05, "loss": 0.2134, "step": 8154 }, { "epoch": 79.07831325301204, "grad_norm": 10.686927795410156, "learning_rate": 6.041262135922331e-05, "loss": 0.1876, "step": 8155 }, { "epoch": 79.08795180722892, "grad_norm": 16.961233139038086, "learning_rate": 6.040776699029127e-05, "loss": 0.5287, "step": 8156 }, { "epoch": 79.09759036144578, "grad_norm": 9.171045303344727, "learning_rate": 6.040291262135923e-05, "loss": 0.3019, "step": 8157 }, { "epoch": 79.10722891566265, "grad_norm": 5.065258502960205, "learning_rate": 6.0398058252427194e-05, "loss": 0.1806, "step": 8158 }, { "epoch": 79.11686746987952, "grad_norm": 5.904909133911133, "learning_rate": 6.039320388349515e-05, "loss": 0.1699, "step": 8159 }, { "epoch": 79.12650602409639, "grad_norm": 7.934677600860596, "learning_rate": 6.038834951456312e-05, "loss": 0.2116, "step": 8160 }, { "epoch": 79.13614457831325, "grad_norm": 18.82634735107422, "learning_rate": 6.0383495145631074e-05, "loss": 0.1562, "step": 8161 }, { "epoch": 79.14578313253013, "grad_norm": 6.538631439208984, "learning_rate": 6.037864077669904e-05, "loss": 0.2155, "step": 8162 }, { "epoch": 79.15542168674699, "grad_norm": 24.57607078552246, "learning_rate": 6.0373786407767e-05, "loss": 0.4174, "step": 8163 }, { "epoch": 79.16506024096385, "grad_norm": 8.831595420837402, "learning_rate": 6.0368932038834954e-05, "loss": 0.2188, "step": 8164 }, { "epoch": 79.17469879518072, "grad_norm": 11.705693244934082, "learning_rate": 6.036407766990292e-05, "loss": 0.1789, "step": 8165 }, { "epoch": 79.1843373493976, "grad_norm": 5.175743103027344, "learning_rate": 6.035922330097088e-05, "loss": 0.1836, "step": 8166 }, { "epoch": 79.19397590361446, "grad_norm": 10.534976959228516, "learning_rate": 6.035436893203884e-05, "loss": 0.5238, "step": 8167 }, { "epoch": 79.20361445783132, "grad_norm": 10.852956771850586, "learning_rate": 6.03495145631068e-05, "loss": 0.4403, "step": 8168 }, { "epoch": 79.2132530120482, "grad_norm": 10.942670822143555, "learning_rate": 6.0344660194174764e-05, "loss": 0.3393, "step": 8169 }, { "epoch": 79.22289156626506, "grad_norm": 16.199893951416016, "learning_rate": 6.033980582524272e-05, "loss": 0.2978, "step": 8170 }, { "epoch": 79.23253012048193, "grad_norm": 14.593478202819824, "learning_rate": 6.0334951456310686e-05, "loss": 0.2614, "step": 8171 }, { "epoch": 79.24216867469879, "grad_norm": 19.585693359375, "learning_rate": 6.0330097087378644e-05, "loss": 0.3498, "step": 8172 }, { "epoch": 79.25180722891567, "grad_norm": 11.150785446166992, "learning_rate": 6.032524271844661e-05, "loss": 0.4659, "step": 8173 }, { "epoch": 79.26144578313253, "grad_norm": 5.441066741943359, "learning_rate": 6.0320388349514566e-05, "loss": 0.3199, "step": 8174 }, { "epoch": 79.2710843373494, "grad_norm": 8.811394691467285, "learning_rate": 6.031553398058253e-05, "loss": 0.2462, "step": 8175 }, { "epoch": 79.28072289156627, "grad_norm": 5.5508527755737305, "learning_rate": 6.031067961165049e-05, "loss": 0.4629, "step": 8176 }, { "epoch": 79.29036144578313, "grad_norm": 4.48264741897583, "learning_rate": 6.030582524271845e-05, "loss": 0.2958, "step": 8177 }, { "epoch": 79.3, "grad_norm": 15.69322681427002, "learning_rate": 6.030097087378641e-05, "loss": 0.5004, "step": 8178 }, { "epoch": 79.30963855421686, "grad_norm": 8.857769966125488, "learning_rate": 6.0296116504854375e-05, "loss": 0.4358, "step": 8179 }, { "epoch": 79.31927710843374, "grad_norm": 11.927854537963867, "learning_rate": 6.029126213592233e-05, "loss": 0.4035, "step": 8180 }, { "epoch": 79.3289156626506, "grad_norm": 14.695650100708008, "learning_rate": 6.028640776699029e-05, "loss": 0.2939, "step": 8181 }, { "epoch": 79.33855421686746, "grad_norm": 19.877574920654297, "learning_rate": 6.0281553398058255e-05, "loss": 0.4832, "step": 8182 }, { "epoch": 79.34819277108434, "grad_norm": 3.501518964767456, "learning_rate": 6.027669902912621e-05, "loss": 0.1686, "step": 8183 }, { "epoch": 79.3578313253012, "grad_norm": 6.802138328552246, "learning_rate": 6.027184466019418e-05, "loss": 0.3801, "step": 8184 }, { "epoch": 79.36746987951807, "grad_norm": 5.8656487464904785, "learning_rate": 6.0266990291262135e-05, "loss": 0.0758, "step": 8185 }, { "epoch": 79.37710843373495, "grad_norm": 4.439849853515625, "learning_rate": 6.02621359223301e-05, "loss": 0.3052, "step": 8186 }, { "epoch": 79.38674698795181, "grad_norm": 10.993745803833008, "learning_rate": 6.025728155339806e-05, "loss": 0.2937, "step": 8187 }, { "epoch": 79.39638554216867, "grad_norm": 7.97095251083374, "learning_rate": 6.025242718446602e-05, "loss": 0.3407, "step": 8188 }, { "epoch": 79.40602409638554, "grad_norm": 18.41283416748047, "learning_rate": 6.024757281553398e-05, "loss": 0.3538, "step": 8189 }, { "epoch": 79.41566265060241, "grad_norm": 27.378461837768555, "learning_rate": 6.0242718446601944e-05, "loss": 0.6042, "step": 8190 }, { "epoch": 79.42530120481928, "grad_norm": 20.998693466186523, "learning_rate": 6.02378640776699e-05, "loss": 0.4951, "step": 8191 }, { "epoch": 79.43493975903614, "grad_norm": 12.133472442626953, "learning_rate": 6.0233009708737867e-05, "loss": 0.4121, "step": 8192 }, { "epoch": 79.44457831325302, "grad_norm": 8.018135070800781, "learning_rate": 6.0228155339805824e-05, "loss": 0.2039, "step": 8193 }, { "epoch": 79.45421686746988, "grad_norm": 7.609623908996582, "learning_rate": 6.022330097087379e-05, "loss": 0.1584, "step": 8194 }, { "epoch": 79.46385542168674, "grad_norm": 18.735454559326172, "learning_rate": 6.0218446601941747e-05, "loss": 0.3309, "step": 8195 }, { "epoch": 79.47349397590361, "grad_norm": 7.292303562164307, "learning_rate": 6.021359223300971e-05, "loss": 0.277, "step": 8196 }, { "epoch": 79.48313253012049, "grad_norm": 9.2860746383667, "learning_rate": 6.020873786407767e-05, "loss": 0.2262, "step": 8197 }, { "epoch": 79.49277108433735, "grad_norm": 11.657293319702148, "learning_rate": 6.020388349514563e-05, "loss": 0.2522, "step": 8198 }, { "epoch": 79.50240963855421, "grad_norm": 8.342989921569824, "learning_rate": 6.019902912621359e-05, "loss": 0.2891, "step": 8199 }, { "epoch": 79.51204819277109, "grad_norm": 6.416651248931885, "learning_rate": 6.019417475728155e-05, "loss": 0.3359, "step": 8200 }, { "epoch": 79.52168674698795, "grad_norm": 10.004833221435547, "learning_rate": 6.0189320388349513e-05, "loss": 0.4275, "step": 8201 }, { "epoch": 79.53132530120482, "grad_norm": 8.202971458435059, "learning_rate": 6.018446601941747e-05, "loss": 0.3618, "step": 8202 }, { "epoch": 79.54096385542168, "grad_norm": 5.372691631317139, "learning_rate": 6.0179611650485436e-05, "loss": 0.293, "step": 8203 }, { "epoch": 79.55060240963856, "grad_norm": 7.26524543762207, "learning_rate": 6.0174757281553393e-05, "loss": 0.299, "step": 8204 }, { "epoch": 79.56024096385542, "grad_norm": 3.316028594970703, "learning_rate": 6.0169902912621365e-05, "loss": 0.1319, "step": 8205 }, { "epoch": 79.56987951807228, "grad_norm": 11.33896255493164, "learning_rate": 6.016504854368933e-05, "loss": 0.6711, "step": 8206 }, { "epoch": 79.57951807228916, "grad_norm": 9.784873008728027, "learning_rate": 6.016019417475729e-05, "loss": 0.2777, "step": 8207 }, { "epoch": 79.58915662650602, "grad_norm": 6.683570384979248, "learning_rate": 6.015533980582525e-05, "loss": 0.1267, "step": 8208 }, { "epoch": 79.59879518072289, "grad_norm": 7.276419162750244, "learning_rate": 6.015048543689321e-05, "loss": 0.2279, "step": 8209 }, { "epoch": 79.60843373493977, "grad_norm": 7.107195854187012, "learning_rate": 6.0145631067961174e-05, "loss": 0.2377, "step": 8210 }, { "epoch": 79.61807228915663, "grad_norm": 9.988988876342773, "learning_rate": 6.014077669902913e-05, "loss": 0.1483, "step": 8211 }, { "epoch": 79.62771084337349, "grad_norm": 10.601417541503906, "learning_rate": 6.0135922330097096e-05, "loss": 0.2805, "step": 8212 }, { "epoch": 79.63734939759036, "grad_norm": 30.38473129272461, "learning_rate": 6.0131067961165054e-05, "loss": 0.4297, "step": 8213 }, { "epoch": 79.64698795180723, "grad_norm": 13.001341819763184, "learning_rate": 6.012621359223302e-05, "loss": 0.2645, "step": 8214 }, { "epoch": 79.6566265060241, "grad_norm": 5.350159645080566, "learning_rate": 6.0121359223300976e-05, "loss": 0.2345, "step": 8215 }, { "epoch": 79.66626506024096, "grad_norm": 9.41901969909668, "learning_rate": 6.011650485436894e-05, "loss": 0.4745, "step": 8216 }, { "epoch": 79.67590361445784, "grad_norm": 13.94814395904541, "learning_rate": 6.01116504854369e-05, "loss": 0.256, "step": 8217 }, { "epoch": 79.6855421686747, "grad_norm": 4.562292098999023, "learning_rate": 6.010679611650486e-05, "loss": 0.1961, "step": 8218 }, { "epoch": 79.69518072289156, "grad_norm": 17.436187744140625, "learning_rate": 6.010194174757282e-05, "loss": 0.2422, "step": 8219 }, { "epoch": 79.70481927710843, "grad_norm": 25.934371948242188, "learning_rate": 6.009708737864078e-05, "loss": 0.5075, "step": 8220 }, { "epoch": 79.7144578313253, "grad_norm": 6.701637268066406, "learning_rate": 6.009223300970874e-05, "loss": 0.32, "step": 8221 }, { "epoch": 79.72409638554217, "grad_norm": 10.54663372039795, "learning_rate": 6.00873786407767e-05, "loss": 0.1576, "step": 8222 }, { "epoch": 79.73373493975903, "grad_norm": 12.804073333740234, "learning_rate": 6.0082524271844665e-05, "loss": 0.307, "step": 8223 }, { "epoch": 79.74337349397591, "grad_norm": 9.819131851196289, "learning_rate": 6.007766990291262e-05, "loss": 0.1752, "step": 8224 }, { "epoch": 79.75301204819277, "grad_norm": 9.72387981414795, "learning_rate": 6.007281553398059e-05, "loss": 0.303, "step": 8225 }, { "epoch": 79.76265060240964, "grad_norm": 7.817143440246582, "learning_rate": 6.0067961165048545e-05, "loss": 0.2837, "step": 8226 }, { "epoch": 79.7722891566265, "grad_norm": 11.453784942626953, "learning_rate": 6.006310679611651e-05, "loss": 0.3937, "step": 8227 }, { "epoch": 79.78192771084338, "grad_norm": 9.230589866638184, "learning_rate": 6.005825242718447e-05, "loss": 0.1802, "step": 8228 }, { "epoch": 79.79156626506024, "grad_norm": 9.681415557861328, "learning_rate": 6.005339805825243e-05, "loss": 0.2728, "step": 8229 }, { "epoch": 79.8012048192771, "grad_norm": 14.865317344665527, "learning_rate": 6.004854368932039e-05, "loss": 0.2151, "step": 8230 }, { "epoch": 79.81084337349398, "grad_norm": 7.845433235168457, "learning_rate": 6.0043689320388355e-05, "loss": 0.28, "step": 8231 }, { "epoch": 79.82048192771084, "grad_norm": 19.79195213317871, "learning_rate": 6.003883495145631e-05, "loss": 0.2989, "step": 8232 }, { "epoch": 79.83012048192771, "grad_norm": 6.513072967529297, "learning_rate": 6.003398058252428e-05, "loss": 0.2416, "step": 8233 }, { "epoch": 79.83975903614459, "grad_norm": 5.431432723999023, "learning_rate": 6.0029126213592235e-05, "loss": 0.2556, "step": 8234 }, { "epoch": 79.84939759036145, "grad_norm": 5.914316654205322, "learning_rate": 6.00242718446602e-05, "loss": 0.2484, "step": 8235 }, { "epoch": 79.85903614457831, "grad_norm": 13.366649627685547, "learning_rate": 6.001941747572816e-05, "loss": 0.3339, "step": 8236 }, { "epoch": 79.86867469879518, "grad_norm": 6.745315074920654, "learning_rate": 6.0014563106796115e-05, "loss": 0.3304, "step": 8237 }, { "epoch": 79.87831325301205, "grad_norm": 7.695306777954102, "learning_rate": 6.000970873786408e-05, "loss": 0.1574, "step": 8238 }, { "epoch": 79.88795180722892, "grad_norm": 11.754264831542969, "learning_rate": 6.000485436893204e-05, "loss": 0.2216, "step": 8239 }, { "epoch": 79.89759036144578, "grad_norm": 8.606245994567871, "learning_rate": 6e-05, "loss": 0.1918, "step": 8240 }, { "epoch": 79.90722891566266, "grad_norm": 7.056763172149658, "learning_rate": 5.999514563106796e-05, "loss": 0.3516, "step": 8241 }, { "epoch": 79.91686746987952, "grad_norm": 16.979686737060547, "learning_rate": 5.9990291262135924e-05, "loss": 0.1969, "step": 8242 }, { "epoch": 79.92650602409638, "grad_norm": 8.125672340393066, "learning_rate": 5.998543689320388e-05, "loss": 0.3434, "step": 8243 }, { "epoch": 79.93614457831325, "grad_norm": 6.676231384277344, "learning_rate": 5.9980582524271846e-05, "loss": 0.3291, "step": 8244 }, { "epoch": 79.94578313253012, "grad_norm": 9.02513313293457, "learning_rate": 5.9975728155339804e-05, "loss": 0.307, "step": 8245 }, { "epoch": 79.95542168674699, "grad_norm": 5.984504222869873, "learning_rate": 5.997087378640777e-05, "loss": 0.1891, "step": 8246 }, { "epoch": 79.96506024096385, "grad_norm": 7.847035884857178, "learning_rate": 5.9966019417475726e-05, "loss": 0.2441, "step": 8247 }, { "epoch": 79.97469879518073, "grad_norm": 16.60121726989746, "learning_rate": 5.996116504854369e-05, "loss": 0.3486, "step": 8248 }, { "epoch": 79.98433734939759, "grad_norm": 8.715173721313477, "learning_rate": 5.995631067961165e-05, "loss": 0.3088, "step": 8249 }, { "epoch": 79.99397590361446, "grad_norm": 8.728009223937988, "learning_rate": 5.995145631067961e-05, "loss": 0.1553, "step": 8250 }, { "epoch": 80.00963855421686, "grad_norm": 22.666593551635742, "learning_rate": 5.994660194174757e-05, "loss": 0.3642, "step": 8251 }, { "epoch": 80.01927710843374, "grad_norm": 8.29188060760498, "learning_rate": 5.9941747572815535e-05, "loss": 0.2131, "step": 8252 }, { "epoch": 80.0289156626506, "grad_norm": 11.066702842712402, "learning_rate": 5.993689320388349e-05, "loss": 0.351, "step": 8253 }, { "epoch": 80.03855421686747, "grad_norm": 6.500953674316406, "learning_rate": 5.993203883495145e-05, "loss": 0.2572, "step": 8254 }, { "epoch": 80.04819277108433, "grad_norm": 11.081644058227539, "learning_rate": 5.9927184466019415e-05, "loss": 0.2349, "step": 8255 }, { "epoch": 80.05783132530121, "grad_norm": 8.996909141540527, "learning_rate": 5.992233009708739e-05, "loss": 0.2335, "step": 8256 }, { "epoch": 80.06746987951807, "grad_norm": 13.461633682250977, "learning_rate": 5.991747572815535e-05, "loss": 0.3272, "step": 8257 }, { "epoch": 80.07710843373494, "grad_norm": 6.092802047729492, "learning_rate": 5.991262135922331e-05, "loss": 0.291, "step": 8258 }, { "epoch": 80.08674698795181, "grad_norm": 10.380788803100586, "learning_rate": 5.990776699029127e-05, "loss": 0.1769, "step": 8259 }, { "epoch": 80.09638554216868, "grad_norm": 12.099583625793457, "learning_rate": 5.990291262135923e-05, "loss": 0.257, "step": 8260 }, { "epoch": 80.10602409638554, "grad_norm": 10.787297248840332, "learning_rate": 5.989805825242719e-05, "loss": 0.3984, "step": 8261 }, { "epoch": 80.1156626506024, "grad_norm": 13.830042839050293, "learning_rate": 5.9893203883495154e-05, "loss": 0.5785, "step": 8262 }, { "epoch": 80.12530120481928, "grad_norm": 5.987460613250732, "learning_rate": 5.988834951456311e-05, "loss": 0.2276, "step": 8263 }, { "epoch": 80.13493975903614, "grad_norm": 7.850924491882324, "learning_rate": 5.9883495145631076e-05, "loss": 0.1637, "step": 8264 }, { "epoch": 80.144578313253, "grad_norm": 7.170277118682861, "learning_rate": 5.9878640776699034e-05, "loss": 0.1969, "step": 8265 }, { "epoch": 80.15421686746988, "grad_norm": 11.64796257019043, "learning_rate": 5.9873786407767e-05, "loss": 0.2991, "step": 8266 }, { "epoch": 80.16385542168675, "grad_norm": 13.801621437072754, "learning_rate": 5.9868932038834956e-05, "loss": 0.5096, "step": 8267 }, { "epoch": 80.17349397590361, "grad_norm": 12.716291427612305, "learning_rate": 5.986407766990292e-05, "loss": 0.415, "step": 8268 }, { "epoch": 80.18313253012049, "grad_norm": 13.061254501342773, "learning_rate": 5.985922330097088e-05, "loss": 0.2965, "step": 8269 }, { "epoch": 80.19277108433735, "grad_norm": 8.105101585388184, "learning_rate": 5.985436893203884e-05, "loss": 0.3041, "step": 8270 }, { "epoch": 80.20240963855422, "grad_norm": 12.102636337280273, "learning_rate": 5.98495145631068e-05, "loss": 0.3003, "step": 8271 }, { "epoch": 80.21204819277108, "grad_norm": 7.0590081214904785, "learning_rate": 5.9844660194174765e-05, "loss": 0.3364, "step": 8272 }, { "epoch": 80.22168674698796, "grad_norm": 12.813553810119629, "learning_rate": 5.983980582524272e-05, "loss": 0.3119, "step": 8273 }, { "epoch": 80.23132530120482, "grad_norm": 9.194072723388672, "learning_rate": 5.983495145631069e-05, "loss": 0.3406, "step": 8274 }, { "epoch": 80.24096385542168, "grad_norm": 9.056380271911621, "learning_rate": 5.9830097087378645e-05, "loss": 0.299, "step": 8275 }, { "epoch": 80.25060240963856, "grad_norm": 10.307931900024414, "learning_rate": 5.98252427184466e-05, "loss": 0.1841, "step": 8276 }, { "epoch": 80.26024096385542, "grad_norm": 11.565746307373047, "learning_rate": 5.982038834951457e-05, "loss": 0.2782, "step": 8277 }, { "epoch": 80.26987951807229, "grad_norm": 7.411890029907227, "learning_rate": 5.9815533980582525e-05, "loss": 0.2042, "step": 8278 }, { "epoch": 80.27951807228915, "grad_norm": 8.023387908935547, "learning_rate": 5.981067961165049e-05, "loss": 0.3119, "step": 8279 }, { "epoch": 80.28915662650603, "grad_norm": 7.765839576721191, "learning_rate": 5.980582524271845e-05, "loss": 0.3358, "step": 8280 }, { "epoch": 80.29879518072289, "grad_norm": 9.9652681350708, "learning_rate": 5.980097087378641e-05, "loss": 0.3284, "step": 8281 }, { "epoch": 80.30843373493975, "grad_norm": 12.113851547241211, "learning_rate": 5.979611650485437e-05, "loss": 0.2319, "step": 8282 }, { "epoch": 80.31807228915663, "grad_norm": 8.568867683410645, "learning_rate": 5.9791262135922334e-05, "loss": 0.1185, "step": 8283 }, { "epoch": 80.3277108433735, "grad_norm": 9.957000732421875, "learning_rate": 5.978640776699029e-05, "loss": 0.3004, "step": 8284 }, { "epoch": 80.33734939759036, "grad_norm": 11.841056823730469, "learning_rate": 5.9781553398058256e-05, "loss": 0.2238, "step": 8285 }, { "epoch": 80.34698795180722, "grad_norm": 14.451016426086426, "learning_rate": 5.9776699029126214e-05, "loss": 0.4208, "step": 8286 }, { "epoch": 80.3566265060241, "grad_norm": 6.328277111053467, "learning_rate": 5.977184466019418e-05, "loss": 0.2722, "step": 8287 }, { "epoch": 80.36626506024096, "grad_norm": 8.482747077941895, "learning_rate": 5.9766990291262137e-05, "loss": 0.2728, "step": 8288 }, { "epoch": 80.37590361445783, "grad_norm": 12.00107479095459, "learning_rate": 5.97621359223301e-05, "loss": 0.396, "step": 8289 }, { "epoch": 80.3855421686747, "grad_norm": 7.055679798126221, "learning_rate": 5.975728155339806e-05, "loss": 0.2859, "step": 8290 }, { "epoch": 80.39518072289157, "grad_norm": 5.593285083770752, "learning_rate": 5.9752427184466017e-05, "loss": 0.209, "step": 8291 }, { "epoch": 80.40481927710843, "grad_norm": 12.235334396362305, "learning_rate": 5.974757281553398e-05, "loss": 0.332, "step": 8292 }, { "epoch": 80.41445783132531, "grad_norm": 8.370499610900879, "learning_rate": 5.974271844660194e-05, "loss": 0.3366, "step": 8293 }, { "epoch": 80.42409638554217, "grad_norm": 11.29161262512207, "learning_rate": 5.9737864077669903e-05, "loss": 0.2229, "step": 8294 }, { "epoch": 80.43373493975903, "grad_norm": 4.633482933044434, "learning_rate": 5.973300970873786e-05, "loss": 0.318, "step": 8295 }, { "epoch": 80.4433734939759, "grad_norm": 7.09244966506958, "learning_rate": 5.9728155339805826e-05, "loss": 0.3283, "step": 8296 }, { "epoch": 80.45301204819278, "grad_norm": 10.914859771728516, "learning_rate": 5.9723300970873783e-05, "loss": 0.1642, "step": 8297 }, { "epoch": 80.46265060240964, "grad_norm": 6.277137756347656, "learning_rate": 5.971844660194175e-05, "loss": 0.2376, "step": 8298 }, { "epoch": 80.4722891566265, "grad_norm": 19.29038429260254, "learning_rate": 5.9713592233009706e-05, "loss": 0.3863, "step": 8299 }, { "epoch": 80.48192771084338, "grad_norm": 7.175030708312988, "learning_rate": 5.970873786407767e-05, "loss": 0.1665, "step": 8300 }, { "epoch": 80.49156626506024, "grad_norm": 10.282286643981934, "learning_rate": 5.970388349514563e-05, "loss": 0.4932, "step": 8301 }, { "epoch": 80.5012048192771, "grad_norm": 8.049354553222656, "learning_rate": 5.969902912621359e-05, "loss": 0.3116, "step": 8302 }, { "epoch": 80.51084337349397, "grad_norm": 15.894403457641602, "learning_rate": 5.969417475728155e-05, "loss": 0.3529, "step": 8303 }, { "epoch": 80.52048192771085, "grad_norm": 8.805912017822266, "learning_rate": 5.9689320388349515e-05, "loss": 0.1841, "step": 8304 }, { "epoch": 80.53012048192771, "grad_norm": 6.819592475891113, "learning_rate": 5.968446601941747e-05, "loss": 0.3458, "step": 8305 }, { "epoch": 80.53975903614457, "grad_norm": 15.68964672088623, "learning_rate": 5.9679611650485444e-05, "loss": 0.6036, "step": 8306 }, { "epoch": 80.54939759036145, "grad_norm": 12.471048355102539, "learning_rate": 5.967475728155341e-05, "loss": 0.2674, "step": 8307 }, { "epoch": 80.55903614457831, "grad_norm": 5.6384429931640625, "learning_rate": 5.9669902912621366e-05, "loss": 0.3317, "step": 8308 }, { "epoch": 80.56867469879518, "grad_norm": 8.964723587036133, "learning_rate": 5.966504854368933e-05, "loss": 0.2907, "step": 8309 }, { "epoch": 80.57831325301204, "grad_norm": 3.9600350856781006, "learning_rate": 5.966019417475729e-05, "loss": 0.1696, "step": 8310 }, { "epoch": 80.58795180722892, "grad_norm": 9.214179039001465, "learning_rate": 5.965533980582525e-05, "loss": 0.1593, "step": 8311 }, { "epoch": 80.59759036144578, "grad_norm": 9.833550453186035, "learning_rate": 5.965048543689321e-05, "loss": 0.4436, "step": 8312 }, { "epoch": 80.60722891566265, "grad_norm": 12.800948143005371, "learning_rate": 5.964563106796117e-05, "loss": 0.4851, "step": 8313 }, { "epoch": 80.61686746987952, "grad_norm": 6.50042200088501, "learning_rate": 5.964077669902913e-05, "loss": 0.229, "step": 8314 }, { "epoch": 80.62650602409639, "grad_norm": 8.36296272277832, "learning_rate": 5.963592233009709e-05, "loss": 0.2138, "step": 8315 }, { "epoch": 80.63614457831325, "grad_norm": 5.04952335357666, "learning_rate": 5.9631067961165055e-05, "loss": 0.1495, "step": 8316 }, { "epoch": 80.64578313253013, "grad_norm": 3.871647834777832, "learning_rate": 5.962621359223301e-05, "loss": 0.2166, "step": 8317 }, { "epoch": 80.65542168674699, "grad_norm": 12.8651123046875, "learning_rate": 5.962135922330098e-05, "loss": 0.494, "step": 8318 }, { "epoch": 80.66506024096385, "grad_norm": 7.614495754241943, "learning_rate": 5.9616504854368935e-05, "loss": 0.1786, "step": 8319 }, { "epoch": 80.67469879518072, "grad_norm": 15.648347854614258, "learning_rate": 5.96116504854369e-05, "loss": 0.1665, "step": 8320 }, { "epoch": 80.6843373493976, "grad_norm": 11.36241340637207, "learning_rate": 5.960679611650486e-05, "loss": 0.2338, "step": 8321 }, { "epoch": 80.69397590361446, "grad_norm": 5.760176658630371, "learning_rate": 5.960194174757282e-05, "loss": 0.252, "step": 8322 }, { "epoch": 80.70361445783132, "grad_norm": 12.305843353271484, "learning_rate": 5.959708737864078e-05, "loss": 0.3398, "step": 8323 }, { "epoch": 80.7132530120482, "grad_norm": 10.532488822937012, "learning_rate": 5.9592233009708745e-05, "loss": 0.2616, "step": 8324 }, { "epoch": 80.72289156626506, "grad_norm": 11.076416015625, "learning_rate": 5.95873786407767e-05, "loss": 0.2427, "step": 8325 }, { "epoch": 80.73253012048193, "grad_norm": 5.302202224731445, "learning_rate": 5.958252427184467e-05, "loss": 0.1871, "step": 8326 }, { "epoch": 80.74216867469879, "grad_norm": 6.991861343383789, "learning_rate": 5.9577669902912625e-05, "loss": 0.267, "step": 8327 }, { "epoch": 80.75180722891567, "grad_norm": 8.757392883300781, "learning_rate": 5.957281553398059e-05, "loss": 0.1919, "step": 8328 }, { "epoch": 80.76144578313253, "grad_norm": 12.028814315795898, "learning_rate": 5.956796116504855e-05, "loss": 0.2808, "step": 8329 }, { "epoch": 80.7710843373494, "grad_norm": 7.641663551330566, "learning_rate": 5.9563106796116505e-05, "loss": 0.4152, "step": 8330 }, { "epoch": 80.78072289156627, "grad_norm": 18.670501708984375, "learning_rate": 5.955825242718447e-05, "loss": 0.3501, "step": 8331 }, { "epoch": 80.79036144578313, "grad_norm": 8.45357608795166, "learning_rate": 5.955339805825243e-05, "loss": 0.1794, "step": 8332 }, { "epoch": 80.8, "grad_norm": 6.495135307312012, "learning_rate": 5.954854368932039e-05, "loss": 0.1675, "step": 8333 }, { "epoch": 80.80963855421686, "grad_norm": 13.185992240905762, "learning_rate": 5.954368932038835e-05, "loss": 0.31, "step": 8334 }, { "epoch": 80.81927710843374, "grad_norm": 15.206729888916016, "learning_rate": 5.9538834951456314e-05, "loss": 0.2258, "step": 8335 }, { "epoch": 80.8289156626506, "grad_norm": 5.060444355010986, "learning_rate": 5.953398058252427e-05, "loss": 0.2539, "step": 8336 }, { "epoch": 80.83855421686746, "grad_norm": 7.860358238220215, "learning_rate": 5.9529126213592236e-05, "loss": 0.2134, "step": 8337 }, { "epoch": 80.84819277108434, "grad_norm": 9.38110637664795, "learning_rate": 5.9524271844660194e-05, "loss": 0.4254, "step": 8338 }, { "epoch": 80.8578313253012, "grad_norm": 7.897110939025879, "learning_rate": 5.951941747572816e-05, "loss": 0.2623, "step": 8339 }, { "epoch": 80.86746987951807, "grad_norm": 4.632669448852539, "learning_rate": 5.9514563106796116e-05, "loss": 0.177, "step": 8340 }, { "epoch": 80.87710843373495, "grad_norm": 10.218603134155273, "learning_rate": 5.950970873786408e-05, "loss": 0.6285, "step": 8341 }, { "epoch": 80.88674698795181, "grad_norm": 16.654573440551758, "learning_rate": 5.950485436893204e-05, "loss": 0.4009, "step": 8342 }, { "epoch": 80.89638554216867, "grad_norm": 9.689457893371582, "learning_rate": 5.95e-05, "loss": 0.1232, "step": 8343 }, { "epoch": 80.90602409638554, "grad_norm": 7.561078071594238, "learning_rate": 5.949514563106796e-05, "loss": 0.1954, "step": 8344 }, { "epoch": 80.91566265060241, "grad_norm": 6.200259208679199, "learning_rate": 5.9490291262135925e-05, "loss": 0.2347, "step": 8345 }, { "epoch": 80.92530120481928, "grad_norm": 6.2299418449401855, "learning_rate": 5.948543689320388e-05, "loss": 0.1763, "step": 8346 }, { "epoch": 80.93493975903614, "grad_norm": 7.670006275177002, "learning_rate": 5.948058252427184e-05, "loss": 0.2644, "step": 8347 }, { "epoch": 80.94457831325302, "grad_norm": 5.924516677856445, "learning_rate": 5.9475728155339805e-05, "loss": 0.1817, "step": 8348 }, { "epoch": 80.95421686746988, "grad_norm": 10.99212646484375, "learning_rate": 5.947087378640776e-05, "loss": 0.244, "step": 8349 }, { "epoch": 80.96385542168674, "grad_norm": 7.707190036773682, "learning_rate": 5.946601941747573e-05, "loss": 0.2086, "step": 8350 }, { "epoch": 80.97349397590361, "grad_norm": 4.206699848175049, "learning_rate": 5.9461165048543685e-05, "loss": 0.0751, "step": 8351 }, { "epoch": 80.98313253012049, "grad_norm": 5.844779968261719, "learning_rate": 5.945631067961165e-05, "loss": 0.2338, "step": 8352 }, { "epoch": 80.99277108433735, "grad_norm": 11.945700645446777, "learning_rate": 5.945145631067961e-05, "loss": 0.5031, "step": 8353 }, { "epoch": 81.00843373493976, "grad_norm": 5.997403144836426, "learning_rate": 5.944660194174757e-05, "loss": 0.1585, "step": 8354 }, { "epoch": 81.01807228915662, "grad_norm": 14.017000198364258, "learning_rate": 5.944174757281553e-05, "loss": 0.2127, "step": 8355 }, { "epoch": 81.0277108433735, "grad_norm": 11.570769309997559, "learning_rate": 5.9436893203883494e-05, "loss": 0.2662, "step": 8356 }, { "epoch": 81.03734939759036, "grad_norm": 3.571803092956543, "learning_rate": 5.9432038834951466e-05, "loss": 0.3045, "step": 8357 }, { "epoch": 81.04698795180722, "grad_norm": 14.896988868713379, "learning_rate": 5.9427184466019423e-05, "loss": 0.2192, "step": 8358 }, { "epoch": 81.0566265060241, "grad_norm": 13.831665992736816, "learning_rate": 5.942233009708739e-05, "loss": 0.4749, "step": 8359 }, { "epoch": 81.06626506024097, "grad_norm": 5.9649763107299805, "learning_rate": 5.9417475728155346e-05, "loss": 0.3967, "step": 8360 }, { "epoch": 81.07590361445783, "grad_norm": 17.37518310546875, "learning_rate": 5.941262135922331e-05, "loss": 0.1704, "step": 8361 }, { "epoch": 81.08554216867469, "grad_norm": 5.755687713623047, "learning_rate": 5.940776699029127e-05, "loss": 0.2166, "step": 8362 }, { "epoch": 81.09518072289157, "grad_norm": 177.57064819335938, "learning_rate": 5.940291262135923e-05, "loss": 0.4742, "step": 8363 }, { "epoch": 81.10481927710843, "grad_norm": 13.59211540222168, "learning_rate": 5.939805825242719e-05, "loss": 0.54, "step": 8364 }, { "epoch": 81.1144578313253, "grad_norm": 42.45346450805664, "learning_rate": 5.9393203883495155e-05, "loss": 0.3924, "step": 8365 }, { "epoch": 81.12409638554217, "grad_norm": 15.511122703552246, "learning_rate": 5.938834951456311e-05, "loss": 0.3523, "step": 8366 }, { "epoch": 81.13373493975904, "grad_norm": 7.059877395629883, "learning_rate": 5.938349514563108e-05, "loss": 0.3044, "step": 8367 }, { "epoch": 81.1433734939759, "grad_norm": 2.5857183933258057, "learning_rate": 5.9378640776699035e-05, "loss": 0.1389, "step": 8368 }, { "epoch": 81.15301204819278, "grad_norm": 10.94770622253418, "learning_rate": 5.937378640776699e-05, "loss": 0.2267, "step": 8369 }, { "epoch": 81.16265060240964, "grad_norm": 5.122956275939941, "learning_rate": 5.936893203883496e-05, "loss": 0.2845, "step": 8370 }, { "epoch": 81.1722891566265, "grad_norm": 8.800579071044922, "learning_rate": 5.9364077669902915e-05, "loss": 0.1076, "step": 8371 }, { "epoch": 81.18192771084337, "grad_norm": 17.23679542541504, "learning_rate": 5.935922330097088e-05, "loss": 0.4403, "step": 8372 }, { "epoch": 81.19156626506025, "grad_norm": 32.54694366455078, "learning_rate": 5.935436893203884e-05, "loss": 0.437, "step": 8373 }, { "epoch": 81.20120481927711, "grad_norm": 4.642674446105957, "learning_rate": 5.93495145631068e-05, "loss": 0.3711, "step": 8374 }, { "epoch": 81.21084337349397, "grad_norm": 37.307777404785156, "learning_rate": 5.934466019417476e-05, "loss": 0.344, "step": 8375 }, { "epoch": 81.22048192771085, "grad_norm": 14.892058372497559, "learning_rate": 5.9339805825242724e-05, "loss": 0.4803, "step": 8376 }, { "epoch": 81.23012048192771, "grad_norm": 5.957064628601074, "learning_rate": 5.933495145631068e-05, "loss": 0.2386, "step": 8377 }, { "epoch": 81.23975903614458, "grad_norm": 13.365514755249023, "learning_rate": 5.9330097087378646e-05, "loss": 0.2176, "step": 8378 }, { "epoch": 81.24939759036144, "grad_norm": 15.467290878295898, "learning_rate": 5.9325242718446604e-05, "loss": 0.2973, "step": 8379 }, { "epoch": 81.25903614457832, "grad_norm": 13.877628326416016, "learning_rate": 5.932038834951457e-05, "loss": 0.1798, "step": 8380 }, { "epoch": 81.26867469879518, "grad_norm": 13.020598411560059, "learning_rate": 5.9315533980582526e-05, "loss": 0.1915, "step": 8381 }, { "epoch": 81.27831325301204, "grad_norm": 61.82933807373047, "learning_rate": 5.931067961165049e-05, "loss": 0.1943, "step": 8382 }, { "epoch": 81.28795180722892, "grad_norm": 16.4002742767334, "learning_rate": 5.930582524271845e-05, "loss": 0.3996, "step": 8383 }, { "epoch": 81.29759036144578, "grad_norm": 38.689937591552734, "learning_rate": 5.930097087378641e-05, "loss": 0.1587, "step": 8384 }, { "epoch": 81.30722891566265, "grad_norm": 23.3496036529541, "learning_rate": 5.929611650485437e-05, "loss": 0.3652, "step": 8385 }, { "epoch": 81.31686746987951, "grad_norm": 2.6302456855773926, "learning_rate": 5.929126213592233e-05, "loss": 0.3201, "step": 8386 }, { "epoch": 81.32650602409639, "grad_norm": 80.5664291381836, "learning_rate": 5.928640776699029e-05, "loss": 0.2888, "step": 8387 }, { "epoch": 81.33614457831325, "grad_norm": 40.02912521362305, "learning_rate": 5.928155339805825e-05, "loss": 0.5088, "step": 8388 }, { "epoch": 81.34578313253012, "grad_norm": 93.311279296875, "learning_rate": 5.9276699029126216e-05, "loss": 0.5492, "step": 8389 }, { "epoch": 81.355421686747, "grad_norm": 7.952503681182861, "learning_rate": 5.927184466019417e-05, "loss": 0.1833, "step": 8390 }, { "epoch": 81.36506024096386, "grad_norm": 3.998264789581299, "learning_rate": 5.926699029126214e-05, "loss": 0.1103, "step": 8391 }, { "epoch": 81.37469879518072, "grad_norm": 9.510334968566895, "learning_rate": 5.9262135922330096e-05, "loss": 0.3261, "step": 8392 }, { "epoch": 81.38433734939758, "grad_norm": 6.8514227867126465, "learning_rate": 5.925728155339806e-05, "loss": 0.2875, "step": 8393 }, { "epoch": 81.39397590361446, "grad_norm": 2.198101043701172, "learning_rate": 5.925242718446602e-05, "loss": 0.1946, "step": 8394 }, { "epoch": 81.40361445783132, "grad_norm": 19.190170288085938, "learning_rate": 5.924757281553398e-05, "loss": 0.1478, "step": 8395 }, { "epoch": 81.41325301204819, "grad_norm": 25.005067825317383, "learning_rate": 5.924271844660194e-05, "loss": 0.2771, "step": 8396 }, { "epoch": 81.42289156626506, "grad_norm": 9.456487655639648, "learning_rate": 5.9237864077669905e-05, "loss": 0.4811, "step": 8397 }, { "epoch": 81.43253012048193, "grad_norm": 31.827308654785156, "learning_rate": 5.923300970873786e-05, "loss": 0.3576, "step": 8398 }, { "epoch": 81.44216867469879, "grad_norm": 3.9458587169647217, "learning_rate": 5.922815533980583e-05, "loss": 0.2649, "step": 8399 }, { "epoch": 81.45180722891567, "grad_norm": 8.663731575012207, "learning_rate": 5.9223300970873785e-05, "loss": 0.4107, "step": 8400 }, { "epoch": 81.46144578313253, "grad_norm": 16.483726501464844, "learning_rate": 5.921844660194175e-05, "loss": 0.1874, "step": 8401 }, { "epoch": 81.4710843373494, "grad_norm": 5.148740768432617, "learning_rate": 5.921359223300971e-05, "loss": 0.2075, "step": 8402 }, { "epoch": 81.48072289156626, "grad_norm": 22.38153076171875, "learning_rate": 5.9208737864077665e-05, "loss": 0.4214, "step": 8403 }, { "epoch": 81.49036144578314, "grad_norm": 4.767266273498535, "learning_rate": 5.920388349514563e-05, "loss": 0.3815, "step": 8404 }, { "epoch": 81.5, "grad_norm": 6.617228031158447, "learning_rate": 5.919902912621359e-05, "loss": 0.1019, "step": 8405 }, { "epoch": 81.50963855421686, "grad_norm": 8.613985061645508, "learning_rate": 5.919417475728155e-05, "loss": 0.3062, "step": 8406 }, { "epoch": 81.51927710843374, "grad_norm": 7.655037879943848, "learning_rate": 5.918932038834952e-05, "loss": 0.228, "step": 8407 }, { "epoch": 81.5289156626506, "grad_norm": 26.995403289794922, "learning_rate": 5.918446601941748e-05, "loss": 0.2419, "step": 8408 }, { "epoch": 81.53855421686747, "grad_norm": 7.098845958709717, "learning_rate": 5.9179611650485445e-05, "loss": 0.2125, "step": 8409 }, { "epoch": 81.54819277108433, "grad_norm": 3.618889808654785, "learning_rate": 5.91747572815534e-05, "loss": 0.1204, "step": 8410 }, { "epoch": 81.55783132530121, "grad_norm": 13.222487449645996, "learning_rate": 5.916990291262137e-05, "loss": 0.4806, "step": 8411 }, { "epoch": 81.56746987951807, "grad_norm": 11.581809043884277, "learning_rate": 5.9165048543689325e-05, "loss": 0.285, "step": 8412 }, { "epoch": 81.57710843373494, "grad_norm": 24.283220291137695, "learning_rate": 5.916019417475729e-05, "loss": 0.374, "step": 8413 }, { "epoch": 81.58674698795181, "grad_norm": 12.890034675598145, "learning_rate": 5.915533980582525e-05, "loss": 0.2991, "step": 8414 }, { "epoch": 81.59638554216868, "grad_norm": 31.019187927246094, "learning_rate": 5.915048543689321e-05, "loss": 0.3413, "step": 8415 }, { "epoch": 81.60602409638554, "grad_norm": 10.127633094787598, "learning_rate": 5.914563106796117e-05, "loss": 0.2622, "step": 8416 }, { "epoch": 81.61566265060242, "grad_norm": 26.043407440185547, "learning_rate": 5.9140776699029134e-05, "loss": 0.5115, "step": 8417 }, { "epoch": 81.62530120481928, "grad_norm": 25.812803268432617, "learning_rate": 5.913592233009709e-05, "loss": 0.487, "step": 8418 }, { "epoch": 81.63493975903614, "grad_norm": 7.615488052368164, "learning_rate": 5.913106796116506e-05, "loss": 0.1822, "step": 8419 }, { "epoch": 81.644578313253, "grad_norm": 14.17885684967041, "learning_rate": 5.9126213592233015e-05, "loss": 0.2119, "step": 8420 }, { "epoch": 81.65421686746988, "grad_norm": 10.53975772857666, "learning_rate": 5.912135922330098e-05, "loss": 0.148, "step": 8421 }, { "epoch": 81.66385542168675, "grad_norm": 8.564537048339844, "learning_rate": 5.911650485436894e-05, "loss": 0.2514, "step": 8422 }, { "epoch": 81.67349397590361, "grad_norm": 4.266730308532715, "learning_rate": 5.91116504854369e-05, "loss": 0.2507, "step": 8423 }, { "epoch": 81.68313253012049, "grad_norm": 4.587274074554443, "learning_rate": 5.910679611650486e-05, "loss": 0.3372, "step": 8424 }, { "epoch": 81.69277108433735, "grad_norm": 20.905467987060547, "learning_rate": 5.910194174757282e-05, "loss": 0.1561, "step": 8425 }, { "epoch": 81.70240963855422, "grad_norm": 22.79741668701172, "learning_rate": 5.909708737864078e-05, "loss": 0.3517, "step": 8426 }, { "epoch": 81.71204819277108, "grad_norm": 7.617430210113525, "learning_rate": 5.909223300970874e-05, "loss": 0.1239, "step": 8427 }, { "epoch": 81.72168674698796, "grad_norm": 9.106727600097656, "learning_rate": 5.9087378640776704e-05, "loss": 0.5207, "step": 8428 }, { "epoch": 81.73132530120482, "grad_norm": 15.66258430480957, "learning_rate": 5.908252427184466e-05, "loss": 0.4386, "step": 8429 }, { "epoch": 81.74096385542168, "grad_norm": 11.264021873474121, "learning_rate": 5.9077669902912626e-05, "loss": 0.1725, "step": 8430 }, { "epoch": 81.75060240963856, "grad_norm": 4.0474934577941895, "learning_rate": 5.9072815533980584e-05, "loss": 0.1995, "step": 8431 }, { "epoch": 81.76024096385542, "grad_norm": 8.609657287597656, "learning_rate": 5.906796116504855e-05, "loss": 0.1992, "step": 8432 }, { "epoch": 81.76987951807229, "grad_norm": 1.1789257526397705, "learning_rate": 5.9063106796116506e-05, "loss": 0.4047, "step": 8433 }, { "epoch": 81.77951807228915, "grad_norm": 19.822242736816406, "learning_rate": 5.905825242718447e-05, "loss": 0.3316, "step": 8434 }, { "epoch": 81.78915662650603, "grad_norm": 52.234989166259766, "learning_rate": 5.905339805825243e-05, "loss": 0.1424, "step": 8435 }, { "epoch": 81.79879518072289, "grad_norm": 20.684364318847656, "learning_rate": 5.904854368932039e-05, "loss": 0.2524, "step": 8436 }, { "epoch": 81.80843373493975, "grad_norm": 1.6225894689559937, "learning_rate": 5.904368932038835e-05, "loss": 0.0984, "step": 8437 }, { "epoch": 81.81807228915663, "grad_norm": 33.93935775756836, "learning_rate": 5.9038834951456315e-05, "loss": 0.3685, "step": 8438 }, { "epoch": 81.8277108433735, "grad_norm": 4.39932107925415, "learning_rate": 5.903398058252427e-05, "loss": 0.238, "step": 8439 }, { "epoch": 81.83734939759036, "grad_norm": 12.191788673400879, "learning_rate": 5.902912621359223e-05, "loss": 0.3162, "step": 8440 }, { "epoch": 81.84698795180722, "grad_norm": 4.375613212585449, "learning_rate": 5.9024271844660195e-05, "loss": 0.3297, "step": 8441 }, { "epoch": 81.8566265060241, "grad_norm": 4.900963306427002, "learning_rate": 5.901941747572815e-05, "loss": 0.2848, "step": 8442 }, { "epoch": 81.86626506024096, "grad_norm": 25.88529396057129, "learning_rate": 5.901456310679612e-05, "loss": 0.3286, "step": 8443 }, { "epoch": 81.87590361445783, "grad_norm": 13.665834426879883, "learning_rate": 5.9009708737864075e-05, "loss": 0.2912, "step": 8444 }, { "epoch": 81.8855421686747, "grad_norm": 25.326007843017578, "learning_rate": 5.900485436893204e-05, "loss": 0.2652, "step": 8445 }, { "epoch": 81.89518072289157, "grad_norm": 7.325789928436279, "learning_rate": 5.9e-05, "loss": 0.4518, "step": 8446 }, { "epoch": 81.90481927710843, "grad_norm": 10.08324909210205, "learning_rate": 5.899514563106796e-05, "loss": 0.282, "step": 8447 }, { "epoch": 81.91445783132531, "grad_norm": 7.778529167175293, "learning_rate": 5.899029126213592e-05, "loss": 0.325, "step": 8448 }, { "epoch": 81.92409638554217, "grad_norm": 10.805318832397461, "learning_rate": 5.8985436893203884e-05, "loss": 0.2892, "step": 8449 }, { "epoch": 81.93373493975903, "grad_norm": 11.40517807006836, "learning_rate": 5.898058252427184e-05, "loss": 0.4432, "step": 8450 }, { "epoch": 81.9433734939759, "grad_norm": 18.25066375732422, "learning_rate": 5.897572815533981e-05, "loss": 0.4244, "step": 8451 }, { "epoch": 81.95301204819278, "grad_norm": 29.96432113647461, "learning_rate": 5.8970873786407764e-05, "loss": 0.2701, "step": 8452 }, { "epoch": 81.96265060240964, "grad_norm": 21.149219512939453, "learning_rate": 5.896601941747573e-05, "loss": 0.162, "step": 8453 }, { "epoch": 81.9722891566265, "grad_norm": 10.814780235290527, "learning_rate": 5.896116504854369e-05, "loss": 0.3439, "step": 8454 }, { "epoch": 81.98192771084338, "grad_norm": 12.947131156921387, "learning_rate": 5.895631067961165e-05, "loss": 0.3311, "step": 8455 }, { "epoch": 81.99156626506024, "grad_norm": 8.602914810180664, "learning_rate": 5.895145631067961e-05, "loss": 0.1633, "step": 8456 }, { "epoch": 82.00722891566265, "grad_norm": 31.642484664916992, "learning_rate": 5.894660194174757e-05, "loss": 0.2296, "step": 8457 }, { "epoch": 82.01686746987951, "grad_norm": 7.913972854614258, "learning_rate": 5.8941747572815545e-05, "loss": 0.2455, "step": 8458 }, { "epoch": 82.02650602409639, "grad_norm": 8.698951721191406, "learning_rate": 5.89368932038835e-05, "loss": 0.199, "step": 8459 }, { "epoch": 82.03614457831326, "grad_norm": 9.256752014160156, "learning_rate": 5.893203883495147e-05, "loss": 0.4158, "step": 8460 }, { "epoch": 82.04578313253012, "grad_norm": 27.717016220092773, "learning_rate": 5.8927184466019425e-05, "loss": 0.4553, "step": 8461 }, { "epoch": 82.05542168674698, "grad_norm": 10.095070838928223, "learning_rate": 5.892233009708738e-05, "loss": 0.3691, "step": 8462 }, { "epoch": 82.06506024096386, "grad_norm": 8.05465030670166, "learning_rate": 5.891747572815535e-05, "loss": 0.2964, "step": 8463 }, { "epoch": 82.07469879518072, "grad_norm": 12.60605239868164, "learning_rate": 5.8912621359223305e-05, "loss": 0.2548, "step": 8464 }, { "epoch": 82.08433734939759, "grad_norm": 3.265469551086426, "learning_rate": 5.890776699029127e-05, "loss": 0.1659, "step": 8465 }, { "epoch": 82.09397590361446, "grad_norm": 1.430861234664917, "learning_rate": 5.890291262135923e-05, "loss": 0.1885, "step": 8466 }, { "epoch": 82.10361445783133, "grad_norm": 6.055777549743652, "learning_rate": 5.889805825242719e-05, "loss": 0.2202, "step": 8467 }, { "epoch": 82.11325301204819, "grad_norm": 4.512850761413574, "learning_rate": 5.889320388349515e-05, "loss": 0.148, "step": 8468 }, { "epoch": 82.12289156626505, "grad_norm": 32.543399810791016, "learning_rate": 5.8888349514563114e-05, "loss": 0.2006, "step": 8469 }, { "epoch": 82.13253012048193, "grad_norm": 9.878402709960938, "learning_rate": 5.888349514563107e-05, "loss": 0.25, "step": 8470 }, { "epoch": 82.1421686746988, "grad_norm": 17.143068313598633, "learning_rate": 5.8878640776699036e-05, "loss": 0.2022, "step": 8471 }, { "epoch": 82.15180722891566, "grad_norm": 22.44188117980957, "learning_rate": 5.8873786407766994e-05, "loss": 0.394, "step": 8472 }, { "epoch": 82.16144578313254, "grad_norm": 13.918437004089355, "learning_rate": 5.886893203883496e-05, "loss": 0.4349, "step": 8473 }, { "epoch": 82.1710843373494, "grad_norm": 11.175570487976074, "learning_rate": 5.8864077669902916e-05, "loss": 0.2686, "step": 8474 }, { "epoch": 82.18072289156626, "grad_norm": 7.951381683349609, "learning_rate": 5.885922330097088e-05, "loss": 0.1831, "step": 8475 }, { "epoch": 82.19036144578314, "grad_norm": 7.005080223083496, "learning_rate": 5.885436893203884e-05, "loss": 0.4263, "step": 8476 }, { "epoch": 82.2, "grad_norm": 17.399206161499023, "learning_rate": 5.88495145631068e-05, "loss": 0.308, "step": 8477 }, { "epoch": 82.20963855421687, "grad_norm": 45.074466705322266, "learning_rate": 5.884466019417476e-05, "loss": 0.2306, "step": 8478 }, { "epoch": 82.21927710843373, "grad_norm": 10.584307670593262, "learning_rate": 5.883980582524272e-05, "loss": 0.3112, "step": 8479 }, { "epoch": 82.2289156626506, "grad_norm": 4.252876281738281, "learning_rate": 5.883495145631068e-05, "loss": 0.2555, "step": 8480 }, { "epoch": 82.23855421686747, "grad_norm": 3.6533308029174805, "learning_rate": 5.883009708737864e-05, "loss": 0.1901, "step": 8481 }, { "epoch": 82.24819277108433, "grad_norm": 19.744029998779297, "learning_rate": 5.8825242718446606e-05, "loss": 0.2655, "step": 8482 }, { "epoch": 82.25783132530121, "grad_norm": 1.5197761058807373, "learning_rate": 5.882038834951456e-05, "loss": 0.1454, "step": 8483 }, { "epoch": 82.26746987951807, "grad_norm": 8.168111801147461, "learning_rate": 5.881553398058253e-05, "loss": 0.1469, "step": 8484 }, { "epoch": 82.27710843373494, "grad_norm": 40.158443450927734, "learning_rate": 5.8810679611650486e-05, "loss": 0.2904, "step": 8485 }, { "epoch": 82.2867469879518, "grad_norm": 2.2813384532928467, "learning_rate": 5.880582524271845e-05, "loss": 0.0484, "step": 8486 }, { "epoch": 82.29638554216868, "grad_norm": 23.389598846435547, "learning_rate": 5.880097087378641e-05, "loss": 0.3752, "step": 8487 }, { "epoch": 82.30602409638554, "grad_norm": 16.9368896484375, "learning_rate": 5.879611650485437e-05, "loss": 0.3026, "step": 8488 }, { "epoch": 82.3156626506024, "grad_norm": 24.370630264282227, "learning_rate": 5.879126213592233e-05, "loss": 0.2382, "step": 8489 }, { "epoch": 82.32530120481928, "grad_norm": 7.0938029289245605, "learning_rate": 5.8786407766990295e-05, "loss": 0.4333, "step": 8490 }, { "epoch": 82.33493975903615, "grad_norm": 22.976240158081055, "learning_rate": 5.878155339805825e-05, "loss": 0.3596, "step": 8491 }, { "epoch": 82.34457831325301, "grad_norm": 14.867966651916504, "learning_rate": 5.877669902912622e-05, "loss": 0.3113, "step": 8492 }, { "epoch": 82.35421686746987, "grad_norm": 12.792379379272461, "learning_rate": 5.8771844660194175e-05, "loss": 0.2013, "step": 8493 }, { "epoch": 82.36385542168675, "grad_norm": 2.941251039505005, "learning_rate": 5.876699029126214e-05, "loss": 0.1779, "step": 8494 }, { "epoch": 82.37349397590361, "grad_norm": 10.551197052001953, "learning_rate": 5.87621359223301e-05, "loss": 0.3721, "step": 8495 }, { "epoch": 82.38313253012048, "grad_norm": 2.28231143951416, "learning_rate": 5.8757281553398055e-05, "loss": 0.233, "step": 8496 }, { "epoch": 82.39277108433735, "grad_norm": 18.117490768432617, "learning_rate": 5.875242718446602e-05, "loss": 0.3778, "step": 8497 }, { "epoch": 82.40240963855422, "grad_norm": 9.9179048538208, "learning_rate": 5.874757281553398e-05, "loss": 0.1372, "step": 8498 }, { "epoch": 82.41204819277108, "grad_norm": 4.931938648223877, "learning_rate": 5.874271844660194e-05, "loss": 0.1981, "step": 8499 }, { "epoch": 82.42168674698796, "grad_norm": 8.699109077453613, "learning_rate": 5.87378640776699e-05, "loss": 0.218, "step": 8500 }, { "epoch": 82.43132530120482, "grad_norm": 17.413619995117188, "learning_rate": 5.8733009708737864e-05, "loss": 0.3241, "step": 8501 }, { "epoch": 82.44096385542169, "grad_norm": 5.358212947845459, "learning_rate": 5.872815533980582e-05, "loss": 0.213, "step": 8502 }, { "epoch": 82.45060240963855, "grad_norm": 29.5333309173584, "learning_rate": 5.8723300970873786e-05, "loss": 0.1308, "step": 8503 }, { "epoch": 82.46024096385543, "grad_norm": 3.123774528503418, "learning_rate": 5.8718446601941744e-05, "loss": 0.1607, "step": 8504 }, { "epoch": 82.46987951807229, "grad_norm": 6.832690238952637, "learning_rate": 5.871359223300971e-05, "loss": 0.3566, "step": 8505 }, { "epoch": 82.47951807228915, "grad_norm": 7.215016841888428, "learning_rate": 5.8708737864077666e-05, "loss": 0.1983, "step": 8506 }, { "epoch": 82.48915662650603, "grad_norm": 10.430563926696777, "learning_rate": 5.870388349514563e-05, "loss": 0.229, "step": 8507 }, { "epoch": 82.4987951807229, "grad_norm": 9.788549423217773, "learning_rate": 5.86990291262136e-05, "loss": 0.2939, "step": 8508 }, { "epoch": 82.50843373493976, "grad_norm": 24.0633487701416, "learning_rate": 5.869417475728156e-05, "loss": 0.4481, "step": 8509 }, { "epoch": 82.51807228915662, "grad_norm": 20.819332122802734, "learning_rate": 5.8689320388349524e-05, "loss": 0.2196, "step": 8510 }, { "epoch": 82.5277108433735, "grad_norm": 20.503908157348633, "learning_rate": 5.868446601941748e-05, "loss": 0.233, "step": 8511 }, { "epoch": 82.53734939759036, "grad_norm": 13.264778137207031, "learning_rate": 5.867961165048545e-05, "loss": 0.1825, "step": 8512 }, { "epoch": 82.54698795180722, "grad_norm": 20.991044998168945, "learning_rate": 5.8674757281553404e-05, "loss": 0.3586, "step": 8513 }, { "epoch": 82.5566265060241, "grad_norm": 3.7422451972961426, "learning_rate": 5.866990291262137e-05, "loss": 0.2729, "step": 8514 }, { "epoch": 82.56626506024097, "grad_norm": 24.972576141357422, "learning_rate": 5.866504854368933e-05, "loss": 0.3248, "step": 8515 }, { "epoch": 82.57590361445783, "grad_norm": 9.422553062438965, "learning_rate": 5.866019417475729e-05, "loss": 0.3333, "step": 8516 }, { "epoch": 82.58554216867469, "grad_norm": 8.215447425842285, "learning_rate": 5.865533980582525e-05, "loss": 0.4823, "step": 8517 }, { "epoch": 82.59518072289157, "grad_norm": 16.342512130737305, "learning_rate": 5.865048543689321e-05, "loss": 0.2233, "step": 8518 }, { "epoch": 82.60481927710843, "grad_norm": 11.128213882446289, "learning_rate": 5.864563106796117e-05, "loss": 0.1608, "step": 8519 }, { "epoch": 82.6144578313253, "grad_norm": 15.33233642578125, "learning_rate": 5.864077669902913e-05, "loss": 0.1507, "step": 8520 }, { "epoch": 82.62409638554217, "grad_norm": 13.181142807006836, "learning_rate": 5.8635922330097094e-05, "loss": 0.3459, "step": 8521 }, { "epoch": 82.63373493975904, "grad_norm": 21.377607345581055, "learning_rate": 5.863106796116505e-05, "loss": 0.4595, "step": 8522 }, { "epoch": 82.6433734939759, "grad_norm": 33.231807708740234, "learning_rate": 5.8626213592233016e-05, "loss": 0.4032, "step": 8523 }, { "epoch": 82.65301204819278, "grad_norm": 12.038553237915039, "learning_rate": 5.8621359223300974e-05, "loss": 0.2747, "step": 8524 }, { "epoch": 82.66265060240964, "grad_norm": 21.171459197998047, "learning_rate": 5.861650485436894e-05, "loss": 0.2937, "step": 8525 }, { "epoch": 82.6722891566265, "grad_norm": 2.155470132827759, "learning_rate": 5.8611650485436896e-05, "loss": 0.2528, "step": 8526 }, { "epoch": 82.68192771084337, "grad_norm": 6.6619977951049805, "learning_rate": 5.860679611650486e-05, "loss": 0.1646, "step": 8527 }, { "epoch": 82.69156626506025, "grad_norm": 7.721588611602783, "learning_rate": 5.860194174757282e-05, "loss": 0.2937, "step": 8528 }, { "epoch": 82.70120481927711, "grad_norm": 3.4257872104644775, "learning_rate": 5.859708737864078e-05, "loss": 0.152, "step": 8529 }, { "epoch": 82.71084337349397, "grad_norm": 19.851688385009766, "learning_rate": 5.859223300970874e-05, "loss": 0.3241, "step": 8530 }, { "epoch": 82.72048192771085, "grad_norm": 14.76008415222168, "learning_rate": 5.8587378640776705e-05, "loss": 0.1849, "step": 8531 }, { "epoch": 82.73012048192771, "grad_norm": 9.040544509887695, "learning_rate": 5.858252427184466e-05, "loss": 0.173, "step": 8532 }, { "epoch": 82.73975903614458, "grad_norm": 30.60487174987793, "learning_rate": 5.857766990291263e-05, "loss": 0.3058, "step": 8533 }, { "epoch": 82.74939759036144, "grad_norm": 29.590957641601562, "learning_rate": 5.8572815533980585e-05, "loss": 0.3069, "step": 8534 }, { "epoch": 82.75903614457832, "grad_norm": 6.6331892013549805, "learning_rate": 5.856796116504854e-05, "loss": 0.4419, "step": 8535 }, { "epoch": 82.76867469879518, "grad_norm": 13.628698348999023, "learning_rate": 5.856310679611651e-05, "loss": 0.395, "step": 8536 }, { "epoch": 82.77831325301204, "grad_norm": 12.064125061035156, "learning_rate": 5.8558252427184465e-05, "loss": 0.2571, "step": 8537 }, { "epoch": 82.78795180722892, "grad_norm": 11.279072761535645, "learning_rate": 5.855339805825243e-05, "loss": 0.5112, "step": 8538 }, { "epoch": 82.79759036144578, "grad_norm": 34.65277862548828, "learning_rate": 5.854854368932039e-05, "loss": 0.4201, "step": 8539 }, { "epoch": 82.80722891566265, "grad_norm": 19.452863693237305, "learning_rate": 5.854368932038835e-05, "loss": 0.4374, "step": 8540 }, { "epoch": 82.81686746987951, "grad_norm": 6.03862190246582, "learning_rate": 5.853883495145631e-05, "loss": 0.2654, "step": 8541 }, { "epoch": 82.82650602409639, "grad_norm": 14.005816459655762, "learning_rate": 5.8533980582524274e-05, "loss": 0.4016, "step": 8542 }, { "epoch": 82.83614457831325, "grad_norm": 21.637950897216797, "learning_rate": 5.852912621359223e-05, "loss": 0.3591, "step": 8543 }, { "epoch": 82.84578313253012, "grad_norm": 16.238597869873047, "learning_rate": 5.8524271844660197e-05, "loss": 0.3528, "step": 8544 }, { "epoch": 82.855421686747, "grad_norm": 31.822734832763672, "learning_rate": 5.8519417475728154e-05, "loss": 0.4127, "step": 8545 }, { "epoch": 82.86506024096386, "grad_norm": 5.276298999786377, "learning_rate": 5.851456310679612e-05, "loss": 0.3795, "step": 8546 }, { "epoch": 82.87469879518072, "grad_norm": 11.796106338500977, "learning_rate": 5.850970873786408e-05, "loss": 0.4507, "step": 8547 }, { "epoch": 82.88433734939758, "grad_norm": 19.339771270751953, "learning_rate": 5.850485436893204e-05, "loss": 0.2325, "step": 8548 }, { "epoch": 82.89397590361446, "grad_norm": 17.656471252441406, "learning_rate": 5.85e-05, "loss": 0.3565, "step": 8549 }, { "epoch": 82.90361445783132, "grad_norm": 9.412932395935059, "learning_rate": 5.8495145631067963e-05, "loss": 0.322, "step": 8550 }, { "epoch": 82.91325301204819, "grad_norm": 24.754831314086914, "learning_rate": 5.849029126213592e-05, "loss": 0.2722, "step": 8551 }, { "epoch": 82.92289156626506, "grad_norm": 30.259151458740234, "learning_rate": 5.848543689320388e-05, "loss": 0.1559, "step": 8552 }, { "epoch": 82.93253012048193, "grad_norm": 2.766942024230957, "learning_rate": 5.8480582524271844e-05, "loss": 0.1732, "step": 8553 }, { "epoch": 82.94216867469879, "grad_norm": 42.802330017089844, "learning_rate": 5.84757281553398e-05, "loss": 0.5394, "step": 8554 }, { "epoch": 82.95180722891567, "grad_norm": 2.7849364280700684, "learning_rate": 5.8470873786407766e-05, "loss": 0.3095, "step": 8555 }, { "epoch": 82.96144578313253, "grad_norm": 16.112459182739258, "learning_rate": 5.8466019417475724e-05, "loss": 0.2741, "step": 8556 }, { "epoch": 82.9710843373494, "grad_norm": 34.48670959472656, "learning_rate": 5.846116504854369e-05, "loss": 0.398, "step": 8557 }, { "epoch": 82.98072289156626, "grad_norm": 12.686930656433105, "learning_rate": 5.8456310679611646e-05, "loss": 0.5026, "step": 8558 }, { "epoch": 82.99036144578314, "grad_norm": 9.235404968261719, "learning_rate": 5.845145631067962e-05, "loss": 0.347, "step": 8559 }, { "epoch": 83.00602409638554, "grad_norm": 3.921591281890869, "learning_rate": 5.844660194174758e-05, "loss": 0.1503, "step": 8560 }, { "epoch": 83.01566265060241, "grad_norm": 8.86647891998291, "learning_rate": 5.844174757281554e-05, "loss": 0.3326, "step": 8561 }, { "epoch": 83.02530120481927, "grad_norm": 40.91661834716797, "learning_rate": 5.8436893203883504e-05, "loss": 0.4172, "step": 8562 }, { "epoch": 83.03493975903615, "grad_norm": 41.54026412963867, "learning_rate": 5.843203883495146e-05, "loss": 0.3881, "step": 8563 }, { "epoch": 83.04457831325301, "grad_norm": 5.714669704437256, "learning_rate": 5.8427184466019426e-05, "loss": 0.3459, "step": 8564 }, { "epoch": 83.05421686746988, "grad_norm": 48.334407806396484, "learning_rate": 5.8422330097087384e-05, "loss": 0.3257, "step": 8565 }, { "epoch": 83.06385542168675, "grad_norm": 16.35714340209961, "learning_rate": 5.841747572815535e-05, "loss": 0.2819, "step": 8566 }, { "epoch": 83.07349397590362, "grad_norm": 33.49429702758789, "learning_rate": 5.8412621359223306e-05, "loss": 0.4973, "step": 8567 }, { "epoch": 83.08313253012048, "grad_norm": 32.08513641357422, "learning_rate": 5.840776699029127e-05, "loss": 0.2885, "step": 8568 }, { "epoch": 83.09277108433734, "grad_norm": 13.72921085357666, "learning_rate": 5.840291262135923e-05, "loss": 0.1654, "step": 8569 }, { "epoch": 83.10240963855422, "grad_norm": 6.831538677215576, "learning_rate": 5.839805825242719e-05, "loss": 0.2865, "step": 8570 }, { "epoch": 83.11204819277108, "grad_norm": 2.430971622467041, "learning_rate": 5.839320388349515e-05, "loss": 0.1649, "step": 8571 }, { "epoch": 83.12168674698795, "grad_norm": 2.9245193004608154, "learning_rate": 5.8388349514563115e-05, "loss": 0.2104, "step": 8572 }, { "epoch": 83.13132530120482, "grad_norm": 9.93771743774414, "learning_rate": 5.838349514563107e-05, "loss": 0.1777, "step": 8573 }, { "epoch": 83.14096385542169, "grad_norm": 6.793925762176514, "learning_rate": 5.837864077669903e-05, "loss": 0.3673, "step": 8574 }, { "epoch": 83.15060240963855, "grad_norm": 47.8244514465332, "learning_rate": 5.8373786407766996e-05, "loss": 0.449, "step": 8575 }, { "epoch": 83.16024096385541, "grad_norm": 6.611181259155273, "learning_rate": 5.836893203883495e-05, "loss": 0.3774, "step": 8576 }, { "epoch": 83.16987951807229, "grad_norm": 5.132513523101807, "learning_rate": 5.836407766990292e-05, "loss": 0.2921, "step": 8577 }, { "epoch": 83.17951807228916, "grad_norm": 21.358104705810547, "learning_rate": 5.8359223300970876e-05, "loss": 0.3962, "step": 8578 }, { "epoch": 83.18915662650602, "grad_norm": 11.017522811889648, "learning_rate": 5.835436893203884e-05, "loss": 0.2706, "step": 8579 }, { "epoch": 83.1987951807229, "grad_norm": 2.9663078784942627, "learning_rate": 5.83495145631068e-05, "loss": 0.2598, "step": 8580 }, { "epoch": 83.20843373493976, "grad_norm": 13.23237419128418, "learning_rate": 5.834466019417476e-05, "loss": 0.306, "step": 8581 }, { "epoch": 83.21807228915662, "grad_norm": 73.27619171142578, "learning_rate": 5.833980582524272e-05, "loss": 0.4019, "step": 8582 }, { "epoch": 83.2277108433735, "grad_norm": 13.774178504943848, "learning_rate": 5.8334951456310685e-05, "loss": 0.5118, "step": 8583 }, { "epoch": 83.23734939759036, "grad_norm": 2.7235679626464844, "learning_rate": 5.833009708737864e-05, "loss": 0.3509, "step": 8584 }, { "epoch": 83.24698795180723, "grad_norm": 76.6046371459961, "learning_rate": 5.832524271844661e-05, "loss": 0.2407, "step": 8585 }, { "epoch": 83.25662650602409, "grad_norm": 24.88450813293457, "learning_rate": 5.8320388349514565e-05, "loss": 0.6373, "step": 8586 }, { "epoch": 83.26626506024097, "grad_norm": 10.537386894226074, "learning_rate": 5.831553398058253e-05, "loss": 0.3306, "step": 8587 }, { "epoch": 83.27590361445783, "grad_norm": 31.13516616821289, "learning_rate": 5.831067961165049e-05, "loss": 0.3649, "step": 8588 }, { "epoch": 83.2855421686747, "grad_norm": 32.278873443603516, "learning_rate": 5.8305825242718445e-05, "loss": 0.454, "step": 8589 }, { "epoch": 83.29518072289157, "grad_norm": 5.733341693878174, "learning_rate": 5.830097087378641e-05, "loss": 0.2118, "step": 8590 }, { "epoch": 83.30481927710844, "grad_norm": 5.21045446395874, "learning_rate": 5.829611650485437e-05, "loss": 0.1712, "step": 8591 }, { "epoch": 83.3144578313253, "grad_norm": 13.26318073272705, "learning_rate": 5.829126213592233e-05, "loss": 0.2132, "step": 8592 }, { "epoch": 83.32409638554216, "grad_norm": 5.924147129058838, "learning_rate": 5.828640776699029e-05, "loss": 0.2976, "step": 8593 }, { "epoch": 83.33373493975904, "grad_norm": 29.247821807861328, "learning_rate": 5.8281553398058254e-05, "loss": 0.2175, "step": 8594 }, { "epoch": 83.3433734939759, "grad_norm": 5.993221282958984, "learning_rate": 5.827669902912621e-05, "loss": 0.3854, "step": 8595 }, { "epoch": 83.35301204819277, "grad_norm": 9.781805038452148, "learning_rate": 5.8271844660194176e-05, "loss": 0.349, "step": 8596 }, { "epoch": 83.36265060240964, "grad_norm": 8.42052936553955, "learning_rate": 5.8266990291262134e-05, "loss": 0.3685, "step": 8597 }, { "epoch": 83.37228915662651, "grad_norm": 6.603639125823975, "learning_rate": 5.82621359223301e-05, "loss": 0.1792, "step": 8598 }, { "epoch": 83.38192771084337, "grad_norm": 12.848904609680176, "learning_rate": 5.8257281553398056e-05, "loss": 0.3582, "step": 8599 }, { "epoch": 83.39156626506023, "grad_norm": 1.5402491092681885, "learning_rate": 5.825242718446602e-05, "loss": 0.1225, "step": 8600 }, { "epoch": 83.40120481927711, "grad_norm": 8.237900733947754, "learning_rate": 5.824757281553398e-05, "loss": 0.3475, "step": 8601 }, { "epoch": 83.41084337349398, "grad_norm": 27.945554733276367, "learning_rate": 5.824271844660194e-05, "loss": 0.3308, "step": 8602 }, { "epoch": 83.42048192771084, "grad_norm": 13.611452102661133, "learning_rate": 5.82378640776699e-05, "loss": 0.2589, "step": 8603 }, { "epoch": 83.43012048192772, "grad_norm": 4.827247142791748, "learning_rate": 5.8233009708737865e-05, "loss": 0.4043, "step": 8604 }, { "epoch": 83.43975903614458, "grad_norm": 12.369620323181152, "learning_rate": 5.822815533980582e-05, "loss": 0.3622, "step": 8605 }, { "epoch": 83.44939759036144, "grad_norm": 16.784446716308594, "learning_rate": 5.822330097087378e-05, "loss": 0.2143, "step": 8606 }, { "epoch": 83.45903614457832, "grad_norm": 7.617030143737793, "learning_rate": 5.8218446601941745e-05, "loss": 0.2516, "step": 8607 }, { "epoch": 83.46867469879518, "grad_norm": 5.841093063354492, "learning_rate": 5.82135922330097e-05, "loss": 0.2732, "step": 8608 }, { "epoch": 83.47831325301205, "grad_norm": 5.928661823272705, "learning_rate": 5.820873786407768e-05, "loss": 0.3444, "step": 8609 }, { "epoch": 83.48795180722891, "grad_norm": 4.354668617248535, "learning_rate": 5.820388349514564e-05, "loss": 0.1724, "step": 8610 }, { "epoch": 83.49759036144579, "grad_norm": 1.888953685760498, "learning_rate": 5.81990291262136e-05, "loss": 0.1438, "step": 8611 }, { "epoch": 83.50722891566265, "grad_norm": 8.537818908691406, "learning_rate": 5.819417475728156e-05, "loss": 0.1771, "step": 8612 }, { "epoch": 83.51686746987951, "grad_norm": 6.214034557342529, "learning_rate": 5.818932038834952e-05, "loss": 0.3428, "step": 8613 }, { "epoch": 83.52650602409639, "grad_norm": 7.440403938293457, "learning_rate": 5.8184466019417484e-05, "loss": 0.2806, "step": 8614 }, { "epoch": 83.53614457831326, "grad_norm": 10.122100830078125, "learning_rate": 5.817961165048544e-05, "loss": 0.2589, "step": 8615 }, { "epoch": 83.54578313253012, "grad_norm": 42.10194778442383, "learning_rate": 5.8174757281553406e-05, "loss": 0.2731, "step": 8616 }, { "epoch": 83.55542168674698, "grad_norm": 6.847631931304932, "learning_rate": 5.8169902912621364e-05, "loss": 0.1857, "step": 8617 }, { "epoch": 83.56506024096386, "grad_norm": 5.108174800872803, "learning_rate": 5.816504854368933e-05, "loss": 0.1349, "step": 8618 }, { "epoch": 83.57469879518072, "grad_norm": 23.509899139404297, "learning_rate": 5.8160194174757286e-05, "loss": 0.2779, "step": 8619 }, { "epoch": 83.58433734939759, "grad_norm": 35.95549774169922, "learning_rate": 5.815533980582525e-05, "loss": 0.2518, "step": 8620 }, { "epoch": 83.59397590361446, "grad_norm": 24.106189727783203, "learning_rate": 5.815048543689321e-05, "loss": 0.3561, "step": 8621 }, { "epoch": 83.60361445783133, "grad_norm": 3.555830478668213, "learning_rate": 5.814563106796117e-05, "loss": 0.3359, "step": 8622 }, { "epoch": 83.61325301204819, "grad_norm": 10.12009048461914, "learning_rate": 5.814077669902913e-05, "loss": 0.315, "step": 8623 }, { "epoch": 83.62289156626505, "grad_norm": 15.368971824645996, "learning_rate": 5.8135922330097095e-05, "loss": 0.1969, "step": 8624 }, { "epoch": 83.63253012048193, "grad_norm": 5.491776466369629, "learning_rate": 5.813106796116505e-05, "loss": 0.2289, "step": 8625 }, { "epoch": 83.6421686746988, "grad_norm": 23.33831787109375, "learning_rate": 5.812621359223302e-05, "loss": 0.1142, "step": 8626 }, { "epoch": 83.65180722891566, "grad_norm": 14.038900375366211, "learning_rate": 5.8121359223300975e-05, "loss": 0.2646, "step": 8627 }, { "epoch": 83.66144578313254, "grad_norm": 3.5450387001037598, "learning_rate": 5.811650485436893e-05, "loss": 0.1608, "step": 8628 }, { "epoch": 83.6710843373494, "grad_norm": 14.528005599975586, "learning_rate": 5.81116504854369e-05, "loss": 0.2963, "step": 8629 }, { "epoch": 83.68072289156626, "grad_norm": 6.602867126464844, "learning_rate": 5.8106796116504855e-05, "loss": 0.1889, "step": 8630 }, { "epoch": 83.69036144578314, "grad_norm": 7.258765697479248, "learning_rate": 5.810194174757282e-05, "loss": 0.4326, "step": 8631 }, { "epoch": 83.7, "grad_norm": 13.671194076538086, "learning_rate": 5.809708737864078e-05, "loss": 0.2174, "step": 8632 }, { "epoch": 83.70963855421687, "grad_norm": 15.224279403686523, "learning_rate": 5.809223300970874e-05, "loss": 0.2793, "step": 8633 }, { "epoch": 83.71927710843373, "grad_norm": 22.683719635009766, "learning_rate": 5.80873786407767e-05, "loss": 0.2161, "step": 8634 }, { "epoch": 83.7289156626506, "grad_norm": 7.792816638946533, "learning_rate": 5.8082524271844664e-05, "loss": 0.3044, "step": 8635 }, { "epoch": 83.73855421686747, "grad_norm": 31.499900817871094, "learning_rate": 5.807766990291262e-05, "loss": 0.2509, "step": 8636 }, { "epoch": 83.74819277108433, "grad_norm": 15.512024879455566, "learning_rate": 5.8072815533980587e-05, "loss": 0.2373, "step": 8637 }, { "epoch": 83.75783132530121, "grad_norm": 6.4777302742004395, "learning_rate": 5.8067961165048544e-05, "loss": 0.3879, "step": 8638 }, { "epoch": 83.76746987951807, "grad_norm": 13.482990264892578, "learning_rate": 5.806310679611651e-05, "loss": 0.2674, "step": 8639 }, { "epoch": 83.77710843373494, "grad_norm": 4.944138526916504, "learning_rate": 5.8058252427184467e-05, "loss": 0.2096, "step": 8640 }, { "epoch": 83.7867469879518, "grad_norm": 9.645505905151367, "learning_rate": 5.805339805825243e-05, "loss": 0.2623, "step": 8641 }, { "epoch": 83.79638554216868, "grad_norm": 6.757660388946533, "learning_rate": 5.804854368932039e-05, "loss": 0.2921, "step": 8642 }, { "epoch": 83.80602409638554, "grad_norm": 22.952346801757812, "learning_rate": 5.8043689320388353e-05, "loss": 0.2463, "step": 8643 }, { "epoch": 83.8156626506024, "grad_norm": 41.244712829589844, "learning_rate": 5.803883495145631e-05, "loss": 0.8992, "step": 8644 }, { "epoch": 83.82530120481928, "grad_norm": 22.51830291748047, "learning_rate": 5.803398058252427e-05, "loss": 0.2566, "step": 8645 }, { "epoch": 83.83493975903615, "grad_norm": 17.709449768066406, "learning_rate": 5.8029126213592233e-05, "loss": 0.2198, "step": 8646 }, { "epoch": 83.84457831325301, "grad_norm": 5.7235002517700195, "learning_rate": 5.802427184466019e-05, "loss": 0.4274, "step": 8647 }, { "epoch": 83.85421686746987, "grad_norm": 7.7638840675354, "learning_rate": 5.8019417475728156e-05, "loss": 0.2512, "step": 8648 }, { "epoch": 83.86385542168675, "grad_norm": 10.656392097473145, "learning_rate": 5.8014563106796114e-05, "loss": 0.3591, "step": 8649 }, { "epoch": 83.87349397590361, "grad_norm": 3.1431570053100586, "learning_rate": 5.800970873786408e-05, "loss": 0.2685, "step": 8650 }, { "epoch": 83.88313253012048, "grad_norm": 14.101252555847168, "learning_rate": 5.8004854368932036e-05, "loss": 0.5602, "step": 8651 }, { "epoch": 83.89277108433735, "grad_norm": 6.632186412811279, "learning_rate": 5.8e-05, "loss": 0.2014, "step": 8652 }, { "epoch": 83.90240963855422, "grad_norm": 28.959287643432617, "learning_rate": 5.799514563106796e-05, "loss": 0.2559, "step": 8653 }, { "epoch": 83.91204819277108, "grad_norm": 4.625728130340576, "learning_rate": 5.799029126213592e-05, "loss": 0.3568, "step": 8654 }, { "epoch": 83.92168674698796, "grad_norm": 30.742244720458984, "learning_rate": 5.798543689320388e-05, "loss": 0.3901, "step": 8655 }, { "epoch": 83.93132530120482, "grad_norm": 23.811386108398438, "learning_rate": 5.7980582524271845e-05, "loss": 0.2026, "step": 8656 }, { "epoch": 83.94096385542169, "grad_norm": 17.245805740356445, "learning_rate": 5.79757281553398e-05, "loss": 0.2437, "step": 8657 }, { "epoch": 83.95060240963855, "grad_norm": 3.6412386894226074, "learning_rate": 5.797087378640777e-05, "loss": 0.0938, "step": 8658 }, { "epoch": 83.96024096385543, "grad_norm": 21.05669593811035, "learning_rate": 5.7966019417475725e-05, "loss": 0.2491, "step": 8659 }, { "epoch": 83.96987951807229, "grad_norm": 10.9464111328125, "learning_rate": 5.7961165048543696e-05, "loss": 0.1646, "step": 8660 }, { "epoch": 83.97951807228915, "grad_norm": 27.50814437866211, "learning_rate": 5.795631067961166e-05, "loss": 0.2402, "step": 8661 }, { "epoch": 83.98915662650603, "grad_norm": 27.990591049194336, "learning_rate": 5.795145631067962e-05, "loss": 0.5944, "step": 8662 }, { "epoch": 84.00481927710844, "grad_norm": 38.236915588378906, "learning_rate": 5.794660194174758e-05, "loss": 0.226, "step": 8663 }, { "epoch": 84.0144578313253, "grad_norm": 13.910752296447754, "learning_rate": 5.794174757281554e-05, "loss": 0.1214, "step": 8664 }, { "epoch": 84.02409638554217, "grad_norm": 4.122941970825195, "learning_rate": 5.7936893203883505e-05, "loss": 0.1338, "step": 8665 }, { "epoch": 84.03373493975904, "grad_norm": 36.823482513427734, "learning_rate": 5.793203883495146e-05, "loss": 0.3119, "step": 8666 }, { "epoch": 84.0433734939759, "grad_norm": 11.150875091552734, "learning_rate": 5.792718446601942e-05, "loss": 0.4196, "step": 8667 }, { "epoch": 84.05301204819277, "grad_norm": 9.26663589477539, "learning_rate": 5.7922330097087385e-05, "loss": 0.3256, "step": 8668 }, { "epoch": 84.06265060240963, "grad_norm": 3.665827512741089, "learning_rate": 5.791747572815534e-05, "loss": 0.183, "step": 8669 }, { "epoch": 84.07228915662651, "grad_norm": 11.477936744689941, "learning_rate": 5.791262135922331e-05, "loss": 0.3056, "step": 8670 }, { "epoch": 84.08192771084337, "grad_norm": 36.90306854248047, "learning_rate": 5.7907766990291265e-05, "loss": 0.1361, "step": 8671 }, { "epoch": 84.09156626506024, "grad_norm": 9.103229522705078, "learning_rate": 5.790291262135923e-05, "loss": 0.4538, "step": 8672 }, { "epoch": 84.10120481927711, "grad_norm": 3.5296530723571777, "learning_rate": 5.789805825242719e-05, "loss": 0.313, "step": 8673 }, { "epoch": 84.11084337349398, "grad_norm": 3.494337797164917, "learning_rate": 5.789320388349515e-05, "loss": 0.1873, "step": 8674 }, { "epoch": 84.12048192771084, "grad_norm": 15.914870262145996, "learning_rate": 5.788834951456311e-05, "loss": 0.343, "step": 8675 }, { "epoch": 84.1301204819277, "grad_norm": 6.4110026359558105, "learning_rate": 5.7883495145631075e-05, "loss": 0.4679, "step": 8676 }, { "epoch": 84.13975903614458, "grad_norm": 19.991243362426758, "learning_rate": 5.787864077669903e-05, "loss": 0.2671, "step": 8677 }, { "epoch": 84.14939759036145, "grad_norm": 32.349525451660156, "learning_rate": 5.7873786407767e-05, "loss": 0.4202, "step": 8678 }, { "epoch": 84.15903614457831, "grad_norm": 16.864648818969727, "learning_rate": 5.7868932038834955e-05, "loss": 0.2569, "step": 8679 }, { "epoch": 84.16867469879519, "grad_norm": 11.076180458068848, "learning_rate": 5.786407766990292e-05, "loss": 0.3149, "step": 8680 }, { "epoch": 84.17831325301205, "grad_norm": 25.912940979003906, "learning_rate": 5.785922330097088e-05, "loss": 0.5866, "step": 8681 }, { "epoch": 84.18795180722891, "grad_norm": 17.064516067504883, "learning_rate": 5.785436893203884e-05, "loss": 0.4492, "step": 8682 }, { "epoch": 84.19759036144578, "grad_norm": 8.594548225402832, "learning_rate": 5.78495145631068e-05, "loss": 0.5657, "step": 8683 }, { "epoch": 84.20722891566265, "grad_norm": 3.811728000640869, "learning_rate": 5.784466019417476e-05, "loss": 0.2844, "step": 8684 }, { "epoch": 84.21686746987952, "grad_norm": 6.788392066955566, "learning_rate": 5.783980582524272e-05, "loss": 0.1628, "step": 8685 }, { "epoch": 84.22650602409638, "grad_norm": 13.511249542236328, "learning_rate": 5.783495145631068e-05, "loss": 0.2957, "step": 8686 }, { "epoch": 84.23614457831326, "grad_norm": 43.25822830200195, "learning_rate": 5.7830097087378644e-05, "loss": 0.4058, "step": 8687 }, { "epoch": 84.24578313253012, "grad_norm": 13.586770057678223, "learning_rate": 5.78252427184466e-05, "loss": 0.1653, "step": 8688 }, { "epoch": 84.25542168674698, "grad_norm": 6.779877662658691, "learning_rate": 5.7820388349514566e-05, "loss": 0.1388, "step": 8689 }, { "epoch": 84.26506024096386, "grad_norm": 7.337469577789307, "learning_rate": 5.7815533980582524e-05, "loss": 0.4271, "step": 8690 }, { "epoch": 84.27469879518073, "grad_norm": 13.47025203704834, "learning_rate": 5.781067961165049e-05, "loss": 0.2731, "step": 8691 }, { "epoch": 84.28433734939759, "grad_norm": 25.36886978149414, "learning_rate": 5.7805825242718446e-05, "loss": 0.5494, "step": 8692 }, { "epoch": 84.29397590361445, "grad_norm": 18.89041519165039, "learning_rate": 5.780097087378641e-05, "loss": 0.3164, "step": 8693 }, { "epoch": 84.30361445783133, "grad_norm": 9.07898235321045, "learning_rate": 5.779611650485437e-05, "loss": 0.429, "step": 8694 }, { "epoch": 84.3132530120482, "grad_norm": 22.186506271362305, "learning_rate": 5.779126213592233e-05, "loss": 0.3243, "step": 8695 }, { "epoch": 84.32289156626506, "grad_norm": 7.220738410949707, "learning_rate": 5.778640776699029e-05, "loss": 0.3308, "step": 8696 }, { "epoch": 84.33253012048193, "grad_norm": 3.8234074115753174, "learning_rate": 5.7781553398058255e-05, "loss": 0.2209, "step": 8697 }, { "epoch": 84.3421686746988, "grad_norm": 7.940439701080322, "learning_rate": 5.777669902912621e-05, "loss": 0.3959, "step": 8698 }, { "epoch": 84.35180722891566, "grad_norm": 12.419965744018555, "learning_rate": 5.777184466019418e-05, "loss": 0.5593, "step": 8699 }, { "epoch": 84.36144578313252, "grad_norm": 13.614517211914062, "learning_rate": 5.7766990291262135e-05, "loss": 0.256, "step": 8700 }, { "epoch": 84.3710843373494, "grad_norm": 17.12227439880371, "learning_rate": 5.776213592233009e-05, "loss": 0.3482, "step": 8701 }, { "epoch": 84.38072289156626, "grad_norm": 11.225646018981934, "learning_rate": 5.775728155339806e-05, "loss": 0.2156, "step": 8702 }, { "epoch": 84.39036144578313, "grad_norm": 13.109512329101562, "learning_rate": 5.7752427184466015e-05, "loss": 0.1287, "step": 8703 }, { "epoch": 84.4, "grad_norm": 5.103185653686523, "learning_rate": 5.774757281553398e-05, "loss": 0.5671, "step": 8704 }, { "epoch": 84.40963855421687, "grad_norm": 3.7720823287963867, "learning_rate": 5.774271844660194e-05, "loss": 0.1496, "step": 8705 }, { "epoch": 84.41927710843373, "grad_norm": 7.353865146636963, "learning_rate": 5.77378640776699e-05, "loss": 0.2253, "step": 8706 }, { "epoch": 84.4289156626506, "grad_norm": 70.42997741699219, "learning_rate": 5.773300970873786e-05, "loss": 0.3739, "step": 8707 }, { "epoch": 84.43855421686747, "grad_norm": 9.974000930786133, "learning_rate": 5.7728155339805825e-05, "loss": 0.2897, "step": 8708 }, { "epoch": 84.44819277108434, "grad_norm": 23.34304428100586, "learning_rate": 5.772330097087378e-05, "loss": 0.3319, "step": 8709 }, { "epoch": 84.4578313253012, "grad_norm": 9.202489852905273, "learning_rate": 5.7718446601941754e-05, "loss": 0.4849, "step": 8710 }, { "epoch": 84.46746987951808, "grad_norm": 16.675350189208984, "learning_rate": 5.771359223300972e-05, "loss": 0.2855, "step": 8711 }, { "epoch": 84.47710843373494, "grad_norm": 3.6237130165100098, "learning_rate": 5.7708737864077676e-05, "loss": 0.2516, "step": 8712 }, { "epoch": 84.4867469879518, "grad_norm": 30.675939559936523, "learning_rate": 5.770388349514564e-05, "loss": 0.2825, "step": 8713 }, { "epoch": 84.49638554216868, "grad_norm": 3.8965861797332764, "learning_rate": 5.76990291262136e-05, "loss": 0.121, "step": 8714 }, { "epoch": 84.50602409638554, "grad_norm": 19.321271896362305, "learning_rate": 5.769417475728156e-05, "loss": 0.2423, "step": 8715 }, { "epoch": 84.51566265060241, "grad_norm": 9.968659400939941, "learning_rate": 5.768932038834952e-05, "loss": 0.271, "step": 8716 }, { "epoch": 84.52530120481927, "grad_norm": 6.895514965057373, "learning_rate": 5.7684466019417485e-05, "loss": 0.3619, "step": 8717 }, { "epoch": 84.53493975903615, "grad_norm": 5.366795539855957, "learning_rate": 5.767961165048544e-05, "loss": 0.2067, "step": 8718 }, { "epoch": 84.54457831325301, "grad_norm": 13.174483299255371, "learning_rate": 5.767475728155341e-05, "loss": 0.4143, "step": 8719 }, { "epoch": 84.55421686746988, "grad_norm": 6.286105632781982, "learning_rate": 5.7669902912621365e-05, "loss": 0.2834, "step": 8720 }, { "epoch": 84.56385542168675, "grad_norm": 15.001124382019043, "learning_rate": 5.766504854368933e-05, "loss": 0.3443, "step": 8721 }, { "epoch": 84.57349397590362, "grad_norm": 7.4698710441589355, "learning_rate": 5.766019417475729e-05, "loss": 0.3411, "step": 8722 }, { "epoch": 84.58313253012048, "grad_norm": 12.729683876037598, "learning_rate": 5.7655339805825245e-05, "loss": 0.3818, "step": 8723 }, { "epoch": 84.59277108433734, "grad_norm": 12.421784400939941, "learning_rate": 5.765048543689321e-05, "loss": 0.2699, "step": 8724 }, { "epoch": 84.60240963855422, "grad_norm": 18.065105438232422, "learning_rate": 5.764563106796117e-05, "loss": 0.3341, "step": 8725 }, { "epoch": 84.61204819277108, "grad_norm": 31.440025329589844, "learning_rate": 5.764077669902913e-05, "loss": 0.5546, "step": 8726 }, { "epoch": 84.62168674698795, "grad_norm": 26.348140716552734, "learning_rate": 5.763592233009709e-05, "loss": 0.5137, "step": 8727 }, { "epoch": 84.63132530120482, "grad_norm": 10.007030487060547, "learning_rate": 5.7631067961165054e-05, "loss": 0.3499, "step": 8728 }, { "epoch": 84.64096385542169, "grad_norm": 5.760046482086182, "learning_rate": 5.762621359223301e-05, "loss": 0.1449, "step": 8729 }, { "epoch": 84.65060240963855, "grad_norm": 15.730484962463379, "learning_rate": 5.7621359223300976e-05, "loss": 0.2424, "step": 8730 }, { "epoch": 84.66024096385541, "grad_norm": 17.965444564819336, "learning_rate": 5.7616504854368934e-05, "loss": 0.2811, "step": 8731 }, { "epoch": 84.66987951807229, "grad_norm": 14.456707000732422, "learning_rate": 5.76116504854369e-05, "loss": 0.2752, "step": 8732 }, { "epoch": 84.67951807228916, "grad_norm": 8.382655143737793, "learning_rate": 5.7606796116504857e-05, "loss": 0.3126, "step": 8733 }, { "epoch": 84.68915662650602, "grad_norm": 5.958867073059082, "learning_rate": 5.760194174757282e-05, "loss": 0.1533, "step": 8734 }, { "epoch": 84.6987951807229, "grad_norm": 24.60144805908203, "learning_rate": 5.759708737864078e-05, "loss": 0.2517, "step": 8735 }, { "epoch": 84.70843373493976, "grad_norm": 9.550114631652832, "learning_rate": 5.759223300970874e-05, "loss": 0.162, "step": 8736 }, { "epoch": 84.71807228915662, "grad_norm": 2.4872047901153564, "learning_rate": 5.75873786407767e-05, "loss": 0.2392, "step": 8737 }, { "epoch": 84.7277108433735, "grad_norm": 3.484584331512451, "learning_rate": 5.758252427184466e-05, "loss": 0.2753, "step": 8738 }, { "epoch": 84.73734939759036, "grad_norm": 8.27023983001709, "learning_rate": 5.7577669902912623e-05, "loss": 0.3746, "step": 8739 }, { "epoch": 84.74698795180723, "grad_norm": 43.349884033203125, "learning_rate": 5.757281553398058e-05, "loss": 0.2415, "step": 8740 }, { "epoch": 84.75662650602409, "grad_norm": 2.968287706375122, "learning_rate": 5.7567961165048546e-05, "loss": 0.1562, "step": 8741 }, { "epoch": 84.76626506024097, "grad_norm": 8.055079460144043, "learning_rate": 5.7563106796116503e-05, "loss": 0.368, "step": 8742 }, { "epoch": 84.77590361445783, "grad_norm": 12.652810096740723, "learning_rate": 5.755825242718447e-05, "loss": 0.1803, "step": 8743 }, { "epoch": 84.7855421686747, "grad_norm": 10.953289031982422, "learning_rate": 5.7553398058252426e-05, "loss": 0.2374, "step": 8744 }, { "epoch": 84.79518072289157, "grad_norm": 6.677964687347412, "learning_rate": 5.754854368932039e-05, "loss": 0.2638, "step": 8745 }, { "epoch": 84.80481927710844, "grad_norm": 16.91888999938965, "learning_rate": 5.754368932038835e-05, "loss": 0.2797, "step": 8746 }, { "epoch": 84.8144578313253, "grad_norm": 6.501942157745361, "learning_rate": 5.753883495145631e-05, "loss": 0.1846, "step": 8747 }, { "epoch": 84.82409638554216, "grad_norm": 10.934601783752441, "learning_rate": 5.753398058252427e-05, "loss": 0.3316, "step": 8748 }, { "epoch": 84.83373493975904, "grad_norm": 46.645748138427734, "learning_rate": 5.7529126213592235e-05, "loss": 0.2335, "step": 8749 }, { "epoch": 84.8433734939759, "grad_norm": 6.573051452636719, "learning_rate": 5.752427184466019e-05, "loss": 0.2209, "step": 8750 }, { "epoch": 84.85301204819277, "grad_norm": 3.1833784580230713, "learning_rate": 5.751941747572816e-05, "loss": 0.2005, "step": 8751 }, { "epoch": 84.86265060240964, "grad_norm": 8.4375, "learning_rate": 5.7514563106796115e-05, "loss": 0.5429, "step": 8752 }, { "epoch": 84.87228915662651, "grad_norm": 7.136817932128906, "learning_rate": 5.750970873786408e-05, "loss": 0.1914, "step": 8753 }, { "epoch": 84.88192771084337, "grad_norm": 13.764749526977539, "learning_rate": 5.750485436893204e-05, "loss": 0.1654, "step": 8754 }, { "epoch": 84.89156626506023, "grad_norm": 43.60764694213867, "learning_rate": 5.7499999999999995e-05, "loss": 0.4412, "step": 8755 }, { "epoch": 84.90120481927711, "grad_norm": 12.93140697479248, "learning_rate": 5.749514563106796e-05, "loss": 0.1921, "step": 8756 }, { "epoch": 84.91084337349398, "grad_norm": 5.286777973175049, "learning_rate": 5.749029126213592e-05, "loss": 0.3648, "step": 8757 }, { "epoch": 84.92048192771084, "grad_norm": 26.05895233154297, "learning_rate": 5.748543689320388e-05, "loss": 0.2544, "step": 8758 }, { "epoch": 84.93012048192772, "grad_norm": 9.393935203552246, "learning_rate": 5.748058252427184e-05, "loss": 0.3972, "step": 8759 }, { "epoch": 84.93975903614458, "grad_norm": 15.63355541229248, "learning_rate": 5.7475728155339804e-05, "loss": 0.3219, "step": 8760 }, { "epoch": 84.94939759036144, "grad_norm": 12.395764350891113, "learning_rate": 5.7470873786407775e-05, "loss": 0.1938, "step": 8761 }, { "epoch": 84.95903614457832, "grad_norm": 29.039587020874023, "learning_rate": 5.746601941747573e-05, "loss": 0.2396, "step": 8762 }, { "epoch": 84.96867469879518, "grad_norm": 7.043245315551758, "learning_rate": 5.74611650485437e-05, "loss": 0.2415, "step": 8763 }, { "epoch": 84.97831325301205, "grad_norm": 4.678742408752441, "learning_rate": 5.7456310679611655e-05, "loss": 0.1124, "step": 8764 }, { "epoch": 84.98795180722891, "grad_norm": 42.387516021728516, "learning_rate": 5.745145631067962e-05, "loss": 0.1685, "step": 8765 }, { "epoch": 85.00361445783132, "grad_norm": 5.246509075164795, "learning_rate": 5.744660194174758e-05, "loss": 0.2969, "step": 8766 }, { "epoch": 85.0132530120482, "grad_norm": 10.031105041503906, "learning_rate": 5.744174757281554e-05, "loss": 0.1581, "step": 8767 }, { "epoch": 85.02289156626506, "grad_norm": 6.801429748535156, "learning_rate": 5.74368932038835e-05, "loss": 0.1624, "step": 8768 }, { "epoch": 85.03253012048192, "grad_norm": 9.102025032043457, "learning_rate": 5.7432038834951465e-05, "loss": 0.3003, "step": 8769 }, { "epoch": 85.0421686746988, "grad_norm": 11.273456573486328, "learning_rate": 5.742718446601942e-05, "loss": 0.223, "step": 8770 }, { "epoch": 85.05180722891566, "grad_norm": 11.083453178405762, "learning_rate": 5.742233009708739e-05, "loss": 0.5215, "step": 8771 }, { "epoch": 85.06144578313253, "grad_norm": 12.349853515625, "learning_rate": 5.7417475728155345e-05, "loss": 0.4919, "step": 8772 }, { "epoch": 85.0710843373494, "grad_norm": 11.452850341796875, "learning_rate": 5.741262135922331e-05, "loss": 0.3312, "step": 8773 }, { "epoch": 85.08072289156627, "grad_norm": 9.132658958435059, "learning_rate": 5.740776699029127e-05, "loss": 0.1865, "step": 8774 }, { "epoch": 85.09036144578313, "grad_norm": 33.9458122253418, "learning_rate": 5.740291262135923e-05, "loss": 0.3715, "step": 8775 }, { "epoch": 85.1, "grad_norm": 8.92958927154541, "learning_rate": 5.739805825242719e-05, "loss": 0.382, "step": 8776 }, { "epoch": 85.10963855421687, "grad_norm": 8.105829238891602, "learning_rate": 5.739320388349515e-05, "loss": 0.2966, "step": 8777 }, { "epoch": 85.11927710843374, "grad_norm": 8.129788398742676, "learning_rate": 5.738834951456311e-05, "loss": 0.28, "step": 8778 }, { "epoch": 85.1289156626506, "grad_norm": 5.695988178253174, "learning_rate": 5.738349514563107e-05, "loss": 0.5419, "step": 8779 }, { "epoch": 85.13855421686748, "grad_norm": 12.024994850158691, "learning_rate": 5.7378640776699034e-05, "loss": 0.2562, "step": 8780 }, { "epoch": 85.14819277108434, "grad_norm": 16.759017944335938, "learning_rate": 5.737378640776699e-05, "loss": 0.3799, "step": 8781 }, { "epoch": 85.1578313253012, "grad_norm": 7.15967321395874, "learning_rate": 5.7368932038834956e-05, "loss": 0.1787, "step": 8782 }, { "epoch": 85.16746987951807, "grad_norm": 8.173938751220703, "learning_rate": 5.7364077669902914e-05, "loss": 0.3758, "step": 8783 }, { "epoch": 85.17710843373494, "grad_norm": 24.531635284423828, "learning_rate": 5.735922330097088e-05, "loss": 0.2865, "step": 8784 }, { "epoch": 85.1867469879518, "grad_norm": 9.069561004638672, "learning_rate": 5.7354368932038836e-05, "loss": 0.182, "step": 8785 }, { "epoch": 85.19638554216867, "grad_norm": 5.859248638153076, "learning_rate": 5.73495145631068e-05, "loss": 0.178, "step": 8786 }, { "epoch": 85.20602409638555, "grad_norm": 12.48189640045166, "learning_rate": 5.734466019417476e-05, "loss": 0.2112, "step": 8787 }, { "epoch": 85.21566265060241, "grad_norm": 18.59886932373047, "learning_rate": 5.733980582524272e-05, "loss": 0.213, "step": 8788 }, { "epoch": 85.22530120481927, "grad_norm": 6.098861217498779, "learning_rate": 5.733495145631068e-05, "loss": 0.2356, "step": 8789 }, { "epoch": 85.23493975903614, "grad_norm": 9.211566925048828, "learning_rate": 5.7330097087378645e-05, "loss": 0.217, "step": 8790 }, { "epoch": 85.24457831325302, "grad_norm": 21.783411026000977, "learning_rate": 5.73252427184466e-05, "loss": 0.4229, "step": 8791 }, { "epoch": 85.25421686746988, "grad_norm": 6.408714294433594, "learning_rate": 5.732038834951457e-05, "loss": 0.1388, "step": 8792 }, { "epoch": 85.26385542168674, "grad_norm": 19.281808853149414, "learning_rate": 5.7315533980582525e-05, "loss": 0.2633, "step": 8793 }, { "epoch": 85.27349397590362, "grad_norm": 13.307767868041992, "learning_rate": 5.731067961165048e-05, "loss": 0.1971, "step": 8794 }, { "epoch": 85.28313253012048, "grad_norm": 22.118589401245117, "learning_rate": 5.730582524271845e-05, "loss": 0.4627, "step": 8795 }, { "epoch": 85.29277108433735, "grad_norm": 7.80518102645874, "learning_rate": 5.7300970873786405e-05, "loss": 0.3689, "step": 8796 }, { "epoch": 85.30240963855422, "grad_norm": 30.260698318481445, "learning_rate": 5.729611650485437e-05, "loss": 0.3446, "step": 8797 }, { "epoch": 85.31204819277109, "grad_norm": 13.24779987335205, "learning_rate": 5.729126213592233e-05, "loss": 0.3746, "step": 8798 }, { "epoch": 85.32168674698795, "grad_norm": 24.363359451293945, "learning_rate": 5.728640776699029e-05, "loss": 0.4142, "step": 8799 }, { "epoch": 85.33132530120481, "grad_norm": 17.74969482421875, "learning_rate": 5.728155339805825e-05, "loss": 0.3967, "step": 8800 }, { "epoch": 85.34096385542169, "grad_norm": 5.056265354156494, "learning_rate": 5.7276699029126214e-05, "loss": 0.2362, "step": 8801 }, { "epoch": 85.35060240963855, "grad_norm": 10.31623363494873, "learning_rate": 5.727184466019417e-05, "loss": 0.2423, "step": 8802 }, { "epoch": 85.36024096385542, "grad_norm": 10.550823211669922, "learning_rate": 5.726699029126214e-05, "loss": 0.2647, "step": 8803 }, { "epoch": 85.3698795180723, "grad_norm": 23.54034996032715, "learning_rate": 5.7262135922330094e-05, "loss": 0.3298, "step": 8804 }, { "epoch": 85.37951807228916, "grad_norm": 9.436307907104492, "learning_rate": 5.725728155339806e-05, "loss": 0.2407, "step": 8805 }, { "epoch": 85.38915662650602, "grad_norm": 18.253976821899414, "learning_rate": 5.725242718446602e-05, "loss": 0.3951, "step": 8806 }, { "epoch": 85.39879518072289, "grad_norm": 7.280372619628906, "learning_rate": 5.724757281553398e-05, "loss": 0.2986, "step": 8807 }, { "epoch": 85.40843373493976, "grad_norm": 25.442626953125, "learning_rate": 5.724271844660194e-05, "loss": 0.2038, "step": 8808 }, { "epoch": 85.41807228915663, "grad_norm": 4.2559943199157715, "learning_rate": 5.7237864077669904e-05, "loss": 0.2396, "step": 8809 }, { "epoch": 85.42771084337349, "grad_norm": 21.814640045166016, "learning_rate": 5.723300970873786e-05, "loss": 0.4518, "step": 8810 }, { "epoch": 85.43734939759037, "grad_norm": 7.609894275665283, "learning_rate": 5.722815533980583e-05, "loss": 0.293, "step": 8811 }, { "epoch": 85.44698795180723, "grad_norm": 2.9932351112365723, "learning_rate": 5.72233009708738e-05, "loss": 0.0896, "step": 8812 }, { "epoch": 85.4566265060241, "grad_norm": 8.521585464477539, "learning_rate": 5.7218446601941755e-05, "loss": 0.2755, "step": 8813 }, { "epoch": 85.46626506024096, "grad_norm": 6.1907734870910645, "learning_rate": 5.721359223300972e-05, "loss": 0.2417, "step": 8814 }, { "epoch": 85.47590361445783, "grad_norm": 18.23073387145996, "learning_rate": 5.720873786407768e-05, "loss": 0.2975, "step": 8815 }, { "epoch": 85.4855421686747, "grad_norm": 7.832311153411865, "learning_rate": 5.7203883495145635e-05, "loss": 0.2291, "step": 8816 }, { "epoch": 85.49518072289156, "grad_norm": 10.760150909423828, "learning_rate": 5.71990291262136e-05, "loss": 0.2978, "step": 8817 }, { "epoch": 85.50481927710844, "grad_norm": 5.609379768371582, "learning_rate": 5.719417475728156e-05, "loss": 0.2037, "step": 8818 }, { "epoch": 85.5144578313253, "grad_norm": 28.34535789489746, "learning_rate": 5.718932038834952e-05, "loss": 0.4845, "step": 8819 }, { "epoch": 85.52409638554217, "grad_norm": 9.04687213897705, "learning_rate": 5.718446601941748e-05, "loss": 0.3839, "step": 8820 }, { "epoch": 85.53373493975904, "grad_norm": 6.43660831451416, "learning_rate": 5.7179611650485444e-05, "loss": 0.3076, "step": 8821 }, { "epoch": 85.5433734939759, "grad_norm": 13.514284133911133, "learning_rate": 5.71747572815534e-05, "loss": 0.2679, "step": 8822 }, { "epoch": 85.55301204819277, "grad_norm": 12.33747386932373, "learning_rate": 5.7169902912621366e-05, "loss": 0.414, "step": 8823 }, { "epoch": 85.56265060240963, "grad_norm": 3.5969903469085693, "learning_rate": 5.7165048543689324e-05, "loss": 0.242, "step": 8824 }, { "epoch": 85.57228915662651, "grad_norm": 21.306901931762695, "learning_rate": 5.716019417475729e-05, "loss": 0.2929, "step": 8825 }, { "epoch": 85.58192771084337, "grad_norm": 20.84523582458496, "learning_rate": 5.7155339805825246e-05, "loss": 0.4551, "step": 8826 }, { "epoch": 85.59156626506024, "grad_norm": 10.749055862426758, "learning_rate": 5.715048543689321e-05, "loss": 0.1958, "step": 8827 }, { "epoch": 85.60120481927711, "grad_norm": 9.254252433776855, "learning_rate": 5.714563106796117e-05, "loss": 0.1978, "step": 8828 }, { "epoch": 85.61084337349398, "grad_norm": 8.053498268127441, "learning_rate": 5.714077669902913e-05, "loss": 0.2932, "step": 8829 }, { "epoch": 85.62048192771084, "grad_norm": 3.985241413116455, "learning_rate": 5.713592233009709e-05, "loss": 0.1856, "step": 8830 }, { "epoch": 85.6301204819277, "grad_norm": 17.23053741455078, "learning_rate": 5.7131067961165056e-05, "loss": 0.297, "step": 8831 }, { "epoch": 85.63975903614458, "grad_norm": 5.040644645690918, "learning_rate": 5.712621359223301e-05, "loss": 0.3776, "step": 8832 }, { "epoch": 85.64939759036145, "grad_norm": 17.49760627746582, "learning_rate": 5.712135922330097e-05, "loss": 0.2966, "step": 8833 }, { "epoch": 85.65903614457831, "grad_norm": 21.654170989990234, "learning_rate": 5.7116504854368936e-05, "loss": 0.2297, "step": 8834 }, { "epoch": 85.66867469879519, "grad_norm": 12.70780086517334, "learning_rate": 5.7111650485436893e-05, "loss": 0.2981, "step": 8835 }, { "epoch": 85.67831325301205, "grad_norm": 3.677690029144287, "learning_rate": 5.710679611650486e-05, "loss": 0.2554, "step": 8836 }, { "epoch": 85.68795180722891, "grad_norm": 25.760690689086914, "learning_rate": 5.7101941747572816e-05, "loss": 0.1126, "step": 8837 }, { "epoch": 85.69759036144578, "grad_norm": 20.582937240600586, "learning_rate": 5.709708737864078e-05, "loss": 0.1833, "step": 8838 }, { "epoch": 85.70722891566265, "grad_norm": 10.493192672729492, "learning_rate": 5.709223300970874e-05, "loss": 0.4385, "step": 8839 }, { "epoch": 85.71686746987952, "grad_norm": 15.691058158874512, "learning_rate": 5.70873786407767e-05, "loss": 0.198, "step": 8840 }, { "epoch": 85.72650602409638, "grad_norm": 6.491084098815918, "learning_rate": 5.708252427184466e-05, "loss": 0.3759, "step": 8841 }, { "epoch": 85.73614457831326, "grad_norm": 10.782658576965332, "learning_rate": 5.7077669902912625e-05, "loss": 0.2598, "step": 8842 }, { "epoch": 85.74578313253012, "grad_norm": 8.33524227142334, "learning_rate": 5.707281553398058e-05, "loss": 0.3503, "step": 8843 }, { "epoch": 85.75542168674698, "grad_norm": 12.760367393493652, "learning_rate": 5.706796116504855e-05, "loss": 0.3913, "step": 8844 }, { "epoch": 85.76506024096386, "grad_norm": 11.498739242553711, "learning_rate": 5.7063106796116505e-05, "loss": 0.3659, "step": 8845 }, { "epoch": 85.77469879518073, "grad_norm": 13.8725004196167, "learning_rate": 5.705825242718447e-05, "loss": 0.3041, "step": 8846 }, { "epoch": 85.78433734939759, "grad_norm": 3.9413955211639404, "learning_rate": 5.705339805825243e-05, "loss": 0.1887, "step": 8847 }, { "epoch": 85.79397590361445, "grad_norm": 16.29010772705078, "learning_rate": 5.704854368932039e-05, "loss": 0.4249, "step": 8848 }, { "epoch": 85.80361445783133, "grad_norm": 10.970616340637207, "learning_rate": 5.704368932038835e-05, "loss": 0.3865, "step": 8849 }, { "epoch": 85.8132530120482, "grad_norm": 14.549543380737305, "learning_rate": 5.703883495145631e-05, "loss": 0.4044, "step": 8850 }, { "epoch": 85.82289156626506, "grad_norm": 6.7625412940979, "learning_rate": 5.703398058252427e-05, "loss": 0.1898, "step": 8851 }, { "epoch": 85.83253012048193, "grad_norm": 10.909255027770996, "learning_rate": 5.702912621359223e-05, "loss": 0.2915, "step": 8852 }, { "epoch": 85.8421686746988, "grad_norm": 40.093624114990234, "learning_rate": 5.7024271844660194e-05, "loss": 0.1775, "step": 8853 }, { "epoch": 85.85180722891566, "grad_norm": 31.336381912231445, "learning_rate": 5.701941747572815e-05, "loss": 0.385, "step": 8854 }, { "epoch": 85.86144578313252, "grad_norm": 6.090850353240967, "learning_rate": 5.7014563106796116e-05, "loss": 0.1862, "step": 8855 }, { "epoch": 85.8710843373494, "grad_norm": 9.518695831298828, "learning_rate": 5.7009708737864074e-05, "loss": 0.3002, "step": 8856 }, { "epoch": 85.88072289156626, "grad_norm": 17.747554779052734, "learning_rate": 5.700485436893204e-05, "loss": 0.2109, "step": 8857 }, { "epoch": 85.89036144578313, "grad_norm": 10.909135818481445, "learning_rate": 5.6999999999999996e-05, "loss": 0.3099, "step": 8858 }, { "epoch": 85.9, "grad_norm": 31.833600997924805, "learning_rate": 5.699514563106796e-05, "loss": 0.222, "step": 8859 }, { "epoch": 85.90963855421687, "grad_norm": 8.267577171325684, "learning_rate": 5.699029126213592e-05, "loss": 0.2785, "step": 8860 }, { "epoch": 85.91927710843373, "grad_norm": 8.713460922241211, "learning_rate": 5.698543689320388e-05, "loss": 0.1284, "step": 8861 }, { "epoch": 85.9289156626506, "grad_norm": 50.02933120727539, "learning_rate": 5.6980582524271854e-05, "loss": 0.4616, "step": 8862 }, { "epoch": 85.93855421686747, "grad_norm": 6.280852317810059, "learning_rate": 5.697572815533981e-05, "loss": 0.2187, "step": 8863 }, { "epoch": 85.94819277108434, "grad_norm": 5.650640964508057, "learning_rate": 5.697087378640778e-05, "loss": 0.3, "step": 8864 }, { "epoch": 85.9578313253012, "grad_norm": 17.489580154418945, "learning_rate": 5.6966019417475735e-05, "loss": 0.4805, "step": 8865 }, { "epoch": 85.96746987951808, "grad_norm": 6.173410892486572, "learning_rate": 5.69611650485437e-05, "loss": 0.1912, "step": 8866 }, { "epoch": 85.97710843373494, "grad_norm": 24.18425750732422, "learning_rate": 5.695631067961166e-05, "loss": 0.3733, "step": 8867 }, { "epoch": 85.9867469879518, "grad_norm": 6.993476867675781, "learning_rate": 5.695145631067962e-05, "loss": 0.1624, "step": 8868 }, { "epoch": 86.00240963855421, "grad_norm": 9.838799476623535, "learning_rate": 5.694660194174758e-05, "loss": 0.1875, "step": 8869 }, { "epoch": 86.01204819277109, "grad_norm": 8.816224098205566, "learning_rate": 5.6941747572815544e-05, "loss": 0.5193, "step": 8870 }, { "epoch": 86.02168674698795, "grad_norm": 11.8206205368042, "learning_rate": 5.69368932038835e-05, "loss": 0.261, "step": 8871 }, { "epoch": 86.03132530120482, "grad_norm": 9.85368537902832, "learning_rate": 5.693203883495146e-05, "loss": 0.2851, "step": 8872 }, { "epoch": 86.04096385542168, "grad_norm": 8.223931312561035, "learning_rate": 5.6927184466019424e-05, "loss": 0.2352, "step": 8873 }, { "epoch": 86.05060240963856, "grad_norm": 7.356831073760986, "learning_rate": 5.692233009708738e-05, "loss": 0.1669, "step": 8874 }, { "epoch": 86.06024096385542, "grad_norm": 12.395709991455078, "learning_rate": 5.6917475728155346e-05, "loss": 0.3308, "step": 8875 }, { "epoch": 86.06987951807228, "grad_norm": 10.619363784790039, "learning_rate": 5.6912621359223304e-05, "loss": 0.3448, "step": 8876 }, { "epoch": 86.07951807228916, "grad_norm": 21.57472038269043, "learning_rate": 5.690776699029127e-05, "loss": 0.2426, "step": 8877 }, { "epoch": 86.08915662650602, "grad_norm": 9.422627449035645, "learning_rate": 5.6902912621359226e-05, "loss": 0.2642, "step": 8878 }, { "epoch": 86.09879518072289, "grad_norm": 8.33051586151123, "learning_rate": 5.689805825242719e-05, "loss": 0.3217, "step": 8879 }, { "epoch": 86.10843373493977, "grad_norm": 8.561267852783203, "learning_rate": 5.689320388349515e-05, "loss": 0.3441, "step": 8880 }, { "epoch": 86.11807228915663, "grad_norm": 8.390742301940918, "learning_rate": 5.688834951456311e-05, "loss": 0.4637, "step": 8881 }, { "epoch": 86.12771084337349, "grad_norm": 30.403474807739258, "learning_rate": 5.688349514563107e-05, "loss": 0.6272, "step": 8882 }, { "epoch": 86.13734939759036, "grad_norm": 5.285812854766846, "learning_rate": 5.6878640776699035e-05, "loss": 0.1771, "step": 8883 }, { "epoch": 86.14698795180723, "grad_norm": 3.968430995941162, "learning_rate": 5.687378640776699e-05, "loss": 0.2034, "step": 8884 }, { "epoch": 86.1566265060241, "grad_norm": 6.721783638000488, "learning_rate": 5.686893203883496e-05, "loss": 0.3034, "step": 8885 }, { "epoch": 86.16626506024096, "grad_norm": 2.5529065132141113, "learning_rate": 5.6864077669902915e-05, "loss": 0.1058, "step": 8886 }, { "epoch": 86.17590361445784, "grad_norm": 24.842987060546875, "learning_rate": 5.685922330097087e-05, "loss": 0.4227, "step": 8887 }, { "epoch": 86.1855421686747, "grad_norm": 4.789358139038086, "learning_rate": 5.685436893203884e-05, "loss": 0.2138, "step": 8888 }, { "epoch": 86.19518072289156, "grad_norm": 10.616525650024414, "learning_rate": 5.6849514563106795e-05, "loss": 0.2969, "step": 8889 }, { "epoch": 86.20481927710843, "grad_norm": 6.664907455444336, "learning_rate": 5.684466019417476e-05, "loss": 0.2468, "step": 8890 }, { "epoch": 86.2144578313253, "grad_norm": 2.578150987625122, "learning_rate": 5.683980582524272e-05, "loss": 0.1711, "step": 8891 }, { "epoch": 86.22409638554217, "grad_norm": 36.83837890625, "learning_rate": 5.683495145631068e-05, "loss": 0.5125, "step": 8892 }, { "epoch": 86.23373493975903, "grad_norm": 8.092252731323242, "learning_rate": 5.683009708737864e-05, "loss": 0.2958, "step": 8893 }, { "epoch": 86.24337349397591, "grad_norm": 13.665780067443848, "learning_rate": 5.6825242718446604e-05, "loss": 0.2374, "step": 8894 }, { "epoch": 86.25301204819277, "grad_norm": 16.062654495239258, "learning_rate": 5.682038834951456e-05, "loss": 0.2144, "step": 8895 }, { "epoch": 86.26265060240964, "grad_norm": 22.68145751953125, "learning_rate": 5.681553398058253e-05, "loss": 0.294, "step": 8896 }, { "epoch": 86.2722891566265, "grad_norm": 11.492441177368164, "learning_rate": 5.6810679611650484e-05, "loss": 0.3429, "step": 8897 }, { "epoch": 86.28192771084338, "grad_norm": 15.74514389038086, "learning_rate": 5.680582524271845e-05, "loss": 0.4211, "step": 8898 }, { "epoch": 86.29156626506024, "grad_norm": 7.77284574508667, "learning_rate": 5.680097087378641e-05, "loss": 0.3303, "step": 8899 }, { "epoch": 86.3012048192771, "grad_norm": 6.005896091461182, "learning_rate": 5.679611650485437e-05, "loss": 0.3294, "step": 8900 }, { "epoch": 86.31084337349398, "grad_norm": 11.074057579040527, "learning_rate": 5.679126213592233e-05, "loss": 0.3793, "step": 8901 }, { "epoch": 86.32048192771084, "grad_norm": 7.191914081573486, "learning_rate": 5.6786407766990294e-05, "loss": 0.2903, "step": 8902 }, { "epoch": 86.33012048192771, "grad_norm": 5.603761196136475, "learning_rate": 5.678155339805825e-05, "loss": 0.1494, "step": 8903 }, { "epoch": 86.33975903614459, "grad_norm": 23.510086059570312, "learning_rate": 5.677669902912621e-05, "loss": 0.3535, "step": 8904 }, { "epoch": 86.34939759036145, "grad_norm": 72.58349609375, "learning_rate": 5.6771844660194174e-05, "loss": 0.2238, "step": 8905 }, { "epoch": 86.35903614457831, "grad_norm": 8.893733978271484, "learning_rate": 5.676699029126213e-05, "loss": 0.3195, "step": 8906 }, { "epoch": 86.36867469879518, "grad_norm": 9.376956939697266, "learning_rate": 5.6762135922330096e-05, "loss": 0.2204, "step": 8907 }, { "epoch": 86.37831325301205, "grad_norm": 3.262458086013794, "learning_rate": 5.6757281553398054e-05, "loss": 0.1034, "step": 8908 }, { "epoch": 86.38795180722892, "grad_norm": 5.231998443603516, "learning_rate": 5.675242718446602e-05, "loss": 0.1896, "step": 8909 }, { "epoch": 86.39759036144578, "grad_norm": 6.928901195526123, "learning_rate": 5.6747572815533976e-05, "loss": 0.2858, "step": 8910 }, { "epoch": 86.40722891566266, "grad_norm": 7.560854911804199, "learning_rate": 5.674271844660194e-05, "loss": 0.2197, "step": 8911 }, { "epoch": 86.41686746987952, "grad_norm": 20.07170867919922, "learning_rate": 5.673786407766991e-05, "loss": 0.4653, "step": 8912 }, { "epoch": 86.42650602409638, "grad_norm": 35.557525634765625, "learning_rate": 5.673300970873787e-05, "loss": 0.3291, "step": 8913 }, { "epoch": 86.43614457831325, "grad_norm": 12.061205863952637, "learning_rate": 5.6728155339805834e-05, "loss": 0.4126, "step": 8914 }, { "epoch": 86.44578313253012, "grad_norm": 23.09503746032715, "learning_rate": 5.672330097087379e-05, "loss": 0.6257, "step": 8915 }, { "epoch": 86.45542168674699, "grad_norm": 5.826838493347168, "learning_rate": 5.6718446601941756e-05, "loss": 0.1748, "step": 8916 }, { "epoch": 86.46506024096385, "grad_norm": 42.94160079956055, "learning_rate": 5.6713592233009714e-05, "loss": 0.466, "step": 8917 }, { "epoch": 86.47469879518073, "grad_norm": 6.401993751525879, "learning_rate": 5.670873786407768e-05, "loss": 0.2058, "step": 8918 }, { "epoch": 86.48433734939759, "grad_norm": 10.838991165161133, "learning_rate": 5.6703883495145636e-05, "loss": 0.2501, "step": 8919 }, { "epoch": 86.49397590361446, "grad_norm": 9.00100040435791, "learning_rate": 5.66990291262136e-05, "loss": 0.3366, "step": 8920 }, { "epoch": 86.50361445783132, "grad_norm": 7.740957260131836, "learning_rate": 5.669417475728156e-05, "loss": 0.3191, "step": 8921 }, { "epoch": 86.5132530120482, "grad_norm": 8.815149307250977, "learning_rate": 5.668932038834952e-05, "loss": 0.264, "step": 8922 }, { "epoch": 86.52289156626506, "grad_norm": 7.811392784118652, "learning_rate": 5.668446601941748e-05, "loss": 0.2017, "step": 8923 }, { "epoch": 86.53253012048192, "grad_norm": 19.430696487426758, "learning_rate": 5.6679611650485446e-05, "loss": 0.4571, "step": 8924 }, { "epoch": 86.5421686746988, "grad_norm": 11.000138282775879, "learning_rate": 5.66747572815534e-05, "loss": 0.3888, "step": 8925 }, { "epoch": 86.55180722891566, "grad_norm": 8.579374313354492, "learning_rate": 5.666990291262136e-05, "loss": 0.3013, "step": 8926 }, { "epoch": 86.56144578313253, "grad_norm": 7.561336994171143, "learning_rate": 5.6665048543689326e-05, "loss": 0.278, "step": 8927 }, { "epoch": 86.5710843373494, "grad_norm": 11.81482982635498, "learning_rate": 5.666019417475728e-05, "loss": 0.2204, "step": 8928 }, { "epoch": 86.58072289156627, "grad_norm": 11.143519401550293, "learning_rate": 5.665533980582525e-05, "loss": 0.1654, "step": 8929 }, { "epoch": 86.59036144578313, "grad_norm": 10.092242240905762, "learning_rate": 5.6650485436893206e-05, "loss": 0.2894, "step": 8930 }, { "epoch": 86.6, "grad_norm": 7.96025276184082, "learning_rate": 5.664563106796117e-05, "loss": 0.2, "step": 8931 }, { "epoch": 86.60963855421687, "grad_norm": 9.386873245239258, "learning_rate": 5.664077669902913e-05, "loss": 0.5018, "step": 8932 }, { "epoch": 86.61927710843374, "grad_norm": 6.62628173828125, "learning_rate": 5.663592233009709e-05, "loss": 0.2629, "step": 8933 }, { "epoch": 86.6289156626506, "grad_norm": 5.0029096603393555, "learning_rate": 5.663106796116505e-05, "loss": 0.1454, "step": 8934 }, { "epoch": 86.63855421686748, "grad_norm": 8.46626091003418, "learning_rate": 5.6626213592233015e-05, "loss": 0.4391, "step": 8935 }, { "epoch": 86.64819277108434, "grad_norm": 8.831945419311523, "learning_rate": 5.662135922330097e-05, "loss": 0.2779, "step": 8936 }, { "epoch": 86.6578313253012, "grad_norm": 6.897026062011719, "learning_rate": 5.661650485436894e-05, "loss": 0.1849, "step": 8937 }, { "epoch": 86.66746987951807, "grad_norm": 27.38191795349121, "learning_rate": 5.6611650485436895e-05, "loss": 0.4417, "step": 8938 }, { "epoch": 86.67710843373494, "grad_norm": 7.984088897705078, "learning_rate": 5.660679611650486e-05, "loss": 0.2049, "step": 8939 }, { "epoch": 86.6867469879518, "grad_norm": 7.452298164367676, "learning_rate": 5.660194174757282e-05, "loss": 0.2613, "step": 8940 }, { "epoch": 86.69638554216867, "grad_norm": 16.540645599365234, "learning_rate": 5.659708737864078e-05, "loss": 0.4196, "step": 8941 }, { "epoch": 86.70602409638555, "grad_norm": 3.9692203998565674, "learning_rate": 5.659223300970874e-05, "loss": 0.1482, "step": 8942 }, { "epoch": 86.71566265060241, "grad_norm": 9.548306465148926, "learning_rate": 5.65873786407767e-05, "loss": 0.1336, "step": 8943 }, { "epoch": 86.72530120481927, "grad_norm": 13.875988960266113, "learning_rate": 5.658252427184466e-05, "loss": 0.3928, "step": 8944 }, { "epoch": 86.73493975903614, "grad_norm": 10.39173412322998, "learning_rate": 5.657766990291262e-05, "loss": 0.4687, "step": 8945 }, { "epoch": 86.74457831325302, "grad_norm": 8.587199211120605, "learning_rate": 5.6572815533980584e-05, "loss": 0.3452, "step": 8946 }, { "epoch": 86.75421686746988, "grad_norm": 24.733823776245117, "learning_rate": 5.656796116504854e-05, "loss": 0.1959, "step": 8947 }, { "epoch": 86.76385542168674, "grad_norm": 20.403953552246094, "learning_rate": 5.6563106796116506e-05, "loss": 0.3012, "step": 8948 }, { "epoch": 86.77349397590362, "grad_norm": 4.336374282836914, "learning_rate": 5.6558252427184464e-05, "loss": 0.112, "step": 8949 }, { "epoch": 86.78313253012048, "grad_norm": 9.510095596313477, "learning_rate": 5.655339805825243e-05, "loss": 0.2394, "step": 8950 }, { "epoch": 86.79277108433735, "grad_norm": 9.408182144165039, "learning_rate": 5.6548543689320386e-05, "loss": 0.3252, "step": 8951 }, { "epoch": 86.80240963855422, "grad_norm": 11.675753593444824, "learning_rate": 5.654368932038835e-05, "loss": 0.1717, "step": 8952 }, { "epoch": 86.81204819277109, "grad_norm": 4.672750949859619, "learning_rate": 5.653883495145631e-05, "loss": 0.198, "step": 8953 }, { "epoch": 86.82168674698795, "grad_norm": 9.397954940795898, "learning_rate": 5.653398058252427e-05, "loss": 0.378, "step": 8954 }, { "epoch": 86.83132530120481, "grad_norm": 9.878802299499512, "learning_rate": 5.652912621359223e-05, "loss": 0.266, "step": 8955 }, { "epoch": 86.84096385542169, "grad_norm": 9.181926727294922, "learning_rate": 5.6524271844660195e-05, "loss": 0.2764, "step": 8956 }, { "epoch": 86.85060240963855, "grad_norm": 10.620288848876953, "learning_rate": 5.651941747572815e-05, "loss": 0.4211, "step": 8957 }, { "epoch": 86.86024096385542, "grad_norm": 5.762754917144775, "learning_rate": 5.651456310679612e-05, "loss": 0.2827, "step": 8958 }, { "epoch": 86.8698795180723, "grad_norm": 8.662117958068848, "learning_rate": 5.6509708737864075e-05, "loss": 0.2454, "step": 8959 }, { "epoch": 86.87951807228916, "grad_norm": 20.553102493286133, "learning_rate": 5.650485436893203e-05, "loss": 0.2917, "step": 8960 }, { "epoch": 86.88915662650602, "grad_norm": 14.960836410522461, "learning_rate": 5.65e-05, "loss": 0.3088, "step": 8961 }, { "epoch": 86.89879518072289, "grad_norm": 5.073113441467285, "learning_rate": 5.6495145631067956e-05, "loss": 0.2094, "step": 8962 }, { "epoch": 86.90843373493976, "grad_norm": 4.9337544441223145, "learning_rate": 5.6490291262135934e-05, "loss": 0.2298, "step": 8963 }, { "epoch": 86.91807228915663, "grad_norm": 8.42979621887207, "learning_rate": 5.648543689320389e-05, "loss": 0.4085, "step": 8964 }, { "epoch": 86.92771084337349, "grad_norm": 13.8262300491333, "learning_rate": 5.648058252427185e-05, "loss": 0.35, "step": 8965 }, { "epoch": 86.93734939759037, "grad_norm": 24.588653564453125, "learning_rate": 5.6475728155339814e-05, "loss": 0.291, "step": 8966 }, { "epoch": 86.94698795180723, "grad_norm": 9.277695655822754, "learning_rate": 5.647087378640777e-05, "loss": 0.2736, "step": 8967 }, { "epoch": 86.9566265060241, "grad_norm": 9.778702735900879, "learning_rate": 5.6466019417475736e-05, "loss": 0.2298, "step": 8968 }, { "epoch": 86.96626506024096, "grad_norm": 4.502128601074219, "learning_rate": 5.6461165048543694e-05, "loss": 0.1448, "step": 8969 }, { "epoch": 86.97590361445783, "grad_norm": 20.956687927246094, "learning_rate": 5.645631067961166e-05, "loss": 0.4353, "step": 8970 }, { "epoch": 86.9855421686747, "grad_norm": 10.879047393798828, "learning_rate": 5.6451456310679616e-05, "loss": 0.3105, "step": 8971 }, { "epoch": 87.0012048192771, "grad_norm": 11.217528343200684, "learning_rate": 5.644660194174758e-05, "loss": 0.3807, "step": 8972 }, { "epoch": 87.01084337349397, "grad_norm": 8.5156888961792, "learning_rate": 5.644174757281554e-05, "loss": 0.1693, "step": 8973 }, { "epoch": 87.02048192771085, "grad_norm": 16.63778305053711, "learning_rate": 5.64368932038835e-05, "loss": 0.5578, "step": 8974 }, { "epoch": 87.03012048192771, "grad_norm": 10.783409118652344, "learning_rate": 5.643203883495146e-05, "loss": 0.2511, "step": 8975 }, { "epoch": 87.03975903614457, "grad_norm": 11.961524963378906, "learning_rate": 5.6427184466019425e-05, "loss": 0.3148, "step": 8976 }, { "epoch": 87.04939759036145, "grad_norm": 12.677860260009766, "learning_rate": 5.642233009708738e-05, "loss": 0.5046, "step": 8977 }, { "epoch": 87.05903614457831, "grad_norm": 3.6009042263031006, "learning_rate": 5.641747572815535e-05, "loss": 0.1156, "step": 8978 }, { "epoch": 87.06867469879518, "grad_norm": 9.243934631347656, "learning_rate": 5.6412621359223305e-05, "loss": 0.1702, "step": 8979 }, { "epoch": 87.07831325301204, "grad_norm": 15.634761810302734, "learning_rate": 5.640776699029127e-05, "loss": 0.3476, "step": 8980 }, { "epoch": 87.08795180722892, "grad_norm": 8.059539794921875, "learning_rate": 5.640291262135923e-05, "loss": 0.3684, "step": 8981 }, { "epoch": 87.09759036144578, "grad_norm": 10.133172988891602, "learning_rate": 5.6398058252427185e-05, "loss": 0.2118, "step": 8982 }, { "epoch": 87.10722891566265, "grad_norm": 13.272324562072754, "learning_rate": 5.639320388349515e-05, "loss": 0.4907, "step": 8983 }, { "epoch": 87.11686746987952, "grad_norm": 8.108715057373047, "learning_rate": 5.638834951456311e-05, "loss": 0.2888, "step": 8984 }, { "epoch": 87.12650602409639, "grad_norm": 17.521896362304688, "learning_rate": 5.638349514563107e-05, "loss": 0.4234, "step": 8985 }, { "epoch": 87.13614457831325, "grad_norm": 9.203536033630371, "learning_rate": 5.637864077669903e-05, "loss": 0.1303, "step": 8986 }, { "epoch": 87.14578313253013, "grad_norm": 33.57626724243164, "learning_rate": 5.6373786407766994e-05, "loss": 0.3682, "step": 8987 }, { "epoch": 87.15542168674699, "grad_norm": 10.33704948425293, "learning_rate": 5.636893203883495e-05, "loss": 0.1505, "step": 8988 }, { "epoch": 87.16506024096385, "grad_norm": 5.964989185333252, "learning_rate": 5.636407766990292e-05, "loss": 0.2706, "step": 8989 }, { "epoch": 87.17469879518072, "grad_norm": 13.286552429199219, "learning_rate": 5.6359223300970874e-05, "loss": 0.3216, "step": 8990 }, { "epoch": 87.1843373493976, "grad_norm": 10.035965919494629, "learning_rate": 5.635436893203884e-05, "loss": 0.3187, "step": 8991 }, { "epoch": 87.19397590361446, "grad_norm": 8.322434425354004, "learning_rate": 5.63495145631068e-05, "loss": 0.1825, "step": 8992 }, { "epoch": 87.20361445783132, "grad_norm": 5.670292854309082, "learning_rate": 5.634466019417476e-05, "loss": 0.1912, "step": 8993 }, { "epoch": 87.2132530120482, "grad_norm": 12.9063081741333, "learning_rate": 5.633980582524272e-05, "loss": 0.234, "step": 8994 }, { "epoch": 87.22289156626506, "grad_norm": 9.531556129455566, "learning_rate": 5.6334951456310683e-05, "loss": 0.1915, "step": 8995 }, { "epoch": 87.23253012048193, "grad_norm": 7.880007743835449, "learning_rate": 5.633009708737864e-05, "loss": 0.2498, "step": 8996 }, { "epoch": 87.24216867469879, "grad_norm": 7.808654308319092, "learning_rate": 5.6325242718446606e-05, "loss": 0.362, "step": 8997 }, { "epoch": 87.25180722891567, "grad_norm": 5.462442874908447, "learning_rate": 5.6320388349514564e-05, "loss": 0.2211, "step": 8998 }, { "epoch": 87.26144578313253, "grad_norm": 16.62613868713379, "learning_rate": 5.631553398058252e-05, "loss": 0.5213, "step": 8999 }, { "epoch": 87.2710843373494, "grad_norm": 8.69662857055664, "learning_rate": 5.6310679611650486e-05, "loss": 0.2418, "step": 9000 }, { "epoch": 87.28072289156627, "grad_norm": 7.159566402435303, "learning_rate": 5.6305825242718444e-05, "loss": 0.3368, "step": 9001 }, { "epoch": 87.29036144578313, "grad_norm": 11.496176719665527, "learning_rate": 5.630097087378641e-05, "loss": 0.5063, "step": 9002 }, { "epoch": 87.3, "grad_norm": 17.020967483520508, "learning_rate": 5.6296116504854366e-05, "loss": 0.3632, "step": 9003 }, { "epoch": 87.30963855421686, "grad_norm": 15.306502342224121, "learning_rate": 5.629126213592233e-05, "loss": 0.2997, "step": 9004 }, { "epoch": 87.31927710843374, "grad_norm": 13.591218948364258, "learning_rate": 5.628640776699029e-05, "loss": 0.1934, "step": 9005 }, { "epoch": 87.3289156626506, "grad_norm": 12.004480361938477, "learning_rate": 5.628155339805825e-05, "loss": 0.3362, "step": 9006 }, { "epoch": 87.33855421686746, "grad_norm": 14.04089069366455, "learning_rate": 5.627669902912621e-05, "loss": 0.4021, "step": 9007 }, { "epoch": 87.34819277108434, "grad_norm": 6.937991619110107, "learning_rate": 5.6271844660194175e-05, "loss": 0.3263, "step": 9008 }, { "epoch": 87.3578313253012, "grad_norm": 14.77341079711914, "learning_rate": 5.626699029126213e-05, "loss": 0.5558, "step": 9009 }, { "epoch": 87.36746987951807, "grad_norm": 10.4163818359375, "learning_rate": 5.62621359223301e-05, "loss": 0.1958, "step": 9010 }, { "epoch": 87.37710843373495, "grad_norm": 11.606749534606934, "learning_rate": 5.6257281553398055e-05, "loss": 0.2649, "step": 9011 }, { "epoch": 87.38674698795181, "grad_norm": 15.921867370605469, "learning_rate": 5.625242718446602e-05, "loss": 0.4518, "step": 9012 }, { "epoch": 87.39638554216867, "grad_norm": 4.94780158996582, "learning_rate": 5.624757281553399e-05, "loss": 0.2368, "step": 9013 }, { "epoch": 87.40602409638554, "grad_norm": 15.631983757019043, "learning_rate": 5.624271844660195e-05, "loss": 0.5048, "step": 9014 }, { "epoch": 87.41566265060241, "grad_norm": 7.3806562423706055, "learning_rate": 5.623786407766991e-05, "loss": 0.2837, "step": 9015 }, { "epoch": 87.42530120481928, "grad_norm": 3.9566421508789062, "learning_rate": 5.623300970873787e-05, "loss": 0.097, "step": 9016 }, { "epoch": 87.43493975903614, "grad_norm": 11.0753173828125, "learning_rate": 5.6228155339805835e-05, "loss": 0.2817, "step": 9017 }, { "epoch": 87.44457831325302, "grad_norm": 17.515546798706055, "learning_rate": 5.622330097087379e-05, "loss": 0.3943, "step": 9018 }, { "epoch": 87.45421686746988, "grad_norm": 7.007698059082031, "learning_rate": 5.621844660194176e-05, "loss": 0.3515, "step": 9019 }, { "epoch": 87.46385542168674, "grad_norm": 4.701643466949463, "learning_rate": 5.6213592233009716e-05, "loss": 0.0904, "step": 9020 }, { "epoch": 87.47349397590361, "grad_norm": 8.722362518310547, "learning_rate": 5.620873786407767e-05, "loss": 0.3082, "step": 9021 }, { "epoch": 87.48313253012049, "grad_norm": 9.78195571899414, "learning_rate": 5.620388349514564e-05, "loss": 0.4277, "step": 9022 }, { "epoch": 87.49277108433735, "grad_norm": 13.005264282226562, "learning_rate": 5.6199029126213596e-05, "loss": 0.4475, "step": 9023 }, { "epoch": 87.50240963855421, "grad_norm": 4.642416477203369, "learning_rate": 5.619417475728156e-05, "loss": 0.1859, "step": 9024 }, { "epoch": 87.51204819277109, "grad_norm": 9.78342342376709, "learning_rate": 5.618932038834952e-05, "loss": 0.2386, "step": 9025 }, { "epoch": 87.52168674698795, "grad_norm": 7.940223693847656, "learning_rate": 5.618446601941748e-05, "loss": 0.1819, "step": 9026 }, { "epoch": 87.53132530120482, "grad_norm": 10.010266304016113, "learning_rate": 5.617961165048544e-05, "loss": 0.2763, "step": 9027 }, { "epoch": 87.54096385542168, "grad_norm": 6.991367340087891, "learning_rate": 5.6174757281553405e-05, "loss": 0.2037, "step": 9028 }, { "epoch": 87.55060240963856, "grad_norm": 19.966758728027344, "learning_rate": 5.616990291262136e-05, "loss": 0.3171, "step": 9029 }, { "epoch": 87.56024096385542, "grad_norm": 12.01840591430664, "learning_rate": 5.616504854368933e-05, "loss": 0.2196, "step": 9030 }, { "epoch": 87.56987951807228, "grad_norm": 8.460615158081055, "learning_rate": 5.6160194174757285e-05, "loss": 0.1911, "step": 9031 }, { "epoch": 87.57951807228916, "grad_norm": 5.956125736236572, "learning_rate": 5.615533980582525e-05, "loss": 0.2779, "step": 9032 }, { "epoch": 87.58915662650602, "grad_norm": 15.068341255187988, "learning_rate": 5.615048543689321e-05, "loss": 0.1994, "step": 9033 }, { "epoch": 87.59879518072289, "grad_norm": 9.092719078063965, "learning_rate": 5.614563106796117e-05, "loss": 0.3923, "step": 9034 }, { "epoch": 87.60843373493977, "grad_norm": 4.005145072937012, "learning_rate": 5.614077669902913e-05, "loss": 0.1988, "step": 9035 }, { "epoch": 87.61807228915663, "grad_norm": 9.966401100158691, "learning_rate": 5.613592233009709e-05, "loss": 0.3467, "step": 9036 }, { "epoch": 87.62771084337349, "grad_norm": 27.065752029418945, "learning_rate": 5.613106796116505e-05, "loss": 0.1397, "step": 9037 }, { "epoch": 87.63734939759036, "grad_norm": 3.3038887977600098, "learning_rate": 5.612621359223301e-05, "loss": 0.0697, "step": 9038 }, { "epoch": 87.64698795180723, "grad_norm": 6.62601900100708, "learning_rate": 5.6121359223300974e-05, "loss": 0.1588, "step": 9039 }, { "epoch": 87.6566265060241, "grad_norm": 6.766843318939209, "learning_rate": 5.611650485436893e-05, "loss": 0.3825, "step": 9040 }, { "epoch": 87.66626506024096, "grad_norm": 8.571300506591797, "learning_rate": 5.6111650485436896e-05, "loss": 0.3887, "step": 9041 }, { "epoch": 87.67590361445784, "grad_norm": 5.385858535766602, "learning_rate": 5.6106796116504854e-05, "loss": 0.2848, "step": 9042 }, { "epoch": 87.6855421686747, "grad_norm": 13.538431167602539, "learning_rate": 5.610194174757282e-05, "loss": 0.4061, "step": 9043 }, { "epoch": 87.69518072289156, "grad_norm": 7.983536720275879, "learning_rate": 5.6097087378640776e-05, "loss": 0.311, "step": 9044 }, { "epoch": 87.70481927710843, "grad_norm": 22.028175354003906, "learning_rate": 5.609223300970874e-05, "loss": 0.3648, "step": 9045 }, { "epoch": 87.7144578313253, "grad_norm": 11.68390941619873, "learning_rate": 5.60873786407767e-05, "loss": 0.1717, "step": 9046 }, { "epoch": 87.72409638554217, "grad_norm": 5.991830348968506, "learning_rate": 5.608252427184466e-05, "loss": 0.3913, "step": 9047 }, { "epoch": 87.73373493975903, "grad_norm": 13.510051727294922, "learning_rate": 5.607766990291262e-05, "loss": 0.3713, "step": 9048 }, { "epoch": 87.74337349397591, "grad_norm": 8.841025352478027, "learning_rate": 5.6072815533980585e-05, "loss": 0.2809, "step": 9049 }, { "epoch": 87.75301204819277, "grad_norm": 22.61967658996582, "learning_rate": 5.606796116504854e-05, "loss": 0.2388, "step": 9050 }, { "epoch": 87.76265060240964, "grad_norm": 10.714054107666016, "learning_rate": 5.606310679611651e-05, "loss": 0.4726, "step": 9051 }, { "epoch": 87.7722891566265, "grad_norm": 7.344820499420166, "learning_rate": 5.6058252427184465e-05, "loss": 0.3347, "step": 9052 }, { "epoch": 87.78192771084338, "grad_norm": 45.51824951171875, "learning_rate": 5.605339805825242e-05, "loss": 0.2805, "step": 9053 }, { "epoch": 87.79156626506024, "grad_norm": 7.622892379760742, "learning_rate": 5.604854368932039e-05, "loss": 0.2489, "step": 9054 }, { "epoch": 87.8012048192771, "grad_norm": 15.490950584411621, "learning_rate": 5.6043689320388345e-05, "loss": 0.1879, "step": 9055 }, { "epoch": 87.81084337349398, "grad_norm": 8.289854049682617, "learning_rate": 5.603883495145631e-05, "loss": 0.2929, "step": 9056 }, { "epoch": 87.82048192771084, "grad_norm": 10.869369506835938, "learning_rate": 5.603398058252427e-05, "loss": 0.3981, "step": 9057 }, { "epoch": 87.83012048192771, "grad_norm": 6.871895790100098, "learning_rate": 5.602912621359223e-05, "loss": 0.3021, "step": 9058 }, { "epoch": 87.83975903614459, "grad_norm": 6.646849632263184, "learning_rate": 5.602427184466019e-05, "loss": 0.228, "step": 9059 }, { "epoch": 87.84939759036145, "grad_norm": 8.92814826965332, "learning_rate": 5.6019417475728155e-05, "loss": 0.1606, "step": 9060 }, { "epoch": 87.85903614457831, "grad_norm": 14.77862548828125, "learning_rate": 5.601456310679611e-05, "loss": 0.4285, "step": 9061 }, { "epoch": 87.86867469879518, "grad_norm": 7.330878734588623, "learning_rate": 5.600970873786408e-05, "loss": 0.2488, "step": 9062 }, { "epoch": 87.87831325301205, "grad_norm": 18.321029663085938, "learning_rate": 5.600485436893205e-05, "loss": 0.3184, "step": 9063 }, { "epoch": 87.88795180722892, "grad_norm": 7.359135627746582, "learning_rate": 5.6000000000000006e-05, "loss": 0.3552, "step": 9064 }, { "epoch": 87.89759036144578, "grad_norm": 8.963680267333984, "learning_rate": 5.599514563106797e-05, "loss": 0.3374, "step": 9065 }, { "epoch": 87.90722891566266, "grad_norm": 5.707785129547119, "learning_rate": 5.599029126213593e-05, "loss": 0.4228, "step": 9066 }, { "epoch": 87.91686746987952, "grad_norm": 11.126642227172852, "learning_rate": 5.598543689320389e-05, "loss": 0.2692, "step": 9067 }, { "epoch": 87.92650602409638, "grad_norm": 5.677391529083252, "learning_rate": 5.598058252427185e-05, "loss": 0.348, "step": 9068 }, { "epoch": 87.93614457831325, "grad_norm": 6.526860237121582, "learning_rate": 5.5975728155339815e-05, "loss": 0.328, "step": 9069 }, { "epoch": 87.94578313253012, "grad_norm": 5.4429426193237305, "learning_rate": 5.597087378640777e-05, "loss": 0.2944, "step": 9070 }, { "epoch": 87.95542168674699, "grad_norm": 14.254103660583496, "learning_rate": 5.596601941747574e-05, "loss": 0.2017, "step": 9071 }, { "epoch": 87.96506024096385, "grad_norm": 3.0142436027526855, "learning_rate": 5.5961165048543695e-05, "loss": 0.1706, "step": 9072 }, { "epoch": 87.97469879518073, "grad_norm": 12.234663963317871, "learning_rate": 5.595631067961166e-05, "loss": 0.1868, "step": 9073 }, { "epoch": 87.98433734939759, "grad_norm": 2.6656579971313477, "learning_rate": 5.595145631067962e-05, "loss": 0.1194, "step": 9074 }, { "epoch": 87.99397590361446, "grad_norm": 7.88320255279541, "learning_rate": 5.5946601941747575e-05, "loss": 0.2307, "step": 9075 }, { "epoch": 88.00963855421686, "grad_norm": 4.305882930755615, "learning_rate": 5.594174757281554e-05, "loss": 0.1686, "step": 9076 }, { "epoch": 88.01927710843374, "grad_norm": 10.267489433288574, "learning_rate": 5.59368932038835e-05, "loss": 0.329, "step": 9077 }, { "epoch": 88.0289156626506, "grad_norm": 9.092705726623535, "learning_rate": 5.593203883495146e-05, "loss": 0.1982, "step": 9078 }, { "epoch": 88.03855421686747, "grad_norm": 11.882242202758789, "learning_rate": 5.592718446601942e-05, "loss": 0.5326, "step": 9079 }, { "epoch": 88.04819277108433, "grad_norm": 8.156070709228516, "learning_rate": 5.5922330097087384e-05, "loss": 0.3911, "step": 9080 }, { "epoch": 88.05783132530121, "grad_norm": 8.931602478027344, "learning_rate": 5.591747572815534e-05, "loss": 0.3193, "step": 9081 }, { "epoch": 88.06746987951807, "grad_norm": 9.485040664672852, "learning_rate": 5.5912621359223307e-05, "loss": 0.4393, "step": 9082 }, { "epoch": 88.07710843373494, "grad_norm": 8.915430068969727, "learning_rate": 5.5907766990291264e-05, "loss": 0.3307, "step": 9083 }, { "epoch": 88.08674698795181, "grad_norm": 9.647208213806152, "learning_rate": 5.590291262135923e-05, "loss": 0.2402, "step": 9084 }, { "epoch": 88.09638554216868, "grad_norm": 7.302329063415527, "learning_rate": 5.5898058252427187e-05, "loss": 0.3904, "step": 9085 }, { "epoch": 88.10602409638554, "grad_norm": 10.747101783752441, "learning_rate": 5.589320388349515e-05, "loss": 0.4317, "step": 9086 }, { "epoch": 88.1156626506024, "grad_norm": 15.384137153625488, "learning_rate": 5.588834951456311e-05, "loss": 0.5882, "step": 9087 }, { "epoch": 88.12530120481928, "grad_norm": 15.589798927307129, "learning_rate": 5.5883495145631073e-05, "loss": 0.4095, "step": 9088 }, { "epoch": 88.13493975903614, "grad_norm": 7.805981159210205, "learning_rate": 5.587864077669903e-05, "loss": 0.2494, "step": 9089 }, { "epoch": 88.144578313253, "grad_norm": 8.629945755004883, "learning_rate": 5.5873786407766996e-05, "loss": 0.2486, "step": 9090 }, { "epoch": 88.15421686746988, "grad_norm": 11.741599082946777, "learning_rate": 5.5868932038834953e-05, "loss": 0.2536, "step": 9091 }, { "epoch": 88.16385542168675, "grad_norm": 7.010340213775635, "learning_rate": 5.586407766990291e-05, "loss": 0.4085, "step": 9092 }, { "epoch": 88.17349397590361, "grad_norm": 22.544618606567383, "learning_rate": 5.5859223300970876e-05, "loss": 0.2733, "step": 9093 }, { "epoch": 88.18313253012049, "grad_norm": 5.483741283416748, "learning_rate": 5.5854368932038834e-05, "loss": 0.1625, "step": 9094 }, { "epoch": 88.19277108433735, "grad_norm": 4.445727348327637, "learning_rate": 5.58495145631068e-05, "loss": 0.177, "step": 9095 }, { "epoch": 88.20240963855422, "grad_norm": 10.875865936279297, "learning_rate": 5.5844660194174756e-05, "loss": 0.4386, "step": 9096 }, { "epoch": 88.21204819277108, "grad_norm": 7.621675491333008, "learning_rate": 5.583980582524272e-05, "loss": 0.1678, "step": 9097 }, { "epoch": 88.22168674698796, "grad_norm": 17.031293869018555, "learning_rate": 5.583495145631068e-05, "loss": 0.3959, "step": 9098 }, { "epoch": 88.23132530120482, "grad_norm": 11.214262008666992, "learning_rate": 5.583009708737864e-05, "loss": 0.2606, "step": 9099 }, { "epoch": 88.24096385542168, "grad_norm": 7.826404094696045, "learning_rate": 5.58252427184466e-05, "loss": 0.482, "step": 9100 }, { "epoch": 88.25060240963856, "grad_norm": 13.977128982543945, "learning_rate": 5.5820388349514565e-05, "loss": 0.4036, "step": 9101 }, { "epoch": 88.26024096385542, "grad_norm": 5.486871719360352, "learning_rate": 5.581553398058252e-05, "loss": 0.1608, "step": 9102 }, { "epoch": 88.26987951807229, "grad_norm": 13.623351097106934, "learning_rate": 5.581067961165049e-05, "loss": 0.3558, "step": 9103 }, { "epoch": 88.27951807228915, "grad_norm": 5.967000961303711, "learning_rate": 5.5805825242718445e-05, "loss": 0.3111, "step": 9104 }, { "epoch": 88.28915662650603, "grad_norm": 10.527069091796875, "learning_rate": 5.580097087378641e-05, "loss": 0.2946, "step": 9105 }, { "epoch": 88.29879518072289, "grad_norm": 7.772378921508789, "learning_rate": 5.579611650485437e-05, "loss": 0.293, "step": 9106 }, { "epoch": 88.30843373493975, "grad_norm": 18.081960678100586, "learning_rate": 5.579126213592233e-05, "loss": 0.3891, "step": 9107 }, { "epoch": 88.31807228915663, "grad_norm": 6.304901599884033, "learning_rate": 5.578640776699029e-05, "loss": 0.2162, "step": 9108 }, { "epoch": 88.3277108433735, "grad_norm": 10.014248847961426, "learning_rate": 5.578155339805825e-05, "loss": 0.392, "step": 9109 }, { "epoch": 88.33734939759036, "grad_norm": 5.602302074432373, "learning_rate": 5.577669902912621e-05, "loss": 0.24, "step": 9110 }, { "epoch": 88.34698795180722, "grad_norm": 16.704166412353516, "learning_rate": 5.577184466019417e-05, "loss": 0.449, "step": 9111 }, { "epoch": 88.3566265060241, "grad_norm": 6.641969203948975, "learning_rate": 5.5766990291262134e-05, "loss": 0.3002, "step": 9112 }, { "epoch": 88.36626506024096, "grad_norm": 11.747577667236328, "learning_rate": 5.576213592233009e-05, "loss": 0.405, "step": 9113 }, { "epoch": 88.37590361445783, "grad_norm": 8.310233116149902, "learning_rate": 5.575728155339806e-05, "loss": 0.2936, "step": 9114 }, { "epoch": 88.3855421686747, "grad_norm": 3.70597767829895, "learning_rate": 5.575242718446603e-05, "loss": 0.1563, "step": 9115 }, { "epoch": 88.39518072289157, "grad_norm": 3.968287229537964, "learning_rate": 5.5747572815533986e-05, "loss": 0.1142, "step": 9116 }, { "epoch": 88.40481927710843, "grad_norm": 8.710672378540039, "learning_rate": 5.574271844660195e-05, "loss": 0.3243, "step": 9117 }, { "epoch": 88.41445783132531, "grad_norm": 6.086416721343994, "learning_rate": 5.573786407766991e-05, "loss": 0.2791, "step": 9118 }, { "epoch": 88.42409638554217, "grad_norm": 15.332172393798828, "learning_rate": 5.573300970873787e-05, "loss": 0.4897, "step": 9119 }, { "epoch": 88.43373493975903, "grad_norm": 3.844327211380005, "learning_rate": 5.572815533980583e-05, "loss": 0.0868, "step": 9120 }, { "epoch": 88.4433734939759, "grad_norm": 9.876908302307129, "learning_rate": 5.5723300970873795e-05, "loss": 0.3923, "step": 9121 }, { "epoch": 88.45301204819278, "grad_norm": 8.427072525024414, "learning_rate": 5.571844660194175e-05, "loss": 0.3378, "step": 9122 }, { "epoch": 88.46265060240964, "grad_norm": 10.853409767150879, "learning_rate": 5.571359223300972e-05, "loss": 0.2308, "step": 9123 }, { "epoch": 88.4722891566265, "grad_norm": 4.398587703704834, "learning_rate": 5.5708737864077675e-05, "loss": 0.1875, "step": 9124 }, { "epoch": 88.48192771084338, "grad_norm": 8.558703422546387, "learning_rate": 5.570388349514564e-05, "loss": 0.298, "step": 9125 }, { "epoch": 88.49156626506024, "grad_norm": 11.181502342224121, "learning_rate": 5.56990291262136e-05, "loss": 0.4062, "step": 9126 }, { "epoch": 88.5012048192771, "grad_norm": 11.422187805175781, "learning_rate": 5.569417475728156e-05, "loss": 0.3674, "step": 9127 }, { "epoch": 88.51084337349397, "grad_norm": 12.167655944824219, "learning_rate": 5.568932038834952e-05, "loss": 0.4823, "step": 9128 }, { "epoch": 88.52048192771085, "grad_norm": 7.177402496337891, "learning_rate": 5.5684466019417484e-05, "loss": 0.1693, "step": 9129 }, { "epoch": 88.53012048192771, "grad_norm": 9.788946151733398, "learning_rate": 5.567961165048544e-05, "loss": 0.25, "step": 9130 }, { "epoch": 88.53975903614457, "grad_norm": 6.855284690856934, "learning_rate": 5.56747572815534e-05, "loss": 0.2097, "step": 9131 }, { "epoch": 88.54939759036145, "grad_norm": 7.339930057525635, "learning_rate": 5.5669902912621364e-05, "loss": 0.2035, "step": 9132 }, { "epoch": 88.55903614457831, "grad_norm": 10.382100105285645, "learning_rate": 5.566504854368932e-05, "loss": 0.2521, "step": 9133 }, { "epoch": 88.56867469879518, "grad_norm": 6.923975944519043, "learning_rate": 5.5660194174757286e-05, "loss": 0.2605, "step": 9134 }, { "epoch": 88.57831325301204, "grad_norm": 4.716518402099609, "learning_rate": 5.5655339805825244e-05, "loss": 0.1201, "step": 9135 }, { "epoch": 88.58795180722892, "grad_norm": 14.17169189453125, "learning_rate": 5.565048543689321e-05, "loss": 0.4801, "step": 9136 }, { "epoch": 88.59759036144578, "grad_norm": 9.435696601867676, "learning_rate": 5.5645631067961166e-05, "loss": 0.3093, "step": 9137 }, { "epoch": 88.60722891566265, "grad_norm": 8.541812896728516, "learning_rate": 5.564077669902913e-05, "loss": 0.2453, "step": 9138 }, { "epoch": 88.61686746987952, "grad_norm": 5.308385848999023, "learning_rate": 5.563592233009709e-05, "loss": 0.2049, "step": 9139 }, { "epoch": 88.62650602409639, "grad_norm": 8.856133460998535, "learning_rate": 5.563106796116505e-05, "loss": 0.4436, "step": 9140 }, { "epoch": 88.63614457831325, "grad_norm": 10.60135269165039, "learning_rate": 5.562621359223301e-05, "loss": 0.3243, "step": 9141 }, { "epoch": 88.64578313253013, "grad_norm": 8.191329956054688, "learning_rate": 5.5621359223300975e-05, "loss": 0.2902, "step": 9142 }, { "epoch": 88.65542168674699, "grad_norm": 4.827840805053711, "learning_rate": 5.561650485436893e-05, "loss": 0.2804, "step": 9143 }, { "epoch": 88.66506024096385, "grad_norm": 6.895505428314209, "learning_rate": 5.56116504854369e-05, "loss": 0.2767, "step": 9144 }, { "epoch": 88.67469879518072, "grad_norm": 5.437589645385742, "learning_rate": 5.5606796116504855e-05, "loss": 0.2825, "step": 9145 }, { "epoch": 88.6843373493976, "grad_norm": 8.280637741088867, "learning_rate": 5.560194174757282e-05, "loss": 0.2871, "step": 9146 }, { "epoch": 88.69397590361446, "grad_norm": 6.427376747131348, "learning_rate": 5.559708737864078e-05, "loss": 0.2903, "step": 9147 }, { "epoch": 88.70361445783132, "grad_norm": 10.083983421325684, "learning_rate": 5.5592233009708735e-05, "loss": 0.343, "step": 9148 }, { "epoch": 88.7132530120482, "grad_norm": 11.715704917907715, "learning_rate": 5.55873786407767e-05, "loss": 0.2959, "step": 9149 }, { "epoch": 88.72289156626506, "grad_norm": 10.239500999450684, "learning_rate": 5.558252427184466e-05, "loss": 0.3772, "step": 9150 }, { "epoch": 88.73253012048193, "grad_norm": 16.232872009277344, "learning_rate": 5.557766990291262e-05, "loss": 0.3228, "step": 9151 }, { "epoch": 88.74216867469879, "grad_norm": 11.959712028503418, "learning_rate": 5.557281553398058e-05, "loss": 0.3458, "step": 9152 }, { "epoch": 88.75180722891567, "grad_norm": 4.599846363067627, "learning_rate": 5.5567961165048545e-05, "loss": 0.1691, "step": 9153 }, { "epoch": 88.76144578313253, "grad_norm": 9.803441047668457, "learning_rate": 5.55631067961165e-05, "loss": 0.1882, "step": 9154 }, { "epoch": 88.7710843373494, "grad_norm": 8.4241361618042, "learning_rate": 5.555825242718447e-05, "loss": 0.265, "step": 9155 }, { "epoch": 88.78072289156627, "grad_norm": 15.610578536987305, "learning_rate": 5.5553398058252425e-05, "loss": 0.504, "step": 9156 }, { "epoch": 88.79036144578313, "grad_norm": 5.463128566741943, "learning_rate": 5.554854368932039e-05, "loss": 0.2556, "step": 9157 }, { "epoch": 88.8, "grad_norm": 3.874396562576294, "learning_rate": 5.554368932038835e-05, "loss": 0.2098, "step": 9158 }, { "epoch": 88.80963855421686, "grad_norm": 10.706232070922852, "learning_rate": 5.553883495145631e-05, "loss": 0.1734, "step": 9159 }, { "epoch": 88.81927710843374, "grad_norm": 6.934715747833252, "learning_rate": 5.553398058252427e-05, "loss": 0.1985, "step": 9160 }, { "epoch": 88.8289156626506, "grad_norm": 8.509519577026367, "learning_rate": 5.5529126213592234e-05, "loss": 0.2201, "step": 9161 }, { "epoch": 88.83855421686746, "grad_norm": 6.1080780029296875, "learning_rate": 5.552427184466019e-05, "loss": 0.2088, "step": 9162 }, { "epoch": 88.84819277108434, "grad_norm": 6.6867523193359375, "learning_rate": 5.5519417475728156e-05, "loss": 0.2895, "step": 9163 }, { "epoch": 88.8578313253012, "grad_norm": 11.707693099975586, "learning_rate": 5.551456310679613e-05, "loss": 0.3185, "step": 9164 }, { "epoch": 88.86746987951807, "grad_norm": 10.016250610351562, "learning_rate": 5.5509708737864085e-05, "loss": 0.342, "step": 9165 }, { "epoch": 88.87710843373495, "grad_norm": 16.20920181274414, "learning_rate": 5.550485436893205e-05, "loss": 0.2414, "step": 9166 }, { "epoch": 88.88674698795181, "grad_norm": 8.800980567932129, "learning_rate": 5.550000000000001e-05, "loss": 0.2188, "step": 9167 }, { "epoch": 88.89638554216867, "grad_norm": 10.093024253845215, "learning_rate": 5.549514563106797e-05, "loss": 0.3434, "step": 9168 }, { "epoch": 88.90602409638554, "grad_norm": 8.378242492675781, "learning_rate": 5.549029126213593e-05, "loss": 0.3457, "step": 9169 }, { "epoch": 88.91566265060241, "grad_norm": 4.869582653045654, "learning_rate": 5.548543689320389e-05, "loss": 0.1112, "step": 9170 }, { "epoch": 88.92530120481928, "grad_norm": 15.296452522277832, "learning_rate": 5.548058252427185e-05, "loss": 0.4698, "step": 9171 }, { "epoch": 88.93493975903614, "grad_norm": 6.4454874992370605, "learning_rate": 5.547572815533981e-05, "loss": 0.3008, "step": 9172 }, { "epoch": 88.94457831325302, "grad_norm": 7.967060089111328, "learning_rate": 5.5470873786407774e-05, "loss": 0.3289, "step": 9173 }, { "epoch": 88.95421686746988, "grad_norm": 8.26943302154541, "learning_rate": 5.546601941747573e-05, "loss": 0.2541, "step": 9174 }, { "epoch": 88.96385542168674, "grad_norm": 13.985998153686523, "learning_rate": 5.5461165048543697e-05, "loss": 0.3357, "step": 9175 }, { "epoch": 88.97349397590361, "grad_norm": 9.711231231689453, "learning_rate": 5.5456310679611654e-05, "loss": 0.394, "step": 9176 }, { "epoch": 88.98313253012049, "grad_norm": 2.9285619258880615, "learning_rate": 5.545145631067962e-05, "loss": 0.0948, "step": 9177 }, { "epoch": 88.99277108433735, "grad_norm": 11.7910737991333, "learning_rate": 5.5446601941747577e-05, "loss": 0.5206, "step": 9178 }, { "epoch": 89.00843373493976, "grad_norm": 2.6727681159973145, "learning_rate": 5.544174757281554e-05, "loss": 0.1531, "step": 9179 }, { "epoch": 89.01807228915662, "grad_norm": 3.556565523147583, "learning_rate": 5.54368932038835e-05, "loss": 0.1675, "step": 9180 }, { "epoch": 89.0277108433735, "grad_norm": 4.40043830871582, "learning_rate": 5.543203883495146e-05, "loss": 0.2108, "step": 9181 }, { "epoch": 89.03734939759036, "grad_norm": 11.984009742736816, "learning_rate": 5.542718446601942e-05, "loss": 0.2131, "step": 9182 }, { "epoch": 89.04698795180722, "grad_norm": 26.674543380737305, "learning_rate": 5.5422330097087386e-05, "loss": 0.1547, "step": 9183 }, { "epoch": 89.0566265060241, "grad_norm": 23.28561019897461, "learning_rate": 5.5417475728155343e-05, "loss": 0.289, "step": 9184 }, { "epoch": 89.06626506024097, "grad_norm": 7.125781536102295, "learning_rate": 5.54126213592233e-05, "loss": 0.2759, "step": 9185 }, { "epoch": 89.07590361445783, "grad_norm": 16.141063690185547, "learning_rate": 5.5407766990291266e-05, "loss": 0.2701, "step": 9186 }, { "epoch": 89.08554216867469, "grad_norm": 25.389741897583008, "learning_rate": 5.5402912621359223e-05, "loss": 0.3837, "step": 9187 }, { "epoch": 89.09518072289157, "grad_norm": 31.117456436157227, "learning_rate": 5.539805825242719e-05, "loss": 0.315, "step": 9188 }, { "epoch": 89.10481927710843, "grad_norm": 8.684453010559082, "learning_rate": 5.5393203883495146e-05, "loss": 0.43, "step": 9189 }, { "epoch": 89.1144578313253, "grad_norm": 9.43671989440918, "learning_rate": 5.538834951456311e-05, "loss": 0.335, "step": 9190 }, { "epoch": 89.12409638554217, "grad_norm": 10.791229248046875, "learning_rate": 5.538349514563107e-05, "loss": 0.4964, "step": 9191 }, { "epoch": 89.13373493975904, "grad_norm": 58.11721420288086, "learning_rate": 5.537864077669903e-05, "loss": 0.2745, "step": 9192 }, { "epoch": 89.1433734939759, "grad_norm": 21.419906616210938, "learning_rate": 5.537378640776699e-05, "loss": 0.4697, "step": 9193 }, { "epoch": 89.15301204819278, "grad_norm": 10.149008750915527, "learning_rate": 5.5368932038834955e-05, "loss": 0.3722, "step": 9194 }, { "epoch": 89.16265060240964, "grad_norm": 39.381561279296875, "learning_rate": 5.536407766990291e-05, "loss": 0.4342, "step": 9195 }, { "epoch": 89.1722891566265, "grad_norm": 19.426002502441406, "learning_rate": 5.535922330097088e-05, "loss": 0.3463, "step": 9196 }, { "epoch": 89.18192771084337, "grad_norm": 24.22134017944336, "learning_rate": 5.5354368932038835e-05, "loss": 0.4178, "step": 9197 }, { "epoch": 89.19156626506025, "grad_norm": 3.108182668685913, "learning_rate": 5.53495145631068e-05, "loss": 0.2679, "step": 9198 }, { "epoch": 89.20120481927711, "grad_norm": 16.352529525756836, "learning_rate": 5.534466019417476e-05, "loss": 0.3095, "step": 9199 }, { "epoch": 89.21084337349397, "grad_norm": 17.400251388549805, "learning_rate": 5.533980582524272e-05, "loss": 0.3314, "step": 9200 }, { "epoch": 89.22048192771085, "grad_norm": 19.47843360900879, "learning_rate": 5.533495145631068e-05, "loss": 0.3341, "step": 9201 }, { "epoch": 89.23012048192771, "grad_norm": 14.624835014343262, "learning_rate": 5.533009708737864e-05, "loss": 0.4561, "step": 9202 }, { "epoch": 89.23975903614458, "grad_norm": 14.577364921569824, "learning_rate": 5.53252427184466e-05, "loss": 0.338, "step": 9203 }, { "epoch": 89.24939759036144, "grad_norm": 19.684114456176758, "learning_rate": 5.532038834951456e-05, "loss": 0.4954, "step": 9204 }, { "epoch": 89.25903614457832, "grad_norm": 10.208205223083496, "learning_rate": 5.5315533980582524e-05, "loss": 0.1372, "step": 9205 }, { "epoch": 89.26867469879518, "grad_norm": 36.38361740112305, "learning_rate": 5.531067961165048e-05, "loss": 0.3253, "step": 9206 }, { "epoch": 89.27831325301204, "grad_norm": 12.473462104797363, "learning_rate": 5.5305825242718446e-05, "loss": 0.2487, "step": 9207 }, { "epoch": 89.28795180722892, "grad_norm": 23.200225830078125, "learning_rate": 5.5300970873786404e-05, "loss": 0.2607, "step": 9208 }, { "epoch": 89.29759036144578, "grad_norm": 19.2752628326416, "learning_rate": 5.529611650485437e-05, "loss": 0.4299, "step": 9209 }, { "epoch": 89.30722891566265, "grad_norm": 18.523698806762695, "learning_rate": 5.5291262135922326e-05, "loss": 0.2502, "step": 9210 }, { "epoch": 89.31686746987951, "grad_norm": 4.1682353019714355, "learning_rate": 5.528640776699029e-05, "loss": 0.1647, "step": 9211 }, { "epoch": 89.32650602409639, "grad_norm": 46.27553176879883, "learning_rate": 5.528155339805825e-05, "loss": 0.3153, "step": 9212 }, { "epoch": 89.33614457831325, "grad_norm": 137.21690368652344, "learning_rate": 5.527669902912621e-05, "loss": 0.4037, "step": 9213 }, { "epoch": 89.34578313253012, "grad_norm": 10.109803199768066, "learning_rate": 5.527184466019417e-05, "loss": 0.2695, "step": 9214 }, { "epoch": 89.355421686747, "grad_norm": 1.5965018272399902, "learning_rate": 5.526699029126214e-05, "loss": 0.1597, "step": 9215 }, { "epoch": 89.36506024096386, "grad_norm": 17.774118423461914, "learning_rate": 5.526213592233011e-05, "loss": 0.481, "step": 9216 }, { "epoch": 89.37469879518072, "grad_norm": 14.467717170715332, "learning_rate": 5.5257281553398065e-05, "loss": 0.3286, "step": 9217 }, { "epoch": 89.38433734939758, "grad_norm": 27.220598220825195, "learning_rate": 5.525242718446603e-05, "loss": 0.296, "step": 9218 }, { "epoch": 89.39397590361446, "grad_norm": 5.385734558105469, "learning_rate": 5.524757281553399e-05, "loss": 0.1531, "step": 9219 }, { "epoch": 89.40361445783132, "grad_norm": 16.612346649169922, "learning_rate": 5.524271844660195e-05, "loss": 0.351, "step": 9220 }, { "epoch": 89.41325301204819, "grad_norm": 15.635683059692383, "learning_rate": 5.523786407766991e-05, "loss": 0.3706, "step": 9221 }, { "epoch": 89.42289156626506, "grad_norm": 6.792923450469971, "learning_rate": 5.5233009708737874e-05, "loss": 0.3833, "step": 9222 }, { "epoch": 89.43253012048193, "grad_norm": 8.083266258239746, "learning_rate": 5.522815533980583e-05, "loss": 0.1888, "step": 9223 }, { "epoch": 89.44216867469879, "grad_norm": 21.23678970336914, "learning_rate": 5.522330097087379e-05, "loss": 0.4172, "step": 9224 }, { "epoch": 89.45180722891567, "grad_norm": 5.126220703125, "learning_rate": 5.5218446601941754e-05, "loss": 0.3394, "step": 9225 }, { "epoch": 89.46144578313253, "grad_norm": 18.408708572387695, "learning_rate": 5.521359223300971e-05, "loss": 0.1737, "step": 9226 }, { "epoch": 89.4710843373494, "grad_norm": 24.332988739013672, "learning_rate": 5.5208737864077676e-05, "loss": 0.2205, "step": 9227 }, { "epoch": 89.48072289156626, "grad_norm": 7.68312931060791, "learning_rate": 5.5203883495145634e-05, "loss": 0.2137, "step": 9228 }, { "epoch": 89.49036144578314, "grad_norm": 22.472272872924805, "learning_rate": 5.51990291262136e-05, "loss": 0.3526, "step": 9229 }, { "epoch": 89.5, "grad_norm": 10.570134162902832, "learning_rate": 5.5194174757281556e-05, "loss": 0.3514, "step": 9230 }, { "epoch": 89.50963855421686, "grad_norm": 37.89182662963867, "learning_rate": 5.518932038834952e-05, "loss": 0.1562, "step": 9231 }, { "epoch": 89.51927710843374, "grad_norm": 12.257719993591309, "learning_rate": 5.518446601941748e-05, "loss": 0.2624, "step": 9232 }, { "epoch": 89.5289156626506, "grad_norm": 3.453744649887085, "learning_rate": 5.517961165048544e-05, "loss": 0.2664, "step": 9233 }, { "epoch": 89.53855421686747, "grad_norm": 12.354257583618164, "learning_rate": 5.51747572815534e-05, "loss": 0.3654, "step": 9234 }, { "epoch": 89.54819277108433, "grad_norm": 12.869074821472168, "learning_rate": 5.5169902912621365e-05, "loss": 0.2554, "step": 9235 }, { "epoch": 89.55783132530121, "grad_norm": 25.722126007080078, "learning_rate": 5.516504854368932e-05, "loss": 0.2377, "step": 9236 }, { "epoch": 89.56746987951807, "grad_norm": 28.629684448242188, "learning_rate": 5.516019417475729e-05, "loss": 0.2885, "step": 9237 }, { "epoch": 89.57710843373494, "grad_norm": 33.085792541503906, "learning_rate": 5.5155339805825245e-05, "loss": 0.2132, "step": 9238 }, { "epoch": 89.58674698795181, "grad_norm": 15.459122657775879, "learning_rate": 5.515048543689321e-05, "loss": 0.2562, "step": 9239 }, { "epoch": 89.59638554216868, "grad_norm": 23.81194305419922, "learning_rate": 5.514563106796117e-05, "loss": 0.3053, "step": 9240 }, { "epoch": 89.60602409638554, "grad_norm": 12.749235153198242, "learning_rate": 5.5140776699029125e-05, "loss": 0.2362, "step": 9241 }, { "epoch": 89.61566265060242, "grad_norm": 11.622879028320312, "learning_rate": 5.513592233009709e-05, "loss": 0.2342, "step": 9242 }, { "epoch": 89.62530120481928, "grad_norm": 36.01693344116211, "learning_rate": 5.513106796116505e-05, "loss": 0.3355, "step": 9243 }, { "epoch": 89.63493975903614, "grad_norm": 5.549153804779053, "learning_rate": 5.512621359223301e-05, "loss": 0.4305, "step": 9244 }, { "epoch": 89.644578313253, "grad_norm": 3.832813262939453, "learning_rate": 5.512135922330097e-05, "loss": 0.0855, "step": 9245 }, { "epoch": 89.65421686746988, "grad_norm": 9.660202026367188, "learning_rate": 5.5116504854368934e-05, "loss": 0.2265, "step": 9246 }, { "epoch": 89.66385542168675, "grad_norm": 3.525500774383545, "learning_rate": 5.511165048543689e-05, "loss": 0.2487, "step": 9247 }, { "epoch": 89.67349397590361, "grad_norm": 35.92377471923828, "learning_rate": 5.510679611650486e-05, "loss": 0.5067, "step": 9248 }, { "epoch": 89.68313253012049, "grad_norm": 5.080034255981445, "learning_rate": 5.5101941747572815e-05, "loss": 0.1918, "step": 9249 }, { "epoch": 89.69277108433735, "grad_norm": 46.67391586303711, "learning_rate": 5.509708737864078e-05, "loss": 0.2654, "step": 9250 }, { "epoch": 89.70240963855422, "grad_norm": 13.155653953552246, "learning_rate": 5.509223300970874e-05, "loss": 0.3457, "step": 9251 }, { "epoch": 89.71204819277108, "grad_norm": 10.483574867248535, "learning_rate": 5.50873786407767e-05, "loss": 0.5168, "step": 9252 }, { "epoch": 89.72168674698796, "grad_norm": 28.09523582458496, "learning_rate": 5.508252427184466e-05, "loss": 0.2434, "step": 9253 }, { "epoch": 89.73132530120482, "grad_norm": 5.768157958984375, "learning_rate": 5.5077669902912624e-05, "loss": 0.3389, "step": 9254 }, { "epoch": 89.74096385542168, "grad_norm": 18.45963478088379, "learning_rate": 5.507281553398058e-05, "loss": 0.386, "step": 9255 }, { "epoch": 89.75060240963856, "grad_norm": 16.15911293029785, "learning_rate": 5.5067961165048546e-05, "loss": 0.213, "step": 9256 }, { "epoch": 89.76024096385542, "grad_norm": 3.06512713432312, "learning_rate": 5.5063106796116504e-05, "loss": 0.367, "step": 9257 }, { "epoch": 89.76987951807229, "grad_norm": 23.603485107421875, "learning_rate": 5.505825242718446e-05, "loss": 0.3432, "step": 9258 }, { "epoch": 89.77951807228915, "grad_norm": 33.82187271118164, "learning_rate": 5.5053398058252426e-05, "loss": 0.4036, "step": 9259 }, { "epoch": 89.78915662650603, "grad_norm": 16.24909782409668, "learning_rate": 5.5048543689320384e-05, "loss": 0.4979, "step": 9260 }, { "epoch": 89.79879518072289, "grad_norm": 12.755800247192383, "learning_rate": 5.504368932038835e-05, "loss": 0.3891, "step": 9261 }, { "epoch": 89.80843373493975, "grad_norm": 51.0694694519043, "learning_rate": 5.5038834951456306e-05, "loss": 0.3092, "step": 9262 }, { "epoch": 89.81807228915663, "grad_norm": 35.95683670043945, "learning_rate": 5.503398058252427e-05, "loss": 0.2976, "step": 9263 }, { "epoch": 89.8277108433735, "grad_norm": 3.6102705001831055, "learning_rate": 5.502912621359223e-05, "loss": 0.1931, "step": 9264 }, { "epoch": 89.83734939759036, "grad_norm": 7.14941930770874, "learning_rate": 5.50242718446602e-05, "loss": 0.3418, "step": 9265 }, { "epoch": 89.84698795180722, "grad_norm": 76.0035171508789, "learning_rate": 5.5019417475728164e-05, "loss": 0.4674, "step": 9266 }, { "epoch": 89.8566265060241, "grad_norm": 9.475584983825684, "learning_rate": 5.501456310679612e-05, "loss": 0.1346, "step": 9267 }, { "epoch": 89.86626506024096, "grad_norm": 24.161640167236328, "learning_rate": 5.5009708737864086e-05, "loss": 0.2818, "step": 9268 }, { "epoch": 89.87590361445783, "grad_norm": 16.88300132751465, "learning_rate": 5.5004854368932044e-05, "loss": 0.2837, "step": 9269 }, { "epoch": 89.8855421686747, "grad_norm": 3.9184439182281494, "learning_rate": 5.500000000000001e-05, "loss": 0.2634, "step": 9270 }, { "epoch": 89.89518072289157, "grad_norm": 10.639509201049805, "learning_rate": 5.4995145631067966e-05, "loss": 0.2762, "step": 9271 }, { "epoch": 89.90481927710843, "grad_norm": 24.46611213684082, "learning_rate": 5.499029126213593e-05, "loss": 0.2229, "step": 9272 }, { "epoch": 89.91445783132531, "grad_norm": 3.7014143466949463, "learning_rate": 5.498543689320389e-05, "loss": 0.3364, "step": 9273 }, { "epoch": 89.92409638554217, "grad_norm": 17.36236000061035, "learning_rate": 5.498058252427185e-05, "loss": 0.3278, "step": 9274 }, { "epoch": 89.93373493975903, "grad_norm": 21.472187042236328, "learning_rate": 5.497572815533981e-05, "loss": 0.3964, "step": 9275 }, { "epoch": 89.9433734939759, "grad_norm": 69.39598846435547, "learning_rate": 5.4970873786407776e-05, "loss": 0.3832, "step": 9276 }, { "epoch": 89.95301204819278, "grad_norm": 17.78062629699707, "learning_rate": 5.496601941747573e-05, "loss": 0.193, "step": 9277 }, { "epoch": 89.96265060240964, "grad_norm": 13.03508186340332, "learning_rate": 5.49611650485437e-05, "loss": 0.2127, "step": 9278 }, { "epoch": 89.9722891566265, "grad_norm": 9.120377540588379, "learning_rate": 5.4956310679611656e-05, "loss": 0.197, "step": 9279 }, { "epoch": 89.98192771084338, "grad_norm": 17.047719955444336, "learning_rate": 5.4951456310679613e-05, "loss": 0.641, "step": 9280 }, { "epoch": 89.99156626506024, "grad_norm": 2.8914194107055664, "learning_rate": 5.494660194174758e-05, "loss": 0.2249, "step": 9281 }, { "epoch": 90.00722891566265, "grad_norm": 15.227267265319824, "learning_rate": 5.4941747572815536e-05, "loss": 0.3182, "step": 9282 }, { "epoch": 90.01686746987951, "grad_norm": 10.610105514526367, "learning_rate": 5.49368932038835e-05, "loss": 0.3168, "step": 9283 }, { "epoch": 90.02650602409639, "grad_norm": 12.662142753601074, "learning_rate": 5.493203883495146e-05, "loss": 0.4769, "step": 9284 }, { "epoch": 90.03614457831326, "grad_norm": 12.409502983093262, "learning_rate": 5.492718446601942e-05, "loss": 0.2267, "step": 9285 }, { "epoch": 90.04578313253012, "grad_norm": 20.459835052490234, "learning_rate": 5.492233009708738e-05, "loss": 0.2922, "step": 9286 }, { "epoch": 90.05542168674698, "grad_norm": 9.88006591796875, "learning_rate": 5.4917475728155345e-05, "loss": 0.3797, "step": 9287 }, { "epoch": 90.06506024096386, "grad_norm": 16.532459259033203, "learning_rate": 5.49126213592233e-05, "loss": 0.4224, "step": 9288 }, { "epoch": 90.07469879518072, "grad_norm": 7.1959404945373535, "learning_rate": 5.490776699029127e-05, "loss": 0.3048, "step": 9289 }, { "epoch": 90.08433734939759, "grad_norm": 79.1798324584961, "learning_rate": 5.4902912621359225e-05, "loss": 0.3535, "step": 9290 }, { "epoch": 90.09397590361446, "grad_norm": 9.053857803344727, "learning_rate": 5.489805825242719e-05, "loss": 0.2694, "step": 9291 }, { "epoch": 90.10361445783133, "grad_norm": 17.079496383666992, "learning_rate": 5.489320388349515e-05, "loss": 0.4449, "step": 9292 }, { "epoch": 90.11325301204819, "grad_norm": 2.9583818912506104, "learning_rate": 5.488834951456311e-05, "loss": 0.2214, "step": 9293 }, { "epoch": 90.12289156626505, "grad_norm": 5.478576183319092, "learning_rate": 5.488349514563107e-05, "loss": 0.3439, "step": 9294 }, { "epoch": 90.13253012048193, "grad_norm": 5.681080341339111, "learning_rate": 5.4878640776699034e-05, "loss": 0.2896, "step": 9295 }, { "epoch": 90.1421686746988, "grad_norm": 13.7312650680542, "learning_rate": 5.487378640776699e-05, "loss": 0.2659, "step": 9296 }, { "epoch": 90.15180722891566, "grad_norm": 5.08053731918335, "learning_rate": 5.486893203883495e-05, "loss": 0.148, "step": 9297 }, { "epoch": 90.16144578313254, "grad_norm": 4.659666538238525, "learning_rate": 5.4864077669902914e-05, "loss": 0.2591, "step": 9298 }, { "epoch": 90.1710843373494, "grad_norm": 7.6842522621154785, "learning_rate": 5.485922330097087e-05, "loss": 0.3349, "step": 9299 }, { "epoch": 90.18072289156626, "grad_norm": 10.471384048461914, "learning_rate": 5.4854368932038836e-05, "loss": 0.1666, "step": 9300 }, { "epoch": 90.19036144578314, "grad_norm": 12.60007095336914, "learning_rate": 5.4849514563106794e-05, "loss": 0.3129, "step": 9301 }, { "epoch": 90.2, "grad_norm": 12.062946319580078, "learning_rate": 5.484466019417476e-05, "loss": 0.1755, "step": 9302 }, { "epoch": 90.20963855421687, "grad_norm": 35.82588195800781, "learning_rate": 5.4839805825242716e-05, "loss": 0.3763, "step": 9303 }, { "epoch": 90.21927710843373, "grad_norm": 11.550466537475586, "learning_rate": 5.483495145631068e-05, "loss": 0.1894, "step": 9304 }, { "epoch": 90.2289156626506, "grad_norm": 15.55325698852539, "learning_rate": 5.483009708737864e-05, "loss": 0.1401, "step": 9305 }, { "epoch": 90.23855421686747, "grad_norm": 17.320253372192383, "learning_rate": 5.48252427184466e-05, "loss": 0.5058, "step": 9306 }, { "epoch": 90.24819277108433, "grad_norm": 64.72832489013672, "learning_rate": 5.482038834951456e-05, "loss": 0.2415, "step": 9307 }, { "epoch": 90.25783132530121, "grad_norm": 11.809393882751465, "learning_rate": 5.4815533980582525e-05, "loss": 0.4398, "step": 9308 }, { "epoch": 90.26746987951807, "grad_norm": 8.660512924194336, "learning_rate": 5.481067961165048e-05, "loss": 0.2617, "step": 9309 }, { "epoch": 90.27710843373494, "grad_norm": 4.923190593719482, "learning_rate": 5.480582524271845e-05, "loss": 0.1748, "step": 9310 }, { "epoch": 90.2867469879518, "grad_norm": 3.8849682807922363, "learning_rate": 5.4800970873786406e-05, "loss": 0.1553, "step": 9311 }, { "epoch": 90.29638554216868, "grad_norm": 21.39128303527832, "learning_rate": 5.479611650485437e-05, "loss": 0.3679, "step": 9312 }, { "epoch": 90.30602409638554, "grad_norm": 17.212940216064453, "learning_rate": 5.479126213592233e-05, "loss": 0.2901, "step": 9313 }, { "epoch": 90.3156626506024, "grad_norm": 6.3224945068359375, "learning_rate": 5.4786407766990286e-05, "loss": 0.1288, "step": 9314 }, { "epoch": 90.32530120481928, "grad_norm": 4.925975799560547, "learning_rate": 5.478155339805825e-05, "loss": 0.1438, "step": 9315 }, { "epoch": 90.33493975903615, "grad_norm": 19.785934448242188, "learning_rate": 5.477669902912622e-05, "loss": 0.2508, "step": 9316 }, { "epoch": 90.34457831325301, "grad_norm": 4.491246223449707, "learning_rate": 5.4771844660194186e-05, "loss": 0.4084, "step": 9317 }, { "epoch": 90.35421686746987, "grad_norm": 21.13955307006836, "learning_rate": 5.4766990291262144e-05, "loss": 0.247, "step": 9318 }, { "epoch": 90.36385542168675, "grad_norm": 7.4765238761901855, "learning_rate": 5.47621359223301e-05, "loss": 0.2856, "step": 9319 }, { "epoch": 90.37349397590361, "grad_norm": 6.050138473510742, "learning_rate": 5.4757281553398066e-05, "loss": 0.3301, "step": 9320 }, { "epoch": 90.38313253012048, "grad_norm": 6.629690170288086, "learning_rate": 5.4752427184466024e-05, "loss": 0.2823, "step": 9321 }, { "epoch": 90.39277108433735, "grad_norm": 18.83538818359375, "learning_rate": 5.474757281553399e-05, "loss": 0.4723, "step": 9322 }, { "epoch": 90.40240963855422, "grad_norm": 6.454766273498535, "learning_rate": 5.4742718446601946e-05, "loss": 0.2582, "step": 9323 }, { "epoch": 90.41204819277108, "grad_norm": 5.302296161651611, "learning_rate": 5.473786407766991e-05, "loss": 0.3503, "step": 9324 }, { "epoch": 90.42168674698796, "grad_norm": 6.567382335662842, "learning_rate": 5.473300970873787e-05, "loss": 0.343, "step": 9325 }, { "epoch": 90.43132530120482, "grad_norm": 4.921490669250488, "learning_rate": 5.472815533980583e-05, "loss": 0.1194, "step": 9326 }, { "epoch": 90.44096385542169, "grad_norm": 19.80912971496582, "learning_rate": 5.472330097087379e-05, "loss": 0.3624, "step": 9327 }, { "epoch": 90.45060240963855, "grad_norm": 34.567604064941406, "learning_rate": 5.4718446601941755e-05, "loss": 0.3392, "step": 9328 }, { "epoch": 90.46024096385543, "grad_norm": 21.36069107055664, "learning_rate": 5.471359223300971e-05, "loss": 0.5349, "step": 9329 }, { "epoch": 90.46987951807229, "grad_norm": 16.155269622802734, "learning_rate": 5.470873786407768e-05, "loss": 0.2415, "step": 9330 }, { "epoch": 90.47951807228915, "grad_norm": 3.298926830291748, "learning_rate": 5.4703883495145635e-05, "loss": 0.3007, "step": 9331 }, { "epoch": 90.48915662650603, "grad_norm": 18.51780891418457, "learning_rate": 5.46990291262136e-05, "loss": 0.217, "step": 9332 }, { "epoch": 90.4987951807229, "grad_norm": 16.153696060180664, "learning_rate": 5.469417475728156e-05, "loss": 0.3338, "step": 9333 }, { "epoch": 90.50843373493976, "grad_norm": 10.779333114624023, "learning_rate": 5.4689320388349515e-05, "loss": 0.5176, "step": 9334 }, { "epoch": 90.51807228915662, "grad_norm": 10.58073902130127, "learning_rate": 5.468446601941748e-05, "loss": 0.3506, "step": 9335 }, { "epoch": 90.5277108433735, "grad_norm": 15.769220352172852, "learning_rate": 5.467961165048544e-05, "loss": 0.4377, "step": 9336 }, { "epoch": 90.53734939759036, "grad_norm": 13.075814247131348, "learning_rate": 5.46747572815534e-05, "loss": 0.2761, "step": 9337 }, { "epoch": 90.54698795180722, "grad_norm": 20.91714096069336, "learning_rate": 5.466990291262136e-05, "loss": 0.2364, "step": 9338 }, { "epoch": 90.5566265060241, "grad_norm": 3.6804637908935547, "learning_rate": 5.4665048543689324e-05, "loss": 0.5186, "step": 9339 }, { "epoch": 90.56626506024097, "grad_norm": 19.68959617614746, "learning_rate": 5.466019417475728e-05, "loss": 0.2393, "step": 9340 }, { "epoch": 90.57590361445783, "grad_norm": 35.55358123779297, "learning_rate": 5.465533980582525e-05, "loss": 0.2068, "step": 9341 }, { "epoch": 90.58554216867469, "grad_norm": 15.949716567993164, "learning_rate": 5.4650485436893204e-05, "loss": 0.3628, "step": 9342 }, { "epoch": 90.59518072289157, "grad_norm": 19.085773468017578, "learning_rate": 5.464563106796117e-05, "loss": 0.3454, "step": 9343 }, { "epoch": 90.60481927710843, "grad_norm": 16.95946502685547, "learning_rate": 5.464077669902913e-05, "loss": 0.2573, "step": 9344 }, { "epoch": 90.6144578313253, "grad_norm": 11.364517211914062, "learning_rate": 5.463592233009709e-05, "loss": 0.2452, "step": 9345 }, { "epoch": 90.62409638554217, "grad_norm": 8.1393461227417, "learning_rate": 5.463106796116505e-05, "loss": 0.2952, "step": 9346 }, { "epoch": 90.63373493975904, "grad_norm": 17.16350555419922, "learning_rate": 5.4626213592233014e-05, "loss": 0.3472, "step": 9347 }, { "epoch": 90.6433734939759, "grad_norm": 19.062376022338867, "learning_rate": 5.462135922330097e-05, "loss": 0.182, "step": 9348 }, { "epoch": 90.65301204819278, "grad_norm": 61.464778900146484, "learning_rate": 5.4616504854368936e-05, "loss": 0.5518, "step": 9349 }, { "epoch": 90.66265060240964, "grad_norm": 9.738019943237305, "learning_rate": 5.4611650485436894e-05, "loss": 0.2716, "step": 9350 }, { "epoch": 90.6722891566265, "grad_norm": 5.245325565338135, "learning_rate": 5.460679611650485e-05, "loss": 0.1628, "step": 9351 }, { "epoch": 90.68192771084337, "grad_norm": 3.1334424018859863, "learning_rate": 5.4601941747572816e-05, "loss": 0.4895, "step": 9352 }, { "epoch": 90.69156626506025, "grad_norm": 7.077677249908447, "learning_rate": 5.4597087378640774e-05, "loss": 0.1784, "step": 9353 }, { "epoch": 90.70120481927711, "grad_norm": 11.028929710388184, "learning_rate": 5.459223300970874e-05, "loss": 0.3013, "step": 9354 }, { "epoch": 90.71084337349397, "grad_norm": 3.5095722675323486, "learning_rate": 5.4587378640776696e-05, "loss": 0.2347, "step": 9355 }, { "epoch": 90.72048192771085, "grad_norm": 4.3478474617004395, "learning_rate": 5.458252427184466e-05, "loss": 0.1335, "step": 9356 }, { "epoch": 90.73012048192771, "grad_norm": 16.42014503479004, "learning_rate": 5.457766990291262e-05, "loss": 0.4262, "step": 9357 }, { "epoch": 90.73975903614458, "grad_norm": 21.50156021118164, "learning_rate": 5.457281553398058e-05, "loss": 0.4726, "step": 9358 }, { "epoch": 90.74939759036144, "grad_norm": 8.175994873046875, "learning_rate": 5.456796116504854e-05, "loss": 0.3801, "step": 9359 }, { "epoch": 90.75903614457832, "grad_norm": 23.727415084838867, "learning_rate": 5.4563106796116505e-05, "loss": 0.2419, "step": 9360 }, { "epoch": 90.76867469879518, "grad_norm": 44.29710006713867, "learning_rate": 5.455825242718446e-05, "loss": 0.1871, "step": 9361 }, { "epoch": 90.77831325301204, "grad_norm": 9.632746696472168, "learning_rate": 5.455339805825243e-05, "loss": 0.312, "step": 9362 }, { "epoch": 90.78795180722892, "grad_norm": 116.99861145019531, "learning_rate": 5.4548543689320385e-05, "loss": 0.5047, "step": 9363 }, { "epoch": 90.79759036144578, "grad_norm": 5.265757083892822, "learning_rate": 5.454368932038835e-05, "loss": 0.2136, "step": 9364 }, { "epoch": 90.80722891566265, "grad_norm": 15.394734382629395, "learning_rate": 5.453883495145631e-05, "loss": 0.2811, "step": 9365 }, { "epoch": 90.81686746987951, "grad_norm": 2.621070384979248, "learning_rate": 5.453398058252428e-05, "loss": 0.2328, "step": 9366 }, { "epoch": 90.82650602409639, "grad_norm": 33.9897575378418, "learning_rate": 5.452912621359224e-05, "loss": 0.2778, "step": 9367 }, { "epoch": 90.83614457831325, "grad_norm": 25.18703269958496, "learning_rate": 5.45242718446602e-05, "loss": 0.2954, "step": 9368 }, { "epoch": 90.84578313253012, "grad_norm": 20.18169403076172, "learning_rate": 5.4519417475728166e-05, "loss": 0.1913, "step": 9369 }, { "epoch": 90.855421686747, "grad_norm": 5.160970211029053, "learning_rate": 5.451456310679612e-05, "loss": 0.2427, "step": 9370 }, { "epoch": 90.86506024096386, "grad_norm": 39.23207092285156, "learning_rate": 5.450970873786409e-05, "loss": 0.4232, "step": 9371 }, { "epoch": 90.87469879518072, "grad_norm": 12.599577903747559, "learning_rate": 5.4504854368932046e-05, "loss": 0.2111, "step": 9372 }, { "epoch": 90.88433734939758, "grad_norm": 11.46425724029541, "learning_rate": 5.45e-05, "loss": 0.3544, "step": 9373 }, { "epoch": 90.89397590361446, "grad_norm": 8.490193367004395, "learning_rate": 5.449514563106797e-05, "loss": 0.1455, "step": 9374 }, { "epoch": 90.90361445783132, "grad_norm": 26.364500045776367, "learning_rate": 5.4490291262135926e-05, "loss": 0.3289, "step": 9375 }, { "epoch": 90.91325301204819, "grad_norm": 34.909061431884766, "learning_rate": 5.448543689320389e-05, "loss": 0.3687, "step": 9376 }, { "epoch": 90.92289156626506, "grad_norm": 5.970771312713623, "learning_rate": 5.448058252427185e-05, "loss": 0.1811, "step": 9377 }, { "epoch": 90.93253012048193, "grad_norm": 41.94123840332031, "learning_rate": 5.447572815533981e-05, "loss": 0.3749, "step": 9378 }, { "epoch": 90.94216867469879, "grad_norm": 5.334604263305664, "learning_rate": 5.447087378640777e-05, "loss": 0.2474, "step": 9379 }, { "epoch": 90.95180722891567, "grad_norm": 11.7218599319458, "learning_rate": 5.4466019417475735e-05, "loss": 0.2713, "step": 9380 }, { "epoch": 90.96144578313253, "grad_norm": 7.5465545654296875, "learning_rate": 5.446116504854369e-05, "loss": 0.2277, "step": 9381 }, { "epoch": 90.9710843373494, "grad_norm": 3.4807136058807373, "learning_rate": 5.445631067961166e-05, "loss": 0.2879, "step": 9382 }, { "epoch": 90.98072289156626, "grad_norm": 1.71554696559906, "learning_rate": 5.4451456310679615e-05, "loss": 0.2433, "step": 9383 }, { "epoch": 90.99036144578314, "grad_norm": 15.632465362548828, "learning_rate": 5.444660194174758e-05, "loss": 0.3329, "step": 9384 }, { "epoch": 91.00602409638554, "grad_norm": 30.47602081298828, "learning_rate": 5.444174757281554e-05, "loss": 0.203, "step": 9385 }, { "epoch": 91.01566265060241, "grad_norm": 6.086615085601807, "learning_rate": 5.44368932038835e-05, "loss": 0.3551, "step": 9386 }, { "epoch": 91.02530120481927, "grad_norm": 24.285493850708008, "learning_rate": 5.443203883495146e-05, "loss": 0.3621, "step": 9387 }, { "epoch": 91.03493975903615, "grad_norm": 16.707157135009766, "learning_rate": 5.4427184466019424e-05, "loss": 0.3832, "step": 9388 }, { "epoch": 91.04457831325301, "grad_norm": 10.355867385864258, "learning_rate": 5.442233009708738e-05, "loss": 0.3766, "step": 9389 }, { "epoch": 91.05421686746988, "grad_norm": 4.383118152618408, "learning_rate": 5.441747572815534e-05, "loss": 0.1921, "step": 9390 }, { "epoch": 91.06385542168675, "grad_norm": 38.66371154785156, "learning_rate": 5.4412621359223304e-05, "loss": 0.3039, "step": 9391 }, { "epoch": 91.07349397590362, "grad_norm": 2.155508518218994, "learning_rate": 5.440776699029126e-05, "loss": 0.181, "step": 9392 }, { "epoch": 91.08313253012048, "grad_norm": 12.38956356048584, "learning_rate": 5.4402912621359226e-05, "loss": 0.2279, "step": 9393 }, { "epoch": 91.09277108433734, "grad_norm": 10.458487510681152, "learning_rate": 5.4398058252427184e-05, "loss": 0.174, "step": 9394 }, { "epoch": 91.10240963855422, "grad_norm": 4.293388366699219, "learning_rate": 5.439320388349515e-05, "loss": 0.3434, "step": 9395 }, { "epoch": 91.11204819277108, "grad_norm": 6.259283065795898, "learning_rate": 5.4388349514563106e-05, "loss": 0.1478, "step": 9396 }, { "epoch": 91.12168674698795, "grad_norm": 9.89138412475586, "learning_rate": 5.438349514563107e-05, "loss": 0.214, "step": 9397 }, { "epoch": 91.13132530120482, "grad_norm": 11.400568008422852, "learning_rate": 5.437864077669903e-05, "loss": 0.4479, "step": 9398 }, { "epoch": 91.14096385542169, "grad_norm": 4.985047340393066, "learning_rate": 5.437378640776699e-05, "loss": 0.3015, "step": 9399 }, { "epoch": 91.15060240963855, "grad_norm": 11.34765625, "learning_rate": 5.436893203883495e-05, "loss": 0.4179, "step": 9400 }, { "epoch": 91.16024096385541, "grad_norm": 13.676046371459961, "learning_rate": 5.4364077669902915e-05, "loss": 0.479, "step": 9401 }, { "epoch": 91.16987951807229, "grad_norm": 7.653842926025391, "learning_rate": 5.435922330097087e-05, "loss": 0.3252, "step": 9402 }, { "epoch": 91.17951807228916, "grad_norm": 7.3762526512146, "learning_rate": 5.435436893203884e-05, "loss": 0.4873, "step": 9403 }, { "epoch": 91.18915662650602, "grad_norm": 7.756486415863037, "learning_rate": 5.4349514563106795e-05, "loss": 0.2148, "step": 9404 }, { "epoch": 91.1987951807229, "grad_norm": 13.870565414428711, "learning_rate": 5.434466019417476e-05, "loss": 0.4141, "step": 9405 }, { "epoch": 91.20843373493976, "grad_norm": 4.510365009307861, "learning_rate": 5.433980582524272e-05, "loss": 0.1151, "step": 9406 }, { "epoch": 91.21807228915662, "grad_norm": 11.79112720489502, "learning_rate": 5.4334951456310676e-05, "loss": 0.153, "step": 9407 }, { "epoch": 91.2277108433735, "grad_norm": 36.11991500854492, "learning_rate": 5.433009708737864e-05, "loss": 0.4622, "step": 9408 }, { "epoch": 91.23734939759036, "grad_norm": 19.79429054260254, "learning_rate": 5.43252427184466e-05, "loss": 0.3788, "step": 9409 }, { "epoch": 91.24698795180723, "grad_norm": 13.168676376342773, "learning_rate": 5.432038834951456e-05, "loss": 0.2405, "step": 9410 }, { "epoch": 91.25662650602409, "grad_norm": 23.945194244384766, "learning_rate": 5.431553398058252e-05, "loss": 0.2202, "step": 9411 }, { "epoch": 91.26626506024097, "grad_norm": 8.055671691894531, "learning_rate": 5.4310679611650485e-05, "loss": 0.2098, "step": 9412 }, { "epoch": 91.27590361445783, "grad_norm": 9.014469146728516, "learning_rate": 5.430582524271844e-05, "loss": 0.1701, "step": 9413 }, { "epoch": 91.2855421686747, "grad_norm": 9.469008445739746, "learning_rate": 5.430097087378641e-05, "loss": 0.269, "step": 9414 }, { "epoch": 91.29518072289157, "grad_norm": 14.031744003295898, "learning_rate": 5.4296116504854365e-05, "loss": 0.3582, "step": 9415 }, { "epoch": 91.30481927710844, "grad_norm": 12.644947052001953, "learning_rate": 5.429126213592233e-05, "loss": 0.326, "step": 9416 }, { "epoch": 91.3144578313253, "grad_norm": 19.499675750732422, "learning_rate": 5.42864077669903e-05, "loss": 0.1437, "step": 9417 }, { "epoch": 91.32409638554216, "grad_norm": 14.708209037780762, "learning_rate": 5.428155339805826e-05, "loss": 0.2096, "step": 9418 }, { "epoch": 91.33373493975904, "grad_norm": 4.272602081298828, "learning_rate": 5.427669902912622e-05, "loss": 0.4523, "step": 9419 }, { "epoch": 91.3433734939759, "grad_norm": 13.369644165039062, "learning_rate": 5.427184466019418e-05, "loss": 0.4088, "step": 9420 }, { "epoch": 91.35301204819277, "grad_norm": 6.569841384887695, "learning_rate": 5.4266990291262145e-05, "loss": 0.2866, "step": 9421 }, { "epoch": 91.36265060240964, "grad_norm": 2.896815538406372, "learning_rate": 5.42621359223301e-05, "loss": 0.1349, "step": 9422 }, { "epoch": 91.37228915662651, "grad_norm": 6.129974365234375, "learning_rate": 5.425728155339807e-05, "loss": 0.1742, "step": 9423 }, { "epoch": 91.38192771084337, "grad_norm": 8.178019523620605, "learning_rate": 5.4252427184466025e-05, "loss": 0.2479, "step": 9424 }, { "epoch": 91.39156626506023, "grad_norm": 53.46937561035156, "learning_rate": 5.424757281553399e-05, "loss": 0.3155, "step": 9425 }, { "epoch": 91.40120481927711, "grad_norm": 28.278141021728516, "learning_rate": 5.424271844660195e-05, "loss": 0.5826, "step": 9426 }, { "epoch": 91.41084337349398, "grad_norm": 16.54629898071289, "learning_rate": 5.423786407766991e-05, "loss": 0.3454, "step": 9427 }, { "epoch": 91.42048192771084, "grad_norm": 33.65668869018555, "learning_rate": 5.423300970873787e-05, "loss": 0.3019, "step": 9428 }, { "epoch": 91.43012048192772, "grad_norm": 33.72221374511719, "learning_rate": 5.422815533980583e-05, "loss": 0.4885, "step": 9429 }, { "epoch": 91.43975903614458, "grad_norm": 9.4906005859375, "learning_rate": 5.422330097087379e-05, "loss": 0.1459, "step": 9430 }, { "epoch": 91.44939759036144, "grad_norm": 20.33003044128418, "learning_rate": 5.421844660194175e-05, "loss": 0.3873, "step": 9431 }, { "epoch": 91.45903614457832, "grad_norm": 10.767390251159668, "learning_rate": 5.4213592233009714e-05, "loss": 0.3156, "step": 9432 }, { "epoch": 91.46867469879518, "grad_norm": 10.68087387084961, "learning_rate": 5.420873786407767e-05, "loss": 0.3064, "step": 9433 }, { "epoch": 91.47831325301205, "grad_norm": 35.038291931152344, "learning_rate": 5.420388349514564e-05, "loss": 0.2372, "step": 9434 }, { "epoch": 91.48795180722891, "grad_norm": 23.803836822509766, "learning_rate": 5.4199029126213594e-05, "loss": 0.2133, "step": 9435 }, { "epoch": 91.49759036144579, "grad_norm": 9.23712158203125, "learning_rate": 5.419417475728156e-05, "loss": 0.3779, "step": 9436 }, { "epoch": 91.50722891566265, "grad_norm": 4.1647138595581055, "learning_rate": 5.418932038834952e-05, "loss": 0.1612, "step": 9437 }, { "epoch": 91.51686746987951, "grad_norm": 6.636584758758545, "learning_rate": 5.418446601941748e-05, "loss": 0.3331, "step": 9438 }, { "epoch": 91.52650602409639, "grad_norm": 45.55376052856445, "learning_rate": 5.417961165048544e-05, "loss": 0.2592, "step": 9439 }, { "epoch": 91.53614457831326, "grad_norm": 8.552164077758789, "learning_rate": 5.4174757281553404e-05, "loss": 0.2482, "step": 9440 }, { "epoch": 91.54578313253012, "grad_norm": 7.460737228393555, "learning_rate": 5.416990291262136e-05, "loss": 0.2947, "step": 9441 }, { "epoch": 91.55542168674698, "grad_norm": 7.871649742126465, "learning_rate": 5.4165048543689326e-05, "loss": 0.1807, "step": 9442 }, { "epoch": 91.56506024096386, "grad_norm": 13.436432838439941, "learning_rate": 5.4160194174757284e-05, "loss": 0.4253, "step": 9443 }, { "epoch": 91.57469879518072, "grad_norm": 3.816162586212158, "learning_rate": 5.415533980582525e-05, "loss": 0.3261, "step": 9444 }, { "epoch": 91.58433734939759, "grad_norm": 27.967876434326172, "learning_rate": 5.4150485436893206e-05, "loss": 0.3545, "step": 9445 }, { "epoch": 91.59397590361446, "grad_norm": 6.393531799316406, "learning_rate": 5.4145631067961164e-05, "loss": 0.3485, "step": 9446 }, { "epoch": 91.60361445783133, "grad_norm": 20.17350959777832, "learning_rate": 5.414077669902913e-05, "loss": 0.295, "step": 9447 }, { "epoch": 91.61325301204819, "grad_norm": 5.508694171905518, "learning_rate": 5.4135922330097086e-05, "loss": 0.163, "step": 9448 }, { "epoch": 91.62289156626505, "grad_norm": 4.547636985778809, "learning_rate": 5.413106796116505e-05, "loss": 0.2575, "step": 9449 }, { "epoch": 91.63253012048193, "grad_norm": 11.205397605895996, "learning_rate": 5.412621359223301e-05, "loss": 0.397, "step": 9450 }, { "epoch": 91.6421686746988, "grad_norm": 7.353526592254639, "learning_rate": 5.412135922330097e-05, "loss": 0.346, "step": 9451 }, { "epoch": 91.65180722891566, "grad_norm": 6.092895030975342, "learning_rate": 5.411650485436893e-05, "loss": 0.2667, "step": 9452 }, { "epoch": 91.66144578313254, "grad_norm": 35.206050872802734, "learning_rate": 5.4111650485436895e-05, "loss": 0.5438, "step": 9453 }, { "epoch": 91.6710843373494, "grad_norm": 4.22556734085083, "learning_rate": 5.410679611650485e-05, "loss": 0.1328, "step": 9454 }, { "epoch": 91.68072289156626, "grad_norm": 9.563387870788574, "learning_rate": 5.410194174757282e-05, "loss": 0.0905, "step": 9455 }, { "epoch": 91.69036144578314, "grad_norm": 56.1880989074707, "learning_rate": 5.4097087378640775e-05, "loss": 0.5528, "step": 9456 }, { "epoch": 91.7, "grad_norm": 6.921177387237549, "learning_rate": 5.409223300970874e-05, "loss": 0.224, "step": 9457 }, { "epoch": 91.70963855421687, "grad_norm": 28.569923400878906, "learning_rate": 5.40873786407767e-05, "loss": 0.3895, "step": 9458 }, { "epoch": 91.71927710843373, "grad_norm": 12.816387176513672, "learning_rate": 5.408252427184466e-05, "loss": 0.2777, "step": 9459 }, { "epoch": 91.7289156626506, "grad_norm": 13.367244720458984, "learning_rate": 5.407766990291262e-05, "loss": 0.3323, "step": 9460 }, { "epoch": 91.73855421686747, "grad_norm": 8.391571998596191, "learning_rate": 5.4072815533980584e-05, "loss": 0.3497, "step": 9461 }, { "epoch": 91.74819277108433, "grad_norm": 13.315845489501953, "learning_rate": 5.406796116504854e-05, "loss": 0.2348, "step": 9462 }, { "epoch": 91.75783132530121, "grad_norm": 13.431177139282227, "learning_rate": 5.40631067961165e-05, "loss": 0.2204, "step": 9463 }, { "epoch": 91.76746987951807, "grad_norm": 4.178961753845215, "learning_rate": 5.4058252427184464e-05, "loss": 0.3014, "step": 9464 }, { "epoch": 91.77710843373494, "grad_norm": 15.489265441894531, "learning_rate": 5.405339805825242e-05, "loss": 0.5638, "step": 9465 }, { "epoch": 91.7867469879518, "grad_norm": 7.825664520263672, "learning_rate": 5.4048543689320387e-05, "loss": 0.3857, "step": 9466 }, { "epoch": 91.79638554216868, "grad_norm": 30.4598388671875, "learning_rate": 5.404368932038836e-05, "loss": 0.3327, "step": 9467 }, { "epoch": 91.80602409638554, "grad_norm": 10.859512329101562, "learning_rate": 5.4038834951456316e-05, "loss": 0.4548, "step": 9468 }, { "epoch": 91.8156626506024, "grad_norm": 18.976205825805664, "learning_rate": 5.403398058252428e-05, "loss": 0.1457, "step": 9469 }, { "epoch": 91.82530120481928, "grad_norm": 6.346792221069336, "learning_rate": 5.402912621359224e-05, "loss": 0.2614, "step": 9470 }, { "epoch": 91.83493975903615, "grad_norm": 6.450075626373291, "learning_rate": 5.40242718446602e-05, "loss": 0.3038, "step": 9471 }, { "epoch": 91.84457831325301, "grad_norm": 8.9028902053833, "learning_rate": 5.401941747572816e-05, "loss": 0.4883, "step": 9472 }, { "epoch": 91.85421686746987, "grad_norm": 6.384052276611328, "learning_rate": 5.4014563106796125e-05, "loss": 0.4233, "step": 9473 }, { "epoch": 91.86385542168675, "grad_norm": 23.45726776123047, "learning_rate": 5.400970873786408e-05, "loss": 0.3901, "step": 9474 }, { "epoch": 91.87349397590361, "grad_norm": 16.114145278930664, "learning_rate": 5.400485436893205e-05, "loss": 0.295, "step": 9475 }, { "epoch": 91.88313253012048, "grad_norm": 9.288652420043945, "learning_rate": 5.4000000000000005e-05, "loss": 0.4969, "step": 9476 }, { "epoch": 91.89277108433735, "grad_norm": 13.854628562927246, "learning_rate": 5.399514563106797e-05, "loss": 0.2882, "step": 9477 }, { "epoch": 91.90240963855422, "grad_norm": 16.408632278442383, "learning_rate": 5.399029126213593e-05, "loss": 0.2975, "step": 9478 }, { "epoch": 91.91204819277108, "grad_norm": 6.540271282196045, "learning_rate": 5.398543689320389e-05, "loss": 0.2055, "step": 9479 }, { "epoch": 91.92168674698796, "grad_norm": 22.340543746948242, "learning_rate": 5.398058252427185e-05, "loss": 0.5522, "step": 9480 }, { "epoch": 91.93132530120482, "grad_norm": 19.3928165435791, "learning_rate": 5.3975728155339814e-05, "loss": 0.6668, "step": 9481 }, { "epoch": 91.94096385542169, "grad_norm": 19.38624382019043, "learning_rate": 5.397087378640777e-05, "loss": 0.3972, "step": 9482 }, { "epoch": 91.95060240963855, "grad_norm": 7.537986755371094, "learning_rate": 5.396601941747573e-05, "loss": 0.1252, "step": 9483 }, { "epoch": 91.96024096385543, "grad_norm": 6.348085403442383, "learning_rate": 5.3961165048543694e-05, "loss": 0.3389, "step": 9484 }, { "epoch": 91.96987951807229, "grad_norm": 39.279319763183594, "learning_rate": 5.395631067961165e-05, "loss": 0.3518, "step": 9485 }, { "epoch": 91.97951807228915, "grad_norm": 14.485138893127441, "learning_rate": 5.3951456310679616e-05, "loss": 0.4317, "step": 9486 }, { "epoch": 91.98915662650603, "grad_norm": 4.868889331817627, "learning_rate": 5.3946601941747574e-05, "loss": 0.1337, "step": 9487 }, { "epoch": 92.00481927710844, "grad_norm": 17.817028045654297, "learning_rate": 5.394174757281554e-05, "loss": 0.2973, "step": 9488 }, { "epoch": 92.0144578313253, "grad_norm": 14.127325057983398, "learning_rate": 5.3936893203883496e-05, "loss": 0.167, "step": 9489 }, { "epoch": 92.02409638554217, "grad_norm": 2.1692044734954834, "learning_rate": 5.393203883495146e-05, "loss": 0.3836, "step": 9490 }, { "epoch": 92.03373493975904, "grad_norm": 35.20198059082031, "learning_rate": 5.392718446601942e-05, "loss": 0.1628, "step": 9491 }, { "epoch": 92.0433734939759, "grad_norm": 12.297026634216309, "learning_rate": 5.392233009708738e-05, "loss": 0.1169, "step": 9492 }, { "epoch": 92.05301204819277, "grad_norm": 3.862605094909668, "learning_rate": 5.391747572815534e-05, "loss": 0.2442, "step": 9493 }, { "epoch": 92.06265060240963, "grad_norm": 2.428557872772217, "learning_rate": 5.3912621359223305e-05, "loss": 0.0824, "step": 9494 }, { "epoch": 92.07228915662651, "grad_norm": 11.774358749389648, "learning_rate": 5.390776699029126e-05, "loss": 0.2108, "step": 9495 }, { "epoch": 92.08192771084337, "grad_norm": 5.219541072845459, "learning_rate": 5.390291262135923e-05, "loss": 0.1673, "step": 9496 }, { "epoch": 92.09156626506024, "grad_norm": 8.355734825134277, "learning_rate": 5.3898058252427185e-05, "loss": 0.2516, "step": 9497 }, { "epoch": 92.10120481927711, "grad_norm": 22.823348999023438, "learning_rate": 5.389320388349515e-05, "loss": 0.1876, "step": 9498 }, { "epoch": 92.11084337349398, "grad_norm": 99.40072631835938, "learning_rate": 5.388834951456311e-05, "loss": 0.3948, "step": 9499 }, { "epoch": 92.12048192771084, "grad_norm": 3.024888753890991, "learning_rate": 5.3883495145631065e-05, "loss": 0.2282, "step": 9500 }, { "epoch": 92.1301204819277, "grad_norm": 12.086542129516602, "learning_rate": 5.387864077669903e-05, "loss": 0.2014, "step": 9501 }, { "epoch": 92.13975903614458, "grad_norm": 6.7168169021606445, "learning_rate": 5.387378640776699e-05, "loss": 0.227, "step": 9502 }, { "epoch": 92.14939759036145, "grad_norm": 18.499492645263672, "learning_rate": 5.386893203883495e-05, "loss": 0.2629, "step": 9503 }, { "epoch": 92.15903614457831, "grad_norm": 8.612943649291992, "learning_rate": 5.386407766990291e-05, "loss": 0.1629, "step": 9504 }, { "epoch": 92.16867469879519, "grad_norm": 24.435840606689453, "learning_rate": 5.3859223300970875e-05, "loss": 0.3914, "step": 9505 }, { "epoch": 92.17831325301205, "grad_norm": 9.802433967590332, "learning_rate": 5.385436893203883e-05, "loss": 0.3413, "step": 9506 }, { "epoch": 92.18795180722891, "grad_norm": 4.99387788772583, "learning_rate": 5.38495145631068e-05, "loss": 0.2308, "step": 9507 }, { "epoch": 92.19759036144578, "grad_norm": 7.9489827156066895, "learning_rate": 5.3844660194174755e-05, "loss": 0.2121, "step": 9508 }, { "epoch": 92.20722891566265, "grad_norm": 17.047405242919922, "learning_rate": 5.383980582524272e-05, "loss": 0.305, "step": 9509 }, { "epoch": 92.21686746987952, "grad_norm": 8.436565399169922, "learning_rate": 5.383495145631068e-05, "loss": 0.2902, "step": 9510 }, { "epoch": 92.22650602409638, "grad_norm": 4.667634963989258, "learning_rate": 5.383009708737864e-05, "loss": 0.2284, "step": 9511 }, { "epoch": 92.23614457831326, "grad_norm": 12.257506370544434, "learning_rate": 5.38252427184466e-05, "loss": 0.5145, "step": 9512 }, { "epoch": 92.24578313253012, "grad_norm": 3.3102049827575684, "learning_rate": 5.3820388349514564e-05, "loss": 0.1547, "step": 9513 }, { "epoch": 92.25542168674698, "grad_norm": 27.311765670776367, "learning_rate": 5.381553398058252e-05, "loss": 0.3323, "step": 9514 }, { "epoch": 92.26506024096386, "grad_norm": 2.6701109409332275, "learning_rate": 5.3810679611650486e-05, "loss": 0.1089, "step": 9515 }, { "epoch": 92.27469879518073, "grad_norm": 25.483051300048828, "learning_rate": 5.3805825242718444e-05, "loss": 0.454, "step": 9516 }, { "epoch": 92.28433734939759, "grad_norm": 21.46025276184082, "learning_rate": 5.38009708737864e-05, "loss": 0.3979, "step": 9517 }, { "epoch": 92.29397590361445, "grad_norm": 6.191549301147461, "learning_rate": 5.379611650485438e-05, "loss": 0.1361, "step": 9518 }, { "epoch": 92.30361445783133, "grad_norm": 13.794833183288574, "learning_rate": 5.379126213592234e-05, "loss": 0.4084, "step": 9519 }, { "epoch": 92.3132530120482, "grad_norm": 12.1937837600708, "learning_rate": 5.37864077669903e-05, "loss": 0.3672, "step": 9520 }, { "epoch": 92.32289156626506, "grad_norm": 9.32162094116211, "learning_rate": 5.378155339805826e-05, "loss": 0.5142, "step": 9521 }, { "epoch": 92.33253012048193, "grad_norm": 4.563716411590576, "learning_rate": 5.377669902912622e-05, "loss": 0.3186, "step": 9522 }, { "epoch": 92.3421686746988, "grad_norm": 4.536680221557617, "learning_rate": 5.377184466019418e-05, "loss": 0.1509, "step": 9523 }, { "epoch": 92.35180722891566, "grad_norm": 8.16970443725586, "learning_rate": 5.376699029126214e-05, "loss": 0.4464, "step": 9524 }, { "epoch": 92.36144578313252, "grad_norm": 18.141958236694336, "learning_rate": 5.3762135922330104e-05, "loss": 0.2952, "step": 9525 }, { "epoch": 92.3710843373494, "grad_norm": 4.3570709228515625, "learning_rate": 5.375728155339806e-05, "loss": 0.1758, "step": 9526 }, { "epoch": 92.38072289156626, "grad_norm": 2.3343565464019775, "learning_rate": 5.3752427184466027e-05, "loss": 0.0517, "step": 9527 }, { "epoch": 92.39036144578313, "grad_norm": 84.10486602783203, "learning_rate": 5.3747572815533984e-05, "loss": 0.3744, "step": 9528 }, { "epoch": 92.4, "grad_norm": 6.71901273727417, "learning_rate": 5.374271844660195e-05, "loss": 0.312, "step": 9529 }, { "epoch": 92.40963855421687, "grad_norm": 5.201056003570557, "learning_rate": 5.373786407766991e-05, "loss": 0.3138, "step": 9530 }, { "epoch": 92.41927710843373, "grad_norm": 3.94098162651062, "learning_rate": 5.373300970873787e-05, "loss": 0.2042, "step": 9531 }, { "epoch": 92.4289156626506, "grad_norm": 22.12367057800293, "learning_rate": 5.372815533980583e-05, "loss": 0.2299, "step": 9532 }, { "epoch": 92.43855421686747, "grad_norm": 8.315445899963379, "learning_rate": 5.3723300970873793e-05, "loss": 0.4651, "step": 9533 }, { "epoch": 92.44819277108434, "grad_norm": 7.5450615882873535, "learning_rate": 5.371844660194175e-05, "loss": 0.2455, "step": 9534 }, { "epoch": 92.4578313253012, "grad_norm": 14.36998462677002, "learning_rate": 5.3713592233009716e-05, "loss": 0.3477, "step": 9535 }, { "epoch": 92.46746987951808, "grad_norm": 4.084775447845459, "learning_rate": 5.3708737864077674e-05, "loss": 0.3109, "step": 9536 }, { "epoch": 92.47710843373494, "grad_norm": 5.5729827880859375, "learning_rate": 5.370388349514564e-05, "loss": 0.2458, "step": 9537 }, { "epoch": 92.4867469879518, "grad_norm": 31.072622299194336, "learning_rate": 5.3699029126213596e-05, "loss": 0.3252, "step": 9538 }, { "epoch": 92.49638554216868, "grad_norm": 14.184931755065918, "learning_rate": 5.3694174757281554e-05, "loss": 0.2572, "step": 9539 }, { "epoch": 92.50602409638554, "grad_norm": 6.786371231079102, "learning_rate": 5.368932038834952e-05, "loss": 0.3043, "step": 9540 }, { "epoch": 92.51566265060241, "grad_norm": 22.925888061523438, "learning_rate": 5.3684466019417476e-05, "loss": 0.3445, "step": 9541 }, { "epoch": 92.52530120481927, "grad_norm": 2.9838366508483887, "learning_rate": 5.367961165048544e-05, "loss": 0.1882, "step": 9542 }, { "epoch": 92.53493975903615, "grad_norm": 3.984658718109131, "learning_rate": 5.36747572815534e-05, "loss": 0.1976, "step": 9543 }, { "epoch": 92.54457831325301, "grad_norm": 3.527364492416382, "learning_rate": 5.366990291262136e-05, "loss": 0.2642, "step": 9544 }, { "epoch": 92.55421686746988, "grad_norm": 10.928604125976562, "learning_rate": 5.366504854368932e-05, "loss": 0.4066, "step": 9545 }, { "epoch": 92.56385542168675, "grad_norm": 18.953271865844727, "learning_rate": 5.3660194174757285e-05, "loss": 0.296, "step": 9546 }, { "epoch": 92.57349397590362, "grad_norm": 14.481697082519531, "learning_rate": 5.365533980582524e-05, "loss": 0.2432, "step": 9547 }, { "epoch": 92.58313253012048, "grad_norm": 11.002917289733887, "learning_rate": 5.365048543689321e-05, "loss": 0.4424, "step": 9548 }, { "epoch": 92.59277108433734, "grad_norm": 6.063878536224365, "learning_rate": 5.3645631067961165e-05, "loss": 0.2058, "step": 9549 }, { "epoch": 92.60240963855422, "grad_norm": 36.50544738769531, "learning_rate": 5.364077669902913e-05, "loss": 0.4064, "step": 9550 }, { "epoch": 92.61204819277108, "grad_norm": 24.335718154907227, "learning_rate": 5.363592233009709e-05, "loss": 0.4197, "step": 9551 }, { "epoch": 92.62168674698795, "grad_norm": 6.651752471923828, "learning_rate": 5.363106796116505e-05, "loss": 0.1658, "step": 9552 }, { "epoch": 92.63132530120482, "grad_norm": 5.5977654457092285, "learning_rate": 5.362621359223301e-05, "loss": 0.268, "step": 9553 }, { "epoch": 92.64096385542169, "grad_norm": 17.90384292602539, "learning_rate": 5.3621359223300974e-05, "loss": 0.3853, "step": 9554 }, { "epoch": 92.65060240963855, "grad_norm": 17.77198600769043, "learning_rate": 5.361650485436893e-05, "loss": 0.2605, "step": 9555 }, { "epoch": 92.66024096385541, "grad_norm": 4.519895076751709, "learning_rate": 5.361165048543689e-05, "loss": 0.1013, "step": 9556 }, { "epoch": 92.66987951807229, "grad_norm": 37.71611404418945, "learning_rate": 5.3606796116504854e-05, "loss": 0.4013, "step": 9557 }, { "epoch": 92.67951807228916, "grad_norm": 15.836501121520996, "learning_rate": 5.360194174757281e-05, "loss": 0.275, "step": 9558 }, { "epoch": 92.68915662650602, "grad_norm": 7.4317426681518555, "learning_rate": 5.3597087378640776e-05, "loss": 0.2424, "step": 9559 }, { "epoch": 92.6987951807229, "grad_norm": 11.601024627685547, "learning_rate": 5.3592233009708734e-05, "loss": 0.3951, "step": 9560 }, { "epoch": 92.70843373493976, "grad_norm": 2.4476261138916016, "learning_rate": 5.35873786407767e-05, "loss": 0.1133, "step": 9561 }, { "epoch": 92.71807228915662, "grad_norm": 5.346960067749023, "learning_rate": 5.3582524271844657e-05, "loss": 0.1306, "step": 9562 }, { "epoch": 92.7277108433735, "grad_norm": 7.157000541687012, "learning_rate": 5.357766990291262e-05, "loss": 0.2869, "step": 9563 }, { "epoch": 92.73734939759036, "grad_norm": 38.26102828979492, "learning_rate": 5.357281553398058e-05, "loss": 0.3231, "step": 9564 }, { "epoch": 92.74698795180723, "grad_norm": 21.335756301879883, "learning_rate": 5.356796116504854e-05, "loss": 0.2642, "step": 9565 }, { "epoch": 92.75662650602409, "grad_norm": 3.1550934314727783, "learning_rate": 5.35631067961165e-05, "loss": 0.0932, "step": 9566 }, { "epoch": 92.76626506024097, "grad_norm": 19.20625877380371, "learning_rate": 5.3558252427184466e-05, "loss": 0.3026, "step": 9567 }, { "epoch": 92.77590361445783, "grad_norm": 8.569758415222168, "learning_rate": 5.355339805825244e-05, "loss": 0.3711, "step": 9568 }, { "epoch": 92.7855421686747, "grad_norm": 7.158308982849121, "learning_rate": 5.3548543689320395e-05, "loss": 0.3931, "step": 9569 }, { "epoch": 92.79518072289157, "grad_norm": 17.390771865844727, "learning_rate": 5.354368932038836e-05, "loss": 0.3125, "step": 9570 }, { "epoch": 92.80481927710844, "grad_norm": 11.162968635559082, "learning_rate": 5.353883495145632e-05, "loss": 0.3417, "step": 9571 }, { "epoch": 92.8144578313253, "grad_norm": 2.9838180541992188, "learning_rate": 5.353398058252428e-05, "loss": 0.1775, "step": 9572 }, { "epoch": 92.82409638554216, "grad_norm": 13.805277824401855, "learning_rate": 5.352912621359224e-05, "loss": 0.3385, "step": 9573 }, { "epoch": 92.83373493975904, "grad_norm": 12.234238624572754, "learning_rate": 5.3524271844660204e-05, "loss": 0.2974, "step": 9574 }, { "epoch": 92.8433734939759, "grad_norm": 6.516094207763672, "learning_rate": 5.351941747572816e-05, "loss": 0.1749, "step": 9575 }, { "epoch": 92.85301204819277, "grad_norm": 54.50382614135742, "learning_rate": 5.3514563106796126e-05, "loss": 0.3641, "step": 9576 }, { "epoch": 92.86265060240964, "grad_norm": 12.19788932800293, "learning_rate": 5.3509708737864084e-05, "loss": 0.3383, "step": 9577 }, { "epoch": 92.87228915662651, "grad_norm": 12.240410804748535, "learning_rate": 5.350485436893204e-05, "loss": 0.322, "step": 9578 }, { "epoch": 92.88192771084337, "grad_norm": 5.90385103225708, "learning_rate": 5.3500000000000006e-05, "loss": 0.3482, "step": 9579 }, { "epoch": 92.89156626506023, "grad_norm": 25.08863639831543, "learning_rate": 5.3495145631067964e-05, "loss": 0.2217, "step": 9580 }, { "epoch": 92.90120481927711, "grad_norm": 8.663228034973145, "learning_rate": 5.349029126213593e-05, "loss": 0.2624, "step": 9581 }, { "epoch": 92.91084337349398, "grad_norm": 2.3461852073669434, "learning_rate": 5.3485436893203886e-05, "loss": 0.0781, "step": 9582 }, { "epoch": 92.92048192771084, "grad_norm": 10.954327583312988, "learning_rate": 5.348058252427185e-05, "loss": 0.3693, "step": 9583 }, { "epoch": 92.93012048192772, "grad_norm": 7.311123371124268, "learning_rate": 5.347572815533981e-05, "loss": 0.2686, "step": 9584 }, { "epoch": 92.93975903614458, "grad_norm": 5.925896167755127, "learning_rate": 5.347087378640777e-05, "loss": 0.1714, "step": 9585 }, { "epoch": 92.94939759036144, "grad_norm": 9.8560791015625, "learning_rate": 5.346601941747573e-05, "loss": 0.1548, "step": 9586 }, { "epoch": 92.95903614457832, "grad_norm": 6.590149402618408, "learning_rate": 5.3461165048543695e-05, "loss": 0.2853, "step": 9587 }, { "epoch": 92.96867469879518, "grad_norm": 14.315564155578613, "learning_rate": 5.345631067961165e-05, "loss": 0.3662, "step": 9588 }, { "epoch": 92.97831325301205, "grad_norm": 7.680940628051758, "learning_rate": 5.345145631067962e-05, "loss": 0.3264, "step": 9589 }, { "epoch": 92.98795180722891, "grad_norm": 4.356922149658203, "learning_rate": 5.3446601941747575e-05, "loss": 0.2781, "step": 9590 }, { "epoch": 93.00361445783132, "grad_norm": 6.051095962524414, "learning_rate": 5.344174757281554e-05, "loss": 0.2151, "step": 9591 }, { "epoch": 93.0132530120482, "grad_norm": 14.39313793182373, "learning_rate": 5.34368932038835e-05, "loss": 0.3941, "step": 9592 }, { "epoch": 93.02289156626506, "grad_norm": 7.541507720947266, "learning_rate": 5.343203883495146e-05, "loss": 0.3227, "step": 9593 }, { "epoch": 93.03253012048192, "grad_norm": 5.148751735687256, "learning_rate": 5.342718446601942e-05, "loss": 0.4259, "step": 9594 }, { "epoch": 93.0421686746988, "grad_norm": 12.090863227844238, "learning_rate": 5.342233009708738e-05, "loss": 0.4054, "step": 9595 }, { "epoch": 93.05180722891566, "grad_norm": 4.387895584106445, "learning_rate": 5.341747572815534e-05, "loss": 0.0699, "step": 9596 }, { "epoch": 93.06144578313253, "grad_norm": 4.402541160583496, "learning_rate": 5.34126213592233e-05, "loss": 0.1748, "step": 9597 }, { "epoch": 93.0710843373494, "grad_norm": 32.82345962524414, "learning_rate": 5.3407766990291265e-05, "loss": 0.5229, "step": 9598 }, { "epoch": 93.08072289156627, "grad_norm": 41.53312301635742, "learning_rate": 5.340291262135922e-05, "loss": 0.4145, "step": 9599 }, { "epoch": 93.09036144578313, "grad_norm": 7.023188591003418, "learning_rate": 5.339805825242719e-05, "loss": 0.3381, "step": 9600 }, { "epoch": 93.1, "grad_norm": 5.837996959686279, "learning_rate": 5.3393203883495145e-05, "loss": 0.1669, "step": 9601 }, { "epoch": 93.10963855421687, "grad_norm": 21.50885772705078, "learning_rate": 5.338834951456311e-05, "loss": 0.3333, "step": 9602 }, { "epoch": 93.11927710843374, "grad_norm": 6.303531169891357, "learning_rate": 5.338349514563107e-05, "loss": 0.3126, "step": 9603 }, { "epoch": 93.1289156626506, "grad_norm": 5.490333080291748, "learning_rate": 5.337864077669903e-05, "loss": 0.2293, "step": 9604 }, { "epoch": 93.13855421686748, "grad_norm": 10.049640655517578, "learning_rate": 5.337378640776699e-05, "loss": 0.3542, "step": 9605 }, { "epoch": 93.14819277108434, "grad_norm": 10.49155044555664, "learning_rate": 5.3368932038834954e-05, "loss": 0.2736, "step": 9606 }, { "epoch": 93.1578313253012, "grad_norm": 7.549398899078369, "learning_rate": 5.336407766990291e-05, "loss": 0.3494, "step": 9607 }, { "epoch": 93.16746987951807, "grad_norm": 31.514184951782227, "learning_rate": 5.3359223300970876e-05, "loss": 0.4597, "step": 9608 }, { "epoch": 93.17710843373494, "grad_norm": 10.500505447387695, "learning_rate": 5.3354368932038834e-05, "loss": 0.2448, "step": 9609 }, { "epoch": 93.1867469879518, "grad_norm": 22.864709854125977, "learning_rate": 5.33495145631068e-05, "loss": 0.6678, "step": 9610 }, { "epoch": 93.19638554216867, "grad_norm": 5.875350475311279, "learning_rate": 5.3344660194174756e-05, "loss": 0.1457, "step": 9611 }, { "epoch": 93.20602409638555, "grad_norm": 4.241982936859131, "learning_rate": 5.3339805825242714e-05, "loss": 0.21, "step": 9612 }, { "epoch": 93.21566265060241, "grad_norm": 20.37126350402832, "learning_rate": 5.333495145631068e-05, "loss": 0.3734, "step": 9613 }, { "epoch": 93.22530120481927, "grad_norm": 5.691227436065674, "learning_rate": 5.3330097087378636e-05, "loss": 0.2511, "step": 9614 }, { "epoch": 93.23493975903614, "grad_norm": 4.837275981903076, "learning_rate": 5.33252427184466e-05, "loss": 0.07, "step": 9615 }, { "epoch": 93.24457831325302, "grad_norm": 14.628503799438477, "learning_rate": 5.332038834951456e-05, "loss": 0.3613, "step": 9616 }, { "epoch": 93.25421686746988, "grad_norm": 4.846022129058838, "learning_rate": 5.331553398058252e-05, "loss": 0.15, "step": 9617 }, { "epoch": 93.26385542168674, "grad_norm": 8.274408340454102, "learning_rate": 5.331067961165048e-05, "loss": 0.3592, "step": 9618 }, { "epoch": 93.27349397590362, "grad_norm": 7.228302478790283, "learning_rate": 5.330582524271845e-05, "loss": 0.2019, "step": 9619 }, { "epoch": 93.28313253012048, "grad_norm": 33.805240631103516, "learning_rate": 5.3300970873786417e-05, "loss": 0.4572, "step": 9620 }, { "epoch": 93.29277108433735, "grad_norm": 12.525428771972656, "learning_rate": 5.3296116504854374e-05, "loss": 0.5345, "step": 9621 }, { "epoch": 93.30240963855422, "grad_norm": 9.543991088867188, "learning_rate": 5.329126213592234e-05, "loss": 0.3061, "step": 9622 }, { "epoch": 93.31204819277109, "grad_norm": 22.160594940185547, "learning_rate": 5.3286407766990297e-05, "loss": 0.1629, "step": 9623 }, { "epoch": 93.32168674698795, "grad_norm": 7.045642852783203, "learning_rate": 5.328155339805826e-05, "loss": 0.306, "step": 9624 }, { "epoch": 93.33132530120481, "grad_norm": 13.698524475097656, "learning_rate": 5.327669902912622e-05, "loss": 0.5117, "step": 9625 }, { "epoch": 93.34096385542169, "grad_norm": 8.218881607055664, "learning_rate": 5.3271844660194183e-05, "loss": 0.2577, "step": 9626 }, { "epoch": 93.35060240963855, "grad_norm": 21.008628845214844, "learning_rate": 5.326699029126214e-05, "loss": 0.4368, "step": 9627 }, { "epoch": 93.36024096385542, "grad_norm": 11.866705894470215, "learning_rate": 5.3262135922330106e-05, "loss": 0.2259, "step": 9628 }, { "epoch": 93.3698795180723, "grad_norm": 16.899709701538086, "learning_rate": 5.3257281553398063e-05, "loss": 0.195, "step": 9629 }, { "epoch": 93.37951807228916, "grad_norm": 6.679144859313965, "learning_rate": 5.325242718446603e-05, "loss": 0.58, "step": 9630 }, { "epoch": 93.38915662650602, "grad_norm": 7.821354389190674, "learning_rate": 5.3247572815533986e-05, "loss": 0.2967, "step": 9631 }, { "epoch": 93.39879518072289, "grad_norm": 7.715855121612549, "learning_rate": 5.3242718446601943e-05, "loss": 0.5163, "step": 9632 }, { "epoch": 93.40843373493976, "grad_norm": 15.667115211486816, "learning_rate": 5.323786407766991e-05, "loss": 0.3749, "step": 9633 }, { "epoch": 93.41807228915663, "grad_norm": 7.5777974128723145, "learning_rate": 5.3233009708737866e-05, "loss": 0.1672, "step": 9634 }, { "epoch": 93.42771084337349, "grad_norm": 5.700135707855225, "learning_rate": 5.322815533980583e-05, "loss": 0.2846, "step": 9635 }, { "epoch": 93.43734939759037, "grad_norm": 2.606287717819214, "learning_rate": 5.322330097087379e-05, "loss": 0.1071, "step": 9636 }, { "epoch": 93.44698795180723, "grad_norm": 6.624941349029541, "learning_rate": 5.321844660194175e-05, "loss": 0.1538, "step": 9637 }, { "epoch": 93.4566265060241, "grad_norm": 13.328597068786621, "learning_rate": 5.321359223300971e-05, "loss": 0.3305, "step": 9638 }, { "epoch": 93.46626506024096, "grad_norm": 18.242931365966797, "learning_rate": 5.3208737864077675e-05, "loss": 0.2766, "step": 9639 }, { "epoch": 93.47590361445783, "grad_norm": 3.3758320808410645, "learning_rate": 5.320388349514563e-05, "loss": 0.1781, "step": 9640 }, { "epoch": 93.4855421686747, "grad_norm": 6.671545505523682, "learning_rate": 5.31990291262136e-05, "loss": 0.2266, "step": 9641 }, { "epoch": 93.49518072289156, "grad_norm": 3.8718128204345703, "learning_rate": 5.3194174757281555e-05, "loss": 0.1388, "step": 9642 }, { "epoch": 93.50481927710844, "grad_norm": 16.291357040405273, "learning_rate": 5.318932038834952e-05, "loss": 0.3884, "step": 9643 }, { "epoch": 93.5144578313253, "grad_norm": 7.4938554763793945, "learning_rate": 5.318446601941748e-05, "loss": 0.1641, "step": 9644 }, { "epoch": 93.52409638554217, "grad_norm": 10.733654975891113, "learning_rate": 5.317961165048544e-05, "loss": 0.2541, "step": 9645 }, { "epoch": 93.53373493975904, "grad_norm": 17.811832427978516, "learning_rate": 5.31747572815534e-05, "loss": 0.4106, "step": 9646 }, { "epoch": 93.5433734939759, "grad_norm": 5.337193965911865, "learning_rate": 5.3169902912621364e-05, "loss": 0.2482, "step": 9647 }, { "epoch": 93.55301204819277, "grad_norm": 34.09040832519531, "learning_rate": 5.316504854368932e-05, "loss": 0.3802, "step": 9648 }, { "epoch": 93.56265060240963, "grad_norm": 2.8817763328552246, "learning_rate": 5.316019417475728e-05, "loss": 0.123, "step": 9649 }, { "epoch": 93.57228915662651, "grad_norm": 7.7277913093566895, "learning_rate": 5.3155339805825244e-05, "loss": 0.2466, "step": 9650 }, { "epoch": 93.58192771084337, "grad_norm": 10.205604553222656, "learning_rate": 5.31504854368932e-05, "loss": 0.2014, "step": 9651 }, { "epoch": 93.59156626506024, "grad_norm": 12.825281143188477, "learning_rate": 5.3145631067961166e-05, "loss": 0.4472, "step": 9652 }, { "epoch": 93.60120481927711, "grad_norm": 16.468215942382812, "learning_rate": 5.3140776699029124e-05, "loss": 0.2624, "step": 9653 }, { "epoch": 93.61084337349398, "grad_norm": 21.992374420166016, "learning_rate": 5.313592233009709e-05, "loss": 0.342, "step": 9654 }, { "epoch": 93.62048192771084, "grad_norm": 23.242698669433594, "learning_rate": 5.3131067961165046e-05, "loss": 0.5181, "step": 9655 }, { "epoch": 93.6301204819277, "grad_norm": 16.328773498535156, "learning_rate": 5.312621359223301e-05, "loss": 0.2309, "step": 9656 }, { "epoch": 93.63975903614458, "grad_norm": 11.879545211791992, "learning_rate": 5.312135922330097e-05, "loss": 0.2345, "step": 9657 }, { "epoch": 93.64939759036145, "grad_norm": 12.62193775177002, "learning_rate": 5.311650485436893e-05, "loss": 0.2179, "step": 9658 }, { "epoch": 93.65903614457831, "grad_norm": 9.961455345153809, "learning_rate": 5.311165048543689e-05, "loss": 0.1899, "step": 9659 }, { "epoch": 93.66867469879519, "grad_norm": 43.102821350097656, "learning_rate": 5.3106796116504856e-05, "loss": 0.4471, "step": 9660 }, { "epoch": 93.67831325301205, "grad_norm": 6.676729679107666, "learning_rate": 5.310194174757281e-05, "loss": 0.3086, "step": 9661 }, { "epoch": 93.68795180722891, "grad_norm": 13.368285179138184, "learning_rate": 5.309708737864078e-05, "loss": 0.3077, "step": 9662 }, { "epoch": 93.69759036144578, "grad_norm": 3.051867961883545, "learning_rate": 5.3092233009708736e-05, "loss": 0.119, "step": 9663 }, { "epoch": 93.70722891566265, "grad_norm": 21.428234100341797, "learning_rate": 5.30873786407767e-05, "loss": 0.2149, "step": 9664 }, { "epoch": 93.71686746987952, "grad_norm": 11.477092742919922, "learning_rate": 5.308252427184466e-05, "loss": 0.2373, "step": 9665 }, { "epoch": 93.72650602409638, "grad_norm": 61.73954391479492, "learning_rate": 5.3077669902912616e-05, "loss": 0.3572, "step": 9666 }, { "epoch": 93.73614457831326, "grad_norm": 9.345561027526855, "learning_rate": 5.307281553398058e-05, "loss": 0.2833, "step": 9667 }, { "epoch": 93.74578313253012, "grad_norm": 12.448492050170898, "learning_rate": 5.306796116504854e-05, "loss": 0.2613, "step": 9668 }, { "epoch": 93.75542168674698, "grad_norm": 10.89980411529541, "learning_rate": 5.3063106796116516e-05, "loss": 0.3265, "step": 9669 }, { "epoch": 93.76506024096386, "grad_norm": 10.573249816894531, "learning_rate": 5.3058252427184474e-05, "loss": 0.3558, "step": 9670 }, { "epoch": 93.77469879518073, "grad_norm": 7.687826156616211, "learning_rate": 5.305339805825243e-05, "loss": 0.3483, "step": 9671 }, { "epoch": 93.78433734939759, "grad_norm": 3.1725754737854004, "learning_rate": 5.3048543689320396e-05, "loss": 0.0782, "step": 9672 }, { "epoch": 93.79397590361445, "grad_norm": 6.240335941314697, "learning_rate": 5.3043689320388354e-05, "loss": 0.1228, "step": 9673 }, { "epoch": 93.80361445783133, "grad_norm": 3.0472218990325928, "learning_rate": 5.303883495145632e-05, "loss": 0.1728, "step": 9674 }, { "epoch": 93.8132530120482, "grad_norm": 7.123762130737305, "learning_rate": 5.3033980582524276e-05, "loss": 0.179, "step": 9675 }, { "epoch": 93.82289156626506, "grad_norm": 3.9238150119781494, "learning_rate": 5.302912621359224e-05, "loss": 0.2008, "step": 9676 }, { "epoch": 93.83253012048193, "grad_norm": 9.063499450683594, "learning_rate": 5.30242718446602e-05, "loss": 0.2798, "step": 9677 }, { "epoch": 93.8421686746988, "grad_norm": 5.562893390655518, "learning_rate": 5.301941747572816e-05, "loss": 0.3489, "step": 9678 }, { "epoch": 93.85180722891566, "grad_norm": 29.258798599243164, "learning_rate": 5.301456310679612e-05, "loss": 0.4839, "step": 9679 }, { "epoch": 93.86144578313252, "grad_norm": 8.179065704345703, "learning_rate": 5.3009708737864085e-05, "loss": 0.1697, "step": 9680 }, { "epoch": 93.8710843373494, "grad_norm": 7.923730373382568, "learning_rate": 5.300485436893204e-05, "loss": 0.349, "step": 9681 }, { "epoch": 93.88072289156626, "grad_norm": 18.03876495361328, "learning_rate": 5.300000000000001e-05, "loss": 0.3601, "step": 9682 }, { "epoch": 93.89036144578313, "grad_norm": 9.110230445861816, "learning_rate": 5.2995145631067965e-05, "loss": 0.2149, "step": 9683 }, { "epoch": 93.9, "grad_norm": 12.097244262695312, "learning_rate": 5.299029126213593e-05, "loss": 0.1493, "step": 9684 }, { "epoch": 93.90963855421687, "grad_norm": 8.21112060546875, "learning_rate": 5.298543689320389e-05, "loss": 0.1977, "step": 9685 }, { "epoch": 93.91927710843373, "grad_norm": 9.061790466308594, "learning_rate": 5.298058252427185e-05, "loss": 0.3027, "step": 9686 }, { "epoch": 93.9289156626506, "grad_norm": 19.427066802978516, "learning_rate": 5.297572815533981e-05, "loss": 0.2852, "step": 9687 }, { "epoch": 93.93855421686747, "grad_norm": 13.034520149230957, "learning_rate": 5.297087378640777e-05, "loss": 0.2025, "step": 9688 }, { "epoch": 93.94819277108434, "grad_norm": 14.326615333557129, "learning_rate": 5.296601941747573e-05, "loss": 0.152, "step": 9689 }, { "epoch": 93.9578313253012, "grad_norm": 15.840863227844238, "learning_rate": 5.296116504854369e-05, "loss": 0.264, "step": 9690 }, { "epoch": 93.96746987951808, "grad_norm": 3.631209373474121, "learning_rate": 5.2956310679611654e-05, "loss": 0.1415, "step": 9691 }, { "epoch": 93.97710843373494, "grad_norm": 4.472979545593262, "learning_rate": 5.295145631067961e-05, "loss": 0.1245, "step": 9692 }, { "epoch": 93.9867469879518, "grad_norm": 6.7571797370910645, "learning_rate": 5.294660194174758e-05, "loss": 0.4149, "step": 9693 }, { "epoch": 94.00240963855421, "grad_norm": 18.722837448120117, "learning_rate": 5.2941747572815535e-05, "loss": 0.5115, "step": 9694 }, { "epoch": 94.01204819277109, "grad_norm": 7.389230251312256, "learning_rate": 5.29368932038835e-05, "loss": 0.1581, "step": 9695 }, { "epoch": 94.02168674698795, "grad_norm": 6.584291458129883, "learning_rate": 5.293203883495146e-05, "loss": 0.3783, "step": 9696 }, { "epoch": 94.03132530120482, "grad_norm": 5.035003185272217, "learning_rate": 5.292718446601942e-05, "loss": 0.1946, "step": 9697 }, { "epoch": 94.04096385542168, "grad_norm": 31.261489868164062, "learning_rate": 5.292233009708738e-05, "loss": 0.4809, "step": 9698 }, { "epoch": 94.05060240963856, "grad_norm": 8.301932334899902, "learning_rate": 5.2917475728155344e-05, "loss": 0.1994, "step": 9699 }, { "epoch": 94.06024096385542, "grad_norm": 5.943978309631348, "learning_rate": 5.29126213592233e-05, "loss": 0.2899, "step": 9700 }, { "epoch": 94.06987951807228, "grad_norm": 17.568693161010742, "learning_rate": 5.2907766990291266e-05, "loss": 0.2356, "step": 9701 }, { "epoch": 94.07951807228916, "grad_norm": 13.983656883239746, "learning_rate": 5.2902912621359224e-05, "loss": 0.3715, "step": 9702 }, { "epoch": 94.08915662650602, "grad_norm": 9.806459426879883, "learning_rate": 5.289805825242719e-05, "loss": 0.214, "step": 9703 }, { "epoch": 94.09879518072289, "grad_norm": 5.710343837738037, "learning_rate": 5.2893203883495146e-05, "loss": 0.2219, "step": 9704 }, { "epoch": 94.10843373493977, "grad_norm": 16.3582820892334, "learning_rate": 5.2888349514563104e-05, "loss": 0.3256, "step": 9705 }, { "epoch": 94.11807228915663, "grad_norm": 4.945910453796387, "learning_rate": 5.288349514563107e-05, "loss": 0.244, "step": 9706 }, { "epoch": 94.12771084337349, "grad_norm": 9.012683868408203, "learning_rate": 5.2878640776699026e-05, "loss": 0.2534, "step": 9707 }, { "epoch": 94.13734939759036, "grad_norm": 16.98406219482422, "learning_rate": 5.287378640776699e-05, "loss": 0.2573, "step": 9708 }, { "epoch": 94.14698795180723, "grad_norm": 7.551938533782959, "learning_rate": 5.286893203883495e-05, "loss": 0.2732, "step": 9709 }, { "epoch": 94.1566265060241, "grad_norm": 10.945993423461914, "learning_rate": 5.286407766990291e-05, "loss": 0.2461, "step": 9710 }, { "epoch": 94.16626506024096, "grad_norm": 9.450111389160156, "learning_rate": 5.285922330097087e-05, "loss": 0.278, "step": 9711 }, { "epoch": 94.17590361445784, "grad_norm": 9.8350248336792, "learning_rate": 5.2854368932038835e-05, "loss": 0.272, "step": 9712 }, { "epoch": 94.1855421686747, "grad_norm": 5.42877197265625, "learning_rate": 5.284951456310679e-05, "loss": 0.2883, "step": 9713 }, { "epoch": 94.19518072289156, "grad_norm": 23.44807243347168, "learning_rate": 5.284466019417476e-05, "loss": 0.4507, "step": 9714 }, { "epoch": 94.20481927710843, "grad_norm": 31.25331687927246, "learning_rate": 5.2839805825242715e-05, "loss": 0.274, "step": 9715 }, { "epoch": 94.2144578313253, "grad_norm": 10.160868644714355, "learning_rate": 5.283495145631068e-05, "loss": 0.2919, "step": 9716 }, { "epoch": 94.22409638554217, "grad_norm": 16.134767532348633, "learning_rate": 5.283009708737864e-05, "loss": 0.5909, "step": 9717 }, { "epoch": 94.23373493975903, "grad_norm": 10.132773399353027, "learning_rate": 5.28252427184466e-05, "loss": 0.496, "step": 9718 }, { "epoch": 94.24337349397591, "grad_norm": 8.830327987670898, "learning_rate": 5.282038834951456e-05, "loss": 0.3659, "step": 9719 }, { "epoch": 94.25301204819277, "grad_norm": 13.309982299804688, "learning_rate": 5.281553398058253e-05, "loss": 0.3108, "step": 9720 }, { "epoch": 94.26265060240964, "grad_norm": 6.512661933898926, "learning_rate": 5.2810679611650496e-05, "loss": 0.2866, "step": 9721 }, { "epoch": 94.2722891566265, "grad_norm": 13.887184143066406, "learning_rate": 5.280582524271845e-05, "loss": 0.4091, "step": 9722 }, { "epoch": 94.28192771084338, "grad_norm": 9.398805618286133, "learning_rate": 5.280097087378642e-05, "loss": 0.2163, "step": 9723 }, { "epoch": 94.29156626506024, "grad_norm": 31.03740119934082, "learning_rate": 5.2796116504854376e-05, "loss": 0.4583, "step": 9724 }, { "epoch": 94.3012048192771, "grad_norm": 11.205217361450195, "learning_rate": 5.279126213592234e-05, "loss": 0.5234, "step": 9725 }, { "epoch": 94.31084337349398, "grad_norm": 5.255236625671387, "learning_rate": 5.27864077669903e-05, "loss": 0.3221, "step": 9726 }, { "epoch": 94.32048192771084, "grad_norm": 11.503043174743652, "learning_rate": 5.2781553398058256e-05, "loss": 0.5066, "step": 9727 }, { "epoch": 94.33012048192771, "grad_norm": 17.914892196655273, "learning_rate": 5.277669902912622e-05, "loss": 0.2996, "step": 9728 }, { "epoch": 94.33975903614459, "grad_norm": 6.6655378341674805, "learning_rate": 5.277184466019418e-05, "loss": 0.2461, "step": 9729 }, { "epoch": 94.34939759036145, "grad_norm": 104.57083892822266, "learning_rate": 5.276699029126214e-05, "loss": 0.1626, "step": 9730 }, { "epoch": 94.35903614457831, "grad_norm": 9.81472110748291, "learning_rate": 5.27621359223301e-05, "loss": 0.2135, "step": 9731 }, { "epoch": 94.36867469879518, "grad_norm": 6.171461582183838, "learning_rate": 5.2757281553398065e-05, "loss": 0.1393, "step": 9732 }, { "epoch": 94.37831325301205, "grad_norm": 44.63363265991211, "learning_rate": 5.275242718446602e-05, "loss": 0.6702, "step": 9733 }, { "epoch": 94.38795180722892, "grad_norm": 7.374697208404541, "learning_rate": 5.274757281553399e-05, "loss": 0.4088, "step": 9734 }, { "epoch": 94.39759036144578, "grad_norm": 16.117124557495117, "learning_rate": 5.2742718446601945e-05, "loss": 0.3895, "step": 9735 }, { "epoch": 94.40722891566266, "grad_norm": 10.624781608581543, "learning_rate": 5.273786407766991e-05, "loss": 0.4852, "step": 9736 }, { "epoch": 94.41686746987952, "grad_norm": 9.27146053314209, "learning_rate": 5.273300970873787e-05, "loss": 0.3137, "step": 9737 }, { "epoch": 94.42650602409638, "grad_norm": 10.493027687072754, "learning_rate": 5.272815533980583e-05, "loss": 0.3215, "step": 9738 }, { "epoch": 94.43614457831325, "grad_norm": 12.82404899597168, "learning_rate": 5.272330097087379e-05, "loss": 0.3179, "step": 9739 }, { "epoch": 94.44578313253012, "grad_norm": 5.721460819244385, "learning_rate": 5.2718446601941754e-05, "loss": 0.2803, "step": 9740 }, { "epoch": 94.45542168674699, "grad_norm": 6.038799285888672, "learning_rate": 5.271359223300971e-05, "loss": 0.2344, "step": 9741 }, { "epoch": 94.46506024096385, "grad_norm": 18.063140869140625, "learning_rate": 5.2708737864077676e-05, "loss": 0.1611, "step": 9742 }, { "epoch": 94.47469879518073, "grad_norm": 8.43608283996582, "learning_rate": 5.2703883495145634e-05, "loss": 0.4227, "step": 9743 }, { "epoch": 94.48433734939759, "grad_norm": 7.466954708099365, "learning_rate": 5.269902912621359e-05, "loss": 0.2693, "step": 9744 }, { "epoch": 94.49397590361446, "grad_norm": 9.359416007995605, "learning_rate": 5.2694174757281556e-05, "loss": 0.2936, "step": 9745 }, { "epoch": 94.50361445783132, "grad_norm": 3.0331640243530273, "learning_rate": 5.2689320388349514e-05, "loss": 0.2761, "step": 9746 }, { "epoch": 94.5132530120482, "grad_norm": 9.007134437561035, "learning_rate": 5.268446601941748e-05, "loss": 0.3653, "step": 9747 }, { "epoch": 94.52289156626506, "grad_norm": 6.131917476654053, "learning_rate": 5.2679611650485436e-05, "loss": 0.1601, "step": 9748 }, { "epoch": 94.53253012048192, "grad_norm": 38.60297393798828, "learning_rate": 5.26747572815534e-05, "loss": 0.2662, "step": 9749 }, { "epoch": 94.5421686746988, "grad_norm": 8.65876579284668, "learning_rate": 5.266990291262136e-05, "loss": 0.2711, "step": 9750 }, { "epoch": 94.55180722891566, "grad_norm": 3.754765033721924, "learning_rate": 5.266504854368932e-05, "loss": 0.092, "step": 9751 }, { "epoch": 94.56144578313253, "grad_norm": 10.366231918334961, "learning_rate": 5.266019417475728e-05, "loss": 0.3894, "step": 9752 }, { "epoch": 94.5710843373494, "grad_norm": 7.678697109222412, "learning_rate": 5.2655339805825246e-05, "loss": 0.3548, "step": 9753 }, { "epoch": 94.58072289156627, "grad_norm": 6.648440361022949, "learning_rate": 5.26504854368932e-05, "loss": 0.141, "step": 9754 }, { "epoch": 94.59036144578313, "grad_norm": 16.1724853515625, "learning_rate": 5.264563106796117e-05, "loss": 0.2033, "step": 9755 }, { "epoch": 94.6, "grad_norm": 9.962491989135742, "learning_rate": 5.2640776699029126e-05, "loss": 0.1942, "step": 9756 }, { "epoch": 94.60963855421687, "grad_norm": 8.147371292114258, "learning_rate": 5.263592233009709e-05, "loss": 0.1546, "step": 9757 }, { "epoch": 94.61927710843374, "grad_norm": 11.035261154174805, "learning_rate": 5.263106796116505e-05, "loss": 0.4199, "step": 9758 }, { "epoch": 94.6289156626506, "grad_norm": 9.089814186096191, "learning_rate": 5.262621359223301e-05, "loss": 0.2718, "step": 9759 }, { "epoch": 94.63855421686748, "grad_norm": 12.49871826171875, "learning_rate": 5.262135922330097e-05, "loss": 0.3021, "step": 9760 }, { "epoch": 94.64819277108434, "grad_norm": 10.540675163269043, "learning_rate": 5.261650485436893e-05, "loss": 0.1699, "step": 9761 }, { "epoch": 94.6578313253012, "grad_norm": 6.400777339935303, "learning_rate": 5.261165048543689e-05, "loss": 0.3579, "step": 9762 }, { "epoch": 94.66746987951807, "grad_norm": 7.015246868133545, "learning_rate": 5.260679611650485e-05, "loss": 0.2191, "step": 9763 }, { "epoch": 94.67710843373494, "grad_norm": 25.829669952392578, "learning_rate": 5.2601941747572815e-05, "loss": 0.3226, "step": 9764 }, { "epoch": 94.6867469879518, "grad_norm": 5.708664417266846, "learning_rate": 5.259708737864077e-05, "loss": 0.3855, "step": 9765 }, { "epoch": 94.69638554216867, "grad_norm": 15.747536659240723, "learning_rate": 5.259223300970874e-05, "loss": 0.2463, "step": 9766 }, { "epoch": 94.70602409638555, "grad_norm": 13.824097633361816, "learning_rate": 5.2587378640776695e-05, "loss": 0.4385, "step": 9767 }, { "epoch": 94.71566265060241, "grad_norm": 5.389925479888916, "learning_rate": 5.258252427184466e-05, "loss": 0.2293, "step": 9768 }, { "epoch": 94.72530120481927, "grad_norm": 9.983600616455078, "learning_rate": 5.257766990291262e-05, "loss": 0.1597, "step": 9769 }, { "epoch": 94.73493975903614, "grad_norm": 7.060722351074219, "learning_rate": 5.257281553398059e-05, "loss": 0.1956, "step": 9770 }, { "epoch": 94.74457831325302, "grad_norm": 6.616520881652832, "learning_rate": 5.256796116504855e-05, "loss": 0.1929, "step": 9771 }, { "epoch": 94.75421686746988, "grad_norm": 3.5644588470458984, "learning_rate": 5.256310679611651e-05, "loss": 0.1095, "step": 9772 }, { "epoch": 94.76385542168674, "grad_norm": 9.894644737243652, "learning_rate": 5.2558252427184475e-05, "loss": 0.2155, "step": 9773 }, { "epoch": 94.77349397590362, "grad_norm": 7.915764331817627, "learning_rate": 5.255339805825243e-05, "loss": 0.3702, "step": 9774 }, { "epoch": 94.78313253012048, "grad_norm": 6.741111755371094, "learning_rate": 5.25485436893204e-05, "loss": 0.264, "step": 9775 }, { "epoch": 94.79277108433735, "grad_norm": 7.035111904144287, "learning_rate": 5.2543689320388355e-05, "loss": 0.2713, "step": 9776 }, { "epoch": 94.80240963855422, "grad_norm": 24.3702335357666, "learning_rate": 5.253883495145632e-05, "loss": 0.3508, "step": 9777 }, { "epoch": 94.81204819277109, "grad_norm": 21.90912437438965, "learning_rate": 5.253398058252428e-05, "loss": 0.2789, "step": 9778 }, { "epoch": 94.82168674698795, "grad_norm": 3.7209274768829346, "learning_rate": 5.252912621359224e-05, "loss": 0.1007, "step": 9779 }, { "epoch": 94.83132530120481, "grad_norm": 6.777889251708984, "learning_rate": 5.25242718446602e-05, "loss": 0.2755, "step": 9780 }, { "epoch": 94.84096385542169, "grad_norm": 6.838619709014893, "learning_rate": 5.251941747572816e-05, "loss": 0.2756, "step": 9781 }, { "epoch": 94.85060240963855, "grad_norm": 10.57164192199707, "learning_rate": 5.251456310679612e-05, "loss": 0.2536, "step": 9782 }, { "epoch": 94.86024096385542, "grad_norm": 8.96718978881836, "learning_rate": 5.250970873786408e-05, "loss": 0.3193, "step": 9783 }, { "epoch": 94.8698795180723, "grad_norm": 9.062126159667969, "learning_rate": 5.2504854368932044e-05, "loss": 0.1376, "step": 9784 }, { "epoch": 94.87951807228916, "grad_norm": 3.6512439250946045, "learning_rate": 5.25e-05, "loss": 0.3286, "step": 9785 }, { "epoch": 94.88915662650602, "grad_norm": 5.3362274169921875, "learning_rate": 5.249514563106797e-05, "loss": 0.2611, "step": 9786 }, { "epoch": 94.89879518072289, "grad_norm": 10.257896423339844, "learning_rate": 5.2490291262135924e-05, "loss": 0.4336, "step": 9787 }, { "epoch": 94.90843373493976, "grad_norm": 6.487163066864014, "learning_rate": 5.248543689320389e-05, "loss": 0.3515, "step": 9788 }, { "epoch": 94.91807228915663, "grad_norm": 5.787095069885254, "learning_rate": 5.248058252427185e-05, "loss": 0.2606, "step": 9789 }, { "epoch": 94.92771084337349, "grad_norm": 8.875121116638184, "learning_rate": 5.247572815533981e-05, "loss": 0.203, "step": 9790 }, { "epoch": 94.93734939759037, "grad_norm": 7.46127462387085, "learning_rate": 5.247087378640777e-05, "loss": 0.3988, "step": 9791 }, { "epoch": 94.94698795180723, "grad_norm": 7.466076374053955, "learning_rate": 5.2466019417475734e-05, "loss": 0.3238, "step": 9792 }, { "epoch": 94.9566265060241, "grad_norm": 8.73588752746582, "learning_rate": 5.246116504854369e-05, "loss": 0.4526, "step": 9793 }, { "epoch": 94.96626506024096, "grad_norm": 14.535372734069824, "learning_rate": 5.2456310679611656e-05, "loss": 0.3629, "step": 9794 }, { "epoch": 94.97590361445783, "grad_norm": 14.529546737670898, "learning_rate": 5.2451456310679614e-05, "loss": 0.3334, "step": 9795 }, { "epoch": 94.9855421686747, "grad_norm": 6.977667808532715, "learning_rate": 5.244660194174758e-05, "loss": 0.2382, "step": 9796 }, { "epoch": 95.0012048192771, "grad_norm": 5.158083438873291, "learning_rate": 5.2441747572815536e-05, "loss": 0.2799, "step": 9797 }, { "epoch": 95.01084337349397, "grad_norm": 6.852070331573486, "learning_rate": 5.2436893203883494e-05, "loss": 0.2472, "step": 9798 }, { "epoch": 95.02048192771085, "grad_norm": 8.710901260375977, "learning_rate": 5.243203883495146e-05, "loss": 0.2558, "step": 9799 }, { "epoch": 95.03012048192771, "grad_norm": 10.117777824401855, "learning_rate": 5.2427184466019416e-05, "loss": 0.4639, "step": 9800 }, { "epoch": 95.03975903614457, "grad_norm": 6.581822395324707, "learning_rate": 5.242233009708738e-05, "loss": 0.1558, "step": 9801 }, { "epoch": 95.04939759036145, "grad_norm": 18.271928787231445, "learning_rate": 5.241747572815534e-05, "loss": 0.3419, "step": 9802 }, { "epoch": 95.05903614457831, "grad_norm": 8.070375442504883, "learning_rate": 5.24126213592233e-05, "loss": 0.3365, "step": 9803 }, { "epoch": 95.06867469879518, "grad_norm": 7.245493412017822, "learning_rate": 5.240776699029126e-05, "loss": 0.3855, "step": 9804 }, { "epoch": 95.07831325301204, "grad_norm": 20.80179786682129, "learning_rate": 5.2402912621359225e-05, "loss": 0.1908, "step": 9805 }, { "epoch": 95.08795180722892, "grad_norm": 3.952888250350952, "learning_rate": 5.239805825242718e-05, "loss": 0.0932, "step": 9806 }, { "epoch": 95.09759036144578, "grad_norm": 3.974470376968384, "learning_rate": 5.239320388349515e-05, "loss": 0.0456, "step": 9807 }, { "epoch": 95.10722891566265, "grad_norm": 7.603262901306152, "learning_rate": 5.2388349514563105e-05, "loss": 0.3132, "step": 9808 }, { "epoch": 95.11686746987952, "grad_norm": 7.138762474060059, "learning_rate": 5.238349514563107e-05, "loss": 0.261, "step": 9809 }, { "epoch": 95.12650602409639, "grad_norm": 11.475706100463867, "learning_rate": 5.237864077669903e-05, "loss": 0.4264, "step": 9810 }, { "epoch": 95.13614457831325, "grad_norm": 6.069722652435303, "learning_rate": 5.237378640776699e-05, "loss": 0.1304, "step": 9811 }, { "epoch": 95.14578313253013, "grad_norm": 13.432783126831055, "learning_rate": 5.236893203883495e-05, "loss": 0.2421, "step": 9812 }, { "epoch": 95.15542168674699, "grad_norm": 11.747928619384766, "learning_rate": 5.2364077669902914e-05, "loss": 0.2997, "step": 9813 }, { "epoch": 95.16506024096385, "grad_norm": 9.498223304748535, "learning_rate": 5.235922330097087e-05, "loss": 0.3609, "step": 9814 }, { "epoch": 95.17469879518072, "grad_norm": 11.63095474243164, "learning_rate": 5.235436893203883e-05, "loss": 0.2623, "step": 9815 }, { "epoch": 95.1843373493976, "grad_norm": 3.5188193321228027, "learning_rate": 5.2349514563106794e-05, "loss": 0.2634, "step": 9816 }, { "epoch": 95.19397590361446, "grad_norm": 12.893824577331543, "learning_rate": 5.234466019417475e-05, "loss": 0.367, "step": 9817 }, { "epoch": 95.20361445783132, "grad_norm": 12.762125015258789, "learning_rate": 5.2339805825242717e-05, "loss": 0.4829, "step": 9818 }, { "epoch": 95.2132530120482, "grad_norm": 5.42217493057251, "learning_rate": 5.2334951456310674e-05, "loss": 0.1972, "step": 9819 }, { "epoch": 95.22289156626506, "grad_norm": 8.748028755187988, "learning_rate": 5.233009708737864e-05, "loss": 0.2483, "step": 9820 }, { "epoch": 95.23253012048193, "grad_norm": 24.211883544921875, "learning_rate": 5.232524271844661e-05, "loss": 0.3102, "step": 9821 }, { "epoch": 95.24216867469879, "grad_norm": 7.453175067901611, "learning_rate": 5.232038834951457e-05, "loss": 0.3191, "step": 9822 }, { "epoch": 95.25180722891567, "grad_norm": 7.714345455169678, "learning_rate": 5.231553398058253e-05, "loss": 0.2549, "step": 9823 }, { "epoch": 95.26144578313253, "grad_norm": 7.954293727874756, "learning_rate": 5.231067961165049e-05, "loss": 0.4785, "step": 9824 }, { "epoch": 95.2710843373494, "grad_norm": 11.588044166564941, "learning_rate": 5.2305825242718455e-05, "loss": 0.3343, "step": 9825 }, { "epoch": 95.28072289156627, "grad_norm": 9.6962890625, "learning_rate": 5.230097087378641e-05, "loss": 0.2762, "step": 9826 }, { "epoch": 95.29036144578313, "grad_norm": 11.37098217010498, "learning_rate": 5.229611650485438e-05, "loss": 0.5888, "step": 9827 }, { "epoch": 95.3, "grad_norm": 5.849062442779541, "learning_rate": 5.2291262135922335e-05, "loss": 0.354, "step": 9828 }, { "epoch": 95.30963855421686, "grad_norm": 12.094768524169922, "learning_rate": 5.22864077669903e-05, "loss": 0.3487, "step": 9829 }, { "epoch": 95.31927710843374, "grad_norm": 20.675851821899414, "learning_rate": 5.228155339805826e-05, "loss": 0.2232, "step": 9830 }, { "epoch": 95.3289156626506, "grad_norm": 16.089662551879883, "learning_rate": 5.227669902912622e-05, "loss": 0.249, "step": 9831 }, { "epoch": 95.33855421686746, "grad_norm": 10.802950859069824, "learning_rate": 5.227184466019418e-05, "loss": 0.4402, "step": 9832 }, { "epoch": 95.34819277108434, "grad_norm": 5.677891731262207, "learning_rate": 5.2266990291262144e-05, "loss": 0.2704, "step": 9833 }, { "epoch": 95.3578313253012, "grad_norm": 6.338758945465088, "learning_rate": 5.22621359223301e-05, "loss": 0.2196, "step": 9834 }, { "epoch": 95.36746987951807, "grad_norm": 22.3062801361084, "learning_rate": 5.2257281553398066e-05, "loss": 0.2732, "step": 9835 }, { "epoch": 95.37710843373495, "grad_norm": 9.763264656066895, "learning_rate": 5.2252427184466024e-05, "loss": 0.3613, "step": 9836 }, { "epoch": 95.38674698795181, "grad_norm": 8.533196449279785, "learning_rate": 5.224757281553398e-05, "loss": 0.2508, "step": 9837 }, { "epoch": 95.39638554216867, "grad_norm": 11.71915340423584, "learning_rate": 5.2242718446601946e-05, "loss": 0.3707, "step": 9838 }, { "epoch": 95.40602409638554, "grad_norm": 8.381943702697754, "learning_rate": 5.2237864077669904e-05, "loss": 0.4163, "step": 9839 }, { "epoch": 95.41566265060241, "grad_norm": 9.003950119018555, "learning_rate": 5.223300970873787e-05, "loss": 0.2802, "step": 9840 }, { "epoch": 95.42530120481928, "grad_norm": 3.1559810638427734, "learning_rate": 5.2228155339805826e-05, "loss": 0.1172, "step": 9841 }, { "epoch": 95.43493975903614, "grad_norm": 11.364280700683594, "learning_rate": 5.222330097087379e-05, "loss": 0.5012, "step": 9842 }, { "epoch": 95.44457831325302, "grad_norm": 16.96478271484375, "learning_rate": 5.221844660194175e-05, "loss": 0.4454, "step": 9843 }, { "epoch": 95.45421686746988, "grad_norm": 7.897555351257324, "learning_rate": 5.221359223300971e-05, "loss": 0.3066, "step": 9844 }, { "epoch": 95.46385542168674, "grad_norm": 22.422752380371094, "learning_rate": 5.220873786407767e-05, "loss": 0.4158, "step": 9845 }, { "epoch": 95.47349397590361, "grad_norm": 9.791089057922363, "learning_rate": 5.2203883495145635e-05, "loss": 0.261, "step": 9846 }, { "epoch": 95.48313253012049, "grad_norm": 21.0196475982666, "learning_rate": 5.219902912621359e-05, "loss": 0.3889, "step": 9847 }, { "epoch": 95.49277108433735, "grad_norm": 11.876989364624023, "learning_rate": 5.219417475728156e-05, "loss": 0.2206, "step": 9848 }, { "epoch": 95.50240963855421, "grad_norm": 6.051084041595459, "learning_rate": 5.2189320388349516e-05, "loss": 0.163, "step": 9849 }, { "epoch": 95.51204819277109, "grad_norm": 7.199100971221924, "learning_rate": 5.218446601941748e-05, "loss": 0.1685, "step": 9850 }, { "epoch": 95.52168674698795, "grad_norm": 13.23175048828125, "learning_rate": 5.217961165048544e-05, "loss": 0.2881, "step": 9851 }, { "epoch": 95.53132530120482, "grad_norm": 7.836386680603027, "learning_rate": 5.21747572815534e-05, "loss": 0.2855, "step": 9852 }, { "epoch": 95.54096385542168, "grad_norm": 6.240805149078369, "learning_rate": 5.216990291262136e-05, "loss": 0.2778, "step": 9853 }, { "epoch": 95.55060240963856, "grad_norm": 40.102989196777344, "learning_rate": 5.216504854368932e-05, "loss": 0.2798, "step": 9854 }, { "epoch": 95.56024096385542, "grad_norm": 7.8872151374816895, "learning_rate": 5.216019417475728e-05, "loss": 0.3662, "step": 9855 }, { "epoch": 95.56987951807228, "grad_norm": 10.670722961425781, "learning_rate": 5.215533980582524e-05, "loss": 0.2225, "step": 9856 }, { "epoch": 95.57951807228916, "grad_norm": 20.138870239257812, "learning_rate": 5.2150485436893205e-05, "loss": 0.2454, "step": 9857 }, { "epoch": 95.58915662650602, "grad_norm": 8.627400398254395, "learning_rate": 5.214563106796116e-05, "loss": 0.4133, "step": 9858 }, { "epoch": 95.59879518072289, "grad_norm": 5.609146595001221, "learning_rate": 5.214077669902913e-05, "loss": 0.1896, "step": 9859 }, { "epoch": 95.60843373493977, "grad_norm": 9.831884384155273, "learning_rate": 5.2135922330097085e-05, "loss": 0.3249, "step": 9860 }, { "epoch": 95.61807228915663, "grad_norm": 14.156925201416016, "learning_rate": 5.213106796116505e-05, "loss": 0.6065, "step": 9861 }, { "epoch": 95.62771084337349, "grad_norm": 14.615570068359375, "learning_rate": 5.212621359223301e-05, "loss": 0.3599, "step": 9862 }, { "epoch": 95.63734939759036, "grad_norm": 8.13647174835205, "learning_rate": 5.212135922330097e-05, "loss": 0.1854, "step": 9863 }, { "epoch": 95.64698795180723, "grad_norm": 5.427962779998779, "learning_rate": 5.211650485436893e-05, "loss": 0.2602, "step": 9864 }, { "epoch": 95.6566265060241, "grad_norm": 11.259066581726074, "learning_rate": 5.2111650485436894e-05, "loss": 0.4072, "step": 9865 }, { "epoch": 95.66626506024096, "grad_norm": 11.755563735961914, "learning_rate": 5.210679611650485e-05, "loss": 0.4984, "step": 9866 }, { "epoch": 95.67590361445784, "grad_norm": 4.611675262451172, "learning_rate": 5.2101941747572816e-05, "loss": 0.2451, "step": 9867 }, { "epoch": 95.6855421686747, "grad_norm": 8.163317680358887, "learning_rate": 5.2097087378640774e-05, "loss": 0.1848, "step": 9868 }, { "epoch": 95.69518072289156, "grad_norm": 17.94063377380371, "learning_rate": 5.209223300970874e-05, "loss": 0.3991, "step": 9869 }, { "epoch": 95.70481927710843, "grad_norm": 7.547468185424805, "learning_rate": 5.2087378640776696e-05, "loss": 0.3956, "step": 9870 }, { "epoch": 95.7144578313253, "grad_norm": 34.92923355102539, "learning_rate": 5.208252427184467e-05, "loss": 0.5558, "step": 9871 }, { "epoch": 95.72409638554217, "grad_norm": 5.52406644821167, "learning_rate": 5.207766990291263e-05, "loss": 0.2703, "step": 9872 }, { "epoch": 95.73373493975903, "grad_norm": 8.360069274902344, "learning_rate": 5.207281553398059e-05, "loss": 0.1761, "step": 9873 }, { "epoch": 95.74337349397591, "grad_norm": 9.410524368286133, "learning_rate": 5.2067961165048554e-05, "loss": 0.3179, "step": 9874 }, { "epoch": 95.75301204819277, "grad_norm": 6.750777244567871, "learning_rate": 5.206310679611651e-05, "loss": 0.1483, "step": 9875 }, { "epoch": 95.76265060240964, "grad_norm": 10.54493522644043, "learning_rate": 5.205825242718447e-05, "loss": 0.5162, "step": 9876 }, { "epoch": 95.7722891566265, "grad_norm": 6.694182395935059, "learning_rate": 5.2053398058252434e-05, "loss": 0.174, "step": 9877 }, { "epoch": 95.78192771084338, "grad_norm": 8.383296966552734, "learning_rate": 5.204854368932039e-05, "loss": 0.2014, "step": 9878 }, { "epoch": 95.79156626506024, "grad_norm": 10.36979866027832, "learning_rate": 5.204368932038836e-05, "loss": 0.1773, "step": 9879 }, { "epoch": 95.8012048192771, "grad_norm": 7.429121971130371, "learning_rate": 5.2038834951456314e-05, "loss": 0.1418, "step": 9880 }, { "epoch": 95.81084337349398, "grad_norm": 8.583670616149902, "learning_rate": 5.203398058252428e-05, "loss": 0.2837, "step": 9881 }, { "epoch": 95.82048192771084, "grad_norm": 9.717113494873047, "learning_rate": 5.202912621359224e-05, "loss": 0.2116, "step": 9882 }, { "epoch": 95.83012048192771, "grad_norm": 5.26271915435791, "learning_rate": 5.20242718446602e-05, "loss": 0.2731, "step": 9883 }, { "epoch": 95.83975903614459, "grad_norm": 6.155111789703369, "learning_rate": 5.201941747572816e-05, "loss": 0.164, "step": 9884 }, { "epoch": 95.84939759036145, "grad_norm": 15.20146656036377, "learning_rate": 5.2014563106796124e-05, "loss": 0.2197, "step": 9885 }, { "epoch": 95.85903614457831, "grad_norm": 22.26154899597168, "learning_rate": 5.200970873786408e-05, "loss": 0.4072, "step": 9886 }, { "epoch": 95.86867469879518, "grad_norm": 8.386452674865723, "learning_rate": 5.2004854368932046e-05, "loss": 0.2888, "step": 9887 }, { "epoch": 95.87831325301205, "grad_norm": 31.16155242919922, "learning_rate": 5.2000000000000004e-05, "loss": 0.2885, "step": 9888 }, { "epoch": 95.88795180722892, "grad_norm": 6.447147369384766, "learning_rate": 5.199514563106797e-05, "loss": 0.3421, "step": 9889 }, { "epoch": 95.89759036144578, "grad_norm": 14.078880310058594, "learning_rate": 5.1990291262135926e-05, "loss": 0.2711, "step": 9890 }, { "epoch": 95.90722891566266, "grad_norm": 7.26760196685791, "learning_rate": 5.198543689320389e-05, "loss": 0.1611, "step": 9891 }, { "epoch": 95.91686746987952, "grad_norm": 4.8231072425842285, "learning_rate": 5.198058252427185e-05, "loss": 0.1549, "step": 9892 }, { "epoch": 95.92650602409638, "grad_norm": 7.849889755249023, "learning_rate": 5.1975728155339806e-05, "loss": 0.3831, "step": 9893 }, { "epoch": 95.93614457831325, "grad_norm": 3.9315707683563232, "learning_rate": 5.197087378640777e-05, "loss": 0.1068, "step": 9894 }, { "epoch": 95.94578313253012, "grad_norm": 7.616885185241699, "learning_rate": 5.196601941747573e-05, "loss": 0.2145, "step": 9895 }, { "epoch": 95.95542168674699, "grad_norm": 5.406538486480713, "learning_rate": 5.196116504854369e-05, "loss": 0.1964, "step": 9896 }, { "epoch": 95.96506024096385, "grad_norm": 14.335104942321777, "learning_rate": 5.195631067961165e-05, "loss": 0.1169, "step": 9897 }, { "epoch": 95.97469879518073, "grad_norm": 38.74266815185547, "learning_rate": 5.1951456310679615e-05, "loss": 0.386, "step": 9898 }, { "epoch": 95.98433734939759, "grad_norm": 9.92883014678955, "learning_rate": 5.194660194174757e-05, "loss": 0.3769, "step": 9899 }, { "epoch": 95.99397590361446, "grad_norm": 10.906198501586914, "learning_rate": 5.194174757281554e-05, "loss": 0.3938, "step": 9900 }, { "epoch": 96.00963855421686, "grad_norm": 9.681983947753906, "learning_rate": 5.1936893203883495e-05, "loss": 0.1678, "step": 9901 }, { "epoch": 96.01927710843374, "grad_norm": 7.616186141967773, "learning_rate": 5.193203883495146e-05, "loss": 0.2842, "step": 9902 }, { "epoch": 96.0289156626506, "grad_norm": 13.733665466308594, "learning_rate": 5.192718446601942e-05, "loss": 0.2042, "step": 9903 }, { "epoch": 96.03855421686747, "grad_norm": 8.219423294067383, "learning_rate": 5.192233009708738e-05, "loss": 0.4496, "step": 9904 }, { "epoch": 96.04819277108433, "grad_norm": 8.913348197937012, "learning_rate": 5.191747572815534e-05, "loss": 0.2412, "step": 9905 }, { "epoch": 96.05783132530121, "grad_norm": 8.106764793395996, "learning_rate": 5.1912621359223304e-05, "loss": 0.4754, "step": 9906 }, { "epoch": 96.06746987951807, "grad_norm": 7.741398811340332, "learning_rate": 5.190776699029126e-05, "loss": 0.243, "step": 9907 }, { "epoch": 96.07710843373494, "grad_norm": 8.008249282836914, "learning_rate": 5.1902912621359226e-05, "loss": 0.3329, "step": 9908 }, { "epoch": 96.08674698795181, "grad_norm": 10.308424949645996, "learning_rate": 5.1898058252427184e-05, "loss": 0.2236, "step": 9909 }, { "epoch": 96.09638554216868, "grad_norm": 8.740864753723145, "learning_rate": 5.189320388349514e-05, "loss": 0.2316, "step": 9910 }, { "epoch": 96.10602409638554, "grad_norm": 1.5133463144302368, "learning_rate": 5.1888349514563107e-05, "loss": 0.0403, "step": 9911 }, { "epoch": 96.1156626506024, "grad_norm": 5.859963893890381, "learning_rate": 5.1883495145631064e-05, "loss": 0.2192, "step": 9912 }, { "epoch": 96.12530120481928, "grad_norm": 8.739137649536133, "learning_rate": 5.187864077669903e-05, "loss": 0.1943, "step": 9913 }, { "epoch": 96.13493975903614, "grad_norm": 8.953062057495117, "learning_rate": 5.1873786407766987e-05, "loss": 0.2102, "step": 9914 }, { "epoch": 96.144578313253, "grad_norm": 27.095436096191406, "learning_rate": 5.186893203883495e-05, "loss": 0.4855, "step": 9915 }, { "epoch": 96.15421686746988, "grad_norm": 12.026534080505371, "learning_rate": 5.186407766990291e-05, "loss": 0.3311, "step": 9916 }, { "epoch": 96.16385542168675, "grad_norm": 8.363348007202148, "learning_rate": 5.1859223300970873e-05, "loss": 0.312, "step": 9917 }, { "epoch": 96.17349397590361, "grad_norm": 8.364023208618164, "learning_rate": 5.185436893203883e-05, "loss": 0.322, "step": 9918 }, { "epoch": 96.18313253012049, "grad_norm": 6.445942401885986, "learning_rate": 5.1849514563106796e-05, "loss": 0.236, "step": 9919 }, { "epoch": 96.19277108433735, "grad_norm": 6.513489246368408, "learning_rate": 5.1844660194174753e-05, "loss": 0.1426, "step": 9920 }, { "epoch": 96.20240963855422, "grad_norm": 12.709409713745117, "learning_rate": 5.183980582524272e-05, "loss": 0.4073, "step": 9921 }, { "epoch": 96.21204819277108, "grad_norm": 9.561139106750488, "learning_rate": 5.183495145631069e-05, "loss": 0.2681, "step": 9922 }, { "epoch": 96.22168674698796, "grad_norm": 4.941762924194336, "learning_rate": 5.183009708737865e-05, "loss": 0.2282, "step": 9923 }, { "epoch": 96.23132530120482, "grad_norm": 12.132017135620117, "learning_rate": 5.182524271844661e-05, "loss": 0.2746, "step": 9924 }, { "epoch": 96.24096385542168, "grad_norm": 6.163456439971924, "learning_rate": 5.182038834951457e-05, "loss": 0.2701, "step": 9925 }, { "epoch": 96.25060240963856, "grad_norm": 9.03679084777832, "learning_rate": 5.1815533980582534e-05, "loss": 0.4254, "step": 9926 }, { "epoch": 96.26024096385542, "grad_norm": 7.502760410308838, "learning_rate": 5.181067961165049e-05, "loss": 0.1785, "step": 9927 }, { "epoch": 96.26987951807229, "grad_norm": 5.5061798095703125, "learning_rate": 5.1805825242718456e-05, "loss": 0.1303, "step": 9928 }, { "epoch": 96.27951807228915, "grad_norm": 18.095930099487305, "learning_rate": 5.1800970873786414e-05, "loss": 0.3856, "step": 9929 }, { "epoch": 96.28915662650603, "grad_norm": 4.328673362731934, "learning_rate": 5.179611650485437e-05, "loss": 0.1517, "step": 9930 }, { "epoch": 96.29879518072289, "grad_norm": 13.861534118652344, "learning_rate": 5.1791262135922336e-05, "loss": 0.2384, "step": 9931 }, { "epoch": 96.30843373493975, "grad_norm": 8.024018287658691, "learning_rate": 5.1786407766990294e-05, "loss": 0.1641, "step": 9932 }, { "epoch": 96.31807228915663, "grad_norm": 6.037250995635986, "learning_rate": 5.178155339805826e-05, "loss": 0.2136, "step": 9933 }, { "epoch": 96.3277108433735, "grad_norm": 9.26766300201416, "learning_rate": 5.1776699029126216e-05, "loss": 0.1552, "step": 9934 }, { "epoch": 96.33734939759036, "grad_norm": 13.352226257324219, "learning_rate": 5.177184466019418e-05, "loss": 0.2584, "step": 9935 }, { "epoch": 96.34698795180722, "grad_norm": 30.548917770385742, "learning_rate": 5.176699029126214e-05, "loss": 0.2988, "step": 9936 }, { "epoch": 96.3566265060241, "grad_norm": 9.918641090393066, "learning_rate": 5.17621359223301e-05, "loss": 0.3411, "step": 9937 }, { "epoch": 96.36626506024096, "grad_norm": 11.916431427001953, "learning_rate": 5.175728155339806e-05, "loss": 0.2105, "step": 9938 }, { "epoch": 96.37590361445783, "grad_norm": 13.252187728881836, "learning_rate": 5.1752427184466025e-05, "loss": 0.3431, "step": 9939 }, { "epoch": 96.3855421686747, "grad_norm": 6.052306652069092, "learning_rate": 5.174757281553398e-05, "loss": 0.2303, "step": 9940 }, { "epoch": 96.39518072289157, "grad_norm": 7.7551350593566895, "learning_rate": 5.174271844660195e-05, "loss": 0.2517, "step": 9941 }, { "epoch": 96.40481927710843, "grad_norm": 10.891019821166992, "learning_rate": 5.1737864077669905e-05, "loss": 0.2402, "step": 9942 }, { "epoch": 96.41445783132531, "grad_norm": 7.993810176849365, "learning_rate": 5.173300970873787e-05, "loss": 0.2187, "step": 9943 }, { "epoch": 96.42409638554217, "grad_norm": 9.344766616821289, "learning_rate": 5.172815533980583e-05, "loss": 0.2771, "step": 9944 }, { "epoch": 96.43373493975903, "grad_norm": 6.269413948059082, "learning_rate": 5.172330097087379e-05, "loss": 0.2419, "step": 9945 }, { "epoch": 96.4433734939759, "grad_norm": 7.057982921600342, "learning_rate": 5.171844660194175e-05, "loss": 0.3199, "step": 9946 }, { "epoch": 96.45301204819278, "grad_norm": 12.9371337890625, "learning_rate": 5.171359223300971e-05, "loss": 0.309, "step": 9947 }, { "epoch": 96.46265060240964, "grad_norm": 6.097620487213135, "learning_rate": 5.170873786407767e-05, "loss": 0.2604, "step": 9948 }, { "epoch": 96.4722891566265, "grad_norm": 17.52774429321289, "learning_rate": 5.170388349514563e-05, "loss": 0.4206, "step": 9949 }, { "epoch": 96.48192771084338, "grad_norm": 13.020968437194824, "learning_rate": 5.1699029126213595e-05, "loss": 0.4299, "step": 9950 }, { "epoch": 96.49156626506024, "grad_norm": 9.561925888061523, "learning_rate": 5.169417475728155e-05, "loss": 0.3951, "step": 9951 }, { "epoch": 96.5012048192771, "grad_norm": 14.843066215515137, "learning_rate": 5.168932038834952e-05, "loss": 0.2337, "step": 9952 }, { "epoch": 96.51084337349397, "grad_norm": 9.708454132080078, "learning_rate": 5.1684466019417475e-05, "loss": 0.1923, "step": 9953 }, { "epoch": 96.52048192771085, "grad_norm": 10.689424514770508, "learning_rate": 5.167961165048544e-05, "loss": 0.3693, "step": 9954 }, { "epoch": 96.53012048192771, "grad_norm": 8.715471267700195, "learning_rate": 5.16747572815534e-05, "loss": 0.3512, "step": 9955 }, { "epoch": 96.53975903614457, "grad_norm": 5.788638114929199, "learning_rate": 5.166990291262136e-05, "loss": 0.2092, "step": 9956 }, { "epoch": 96.54939759036145, "grad_norm": 7.765910625457764, "learning_rate": 5.166504854368932e-05, "loss": 0.1429, "step": 9957 }, { "epoch": 96.55903614457831, "grad_norm": 10.194153785705566, "learning_rate": 5.1660194174757284e-05, "loss": 0.3955, "step": 9958 }, { "epoch": 96.56867469879518, "grad_norm": 8.628966331481934, "learning_rate": 5.165533980582524e-05, "loss": 0.4268, "step": 9959 }, { "epoch": 96.57831325301204, "grad_norm": 15.414769172668457, "learning_rate": 5.1650485436893206e-05, "loss": 0.3308, "step": 9960 }, { "epoch": 96.58795180722892, "grad_norm": 7.654146671295166, "learning_rate": 5.1645631067961164e-05, "loss": 0.2379, "step": 9961 }, { "epoch": 96.59759036144578, "grad_norm": 7.9200921058654785, "learning_rate": 5.164077669902913e-05, "loss": 0.1207, "step": 9962 }, { "epoch": 96.60722891566265, "grad_norm": 18.26259422302246, "learning_rate": 5.1635922330097086e-05, "loss": 0.2002, "step": 9963 }, { "epoch": 96.61686746987952, "grad_norm": 9.81106185913086, "learning_rate": 5.1631067961165044e-05, "loss": 0.3261, "step": 9964 }, { "epoch": 96.62650602409639, "grad_norm": 8.900435447692871, "learning_rate": 5.162621359223301e-05, "loss": 0.2345, "step": 9965 }, { "epoch": 96.63614457831325, "grad_norm": 6.918953895568848, "learning_rate": 5.1621359223300966e-05, "loss": 0.2679, "step": 9966 }, { "epoch": 96.64578313253013, "grad_norm": 5.550777435302734, "learning_rate": 5.161650485436893e-05, "loss": 0.1341, "step": 9967 }, { "epoch": 96.65542168674699, "grad_norm": 15.252979278564453, "learning_rate": 5.161165048543689e-05, "loss": 0.1052, "step": 9968 }, { "epoch": 96.66506024096385, "grad_norm": 13.112597465515137, "learning_rate": 5.160679611650485e-05, "loss": 0.5678, "step": 9969 }, { "epoch": 96.67469879518072, "grad_norm": 9.998189926147461, "learning_rate": 5.160194174757281e-05, "loss": 0.243, "step": 9970 }, { "epoch": 96.6843373493976, "grad_norm": 8.86978816986084, "learning_rate": 5.1597087378640775e-05, "loss": 0.2168, "step": 9971 }, { "epoch": 96.69397590361446, "grad_norm": 9.269722938537598, "learning_rate": 5.1592233009708747e-05, "loss": 0.3494, "step": 9972 }, { "epoch": 96.70361445783132, "grad_norm": 16.79107666015625, "learning_rate": 5.1587378640776704e-05, "loss": 0.5806, "step": 9973 }, { "epoch": 96.7132530120482, "grad_norm": 11.986093521118164, "learning_rate": 5.158252427184467e-05, "loss": 0.4893, "step": 9974 }, { "epoch": 96.72289156626506, "grad_norm": 14.172356605529785, "learning_rate": 5.157766990291263e-05, "loss": 0.2921, "step": 9975 }, { "epoch": 96.73253012048193, "grad_norm": 6.852723598480225, "learning_rate": 5.157281553398059e-05, "loss": 0.3011, "step": 9976 }, { "epoch": 96.74216867469879, "grad_norm": 8.28948974609375, "learning_rate": 5.156796116504855e-05, "loss": 0.1321, "step": 9977 }, { "epoch": 96.75180722891567, "grad_norm": 10.289970397949219, "learning_rate": 5.1563106796116513e-05, "loss": 0.4101, "step": 9978 }, { "epoch": 96.76144578313253, "grad_norm": 9.115139961242676, "learning_rate": 5.155825242718447e-05, "loss": 0.1525, "step": 9979 }, { "epoch": 96.7710843373494, "grad_norm": 7.7702765464782715, "learning_rate": 5.1553398058252436e-05, "loss": 0.383, "step": 9980 }, { "epoch": 96.78072289156627, "grad_norm": 10.425809860229492, "learning_rate": 5.1548543689320394e-05, "loss": 0.3171, "step": 9981 }, { "epoch": 96.79036144578313, "grad_norm": 8.091211318969727, "learning_rate": 5.154368932038836e-05, "loss": 0.3284, "step": 9982 }, { "epoch": 96.8, "grad_norm": 11.894079208374023, "learning_rate": 5.1538834951456316e-05, "loss": 0.4028, "step": 9983 }, { "epoch": 96.80963855421686, "grad_norm": 10.629579544067383, "learning_rate": 5.153398058252428e-05, "loss": 0.3187, "step": 9984 }, { "epoch": 96.81927710843374, "grad_norm": 9.146539688110352, "learning_rate": 5.152912621359224e-05, "loss": 0.3351, "step": 9985 }, { "epoch": 96.8289156626506, "grad_norm": 9.046882629394531, "learning_rate": 5.1524271844660196e-05, "loss": 0.3456, "step": 9986 }, { "epoch": 96.83855421686746, "grad_norm": 11.286426544189453, "learning_rate": 5.151941747572816e-05, "loss": 0.1664, "step": 9987 }, { "epoch": 96.84819277108434, "grad_norm": 10.050975799560547, "learning_rate": 5.151456310679612e-05, "loss": 0.1322, "step": 9988 }, { "epoch": 96.8578313253012, "grad_norm": 6.676712989807129, "learning_rate": 5.150970873786408e-05, "loss": 0.3072, "step": 9989 }, { "epoch": 96.86746987951807, "grad_norm": 6.5277581214904785, "learning_rate": 5.150485436893204e-05, "loss": 0.1944, "step": 9990 }, { "epoch": 96.87710843373495, "grad_norm": 8.465824127197266, "learning_rate": 5.1500000000000005e-05, "loss": 0.2284, "step": 9991 }, { "epoch": 96.88674698795181, "grad_norm": 13.295183181762695, "learning_rate": 5.149514563106796e-05, "loss": 0.378, "step": 9992 }, { "epoch": 96.89638554216867, "grad_norm": 4.543571472167969, "learning_rate": 5.149029126213593e-05, "loss": 0.1197, "step": 9993 }, { "epoch": 96.90602409638554, "grad_norm": 7.569888114929199, "learning_rate": 5.1485436893203885e-05, "loss": 0.1925, "step": 9994 }, { "epoch": 96.91566265060241, "grad_norm": 10.204425811767578, "learning_rate": 5.148058252427185e-05, "loss": 0.2734, "step": 9995 }, { "epoch": 96.92530120481928, "grad_norm": 18.5592041015625, "learning_rate": 5.147572815533981e-05, "loss": 0.4027, "step": 9996 }, { "epoch": 96.93493975903614, "grad_norm": 10.114943504333496, "learning_rate": 5.147087378640777e-05, "loss": 0.3033, "step": 9997 }, { "epoch": 96.94457831325302, "grad_norm": 9.079584121704102, "learning_rate": 5.146601941747573e-05, "loss": 0.2189, "step": 9998 }, { "epoch": 96.95421686746988, "grad_norm": 11.393102645874023, "learning_rate": 5.1461165048543694e-05, "loss": 0.23, "step": 9999 }, { "epoch": 96.96385542168674, "grad_norm": 10.833122253417969, "learning_rate": 5.145631067961165e-05, "loss": 0.4964, "step": 10000 }, { "epoch": 96.97349397590361, "grad_norm": 6.865086078643799, "learning_rate": 5.1451456310679616e-05, "loss": 0.1391, "step": 10001 }, { "epoch": 96.98313253012049, "grad_norm": 7.726236820220947, "learning_rate": 5.1446601941747574e-05, "loss": 0.213, "step": 10002 }, { "epoch": 96.99277108433735, "grad_norm": 12.727323532104492, "learning_rate": 5.144174757281553e-05, "loss": 0.3076, "step": 10003 }, { "epoch": 97.00843373493976, "grad_norm": 18.474916458129883, "learning_rate": 5.1436893203883496e-05, "loss": 0.2245, "step": 10004 }, { "epoch": 97.01807228915662, "grad_norm": 17.435001373291016, "learning_rate": 5.1432038834951454e-05, "loss": 0.2002, "step": 10005 }, { "epoch": 97.0277108433735, "grad_norm": 3.8189544677734375, "learning_rate": 5.142718446601942e-05, "loss": 0.1569, "step": 10006 }, { "epoch": 97.03734939759036, "grad_norm": 13.159899711608887, "learning_rate": 5.1422330097087377e-05, "loss": 0.1292, "step": 10007 }, { "epoch": 97.04698795180722, "grad_norm": 22.9268741607666, "learning_rate": 5.141747572815534e-05, "loss": 0.2713, "step": 10008 }, { "epoch": 97.0566265060241, "grad_norm": 8.06573486328125, "learning_rate": 5.14126213592233e-05, "loss": 0.1571, "step": 10009 }, { "epoch": 97.06626506024097, "grad_norm": 32.408348083496094, "learning_rate": 5.140776699029126e-05, "loss": 0.344, "step": 10010 }, { "epoch": 97.07590361445783, "grad_norm": 5.819936275482178, "learning_rate": 5.140291262135922e-05, "loss": 0.1947, "step": 10011 }, { "epoch": 97.08554216867469, "grad_norm": 7.327682018280029, "learning_rate": 5.1398058252427186e-05, "loss": 0.3176, "step": 10012 }, { "epoch": 97.09518072289157, "grad_norm": 4.926824569702148, "learning_rate": 5.1393203883495143e-05, "loss": 0.2896, "step": 10013 }, { "epoch": 97.10481927710843, "grad_norm": 70.73949432373047, "learning_rate": 5.138834951456311e-05, "loss": 0.4084, "step": 10014 }, { "epoch": 97.1144578313253, "grad_norm": 11.049256324768066, "learning_rate": 5.1383495145631066e-05, "loss": 0.306, "step": 10015 }, { "epoch": 97.12409638554217, "grad_norm": 23.220325469970703, "learning_rate": 5.137864077669903e-05, "loss": 0.3546, "step": 10016 }, { "epoch": 97.13373493975904, "grad_norm": 14.937435150146484, "learning_rate": 5.137378640776699e-05, "loss": 0.265, "step": 10017 }, { "epoch": 97.1433734939759, "grad_norm": 13.057735443115234, "learning_rate": 5.136893203883495e-05, "loss": 0.314, "step": 10018 }, { "epoch": 97.15301204819278, "grad_norm": 8.658769607543945, "learning_rate": 5.136407766990291e-05, "loss": 0.552, "step": 10019 }, { "epoch": 97.16265060240964, "grad_norm": 10.18995189666748, "learning_rate": 5.135922330097087e-05, "loss": 0.3228, "step": 10020 }, { "epoch": 97.1722891566265, "grad_norm": 13.913398742675781, "learning_rate": 5.135436893203883e-05, "loss": 0.123, "step": 10021 }, { "epoch": 97.18192771084337, "grad_norm": 3.3894693851470947, "learning_rate": 5.134951456310679e-05, "loss": 0.1799, "step": 10022 }, { "epoch": 97.19156626506025, "grad_norm": 2.790172576904297, "learning_rate": 5.134466019417477e-05, "loss": 0.1773, "step": 10023 }, { "epoch": 97.20120481927711, "grad_norm": 18.69957733154297, "learning_rate": 5.1339805825242726e-05, "loss": 0.4697, "step": 10024 }, { "epoch": 97.21084337349397, "grad_norm": 16.568254470825195, "learning_rate": 5.1334951456310684e-05, "loss": 0.29, "step": 10025 }, { "epoch": 97.22048192771085, "grad_norm": 12.55515193939209, "learning_rate": 5.133009708737865e-05, "loss": 0.2853, "step": 10026 }, { "epoch": 97.23012048192771, "grad_norm": 12.92197036743164, "learning_rate": 5.1325242718446606e-05, "loss": 0.2984, "step": 10027 }, { "epoch": 97.23975903614458, "grad_norm": 23.229490280151367, "learning_rate": 5.132038834951457e-05, "loss": 0.2803, "step": 10028 }, { "epoch": 97.24939759036144, "grad_norm": 6.095532417297363, "learning_rate": 5.131553398058253e-05, "loss": 0.1118, "step": 10029 }, { "epoch": 97.25903614457832, "grad_norm": 4.579070568084717, "learning_rate": 5.131067961165049e-05, "loss": 0.3984, "step": 10030 }, { "epoch": 97.26867469879518, "grad_norm": 24.294187545776367, "learning_rate": 5.130582524271845e-05, "loss": 0.3355, "step": 10031 }, { "epoch": 97.27831325301204, "grad_norm": 14.413946151733398, "learning_rate": 5.1300970873786415e-05, "loss": 0.307, "step": 10032 }, { "epoch": 97.28795180722892, "grad_norm": 3.248934268951416, "learning_rate": 5.129611650485437e-05, "loss": 0.1836, "step": 10033 }, { "epoch": 97.29759036144578, "grad_norm": 41.380069732666016, "learning_rate": 5.129126213592234e-05, "loss": 0.7152, "step": 10034 }, { "epoch": 97.30722891566265, "grad_norm": 18.477996826171875, "learning_rate": 5.1286407766990295e-05, "loss": 0.1476, "step": 10035 }, { "epoch": 97.31686746987951, "grad_norm": 9.595105171203613, "learning_rate": 5.128155339805826e-05, "loss": 0.2901, "step": 10036 }, { "epoch": 97.32650602409639, "grad_norm": 5.103181838989258, "learning_rate": 5.127669902912622e-05, "loss": 0.3094, "step": 10037 }, { "epoch": 97.33614457831325, "grad_norm": 15.541440963745117, "learning_rate": 5.127184466019418e-05, "loss": 0.3051, "step": 10038 }, { "epoch": 97.34578313253012, "grad_norm": 7.088634490966797, "learning_rate": 5.126699029126214e-05, "loss": 0.1712, "step": 10039 }, { "epoch": 97.355421686747, "grad_norm": 3.538806915283203, "learning_rate": 5.1262135922330105e-05, "loss": 0.2949, "step": 10040 }, { "epoch": 97.36506024096386, "grad_norm": 38.343807220458984, "learning_rate": 5.125728155339806e-05, "loss": 0.287, "step": 10041 }, { "epoch": 97.37469879518072, "grad_norm": 5.180109977722168, "learning_rate": 5.125242718446602e-05, "loss": 0.2085, "step": 10042 }, { "epoch": 97.38433734939758, "grad_norm": 5.563335418701172, "learning_rate": 5.1247572815533985e-05, "loss": 0.2334, "step": 10043 }, { "epoch": 97.39397590361446, "grad_norm": 18.859859466552734, "learning_rate": 5.124271844660194e-05, "loss": 0.5026, "step": 10044 }, { "epoch": 97.40361445783132, "grad_norm": 11.555105209350586, "learning_rate": 5.123786407766991e-05, "loss": 0.5206, "step": 10045 }, { "epoch": 97.41325301204819, "grad_norm": 23.12689971923828, "learning_rate": 5.1233009708737865e-05, "loss": 0.2382, "step": 10046 }, { "epoch": 97.42289156626506, "grad_norm": 20.13006019592285, "learning_rate": 5.122815533980583e-05, "loss": 0.2898, "step": 10047 }, { "epoch": 97.43253012048193, "grad_norm": 20.91410255432129, "learning_rate": 5.122330097087379e-05, "loss": 0.2507, "step": 10048 }, { "epoch": 97.44216867469879, "grad_norm": 23.65647315979004, "learning_rate": 5.121844660194175e-05, "loss": 0.228, "step": 10049 }, { "epoch": 97.45180722891567, "grad_norm": 32.67561721801758, "learning_rate": 5.121359223300971e-05, "loss": 0.4648, "step": 10050 }, { "epoch": 97.46144578313253, "grad_norm": 7.145797252655029, "learning_rate": 5.1208737864077674e-05, "loss": 0.2256, "step": 10051 }, { "epoch": 97.4710843373494, "grad_norm": 8.158042907714844, "learning_rate": 5.120388349514563e-05, "loss": 0.3012, "step": 10052 }, { "epoch": 97.48072289156626, "grad_norm": 7.802107334136963, "learning_rate": 5.1199029126213596e-05, "loss": 0.4799, "step": 10053 }, { "epoch": 97.49036144578314, "grad_norm": 20.82267189025879, "learning_rate": 5.1194174757281554e-05, "loss": 0.4002, "step": 10054 }, { "epoch": 97.5, "grad_norm": 3.7326173782348633, "learning_rate": 5.118932038834952e-05, "loss": 0.3297, "step": 10055 }, { "epoch": 97.50963855421686, "grad_norm": 15.473671913146973, "learning_rate": 5.1184466019417476e-05, "loss": 0.336, "step": 10056 }, { "epoch": 97.51927710843374, "grad_norm": 9.664947509765625, "learning_rate": 5.117961165048544e-05, "loss": 0.1298, "step": 10057 }, { "epoch": 97.5289156626506, "grad_norm": 34.39376449584961, "learning_rate": 5.11747572815534e-05, "loss": 0.3341, "step": 10058 }, { "epoch": 97.53855421686747, "grad_norm": 27.8770694732666, "learning_rate": 5.1169902912621356e-05, "loss": 0.1418, "step": 10059 }, { "epoch": 97.54819277108433, "grad_norm": 8.260767936706543, "learning_rate": 5.116504854368932e-05, "loss": 0.3078, "step": 10060 }, { "epoch": 97.55783132530121, "grad_norm": 14.039250373840332, "learning_rate": 5.116019417475728e-05, "loss": 0.4226, "step": 10061 }, { "epoch": 97.56746987951807, "grad_norm": 7.039376735687256, "learning_rate": 5.115533980582524e-05, "loss": 0.0982, "step": 10062 }, { "epoch": 97.57710843373494, "grad_norm": 12.853157997131348, "learning_rate": 5.11504854368932e-05, "loss": 0.2754, "step": 10063 }, { "epoch": 97.58674698795181, "grad_norm": 11.067826271057129, "learning_rate": 5.1145631067961165e-05, "loss": 0.1456, "step": 10064 }, { "epoch": 97.59638554216868, "grad_norm": 22.35784912109375, "learning_rate": 5.114077669902912e-05, "loss": 0.3061, "step": 10065 }, { "epoch": 97.60602409638554, "grad_norm": 4.981251239776611, "learning_rate": 5.113592233009709e-05, "loss": 0.3995, "step": 10066 }, { "epoch": 97.61566265060242, "grad_norm": 7.392004489898682, "learning_rate": 5.1131067961165045e-05, "loss": 0.3441, "step": 10067 }, { "epoch": 97.62530120481928, "grad_norm": 43.27710723876953, "learning_rate": 5.112621359223301e-05, "loss": 0.4153, "step": 10068 }, { "epoch": 97.63493975903614, "grad_norm": 24.224531173706055, "learning_rate": 5.112135922330097e-05, "loss": 0.2262, "step": 10069 }, { "epoch": 97.644578313253, "grad_norm": 3.472140073776245, "learning_rate": 5.111650485436893e-05, "loss": 0.2702, "step": 10070 }, { "epoch": 97.65421686746988, "grad_norm": 7.853805065155029, "learning_rate": 5.111165048543689e-05, "loss": 0.3195, "step": 10071 }, { "epoch": 97.66385542168675, "grad_norm": 26.17398452758789, "learning_rate": 5.1106796116504854e-05, "loss": 0.197, "step": 10072 }, { "epoch": 97.67349397590361, "grad_norm": 3.561830520629883, "learning_rate": 5.1101941747572826e-05, "loss": 0.1828, "step": 10073 }, { "epoch": 97.68313253012049, "grad_norm": 8.1138916015625, "learning_rate": 5.1097087378640783e-05, "loss": 0.1492, "step": 10074 }, { "epoch": 97.69277108433735, "grad_norm": 14.289551734924316, "learning_rate": 5.109223300970875e-05, "loss": 0.3802, "step": 10075 }, { "epoch": 97.70240963855422, "grad_norm": 8.827923774719238, "learning_rate": 5.1087378640776706e-05, "loss": 0.2102, "step": 10076 }, { "epoch": 97.71204819277108, "grad_norm": 17.822816848754883, "learning_rate": 5.108252427184467e-05, "loss": 0.2759, "step": 10077 }, { "epoch": 97.72168674698796, "grad_norm": 29.836320877075195, "learning_rate": 5.107766990291263e-05, "loss": 0.2, "step": 10078 }, { "epoch": 97.73132530120482, "grad_norm": 19.039745330810547, "learning_rate": 5.1072815533980586e-05, "loss": 0.2379, "step": 10079 }, { "epoch": 97.74096385542168, "grad_norm": 21.755435943603516, "learning_rate": 5.106796116504855e-05, "loss": 0.2309, "step": 10080 }, { "epoch": 97.75060240963856, "grad_norm": 2.351076126098633, "learning_rate": 5.106310679611651e-05, "loss": 0.2207, "step": 10081 }, { "epoch": 97.76024096385542, "grad_norm": 3.8445491790771484, "learning_rate": 5.105825242718447e-05, "loss": 0.2849, "step": 10082 }, { "epoch": 97.76987951807229, "grad_norm": 8.83578872680664, "learning_rate": 5.105339805825243e-05, "loss": 0.2353, "step": 10083 }, { "epoch": 97.77951807228915, "grad_norm": 25.126649856567383, "learning_rate": 5.1048543689320395e-05, "loss": 0.144, "step": 10084 }, { "epoch": 97.78915662650603, "grad_norm": 55.76144790649414, "learning_rate": 5.104368932038835e-05, "loss": 0.3273, "step": 10085 }, { "epoch": 97.79879518072289, "grad_norm": 22.719879150390625, "learning_rate": 5.103883495145632e-05, "loss": 0.1825, "step": 10086 }, { "epoch": 97.80843373493975, "grad_norm": 3.4946115016937256, "learning_rate": 5.1033980582524275e-05, "loss": 0.2692, "step": 10087 }, { "epoch": 97.81807228915663, "grad_norm": 16.455528259277344, "learning_rate": 5.102912621359224e-05, "loss": 0.3341, "step": 10088 }, { "epoch": 97.8277108433735, "grad_norm": 6.357410907745361, "learning_rate": 5.10242718446602e-05, "loss": 0.1757, "step": 10089 }, { "epoch": 97.83734939759036, "grad_norm": 16.728755950927734, "learning_rate": 5.101941747572816e-05, "loss": 0.1923, "step": 10090 }, { "epoch": 97.84698795180722, "grad_norm": 24.2420654296875, "learning_rate": 5.101456310679612e-05, "loss": 0.282, "step": 10091 }, { "epoch": 97.8566265060241, "grad_norm": 106.06721496582031, "learning_rate": 5.1009708737864084e-05, "loss": 0.3088, "step": 10092 }, { "epoch": 97.86626506024096, "grad_norm": 24.20090103149414, "learning_rate": 5.100485436893204e-05, "loss": 0.2753, "step": 10093 }, { "epoch": 97.87590361445783, "grad_norm": 33.32905960083008, "learning_rate": 5.1000000000000006e-05, "loss": 0.3399, "step": 10094 }, { "epoch": 97.8855421686747, "grad_norm": 6.919022083282471, "learning_rate": 5.0995145631067964e-05, "loss": 0.4042, "step": 10095 }, { "epoch": 97.89518072289157, "grad_norm": 33.34011459350586, "learning_rate": 5.099029126213592e-05, "loss": 0.349, "step": 10096 }, { "epoch": 97.90481927710843, "grad_norm": 5.974413871765137, "learning_rate": 5.0985436893203886e-05, "loss": 0.1343, "step": 10097 }, { "epoch": 97.91445783132531, "grad_norm": 18.19322967529297, "learning_rate": 5.0980582524271844e-05, "loss": 0.3506, "step": 10098 }, { "epoch": 97.92409638554217, "grad_norm": 5.2252302169799805, "learning_rate": 5.097572815533981e-05, "loss": 0.2712, "step": 10099 }, { "epoch": 97.93373493975903, "grad_norm": 58.58637619018555, "learning_rate": 5.0970873786407766e-05, "loss": 0.5191, "step": 10100 }, { "epoch": 97.9433734939759, "grad_norm": 8.86467456817627, "learning_rate": 5.096601941747573e-05, "loss": 0.1906, "step": 10101 }, { "epoch": 97.95301204819278, "grad_norm": 2.881122589111328, "learning_rate": 5.096116504854369e-05, "loss": 0.2314, "step": 10102 }, { "epoch": 97.96265060240964, "grad_norm": 13.280117988586426, "learning_rate": 5.095631067961165e-05, "loss": 0.1955, "step": 10103 }, { "epoch": 97.9722891566265, "grad_norm": 18.098255157470703, "learning_rate": 5.095145631067961e-05, "loss": 0.2278, "step": 10104 }, { "epoch": 97.98192771084338, "grad_norm": 16.89056968688965, "learning_rate": 5.0946601941747576e-05, "loss": 0.3972, "step": 10105 }, { "epoch": 97.99156626506024, "grad_norm": 49.97652816772461, "learning_rate": 5.094174757281553e-05, "loss": 0.2055, "step": 10106 }, { "epoch": 98.00722891566265, "grad_norm": 15.705406188964844, "learning_rate": 5.09368932038835e-05, "loss": 0.1969, "step": 10107 }, { "epoch": 98.01686746987951, "grad_norm": 23.890642166137695, "learning_rate": 5.0932038834951456e-05, "loss": 0.3168, "step": 10108 }, { "epoch": 98.02650602409639, "grad_norm": 15.747862815856934, "learning_rate": 5.092718446601942e-05, "loss": 0.3921, "step": 10109 }, { "epoch": 98.03614457831326, "grad_norm": 22.00709342956543, "learning_rate": 5.092233009708738e-05, "loss": 0.0866, "step": 10110 }, { "epoch": 98.04578313253012, "grad_norm": 5.926250457763672, "learning_rate": 5.091747572815534e-05, "loss": 0.166, "step": 10111 }, { "epoch": 98.05542168674698, "grad_norm": 4.247297286987305, "learning_rate": 5.09126213592233e-05, "loss": 0.1294, "step": 10112 }, { "epoch": 98.06506024096386, "grad_norm": 3.293692111968994, "learning_rate": 5.090776699029126e-05, "loss": 0.2278, "step": 10113 }, { "epoch": 98.07469879518072, "grad_norm": 4.363041877746582, "learning_rate": 5.090291262135922e-05, "loss": 0.1884, "step": 10114 }, { "epoch": 98.08433734939759, "grad_norm": 251.21006774902344, "learning_rate": 5.089805825242718e-05, "loss": 0.4517, "step": 10115 }, { "epoch": 98.09397590361446, "grad_norm": 41.77937698364258, "learning_rate": 5.0893203883495145e-05, "loss": 0.4863, "step": 10116 }, { "epoch": 98.10361445783133, "grad_norm": 3.98280668258667, "learning_rate": 5.08883495145631e-05, "loss": 0.2355, "step": 10117 }, { "epoch": 98.11325301204819, "grad_norm": 5.513203144073486, "learning_rate": 5.088349514563107e-05, "loss": 0.2749, "step": 10118 }, { "epoch": 98.12289156626505, "grad_norm": 8.410531044006348, "learning_rate": 5.0878640776699025e-05, "loss": 0.199, "step": 10119 }, { "epoch": 98.13253012048193, "grad_norm": 12.537510871887207, "learning_rate": 5.087378640776699e-05, "loss": 0.2895, "step": 10120 }, { "epoch": 98.1421686746988, "grad_norm": 9.82150650024414, "learning_rate": 5.086893203883495e-05, "loss": 0.1733, "step": 10121 }, { "epoch": 98.15180722891566, "grad_norm": 1.7726107835769653, "learning_rate": 5.086407766990291e-05, "loss": 0.1736, "step": 10122 }, { "epoch": 98.16144578313254, "grad_norm": 28.061342239379883, "learning_rate": 5.085922330097087e-05, "loss": 0.2897, "step": 10123 }, { "epoch": 98.1710843373494, "grad_norm": 33.076568603515625, "learning_rate": 5.085436893203884e-05, "loss": 0.3505, "step": 10124 }, { "epoch": 98.18072289156626, "grad_norm": 38.45525360107422, "learning_rate": 5.0849514563106805e-05, "loss": 0.2175, "step": 10125 }, { "epoch": 98.19036144578314, "grad_norm": 14.992199897766113, "learning_rate": 5.084466019417476e-05, "loss": 0.4688, "step": 10126 }, { "epoch": 98.2, "grad_norm": 7.246948719024658, "learning_rate": 5.083980582524273e-05, "loss": 0.3042, "step": 10127 }, { "epoch": 98.20963855421687, "grad_norm": 7.101062297821045, "learning_rate": 5.0834951456310685e-05, "loss": 0.307, "step": 10128 }, { "epoch": 98.21927710843373, "grad_norm": 15.19758415222168, "learning_rate": 5.083009708737865e-05, "loss": 0.3598, "step": 10129 }, { "epoch": 98.2289156626506, "grad_norm": 18.983598709106445, "learning_rate": 5.082524271844661e-05, "loss": 0.3836, "step": 10130 }, { "epoch": 98.23855421686747, "grad_norm": 11.920469284057617, "learning_rate": 5.082038834951457e-05, "loss": 0.3508, "step": 10131 }, { "epoch": 98.24819277108433, "grad_norm": 14.012733459472656, "learning_rate": 5.081553398058253e-05, "loss": 0.232, "step": 10132 }, { "epoch": 98.25783132530121, "grad_norm": 11.954384803771973, "learning_rate": 5.0810679611650494e-05, "loss": 0.2198, "step": 10133 }, { "epoch": 98.26746987951807, "grad_norm": 24.509729385375977, "learning_rate": 5.080582524271845e-05, "loss": 0.2439, "step": 10134 }, { "epoch": 98.27710843373494, "grad_norm": 12.742685317993164, "learning_rate": 5.080097087378641e-05, "loss": 0.277, "step": 10135 }, { "epoch": 98.2867469879518, "grad_norm": 6.687498092651367, "learning_rate": 5.0796116504854374e-05, "loss": 0.2294, "step": 10136 }, { "epoch": 98.29638554216868, "grad_norm": 18.63869285583496, "learning_rate": 5.079126213592233e-05, "loss": 0.1578, "step": 10137 }, { "epoch": 98.30602409638554, "grad_norm": 15.524292945861816, "learning_rate": 5.07864077669903e-05, "loss": 0.4838, "step": 10138 }, { "epoch": 98.3156626506024, "grad_norm": 19.803239822387695, "learning_rate": 5.0781553398058255e-05, "loss": 0.2353, "step": 10139 }, { "epoch": 98.32530120481928, "grad_norm": 19.808942794799805, "learning_rate": 5.077669902912622e-05, "loss": 0.2306, "step": 10140 }, { "epoch": 98.33493975903615, "grad_norm": 5.234460830688477, "learning_rate": 5.077184466019418e-05, "loss": 0.1547, "step": 10141 }, { "epoch": 98.34457831325301, "grad_norm": 10.646251678466797, "learning_rate": 5.076699029126214e-05, "loss": 0.1544, "step": 10142 }, { "epoch": 98.35421686746987, "grad_norm": 4.577619552612305, "learning_rate": 5.07621359223301e-05, "loss": 0.3407, "step": 10143 }, { "epoch": 98.36385542168675, "grad_norm": 6.860701560974121, "learning_rate": 5.0757281553398064e-05, "loss": 0.2617, "step": 10144 }, { "epoch": 98.37349397590361, "grad_norm": 6.691125392913818, "learning_rate": 5.075242718446602e-05, "loss": 0.2467, "step": 10145 }, { "epoch": 98.38313253012048, "grad_norm": 42.962806701660156, "learning_rate": 5.0747572815533986e-05, "loss": 0.4634, "step": 10146 }, { "epoch": 98.39277108433735, "grad_norm": 19.03033447265625, "learning_rate": 5.0742718446601944e-05, "loss": 0.5173, "step": 10147 }, { "epoch": 98.40240963855422, "grad_norm": 12.753935813903809, "learning_rate": 5.073786407766991e-05, "loss": 0.2352, "step": 10148 }, { "epoch": 98.41204819277108, "grad_norm": 46.91583251953125, "learning_rate": 5.0733009708737866e-05, "loss": 0.1908, "step": 10149 }, { "epoch": 98.42168674698796, "grad_norm": 2.906348705291748, "learning_rate": 5.072815533980583e-05, "loss": 0.2151, "step": 10150 }, { "epoch": 98.43132530120482, "grad_norm": 15.889073371887207, "learning_rate": 5.072330097087379e-05, "loss": 0.3212, "step": 10151 }, { "epoch": 98.44096385542169, "grad_norm": 14.995867729187012, "learning_rate": 5.0718446601941746e-05, "loss": 0.1678, "step": 10152 }, { "epoch": 98.45060240963855, "grad_norm": 5.042940139770508, "learning_rate": 5.071359223300971e-05, "loss": 0.2885, "step": 10153 }, { "epoch": 98.46024096385543, "grad_norm": 14.167367935180664, "learning_rate": 5.070873786407767e-05, "loss": 0.2488, "step": 10154 }, { "epoch": 98.46987951807229, "grad_norm": 10.83606243133545, "learning_rate": 5.070388349514563e-05, "loss": 0.2047, "step": 10155 }, { "epoch": 98.47951807228915, "grad_norm": 19.014537811279297, "learning_rate": 5.069902912621359e-05, "loss": 0.1943, "step": 10156 }, { "epoch": 98.48915662650603, "grad_norm": 43.42509841918945, "learning_rate": 5.0694174757281555e-05, "loss": 0.2564, "step": 10157 }, { "epoch": 98.4987951807229, "grad_norm": 9.7324857711792, "learning_rate": 5.068932038834951e-05, "loss": 0.4554, "step": 10158 }, { "epoch": 98.50843373493976, "grad_norm": 57.721839904785156, "learning_rate": 5.068446601941748e-05, "loss": 0.2218, "step": 10159 }, { "epoch": 98.51807228915662, "grad_norm": 0.7801135182380676, "learning_rate": 5.0679611650485435e-05, "loss": 0.162, "step": 10160 }, { "epoch": 98.5277108433735, "grad_norm": 2.0077154636383057, "learning_rate": 5.06747572815534e-05, "loss": 0.3421, "step": 10161 }, { "epoch": 98.53734939759036, "grad_norm": 21.922000885009766, "learning_rate": 5.066990291262136e-05, "loss": 0.3375, "step": 10162 }, { "epoch": 98.54698795180722, "grad_norm": 11.591272354125977, "learning_rate": 5.066504854368932e-05, "loss": 0.254, "step": 10163 }, { "epoch": 98.5566265060241, "grad_norm": 5.485836029052734, "learning_rate": 5.066019417475728e-05, "loss": 0.2895, "step": 10164 }, { "epoch": 98.56626506024097, "grad_norm": 3.3946664333343506, "learning_rate": 5.0655339805825244e-05, "loss": 0.3014, "step": 10165 }, { "epoch": 98.57590361445783, "grad_norm": 69.62387084960938, "learning_rate": 5.06504854368932e-05, "loss": 0.1107, "step": 10166 }, { "epoch": 98.58554216867469, "grad_norm": 15.032200813293457, "learning_rate": 5.064563106796117e-05, "loss": 0.2155, "step": 10167 }, { "epoch": 98.59518072289157, "grad_norm": 27.499919891357422, "learning_rate": 5.0640776699029124e-05, "loss": 0.2217, "step": 10168 }, { "epoch": 98.60481927710843, "grad_norm": 7.440958499908447, "learning_rate": 5.063592233009708e-05, "loss": 0.2494, "step": 10169 }, { "epoch": 98.6144578313253, "grad_norm": 28.518966674804688, "learning_rate": 5.063106796116505e-05, "loss": 0.2703, "step": 10170 }, { "epoch": 98.62409638554217, "grad_norm": 23.766361236572266, "learning_rate": 5.0626213592233004e-05, "loss": 0.3599, "step": 10171 }, { "epoch": 98.63373493975904, "grad_norm": 13.11489486694336, "learning_rate": 5.062135922330097e-05, "loss": 0.4015, "step": 10172 }, { "epoch": 98.6433734939759, "grad_norm": 15.000395774841309, "learning_rate": 5.061650485436893e-05, "loss": 0.3558, "step": 10173 }, { "epoch": 98.65301204819278, "grad_norm": 22.230566024780273, "learning_rate": 5.06116504854369e-05, "loss": 0.1762, "step": 10174 }, { "epoch": 98.66265060240964, "grad_norm": 3.9424829483032227, "learning_rate": 5.060679611650486e-05, "loss": 0.2024, "step": 10175 }, { "epoch": 98.6722891566265, "grad_norm": 5.017570495605469, "learning_rate": 5.060194174757282e-05, "loss": 0.2387, "step": 10176 }, { "epoch": 98.68192771084337, "grad_norm": 13.066815376281738, "learning_rate": 5.0597087378640785e-05, "loss": 0.5531, "step": 10177 }, { "epoch": 98.69156626506025, "grad_norm": 14.430408477783203, "learning_rate": 5.059223300970874e-05, "loss": 0.4905, "step": 10178 }, { "epoch": 98.70120481927711, "grad_norm": 5.477048873901367, "learning_rate": 5.058737864077671e-05, "loss": 0.3051, "step": 10179 }, { "epoch": 98.71084337349397, "grad_norm": 11.340863227844238, "learning_rate": 5.0582524271844665e-05, "loss": 0.2526, "step": 10180 }, { "epoch": 98.72048192771085, "grad_norm": 108.86710357666016, "learning_rate": 5.057766990291263e-05, "loss": 0.2347, "step": 10181 }, { "epoch": 98.73012048192771, "grad_norm": 13.786905288696289, "learning_rate": 5.057281553398059e-05, "loss": 0.359, "step": 10182 }, { "epoch": 98.73975903614458, "grad_norm": 10.694101333618164, "learning_rate": 5.056796116504855e-05, "loss": 0.3369, "step": 10183 }, { "epoch": 98.74939759036144, "grad_norm": 11.010004043579102, "learning_rate": 5.056310679611651e-05, "loss": 0.4293, "step": 10184 }, { "epoch": 98.75903614457832, "grad_norm": 7.030365467071533, "learning_rate": 5.0558252427184474e-05, "loss": 0.1708, "step": 10185 }, { "epoch": 98.76867469879518, "grad_norm": 15.817442893981934, "learning_rate": 5.055339805825243e-05, "loss": 0.256, "step": 10186 }, { "epoch": 98.77831325301204, "grad_norm": 20.477691650390625, "learning_rate": 5.0548543689320396e-05, "loss": 0.3165, "step": 10187 }, { "epoch": 98.78795180722892, "grad_norm": 8.844436645507812, "learning_rate": 5.0543689320388354e-05, "loss": 0.2176, "step": 10188 }, { "epoch": 98.79759036144578, "grad_norm": 20.784456253051758, "learning_rate": 5.053883495145632e-05, "loss": 0.3273, "step": 10189 }, { "epoch": 98.80722891566265, "grad_norm": 6.0977864265441895, "learning_rate": 5.0533980582524276e-05, "loss": 0.3866, "step": 10190 }, { "epoch": 98.81686746987951, "grad_norm": 12.665335655212402, "learning_rate": 5.0529126213592234e-05, "loss": 0.4688, "step": 10191 }, { "epoch": 98.82650602409639, "grad_norm": 13.150131225585938, "learning_rate": 5.05242718446602e-05, "loss": 0.4041, "step": 10192 }, { "epoch": 98.83614457831325, "grad_norm": 10.415460586547852, "learning_rate": 5.0519417475728156e-05, "loss": 0.2918, "step": 10193 }, { "epoch": 98.84578313253012, "grad_norm": 7.9015607833862305, "learning_rate": 5.051456310679612e-05, "loss": 0.3277, "step": 10194 }, { "epoch": 98.855421686747, "grad_norm": 2.4357950687408447, "learning_rate": 5.050970873786408e-05, "loss": 0.1887, "step": 10195 }, { "epoch": 98.86506024096386, "grad_norm": 7.699656009674072, "learning_rate": 5.050485436893204e-05, "loss": 0.296, "step": 10196 }, { "epoch": 98.87469879518072, "grad_norm": 4.109846591949463, "learning_rate": 5.05e-05, "loss": 0.2912, "step": 10197 }, { "epoch": 98.88433734939758, "grad_norm": 11.305110931396484, "learning_rate": 5.0495145631067966e-05, "loss": 0.268, "step": 10198 }, { "epoch": 98.89397590361446, "grad_norm": 8.73231315612793, "learning_rate": 5.049029126213592e-05, "loss": 0.1476, "step": 10199 }, { "epoch": 98.90361445783132, "grad_norm": 15.355653762817383, "learning_rate": 5.048543689320389e-05, "loss": 0.2451, "step": 10200 }, { "epoch": 98.91325301204819, "grad_norm": 77.68942260742188, "learning_rate": 5.0480582524271846e-05, "loss": 0.3474, "step": 10201 }, { "epoch": 98.92289156626506, "grad_norm": 5.794362545013428, "learning_rate": 5.047572815533981e-05, "loss": 0.2547, "step": 10202 }, { "epoch": 98.93253012048193, "grad_norm": 49.32490539550781, "learning_rate": 5.047087378640777e-05, "loss": 0.3018, "step": 10203 }, { "epoch": 98.94216867469879, "grad_norm": 30.32761001586914, "learning_rate": 5.046601941747573e-05, "loss": 0.3926, "step": 10204 }, { "epoch": 98.95180722891567, "grad_norm": 9.195537567138672, "learning_rate": 5.046116504854369e-05, "loss": 0.2041, "step": 10205 }, { "epoch": 98.96144578313253, "grad_norm": 12.470454216003418, "learning_rate": 5.0456310679611655e-05, "loss": 0.443, "step": 10206 }, { "epoch": 98.9710843373494, "grad_norm": 9.522775650024414, "learning_rate": 5.045145631067961e-05, "loss": 0.2248, "step": 10207 }, { "epoch": 98.98072289156626, "grad_norm": 18.616905212402344, "learning_rate": 5.044660194174757e-05, "loss": 0.1784, "step": 10208 }, { "epoch": 98.99036144578314, "grad_norm": 14.384087562561035, "learning_rate": 5.0441747572815535e-05, "loss": 0.2055, "step": 10209 }, { "epoch": 99.00602409638554, "grad_norm": 3.0407583713531494, "learning_rate": 5.043689320388349e-05, "loss": 0.3992, "step": 10210 }, { "epoch": 99.01566265060241, "grad_norm": 15.289557456970215, "learning_rate": 5.043203883495146e-05, "loss": 0.344, "step": 10211 }, { "epoch": 99.02530120481927, "grad_norm": 22.276342391967773, "learning_rate": 5.0427184466019415e-05, "loss": 0.1556, "step": 10212 }, { "epoch": 99.03493975903615, "grad_norm": 8.211430549621582, "learning_rate": 5.042233009708738e-05, "loss": 0.2548, "step": 10213 }, { "epoch": 99.04457831325301, "grad_norm": 15.293793678283691, "learning_rate": 5.041747572815534e-05, "loss": 0.4648, "step": 10214 }, { "epoch": 99.05421686746988, "grad_norm": 5.589765548706055, "learning_rate": 5.04126213592233e-05, "loss": 0.1193, "step": 10215 }, { "epoch": 99.06385542168675, "grad_norm": 20.24909782409668, "learning_rate": 5.040776699029126e-05, "loss": 0.3324, "step": 10216 }, { "epoch": 99.07349397590362, "grad_norm": 7.184384822845459, "learning_rate": 5.0402912621359224e-05, "loss": 0.2869, "step": 10217 }, { "epoch": 99.08313253012048, "grad_norm": 15.15989875793457, "learning_rate": 5.039805825242718e-05, "loss": 0.3088, "step": 10218 }, { "epoch": 99.09277108433734, "grad_norm": 6.578237533569336, "learning_rate": 5.0393203883495146e-05, "loss": 0.2614, "step": 10219 }, { "epoch": 99.10240963855422, "grad_norm": 29.04680633544922, "learning_rate": 5.0388349514563104e-05, "loss": 0.3587, "step": 10220 }, { "epoch": 99.11204819277108, "grad_norm": 6.21566915512085, "learning_rate": 5.038349514563107e-05, "loss": 0.4793, "step": 10221 }, { "epoch": 99.12168674698795, "grad_norm": 33.70033645629883, "learning_rate": 5.0378640776699026e-05, "loss": 0.1265, "step": 10222 }, { "epoch": 99.13132530120482, "grad_norm": 19.365385055541992, "learning_rate": 5.037378640776699e-05, "loss": 0.24, "step": 10223 }, { "epoch": 99.14096385542169, "grad_norm": 25.330293655395508, "learning_rate": 5.036893203883495e-05, "loss": 0.5246, "step": 10224 }, { "epoch": 99.15060240963855, "grad_norm": 21.54415512084961, "learning_rate": 5.036407766990292e-05, "loss": 0.3705, "step": 10225 }, { "epoch": 99.16024096385541, "grad_norm": 29.764204025268555, "learning_rate": 5.0359223300970884e-05, "loss": 0.2823, "step": 10226 }, { "epoch": 99.16987951807229, "grad_norm": 17.496767044067383, "learning_rate": 5.035436893203884e-05, "loss": 0.1974, "step": 10227 }, { "epoch": 99.17951807228916, "grad_norm": 10.6709623336792, "learning_rate": 5.03495145631068e-05, "loss": 0.2766, "step": 10228 }, { "epoch": 99.18915662650602, "grad_norm": 13.397512435913086, "learning_rate": 5.0344660194174764e-05, "loss": 0.2846, "step": 10229 }, { "epoch": 99.1987951807229, "grad_norm": 35.866722106933594, "learning_rate": 5.033980582524272e-05, "loss": 0.4781, "step": 10230 }, { "epoch": 99.20843373493976, "grad_norm": 28.519716262817383, "learning_rate": 5.033495145631069e-05, "loss": 0.4926, "step": 10231 }, { "epoch": 99.21807228915662, "grad_norm": 12.031627655029297, "learning_rate": 5.0330097087378644e-05, "loss": 0.4015, "step": 10232 }, { "epoch": 99.2277108433735, "grad_norm": 20.249103546142578, "learning_rate": 5.032524271844661e-05, "loss": 0.3955, "step": 10233 }, { "epoch": 99.23734939759036, "grad_norm": 12.329541206359863, "learning_rate": 5.032038834951457e-05, "loss": 0.1637, "step": 10234 }, { "epoch": 99.24698795180723, "grad_norm": 9.162304878234863, "learning_rate": 5.031553398058253e-05, "loss": 0.3071, "step": 10235 }, { "epoch": 99.25662650602409, "grad_norm": 21.58061981201172, "learning_rate": 5.031067961165049e-05, "loss": 0.4493, "step": 10236 }, { "epoch": 99.26626506024097, "grad_norm": 14.322344779968262, "learning_rate": 5.0305825242718454e-05, "loss": 0.2068, "step": 10237 }, { "epoch": 99.27590361445783, "grad_norm": 3.887256622314453, "learning_rate": 5.030097087378641e-05, "loss": 0.2261, "step": 10238 }, { "epoch": 99.2855421686747, "grad_norm": 4.708380222320557, "learning_rate": 5.0296116504854376e-05, "loss": 0.1556, "step": 10239 }, { "epoch": 99.29518072289157, "grad_norm": 15.503280639648438, "learning_rate": 5.0291262135922334e-05, "loss": 0.2873, "step": 10240 }, { "epoch": 99.30481927710844, "grad_norm": 22.50484275817871, "learning_rate": 5.02864077669903e-05, "loss": 0.2473, "step": 10241 }, { "epoch": 99.3144578313253, "grad_norm": 5.107472896575928, "learning_rate": 5.0281553398058256e-05, "loss": 0.2196, "step": 10242 }, { "epoch": 99.32409638554216, "grad_norm": 12.278064727783203, "learning_rate": 5.027669902912622e-05, "loss": 0.3889, "step": 10243 }, { "epoch": 99.33373493975904, "grad_norm": 7.192921161651611, "learning_rate": 5.027184466019418e-05, "loss": 0.3233, "step": 10244 }, { "epoch": 99.3433734939759, "grad_norm": 20.277381896972656, "learning_rate": 5.0266990291262136e-05, "loss": 0.3655, "step": 10245 }, { "epoch": 99.35301204819277, "grad_norm": 8.007027626037598, "learning_rate": 5.02621359223301e-05, "loss": 0.2296, "step": 10246 }, { "epoch": 99.36265060240964, "grad_norm": 11.350903511047363, "learning_rate": 5.025728155339806e-05, "loss": 0.2993, "step": 10247 }, { "epoch": 99.37228915662651, "grad_norm": 12.204854965209961, "learning_rate": 5.025242718446602e-05, "loss": 0.2599, "step": 10248 }, { "epoch": 99.38192771084337, "grad_norm": 10.23876953125, "learning_rate": 5.024757281553398e-05, "loss": 0.4221, "step": 10249 }, { "epoch": 99.39156626506023, "grad_norm": 18.297943115234375, "learning_rate": 5.0242718446601945e-05, "loss": 0.2546, "step": 10250 }, { "epoch": 99.40120481927711, "grad_norm": 5.441380023956299, "learning_rate": 5.02378640776699e-05, "loss": 0.3317, "step": 10251 }, { "epoch": 99.41084337349398, "grad_norm": 26.162822723388672, "learning_rate": 5.023300970873787e-05, "loss": 0.3829, "step": 10252 }, { "epoch": 99.42048192771084, "grad_norm": 21.775848388671875, "learning_rate": 5.0228155339805825e-05, "loss": 0.2145, "step": 10253 }, { "epoch": 99.43012048192772, "grad_norm": 12.920827865600586, "learning_rate": 5.022330097087379e-05, "loss": 0.2516, "step": 10254 }, { "epoch": 99.43975903614458, "grad_norm": 10.622918128967285, "learning_rate": 5.021844660194175e-05, "loss": 0.3206, "step": 10255 }, { "epoch": 99.44939759036144, "grad_norm": 81.078857421875, "learning_rate": 5.021359223300971e-05, "loss": 0.4835, "step": 10256 }, { "epoch": 99.45903614457832, "grad_norm": 5.063800811767578, "learning_rate": 5.020873786407767e-05, "loss": 0.2104, "step": 10257 }, { "epoch": 99.46867469879518, "grad_norm": 34.362449645996094, "learning_rate": 5.0203883495145634e-05, "loss": 0.3073, "step": 10258 }, { "epoch": 99.47831325301205, "grad_norm": 23.74051856994629, "learning_rate": 5.019902912621359e-05, "loss": 0.3603, "step": 10259 }, { "epoch": 99.48795180722891, "grad_norm": 25.31815528869629, "learning_rate": 5.0194174757281557e-05, "loss": 0.1923, "step": 10260 }, { "epoch": 99.49759036144579, "grad_norm": 31.708106994628906, "learning_rate": 5.0189320388349514e-05, "loss": 0.4736, "step": 10261 }, { "epoch": 99.50722891566265, "grad_norm": 11.820638656616211, "learning_rate": 5.018446601941747e-05, "loss": 0.3465, "step": 10262 }, { "epoch": 99.51686746987951, "grad_norm": 2.550055742263794, "learning_rate": 5.017961165048544e-05, "loss": 0.1843, "step": 10263 }, { "epoch": 99.52650602409639, "grad_norm": 41.071258544921875, "learning_rate": 5.0174757281553394e-05, "loss": 0.3489, "step": 10264 }, { "epoch": 99.53614457831326, "grad_norm": 105.49179077148438, "learning_rate": 5.016990291262136e-05, "loss": 0.2223, "step": 10265 }, { "epoch": 99.54578313253012, "grad_norm": 18.188079833984375, "learning_rate": 5.016504854368932e-05, "loss": 0.405, "step": 10266 }, { "epoch": 99.55542168674698, "grad_norm": 23.86577033996582, "learning_rate": 5.016019417475728e-05, "loss": 0.4025, "step": 10267 }, { "epoch": 99.56506024096386, "grad_norm": 43.770416259765625, "learning_rate": 5.015533980582524e-05, "loss": 0.4344, "step": 10268 }, { "epoch": 99.57469879518072, "grad_norm": 20.383134841918945, "learning_rate": 5.0150485436893203e-05, "loss": 0.4818, "step": 10269 }, { "epoch": 99.58433734939759, "grad_norm": 2.2277021408081055, "learning_rate": 5.014563106796116e-05, "loss": 0.2383, "step": 10270 }, { "epoch": 99.59397590361446, "grad_norm": 11.992202758789062, "learning_rate": 5.0140776699029126e-05, "loss": 0.4608, "step": 10271 }, { "epoch": 99.60361445783133, "grad_norm": 18.262548446655273, "learning_rate": 5.0135922330097084e-05, "loss": 0.2664, "step": 10272 }, { "epoch": 99.61325301204819, "grad_norm": 12.446698188781738, "learning_rate": 5.013106796116505e-05, "loss": 0.4658, "step": 10273 }, { "epoch": 99.62289156626505, "grad_norm": 17.0487060546875, "learning_rate": 5.0126213592233006e-05, "loss": 0.2589, "step": 10274 }, { "epoch": 99.63253012048193, "grad_norm": 18.738454818725586, "learning_rate": 5.012135922330098e-05, "loss": 0.322, "step": 10275 }, { "epoch": 99.6421686746988, "grad_norm": 17.053020477294922, "learning_rate": 5.011650485436894e-05, "loss": 0.3798, "step": 10276 }, { "epoch": 99.65180722891566, "grad_norm": 8.550875663757324, "learning_rate": 5.01116504854369e-05, "loss": 0.3674, "step": 10277 }, { "epoch": 99.66144578313254, "grad_norm": 2.593705415725708, "learning_rate": 5.0106796116504864e-05, "loss": 0.1492, "step": 10278 }, { "epoch": 99.6710843373494, "grad_norm": 5.921290874481201, "learning_rate": 5.010194174757282e-05, "loss": 0.2451, "step": 10279 }, { "epoch": 99.68072289156626, "grad_norm": 5.023929119110107, "learning_rate": 5.0097087378640786e-05, "loss": 0.2437, "step": 10280 }, { "epoch": 99.69036144578314, "grad_norm": 187.36349487304688, "learning_rate": 5.0092233009708744e-05, "loss": 0.275, "step": 10281 }, { "epoch": 99.7, "grad_norm": 7.750641345977783, "learning_rate": 5.008737864077671e-05, "loss": 0.399, "step": 10282 }, { "epoch": 99.70963855421687, "grad_norm": 6.5467400550842285, "learning_rate": 5.0082524271844666e-05, "loss": 0.2843, "step": 10283 }, { "epoch": 99.71927710843373, "grad_norm": 25.601112365722656, "learning_rate": 5.0077669902912624e-05, "loss": 0.3455, "step": 10284 }, { "epoch": 99.7289156626506, "grad_norm": 6.05398416519165, "learning_rate": 5.007281553398059e-05, "loss": 0.2157, "step": 10285 }, { "epoch": 99.73855421686747, "grad_norm": 13.204821586608887, "learning_rate": 5.0067961165048546e-05, "loss": 0.1857, "step": 10286 }, { "epoch": 99.74819277108433, "grad_norm": 7.872300148010254, "learning_rate": 5.006310679611651e-05, "loss": 0.2241, "step": 10287 }, { "epoch": 99.75783132530121, "grad_norm": 5.0233635902404785, "learning_rate": 5.005825242718447e-05, "loss": 0.4913, "step": 10288 }, { "epoch": 99.76746987951807, "grad_norm": 36.7023811340332, "learning_rate": 5.005339805825243e-05, "loss": 0.1086, "step": 10289 }, { "epoch": 99.77710843373494, "grad_norm": 27.45355987548828, "learning_rate": 5.004854368932039e-05, "loss": 0.237, "step": 10290 }, { "epoch": 99.7867469879518, "grad_norm": 7.7363080978393555, "learning_rate": 5.0043689320388355e-05, "loss": 0.3441, "step": 10291 }, { "epoch": 99.79638554216868, "grad_norm": 7.5347723960876465, "learning_rate": 5.003883495145631e-05, "loss": 0.2147, "step": 10292 }, { "epoch": 99.80602409638554, "grad_norm": 13.56844425201416, "learning_rate": 5.003398058252428e-05, "loss": 0.1584, "step": 10293 }, { "epoch": 99.8156626506024, "grad_norm": 22.849288940429688, "learning_rate": 5.0029126213592236e-05, "loss": 0.4147, "step": 10294 }, { "epoch": 99.82530120481928, "grad_norm": 15.143835067749023, "learning_rate": 5.00242718446602e-05, "loss": 0.2681, "step": 10295 }, { "epoch": 99.83493975903615, "grad_norm": 8.001534461975098, "learning_rate": 5.001941747572816e-05, "loss": 0.2855, "step": 10296 }, { "epoch": 99.84457831325301, "grad_norm": 26.03980255126953, "learning_rate": 5.001456310679612e-05, "loss": 0.2404, "step": 10297 }, { "epoch": 99.85421686746987, "grad_norm": 23.635061264038086, "learning_rate": 5.000970873786408e-05, "loss": 0.3819, "step": 10298 }, { "epoch": 99.86385542168675, "grad_norm": 10.620211601257324, "learning_rate": 5.0004854368932045e-05, "loss": 0.0948, "step": 10299 }, { "epoch": 99.87349397590361, "grad_norm": 31.49822425842285, "learning_rate": 5e-05, "loss": 0.3161, "step": 10300 }, { "epoch": 99.88313253012048, "grad_norm": 14.97055435180664, "learning_rate": 4.999514563106796e-05, "loss": 0.4177, "step": 10301 }, { "epoch": 99.89277108433735, "grad_norm": 16.945171356201172, "learning_rate": 4.9990291262135925e-05, "loss": 0.3697, "step": 10302 }, { "epoch": 99.90240963855422, "grad_norm": 32.65433883666992, "learning_rate": 4.998543689320388e-05, "loss": 0.3822, "step": 10303 }, { "epoch": 99.91204819277108, "grad_norm": 4.431458950042725, "learning_rate": 4.998058252427185e-05, "loss": 0.2859, "step": 10304 }, { "epoch": 99.92168674698796, "grad_norm": 11.537339210510254, "learning_rate": 4.9975728155339805e-05, "loss": 0.4311, "step": 10305 }, { "epoch": 99.93132530120482, "grad_norm": 1.5994890928268433, "learning_rate": 4.997087378640777e-05, "loss": 0.1735, "step": 10306 }, { "epoch": 99.94096385542169, "grad_norm": 18.89168357849121, "learning_rate": 4.996601941747573e-05, "loss": 0.3432, "step": 10307 }, { "epoch": 99.95060240963855, "grad_norm": 9.35467529296875, "learning_rate": 4.996116504854369e-05, "loss": 0.2045, "step": 10308 }, { "epoch": 99.96024096385543, "grad_norm": 37.16474151611328, "learning_rate": 4.995631067961165e-05, "loss": 0.4262, "step": 10309 }, { "epoch": 99.96987951807229, "grad_norm": 8.276409149169922, "learning_rate": 4.9951456310679614e-05, "loss": 0.2621, "step": 10310 }, { "epoch": 99.97951807228915, "grad_norm": 8.146198272705078, "learning_rate": 4.994660194174757e-05, "loss": 0.2037, "step": 10311 }, { "epoch": 99.98915662650603, "grad_norm": 10.162540435791016, "learning_rate": 4.9941747572815536e-05, "loss": 0.2896, "step": 10312 }, { "epoch": 100.00481927710844, "grad_norm": 11.20681381225586, "learning_rate": 4.99368932038835e-05, "loss": 0.1588, "step": 10313 }, { "epoch": 100.0144578313253, "grad_norm": 16.99727439880371, "learning_rate": 4.993203883495146e-05, "loss": 0.3402, "step": 10314 }, { "epoch": 100.02409638554217, "grad_norm": 10.37129020690918, "learning_rate": 4.992718446601942e-05, "loss": 0.3685, "step": 10315 }, { "epoch": 100.03373493975904, "grad_norm": 14.748259544372559, "learning_rate": 4.992233009708738e-05, "loss": 0.3492, "step": 10316 }, { "epoch": 100.0433734939759, "grad_norm": 6.249054431915283, "learning_rate": 4.9917475728155345e-05, "loss": 0.2295, "step": 10317 }, { "epoch": 100.05301204819277, "grad_norm": 37.38815689086914, "learning_rate": 4.99126213592233e-05, "loss": 0.5208, "step": 10318 }, { "epoch": 100.06265060240963, "grad_norm": 3.5261058807373047, "learning_rate": 4.990776699029127e-05, "loss": 0.2854, "step": 10319 }, { "epoch": 100.07228915662651, "grad_norm": 18.850095748901367, "learning_rate": 4.9902912621359225e-05, "loss": 0.241, "step": 10320 }, { "epoch": 100.08192771084337, "grad_norm": 7.891079902648926, "learning_rate": 4.989805825242719e-05, "loss": 0.3441, "step": 10321 }, { "epoch": 100.09156626506024, "grad_norm": 20.372432708740234, "learning_rate": 4.989320388349515e-05, "loss": 0.3967, "step": 10322 }, { "epoch": 100.10120481927711, "grad_norm": 6.859499931335449, "learning_rate": 4.988834951456311e-05, "loss": 0.2676, "step": 10323 }, { "epoch": 100.11084337349398, "grad_norm": 14.228832244873047, "learning_rate": 4.988349514563107e-05, "loss": 0.4526, "step": 10324 }, { "epoch": 100.12048192771084, "grad_norm": 35.610443115234375, "learning_rate": 4.9878640776699034e-05, "loss": 0.3085, "step": 10325 }, { "epoch": 100.1301204819277, "grad_norm": 6.694357395172119, "learning_rate": 4.987378640776699e-05, "loss": 0.1549, "step": 10326 }, { "epoch": 100.13975903614458, "grad_norm": 15.032203674316406, "learning_rate": 4.986893203883496e-05, "loss": 0.2667, "step": 10327 }, { "epoch": 100.14939759036145, "grad_norm": 7.816946506500244, "learning_rate": 4.9864077669902914e-05, "loss": 0.1861, "step": 10328 }, { "epoch": 100.15903614457831, "grad_norm": 20.975872039794922, "learning_rate": 4.985922330097087e-05, "loss": 0.3885, "step": 10329 }, { "epoch": 100.16867469879519, "grad_norm": 3.9075334072113037, "learning_rate": 4.985436893203884e-05, "loss": 0.2656, "step": 10330 }, { "epoch": 100.17831325301205, "grad_norm": 32.98542022705078, "learning_rate": 4.9849514563106795e-05, "loss": 0.3614, "step": 10331 }, { "epoch": 100.18795180722891, "grad_norm": 7.1740593910217285, "learning_rate": 4.984466019417476e-05, "loss": 0.1443, "step": 10332 }, { "epoch": 100.19759036144578, "grad_norm": 33.724178314208984, "learning_rate": 4.983980582524272e-05, "loss": 0.6341, "step": 10333 }, { "epoch": 100.20722891566265, "grad_norm": 13.695966720581055, "learning_rate": 4.983495145631068e-05, "loss": 0.3853, "step": 10334 }, { "epoch": 100.21686746987952, "grad_norm": 8.876741409301758, "learning_rate": 4.983009708737864e-05, "loss": 0.2276, "step": 10335 }, { "epoch": 100.22650602409638, "grad_norm": 1.3411427736282349, "learning_rate": 4.9825242718446604e-05, "loss": 0.0849, "step": 10336 }, { "epoch": 100.23614457831326, "grad_norm": 7.736196994781494, "learning_rate": 4.982038834951456e-05, "loss": 0.2963, "step": 10337 }, { "epoch": 100.24578313253012, "grad_norm": 6.858038425445557, "learning_rate": 4.981553398058253e-05, "loss": 0.3732, "step": 10338 }, { "epoch": 100.25542168674698, "grad_norm": 13.569236755371094, "learning_rate": 4.981067961165049e-05, "loss": 0.3557, "step": 10339 }, { "epoch": 100.26506024096386, "grad_norm": 29.112390518188477, "learning_rate": 4.980582524271845e-05, "loss": 0.2431, "step": 10340 }, { "epoch": 100.27469879518073, "grad_norm": 5.064362525939941, "learning_rate": 4.980097087378641e-05, "loss": 0.3368, "step": 10341 }, { "epoch": 100.28433734939759, "grad_norm": 6.832089424133301, "learning_rate": 4.979611650485437e-05, "loss": 0.3021, "step": 10342 }, { "epoch": 100.29397590361445, "grad_norm": 7.681565284729004, "learning_rate": 4.9791262135922335e-05, "loss": 0.2671, "step": 10343 }, { "epoch": 100.30361445783133, "grad_norm": 4.192964553833008, "learning_rate": 4.978640776699029e-05, "loss": 0.276, "step": 10344 }, { "epoch": 100.3132530120482, "grad_norm": 11.032265663146973, "learning_rate": 4.978155339805826e-05, "loss": 0.3468, "step": 10345 }, { "epoch": 100.32289156626506, "grad_norm": 15.295228004455566, "learning_rate": 4.9776699029126215e-05, "loss": 0.4103, "step": 10346 }, { "epoch": 100.33253012048193, "grad_norm": 24.55428695678711, "learning_rate": 4.977184466019418e-05, "loss": 0.4022, "step": 10347 }, { "epoch": 100.3421686746988, "grad_norm": 9.506606101989746, "learning_rate": 4.976699029126214e-05, "loss": 0.279, "step": 10348 }, { "epoch": 100.35180722891566, "grad_norm": 5.337984085083008, "learning_rate": 4.97621359223301e-05, "loss": 0.3021, "step": 10349 }, { "epoch": 100.36144578313252, "grad_norm": 9.458174705505371, "learning_rate": 4.975728155339806e-05, "loss": 0.2021, "step": 10350 }, { "epoch": 100.3710843373494, "grad_norm": 19.079713821411133, "learning_rate": 4.9752427184466024e-05, "loss": 0.1953, "step": 10351 }, { "epoch": 100.38072289156626, "grad_norm": 15.672944068908691, "learning_rate": 4.974757281553398e-05, "loss": 0.2188, "step": 10352 }, { "epoch": 100.39036144578313, "grad_norm": 19.98890495300293, "learning_rate": 4.9742718446601947e-05, "loss": 0.4478, "step": 10353 }, { "epoch": 100.4, "grad_norm": 10.313115119934082, "learning_rate": 4.9737864077669904e-05, "loss": 0.2301, "step": 10354 }, { "epoch": 100.40963855421687, "grad_norm": 8.135464668273926, "learning_rate": 4.973300970873787e-05, "loss": 0.3766, "step": 10355 }, { "epoch": 100.41927710843373, "grad_norm": 5.232161045074463, "learning_rate": 4.9728155339805827e-05, "loss": 0.3002, "step": 10356 }, { "epoch": 100.4289156626506, "grad_norm": 6.578525543212891, "learning_rate": 4.9723300970873784e-05, "loss": 0.1993, "step": 10357 }, { "epoch": 100.43855421686747, "grad_norm": 3.8501267433166504, "learning_rate": 4.971844660194175e-05, "loss": 0.0627, "step": 10358 }, { "epoch": 100.44819277108434, "grad_norm": 9.431326866149902, "learning_rate": 4.9713592233009707e-05, "loss": 0.2979, "step": 10359 }, { "epoch": 100.4578313253012, "grad_norm": 19.32864761352539, "learning_rate": 4.970873786407767e-05, "loss": 0.2122, "step": 10360 }, { "epoch": 100.46746987951808, "grad_norm": 7.53866720199585, "learning_rate": 4.970388349514563e-05, "loss": 0.1877, "step": 10361 }, { "epoch": 100.47710843373494, "grad_norm": 15.667336463928223, "learning_rate": 4.9699029126213593e-05, "loss": 0.297, "step": 10362 }, { "epoch": 100.4867469879518, "grad_norm": 11.692089080810547, "learning_rate": 4.969417475728155e-05, "loss": 0.229, "step": 10363 }, { "epoch": 100.49638554216868, "grad_norm": 13.432847023010254, "learning_rate": 4.968932038834952e-05, "loss": 0.6856, "step": 10364 }, { "epoch": 100.50602409638554, "grad_norm": 16.88568878173828, "learning_rate": 4.968446601941748e-05, "loss": 0.1924, "step": 10365 }, { "epoch": 100.51566265060241, "grad_norm": 20.71065330505371, "learning_rate": 4.9679611650485445e-05, "loss": 0.3104, "step": 10366 }, { "epoch": 100.52530120481927, "grad_norm": 13.789261817932129, "learning_rate": 4.96747572815534e-05, "loss": 0.1826, "step": 10367 }, { "epoch": 100.53493975903615, "grad_norm": 17.93990707397461, "learning_rate": 4.966990291262136e-05, "loss": 0.3752, "step": 10368 }, { "epoch": 100.54457831325301, "grad_norm": 7.504151821136475, "learning_rate": 4.9665048543689325e-05, "loss": 0.3029, "step": 10369 }, { "epoch": 100.55421686746988, "grad_norm": 10.549983978271484, "learning_rate": 4.966019417475728e-05, "loss": 0.2913, "step": 10370 }, { "epoch": 100.56385542168675, "grad_norm": 16.14238739013672, "learning_rate": 4.965533980582525e-05, "loss": 0.5251, "step": 10371 }, { "epoch": 100.57349397590362, "grad_norm": 5.937098026275635, "learning_rate": 4.9650485436893205e-05, "loss": 0.2034, "step": 10372 }, { "epoch": 100.58313253012048, "grad_norm": 11.844659805297852, "learning_rate": 4.964563106796117e-05, "loss": 0.1265, "step": 10373 }, { "epoch": 100.59277108433734, "grad_norm": 18.856325149536133, "learning_rate": 4.964077669902913e-05, "loss": 0.2049, "step": 10374 }, { "epoch": 100.60240963855422, "grad_norm": 28.490997314453125, "learning_rate": 4.963592233009709e-05, "loss": 0.4324, "step": 10375 }, { "epoch": 100.61204819277108, "grad_norm": 20.226736068725586, "learning_rate": 4.963106796116505e-05, "loss": 0.4363, "step": 10376 }, { "epoch": 100.62168674698795, "grad_norm": 28.726041793823242, "learning_rate": 4.9626213592233014e-05, "loss": 0.3479, "step": 10377 }, { "epoch": 100.63132530120482, "grad_norm": 20.110355377197266, "learning_rate": 4.962135922330097e-05, "loss": 0.2707, "step": 10378 }, { "epoch": 100.64096385542169, "grad_norm": 25.09418487548828, "learning_rate": 4.9616504854368936e-05, "loss": 0.2904, "step": 10379 }, { "epoch": 100.65060240963855, "grad_norm": 33.57555389404297, "learning_rate": 4.9611650485436894e-05, "loss": 0.2346, "step": 10380 }, { "epoch": 100.66024096385541, "grad_norm": 11.84870719909668, "learning_rate": 4.960679611650486e-05, "loss": 0.2603, "step": 10381 }, { "epoch": 100.66987951807229, "grad_norm": 24.947229385375977, "learning_rate": 4.9601941747572816e-05, "loss": 0.2494, "step": 10382 }, { "epoch": 100.67951807228916, "grad_norm": 5.784005641937256, "learning_rate": 4.9597087378640774e-05, "loss": 0.2139, "step": 10383 }, { "epoch": 100.68915662650602, "grad_norm": 19.746803283691406, "learning_rate": 4.959223300970874e-05, "loss": 0.5046, "step": 10384 }, { "epoch": 100.6987951807229, "grad_norm": 7.150199890136719, "learning_rate": 4.9587378640776696e-05, "loss": 0.1305, "step": 10385 }, { "epoch": 100.70843373493976, "grad_norm": 8.034371376037598, "learning_rate": 4.958252427184466e-05, "loss": 0.296, "step": 10386 }, { "epoch": 100.71807228915662, "grad_norm": 5.0515618324279785, "learning_rate": 4.957766990291262e-05, "loss": 0.2297, "step": 10387 }, { "epoch": 100.7277108433735, "grad_norm": 25.398054122924805, "learning_rate": 4.957281553398058e-05, "loss": 0.4062, "step": 10388 }, { "epoch": 100.73734939759036, "grad_norm": 11.36060905456543, "learning_rate": 4.956796116504855e-05, "loss": 0.3157, "step": 10389 }, { "epoch": 100.74698795180723, "grad_norm": 13.019286155700684, "learning_rate": 4.956310679611651e-05, "loss": 0.3716, "step": 10390 }, { "epoch": 100.75662650602409, "grad_norm": 12.227917671203613, "learning_rate": 4.955825242718447e-05, "loss": 0.2726, "step": 10391 }, { "epoch": 100.76626506024097, "grad_norm": 8.412135124206543, "learning_rate": 4.9553398058252435e-05, "loss": 0.5332, "step": 10392 }, { "epoch": 100.77590361445783, "grad_norm": 4.767295837402344, "learning_rate": 4.954854368932039e-05, "loss": 0.1521, "step": 10393 }, { "epoch": 100.7855421686747, "grad_norm": 7.241024971008301, "learning_rate": 4.954368932038835e-05, "loss": 0.3876, "step": 10394 }, { "epoch": 100.79518072289157, "grad_norm": 18.567724227905273, "learning_rate": 4.9538834951456315e-05, "loss": 0.3296, "step": 10395 }, { "epoch": 100.80481927710844, "grad_norm": 15.345973014831543, "learning_rate": 4.953398058252427e-05, "loss": 0.2433, "step": 10396 }, { "epoch": 100.8144578313253, "grad_norm": 4.777826309204102, "learning_rate": 4.952912621359224e-05, "loss": 0.3427, "step": 10397 }, { "epoch": 100.82409638554216, "grad_norm": 4.6297078132629395, "learning_rate": 4.9524271844660195e-05, "loss": 0.194, "step": 10398 }, { "epoch": 100.83373493975904, "grad_norm": 5.82307767868042, "learning_rate": 4.951941747572816e-05, "loss": 0.2689, "step": 10399 }, { "epoch": 100.8433734939759, "grad_norm": 78.31908416748047, "learning_rate": 4.951456310679612e-05, "loss": 0.2422, "step": 10400 }, { "epoch": 100.85301204819277, "grad_norm": 9.345744132995605, "learning_rate": 4.950970873786408e-05, "loss": 0.2983, "step": 10401 }, { "epoch": 100.86265060240964, "grad_norm": 5.732773780822754, "learning_rate": 4.950485436893204e-05, "loss": 0.0967, "step": 10402 }, { "epoch": 100.87228915662651, "grad_norm": 34.058189392089844, "learning_rate": 4.9500000000000004e-05, "loss": 0.189, "step": 10403 }, { "epoch": 100.88192771084337, "grad_norm": 17.226131439208984, "learning_rate": 4.949514563106796e-05, "loss": 0.4697, "step": 10404 }, { "epoch": 100.89156626506023, "grad_norm": 10.70313835144043, "learning_rate": 4.9490291262135926e-05, "loss": 0.3719, "step": 10405 }, { "epoch": 100.90120481927711, "grad_norm": 6.089843273162842, "learning_rate": 4.9485436893203884e-05, "loss": 0.2756, "step": 10406 }, { "epoch": 100.91084337349398, "grad_norm": 18.53067398071289, "learning_rate": 4.948058252427185e-05, "loss": 0.1733, "step": 10407 }, { "epoch": 100.92048192771084, "grad_norm": 6.549394607543945, "learning_rate": 4.9475728155339806e-05, "loss": 0.2314, "step": 10408 }, { "epoch": 100.93012048192772, "grad_norm": 20.3056640625, "learning_rate": 4.947087378640777e-05, "loss": 0.2531, "step": 10409 }, { "epoch": 100.93975903614458, "grad_norm": 8.779519081115723, "learning_rate": 4.946601941747573e-05, "loss": 0.1208, "step": 10410 }, { "epoch": 100.94939759036144, "grad_norm": 20.624713897705078, "learning_rate": 4.9461165048543686e-05, "loss": 0.092, "step": 10411 }, { "epoch": 100.95903614457832, "grad_norm": 14.45573902130127, "learning_rate": 4.945631067961165e-05, "loss": 0.4835, "step": 10412 }, { "epoch": 100.96867469879518, "grad_norm": 22.980302810668945, "learning_rate": 4.945145631067961e-05, "loss": 0.2186, "step": 10413 }, { "epoch": 100.97831325301205, "grad_norm": 22.16228675842285, "learning_rate": 4.944660194174758e-05, "loss": 0.4778, "step": 10414 }, { "epoch": 100.98795180722891, "grad_norm": 12.942910194396973, "learning_rate": 4.944174757281554e-05, "loss": 0.391, "step": 10415 }, { "epoch": 101.00361445783132, "grad_norm": 5.156022071838379, "learning_rate": 4.94368932038835e-05, "loss": 0.183, "step": 10416 }, { "epoch": 101.0132530120482, "grad_norm": 17.101966857910156, "learning_rate": 4.943203883495146e-05, "loss": 0.2835, "step": 10417 }, { "epoch": 101.02289156626506, "grad_norm": 14.086247444152832, "learning_rate": 4.9427184466019424e-05, "loss": 0.4185, "step": 10418 }, { "epoch": 101.03253012048192, "grad_norm": 6.556939125061035, "learning_rate": 4.942233009708738e-05, "loss": 0.3511, "step": 10419 }, { "epoch": 101.0421686746988, "grad_norm": 9.692699432373047, "learning_rate": 4.941747572815535e-05, "loss": 0.1913, "step": 10420 }, { "epoch": 101.05180722891566, "grad_norm": 18.577943801879883, "learning_rate": 4.9412621359223304e-05, "loss": 0.2158, "step": 10421 }, { "epoch": 101.06144578313253, "grad_norm": 19.276386260986328, "learning_rate": 4.940776699029126e-05, "loss": 0.3673, "step": 10422 }, { "epoch": 101.0710843373494, "grad_norm": 24.56143569946289, "learning_rate": 4.940291262135923e-05, "loss": 0.3539, "step": 10423 }, { "epoch": 101.08072289156627, "grad_norm": 8.111818313598633, "learning_rate": 4.9398058252427184e-05, "loss": 0.2213, "step": 10424 }, { "epoch": 101.09036144578313, "grad_norm": 19.007461547851562, "learning_rate": 4.939320388349515e-05, "loss": 0.4021, "step": 10425 }, { "epoch": 101.1, "grad_norm": 5.323831558227539, "learning_rate": 4.938834951456311e-05, "loss": 0.2166, "step": 10426 }, { "epoch": 101.10963855421687, "grad_norm": 7.0980987548828125, "learning_rate": 4.938349514563107e-05, "loss": 0.4474, "step": 10427 }, { "epoch": 101.11927710843374, "grad_norm": 14.693994522094727, "learning_rate": 4.937864077669903e-05, "loss": 0.2795, "step": 10428 }, { "epoch": 101.1289156626506, "grad_norm": 10.300163269042969, "learning_rate": 4.9373786407766994e-05, "loss": 0.4, "step": 10429 }, { "epoch": 101.13855421686748, "grad_norm": 11.221602439880371, "learning_rate": 4.936893203883495e-05, "loss": 0.5284, "step": 10430 }, { "epoch": 101.14819277108434, "grad_norm": 40.22843551635742, "learning_rate": 4.9364077669902916e-05, "loss": 0.4397, "step": 10431 }, { "epoch": 101.1578313253012, "grad_norm": 12.533644676208496, "learning_rate": 4.9359223300970874e-05, "loss": 0.2127, "step": 10432 }, { "epoch": 101.16746987951807, "grad_norm": 9.23444938659668, "learning_rate": 4.935436893203884e-05, "loss": 0.2888, "step": 10433 }, { "epoch": 101.17710843373494, "grad_norm": 9.388084411621094, "learning_rate": 4.9349514563106796e-05, "loss": 0.1147, "step": 10434 }, { "epoch": 101.1867469879518, "grad_norm": 11.129060745239258, "learning_rate": 4.934466019417476e-05, "loss": 0.3486, "step": 10435 }, { "epoch": 101.19638554216867, "grad_norm": 9.999528884887695, "learning_rate": 4.933980582524272e-05, "loss": 0.1539, "step": 10436 }, { "epoch": 101.20602409638555, "grad_norm": 7.5681610107421875, "learning_rate": 4.933495145631068e-05, "loss": 0.234, "step": 10437 }, { "epoch": 101.21566265060241, "grad_norm": 17.49942398071289, "learning_rate": 4.933009708737864e-05, "loss": 0.2945, "step": 10438 }, { "epoch": 101.22530120481927, "grad_norm": 11.455878257751465, "learning_rate": 4.9325242718446605e-05, "loss": 0.2682, "step": 10439 }, { "epoch": 101.23493975903614, "grad_norm": 6.0290679931640625, "learning_rate": 4.932038834951457e-05, "loss": 0.2535, "step": 10440 }, { "epoch": 101.24457831325302, "grad_norm": 35.96688461303711, "learning_rate": 4.931553398058253e-05, "loss": 0.3272, "step": 10441 }, { "epoch": 101.25421686746988, "grad_norm": 41.34259033203125, "learning_rate": 4.931067961165049e-05, "loss": 0.421, "step": 10442 }, { "epoch": 101.26385542168674, "grad_norm": 16.72174072265625, "learning_rate": 4.930582524271845e-05, "loss": 0.2965, "step": 10443 }, { "epoch": 101.27349397590362, "grad_norm": 5.710700035095215, "learning_rate": 4.9300970873786414e-05, "loss": 0.2549, "step": 10444 }, { "epoch": 101.28313253012048, "grad_norm": 9.33931827545166, "learning_rate": 4.929611650485437e-05, "loss": 0.4067, "step": 10445 }, { "epoch": 101.29277108433735, "grad_norm": 33.334232330322266, "learning_rate": 4.9291262135922336e-05, "loss": 0.2165, "step": 10446 }, { "epoch": 101.30240963855422, "grad_norm": 18.611600875854492, "learning_rate": 4.9286407766990294e-05, "loss": 0.2239, "step": 10447 }, { "epoch": 101.31204819277109, "grad_norm": 4.5863566398620605, "learning_rate": 4.928155339805826e-05, "loss": 0.1371, "step": 10448 }, { "epoch": 101.32168674698795, "grad_norm": 47.1357421875, "learning_rate": 4.9276699029126216e-05, "loss": 0.2996, "step": 10449 }, { "epoch": 101.33132530120481, "grad_norm": 4.730987548828125, "learning_rate": 4.9271844660194174e-05, "loss": 0.3055, "step": 10450 }, { "epoch": 101.34096385542169, "grad_norm": 6.770876884460449, "learning_rate": 4.926699029126214e-05, "loss": 0.1115, "step": 10451 }, { "epoch": 101.35060240963855, "grad_norm": 7.38093376159668, "learning_rate": 4.9262135922330097e-05, "loss": 0.1924, "step": 10452 }, { "epoch": 101.36024096385542, "grad_norm": 3.8864941596984863, "learning_rate": 4.925728155339806e-05, "loss": 0.1746, "step": 10453 }, { "epoch": 101.3698795180723, "grad_norm": 2.9113781452178955, "learning_rate": 4.925242718446602e-05, "loss": 0.2535, "step": 10454 }, { "epoch": 101.37951807228916, "grad_norm": 7.684232234954834, "learning_rate": 4.924757281553398e-05, "loss": 0.3255, "step": 10455 }, { "epoch": 101.38915662650602, "grad_norm": 5.809730529785156, "learning_rate": 4.924271844660194e-05, "loss": 0.3066, "step": 10456 }, { "epoch": 101.39879518072289, "grad_norm": 10.307433128356934, "learning_rate": 4.9237864077669906e-05, "loss": 0.1288, "step": 10457 }, { "epoch": 101.40843373493976, "grad_norm": 5.968269348144531, "learning_rate": 4.9233009708737863e-05, "loss": 0.3904, "step": 10458 }, { "epoch": 101.41807228915663, "grad_norm": 20.204383850097656, "learning_rate": 4.922815533980583e-05, "loss": 0.279, "step": 10459 }, { "epoch": 101.42771084337349, "grad_norm": 12.294118881225586, "learning_rate": 4.9223300970873786e-05, "loss": 0.4123, "step": 10460 }, { "epoch": 101.43734939759037, "grad_norm": 9.175800323486328, "learning_rate": 4.921844660194175e-05, "loss": 0.2156, "step": 10461 }, { "epoch": 101.44698795180723, "grad_norm": 8.209308624267578, "learning_rate": 4.921359223300971e-05, "loss": 0.2937, "step": 10462 }, { "epoch": 101.4566265060241, "grad_norm": 15.320890426635742, "learning_rate": 4.920873786407767e-05, "loss": 0.3013, "step": 10463 }, { "epoch": 101.46626506024096, "grad_norm": 7.845851421356201, "learning_rate": 4.920388349514563e-05, "loss": 0.2252, "step": 10464 }, { "epoch": 101.47590361445783, "grad_norm": 5.93715238571167, "learning_rate": 4.9199029126213595e-05, "loss": 0.2631, "step": 10465 }, { "epoch": 101.4855421686747, "grad_norm": 10.34353256225586, "learning_rate": 4.919417475728156e-05, "loss": 0.31, "step": 10466 }, { "epoch": 101.49518072289156, "grad_norm": 12.451098442077637, "learning_rate": 4.918932038834952e-05, "loss": 0.3899, "step": 10467 }, { "epoch": 101.50481927710844, "grad_norm": 12.636528015136719, "learning_rate": 4.918446601941748e-05, "loss": 0.4665, "step": 10468 }, { "epoch": 101.5144578313253, "grad_norm": 8.515251159667969, "learning_rate": 4.917961165048544e-05, "loss": 0.2121, "step": 10469 }, { "epoch": 101.52409638554217, "grad_norm": 8.010981559753418, "learning_rate": 4.9174757281553404e-05, "loss": 0.318, "step": 10470 }, { "epoch": 101.53373493975904, "grad_norm": 6.103548526763916, "learning_rate": 4.916990291262136e-05, "loss": 0.2136, "step": 10471 }, { "epoch": 101.5433734939759, "grad_norm": 8.642197608947754, "learning_rate": 4.9165048543689326e-05, "loss": 0.4086, "step": 10472 }, { "epoch": 101.55301204819277, "grad_norm": 5.545072078704834, "learning_rate": 4.9160194174757284e-05, "loss": 0.268, "step": 10473 }, { "epoch": 101.56265060240963, "grad_norm": 4.40657901763916, "learning_rate": 4.915533980582525e-05, "loss": 0.238, "step": 10474 }, { "epoch": 101.57228915662651, "grad_norm": 11.475276947021484, "learning_rate": 4.9150485436893206e-05, "loss": 0.202, "step": 10475 }, { "epoch": 101.58192771084337, "grad_norm": 9.217116355895996, "learning_rate": 4.914563106796117e-05, "loss": 0.355, "step": 10476 }, { "epoch": 101.59156626506024, "grad_norm": 15.230427742004395, "learning_rate": 4.914077669902913e-05, "loss": 0.1465, "step": 10477 }, { "epoch": 101.60120481927711, "grad_norm": 19.451902389526367, "learning_rate": 4.9135922330097086e-05, "loss": 0.144, "step": 10478 }, { "epoch": 101.61084337349398, "grad_norm": 6.2453789710998535, "learning_rate": 4.913106796116505e-05, "loss": 0.213, "step": 10479 }, { "epoch": 101.62048192771084, "grad_norm": 13.439613342285156, "learning_rate": 4.912621359223301e-05, "loss": 0.3565, "step": 10480 }, { "epoch": 101.6301204819277, "grad_norm": 10.789891242980957, "learning_rate": 4.912135922330097e-05, "loss": 0.254, "step": 10481 }, { "epoch": 101.63975903614458, "grad_norm": 13.189665794372559, "learning_rate": 4.911650485436893e-05, "loss": 0.5131, "step": 10482 }, { "epoch": 101.64939759036145, "grad_norm": 11.825244903564453, "learning_rate": 4.9111650485436895e-05, "loss": 0.2497, "step": 10483 }, { "epoch": 101.65903614457831, "grad_norm": 10.01165771484375, "learning_rate": 4.910679611650485e-05, "loss": 0.3104, "step": 10484 }, { "epoch": 101.66867469879519, "grad_norm": 4.841965675354004, "learning_rate": 4.910194174757282e-05, "loss": 0.2396, "step": 10485 }, { "epoch": 101.67831325301205, "grad_norm": 20.718046188354492, "learning_rate": 4.9097087378640776e-05, "loss": 0.3077, "step": 10486 }, { "epoch": 101.68795180722891, "grad_norm": 7.751186847686768, "learning_rate": 4.909223300970874e-05, "loss": 0.1807, "step": 10487 }, { "epoch": 101.69759036144578, "grad_norm": 13.60242748260498, "learning_rate": 4.90873786407767e-05, "loss": 0.2942, "step": 10488 }, { "epoch": 101.70722891566265, "grad_norm": 7.9397149085998535, "learning_rate": 4.908252427184466e-05, "loss": 0.2928, "step": 10489 }, { "epoch": 101.71686746987952, "grad_norm": 26.81671714782715, "learning_rate": 4.907766990291263e-05, "loss": 0.2084, "step": 10490 }, { "epoch": 101.72650602409638, "grad_norm": 10.126042366027832, "learning_rate": 4.9072815533980585e-05, "loss": 0.2086, "step": 10491 }, { "epoch": 101.73614457831326, "grad_norm": 9.274029731750488, "learning_rate": 4.906796116504855e-05, "loss": 0.1406, "step": 10492 }, { "epoch": 101.74578313253012, "grad_norm": 9.292333602905273, "learning_rate": 4.906310679611651e-05, "loss": 0.1457, "step": 10493 }, { "epoch": 101.75542168674698, "grad_norm": 6.121618270874023, "learning_rate": 4.905825242718447e-05, "loss": 0.1951, "step": 10494 }, { "epoch": 101.76506024096386, "grad_norm": 11.610795974731445, "learning_rate": 4.905339805825243e-05, "loss": 0.2555, "step": 10495 }, { "epoch": 101.77469879518073, "grad_norm": 5.596935749053955, "learning_rate": 4.9048543689320394e-05, "loss": 0.3155, "step": 10496 }, { "epoch": 101.78433734939759, "grad_norm": 10.037004470825195, "learning_rate": 4.904368932038835e-05, "loss": 0.4657, "step": 10497 }, { "epoch": 101.79397590361445, "grad_norm": 2.9880542755126953, "learning_rate": 4.9038834951456316e-05, "loss": 0.3364, "step": 10498 }, { "epoch": 101.80361445783133, "grad_norm": 4.739204406738281, "learning_rate": 4.9033980582524274e-05, "loss": 0.1836, "step": 10499 }, { "epoch": 101.8132530120482, "grad_norm": 6.614352226257324, "learning_rate": 4.902912621359224e-05, "loss": 0.1707, "step": 10500 }, { "epoch": 101.82289156626506, "grad_norm": 26.009944915771484, "learning_rate": 4.9024271844660196e-05, "loss": 0.376, "step": 10501 }, { "epoch": 101.83253012048193, "grad_norm": 4.225818634033203, "learning_rate": 4.901941747572816e-05, "loss": 0.2683, "step": 10502 }, { "epoch": 101.8421686746988, "grad_norm": 9.760170936584473, "learning_rate": 4.901456310679612e-05, "loss": 0.1809, "step": 10503 }, { "epoch": 101.85180722891566, "grad_norm": 14.220972061157227, "learning_rate": 4.900970873786408e-05, "loss": 0.3004, "step": 10504 }, { "epoch": 101.86144578313252, "grad_norm": 11.56533145904541, "learning_rate": 4.900485436893204e-05, "loss": 0.2756, "step": 10505 }, { "epoch": 101.8710843373494, "grad_norm": 9.851655006408691, "learning_rate": 4.9e-05, "loss": 0.3077, "step": 10506 }, { "epoch": 101.88072289156626, "grad_norm": 6.045978546142578, "learning_rate": 4.899514563106796e-05, "loss": 0.2037, "step": 10507 }, { "epoch": 101.89036144578313, "grad_norm": 19.409503936767578, "learning_rate": 4.899029126213592e-05, "loss": 0.432, "step": 10508 }, { "epoch": 101.9, "grad_norm": 45.63804626464844, "learning_rate": 4.8985436893203885e-05, "loss": 0.251, "step": 10509 }, { "epoch": 101.90963855421687, "grad_norm": 6.581319332122803, "learning_rate": 4.898058252427184e-05, "loss": 0.226, "step": 10510 }, { "epoch": 101.91927710843373, "grad_norm": 14.014259338378906, "learning_rate": 4.897572815533981e-05, "loss": 0.3534, "step": 10511 }, { "epoch": 101.9289156626506, "grad_norm": 8.159643173217773, "learning_rate": 4.8970873786407765e-05, "loss": 0.2691, "step": 10512 }, { "epoch": 101.93855421686747, "grad_norm": 2.902876377105713, "learning_rate": 4.896601941747573e-05, "loss": 0.2019, "step": 10513 }, { "epoch": 101.94819277108434, "grad_norm": 10.897943496704102, "learning_rate": 4.896116504854369e-05, "loss": 0.2637, "step": 10514 }, { "epoch": 101.9578313253012, "grad_norm": 44.46641540527344, "learning_rate": 4.895631067961166e-05, "loss": 0.2574, "step": 10515 }, { "epoch": 101.96746987951808, "grad_norm": 8.442289352416992, "learning_rate": 4.895145631067962e-05, "loss": 0.2461, "step": 10516 }, { "epoch": 101.97710843373494, "grad_norm": 3.5624499320983887, "learning_rate": 4.8946601941747574e-05, "loss": 0.2616, "step": 10517 }, { "epoch": 101.9867469879518, "grad_norm": 10.796146392822266, "learning_rate": 4.894174757281554e-05, "loss": 0.2313, "step": 10518 }, { "epoch": 102.00240963855421, "grad_norm": 33.50489044189453, "learning_rate": 4.89368932038835e-05, "loss": 0.3172, "step": 10519 }, { "epoch": 102.01204819277109, "grad_norm": 16.99161148071289, "learning_rate": 4.893203883495146e-05, "loss": 0.1869, "step": 10520 }, { "epoch": 102.02168674698795, "grad_norm": 5.019536018371582, "learning_rate": 4.892718446601942e-05, "loss": 0.1719, "step": 10521 }, { "epoch": 102.03132530120482, "grad_norm": 33.39769744873047, "learning_rate": 4.8922330097087384e-05, "loss": 0.2889, "step": 10522 }, { "epoch": 102.04096385542168, "grad_norm": 7.2275872230529785, "learning_rate": 4.891747572815534e-05, "loss": 0.1194, "step": 10523 }, { "epoch": 102.05060240963856, "grad_norm": 9.648988723754883, "learning_rate": 4.8912621359223306e-05, "loss": 0.2999, "step": 10524 }, { "epoch": 102.06024096385542, "grad_norm": 33.51769256591797, "learning_rate": 4.8907766990291264e-05, "loss": 0.0924, "step": 10525 }, { "epoch": 102.06987951807228, "grad_norm": 9.226512908935547, "learning_rate": 4.890291262135923e-05, "loss": 0.33, "step": 10526 }, { "epoch": 102.07951807228916, "grad_norm": 6.386602401733398, "learning_rate": 4.8898058252427186e-05, "loss": 0.2477, "step": 10527 }, { "epoch": 102.08915662650602, "grad_norm": 6.949314117431641, "learning_rate": 4.889320388349515e-05, "loss": 0.1646, "step": 10528 }, { "epoch": 102.09879518072289, "grad_norm": 6.935786247253418, "learning_rate": 4.888834951456311e-05, "loss": 0.1862, "step": 10529 }, { "epoch": 102.10843373493977, "grad_norm": 21.597423553466797, "learning_rate": 4.888349514563107e-05, "loss": 0.377, "step": 10530 }, { "epoch": 102.11807228915663, "grad_norm": 6.707244873046875, "learning_rate": 4.887864077669903e-05, "loss": 0.1894, "step": 10531 }, { "epoch": 102.12771084337349, "grad_norm": 10.035449981689453, "learning_rate": 4.887378640776699e-05, "loss": 0.2965, "step": 10532 }, { "epoch": 102.13734939759036, "grad_norm": 7.052798271179199, "learning_rate": 4.886893203883495e-05, "loss": 0.2183, "step": 10533 }, { "epoch": 102.14698795180723, "grad_norm": 7.627194404602051, "learning_rate": 4.886407766990291e-05, "loss": 0.1318, "step": 10534 }, { "epoch": 102.1566265060241, "grad_norm": 6.086745262145996, "learning_rate": 4.8859223300970875e-05, "loss": 0.211, "step": 10535 }, { "epoch": 102.16626506024096, "grad_norm": 5.058073997497559, "learning_rate": 4.885436893203883e-05, "loss": 0.1206, "step": 10536 }, { "epoch": 102.17590361445784, "grad_norm": 14.297196388244629, "learning_rate": 4.88495145631068e-05, "loss": 0.2394, "step": 10537 }, { "epoch": 102.1855421686747, "grad_norm": 15.061767578125, "learning_rate": 4.8844660194174755e-05, "loss": 0.315, "step": 10538 }, { "epoch": 102.19518072289156, "grad_norm": 34.5284538269043, "learning_rate": 4.883980582524272e-05, "loss": 0.3781, "step": 10539 }, { "epoch": 102.20481927710843, "grad_norm": 5.53499698638916, "learning_rate": 4.8834951456310684e-05, "loss": 0.1908, "step": 10540 }, { "epoch": 102.2144578313253, "grad_norm": 13.38656234741211, "learning_rate": 4.883009708737865e-05, "loss": 0.3346, "step": 10541 }, { "epoch": 102.22409638554217, "grad_norm": 3.0670368671417236, "learning_rate": 4.8825242718446606e-05, "loss": 0.1376, "step": 10542 }, { "epoch": 102.23373493975903, "grad_norm": 15.153250694274902, "learning_rate": 4.8820388349514564e-05, "loss": 0.402, "step": 10543 }, { "epoch": 102.24337349397591, "grad_norm": 6.746090412139893, "learning_rate": 4.881553398058253e-05, "loss": 0.4287, "step": 10544 }, { "epoch": 102.25301204819277, "grad_norm": 8.18857479095459, "learning_rate": 4.8810679611650486e-05, "loss": 0.4019, "step": 10545 }, { "epoch": 102.26265060240964, "grad_norm": 10.676623344421387, "learning_rate": 4.880582524271845e-05, "loss": 0.3209, "step": 10546 }, { "epoch": 102.2722891566265, "grad_norm": 14.713613510131836, "learning_rate": 4.880097087378641e-05, "loss": 0.2146, "step": 10547 }, { "epoch": 102.28192771084338, "grad_norm": 15.18364143371582, "learning_rate": 4.879611650485437e-05, "loss": 0.16, "step": 10548 }, { "epoch": 102.29156626506024, "grad_norm": 21.755155563354492, "learning_rate": 4.879126213592233e-05, "loss": 0.3309, "step": 10549 }, { "epoch": 102.3012048192771, "grad_norm": 8.88045597076416, "learning_rate": 4.8786407766990296e-05, "loss": 0.3633, "step": 10550 }, { "epoch": 102.31084337349398, "grad_norm": 3.934389591217041, "learning_rate": 4.878155339805825e-05, "loss": 0.1605, "step": 10551 }, { "epoch": 102.32048192771084, "grad_norm": 11.538455963134766, "learning_rate": 4.877669902912622e-05, "loss": 0.2171, "step": 10552 }, { "epoch": 102.33012048192771, "grad_norm": 7.233488082885742, "learning_rate": 4.8771844660194176e-05, "loss": 0.1225, "step": 10553 }, { "epoch": 102.33975903614459, "grad_norm": 6.5146307945251465, "learning_rate": 4.876699029126214e-05, "loss": 0.2232, "step": 10554 }, { "epoch": 102.34939759036145, "grad_norm": 5.660515308380127, "learning_rate": 4.87621359223301e-05, "loss": 0.2767, "step": 10555 }, { "epoch": 102.35903614457831, "grad_norm": 19.311067581176758, "learning_rate": 4.875728155339806e-05, "loss": 0.3024, "step": 10556 }, { "epoch": 102.36867469879518, "grad_norm": 5.605024337768555, "learning_rate": 4.875242718446602e-05, "loss": 0.2482, "step": 10557 }, { "epoch": 102.37831325301205, "grad_norm": 5.825705528259277, "learning_rate": 4.8747572815533985e-05, "loss": 0.154, "step": 10558 }, { "epoch": 102.38795180722892, "grad_norm": 5.070528507232666, "learning_rate": 4.874271844660194e-05, "loss": 0.1398, "step": 10559 }, { "epoch": 102.39759036144578, "grad_norm": 14.79703140258789, "learning_rate": 4.87378640776699e-05, "loss": 0.255, "step": 10560 }, { "epoch": 102.40722891566266, "grad_norm": 12.797562599182129, "learning_rate": 4.8733009708737865e-05, "loss": 0.3499, "step": 10561 }, { "epoch": 102.41686746987952, "grad_norm": 32.34223556518555, "learning_rate": 4.872815533980582e-05, "loss": 0.2609, "step": 10562 }, { "epoch": 102.42650602409638, "grad_norm": 10.837564468383789, "learning_rate": 4.872330097087379e-05, "loss": 0.3345, "step": 10563 }, { "epoch": 102.43614457831325, "grad_norm": 8.655692100524902, "learning_rate": 4.8718446601941745e-05, "loss": 0.3473, "step": 10564 }, { "epoch": 102.44578313253012, "grad_norm": 14.509407997131348, "learning_rate": 4.871359223300971e-05, "loss": 0.345, "step": 10565 }, { "epoch": 102.45542168674699, "grad_norm": 9.561939239501953, "learning_rate": 4.8708737864077674e-05, "loss": 0.1639, "step": 10566 }, { "epoch": 102.46506024096385, "grad_norm": 9.065958023071289, "learning_rate": 4.870388349514564e-05, "loss": 0.2723, "step": 10567 }, { "epoch": 102.47469879518073, "grad_norm": 2.689343214035034, "learning_rate": 4.8699029126213596e-05, "loss": 0.112, "step": 10568 }, { "epoch": 102.48433734939759, "grad_norm": 17.026052474975586, "learning_rate": 4.869417475728156e-05, "loss": 0.2391, "step": 10569 }, { "epoch": 102.49397590361446, "grad_norm": 4.658711910247803, "learning_rate": 4.868932038834952e-05, "loss": 0.2418, "step": 10570 }, { "epoch": 102.50361445783132, "grad_norm": 24.925434112548828, "learning_rate": 4.8684466019417476e-05, "loss": 0.2823, "step": 10571 }, { "epoch": 102.5132530120482, "grad_norm": 45.34404373168945, "learning_rate": 4.867961165048544e-05, "loss": 0.3803, "step": 10572 }, { "epoch": 102.52289156626506, "grad_norm": 10.754349708557129, "learning_rate": 4.86747572815534e-05, "loss": 0.1053, "step": 10573 }, { "epoch": 102.53253012048192, "grad_norm": 26.3610897064209, "learning_rate": 4.866990291262136e-05, "loss": 0.3967, "step": 10574 }, { "epoch": 102.5421686746988, "grad_norm": 6.813853740692139, "learning_rate": 4.866504854368932e-05, "loss": 0.2918, "step": 10575 }, { "epoch": 102.55180722891566, "grad_norm": 13.60213565826416, "learning_rate": 4.8660194174757285e-05, "loss": 0.3521, "step": 10576 }, { "epoch": 102.56144578313253, "grad_norm": 4.385020732879639, "learning_rate": 4.865533980582524e-05, "loss": 0.0648, "step": 10577 }, { "epoch": 102.5710843373494, "grad_norm": 7.080018997192383, "learning_rate": 4.865048543689321e-05, "loss": 0.2285, "step": 10578 }, { "epoch": 102.58072289156627, "grad_norm": 10.06888198852539, "learning_rate": 4.8645631067961165e-05, "loss": 0.4788, "step": 10579 }, { "epoch": 102.59036144578313, "grad_norm": 5.95328950881958, "learning_rate": 4.864077669902913e-05, "loss": 0.0895, "step": 10580 }, { "epoch": 102.6, "grad_norm": 9.05016040802002, "learning_rate": 4.863592233009709e-05, "loss": 0.3528, "step": 10581 }, { "epoch": 102.60963855421687, "grad_norm": 9.023418426513672, "learning_rate": 4.863106796116505e-05, "loss": 0.154, "step": 10582 }, { "epoch": 102.61927710843374, "grad_norm": 7.777126312255859, "learning_rate": 4.862621359223301e-05, "loss": 0.3272, "step": 10583 }, { "epoch": 102.6289156626506, "grad_norm": 29.277891159057617, "learning_rate": 4.8621359223300975e-05, "loss": 0.3589, "step": 10584 }, { "epoch": 102.63855421686748, "grad_norm": 8.932601928710938, "learning_rate": 4.861650485436893e-05, "loss": 0.2726, "step": 10585 }, { "epoch": 102.64819277108434, "grad_norm": 13.082597732543945, "learning_rate": 4.86116504854369e-05, "loss": 0.2159, "step": 10586 }, { "epoch": 102.6578313253012, "grad_norm": 12.723052024841309, "learning_rate": 4.8606796116504855e-05, "loss": 0.1927, "step": 10587 }, { "epoch": 102.66746987951807, "grad_norm": 12.64748477935791, "learning_rate": 4.860194174757281e-05, "loss": 0.2162, "step": 10588 }, { "epoch": 102.67710843373494, "grad_norm": 7.311245441436768, "learning_rate": 4.859708737864078e-05, "loss": 0.2135, "step": 10589 }, { "epoch": 102.6867469879518, "grad_norm": 3.168067693710327, "learning_rate": 4.8592233009708735e-05, "loss": 0.3369, "step": 10590 }, { "epoch": 102.69638554216867, "grad_norm": 15.725028038024902, "learning_rate": 4.8587378640776706e-05, "loss": 0.2831, "step": 10591 }, { "epoch": 102.70602409638555, "grad_norm": 7.007607936859131, "learning_rate": 4.8582524271844664e-05, "loss": 0.1882, "step": 10592 }, { "epoch": 102.71566265060241, "grad_norm": 9.522270202636719, "learning_rate": 4.857766990291263e-05, "loss": 0.2747, "step": 10593 }, { "epoch": 102.72530120481927, "grad_norm": 8.8903169631958, "learning_rate": 4.8572815533980586e-05, "loss": 0.1896, "step": 10594 }, { "epoch": 102.73493975903614, "grad_norm": 14.724699974060059, "learning_rate": 4.856796116504855e-05, "loss": 0.3048, "step": 10595 }, { "epoch": 102.74457831325302, "grad_norm": 16.245176315307617, "learning_rate": 4.856310679611651e-05, "loss": 0.1554, "step": 10596 }, { "epoch": 102.75421686746988, "grad_norm": 12.426590919494629, "learning_rate": 4.855825242718447e-05, "loss": 0.2935, "step": 10597 }, { "epoch": 102.76385542168674, "grad_norm": 22.8065185546875, "learning_rate": 4.855339805825243e-05, "loss": 0.4762, "step": 10598 }, { "epoch": 102.77349397590362, "grad_norm": 8.94581127166748, "learning_rate": 4.854854368932039e-05, "loss": 0.2695, "step": 10599 }, { "epoch": 102.78313253012048, "grad_norm": 21.76180076599121, "learning_rate": 4.854368932038835e-05, "loss": 0.3983, "step": 10600 }, { "epoch": 102.79277108433735, "grad_norm": 6.731939792633057, "learning_rate": 4.853883495145631e-05, "loss": 0.3558, "step": 10601 }, { "epoch": 102.80240963855422, "grad_norm": 7.82069206237793, "learning_rate": 4.8533980582524275e-05, "loss": 0.2475, "step": 10602 }, { "epoch": 102.81204819277109, "grad_norm": 16.352109909057617, "learning_rate": 4.852912621359223e-05, "loss": 0.1391, "step": 10603 }, { "epoch": 102.82168674698795, "grad_norm": 12.933804512023926, "learning_rate": 4.85242718446602e-05, "loss": 0.3777, "step": 10604 }, { "epoch": 102.83132530120481, "grad_norm": 25.1552791595459, "learning_rate": 4.8519417475728155e-05, "loss": 0.3247, "step": 10605 }, { "epoch": 102.84096385542169, "grad_norm": 23.10591697692871, "learning_rate": 4.851456310679612e-05, "loss": 0.3218, "step": 10606 }, { "epoch": 102.85060240963855, "grad_norm": 20.217662811279297, "learning_rate": 4.850970873786408e-05, "loss": 0.3513, "step": 10607 }, { "epoch": 102.86024096385542, "grad_norm": 6.662800312042236, "learning_rate": 4.850485436893204e-05, "loss": 0.314, "step": 10608 }, { "epoch": 102.8698795180723, "grad_norm": 12.774943351745605, "learning_rate": 4.85e-05, "loss": 0.2541, "step": 10609 }, { "epoch": 102.87951807228916, "grad_norm": 6.404486656188965, "learning_rate": 4.8495145631067964e-05, "loss": 0.2063, "step": 10610 }, { "epoch": 102.88915662650602, "grad_norm": 13.61378288269043, "learning_rate": 4.849029126213592e-05, "loss": 0.3585, "step": 10611 }, { "epoch": 102.89879518072289, "grad_norm": 28.825645446777344, "learning_rate": 4.848543689320389e-05, "loss": 0.257, "step": 10612 }, { "epoch": 102.90843373493976, "grad_norm": 7.617615699768066, "learning_rate": 4.8480582524271844e-05, "loss": 0.4209, "step": 10613 }, { "epoch": 102.91807228915663, "grad_norm": 20.16839599609375, "learning_rate": 4.847572815533981e-05, "loss": 0.3618, "step": 10614 }, { "epoch": 102.92771084337349, "grad_norm": 6.52932071685791, "learning_rate": 4.847087378640777e-05, "loss": 0.1999, "step": 10615 }, { "epoch": 102.93734939759037, "grad_norm": 6.996838569641113, "learning_rate": 4.846601941747573e-05, "loss": 0.2512, "step": 10616 }, { "epoch": 102.94698795180723, "grad_norm": 8.587302207946777, "learning_rate": 4.8461165048543696e-05, "loss": 0.4171, "step": 10617 }, { "epoch": 102.9566265060241, "grad_norm": 10.101849555969238, "learning_rate": 4.8456310679611654e-05, "loss": 0.1782, "step": 10618 }, { "epoch": 102.96626506024096, "grad_norm": 4.550143718719482, "learning_rate": 4.845145631067962e-05, "loss": 0.1768, "step": 10619 }, { "epoch": 102.97590361445783, "grad_norm": 6.247494697570801, "learning_rate": 4.8446601941747576e-05, "loss": 0.165, "step": 10620 }, { "epoch": 102.9855421686747, "grad_norm": 18.718050003051758, "learning_rate": 4.844174757281554e-05, "loss": 0.5353, "step": 10621 }, { "epoch": 103.0012048192771, "grad_norm": 19.46883201599121, "learning_rate": 4.84368932038835e-05, "loss": 0.2778, "step": 10622 }, { "epoch": 103.01084337349397, "grad_norm": 2.915189027786255, "learning_rate": 4.843203883495146e-05, "loss": 0.0937, "step": 10623 }, { "epoch": 103.02048192771085, "grad_norm": 9.615815162658691, "learning_rate": 4.842718446601942e-05, "loss": 0.3067, "step": 10624 }, { "epoch": 103.03012048192771, "grad_norm": 9.789678573608398, "learning_rate": 4.8422330097087385e-05, "loss": 0.3966, "step": 10625 }, { "epoch": 103.03975903614457, "grad_norm": 4.93449068069458, "learning_rate": 4.841747572815534e-05, "loss": 0.2552, "step": 10626 }, { "epoch": 103.04939759036145, "grad_norm": 17.768001556396484, "learning_rate": 4.84126213592233e-05, "loss": 0.3585, "step": 10627 }, { "epoch": 103.05903614457831, "grad_norm": 18.498430252075195, "learning_rate": 4.8407766990291265e-05, "loss": 0.4942, "step": 10628 }, { "epoch": 103.06867469879518, "grad_norm": 6.463723182678223, "learning_rate": 4.840291262135922e-05, "loss": 0.3907, "step": 10629 }, { "epoch": 103.07831325301204, "grad_norm": 8.127481460571289, "learning_rate": 4.839805825242719e-05, "loss": 0.2796, "step": 10630 }, { "epoch": 103.08795180722892, "grad_norm": 20.24444007873535, "learning_rate": 4.8393203883495145e-05, "loss": 0.263, "step": 10631 }, { "epoch": 103.09759036144578, "grad_norm": 7.528254508972168, "learning_rate": 4.838834951456311e-05, "loss": 0.2485, "step": 10632 }, { "epoch": 103.10722891566265, "grad_norm": 8.512696266174316, "learning_rate": 4.838349514563107e-05, "loss": 0.2449, "step": 10633 }, { "epoch": 103.11686746987952, "grad_norm": 5.878077030181885, "learning_rate": 4.837864077669903e-05, "loss": 0.0939, "step": 10634 }, { "epoch": 103.12650602409639, "grad_norm": 14.408198356628418, "learning_rate": 4.837378640776699e-05, "loss": 0.27, "step": 10635 }, { "epoch": 103.13614457831325, "grad_norm": 19.543014526367188, "learning_rate": 4.8368932038834954e-05, "loss": 0.2547, "step": 10636 }, { "epoch": 103.14578313253013, "grad_norm": 9.31975269317627, "learning_rate": 4.836407766990291e-05, "loss": 0.34, "step": 10637 }, { "epoch": 103.15542168674699, "grad_norm": 19.408578872680664, "learning_rate": 4.8359223300970876e-05, "loss": 0.3363, "step": 10638 }, { "epoch": 103.16506024096385, "grad_norm": 28.95340919494629, "learning_rate": 4.8354368932038834e-05, "loss": 0.2728, "step": 10639 }, { "epoch": 103.17469879518072, "grad_norm": 6.2682108879089355, "learning_rate": 4.83495145631068e-05, "loss": 0.1082, "step": 10640 }, { "epoch": 103.1843373493976, "grad_norm": 4.444620609283447, "learning_rate": 4.834466019417476e-05, "loss": 0.1439, "step": 10641 }, { "epoch": 103.19397590361446, "grad_norm": 13.922284126281738, "learning_rate": 4.833980582524272e-05, "loss": 0.3862, "step": 10642 }, { "epoch": 103.20361445783132, "grad_norm": 7.273313522338867, "learning_rate": 4.8334951456310686e-05, "loss": 0.1352, "step": 10643 }, { "epoch": 103.2132530120482, "grad_norm": 17.102886199951172, "learning_rate": 4.833009708737864e-05, "loss": 0.3517, "step": 10644 }, { "epoch": 103.22289156626506, "grad_norm": 4.588693618774414, "learning_rate": 4.832524271844661e-05, "loss": 0.1558, "step": 10645 }, { "epoch": 103.23253012048193, "grad_norm": 8.011946678161621, "learning_rate": 4.8320388349514566e-05, "loss": 0.2734, "step": 10646 }, { "epoch": 103.24216867469879, "grad_norm": 23.303890228271484, "learning_rate": 4.831553398058253e-05, "loss": 0.2314, "step": 10647 }, { "epoch": 103.25180722891567, "grad_norm": 7.544207572937012, "learning_rate": 4.831067961165049e-05, "loss": 0.1496, "step": 10648 }, { "epoch": 103.26144578313253, "grad_norm": 20.387380599975586, "learning_rate": 4.830582524271845e-05, "loss": 0.3533, "step": 10649 }, { "epoch": 103.2710843373494, "grad_norm": 26.571149826049805, "learning_rate": 4.830097087378641e-05, "loss": 0.2357, "step": 10650 }, { "epoch": 103.28072289156627, "grad_norm": 14.000056266784668, "learning_rate": 4.8296116504854375e-05, "loss": 0.3023, "step": 10651 }, { "epoch": 103.29036144578313, "grad_norm": 9.768798828125, "learning_rate": 4.829126213592233e-05, "loss": 0.3042, "step": 10652 }, { "epoch": 103.3, "grad_norm": 12.797907829284668, "learning_rate": 4.82864077669903e-05, "loss": 0.219, "step": 10653 }, { "epoch": 103.30963855421686, "grad_norm": 9.39315128326416, "learning_rate": 4.8281553398058255e-05, "loss": 0.152, "step": 10654 }, { "epoch": 103.31927710843374, "grad_norm": 9.985252380371094, "learning_rate": 4.827669902912621e-05, "loss": 0.4006, "step": 10655 }, { "epoch": 103.3289156626506, "grad_norm": 13.329291343688965, "learning_rate": 4.827184466019418e-05, "loss": 0.3788, "step": 10656 }, { "epoch": 103.33855421686746, "grad_norm": 9.549589157104492, "learning_rate": 4.8266990291262135e-05, "loss": 0.2814, "step": 10657 }, { "epoch": 103.34819277108434, "grad_norm": 12.217074394226074, "learning_rate": 4.82621359223301e-05, "loss": 0.2816, "step": 10658 }, { "epoch": 103.3578313253012, "grad_norm": 14.220982551574707, "learning_rate": 4.825728155339806e-05, "loss": 0.2258, "step": 10659 }, { "epoch": 103.36746987951807, "grad_norm": 7.621799945831299, "learning_rate": 4.825242718446602e-05, "loss": 0.2919, "step": 10660 }, { "epoch": 103.37710843373495, "grad_norm": 8.810982704162598, "learning_rate": 4.824757281553398e-05, "loss": 0.2263, "step": 10661 }, { "epoch": 103.38674698795181, "grad_norm": 11.278829574584961, "learning_rate": 4.8242718446601944e-05, "loss": 0.3477, "step": 10662 }, { "epoch": 103.39638554216867, "grad_norm": 6.333271026611328, "learning_rate": 4.82378640776699e-05, "loss": 0.1845, "step": 10663 }, { "epoch": 103.40602409638554, "grad_norm": 20.89371681213379, "learning_rate": 4.8233009708737866e-05, "loss": 0.3779, "step": 10664 }, { "epoch": 103.41566265060241, "grad_norm": 8.041061401367188, "learning_rate": 4.8228155339805824e-05, "loss": 0.3398, "step": 10665 }, { "epoch": 103.42530120481928, "grad_norm": 9.303934097290039, "learning_rate": 4.822330097087379e-05, "loss": 0.2718, "step": 10666 }, { "epoch": 103.43493975903614, "grad_norm": 12.026070594787598, "learning_rate": 4.821844660194175e-05, "loss": 0.2169, "step": 10667 }, { "epoch": 103.44457831325302, "grad_norm": 10.383219718933105, "learning_rate": 4.821359223300971e-05, "loss": 0.3665, "step": 10668 }, { "epoch": 103.45421686746988, "grad_norm": 20.802576065063477, "learning_rate": 4.8208737864077675e-05, "loss": 0.4281, "step": 10669 }, { "epoch": 103.46385542168674, "grad_norm": 11.084781646728516, "learning_rate": 4.820388349514563e-05, "loss": 0.279, "step": 10670 }, { "epoch": 103.47349397590361, "grad_norm": 10.493735313415527, "learning_rate": 4.81990291262136e-05, "loss": 0.3261, "step": 10671 }, { "epoch": 103.48313253012049, "grad_norm": 7.298264980316162, "learning_rate": 4.8194174757281555e-05, "loss": 0.192, "step": 10672 }, { "epoch": 103.49277108433735, "grad_norm": 16.91525650024414, "learning_rate": 4.818932038834952e-05, "loss": 0.2757, "step": 10673 }, { "epoch": 103.50240963855421, "grad_norm": 7.729198455810547, "learning_rate": 4.818446601941748e-05, "loss": 0.3526, "step": 10674 }, { "epoch": 103.51204819277109, "grad_norm": 9.501959800720215, "learning_rate": 4.817961165048544e-05, "loss": 0.427, "step": 10675 }, { "epoch": 103.52168674698795, "grad_norm": 12.912260055541992, "learning_rate": 4.81747572815534e-05, "loss": 0.2199, "step": 10676 }, { "epoch": 103.53132530120482, "grad_norm": 7.752274990081787, "learning_rate": 4.8169902912621365e-05, "loss": 0.1227, "step": 10677 }, { "epoch": 103.54096385542168, "grad_norm": 17.22664451599121, "learning_rate": 4.816504854368932e-05, "loss": 0.2324, "step": 10678 }, { "epoch": 103.55060240963856, "grad_norm": 8.5401029586792, "learning_rate": 4.816019417475729e-05, "loss": 0.1576, "step": 10679 }, { "epoch": 103.56024096385542, "grad_norm": 11.107245445251465, "learning_rate": 4.8155339805825245e-05, "loss": 0.3483, "step": 10680 }, { "epoch": 103.56987951807228, "grad_norm": 12.598564147949219, "learning_rate": 4.81504854368932e-05, "loss": 0.2685, "step": 10681 }, { "epoch": 103.57951807228916, "grad_norm": 12.495838165283203, "learning_rate": 4.814563106796117e-05, "loss": 0.2522, "step": 10682 }, { "epoch": 103.58915662650602, "grad_norm": 15.644661903381348, "learning_rate": 4.8140776699029125e-05, "loss": 0.3394, "step": 10683 }, { "epoch": 103.59879518072289, "grad_norm": 19.15281105041504, "learning_rate": 4.813592233009709e-05, "loss": 0.3086, "step": 10684 }, { "epoch": 103.60843373493977, "grad_norm": 7.945707321166992, "learning_rate": 4.813106796116505e-05, "loss": 0.2983, "step": 10685 }, { "epoch": 103.61807228915663, "grad_norm": 5.068421840667725, "learning_rate": 4.812621359223301e-05, "loss": 0.2337, "step": 10686 }, { "epoch": 103.62771084337349, "grad_norm": 8.548773765563965, "learning_rate": 4.812135922330097e-05, "loss": 0.3189, "step": 10687 }, { "epoch": 103.63734939759036, "grad_norm": 11.512382507324219, "learning_rate": 4.8116504854368934e-05, "loss": 0.1772, "step": 10688 }, { "epoch": 103.64698795180723, "grad_norm": 8.065333366394043, "learning_rate": 4.811165048543689e-05, "loss": 0.1769, "step": 10689 }, { "epoch": 103.6566265060241, "grad_norm": 4.302774429321289, "learning_rate": 4.8106796116504856e-05, "loss": 0.2281, "step": 10690 }, { "epoch": 103.66626506024096, "grad_norm": 8.289787292480469, "learning_rate": 4.8101941747572814e-05, "loss": 0.1134, "step": 10691 }, { "epoch": 103.67590361445784, "grad_norm": 9.245046615600586, "learning_rate": 4.809708737864078e-05, "loss": 0.3369, "step": 10692 }, { "epoch": 103.6855421686747, "grad_norm": 14.6909818649292, "learning_rate": 4.809223300970874e-05, "loss": 0.1772, "step": 10693 }, { "epoch": 103.69518072289156, "grad_norm": 14.47940731048584, "learning_rate": 4.80873786407767e-05, "loss": 0.2315, "step": 10694 }, { "epoch": 103.70481927710843, "grad_norm": 16.42482566833496, "learning_rate": 4.8082524271844665e-05, "loss": 0.2781, "step": 10695 }, { "epoch": 103.7144578313253, "grad_norm": 6.055596351623535, "learning_rate": 4.807766990291262e-05, "loss": 0.2595, "step": 10696 }, { "epoch": 103.72409638554217, "grad_norm": 4.255891799926758, "learning_rate": 4.807281553398059e-05, "loss": 0.1848, "step": 10697 }, { "epoch": 103.73373493975903, "grad_norm": 9.7466402053833, "learning_rate": 4.8067961165048545e-05, "loss": 0.1875, "step": 10698 }, { "epoch": 103.74337349397591, "grad_norm": 8.852490425109863, "learning_rate": 4.806310679611651e-05, "loss": 0.1677, "step": 10699 }, { "epoch": 103.75301204819277, "grad_norm": 5.434088230133057, "learning_rate": 4.805825242718447e-05, "loss": 0.1392, "step": 10700 }, { "epoch": 103.76265060240964, "grad_norm": 8.80687141418457, "learning_rate": 4.805339805825243e-05, "loss": 0.2101, "step": 10701 }, { "epoch": 103.7722891566265, "grad_norm": 10.760272026062012, "learning_rate": 4.804854368932039e-05, "loss": 0.317, "step": 10702 }, { "epoch": 103.78192771084338, "grad_norm": 9.768549919128418, "learning_rate": 4.8043689320388354e-05, "loss": 0.3077, "step": 10703 }, { "epoch": 103.79156626506024, "grad_norm": 9.421579360961914, "learning_rate": 4.803883495145631e-05, "loss": 0.3558, "step": 10704 }, { "epoch": 103.8012048192771, "grad_norm": 13.301928520202637, "learning_rate": 4.8033980582524277e-05, "loss": 0.1714, "step": 10705 }, { "epoch": 103.81084337349398, "grad_norm": 9.223343849182129, "learning_rate": 4.8029126213592234e-05, "loss": 0.2629, "step": 10706 }, { "epoch": 103.82048192771084, "grad_norm": 19.24188232421875, "learning_rate": 4.80242718446602e-05, "loss": 0.378, "step": 10707 }, { "epoch": 103.83012048192771, "grad_norm": 5.5693488121032715, "learning_rate": 4.801941747572816e-05, "loss": 0.2106, "step": 10708 }, { "epoch": 103.83975903614459, "grad_norm": 63.11090087890625, "learning_rate": 4.8014563106796114e-05, "loss": 0.3125, "step": 10709 }, { "epoch": 103.84939759036145, "grad_norm": 9.450946807861328, "learning_rate": 4.800970873786408e-05, "loss": 0.3883, "step": 10710 }, { "epoch": 103.85903614457831, "grad_norm": 10.53048038482666, "learning_rate": 4.800485436893204e-05, "loss": 0.2628, "step": 10711 }, { "epoch": 103.86867469879518, "grad_norm": 11.446913719177246, "learning_rate": 4.8e-05, "loss": 0.2834, "step": 10712 }, { "epoch": 103.87831325301205, "grad_norm": 7.528823375701904, "learning_rate": 4.799514563106796e-05, "loss": 0.2224, "step": 10713 }, { "epoch": 103.88795180722892, "grad_norm": 20.666637420654297, "learning_rate": 4.7990291262135924e-05, "loss": 0.3494, "step": 10714 }, { "epoch": 103.89759036144578, "grad_norm": 8.545727729797363, "learning_rate": 4.798543689320388e-05, "loss": 0.1392, "step": 10715 }, { "epoch": 103.90722891566266, "grad_norm": 4.679541110992432, "learning_rate": 4.7980582524271846e-05, "loss": 0.1226, "step": 10716 }, { "epoch": 103.91686746987952, "grad_norm": 12.098310470581055, "learning_rate": 4.797572815533981e-05, "loss": 0.2772, "step": 10717 }, { "epoch": 103.92650602409638, "grad_norm": 8.39507007598877, "learning_rate": 4.7970873786407775e-05, "loss": 0.2853, "step": 10718 }, { "epoch": 103.93614457831325, "grad_norm": 26.034626007080078, "learning_rate": 4.796601941747573e-05, "loss": 0.1686, "step": 10719 }, { "epoch": 103.94578313253012, "grad_norm": 15.256402015686035, "learning_rate": 4.796116504854369e-05, "loss": 0.1685, "step": 10720 }, { "epoch": 103.95542168674699, "grad_norm": 7.491292476654053, "learning_rate": 4.7956310679611655e-05, "loss": 0.2151, "step": 10721 }, { "epoch": 103.96506024096385, "grad_norm": 16.798160552978516, "learning_rate": 4.795145631067961e-05, "loss": 0.2077, "step": 10722 }, { "epoch": 103.97469879518073, "grad_norm": 9.325608253479004, "learning_rate": 4.794660194174758e-05, "loss": 0.1614, "step": 10723 }, { "epoch": 103.98433734939759, "grad_norm": 4.2639617919921875, "learning_rate": 4.7941747572815535e-05, "loss": 0.1315, "step": 10724 }, { "epoch": 103.99397590361446, "grad_norm": 12.383275985717773, "learning_rate": 4.79368932038835e-05, "loss": 0.2685, "step": 10725 }, { "epoch": 104.00963855421686, "grad_norm": 6.446348667144775, "learning_rate": 4.793203883495146e-05, "loss": 0.1355, "step": 10726 }, { "epoch": 104.01927710843374, "grad_norm": 8.628710746765137, "learning_rate": 4.792718446601942e-05, "loss": 0.2207, "step": 10727 }, { "epoch": 104.0289156626506, "grad_norm": 5.186100959777832, "learning_rate": 4.792233009708738e-05, "loss": 0.1539, "step": 10728 }, { "epoch": 104.03855421686747, "grad_norm": 10.316165924072266, "learning_rate": 4.7917475728155344e-05, "loss": 0.3031, "step": 10729 }, { "epoch": 104.04819277108433, "grad_norm": 8.247735977172852, "learning_rate": 4.79126213592233e-05, "loss": 0.247, "step": 10730 }, { "epoch": 104.05783132530121, "grad_norm": 11.906643867492676, "learning_rate": 4.7907766990291266e-05, "loss": 0.4228, "step": 10731 }, { "epoch": 104.06746987951807, "grad_norm": 12.128225326538086, "learning_rate": 4.7902912621359224e-05, "loss": 0.3205, "step": 10732 }, { "epoch": 104.07710843373494, "grad_norm": 15.765570640563965, "learning_rate": 4.789805825242719e-05, "loss": 0.3823, "step": 10733 }, { "epoch": 104.08674698795181, "grad_norm": 5.590147018432617, "learning_rate": 4.7893203883495146e-05, "loss": 0.192, "step": 10734 }, { "epoch": 104.09638554216868, "grad_norm": 10.321494102478027, "learning_rate": 4.788834951456311e-05, "loss": 0.3072, "step": 10735 }, { "epoch": 104.10602409638554, "grad_norm": 11.722005844116211, "learning_rate": 4.788349514563107e-05, "loss": 0.3599, "step": 10736 }, { "epoch": 104.1156626506024, "grad_norm": 5.030894756317139, "learning_rate": 4.7878640776699026e-05, "loss": 0.197, "step": 10737 }, { "epoch": 104.12530120481928, "grad_norm": 18.321475982666016, "learning_rate": 4.787378640776699e-05, "loss": 0.2955, "step": 10738 }, { "epoch": 104.13493975903614, "grad_norm": 6.513418674468994, "learning_rate": 4.786893203883495e-05, "loss": 0.3199, "step": 10739 }, { "epoch": 104.144578313253, "grad_norm": 10.646029472351074, "learning_rate": 4.786407766990291e-05, "loss": 0.1587, "step": 10740 }, { "epoch": 104.15421686746988, "grad_norm": 10.889897346496582, "learning_rate": 4.785922330097087e-05, "loss": 0.1856, "step": 10741 }, { "epoch": 104.16385542168675, "grad_norm": 5.439764976501465, "learning_rate": 4.785436893203884e-05, "loss": 0.1228, "step": 10742 }, { "epoch": 104.17349397590361, "grad_norm": 7.557501316070557, "learning_rate": 4.78495145631068e-05, "loss": 0.3456, "step": 10743 }, { "epoch": 104.18313253012049, "grad_norm": 23.098148345947266, "learning_rate": 4.7844660194174765e-05, "loss": 0.2084, "step": 10744 }, { "epoch": 104.19277108433735, "grad_norm": 9.371009826660156, "learning_rate": 4.783980582524272e-05, "loss": 0.4247, "step": 10745 }, { "epoch": 104.20240963855422, "grad_norm": 71.0131607055664, "learning_rate": 4.783495145631069e-05, "loss": 0.5304, "step": 10746 }, { "epoch": 104.21204819277108, "grad_norm": 10.03200912475586, "learning_rate": 4.7830097087378645e-05, "loss": 0.3334, "step": 10747 }, { "epoch": 104.22168674698796, "grad_norm": 8.741973876953125, "learning_rate": 4.78252427184466e-05, "loss": 0.3502, "step": 10748 }, { "epoch": 104.23132530120482, "grad_norm": 22.603923797607422, "learning_rate": 4.782038834951457e-05, "loss": 0.292, "step": 10749 }, { "epoch": 104.24096385542168, "grad_norm": 8.890054702758789, "learning_rate": 4.7815533980582525e-05, "loss": 0.3067, "step": 10750 }, { "epoch": 104.25060240963856, "grad_norm": 10.0189847946167, "learning_rate": 4.781067961165049e-05, "loss": 0.1955, "step": 10751 }, { "epoch": 104.26024096385542, "grad_norm": 5.351115703582764, "learning_rate": 4.780582524271845e-05, "loss": 0.1634, "step": 10752 }, { "epoch": 104.26987951807229, "grad_norm": 9.370790481567383, "learning_rate": 4.780097087378641e-05, "loss": 0.2229, "step": 10753 }, { "epoch": 104.27951807228915, "grad_norm": 10.814506530761719, "learning_rate": 4.779611650485437e-05, "loss": 0.2858, "step": 10754 }, { "epoch": 104.28915662650603, "grad_norm": 27.106475830078125, "learning_rate": 4.7791262135922334e-05, "loss": 0.1277, "step": 10755 }, { "epoch": 104.29879518072289, "grad_norm": 25.015045166015625, "learning_rate": 4.778640776699029e-05, "loss": 0.3017, "step": 10756 }, { "epoch": 104.30843373493975, "grad_norm": 16.62014389038086, "learning_rate": 4.7781553398058256e-05, "loss": 0.2522, "step": 10757 }, { "epoch": 104.31807228915663, "grad_norm": 8.465315818786621, "learning_rate": 4.7776699029126214e-05, "loss": 0.1422, "step": 10758 }, { "epoch": 104.3277108433735, "grad_norm": 16.103361129760742, "learning_rate": 4.777184466019418e-05, "loss": 0.5714, "step": 10759 }, { "epoch": 104.33734939759036, "grad_norm": 13.621963500976562, "learning_rate": 4.7766990291262136e-05, "loss": 0.4646, "step": 10760 }, { "epoch": 104.34698795180722, "grad_norm": 8.910385131835938, "learning_rate": 4.77621359223301e-05, "loss": 0.3454, "step": 10761 }, { "epoch": 104.3566265060241, "grad_norm": 13.676155090332031, "learning_rate": 4.775728155339806e-05, "loss": 0.5297, "step": 10762 }, { "epoch": 104.36626506024096, "grad_norm": 7.6842942237854, "learning_rate": 4.775242718446602e-05, "loss": 0.2829, "step": 10763 }, { "epoch": 104.37590361445783, "grad_norm": 10.804189682006836, "learning_rate": 4.774757281553398e-05, "loss": 0.2832, "step": 10764 }, { "epoch": 104.3855421686747, "grad_norm": 14.50539779663086, "learning_rate": 4.774271844660194e-05, "loss": 0.2818, "step": 10765 }, { "epoch": 104.39518072289157, "grad_norm": 7.247004508972168, "learning_rate": 4.77378640776699e-05, "loss": 0.1221, "step": 10766 }, { "epoch": 104.40481927710843, "grad_norm": 4.949916839599609, "learning_rate": 4.773300970873786e-05, "loss": 0.2215, "step": 10767 }, { "epoch": 104.41445783132531, "grad_norm": 6.04191780090332, "learning_rate": 4.772815533980583e-05, "loss": 0.2863, "step": 10768 }, { "epoch": 104.42409638554217, "grad_norm": 12.47606372833252, "learning_rate": 4.772330097087379e-05, "loss": 0.2971, "step": 10769 }, { "epoch": 104.43373493975903, "grad_norm": 10.362930297851562, "learning_rate": 4.7718446601941754e-05, "loss": 0.2118, "step": 10770 }, { "epoch": 104.4433734939759, "grad_norm": 11.482808113098145, "learning_rate": 4.771359223300971e-05, "loss": 0.3126, "step": 10771 }, { "epoch": 104.45301204819278, "grad_norm": 13.632397651672363, "learning_rate": 4.770873786407768e-05, "loss": 0.3541, "step": 10772 }, { "epoch": 104.46265060240964, "grad_norm": 9.461710929870605, "learning_rate": 4.7703883495145634e-05, "loss": 0.4108, "step": 10773 }, { "epoch": 104.4722891566265, "grad_norm": 10.247410774230957, "learning_rate": 4.76990291262136e-05, "loss": 0.3651, "step": 10774 }, { "epoch": 104.48192771084338, "grad_norm": 7.217814922332764, "learning_rate": 4.769417475728156e-05, "loss": 0.1676, "step": 10775 }, { "epoch": 104.49156626506024, "grad_norm": 8.299202919006348, "learning_rate": 4.7689320388349515e-05, "loss": 0.2781, "step": 10776 }, { "epoch": 104.5012048192771, "grad_norm": 7.296413898468018, "learning_rate": 4.768446601941748e-05, "loss": 0.2535, "step": 10777 }, { "epoch": 104.51084337349397, "grad_norm": 15.80495548248291, "learning_rate": 4.767961165048544e-05, "loss": 0.477, "step": 10778 }, { "epoch": 104.52048192771085, "grad_norm": 4.760944843292236, "learning_rate": 4.76747572815534e-05, "loss": 0.1627, "step": 10779 }, { "epoch": 104.53012048192771, "grad_norm": 4.192602157592773, "learning_rate": 4.766990291262136e-05, "loss": 0.1458, "step": 10780 }, { "epoch": 104.53975903614457, "grad_norm": 5.006908416748047, "learning_rate": 4.7665048543689324e-05, "loss": 0.1423, "step": 10781 }, { "epoch": 104.54939759036145, "grad_norm": 10.40964412689209, "learning_rate": 4.766019417475728e-05, "loss": 0.3104, "step": 10782 }, { "epoch": 104.55903614457831, "grad_norm": 13.066314697265625, "learning_rate": 4.7655339805825246e-05, "loss": 0.1701, "step": 10783 }, { "epoch": 104.56867469879518, "grad_norm": 30.1865291595459, "learning_rate": 4.7650485436893204e-05, "loss": 0.3476, "step": 10784 }, { "epoch": 104.57831325301204, "grad_norm": 10.24094009399414, "learning_rate": 4.764563106796117e-05, "loss": 0.5784, "step": 10785 }, { "epoch": 104.58795180722892, "grad_norm": 9.56742000579834, "learning_rate": 4.7640776699029126e-05, "loss": 0.1682, "step": 10786 }, { "epoch": 104.59759036144578, "grad_norm": 5.713899612426758, "learning_rate": 4.763592233009709e-05, "loss": 0.1822, "step": 10787 }, { "epoch": 104.60722891566265, "grad_norm": 8.932848930358887, "learning_rate": 4.763106796116505e-05, "loss": 0.2785, "step": 10788 }, { "epoch": 104.61686746987952, "grad_norm": 7.142637252807617, "learning_rate": 4.762621359223301e-05, "loss": 0.2073, "step": 10789 }, { "epoch": 104.62650602409639, "grad_norm": 11.381203651428223, "learning_rate": 4.762135922330097e-05, "loss": 0.343, "step": 10790 }, { "epoch": 104.63614457831325, "grad_norm": 28.779544830322266, "learning_rate": 4.7616504854368935e-05, "loss": 0.3591, "step": 10791 }, { "epoch": 104.64578313253013, "grad_norm": 4.138125419616699, "learning_rate": 4.761165048543689e-05, "loss": 0.1715, "step": 10792 }, { "epoch": 104.65542168674699, "grad_norm": 8.26217269897461, "learning_rate": 4.760679611650486e-05, "loss": 0.3413, "step": 10793 }, { "epoch": 104.66506024096385, "grad_norm": 10.31789493560791, "learning_rate": 4.760194174757282e-05, "loss": 0.1853, "step": 10794 }, { "epoch": 104.67469879518072, "grad_norm": 7.874306678771973, "learning_rate": 4.759708737864078e-05, "loss": 0.3258, "step": 10795 }, { "epoch": 104.6843373493976, "grad_norm": 8.599212646484375, "learning_rate": 4.7592233009708744e-05, "loss": 0.3738, "step": 10796 }, { "epoch": 104.69397590361446, "grad_norm": 6.206963539123535, "learning_rate": 4.75873786407767e-05, "loss": 0.3544, "step": 10797 }, { "epoch": 104.70361445783132, "grad_norm": 10.761083602905273, "learning_rate": 4.7582524271844667e-05, "loss": 0.2554, "step": 10798 }, { "epoch": 104.7132530120482, "grad_norm": 7.596444606781006, "learning_rate": 4.7577669902912624e-05, "loss": 0.1871, "step": 10799 }, { "epoch": 104.72289156626506, "grad_norm": 7.613351821899414, "learning_rate": 4.757281553398059e-05, "loss": 0.2236, "step": 10800 }, { "epoch": 104.73253012048193, "grad_norm": 9.733405113220215, "learning_rate": 4.7567961165048547e-05, "loss": 0.4433, "step": 10801 }, { "epoch": 104.74216867469879, "grad_norm": 17.117610931396484, "learning_rate": 4.756310679611651e-05, "loss": 0.2524, "step": 10802 }, { "epoch": 104.75180722891567, "grad_norm": 10.964871406555176, "learning_rate": 4.755825242718447e-05, "loss": 0.3777, "step": 10803 }, { "epoch": 104.76144578313253, "grad_norm": 10.241493225097656, "learning_rate": 4.755339805825243e-05, "loss": 0.3634, "step": 10804 }, { "epoch": 104.7710843373494, "grad_norm": 16.21858024597168, "learning_rate": 4.754854368932039e-05, "loss": 0.4052, "step": 10805 }, { "epoch": 104.78072289156627, "grad_norm": 14.178343772888184, "learning_rate": 4.754368932038835e-05, "loss": 0.4804, "step": 10806 }, { "epoch": 104.79036144578313, "grad_norm": 8.526557922363281, "learning_rate": 4.7538834951456313e-05, "loss": 0.2965, "step": 10807 }, { "epoch": 104.8, "grad_norm": 4.112565994262695, "learning_rate": 4.753398058252427e-05, "loss": 0.1618, "step": 10808 }, { "epoch": 104.80963855421686, "grad_norm": 10.517923355102539, "learning_rate": 4.7529126213592236e-05, "loss": 0.3598, "step": 10809 }, { "epoch": 104.81927710843374, "grad_norm": 4.745139122009277, "learning_rate": 4.7524271844660193e-05, "loss": 0.2524, "step": 10810 }, { "epoch": 104.8289156626506, "grad_norm": 3.9883971214294434, "learning_rate": 4.751941747572816e-05, "loss": 0.1316, "step": 10811 }, { "epoch": 104.83855421686746, "grad_norm": 7.995724201202393, "learning_rate": 4.7514563106796116e-05, "loss": 0.2182, "step": 10812 }, { "epoch": 104.84819277108434, "grad_norm": 12.038457870483398, "learning_rate": 4.750970873786408e-05, "loss": 0.5661, "step": 10813 }, { "epoch": 104.8578313253012, "grad_norm": 10.165276527404785, "learning_rate": 4.750485436893204e-05, "loss": 0.3074, "step": 10814 }, { "epoch": 104.86746987951807, "grad_norm": 23.027904510498047, "learning_rate": 4.75e-05, "loss": 0.1728, "step": 10815 }, { "epoch": 104.87710843373495, "grad_norm": 5.180055141448975, "learning_rate": 4.749514563106796e-05, "loss": 0.1002, "step": 10816 }, { "epoch": 104.88674698795181, "grad_norm": 6.875192165374756, "learning_rate": 4.7490291262135925e-05, "loss": 0.331, "step": 10817 }, { "epoch": 104.89638554216867, "grad_norm": 7.958973407745361, "learning_rate": 4.748543689320389e-05, "loss": 0.1997, "step": 10818 }, { "epoch": 104.90602409638554, "grad_norm": 18.057313919067383, "learning_rate": 4.748058252427185e-05, "loss": 0.3146, "step": 10819 }, { "epoch": 104.91566265060241, "grad_norm": 5.096827507019043, "learning_rate": 4.747572815533981e-05, "loss": 0.1881, "step": 10820 }, { "epoch": 104.92530120481928, "grad_norm": 14.740927696228027, "learning_rate": 4.747087378640777e-05, "loss": 0.1572, "step": 10821 }, { "epoch": 104.93493975903614, "grad_norm": 12.11034870147705, "learning_rate": 4.7466019417475734e-05, "loss": 0.3317, "step": 10822 }, { "epoch": 104.94457831325302, "grad_norm": 7.971421241760254, "learning_rate": 4.746116504854369e-05, "loss": 0.1632, "step": 10823 }, { "epoch": 104.95421686746988, "grad_norm": 8.725985527038574, "learning_rate": 4.7456310679611656e-05, "loss": 0.3365, "step": 10824 }, { "epoch": 104.96385542168674, "grad_norm": 6.32071590423584, "learning_rate": 4.7451456310679614e-05, "loss": 0.2244, "step": 10825 }, { "epoch": 104.97349397590361, "grad_norm": 19.158905029296875, "learning_rate": 4.744660194174758e-05, "loss": 0.3077, "step": 10826 }, { "epoch": 104.98313253012049, "grad_norm": 7.525883674621582, "learning_rate": 4.7441747572815536e-05, "loss": 0.293, "step": 10827 }, { "epoch": 104.99277108433735, "grad_norm": 7.233387470245361, "learning_rate": 4.74368932038835e-05, "loss": 0.3053, "step": 10828 }, { "epoch": 105.00843373493976, "grad_norm": 3.0401010513305664, "learning_rate": 4.743203883495146e-05, "loss": 0.4591, "step": 10829 }, { "epoch": 105.01807228915662, "grad_norm": 18.749420166015625, "learning_rate": 4.7427184466019416e-05, "loss": 0.3544, "step": 10830 }, { "epoch": 105.0277108433735, "grad_norm": 8.019235610961914, "learning_rate": 4.742233009708738e-05, "loss": 0.3223, "step": 10831 }, { "epoch": 105.03734939759036, "grad_norm": 18.394865036010742, "learning_rate": 4.741747572815534e-05, "loss": 0.2757, "step": 10832 }, { "epoch": 105.04698795180722, "grad_norm": 24.128219604492188, "learning_rate": 4.74126213592233e-05, "loss": 0.2225, "step": 10833 }, { "epoch": 105.0566265060241, "grad_norm": 16.63559341430664, "learning_rate": 4.740776699029126e-05, "loss": 0.3724, "step": 10834 }, { "epoch": 105.06626506024097, "grad_norm": 3.6555216312408447, "learning_rate": 4.7402912621359226e-05, "loss": 0.1553, "step": 10835 }, { "epoch": 105.07590361445783, "grad_norm": 10.283134460449219, "learning_rate": 4.739805825242718e-05, "loss": 0.2882, "step": 10836 }, { "epoch": 105.08554216867469, "grad_norm": 2.3467745780944824, "learning_rate": 4.739320388349515e-05, "loss": 0.1823, "step": 10837 }, { "epoch": 105.09518072289157, "grad_norm": 24.64100456237793, "learning_rate": 4.7388349514563106e-05, "loss": 0.2911, "step": 10838 }, { "epoch": 105.10481927710843, "grad_norm": 3.6910183429718018, "learning_rate": 4.738349514563107e-05, "loss": 0.0856, "step": 10839 }, { "epoch": 105.1144578313253, "grad_norm": 1.1413840055465698, "learning_rate": 4.737864077669903e-05, "loss": 0.4043, "step": 10840 }, { "epoch": 105.12409638554217, "grad_norm": 51.13977813720703, "learning_rate": 4.737378640776699e-05, "loss": 0.2354, "step": 10841 }, { "epoch": 105.13373493975904, "grad_norm": 77.59046173095703, "learning_rate": 4.736893203883495e-05, "loss": 0.1507, "step": 10842 }, { "epoch": 105.1433734939759, "grad_norm": 18.130128860473633, "learning_rate": 4.7364077669902915e-05, "loss": 0.2621, "step": 10843 }, { "epoch": 105.15301204819278, "grad_norm": 4.048333168029785, "learning_rate": 4.735922330097088e-05, "loss": 0.2681, "step": 10844 }, { "epoch": 105.16265060240964, "grad_norm": 26.758831024169922, "learning_rate": 4.735436893203884e-05, "loss": 0.2669, "step": 10845 }, { "epoch": 105.1722891566265, "grad_norm": 26.438894271850586, "learning_rate": 4.73495145631068e-05, "loss": 0.4282, "step": 10846 }, { "epoch": 105.18192771084337, "grad_norm": 5.629518032073975, "learning_rate": 4.734466019417476e-05, "loss": 0.1743, "step": 10847 }, { "epoch": 105.19156626506025, "grad_norm": 8.108202934265137, "learning_rate": 4.7339805825242724e-05, "loss": 0.1379, "step": 10848 }, { "epoch": 105.20120481927711, "grad_norm": 6.324591636657715, "learning_rate": 4.733495145631068e-05, "loss": 0.2322, "step": 10849 }, { "epoch": 105.21084337349397, "grad_norm": 6.598320484161377, "learning_rate": 4.7330097087378646e-05, "loss": 0.3915, "step": 10850 }, { "epoch": 105.22048192771085, "grad_norm": 12.722588539123535, "learning_rate": 4.7325242718446604e-05, "loss": 0.2343, "step": 10851 }, { "epoch": 105.23012048192771, "grad_norm": 2.3372857570648193, "learning_rate": 4.732038834951457e-05, "loss": 0.2221, "step": 10852 }, { "epoch": 105.23975903614458, "grad_norm": 1.8724594116210938, "learning_rate": 4.7315533980582526e-05, "loss": 0.0965, "step": 10853 }, { "epoch": 105.24939759036144, "grad_norm": 28.246015548706055, "learning_rate": 4.731067961165049e-05, "loss": 0.3076, "step": 10854 }, { "epoch": 105.25903614457832, "grad_norm": 12.436749458312988, "learning_rate": 4.730582524271845e-05, "loss": 0.2862, "step": 10855 }, { "epoch": 105.26867469879518, "grad_norm": 4.876667022705078, "learning_rate": 4.730097087378641e-05, "loss": 0.1051, "step": 10856 }, { "epoch": 105.27831325301204, "grad_norm": 15.588998794555664, "learning_rate": 4.729611650485437e-05, "loss": 0.3857, "step": 10857 }, { "epoch": 105.28795180722892, "grad_norm": 3.5897560119628906, "learning_rate": 4.729126213592233e-05, "loss": 0.14, "step": 10858 }, { "epoch": 105.29759036144578, "grad_norm": 2.681807279586792, "learning_rate": 4.728640776699029e-05, "loss": 0.213, "step": 10859 }, { "epoch": 105.30722891566265, "grad_norm": 10.569912910461426, "learning_rate": 4.728155339805825e-05, "loss": 0.3694, "step": 10860 }, { "epoch": 105.31686746987951, "grad_norm": 32.67365646362305, "learning_rate": 4.7276699029126215e-05, "loss": 0.2589, "step": 10861 }, { "epoch": 105.32650602409639, "grad_norm": 56.49730682373047, "learning_rate": 4.727184466019417e-05, "loss": 0.2822, "step": 10862 }, { "epoch": 105.33614457831325, "grad_norm": 16.777225494384766, "learning_rate": 4.726699029126214e-05, "loss": 0.2448, "step": 10863 }, { "epoch": 105.34578313253012, "grad_norm": 2.521500587463379, "learning_rate": 4.7262135922330095e-05, "loss": 0.2196, "step": 10864 }, { "epoch": 105.355421686747, "grad_norm": 18.131166458129883, "learning_rate": 4.725728155339806e-05, "loss": 0.2569, "step": 10865 }, { "epoch": 105.36506024096386, "grad_norm": 62.17967987060547, "learning_rate": 4.725242718446602e-05, "loss": 0.3606, "step": 10866 }, { "epoch": 105.37469879518072, "grad_norm": 24.425811767578125, "learning_rate": 4.724757281553398e-05, "loss": 0.1997, "step": 10867 }, { "epoch": 105.38433734939758, "grad_norm": 11.45864486694336, "learning_rate": 4.724271844660194e-05, "loss": 0.4698, "step": 10868 }, { "epoch": 105.39397590361446, "grad_norm": 12.84200382232666, "learning_rate": 4.7237864077669904e-05, "loss": 0.1556, "step": 10869 }, { "epoch": 105.40361445783132, "grad_norm": 2.0630221366882324, "learning_rate": 4.723300970873787e-05, "loss": 0.231, "step": 10870 }, { "epoch": 105.41325301204819, "grad_norm": 21.440237045288086, "learning_rate": 4.722815533980583e-05, "loss": 0.3658, "step": 10871 }, { "epoch": 105.42289156626506, "grad_norm": 3.1054794788360596, "learning_rate": 4.722330097087379e-05, "loss": 0.2664, "step": 10872 }, { "epoch": 105.43253012048193, "grad_norm": 15.525129318237305, "learning_rate": 4.721844660194175e-05, "loss": 0.2359, "step": 10873 }, { "epoch": 105.44216867469879, "grad_norm": 4.940255165100098, "learning_rate": 4.7213592233009714e-05, "loss": 0.2193, "step": 10874 }, { "epoch": 105.45180722891567, "grad_norm": 1.4059748649597168, "learning_rate": 4.720873786407767e-05, "loss": 0.1916, "step": 10875 }, { "epoch": 105.46144578313253, "grad_norm": 35.65149688720703, "learning_rate": 4.7203883495145636e-05, "loss": 0.1818, "step": 10876 }, { "epoch": 105.4710843373494, "grad_norm": 3.350175619125366, "learning_rate": 4.7199029126213594e-05, "loss": 0.1454, "step": 10877 }, { "epoch": 105.48072289156626, "grad_norm": 17.507030487060547, "learning_rate": 4.719417475728156e-05, "loss": 0.3201, "step": 10878 }, { "epoch": 105.49036144578314, "grad_norm": 27.41139793395996, "learning_rate": 4.7189320388349516e-05, "loss": 0.2162, "step": 10879 }, { "epoch": 105.5, "grad_norm": 5.042195796966553, "learning_rate": 4.718446601941748e-05, "loss": 0.1781, "step": 10880 }, { "epoch": 105.50963855421686, "grad_norm": 12.36188793182373, "learning_rate": 4.717961165048544e-05, "loss": 0.3303, "step": 10881 }, { "epoch": 105.51927710843374, "grad_norm": 93.2942123413086, "learning_rate": 4.71747572815534e-05, "loss": 0.3054, "step": 10882 }, { "epoch": 105.5289156626506, "grad_norm": 7.38175106048584, "learning_rate": 4.716990291262136e-05, "loss": 0.1517, "step": 10883 }, { "epoch": 105.53855421686747, "grad_norm": 8.063530921936035, "learning_rate": 4.7165048543689325e-05, "loss": 0.3593, "step": 10884 }, { "epoch": 105.54819277108433, "grad_norm": 11.354021072387695, "learning_rate": 4.716019417475728e-05, "loss": 0.2407, "step": 10885 }, { "epoch": 105.55783132530121, "grad_norm": 16.956798553466797, "learning_rate": 4.715533980582524e-05, "loss": 0.3375, "step": 10886 }, { "epoch": 105.56746987951807, "grad_norm": 33.31001663208008, "learning_rate": 4.7150485436893205e-05, "loss": 0.2534, "step": 10887 }, { "epoch": 105.57710843373494, "grad_norm": 19.44814109802246, "learning_rate": 4.714563106796116e-05, "loss": 0.5114, "step": 10888 }, { "epoch": 105.58674698795181, "grad_norm": 21.17707633972168, "learning_rate": 4.714077669902913e-05, "loss": 0.4357, "step": 10889 }, { "epoch": 105.59638554216868, "grad_norm": 55.4346809387207, "learning_rate": 4.7135922330097085e-05, "loss": 0.3137, "step": 10890 }, { "epoch": 105.60602409638554, "grad_norm": 8.061688423156738, "learning_rate": 4.713106796116505e-05, "loss": 0.3963, "step": 10891 }, { "epoch": 105.61566265060242, "grad_norm": 16.639354705810547, "learning_rate": 4.712621359223301e-05, "loss": 0.4055, "step": 10892 }, { "epoch": 105.62530120481928, "grad_norm": 7.67349100112915, "learning_rate": 4.712135922330097e-05, "loss": 0.1967, "step": 10893 }, { "epoch": 105.63493975903614, "grad_norm": 24.003704071044922, "learning_rate": 4.7116504854368937e-05, "loss": 0.5524, "step": 10894 }, { "epoch": 105.644578313253, "grad_norm": 14.29478645324707, "learning_rate": 4.71116504854369e-05, "loss": 0.4418, "step": 10895 }, { "epoch": 105.65421686746988, "grad_norm": 3.1032936573028564, "learning_rate": 4.710679611650486e-05, "loss": 0.3448, "step": 10896 }, { "epoch": 105.66385542168675, "grad_norm": 38.42007827758789, "learning_rate": 4.7101941747572817e-05, "loss": 0.3033, "step": 10897 }, { "epoch": 105.67349397590361, "grad_norm": 13.009795188903809, "learning_rate": 4.709708737864078e-05, "loss": 0.1876, "step": 10898 }, { "epoch": 105.68313253012049, "grad_norm": 1.0239044427871704, "learning_rate": 4.709223300970874e-05, "loss": 0.2118, "step": 10899 }, { "epoch": 105.69277108433735, "grad_norm": 7.384618759155273, "learning_rate": 4.7087378640776703e-05, "loss": 0.1315, "step": 10900 }, { "epoch": 105.70240963855422, "grad_norm": 44.963714599609375, "learning_rate": 4.708252427184466e-05, "loss": 0.302, "step": 10901 }, { "epoch": 105.71204819277108, "grad_norm": 3.7655789852142334, "learning_rate": 4.7077669902912626e-05, "loss": 0.2204, "step": 10902 }, { "epoch": 105.72168674698796, "grad_norm": 52.834808349609375, "learning_rate": 4.7072815533980583e-05, "loss": 0.4114, "step": 10903 }, { "epoch": 105.73132530120482, "grad_norm": 15.915654182434082, "learning_rate": 4.706796116504855e-05, "loss": 0.3304, "step": 10904 }, { "epoch": 105.74096385542168, "grad_norm": 52.6992301940918, "learning_rate": 4.7063106796116506e-05, "loss": 0.2721, "step": 10905 }, { "epoch": 105.75060240963856, "grad_norm": 8.036214828491211, "learning_rate": 4.705825242718447e-05, "loss": 0.2972, "step": 10906 }, { "epoch": 105.76024096385542, "grad_norm": 14.203340530395508, "learning_rate": 4.705339805825243e-05, "loss": 0.1884, "step": 10907 }, { "epoch": 105.76987951807229, "grad_norm": 42.410797119140625, "learning_rate": 4.704854368932039e-05, "loss": 0.5864, "step": 10908 }, { "epoch": 105.77951807228915, "grad_norm": 7.015756130218506, "learning_rate": 4.704368932038835e-05, "loss": 0.2891, "step": 10909 }, { "epoch": 105.78915662650603, "grad_norm": 23.580541610717773, "learning_rate": 4.7038834951456315e-05, "loss": 0.2734, "step": 10910 }, { "epoch": 105.79879518072289, "grad_norm": 22.760833740234375, "learning_rate": 4.703398058252427e-05, "loss": 0.122, "step": 10911 }, { "epoch": 105.80843373493975, "grad_norm": 18.634153366088867, "learning_rate": 4.702912621359224e-05, "loss": 0.3623, "step": 10912 }, { "epoch": 105.81807228915663, "grad_norm": 18.884855270385742, "learning_rate": 4.7024271844660195e-05, "loss": 0.2386, "step": 10913 }, { "epoch": 105.8277108433735, "grad_norm": 17.52170753479004, "learning_rate": 4.701941747572815e-05, "loss": 0.3815, "step": 10914 }, { "epoch": 105.83734939759036, "grad_norm": 15.705698013305664, "learning_rate": 4.701456310679612e-05, "loss": 0.2195, "step": 10915 }, { "epoch": 105.84698795180722, "grad_norm": 26.393922805786133, "learning_rate": 4.7009708737864075e-05, "loss": 0.4289, "step": 10916 }, { "epoch": 105.8566265060241, "grad_norm": 18.322555541992188, "learning_rate": 4.700485436893204e-05, "loss": 0.1694, "step": 10917 }, { "epoch": 105.86626506024096, "grad_norm": 18.41857147216797, "learning_rate": 4.7e-05, "loss": 0.2433, "step": 10918 }, { "epoch": 105.87590361445783, "grad_norm": 11.514947891235352, "learning_rate": 4.699514563106797e-05, "loss": 0.2298, "step": 10919 }, { "epoch": 105.8855421686747, "grad_norm": 5.501639366149902, "learning_rate": 4.6990291262135926e-05, "loss": 0.1714, "step": 10920 }, { "epoch": 105.89518072289157, "grad_norm": 20.583024978637695, "learning_rate": 4.698543689320389e-05, "loss": 0.2476, "step": 10921 }, { "epoch": 105.90481927710843, "grad_norm": 2.001477003097534, "learning_rate": 4.698058252427185e-05, "loss": 0.19, "step": 10922 }, { "epoch": 105.91445783132531, "grad_norm": 84.6719970703125, "learning_rate": 4.697572815533981e-05, "loss": 0.4612, "step": 10923 }, { "epoch": 105.92409638554217, "grad_norm": 23.053386688232422, "learning_rate": 4.697087378640777e-05, "loss": 0.3473, "step": 10924 }, { "epoch": 105.93373493975903, "grad_norm": 4.076059341430664, "learning_rate": 4.696601941747573e-05, "loss": 0.1865, "step": 10925 }, { "epoch": 105.9433734939759, "grad_norm": 5.347890853881836, "learning_rate": 4.696116504854369e-05, "loss": 0.3978, "step": 10926 }, { "epoch": 105.95301204819278, "grad_norm": 14.260759353637695, "learning_rate": 4.695631067961165e-05, "loss": 0.1714, "step": 10927 }, { "epoch": 105.96265060240964, "grad_norm": 13.732428550720215, "learning_rate": 4.6951456310679615e-05, "loss": 0.1582, "step": 10928 }, { "epoch": 105.9722891566265, "grad_norm": 24.95343017578125, "learning_rate": 4.694660194174757e-05, "loss": 0.2673, "step": 10929 }, { "epoch": 105.98192771084338, "grad_norm": 12.615499496459961, "learning_rate": 4.694174757281554e-05, "loss": 0.2213, "step": 10930 }, { "epoch": 105.99156626506024, "grad_norm": 21.26262092590332, "learning_rate": 4.6936893203883496e-05, "loss": 0.3155, "step": 10931 }, { "epoch": 106.00722891566265, "grad_norm": 40.3216667175293, "learning_rate": 4.693203883495146e-05, "loss": 0.1167, "step": 10932 }, { "epoch": 106.01686746987951, "grad_norm": 10.365814208984375, "learning_rate": 4.692718446601942e-05, "loss": 0.2892, "step": 10933 }, { "epoch": 106.02650602409639, "grad_norm": 35.466590881347656, "learning_rate": 4.692233009708738e-05, "loss": 0.2041, "step": 10934 }, { "epoch": 106.03614457831326, "grad_norm": 24.90682029724121, "learning_rate": 4.691747572815534e-05, "loss": 0.3669, "step": 10935 }, { "epoch": 106.04578313253012, "grad_norm": 24.02046775817871, "learning_rate": 4.6912621359223305e-05, "loss": 0.2811, "step": 10936 }, { "epoch": 106.05542168674698, "grad_norm": 7.98244047164917, "learning_rate": 4.690776699029126e-05, "loss": 0.1016, "step": 10937 }, { "epoch": 106.06506024096386, "grad_norm": 106.9837417602539, "learning_rate": 4.690291262135923e-05, "loss": 0.3489, "step": 10938 }, { "epoch": 106.07469879518072, "grad_norm": 17.75544548034668, "learning_rate": 4.6898058252427185e-05, "loss": 0.3423, "step": 10939 }, { "epoch": 106.08433734939759, "grad_norm": 56.62028503417969, "learning_rate": 4.689320388349515e-05, "loss": 0.304, "step": 10940 }, { "epoch": 106.09397590361446, "grad_norm": 7.020153045654297, "learning_rate": 4.688834951456311e-05, "loss": 0.1487, "step": 10941 }, { "epoch": 106.10361445783133, "grad_norm": 17.06193733215332, "learning_rate": 4.6883495145631065e-05, "loss": 0.236, "step": 10942 }, { "epoch": 106.11325301204819, "grad_norm": 27.539554595947266, "learning_rate": 4.687864077669903e-05, "loss": 0.4021, "step": 10943 }, { "epoch": 106.12289156626505, "grad_norm": 7.175328254699707, "learning_rate": 4.6873786407766994e-05, "loss": 0.1299, "step": 10944 }, { "epoch": 106.13253012048193, "grad_norm": 13.363492012023926, "learning_rate": 4.686893203883496e-05, "loss": 0.291, "step": 10945 }, { "epoch": 106.1421686746988, "grad_norm": 36.75165557861328, "learning_rate": 4.6864077669902916e-05, "loss": 0.3207, "step": 10946 }, { "epoch": 106.15180722891566, "grad_norm": 17.11219596862793, "learning_rate": 4.685922330097088e-05, "loss": 0.2132, "step": 10947 }, { "epoch": 106.16144578313254, "grad_norm": 14.504891395568848, "learning_rate": 4.685436893203884e-05, "loss": 0.2853, "step": 10948 }, { "epoch": 106.1710843373494, "grad_norm": 32.14700698852539, "learning_rate": 4.68495145631068e-05, "loss": 0.3964, "step": 10949 }, { "epoch": 106.18072289156626, "grad_norm": 18.22654914855957, "learning_rate": 4.684466019417476e-05, "loss": 0.326, "step": 10950 }, { "epoch": 106.19036144578314, "grad_norm": 8.619966506958008, "learning_rate": 4.6839805825242725e-05, "loss": 0.3298, "step": 10951 }, { "epoch": 106.2, "grad_norm": 25.25734519958496, "learning_rate": 4.683495145631068e-05, "loss": 0.2727, "step": 10952 }, { "epoch": 106.20963855421687, "grad_norm": 35.05587387084961, "learning_rate": 4.683009708737864e-05, "loss": 0.3262, "step": 10953 }, { "epoch": 106.21927710843373, "grad_norm": 17.464759826660156, "learning_rate": 4.6825242718446605e-05, "loss": 0.2423, "step": 10954 }, { "epoch": 106.2289156626506, "grad_norm": 16.96718978881836, "learning_rate": 4.682038834951456e-05, "loss": 0.2513, "step": 10955 }, { "epoch": 106.23855421686747, "grad_norm": 6.227656841278076, "learning_rate": 4.681553398058253e-05, "loss": 0.1894, "step": 10956 }, { "epoch": 106.24819277108433, "grad_norm": 17.39470863342285, "learning_rate": 4.6810679611650485e-05, "loss": 0.4085, "step": 10957 }, { "epoch": 106.25783132530121, "grad_norm": 19.725549697875977, "learning_rate": 4.680582524271845e-05, "loss": 0.306, "step": 10958 }, { "epoch": 106.26746987951807, "grad_norm": 20.487497329711914, "learning_rate": 4.680097087378641e-05, "loss": 0.3616, "step": 10959 }, { "epoch": 106.27710843373494, "grad_norm": 139.16944885253906, "learning_rate": 4.679611650485437e-05, "loss": 0.2311, "step": 10960 }, { "epoch": 106.2867469879518, "grad_norm": 9.389412879943848, "learning_rate": 4.679126213592233e-05, "loss": 0.4275, "step": 10961 }, { "epoch": 106.29638554216868, "grad_norm": 10.35302448272705, "learning_rate": 4.6786407766990294e-05, "loss": 0.1897, "step": 10962 }, { "epoch": 106.30602409638554, "grad_norm": 32.333045959472656, "learning_rate": 4.678155339805825e-05, "loss": 0.3163, "step": 10963 }, { "epoch": 106.3156626506024, "grad_norm": 61.042476654052734, "learning_rate": 4.677669902912622e-05, "loss": 0.426, "step": 10964 }, { "epoch": 106.32530120481928, "grad_norm": 18.850521087646484, "learning_rate": 4.6771844660194174e-05, "loss": 0.3737, "step": 10965 }, { "epoch": 106.33493975903615, "grad_norm": 340.61285400390625, "learning_rate": 4.676699029126214e-05, "loss": 0.328, "step": 10966 }, { "epoch": 106.34457831325301, "grad_norm": 8.139431953430176, "learning_rate": 4.67621359223301e-05, "loss": 0.3138, "step": 10967 }, { "epoch": 106.35421686746987, "grad_norm": 90.995361328125, "learning_rate": 4.675728155339806e-05, "loss": 0.2579, "step": 10968 }, { "epoch": 106.36385542168675, "grad_norm": 57.303977966308594, "learning_rate": 4.6752427184466026e-05, "loss": 0.5529, "step": 10969 }, { "epoch": 106.37349397590361, "grad_norm": 7.472755432128906, "learning_rate": 4.6747572815533984e-05, "loss": 0.2097, "step": 10970 }, { "epoch": 106.38313253012048, "grad_norm": 25.899169921875, "learning_rate": 4.674271844660195e-05, "loss": 0.3185, "step": 10971 }, { "epoch": 106.39277108433735, "grad_norm": 4.247406959533691, "learning_rate": 4.6737864077669906e-05, "loss": 0.2008, "step": 10972 }, { "epoch": 106.40240963855422, "grad_norm": 3.5350594520568848, "learning_rate": 4.673300970873787e-05, "loss": 0.2442, "step": 10973 }, { "epoch": 106.41204819277108, "grad_norm": 41.79710006713867, "learning_rate": 4.672815533980583e-05, "loss": 0.2715, "step": 10974 }, { "epoch": 106.42168674698796, "grad_norm": 22.169483184814453, "learning_rate": 4.672330097087379e-05, "loss": 0.1799, "step": 10975 }, { "epoch": 106.43132530120482, "grad_norm": 17.172250747680664, "learning_rate": 4.671844660194175e-05, "loss": 0.1162, "step": 10976 }, { "epoch": 106.44096385542169, "grad_norm": 21.377094268798828, "learning_rate": 4.6713592233009715e-05, "loss": 0.3224, "step": 10977 }, { "epoch": 106.45060240963855, "grad_norm": 22.09234046936035, "learning_rate": 4.670873786407767e-05, "loss": 0.1428, "step": 10978 }, { "epoch": 106.46024096385543, "grad_norm": 17.612367630004883, "learning_rate": 4.670388349514563e-05, "loss": 0.3543, "step": 10979 }, { "epoch": 106.46987951807229, "grad_norm": 15.36900806427002, "learning_rate": 4.6699029126213595e-05, "loss": 0.2414, "step": 10980 }, { "epoch": 106.47951807228915, "grad_norm": 83.56324768066406, "learning_rate": 4.669417475728155e-05, "loss": 0.2436, "step": 10981 }, { "epoch": 106.48915662650603, "grad_norm": 27.423837661743164, "learning_rate": 4.668932038834952e-05, "loss": 0.3603, "step": 10982 }, { "epoch": 106.4987951807229, "grad_norm": 21.821327209472656, "learning_rate": 4.6684466019417475e-05, "loss": 0.5387, "step": 10983 }, { "epoch": 106.50843373493976, "grad_norm": 4.769631862640381, "learning_rate": 4.667961165048544e-05, "loss": 0.4644, "step": 10984 }, { "epoch": 106.51807228915662, "grad_norm": 276.4830322265625, "learning_rate": 4.66747572815534e-05, "loss": 0.3599, "step": 10985 }, { "epoch": 106.5277108433735, "grad_norm": 18.36460304260254, "learning_rate": 4.666990291262136e-05, "loss": 0.2525, "step": 10986 }, { "epoch": 106.53734939759036, "grad_norm": 21.14687728881836, "learning_rate": 4.666504854368932e-05, "loss": 0.1688, "step": 10987 }, { "epoch": 106.54698795180722, "grad_norm": 6.206833362579346, "learning_rate": 4.6660194174757284e-05, "loss": 0.2033, "step": 10988 }, { "epoch": 106.5566265060241, "grad_norm": 26.537952423095703, "learning_rate": 4.665533980582524e-05, "loss": 0.3039, "step": 10989 }, { "epoch": 106.56626506024097, "grad_norm": 21.823673248291016, "learning_rate": 4.6650485436893207e-05, "loss": 0.19, "step": 10990 }, { "epoch": 106.57590361445783, "grad_norm": 44.38802719116211, "learning_rate": 4.6645631067961164e-05, "loss": 0.3072, "step": 10991 }, { "epoch": 106.58554216867469, "grad_norm": 3.927640438079834, "learning_rate": 4.664077669902913e-05, "loss": 0.2997, "step": 10992 }, { "epoch": 106.59518072289157, "grad_norm": 11.500994682312012, "learning_rate": 4.6635922330097087e-05, "loss": 0.1007, "step": 10993 }, { "epoch": 106.60481927710843, "grad_norm": 7.234886646270752, "learning_rate": 4.663106796116505e-05, "loss": 0.3192, "step": 10994 }, { "epoch": 106.6144578313253, "grad_norm": 16.7291316986084, "learning_rate": 4.6626213592233016e-05, "loss": 0.3708, "step": 10995 }, { "epoch": 106.62409638554217, "grad_norm": 46.551448822021484, "learning_rate": 4.662135922330097e-05, "loss": 0.545, "step": 10996 }, { "epoch": 106.63373493975904, "grad_norm": 5.537012577056885, "learning_rate": 4.661650485436894e-05, "loss": 0.1, "step": 10997 }, { "epoch": 106.6433734939759, "grad_norm": 13.774401664733887, "learning_rate": 4.6611650485436896e-05, "loss": 0.2131, "step": 10998 }, { "epoch": 106.65301204819278, "grad_norm": 22.870166778564453, "learning_rate": 4.660679611650486e-05, "loss": 0.1942, "step": 10999 }, { "epoch": 106.66265060240964, "grad_norm": 23.394147872924805, "learning_rate": 4.660194174757282e-05, "loss": 0.3018, "step": 11000 }, { "epoch": 106.6722891566265, "grad_norm": 13.247551918029785, "learning_rate": 4.659708737864078e-05, "loss": 0.4402, "step": 11001 }, { "epoch": 106.68192771084337, "grad_norm": 42.332725524902344, "learning_rate": 4.659223300970874e-05, "loss": 0.2506, "step": 11002 }, { "epoch": 106.69156626506025, "grad_norm": 4.97349214553833, "learning_rate": 4.6587378640776705e-05, "loss": 0.3409, "step": 11003 }, { "epoch": 106.70120481927711, "grad_norm": 38.249629974365234, "learning_rate": 4.658252427184466e-05, "loss": 0.308, "step": 11004 }, { "epoch": 106.71084337349397, "grad_norm": 15.501051902770996, "learning_rate": 4.657766990291263e-05, "loss": 0.2079, "step": 11005 }, { "epoch": 106.72048192771085, "grad_norm": 2.974123001098633, "learning_rate": 4.6572815533980585e-05, "loss": 0.2387, "step": 11006 }, { "epoch": 106.73012048192771, "grad_norm": 14.458703994750977, "learning_rate": 4.656796116504854e-05, "loss": 0.2672, "step": 11007 }, { "epoch": 106.73975903614458, "grad_norm": 20.75120735168457, "learning_rate": 4.656310679611651e-05, "loss": 0.2841, "step": 11008 }, { "epoch": 106.74939759036144, "grad_norm": 3.6673662662506104, "learning_rate": 4.6558252427184465e-05, "loss": 0.1705, "step": 11009 }, { "epoch": 106.75903614457832, "grad_norm": 12.258028984069824, "learning_rate": 4.655339805825243e-05, "loss": 0.2109, "step": 11010 }, { "epoch": 106.76867469879518, "grad_norm": 20.494497299194336, "learning_rate": 4.654854368932039e-05, "loss": 0.183, "step": 11011 }, { "epoch": 106.77831325301204, "grad_norm": 41.95195770263672, "learning_rate": 4.654368932038835e-05, "loss": 0.3156, "step": 11012 }, { "epoch": 106.78795180722892, "grad_norm": 12.573421478271484, "learning_rate": 4.653883495145631e-05, "loss": 0.4269, "step": 11013 }, { "epoch": 106.79759036144578, "grad_norm": 9.217074394226074, "learning_rate": 4.6533980582524274e-05, "loss": 0.2383, "step": 11014 }, { "epoch": 106.80722891566265, "grad_norm": 4.181114673614502, "learning_rate": 4.652912621359223e-05, "loss": 0.2355, "step": 11015 }, { "epoch": 106.81686746987951, "grad_norm": 32.5872917175293, "learning_rate": 4.6524271844660196e-05, "loss": 0.4413, "step": 11016 }, { "epoch": 106.82650602409639, "grad_norm": 14.026568412780762, "learning_rate": 4.6519417475728154e-05, "loss": 0.3595, "step": 11017 }, { "epoch": 106.83614457831325, "grad_norm": 10.488494873046875, "learning_rate": 4.651456310679612e-05, "loss": 0.301, "step": 11018 }, { "epoch": 106.84578313253012, "grad_norm": 24.645980834960938, "learning_rate": 4.6509708737864076e-05, "loss": 0.2318, "step": 11019 }, { "epoch": 106.855421686747, "grad_norm": 4.783150672912598, "learning_rate": 4.650485436893204e-05, "loss": 0.1347, "step": 11020 }, { "epoch": 106.86506024096386, "grad_norm": 7.115112781524658, "learning_rate": 4.6500000000000005e-05, "loss": 0.2011, "step": 11021 }, { "epoch": 106.87469879518072, "grad_norm": 4.336332321166992, "learning_rate": 4.649514563106796e-05, "loss": 0.3155, "step": 11022 }, { "epoch": 106.88433734939758, "grad_norm": 36.43291091918945, "learning_rate": 4.649029126213593e-05, "loss": 0.1924, "step": 11023 }, { "epoch": 106.89397590361446, "grad_norm": 4.163760185241699, "learning_rate": 4.6485436893203885e-05, "loss": 0.2311, "step": 11024 }, { "epoch": 106.90361445783132, "grad_norm": 35.66936492919922, "learning_rate": 4.648058252427185e-05, "loss": 0.2369, "step": 11025 }, { "epoch": 106.91325301204819, "grad_norm": 16.740869522094727, "learning_rate": 4.647572815533981e-05, "loss": 0.2906, "step": 11026 }, { "epoch": 106.92289156626506, "grad_norm": 33.21320343017578, "learning_rate": 4.647087378640777e-05, "loss": 0.2689, "step": 11027 }, { "epoch": 106.93253012048193, "grad_norm": 17.640777587890625, "learning_rate": 4.646601941747573e-05, "loss": 0.4387, "step": 11028 }, { "epoch": 106.94216867469879, "grad_norm": 14.78770637512207, "learning_rate": 4.6461165048543695e-05, "loss": 0.2228, "step": 11029 }, { "epoch": 106.95180722891567, "grad_norm": 9.751997947692871, "learning_rate": 4.645631067961165e-05, "loss": 0.3187, "step": 11030 }, { "epoch": 106.96144578313253, "grad_norm": 7.700796127319336, "learning_rate": 4.645145631067962e-05, "loss": 0.2057, "step": 11031 }, { "epoch": 106.9710843373494, "grad_norm": 53.49541473388672, "learning_rate": 4.6446601941747575e-05, "loss": 0.3609, "step": 11032 }, { "epoch": 106.98072289156626, "grad_norm": 4.863448143005371, "learning_rate": 4.644174757281554e-05, "loss": 0.2213, "step": 11033 }, { "epoch": 106.99036144578314, "grad_norm": 12.484920501708984, "learning_rate": 4.64368932038835e-05, "loss": 0.1871, "step": 11034 }, { "epoch": 107.00602409638554, "grad_norm": 29.93056869506836, "learning_rate": 4.6432038834951455e-05, "loss": 0.4153, "step": 11035 }, { "epoch": 107.01566265060241, "grad_norm": 28.80675506591797, "learning_rate": 4.642718446601942e-05, "loss": 0.2492, "step": 11036 }, { "epoch": 107.02530120481927, "grad_norm": 18.099828720092773, "learning_rate": 4.642233009708738e-05, "loss": 0.177, "step": 11037 }, { "epoch": 107.03493975903615, "grad_norm": 1.7242918014526367, "learning_rate": 4.641747572815534e-05, "loss": 0.131, "step": 11038 }, { "epoch": 107.04457831325301, "grad_norm": 5.703797340393066, "learning_rate": 4.64126213592233e-05, "loss": 0.2554, "step": 11039 }, { "epoch": 107.05421686746988, "grad_norm": 15.219343185424805, "learning_rate": 4.6407766990291264e-05, "loss": 0.1879, "step": 11040 }, { "epoch": 107.06385542168675, "grad_norm": 31.97216796875, "learning_rate": 4.640291262135922e-05, "loss": 0.2565, "step": 11041 }, { "epoch": 107.07349397590362, "grad_norm": 14.019489288330078, "learning_rate": 4.6398058252427186e-05, "loss": 0.3104, "step": 11042 }, { "epoch": 107.08313253012048, "grad_norm": 38.54420471191406, "learning_rate": 4.6393203883495144e-05, "loss": 0.2158, "step": 11043 }, { "epoch": 107.09277108433734, "grad_norm": 34.08744812011719, "learning_rate": 4.638834951456311e-05, "loss": 0.189, "step": 11044 }, { "epoch": 107.10240963855422, "grad_norm": 15.3743896484375, "learning_rate": 4.638349514563107e-05, "loss": 0.3322, "step": 11045 }, { "epoch": 107.11204819277108, "grad_norm": 52.94757080078125, "learning_rate": 4.637864077669903e-05, "loss": 0.3119, "step": 11046 }, { "epoch": 107.12168674698795, "grad_norm": 111.05562591552734, "learning_rate": 4.6373786407766995e-05, "loss": 0.219, "step": 11047 }, { "epoch": 107.13132530120482, "grad_norm": 4.824398517608643, "learning_rate": 4.636893203883495e-05, "loss": 0.2573, "step": 11048 }, { "epoch": 107.14096385542169, "grad_norm": 28.326021194458008, "learning_rate": 4.636407766990292e-05, "loss": 0.2813, "step": 11049 }, { "epoch": 107.15060240963855, "grad_norm": 17.574522018432617, "learning_rate": 4.6359223300970875e-05, "loss": 0.2382, "step": 11050 }, { "epoch": 107.16024096385541, "grad_norm": 38.6379508972168, "learning_rate": 4.635436893203884e-05, "loss": 0.131, "step": 11051 }, { "epoch": 107.16987951807229, "grad_norm": 65.437744140625, "learning_rate": 4.63495145631068e-05, "loss": 0.3086, "step": 11052 }, { "epoch": 107.17951807228916, "grad_norm": 8.250740051269531, "learning_rate": 4.634466019417476e-05, "loss": 0.3635, "step": 11053 }, { "epoch": 107.18915662650602, "grad_norm": 47.31261444091797, "learning_rate": 4.633980582524272e-05, "loss": 0.525, "step": 11054 }, { "epoch": 107.1987951807229, "grad_norm": 13.855182647705078, "learning_rate": 4.6334951456310684e-05, "loss": 0.472, "step": 11055 }, { "epoch": 107.20843373493976, "grad_norm": 14.491270065307617, "learning_rate": 4.633009708737864e-05, "loss": 0.3388, "step": 11056 }, { "epoch": 107.21807228915662, "grad_norm": 3.3479576110839844, "learning_rate": 4.632524271844661e-05, "loss": 0.1587, "step": 11057 }, { "epoch": 107.2277108433735, "grad_norm": 4.347441673278809, "learning_rate": 4.6320388349514564e-05, "loss": 0.2162, "step": 11058 }, { "epoch": 107.23734939759036, "grad_norm": 20.533300399780273, "learning_rate": 4.631553398058253e-05, "loss": 0.4848, "step": 11059 }, { "epoch": 107.24698795180723, "grad_norm": 16.563730239868164, "learning_rate": 4.631067961165049e-05, "loss": 0.198, "step": 11060 }, { "epoch": 107.25662650602409, "grad_norm": 61.60908126831055, "learning_rate": 4.630582524271845e-05, "loss": 0.1033, "step": 11061 }, { "epoch": 107.26626506024097, "grad_norm": 8.853464126586914, "learning_rate": 4.630097087378641e-05, "loss": 0.2356, "step": 11062 }, { "epoch": 107.27590361445783, "grad_norm": 7.636504173278809, "learning_rate": 4.629611650485437e-05, "loss": 0.2313, "step": 11063 }, { "epoch": 107.2855421686747, "grad_norm": 17.375322341918945, "learning_rate": 4.629126213592233e-05, "loss": 0.2212, "step": 11064 }, { "epoch": 107.29518072289157, "grad_norm": 11.697760581970215, "learning_rate": 4.628640776699029e-05, "loss": 0.174, "step": 11065 }, { "epoch": 107.30481927710844, "grad_norm": 5.163486480712891, "learning_rate": 4.6281553398058254e-05, "loss": 0.2022, "step": 11066 }, { "epoch": 107.3144578313253, "grad_norm": 3.976670503616333, "learning_rate": 4.627669902912621e-05, "loss": 0.1245, "step": 11067 }, { "epoch": 107.32409638554216, "grad_norm": 7.910069465637207, "learning_rate": 4.6271844660194176e-05, "loss": 0.2996, "step": 11068 }, { "epoch": 107.33373493975904, "grad_norm": 2.595533847808838, "learning_rate": 4.6266990291262134e-05, "loss": 0.147, "step": 11069 }, { "epoch": 107.3433734939759, "grad_norm": 71.67758178710938, "learning_rate": 4.6262135922330105e-05, "loss": 0.576, "step": 11070 }, { "epoch": 107.35301204819277, "grad_norm": 7.883484363555908, "learning_rate": 4.625728155339806e-05, "loss": 0.156, "step": 11071 }, { "epoch": 107.36265060240964, "grad_norm": 5.771870136260986, "learning_rate": 4.625242718446603e-05, "loss": 0.1254, "step": 11072 }, { "epoch": 107.37228915662651, "grad_norm": 20.37782096862793, "learning_rate": 4.6247572815533985e-05, "loss": 0.209, "step": 11073 }, { "epoch": 107.38192771084337, "grad_norm": 18.743335723876953, "learning_rate": 4.624271844660194e-05, "loss": 0.2028, "step": 11074 }, { "epoch": 107.39156626506023, "grad_norm": 30.512088775634766, "learning_rate": 4.623786407766991e-05, "loss": 0.4001, "step": 11075 }, { "epoch": 107.40120481927711, "grad_norm": 6.373748302459717, "learning_rate": 4.6233009708737865e-05, "loss": 0.4882, "step": 11076 }, { "epoch": 107.41084337349398, "grad_norm": 17.21708869934082, "learning_rate": 4.622815533980583e-05, "loss": 0.3994, "step": 11077 }, { "epoch": 107.42048192771084, "grad_norm": 19.72372817993164, "learning_rate": 4.622330097087379e-05, "loss": 0.3442, "step": 11078 }, { "epoch": 107.43012048192772, "grad_norm": 9.747833251953125, "learning_rate": 4.621844660194175e-05, "loss": 0.307, "step": 11079 }, { "epoch": 107.43975903614458, "grad_norm": 41.38906478881836, "learning_rate": 4.621359223300971e-05, "loss": 0.2997, "step": 11080 }, { "epoch": 107.44939759036144, "grad_norm": 21.216995239257812, "learning_rate": 4.6208737864077674e-05, "loss": 0.4162, "step": 11081 }, { "epoch": 107.45903614457832, "grad_norm": 4.35911226272583, "learning_rate": 4.620388349514563e-05, "loss": 0.2418, "step": 11082 }, { "epoch": 107.46867469879518, "grad_norm": 8.238509178161621, "learning_rate": 4.6199029126213596e-05, "loss": 0.3891, "step": 11083 }, { "epoch": 107.47831325301205, "grad_norm": 8.484108924865723, "learning_rate": 4.6194174757281554e-05, "loss": 0.1998, "step": 11084 }, { "epoch": 107.48795180722891, "grad_norm": 50.69864273071289, "learning_rate": 4.618932038834952e-05, "loss": 0.2571, "step": 11085 }, { "epoch": 107.49759036144579, "grad_norm": 14.744172096252441, "learning_rate": 4.6184466019417476e-05, "loss": 0.1845, "step": 11086 }, { "epoch": 107.50722891566265, "grad_norm": 6.5789794921875, "learning_rate": 4.617961165048544e-05, "loss": 0.1756, "step": 11087 }, { "epoch": 107.51686746987951, "grad_norm": 3.516972780227661, "learning_rate": 4.61747572815534e-05, "loss": 0.1664, "step": 11088 }, { "epoch": 107.52650602409639, "grad_norm": 4.054778099060059, "learning_rate": 4.616990291262136e-05, "loss": 0.102, "step": 11089 }, { "epoch": 107.53614457831326, "grad_norm": 110.36363983154297, "learning_rate": 4.616504854368932e-05, "loss": 0.3233, "step": 11090 }, { "epoch": 107.54578313253012, "grad_norm": 4.26071310043335, "learning_rate": 4.616019417475728e-05, "loss": 0.2656, "step": 11091 }, { "epoch": 107.55542168674698, "grad_norm": 5.556873321533203, "learning_rate": 4.615533980582524e-05, "loss": 0.1483, "step": 11092 }, { "epoch": 107.56506024096386, "grad_norm": 5.451356887817383, "learning_rate": 4.61504854368932e-05, "loss": 0.1231, "step": 11093 }, { "epoch": 107.57469879518072, "grad_norm": 20.51471519470215, "learning_rate": 4.6145631067961166e-05, "loss": 0.2266, "step": 11094 }, { "epoch": 107.58433734939759, "grad_norm": 22.021615982055664, "learning_rate": 4.6140776699029123e-05, "loss": 0.2047, "step": 11095 }, { "epoch": 107.59397590361446, "grad_norm": 4.9078569412231445, "learning_rate": 4.6135922330097095e-05, "loss": 0.2196, "step": 11096 }, { "epoch": 107.60361445783133, "grad_norm": 13.093103408813477, "learning_rate": 4.613106796116505e-05, "loss": 0.2179, "step": 11097 }, { "epoch": 107.61325301204819, "grad_norm": 3.284479856491089, "learning_rate": 4.612621359223302e-05, "loss": 0.1977, "step": 11098 }, { "epoch": 107.62289156626505, "grad_norm": 7.731938362121582, "learning_rate": 4.6121359223300975e-05, "loss": 0.167, "step": 11099 }, { "epoch": 107.63253012048193, "grad_norm": 14.941903114318848, "learning_rate": 4.611650485436894e-05, "loss": 0.1157, "step": 11100 }, { "epoch": 107.6421686746988, "grad_norm": 105.52275848388672, "learning_rate": 4.61116504854369e-05, "loss": 0.5391, "step": 11101 }, { "epoch": 107.65180722891566, "grad_norm": 70.89588928222656, "learning_rate": 4.6106796116504855e-05, "loss": 0.319, "step": 11102 }, { "epoch": 107.66144578313254, "grad_norm": 6.324420928955078, "learning_rate": 4.610194174757282e-05, "loss": 0.2181, "step": 11103 }, { "epoch": 107.6710843373494, "grad_norm": 30.74163055419922, "learning_rate": 4.609708737864078e-05, "loss": 0.253, "step": 11104 }, { "epoch": 107.68072289156626, "grad_norm": 35.26338577270508, "learning_rate": 4.609223300970874e-05, "loss": 0.3204, "step": 11105 }, { "epoch": 107.69036144578314, "grad_norm": 3.451414108276367, "learning_rate": 4.60873786407767e-05, "loss": 0.235, "step": 11106 }, { "epoch": 107.7, "grad_norm": 29.390762329101562, "learning_rate": 4.6082524271844664e-05, "loss": 0.1677, "step": 11107 }, { "epoch": 107.70963855421687, "grad_norm": 19.21241569519043, "learning_rate": 4.607766990291262e-05, "loss": 0.2047, "step": 11108 }, { "epoch": 107.71927710843373, "grad_norm": 11.166460037231445, "learning_rate": 4.6072815533980586e-05, "loss": 0.2516, "step": 11109 }, { "epoch": 107.7289156626506, "grad_norm": 10.653226852416992, "learning_rate": 4.6067961165048544e-05, "loss": 0.1859, "step": 11110 }, { "epoch": 107.73855421686747, "grad_norm": 14.511374473571777, "learning_rate": 4.606310679611651e-05, "loss": 0.2989, "step": 11111 }, { "epoch": 107.74819277108433, "grad_norm": 10.942553520202637, "learning_rate": 4.6058252427184466e-05, "loss": 0.2144, "step": 11112 }, { "epoch": 107.75783132530121, "grad_norm": 15.086186408996582, "learning_rate": 4.605339805825243e-05, "loss": 0.3165, "step": 11113 }, { "epoch": 107.76746987951807, "grad_norm": 10.229262351989746, "learning_rate": 4.604854368932039e-05, "loss": 0.3907, "step": 11114 }, { "epoch": 107.77710843373494, "grad_norm": 10.281782150268555, "learning_rate": 4.604368932038835e-05, "loss": 0.254, "step": 11115 }, { "epoch": 107.7867469879518, "grad_norm": 26.99677848815918, "learning_rate": 4.603883495145631e-05, "loss": 0.1892, "step": 11116 }, { "epoch": 107.79638554216868, "grad_norm": 21.930078506469727, "learning_rate": 4.6033980582524275e-05, "loss": 0.2191, "step": 11117 }, { "epoch": 107.80602409638554, "grad_norm": 29.979171752929688, "learning_rate": 4.602912621359223e-05, "loss": 0.3792, "step": 11118 }, { "epoch": 107.8156626506024, "grad_norm": 6.558770656585693, "learning_rate": 4.602427184466019e-05, "loss": 0.2835, "step": 11119 }, { "epoch": 107.82530120481928, "grad_norm": 9.423977851867676, "learning_rate": 4.6019417475728155e-05, "loss": 0.4544, "step": 11120 }, { "epoch": 107.83493975903615, "grad_norm": 69.08271789550781, "learning_rate": 4.601456310679612e-05, "loss": 0.4582, "step": 11121 }, { "epoch": 107.84457831325301, "grad_norm": 10.859062194824219, "learning_rate": 4.6009708737864085e-05, "loss": 0.1995, "step": 11122 }, { "epoch": 107.85421686746987, "grad_norm": 11.94498062133789, "learning_rate": 4.600485436893204e-05, "loss": 0.2533, "step": 11123 }, { "epoch": 107.86385542168675, "grad_norm": 4.683821678161621, "learning_rate": 4.600000000000001e-05, "loss": 0.3109, "step": 11124 }, { "epoch": 107.87349397590361, "grad_norm": 9.875345230102539, "learning_rate": 4.5995145631067965e-05, "loss": 0.2217, "step": 11125 }, { "epoch": 107.88313253012048, "grad_norm": 4.735963344573975, "learning_rate": 4.599029126213593e-05, "loss": 0.1851, "step": 11126 }, { "epoch": 107.89277108433735, "grad_norm": 75.20916748046875, "learning_rate": 4.598543689320389e-05, "loss": 0.3527, "step": 11127 }, { "epoch": 107.90240963855422, "grad_norm": 21.543895721435547, "learning_rate": 4.5980582524271845e-05, "loss": 0.2671, "step": 11128 }, { "epoch": 107.91204819277108, "grad_norm": 16.371315002441406, "learning_rate": 4.597572815533981e-05, "loss": 0.2171, "step": 11129 }, { "epoch": 107.92168674698796, "grad_norm": 51.566410064697266, "learning_rate": 4.597087378640777e-05, "loss": 0.2665, "step": 11130 }, { "epoch": 107.93132530120482, "grad_norm": 7.396758556365967, "learning_rate": 4.596601941747573e-05, "loss": 0.3632, "step": 11131 }, { "epoch": 107.94096385542169, "grad_norm": 13.349499702453613, "learning_rate": 4.596116504854369e-05, "loss": 0.4715, "step": 11132 }, { "epoch": 107.95060240963855, "grad_norm": 27.430999755859375, "learning_rate": 4.5956310679611654e-05, "loss": 0.4737, "step": 11133 }, { "epoch": 107.96024096385543, "grad_norm": 33.812400817871094, "learning_rate": 4.595145631067961e-05, "loss": 0.3158, "step": 11134 }, { "epoch": 107.96987951807229, "grad_norm": 28.421648025512695, "learning_rate": 4.5946601941747576e-05, "loss": 0.2403, "step": 11135 }, { "epoch": 107.97951807228915, "grad_norm": 31.878070831298828, "learning_rate": 4.5941747572815534e-05, "loss": 0.4579, "step": 11136 }, { "epoch": 107.98915662650603, "grad_norm": 31.10161781311035, "learning_rate": 4.59368932038835e-05, "loss": 0.2165, "step": 11137 }, { "epoch": 108.00481927710844, "grad_norm": 10.932039260864258, "learning_rate": 4.5932038834951456e-05, "loss": 0.4143, "step": 11138 }, { "epoch": 108.0144578313253, "grad_norm": 19.98540496826172, "learning_rate": 4.592718446601942e-05, "loss": 0.2935, "step": 11139 }, { "epoch": 108.02409638554217, "grad_norm": 13.02666187286377, "learning_rate": 4.592233009708738e-05, "loss": 0.3214, "step": 11140 }, { "epoch": 108.03373493975904, "grad_norm": 45.03093338012695, "learning_rate": 4.591747572815534e-05, "loss": 0.2419, "step": 11141 }, { "epoch": 108.0433734939759, "grad_norm": 10.093994140625, "learning_rate": 4.59126213592233e-05, "loss": 0.1404, "step": 11142 }, { "epoch": 108.05301204819277, "grad_norm": 7.489977836608887, "learning_rate": 4.5907766990291265e-05, "loss": 0.4262, "step": 11143 }, { "epoch": 108.06265060240963, "grad_norm": 29.100343704223633, "learning_rate": 4.590291262135922e-05, "loss": 0.4918, "step": 11144 }, { "epoch": 108.07228915662651, "grad_norm": 6.63843297958374, "learning_rate": 4.589805825242718e-05, "loss": 0.4203, "step": 11145 }, { "epoch": 108.08192771084337, "grad_norm": 23.007169723510742, "learning_rate": 4.589320388349515e-05, "loss": 0.2815, "step": 11146 }, { "epoch": 108.09156626506024, "grad_norm": 26.938535690307617, "learning_rate": 4.588834951456311e-05, "loss": 0.3681, "step": 11147 }, { "epoch": 108.10120481927711, "grad_norm": 2.9213814735412598, "learning_rate": 4.5883495145631074e-05, "loss": 0.1313, "step": 11148 }, { "epoch": 108.11084337349398, "grad_norm": 21.031721115112305, "learning_rate": 4.587864077669903e-05, "loss": 0.2561, "step": 11149 }, { "epoch": 108.12048192771084, "grad_norm": 18.019126892089844, "learning_rate": 4.5873786407766997e-05, "loss": 0.364, "step": 11150 }, { "epoch": 108.1301204819277, "grad_norm": 10.043051719665527, "learning_rate": 4.5868932038834954e-05, "loss": 0.2061, "step": 11151 }, { "epoch": 108.13975903614458, "grad_norm": 7.743768692016602, "learning_rate": 4.586407766990292e-05, "loss": 0.1864, "step": 11152 }, { "epoch": 108.14939759036145, "grad_norm": 24.257478713989258, "learning_rate": 4.585922330097088e-05, "loss": 0.2516, "step": 11153 }, { "epoch": 108.15903614457831, "grad_norm": 8.861090660095215, "learning_rate": 4.585436893203884e-05, "loss": 0.2636, "step": 11154 }, { "epoch": 108.16867469879519, "grad_norm": 6.680365562438965, "learning_rate": 4.58495145631068e-05, "loss": 0.2239, "step": 11155 }, { "epoch": 108.17831325301205, "grad_norm": 12.971166610717773, "learning_rate": 4.584466019417476e-05, "loss": 0.4429, "step": 11156 }, { "epoch": 108.18795180722891, "grad_norm": 17.32764434814453, "learning_rate": 4.583980582524272e-05, "loss": 0.4404, "step": 11157 }, { "epoch": 108.19759036144578, "grad_norm": 4.277747631072998, "learning_rate": 4.583495145631068e-05, "loss": 0.3787, "step": 11158 }, { "epoch": 108.20722891566265, "grad_norm": 7.66750431060791, "learning_rate": 4.5830097087378644e-05, "loss": 0.0952, "step": 11159 }, { "epoch": 108.21686746987952, "grad_norm": 47.086517333984375, "learning_rate": 4.58252427184466e-05, "loss": 0.1809, "step": 11160 }, { "epoch": 108.22650602409638, "grad_norm": 7.45754337310791, "learning_rate": 4.5820388349514566e-05, "loss": 0.1728, "step": 11161 }, { "epoch": 108.23614457831326, "grad_norm": 36.911922454833984, "learning_rate": 4.5815533980582524e-05, "loss": 0.3934, "step": 11162 }, { "epoch": 108.24578313253012, "grad_norm": 6.278107166290283, "learning_rate": 4.581067961165049e-05, "loss": 0.2633, "step": 11163 }, { "epoch": 108.25542168674698, "grad_norm": 4.0208210945129395, "learning_rate": 4.5805825242718446e-05, "loss": 0.2079, "step": 11164 }, { "epoch": 108.26506024096386, "grad_norm": 57.50072479248047, "learning_rate": 4.580097087378641e-05, "loss": 0.383, "step": 11165 }, { "epoch": 108.27469879518073, "grad_norm": 33.39020919799805, "learning_rate": 4.579611650485437e-05, "loss": 0.2642, "step": 11166 }, { "epoch": 108.28433734939759, "grad_norm": 8.101673126220703, "learning_rate": 4.579126213592233e-05, "loss": 0.3103, "step": 11167 }, { "epoch": 108.29397590361445, "grad_norm": 12.085634231567383, "learning_rate": 4.578640776699029e-05, "loss": 0.2861, "step": 11168 }, { "epoch": 108.30361445783133, "grad_norm": 20.603015899658203, "learning_rate": 4.5781553398058255e-05, "loss": 0.2243, "step": 11169 }, { "epoch": 108.3132530120482, "grad_norm": 7.66597843170166, "learning_rate": 4.577669902912621e-05, "loss": 0.2248, "step": 11170 }, { "epoch": 108.32289156626506, "grad_norm": 6.337036609649658, "learning_rate": 4.577184466019418e-05, "loss": 0.2649, "step": 11171 }, { "epoch": 108.33253012048193, "grad_norm": 9.086749076843262, "learning_rate": 4.576699029126214e-05, "loss": 0.1898, "step": 11172 }, { "epoch": 108.3421686746988, "grad_norm": 33.33723831176758, "learning_rate": 4.57621359223301e-05, "loss": 0.1765, "step": 11173 }, { "epoch": 108.35180722891566, "grad_norm": 25.82868766784668, "learning_rate": 4.5757281553398064e-05, "loss": 0.3854, "step": 11174 }, { "epoch": 108.36144578313252, "grad_norm": 11.304769515991211, "learning_rate": 4.575242718446602e-05, "loss": 0.2891, "step": 11175 }, { "epoch": 108.3710843373494, "grad_norm": 5.425894737243652, "learning_rate": 4.5747572815533986e-05, "loss": 0.2619, "step": 11176 }, { "epoch": 108.38072289156626, "grad_norm": 7.0526885986328125, "learning_rate": 4.5742718446601944e-05, "loss": 0.1563, "step": 11177 }, { "epoch": 108.39036144578313, "grad_norm": 15.698455810546875, "learning_rate": 4.573786407766991e-05, "loss": 0.3388, "step": 11178 }, { "epoch": 108.4, "grad_norm": 9.152106285095215, "learning_rate": 4.5733009708737866e-05, "loss": 0.2695, "step": 11179 }, { "epoch": 108.40963855421687, "grad_norm": 6.799868106842041, "learning_rate": 4.572815533980583e-05, "loss": 0.3, "step": 11180 }, { "epoch": 108.41927710843373, "grad_norm": 11.144774436950684, "learning_rate": 4.572330097087379e-05, "loss": 0.4668, "step": 11181 }, { "epoch": 108.4289156626506, "grad_norm": 3.7836432456970215, "learning_rate": 4.571844660194175e-05, "loss": 0.1948, "step": 11182 }, { "epoch": 108.43855421686747, "grad_norm": 5.9712982177734375, "learning_rate": 4.571359223300971e-05, "loss": 0.2132, "step": 11183 }, { "epoch": 108.44819277108434, "grad_norm": 27.524124145507812, "learning_rate": 4.570873786407767e-05, "loss": 0.3414, "step": 11184 }, { "epoch": 108.4578313253012, "grad_norm": 5.156991481781006, "learning_rate": 4.570388349514563e-05, "loss": 0.2643, "step": 11185 }, { "epoch": 108.46746987951808, "grad_norm": 11.182608604431152, "learning_rate": 4.569902912621359e-05, "loss": 0.2597, "step": 11186 }, { "epoch": 108.47710843373494, "grad_norm": 9.441567420959473, "learning_rate": 4.5694174757281556e-05, "loss": 0.2607, "step": 11187 }, { "epoch": 108.4867469879518, "grad_norm": 3.833472967147827, "learning_rate": 4.568932038834951e-05, "loss": 0.0922, "step": 11188 }, { "epoch": 108.49638554216868, "grad_norm": 5.539128303527832, "learning_rate": 4.568446601941748e-05, "loss": 0.3904, "step": 11189 }, { "epoch": 108.50602409638554, "grad_norm": 10.052135467529297, "learning_rate": 4.5679611650485436e-05, "loss": 0.2959, "step": 11190 }, { "epoch": 108.51566265060241, "grad_norm": 7.668038845062256, "learning_rate": 4.56747572815534e-05, "loss": 0.325, "step": 11191 }, { "epoch": 108.52530120481927, "grad_norm": 4.972437381744385, "learning_rate": 4.566990291262136e-05, "loss": 0.3941, "step": 11192 }, { "epoch": 108.53493975903615, "grad_norm": 8.569015502929688, "learning_rate": 4.566504854368932e-05, "loss": 0.4619, "step": 11193 }, { "epoch": 108.54457831325301, "grad_norm": 14.130305290222168, "learning_rate": 4.566019417475728e-05, "loss": 0.3181, "step": 11194 }, { "epoch": 108.55421686746988, "grad_norm": 14.298789978027344, "learning_rate": 4.5655339805825245e-05, "loss": 0.2304, "step": 11195 }, { "epoch": 108.56385542168675, "grad_norm": 5.880228042602539, "learning_rate": 4.56504854368932e-05, "loss": 0.2315, "step": 11196 }, { "epoch": 108.57349397590362, "grad_norm": 8.129633903503418, "learning_rate": 4.564563106796117e-05, "loss": 0.2408, "step": 11197 }, { "epoch": 108.58313253012048, "grad_norm": 26.08336639404297, "learning_rate": 4.564077669902913e-05, "loss": 0.351, "step": 11198 }, { "epoch": 108.59277108433734, "grad_norm": 4.798615455627441, "learning_rate": 4.563592233009709e-05, "loss": 0.2073, "step": 11199 }, { "epoch": 108.60240963855422, "grad_norm": 3.3364133834838867, "learning_rate": 4.5631067961165054e-05, "loss": 0.1723, "step": 11200 }, { "epoch": 108.61204819277108, "grad_norm": 16.963703155517578, "learning_rate": 4.562621359223301e-05, "loss": 0.4722, "step": 11201 }, { "epoch": 108.62168674698795, "grad_norm": 7.684924125671387, "learning_rate": 4.5621359223300976e-05, "loss": 0.3945, "step": 11202 }, { "epoch": 108.63132530120482, "grad_norm": 13.372913360595703, "learning_rate": 4.5616504854368934e-05, "loss": 0.3158, "step": 11203 }, { "epoch": 108.64096385542169, "grad_norm": 8.746904373168945, "learning_rate": 4.56116504854369e-05, "loss": 0.339, "step": 11204 }, { "epoch": 108.65060240963855, "grad_norm": 46.96705627441406, "learning_rate": 4.5606796116504856e-05, "loss": 0.2487, "step": 11205 }, { "epoch": 108.66024096385541, "grad_norm": 16.783161163330078, "learning_rate": 4.560194174757282e-05, "loss": 0.2824, "step": 11206 }, { "epoch": 108.66987951807229, "grad_norm": 6.151580810546875, "learning_rate": 4.559708737864078e-05, "loss": 0.197, "step": 11207 }, { "epoch": 108.67951807228916, "grad_norm": 9.76486873626709, "learning_rate": 4.559223300970874e-05, "loss": 0.1202, "step": 11208 }, { "epoch": 108.68915662650602, "grad_norm": 13.528465270996094, "learning_rate": 4.55873786407767e-05, "loss": 0.5023, "step": 11209 }, { "epoch": 108.6987951807229, "grad_norm": 5.6218791007995605, "learning_rate": 4.5582524271844665e-05, "loss": 0.1239, "step": 11210 }, { "epoch": 108.70843373493976, "grad_norm": 8.86724853515625, "learning_rate": 4.557766990291262e-05, "loss": 0.3794, "step": 11211 }, { "epoch": 108.71807228915662, "grad_norm": 33.84156799316406, "learning_rate": 4.557281553398058e-05, "loss": 0.4014, "step": 11212 }, { "epoch": 108.7277108433735, "grad_norm": 18.558095932006836, "learning_rate": 4.5567961165048545e-05, "loss": 0.3411, "step": 11213 }, { "epoch": 108.73734939759036, "grad_norm": 9.056071281433105, "learning_rate": 4.55631067961165e-05, "loss": 0.272, "step": 11214 }, { "epoch": 108.74698795180723, "grad_norm": 9.606773376464844, "learning_rate": 4.555825242718447e-05, "loss": 0.1433, "step": 11215 }, { "epoch": 108.75662650602409, "grad_norm": 23.50897789001465, "learning_rate": 4.5553398058252425e-05, "loss": 0.409, "step": 11216 }, { "epoch": 108.76626506024097, "grad_norm": 12.26097297668457, "learning_rate": 4.554854368932039e-05, "loss": 0.238, "step": 11217 }, { "epoch": 108.77590361445783, "grad_norm": 16.470972061157227, "learning_rate": 4.554368932038835e-05, "loss": 0.3032, "step": 11218 }, { "epoch": 108.7855421686747, "grad_norm": 31.081253051757812, "learning_rate": 4.553883495145631e-05, "loss": 0.3526, "step": 11219 }, { "epoch": 108.79518072289157, "grad_norm": 12.275836944580078, "learning_rate": 4.553398058252427e-05, "loss": 0.1457, "step": 11220 }, { "epoch": 108.80481927710844, "grad_norm": 6.4004364013671875, "learning_rate": 4.5529126213592235e-05, "loss": 0.2487, "step": 11221 }, { "epoch": 108.8144578313253, "grad_norm": 20.004817962646484, "learning_rate": 4.55242718446602e-05, "loss": 0.2895, "step": 11222 }, { "epoch": 108.82409638554216, "grad_norm": 6.9761271476745605, "learning_rate": 4.551941747572816e-05, "loss": 0.177, "step": 11223 }, { "epoch": 108.83373493975904, "grad_norm": 6.001438617706299, "learning_rate": 4.551456310679612e-05, "loss": 0.4066, "step": 11224 }, { "epoch": 108.8433734939759, "grad_norm": 32.343421936035156, "learning_rate": 4.550970873786408e-05, "loss": 0.2743, "step": 11225 }, { "epoch": 108.85301204819277, "grad_norm": 15.296836853027344, "learning_rate": 4.5504854368932044e-05, "loss": 0.2373, "step": 11226 }, { "epoch": 108.86265060240964, "grad_norm": 15.618605613708496, "learning_rate": 4.55e-05, "loss": 0.338, "step": 11227 }, { "epoch": 108.87228915662651, "grad_norm": 13.63076114654541, "learning_rate": 4.5495145631067966e-05, "loss": 0.2891, "step": 11228 }, { "epoch": 108.88192771084337, "grad_norm": 27.908788681030273, "learning_rate": 4.5490291262135924e-05, "loss": 0.3566, "step": 11229 }, { "epoch": 108.89156626506023, "grad_norm": 6.926332473754883, "learning_rate": 4.548543689320389e-05, "loss": 0.3371, "step": 11230 }, { "epoch": 108.90120481927711, "grad_norm": 6.498974323272705, "learning_rate": 4.5480582524271846e-05, "loss": 0.3727, "step": 11231 }, { "epoch": 108.91084337349398, "grad_norm": 25.5873966217041, "learning_rate": 4.547572815533981e-05, "loss": 0.2346, "step": 11232 }, { "epoch": 108.92048192771084, "grad_norm": 6.19932222366333, "learning_rate": 4.547087378640777e-05, "loss": 0.2755, "step": 11233 }, { "epoch": 108.93012048192772, "grad_norm": 10.355878829956055, "learning_rate": 4.546601941747573e-05, "loss": 0.2739, "step": 11234 }, { "epoch": 108.93975903614458, "grad_norm": 17.93820571899414, "learning_rate": 4.546116504854369e-05, "loss": 0.4793, "step": 11235 }, { "epoch": 108.94939759036144, "grad_norm": 14.583374977111816, "learning_rate": 4.5456310679611655e-05, "loss": 0.3038, "step": 11236 }, { "epoch": 108.95903614457832, "grad_norm": 13.83655834197998, "learning_rate": 4.545145631067961e-05, "loss": 0.2757, "step": 11237 }, { "epoch": 108.96867469879518, "grad_norm": 16.728321075439453, "learning_rate": 4.544660194174758e-05, "loss": 0.2262, "step": 11238 }, { "epoch": 108.97831325301205, "grad_norm": 7.708146095275879, "learning_rate": 4.5441747572815535e-05, "loss": 0.2477, "step": 11239 }, { "epoch": 108.98795180722891, "grad_norm": 7.219954967498779, "learning_rate": 4.543689320388349e-05, "loss": 0.1654, "step": 11240 }, { "epoch": 109.00361445783132, "grad_norm": 8.981738090515137, "learning_rate": 4.543203883495146e-05, "loss": 0.1173, "step": 11241 }, { "epoch": 109.0132530120482, "grad_norm": 12.22797679901123, "learning_rate": 4.5427184466019415e-05, "loss": 0.4376, "step": 11242 }, { "epoch": 109.02289156626506, "grad_norm": 10.719744682312012, "learning_rate": 4.542233009708738e-05, "loss": 0.2375, "step": 11243 }, { "epoch": 109.03253012048192, "grad_norm": 7.324024677276611, "learning_rate": 4.541747572815534e-05, "loss": 0.2985, "step": 11244 }, { "epoch": 109.0421686746988, "grad_norm": 20.449996948242188, "learning_rate": 4.54126213592233e-05, "loss": 0.1747, "step": 11245 }, { "epoch": 109.05180722891566, "grad_norm": 10.426379203796387, "learning_rate": 4.540776699029126e-05, "loss": 0.1817, "step": 11246 }, { "epoch": 109.06144578313253, "grad_norm": 9.610067367553711, "learning_rate": 4.540291262135923e-05, "loss": 0.1166, "step": 11247 }, { "epoch": 109.0710843373494, "grad_norm": 6.711311340332031, "learning_rate": 4.539805825242719e-05, "loss": 0.1292, "step": 11248 }, { "epoch": 109.08072289156627, "grad_norm": 9.484036445617676, "learning_rate": 4.5393203883495153e-05, "loss": 0.3069, "step": 11249 }, { "epoch": 109.09036144578313, "grad_norm": 10.816685676574707, "learning_rate": 4.538834951456311e-05, "loss": 0.2391, "step": 11250 }, { "epoch": 109.1, "grad_norm": 12.768153190612793, "learning_rate": 4.538349514563107e-05, "loss": 0.3822, "step": 11251 }, { "epoch": 109.10963855421687, "grad_norm": 32.01346206665039, "learning_rate": 4.5378640776699033e-05, "loss": 0.2263, "step": 11252 }, { "epoch": 109.11927710843374, "grad_norm": 24.05876922607422, "learning_rate": 4.537378640776699e-05, "loss": 0.2486, "step": 11253 }, { "epoch": 109.1289156626506, "grad_norm": 19.45402717590332, "learning_rate": 4.5368932038834956e-05, "loss": 0.4048, "step": 11254 }, { "epoch": 109.13855421686748, "grad_norm": 8.024761199951172, "learning_rate": 4.5364077669902914e-05, "loss": 0.2266, "step": 11255 }, { "epoch": 109.14819277108434, "grad_norm": 25.873374938964844, "learning_rate": 4.535922330097088e-05, "loss": 0.2942, "step": 11256 }, { "epoch": 109.1578313253012, "grad_norm": 5.802104949951172, "learning_rate": 4.5354368932038836e-05, "loss": 0.3335, "step": 11257 }, { "epoch": 109.16746987951807, "grad_norm": 31.440832138061523, "learning_rate": 4.53495145631068e-05, "loss": 0.2082, "step": 11258 }, { "epoch": 109.17710843373494, "grad_norm": 10.363061904907227, "learning_rate": 4.534466019417476e-05, "loss": 0.342, "step": 11259 }, { "epoch": 109.1867469879518, "grad_norm": 44.663082122802734, "learning_rate": 4.533980582524272e-05, "loss": 0.3097, "step": 11260 }, { "epoch": 109.19638554216867, "grad_norm": 6.930891513824463, "learning_rate": 4.533495145631068e-05, "loss": 0.2686, "step": 11261 }, { "epoch": 109.20602409638555, "grad_norm": 10.626091957092285, "learning_rate": 4.5330097087378645e-05, "loss": 0.4014, "step": 11262 }, { "epoch": 109.21566265060241, "grad_norm": 17.122617721557617, "learning_rate": 4.53252427184466e-05, "loss": 0.4321, "step": 11263 }, { "epoch": 109.22530120481927, "grad_norm": 10.878271102905273, "learning_rate": 4.532038834951457e-05, "loss": 0.2495, "step": 11264 }, { "epoch": 109.23493975903614, "grad_norm": 8.0449800491333, "learning_rate": 4.5315533980582525e-05, "loss": 0.1695, "step": 11265 }, { "epoch": 109.24457831325302, "grad_norm": 7.014760971069336, "learning_rate": 4.531067961165049e-05, "loss": 0.3036, "step": 11266 }, { "epoch": 109.25421686746988, "grad_norm": 27.01035499572754, "learning_rate": 4.530582524271845e-05, "loss": 0.5748, "step": 11267 }, { "epoch": 109.26385542168674, "grad_norm": 208.32846069335938, "learning_rate": 4.5300970873786405e-05, "loss": 0.1827, "step": 11268 }, { "epoch": 109.27349397590362, "grad_norm": 8.163400650024414, "learning_rate": 4.529611650485437e-05, "loss": 0.5097, "step": 11269 }, { "epoch": 109.28313253012048, "grad_norm": 29.21881866455078, "learning_rate": 4.529126213592233e-05, "loss": 0.1835, "step": 11270 }, { "epoch": 109.29277108433735, "grad_norm": 9.192310333251953, "learning_rate": 4.528640776699029e-05, "loss": 0.2685, "step": 11271 }, { "epoch": 109.30240963855422, "grad_norm": 27.712854385375977, "learning_rate": 4.5281553398058256e-05, "loss": 0.2166, "step": 11272 }, { "epoch": 109.31204819277109, "grad_norm": 18.905773162841797, "learning_rate": 4.527669902912622e-05, "loss": 0.398, "step": 11273 }, { "epoch": 109.32168674698795, "grad_norm": 6.588854789733887, "learning_rate": 4.527184466019418e-05, "loss": 0.2405, "step": 11274 }, { "epoch": 109.33132530120481, "grad_norm": 9.035293579101562, "learning_rate": 4.526699029126214e-05, "loss": 0.457, "step": 11275 }, { "epoch": 109.34096385542169, "grad_norm": 23.338558197021484, "learning_rate": 4.52621359223301e-05, "loss": 0.4068, "step": 11276 }, { "epoch": 109.35060240963855, "grad_norm": 10.395730018615723, "learning_rate": 4.525728155339806e-05, "loss": 0.2184, "step": 11277 }, { "epoch": 109.36024096385542, "grad_norm": 5.747894763946533, "learning_rate": 4.525242718446602e-05, "loss": 0.1298, "step": 11278 }, { "epoch": 109.3698795180723, "grad_norm": 8.107144355773926, "learning_rate": 4.524757281553398e-05, "loss": 0.2178, "step": 11279 }, { "epoch": 109.37951807228916, "grad_norm": 10.7238187789917, "learning_rate": 4.5242718446601946e-05, "loss": 0.3706, "step": 11280 }, { "epoch": 109.38915662650602, "grad_norm": 5.5480194091796875, "learning_rate": 4.52378640776699e-05, "loss": 0.22, "step": 11281 }, { "epoch": 109.39879518072289, "grad_norm": 10.153289794921875, "learning_rate": 4.523300970873787e-05, "loss": 0.3629, "step": 11282 }, { "epoch": 109.40843373493976, "grad_norm": 14.798465728759766, "learning_rate": 4.5228155339805826e-05, "loss": 0.2219, "step": 11283 }, { "epoch": 109.41807228915663, "grad_norm": 4.756551742553711, "learning_rate": 4.522330097087379e-05, "loss": 0.1873, "step": 11284 }, { "epoch": 109.42771084337349, "grad_norm": 5.225057125091553, "learning_rate": 4.521844660194175e-05, "loss": 0.1688, "step": 11285 }, { "epoch": 109.43734939759037, "grad_norm": 24.421085357666016, "learning_rate": 4.521359223300971e-05, "loss": 0.2961, "step": 11286 }, { "epoch": 109.44698795180723, "grad_norm": 16.78113555908203, "learning_rate": 4.520873786407767e-05, "loss": 0.3051, "step": 11287 }, { "epoch": 109.4566265060241, "grad_norm": 8.171030044555664, "learning_rate": 4.5203883495145635e-05, "loss": 0.2664, "step": 11288 }, { "epoch": 109.46626506024096, "grad_norm": 17.921361923217773, "learning_rate": 4.519902912621359e-05, "loss": 0.2352, "step": 11289 }, { "epoch": 109.47590361445783, "grad_norm": 7.3994574546813965, "learning_rate": 4.519417475728156e-05, "loss": 0.2427, "step": 11290 }, { "epoch": 109.4855421686747, "grad_norm": 9.009767532348633, "learning_rate": 4.5189320388349515e-05, "loss": 0.4302, "step": 11291 }, { "epoch": 109.49518072289156, "grad_norm": 8.810751914978027, "learning_rate": 4.518446601941748e-05, "loss": 0.3293, "step": 11292 }, { "epoch": 109.50481927710844, "grad_norm": 14.934795379638672, "learning_rate": 4.517961165048544e-05, "loss": 0.4244, "step": 11293 }, { "epoch": 109.5144578313253, "grad_norm": 22.042261123657227, "learning_rate": 4.5174757281553395e-05, "loss": 0.0867, "step": 11294 }, { "epoch": 109.52409638554217, "grad_norm": 20.62059211730957, "learning_rate": 4.516990291262136e-05, "loss": 0.4289, "step": 11295 }, { "epoch": 109.53373493975904, "grad_norm": 52.766143798828125, "learning_rate": 4.516504854368932e-05, "loss": 0.1904, "step": 11296 }, { "epoch": 109.5433734939759, "grad_norm": 25.809322357177734, "learning_rate": 4.516019417475728e-05, "loss": 0.2189, "step": 11297 }, { "epoch": 109.55301204819277, "grad_norm": 12.1113862991333, "learning_rate": 4.5155339805825246e-05, "loss": 0.3161, "step": 11298 }, { "epoch": 109.56265060240963, "grad_norm": 187.0673370361328, "learning_rate": 4.515048543689321e-05, "loss": 0.2485, "step": 11299 }, { "epoch": 109.57228915662651, "grad_norm": 30.413881301879883, "learning_rate": 4.514563106796117e-05, "loss": 0.312, "step": 11300 }, { "epoch": 109.58192771084337, "grad_norm": 70.71604919433594, "learning_rate": 4.514077669902913e-05, "loss": 0.2632, "step": 11301 }, { "epoch": 109.59156626506024, "grad_norm": 6.3942461013793945, "learning_rate": 4.513592233009709e-05, "loss": 0.3728, "step": 11302 }, { "epoch": 109.60120481927711, "grad_norm": 6.941219806671143, "learning_rate": 4.5131067961165055e-05, "loss": 0.2199, "step": 11303 }, { "epoch": 109.61084337349398, "grad_norm": 24.386621475219727, "learning_rate": 4.512621359223301e-05, "loss": 0.3378, "step": 11304 }, { "epoch": 109.62048192771084, "grad_norm": 2.5203027725219727, "learning_rate": 4.512135922330097e-05, "loss": 0.1907, "step": 11305 }, { "epoch": 109.6301204819277, "grad_norm": 34.753700256347656, "learning_rate": 4.5116504854368935e-05, "loss": 0.2337, "step": 11306 }, { "epoch": 109.63975903614458, "grad_norm": 13.66653823852539, "learning_rate": 4.511165048543689e-05, "loss": 0.3614, "step": 11307 }, { "epoch": 109.64939759036145, "grad_norm": 10.594861030578613, "learning_rate": 4.510679611650486e-05, "loss": 0.2016, "step": 11308 }, { "epoch": 109.65903614457831, "grad_norm": 7.558387756347656, "learning_rate": 4.5101941747572815e-05, "loss": 0.2679, "step": 11309 }, { "epoch": 109.66867469879519, "grad_norm": 13.902486801147461, "learning_rate": 4.509708737864078e-05, "loss": 0.3579, "step": 11310 }, { "epoch": 109.67831325301205, "grad_norm": 15.184296607971191, "learning_rate": 4.509223300970874e-05, "loss": 0.2157, "step": 11311 }, { "epoch": 109.68795180722891, "grad_norm": 43.61363220214844, "learning_rate": 4.50873786407767e-05, "loss": 0.2276, "step": 11312 }, { "epoch": 109.69759036144578, "grad_norm": 11.757699966430664, "learning_rate": 4.508252427184466e-05, "loss": 0.3243, "step": 11313 }, { "epoch": 109.70722891566265, "grad_norm": 5.625895977020264, "learning_rate": 4.5077669902912625e-05, "loss": 0.1185, "step": 11314 }, { "epoch": 109.71686746987952, "grad_norm": 15.81208610534668, "learning_rate": 4.507281553398058e-05, "loss": 0.1759, "step": 11315 }, { "epoch": 109.72650602409638, "grad_norm": 51.706748962402344, "learning_rate": 4.506796116504855e-05, "loss": 0.2155, "step": 11316 }, { "epoch": 109.73614457831326, "grad_norm": 34.96747970581055, "learning_rate": 4.5063106796116505e-05, "loss": 0.3664, "step": 11317 }, { "epoch": 109.74578313253012, "grad_norm": 22.162302017211914, "learning_rate": 4.505825242718447e-05, "loss": 0.2142, "step": 11318 }, { "epoch": 109.75542168674698, "grad_norm": 9.659037590026855, "learning_rate": 4.505339805825243e-05, "loss": 0.3693, "step": 11319 }, { "epoch": 109.76506024096386, "grad_norm": 37.717166900634766, "learning_rate": 4.504854368932039e-05, "loss": 0.1911, "step": 11320 }, { "epoch": 109.77469879518073, "grad_norm": 9.083321571350098, "learning_rate": 4.504368932038835e-05, "loss": 0.2736, "step": 11321 }, { "epoch": 109.78433734939759, "grad_norm": 30.630496978759766, "learning_rate": 4.503883495145631e-05, "loss": 0.3413, "step": 11322 }, { "epoch": 109.79397590361445, "grad_norm": 10.053426742553711, "learning_rate": 4.503398058252428e-05, "loss": 0.2308, "step": 11323 }, { "epoch": 109.80361445783133, "grad_norm": 13.119227409362793, "learning_rate": 4.5029126213592236e-05, "loss": 0.1516, "step": 11324 }, { "epoch": 109.8132530120482, "grad_norm": 12.986056327819824, "learning_rate": 4.50242718446602e-05, "loss": 0.1884, "step": 11325 }, { "epoch": 109.82289156626506, "grad_norm": 8.90785026550293, "learning_rate": 4.501941747572816e-05, "loss": 0.2279, "step": 11326 }, { "epoch": 109.83253012048193, "grad_norm": 20.372575759887695, "learning_rate": 4.501456310679612e-05, "loss": 0.4881, "step": 11327 }, { "epoch": 109.8421686746988, "grad_norm": 39.24524688720703, "learning_rate": 4.500970873786408e-05, "loss": 0.5312, "step": 11328 }, { "epoch": 109.85180722891566, "grad_norm": 5.120588302612305, "learning_rate": 4.5004854368932045e-05, "loss": 0.2485, "step": 11329 }, { "epoch": 109.86144578313252, "grad_norm": 9.261487007141113, "learning_rate": 4.5e-05, "loss": 0.1777, "step": 11330 }, { "epoch": 109.8710843373494, "grad_norm": 11.813562393188477, "learning_rate": 4.499514563106797e-05, "loss": 0.1808, "step": 11331 }, { "epoch": 109.88072289156626, "grad_norm": 14.07006549835205, "learning_rate": 4.4990291262135925e-05, "loss": 0.2715, "step": 11332 }, { "epoch": 109.89036144578313, "grad_norm": 8.867448806762695, "learning_rate": 4.498543689320388e-05, "loss": 0.3402, "step": 11333 }, { "epoch": 109.9, "grad_norm": 6.392765998840332, "learning_rate": 4.498058252427185e-05, "loss": 0.4167, "step": 11334 }, { "epoch": 109.90963855421687, "grad_norm": 11.993390083312988, "learning_rate": 4.4975728155339805e-05, "loss": 0.3349, "step": 11335 }, { "epoch": 109.91927710843373, "grad_norm": 11.731827735900879, "learning_rate": 4.497087378640777e-05, "loss": 0.2827, "step": 11336 }, { "epoch": 109.9289156626506, "grad_norm": 9.324227333068848, "learning_rate": 4.496601941747573e-05, "loss": 0.3163, "step": 11337 }, { "epoch": 109.93855421686747, "grad_norm": 4.4041428565979, "learning_rate": 4.496116504854369e-05, "loss": 0.3353, "step": 11338 }, { "epoch": 109.94819277108434, "grad_norm": 7.460310459136963, "learning_rate": 4.495631067961165e-05, "loss": 0.2524, "step": 11339 }, { "epoch": 109.9578313253012, "grad_norm": 28.88474464416504, "learning_rate": 4.4951456310679614e-05, "loss": 0.4535, "step": 11340 }, { "epoch": 109.96746987951808, "grad_norm": 6.164007186889648, "learning_rate": 4.494660194174757e-05, "loss": 0.2232, "step": 11341 }, { "epoch": 109.97710843373494, "grad_norm": 26.892751693725586, "learning_rate": 4.4941747572815537e-05, "loss": 0.3391, "step": 11342 }, { "epoch": 109.9867469879518, "grad_norm": 18.822153091430664, "learning_rate": 4.4936893203883494e-05, "loss": 0.3782, "step": 11343 }, { "epoch": 110.00240963855421, "grad_norm": 6.459123611450195, "learning_rate": 4.493203883495146e-05, "loss": 0.3125, "step": 11344 }, { "epoch": 110.01204819277109, "grad_norm": 3.4549219608306885, "learning_rate": 4.492718446601942e-05, "loss": 0.188, "step": 11345 }, { "epoch": 110.02168674698795, "grad_norm": 29.193710327148438, "learning_rate": 4.492233009708738e-05, "loss": 0.3788, "step": 11346 }, { "epoch": 110.03132530120482, "grad_norm": 14.536090850830078, "learning_rate": 4.491747572815534e-05, "loss": 0.2923, "step": 11347 }, { "epoch": 110.04096385542168, "grad_norm": 13.778714179992676, "learning_rate": 4.4912621359223303e-05, "loss": 0.4171, "step": 11348 }, { "epoch": 110.05060240963856, "grad_norm": 23.49952507019043, "learning_rate": 4.490776699029127e-05, "loss": 0.3123, "step": 11349 }, { "epoch": 110.06024096385542, "grad_norm": 8.637740135192871, "learning_rate": 4.4902912621359226e-05, "loss": 0.2064, "step": 11350 }, { "epoch": 110.06987951807228, "grad_norm": 28.045164108276367, "learning_rate": 4.489805825242719e-05, "loss": 0.2869, "step": 11351 }, { "epoch": 110.07951807228916, "grad_norm": 35.435646057128906, "learning_rate": 4.489320388349515e-05, "loss": 0.255, "step": 11352 }, { "epoch": 110.08915662650602, "grad_norm": 14.055306434631348, "learning_rate": 4.488834951456311e-05, "loss": 0.2582, "step": 11353 }, { "epoch": 110.09879518072289, "grad_norm": 22.3721866607666, "learning_rate": 4.488349514563107e-05, "loss": 0.4614, "step": 11354 }, { "epoch": 110.10843373493977, "grad_norm": 10.702208518981934, "learning_rate": 4.4878640776699035e-05, "loss": 0.3956, "step": 11355 }, { "epoch": 110.11807228915663, "grad_norm": 10.653366088867188, "learning_rate": 4.487378640776699e-05, "loss": 0.2792, "step": 11356 }, { "epoch": 110.12771084337349, "grad_norm": 23.774259567260742, "learning_rate": 4.486893203883496e-05, "loss": 0.3357, "step": 11357 }, { "epoch": 110.13734939759036, "grad_norm": 15.039008140563965, "learning_rate": 4.4864077669902915e-05, "loss": 0.5142, "step": 11358 }, { "epoch": 110.14698795180723, "grad_norm": 19.308507919311523, "learning_rate": 4.485922330097088e-05, "loss": 0.2491, "step": 11359 }, { "epoch": 110.1566265060241, "grad_norm": 7.683399677276611, "learning_rate": 4.485436893203884e-05, "loss": 0.1973, "step": 11360 }, { "epoch": 110.16626506024096, "grad_norm": 18.24204444885254, "learning_rate": 4.4849514563106795e-05, "loss": 0.3772, "step": 11361 }, { "epoch": 110.17590361445784, "grad_norm": 5.9255452156066895, "learning_rate": 4.484466019417476e-05, "loss": 0.2556, "step": 11362 }, { "epoch": 110.1855421686747, "grad_norm": 5.855328559875488, "learning_rate": 4.483980582524272e-05, "loss": 0.2049, "step": 11363 }, { "epoch": 110.19518072289156, "grad_norm": 16.501415252685547, "learning_rate": 4.483495145631068e-05, "loss": 0.4208, "step": 11364 }, { "epoch": 110.20481927710843, "grad_norm": 12.1747407913208, "learning_rate": 4.483009708737864e-05, "loss": 0.2939, "step": 11365 }, { "epoch": 110.2144578313253, "grad_norm": 13.64267635345459, "learning_rate": 4.4825242718446604e-05, "loss": 0.3218, "step": 11366 }, { "epoch": 110.22409638554217, "grad_norm": 16.968664169311523, "learning_rate": 4.482038834951456e-05, "loss": 0.2744, "step": 11367 }, { "epoch": 110.23373493975903, "grad_norm": 9.501226425170898, "learning_rate": 4.4815533980582526e-05, "loss": 0.2482, "step": 11368 }, { "epoch": 110.24337349397591, "grad_norm": 12.626330375671387, "learning_rate": 4.4810679611650484e-05, "loss": 0.3112, "step": 11369 }, { "epoch": 110.25301204819277, "grad_norm": 11.08218765258789, "learning_rate": 4.480582524271845e-05, "loss": 0.4544, "step": 11370 }, { "epoch": 110.26265060240964, "grad_norm": 10.181961059570312, "learning_rate": 4.4800970873786406e-05, "loss": 0.3001, "step": 11371 }, { "epoch": 110.2722891566265, "grad_norm": 21.81587791442871, "learning_rate": 4.479611650485437e-05, "loss": 0.3023, "step": 11372 }, { "epoch": 110.28192771084338, "grad_norm": 21.258098602294922, "learning_rate": 4.4791262135922335e-05, "loss": 0.3628, "step": 11373 }, { "epoch": 110.29156626506024, "grad_norm": 16.647436141967773, "learning_rate": 4.478640776699029e-05, "loss": 0.1932, "step": 11374 }, { "epoch": 110.3012048192771, "grad_norm": 14.42724323272705, "learning_rate": 4.478155339805826e-05, "loss": 0.2836, "step": 11375 }, { "epoch": 110.31084337349398, "grad_norm": 5.430622100830078, "learning_rate": 4.4776699029126216e-05, "loss": 0.2382, "step": 11376 }, { "epoch": 110.32048192771084, "grad_norm": 3.5313363075256348, "learning_rate": 4.477184466019418e-05, "loss": 0.1805, "step": 11377 }, { "epoch": 110.33012048192771, "grad_norm": 9.151607513427734, "learning_rate": 4.476699029126214e-05, "loss": 0.2501, "step": 11378 }, { "epoch": 110.33975903614459, "grad_norm": 10.611799240112305, "learning_rate": 4.47621359223301e-05, "loss": 0.2146, "step": 11379 }, { "epoch": 110.34939759036145, "grad_norm": 11.167816162109375, "learning_rate": 4.475728155339806e-05, "loss": 0.3233, "step": 11380 }, { "epoch": 110.35903614457831, "grad_norm": 9.686898231506348, "learning_rate": 4.4752427184466025e-05, "loss": 0.165, "step": 11381 }, { "epoch": 110.36867469879518, "grad_norm": 14.508933067321777, "learning_rate": 4.474757281553398e-05, "loss": 0.3142, "step": 11382 }, { "epoch": 110.37831325301205, "grad_norm": 10.712037086486816, "learning_rate": 4.474271844660195e-05, "loss": 0.3358, "step": 11383 }, { "epoch": 110.38795180722892, "grad_norm": 26.75946807861328, "learning_rate": 4.4737864077669905e-05, "loss": 0.1926, "step": 11384 }, { "epoch": 110.39759036144578, "grad_norm": 7.169553756713867, "learning_rate": 4.473300970873787e-05, "loss": 0.1807, "step": 11385 }, { "epoch": 110.40722891566266, "grad_norm": 7.619318008422852, "learning_rate": 4.472815533980583e-05, "loss": 0.2595, "step": 11386 }, { "epoch": 110.41686746987952, "grad_norm": 9.441082954406738, "learning_rate": 4.472330097087379e-05, "loss": 0.3491, "step": 11387 }, { "epoch": 110.42650602409638, "grad_norm": 4.678426742553711, "learning_rate": 4.471844660194175e-05, "loss": 0.2888, "step": 11388 }, { "epoch": 110.43614457831325, "grad_norm": 10.942955017089844, "learning_rate": 4.471359223300971e-05, "loss": 0.5209, "step": 11389 }, { "epoch": 110.44578313253012, "grad_norm": 8.484807968139648, "learning_rate": 4.470873786407767e-05, "loss": 0.3876, "step": 11390 }, { "epoch": 110.45542168674699, "grad_norm": 18.58946990966797, "learning_rate": 4.470388349514563e-05, "loss": 0.2463, "step": 11391 }, { "epoch": 110.46506024096385, "grad_norm": 7.357611179351807, "learning_rate": 4.4699029126213594e-05, "loss": 0.2074, "step": 11392 }, { "epoch": 110.47469879518073, "grad_norm": 6.458904266357422, "learning_rate": 4.469417475728155e-05, "loss": 0.2663, "step": 11393 }, { "epoch": 110.48433734939759, "grad_norm": 10.455294609069824, "learning_rate": 4.4689320388349516e-05, "loss": 0.2592, "step": 11394 }, { "epoch": 110.49397590361446, "grad_norm": 11.243141174316406, "learning_rate": 4.4684466019417474e-05, "loss": 0.182, "step": 11395 }, { "epoch": 110.50361445783132, "grad_norm": 17.972698211669922, "learning_rate": 4.467961165048544e-05, "loss": 0.3093, "step": 11396 }, { "epoch": 110.5132530120482, "grad_norm": 39.63115692138672, "learning_rate": 4.4674757281553396e-05, "loss": 0.37, "step": 11397 }, { "epoch": 110.52289156626506, "grad_norm": 17.736942291259766, "learning_rate": 4.466990291262136e-05, "loss": 0.5469, "step": 11398 }, { "epoch": 110.53253012048192, "grad_norm": 9.207321166992188, "learning_rate": 4.4665048543689325e-05, "loss": 0.3547, "step": 11399 }, { "epoch": 110.5421686746988, "grad_norm": 28.030261993408203, "learning_rate": 4.466019417475728e-05, "loss": 0.4354, "step": 11400 }, { "epoch": 110.55180722891566, "grad_norm": 20.277729034423828, "learning_rate": 4.465533980582525e-05, "loss": 0.3606, "step": 11401 }, { "epoch": 110.56144578313253, "grad_norm": 10.83202838897705, "learning_rate": 4.4650485436893205e-05, "loss": 0.3248, "step": 11402 }, { "epoch": 110.5710843373494, "grad_norm": 5.501221179962158, "learning_rate": 4.464563106796117e-05, "loss": 0.2411, "step": 11403 }, { "epoch": 110.58072289156627, "grad_norm": 26.36883544921875, "learning_rate": 4.464077669902913e-05, "loss": 0.4232, "step": 11404 }, { "epoch": 110.59036144578313, "grad_norm": 9.394774436950684, "learning_rate": 4.463592233009709e-05, "loss": 0.186, "step": 11405 }, { "epoch": 110.6, "grad_norm": 5.9397687911987305, "learning_rate": 4.463106796116505e-05, "loss": 0.147, "step": 11406 }, { "epoch": 110.60963855421687, "grad_norm": 12.752705574035645, "learning_rate": 4.4626213592233014e-05, "loss": 0.2041, "step": 11407 }, { "epoch": 110.61927710843374, "grad_norm": 3.920658588409424, "learning_rate": 4.462135922330097e-05, "loss": 0.2787, "step": 11408 }, { "epoch": 110.6289156626506, "grad_norm": 11.2828950881958, "learning_rate": 4.461650485436894e-05, "loss": 0.1913, "step": 11409 }, { "epoch": 110.63855421686748, "grad_norm": 13.288468360900879, "learning_rate": 4.4611650485436894e-05, "loss": 0.2031, "step": 11410 }, { "epoch": 110.64819277108434, "grad_norm": 8.722664833068848, "learning_rate": 4.460679611650486e-05, "loss": 0.2244, "step": 11411 }, { "epoch": 110.6578313253012, "grad_norm": 15.138320922851562, "learning_rate": 4.460194174757282e-05, "loss": 0.3756, "step": 11412 }, { "epoch": 110.66746987951807, "grad_norm": 10.029916763305664, "learning_rate": 4.459708737864078e-05, "loss": 0.2472, "step": 11413 }, { "epoch": 110.67710843373494, "grad_norm": 7.268595218658447, "learning_rate": 4.459223300970874e-05, "loss": 0.2195, "step": 11414 }, { "epoch": 110.6867469879518, "grad_norm": 6.136338710784912, "learning_rate": 4.4587378640776704e-05, "loss": 0.2289, "step": 11415 }, { "epoch": 110.69638554216867, "grad_norm": 6.619226932525635, "learning_rate": 4.458252427184466e-05, "loss": 0.1443, "step": 11416 }, { "epoch": 110.70602409638555, "grad_norm": 12.33488941192627, "learning_rate": 4.457766990291262e-05, "loss": 0.1566, "step": 11417 }, { "epoch": 110.71566265060241, "grad_norm": 15.51356029510498, "learning_rate": 4.4572815533980584e-05, "loss": 0.3886, "step": 11418 }, { "epoch": 110.72530120481927, "grad_norm": 5.209867000579834, "learning_rate": 4.456796116504854e-05, "loss": 0.169, "step": 11419 }, { "epoch": 110.73493975903614, "grad_norm": 8.561202049255371, "learning_rate": 4.4563106796116506e-05, "loss": 0.3304, "step": 11420 }, { "epoch": 110.74457831325302, "grad_norm": 8.137664794921875, "learning_rate": 4.4558252427184464e-05, "loss": 0.3393, "step": 11421 }, { "epoch": 110.75421686746988, "grad_norm": 17.90236473083496, "learning_rate": 4.455339805825243e-05, "loss": 0.4173, "step": 11422 }, { "epoch": 110.76385542168674, "grad_norm": 16.871938705444336, "learning_rate": 4.4548543689320386e-05, "loss": 0.2125, "step": 11423 }, { "epoch": 110.77349397590362, "grad_norm": 20.592504501342773, "learning_rate": 4.454368932038836e-05, "loss": 0.3679, "step": 11424 }, { "epoch": 110.78313253012048, "grad_norm": 23.025299072265625, "learning_rate": 4.4538834951456315e-05, "loss": 0.4157, "step": 11425 }, { "epoch": 110.79277108433735, "grad_norm": 9.611645698547363, "learning_rate": 4.453398058252427e-05, "loss": 0.1558, "step": 11426 }, { "epoch": 110.80240963855422, "grad_norm": 5.573014259338379, "learning_rate": 4.452912621359224e-05, "loss": 0.2172, "step": 11427 }, { "epoch": 110.81204819277109, "grad_norm": 41.52082824707031, "learning_rate": 4.4524271844660195e-05, "loss": 0.154, "step": 11428 }, { "epoch": 110.82168674698795, "grad_norm": 10.863691329956055, "learning_rate": 4.451941747572816e-05, "loss": 0.3094, "step": 11429 }, { "epoch": 110.83132530120481, "grad_norm": 9.93355655670166, "learning_rate": 4.451456310679612e-05, "loss": 0.2133, "step": 11430 }, { "epoch": 110.84096385542169, "grad_norm": 12.190163612365723, "learning_rate": 4.450970873786408e-05, "loss": 0.338, "step": 11431 }, { "epoch": 110.85060240963855, "grad_norm": 7.207610607147217, "learning_rate": 4.450485436893204e-05, "loss": 0.255, "step": 11432 }, { "epoch": 110.86024096385542, "grad_norm": 17.262462615966797, "learning_rate": 4.4500000000000004e-05, "loss": 0.2721, "step": 11433 }, { "epoch": 110.8698795180723, "grad_norm": 7.20818567276001, "learning_rate": 4.449514563106796e-05, "loss": 0.258, "step": 11434 }, { "epoch": 110.87951807228916, "grad_norm": 15.392395973205566, "learning_rate": 4.4490291262135927e-05, "loss": 0.3392, "step": 11435 }, { "epoch": 110.88915662650602, "grad_norm": 12.323324203491211, "learning_rate": 4.4485436893203884e-05, "loss": 0.1852, "step": 11436 }, { "epoch": 110.89879518072289, "grad_norm": 19.324541091918945, "learning_rate": 4.448058252427185e-05, "loss": 0.3063, "step": 11437 }, { "epoch": 110.90843373493976, "grad_norm": 3.47314715385437, "learning_rate": 4.4475728155339807e-05, "loss": 0.1745, "step": 11438 }, { "epoch": 110.91807228915663, "grad_norm": 5.2440032958984375, "learning_rate": 4.447087378640777e-05, "loss": 0.1112, "step": 11439 }, { "epoch": 110.92771084337349, "grad_norm": 12.688708305358887, "learning_rate": 4.446601941747573e-05, "loss": 0.2043, "step": 11440 }, { "epoch": 110.93734939759037, "grad_norm": 7.235675811767578, "learning_rate": 4.4461165048543693e-05, "loss": 0.266, "step": 11441 }, { "epoch": 110.94698795180723, "grad_norm": 7.765335559844971, "learning_rate": 4.445631067961165e-05, "loss": 0.3186, "step": 11442 }, { "epoch": 110.9566265060241, "grad_norm": 5.533685207366943, "learning_rate": 4.445145631067961e-05, "loss": 0.1355, "step": 11443 }, { "epoch": 110.96626506024096, "grad_norm": 21.050804138183594, "learning_rate": 4.4446601941747573e-05, "loss": 0.192, "step": 11444 }, { "epoch": 110.97590361445783, "grad_norm": 6.826013565063477, "learning_rate": 4.444174757281553e-05, "loss": 0.1246, "step": 11445 }, { "epoch": 110.9855421686747, "grad_norm": 16.185585021972656, "learning_rate": 4.4436893203883496e-05, "loss": 0.2244, "step": 11446 }, { "epoch": 111.0012048192771, "grad_norm": 13.171528816223145, "learning_rate": 4.4432038834951453e-05, "loss": 0.4703, "step": 11447 }, { "epoch": 111.01084337349397, "grad_norm": 17.151630401611328, "learning_rate": 4.442718446601942e-05, "loss": 0.3667, "step": 11448 }, { "epoch": 111.02048192771085, "grad_norm": 8.735919952392578, "learning_rate": 4.442233009708738e-05, "loss": 0.1006, "step": 11449 }, { "epoch": 111.03012048192771, "grad_norm": 8.193463325500488, "learning_rate": 4.441747572815535e-05, "loss": 0.3525, "step": 11450 }, { "epoch": 111.03975903614457, "grad_norm": 18.857404708862305, "learning_rate": 4.4412621359223305e-05, "loss": 0.4582, "step": 11451 }, { "epoch": 111.04939759036145, "grad_norm": 16.23599624633789, "learning_rate": 4.440776699029127e-05, "loss": 0.188, "step": 11452 }, { "epoch": 111.05903614457831, "grad_norm": 19.628124237060547, "learning_rate": 4.440291262135923e-05, "loss": 0.3922, "step": 11453 }, { "epoch": 111.06867469879518, "grad_norm": 22.61017608642578, "learning_rate": 4.4398058252427185e-05, "loss": 0.1977, "step": 11454 }, { "epoch": 111.07831325301204, "grad_norm": 7.014334678649902, "learning_rate": 4.439320388349515e-05, "loss": 0.203, "step": 11455 }, { "epoch": 111.08795180722892, "grad_norm": 8.48859691619873, "learning_rate": 4.438834951456311e-05, "loss": 0.2981, "step": 11456 }, { "epoch": 111.09759036144578, "grad_norm": 12.148578643798828, "learning_rate": 4.438349514563107e-05, "loss": 0.1686, "step": 11457 }, { "epoch": 111.10722891566265, "grad_norm": 9.126479148864746, "learning_rate": 4.437864077669903e-05, "loss": 0.2887, "step": 11458 }, { "epoch": 111.11686746987952, "grad_norm": 12.799355506896973, "learning_rate": 4.4373786407766994e-05, "loss": 0.2086, "step": 11459 }, { "epoch": 111.12650602409639, "grad_norm": 43.50572967529297, "learning_rate": 4.436893203883495e-05, "loss": 0.3024, "step": 11460 }, { "epoch": 111.13614457831325, "grad_norm": 29.138134002685547, "learning_rate": 4.4364077669902916e-05, "loss": 0.2595, "step": 11461 }, { "epoch": 111.14578313253013, "grad_norm": 5.743225574493408, "learning_rate": 4.4359223300970874e-05, "loss": 0.1684, "step": 11462 }, { "epoch": 111.15542168674699, "grad_norm": 15.167994499206543, "learning_rate": 4.435436893203884e-05, "loss": 0.2288, "step": 11463 }, { "epoch": 111.16506024096385, "grad_norm": 32.74188995361328, "learning_rate": 4.4349514563106796e-05, "loss": 0.3009, "step": 11464 }, { "epoch": 111.17469879518072, "grad_norm": 9.076455116271973, "learning_rate": 4.434466019417476e-05, "loss": 0.2585, "step": 11465 }, { "epoch": 111.1843373493976, "grad_norm": 4.670811176300049, "learning_rate": 4.433980582524272e-05, "loss": 0.0999, "step": 11466 }, { "epoch": 111.19397590361446, "grad_norm": 6.010694980621338, "learning_rate": 4.433495145631068e-05, "loss": 0.1333, "step": 11467 }, { "epoch": 111.20361445783132, "grad_norm": 6.842982292175293, "learning_rate": 4.433009708737864e-05, "loss": 0.2921, "step": 11468 }, { "epoch": 111.2132530120482, "grad_norm": 7.35243034362793, "learning_rate": 4.4325242718446605e-05, "loss": 0.4031, "step": 11469 }, { "epoch": 111.22289156626506, "grad_norm": 8.64172649383545, "learning_rate": 4.432038834951456e-05, "loss": 0.2305, "step": 11470 }, { "epoch": 111.23253012048193, "grad_norm": 3.59721302986145, "learning_rate": 4.431553398058252e-05, "loss": 0.0837, "step": 11471 }, { "epoch": 111.24216867469879, "grad_norm": 6.502669334411621, "learning_rate": 4.4310679611650486e-05, "loss": 0.4297, "step": 11472 }, { "epoch": 111.25180722891567, "grad_norm": 10.358454704284668, "learning_rate": 4.430582524271844e-05, "loss": 0.2702, "step": 11473 }, { "epoch": 111.26144578313253, "grad_norm": 21.499969482421875, "learning_rate": 4.4300970873786415e-05, "loss": 0.3743, "step": 11474 }, { "epoch": 111.2710843373494, "grad_norm": 20.026145935058594, "learning_rate": 4.429611650485437e-05, "loss": 0.3674, "step": 11475 }, { "epoch": 111.28072289156627, "grad_norm": 12.104507446289062, "learning_rate": 4.429126213592234e-05, "loss": 0.371, "step": 11476 }, { "epoch": 111.29036144578313, "grad_norm": 4.436043739318848, "learning_rate": 4.4286407766990295e-05, "loss": 0.1968, "step": 11477 }, { "epoch": 111.3, "grad_norm": 12.546531677246094, "learning_rate": 4.428155339805826e-05, "loss": 0.2071, "step": 11478 }, { "epoch": 111.30963855421686, "grad_norm": 3.6614151000976562, "learning_rate": 4.427669902912622e-05, "loss": 0.2474, "step": 11479 }, { "epoch": 111.31927710843374, "grad_norm": 16.055614471435547, "learning_rate": 4.427184466019418e-05, "loss": 0.372, "step": 11480 }, { "epoch": 111.3289156626506, "grad_norm": 13.03945541381836, "learning_rate": 4.426699029126214e-05, "loss": 0.2035, "step": 11481 }, { "epoch": 111.33855421686746, "grad_norm": 20.961286544799805, "learning_rate": 4.42621359223301e-05, "loss": 0.2672, "step": 11482 }, { "epoch": 111.34819277108434, "grad_norm": 13.17386245727539, "learning_rate": 4.425728155339806e-05, "loss": 0.4163, "step": 11483 }, { "epoch": 111.3578313253012, "grad_norm": 16.05397605895996, "learning_rate": 4.425242718446602e-05, "loss": 0.3106, "step": 11484 }, { "epoch": 111.36746987951807, "grad_norm": 68.46527099609375, "learning_rate": 4.4247572815533984e-05, "loss": 0.3677, "step": 11485 }, { "epoch": 111.37710843373495, "grad_norm": 17.76902961730957, "learning_rate": 4.424271844660194e-05, "loss": 0.2525, "step": 11486 }, { "epoch": 111.38674698795181, "grad_norm": 10.846049308776855, "learning_rate": 4.4237864077669906e-05, "loss": 0.2861, "step": 11487 }, { "epoch": 111.39638554216867, "grad_norm": 12.879969596862793, "learning_rate": 4.4233009708737864e-05, "loss": 0.4056, "step": 11488 }, { "epoch": 111.40602409638554, "grad_norm": 13.291007995605469, "learning_rate": 4.422815533980583e-05, "loss": 0.2607, "step": 11489 }, { "epoch": 111.41566265060241, "grad_norm": 8.229225158691406, "learning_rate": 4.4223300970873786e-05, "loss": 0.1839, "step": 11490 }, { "epoch": 111.42530120481928, "grad_norm": 7.162106513977051, "learning_rate": 4.421844660194175e-05, "loss": 0.1753, "step": 11491 }, { "epoch": 111.43493975903614, "grad_norm": 14.252652168273926, "learning_rate": 4.421359223300971e-05, "loss": 0.3962, "step": 11492 }, { "epoch": 111.44457831325302, "grad_norm": 4.715782165527344, "learning_rate": 4.420873786407767e-05, "loss": 0.1645, "step": 11493 }, { "epoch": 111.45421686746988, "grad_norm": 7.53314733505249, "learning_rate": 4.420388349514563e-05, "loss": 0.369, "step": 11494 }, { "epoch": 111.46385542168674, "grad_norm": 4.4470930099487305, "learning_rate": 4.4199029126213595e-05, "loss": 0.1732, "step": 11495 }, { "epoch": 111.47349397590361, "grad_norm": 9.830524444580078, "learning_rate": 4.419417475728155e-05, "loss": 0.1509, "step": 11496 }, { "epoch": 111.48313253012049, "grad_norm": 16.202877044677734, "learning_rate": 4.418932038834952e-05, "loss": 0.3399, "step": 11497 }, { "epoch": 111.49277108433735, "grad_norm": 10.249755859375, "learning_rate": 4.4184466019417475e-05, "loss": 0.2375, "step": 11498 }, { "epoch": 111.50240963855421, "grad_norm": 9.071121215820312, "learning_rate": 4.417961165048543e-05, "loss": 0.2146, "step": 11499 }, { "epoch": 111.51204819277109, "grad_norm": 13.751313209533691, "learning_rate": 4.4174757281553404e-05, "loss": 0.3344, "step": 11500 }, { "epoch": 111.52168674698795, "grad_norm": 6.08500862121582, "learning_rate": 4.416990291262136e-05, "loss": 0.2159, "step": 11501 }, { "epoch": 111.53132530120482, "grad_norm": 8.286429405212402, "learning_rate": 4.416504854368933e-05, "loss": 0.2465, "step": 11502 }, { "epoch": 111.54096385542168, "grad_norm": 10.788642883300781, "learning_rate": 4.4160194174757284e-05, "loss": 0.2825, "step": 11503 }, { "epoch": 111.55060240963856, "grad_norm": 7.64822244644165, "learning_rate": 4.415533980582525e-05, "loss": 0.2073, "step": 11504 }, { "epoch": 111.56024096385542, "grad_norm": 7.961634635925293, "learning_rate": 4.415048543689321e-05, "loss": 0.211, "step": 11505 }, { "epoch": 111.56987951807228, "grad_norm": 11.464951515197754, "learning_rate": 4.414563106796117e-05, "loss": 0.2657, "step": 11506 }, { "epoch": 111.57951807228916, "grad_norm": 10.545488357543945, "learning_rate": 4.414077669902913e-05, "loss": 0.2373, "step": 11507 }, { "epoch": 111.58915662650602, "grad_norm": 8.847840309143066, "learning_rate": 4.4135922330097094e-05, "loss": 0.296, "step": 11508 }, { "epoch": 111.59879518072289, "grad_norm": 16.45406723022461, "learning_rate": 4.413106796116505e-05, "loss": 0.2779, "step": 11509 }, { "epoch": 111.60843373493977, "grad_norm": 7.138360977172852, "learning_rate": 4.412621359223301e-05, "loss": 0.2357, "step": 11510 }, { "epoch": 111.61807228915663, "grad_norm": 8.151233673095703, "learning_rate": 4.4121359223300974e-05, "loss": 0.2743, "step": 11511 }, { "epoch": 111.62771084337349, "grad_norm": 11.325064659118652, "learning_rate": 4.411650485436893e-05, "loss": 0.2372, "step": 11512 }, { "epoch": 111.63734939759036, "grad_norm": 8.0836820602417, "learning_rate": 4.4111650485436896e-05, "loss": 0.3087, "step": 11513 }, { "epoch": 111.64698795180723, "grad_norm": 6.280135154724121, "learning_rate": 4.4106796116504854e-05, "loss": 0.2044, "step": 11514 }, { "epoch": 111.6566265060241, "grad_norm": 15.551056861877441, "learning_rate": 4.410194174757282e-05, "loss": 0.3394, "step": 11515 }, { "epoch": 111.66626506024096, "grad_norm": 9.201738357543945, "learning_rate": 4.4097087378640776e-05, "loss": 0.2669, "step": 11516 }, { "epoch": 111.67590361445784, "grad_norm": 9.269996643066406, "learning_rate": 4.409223300970874e-05, "loss": 0.2469, "step": 11517 }, { "epoch": 111.6855421686747, "grad_norm": 5.062063217163086, "learning_rate": 4.40873786407767e-05, "loss": 0.1824, "step": 11518 }, { "epoch": 111.69518072289156, "grad_norm": 7.409211158752441, "learning_rate": 4.408252427184466e-05, "loss": 0.1925, "step": 11519 }, { "epoch": 111.70481927710843, "grad_norm": 14.729496002197266, "learning_rate": 4.407766990291262e-05, "loss": 0.2247, "step": 11520 }, { "epoch": 111.7144578313253, "grad_norm": 19.88703727722168, "learning_rate": 4.4072815533980585e-05, "loss": 0.4583, "step": 11521 }, { "epoch": 111.72409638554217, "grad_norm": 6.856152534484863, "learning_rate": 4.406796116504854e-05, "loss": 0.1917, "step": 11522 }, { "epoch": 111.73373493975903, "grad_norm": 13.67093276977539, "learning_rate": 4.406310679611651e-05, "loss": 0.4364, "step": 11523 }, { "epoch": 111.74337349397591, "grad_norm": 9.763291358947754, "learning_rate": 4.4058252427184465e-05, "loss": 0.2898, "step": 11524 }, { "epoch": 111.75301204819277, "grad_norm": 12.505522727966309, "learning_rate": 4.405339805825243e-05, "loss": 0.3106, "step": 11525 }, { "epoch": 111.76265060240964, "grad_norm": 17.2244930267334, "learning_rate": 4.4048543689320394e-05, "loss": 0.2178, "step": 11526 }, { "epoch": 111.7722891566265, "grad_norm": 30.971221923828125, "learning_rate": 4.404368932038835e-05, "loss": 0.3976, "step": 11527 }, { "epoch": 111.78192771084338, "grad_norm": 10.677275657653809, "learning_rate": 4.4038834951456316e-05, "loss": 0.333, "step": 11528 }, { "epoch": 111.79156626506024, "grad_norm": 9.019792556762695, "learning_rate": 4.4033980582524274e-05, "loss": 0.361, "step": 11529 }, { "epoch": 111.8012048192771, "grad_norm": 19.926498413085938, "learning_rate": 4.402912621359224e-05, "loss": 0.2359, "step": 11530 }, { "epoch": 111.81084337349398, "grad_norm": 8.966336250305176, "learning_rate": 4.4024271844660197e-05, "loss": 0.1879, "step": 11531 }, { "epoch": 111.82048192771084, "grad_norm": 8.714383125305176, "learning_rate": 4.401941747572816e-05, "loss": 0.1723, "step": 11532 }, { "epoch": 111.83012048192771, "grad_norm": 33.1547966003418, "learning_rate": 4.401456310679612e-05, "loss": 0.3611, "step": 11533 }, { "epoch": 111.83975903614459, "grad_norm": 12.738967895507812, "learning_rate": 4.400970873786408e-05, "loss": 0.3498, "step": 11534 }, { "epoch": 111.84939759036145, "grad_norm": 10.8787841796875, "learning_rate": 4.400485436893204e-05, "loss": 0.3269, "step": 11535 }, { "epoch": 111.85903614457831, "grad_norm": 13.485841751098633, "learning_rate": 4.4000000000000006e-05, "loss": 0.3423, "step": 11536 }, { "epoch": 111.86867469879518, "grad_norm": 16.65818214416504, "learning_rate": 4.3995145631067963e-05, "loss": 0.2081, "step": 11537 }, { "epoch": 111.87831325301205, "grad_norm": 9.34505558013916, "learning_rate": 4.399029126213592e-05, "loss": 0.3017, "step": 11538 }, { "epoch": 111.88795180722892, "grad_norm": 12.396233558654785, "learning_rate": 4.3985436893203886e-05, "loss": 0.2611, "step": 11539 }, { "epoch": 111.89759036144578, "grad_norm": 18.63768196105957, "learning_rate": 4.3980582524271843e-05, "loss": 0.3004, "step": 11540 }, { "epoch": 111.90722891566266, "grad_norm": 33.380279541015625, "learning_rate": 4.397572815533981e-05, "loss": 0.4165, "step": 11541 }, { "epoch": 111.91686746987952, "grad_norm": 8.508439064025879, "learning_rate": 4.3970873786407766e-05, "loss": 0.1986, "step": 11542 }, { "epoch": 111.92650602409638, "grad_norm": 9.131903648376465, "learning_rate": 4.396601941747573e-05, "loss": 0.338, "step": 11543 }, { "epoch": 111.93614457831325, "grad_norm": 6.138891220092773, "learning_rate": 4.396116504854369e-05, "loss": 0.1309, "step": 11544 }, { "epoch": 111.94578313253012, "grad_norm": 11.154889106750488, "learning_rate": 4.395631067961165e-05, "loss": 0.331, "step": 11545 }, { "epoch": 111.95542168674699, "grad_norm": 15.770403861999512, "learning_rate": 4.395145631067961e-05, "loss": 0.4056, "step": 11546 }, { "epoch": 111.96506024096385, "grad_norm": 22.26439666748047, "learning_rate": 4.3946601941747575e-05, "loss": 0.5067, "step": 11547 }, { "epoch": 111.97469879518073, "grad_norm": 7.728394031524658, "learning_rate": 4.394174757281553e-05, "loss": 0.2741, "step": 11548 }, { "epoch": 111.98433734939759, "grad_norm": 7.513891696929932, "learning_rate": 4.39368932038835e-05, "loss": 0.1815, "step": 11549 }, { "epoch": 111.99397590361446, "grad_norm": 8.23629093170166, "learning_rate": 4.393203883495146e-05, "loss": 0.2311, "step": 11550 }, { "epoch": 112.00963855421686, "grad_norm": 20.47525405883789, "learning_rate": 4.392718446601942e-05, "loss": 0.357, "step": 11551 }, { "epoch": 112.01927710843374, "grad_norm": 14.760950088500977, "learning_rate": 4.3922330097087384e-05, "loss": 0.3438, "step": 11552 }, { "epoch": 112.0289156626506, "grad_norm": 7.131338119506836, "learning_rate": 4.391747572815534e-05, "loss": 0.1441, "step": 11553 }, { "epoch": 112.03855421686747, "grad_norm": 12.317309379577637, "learning_rate": 4.3912621359223306e-05, "loss": 0.363, "step": 11554 }, { "epoch": 112.04819277108433, "grad_norm": 10.476770401000977, "learning_rate": 4.3907766990291264e-05, "loss": 0.3233, "step": 11555 }, { "epoch": 112.05783132530121, "grad_norm": 12.370573043823242, "learning_rate": 4.390291262135923e-05, "loss": 0.271, "step": 11556 }, { "epoch": 112.06746987951807, "grad_norm": 13.482396125793457, "learning_rate": 4.3898058252427186e-05, "loss": 0.4482, "step": 11557 }, { "epoch": 112.07710843373494, "grad_norm": 12.244770050048828, "learning_rate": 4.389320388349515e-05, "loss": 0.2125, "step": 11558 }, { "epoch": 112.08674698795181, "grad_norm": 9.456927299499512, "learning_rate": 4.388834951456311e-05, "loss": 0.2822, "step": 11559 }, { "epoch": 112.09638554216868, "grad_norm": 11.081857681274414, "learning_rate": 4.388349514563107e-05, "loss": 0.3641, "step": 11560 }, { "epoch": 112.10602409638554, "grad_norm": 18.80219268798828, "learning_rate": 4.387864077669903e-05, "loss": 0.432, "step": 11561 }, { "epoch": 112.1156626506024, "grad_norm": 9.472981452941895, "learning_rate": 4.3873786407766995e-05, "loss": 0.2221, "step": 11562 }, { "epoch": 112.12530120481928, "grad_norm": 9.848725318908691, "learning_rate": 4.386893203883495e-05, "loss": 0.3494, "step": 11563 }, { "epoch": 112.13493975903614, "grad_norm": 18.617555618286133, "learning_rate": 4.386407766990292e-05, "loss": 0.3544, "step": 11564 }, { "epoch": 112.144578313253, "grad_norm": 7.1050705909729, "learning_rate": 4.3859223300970875e-05, "loss": 0.1635, "step": 11565 }, { "epoch": 112.15421686746988, "grad_norm": 12.80691909790039, "learning_rate": 4.385436893203883e-05, "loss": 0.4004, "step": 11566 }, { "epoch": 112.16385542168675, "grad_norm": 9.950650215148926, "learning_rate": 4.38495145631068e-05, "loss": 0.2498, "step": 11567 }, { "epoch": 112.17349397590361, "grad_norm": 8.872031211853027, "learning_rate": 4.3844660194174756e-05, "loss": 0.1573, "step": 11568 }, { "epoch": 112.18313253012049, "grad_norm": 12.636544227600098, "learning_rate": 4.383980582524272e-05, "loss": 0.3829, "step": 11569 }, { "epoch": 112.19277108433735, "grad_norm": 15.313812255859375, "learning_rate": 4.383495145631068e-05, "loss": 0.2796, "step": 11570 }, { "epoch": 112.20240963855422, "grad_norm": 7.171340465545654, "learning_rate": 4.383009708737864e-05, "loss": 0.1853, "step": 11571 }, { "epoch": 112.21204819277108, "grad_norm": 10.115686416625977, "learning_rate": 4.38252427184466e-05, "loss": 0.3666, "step": 11572 }, { "epoch": 112.22168674698796, "grad_norm": 12.900858879089355, "learning_rate": 4.3820388349514565e-05, "loss": 0.2601, "step": 11573 }, { "epoch": 112.23132530120482, "grad_norm": 9.539429664611816, "learning_rate": 4.381553398058252e-05, "loss": 0.3144, "step": 11574 }, { "epoch": 112.24096385542168, "grad_norm": 16.262269973754883, "learning_rate": 4.381067961165049e-05, "loss": 0.3888, "step": 11575 }, { "epoch": 112.25060240963856, "grad_norm": 13.397838592529297, "learning_rate": 4.380582524271845e-05, "loss": 0.405, "step": 11576 }, { "epoch": 112.26024096385542, "grad_norm": 8.399374961853027, "learning_rate": 4.380097087378641e-05, "loss": 0.1781, "step": 11577 }, { "epoch": 112.26987951807229, "grad_norm": 11.671935081481934, "learning_rate": 4.3796116504854374e-05, "loss": 0.3231, "step": 11578 }, { "epoch": 112.27951807228915, "grad_norm": 11.6420259475708, "learning_rate": 4.379126213592233e-05, "loss": 0.4043, "step": 11579 }, { "epoch": 112.28915662650603, "grad_norm": 5.5179338455200195, "learning_rate": 4.3786407766990296e-05, "loss": 0.1761, "step": 11580 }, { "epoch": 112.29879518072289, "grad_norm": 10.11072063446045, "learning_rate": 4.3781553398058254e-05, "loss": 0.3325, "step": 11581 }, { "epoch": 112.30843373493975, "grad_norm": 11.326247215270996, "learning_rate": 4.377669902912622e-05, "loss": 0.3962, "step": 11582 }, { "epoch": 112.31807228915663, "grad_norm": 8.606353759765625, "learning_rate": 4.3771844660194176e-05, "loss": 0.4308, "step": 11583 }, { "epoch": 112.3277108433735, "grad_norm": 4.9340925216674805, "learning_rate": 4.376699029126214e-05, "loss": 0.2332, "step": 11584 }, { "epoch": 112.33734939759036, "grad_norm": 8.506102561950684, "learning_rate": 4.37621359223301e-05, "loss": 0.3561, "step": 11585 }, { "epoch": 112.34698795180722, "grad_norm": 17.009145736694336, "learning_rate": 4.375728155339806e-05, "loss": 0.2605, "step": 11586 }, { "epoch": 112.3566265060241, "grad_norm": 4.792527198791504, "learning_rate": 4.375242718446602e-05, "loss": 0.1822, "step": 11587 }, { "epoch": 112.36626506024096, "grad_norm": 14.505626678466797, "learning_rate": 4.3747572815533985e-05, "loss": 0.3451, "step": 11588 }, { "epoch": 112.37590361445783, "grad_norm": 7.125392436981201, "learning_rate": 4.374271844660194e-05, "loss": 0.1936, "step": 11589 }, { "epoch": 112.3855421686747, "grad_norm": 10.111580848693848, "learning_rate": 4.373786407766991e-05, "loss": 0.5045, "step": 11590 }, { "epoch": 112.39518072289157, "grad_norm": 6.137942790985107, "learning_rate": 4.3733009708737865e-05, "loss": 0.1872, "step": 11591 }, { "epoch": 112.40481927710843, "grad_norm": 12.808159828186035, "learning_rate": 4.372815533980582e-05, "loss": 0.2938, "step": 11592 }, { "epoch": 112.41445783132531, "grad_norm": 9.314494132995605, "learning_rate": 4.372330097087379e-05, "loss": 0.1616, "step": 11593 }, { "epoch": 112.42409638554217, "grad_norm": 10.205408096313477, "learning_rate": 4.3718446601941745e-05, "loss": 0.2704, "step": 11594 }, { "epoch": 112.43373493975903, "grad_norm": 10.652618408203125, "learning_rate": 4.371359223300971e-05, "loss": 0.3868, "step": 11595 }, { "epoch": 112.4433734939759, "grad_norm": 15.14078140258789, "learning_rate": 4.370873786407767e-05, "loss": 0.2188, "step": 11596 }, { "epoch": 112.45301204819278, "grad_norm": 17.3280029296875, "learning_rate": 4.370388349514563e-05, "loss": 0.1064, "step": 11597 }, { "epoch": 112.46265060240964, "grad_norm": 13.106956481933594, "learning_rate": 4.369902912621359e-05, "loss": 0.3778, "step": 11598 }, { "epoch": 112.4722891566265, "grad_norm": 7.40949010848999, "learning_rate": 4.3694174757281554e-05, "loss": 0.3112, "step": 11599 }, { "epoch": 112.48192771084338, "grad_norm": 6.614086151123047, "learning_rate": 4.368932038834951e-05, "loss": 0.3051, "step": 11600 }, { "epoch": 112.49156626506024, "grad_norm": 6.617210388183594, "learning_rate": 4.3684466019417483e-05, "loss": 0.1822, "step": 11601 }, { "epoch": 112.5012048192771, "grad_norm": 8.769858360290527, "learning_rate": 4.367961165048544e-05, "loss": 0.3393, "step": 11602 }, { "epoch": 112.51084337349397, "grad_norm": 7.234729766845703, "learning_rate": 4.36747572815534e-05, "loss": 0.1783, "step": 11603 }, { "epoch": 112.52048192771085, "grad_norm": 12.864740371704102, "learning_rate": 4.3669902912621364e-05, "loss": 0.2753, "step": 11604 }, { "epoch": 112.53012048192771, "grad_norm": 19.218997955322266, "learning_rate": 4.366504854368932e-05, "loss": 0.1856, "step": 11605 }, { "epoch": 112.53975903614457, "grad_norm": 7.153847694396973, "learning_rate": 4.3660194174757286e-05, "loss": 0.1872, "step": 11606 }, { "epoch": 112.54939759036145, "grad_norm": 4.387037754058838, "learning_rate": 4.3655339805825244e-05, "loss": 0.1953, "step": 11607 }, { "epoch": 112.55903614457831, "grad_norm": 8.835776329040527, "learning_rate": 4.365048543689321e-05, "loss": 0.2555, "step": 11608 }, { "epoch": 112.56867469879518, "grad_norm": 12.190001487731934, "learning_rate": 4.3645631067961166e-05, "loss": 0.4244, "step": 11609 }, { "epoch": 112.57831325301204, "grad_norm": 17.247934341430664, "learning_rate": 4.364077669902913e-05, "loss": 0.5825, "step": 11610 }, { "epoch": 112.58795180722892, "grad_norm": 15.218704223632812, "learning_rate": 4.363592233009709e-05, "loss": 0.304, "step": 11611 }, { "epoch": 112.59759036144578, "grad_norm": 8.813716888427734, "learning_rate": 4.363106796116505e-05, "loss": 0.163, "step": 11612 }, { "epoch": 112.60722891566265, "grad_norm": 20.364194869995117, "learning_rate": 4.362621359223301e-05, "loss": 0.2989, "step": 11613 }, { "epoch": 112.61686746987952, "grad_norm": 17.664485931396484, "learning_rate": 4.3621359223300975e-05, "loss": 0.2079, "step": 11614 }, { "epoch": 112.62650602409639, "grad_norm": 10.286751747131348, "learning_rate": 4.361650485436893e-05, "loss": 0.2417, "step": 11615 }, { "epoch": 112.63614457831325, "grad_norm": 8.27118968963623, "learning_rate": 4.36116504854369e-05, "loss": 0.2767, "step": 11616 }, { "epoch": 112.64578313253013, "grad_norm": 8.307668685913086, "learning_rate": 4.3606796116504855e-05, "loss": 0.2337, "step": 11617 }, { "epoch": 112.65542168674699, "grad_norm": 5.447085380554199, "learning_rate": 4.360194174757282e-05, "loss": 0.2508, "step": 11618 }, { "epoch": 112.66506024096385, "grad_norm": 6.537261009216309, "learning_rate": 4.359708737864078e-05, "loss": 0.134, "step": 11619 }, { "epoch": 112.67469879518072, "grad_norm": 3.87377667427063, "learning_rate": 4.3592233009708735e-05, "loss": 0.061, "step": 11620 }, { "epoch": 112.6843373493976, "grad_norm": 10.674099922180176, "learning_rate": 4.35873786407767e-05, "loss": 0.3074, "step": 11621 }, { "epoch": 112.69397590361446, "grad_norm": 6.325625896453857, "learning_rate": 4.358252427184466e-05, "loss": 0.244, "step": 11622 }, { "epoch": 112.70361445783132, "grad_norm": 25.588518142700195, "learning_rate": 4.357766990291262e-05, "loss": 0.327, "step": 11623 }, { "epoch": 112.7132530120482, "grad_norm": 9.492382049560547, "learning_rate": 4.357281553398058e-05, "loss": 0.3237, "step": 11624 }, { "epoch": 112.72289156626506, "grad_norm": 5.940526485443115, "learning_rate": 4.3567961165048544e-05, "loss": 0.2144, "step": 11625 }, { "epoch": 112.73253012048193, "grad_norm": 9.370259284973145, "learning_rate": 4.356310679611651e-05, "loss": 0.2564, "step": 11626 }, { "epoch": 112.74216867469879, "grad_norm": 14.785262107849121, "learning_rate": 4.355825242718447e-05, "loss": 0.4109, "step": 11627 }, { "epoch": 112.75180722891567, "grad_norm": 10.223626136779785, "learning_rate": 4.355339805825243e-05, "loss": 0.3222, "step": 11628 }, { "epoch": 112.76144578313253, "grad_norm": 6.376702308654785, "learning_rate": 4.3548543689320396e-05, "loss": 0.1978, "step": 11629 }, { "epoch": 112.7710843373494, "grad_norm": 18.11139488220215, "learning_rate": 4.354368932038835e-05, "loss": 0.2865, "step": 11630 }, { "epoch": 112.78072289156627, "grad_norm": 7.200881481170654, "learning_rate": 4.353883495145631e-05, "loss": 0.1439, "step": 11631 }, { "epoch": 112.79036144578313, "grad_norm": 4.124243259429932, "learning_rate": 4.3533980582524276e-05, "loss": 0.0679, "step": 11632 }, { "epoch": 112.8, "grad_norm": 8.958452224731445, "learning_rate": 4.352912621359223e-05, "loss": 0.1959, "step": 11633 }, { "epoch": 112.80963855421686, "grad_norm": 7.953498363494873, "learning_rate": 4.35242718446602e-05, "loss": 0.2194, "step": 11634 }, { "epoch": 112.81927710843374, "grad_norm": 8.172783851623535, "learning_rate": 4.3519417475728156e-05, "loss": 0.3113, "step": 11635 }, { "epoch": 112.8289156626506, "grad_norm": 20.091421127319336, "learning_rate": 4.351456310679612e-05, "loss": 0.4089, "step": 11636 }, { "epoch": 112.83855421686746, "grad_norm": 16.925979614257812, "learning_rate": 4.350970873786408e-05, "loss": 0.2171, "step": 11637 }, { "epoch": 112.84819277108434, "grad_norm": 13.546270370483398, "learning_rate": 4.350485436893204e-05, "loss": 0.4896, "step": 11638 }, { "epoch": 112.8578313253012, "grad_norm": 8.999610900878906, "learning_rate": 4.35e-05, "loss": 0.2425, "step": 11639 }, { "epoch": 112.86746987951807, "grad_norm": 10.12397289276123, "learning_rate": 4.3495145631067965e-05, "loss": 0.2179, "step": 11640 }, { "epoch": 112.87710843373495, "grad_norm": 8.088239669799805, "learning_rate": 4.349029126213592e-05, "loss": 0.246, "step": 11641 }, { "epoch": 112.88674698795181, "grad_norm": 8.89258098602295, "learning_rate": 4.348543689320389e-05, "loss": 0.2943, "step": 11642 }, { "epoch": 112.89638554216867, "grad_norm": 17.808149337768555, "learning_rate": 4.3480582524271845e-05, "loss": 0.3037, "step": 11643 }, { "epoch": 112.90602409638554, "grad_norm": 15.068788528442383, "learning_rate": 4.347572815533981e-05, "loss": 0.2259, "step": 11644 }, { "epoch": 112.91566265060241, "grad_norm": 10.008804321289062, "learning_rate": 4.347087378640777e-05, "loss": 0.19, "step": 11645 }, { "epoch": 112.92530120481928, "grad_norm": 5.935588836669922, "learning_rate": 4.346601941747573e-05, "loss": 0.114, "step": 11646 }, { "epoch": 112.93493975903614, "grad_norm": 15.514581680297852, "learning_rate": 4.346116504854369e-05, "loss": 0.2907, "step": 11647 }, { "epoch": 112.94457831325302, "grad_norm": 11.876873016357422, "learning_rate": 4.345631067961165e-05, "loss": 0.4056, "step": 11648 }, { "epoch": 112.95421686746988, "grad_norm": 7.84898042678833, "learning_rate": 4.345145631067961e-05, "loss": 0.2362, "step": 11649 }, { "epoch": 112.96385542168674, "grad_norm": 8.229537963867188, "learning_rate": 4.344660194174757e-05, "loss": 0.2485, "step": 11650 }, { "epoch": 112.97349397590361, "grad_norm": 5.399066925048828, "learning_rate": 4.344174757281554e-05, "loss": 0.1543, "step": 11651 }, { "epoch": 112.98313253012049, "grad_norm": 6.134361743927002, "learning_rate": 4.34368932038835e-05, "loss": 0.1752, "step": 11652 }, { "epoch": 112.99277108433735, "grad_norm": 29.425159454345703, "learning_rate": 4.343203883495146e-05, "loss": 0.1505, "step": 11653 }, { "epoch": 113.00843373493976, "grad_norm": 15.49980354309082, "learning_rate": 4.342718446601942e-05, "loss": 0.3476, "step": 11654 }, { "epoch": 113.01807228915662, "grad_norm": 28.614526748657227, "learning_rate": 4.3422330097087385e-05, "loss": 0.2512, "step": 11655 }, { "epoch": 113.0277108433735, "grad_norm": 37.26686477661133, "learning_rate": 4.341747572815534e-05, "loss": 0.2713, "step": 11656 }, { "epoch": 113.03734939759036, "grad_norm": 16.899019241333008, "learning_rate": 4.341262135922331e-05, "loss": 0.1261, "step": 11657 }, { "epoch": 113.04698795180722, "grad_norm": 5.619399070739746, "learning_rate": 4.3407766990291265e-05, "loss": 0.1778, "step": 11658 }, { "epoch": 113.0566265060241, "grad_norm": 16.201313018798828, "learning_rate": 4.340291262135922e-05, "loss": 0.134, "step": 11659 }, { "epoch": 113.06626506024097, "grad_norm": 18.981504440307617, "learning_rate": 4.339805825242719e-05, "loss": 0.1612, "step": 11660 }, { "epoch": 113.07590361445783, "grad_norm": 6.329145908355713, "learning_rate": 4.3393203883495145e-05, "loss": 0.3416, "step": 11661 }, { "epoch": 113.08554216867469, "grad_norm": 30.9775390625, "learning_rate": 4.338834951456311e-05, "loss": 0.1322, "step": 11662 }, { "epoch": 113.09518072289157, "grad_norm": 41.75481414794922, "learning_rate": 4.338349514563107e-05, "loss": 0.2789, "step": 11663 }, { "epoch": 113.10481927710843, "grad_norm": 12.002090454101562, "learning_rate": 4.337864077669903e-05, "loss": 0.4814, "step": 11664 }, { "epoch": 113.1144578313253, "grad_norm": 5.345730781555176, "learning_rate": 4.337378640776699e-05, "loss": 0.1287, "step": 11665 }, { "epoch": 113.12409638554217, "grad_norm": 48.1765251159668, "learning_rate": 4.3368932038834955e-05, "loss": 0.4087, "step": 11666 }, { "epoch": 113.13373493975904, "grad_norm": 57.75556564331055, "learning_rate": 4.336407766990291e-05, "loss": 0.3907, "step": 11667 }, { "epoch": 113.1433734939759, "grad_norm": 21.314800262451172, "learning_rate": 4.335922330097088e-05, "loss": 0.1541, "step": 11668 }, { "epoch": 113.15301204819278, "grad_norm": 18.774749755859375, "learning_rate": 4.3354368932038835e-05, "loss": 0.2532, "step": 11669 }, { "epoch": 113.16265060240964, "grad_norm": 9.088595390319824, "learning_rate": 4.33495145631068e-05, "loss": 0.3009, "step": 11670 }, { "epoch": 113.1722891566265, "grad_norm": 11.58914566040039, "learning_rate": 4.334466019417476e-05, "loss": 0.4593, "step": 11671 }, { "epoch": 113.18192771084337, "grad_norm": 24.460811614990234, "learning_rate": 4.333980582524272e-05, "loss": 0.2005, "step": 11672 }, { "epoch": 113.19156626506025, "grad_norm": 42.56460952758789, "learning_rate": 4.333495145631068e-05, "loss": 0.2143, "step": 11673 }, { "epoch": 113.20120481927711, "grad_norm": 20.260042190551758, "learning_rate": 4.3330097087378644e-05, "loss": 0.2605, "step": 11674 }, { "epoch": 113.21084337349397, "grad_norm": 3.441039800643921, "learning_rate": 4.33252427184466e-05, "loss": 0.2148, "step": 11675 }, { "epoch": 113.22048192771085, "grad_norm": 16.059730529785156, "learning_rate": 4.3320388349514566e-05, "loss": 0.2271, "step": 11676 }, { "epoch": 113.23012048192771, "grad_norm": 23.297277450561523, "learning_rate": 4.331553398058253e-05, "loss": 0.2349, "step": 11677 }, { "epoch": 113.23975903614458, "grad_norm": 6.220603942871094, "learning_rate": 4.331067961165049e-05, "loss": 0.1557, "step": 11678 }, { "epoch": 113.24939759036144, "grad_norm": 5.541561126708984, "learning_rate": 4.330582524271845e-05, "loss": 0.2601, "step": 11679 }, { "epoch": 113.25903614457832, "grad_norm": 17.101545333862305, "learning_rate": 4.330097087378641e-05, "loss": 0.162, "step": 11680 }, { "epoch": 113.26867469879518, "grad_norm": 5.069579124450684, "learning_rate": 4.3296116504854375e-05, "loss": 0.3421, "step": 11681 }, { "epoch": 113.27831325301204, "grad_norm": 20.52886962890625, "learning_rate": 4.329126213592233e-05, "loss": 0.199, "step": 11682 }, { "epoch": 113.28795180722892, "grad_norm": 13.854267120361328, "learning_rate": 4.32864077669903e-05, "loss": 0.3582, "step": 11683 }, { "epoch": 113.29759036144578, "grad_norm": 7.7356109619140625, "learning_rate": 4.3281553398058255e-05, "loss": 0.2434, "step": 11684 }, { "epoch": 113.30722891566265, "grad_norm": 12.470301628112793, "learning_rate": 4.327669902912622e-05, "loss": 0.4307, "step": 11685 }, { "epoch": 113.31686746987951, "grad_norm": 13.835786819458008, "learning_rate": 4.327184466019418e-05, "loss": 0.2581, "step": 11686 }, { "epoch": 113.32650602409639, "grad_norm": 13.876635551452637, "learning_rate": 4.3266990291262135e-05, "loss": 0.2804, "step": 11687 }, { "epoch": 113.33614457831325, "grad_norm": 14.0896577835083, "learning_rate": 4.32621359223301e-05, "loss": 0.3904, "step": 11688 }, { "epoch": 113.34578313253012, "grad_norm": 8.980819702148438, "learning_rate": 4.325728155339806e-05, "loss": 0.2934, "step": 11689 }, { "epoch": 113.355421686747, "grad_norm": 1.4058983325958252, "learning_rate": 4.325242718446602e-05, "loss": 0.5132, "step": 11690 }, { "epoch": 113.36506024096386, "grad_norm": 7.864785194396973, "learning_rate": 4.324757281553398e-05, "loss": 0.1616, "step": 11691 }, { "epoch": 113.37469879518072, "grad_norm": 21.46564483642578, "learning_rate": 4.3242718446601944e-05, "loss": 0.2858, "step": 11692 }, { "epoch": 113.38433734939758, "grad_norm": 44.17003631591797, "learning_rate": 4.32378640776699e-05, "loss": 0.1537, "step": 11693 }, { "epoch": 113.39397590361446, "grad_norm": 23.11552619934082, "learning_rate": 4.323300970873787e-05, "loss": 0.1032, "step": 11694 }, { "epoch": 113.40361445783132, "grad_norm": 51.383056640625, "learning_rate": 4.3228155339805824e-05, "loss": 0.3127, "step": 11695 }, { "epoch": 113.41325301204819, "grad_norm": 37.09140396118164, "learning_rate": 4.322330097087379e-05, "loss": 0.3009, "step": 11696 }, { "epoch": 113.42289156626506, "grad_norm": 5.977446556091309, "learning_rate": 4.321844660194175e-05, "loss": 0.0939, "step": 11697 }, { "epoch": 113.43253012048193, "grad_norm": 29.549745559692383, "learning_rate": 4.321359223300971e-05, "loss": 0.427, "step": 11698 }, { "epoch": 113.44216867469879, "grad_norm": 25.230993270874023, "learning_rate": 4.320873786407767e-05, "loss": 0.2151, "step": 11699 }, { "epoch": 113.45180722891567, "grad_norm": 23.393266677856445, "learning_rate": 4.3203883495145634e-05, "loss": 0.3022, "step": 11700 }, { "epoch": 113.46144578313253, "grad_norm": 12.741997718811035, "learning_rate": 4.319902912621359e-05, "loss": 0.1761, "step": 11701 }, { "epoch": 113.4710843373494, "grad_norm": 30.71232795715332, "learning_rate": 4.3194174757281556e-05, "loss": 0.2865, "step": 11702 }, { "epoch": 113.48072289156626, "grad_norm": 5.2498040199279785, "learning_rate": 4.318932038834952e-05, "loss": 0.1018, "step": 11703 }, { "epoch": 113.49036144578314, "grad_norm": 73.23394012451172, "learning_rate": 4.318446601941748e-05, "loss": 0.2777, "step": 11704 }, { "epoch": 113.5, "grad_norm": 13.415728569030762, "learning_rate": 4.317961165048544e-05, "loss": 0.0807, "step": 11705 }, { "epoch": 113.50963855421686, "grad_norm": 52.848777770996094, "learning_rate": 4.31747572815534e-05, "loss": 0.2994, "step": 11706 }, { "epoch": 113.51927710843374, "grad_norm": 10.161457061767578, "learning_rate": 4.3169902912621365e-05, "loss": 0.4468, "step": 11707 }, { "epoch": 113.5289156626506, "grad_norm": 12.681513786315918, "learning_rate": 4.316504854368932e-05, "loss": 0.2884, "step": 11708 }, { "epoch": 113.53855421686747, "grad_norm": 15.969338417053223, "learning_rate": 4.316019417475729e-05, "loss": 0.1443, "step": 11709 }, { "epoch": 113.54819277108433, "grad_norm": 11.460392951965332, "learning_rate": 4.3155339805825245e-05, "loss": 0.4295, "step": 11710 }, { "epoch": 113.55783132530121, "grad_norm": 16.1940975189209, "learning_rate": 4.315048543689321e-05, "loss": 0.1605, "step": 11711 }, { "epoch": 113.56746987951807, "grad_norm": 28.87969970703125, "learning_rate": 4.314563106796117e-05, "loss": 0.2666, "step": 11712 }, { "epoch": 113.57710843373494, "grad_norm": 19.303573608398438, "learning_rate": 4.314077669902913e-05, "loss": 0.428, "step": 11713 }, { "epoch": 113.58674698795181, "grad_norm": 36.55034637451172, "learning_rate": 4.313592233009709e-05, "loss": 0.3006, "step": 11714 }, { "epoch": 113.59638554216868, "grad_norm": 11.916744232177734, "learning_rate": 4.313106796116505e-05, "loss": 0.1884, "step": 11715 }, { "epoch": 113.60602409638554, "grad_norm": 3.9449360370635986, "learning_rate": 4.312621359223301e-05, "loss": 0.2227, "step": 11716 }, { "epoch": 113.61566265060242, "grad_norm": 8.916404724121094, "learning_rate": 4.312135922330097e-05, "loss": 0.195, "step": 11717 }, { "epoch": 113.62530120481928, "grad_norm": 28.680503845214844, "learning_rate": 4.3116504854368934e-05, "loss": 0.2958, "step": 11718 }, { "epoch": 113.63493975903614, "grad_norm": 9.209099769592285, "learning_rate": 4.311165048543689e-05, "loss": 0.2541, "step": 11719 }, { "epoch": 113.644578313253, "grad_norm": 21.010377883911133, "learning_rate": 4.3106796116504856e-05, "loss": 0.3879, "step": 11720 }, { "epoch": 113.65421686746988, "grad_norm": 28.83810806274414, "learning_rate": 4.3101941747572814e-05, "loss": 0.2826, "step": 11721 }, { "epoch": 113.66385542168675, "grad_norm": 17.842397689819336, "learning_rate": 4.309708737864078e-05, "loss": 0.3236, "step": 11722 }, { "epoch": 113.67349397590361, "grad_norm": 21.167007446289062, "learning_rate": 4.3092233009708736e-05, "loss": 0.0737, "step": 11723 }, { "epoch": 113.68313253012049, "grad_norm": 12.448017120361328, "learning_rate": 4.30873786407767e-05, "loss": 0.2906, "step": 11724 }, { "epoch": 113.69277108433735, "grad_norm": 15.709203720092773, "learning_rate": 4.308252427184466e-05, "loss": 0.3929, "step": 11725 }, { "epoch": 113.70240963855422, "grad_norm": 38.6434211730957, "learning_rate": 4.307766990291262e-05, "loss": 0.3531, "step": 11726 }, { "epoch": 113.71204819277108, "grad_norm": 15.37885856628418, "learning_rate": 4.307281553398059e-05, "loss": 0.2107, "step": 11727 }, { "epoch": 113.72168674698796, "grad_norm": 16.51287841796875, "learning_rate": 4.3067961165048546e-05, "loss": 0.2718, "step": 11728 }, { "epoch": 113.73132530120482, "grad_norm": 12.551518440246582, "learning_rate": 4.306310679611651e-05, "loss": 0.198, "step": 11729 }, { "epoch": 113.74096385542168, "grad_norm": 28.997846603393555, "learning_rate": 4.305825242718447e-05, "loss": 0.327, "step": 11730 }, { "epoch": 113.75060240963856, "grad_norm": 33.158958435058594, "learning_rate": 4.305339805825243e-05, "loss": 0.2479, "step": 11731 }, { "epoch": 113.76024096385542, "grad_norm": 47.78834533691406, "learning_rate": 4.304854368932039e-05, "loss": 0.4832, "step": 11732 }, { "epoch": 113.76987951807229, "grad_norm": 16.905542373657227, "learning_rate": 4.3043689320388355e-05, "loss": 0.4306, "step": 11733 }, { "epoch": 113.77951807228915, "grad_norm": 7.036468982696533, "learning_rate": 4.303883495145631e-05, "loss": 0.2812, "step": 11734 }, { "epoch": 113.78915662650603, "grad_norm": 24.72568702697754, "learning_rate": 4.303398058252428e-05, "loss": 0.1819, "step": 11735 }, { "epoch": 113.79879518072289, "grad_norm": 18.78890609741211, "learning_rate": 4.3029126213592235e-05, "loss": 0.2061, "step": 11736 }, { "epoch": 113.80843373493975, "grad_norm": 11.751331329345703, "learning_rate": 4.30242718446602e-05, "loss": 0.2659, "step": 11737 }, { "epoch": 113.81807228915663, "grad_norm": 21.880138397216797, "learning_rate": 4.301941747572816e-05, "loss": 0.3581, "step": 11738 }, { "epoch": 113.8277108433735, "grad_norm": 39.73047637939453, "learning_rate": 4.301456310679612e-05, "loss": 0.3127, "step": 11739 }, { "epoch": 113.83734939759036, "grad_norm": 21.020986557006836, "learning_rate": 4.300970873786408e-05, "loss": 0.3943, "step": 11740 }, { "epoch": 113.84698795180722, "grad_norm": 5.521303176879883, "learning_rate": 4.300485436893204e-05, "loss": 0.2153, "step": 11741 }, { "epoch": 113.8566265060241, "grad_norm": 21.705238342285156, "learning_rate": 4.3e-05, "loss": 0.2355, "step": 11742 }, { "epoch": 113.86626506024096, "grad_norm": 53.0660285949707, "learning_rate": 4.299514563106796e-05, "loss": 0.275, "step": 11743 }, { "epoch": 113.87590361445783, "grad_norm": 28.274974822998047, "learning_rate": 4.2990291262135924e-05, "loss": 0.1728, "step": 11744 }, { "epoch": 113.8855421686747, "grad_norm": 3.988217353820801, "learning_rate": 4.298543689320388e-05, "loss": 0.1773, "step": 11745 }, { "epoch": 113.89518072289157, "grad_norm": 19.697036743164062, "learning_rate": 4.2980582524271846e-05, "loss": 0.064, "step": 11746 }, { "epoch": 113.90481927710843, "grad_norm": 45.25316619873047, "learning_rate": 4.2975728155339804e-05, "loss": 0.4898, "step": 11747 }, { "epoch": 113.91445783132531, "grad_norm": 29.96360206604004, "learning_rate": 4.297087378640777e-05, "loss": 0.1758, "step": 11748 }, { "epoch": 113.92409638554217, "grad_norm": 19.969100952148438, "learning_rate": 4.2966019417475726e-05, "loss": 0.1804, "step": 11749 }, { "epoch": 113.93373493975903, "grad_norm": 19.027359008789062, "learning_rate": 4.296116504854369e-05, "loss": 0.296, "step": 11750 }, { "epoch": 113.9433734939759, "grad_norm": 2.487485885620117, "learning_rate": 4.295631067961165e-05, "loss": 0.224, "step": 11751 }, { "epoch": 113.95301204819278, "grad_norm": 12.412528038024902, "learning_rate": 4.295145631067961e-05, "loss": 0.3639, "step": 11752 }, { "epoch": 113.96265060240964, "grad_norm": 49.352596282958984, "learning_rate": 4.294660194174758e-05, "loss": 0.2102, "step": 11753 }, { "epoch": 113.9722891566265, "grad_norm": 1.5110994577407837, "learning_rate": 4.2941747572815535e-05, "loss": 0.2015, "step": 11754 }, { "epoch": 113.98192771084338, "grad_norm": 34.03569412231445, "learning_rate": 4.29368932038835e-05, "loss": 0.4093, "step": 11755 }, { "epoch": 113.99156626506024, "grad_norm": 3.0332720279693604, "learning_rate": 4.293203883495146e-05, "loss": 0.2794, "step": 11756 }, { "epoch": 114.00722891566265, "grad_norm": 5.939467430114746, "learning_rate": 4.292718446601942e-05, "loss": 0.3068, "step": 11757 }, { "epoch": 114.01686746987951, "grad_norm": 12.910226821899414, "learning_rate": 4.292233009708738e-05, "loss": 0.2046, "step": 11758 }, { "epoch": 114.02650602409639, "grad_norm": 15.271279335021973, "learning_rate": 4.2917475728155345e-05, "loss": 0.3226, "step": 11759 }, { "epoch": 114.03614457831326, "grad_norm": 33.25260543823242, "learning_rate": 4.29126213592233e-05, "loss": 0.1408, "step": 11760 }, { "epoch": 114.04578313253012, "grad_norm": 4.345481872558594, "learning_rate": 4.290776699029127e-05, "loss": 0.142, "step": 11761 }, { "epoch": 114.05542168674698, "grad_norm": 11.636807441711426, "learning_rate": 4.2902912621359225e-05, "loss": 0.3581, "step": 11762 }, { "epoch": 114.06506024096386, "grad_norm": 10.222189903259277, "learning_rate": 4.289805825242719e-05, "loss": 0.318, "step": 11763 }, { "epoch": 114.07469879518072, "grad_norm": 10.998208999633789, "learning_rate": 4.289320388349515e-05, "loss": 0.2329, "step": 11764 }, { "epoch": 114.08433734939759, "grad_norm": 45.307472229003906, "learning_rate": 4.288834951456311e-05, "loss": 0.4008, "step": 11765 }, { "epoch": 114.09397590361446, "grad_norm": 17.762245178222656, "learning_rate": 4.288349514563107e-05, "loss": 0.2653, "step": 11766 }, { "epoch": 114.10361445783133, "grad_norm": 5.858060836791992, "learning_rate": 4.2878640776699034e-05, "loss": 0.2536, "step": 11767 }, { "epoch": 114.11325301204819, "grad_norm": 2.88796067237854, "learning_rate": 4.287378640776699e-05, "loss": 0.1914, "step": 11768 }, { "epoch": 114.12289156626505, "grad_norm": 3.638362169265747, "learning_rate": 4.286893203883495e-05, "loss": 0.1357, "step": 11769 }, { "epoch": 114.13253012048193, "grad_norm": 35.346923828125, "learning_rate": 4.2864077669902914e-05, "loss": 0.1941, "step": 11770 }, { "epoch": 114.1421686746988, "grad_norm": 35.1187858581543, "learning_rate": 4.285922330097087e-05, "loss": 0.2699, "step": 11771 }, { "epoch": 114.15180722891566, "grad_norm": 28.678421020507812, "learning_rate": 4.2854368932038836e-05, "loss": 0.311, "step": 11772 }, { "epoch": 114.16144578313254, "grad_norm": 12.992177963256836, "learning_rate": 4.2849514563106794e-05, "loss": 0.208, "step": 11773 }, { "epoch": 114.1710843373494, "grad_norm": 7.36403226852417, "learning_rate": 4.284466019417476e-05, "loss": 0.3757, "step": 11774 }, { "epoch": 114.18072289156626, "grad_norm": 15.848158836364746, "learning_rate": 4.2839805825242716e-05, "loss": 0.3233, "step": 11775 }, { "epoch": 114.19036144578314, "grad_norm": 13.490190505981445, "learning_rate": 4.283495145631068e-05, "loss": 0.2152, "step": 11776 }, { "epoch": 114.2, "grad_norm": 31.820798873901367, "learning_rate": 4.2830097087378645e-05, "loss": 0.4359, "step": 11777 }, { "epoch": 114.20963855421687, "grad_norm": 2.7983548641204834, "learning_rate": 4.282524271844661e-05, "loss": 0.2901, "step": 11778 }, { "epoch": 114.21927710843373, "grad_norm": 17.878843307495117, "learning_rate": 4.282038834951457e-05, "loss": 0.1941, "step": 11779 }, { "epoch": 114.2289156626506, "grad_norm": 6.634676933288574, "learning_rate": 4.2815533980582525e-05, "loss": 0.3095, "step": 11780 }, { "epoch": 114.23855421686747, "grad_norm": 15.85443115234375, "learning_rate": 4.281067961165049e-05, "loss": 0.4151, "step": 11781 }, { "epoch": 114.24819277108433, "grad_norm": 28.332094192504883, "learning_rate": 4.280582524271845e-05, "loss": 0.4201, "step": 11782 }, { "epoch": 114.25783132530121, "grad_norm": 90.91315460205078, "learning_rate": 4.280097087378641e-05, "loss": 0.2362, "step": 11783 }, { "epoch": 114.26746987951807, "grad_norm": 34.92811965942383, "learning_rate": 4.279611650485437e-05, "loss": 0.2293, "step": 11784 }, { "epoch": 114.27710843373494, "grad_norm": 15.959761619567871, "learning_rate": 4.2791262135922334e-05, "loss": 0.2064, "step": 11785 }, { "epoch": 114.2867469879518, "grad_norm": 13.911517143249512, "learning_rate": 4.278640776699029e-05, "loss": 0.281, "step": 11786 }, { "epoch": 114.29638554216868, "grad_norm": 8.309896469116211, "learning_rate": 4.2781553398058257e-05, "loss": 0.2862, "step": 11787 }, { "epoch": 114.30602409638554, "grad_norm": 35.65857696533203, "learning_rate": 4.2776699029126214e-05, "loss": 0.3934, "step": 11788 }, { "epoch": 114.3156626506024, "grad_norm": 16.690887451171875, "learning_rate": 4.277184466019418e-05, "loss": 0.3278, "step": 11789 }, { "epoch": 114.32530120481928, "grad_norm": 30.739601135253906, "learning_rate": 4.276699029126214e-05, "loss": 0.4947, "step": 11790 }, { "epoch": 114.33493975903615, "grad_norm": 1.7056273221969604, "learning_rate": 4.27621359223301e-05, "loss": 0.4277, "step": 11791 }, { "epoch": 114.34457831325301, "grad_norm": 5.2190423011779785, "learning_rate": 4.275728155339806e-05, "loss": 0.2164, "step": 11792 }, { "epoch": 114.35421686746987, "grad_norm": 20.811214447021484, "learning_rate": 4.2752427184466023e-05, "loss": 0.1472, "step": 11793 }, { "epoch": 114.36385542168675, "grad_norm": 25.158220291137695, "learning_rate": 4.274757281553398e-05, "loss": 0.4212, "step": 11794 }, { "epoch": 114.37349397590361, "grad_norm": 6.339861869812012, "learning_rate": 4.2742718446601946e-05, "loss": 0.2461, "step": 11795 }, { "epoch": 114.38313253012048, "grad_norm": 6.850883960723877, "learning_rate": 4.2737864077669904e-05, "loss": 0.3271, "step": 11796 }, { "epoch": 114.39277108433735, "grad_norm": 28.06508445739746, "learning_rate": 4.273300970873786e-05, "loss": 0.2908, "step": 11797 }, { "epoch": 114.40240963855422, "grad_norm": 33.47550964355469, "learning_rate": 4.2728155339805826e-05, "loss": 0.3319, "step": 11798 }, { "epoch": 114.41204819277108, "grad_norm": 61.969120025634766, "learning_rate": 4.2723300970873784e-05, "loss": 0.3146, "step": 11799 }, { "epoch": 114.42168674698796, "grad_norm": 45.677268981933594, "learning_rate": 4.271844660194175e-05, "loss": 0.3568, "step": 11800 }, { "epoch": 114.43132530120482, "grad_norm": 19.67051887512207, "learning_rate": 4.2713592233009706e-05, "loss": 0.4429, "step": 11801 }, { "epoch": 114.44096385542169, "grad_norm": 34.42741775512695, "learning_rate": 4.270873786407767e-05, "loss": 0.2475, "step": 11802 }, { "epoch": 114.45060240963855, "grad_norm": 6.868599891662598, "learning_rate": 4.2703883495145635e-05, "loss": 0.2376, "step": 11803 }, { "epoch": 114.46024096385543, "grad_norm": 23.421815872192383, "learning_rate": 4.26990291262136e-05, "loss": 0.268, "step": 11804 }, { "epoch": 114.46987951807229, "grad_norm": 5.272761821746826, "learning_rate": 4.269417475728156e-05, "loss": 0.1275, "step": 11805 }, { "epoch": 114.47951807228915, "grad_norm": 128.1632843017578, "learning_rate": 4.268932038834952e-05, "loss": 0.3162, "step": 11806 }, { "epoch": 114.48915662650603, "grad_norm": 41.74913787841797, "learning_rate": 4.268446601941748e-05, "loss": 0.4777, "step": 11807 }, { "epoch": 114.4987951807229, "grad_norm": 5.589845180511475, "learning_rate": 4.267961165048544e-05, "loss": 0.2169, "step": 11808 }, { "epoch": 114.50843373493976, "grad_norm": 25.798368453979492, "learning_rate": 4.26747572815534e-05, "loss": 0.2366, "step": 11809 }, { "epoch": 114.51807228915662, "grad_norm": 8.485260009765625, "learning_rate": 4.266990291262136e-05, "loss": 0.2555, "step": 11810 }, { "epoch": 114.5277108433735, "grad_norm": 26.82977867126465, "learning_rate": 4.2665048543689324e-05, "loss": 0.3826, "step": 11811 }, { "epoch": 114.53734939759036, "grad_norm": 16.853172302246094, "learning_rate": 4.266019417475728e-05, "loss": 0.2841, "step": 11812 }, { "epoch": 114.54698795180722, "grad_norm": 44.57697677612305, "learning_rate": 4.2655339805825246e-05, "loss": 0.3994, "step": 11813 }, { "epoch": 114.5566265060241, "grad_norm": 11.828595161437988, "learning_rate": 4.2650485436893204e-05, "loss": 0.4901, "step": 11814 }, { "epoch": 114.56626506024097, "grad_norm": 8.605745315551758, "learning_rate": 4.264563106796117e-05, "loss": 0.3901, "step": 11815 }, { "epoch": 114.57590361445783, "grad_norm": 15.40347957611084, "learning_rate": 4.2640776699029126e-05, "loss": 0.2948, "step": 11816 }, { "epoch": 114.58554216867469, "grad_norm": 19.03851890563965, "learning_rate": 4.263592233009709e-05, "loss": 0.1914, "step": 11817 }, { "epoch": 114.59518072289157, "grad_norm": 2.3314476013183594, "learning_rate": 4.263106796116505e-05, "loss": 0.2845, "step": 11818 }, { "epoch": 114.60481927710843, "grad_norm": 8.246538162231445, "learning_rate": 4.262621359223301e-05, "loss": 0.2498, "step": 11819 }, { "epoch": 114.6144578313253, "grad_norm": 7.723062038421631, "learning_rate": 4.262135922330097e-05, "loss": 0.2047, "step": 11820 }, { "epoch": 114.62409638554217, "grad_norm": 7.852945327758789, "learning_rate": 4.2616504854368936e-05, "loss": 0.2572, "step": 11821 }, { "epoch": 114.63373493975904, "grad_norm": 15.73487663269043, "learning_rate": 4.261165048543689e-05, "loss": 0.1546, "step": 11822 }, { "epoch": 114.6433734939759, "grad_norm": 29.70379638671875, "learning_rate": 4.260679611650486e-05, "loss": 0.2234, "step": 11823 }, { "epoch": 114.65301204819278, "grad_norm": 13.29513931274414, "learning_rate": 4.2601941747572816e-05, "loss": 0.3544, "step": 11824 }, { "epoch": 114.66265060240964, "grad_norm": 6.065560817718506, "learning_rate": 4.259708737864077e-05, "loss": 0.5275, "step": 11825 }, { "epoch": 114.6722891566265, "grad_norm": 13.330816268920898, "learning_rate": 4.259223300970874e-05, "loss": 0.1852, "step": 11826 }, { "epoch": 114.68192771084337, "grad_norm": 6.158689975738525, "learning_rate": 4.2587378640776696e-05, "loss": 0.2267, "step": 11827 }, { "epoch": 114.69156626506025, "grad_norm": 2.977548122406006, "learning_rate": 4.258252427184467e-05, "loss": 0.1404, "step": 11828 }, { "epoch": 114.70120481927711, "grad_norm": 8.644643783569336, "learning_rate": 4.2577669902912625e-05, "loss": 0.1811, "step": 11829 }, { "epoch": 114.71084337349397, "grad_norm": 15.763404846191406, "learning_rate": 4.257281553398059e-05, "loss": 0.2248, "step": 11830 }, { "epoch": 114.72048192771085, "grad_norm": 12.922475814819336, "learning_rate": 4.256796116504855e-05, "loss": 0.1832, "step": 11831 }, { "epoch": 114.73012048192771, "grad_norm": 24.644756317138672, "learning_rate": 4.256310679611651e-05, "loss": 0.3185, "step": 11832 }, { "epoch": 114.73975903614458, "grad_norm": 5.354084014892578, "learning_rate": 4.255825242718447e-05, "loss": 0.2302, "step": 11833 }, { "epoch": 114.74939759036144, "grad_norm": 31.45195960998535, "learning_rate": 4.2553398058252434e-05, "loss": 0.2757, "step": 11834 }, { "epoch": 114.75903614457832, "grad_norm": 27.149534225463867, "learning_rate": 4.254854368932039e-05, "loss": 0.4001, "step": 11835 }, { "epoch": 114.76867469879518, "grad_norm": 4.495712757110596, "learning_rate": 4.254368932038835e-05, "loss": 0.2447, "step": 11836 }, { "epoch": 114.77831325301204, "grad_norm": 55.42075729370117, "learning_rate": 4.2538834951456314e-05, "loss": 0.4833, "step": 11837 }, { "epoch": 114.78795180722892, "grad_norm": 26.424253463745117, "learning_rate": 4.253398058252427e-05, "loss": 0.3602, "step": 11838 }, { "epoch": 114.79759036144578, "grad_norm": 9.60529613494873, "learning_rate": 4.2529126213592236e-05, "loss": 0.214, "step": 11839 }, { "epoch": 114.80722891566265, "grad_norm": 2.9607160091400146, "learning_rate": 4.2524271844660194e-05, "loss": 0.1679, "step": 11840 }, { "epoch": 114.81686746987951, "grad_norm": 10.225025177001953, "learning_rate": 4.251941747572816e-05, "loss": 0.2603, "step": 11841 }, { "epoch": 114.82650602409639, "grad_norm": 4.884075164794922, "learning_rate": 4.2514563106796116e-05, "loss": 0.2922, "step": 11842 }, { "epoch": 114.83614457831325, "grad_norm": 9.847532272338867, "learning_rate": 4.250970873786408e-05, "loss": 0.2064, "step": 11843 }, { "epoch": 114.84578313253012, "grad_norm": 30.20549774169922, "learning_rate": 4.250485436893204e-05, "loss": 0.6298, "step": 11844 }, { "epoch": 114.855421686747, "grad_norm": 11.683320999145508, "learning_rate": 4.25e-05, "loss": 0.2062, "step": 11845 }, { "epoch": 114.86506024096386, "grad_norm": 10.579000473022461, "learning_rate": 4.249514563106796e-05, "loss": 0.321, "step": 11846 }, { "epoch": 114.87469879518072, "grad_norm": 24.808523178100586, "learning_rate": 4.2490291262135925e-05, "loss": 0.1683, "step": 11847 }, { "epoch": 114.88433734939758, "grad_norm": 10.533463478088379, "learning_rate": 4.248543689320388e-05, "loss": 0.4535, "step": 11848 }, { "epoch": 114.89397590361446, "grad_norm": 6.310844421386719, "learning_rate": 4.248058252427185e-05, "loss": 0.2133, "step": 11849 }, { "epoch": 114.90361445783132, "grad_norm": 44.2099609375, "learning_rate": 4.2475728155339805e-05, "loss": 0.2771, "step": 11850 }, { "epoch": 114.91325301204819, "grad_norm": 6.525218963623047, "learning_rate": 4.247087378640777e-05, "loss": 0.2553, "step": 11851 }, { "epoch": 114.92289156626506, "grad_norm": 23.06558609008789, "learning_rate": 4.246601941747573e-05, "loss": 0.3049, "step": 11852 }, { "epoch": 114.93253012048193, "grad_norm": 23.467178344726562, "learning_rate": 4.246116504854369e-05, "loss": 0.349, "step": 11853 }, { "epoch": 114.94216867469879, "grad_norm": 25.843355178833008, "learning_rate": 4.245631067961166e-05, "loss": 0.3998, "step": 11854 }, { "epoch": 114.95180722891567, "grad_norm": 3.8632307052612305, "learning_rate": 4.2451456310679615e-05, "loss": 0.3005, "step": 11855 }, { "epoch": 114.96144578313253, "grad_norm": 18.291336059570312, "learning_rate": 4.244660194174758e-05, "loss": 0.2681, "step": 11856 }, { "epoch": 114.9710843373494, "grad_norm": 6.6445746421813965, "learning_rate": 4.244174757281554e-05, "loss": 0.2298, "step": 11857 }, { "epoch": 114.98072289156626, "grad_norm": 15.942112922668457, "learning_rate": 4.24368932038835e-05, "loss": 0.3374, "step": 11858 }, { "epoch": 114.99036144578314, "grad_norm": 10.683100700378418, "learning_rate": 4.243203883495146e-05, "loss": 0.3398, "step": 11859 }, { "epoch": 115.00602409638554, "grad_norm": 43.54426193237305, "learning_rate": 4.2427184466019424e-05, "loss": 0.3847, "step": 11860 }, { "epoch": 115.01566265060241, "grad_norm": 26.355051040649414, "learning_rate": 4.242233009708738e-05, "loss": 0.4901, "step": 11861 }, { "epoch": 115.02530120481927, "grad_norm": 5.292012691497803, "learning_rate": 4.2417475728155346e-05, "loss": 0.2997, "step": 11862 }, { "epoch": 115.03493975903615, "grad_norm": 22.19733428955078, "learning_rate": 4.2412621359223304e-05, "loss": 0.2666, "step": 11863 }, { "epoch": 115.04457831325301, "grad_norm": 18.616636276245117, "learning_rate": 4.240776699029126e-05, "loss": 0.4569, "step": 11864 }, { "epoch": 115.05421686746988, "grad_norm": 20.654510498046875, "learning_rate": 4.2402912621359226e-05, "loss": 0.1671, "step": 11865 }, { "epoch": 115.06385542168675, "grad_norm": 6.4011125564575195, "learning_rate": 4.2398058252427184e-05, "loss": 0.0981, "step": 11866 }, { "epoch": 115.07349397590362, "grad_norm": 9.011500358581543, "learning_rate": 4.239320388349515e-05, "loss": 0.3238, "step": 11867 }, { "epoch": 115.08313253012048, "grad_norm": 14.093372344970703, "learning_rate": 4.2388349514563106e-05, "loss": 0.2886, "step": 11868 }, { "epoch": 115.09277108433734, "grad_norm": 5.287201881408691, "learning_rate": 4.238349514563107e-05, "loss": 0.0774, "step": 11869 }, { "epoch": 115.10240963855422, "grad_norm": 14.50005054473877, "learning_rate": 4.237864077669903e-05, "loss": 0.2969, "step": 11870 }, { "epoch": 115.11204819277108, "grad_norm": 35.424800872802734, "learning_rate": 4.237378640776699e-05, "loss": 0.3811, "step": 11871 }, { "epoch": 115.12168674698795, "grad_norm": 10.618531227111816, "learning_rate": 4.236893203883495e-05, "loss": 0.2038, "step": 11872 }, { "epoch": 115.13132530120482, "grad_norm": 12.548810005187988, "learning_rate": 4.2364077669902915e-05, "loss": 0.278, "step": 11873 }, { "epoch": 115.14096385542169, "grad_norm": 4.594973087310791, "learning_rate": 4.235922330097087e-05, "loss": 0.3101, "step": 11874 }, { "epoch": 115.15060240963855, "grad_norm": 15.445226669311523, "learning_rate": 4.235436893203884e-05, "loss": 0.3015, "step": 11875 }, { "epoch": 115.16024096385541, "grad_norm": 9.490171432495117, "learning_rate": 4.2349514563106795e-05, "loss": 0.1807, "step": 11876 }, { "epoch": 115.16987951807229, "grad_norm": 12.570267677307129, "learning_rate": 4.234466019417476e-05, "loss": 0.3745, "step": 11877 }, { "epoch": 115.17951807228916, "grad_norm": 2.8482649326324463, "learning_rate": 4.2339805825242724e-05, "loss": 0.3575, "step": 11878 }, { "epoch": 115.18915662650602, "grad_norm": 29.014026641845703, "learning_rate": 4.233495145631068e-05, "loss": 0.2839, "step": 11879 }, { "epoch": 115.1987951807229, "grad_norm": 14.569732666015625, "learning_rate": 4.2330097087378647e-05, "loss": 0.0994, "step": 11880 }, { "epoch": 115.20843373493976, "grad_norm": 27.80503273010254, "learning_rate": 4.2325242718446604e-05, "loss": 0.3402, "step": 11881 }, { "epoch": 115.21807228915662, "grad_norm": 11.206611633300781, "learning_rate": 4.232038834951457e-05, "loss": 0.196, "step": 11882 }, { "epoch": 115.2277108433735, "grad_norm": 5.0448527336120605, "learning_rate": 4.2315533980582527e-05, "loss": 0.1838, "step": 11883 }, { "epoch": 115.23734939759036, "grad_norm": 8.393877983093262, "learning_rate": 4.231067961165049e-05, "loss": 0.2956, "step": 11884 }, { "epoch": 115.24698795180723, "grad_norm": 9.284196853637695, "learning_rate": 4.230582524271845e-05, "loss": 0.4102, "step": 11885 }, { "epoch": 115.25662650602409, "grad_norm": 9.633742332458496, "learning_rate": 4.2300970873786413e-05, "loss": 0.3587, "step": 11886 }, { "epoch": 115.26626506024097, "grad_norm": 6.2352776527404785, "learning_rate": 4.229611650485437e-05, "loss": 0.1159, "step": 11887 }, { "epoch": 115.27590361445783, "grad_norm": 13.93200969696045, "learning_rate": 4.2291262135922336e-05, "loss": 0.3965, "step": 11888 }, { "epoch": 115.2855421686747, "grad_norm": 7.557431697845459, "learning_rate": 4.2286407766990293e-05, "loss": 0.1501, "step": 11889 }, { "epoch": 115.29518072289157, "grad_norm": 3.325404644012451, "learning_rate": 4.228155339805825e-05, "loss": 0.0458, "step": 11890 }, { "epoch": 115.30481927710844, "grad_norm": 6.941705226898193, "learning_rate": 4.2276699029126216e-05, "loss": 0.2305, "step": 11891 }, { "epoch": 115.3144578313253, "grad_norm": 10.329063415527344, "learning_rate": 4.2271844660194174e-05, "loss": 0.3207, "step": 11892 }, { "epoch": 115.32409638554216, "grad_norm": 22.810745239257812, "learning_rate": 4.226699029126214e-05, "loss": 0.3145, "step": 11893 }, { "epoch": 115.33373493975904, "grad_norm": 19.08848762512207, "learning_rate": 4.2262135922330096e-05, "loss": 0.22, "step": 11894 }, { "epoch": 115.3433734939759, "grad_norm": 23.857921600341797, "learning_rate": 4.225728155339806e-05, "loss": 0.098, "step": 11895 }, { "epoch": 115.35301204819277, "grad_norm": 10.918441772460938, "learning_rate": 4.225242718446602e-05, "loss": 0.2549, "step": 11896 }, { "epoch": 115.36265060240964, "grad_norm": 3.1550776958465576, "learning_rate": 4.224757281553398e-05, "loss": 0.1825, "step": 11897 }, { "epoch": 115.37228915662651, "grad_norm": 16.090299606323242, "learning_rate": 4.224271844660194e-05, "loss": 0.3263, "step": 11898 }, { "epoch": 115.38192771084337, "grad_norm": 5.0662007331848145, "learning_rate": 4.2237864077669905e-05, "loss": 0.2921, "step": 11899 }, { "epoch": 115.39156626506023, "grad_norm": 5.659623622894287, "learning_rate": 4.223300970873786e-05, "loss": 0.0646, "step": 11900 }, { "epoch": 115.40120481927711, "grad_norm": 11.314363479614258, "learning_rate": 4.222815533980583e-05, "loss": 0.3656, "step": 11901 }, { "epoch": 115.41084337349398, "grad_norm": 1.924859881401062, "learning_rate": 4.2223300970873785e-05, "loss": 0.1917, "step": 11902 }, { "epoch": 115.42048192771084, "grad_norm": 2.503182888031006, "learning_rate": 4.221844660194175e-05, "loss": 0.122, "step": 11903 }, { "epoch": 115.43012048192772, "grad_norm": 25.88618278503418, "learning_rate": 4.2213592233009714e-05, "loss": 0.2341, "step": 11904 }, { "epoch": 115.43975903614458, "grad_norm": 13.106340408325195, "learning_rate": 4.220873786407767e-05, "loss": 0.2629, "step": 11905 }, { "epoch": 115.44939759036144, "grad_norm": 47.3943977355957, "learning_rate": 4.2203883495145636e-05, "loss": 0.3561, "step": 11906 }, { "epoch": 115.45903614457832, "grad_norm": 27.927900314331055, "learning_rate": 4.2199029126213594e-05, "loss": 0.4886, "step": 11907 }, { "epoch": 115.46867469879518, "grad_norm": 7.261203289031982, "learning_rate": 4.219417475728156e-05, "loss": 0.3851, "step": 11908 }, { "epoch": 115.47831325301205, "grad_norm": 7.028786659240723, "learning_rate": 4.2189320388349516e-05, "loss": 0.3052, "step": 11909 }, { "epoch": 115.48795180722891, "grad_norm": 7.4943928718566895, "learning_rate": 4.218446601941748e-05, "loss": 0.2254, "step": 11910 }, { "epoch": 115.49759036144579, "grad_norm": 7.997652530670166, "learning_rate": 4.217961165048544e-05, "loss": 0.1809, "step": 11911 }, { "epoch": 115.50722891566265, "grad_norm": 6.632254600524902, "learning_rate": 4.21747572815534e-05, "loss": 0.2796, "step": 11912 }, { "epoch": 115.51686746987951, "grad_norm": 10.051952362060547, "learning_rate": 4.216990291262136e-05, "loss": 0.2101, "step": 11913 }, { "epoch": 115.52650602409639, "grad_norm": 13.00368595123291, "learning_rate": 4.2165048543689325e-05, "loss": 0.3129, "step": 11914 }, { "epoch": 115.53614457831326, "grad_norm": 14.062357902526855, "learning_rate": 4.216019417475728e-05, "loss": 0.3233, "step": 11915 }, { "epoch": 115.54578313253012, "grad_norm": 29.735715866088867, "learning_rate": 4.215533980582525e-05, "loss": 0.3519, "step": 11916 }, { "epoch": 115.55542168674698, "grad_norm": 6.364007949829102, "learning_rate": 4.2150485436893206e-05, "loss": 0.219, "step": 11917 }, { "epoch": 115.56506024096386, "grad_norm": 19.15200424194336, "learning_rate": 4.214563106796116e-05, "loss": 0.3838, "step": 11918 }, { "epoch": 115.57469879518072, "grad_norm": 7.379577159881592, "learning_rate": 4.214077669902913e-05, "loss": 0.4859, "step": 11919 }, { "epoch": 115.58433734939759, "grad_norm": 20.033119201660156, "learning_rate": 4.2135922330097086e-05, "loss": 0.3736, "step": 11920 }, { "epoch": 115.59397590361446, "grad_norm": 4.225557804107666, "learning_rate": 4.213106796116505e-05, "loss": 0.2039, "step": 11921 }, { "epoch": 115.60361445783133, "grad_norm": 11.218701362609863, "learning_rate": 4.212621359223301e-05, "loss": 0.1927, "step": 11922 }, { "epoch": 115.61325301204819, "grad_norm": 8.424715042114258, "learning_rate": 4.212135922330097e-05, "loss": 0.1181, "step": 11923 }, { "epoch": 115.62289156626505, "grad_norm": 6.471581935882568, "learning_rate": 4.211650485436893e-05, "loss": 0.2579, "step": 11924 }, { "epoch": 115.63253012048193, "grad_norm": 4.706819534301758, "learning_rate": 4.2111650485436895e-05, "loss": 0.2205, "step": 11925 }, { "epoch": 115.6421686746988, "grad_norm": 13.431238174438477, "learning_rate": 4.210679611650485e-05, "loss": 0.2244, "step": 11926 }, { "epoch": 115.65180722891566, "grad_norm": 6.963510990142822, "learning_rate": 4.210194174757282e-05, "loss": 0.1556, "step": 11927 }, { "epoch": 115.66144578313254, "grad_norm": 1.6845489740371704, "learning_rate": 4.2097087378640775e-05, "loss": 0.0736, "step": 11928 }, { "epoch": 115.6710843373494, "grad_norm": 12.585618019104004, "learning_rate": 4.209223300970874e-05, "loss": 0.2024, "step": 11929 }, { "epoch": 115.68072289156626, "grad_norm": 11.553854942321777, "learning_rate": 4.2087378640776704e-05, "loss": 0.1489, "step": 11930 }, { "epoch": 115.69036144578314, "grad_norm": 7.95500373840332, "learning_rate": 4.208252427184466e-05, "loss": 0.3196, "step": 11931 }, { "epoch": 115.7, "grad_norm": 6.619874000549316, "learning_rate": 4.2077669902912626e-05, "loss": 0.1586, "step": 11932 }, { "epoch": 115.70963855421687, "grad_norm": 15.586009979248047, "learning_rate": 4.2072815533980584e-05, "loss": 0.1401, "step": 11933 }, { "epoch": 115.71927710843373, "grad_norm": 8.902055740356445, "learning_rate": 4.206796116504855e-05, "loss": 0.1641, "step": 11934 }, { "epoch": 115.7289156626506, "grad_norm": 13.410812377929688, "learning_rate": 4.2063106796116506e-05, "loss": 0.2704, "step": 11935 }, { "epoch": 115.73855421686747, "grad_norm": 24.015775680541992, "learning_rate": 4.205825242718447e-05, "loss": 0.2158, "step": 11936 }, { "epoch": 115.74819277108433, "grad_norm": 14.76106071472168, "learning_rate": 4.205339805825243e-05, "loss": 0.3326, "step": 11937 }, { "epoch": 115.75783132530121, "grad_norm": 38.16150665283203, "learning_rate": 4.204854368932039e-05, "loss": 0.3081, "step": 11938 }, { "epoch": 115.76746987951807, "grad_norm": 27.841825485229492, "learning_rate": 4.204368932038835e-05, "loss": 0.1956, "step": 11939 }, { "epoch": 115.77710843373494, "grad_norm": 22.231966018676758, "learning_rate": 4.2038834951456315e-05, "loss": 0.2929, "step": 11940 }, { "epoch": 115.7867469879518, "grad_norm": 15.30441665649414, "learning_rate": 4.203398058252427e-05, "loss": 0.387, "step": 11941 }, { "epoch": 115.79638554216868, "grad_norm": 34.952789306640625, "learning_rate": 4.202912621359224e-05, "loss": 0.4256, "step": 11942 }, { "epoch": 115.80602409638554, "grad_norm": 9.727755546569824, "learning_rate": 4.2024271844660195e-05, "loss": 0.5593, "step": 11943 }, { "epoch": 115.8156626506024, "grad_norm": 1.8681950569152832, "learning_rate": 4.201941747572816e-05, "loss": 0.0776, "step": 11944 }, { "epoch": 115.82530120481928, "grad_norm": 2.830652952194214, "learning_rate": 4.201456310679612e-05, "loss": 0.0979, "step": 11945 }, { "epoch": 115.83493975903615, "grad_norm": 5.156340599060059, "learning_rate": 4.2009708737864075e-05, "loss": 0.0621, "step": 11946 }, { "epoch": 115.84457831325301, "grad_norm": 31.559236526489258, "learning_rate": 4.200485436893204e-05, "loss": 0.3681, "step": 11947 }, { "epoch": 115.85421686746987, "grad_norm": 5.404388427734375, "learning_rate": 4.2e-05, "loss": 0.2678, "step": 11948 }, { "epoch": 115.86385542168675, "grad_norm": 8.561423301696777, "learning_rate": 4.199514563106796e-05, "loss": 0.1998, "step": 11949 }, { "epoch": 115.87349397590361, "grad_norm": 4.63213586807251, "learning_rate": 4.199029126213592e-05, "loss": 0.1544, "step": 11950 }, { "epoch": 115.88313253012048, "grad_norm": 6.4007248878479, "learning_rate": 4.1985436893203885e-05, "loss": 0.2926, "step": 11951 }, { "epoch": 115.89277108433735, "grad_norm": 8.98071575164795, "learning_rate": 4.198058252427184e-05, "loss": 0.2381, "step": 11952 }, { "epoch": 115.90240963855422, "grad_norm": 12.553323745727539, "learning_rate": 4.197572815533981e-05, "loss": 0.2852, "step": 11953 }, { "epoch": 115.91204819277108, "grad_norm": 5.202991962432861, "learning_rate": 4.197087378640777e-05, "loss": 0.2507, "step": 11954 }, { "epoch": 115.92168674698796, "grad_norm": 41.67169189453125, "learning_rate": 4.1966019417475736e-05, "loss": 0.184, "step": 11955 }, { "epoch": 115.93132530120482, "grad_norm": 19.5996150970459, "learning_rate": 4.1961165048543694e-05, "loss": 0.6967, "step": 11956 }, { "epoch": 115.94096385542169, "grad_norm": 9.14106559753418, "learning_rate": 4.195631067961165e-05, "loss": 0.2116, "step": 11957 }, { "epoch": 115.95060240963855, "grad_norm": 4.838225841522217, "learning_rate": 4.1951456310679616e-05, "loss": 0.2717, "step": 11958 }, { "epoch": 115.96024096385543, "grad_norm": 18.361896514892578, "learning_rate": 4.1946601941747574e-05, "loss": 0.4246, "step": 11959 }, { "epoch": 115.96987951807229, "grad_norm": 13.231897354125977, "learning_rate": 4.194174757281554e-05, "loss": 0.3304, "step": 11960 }, { "epoch": 115.97951807228915, "grad_norm": 4.730155944824219, "learning_rate": 4.1936893203883496e-05, "loss": 0.3532, "step": 11961 }, { "epoch": 115.98915662650603, "grad_norm": 6.177082538604736, "learning_rate": 4.193203883495146e-05, "loss": 0.2214, "step": 11962 }, { "epoch": 116.00481927710844, "grad_norm": 7.898014545440674, "learning_rate": 4.192718446601942e-05, "loss": 0.2484, "step": 11963 }, { "epoch": 116.0144578313253, "grad_norm": 2.2751550674438477, "learning_rate": 4.192233009708738e-05, "loss": 0.1779, "step": 11964 }, { "epoch": 116.02409638554217, "grad_norm": 16.227718353271484, "learning_rate": 4.191747572815534e-05, "loss": 0.2827, "step": 11965 }, { "epoch": 116.03373493975904, "grad_norm": 9.73901653289795, "learning_rate": 4.1912621359223305e-05, "loss": 0.2619, "step": 11966 }, { "epoch": 116.0433734939759, "grad_norm": 32.007144927978516, "learning_rate": 4.190776699029126e-05, "loss": 0.4976, "step": 11967 }, { "epoch": 116.05301204819277, "grad_norm": 14.009305000305176, "learning_rate": 4.190291262135923e-05, "loss": 0.3049, "step": 11968 }, { "epoch": 116.06265060240963, "grad_norm": 11.16700267791748, "learning_rate": 4.1898058252427185e-05, "loss": 0.2815, "step": 11969 }, { "epoch": 116.07228915662651, "grad_norm": 12.51284408569336, "learning_rate": 4.189320388349515e-05, "loss": 0.1872, "step": 11970 }, { "epoch": 116.08192771084337, "grad_norm": 6.076000690460205, "learning_rate": 4.188834951456311e-05, "loss": 0.2086, "step": 11971 }, { "epoch": 116.09156626506024, "grad_norm": 9.983617782592773, "learning_rate": 4.188349514563107e-05, "loss": 0.2573, "step": 11972 }, { "epoch": 116.10120481927711, "grad_norm": 16.1474666595459, "learning_rate": 4.187864077669903e-05, "loss": 0.1755, "step": 11973 }, { "epoch": 116.11084337349398, "grad_norm": 21.683977127075195, "learning_rate": 4.187378640776699e-05, "loss": 0.4304, "step": 11974 }, { "epoch": 116.12048192771084, "grad_norm": 9.398666381835938, "learning_rate": 4.186893203883495e-05, "loss": 0.3499, "step": 11975 }, { "epoch": 116.1301204819277, "grad_norm": 24.291257858276367, "learning_rate": 4.186407766990291e-05, "loss": 0.2992, "step": 11976 }, { "epoch": 116.13975903614458, "grad_norm": 15.428595542907715, "learning_rate": 4.1859223300970874e-05, "loss": 0.3074, "step": 11977 }, { "epoch": 116.14939759036145, "grad_norm": 4.021728992462158, "learning_rate": 4.185436893203883e-05, "loss": 0.2904, "step": 11978 }, { "epoch": 116.15903614457831, "grad_norm": 8.467785835266113, "learning_rate": 4.18495145631068e-05, "loss": 0.2125, "step": 11979 }, { "epoch": 116.16867469879519, "grad_norm": 78.32848358154297, "learning_rate": 4.184466019417476e-05, "loss": 0.2277, "step": 11980 }, { "epoch": 116.17831325301205, "grad_norm": 10.707297325134277, "learning_rate": 4.1839805825242726e-05, "loss": 0.2151, "step": 11981 }, { "epoch": 116.18795180722891, "grad_norm": 7.815500736236572, "learning_rate": 4.1834951456310683e-05, "loss": 0.239, "step": 11982 }, { "epoch": 116.19759036144578, "grad_norm": 3.2166736125946045, "learning_rate": 4.183009708737865e-05, "loss": 0.0564, "step": 11983 }, { "epoch": 116.20722891566265, "grad_norm": 16.054935455322266, "learning_rate": 4.1825242718446606e-05, "loss": 0.4107, "step": 11984 }, { "epoch": 116.21686746987952, "grad_norm": 7.942381858825684, "learning_rate": 4.1820388349514563e-05, "loss": 0.137, "step": 11985 }, { "epoch": 116.22650602409638, "grad_norm": 6.61617374420166, "learning_rate": 4.181553398058253e-05, "loss": 0.253, "step": 11986 }, { "epoch": 116.23614457831326, "grad_norm": 5.814845561981201, "learning_rate": 4.1810679611650486e-05, "loss": 0.2973, "step": 11987 }, { "epoch": 116.24578313253012, "grad_norm": 6.543908596038818, "learning_rate": 4.180582524271845e-05, "loss": 0.1991, "step": 11988 }, { "epoch": 116.25542168674698, "grad_norm": 28.841354370117188, "learning_rate": 4.180097087378641e-05, "loss": 0.2516, "step": 11989 }, { "epoch": 116.26506024096386, "grad_norm": 52.60383987426758, "learning_rate": 4.179611650485437e-05, "loss": 0.2568, "step": 11990 }, { "epoch": 116.27469879518073, "grad_norm": 20.31248664855957, "learning_rate": 4.179126213592233e-05, "loss": 0.2362, "step": 11991 }, { "epoch": 116.28433734939759, "grad_norm": 7.968570232391357, "learning_rate": 4.1786407766990295e-05, "loss": 0.2219, "step": 11992 }, { "epoch": 116.29397590361445, "grad_norm": 4.651917457580566, "learning_rate": 4.178155339805825e-05, "loss": 0.1166, "step": 11993 }, { "epoch": 116.30361445783133, "grad_norm": 8.149742126464844, "learning_rate": 4.177669902912622e-05, "loss": 0.2663, "step": 11994 }, { "epoch": 116.3132530120482, "grad_norm": 39.327789306640625, "learning_rate": 4.1771844660194175e-05, "loss": 0.1399, "step": 11995 }, { "epoch": 116.32289156626506, "grad_norm": 28.431760787963867, "learning_rate": 4.176699029126214e-05, "loss": 0.2778, "step": 11996 }, { "epoch": 116.33253012048193, "grad_norm": 8.550097465515137, "learning_rate": 4.17621359223301e-05, "loss": 0.2855, "step": 11997 }, { "epoch": 116.3421686746988, "grad_norm": 46.80677032470703, "learning_rate": 4.175728155339806e-05, "loss": 0.2566, "step": 11998 }, { "epoch": 116.35180722891566, "grad_norm": 19.41022300720215, "learning_rate": 4.175242718446602e-05, "loss": 0.3505, "step": 11999 }, { "epoch": 116.36144578313252, "grad_norm": 23.008968353271484, "learning_rate": 4.1747572815533984e-05, "loss": 0.2621, "step": 12000 }, { "epoch": 116.3710843373494, "grad_norm": 5.8551483154296875, "learning_rate": 4.174271844660194e-05, "loss": 0.3085, "step": 12001 }, { "epoch": 116.38072289156626, "grad_norm": 6.530857563018799, "learning_rate": 4.17378640776699e-05, "loss": 0.3153, "step": 12002 }, { "epoch": 116.39036144578313, "grad_norm": 24.85352325439453, "learning_rate": 4.1733009708737864e-05, "loss": 0.1953, "step": 12003 }, { "epoch": 116.4, "grad_norm": 21.153127670288086, "learning_rate": 4.172815533980582e-05, "loss": 0.5291, "step": 12004 }, { "epoch": 116.40963855421687, "grad_norm": 7.6288557052612305, "learning_rate": 4.172330097087379e-05, "loss": 0.1228, "step": 12005 }, { "epoch": 116.41927710843373, "grad_norm": 12.175271987915039, "learning_rate": 4.171844660194175e-05, "loss": 0.2603, "step": 12006 }, { "epoch": 116.4289156626506, "grad_norm": 36.057193756103516, "learning_rate": 4.1713592233009715e-05, "loss": 0.2679, "step": 12007 }, { "epoch": 116.43855421686747, "grad_norm": 30.259145736694336, "learning_rate": 4.170873786407767e-05, "loss": 0.1302, "step": 12008 }, { "epoch": 116.44819277108434, "grad_norm": 8.325743675231934, "learning_rate": 4.170388349514564e-05, "loss": 0.3637, "step": 12009 }, { "epoch": 116.4578313253012, "grad_norm": 9.265752792358398, "learning_rate": 4.1699029126213595e-05, "loss": 0.3405, "step": 12010 }, { "epoch": 116.46746987951808, "grad_norm": 4.613616466522217, "learning_rate": 4.169417475728156e-05, "loss": 0.1385, "step": 12011 }, { "epoch": 116.47710843373494, "grad_norm": 9.947588920593262, "learning_rate": 4.168932038834952e-05, "loss": 0.1858, "step": 12012 }, { "epoch": 116.4867469879518, "grad_norm": 33.374969482421875, "learning_rate": 4.1684466019417476e-05, "loss": 0.5227, "step": 12013 }, { "epoch": 116.49638554216868, "grad_norm": 12.248734474182129, "learning_rate": 4.167961165048544e-05, "loss": 0.4901, "step": 12014 }, { "epoch": 116.50602409638554, "grad_norm": 4.7658915519714355, "learning_rate": 4.16747572815534e-05, "loss": 0.1054, "step": 12015 }, { "epoch": 116.51566265060241, "grad_norm": 14.488542556762695, "learning_rate": 4.166990291262136e-05, "loss": 0.1971, "step": 12016 }, { "epoch": 116.52530120481927, "grad_norm": 22.0743408203125, "learning_rate": 4.166504854368932e-05, "loss": 0.5575, "step": 12017 }, { "epoch": 116.53493975903615, "grad_norm": 5.306410789489746, "learning_rate": 4.1660194174757285e-05, "loss": 0.1971, "step": 12018 }, { "epoch": 116.54457831325301, "grad_norm": 4.3128662109375, "learning_rate": 4.165533980582524e-05, "loss": 0.1385, "step": 12019 }, { "epoch": 116.55421686746988, "grad_norm": 7.594738006591797, "learning_rate": 4.165048543689321e-05, "loss": 0.2762, "step": 12020 }, { "epoch": 116.56385542168675, "grad_norm": 17.39535903930664, "learning_rate": 4.1645631067961165e-05, "loss": 0.5313, "step": 12021 }, { "epoch": 116.57349397590362, "grad_norm": 9.45261287689209, "learning_rate": 4.164077669902913e-05, "loss": 0.3708, "step": 12022 }, { "epoch": 116.58313253012048, "grad_norm": 10.180980682373047, "learning_rate": 4.163592233009709e-05, "loss": 0.1687, "step": 12023 }, { "epoch": 116.59277108433734, "grad_norm": 14.16394329071045, "learning_rate": 4.163106796116505e-05, "loss": 0.1786, "step": 12024 }, { "epoch": 116.60240963855422, "grad_norm": 17.268274307250977, "learning_rate": 4.162621359223301e-05, "loss": 0.5227, "step": 12025 }, { "epoch": 116.61204819277108, "grad_norm": 9.133363723754883, "learning_rate": 4.1621359223300974e-05, "loss": 0.3801, "step": 12026 }, { "epoch": 116.62168674698795, "grad_norm": 20.327064514160156, "learning_rate": 4.161650485436893e-05, "loss": 0.1709, "step": 12027 }, { "epoch": 116.63132530120482, "grad_norm": 12.83711051940918, "learning_rate": 4.161165048543689e-05, "loss": 0.434, "step": 12028 }, { "epoch": 116.64096385542169, "grad_norm": 20.62217903137207, "learning_rate": 4.1606796116504854e-05, "loss": 0.418, "step": 12029 }, { "epoch": 116.65060240963855, "grad_norm": 6.559224605560303, "learning_rate": 4.160194174757282e-05, "loss": 0.196, "step": 12030 }, { "epoch": 116.66024096385541, "grad_norm": 19.907583236694336, "learning_rate": 4.159708737864078e-05, "loss": 0.4693, "step": 12031 }, { "epoch": 116.66987951807229, "grad_norm": 12.387679100036621, "learning_rate": 4.159223300970874e-05, "loss": 0.2882, "step": 12032 }, { "epoch": 116.67951807228916, "grad_norm": 16.528240203857422, "learning_rate": 4.1587378640776705e-05, "loss": 0.4048, "step": 12033 }, { "epoch": 116.68915662650602, "grad_norm": 7.314324855804443, "learning_rate": 4.158252427184466e-05, "loss": 0.2258, "step": 12034 }, { "epoch": 116.6987951807229, "grad_norm": 46.19588851928711, "learning_rate": 4.157766990291263e-05, "loss": 0.3944, "step": 12035 }, { "epoch": 116.70843373493976, "grad_norm": 9.115070343017578, "learning_rate": 4.1572815533980585e-05, "loss": 0.2613, "step": 12036 }, { "epoch": 116.71807228915662, "grad_norm": 10.11548137664795, "learning_rate": 4.156796116504855e-05, "loss": 0.2601, "step": 12037 }, { "epoch": 116.7277108433735, "grad_norm": 17.51605796813965, "learning_rate": 4.156310679611651e-05, "loss": 0.1616, "step": 12038 }, { "epoch": 116.73734939759036, "grad_norm": 22.89325714111328, "learning_rate": 4.1558252427184465e-05, "loss": 0.3427, "step": 12039 }, { "epoch": 116.74698795180723, "grad_norm": 16.214706420898438, "learning_rate": 4.155339805825243e-05, "loss": 0.1089, "step": 12040 }, { "epoch": 116.75662650602409, "grad_norm": 55.28559494018555, "learning_rate": 4.154854368932039e-05, "loss": 0.3715, "step": 12041 }, { "epoch": 116.76626506024097, "grad_norm": 140.01918029785156, "learning_rate": 4.154368932038835e-05, "loss": 0.5396, "step": 12042 }, { "epoch": 116.77590361445783, "grad_norm": 8.186488151550293, "learning_rate": 4.153883495145631e-05, "loss": 0.1731, "step": 12043 }, { "epoch": 116.7855421686747, "grad_norm": 7.423877239227295, "learning_rate": 4.1533980582524274e-05, "loss": 0.2563, "step": 12044 }, { "epoch": 116.79518072289157, "grad_norm": 20.579771041870117, "learning_rate": 4.152912621359223e-05, "loss": 0.2552, "step": 12045 }, { "epoch": 116.80481927710844, "grad_norm": 7.8761820793151855, "learning_rate": 4.15242718446602e-05, "loss": 0.3585, "step": 12046 }, { "epoch": 116.8144578313253, "grad_norm": 16.32866859436035, "learning_rate": 4.1519417475728154e-05, "loss": 0.312, "step": 12047 }, { "epoch": 116.82409638554216, "grad_norm": 34.82905197143555, "learning_rate": 4.151456310679612e-05, "loss": 0.3266, "step": 12048 }, { "epoch": 116.83373493975904, "grad_norm": 24.512680053710938, "learning_rate": 4.150970873786408e-05, "loss": 0.3021, "step": 12049 }, { "epoch": 116.8433734939759, "grad_norm": 14.166984558105469, "learning_rate": 4.150485436893204e-05, "loss": 0.3117, "step": 12050 }, { "epoch": 116.85301204819277, "grad_norm": 27.24882698059082, "learning_rate": 4.15e-05, "loss": 0.2339, "step": 12051 }, { "epoch": 116.86265060240964, "grad_norm": 17.4271183013916, "learning_rate": 4.1495145631067964e-05, "loss": 0.1342, "step": 12052 }, { "epoch": 116.87228915662651, "grad_norm": 5.36591100692749, "learning_rate": 4.149029126213592e-05, "loss": 0.2498, "step": 12053 }, { "epoch": 116.88192771084337, "grad_norm": 16.419652938842773, "learning_rate": 4.1485436893203886e-05, "loss": 0.2699, "step": 12054 }, { "epoch": 116.89156626506023, "grad_norm": 9.438176155090332, "learning_rate": 4.148058252427185e-05, "loss": 0.1818, "step": 12055 }, { "epoch": 116.90120481927711, "grad_norm": 14.656848907470703, "learning_rate": 4.147572815533981e-05, "loss": 0.293, "step": 12056 }, { "epoch": 116.91084337349398, "grad_norm": 8.418658256530762, "learning_rate": 4.147087378640777e-05, "loss": 0.1476, "step": 12057 }, { "epoch": 116.92048192771084, "grad_norm": 16.958084106445312, "learning_rate": 4.146601941747573e-05, "loss": 0.2408, "step": 12058 }, { "epoch": 116.93012048192772, "grad_norm": 109.23967742919922, "learning_rate": 4.1461165048543695e-05, "loss": 0.2828, "step": 12059 }, { "epoch": 116.93975903614458, "grad_norm": 9.019365310668945, "learning_rate": 4.145631067961165e-05, "loss": 0.2124, "step": 12060 }, { "epoch": 116.94939759036144, "grad_norm": 20.950910568237305, "learning_rate": 4.145145631067962e-05, "loss": 0.3263, "step": 12061 }, { "epoch": 116.95903614457832, "grad_norm": 6.205526351928711, "learning_rate": 4.1446601941747575e-05, "loss": 0.2314, "step": 12062 }, { "epoch": 116.96867469879518, "grad_norm": 10.383947372436523, "learning_rate": 4.144174757281554e-05, "loss": 0.2297, "step": 12063 }, { "epoch": 116.97831325301205, "grad_norm": 35.967079162597656, "learning_rate": 4.14368932038835e-05, "loss": 0.4367, "step": 12064 }, { "epoch": 116.98795180722891, "grad_norm": 3.82863187789917, "learning_rate": 4.143203883495146e-05, "loss": 0.0959, "step": 12065 }, { "epoch": 117.00361445783132, "grad_norm": 6.032733917236328, "learning_rate": 4.142718446601942e-05, "loss": 0.0556, "step": 12066 }, { "epoch": 117.0132530120482, "grad_norm": 4.337524890899658, "learning_rate": 4.142233009708738e-05, "loss": 0.2211, "step": 12067 }, { "epoch": 117.02289156626506, "grad_norm": 25.414608001708984, "learning_rate": 4.141747572815534e-05, "loss": 0.2832, "step": 12068 }, { "epoch": 117.03253012048192, "grad_norm": 24.801776885986328, "learning_rate": 4.14126213592233e-05, "loss": 0.438, "step": 12069 }, { "epoch": 117.0421686746988, "grad_norm": 41.892242431640625, "learning_rate": 4.1407766990291264e-05, "loss": 0.243, "step": 12070 }, { "epoch": 117.05180722891566, "grad_norm": 4.503481388092041, "learning_rate": 4.140291262135922e-05, "loss": 0.1188, "step": 12071 }, { "epoch": 117.06144578313253, "grad_norm": 21.76772689819336, "learning_rate": 4.1398058252427187e-05, "loss": 0.2126, "step": 12072 }, { "epoch": 117.0710843373494, "grad_norm": 10.234580039978027, "learning_rate": 4.1393203883495144e-05, "loss": 0.5314, "step": 12073 }, { "epoch": 117.08072289156627, "grad_norm": 32.95438766479492, "learning_rate": 4.138834951456311e-05, "loss": 0.3933, "step": 12074 }, { "epoch": 117.09036144578313, "grad_norm": 13.999115943908691, "learning_rate": 4.1383495145631067e-05, "loss": 0.1969, "step": 12075 }, { "epoch": 117.1, "grad_norm": 10.850041389465332, "learning_rate": 4.137864077669903e-05, "loss": 0.348, "step": 12076 }, { "epoch": 117.10963855421687, "grad_norm": 26.07889175415039, "learning_rate": 4.137378640776699e-05, "loss": 0.1222, "step": 12077 }, { "epoch": 117.11927710843374, "grad_norm": 39.293853759765625, "learning_rate": 4.1368932038834953e-05, "loss": 0.2901, "step": 12078 }, { "epoch": 117.1289156626506, "grad_norm": 9.151451110839844, "learning_rate": 4.136407766990291e-05, "loss": 0.3111, "step": 12079 }, { "epoch": 117.13855421686748, "grad_norm": 23.58949089050293, "learning_rate": 4.1359223300970876e-05, "loss": 0.1471, "step": 12080 }, { "epoch": 117.14819277108434, "grad_norm": 10.253584861755371, "learning_rate": 4.135436893203884e-05, "loss": 0.1777, "step": 12081 }, { "epoch": 117.1578313253012, "grad_norm": 40.8918571472168, "learning_rate": 4.13495145631068e-05, "loss": 0.4026, "step": 12082 }, { "epoch": 117.16746987951807, "grad_norm": 11.878973007202148, "learning_rate": 4.134466019417476e-05, "loss": 0.2742, "step": 12083 }, { "epoch": 117.17710843373494, "grad_norm": 32.94932174682617, "learning_rate": 4.133980582524272e-05, "loss": 0.5215, "step": 12084 }, { "epoch": 117.1867469879518, "grad_norm": 5.308163166046143, "learning_rate": 4.1334951456310685e-05, "loss": 0.0768, "step": 12085 }, { "epoch": 117.19638554216867, "grad_norm": 9.054718971252441, "learning_rate": 4.133009708737864e-05, "loss": 0.345, "step": 12086 }, { "epoch": 117.20602409638555, "grad_norm": 9.829017639160156, "learning_rate": 4.132524271844661e-05, "loss": 0.4202, "step": 12087 }, { "epoch": 117.21566265060241, "grad_norm": 33.978641510009766, "learning_rate": 4.1320388349514565e-05, "loss": 0.3961, "step": 12088 }, { "epoch": 117.22530120481927, "grad_norm": 17.17921257019043, "learning_rate": 4.131553398058253e-05, "loss": 0.2248, "step": 12089 }, { "epoch": 117.23493975903614, "grad_norm": 11.035411834716797, "learning_rate": 4.131067961165049e-05, "loss": 0.257, "step": 12090 }, { "epoch": 117.24457831325302, "grad_norm": 13.424345970153809, "learning_rate": 4.130582524271845e-05, "loss": 0.3638, "step": 12091 }, { "epoch": 117.25421686746988, "grad_norm": 8.011465072631836, "learning_rate": 4.130097087378641e-05, "loss": 0.4558, "step": 12092 }, { "epoch": 117.26385542168674, "grad_norm": 25.24043083190918, "learning_rate": 4.1296116504854374e-05, "loss": 0.4079, "step": 12093 }, { "epoch": 117.27349397590362, "grad_norm": 6.108923435211182, "learning_rate": 4.129126213592233e-05, "loss": 0.1663, "step": 12094 }, { "epoch": 117.28313253012048, "grad_norm": 9.601156234741211, "learning_rate": 4.128640776699029e-05, "loss": 0.282, "step": 12095 }, { "epoch": 117.29277108433735, "grad_norm": 12.47964859008789, "learning_rate": 4.1281553398058254e-05, "loss": 0.3795, "step": 12096 }, { "epoch": 117.30240963855422, "grad_norm": 23.588197708129883, "learning_rate": 4.127669902912621e-05, "loss": 0.3293, "step": 12097 }, { "epoch": 117.31204819277109, "grad_norm": 43.302818298339844, "learning_rate": 4.1271844660194176e-05, "loss": 0.3637, "step": 12098 }, { "epoch": 117.32168674698795, "grad_norm": 31.18783187866211, "learning_rate": 4.1266990291262134e-05, "loss": 0.4553, "step": 12099 }, { "epoch": 117.33132530120481, "grad_norm": 6.914172649383545, "learning_rate": 4.12621359223301e-05, "loss": 0.1961, "step": 12100 }, { "epoch": 117.34096385542169, "grad_norm": 9.082833290100098, "learning_rate": 4.1257281553398056e-05, "loss": 0.2046, "step": 12101 }, { "epoch": 117.35060240963855, "grad_norm": 5.2683491706848145, "learning_rate": 4.125242718446602e-05, "loss": 0.1307, "step": 12102 }, { "epoch": 117.36024096385542, "grad_norm": 39.67319869995117, "learning_rate": 4.124757281553398e-05, "loss": 0.527, "step": 12103 }, { "epoch": 117.3698795180723, "grad_norm": 25.609167098999023, "learning_rate": 4.124271844660194e-05, "loss": 0.3383, "step": 12104 }, { "epoch": 117.37951807228916, "grad_norm": 8.34344482421875, "learning_rate": 4.12378640776699e-05, "loss": 0.2626, "step": 12105 }, { "epoch": 117.38915662650602, "grad_norm": 9.436599731445312, "learning_rate": 4.1233009708737865e-05, "loss": 0.1876, "step": 12106 }, { "epoch": 117.39879518072289, "grad_norm": 14.71201229095459, "learning_rate": 4.122815533980583e-05, "loss": 0.1856, "step": 12107 }, { "epoch": 117.40843373493976, "grad_norm": 7.823895454406738, "learning_rate": 4.122330097087379e-05, "loss": 0.1852, "step": 12108 }, { "epoch": 117.41807228915663, "grad_norm": 15.514516830444336, "learning_rate": 4.121844660194175e-05, "loss": 0.358, "step": 12109 }, { "epoch": 117.42771084337349, "grad_norm": 23.778575897216797, "learning_rate": 4.121359223300971e-05, "loss": 0.1929, "step": 12110 }, { "epoch": 117.43734939759037, "grad_norm": 32.157066345214844, "learning_rate": 4.1208737864077675e-05, "loss": 0.3421, "step": 12111 }, { "epoch": 117.44698795180723, "grad_norm": 16.258996963500977, "learning_rate": 4.120388349514563e-05, "loss": 0.3194, "step": 12112 }, { "epoch": 117.4566265060241, "grad_norm": 6.823450088500977, "learning_rate": 4.11990291262136e-05, "loss": 0.1679, "step": 12113 }, { "epoch": 117.46626506024096, "grad_norm": 4.111721992492676, "learning_rate": 4.1194174757281555e-05, "loss": 0.0914, "step": 12114 }, { "epoch": 117.47590361445783, "grad_norm": 7.877783298492432, "learning_rate": 4.118932038834952e-05, "loss": 0.0816, "step": 12115 }, { "epoch": 117.4855421686747, "grad_norm": 7.950739860534668, "learning_rate": 4.118446601941748e-05, "loss": 0.1336, "step": 12116 }, { "epoch": 117.49518072289156, "grad_norm": 24.69839096069336, "learning_rate": 4.117961165048544e-05, "loss": 0.1943, "step": 12117 }, { "epoch": 117.50481927710844, "grad_norm": 11.578377723693848, "learning_rate": 4.11747572815534e-05, "loss": 0.358, "step": 12118 }, { "epoch": 117.5144578313253, "grad_norm": 9.641468048095703, "learning_rate": 4.1169902912621364e-05, "loss": 0.5429, "step": 12119 }, { "epoch": 117.52409638554217, "grad_norm": 23.920047760009766, "learning_rate": 4.116504854368932e-05, "loss": 0.085, "step": 12120 }, { "epoch": 117.53373493975904, "grad_norm": 4.563467025756836, "learning_rate": 4.1160194174757286e-05, "loss": 0.2054, "step": 12121 }, { "epoch": 117.5433734939759, "grad_norm": 9.678982734680176, "learning_rate": 4.1155339805825244e-05, "loss": 0.2296, "step": 12122 }, { "epoch": 117.55301204819277, "grad_norm": 10.9262113571167, "learning_rate": 4.11504854368932e-05, "loss": 0.1808, "step": 12123 }, { "epoch": 117.56265060240963, "grad_norm": 10.658702850341797, "learning_rate": 4.1145631067961166e-05, "loss": 0.5549, "step": 12124 }, { "epoch": 117.57228915662651, "grad_norm": 11.411858558654785, "learning_rate": 4.1140776699029124e-05, "loss": 0.1131, "step": 12125 }, { "epoch": 117.58192771084337, "grad_norm": 15.49441909790039, "learning_rate": 4.113592233009709e-05, "loss": 0.3861, "step": 12126 }, { "epoch": 117.59156626506024, "grad_norm": 7.719172954559326, "learning_rate": 4.1131067961165046e-05, "loss": 0.1961, "step": 12127 }, { "epoch": 117.60120481927711, "grad_norm": 24.298969268798828, "learning_rate": 4.112621359223301e-05, "loss": 0.2956, "step": 12128 }, { "epoch": 117.61084337349398, "grad_norm": 4.9531683921813965, "learning_rate": 4.112135922330097e-05, "loss": 0.1301, "step": 12129 }, { "epoch": 117.62048192771084, "grad_norm": 8.538772583007812, "learning_rate": 4.111650485436893e-05, "loss": 0.1944, "step": 12130 }, { "epoch": 117.6301204819277, "grad_norm": 13.75998306274414, "learning_rate": 4.11116504854369e-05, "loss": 0.2961, "step": 12131 }, { "epoch": 117.63975903614458, "grad_norm": 16.940641403198242, "learning_rate": 4.110679611650486e-05, "loss": 0.2383, "step": 12132 }, { "epoch": 117.64939759036145, "grad_norm": 13.922576904296875, "learning_rate": 4.110194174757282e-05, "loss": 0.2856, "step": 12133 }, { "epoch": 117.65903614457831, "grad_norm": 12.972450256347656, "learning_rate": 4.109708737864078e-05, "loss": 0.2816, "step": 12134 }, { "epoch": 117.66867469879519, "grad_norm": 9.124469757080078, "learning_rate": 4.109223300970874e-05, "loss": 0.2965, "step": 12135 }, { "epoch": 117.67831325301205, "grad_norm": 38.997840881347656, "learning_rate": 4.10873786407767e-05, "loss": 0.1351, "step": 12136 }, { "epoch": 117.68795180722891, "grad_norm": 3.3888556957244873, "learning_rate": 4.1082524271844664e-05, "loss": 0.3088, "step": 12137 }, { "epoch": 117.69759036144578, "grad_norm": 28.847015380859375, "learning_rate": 4.107766990291262e-05, "loss": 0.4205, "step": 12138 }, { "epoch": 117.70722891566265, "grad_norm": 20.302677154541016, "learning_rate": 4.107281553398059e-05, "loss": 0.3447, "step": 12139 }, { "epoch": 117.71686746987952, "grad_norm": 13.201155662536621, "learning_rate": 4.1067961165048544e-05, "loss": 0.3799, "step": 12140 }, { "epoch": 117.72650602409638, "grad_norm": 8.299612998962402, "learning_rate": 4.106310679611651e-05, "loss": 0.2088, "step": 12141 }, { "epoch": 117.73614457831326, "grad_norm": 29.513124465942383, "learning_rate": 4.105825242718447e-05, "loss": 0.4238, "step": 12142 }, { "epoch": 117.74578313253012, "grad_norm": 23.7038516998291, "learning_rate": 4.105339805825243e-05, "loss": 0.5112, "step": 12143 }, { "epoch": 117.75542168674698, "grad_norm": 14.856120109558105, "learning_rate": 4.104854368932039e-05, "loss": 0.2736, "step": 12144 }, { "epoch": 117.76506024096386, "grad_norm": 13.435189247131348, "learning_rate": 4.1043689320388354e-05, "loss": 0.3369, "step": 12145 }, { "epoch": 117.77469879518073, "grad_norm": 6.927172660827637, "learning_rate": 4.103883495145631e-05, "loss": 0.214, "step": 12146 }, { "epoch": 117.78433734939759, "grad_norm": 17.036781311035156, "learning_rate": 4.1033980582524276e-05, "loss": 0.1835, "step": 12147 }, { "epoch": 117.79397590361445, "grad_norm": 28.851388931274414, "learning_rate": 4.1029126213592234e-05, "loss": 0.2473, "step": 12148 }, { "epoch": 117.80361445783133, "grad_norm": 106.86985778808594, "learning_rate": 4.10242718446602e-05, "loss": 0.3673, "step": 12149 }, { "epoch": 117.8132530120482, "grad_norm": 5.5102033615112305, "learning_rate": 4.1019417475728156e-05, "loss": 0.2743, "step": 12150 }, { "epoch": 117.82289156626506, "grad_norm": 7.884368419647217, "learning_rate": 4.1014563106796114e-05, "loss": 0.2512, "step": 12151 }, { "epoch": 117.83253012048193, "grad_norm": 33.964134216308594, "learning_rate": 4.100970873786408e-05, "loss": 0.3605, "step": 12152 }, { "epoch": 117.8421686746988, "grad_norm": 26.26729965209961, "learning_rate": 4.1004854368932036e-05, "loss": 0.3916, "step": 12153 }, { "epoch": 117.85180722891566, "grad_norm": 13.843655586242676, "learning_rate": 4.1e-05, "loss": 0.261, "step": 12154 }, { "epoch": 117.86144578313252, "grad_norm": 6.093344211578369, "learning_rate": 4.099514563106796e-05, "loss": 0.122, "step": 12155 }, { "epoch": 117.8710843373494, "grad_norm": 10.848584175109863, "learning_rate": 4.099029126213593e-05, "loss": 0.2728, "step": 12156 }, { "epoch": 117.88072289156626, "grad_norm": 36.96853256225586, "learning_rate": 4.098543689320389e-05, "loss": 0.3508, "step": 12157 }, { "epoch": 117.89036144578313, "grad_norm": 8.715523719787598, "learning_rate": 4.098058252427185e-05, "loss": 0.2425, "step": 12158 }, { "epoch": 117.9, "grad_norm": 11.20429515838623, "learning_rate": 4.097572815533981e-05, "loss": 0.1392, "step": 12159 }, { "epoch": 117.90963855421687, "grad_norm": 6.161810874938965, "learning_rate": 4.0970873786407774e-05, "loss": 0.2057, "step": 12160 }, { "epoch": 117.91927710843373, "grad_norm": 12.52878189086914, "learning_rate": 4.096601941747573e-05, "loss": 0.2635, "step": 12161 }, { "epoch": 117.9289156626506, "grad_norm": 9.776705741882324, "learning_rate": 4.096116504854369e-05, "loss": 0.2069, "step": 12162 }, { "epoch": 117.93855421686747, "grad_norm": 9.135459899902344, "learning_rate": 4.0956310679611654e-05, "loss": 0.2876, "step": 12163 }, { "epoch": 117.94819277108434, "grad_norm": 9.380006790161133, "learning_rate": 4.095145631067961e-05, "loss": 0.2058, "step": 12164 }, { "epoch": 117.9578313253012, "grad_norm": 100.90375518798828, "learning_rate": 4.0946601941747576e-05, "loss": 0.3576, "step": 12165 }, { "epoch": 117.96746987951808, "grad_norm": 8.317312240600586, "learning_rate": 4.0941747572815534e-05, "loss": 0.1507, "step": 12166 }, { "epoch": 117.97710843373494, "grad_norm": 10.472789764404297, "learning_rate": 4.09368932038835e-05, "loss": 0.2356, "step": 12167 }, { "epoch": 117.9867469879518, "grad_norm": 9.374580383300781, "learning_rate": 4.0932038834951457e-05, "loss": 0.2603, "step": 12168 }, { "epoch": 118.00240963855421, "grad_norm": 17.02294921875, "learning_rate": 4.092718446601942e-05, "loss": 0.2895, "step": 12169 }, { "epoch": 118.01204819277109, "grad_norm": 5.362137317657471, "learning_rate": 4.092233009708738e-05, "loss": 0.0972, "step": 12170 }, { "epoch": 118.02168674698795, "grad_norm": 84.6498794555664, "learning_rate": 4.091747572815534e-05, "loss": 0.2914, "step": 12171 }, { "epoch": 118.03132530120482, "grad_norm": 14.844651222229004, "learning_rate": 4.09126213592233e-05, "loss": 0.527, "step": 12172 }, { "epoch": 118.04096385542168, "grad_norm": 8.39428997039795, "learning_rate": 4.0907766990291266e-05, "loss": 0.3094, "step": 12173 }, { "epoch": 118.05060240963856, "grad_norm": 47.789676666259766, "learning_rate": 4.0902912621359223e-05, "loss": 0.3594, "step": 12174 }, { "epoch": 118.06024096385542, "grad_norm": 13.974617958068848, "learning_rate": 4.089805825242719e-05, "loss": 0.2094, "step": 12175 }, { "epoch": 118.06987951807228, "grad_norm": 4.060812473297119, "learning_rate": 4.0893203883495146e-05, "loss": 0.1561, "step": 12176 }, { "epoch": 118.07951807228916, "grad_norm": 5.109011650085449, "learning_rate": 4.0888349514563103e-05, "loss": 0.3582, "step": 12177 }, { "epoch": 118.08915662650602, "grad_norm": 3.536860942840576, "learning_rate": 4.088349514563107e-05, "loss": 0.3482, "step": 12178 }, { "epoch": 118.09879518072289, "grad_norm": 11.753256797790527, "learning_rate": 4.0878640776699026e-05, "loss": 0.2119, "step": 12179 }, { "epoch": 118.10843373493977, "grad_norm": 24.406030654907227, "learning_rate": 4.087378640776699e-05, "loss": 0.2328, "step": 12180 }, { "epoch": 118.11807228915663, "grad_norm": 13.895880699157715, "learning_rate": 4.0868932038834955e-05, "loss": 0.2848, "step": 12181 }, { "epoch": 118.12771084337349, "grad_norm": 10.2649564743042, "learning_rate": 4.086407766990292e-05, "loss": 0.1309, "step": 12182 }, { "epoch": 118.13734939759036, "grad_norm": 31.714454650878906, "learning_rate": 4.085922330097088e-05, "loss": 0.1553, "step": 12183 }, { "epoch": 118.14698795180723, "grad_norm": 5.228475570678711, "learning_rate": 4.085436893203884e-05, "loss": 0.1945, "step": 12184 }, { "epoch": 118.1566265060241, "grad_norm": 11.457921981811523, "learning_rate": 4.08495145631068e-05, "loss": 0.3168, "step": 12185 }, { "epoch": 118.16626506024096, "grad_norm": 9.427989959716797, "learning_rate": 4.0844660194174764e-05, "loss": 0.1384, "step": 12186 }, { "epoch": 118.17590361445784, "grad_norm": 18.486318588256836, "learning_rate": 4.083980582524272e-05, "loss": 0.3282, "step": 12187 }, { "epoch": 118.1855421686747, "grad_norm": 3.828765869140625, "learning_rate": 4.083495145631068e-05, "loss": 0.07, "step": 12188 }, { "epoch": 118.19518072289156, "grad_norm": 2.617494583129883, "learning_rate": 4.0830097087378644e-05, "loss": 0.0994, "step": 12189 }, { "epoch": 118.20481927710843, "grad_norm": 7.928809642791748, "learning_rate": 4.08252427184466e-05, "loss": 0.3668, "step": 12190 }, { "epoch": 118.2144578313253, "grad_norm": 7.746540069580078, "learning_rate": 4.0820388349514566e-05, "loss": 0.2296, "step": 12191 }, { "epoch": 118.22409638554217, "grad_norm": 8.83425235748291, "learning_rate": 4.0815533980582524e-05, "loss": 0.1062, "step": 12192 }, { "epoch": 118.23373493975903, "grad_norm": 17.949867248535156, "learning_rate": 4.081067961165049e-05, "loss": 0.4212, "step": 12193 }, { "epoch": 118.24337349397591, "grad_norm": 7.784251689910889, "learning_rate": 4.0805825242718446e-05, "loss": 0.1658, "step": 12194 }, { "epoch": 118.25301204819277, "grad_norm": 14.186003684997559, "learning_rate": 4.080097087378641e-05, "loss": 0.1057, "step": 12195 }, { "epoch": 118.26265060240964, "grad_norm": 10.999797821044922, "learning_rate": 4.079611650485437e-05, "loss": 0.3411, "step": 12196 }, { "epoch": 118.2722891566265, "grad_norm": 14.55268669128418, "learning_rate": 4.079126213592233e-05, "loss": 0.4612, "step": 12197 }, { "epoch": 118.28192771084338, "grad_norm": 6.853182315826416, "learning_rate": 4.078640776699029e-05, "loss": 0.3568, "step": 12198 }, { "epoch": 118.29156626506024, "grad_norm": 26.509275436401367, "learning_rate": 4.0781553398058255e-05, "loss": 0.3792, "step": 12199 }, { "epoch": 118.3012048192771, "grad_norm": 8.222095489501953, "learning_rate": 4.077669902912621e-05, "loss": 0.3843, "step": 12200 }, { "epoch": 118.31084337349398, "grad_norm": 16.609418869018555, "learning_rate": 4.077184466019418e-05, "loss": 0.3995, "step": 12201 }, { "epoch": 118.32048192771084, "grad_norm": 9.537190437316895, "learning_rate": 4.0766990291262135e-05, "loss": 0.236, "step": 12202 }, { "epoch": 118.33012048192771, "grad_norm": 10.758931159973145, "learning_rate": 4.07621359223301e-05, "loss": 0.2609, "step": 12203 }, { "epoch": 118.33975903614459, "grad_norm": 20.87431526184082, "learning_rate": 4.075728155339806e-05, "loss": 0.3382, "step": 12204 }, { "epoch": 118.34939759036145, "grad_norm": 7.129364967346191, "learning_rate": 4.0752427184466016e-05, "loss": 0.2592, "step": 12205 }, { "epoch": 118.35903614457831, "grad_norm": 7.443317890167236, "learning_rate": 4.074757281553398e-05, "loss": 0.2269, "step": 12206 }, { "epoch": 118.36867469879518, "grad_norm": 13.409738540649414, "learning_rate": 4.0742718446601945e-05, "loss": 0.2512, "step": 12207 }, { "epoch": 118.37831325301205, "grad_norm": 16.735376358032227, "learning_rate": 4.073786407766991e-05, "loss": 0.4737, "step": 12208 }, { "epoch": 118.38795180722892, "grad_norm": 11.316569328308105, "learning_rate": 4.073300970873787e-05, "loss": 0.4858, "step": 12209 }, { "epoch": 118.39759036144578, "grad_norm": 21.486112594604492, "learning_rate": 4.072815533980583e-05, "loss": 0.2507, "step": 12210 }, { "epoch": 118.40722891566266, "grad_norm": 2.6651899814605713, "learning_rate": 4.072330097087379e-05, "loss": 0.1001, "step": 12211 }, { "epoch": 118.41686746987952, "grad_norm": 10.16530990600586, "learning_rate": 4.0718446601941754e-05, "loss": 0.1484, "step": 12212 }, { "epoch": 118.42650602409638, "grad_norm": 49.275634765625, "learning_rate": 4.071359223300971e-05, "loss": 0.2924, "step": 12213 }, { "epoch": 118.43614457831325, "grad_norm": 1.882811427116394, "learning_rate": 4.0708737864077676e-05, "loss": 0.0605, "step": 12214 }, { "epoch": 118.44578313253012, "grad_norm": 85.38188934326172, "learning_rate": 4.0703883495145634e-05, "loss": 0.36, "step": 12215 }, { "epoch": 118.45542168674699, "grad_norm": 7.406490325927734, "learning_rate": 4.069902912621359e-05, "loss": 0.3369, "step": 12216 }, { "epoch": 118.46506024096385, "grad_norm": 12.851248741149902, "learning_rate": 4.0694174757281556e-05, "loss": 0.4032, "step": 12217 }, { "epoch": 118.47469879518073, "grad_norm": 13.509451866149902, "learning_rate": 4.0689320388349514e-05, "loss": 0.1348, "step": 12218 }, { "epoch": 118.48433734939759, "grad_norm": 32.25581359863281, "learning_rate": 4.068446601941748e-05, "loss": 0.3141, "step": 12219 }, { "epoch": 118.49397590361446, "grad_norm": 19.633750915527344, "learning_rate": 4.0679611650485436e-05, "loss": 0.3169, "step": 12220 }, { "epoch": 118.50361445783132, "grad_norm": 6.953036785125732, "learning_rate": 4.06747572815534e-05, "loss": 0.3253, "step": 12221 }, { "epoch": 118.5132530120482, "grad_norm": 15.191242218017578, "learning_rate": 4.066990291262136e-05, "loss": 0.4004, "step": 12222 }, { "epoch": 118.52289156626506, "grad_norm": 10.11046314239502, "learning_rate": 4.066504854368932e-05, "loss": 0.2181, "step": 12223 }, { "epoch": 118.53253012048192, "grad_norm": 9.362147331237793, "learning_rate": 4.066019417475728e-05, "loss": 0.1471, "step": 12224 }, { "epoch": 118.5421686746988, "grad_norm": 19.1507568359375, "learning_rate": 4.0655339805825245e-05, "loss": 0.2467, "step": 12225 }, { "epoch": 118.55180722891566, "grad_norm": 18.04278564453125, "learning_rate": 4.06504854368932e-05, "loss": 0.263, "step": 12226 }, { "epoch": 118.56144578313253, "grad_norm": 3.923862934112549, "learning_rate": 4.064563106796117e-05, "loss": 0.1532, "step": 12227 }, { "epoch": 118.5710843373494, "grad_norm": 22.72645378112793, "learning_rate": 4.0640776699029125e-05, "loss": 0.342, "step": 12228 }, { "epoch": 118.58072289156627, "grad_norm": 12.1283540725708, "learning_rate": 4.063592233009709e-05, "loss": 0.3599, "step": 12229 }, { "epoch": 118.59036144578313, "grad_norm": 10.482503890991211, "learning_rate": 4.063106796116505e-05, "loss": 0.2025, "step": 12230 }, { "epoch": 118.6, "grad_norm": 9.826837539672852, "learning_rate": 4.062621359223301e-05, "loss": 0.3431, "step": 12231 }, { "epoch": 118.60963855421687, "grad_norm": 18.880937576293945, "learning_rate": 4.062135922330098e-05, "loss": 0.3423, "step": 12232 }, { "epoch": 118.61927710843374, "grad_norm": 7.021444797515869, "learning_rate": 4.0616504854368934e-05, "loss": 0.4836, "step": 12233 }, { "epoch": 118.6289156626506, "grad_norm": 13.275586128234863, "learning_rate": 4.06116504854369e-05, "loss": 0.277, "step": 12234 }, { "epoch": 118.63855421686748, "grad_norm": 19.855213165283203, "learning_rate": 4.060679611650486e-05, "loss": 0.2311, "step": 12235 }, { "epoch": 118.64819277108434, "grad_norm": 15.30740737915039, "learning_rate": 4.060194174757282e-05, "loss": 0.4664, "step": 12236 }, { "epoch": 118.6578313253012, "grad_norm": 5.802502632141113, "learning_rate": 4.059708737864078e-05, "loss": 0.2447, "step": 12237 }, { "epoch": 118.66746987951807, "grad_norm": 9.446958541870117, "learning_rate": 4.0592233009708743e-05, "loss": 0.2591, "step": 12238 }, { "epoch": 118.67710843373494, "grad_norm": 9.698037147521973, "learning_rate": 4.05873786407767e-05, "loss": 0.3156, "step": 12239 }, { "epoch": 118.6867469879518, "grad_norm": 16.887842178344727, "learning_rate": 4.0582524271844666e-05, "loss": 0.2308, "step": 12240 }, { "epoch": 118.69638554216867, "grad_norm": 4.5991668701171875, "learning_rate": 4.0577669902912624e-05, "loss": 0.1715, "step": 12241 }, { "epoch": 118.70602409638555, "grad_norm": 17.93478775024414, "learning_rate": 4.057281553398059e-05, "loss": 0.2151, "step": 12242 }, { "epoch": 118.71566265060241, "grad_norm": 11.252588272094727, "learning_rate": 4.0567961165048546e-05, "loss": 0.2793, "step": 12243 }, { "epoch": 118.72530120481927, "grad_norm": 18.723678588867188, "learning_rate": 4.0563106796116504e-05, "loss": 0.2393, "step": 12244 }, { "epoch": 118.73493975903614, "grad_norm": 7.649742603302002, "learning_rate": 4.055825242718447e-05, "loss": 0.2505, "step": 12245 }, { "epoch": 118.74457831325302, "grad_norm": 17.323925018310547, "learning_rate": 4.0553398058252426e-05, "loss": 0.2167, "step": 12246 }, { "epoch": 118.75421686746988, "grad_norm": 7.6105427742004395, "learning_rate": 4.054854368932039e-05, "loss": 0.1036, "step": 12247 }, { "epoch": 118.76385542168674, "grad_norm": 17.510744094848633, "learning_rate": 4.054368932038835e-05, "loss": 0.2764, "step": 12248 }, { "epoch": 118.77349397590362, "grad_norm": 12.130193710327148, "learning_rate": 4.053883495145631e-05, "loss": 0.2467, "step": 12249 }, { "epoch": 118.78313253012048, "grad_norm": 7.386335372924805, "learning_rate": 4.053398058252427e-05, "loss": 0.4824, "step": 12250 }, { "epoch": 118.79277108433735, "grad_norm": 7.666455268859863, "learning_rate": 4.0529126213592235e-05, "loss": 0.2803, "step": 12251 }, { "epoch": 118.80240963855422, "grad_norm": 14.872435569763184, "learning_rate": 4.052427184466019e-05, "loss": 0.4919, "step": 12252 }, { "epoch": 118.81204819277109, "grad_norm": 18.89453125, "learning_rate": 4.051941747572816e-05, "loss": 0.3974, "step": 12253 }, { "epoch": 118.82168674698795, "grad_norm": 8.4214448928833, "learning_rate": 4.0514563106796115e-05, "loss": 0.3922, "step": 12254 }, { "epoch": 118.83132530120481, "grad_norm": 12.551487922668457, "learning_rate": 4.050970873786408e-05, "loss": 0.1814, "step": 12255 }, { "epoch": 118.84096385542169, "grad_norm": 21.606369018554688, "learning_rate": 4.050485436893204e-05, "loss": 0.2541, "step": 12256 }, { "epoch": 118.85060240963855, "grad_norm": 33.004329681396484, "learning_rate": 4.05e-05, "loss": 0.3636, "step": 12257 }, { "epoch": 118.86024096385542, "grad_norm": 25.9858455657959, "learning_rate": 4.0495145631067966e-05, "loss": 0.2549, "step": 12258 }, { "epoch": 118.8698795180723, "grad_norm": 6.423616409301758, "learning_rate": 4.0490291262135924e-05, "loss": 0.1867, "step": 12259 }, { "epoch": 118.87951807228916, "grad_norm": 5.695998191833496, "learning_rate": 4.048543689320389e-05, "loss": 0.1161, "step": 12260 }, { "epoch": 118.88915662650602, "grad_norm": 14.357406616210938, "learning_rate": 4.0480582524271846e-05, "loss": 0.2108, "step": 12261 }, { "epoch": 118.89879518072289, "grad_norm": 6.277403831481934, "learning_rate": 4.047572815533981e-05, "loss": 0.2911, "step": 12262 }, { "epoch": 118.90843373493976, "grad_norm": 6.959203720092773, "learning_rate": 4.047087378640777e-05, "loss": 0.1944, "step": 12263 }, { "epoch": 118.91807228915663, "grad_norm": 14.038780212402344, "learning_rate": 4.046601941747573e-05, "loss": 0.2303, "step": 12264 }, { "epoch": 118.92771084337349, "grad_norm": 6.788216590881348, "learning_rate": 4.046116504854369e-05, "loss": 0.1889, "step": 12265 }, { "epoch": 118.93734939759037, "grad_norm": 6.872410774230957, "learning_rate": 4.0456310679611656e-05, "loss": 0.3901, "step": 12266 }, { "epoch": 118.94698795180723, "grad_norm": 9.062179565429688, "learning_rate": 4.045145631067961e-05, "loss": 0.392, "step": 12267 }, { "epoch": 118.9566265060241, "grad_norm": 7.87485933303833, "learning_rate": 4.044660194174758e-05, "loss": 0.279, "step": 12268 }, { "epoch": 118.96626506024096, "grad_norm": 12.942323684692383, "learning_rate": 4.0441747572815536e-05, "loss": 0.2731, "step": 12269 }, { "epoch": 118.97590361445783, "grad_norm": 10.789268493652344, "learning_rate": 4.04368932038835e-05, "loss": 0.1854, "step": 12270 }, { "epoch": 118.9855421686747, "grad_norm": 14.317121505737305, "learning_rate": 4.043203883495146e-05, "loss": 0.2398, "step": 12271 }, { "epoch": 119.0012048192771, "grad_norm": 11.737020492553711, "learning_rate": 4.0427184466019416e-05, "loss": 0.1952, "step": 12272 }, { "epoch": 119.01084337349397, "grad_norm": 22.468029022216797, "learning_rate": 4.042233009708738e-05, "loss": 0.2785, "step": 12273 }, { "epoch": 119.02048192771085, "grad_norm": 7.969353675842285, "learning_rate": 4.041747572815534e-05, "loss": 0.3287, "step": 12274 }, { "epoch": 119.03012048192771, "grad_norm": 18.418127059936523, "learning_rate": 4.04126213592233e-05, "loss": 0.342, "step": 12275 }, { "epoch": 119.03975903614457, "grad_norm": 6.997652530670166, "learning_rate": 4.040776699029126e-05, "loss": 0.1783, "step": 12276 }, { "epoch": 119.04939759036145, "grad_norm": 12.677051544189453, "learning_rate": 4.0402912621359225e-05, "loss": 0.2973, "step": 12277 }, { "epoch": 119.05903614457831, "grad_norm": 11.611520767211914, "learning_rate": 4.039805825242718e-05, "loss": 0.323, "step": 12278 }, { "epoch": 119.06867469879518, "grad_norm": 24.765850067138672, "learning_rate": 4.039320388349515e-05, "loss": 0.1174, "step": 12279 }, { "epoch": 119.07831325301204, "grad_norm": 11.21017074584961, "learning_rate": 4.0388349514563105e-05, "loss": 0.4437, "step": 12280 }, { "epoch": 119.08795180722892, "grad_norm": 38.141319274902344, "learning_rate": 4.038349514563107e-05, "loss": 0.3167, "step": 12281 }, { "epoch": 119.09759036144578, "grad_norm": 9.39698314666748, "learning_rate": 4.0378640776699034e-05, "loss": 0.2248, "step": 12282 }, { "epoch": 119.10722891566265, "grad_norm": 23.692665100097656, "learning_rate": 4.037378640776699e-05, "loss": 0.2389, "step": 12283 }, { "epoch": 119.11686746987952, "grad_norm": 11.653297424316406, "learning_rate": 4.0368932038834956e-05, "loss": 0.2156, "step": 12284 }, { "epoch": 119.12650602409639, "grad_norm": 16.786762237548828, "learning_rate": 4.0364077669902914e-05, "loss": 0.4421, "step": 12285 }, { "epoch": 119.13614457831325, "grad_norm": 14.686807632446289, "learning_rate": 4.035922330097088e-05, "loss": 0.3976, "step": 12286 }, { "epoch": 119.14578313253013, "grad_norm": 9.048186302185059, "learning_rate": 4.0354368932038836e-05, "loss": 0.3483, "step": 12287 }, { "epoch": 119.15542168674699, "grad_norm": 42.889095306396484, "learning_rate": 4.03495145631068e-05, "loss": 0.325, "step": 12288 }, { "epoch": 119.16506024096385, "grad_norm": 9.212735176086426, "learning_rate": 4.034466019417476e-05, "loss": 0.3912, "step": 12289 }, { "epoch": 119.17469879518072, "grad_norm": 12.466646194458008, "learning_rate": 4.033980582524272e-05, "loss": 0.3025, "step": 12290 }, { "epoch": 119.1843373493976, "grad_norm": 16.823837280273438, "learning_rate": 4.033495145631068e-05, "loss": 0.1493, "step": 12291 }, { "epoch": 119.19397590361446, "grad_norm": 51.36661148071289, "learning_rate": 4.0330097087378645e-05, "loss": 0.3622, "step": 12292 }, { "epoch": 119.20361445783132, "grad_norm": 7.192019462585449, "learning_rate": 4.03252427184466e-05, "loss": 0.1582, "step": 12293 }, { "epoch": 119.2132530120482, "grad_norm": 9.872361183166504, "learning_rate": 4.032038834951457e-05, "loss": 0.4073, "step": 12294 }, { "epoch": 119.22289156626506, "grad_norm": 17.076045989990234, "learning_rate": 4.0315533980582525e-05, "loss": 0.1473, "step": 12295 }, { "epoch": 119.23253012048193, "grad_norm": 21.852642059326172, "learning_rate": 4.031067961165049e-05, "loss": 0.3938, "step": 12296 }, { "epoch": 119.24216867469879, "grad_norm": 12.833606719970703, "learning_rate": 4.030582524271845e-05, "loss": 0.1675, "step": 12297 }, { "epoch": 119.25180722891567, "grad_norm": 12.911393165588379, "learning_rate": 4.030097087378641e-05, "loss": 0.2833, "step": 12298 }, { "epoch": 119.26144578313253, "grad_norm": 20.80387306213379, "learning_rate": 4.029611650485437e-05, "loss": 0.1925, "step": 12299 }, { "epoch": 119.2710843373494, "grad_norm": 7.321310520172119, "learning_rate": 4.029126213592233e-05, "loss": 0.3292, "step": 12300 }, { "epoch": 119.28072289156627, "grad_norm": 13.233144760131836, "learning_rate": 4.028640776699029e-05, "loss": 0.3378, "step": 12301 }, { "epoch": 119.29036144578313, "grad_norm": 12.603589057922363, "learning_rate": 4.028155339805825e-05, "loss": 0.4434, "step": 12302 }, { "epoch": 119.3, "grad_norm": 9.77486515045166, "learning_rate": 4.0276699029126215e-05, "loss": 0.3955, "step": 12303 }, { "epoch": 119.30963855421686, "grad_norm": 16.809904098510742, "learning_rate": 4.027184466019417e-05, "loss": 0.2506, "step": 12304 }, { "epoch": 119.31927710843374, "grad_norm": 11.492880821228027, "learning_rate": 4.026699029126214e-05, "loss": 0.379, "step": 12305 }, { "epoch": 119.3289156626506, "grad_norm": 4.880222320556641, "learning_rate": 4.0262135922330095e-05, "loss": 0.1103, "step": 12306 }, { "epoch": 119.33855421686746, "grad_norm": 8.532788276672363, "learning_rate": 4.0257281553398066e-05, "loss": 0.3719, "step": 12307 }, { "epoch": 119.34819277108434, "grad_norm": 23.05105209350586, "learning_rate": 4.0252427184466024e-05, "loss": 0.2524, "step": 12308 }, { "epoch": 119.3578313253012, "grad_norm": 12.735943794250488, "learning_rate": 4.024757281553399e-05, "loss": 0.321, "step": 12309 }, { "epoch": 119.36746987951807, "grad_norm": 13.639213562011719, "learning_rate": 4.0242718446601946e-05, "loss": 0.3647, "step": 12310 }, { "epoch": 119.37710843373495, "grad_norm": 9.11318302154541, "learning_rate": 4.0237864077669904e-05, "loss": 0.1731, "step": 12311 }, { "epoch": 119.38674698795181, "grad_norm": 13.065736770629883, "learning_rate": 4.023300970873787e-05, "loss": 0.3051, "step": 12312 }, { "epoch": 119.39638554216867, "grad_norm": 11.753718376159668, "learning_rate": 4.0228155339805826e-05, "loss": 0.4694, "step": 12313 }, { "epoch": 119.40602409638554, "grad_norm": 4.0159502029418945, "learning_rate": 4.022330097087379e-05, "loss": 0.2836, "step": 12314 }, { "epoch": 119.41566265060241, "grad_norm": 8.840221405029297, "learning_rate": 4.021844660194175e-05, "loss": 0.2608, "step": 12315 }, { "epoch": 119.42530120481928, "grad_norm": 6.751776218414307, "learning_rate": 4.021359223300971e-05, "loss": 0.0739, "step": 12316 }, { "epoch": 119.43493975903614, "grad_norm": 10.89313793182373, "learning_rate": 4.020873786407767e-05, "loss": 0.3134, "step": 12317 }, { "epoch": 119.44457831325302, "grad_norm": 6.104720115661621, "learning_rate": 4.0203883495145635e-05, "loss": 0.2141, "step": 12318 }, { "epoch": 119.45421686746988, "grad_norm": 33.90352249145508, "learning_rate": 4.019902912621359e-05, "loss": 0.2166, "step": 12319 }, { "epoch": 119.46385542168674, "grad_norm": 7.1049323081970215, "learning_rate": 4.019417475728156e-05, "loss": 0.1806, "step": 12320 }, { "epoch": 119.47349397590361, "grad_norm": 75.889404296875, "learning_rate": 4.0189320388349515e-05, "loss": 0.3877, "step": 12321 }, { "epoch": 119.48313253012049, "grad_norm": 12.714088439941406, "learning_rate": 4.018446601941748e-05, "loss": 0.2279, "step": 12322 }, { "epoch": 119.49277108433735, "grad_norm": 17.9986572265625, "learning_rate": 4.017961165048544e-05, "loss": 0.2582, "step": 12323 }, { "epoch": 119.50240963855421, "grad_norm": 10.611750602722168, "learning_rate": 4.01747572815534e-05, "loss": 0.2326, "step": 12324 }, { "epoch": 119.51204819277109, "grad_norm": 3.674668312072754, "learning_rate": 4.016990291262136e-05, "loss": 0.1445, "step": 12325 }, { "epoch": 119.52168674698795, "grad_norm": 44.45125198364258, "learning_rate": 4.016504854368932e-05, "loss": 0.3689, "step": 12326 }, { "epoch": 119.53132530120482, "grad_norm": 12.497824668884277, "learning_rate": 4.016019417475728e-05, "loss": 0.2008, "step": 12327 }, { "epoch": 119.54096385542168, "grad_norm": 9.81850528717041, "learning_rate": 4.015533980582524e-05, "loss": 0.2693, "step": 12328 }, { "epoch": 119.55060240963856, "grad_norm": 17.819299697875977, "learning_rate": 4.0150485436893204e-05, "loss": 0.1032, "step": 12329 }, { "epoch": 119.56024096385542, "grad_norm": 8.207907676696777, "learning_rate": 4.014563106796116e-05, "loss": 0.2893, "step": 12330 }, { "epoch": 119.56987951807228, "grad_norm": 6.355217456817627, "learning_rate": 4.014077669902913e-05, "loss": 0.2334, "step": 12331 }, { "epoch": 119.57951807228916, "grad_norm": 6.337828159332275, "learning_rate": 4.0135922330097084e-05, "loss": 0.1527, "step": 12332 }, { "epoch": 119.58915662650602, "grad_norm": 9.97314167022705, "learning_rate": 4.0131067961165056e-05, "loss": 0.2949, "step": 12333 }, { "epoch": 119.59879518072289, "grad_norm": 12.01513957977295, "learning_rate": 4.0126213592233013e-05, "loss": 0.4074, "step": 12334 }, { "epoch": 119.60843373493977, "grad_norm": 4.962522029876709, "learning_rate": 4.012135922330098e-05, "loss": 0.1569, "step": 12335 }, { "epoch": 119.61807228915663, "grad_norm": 7.335052013397217, "learning_rate": 4.0116504854368936e-05, "loss": 0.3296, "step": 12336 }, { "epoch": 119.62771084337349, "grad_norm": 9.47095012664795, "learning_rate": 4.0111650485436894e-05, "loss": 0.1601, "step": 12337 }, { "epoch": 119.63734939759036, "grad_norm": 10.153072357177734, "learning_rate": 4.010679611650486e-05, "loss": 0.2331, "step": 12338 }, { "epoch": 119.64698795180723, "grad_norm": 4.638310432434082, "learning_rate": 4.0101941747572816e-05, "loss": 0.2038, "step": 12339 }, { "epoch": 119.6566265060241, "grad_norm": 5.100026607513428, "learning_rate": 4.009708737864078e-05, "loss": 0.2657, "step": 12340 }, { "epoch": 119.66626506024096, "grad_norm": 10.541115760803223, "learning_rate": 4.009223300970874e-05, "loss": 0.3765, "step": 12341 }, { "epoch": 119.67590361445784, "grad_norm": 20.30765724182129, "learning_rate": 4.00873786407767e-05, "loss": 0.293, "step": 12342 }, { "epoch": 119.6855421686747, "grad_norm": 21.19735336303711, "learning_rate": 4.008252427184466e-05, "loss": 0.2671, "step": 12343 }, { "epoch": 119.69518072289156, "grad_norm": 5.290363788604736, "learning_rate": 4.0077669902912625e-05, "loss": 0.1524, "step": 12344 }, { "epoch": 119.70481927710843, "grad_norm": 16.181074142456055, "learning_rate": 4.007281553398058e-05, "loss": 0.3275, "step": 12345 }, { "epoch": 119.7144578313253, "grad_norm": 8.012035369873047, "learning_rate": 4.006796116504855e-05, "loss": 0.2288, "step": 12346 }, { "epoch": 119.72409638554217, "grad_norm": 15.35506534576416, "learning_rate": 4.0063106796116505e-05, "loss": 0.3456, "step": 12347 }, { "epoch": 119.73373493975903, "grad_norm": 9.966805458068848, "learning_rate": 4.005825242718447e-05, "loss": 0.2253, "step": 12348 }, { "epoch": 119.74337349397591, "grad_norm": 9.666563987731934, "learning_rate": 4.005339805825243e-05, "loss": 0.4424, "step": 12349 }, { "epoch": 119.75301204819277, "grad_norm": 11.16111946105957, "learning_rate": 4.004854368932039e-05, "loss": 0.4191, "step": 12350 }, { "epoch": 119.76265060240964, "grad_norm": 4.460242748260498, "learning_rate": 4.004368932038835e-05, "loss": 0.2063, "step": 12351 }, { "epoch": 119.7722891566265, "grad_norm": 12.923641204833984, "learning_rate": 4.0038834951456314e-05, "loss": 0.2501, "step": 12352 }, { "epoch": 119.78192771084338, "grad_norm": 8.124080657958984, "learning_rate": 4.003398058252427e-05, "loss": 0.2637, "step": 12353 }, { "epoch": 119.79156626506024, "grad_norm": 10.243680953979492, "learning_rate": 4.002912621359223e-05, "loss": 0.371, "step": 12354 }, { "epoch": 119.8012048192771, "grad_norm": 10.19451904296875, "learning_rate": 4.0024271844660194e-05, "loss": 0.2237, "step": 12355 }, { "epoch": 119.81084337349398, "grad_norm": 19.787033081054688, "learning_rate": 4.001941747572815e-05, "loss": 0.189, "step": 12356 }, { "epoch": 119.82048192771084, "grad_norm": 20.147865295410156, "learning_rate": 4.0014563106796116e-05, "loss": 0.2688, "step": 12357 }, { "epoch": 119.83012048192771, "grad_norm": 12.715520858764648, "learning_rate": 4.000970873786408e-05, "loss": 0.2522, "step": 12358 }, { "epoch": 119.83975903614459, "grad_norm": 18.559589385986328, "learning_rate": 4.0004854368932046e-05, "loss": 0.342, "step": 12359 }, { "epoch": 119.84939759036145, "grad_norm": 25.534656524658203, "learning_rate": 4e-05, "loss": 0.372, "step": 12360 }, { "epoch": 119.85903614457831, "grad_norm": 7.865384101867676, "learning_rate": 3.999514563106797e-05, "loss": 0.2011, "step": 12361 }, { "epoch": 119.86867469879518, "grad_norm": 7.342955112457275, "learning_rate": 3.9990291262135926e-05, "loss": 0.1895, "step": 12362 }, { "epoch": 119.87831325301205, "grad_norm": 18.430709838867188, "learning_rate": 3.998543689320389e-05, "loss": 0.2039, "step": 12363 }, { "epoch": 119.88795180722892, "grad_norm": 5.460830211639404, "learning_rate": 3.998058252427185e-05, "loss": 0.2811, "step": 12364 }, { "epoch": 119.89759036144578, "grad_norm": 25.525184631347656, "learning_rate": 3.9975728155339806e-05, "loss": 0.5573, "step": 12365 }, { "epoch": 119.90722891566266, "grad_norm": 10.724197387695312, "learning_rate": 3.997087378640777e-05, "loss": 0.3294, "step": 12366 }, { "epoch": 119.91686746987952, "grad_norm": 13.33545970916748, "learning_rate": 3.996601941747573e-05, "loss": 0.2878, "step": 12367 }, { "epoch": 119.92650602409638, "grad_norm": 13.050432205200195, "learning_rate": 3.996116504854369e-05, "loss": 0.2939, "step": 12368 }, { "epoch": 119.93614457831325, "grad_norm": 18.121299743652344, "learning_rate": 3.995631067961165e-05, "loss": 0.2984, "step": 12369 }, { "epoch": 119.94578313253012, "grad_norm": 9.061697006225586, "learning_rate": 3.9951456310679615e-05, "loss": 0.1552, "step": 12370 }, { "epoch": 119.95542168674699, "grad_norm": 13.205906867980957, "learning_rate": 3.994660194174757e-05, "loss": 0.3699, "step": 12371 }, { "epoch": 119.96506024096385, "grad_norm": 11.416777610778809, "learning_rate": 3.994174757281554e-05, "loss": 0.4603, "step": 12372 }, { "epoch": 119.97469879518073, "grad_norm": 22.041954040527344, "learning_rate": 3.9936893203883495e-05, "loss": 0.3634, "step": 12373 }, { "epoch": 119.98433734939759, "grad_norm": 8.402372360229492, "learning_rate": 3.993203883495146e-05, "loss": 0.23, "step": 12374 }, { "epoch": 119.99397590361446, "grad_norm": 32.26437759399414, "learning_rate": 3.992718446601942e-05, "loss": 0.2466, "step": 12375 }, { "epoch": 120.00963855421686, "grad_norm": 13.076411247253418, "learning_rate": 3.992233009708738e-05, "loss": 0.1862, "step": 12376 }, { "epoch": 120.01927710843374, "grad_norm": 10.010045051574707, "learning_rate": 3.991747572815534e-05, "loss": 0.1066, "step": 12377 }, { "epoch": 120.0289156626506, "grad_norm": 7.701850891113281, "learning_rate": 3.9912621359223304e-05, "loss": 0.1401, "step": 12378 }, { "epoch": 120.03855421686747, "grad_norm": 16.128421783447266, "learning_rate": 3.990776699029126e-05, "loss": 0.3893, "step": 12379 }, { "epoch": 120.04819277108433, "grad_norm": 8.693714141845703, "learning_rate": 3.9902912621359226e-05, "loss": 0.3589, "step": 12380 }, { "epoch": 120.05783132530121, "grad_norm": 9.27918529510498, "learning_rate": 3.9898058252427184e-05, "loss": 0.2448, "step": 12381 }, { "epoch": 120.06746987951807, "grad_norm": 7.68749475479126, "learning_rate": 3.989320388349514e-05, "loss": 0.34, "step": 12382 }, { "epoch": 120.07710843373494, "grad_norm": 13.177156448364258, "learning_rate": 3.988834951456311e-05, "loss": 0.1953, "step": 12383 }, { "epoch": 120.08674698795181, "grad_norm": 20.960857391357422, "learning_rate": 3.988349514563107e-05, "loss": 0.3721, "step": 12384 }, { "epoch": 120.09638554216868, "grad_norm": 14.176597595214844, "learning_rate": 3.9878640776699035e-05, "loss": 0.3213, "step": 12385 }, { "epoch": 120.10602409638554, "grad_norm": 9.186463356018066, "learning_rate": 3.987378640776699e-05, "loss": 0.327, "step": 12386 }, { "epoch": 120.1156626506024, "grad_norm": 9.921523094177246, "learning_rate": 3.986893203883496e-05, "loss": 0.1655, "step": 12387 }, { "epoch": 120.12530120481928, "grad_norm": 18.96363067626953, "learning_rate": 3.9864077669902915e-05, "loss": 0.2442, "step": 12388 }, { "epoch": 120.13493975903614, "grad_norm": 13.695733070373535, "learning_rate": 3.985922330097088e-05, "loss": 0.3099, "step": 12389 }, { "epoch": 120.144578313253, "grad_norm": 18.632707595825195, "learning_rate": 3.985436893203884e-05, "loss": 0.3904, "step": 12390 }, { "epoch": 120.15421686746988, "grad_norm": 237.08450317382812, "learning_rate": 3.98495145631068e-05, "loss": 0.2581, "step": 12391 }, { "epoch": 120.16385542168675, "grad_norm": 16.394474029541016, "learning_rate": 3.984466019417476e-05, "loss": 0.408, "step": 12392 }, { "epoch": 120.17349397590361, "grad_norm": 10.43988037109375, "learning_rate": 3.983980582524272e-05, "loss": 0.2728, "step": 12393 }, { "epoch": 120.18313253012049, "grad_norm": 8.70490837097168, "learning_rate": 3.983495145631068e-05, "loss": 0.2614, "step": 12394 }, { "epoch": 120.19277108433735, "grad_norm": 7.321126461029053, "learning_rate": 3.983009708737864e-05, "loss": 0.1288, "step": 12395 }, { "epoch": 120.20240963855422, "grad_norm": 7.4343976974487305, "learning_rate": 3.9825242718446605e-05, "loss": 0.3304, "step": 12396 }, { "epoch": 120.21204819277108, "grad_norm": 4.600132942199707, "learning_rate": 3.982038834951456e-05, "loss": 0.1048, "step": 12397 }, { "epoch": 120.22168674698796, "grad_norm": 10.724410057067871, "learning_rate": 3.981553398058253e-05, "loss": 0.2424, "step": 12398 }, { "epoch": 120.23132530120482, "grad_norm": 8.888888359069824, "learning_rate": 3.9810679611650485e-05, "loss": 0.3493, "step": 12399 }, { "epoch": 120.24096385542168, "grad_norm": 9.425471305847168, "learning_rate": 3.980582524271845e-05, "loss": 0.3551, "step": 12400 }, { "epoch": 120.25060240963856, "grad_norm": 12.742926597595215, "learning_rate": 3.980097087378641e-05, "loss": 0.294, "step": 12401 }, { "epoch": 120.26024096385542, "grad_norm": 12.40538215637207, "learning_rate": 3.979611650485437e-05, "loss": 0.2253, "step": 12402 }, { "epoch": 120.26987951807229, "grad_norm": 13.660125732421875, "learning_rate": 3.979126213592233e-05, "loss": 0.3434, "step": 12403 }, { "epoch": 120.27951807228915, "grad_norm": 8.598139762878418, "learning_rate": 3.9786407766990294e-05, "loss": 0.3263, "step": 12404 }, { "epoch": 120.28915662650603, "grad_norm": 4.61914587020874, "learning_rate": 3.978155339805825e-05, "loss": 0.0944, "step": 12405 }, { "epoch": 120.29879518072289, "grad_norm": 8.367226600646973, "learning_rate": 3.9776699029126216e-05, "loss": 0.2062, "step": 12406 }, { "epoch": 120.30843373493975, "grad_norm": 8.565376281738281, "learning_rate": 3.9771844660194174e-05, "loss": 0.2795, "step": 12407 }, { "epoch": 120.31807228915663, "grad_norm": 2.6633925437927246, "learning_rate": 3.976699029126214e-05, "loss": 0.0817, "step": 12408 }, { "epoch": 120.3277108433735, "grad_norm": 4.829123497009277, "learning_rate": 3.97621359223301e-05, "loss": 0.0902, "step": 12409 }, { "epoch": 120.33734939759036, "grad_norm": 13.377155303955078, "learning_rate": 3.975728155339806e-05, "loss": 0.3362, "step": 12410 }, { "epoch": 120.34698795180722, "grad_norm": 7.260237693786621, "learning_rate": 3.9752427184466025e-05, "loss": 0.2036, "step": 12411 }, { "epoch": 120.3566265060241, "grad_norm": 6.6827778816223145, "learning_rate": 3.974757281553398e-05, "loss": 0.1612, "step": 12412 }, { "epoch": 120.36626506024096, "grad_norm": 14.690691947937012, "learning_rate": 3.974271844660195e-05, "loss": 0.2656, "step": 12413 }, { "epoch": 120.37590361445783, "grad_norm": 8.082752227783203, "learning_rate": 3.9737864077669905e-05, "loss": 0.3456, "step": 12414 }, { "epoch": 120.3855421686747, "grad_norm": 15.007895469665527, "learning_rate": 3.973300970873787e-05, "loss": 0.4286, "step": 12415 }, { "epoch": 120.39518072289157, "grad_norm": 6.357365608215332, "learning_rate": 3.972815533980583e-05, "loss": 0.1988, "step": 12416 }, { "epoch": 120.40481927710843, "grad_norm": 7.497210502624512, "learning_rate": 3.972330097087379e-05, "loss": 0.2226, "step": 12417 }, { "epoch": 120.41445783132531, "grad_norm": 13.085694313049316, "learning_rate": 3.971844660194175e-05, "loss": 0.3534, "step": 12418 }, { "epoch": 120.42409638554217, "grad_norm": 8.344110488891602, "learning_rate": 3.9713592233009714e-05, "loss": 0.4435, "step": 12419 }, { "epoch": 120.43373493975903, "grad_norm": 13.548430442810059, "learning_rate": 3.970873786407767e-05, "loss": 0.3153, "step": 12420 }, { "epoch": 120.4433734939759, "grad_norm": 16.5249080657959, "learning_rate": 3.970388349514563e-05, "loss": 0.2887, "step": 12421 }, { "epoch": 120.45301204819278, "grad_norm": 10.283411026000977, "learning_rate": 3.9699029126213594e-05, "loss": 0.3427, "step": 12422 }, { "epoch": 120.46265060240964, "grad_norm": 12.58071231842041, "learning_rate": 3.969417475728155e-05, "loss": 0.3682, "step": 12423 }, { "epoch": 120.4722891566265, "grad_norm": 7.668917179107666, "learning_rate": 3.9689320388349517e-05, "loss": 0.2468, "step": 12424 }, { "epoch": 120.48192771084338, "grad_norm": 15.532549858093262, "learning_rate": 3.9684466019417474e-05, "loss": 0.4121, "step": 12425 }, { "epoch": 120.49156626506024, "grad_norm": 23.495512008666992, "learning_rate": 3.967961165048544e-05, "loss": 0.5062, "step": 12426 }, { "epoch": 120.5012048192771, "grad_norm": 8.236943244934082, "learning_rate": 3.96747572815534e-05, "loss": 0.1971, "step": 12427 }, { "epoch": 120.51084337349397, "grad_norm": 14.36717414855957, "learning_rate": 3.966990291262136e-05, "loss": 0.4716, "step": 12428 }, { "epoch": 120.52048192771085, "grad_norm": 6.524432182312012, "learning_rate": 3.966504854368932e-05, "loss": 0.2193, "step": 12429 }, { "epoch": 120.53012048192771, "grad_norm": 15.70709228515625, "learning_rate": 3.9660194174757283e-05, "loss": 0.2821, "step": 12430 }, { "epoch": 120.53975903614457, "grad_norm": 9.914348602294922, "learning_rate": 3.965533980582524e-05, "loss": 0.146, "step": 12431 }, { "epoch": 120.54939759036145, "grad_norm": 10.032519340515137, "learning_rate": 3.9650485436893206e-05, "loss": 0.2708, "step": 12432 }, { "epoch": 120.55903614457831, "grad_norm": 14.102745056152344, "learning_rate": 3.9645631067961164e-05, "loss": 0.429, "step": 12433 }, { "epoch": 120.56867469879518, "grad_norm": 11.613273620605469, "learning_rate": 3.964077669902913e-05, "loss": 0.3428, "step": 12434 }, { "epoch": 120.57831325301204, "grad_norm": 15.438554763793945, "learning_rate": 3.963592233009709e-05, "loss": 0.4533, "step": 12435 }, { "epoch": 120.58795180722892, "grad_norm": 13.083329200744629, "learning_rate": 3.963106796116505e-05, "loss": 0.2477, "step": 12436 }, { "epoch": 120.59759036144578, "grad_norm": 6.666476726531982, "learning_rate": 3.9626213592233015e-05, "loss": 0.2278, "step": 12437 }, { "epoch": 120.60722891566265, "grad_norm": 8.211467742919922, "learning_rate": 3.962135922330097e-05, "loss": 0.1895, "step": 12438 }, { "epoch": 120.61686746987952, "grad_norm": 7.668907642364502, "learning_rate": 3.961650485436894e-05, "loss": 0.1872, "step": 12439 }, { "epoch": 120.62650602409639, "grad_norm": 12.714082717895508, "learning_rate": 3.9611650485436895e-05, "loss": 0.2422, "step": 12440 }, { "epoch": 120.63614457831325, "grad_norm": 25.014144897460938, "learning_rate": 3.960679611650486e-05, "loss": 0.398, "step": 12441 }, { "epoch": 120.64578313253013, "grad_norm": 18.1350040435791, "learning_rate": 3.960194174757282e-05, "loss": 0.2539, "step": 12442 }, { "epoch": 120.65542168674699, "grad_norm": 12.15112590789795, "learning_rate": 3.959708737864078e-05, "loss": 0.4554, "step": 12443 }, { "epoch": 120.66506024096385, "grad_norm": 9.108036041259766, "learning_rate": 3.959223300970874e-05, "loss": 0.3743, "step": 12444 }, { "epoch": 120.67469879518072, "grad_norm": 7.503164291381836, "learning_rate": 3.9587378640776704e-05, "loss": 0.3624, "step": 12445 }, { "epoch": 120.6843373493976, "grad_norm": 20.54108238220215, "learning_rate": 3.958252427184466e-05, "loss": 0.3506, "step": 12446 }, { "epoch": 120.69397590361446, "grad_norm": 7.423421382904053, "learning_rate": 3.9577669902912626e-05, "loss": 0.2249, "step": 12447 }, { "epoch": 120.70361445783132, "grad_norm": 6.8880109786987305, "learning_rate": 3.9572815533980584e-05, "loss": 0.2744, "step": 12448 }, { "epoch": 120.7132530120482, "grad_norm": 12.219152450561523, "learning_rate": 3.956796116504854e-05, "loss": 0.3523, "step": 12449 }, { "epoch": 120.72289156626506, "grad_norm": 15.972725868225098, "learning_rate": 3.9563106796116506e-05, "loss": 0.3889, "step": 12450 }, { "epoch": 120.73253012048193, "grad_norm": 8.169877052307129, "learning_rate": 3.9558252427184464e-05, "loss": 0.203, "step": 12451 }, { "epoch": 120.74216867469879, "grad_norm": 10.9303617477417, "learning_rate": 3.955339805825243e-05, "loss": 0.2214, "step": 12452 }, { "epoch": 120.75180722891567, "grad_norm": 9.0469388961792, "learning_rate": 3.9548543689320386e-05, "loss": 0.1833, "step": 12453 }, { "epoch": 120.76144578313253, "grad_norm": 14.901047706604004, "learning_rate": 3.954368932038835e-05, "loss": 0.323, "step": 12454 }, { "epoch": 120.7710843373494, "grad_norm": 10.264725685119629, "learning_rate": 3.953883495145631e-05, "loss": 0.3744, "step": 12455 }, { "epoch": 120.78072289156627, "grad_norm": 9.86967658996582, "learning_rate": 3.953398058252427e-05, "loss": 0.2928, "step": 12456 }, { "epoch": 120.79036144578313, "grad_norm": 20.472795486450195, "learning_rate": 3.952912621359223e-05, "loss": 0.284, "step": 12457 }, { "epoch": 120.8, "grad_norm": 9.69858169555664, "learning_rate": 3.9524271844660196e-05, "loss": 0.1929, "step": 12458 }, { "epoch": 120.80963855421686, "grad_norm": 14.138067245483398, "learning_rate": 3.951941747572816e-05, "loss": 0.3764, "step": 12459 }, { "epoch": 120.81927710843374, "grad_norm": 16.5246639251709, "learning_rate": 3.951456310679612e-05, "loss": 0.3337, "step": 12460 }, { "epoch": 120.8289156626506, "grad_norm": 8.393622398376465, "learning_rate": 3.950970873786408e-05, "loss": 0.3261, "step": 12461 }, { "epoch": 120.83855421686746, "grad_norm": 8.540488243103027, "learning_rate": 3.950485436893204e-05, "loss": 0.2659, "step": 12462 }, { "epoch": 120.84819277108434, "grad_norm": 15.241377830505371, "learning_rate": 3.9500000000000005e-05, "loss": 0.4294, "step": 12463 }, { "epoch": 120.8578313253012, "grad_norm": 21.088973999023438, "learning_rate": 3.949514563106796e-05, "loss": 0.281, "step": 12464 }, { "epoch": 120.86746987951807, "grad_norm": 18.382902145385742, "learning_rate": 3.949029126213593e-05, "loss": 0.3751, "step": 12465 }, { "epoch": 120.87710843373495, "grad_norm": 34.954063415527344, "learning_rate": 3.9485436893203885e-05, "loss": 0.3799, "step": 12466 }, { "epoch": 120.88674698795181, "grad_norm": 27.12335968017578, "learning_rate": 3.948058252427185e-05, "loss": 0.1739, "step": 12467 }, { "epoch": 120.89638554216867, "grad_norm": 10.86914348602295, "learning_rate": 3.947572815533981e-05, "loss": 0.335, "step": 12468 }, { "epoch": 120.90602409638554, "grad_norm": 9.357386589050293, "learning_rate": 3.947087378640777e-05, "loss": 0.2629, "step": 12469 }, { "epoch": 120.91566265060241, "grad_norm": 15.99091625213623, "learning_rate": 3.946601941747573e-05, "loss": 0.4933, "step": 12470 }, { "epoch": 120.92530120481928, "grad_norm": 17.17771339416504, "learning_rate": 3.9461165048543694e-05, "loss": 0.2801, "step": 12471 }, { "epoch": 120.93493975903614, "grad_norm": 6.823934555053711, "learning_rate": 3.945631067961165e-05, "loss": 0.2115, "step": 12472 }, { "epoch": 120.94457831325302, "grad_norm": 15.088339805603027, "learning_rate": 3.9451456310679616e-05, "loss": 0.2296, "step": 12473 }, { "epoch": 120.95421686746988, "grad_norm": 10.407918930053711, "learning_rate": 3.9446601941747574e-05, "loss": 0.5455, "step": 12474 }, { "epoch": 120.96385542168674, "grad_norm": 15.1309175491333, "learning_rate": 3.944174757281553e-05, "loss": 0.2069, "step": 12475 }, { "epoch": 120.97349397590361, "grad_norm": 10.813590049743652, "learning_rate": 3.9436893203883496e-05, "loss": 0.2244, "step": 12476 }, { "epoch": 120.98313253012049, "grad_norm": 2.925908088684082, "learning_rate": 3.9432038834951454e-05, "loss": 0.0899, "step": 12477 }, { "epoch": 120.99277108433735, "grad_norm": 10.506664276123047, "learning_rate": 3.942718446601942e-05, "loss": 0.2466, "step": 12478 }, { "epoch": 121.00843373493976, "grad_norm": 40.675025939941406, "learning_rate": 3.9422330097087376e-05, "loss": 0.3193, "step": 12479 }, { "epoch": 121.01807228915662, "grad_norm": 35.35495376586914, "learning_rate": 3.941747572815534e-05, "loss": 0.3445, "step": 12480 }, { "epoch": 121.0277108433735, "grad_norm": 2.181734800338745, "learning_rate": 3.94126213592233e-05, "loss": 0.1779, "step": 12481 }, { "epoch": 121.03734939759036, "grad_norm": 3.974637269973755, "learning_rate": 3.940776699029126e-05, "loss": 0.1678, "step": 12482 }, { "epoch": 121.04698795180722, "grad_norm": 13.066960334777832, "learning_rate": 3.940291262135922e-05, "loss": 0.417, "step": 12483 }, { "epoch": 121.0566265060241, "grad_norm": 23.613386154174805, "learning_rate": 3.939805825242719e-05, "loss": 0.3515, "step": 12484 }, { "epoch": 121.06626506024097, "grad_norm": 22.311922073364258, "learning_rate": 3.939320388349515e-05, "loss": 0.3427, "step": 12485 }, { "epoch": 121.07590361445783, "grad_norm": 5.182096004486084, "learning_rate": 3.938834951456311e-05, "loss": 0.1339, "step": 12486 }, { "epoch": 121.08554216867469, "grad_norm": 14.178266525268555, "learning_rate": 3.938349514563107e-05, "loss": 0.2734, "step": 12487 }, { "epoch": 121.09518072289157, "grad_norm": 29.22556495666504, "learning_rate": 3.937864077669903e-05, "loss": 0.2744, "step": 12488 }, { "epoch": 121.10481927710843, "grad_norm": 6.6714911460876465, "learning_rate": 3.9373786407766994e-05, "loss": 0.2971, "step": 12489 }, { "epoch": 121.1144578313253, "grad_norm": 23.02271270751953, "learning_rate": 3.936893203883495e-05, "loss": 0.1574, "step": 12490 }, { "epoch": 121.12409638554217, "grad_norm": 22.89400863647461, "learning_rate": 3.936407766990292e-05, "loss": 0.2116, "step": 12491 }, { "epoch": 121.13373493975904, "grad_norm": 6.6745500564575195, "learning_rate": 3.9359223300970875e-05, "loss": 0.2043, "step": 12492 }, { "epoch": 121.1433734939759, "grad_norm": 0.9138469696044922, "learning_rate": 3.935436893203884e-05, "loss": 0.2515, "step": 12493 }, { "epoch": 121.15301204819278, "grad_norm": 26.588441848754883, "learning_rate": 3.93495145631068e-05, "loss": 0.3753, "step": 12494 }, { "epoch": 121.16265060240964, "grad_norm": 39.5140380859375, "learning_rate": 3.934466019417476e-05, "loss": 0.1533, "step": 12495 }, { "epoch": 121.1722891566265, "grad_norm": 9.305403709411621, "learning_rate": 3.933980582524272e-05, "loss": 0.1546, "step": 12496 }, { "epoch": 121.18192771084337, "grad_norm": 17.048755645751953, "learning_rate": 3.9334951456310684e-05, "loss": 0.1721, "step": 12497 }, { "epoch": 121.19156626506025, "grad_norm": 10.2518310546875, "learning_rate": 3.933009708737864e-05, "loss": 0.2237, "step": 12498 }, { "epoch": 121.20120481927711, "grad_norm": 4.431344032287598, "learning_rate": 3.9325242718446606e-05, "loss": 0.1434, "step": 12499 }, { "epoch": 121.21084337349397, "grad_norm": 37.945762634277344, "learning_rate": 3.9320388349514564e-05, "loss": 0.3805, "step": 12500 }, { "epoch": 121.22048192771085, "grad_norm": 45.75213623046875, "learning_rate": 3.931553398058253e-05, "loss": 0.315, "step": 12501 }, { "epoch": 121.23012048192771, "grad_norm": 41.363773345947266, "learning_rate": 3.9310679611650486e-05, "loss": 0.3304, "step": 12502 }, { "epoch": 121.23975903614458, "grad_norm": 6.531567573547363, "learning_rate": 3.9305825242718444e-05, "loss": 0.2367, "step": 12503 }, { "epoch": 121.24939759036144, "grad_norm": 18.785432815551758, "learning_rate": 3.930097087378641e-05, "loss": 0.3707, "step": 12504 }, { "epoch": 121.25903614457832, "grad_norm": 27.399940490722656, "learning_rate": 3.9296116504854366e-05, "loss": 0.2165, "step": 12505 }, { "epoch": 121.26867469879518, "grad_norm": 33.2125244140625, "learning_rate": 3.929126213592233e-05, "loss": 0.2795, "step": 12506 }, { "epoch": 121.27831325301204, "grad_norm": 9.95695686340332, "learning_rate": 3.928640776699029e-05, "loss": 0.1096, "step": 12507 }, { "epoch": 121.28795180722892, "grad_norm": 23.060007095336914, "learning_rate": 3.928155339805825e-05, "loss": 0.2717, "step": 12508 }, { "epoch": 121.29759036144578, "grad_norm": 29.699480056762695, "learning_rate": 3.927669902912622e-05, "loss": 0.2537, "step": 12509 }, { "epoch": 121.30722891566265, "grad_norm": 7.0530290603637695, "learning_rate": 3.927184466019418e-05, "loss": 0.2162, "step": 12510 }, { "epoch": 121.31686746987951, "grad_norm": 4.907631874084473, "learning_rate": 3.926699029126214e-05, "loss": 0.2528, "step": 12511 }, { "epoch": 121.32650602409639, "grad_norm": 32.75107955932617, "learning_rate": 3.9262135922330104e-05, "loss": 0.1942, "step": 12512 }, { "epoch": 121.33614457831325, "grad_norm": 40.32016372680664, "learning_rate": 3.925728155339806e-05, "loss": 0.2987, "step": 12513 }, { "epoch": 121.34578313253012, "grad_norm": 19.767412185668945, "learning_rate": 3.925242718446602e-05, "loss": 0.2968, "step": 12514 }, { "epoch": 121.355421686747, "grad_norm": 35.977848052978516, "learning_rate": 3.9247572815533984e-05, "loss": 0.176, "step": 12515 }, { "epoch": 121.36506024096386, "grad_norm": 16.476730346679688, "learning_rate": 3.924271844660194e-05, "loss": 0.3463, "step": 12516 }, { "epoch": 121.37469879518072, "grad_norm": 50.07108688354492, "learning_rate": 3.9237864077669907e-05, "loss": 0.2874, "step": 12517 }, { "epoch": 121.38433734939758, "grad_norm": 17.37525749206543, "learning_rate": 3.9233009708737864e-05, "loss": 0.2302, "step": 12518 }, { "epoch": 121.39397590361446, "grad_norm": 24.141695022583008, "learning_rate": 3.922815533980583e-05, "loss": 0.1953, "step": 12519 }, { "epoch": 121.40361445783132, "grad_norm": 0.9970692992210388, "learning_rate": 3.9223300970873787e-05, "loss": 0.1198, "step": 12520 }, { "epoch": 121.41325301204819, "grad_norm": 10.94897174835205, "learning_rate": 3.921844660194175e-05, "loss": 0.2214, "step": 12521 }, { "epoch": 121.42289156626506, "grad_norm": 21.516735076904297, "learning_rate": 3.921359223300971e-05, "loss": 0.4489, "step": 12522 }, { "epoch": 121.43253012048193, "grad_norm": 9.550699234008789, "learning_rate": 3.9208737864077673e-05, "loss": 0.2811, "step": 12523 }, { "epoch": 121.44216867469879, "grad_norm": 4.492860317230225, "learning_rate": 3.920388349514563e-05, "loss": 0.2146, "step": 12524 }, { "epoch": 121.45180722891567, "grad_norm": 7.176884174346924, "learning_rate": 3.9199029126213596e-05, "loss": 0.3046, "step": 12525 }, { "epoch": 121.46144578313253, "grad_norm": 25.344005584716797, "learning_rate": 3.9194174757281553e-05, "loss": 0.4553, "step": 12526 }, { "epoch": 121.4710843373494, "grad_norm": 27.943313598632812, "learning_rate": 3.918932038834952e-05, "loss": 0.336, "step": 12527 }, { "epoch": 121.48072289156626, "grad_norm": 7.612053394317627, "learning_rate": 3.9184466019417476e-05, "loss": 0.2349, "step": 12528 }, { "epoch": 121.49036144578314, "grad_norm": 11.54809856414795, "learning_rate": 3.917961165048544e-05, "loss": 0.3242, "step": 12529 }, { "epoch": 121.5, "grad_norm": 9.96517562866211, "learning_rate": 3.91747572815534e-05, "loss": 0.1713, "step": 12530 }, { "epoch": 121.50963855421686, "grad_norm": 7.754334449768066, "learning_rate": 3.9169902912621356e-05, "loss": 0.4455, "step": 12531 }, { "epoch": 121.51927710843374, "grad_norm": 10.552295684814453, "learning_rate": 3.916504854368932e-05, "loss": 0.2711, "step": 12532 }, { "epoch": 121.5289156626506, "grad_norm": 41.471107482910156, "learning_rate": 3.916019417475728e-05, "loss": 0.2217, "step": 12533 }, { "epoch": 121.53855421686747, "grad_norm": 5.666680335998535, "learning_rate": 3.915533980582524e-05, "loss": 0.1855, "step": 12534 }, { "epoch": 121.54819277108433, "grad_norm": 53.92522048950195, "learning_rate": 3.915048543689321e-05, "loss": 0.2917, "step": 12535 }, { "epoch": 121.55783132530121, "grad_norm": 25.90814971923828, "learning_rate": 3.914563106796117e-05, "loss": 0.21, "step": 12536 }, { "epoch": 121.56746987951807, "grad_norm": 5.647477149963379, "learning_rate": 3.914077669902913e-05, "loss": 0.2567, "step": 12537 }, { "epoch": 121.57710843373494, "grad_norm": 50.12733459472656, "learning_rate": 3.9135922330097094e-05, "loss": 0.2993, "step": 12538 }, { "epoch": 121.58674698795181, "grad_norm": 12.073898315429688, "learning_rate": 3.913106796116505e-05, "loss": 0.2186, "step": 12539 }, { "epoch": 121.59638554216868, "grad_norm": 4.100941181182861, "learning_rate": 3.9126213592233016e-05, "loss": 0.3137, "step": 12540 }, { "epoch": 121.60602409638554, "grad_norm": 31.557798385620117, "learning_rate": 3.9121359223300974e-05, "loss": 0.4367, "step": 12541 }, { "epoch": 121.61566265060242, "grad_norm": 16.820838928222656, "learning_rate": 3.911650485436893e-05, "loss": 0.2319, "step": 12542 }, { "epoch": 121.62530120481928, "grad_norm": 35.604801177978516, "learning_rate": 3.9111650485436896e-05, "loss": 0.1984, "step": 12543 }, { "epoch": 121.63493975903614, "grad_norm": 48.454742431640625, "learning_rate": 3.9106796116504854e-05, "loss": 0.3153, "step": 12544 }, { "epoch": 121.644578313253, "grad_norm": 3.661595582962036, "learning_rate": 3.910194174757282e-05, "loss": 0.2673, "step": 12545 }, { "epoch": 121.65421686746988, "grad_norm": 37.28461456298828, "learning_rate": 3.9097087378640776e-05, "loss": 0.3158, "step": 12546 }, { "epoch": 121.66385542168675, "grad_norm": 2.8120639324188232, "learning_rate": 3.909223300970874e-05, "loss": 0.3543, "step": 12547 }, { "epoch": 121.67349397590361, "grad_norm": 17.037893295288086, "learning_rate": 3.90873786407767e-05, "loss": 0.3981, "step": 12548 }, { "epoch": 121.68313253012049, "grad_norm": 17.052879333496094, "learning_rate": 3.908252427184466e-05, "loss": 0.414, "step": 12549 }, { "epoch": 121.69277108433735, "grad_norm": 12.540226936340332, "learning_rate": 3.907766990291262e-05, "loss": 0.1883, "step": 12550 }, { "epoch": 121.70240963855422, "grad_norm": 12.051213264465332, "learning_rate": 3.9072815533980585e-05, "loss": 0.2534, "step": 12551 }, { "epoch": 121.71204819277108, "grad_norm": 15.16032600402832, "learning_rate": 3.906796116504854e-05, "loss": 0.2957, "step": 12552 }, { "epoch": 121.72168674698796, "grad_norm": 13.081280708312988, "learning_rate": 3.906310679611651e-05, "loss": 0.2272, "step": 12553 }, { "epoch": 121.73132530120482, "grad_norm": 86.03861236572266, "learning_rate": 3.9058252427184466e-05, "loss": 0.1903, "step": 12554 }, { "epoch": 121.74096385542168, "grad_norm": 23.68600082397461, "learning_rate": 3.905339805825243e-05, "loss": 0.3363, "step": 12555 }, { "epoch": 121.75060240963856, "grad_norm": 39.2659797668457, "learning_rate": 3.904854368932039e-05, "loss": 0.4094, "step": 12556 }, { "epoch": 121.76024096385542, "grad_norm": 30.198482513427734, "learning_rate": 3.904368932038835e-05, "loss": 0.3828, "step": 12557 }, { "epoch": 121.76987951807229, "grad_norm": 9.481846809387207, "learning_rate": 3.903883495145631e-05, "loss": 0.0857, "step": 12558 }, { "epoch": 121.77951807228915, "grad_norm": 12.353421211242676, "learning_rate": 3.903398058252427e-05, "loss": 0.207, "step": 12559 }, { "epoch": 121.78915662650603, "grad_norm": 38.990840911865234, "learning_rate": 3.902912621359224e-05, "loss": 0.2002, "step": 12560 }, { "epoch": 121.79879518072289, "grad_norm": 32.23558044433594, "learning_rate": 3.90242718446602e-05, "loss": 0.4628, "step": 12561 }, { "epoch": 121.80843373493975, "grad_norm": 76.12162017822266, "learning_rate": 3.901941747572816e-05, "loss": 0.3411, "step": 12562 }, { "epoch": 121.81807228915663, "grad_norm": 6.014283180236816, "learning_rate": 3.901456310679612e-05, "loss": 0.1212, "step": 12563 }, { "epoch": 121.8277108433735, "grad_norm": 32.30852127075195, "learning_rate": 3.9009708737864084e-05, "loss": 0.3007, "step": 12564 }, { "epoch": 121.83734939759036, "grad_norm": 30.213420867919922, "learning_rate": 3.900485436893204e-05, "loss": 0.2181, "step": 12565 }, { "epoch": 121.84698795180722, "grad_norm": 11.407498359680176, "learning_rate": 3.9000000000000006e-05, "loss": 0.0699, "step": 12566 }, { "epoch": 121.8566265060241, "grad_norm": 29.88514518737793, "learning_rate": 3.8995145631067964e-05, "loss": 0.2014, "step": 12567 }, { "epoch": 121.86626506024096, "grad_norm": 20.045242309570312, "learning_rate": 3.899029126213593e-05, "loss": 0.2874, "step": 12568 }, { "epoch": 121.87590361445783, "grad_norm": 52.6353759765625, "learning_rate": 3.8985436893203886e-05, "loss": 0.2453, "step": 12569 }, { "epoch": 121.8855421686747, "grad_norm": 12.348245620727539, "learning_rate": 3.8980582524271844e-05, "loss": 0.287, "step": 12570 }, { "epoch": 121.89518072289157, "grad_norm": 23.07549285888672, "learning_rate": 3.897572815533981e-05, "loss": 0.2465, "step": 12571 }, { "epoch": 121.90481927710843, "grad_norm": 10.906366348266602, "learning_rate": 3.8970873786407766e-05, "loss": 0.3338, "step": 12572 }, { "epoch": 121.91445783132531, "grad_norm": 5.926662445068359, "learning_rate": 3.896601941747573e-05, "loss": 0.3842, "step": 12573 }, { "epoch": 121.92409638554217, "grad_norm": 9.749215126037598, "learning_rate": 3.896116504854369e-05, "loss": 0.4785, "step": 12574 }, { "epoch": 121.93373493975903, "grad_norm": 5.407838821411133, "learning_rate": 3.895631067961165e-05, "loss": 0.3465, "step": 12575 }, { "epoch": 121.9433734939759, "grad_norm": 11.950566291809082, "learning_rate": 3.895145631067961e-05, "loss": 0.1538, "step": 12576 }, { "epoch": 121.95301204819278, "grad_norm": 12.712581634521484, "learning_rate": 3.8946601941747575e-05, "loss": 0.347, "step": 12577 }, { "epoch": 121.96265060240964, "grad_norm": 10.176488876342773, "learning_rate": 3.894174757281553e-05, "loss": 0.4692, "step": 12578 }, { "epoch": 121.9722891566265, "grad_norm": 3.741823673248291, "learning_rate": 3.89368932038835e-05, "loss": 0.1333, "step": 12579 }, { "epoch": 121.98192771084338, "grad_norm": 47.13869857788086, "learning_rate": 3.8932038834951455e-05, "loss": 0.2613, "step": 12580 }, { "epoch": 121.99156626506024, "grad_norm": 6.242575168609619, "learning_rate": 3.892718446601942e-05, "loss": 0.5, "step": 12581 }, { "epoch": 122.00722891566265, "grad_norm": 52.81376266479492, "learning_rate": 3.892233009708738e-05, "loss": 0.3904, "step": 12582 }, { "epoch": 122.01686746987951, "grad_norm": 11.20483684539795, "learning_rate": 3.891747572815534e-05, "loss": 0.2589, "step": 12583 }, { "epoch": 122.02650602409639, "grad_norm": 25.10465431213379, "learning_rate": 3.89126213592233e-05, "loss": 0.2914, "step": 12584 }, { "epoch": 122.03614457831326, "grad_norm": 25.18712615966797, "learning_rate": 3.8907766990291264e-05, "loss": 0.4807, "step": 12585 }, { "epoch": 122.04578313253012, "grad_norm": 10.592320442199707, "learning_rate": 3.890291262135923e-05, "loss": 0.4823, "step": 12586 }, { "epoch": 122.05542168674698, "grad_norm": 10.427434921264648, "learning_rate": 3.889805825242719e-05, "loss": 0.319, "step": 12587 }, { "epoch": 122.06506024096386, "grad_norm": 1.8164278268814087, "learning_rate": 3.889320388349515e-05, "loss": 0.3153, "step": 12588 }, { "epoch": 122.07469879518072, "grad_norm": 6.062016010284424, "learning_rate": 3.888834951456311e-05, "loss": 0.2957, "step": 12589 }, { "epoch": 122.08433734939759, "grad_norm": 5.0848541259765625, "learning_rate": 3.8883495145631074e-05, "loss": 0.2298, "step": 12590 }, { "epoch": 122.09397590361446, "grad_norm": 14.367805480957031, "learning_rate": 3.887864077669903e-05, "loss": 0.2922, "step": 12591 }, { "epoch": 122.10361445783133, "grad_norm": 6.226535320281982, "learning_rate": 3.8873786407766996e-05, "loss": 0.1135, "step": 12592 }, { "epoch": 122.11325301204819, "grad_norm": 30.528156280517578, "learning_rate": 3.8868932038834954e-05, "loss": 0.2599, "step": 12593 }, { "epoch": 122.12289156626505, "grad_norm": 3.517000675201416, "learning_rate": 3.886407766990292e-05, "loss": 0.1108, "step": 12594 }, { "epoch": 122.13253012048193, "grad_norm": 4.8135199546813965, "learning_rate": 3.8859223300970876e-05, "loss": 0.3338, "step": 12595 }, { "epoch": 122.1421686746988, "grad_norm": 11.346604347229004, "learning_rate": 3.885436893203884e-05, "loss": 0.2418, "step": 12596 }, { "epoch": 122.15180722891566, "grad_norm": 17.54004669189453, "learning_rate": 3.88495145631068e-05, "loss": 0.3269, "step": 12597 }, { "epoch": 122.16144578313254, "grad_norm": 42.55819320678711, "learning_rate": 3.8844660194174756e-05, "loss": 0.1265, "step": 12598 }, { "epoch": 122.1710843373494, "grad_norm": 3.404390811920166, "learning_rate": 3.883980582524272e-05, "loss": 0.1154, "step": 12599 }, { "epoch": 122.18072289156626, "grad_norm": 15.403741836547852, "learning_rate": 3.883495145631068e-05, "loss": 0.3756, "step": 12600 }, { "epoch": 122.19036144578314, "grad_norm": 2.454627275466919, "learning_rate": 3.883009708737864e-05, "loss": 0.2774, "step": 12601 }, { "epoch": 122.2, "grad_norm": 9.759552001953125, "learning_rate": 3.88252427184466e-05, "loss": 0.28, "step": 12602 }, { "epoch": 122.20963855421687, "grad_norm": 36.347381591796875, "learning_rate": 3.8820388349514565e-05, "loss": 0.3606, "step": 12603 }, { "epoch": 122.21927710843373, "grad_norm": 1.0332708358764648, "learning_rate": 3.881553398058252e-05, "loss": 0.0606, "step": 12604 }, { "epoch": 122.2289156626506, "grad_norm": 21.46749496459961, "learning_rate": 3.881067961165049e-05, "loss": 0.3589, "step": 12605 }, { "epoch": 122.23855421686747, "grad_norm": 15.22684097290039, "learning_rate": 3.8805825242718445e-05, "loss": 0.2123, "step": 12606 }, { "epoch": 122.24819277108433, "grad_norm": 6.063408374786377, "learning_rate": 3.880097087378641e-05, "loss": 0.2538, "step": 12607 }, { "epoch": 122.25783132530121, "grad_norm": 13.250359535217285, "learning_rate": 3.879611650485437e-05, "loss": 0.209, "step": 12608 }, { "epoch": 122.26746987951807, "grad_norm": 7.03565788269043, "learning_rate": 3.879126213592233e-05, "loss": 0.1939, "step": 12609 }, { "epoch": 122.27710843373494, "grad_norm": 6.182043075561523, "learning_rate": 3.8786407766990296e-05, "loss": 0.291, "step": 12610 }, { "epoch": 122.2867469879518, "grad_norm": 15.30009651184082, "learning_rate": 3.8781553398058254e-05, "loss": 0.2126, "step": 12611 }, { "epoch": 122.29638554216868, "grad_norm": 18.949451446533203, "learning_rate": 3.877669902912622e-05, "loss": 0.228, "step": 12612 }, { "epoch": 122.30602409638554, "grad_norm": 23.84302520751953, "learning_rate": 3.8771844660194177e-05, "loss": 0.2475, "step": 12613 }, { "epoch": 122.3156626506024, "grad_norm": 20.3449764251709, "learning_rate": 3.876699029126214e-05, "loss": 0.2264, "step": 12614 }, { "epoch": 122.32530120481928, "grad_norm": 46.97163772583008, "learning_rate": 3.87621359223301e-05, "loss": 0.2522, "step": 12615 }, { "epoch": 122.33493975903615, "grad_norm": 2.9617791175842285, "learning_rate": 3.875728155339806e-05, "loss": 0.2844, "step": 12616 }, { "epoch": 122.34457831325301, "grad_norm": 13.981978416442871, "learning_rate": 3.875242718446602e-05, "loss": 0.3153, "step": 12617 }, { "epoch": 122.35421686746987, "grad_norm": 31.83401107788086, "learning_rate": 3.8747572815533986e-05, "loss": 0.2551, "step": 12618 }, { "epoch": 122.36385542168675, "grad_norm": 36.748809814453125, "learning_rate": 3.8742718446601943e-05, "loss": 0.4124, "step": 12619 }, { "epoch": 122.37349397590361, "grad_norm": 6.178593635559082, "learning_rate": 3.873786407766991e-05, "loss": 0.1552, "step": 12620 }, { "epoch": 122.38313253012048, "grad_norm": 51.67578887939453, "learning_rate": 3.8733009708737866e-05, "loss": 0.2879, "step": 12621 }, { "epoch": 122.39277108433735, "grad_norm": 8.631635665893555, "learning_rate": 3.872815533980583e-05, "loss": 0.3728, "step": 12622 }, { "epoch": 122.40240963855422, "grad_norm": 16.033910751342773, "learning_rate": 3.872330097087379e-05, "loss": 0.4061, "step": 12623 }, { "epoch": 122.41204819277108, "grad_norm": 5.905800819396973, "learning_rate": 3.8718446601941746e-05, "loss": 0.1798, "step": 12624 }, { "epoch": 122.42168674698796, "grad_norm": 36.801639556884766, "learning_rate": 3.871359223300971e-05, "loss": 0.2879, "step": 12625 }, { "epoch": 122.43132530120482, "grad_norm": 4.324098587036133, "learning_rate": 3.870873786407767e-05, "loss": 0.2491, "step": 12626 }, { "epoch": 122.44096385542169, "grad_norm": 13.578490257263184, "learning_rate": 3.870388349514563e-05, "loss": 0.263, "step": 12627 }, { "epoch": 122.45060240963855, "grad_norm": 6.025689125061035, "learning_rate": 3.869902912621359e-05, "loss": 0.1179, "step": 12628 }, { "epoch": 122.46024096385543, "grad_norm": 11.917562484741211, "learning_rate": 3.8694174757281555e-05, "loss": 0.2248, "step": 12629 }, { "epoch": 122.46987951807229, "grad_norm": 40.052371978759766, "learning_rate": 3.868932038834951e-05, "loss": 0.2292, "step": 12630 }, { "epoch": 122.47951807228915, "grad_norm": 7.5200910568237305, "learning_rate": 3.868446601941748e-05, "loss": 0.4007, "step": 12631 }, { "epoch": 122.48915662650603, "grad_norm": 35.502349853515625, "learning_rate": 3.8679611650485435e-05, "loss": 0.3084, "step": 12632 }, { "epoch": 122.4987951807229, "grad_norm": 8.036368370056152, "learning_rate": 3.86747572815534e-05, "loss": 0.2174, "step": 12633 }, { "epoch": 122.50843373493976, "grad_norm": 2.324869394302368, "learning_rate": 3.866990291262136e-05, "loss": 0.1428, "step": 12634 }, { "epoch": 122.51807228915662, "grad_norm": 5.764039993286133, "learning_rate": 3.866504854368932e-05, "loss": 0.2255, "step": 12635 }, { "epoch": 122.5277108433735, "grad_norm": 25.063232421875, "learning_rate": 3.8660194174757286e-05, "loss": 0.3843, "step": 12636 }, { "epoch": 122.53734939759036, "grad_norm": 14.931282997131348, "learning_rate": 3.8655339805825244e-05, "loss": 0.3091, "step": 12637 }, { "epoch": 122.54698795180722, "grad_norm": 30.216796875, "learning_rate": 3.865048543689321e-05, "loss": 0.5248, "step": 12638 }, { "epoch": 122.5566265060241, "grad_norm": 25.79680633544922, "learning_rate": 3.8645631067961166e-05, "loss": 0.2777, "step": 12639 }, { "epoch": 122.56626506024097, "grad_norm": 2.1672494411468506, "learning_rate": 3.864077669902913e-05, "loss": 0.0633, "step": 12640 }, { "epoch": 122.57590361445783, "grad_norm": 35.90609359741211, "learning_rate": 3.863592233009709e-05, "loss": 0.2637, "step": 12641 }, { "epoch": 122.58554216867469, "grad_norm": 31.818017959594727, "learning_rate": 3.863106796116505e-05, "loss": 0.4653, "step": 12642 }, { "epoch": 122.59518072289157, "grad_norm": 15.49740982055664, "learning_rate": 3.862621359223301e-05, "loss": 0.2371, "step": 12643 }, { "epoch": 122.60481927710843, "grad_norm": 34.565765380859375, "learning_rate": 3.8621359223300975e-05, "loss": 0.3443, "step": 12644 }, { "epoch": 122.6144578313253, "grad_norm": 5.066173553466797, "learning_rate": 3.861650485436893e-05, "loss": 0.2622, "step": 12645 }, { "epoch": 122.62409638554217, "grad_norm": 19.04861068725586, "learning_rate": 3.86116504854369e-05, "loss": 0.342, "step": 12646 }, { "epoch": 122.63373493975904, "grad_norm": 23.100358963012695, "learning_rate": 3.8606796116504855e-05, "loss": 0.2202, "step": 12647 }, { "epoch": 122.6433734939759, "grad_norm": 5.496893882751465, "learning_rate": 3.860194174757282e-05, "loss": 0.3248, "step": 12648 }, { "epoch": 122.65301204819278, "grad_norm": 17.03849220275879, "learning_rate": 3.859708737864078e-05, "loss": 0.2381, "step": 12649 }, { "epoch": 122.66265060240964, "grad_norm": 38.5770378112793, "learning_rate": 3.859223300970874e-05, "loss": 0.518, "step": 12650 }, { "epoch": 122.6722891566265, "grad_norm": 14.142118453979492, "learning_rate": 3.85873786407767e-05, "loss": 0.2247, "step": 12651 }, { "epoch": 122.68192771084337, "grad_norm": 30.12279510498047, "learning_rate": 3.858252427184466e-05, "loss": 0.173, "step": 12652 }, { "epoch": 122.69156626506025, "grad_norm": 10.584847450256348, "learning_rate": 3.857766990291262e-05, "loss": 0.3319, "step": 12653 }, { "epoch": 122.70120481927711, "grad_norm": 35.122459411621094, "learning_rate": 3.857281553398058e-05, "loss": 0.1838, "step": 12654 }, { "epoch": 122.71084337349397, "grad_norm": 46.0012321472168, "learning_rate": 3.8567961165048545e-05, "loss": 0.3008, "step": 12655 }, { "epoch": 122.72048192771085, "grad_norm": 6.1527605056762695, "learning_rate": 3.85631067961165e-05, "loss": 0.1884, "step": 12656 }, { "epoch": 122.73012048192771, "grad_norm": 22.81534194946289, "learning_rate": 3.855825242718447e-05, "loss": 0.2476, "step": 12657 }, { "epoch": 122.73975903614458, "grad_norm": 21.469926834106445, "learning_rate": 3.8553398058252425e-05, "loss": 0.2985, "step": 12658 }, { "epoch": 122.74939759036144, "grad_norm": 14.040218353271484, "learning_rate": 3.854854368932039e-05, "loss": 0.3267, "step": 12659 }, { "epoch": 122.75903614457832, "grad_norm": 5.411288738250732, "learning_rate": 3.854368932038835e-05, "loss": 0.1728, "step": 12660 }, { "epoch": 122.76867469879518, "grad_norm": 5.688112735748291, "learning_rate": 3.853883495145632e-05, "loss": 0.453, "step": 12661 }, { "epoch": 122.77831325301204, "grad_norm": 31.384037017822266, "learning_rate": 3.8533980582524276e-05, "loss": 0.2778, "step": 12662 }, { "epoch": 122.78795180722892, "grad_norm": 11.037454605102539, "learning_rate": 3.8529126213592234e-05, "loss": 0.2858, "step": 12663 }, { "epoch": 122.79759036144578, "grad_norm": 28.14731788635254, "learning_rate": 3.85242718446602e-05, "loss": 0.3295, "step": 12664 }, { "epoch": 122.80722891566265, "grad_norm": 10.874970436096191, "learning_rate": 3.8519417475728156e-05, "loss": 0.2935, "step": 12665 }, { "epoch": 122.81686746987951, "grad_norm": 27.892620086669922, "learning_rate": 3.851456310679612e-05, "loss": 0.524, "step": 12666 }, { "epoch": 122.82650602409639, "grad_norm": 4.0396199226379395, "learning_rate": 3.850970873786408e-05, "loss": 0.2703, "step": 12667 }, { "epoch": 122.83614457831325, "grad_norm": 11.608145713806152, "learning_rate": 3.850485436893204e-05, "loss": 0.3609, "step": 12668 }, { "epoch": 122.84578313253012, "grad_norm": 75.87630462646484, "learning_rate": 3.85e-05, "loss": 0.1187, "step": 12669 }, { "epoch": 122.855421686747, "grad_norm": 2.5006344318389893, "learning_rate": 3.8495145631067965e-05, "loss": 0.062, "step": 12670 }, { "epoch": 122.86506024096386, "grad_norm": 11.30947208404541, "learning_rate": 3.849029126213592e-05, "loss": 0.3093, "step": 12671 }, { "epoch": 122.87469879518072, "grad_norm": 35.45172882080078, "learning_rate": 3.848543689320389e-05, "loss": 0.4428, "step": 12672 }, { "epoch": 122.88433734939758, "grad_norm": 18.32291030883789, "learning_rate": 3.8480582524271845e-05, "loss": 0.3334, "step": 12673 }, { "epoch": 122.89397590361446, "grad_norm": 29.193788528442383, "learning_rate": 3.847572815533981e-05, "loss": 0.208, "step": 12674 }, { "epoch": 122.90361445783132, "grad_norm": 9.480545997619629, "learning_rate": 3.847087378640777e-05, "loss": 0.2724, "step": 12675 }, { "epoch": 122.91325301204819, "grad_norm": 79.94503021240234, "learning_rate": 3.846601941747573e-05, "loss": 0.1929, "step": 12676 }, { "epoch": 122.92289156626506, "grad_norm": 27.97228240966797, "learning_rate": 3.846116504854369e-05, "loss": 0.374, "step": 12677 }, { "epoch": 122.93253012048193, "grad_norm": 4.9847893714904785, "learning_rate": 3.8456310679611654e-05, "loss": 0.2342, "step": 12678 }, { "epoch": 122.94216867469879, "grad_norm": 74.81610107421875, "learning_rate": 3.845145631067961e-05, "loss": 0.451, "step": 12679 }, { "epoch": 122.95180722891567, "grad_norm": 16.96931266784668, "learning_rate": 3.844660194174757e-05, "loss": 0.2525, "step": 12680 }, { "epoch": 122.96144578313253, "grad_norm": 10.199769973754883, "learning_rate": 3.8441747572815534e-05, "loss": 0.338, "step": 12681 }, { "epoch": 122.9710843373494, "grad_norm": 12.663480758666992, "learning_rate": 3.843689320388349e-05, "loss": 0.3117, "step": 12682 }, { "epoch": 122.98072289156626, "grad_norm": 33.63023376464844, "learning_rate": 3.843203883495146e-05, "loss": 0.1743, "step": 12683 }, { "epoch": 122.99036144578314, "grad_norm": 12.893303871154785, "learning_rate": 3.8427184466019414e-05, "loss": 0.1295, "step": 12684 }, { "epoch": 123.00602409638554, "grad_norm": 5.700934410095215, "learning_rate": 3.842233009708738e-05, "loss": 0.1387, "step": 12685 }, { "epoch": 123.01566265060241, "grad_norm": 7.74231481552124, "learning_rate": 3.8417475728155344e-05, "loss": 0.2586, "step": 12686 }, { "epoch": 123.02530120481927, "grad_norm": 5.878788948059082, "learning_rate": 3.841262135922331e-05, "loss": 0.3734, "step": 12687 }, { "epoch": 123.03493975903615, "grad_norm": 62.12827682495117, "learning_rate": 3.8407766990291266e-05, "loss": 0.3169, "step": 12688 }, { "epoch": 123.04457831325301, "grad_norm": 39.688968658447266, "learning_rate": 3.840291262135923e-05, "loss": 0.2949, "step": 12689 }, { "epoch": 123.05421686746988, "grad_norm": 17.223812103271484, "learning_rate": 3.839805825242719e-05, "loss": 0.2844, "step": 12690 }, { "epoch": 123.06385542168675, "grad_norm": 28.84678840637207, "learning_rate": 3.8393203883495146e-05, "loss": 0.3032, "step": 12691 }, { "epoch": 123.07349397590362, "grad_norm": 7.425466537475586, "learning_rate": 3.838834951456311e-05, "loss": 0.4205, "step": 12692 }, { "epoch": 123.08313253012048, "grad_norm": 13.026017189025879, "learning_rate": 3.838349514563107e-05, "loss": 0.4495, "step": 12693 }, { "epoch": 123.09277108433734, "grad_norm": 3.572874069213867, "learning_rate": 3.837864077669903e-05, "loss": 0.0791, "step": 12694 }, { "epoch": 123.10240963855422, "grad_norm": 8.204297065734863, "learning_rate": 3.837378640776699e-05, "loss": 0.3275, "step": 12695 }, { "epoch": 123.11204819277108, "grad_norm": 12.17597770690918, "learning_rate": 3.8368932038834955e-05, "loss": 0.3499, "step": 12696 }, { "epoch": 123.12168674698795, "grad_norm": 8.986260414123535, "learning_rate": 3.836407766990291e-05, "loss": 0.3845, "step": 12697 }, { "epoch": 123.13132530120482, "grad_norm": 36.07024002075195, "learning_rate": 3.835922330097088e-05, "loss": 0.2309, "step": 12698 }, { "epoch": 123.14096385542169, "grad_norm": 10.589327812194824, "learning_rate": 3.8354368932038835e-05, "loss": 0.2634, "step": 12699 }, { "epoch": 123.15060240963855, "grad_norm": 13.384554862976074, "learning_rate": 3.83495145631068e-05, "loss": 0.415, "step": 12700 }, { "epoch": 123.16024096385541, "grad_norm": 5.4537200927734375, "learning_rate": 3.834466019417476e-05, "loss": 0.2602, "step": 12701 }, { "epoch": 123.16987951807229, "grad_norm": 6.800747394561768, "learning_rate": 3.833980582524272e-05, "loss": 0.1727, "step": 12702 }, { "epoch": 123.17951807228916, "grad_norm": 54.64817428588867, "learning_rate": 3.833495145631068e-05, "loss": 0.3453, "step": 12703 }, { "epoch": 123.18915662650602, "grad_norm": 26.30991554260254, "learning_rate": 3.8330097087378644e-05, "loss": 0.2648, "step": 12704 }, { "epoch": 123.1987951807229, "grad_norm": 16.107337951660156, "learning_rate": 3.83252427184466e-05, "loss": 0.1427, "step": 12705 }, { "epoch": 123.20843373493976, "grad_norm": 69.28977966308594, "learning_rate": 3.8320388349514566e-05, "loss": 0.2917, "step": 12706 }, { "epoch": 123.21807228915662, "grad_norm": 11.385780334472656, "learning_rate": 3.8315533980582524e-05, "loss": 0.3006, "step": 12707 }, { "epoch": 123.2277108433735, "grad_norm": 72.35074615478516, "learning_rate": 3.831067961165048e-05, "loss": 0.5309, "step": 12708 }, { "epoch": 123.23734939759036, "grad_norm": 20.222978591918945, "learning_rate": 3.8305825242718447e-05, "loss": 0.3961, "step": 12709 }, { "epoch": 123.24698795180723, "grad_norm": 91.4991683959961, "learning_rate": 3.8300970873786404e-05, "loss": 0.2216, "step": 12710 }, { "epoch": 123.25662650602409, "grad_norm": 15.827380180358887, "learning_rate": 3.8296116504854376e-05, "loss": 0.3345, "step": 12711 }, { "epoch": 123.26626506024097, "grad_norm": 34.70215606689453, "learning_rate": 3.829126213592233e-05, "loss": 0.3589, "step": 12712 }, { "epoch": 123.27590361445783, "grad_norm": 4.760279655456543, "learning_rate": 3.82864077669903e-05, "loss": 0.1443, "step": 12713 }, { "epoch": 123.2855421686747, "grad_norm": 44.85348892211914, "learning_rate": 3.8281553398058256e-05, "loss": 0.3015, "step": 12714 }, { "epoch": 123.29518072289157, "grad_norm": 11.01491641998291, "learning_rate": 3.827669902912622e-05, "loss": 0.4561, "step": 12715 }, { "epoch": 123.30481927710844, "grad_norm": 47.53772735595703, "learning_rate": 3.827184466019418e-05, "loss": 0.2283, "step": 12716 }, { "epoch": 123.3144578313253, "grad_norm": 13.500657081604004, "learning_rate": 3.826699029126214e-05, "loss": 0.3218, "step": 12717 }, { "epoch": 123.32409638554216, "grad_norm": 21.067869186401367, "learning_rate": 3.82621359223301e-05, "loss": 0.427, "step": 12718 }, { "epoch": 123.33373493975904, "grad_norm": 50.66509246826172, "learning_rate": 3.825728155339806e-05, "loss": 0.142, "step": 12719 }, { "epoch": 123.3433734939759, "grad_norm": 6.2318572998046875, "learning_rate": 3.825242718446602e-05, "loss": 0.3569, "step": 12720 }, { "epoch": 123.35301204819277, "grad_norm": 23.88001823425293, "learning_rate": 3.824757281553398e-05, "loss": 0.2989, "step": 12721 }, { "epoch": 123.36265060240964, "grad_norm": 11.602968215942383, "learning_rate": 3.8242718446601945e-05, "loss": 0.4099, "step": 12722 }, { "epoch": 123.37228915662651, "grad_norm": 5.796635627746582, "learning_rate": 3.82378640776699e-05, "loss": 0.1444, "step": 12723 }, { "epoch": 123.38192771084337, "grad_norm": 88.26774597167969, "learning_rate": 3.823300970873787e-05, "loss": 0.2693, "step": 12724 }, { "epoch": 123.39156626506023, "grad_norm": 7.49353551864624, "learning_rate": 3.8228155339805825e-05, "loss": 0.223, "step": 12725 }, { "epoch": 123.40120481927711, "grad_norm": 6.091370105743408, "learning_rate": 3.822330097087379e-05, "loss": 0.3205, "step": 12726 }, { "epoch": 123.41084337349398, "grad_norm": 20.957117080688477, "learning_rate": 3.821844660194175e-05, "loss": 0.3845, "step": 12727 }, { "epoch": 123.42048192771084, "grad_norm": 4.153534412384033, "learning_rate": 3.821359223300971e-05, "loss": 0.1712, "step": 12728 }, { "epoch": 123.43012048192772, "grad_norm": 19.239501953125, "learning_rate": 3.820873786407767e-05, "loss": 0.3751, "step": 12729 }, { "epoch": 123.43975903614458, "grad_norm": 6.005833625793457, "learning_rate": 3.8203883495145634e-05, "loss": 0.168, "step": 12730 }, { "epoch": 123.44939759036144, "grad_norm": 4.7965168952941895, "learning_rate": 3.819902912621359e-05, "loss": 0.082, "step": 12731 }, { "epoch": 123.45903614457832, "grad_norm": 17.925220489501953, "learning_rate": 3.8194174757281556e-05, "loss": 0.244, "step": 12732 }, { "epoch": 123.46867469879518, "grad_norm": 24.398977279663086, "learning_rate": 3.8189320388349514e-05, "loss": 0.2261, "step": 12733 }, { "epoch": 123.47831325301205, "grad_norm": 14.50175952911377, "learning_rate": 3.818446601941748e-05, "loss": 0.3486, "step": 12734 }, { "epoch": 123.48795180722891, "grad_norm": 15.119154930114746, "learning_rate": 3.8179611650485436e-05, "loss": 0.331, "step": 12735 }, { "epoch": 123.49759036144579, "grad_norm": 42.82590103149414, "learning_rate": 3.8174757281553394e-05, "loss": 0.405, "step": 12736 }, { "epoch": 123.50722891566265, "grad_norm": 207.3731689453125, "learning_rate": 3.8169902912621365e-05, "loss": 0.3507, "step": 12737 }, { "epoch": 123.51686746987951, "grad_norm": 5.365285396575928, "learning_rate": 3.816504854368932e-05, "loss": 0.1739, "step": 12738 }, { "epoch": 123.52650602409639, "grad_norm": 7.190728664398193, "learning_rate": 3.816019417475729e-05, "loss": 0.2697, "step": 12739 }, { "epoch": 123.53614457831326, "grad_norm": 14.040989875793457, "learning_rate": 3.8155339805825245e-05, "loss": 0.2214, "step": 12740 }, { "epoch": 123.54578313253012, "grad_norm": 6.14196252822876, "learning_rate": 3.815048543689321e-05, "loss": 0.09, "step": 12741 }, { "epoch": 123.55542168674698, "grad_norm": 11.432974815368652, "learning_rate": 3.814563106796117e-05, "loss": 0.1017, "step": 12742 }, { "epoch": 123.56506024096386, "grad_norm": 2.916457176208496, "learning_rate": 3.814077669902913e-05, "loss": 0.1767, "step": 12743 }, { "epoch": 123.57469879518072, "grad_norm": 2.8321104049682617, "learning_rate": 3.813592233009709e-05, "loss": 0.3199, "step": 12744 }, { "epoch": 123.58433734939759, "grad_norm": 10.701370239257812, "learning_rate": 3.8131067961165055e-05, "loss": 0.2104, "step": 12745 }, { "epoch": 123.59397590361446, "grad_norm": 6.545361518859863, "learning_rate": 3.812621359223301e-05, "loss": 0.2752, "step": 12746 }, { "epoch": 123.60361445783133, "grad_norm": 9.366247177124023, "learning_rate": 3.812135922330097e-05, "loss": 0.2907, "step": 12747 }, { "epoch": 123.61325301204819, "grad_norm": 12.100747108459473, "learning_rate": 3.8116504854368935e-05, "loss": 0.1763, "step": 12748 }, { "epoch": 123.62289156626505, "grad_norm": 173.09146118164062, "learning_rate": 3.811165048543689e-05, "loss": 0.106, "step": 12749 }, { "epoch": 123.63253012048193, "grad_norm": 2.9995908737182617, "learning_rate": 3.810679611650486e-05, "loss": 0.1956, "step": 12750 }, { "epoch": 123.6421686746988, "grad_norm": 12.21308708190918, "learning_rate": 3.8101941747572815e-05, "loss": 0.1077, "step": 12751 }, { "epoch": 123.65180722891566, "grad_norm": 7.70928430557251, "learning_rate": 3.809708737864078e-05, "loss": 0.2187, "step": 12752 }, { "epoch": 123.66144578313254, "grad_norm": 21.528581619262695, "learning_rate": 3.809223300970874e-05, "loss": 0.4261, "step": 12753 }, { "epoch": 123.6710843373494, "grad_norm": 9.278351783752441, "learning_rate": 3.80873786407767e-05, "loss": 0.1836, "step": 12754 }, { "epoch": 123.68072289156626, "grad_norm": 11.036712646484375, "learning_rate": 3.808252427184466e-05, "loss": 0.2219, "step": 12755 }, { "epoch": 123.69036144578314, "grad_norm": 18.42296600341797, "learning_rate": 3.8077669902912624e-05, "loss": 0.3869, "step": 12756 }, { "epoch": 123.7, "grad_norm": 6.880377292633057, "learning_rate": 3.807281553398058e-05, "loss": 0.2279, "step": 12757 }, { "epoch": 123.70963855421687, "grad_norm": 16.155967712402344, "learning_rate": 3.8067961165048546e-05, "loss": 0.2803, "step": 12758 }, { "epoch": 123.71927710843373, "grad_norm": 28.608478546142578, "learning_rate": 3.8063106796116504e-05, "loss": 0.2891, "step": 12759 }, { "epoch": 123.7289156626506, "grad_norm": 9.127339363098145, "learning_rate": 3.805825242718447e-05, "loss": 0.3597, "step": 12760 }, { "epoch": 123.73855421686747, "grad_norm": 7.902918815612793, "learning_rate": 3.8053398058252426e-05, "loss": 0.2409, "step": 12761 }, { "epoch": 123.74819277108433, "grad_norm": 27.243833541870117, "learning_rate": 3.804854368932039e-05, "loss": 0.501, "step": 12762 }, { "epoch": 123.75783132530121, "grad_norm": 11.540994644165039, "learning_rate": 3.8043689320388355e-05, "loss": 0.29, "step": 12763 }, { "epoch": 123.76746987951807, "grad_norm": 7.563014984130859, "learning_rate": 3.803883495145631e-05, "loss": 0.3151, "step": 12764 }, { "epoch": 123.77710843373494, "grad_norm": 4.574219703674316, "learning_rate": 3.803398058252428e-05, "loss": 0.5799, "step": 12765 }, { "epoch": 123.7867469879518, "grad_norm": 41.21427917480469, "learning_rate": 3.8029126213592235e-05, "loss": 0.2567, "step": 12766 }, { "epoch": 123.79638554216868, "grad_norm": 47.84645462036133, "learning_rate": 3.80242718446602e-05, "loss": 0.4064, "step": 12767 }, { "epoch": 123.80602409638554, "grad_norm": 16.14522933959961, "learning_rate": 3.801941747572816e-05, "loss": 0.2379, "step": 12768 }, { "epoch": 123.8156626506024, "grad_norm": 17.120447158813477, "learning_rate": 3.801456310679612e-05, "loss": 0.2401, "step": 12769 }, { "epoch": 123.82530120481928, "grad_norm": 2.5437204837799072, "learning_rate": 3.800970873786408e-05, "loss": 0.1861, "step": 12770 }, { "epoch": 123.83493975903615, "grad_norm": 25.479143142700195, "learning_rate": 3.8004854368932044e-05, "loss": 0.2704, "step": 12771 }, { "epoch": 123.84457831325301, "grad_norm": 32.60356903076172, "learning_rate": 3.8e-05, "loss": 0.382, "step": 12772 }, { "epoch": 123.85421686746987, "grad_norm": 10.456720352172852, "learning_rate": 3.799514563106796e-05, "loss": 0.2389, "step": 12773 }, { "epoch": 123.86385542168675, "grad_norm": 8.093038558959961, "learning_rate": 3.7990291262135924e-05, "loss": 0.2539, "step": 12774 }, { "epoch": 123.87349397590361, "grad_norm": 10.589825630187988, "learning_rate": 3.798543689320388e-05, "loss": 0.2653, "step": 12775 }, { "epoch": 123.88313253012048, "grad_norm": 23.001001358032227, "learning_rate": 3.798058252427185e-05, "loss": 0.2864, "step": 12776 }, { "epoch": 123.89277108433735, "grad_norm": 59.64451217651367, "learning_rate": 3.7975728155339804e-05, "loss": 0.2668, "step": 12777 }, { "epoch": 123.90240963855422, "grad_norm": 66.60104370117188, "learning_rate": 3.797087378640777e-05, "loss": 0.3919, "step": 12778 }, { "epoch": 123.91204819277108, "grad_norm": 57.63865661621094, "learning_rate": 3.796601941747573e-05, "loss": 0.3227, "step": 12779 }, { "epoch": 123.92168674698796, "grad_norm": 15.907953262329102, "learning_rate": 3.796116504854369e-05, "loss": 0.2917, "step": 12780 }, { "epoch": 123.93132530120482, "grad_norm": 27.342206954956055, "learning_rate": 3.795631067961165e-05, "loss": 0.3536, "step": 12781 }, { "epoch": 123.94096385542169, "grad_norm": 11.367473602294922, "learning_rate": 3.7951456310679614e-05, "loss": 0.1332, "step": 12782 }, { "epoch": 123.95060240963855, "grad_norm": 6.766629695892334, "learning_rate": 3.794660194174757e-05, "loss": 0.19, "step": 12783 }, { "epoch": 123.96024096385543, "grad_norm": 7.963171005249023, "learning_rate": 3.7941747572815536e-05, "loss": 0.2715, "step": 12784 }, { "epoch": 123.96987951807229, "grad_norm": 3.7111687660217285, "learning_rate": 3.7936893203883494e-05, "loss": 0.3242, "step": 12785 }, { "epoch": 123.97951807228915, "grad_norm": 2.6880714893341064, "learning_rate": 3.793203883495146e-05, "loss": 0.2267, "step": 12786 }, { "epoch": 123.98915662650603, "grad_norm": 5.993680953979492, "learning_rate": 3.792718446601942e-05, "loss": 0.2278, "step": 12787 }, { "epoch": 124.00481927710844, "grad_norm": 5.564809322357178, "learning_rate": 3.792233009708738e-05, "loss": 0.2826, "step": 12788 }, { "epoch": 124.0144578313253, "grad_norm": 19.147071838378906, "learning_rate": 3.7917475728155345e-05, "loss": 0.4252, "step": 12789 }, { "epoch": 124.02409638554217, "grad_norm": 7.665487289428711, "learning_rate": 3.79126213592233e-05, "loss": 0.2019, "step": 12790 }, { "epoch": 124.03373493975904, "grad_norm": 10.418937683105469, "learning_rate": 3.790776699029127e-05, "loss": 0.1746, "step": 12791 }, { "epoch": 124.0433734939759, "grad_norm": 6.309469699859619, "learning_rate": 3.7902912621359225e-05, "loss": 0.2702, "step": 12792 }, { "epoch": 124.05301204819277, "grad_norm": 10.525485038757324, "learning_rate": 3.789805825242719e-05, "loss": 0.3486, "step": 12793 }, { "epoch": 124.06265060240963, "grad_norm": 5.548995018005371, "learning_rate": 3.789320388349515e-05, "loss": 0.3334, "step": 12794 }, { "epoch": 124.07228915662651, "grad_norm": 18.298433303833008, "learning_rate": 3.788834951456311e-05, "loss": 0.2249, "step": 12795 }, { "epoch": 124.08192771084337, "grad_norm": 10.491161346435547, "learning_rate": 3.788349514563107e-05, "loss": 0.3389, "step": 12796 }, { "epoch": 124.09156626506024, "grad_norm": 5.238959312438965, "learning_rate": 3.7878640776699034e-05, "loss": 0.2519, "step": 12797 }, { "epoch": 124.10120481927711, "grad_norm": 8.104669570922852, "learning_rate": 3.787378640776699e-05, "loss": 0.1854, "step": 12798 }, { "epoch": 124.11084337349398, "grad_norm": 18.557594299316406, "learning_rate": 3.7868932038834956e-05, "loss": 0.3508, "step": 12799 }, { "epoch": 124.12048192771084, "grad_norm": 21.60923957824707, "learning_rate": 3.7864077669902914e-05, "loss": 0.1885, "step": 12800 }, { "epoch": 124.1301204819277, "grad_norm": 45.81167221069336, "learning_rate": 3.785922330097087e-05, "loss": 0.2858, "step": 12801 }, { "epoch": 124.13975903614458, "grad_norm": 24.61841583251953, "learning_rate": 3.7854368932038836e-05, "loss": 0.2129, "step": 12802 }, { "epoch": 124.14939759036145, "grad_norm": 4.344115257263184, "learning_rate": 3.7849514563106794e-05, "loss": 0.1587, "step": 12803 }, { "epoch": 124.15903614457831, "grad_norm": 42.958534240722656, "learning_rate": 3.784466019417476e-05, "loss": 0.6224, "step": 12804 }, { "epoch": 124.16867469879519, "grad_norm": 19.379323959350586, "learning_rate": 3.7839805825242717e-05, "loss": 0.3883, "step": 12805 }, { "epoch": 124.17831325301205, "grad_norm": 31.67873764038086, "learning_rate": 3.783495145631068e-05, "loss": 0.4122, "step": 12806 }, { "epoch": 124.18795180722891, "grad_norm": 14.680257797241211, "learning_rate": 3.783009708737864e-05, "loss": 0.3918, "step": 12807 }, { "epoch": 124.19759036144578, "grad_norm": 24.48920249938965, "learning_rate": 3.78252427184466e-05, "loss": 0.2439, "step": 12808 }, { "epoch": 124.20722891566265, "grad_norm": 7.2321953773498535, "learning_rate": 3.782038834951456e-05, "loss": 0.215, "step": 12809 }, { "epoch": 124.21686746987952, "grad_norm": 6.46204137802124, "learning_rate": 3.7815533980582526e-05, "loss": 0.1446, "step": 12810 }, { "epoch": 124.22650602409638, "grad_norm": 55.909908294677734, "learning_rate": 3.781067961165048e-05, "loss": 0.4449, "step": 12811 }, { "epoch": 124.23614457831326, "grad_norm": 9.993614196777344, "learning_rate": 3.780582524271845e-05, "loss": 0.1978, "step": 12812 }, { "epoch": 124.24578313253012, "grad_norm": 45.379302978515625, "learning_rate": 3.780097087378641e-05, "loss": 0.188, "step": 12813 }, { "epoch": 124.25542168674698, "grad_norm": 5.531102657318115, "learning_rate": 3.779611650485437e-05, "loss": 0.1289, "step": 12814 }, { "epoch": 124.26506024096386, "grad_norm": 13.592552185058594, "learning_rate": 3.7791262135922335e-05, "loss": 0.1564, "step": 12815 }, { "epoch": 124.27469879518073, "grad_norm": 9.63017463684082, "learning_rate": 3.778640776699029e-05, "loss": 0.3634, "step": 12816 }, { "epoch": 124.28433734939759, "grad_norm": 18.395898818969727, "learning_rate": 3.778155339805826e-05, "loss": 0.1817, "step": 12817 }, { "epoch": 124.29397590361445, "grad_norm": 27.62847900390625, "learning_rate": 3.7776699029126215e-05, "loss": 0.5005, "step": 12818 }, { "epoch": 124.30361445783133, "grad_norm": 15.601129531860352, "learning_rate": 3.777184466019418e-05, "loss": 0.1473, "step": 12819 }, { "epoch": 124.3132530120482, "grad_norm": 5.013331413269043, "learning_rate": 3.776699029126214e-05, "loss": 0.3448, "step": 12820 }, { "epoch": 124.32289156626506, "grad_norm": 17.615766525268555, "learning_rate": 3.77621359223301e-05, "loss": 0.4265, "step": 12821 }, { "epoch": 124.33253012048193, "grad_norm": 5.613029956817627, "learning_rate": 3.775728155339806e-05, "loss": 0.0903, "step": 12822 }, { "epoch": 124.3421686746988, "grad_norm": 5.466037750244141, "learning_rate": 3.7752427184466024e-05, "loss": 0.1036, "step": 12823 }, { "epoch": 124.35180722891566, "grad_norm": 16.318344116210938, "learning_rate": 3.774757281553398e-05, "loss": 0.3336, "step": 12824 }, { "epoch": 124.36144578313252, "grad_norm": 15.85391616821289, "learning_rate": 3.7742718446601946e-05, "loss": 0.2098, "step": 12825 }, { "epoch": 124.3710843373494, "grad_norm": 9.6675386428833, "learning_rate": 3.7737864077669904e-05, "loss": 0.2106, "step": 12826 }, { "epoch": 124.38072289156626, "grad_norm": 5.693109035491943, "learning_rate": 3.773300970873787e-05, "loss": 0.2449, "step": 12827 }, { "epoch": 124.39036144578313, "grad_norm": 21.473318099975586, "learning_rate": 3.7728155339805826e-05, "loss": 0.2677, "step": 12828 }, { "epoch": 124.4, "grad_norm": 24.219873428344727, "learning_rate": 3.7723300970873784e-05, "loss": 0.3665, "step": 12829 }, { "epoch": 124.40963855421687, "grad_norm": 9.483853340148926, "learning_rate": 3.771844660194175e-05, "loss": 0.323, "step": 12830 }, { "epoch": 124.41927710843373, "grad_norm": 12.528536796569824, "learning_rate": 3.7713592233009706e-05, "loss": 0.3383, "step": 12831 }, { "epoch": 124.4289156626506, "grad_norm": 13.003471374511719, "learning_rate": 3.770873786407767e-05, "loss": 0.377, "step": 12832 }, { "epoch": 124.43855421686747, "grad_norm": 5.957798480987549, "learning_rate": 3.770388349514563e-05, "loss": 0.1884, "step": 12833 }, { "epoch": 124.44819277108434, "grad_norm": 4.594636917114258, "learning_rate": 3.769902912621359e-05, "loss": 0.3321, "step": 12834 }, { "epoch": 124.4578313253012, "grad_norm": 6.246277809143066, "learning_rate": 3.769417475728155e-05, "loss": 0.27, "step": 12835 }, { "epoch": 124.46746987951808, "grad_norm": 37.040626525878906, "learning_rate": 3.7689320388349515e-05, "loss": 0.3044, "step": 12836 }, { "epoch": 124.47710843373494, "grad_norm": 8.372760772705078, "learning_rate": 3.768446601941747e-05, "loss": 0.3205, "step": 12837 }, { "epoch": 124.4867469879518, "grad_norm": 12.901739120483398, "learning_rate": 3.7679611650485444e-05, "loss": 0.2438, "step": 12838 }, { "epoch": 124.49638554216868, "grad_norm": 11.33378791809082, "learning_rate": 3.76747572815534e-05, "loss": 0.4643, "step": 12839 }, { "epoch": 124.50602409638554, "grad_norm": 9.741265296936035, "learning_rate": 3.766990291262136e-05, "loss": 0.2243, "step": 12840 }, { "epoch": 124.51566265060241, "grad_norm": 13.426653861999512, "learning_rate": 3.7665048543689325e-05, "loss": 0.2101, "step": 12841 }, { "epoch": 124.52530120481927, "grad_norm": 28.08094024658203, "learning_rate": 3.766019417475728e-05, "loss": 0.2318, "step": 12842 }, { "epoch": 124.53493975903615, "grad_norm": 3.30825138092041, "learning_rate": 3.765533980582525e-05, "loss": 0.1617, "step": 12843 }, { "epoch": 124.54457831325301, "grad_norm": 6.91392183303833, "learning_rate": 3.7650485436893205e-05, "loss": 0.3804, "step": 12844 }, { "epoch": 124.55421686746988, "grad_norm": 8.394083023071289, "learning_rate": 3.764563106796117e-05, "loss": 0.3095, "step": 12845 }, { "epoch": 124.56385542168675, "grad_norm": 11.229416847229004, "learning_rate": 3.764077669902913e-05, "loss": 0.3509, "step": 12846 }, { "epoch": 124.57349397590362, "grad_norm": 37.053009033203125, "learning_rate": 3.763592233009709e-05, "loss": 0.2387, "step": 12847 }, { "epoch": 124.58313253012048, "grad_norm": 26.9038143157959, "learning_rate": 3.763106796116505e-05, "loss": 0.2549, "step": 12848 }, { "epoch": 124.59277108433734, "grad_norm": 17.676034927368164, "learning_rate": 3.7626213592233014e-05, "loss": 0.1052, "step": 12849 }, { "epoch": 124.60240963855422, "grad_norm": 3.1332364082336426, "learning_rate": 3.762135922330097e-05, "loss": 0.1427, "step": 12850 }, { "epoch": 124.61204819277108, "grad_norm": 10.793281555175781, "learning_rate": 3.7616504854368936e-05, "loss": 0.4326, "step": 12851 }, { "epoch": 124.62168674698795, "grad_norm": 48.7520751953125, "learning_rate": 3.7611650485436894e-05, "loss": 0.413, "step": 12852 }, { "epoch": 124.63132530120482, "grad_norm": 6.437780857086182, "learning_rate": 3.760679611650486e-05, "loss": 0.3681, "step": 12853 }, { "epoch": 124.64096385542169, "grad_norm": 22.67789649963379, "learning_rate": 3.7601941747572816e-05, "loss": 0.3024, "step": 12854 }, { "epoch": 124.65060240963855, "grad_norm": 15.246210098266602, "learning_rate": 3.759708737864078e-05, "loss": 0.2959, "step": 12855 }, { "epoch": 124.66024096385541, "grad_norm": 2.6067590713500977, "learning_rate": 3.759223300970874e-05, "loss": 0.1858, "step": 12856 }, { "epoch": 124.66987951807229, "grad_norm": 7.2276692390441895, "learning_rate": 3.7587378640776696e-05, "loss": 0.1231, "step": 12857 }, { "epoch": 124.67951807228916, "grad_norm": 29.76915168762207, "learning_rate": 3.758252427184466e-05, "loss": 0.223, "step": 12858 }, { "epoch": 124.68915662650602, "grad_norm": 34.01955795288086, "learning_rate": 3.757766990291262e-05, "loss": 0.2185, "step": 12859 }, { "epoch": 124.6987951807229, "grad_norm": 6.746201038360596, "learning_rate": 3.757281553398058e-05, "loss": 0.1453, "step": 12860 }, { "epoch": 124.70843373493976, "grad_norm": 38.99425506591797, "learning_rate": 3.756796116504854e-05, "loss": 0.6097, "step": 12861 }, { "epoch": 124.71807228915662, "grad_norm": 10.43294906616211, "learning_rate": 3.7563106796116505e-05, "loss": 0.2172, "step": 12862 }, { "epoch": 124.7277108433735, "grad_norm": 6.036623954772949, "learning_rate": 3.755825242718447e-05, "loss": 0.172, "step": 12863 }, { "epoch": 124.73734939759036, "grad_norm": 20.243026733398438, "learning_rate": 3.7553398058252434e-05, "loss": 0.2336, "step": 12864 }, { "epoch": 124.74698795180723, "grad_norm": 73.95281982421875, "learning_rate": 3.754854368932039e-05, "loss": 0.3396, "step": 12865 }, { "epoch": 124.75662650602409, "grad_norm": 13.516914367675781, "learning_rate": 3.7543689320388357e-05, "loss": 0.236, "step": 12866 }, { "epoch": 124.76626506024097, "grad_norm": 12.63647174835205, "learning_rate": 3.7538834951456314e-05, "loss": 0.1385, "step": 12867 }, { "epoch": 124.77590361445783, "grad_norm": 31.817838668823242, "learning_rate": 3.753398058252427e-05, "loss": 0.1724, "step": 12868 }, { "epoch": 124.7855421686747, "grad_norm": 30.0836181640625, "learning_rate": 3.752912621359224e-05, "loss": 0.4409, "step": 12869 }, { "epoch": 124.79518072289157, "grad_norm": 22.347412109375, "learning_rate": 3.7524271844660194e-05, "loss": 0.3907, "step": 12870 }, { "epoch": 124.80481927710844, "grad_norm": 5.0272297859191895, "learning_rate": 3.751941747572816e-05, "loss": 0.1384, "step": 12871 }, { "epoch": 124.8144578313253, "grad_norm": 8.304793357849121, "learning_rate": 3.751456310679612e-05, "loss": 0.2626, "step": 12872 }, { "epoch": 124.82409638554216, "grad_norm": 18.98977279663086, "learning_rate": 3.750970873786408e-05, "loss": 0.2993, "step": 12873 }, { "epoch": 124.83373493975904, "grad_norm": 16.76108169555664, "learning_rate": 3.750485436893204e-05, "loss": 0.2233, "step": 12874 }, { "epoch": 124.8433734939759, "grad_norm": 4.625492572784424, "learning_rate": 3.7500000000000003e-05, "loss": 0.1269, "step": 12875 }, { "epoch": 124.85301204819277, "grad_norm": 50.11437225341797, "learning_rate": 3.749514563106796e-05, "loss": 0.3943, "step": 12876 }, { "epoch": 124.86265060240964, "grad_norm": 12.787137031555176, "learning_rate": 3.7490291262135926e-05, "loss": 0.3258, "step": 12877 }, { "epoch": 124.87228915662651, "grad_norm": 11.193039894104004, "learning_rate": 3.7485436893203884e-05, "loss": 0.1603, "step": 12878 }, { "epoch": 124.88192771084337, "grad_norm": 62.494537353515625, "learning_rate": 3.748058252427185e-05, "loss": 0.4211, "step": 12879 }, { "epoch": 124.89156626506023, "grad_norm": 32.532928466796875, "learning_rate": 3.7475728155339806e-05, "loss": 0.2603, "step": 12880 }, { "epoch": 124.90120481927711, "grad_norm": 14.131011009216309, "learning_rate": 3.747087378640777e-05, "loss": 0.1886, "step": 12881 }, { "epoch": 124.91084337349398, "grad_norm": 21.042451858520508, "learning_rate": 3.746601941747573e-05, "loss": 0.2944, "step": 12882 }, { "epoch": 124.92048192771084, "grad_norm": 44.1940803527832, "learning_rate": 3.746116504854369e-05, "loss": 0.1991, "step": 12883 }, { "epoch": 124.93012048192772, "grad_norm": 6.167780876159668, "learning_rate": 3.745631067961165e-05, "loss": 0.1839, "step": 12884 }, { "epoch": 124.93975903614458, "grad_norm": 9.704265594482422, "learning_rate": 3.745145631067961e-05, "loss": 0.3212, "step": 12885 }, { "epoch": 124.94939759036144, "grad_norm": 9.708614349365234, "learning_rate": 3.744660194174757e-05, "loss": 0.3192, "step": 12886 }, { "epoch": 124.95903614457832, "grad_norm": 45.15131759643555, "learning_rate": 3.744174757281553e-05, "loss": 0.4566, "step": 12887 }, { "epoch": 124.96867469879518, "grad_norm": 25.31821060180664, "learning_rate": 3.74368932038835e-05, "loss": 0.2474, "step": 12888 }, { "epoch": 124.97831325301205, "grad_norm": 4.636997699737549, "learning_rate": 3.743203883495146e-05, "loss": 0.1111, "step": 12889 }, { "epoch": 124.98795180722891, "grad_norm": 3.7529006004333496, "learning_rate": 3.7427184466019424e-05, "loss": 0.2368, "step": 12890 }, { "epoch": 125.00361445783132, "grad_norm": 11.429085731506348, "learning_rate": 3.742233009708738e-05, "loss": 0.1979, "step": 12891 }, { "epoch": 125.0132530120482, "grad_norm": 3.55429744720459, "learning_rate": 3.7417475728155346e-05, "loss": 0.1264, "step": 12892 }, { "epoch": 125.02289156626506, "grad_norm": 27.63459014892578, "learning_rate": 3.7412621359223304e-05, "loss": 0.4302, "step": 12893 }, { "epoch": 125.03253012048192, "grad_norm": 28.725467681884766, "learning_rate": 3.740776699029127e-05, "loss": 0.2689, "step": 12894 }, { "epoch": 125.0421686746988, "grad_norm": 14.584510803222656, "learning_rate": 3.7402912621359226e-05, "loss": 0.3168, "step": 12895 }, { "epoch": 125.05180722891566, "grad_norm": 18.2181396484375, "learning_rate": 3.7398058252427184e-05, "loss": 0.359, "step": 12896 }, { "epoch": 125.06144578313253, "grad_norm": 10.514922142028809, "learning_rate": 3.739320388349515e-05, "loss": 0.3131, "step": 12897 }, { "epoch": 125.0710843373494, "grad_norm": 18.60801124572754, "learning_rate": 3.7388349514563106e-05, "loss": 0.2857, "step": 12898 }, { "epoch": 125.08072289156627, "grad_norm": 13.824004173278809, "learning_rate": 3.738349514563107e-05, "loss": 0.2813, "step": 12899 }, { "epoch": 125.09036144578313, "grad_norm": 17.448558807373047, "learning_rate": 3.737864077669903e-05, "loss": 0.1104, "step": 12900 }, { "epoch": 125.1, "grad_norm": 6.623231887817383, "learning_rate": 3.737378640776699e-05, "loss": 0.1082, "step": 12901 }, { "epoch": 125.10963855421687, "grad_norm": 15.17015552520752, "learning_rate": 3.736893203883495e-05, "loss": 0.1023, "step": 12902 }, { "epoch": 125.11927710843374, "grad_norm": 6.655190467834473, "learning_rate": 3.7364077669902916e-05, "loss": 0.1892, "step": 12903 }, { "epoch": 125.1289156626506, "grad_norm": 6.593975067138672, "learning_rate": 3.735922330097087e-05, "loss": 0.1927, "step": 12904 }, { "epoch": 125.13855421686748, "grad_norm": 4.767807960510254, "learning_rate": 3.735436893203884e-05, "loss": 0.1719, "step": 12905 }, { "epoch": 125.14819277108434, "grad_norm": 15.251385688781738, "learning_rate": 3.7349514563106796e-05, "loss": 0.5305, "step": 12906 }, { "epoch": 125.1578313253012, "grad_norm": 5.219505310058594, "learning_rate": 3.734466019417476e-05, "loss": 0.1539, "step": 12907 }, { "epoch": 125.16746987951807, "grad_norm": 11.149626731872559, "learning_rate": 3.733980582524272e-05, "loss": 0.3278, "step": 12908 }, { "epoch": 125.17710843373494, "grad_norm": 8.142646789550781, "learning_rate": 3.733495145631068e-05, "loss": 0.4398, "step": 12909 }, { "epoch": 125.1867469879518, "grad_norm": 28.08068084716797, "learning_rate": 3.733009708737864e-05, "loss": 0.3692, "step": 12910 }, { "epoch": 125.19638554216867, "grad_norm": 268.7976379394531, "learning_rate": 3.7325242718446605e-05, "loss": 0.168, "step": 12911 }, { "epoch": 125.20602409638555, "grad_norm": 11.664316177368164, "learning_rate": 3.732038834951456e-05, "loss": 0.3585, "step": 12912 }, { "epoch": 125.21566265060241, "grad_norm": 12.090808868408203, "learning_rate": 3.731553398058253e-05, "loss": 0.1835, "step": 12913 }, { "epoch": 125.22530120481927, "grad_norm": 10.660956382751465, "learning_rate": 3.731067961165049e-05, "loss": 0.3095, "step": 12914 }, { "epoch": 125.23493975903614, "grad_norm": 15.05687427520752, "learning_rate": 3.730582524271845e-05, "loss": 0.3653, "step": 12915 }, { "epoch": 125.24457831325302, "grad_norm": 7.672616481781006, "learning_rate": 3.7300970873786414e-05, "loss": 0.1161, "step": 12916 }, { "epoch": 125.25421686746988, "grad_norm": 19.22806739807129, "learning_rate": 3.729611650485437e-05, "loss": 0.3671, "step": 12917 }, { "epoch": 125.26385542168674, "grad_norm": 9.839057922363281, "learning_rate": 3.7291262135922336e-05, "loss": 0.2072, "step": 12918 }, { "epoch": 125.27349397590362, "grad_norm": 7.833794116973877, "learning_rate": 3.7286407766990294e-05, "loss": 0.2247, "step": 12919 }, { "epoch": 125.28313253012048, "grad_norm": 7.1704325675964355, "learning_rate": 3.728155339805826e-05, "loss": 0.2805, "step": 12920 }, { "epoch": 125.29277108433735, "grad_norm": 33.38458251953125, "learning_rate": 3.7276699029126216e-05, "loss": 0.27, "step": 12921 }, { "epoch": 125.30240963855422, "grad_norm": 2.9146432876586914, "learning_rate": 3.7271844660194174e-05, "loss": 0.081, "step": 12922 }, { "epoch": 125.31204819277109, "grad_norm": 20.871566772460938, "learning_rate": 3.726699029126214e-05, "loss": 0.3674, "step": 12923 }, { "epoch": 125.32168674698795, "grad_norm": 3.6503138542175293, "learning_rate": 3.7262135922330096e-05, "loss": 0.1649, "step": 12924 }, { "epoch": 125.33132530120481, "grad_norm": 8.606170654296875, "learning_rate": 3.725728155339806e-05, "loss": 0.3275, "step": 12925 }, { "epoch": 125.34096385542169, "grad_norm": 7.85015344619751, "learning_rate": 3.725242718446602e-05, "loss": 0.1052, "step": 12926 }, { "epoch": 125.35060240963855, "grad_norm": 9.374985694885254, "learning_rate": 3.724757281553398e-05, "loss": 0.3181, "step": 12927 }, { "epoch": 125.36024096385542, "grad_norm": 5.873442649841309, "learning_rate": 3.724271844660194e-05, "loss": 0.2027, "step": 12928 }, { "epoch": 125.3698795180723, "grad_norm": 7.21783447265625, "learning_rate": 3.7237864077669905e-05, "loss": 0.2346, "step": 12929 }, { "epoch": 125.37951807228916, "grad_norm": 14.32852840423584, "learning_rate": 3.723300970873786e-05, "loss": 0.2119, "step": 12930 }, { "epoch": 125.38915662650602, "grad_norm": 10.548700332641602, "learning_rate": 3.722815533980583e-05, "loss": 0.2814, "step": 12931 }, { "epoch": 125.39879518072289, "grad_norm": 33.697776794433594, "learning_rate": 3.7223300970873785e-05, "loss": 0.2457, "step": 12932 }, { "epoch": 125.40843373493976, "grad_norm": 33.15540313720703, "learning_rate": 3.721844660194175e-05, "loss": 0.4135, "step": 12933 }, { "epoch": 125.41807228915663, "grad_norm": 16.320322036743164, "learning_rate": 3.721359223300971e-05, "loss": 0.3647, "step": 12934 }, { "epoch": 125.42771084337349, "grad_norm": 6.811741828918457, "learning_rate": 3.720873786407767e-05, "loss": 0.1831, "step": 12935 }, { "epoch": 125.43734939759037, "grad_norm": 22.321853637695312, "learning_rate": 3.720388349514563e-05, "loss": 0.2053, "step": 12936 }, { "epoch": 125.44698795180723, "grad_norm": 12.16635513305664, "learning_rate": 3.7199029126213595e-05, "loss": 0.2391, "step": 12937 }, { "epoch": 125.4566265060241, "grad_norm": 12.647753715515137, "learning_rate": 3.719417475728155e-05, "loss": 0.2899, "step": 12938 }, { "epoch": 125.46626506024096, "grad_norm": 12.395706176757812, "learning_rate": 3.718932038834952e-05, "loss": 0.1742, "step": 12939 }, { "epoch": 125.47590361445783, "grad_norm": 23.736692428588867, "learning_rate": 3.718446601941748e-05, "loss": 0.2098, "step": 12940 }, { "epoch": 125.4855421686747, "grad_norm": 5.964502811431885, "learning_rate": 3.717961165048544e-05, "loss": 0.1886, "step": 12941 }, { "epoch": 125.49518072289156, "grad_norm": 26.83184814453125, "learning_rate": 3.7174757281553404e-05, "loss": 0.2481, "step": 12942 }, { "epoch": 125.50481927710844, "grad_norm": 13.189950942993164, "learning_rate": 3.716990291262136e-05, "loss": 0.1911, "step": 12943 }, { "epoch": 125.5144578313253, "grad_norm": 5.547135353088379, "learning_rate": 3.7165048543689326e-05, "loss": 0.2402, "step": 12944 }, { "epoch": 125.52409638554217, "grad_norm": 46.79204559326172, "learning_rate": 3.7160194174757284e-05, "loss": 0.2872, "step": 12945 }, { "epoch": 125.53373493975904, "grad_norm": 21.630443572998047, "learning_rate": 3.715533980582525e-05, "loss": 0.2547, "step": 12946 }, { "epoch": 125.5433734939759, "grad_norm": 9.202899932861328, "learning_rate": 3.7150485436893206e-05, "loss": 0.3083, "step": 12947 }, { "epoch": 125.55301204819277, "grad_norm": 17.93846893310547, "learning_rate": 3.714563106796117e-05, "loss": 0.4036, "step": 12948 }, { "epoch": 125.56265060240963, "grad_norm": 5.561625003814697, "learning_rate": 3.714077669902913e-05, "loss": 0.3459, "step": 12949 }, { "epoch": 125.57228915662651, "grad_norm": 3.599125862121582, "learning_rate": 3.7135922330097086e-05, "loss": 0.078, "step": 12950 }, { "epoch": 125.58192771084337, "grad_norm": 13.227073669433594, "learning_rate": 3.713106796116505e-05, "loss": 0.1621, "step": 12951 }, { "epoch": 125.59156626506024, "grad_norm": 7.8355278968811035, "learning_rate": 3.712621359223301e-05, "loss": 0.1471, "step": 12952 }, { "epoch": 125.60120481927711, "grad_norm": 42.148902893066406, "learning_rate": 3.712135922330097e-05, "loss": 0.239, "step": 12953 }, { "epoch": 125.61084337349398, "grad_norm": 12.780781745910645, "learning_rate": 3.711650485436893e-05, "loss": 0.2671, "step": 12954 }, { "epoch": 125.62048192771084, "grad_norm": 24.28950309753418, "learning_rate": 3.7111650485436895e-05, "loss": 0.4917, "step": 12955 }, { "epoch": 125.6301204819277, "grad_norm": 8.33408260345459, "learning_rate": 3.710679611650485e-05, "loss": 0.2777, "step": 12956 }, { "epoch": 125.63975903614458, "grad_norm": 19.544986724853516, "learning_rate": 3.710194174757282e-05, "loss": 0.2431, "step": 12957 }, { "epoch": 125.64939759036145, "grad_norm": 5.213543891906738, "learning_rate": 3.7097087378640775e-05, "loss": 0.2561, "step": 12958 }, { "epoch": 125.65903614457831, "grad_norm": 16.98373794555664, "learning_rate": 3.709223300970874e-05, "loss": 0.3088, "step": 12959 }, { "epoch": 125.66867469879519, "grad_norm": 9.921476364135742, "learning_rate": 3.70873786407767e-05, "loss": 0.1098, "step": 12960 }, { "epoch": 125.67831325301205, "grad_norm": 10.357609748840332, "learning_rate": 3.708252427184466e-05, "loss": 0.3892, "step": 12961 }, { "epoch": 125.68795180722891, "grad_norm": 36.8401985168457, "learning_rate": 3.707766990291262e-05, "loss": 0.2888, "step": 12962 }, { "epoch": 125.69759036144578, "grad_norm": 55.17753219604492, "learning_rate": 3.7072815533980584e-05, "loss": 0.3184, "step": 12963 }, { "epoch": 125.70722891566265, "grad_norm": 8.732868194580078, "learning_rate": 3.706796116504855e-05, "loss": 0.1719, "step": 12964 }, { "epoch": 125.71686746987952, "grad_norm": 11.919273376464844, "learning_rate": 3.7063106796116507e-05, "loss": 0.2934, "step": 12965 }, { "epoch": 125.72650602409638, "grad_norm": 5.781925678253174, "learning_rate": 3.705825242718447e-05, "loss": 0.2853, "step": 12966 }, { "epoch": 125.73614457831326, "grad_norm": 20.619356155395508, "learning_rate": 3.705339805825243e-05, "loss": 0.3686, "step": 12967 }, { "epoch": 125.74578313253012, "grad_norm": 28.22789192199707, "learning_rate": 3.7048543689320393e-05, "loss": 0.2566, "step": 12968 }, { "epoch": 125.75542168674698, "grad_norm": 13.824536323547363, "learning_rate": 3.704368932038835e-05, "loss": 0.2133, "step": 12969 }, { "epoch": 125.76506024096386, "grad_norm": 28.439809799194336, "learning_rate": 3.7038834951456316e-05, "loss": 0.2991, "step": 12970 }, { "epoch": 125.77469879518073, "grad_norm": 15.029703140258789, "learning_rate": 3.7033980582524273e-05, "loss": 0.2066, "step": 12971 }, { "epoch": 125.78433734939759, "grad_norm": 17.096254348754883, "learning_rate": 3.702912621359224e-05, "loss": 0.2349, "step": 12972 }, { "epoch": 125.79397590361445, "grad_norm": 58.931976318359375, "learning_rate": 3.7024271844660196e-05, "loss": 0.2955, "step": 12973 }, { "epoch": 125.80361445783133, "grad_norm": 40.58903884887695, "learning_rate": 3.701941747572816e-05, "loss": 0.2347, "step": 12974 }, { "epoch": 125.8132530120482, "grad_norm": 5.192508697509766, "learning_rate": 3.701456310679612e-05, "loss": 0.3024, "step": 12975 }, { "epoch": 125.82289156626506, "grad_norm": 6.70398473739624, "learning_rate": 3.700970873786408e-05, "loss": 0.1448, "step": 12976 }, { "epoch": 125.83253012048193, "grad_norm": 34.97957992553711, "learning_rate": 3.700485436893204e-05, "loss": 0.3752, "step": 12977 }, { "epoch": 125.8421686746988, "grad_norm": 38.248783111572266, "learning_rate": 3.7e-05, "loss": 0.1679, "step": 12978 }, { "epoch": 125.85180722891566, "grad_norm": 13.045672416687012, "learning_rate": 3.699514563106796e-05, "loss": 0.2611, "step": 12979 }, { "epoch": 125.86144578313252, "grad_norm": 48.29740905761719, "learning_rate": 3.699029126213592e-05, "loss": 0.3849, "step": 12980 }, { "epoch": 125.8710843373494, "grad_norm": 12.411930084228516, "learning_rate": 3.6985436893203885e-05, "loss": 0.2206, "step": 12981 }, { "epoch": 125.88072289156626, "grad_norm": 14.942034721374512, "learning_rate": 3.698058252427184e-05, "loss": 0.4376, "step": 12982 }, { "epoch": 125.89036144578313, "grad_norm": 11.955686569213867, "learning_rate": 3.697572815533981e-05, "loss": 0.3574, "step": 12983 }, { "epoch": 125.9, "grad_norm": 20.079601287841797, "learning_rate": 3.6970873786407765e-05, "loss": 0.2504, "step": 12984 }, { "epoch": 125.90963855421687, "grad_norm": 18.660621643066406, "learning_rate": 3.696601941747573e-05, "loss": 0.3825, "step": 12985 }, { "epoch": 125.91927710843373, "grad_norm": 47.58332443237305, "learning_rate": 3.696116504854369e-05, "loss": 0.3478, "step": 12986 }, { "epoch": 125.9289156626506, "grad_norm": 8.752230644226074, "learning_rate": 3.695631067961165e-05, "loss": 0.3489, "step": 12987 }, { "epoch": 125.93855421686747, "grad_norm": 23.157854080200195, "learning_rate": 3.695145631067961e-05, "loss": 0.3958, "step": 12988 }, { "epoch": 125.94819277108434, "grad_norm": 35.61830520629883, "learning_rate": 3.6946601941747574e-05, "loss": 0.3, "step": 12989 }, { "epoch": 125.9578313253012, "grad_norm": 6.828128337860107, "learning_rate": 3.694174757281554e-05, "loss": 0.1763, "step": 12990 }, { "epoch": 125.96746987951808, "grad_norm": 24.46596908569336, "learning_rate": 3.6936893203883496e-05, "loss": 0.4463, "step": 12991 }, { "epoch": 125.97710843373494, "grad_norm": 17.978487014770508, "learning_rate": 3.693203883495146e-05, "loss": 0.1393, "step": 12992 }, { "epoch": 125.9867469879518, "grad_norm": 8.26870059967041, "learning_rate": 3.692718446601942e-05, "loss": 0.0744, "step": 12993 }, { "epoch": 126.00240963855421, "grad_norm": 10.582075119018555, "learning_rate": 3.692233009708738e-05, "loss": 0.3377, "step": 12994 }, { "epoch": 126.01204819277109, "grad_norm": 19.343822479248047, "learning_rate": 3.691747572815534e-05, "loss": 0.1969, "step": 12995 }, { "epoch": 126.02168674698795, "grad_norm": 8.769831657409668, "learning_rate": 3.6912621359223306e-05, "loss": 0.4561, "step": 12996 }, { "epoch": 126.03132530120482, "grad_norm": 20.99102020263672, "learning_rate": 3.690776699029126e-05, "loss": 0.3134, "step": 12997 }, { "epoch": 126.04096385542168, "grad_norm": 11.476859092712402, "learning_rate": 3.690291262135923e-05, "loss": 0.4403, "step": 12998 }, { "epoch": 126.05060240963856, "grad_norm": 10.936871528625488, "learning_rate": 3.6898058252427186e-05, "loss": 0.4376, "step": 12999 }, { "epoch": 126.06024096385542, "grad_norm": 23.012666702270508, "learning_rate": 3.689320388349515e-05, "loss": 0.3726, "step": 13000 }, { "epoch": 126.06987951807228, "grad_norm": 21.991273880004883, "learning_rate": 3.688834951456311e-05, "loss": 0.3258, "step": 13001 }, { "epoch": 126.07951807228916, "grad_norm": 27.416622161865234, "learning_rate": 3.688349514563107e-05, "loss": 0.1363, "step": 13002 }, { "epoch": 126.08915662650602, "grad_norm": 2.1073248386383057, "learning_rate": 3.687864077669903e-05, "loss": 0.1314, "step": 13003 }, { "epoch": 126.09879518072289, "grad_norm": 10.349587440490723, "learning_rate": 3.6873786407766995e-05, "loss": 0.278, "step": 13004 }, { "epoch": 126.10843373493977, "grad_norm": 9.979388236999512, "learning_rate": 3.686893203883495e-05, "loss": 0.2726, "step": 13005 }, { "epoch": 126.11807228915663, "grad_norm": 48.64801788330078, "learning_rate": 3.686407766990291e-05, "loss": 0.4278, "step": 13006 }, { "epoch": 126.12771084337349, "grad_norm": 15.133190155029297, "learning_rate": 3.6859223300970875e-05, "loss": 0.2137, "step": 13007 }, { "epoch": 126.13734939759036, "grad_norm": 17.749080657958984, "learning_rate": 3.685436893203883e-05, "loss": 0.3428, "step": 13008 }, { "epoch": 126.14698795180723, "grad_norm": 12.481489181518555, "learning_rate": 3.68495145631068e-05, "loss": 0.1682, "step": 13009 }, { "epoch": 126.1566265060241, "grad_norm": 16.211421966552734, "learning_rate": 3.6844660194174755e-05, "loss": 0.2582, "step": 13010 }, { "epoch": 126.16626506024096, "grad_norm": 11.36069107055664, "learning_rate": 3.683980582524272e-05, "loss": 0.2402, "step": 13011 }, { "epoch": 126.17590361445784, "grad_norm": 10.356258392333984, "learning_rate": 3.683495145631068e-05, "loss": 0.4871, "step": 13012 }, { "epoch": 126.1855421686747, "grad_norm": 7.509084701538086, "learning_rate": 3.683009708737864e-05, "loss": 0.1743, "step": 13013 }, { "epoch": 126.19518072289156, "grad_norm": 40.74995040893555, "learning_rate": 3.6825242718446606e-05, "loss": 0.2268, "step": 13014 }, { "epoch": 126.20481927710843, "grad_norm": 7.591060638427734, "learning_rate": 3.682038834951457e-05, "loss": 0.3298, "step": 13015 }, { "epoch": 126.2144578313253, "grad_norm": 11.99166488647461, "learning_rate": 3.681553398058253e-05, "loss": 0.2442, "step": 13016 }, { "epoch": 126.22409638554217, "grad_norm": 8.964927673339844, "learning_rate": 3.6810679611650486e-05, "loss": 0.36, "step": 13017 }, { "epoch": 126.23373493975903, "grad_norm": 17.540027618408203, "learning_rate": 3.680582524271845e-05, "loss": 0.4884, "step": 13018 }, { "epoch": 126.24337349397591, "grad_norm": 9.40014934539795, "learning_rate": 3.680097087378641e-05, "loss": 0.2715, "step": 13019 }, { "epoch": 126.25301204819277, "grad_norm": 3.159759521484375, "learning_rate": 3.679611650485437e-05, "loss": 0.1286, "step": 13020 }, { "epoch": 126.26265060240964, "grad_norm": 8.334954261779785, "learning_rate": 3.679126213592233e-05, "loss": 0.1721, "step": 13021 }, { "epoch": 126.2722891566265, "grad_norm": 9.600896835327148, "learning_rate": 3.6786407766990295e-05, "loss": 0.3901, "step": 13022 }, { "epoch": 126.28192771084338, "grad_norm": 7.581765174865723, "learning_rate": 3.678155339805825e-05, "loss": 0.2519, "step": 13023 }, { "epoch": 126.29156626506024, "grad_norm": 7.306560039520264, "learning_rate": 3.677669902912622e-05, "loss": 0.0939, "step": 13024 }, { "epoch": 126.3012048192771, "grad_norm": 13.548121452331543, "learning_rate": 3.6771844660194175e-05, "loss": 0.2647, "step": 13025 }, { "epoch": 126.31084337349398, "grad_norm": 6.920067310333252, "learning_rate": 3.676699029126214e-05, "loss": 0.2403, "step": 13026 }, { "epoch": 126.32048192771084, "grad_norm": 5.94782018661499, "learning_rate": 3.67621359223301e-05, "loss": 0.247, "step": 13027 }, { "epoch": 126.33012048192771, "grad_norm": 15.32343864440918, "learning_rate": 3.675728155339806e-05, "loss": 0.2849, "step": 13028 }, { "epoch": 126.33975903614459, "grad_norm": 3.2154250144958496, "learning_rate": 3.675242718446602e-05, "loss": 0.1801, "step": 13029 }, { "epoch": 126.34939759036145, "grad_norm": 26.889793395996094, "learning_rate": 3.6747572815533984e-05, "loss": 0.2626, "step": 13030 }, { "epoch": 126.35903614457831, "grad_norm": 13.26685905456543, "learning_rate": 3.674271844660194e-05, "loss": 0.3696, "step": 13031 }, { "epoch": 126.36867469879518, "grad_norm": 23.954177856445312, "learning_rate": 3.673786407766991e-05, "loss": 0.4058, "step": 13032 }, { "epoch": 126.37831325301205, "grad_norm": 25.877225875854492, "learning_rate": 3.6733009708737865e-05, "loss": 0.323, "step": 13033 }, { "epoch": 126.38795180722892, "grad_norm": 5.6421895027160645, "learning_rate": 3.672815533980582e-05, "loss": 0.2591, "step": 13034 }, { "epoch": 126.39759036144578, "grad_norm": 23.483121871948242, "learning_rate": 3.672330097087379e-05, "loss": 0.1494, "step": 13035 }, { "epoch": 126.40722891566266, "grad_norm": 11.637085914611816, "learning_rate": 3.6718446601941745e-05, "loss": 0.3321, "step": 13036 }, { "epoch": 126.41686746987952, "grad_norm": 11.380674362182617, "learning_rate": 3.671359223300971e-05, "loss": 0.2787, "step": 13037 }, { "epoch": 126.42650602409638, "grad_norm": 7.106183052062988, "learning_rate": 3.670873786407767e-05, "loss": 0.1726, "step": 13038 }, { "epoch": 126.43614457831325, "grad_norm": 17.243961334228516, "learning_rate": 3.670388349514563e-05, "loss": 0.1693, "step": 13039 }, { "epoch": 126.44578313253012, "grad_norm": 9.117515563964844, "learning_rate": 3.6699029126213596e-05, "loss": 0.2539, "step": 13040 }, { "epoch": 126.45542168674699, "grad_norm": 8.819332122802734, "learning_rate": 3.669417475728156e-05, "loss": 0.2603, "step": 13041 }, { "epoch": 126.46506024096385, "grad_norm": 6.741176128387451, "learning_rate": 3.668932038834952e-05, "loss": 0.1608, "step": 13042 }, { "epoch": 126.47469879518073, "grad_norm": 15.439297676086426, "learning_rate": 3.668446601941748e-05, "loss": 0.2308, "step": 13043 }, { "epoch": 126.48433734939759, "grad_norm": 9.816316604614258, "learning_rate": 3.667961165048544e-05, "loss": 0.2276, "step": 13044 }, { "epoch": 126.49397590361446, "grad_norm": 4.795786380767822, "learning_rate": 3.66747572815534e-05, "loss": 0.1509, "step": 13045 }, { "epoch": 126.50361445783132, "grad_norm": 15.062753677368164, "learning_rate": 3.666990291262136e-05, "loss": 0.2841, "step": 13046 }, { "epoch": 126.5132530120482, "grad_norm": 13.077640533447266, "learning_rate": 3.666504854368932e-05, "loss": 0.3753, "step": 13047 }, { "epoch": 126.52289156626506, "grad_norm": 70.49852752685547, "learning_rate": 3.6660194174757285e-05, "loss": 0.3326, "step": 13048 }, { "epoch": 126.53253012048192, "grad_norm": 10.827216148376465, "learning_rate": 3.665533980582524e-05, "loss": 0.1406, "step": 13049 }, { "epoch": 126.5421686746988, "grad_norm": 18.32953453063965, "learning_rate": 3.665048543689321e-05, "loss": 0.3269, "step": 13050 }, { "epoch": 126.55180722891566, "grad_norm": 27.458524703979492, "learning_rate": 3.6645631067961165e-05, "loss": 0.3297, "step": 13051 }, { "epoch": 126.56144578313253, "grad_norm": 8.87010383605957, "learning_rate": 3.664077669902913e-05, "loss": 0.1531, "step": 13052 }, { "epoch": 126.5710843373494, "grad_norm": 7.095170021057129, "learning_rate": 3.663592233009709e-05, "loss": 0.3059, "step": 13053 }, { "epoch": 126.58072289156627, "grad_norm": 11.346012115478516, "learning_rate": 3.663106796116505e-05, "loss": 0.288, "step": 13054 }, { "epoch": 126.59036144578313, "grad_norm": 13.166048049926758, "learning_rate": 3.662621359223301e-05, "loss": 0.3438, "step": 13055 }, { "epoch": 126.6, "grad_norm": 16.31412124633789, "learning_rate": 3.6621359223300974e-05, "loss": 0.2322, "step": 13056 }, { "epoch": 126.60963855421687, "grad_norm": 5.727962017059326, "learning_rate": 3.661650485436893e-05, "loss": 0.4854, "step": 13057 }, { "epoch": 126.61927710843374, "grad_norm": 13.67831039428711, "learning_rate": 3.6611650485436897e-05, "loss": 0.3707, "step": 13058 }, { "epoch": 126.6289156626506, "grad_norm": 13.500849723815918, "learning_rate": 3.6606796116504854e-05, "loss": 0.4662, "step": 13059 }, { "epoch": 126.63855421686748, "grad_norm": 9.200284957885742, "learning_rate": 3.660194174757282e-05, "loss": 0.3267, "step": 13060 }, { "epoch": 126.64819277108434, "grad_norm": 13.658729553222656, "learning_rate": 3.6597087378640777e-05, "loss": 0.1949, "step": 13061 }, { "epoch": 126.6578313253012, "grad_norm": 12.204858779907227, "learning_rate": 3.6592233009708734e-05, "loss": 0.3514, "step": 13062 }, { "epoch": 126.66746987951807, "grad_norm": 22.661739349365234, "learning_rate": 3.65873786407767e-05, "loss": 0.5092, "step": 13063 }, { "epoch": 126.67710843373494, "grad_norm": 35.576683044433594, "learning_rate": 3.658252427184466e-05, "loss": 0.3931, "step": 13064 }, { "epoch": 126.6867469879518, "grad_norm": 34.96669006347656, "learning_rate": 3.657766990291263e-05, "loss": 0.6138, "step": 13065 }, { "epoch": 126.69638554216867, "grad_norm": 12.575641632080078, "learning_rate": 3.6572815533980586e-05, "loss": 0.3571, "step": 13066 }, { "epoch": 126.70602409638555, "grad_norm": 6.966734409332275, "learning_rate": 3.656796116504855e-05, "loss": 0.2171, "step": 13067 }, { "epoch": 126.71566265060241, "grad_norm": 18.788814544677734, "learning_rate": 3.656310679611651e-05, "loss": 0.2295, "step": 13068 }, { "epoch": 126.72530120481927, "grad_norm": 12.48347282409668, "learning_rate": 3.655825242718447e-05, "loss": 0.3865, "step": 13069 }, { "epoch": 126.73493975903614, "grad_norm": 6.236184120178223, "learning_rate": 3.655339805825243e-05, "loss": 0.2004, "step": 13070 }, { "epoch": 126.74457831325302, "grad_norm": 10.196616172790527, "learning_rate": 3.654854368932039e-05, "loss": 0.3728, "step": 13071 }, { "epoch": 126.75421686746988, "grad_norm": 6.900184631347656, "learning_rate": 3.654368932038835e-05, "loss": 0.203, "step": 13072 }, { "epoch": 126.76385542168674, "grad_norm": 21.29669189453125, "learning_rate": 3.653883495145631e-05, "loss": 0.3459, "step": 13073 }, { "epoch": 126.77349397590362, "grad_norm": 10.035436630249023, "learning_rate": 3.6533980582524275e-05, "loss": 0.3395, "step": 13074 }, { "epoch": 126.78313253012048, "grad_norm": 4.54284143447876, "learning_rate": 3.652912621359223e-05, "loss": 0.1473, "step": 13075 }, { "epoch": 126.79277108433735, "grad_norm": 14.8949556350708, "learning_rate": 3.65242718446602e-05, "loss": 0.1757, "step": 13076 }, { "epoch": 126.80240963855422, "grad_norm": 8.61428451538086, "learning_rate": 3.6519417475728155e-05, "loss": 0.1823, "step": 13077 }, { "epoch": 126.81204819277109, "grad_norm": 12.474749565124512, "learning_rate": 3.651456310679612e-05, "loss": 0.2406, "step": 13078 }, { "epoch": 126.82168674698795, "grad_norm": 14.873189926147461, "learning_rate": 3.650970873786408e-05, "loss": 0.5049, "step": 13079 }, { "epoch": 126.83132530120481, "grad_norm": 13.767119407653809, "learning_rate": 3.650485436893204e-05, "loss": 0.3404, "step": 13080 }, { "epoch": 126.84096385542169, "grad_norm": 7.188912391662598, "learning_rate": 3.65e-05, "loss": 0.2622, "step": 13081 }, { "epoch": 126.85060240963855, "grad_norm": 8.804780960083008, "learning_rate": 3.6495145631067964e-05, "loss": 0.2249, "step": 13082 }, { "epoch": 126.86024096385542, "grad_norm": 23.02787971496582, "learning_rate": 3.649029126213592e-05, "loss": 0.5101, "step": 13083 }, { "epoch": 126.8698795180723, "grad_norm": 24.9980411529541, "learning_rate": 3.6485436893203886e-05, "loss": 0.431, "step": 13084 }, { "epoch": 126.87951807228916, "grad_norm": 32.37935256958008, "learning_rate": 3.6480582524271844e-05, "loss": 0.2462, "step": 13085 }, { "epoch": 126.88915662650602, "grad_norm": 13.2980375289917, "learning_rate": 3.647572815533981e-05, "loss": 0.2634, "step": 13086 }, { "epoch": 126.89879518072289, "grad_norm": 14.68677806854248, "learning_rate": 3.6470873786407766e-05, "loss": 0.3284, "step": 13087 }, { "epoch": 126.90843373493976, "grad_norm": 45.16200637817383, "learning_rate": 3.6466019417475724e-05, "loss": 0.1698, "step": 13088 }, { "epoch": 126.91807228915663, "grad_norm": 29.54915428161621, "learning_rate": 3.646116504854369e-05, "loss": 0.2126, "step": 13089 }, { "epoch": 126.92771084337349, "grad_norm": 23.013118743896484, "learning_rate": 3.645631067961165e-05, "loss": 0.4435, "step": 13090 }, { "epoch": 126.93734939759037, "grad_norm": 12.591145515441895, "learning_rate": 3.645145631067962e-05, "loss": 0.4703, "step": 13091 }, { "epoch": 126.94698795180723, "grad_norm": 7.43541145324707, "learning_rate": 3.6446601941747576e-05, "loss": 0.2285, "step": 13092 }, { "epoch": 126.9566265060241, "grad_norm": 20.180391311645508, "learning_rate": 3.644174757281554e-05, "loss": 0.2779, "step": 13093 }, { "epoch": 126.96626506024096, "grad_norm": 6.485978126525879, "learning_rate": 3.64368932038835e-05, "loss": 0.223, "step": 13094 }, { "epoch": 126.97590361445783, "grad_norm": 24.808168411254883, "learning_rate": 3.643203883495146e-05, "loss": 0.424, "step": 13095 }, { "epoch": 126.9855421686747, "grad_norm": 5.872963905334473, "learning_rate": 3.642718446601942e-05, "loss": 0.1559, "step": 13096 }, { "epoch": 127.0012048192771, "grad_norm": 12.871744155883789, "learning_rate": 3.6422330097087385e-05, "loss": 0.208, "step": 13097 }, { "epoch": 127.01084337349397, "grad_norm": 14.49215316772461, "learning_rate": 3.641747572815534e-05, "loss": 0.4951, "step": 13098 }, { "epoch": 127.02048192771085, "grad_norm": 13.969684600830078, "learning_rate": 3.64126213592233e-05, "loss": 0.4779, "step": 13099 }, { "epoch": 127.03012048192771, "grad_norm": 5.042593002319336, "learning_rate": 3.6407766990291265e-05, "loss": 0.1617, "step": 13100 }, { "epoch": 127.03975903614457, "grad_norm": 6.225045204162598, "learning_rate": 3.640291262135922e-05, "loss": 0.1877, "step": 13101 }, { "epoch": 127.04939759036145, "grad_norm": 10.857812881469727, "learning_rate": 3.639805825242719e-05, "loss": 0.3807, "step": 13102 }, { "epoch": 127.05903614457831, "grad_norm": 12.139082908630371, "learning_rate": 3.6393203883495145e-05, "loss": 0.646, "step": 13103 }, { "epoch": 127.06867469879518, "grad_norm": 75.11239624023438, "learning_rate": 3.638834951456311e-05, "loss": 0.2668, "step": 13104 }, { "epoch": 127.07831325301204, "grad_norm": 17.398387908935547, "learning_rate": 3.638349514563107e-05, "loss": 0.2864, "step": 13105 }, { "epoch": 127.08795180722892, "grad_norm": 13.31171703338623, "learning_rate": 3.637864077669903e-05, "loss": 0.4791, "step": 13106 }, { "epoch": 127.09759036144578, "grad_norm": 7.252829074859619, "learning_rate": 3.637378640776699e-05, "loss": 0.1819, "step": 13107 }, { "epoch": 127.10722891566265, "grad_norm": 11.06078815460205, "learning_rate": 3.6368932038834954e-05, "loss": 0.1426, "step": 13108 }, { "epoch": 127.11686746987952, "grad_norm": 10.961111068725586, "learning_rate": 3.636407766990291e-05, "loss": 0.32, "step": 13109 }, { "epoch": 127.12650602409639, "grad_norm": 6.308187484741211, "learning_rate": 3.6359223300970876e-05, "loss": 0.2048, "step": 13110 }, { "epoch": 127.13614457831325, "grad_norm": 3.948267936706543, "learning_rate": 3.6354368932038834e-05, "loss": 0.1162, "step": 13111 }, { "epoch": 127.14578313253013, "grad_norm": 12.104110717773438, "learning_rate": 3.63495145631068e-05, "loss": 0.2829, "step": 13112 }, { "epoch": 127.15542168674699, "grad_norm": 13.150528907775879, "learning_rate": 3.6344660194174756e-05, "loss": 0.4184, "step": 13113 }, { "epoch": 127.16506024096385, "grad_norm": 6.273075580596924, "learning_rate": 3.633980582524272e-05, "loss": 0.1175, "step": 13114 }, { "epoch": 127.17469879518072, "grad_norm": 7.4667463302612305, "learning_rate": 3.6334951456310685e-05, "loss": 0.2087, "step": 13115 }, { "epoch": 127.1843373493976, "grad_norm": 12.152153015136719, "learning_rate": 3.633009708737864e-05, "loss": 0.2431, "step": 13116 }, { "epoch": 127.19397590361446, "grad_norm": 7.959447860717773, "learning_rate": 3.632524271844661e-05, "loss": 0.1919, "step": 13117 }, { "epoch": 127.20361445783132, "grad_norm": 13.51253890991211, "learning_rate": 3.6320388349514565e-05, "loss": 0.3873, "step": 13118 }, { "epoch": 127.2132530120482, "grad_norm": 11.70482349395752, "learning_rate": 3.631553398058253e-05, "loss": 0.4733, "step": 13119 }, { "epoch": 127.22289156626506, "grad_norm": 11.30787467956543, "learning_rate": 3.631067961165049e-05, "loss": 0.4747, "step": 13120 }, { "epoch": 127.23253012048193, "grad_norm": 7.909267425537109, "learning_rate": 3.630582524271845e-05, "loss": 0.1417, "step": 13121 }, { "epoch": 127.24216867469879, "grad_norm": 9.48434066772461, "learning_rate": 3.630097087378641e-05, "loss": 0.2885, "step": 13122 }, { "epoch": 127.25180722891567, "grad_norm": 10.49758529663086, "learning_rate": 3.6296116504854374e-05, "loss": 0.3363, "step": 13123 }, { "epoch": 127.26144578313253, "grad_norm": 6.131147861480713, "learning_rate": 3.629126213592233e-05, "loss": 0.2609, "step": 13124 }, { "epoch": 127.2710843373494, "grad_norm": 4.709558486938477, "learning_rate": 3.62864077669903e-05, "loss": 0.0916, "step": 13125 }, { "epoch": 127.28072289156627, "grad_norm": 3.606057643890381, "learning_rate": 3.6281553398058254e-05, "loss": 0.1067, "step": 13126 }, { "epoch": 127.29036144578313, "grad_norm": 5.411302089691162, "learning_rate": 3.627669902912621e-05, "loss": 0.1461, "step": 13127 }, { "epoch": 127.3, "grad_norm": 9.046398162841797, "learning_rate": 3.627184466019418e-05, "loss": 0.2063, "step": 13128 }, { "epoch": 127.30963855421686, "grad_norm": 8.120741844177246, "learning_rate": 3.6266990291262135e-05, "loss": 0.2184, "step": 13129 }, { "epoch": 127.31927710843374, "grad_norm": 5.085209369659424, "learning_rate": 3.62621359223301e-05, "loss": 0.1967, "step": 13130 }, { "epoch": 127.3289156626506, "grad_norm": 16.734703063964844, "learning_rate": 3.625728155339806e-05, "loss": 0.1886, "step": 13131 }, { "epoch": 127.33855421686746, "grad_norm": 7.213728427886963, "learning_rate": 3.625242718446602e-05, "loss": 0.442, "step": 13132 }, { "epoch": 127.34819277108434, "grad_norm": 6.636667251586914, "learning_rate": 3.624757281553398e-05, "loss": 0.0987, "step": 13133 }, { "epoch": 127.3578313253012, "grad_norm": 13.097271919250488, "learning_rate": 3.6242718446601944e-05, "loss": 0.1776, "step": 13134 }, { "epoch": 127.36746987951807, "grad_norm": 12.305283546447754, "learning_rate": 3.62378640776699e-05, "loss": 0.3082, "step": 13135 }, { "epoch": 127.37710843373495, "grad_norm": 5.2034101486206055, "learning_rate": 3.6233009708737866e-05, "loss": 0.1097, "step": 13136 }, { "epoch": 127.38674698795181, "grad_norm": 7.126510143280029, "learning_rate": 3.6228155339805824e-05, "loss": 0.1587, "step": 13137 }, { "epoch": 127.39638554216867, "grad_norm": 10.026192665100098, "learning_rate": 3.622330097087379e-05, "loss": 0.238, "step": 13138 }, { "epoch": 127.40602409638554, "grad_norm": 13.27188777923584, "learning_rate": 3.6218446601941746e-05, "loss": 0.1957, "step": 13139 }, { "epoch": 127.41566265060241, "grad_norm": 28.8420352935791, "learning_rate": 3.621359223300971e-05, "loss": 0.308, "step": 13140 }, { "epoch": 127.42530120481928, "grad_norm": 14.020944595336914, "learning_rate": 3.6208737864077675e-05, "loss": 0.3582, "step": 13141 }, { "epoch": 127.43493975903614, "grad_norm": 6.589940547943115, "learning_rate": 3.620388349514563e-05, "loss": 0.2269, "step": 13142 }, { "epoch": 127.44457831325302, "grad_norm": 6.582858085632324, "learning_rate": 3.61990291262136e-05, "loss": 0.2021, "step": 13143 }, { "epoch": 127.45421686746988, "grad_norm": 20.438108444213867, "learning_rate": 3.6194174757281555e-05, "loss": 0.3656, "step": 13144 }, { "epoch": 127.46385542168674, "grad_norm": 39.83004379272461, "learning_rate": 3.618932038834952e-05, "loss": 0.3161, "step": 13145 }, { "epoch": 127.47349397590361, "grad_norm": 11.370085716247559, "learning_rate": 3.618446601941748e-05, "loss": 0.3154, "step": 13146 }, { "epoch": 127.48313253012049, "grad_norm": 6.257630348205566, "learning_rate": 3.617961165048544e-05, "loss": 0.1231, "step": 13147 }, { "epoch": 127.49277108433735, "grad_norm": 13.5629243850708, "learning_rate": 3.61747572815534e-05, "loss": 0.2377, "step": 13148 }, { "epoch": 127.50240963855421, "grad_norm": 9.090927124023438, "learning_rate": 3.6169902912621364e-05, "loss": 0.2771, "step": 13149 }, { "epoch": 127.51204819277109, "grad_norm": 20.191226959228516, "learning_rate": 3.616504854368932e-05, "loss": 0.4505, "step": 13150 }, { "epoch": 127.52168674698795, "grad_norm": 18.538360595703125, "learning_rate": 3.6160194174757286e-05, "loss": 0.5773, "step": 13151 }, { "epoch": 127.53132530120482, "grad_norm": 5.243191719055176, "learning_rate": 3.6155339805825244e-05, "loss": 0.1094, "step": 13152 }, { "epoch": 127.54096385542168, "grad_norm": 6.722772598266602, "learning_rate": 3.615048543689321e-05, "loss": 0.2463, "step": 13153 }, { "epoch": 127.55060240963856, "grad_norm": 7.776521682739258, "learning_rate": 3.6145631067961167e-05, "loss": 0.2088, "step": 13154 }, { "epoch": 127.56024096385542, "grad_norm": 8.515296936035156, "learning_rate": 3.6140776699029124e-05, "loss": 0.2569, "step": 13155 }, { "epoch": 127.56987951807228, "grad_norm": 5.091420650482178, "learning_rate": 3.613592233009709e-05, "loss": 0.218, "step": 13156 }, { "epoch": 127.57951807228916, "grad_norm": 18.05889320373535, "learning_rate": 3.6131067961165047e-05, "loss": 0.181, "step": 13157 }, { "epoch": 127.58915662650602, "grad_norm": 10.733321189880371, "learning_rate": 3.612621359223301e-05, "loss": 0.3997, "step": 13158 }, { "epoch": 127.59879518072289, "grad_norm": 33.658573150634766, "learning_rate": 3.612135922330097e-05, "loss": 0.3039, "step": 13159 }, { "epoch": 127.60843373493977, "grad_norm": 12.79481029510498, "learning_rate": 3.6116504854368933e-05, "loss": 0.1006, "step": 13160 }, { "epoch": 127.61807228915663, "grad_norm": 10.741679191589355, "learning_rate": 3.611165048543689e-05, "loss": 0.3295, "step": 13161 }, { "epoch": 127.62771084337349, "grad_norm": 14.686098098754883, "learning_rate": 3.6106796116504856e-05, "loss": 0.3231, "step": 13162 }, { "epoch": 127.63734939759036, "grad_norm": 19.647235870361328, "learning_rate": 3.6101941747572813e-05, "loss": 0.2196, "step": 13163 }, { "epoch": 127.64698795180723, "grad_norm": 9.43048095703125, "learning_rate": 3.609708737864078e-05, "loss": 0.2326, "step": 13164 }, { "epoch": 127.6566265060241, "grad_norm": 13.888028144836426, "learning_rate": 3.6092233009708736e-05, "loss": 0.2574, "step": 13165 }, { "epoch": 127.66626506024096, "grad_norm": 5.717218399047852, "learning_rate": 3.60873786407767e-05, "loss": 0.1789, "step": 13166 }, { "epoch": 127.67590361445784, "grad_norm": 9.83895492553711, "learning_rate": 3.6082524271844665e-05, "loss": 0.4309, "step": 13167 }, { "epoch": 127.6855421686747, "grad_norm": 15.602463722229004, "learning_rate": 3.607766990291262e-05, "loss": 0.401, "step": 13168 }, { "epoch": 127.69518072289156, "grad_norm": 22.76949119567871, "learning_rate": 3.607281553398059e-05, "loss": 0.3867, "step": 13169 }, { "epoch": 127.70481927710843, "grad_norm": 8.12835693359375, "learning_rate": 3.6067961165048545e-05, "loss": 0.1856, "step": 13170 }, { "epoch": 127.7144578313253, "grad_norm": 8.46259880065918, "learning_rate": 3.606310679611651e-05, "loss": 0.3069, "step": 13171 }, { "epoch": 127.72409638554217, "grad_norm": 14.632832527160645, "learning_rate": 3.605825242718447e-05, "loss": 0.4138, "step": 13172 }, { "epoch": 127.73373493975903, "grad_norm": 15.265703201293945, "learning_rate": 3.605339805825243e-05, "loss": 0.2565, "step": 13173 }, { "epoch": 127.74337349397591, "grad_norm": 15.038838386535645, "learning_rate": 3.604854368932039e-05, "loss": 0.3387, "step": 13174 }, { "epoch": 127.75301204819277, "grad_norm": 7.163206577301025, "learning_rate": 3.6043689320388354e-05, "loss": 0.2796, "step": 13175 }, { "epoch": 127.76265060240964, "grad_norm": 12.68705940246582, "learning_rate": 3.603883495145631e-05, "loss": 0.4306, "step": 13176 }, { "epoch": 127.7722891566265, "grad_norm": 22.694820404052734, "learning_rate": 3.6033980582524276e-05, "loss": 0.2042, "step": 13177 }, { "epoch": 127.78192771084338, "grad_norm": 9.853740692138672, "learning_rate": 3.6029126213592234e-05, "loss": 0.3478, "step": 13178 }, { "epoch": 127.79156626506024, "grad_norm": 11.919416427612305, "learning_rate": 3.60242718446602e-05, "loss": 0.496, "step": 13179 }, { "epoch": 127.8012048192771, "grad_norm": 21.404558181762695, "learning_rate": 3.6019417475728156e-05, "loss": 0.2671, "step": 13180 }, { "epoch": 127.81084337349398, "grad_norm": 12.704216003417969, "learning_rate": 3.601456310679612e-05, "loss": 0.2137, "step": 13181 }, { "epoch": 127.82048192771084, "grad_norm": 13.078060150146484, "learning_rate": 3.600970873786408e-05, "loss": 0.2737, "step": 13182 }, { "epoch": 127.83012048192771, "grad_norm": 8.1673002243042, "learning_rate": 3.6004854368932036e-05, "loss": 0.3023, "step": 13183 }, { "epoch": 127.83975903614459, "grad_norm": 12.405457496643066, "learning_rate": 3.6e-05, "loss": 0.1783, "step": 13184 }, { "epoch": 127.84939759036145, "grad_norm": 10.008768081665039, "learning_rate": 3.599514563106796e-05, "loss": 0.2584, "step": 13185 }, { "epoch": 127.85903614457831, "grad_norm": 11.344829559326172, "learning_rate": 3.599029126213592e-05, "loss": 0.257, "step": 13186 }, { "epoch": 127.86867469879518, "grad_norm": 8.041621208190918, "learning_rate": 3.598543689320388e-05, "loss": 0.2666, "step": 13187 }, { "epoch": 127.87831325301205, "grad_norm": 8.839455604553223, "learning_rate": 3.5980582524271845e-05, "loss": 0.2107, "step": 13188 }, { "epoch": 127.88795180722892, "grad_norm": 30.73690414428711, "learning_rate": 3.59757281553398e-05, "loss": 0.2445, "step": 13189 }, { "epoch": 127.89759036144578, "grad_norm": 19.451763153076172, "learning_rate": 3.597087378640777e-05, "loss": 0.3414, "step": 13190 }, { "epoch": 127.90722891566266, "grad_norm": 10.814473152160645, "learning_rate": 3.596601941747573e-05, "loss": 0.5137, "step": 13191 }, { "epoch": 127.91686746987952, "grad_norm": 18.061594009399414, "learning_rate": 3.59611650485437e-05, "loss": 0.1246, "step": 13192 }, { "epoch": 127.92650602409638, "grad_norm": 6.39409065246582, "learning_rate": 3.5956310679611655e-05, "loss": 0.2811, "step": 13193 }, { "epoch": 127.93614457831325, "grad_norm": 4.587752342224121, "learning_rate": 3.595145631067961e-05, "loss": 0.1435, "step": 13194 }, { "epoch": 127.94578313253012, "grad_norm": 11.398616790771484, "learning_rate": 3.594660194174758e-05, "loss": 0.2561, "step": 13195 }, { "epoch": 127.95542168674699, "grad_norm": 7.5609211921691895, "learning_rate": 3.5941747572815535e-05, "loss": 0.3, "step": 13196 }, { "epoch": 127.96506024096385, "grad_norm": 7.146437168121338, "learning_rate": 3.59368932038835e-05, "loss": 0.2376, "step": 13197 }, { "epoch": 127.97469879518073, "grad_norm": 8.703146934509277, "learning_rate": 3.593203883495146e-05, "loss": 0.2805, "step": 13198 }, { "epoch": 127.98433734939759, "grad_norm": 6.674076080322266, "learning_rate": 3.592718446601942e-05, "loss": 0.2727, "step": 13199 }, { "epoch": 127.99397590361446, "grad_norm": 9.340500831604004, "learning_rate": 3.592233009708738e-05, "loss": 0.3981, "step": 13200 }, { "epoch": 128.00963855421688, "grad_norm": 10.092182159423828, "learning_rate": 3.5917475728155344e-05, "loss": 0.3091, "step": 13201 }, { "epoch": 128.01927710843373, "grad_norm": 4.715688228607178, "learning_rate": 3.59126213592233e-05, "loss": 0.0981, "step": 13202 }, { "epoch": 128.0289156626506, "grad_norm": 15.825118064880371, "learning_rate": 3.5907766990291266e-05, "loss": 0.4763, "step": 13203 }, { "epoch": 128.03855421686748, "grad_norm": 9.448943138122559, "learning_rate": 3.5902912621359224e-05, "loss": 0.2969, "step": 13204 }, { "epoch": 128.04819277108433, "grad_norm": 16.698183059692383, "learning_rate": 3.589805825242719e-05, "loss": 0.5289, "step": 13205 }, { "epoch": 128.0578313253012, "grad_norm": 10.79308032989502, "learning_rate": 3.5893203883495146e-05, "loss": 0.4089, "step": 13206 }, { "epoch": 128.0674698795181, "grad_norm": 15.590380668640137, "learning_rate": 3.588834951456311e-05, "loss": 0.5009, "step": 13207 }, { "epoch": 128.07710843373494, "grad_norm": 11.803133010864258, "learning_rate": 3.588349514563107e-05, "loss": 0.2445, "step": 13208 }, { "epoch": 128.0867469879518, "grad_norm": 8.9856595993042, "learning_rate": 3.587864077669903e-05, "loss": 0.2369, "step": 13209 }, { "epoch": 128.09638554216866, "grad_norm": 22.924436569213867, "learning_rate": 3.587378640776699e-05, "loss": 0.204, "step": 13210 }, { "epoch": 128.10602409638554, "grad_norm": 14.749564170837402, "learning_rate": 3.586893203883495e-05, "loss": 0.4783, "step": 13211 }, { "epoch": 128.11566265060242, "grad_norm": 13.619800567626953, "learning_rate": 3.586407766990291e-05, "loss": 0.2707, "step": 13212 }, { "epoch": 128.12530120481927, "grad_norm": 10.830492973327637, "learning_rate": 3.585922330097087e-05, "loss": 0.188, "step": 13213 }, { "epoch": 128.13493975903614, "grad_norm": 9.492607116699219, "learning_rate": 3.5854368932038835e-05, "loss": 0.343, "step": 13214 }, { "epoch": 128.14457831325302, "grad_norm": 12.346602439880371, "learning_rate": 3.584951456310679e-05, "loss": 0.2931, "step": 13215 }, { "epoch": 128.15421686746987, "grad_norm": 13.205413818359375, "learning_rate": 3.5844660194174764e-05, "loss": 0.2707, "step": 13216 }, { "epoch": 128.16385542168675, "grad_norm": 9.707411766052246, "learning_rate": 3.583980582524272e-05, "loss": 0.2928, "step": 13217 }, { "epoch": 128.17349397590363, "grad_norm": 11.42790412902832, "learning_rate": 3.583495145631069e-05, "loss": 0.4538, "step": 13218 }, { "epoch": 128.18313253012047, "grad_norm": 20.862627029418945, "learning_rate": 3.5830097087378644e-05, "loss": 0.2959, "step": 13219 }, { "epoch": 128.19277108433735, "grad_norm": 10.397730827331543, "learning_rate": 3.58252427184466e-05, "loss": 0.2677, "step": 13220 }, { "epoch": 128.20240963855423, "grad_norm": 10.002286911010742, "learning_rate": 3.582038834951457e-05, "loss": 0.3076, "step": 13221 }, { "epoch": 128.21204819277108, "grad_norm": 5.697789669036865, "learning_rate": 3.5815533980582524e-05, "loss": 0.2388, "step": 13222 }, { "epoch": 128.22168674698796, "grad_norm": 18.2866268157959, "learning_rate": 3.581067961165049e-05, "loss": 0.2194, "step": 13223 }, { "epoch": 128.2313253012048, "grad_norm": 5.4825592041015625, "learning_rate": 3.580582524271845e-05, "loss": 0.1795, "step": 13224 }, { "epoch": 128.24096385542168, "grad_norm": 11.597898483276367, "learning_rate": 3.580097087378641e-05, "loss": 0.2233, "step": 13225 }, { "epoch": 128.25060240963856, "grad_norm": 7.024440288543701, "learning_rate": 3.579611650485437e-05, "loss": 0.1698, "step": 13226 }, { "epoch": 128.2602409638554, "grad_norm": 18.62184715270996, "learning_rate": 3.5791262135922334e-05, "loss": 0.4754, "step": 13227 }, { "epoch": 128.2698795180723, "grad_norm": 8.9465913772583, "learning_rate": 3.578640776699029e-05, "loss": 0.2957, "step": 13228 }, { "epoch": 128.27951807228916, "grad_norm": 18.729259490966797, "learning_rate": 3.5781553398058256e-05, "loss": 0.2275, "step": 13229 }, { "epoch": 128.289156626506, "grad_norm": 19.604726791381836, "learning_rate": 3.5776699029126214e-05, "loss": 0.3339, "step": 13230 }, { "epoch": 128.2987951807229, "grad_norm": 11.602655410766602, "learning_rate": 3.577184466019418e-05, "loss": 0.1156, "step": 13231 }, { "epoch": 128.30843373493977, "grad_norm": 8.721965789794922, "learning_rate": 3.5766990291262136e-05, "loss": 0.3822, "step": 13232 }, { "epoch": 128.31807228915662, "grad_norm": 9.083239555358887, "learning_rate": 3.57621359223301e-05, "loss": 0.3563, "step": 13233 }, { "epoch": 128.3277108433735, "grad_norm": 8.806124687194824, "learning_rate": 3.575728155339806e-05, "loss": 0.2055, "step": 13234 }, { "epoch": 128.33734939759037, "grad_norm": 8.124785423278809, "learning_rate": 3.575242718446602e-05, "loss": 0.2728, "step": 13235 }, { "epoch": 128.34698795180722, "grad_norm": 5.37874174118042, "learning_rate": 3.574757281553398e-05, "loss": 0.1408, "step": 13236 }, { "epoch": 128.3566265060241, "grad_norm": 22.37471580505371, "learning_rate": 3.574271844660194e-05, "loss": 0.2944, "step": 13237 }, { "epoch": 128.36626506024098, "grad_norm": 8.75890064239502, "learning_rate": 3.57378640776699e-05, "loss": 0.2051, "step": 13238 }, { "epoch": 128.37590361445783, "grad_norm": 5.009967803955078, "learning_rate": 3.573300970873786e-05, "loss": 0.1696, "step": 13239 }, { "epoch": 128.3855421686747, "grad_norm": 9.564865112304688, "learning_rate": 3.5728155339805825e-05, "loss": 0.2833, "step": 13240 }, { "epoch": 128.39518072289155, "grad_norm": 11.38742733001709, "learning_rate": 3.572330097087378e-05, "loss": 0.2663, "step": 13241 }, { "epoch": 128.40481927710843, "grad_norm": 10.23438549041748, "learning_rate": 3.5718446601941754e-05, "loss": 0.3955, "step": 13242 }, { "epoch": 128.4144578313253, "grad_norm": 114.91336059570312, "learning_rate": 3.571359223300971e-05, "loss": 0.5087, "step": 13243 }, { "epoch": 128.42409638554216, "grad_norm": 8.166593551635742, "learning_rate": 3.5708737864077676e-05, "loss": 0.278, "step": 13244 }, { "epoch": 128.43373493975903, "grad_norm": 7.744853973388672, "learning_rate": 3.5703883495145634e-05, "loss": 0.3486, "step": 13245 }, { "epoch": 128.4433734939759, "grad_norm": 8.563820838928223, "learning_rate": 3.56990291262136e-05, "loss": 0.4143, "step": 13246 }, { "epoch": 128.45301204819276, "grad_norm": 4.861521244049072, "learning_rate": 3.5694174757281556e-05, "loss": 0.0638, "step": 13247 }, { "epoch": 128.46265060240964, "grad_norm": 6.897616386413574, "learning_rate": 3.5689320388349514e-05, "loss": 0.2912, "step": 13248 }, { "epoch": 128.47228915662652, "grad_norm": 14.744654655456543, "learning_rate": 3.568446601941748e-05, "loss": 0.4868, "step": 13249 }, { "epoch": 128.48192771084337, "grad_norm": 6.679169654846191, "learning_rate": 3.5679611650485437e-05, "loss": 0.1415, "step": 13250 }, { "epoch": 128.49156626506024, "grad_norm": 9.01889705657959, "learning_rate": 3.56747572815534e-05, "loss": 0.3654, "step": 13251 }, { "epoch": 128.50120481927712, "grad_norm": 11.808213233947754, "learning_rate": 3.566990291262136e-05, "loss": 0.3964, "step": 13252 }, { "epoch": 128.51084337349397, "grad_norm": 11.134289741516113, "learning_rate": 3.566504854368932e-05, "loss": 0.2945, "step": 13253 }, { "epoch": 128.52048192771085, "grad_norm": 20.549951553344727, "learning_rate": 3.566019417475728e-05, "loss": 0.4294, "step": 13254 }, { "epoch": 128.53012048192772, "grad_norm": 7.269767761230469, "learning_rate": 3.5655339805825246e-05, "loss": 0.1852, "step": 13255 }, { "epoch": 128.53975903614457, "grad_norm": 8.056336402893066, "learning_rate": 3.5650485436893203e-05, "loss": 0.2731, "step": 13256 }, { "epoch": 128.54939759036145, "grad_norm": 18.91197967529297, "learning_rate": 3.564563106796117e-05, "loss": 0.2726, "step": 13257 }, { "epoch": 128.5590361445783, "grad_norm": 19.960664749145508, "learning_rate": 3.5640776699029126e-05, "loss": 0.2252, "step": 13258 }, { "epoch": 128.56867469879518, "grad_norm": 14.6453275680542, "learning_rate": 3.563592233009709e-05, "loss": 0.2983, "step": 13259 }, { "epoch": 128.57831325301206, "grad_norm": 9.763339042663574, "learning_rate": 3.563106796116505e-05, "loss": 0.1678, "step": 13260 }, { "epoch": 128.5879518072289, "grad_norm": 12.98136043548584, "learning_rate": 3.562621359223301e-05, "loss": 0.4097, "step": 13261 }, { "epoch": 128.59759036144578, "grad_norm": 7.180098056793213, "learning_rate": 3.562135922330097e-05, "loss": 0.2514, "step": 13262 }, { "epoch": 128.60722891566266, "grad_norm": 11.57660961151123, "learning_rate": 3.5616504854368935e-05, "loss": 0.1981, "step": 13263 }, { "epoch": 128.6168674698795, "grad_norm": 18.644432067871094, "learning_rate": 3.561165048543689e-05, "loss": 0.2978, "step": 13264 }, { "epoch": 128.6265060240964, "grad_norm": 10.878772735595703, "learning_rate": 3.560679611650485e-05, "loss": 0.2596, "step": 13265 }, { "epoch": 128.63614457831326, "grad_norm": 9.78213882446289, "learning_rate": 3.5601941747572815e-05, "loss": 0.3374, "step": 13266 }, { "epoch": 128.6457831325301, "grad_norm": 6.866717338562012, "learning_rate": 3.559708737864078e-05, "loss": 0.1817, "step": 13267 }, { "epoch": 128.655421686747, "grad_norm": 22.098777770996094, "learning_rate": 3.5592233009708744e-05, "loss": 0.3084, "step": 13268 }, { "epoch": 128.66506024096387, "grad_norm": 15.48529052734375, "learning_rate": 3.55873786407767e-05, "loss": 0.4819, "step": 13269 }, { "epoch": 128.67469879518072, "grad_norm": 5.408236026763916, "learning_rate": 3.5582524271844666e-05, "loss": 0.1679, "step": 13270 }, { "epoch": 128.6843373493976, "grad_norm": 9.8185396194458, "learning_rate": 3.5577669902912624e-05, "loss": 0.331, "step": 13271 }, { "epoch": 128.69397590361444, "grad_norm": 21.248851776123047, "learning_rate": 3.557281553398059e-05, "loss": 0.282, "step": 13272 }, { "epoch": 128.70361445783132, "grad_norm": 8.793622970581055, "learning_rate": 3.5567961165048546e-05, "loss": 0.3094, "step": 13273 }, { "epoch": 128.7132530120482, "grad_norm": 10.196440696716309, "learning_rate": 3.556310679611651e-05, "loss": 0.254, "step": 13274 }, { "epoch": 128.72289156626505, "grad_norm": 6.369532585144043, "learning_rate": 3.555825242718447e-05, "loss": 0.1556, "step": 13275 }, { "epoch": 128.73253012048193, "grad_norm": 20.78877067565918, "learning_rate": 3.5553398058252426e-05, "loss": 0.3102, "step": 13276 }, { "epoch": 128.7421686746988, "grad_norm": 10.96828556060791, "learning_rate": 3.554854368932039e-05, "loss": 0.2029, "step": 13277 }, { "epoch": 128.75180722891565, "grad_norm": 11.858951568603516, "learning_rate": 3.554368932038835e-05, "loss": 0.2912, "step": 13278 }, { "epoch": 128.76144578313253, "grad_norm": 13.101898193359375, "learning_rate": 3.553883495145631e-05, "loss": 0.1667, "step": 13279 }, { "epoch": 128.7710843373494, "grad_norm": 24.224525451660156, "learning_rate": 3.553398058252427e-05, "loss": 0.4283, "step": 13280 }, { "epoch": 128.78072289156626, "grad_norm": 5.508083343505859, "learning_rate": 3.5529126213592235e-05, "loss": 0.1405, "step": 13281 }, { "epoch": 128.79036144578313, "grad_norm": 9.29611873626709, "learning_rate": 3.552427184466019e-05, "loss": 0.2065, "step": 13282 }, { "epoch": 128.8, "grad_norm": 8.382217407226562, "learning_rate": 3.551941747572816e-05, "loss": 0.3676, "step": 13283 }, { "epoch": 128.80963855421686, "grad_norm": 13.38472843170166, "learning_rate": 3.5514563106796115e-05, "loss": 0.1736, "step": 13284 }, { "epoch": 128.81927710843374, "grad_norm": 14.797640800476074, "learning_rate": 3.550970873786408e-05, "loss": 0.3488, "step": 13285 }, { "epoch": 128.82891566265062, "grad_norm": 10.665985107421875, "learning_rate": 3.550485436893204e-05, "loss": 0.2122, "step": 13286 }, { "epoch": 128.83855421686746, "grad_norm": 11.4380464553833, "learning_rate": 3.55e-05, "loss": 0.4237, "step": 13287 }, { "epoch": 128.84819277108434, "grad_norm": 6.2103657722473145, "learning_rate": 3.549514563106796e-05, "loss": 0.3067, "step": 13288 }, { "epoch": 128.8578313253012, "grad_norm": 11.2964448928833, "learning_rate": 3.5490291262135925e-05, "loss": 0.3257, "step": 13289 }, { "epoch": 128.86746987951807, "grad_norm": 7.113888263702393, "learning_rate": 3.548543689320388e-05, "loss": 0.327, "step": 13290 }, { "epoch": 128.87710843373495, "grad_norm": 9.68271255493164, "learning_rate": 3.548058252427185e-05, "loss": 0.6449, "step": 13291 }, { "epoch": 128.8867469879518, "grad_norm": 13.656621932983398, "learning_rate": 3.547572815533981e-05, "loss": 0.2616, "step": 13292 }, { "epoch": 128.89638554216867, "grad_norm": 11.463786125183105, "learning_rate": 3.547087378640777e-05, "loss": 0.2435, "step": 13293 }, { "epoch": 128.90602409638555, "grad_norm": 8.39483642578125, "learning_rate": 3.5466019417475734e-05, "loss": 0.2725, "step": 13294 }, { "epoch": 128.9156626506024, "grad_norm": 6.943198204040527, "learning_rate": 3.546116504854369e-05, "loss": 0.1915, "step": 13295 }, { "epoch": 128.92530120481928, "grad_norm": 7.842589378356934, "learning_rate": 3.5456310679611656e-05, "loss": 0.2207, "step": 13296 }, { "epoch": 128.93493975903615, "grad_norm": 12.887874603271484, "learning_rate": 3.5451456310679614e-05, "loss": 0.3448, "step": 13297 }, { "epoch": 128.944578313253, "grad_norm": 12.964615821838379, "learning_rate": 3.544660194174758e-05, "loss": 0.1748, "step": 13298 }, { "epoch": 128.95421686746988, "grad_norm": 8.399577140808105, "learning_rate": 3.5441747572815536e-05, "loss": 0.2335, "step": 13299 }, { "epoch": 128.96385542168676, "grad_norm": 12.454742431640625, "learning_rate": 3.54368932038835e-05, "loss": 0.2848, "step": 13300 }, { "epoch": 128.9734939759036, "grad_norm": 9.841014862060547, "learning_rate": 3.543203883495146e-05, "loss": 0.446, "step": 13301 }, { "epoch": 128.98313253012049, "grad_norm": 9.012500762939453, "learning_rate": 3.542718446601942e-05, "loss": 0.188, "step": 13302 }, { "epoch": 128.99277108433736, "grad_norm": 15.620477676391602, "learning_rate": 3.542233009708738e-05, "loss": 0.4039, "step": 13303 }, { "epoch": 129.00843373493976, "grad_norm": 52.431968688964844, "learning_rate": 3.541747572815534e-05, "loss": 0.5461, "step": 13304 }, { "epoch": 129.01807228915663, "grad_norm": 19.164037704467773, "learning_rate": 3.54126213592233e-05, "loss": 0.1845, "step": 13305 }, { "epoch": 129.02771084337348, "grad_norm": 5.118994235992432, "learning_rate": 3.540776699029126e-05, "loss": 0.2346, "step": 13306 }, { "epoch": 129.03734939759036, "grad_norm": 14.863839149475098, "learning_rate": 3.5402912621359225e-05, "loss": 0.3436, "step": 13307 }, { "epoch": 129.04698795180724, "grad_norm": 21.447509765625, "learning_rate": 3.539805825242718e-05, "loss": 0.3416, "step": 13308 }, { "epoch": 129.0566265060241, "grad_norm": 3.8132500648498535, "learning_rate": 3.539320388349515e-05, "loss": 0.1947, "step": 13309 }, { "epoch": 129.06626506024097, "grad_norm": 22.2407283782959, "learning_rate": 3.5388349514563105e-05, "loss": 0.4379, "step": 13310 }, { "epoch": 129.07590361445784, "grad_norm": 8.968184471130371, "learning_rate": 3.538349514563107e-05, "loss": 0.3543, "step": 13311 }, { "epoch": 129.0855421686747, "grad_norm": 12.390690803527832, "learning_rate": 3.537864077669903e-05, "loss": 0.1175, "step": 13312 }, { "epoch": 129.09518072289157, "grad_norm": 11.771187782287598, "learning_rate": 3.537378640776699e-05, "loss": 0.1846, "step": 13313 }, { "epoch": 129.10481927710845, "grad_norm": 55.88069152832031, "learning_rate": 3.536893203883495e-05, "loss": 0.3578, "step": 13314 }, { "epoch": 129.1144578313253, "grad_norm": 41.46791076660156, "learning_rate": 3.5364077669902914e-05, "loss": 0.2215, "step": 13315 }, { "epoch": 129.12409638554217, "grad_norm": 6.2523193359375, "learning_rate": 3.535922330097087e-05, "loss": 0.4223, "step": 13316 }, { "epoch": 129.13373493975902, "grad_norm": 19.44695472717285, "learning_rate": 3.535436893203884e-05, "loss": 0.2364, "step": 13317 }, { "epoch": 129.1433734939759, "grad_norm": 72.20191192626953, "learning_rate": 3.53495145631068e-05, "loss": 0.3798, "step": 13318 }, { "epoch": 129.15301204819278, "grad_norm": 3.696922779083252, "learning_rate": 3.534466019417476e-05, "loss": 0.259, "step": 13319 }, { "epoch": 129.16265060240963, "grad_norm": 29.84897232055664, "learning_rate": 3.5339805825242724e-05, "loss": 0.2535, "step": 13320 }, { "epoch": 129.1722891566265, "grad_norm": 2.7088258266448975, "learning_rate": 3.533495145631068e-05, "loss": 0.4843, "step": 13321 }, { "epoch": 129.18192771084338, "grad_norm": 14.620340347290039, "learning_rate": 3.5330097087378646e-05, "loss": 0.1296, "step": 13322 }, { "epoch": 129.19156626506023, "grad_norm": 30.349170684814453, "learning_rate": 3.5325242718446604e-05, "loss": 0.3633, "step": 13323 }, { "epoch": 129.2012048192771, "grad_norm": 4.250511169433594, "learning_rate": 3.532038834951457e-05, "loss": 0.2547, "step": 13324 }, { "epoch": 129.210843373494, "grad_norm": 36.26278305053711, "learning_rate": 3.5315533980582526e-05, "loss": 0.3603, "step": 13325 }, { "epoch": 129.22048192771084, "grad_norm": 17.889591217041016, "learning_rate": 3.531067961165049e-05, "loss": 0.4693, "step": 13326 }, { "epoch": 129.2301204819277, "grad_norm": 3.648869276046753, "learning_rate": 3.530582524271845e-05, "loss": 0.2977, "step": 13327 }, { "epoch": 129.2397590361446, "grad_norm": 37.49250411987305, "learning_rate": 3.530097087378641e-05, "loss": 0.5192, "step": 13328 }, { "epoch": 129.24939759036144, "grad_norm": 15.446989059448242, "learning_rate": 3.529611650485437e-05, "loss": 0.3006, "step": 13329 }, { "epoch": 129.25903614457832, "grad_norm": 18.876781463623047, "learning_rate": 3.5291262135922335e-05, "loss": 0.4468, "step": 13330 }, { "epoch": 129.2686746987952, "grad_norm": 9.46529483795166, "learning_rate": 3.528640776699029e-05, "loss": 0.3333, "step": 13331 }, { "epoch": 129.27831325301204, "grad_norm": 5.050840377807617, "learning_rate": 3.528155339805825e-05, "loss": 0.1735, "step": 13332 }, { "epoch": 129.28795180722892, "grad_norm": 16.284849166870117, "learning_rate": 3.5276699029126215e-05, "loss": 0.2396, "step": 13333 }, { "epoch": 129.29759036144577, "grad_norm": 17.22430419921875, "learning_rate": 3.527184466019417e-05, "loss": 0.2124, "step": 13334 }, { "epoch": 129.30722891566265, "grad_norm": 7.562819004058838, "learning_rate": 3.526699029126214e-05, "loss": 0.2566, "step": 13335 }, { "epoch": 129.31686746987953, "grad_norm": 4.242676734924316, "learning_rate": 3.5262135922330095e-05, "loss": 0.1479, "step": 13336 }, { "epoch": 129.32650602409637, "grad_norm": 40.381534576416016, "learning_rate": 3.525728155339806e-05, "loss": 0.2326, "step": 13337 }, { "epoch": 129.33614457831325, "grad_norm": 9.434775352478027, "learning_rate": 3.525242718446602e-05, "loss": 0.3094, "step": 13338 }, { "epoch": 129.34578313253013, "grad_norm": 13.161039352416992, "learning_rate": 3.524757281553398e-05, "loss": 0.193, "step": 13339 }, { "epoch": 129.35542168674698, "grad_norm": 17.032052993774414, "learning_rate": 3.524271844660194e-05, "loss": 0.3507, "step": 13340 }, { "epoch": 129.36506024096386, "grad_norm": 4.134231090545654, "learning_rate": 3.5237864077669904e-05, "loss": 0.2486, "step": 13341 }, { "epoch": 129.37469879518073, "grad_norm": 8.666984558105469, "learning_rate": 3.523300970873786e-05, "loss": 0.3107, "step": 13342 }, { "epoch": 129.38433734939758, "grad_norm": 11.239477157592773, "learning_rate": 3.5228155339805826e-05, "loss": 0.1803, "step": 13343 }, { "epoch": 129.39397590361446, "grad_norm": 31.01139259338379, "learning_rate": 3.522330097087379e-05, "loss": 0.1644, "step": 13344 }, { "epoch": 129.40361445783134, "grad_norm": 10.982146263122559, "learning_rate": 3.521844660194175e-05, "loss": 0.3528, "step": 13345 }, { "epoch": 129.4132530120482, "grad_norm": 6.6284613609313965, "learning_rate": 3.521359223300971e-05, "loss": 0.3128, "step": 13346 }, { "epoch": 129.42289156626506, "grad_norm": 44.04011154174805, "learning_rate": 3.520873786407767e-05, "loss": 0.39, "step": 13347 }, { "epoch": 129.4325301204819, "grad_norm": 19.88168716430664, "learning_rate": 3.5203883495145636e-05, "loss": 0.2757, "step": 13348 }, { "epoch": 129.4421686746988, "grad_norm": 37.45218276977539, "learning_rate": 3.519902912621359e-05, "loss": 0.3123, "step": 13349 }, { "epoch": 129.45180722891567, "grad_norm": 20.959264755249023, "learning_rate": 3.519417475728156e-05, "loss": 0.3252, "step": 13350 }, { "epoch": 129.46144578313252, "grad_norm": 33.121360778808594, "learning_rate": 3.5189320388349516e-05, "loss": 0.393, "step": 13351 }, { "epoch": 129.4710843373494, "grad_norm": 25.253887176513672, "learning_rate": 3.518446601941748e-05, "loss": 0.0834, "step": 13352 }, { "epoch": 129.48072289156627, "grad_norm": 35.258113861083984, "learning_rate": 3.517961165048544e-05, "loss": 0.3075, "step": 13353 }, { "epoch": 129.49036144578312, "grad_norm": 7.1450982093811035, "learning_rate": 3.51747572815534e-05, "loss": 0.1088, "step": 13354 }, { "epoch": 129.5, "grad_norm": 3.8429768085479736, "learning_rate": 3.516990291262136e-05, "loss": 0.242, "step": 13355 }, { "epoch": 129.50963855421688, "grad_norm": 18.459590911865234, "learning_rate": 3.5165048543689325e-05, "loss": 0.3311, "step": 13356 }, { "epoch": 129.51927710843373, "grad_norm": 17.269264221191406, "learning_rate": 3.516019417475728e-05, "loss": 0.1421, "step": 13357 }, { "epoch": 129.5289156626506, "grad_norm": 31.157609939575195, "learning_rate": 3.515533980582525e-05, "loss": 0.2493, "step": 13358 }, { "epoch": 129.53855421686748, "grad_norm": 2.0764811038970947, "learning_rate": 3.5150485436893205e-05, "loss": 0.1208, "step": 13359 }, { "epoch": 129.54819277108433, "grad_norm": 16.211505889892578, "learning_rate": 3.514563106796116e-05, "loss": 0.2186, "step": 13360 }, { "epoch": 129.5578313253012, "grad_norm": 1.19833505153656, "learning_rate": 3.514077669902913e-05, "loss": 0.1941, "step": 13361 }, { "epoch": 129.5674698795181, "grad_norm": 7.623625755310059, "learning_rate": 3.5135922330097085e-05, "loss": 0.2087, "step": 13362 }, { "epoch": 129.57710843373494, "grad_norm": 61.63633728027344, "learning_rate": 3.513106796116505e-05, "loss": 0.4011, "step": 13363 }, { "epoch": 129.5867469879518, "grad_norm": 8.401375770568848, "learning_rate": 3.512621359223301e-05, "loss": 0.4541, "step": 13364 }, { "epoch": 129.59638554216866, "grad_norm": 32.109291076660156, "learning_rate": 3.512135922330097e-05, "loss": 0.1523, "step": 13365 }, { "epoch": 129.60602409638554, "grad_norm": 23.265785217285156, "learning_rate": 3.511650485436893e-05, "loss": 0.3212, "step": 13366 }, { "epoch": 129.61566265060242, "grad_norm": 27.324234008789062, "learning_rate": 3.5111650485436894e-05, "loss": 0.1334, "step": 13367 }, { "epoch": 129.62530120481927, "grad_norm": 14.472853660583496, "learning_rate": 3.510679611650486e-05, "loss": 0.4342, "step": 13368 }, { "epoch": 129.63493975903614, "grad_norm": 25.54888343811035, "learning_rate": 3.5101941747572816e-05, "loss": 0.2804, "step": 13369 }, { "epoch": 129.64457831325302, "grad_norm": 37.60503005981445, "learning_rate": 3.509708737864078e-05, "loss": 0.3124, "step": 13370 }, { "epoch": 129.65421686746987, "grad_norm": 50.549339294433594, "learning_rate": 3.509223300970874e-05, "loss": 0.4997, "step": 13371 }, { "epoch": 129.66385542168675, "grad_norm": 40.42546463012695, "learning_rate": 3.50873786407767e-05, "loss": 0.3204, "step": 13372 }, { "epoch": 129.67349397590363, "grad_norm": 2.734468460083008, "learning_rate": 3.508252427184466e-05, "loss": 0.1545, "step": 13373 }, { "epoch": 129.68313253012047, "grad_norm": 30.653564453125, "learning_rate": 3.5077669902912625e-05, "loss": 0.3548, "step": 13374 }, { "epoch": 129.69277108433735, "grad_norm": 9.778117179870605, "learning_rate": 3.507281553398058e-05, "loss": 0.4173, "step": 13375 }, { "epoch": 129.70240963855423, "grad_norm": 12.795156478881836, "learning_rate": 3.506796116504855e-05, "loss": 0.4244, "step": 13376 }, { "epoch": 129.71204819277108, "grad_norm": 57.06296157836914, "learning_rate": 3.5063106796116505e-05, "loss": 0.4718, "step": 13377 }, { "epoch": 129.72168674698796, "grad_norm": 3.301607370376587, "learning_rate": 3.505825242718447e-05, "loss": 0.2992, "step": 13378 }, { "epoch": 129.73132530120483, "grad_norm": 79.76304626464844, "learning_rate": 3.505339805825243e-05, "loss": 0.405, "step": 13379 }, { "epoch": 129.74096385542168, "grad_norm": 18.346023559570312, "learning_rate": 3.504854368932039e-05, "loss": 0.1964, "step": 13380 }, { "epoch": 129.75060240963856, "grad_norm": 9.156845092773438, "learning_rate": 3.504368932038835e-05, "loss": 0.1206, "step": 13381 }, { "epoch": 129.7602409638554, "grad_norm": 4.344099044799805, "learning_rate": 3.5038834951456315e-05, "loss": 0.0855, "step": 13382 }, { "epoch": 129.7698795180723, "grad_norm": 4.223067283630371, "learning_rate": 3.503398058252427e-05, "loss": 0.3586, "step": 13383 }, { "epoch": 129.77951807228916, "grad_norm": 21.748720169067383, "learning_rate": 3.502912621359224e-05, "loss": 0.2198, "step": 13384 }, { "epoch": 129.789156626506, "grad_norm": 29.84191131591797, "learning_rate": 3.5024271844660195e-05, "loss": 0.2417, "step": 13385 }, { "epoch": 129.7987951807229, "grad_norm": 121.43553924560547, "learning_rate": 3.501941747572815e-05, "loss": 0.3069, "step": 13386 }, { "epoch": 129.80843373493977, "grad_norm": 9.956214904785156, "learning_rate": 3.501456310679612e-05, "loss": 0.4176, "step": 13387 }, { "epoch": 129.81807228915662, "grad_norm": 35.91322708129883, "learning_rate": 3.5009708737864075e-05, "loss": 0.2399, "step": 13388 }, { "epoch": 129.8277108433735, "grad_norm": 2.3225221633911133, "learning_rate": 3.500485436893204e-05, "loss": 0.0838, "step": 13389 }, { "epoch": 129.83734939759037, "grad_norm": 30.13508415222168, "learning_rate": 3.5e-05, "loss": 0.2189, "step": 13390 }, { "epoch": 129.84698795180722, "grad_norm": 20.69898223876953, "learning_rate": 3.499514563106796e-05, "loss": 0.2964, "step": 13391 }, { "epoch": 129.8566265060241, "grad_norm": 39.03093719482422, "learning_rate": 3.499029126213592e-05, "loss": 0.115, "step": 13392 }, { "epoch": 129.86626506024098, "grad_norm": 31.00200843811035, "learning_rate": 3.498543689320389e-05, "loss": 0.2317, "step": 13393 }, { "epoch": 129.87590361445783, "grad_norm": 3.382384777069092, "learning_rate": 3.498058252427185e-05, "loss": 0.2076, "step": 13394 }, { "epoch": 129.8855421686747, "grad_norm": 8.35518741607666, "learning_rate": 3.497572815533981e-05, "loss": 0.2784, "step": 13395 }, { "epoch": 129.89518072289155, "grad_norm": 10.0473051071167, "learning_rate": 3.497087378640777e-05, "loss": 0.3421, "step": 13396 }, { "epoch": 129.90481927710843, "grad_norm": 7.348169803619385, "learning_rate": 3.496601941747573e-05, "loss": 0.2286, "step": 13397 }, { "epoch": 129.9144578313253, "grad_norm": 36.48930740356445, "learning_rate": 3.496116504854369e-05, "loss": 0.1549, "step": 13398 }, { "epoch": 129.92409638554216, "grad_norm": 18.524431228637695, "learning_rate": 3.495631067961165e-05, "loss": 0.2867, "step": 13399 }, { "epoch": 129.93373493975903, "grad_norm": 4.575492858886719, "learning_rate": 3.4951456310679615e-05, "loss": 0.1981, "step": 13400 }, { "epoch": 129.9433734939759, "grad_norm": 10.202549934387207, "learning_rate": 3.494660194174757e-05, "loss": 0.4535, "step": 13401 }, { "epoch": 129.95301204819276, "grad_norm": 21.72041893005371, "learning_rate": 3.494174757281554e-05, "loss": 0.1763, "step": 13402 }, { "epoch": 129.96265060240964, "grad_norm": 12.1241455078125, "learning_rate": 3.4936893203883495e-05, "loss": 0.4911, "step": 13403 }, { "epoch": 129.97228915662652, "grad_norm": 7.237679958343506, "learning_rate": 3.493203883495146e-05, "loss": 0.1998, "step": 13404 }, { "epoch": 129.98192771084337, "grad_norm": 25.077287673950195, "learning_rate": 3.492718446601942e-05, "loss": 0.2093, "step": 13405 }, { "epoch": 129.99156626506024, "grad_norm": 15.998077392578125, "learning_rate": 3.492233009708738e-05, "loss": 0.2097, "step": 13406 }, { "epoch": 130.00722891566264, "grad_norm": 30.42738151550293, "learning_rate": 3.491747572815534e-05, "loss": 0.2288, "step": 13407 }, { "epoch": 130.01686746987951, "grad_norm": 2.0072665214538574, "learning_rate": 3.4912621359223304e-05, "loss": 0.154, "step": 13408 }, { "epoch": 130.0265060240964, "grad_norm": 13.236557960510254, "learning_rate": 3.490776699029126e-05, "loss": 0.1821, "step": 13409 }, { "epoch": 130.03614457831324, "grad_norm": 12.82813549041748, "learning_rate": 3.490291262135923e-05, "loss": 0.2768, "step": 13410 }, { "epoch": 130.04578313253012, "grad_norm": 23.442440032958984, "learning_rate": 3.4898058252427184e-05, "loss": 0.3035, "step": 13411 }, { "epoch": 130.055421686747, "grad_norm": 19.783754348754883, "learning_rate": 3.489320388349515e-05, "loss": 0.1716, "step": 13412 }, { "epoch": 130.06506024096385, "grad_norm": 18.92491340637207, "learning_rate": 3.488834951456311e-05, "loss": 0.4039, "step": 13413 }, { "epoch": 130.07469879518072, "grad_norm": 15.465554237365723, "learning_rate": 3.4883495145631064e-05, "loss": 0.2756, "step": 13414 }, { "epoch": 130.0843373493976, "grad_norm": 36.82255935668945, "learning_rate": 3.487864077669903e-05, "loss": 0.2955, "step": 13415 }, { "epoch": 130.09397590361445, "grad_norm": 23.905603408813477, "learning_rate": 3.487378640776699e-05, "loss": 0.4487, "step": 13416 }, { "epoch": 130.10361445783133, "grad_norm": 13.268571853637695, "learning_rate": 3.486893203883495e-05, "loss": 0.2266, "step": 13417 }, { "epoch": 130.1132530120482, "grad_norm": 7.879053592681885, "learning_rate": 3.4864077669902916e-05, "loss": 0.1762, "step": 13418 }, { "epoch": 130.12289156626505, "grad_norm": 7.166172981262207, "learning_rate": 3.485922330097088e-05, "loss": 0.3708, "step": 13419 }, { "epoch": 130.13253012048193, "grad_norm": 74.74769592285156, "learning_rate": 3.485436893203884e-05, "loss": 0.2355, "step": 13420 }, { "epoch": 130.1421686746988, "grad_norm": 9.723954200744629, "learning_rate": 3.48495145631068e-05, "loss": 0.354, "step": 13421 }, { "epoch": 130.15180722891566, "grad_norm": 5.077507495880127, "learning_rate": 3.484466019417476e-05, "loss": 0.2507, "step": 13422 }, { "epoch": 130.16144578313254, "grad_norm": 8.449323654174805, "learning_rate": 3.4839805825242725e-05, "loss": 0.2784, "step": 13423 }, { "epoch": 130.17108433734938, "grad_norm": 5.532892227172852, "learning_rate": 3.483495145631068e-05, "loss": 0.1888, "step": 13424 }, { "epoch": 130.18072289156626, "grad_norm": 4.035168647766113, "learning_rate": 3.483009708737864e-05, "loss": 0.0635, "step": 13425 }, { "epoch": 130.19036144578314, "grad_norm": 31.715810775756836, "learning_rate": 3.4825242718446605e-05, "loss": 0.1654, "step": 13426 }, { "epoch": 130.2, "grad_norm": 35.249412536621094, "learning_rate": 3.482038834951456e-05, "loss": 0.2975, "step": 13427 }, { "epoch": 130.20963855421687, "grad_norm": 17.527000427246094, "learning_rate": 3.481553398058253e-05, "loss": 0.3038, "step": 13428 }, { "epoch": 130.21927710843374, "grad_norm": 36.52092361450195, "learning_rate": 3.4810679611650485e-05, "loss": 0.0766, "step": 13429 }, { "epoch": 130.2289156626506, "grad_norm": 12.28598403930664, "learning_rate": 3.480582524271845e-05, "loss": 0.3222, "step": 13430 }, { "epoch": 130.23855421686747, "grad_norm": 12.047454833984375, "learning_rate": 3.480097087378641e-05, "loss": 0.1892, "step": 13431 }, { "epoch": 130.24819277108435, "grad_norm": 5.160847187042236, "learning_rate": 3.479611650485437e-05, "loss": 0.198, "step": 13432 }, { "epoch": 130.2578313253012, "grad_norm": 65.10391235351562, "learning_rate": 3.479126213592233e-05, "loss": 0.4313, "step": 13433 }, { "epoch": 130.26746987951807, "grad_norm": 12.186482429504395, "learning_rate": 3.4786407766990294e-05, "loss": 0.1856, "step": 13434 }, { "epoch": 130.27710843373495, "grad_norm": 4.850852966308594, "learning_rate": 3.478155339805825e-05, "loss": 0.4041, "step": 13435 }, { "epoch": 130.2867469879518, "grad_norm": 7.483400344848633, "learning_rate": 3.4776699029126216e-05, "loss": 0.1519, "step": 13436 }, { "epoch": 130.29638554216868, "grad_norm": 12.153766632080078, "learning_rate": 3.4771844660194174e-05, "loss": 0.4377, "step": 13437 }, { "epoch": 130.30602409638556, "grad_norm": 19.572725296020508, "learning_rate": 3.476699029126214e-05, "loss": 0.2704, "step": 13438 }, { "epoch": 130.3156626506024, "grad_norm": 36.18860626220703, "learning_rate": 3.4762135922330096e-05, "loss": 0.2881, "step": 13439 }, { "epoch": 130.32530120481928, "grad_norm": 9.119266510009766, "learning_rate": 3.475728155339806e-05, "loss": 0.3643, "step": 13440 }, { "epoch": 130.33493975903613, "grad_norm": 91.88855743408203, "learning_rate": 3.475242718446602e-05, "loss": 0.1773, "step": 13441 }, { "epoch": 130.344578313253, "grad_norm": 22.106351852416992, "learning_rate": 3.4747572815533977e-05, "loss": 0.2278, "step": 13442 }, { "epoch": 130.3542168674699, "grad_norm": 20.78053092956543, "learning_rate": 3.474271844660194e-05, "loss": 0.2918, "step": 13443 }, { "epoch": 130.36385542168674, "grad_norm": 69.97431945800781, "learning_rate": 3.4737864077669906e-05, "loss": 0.2446, "step": 13444 }, { "epoch": 130.3734939759036, "grad_norm": 3.713698148727417, "learning_rate": 3.473300970873787e-05, "loss": 0.2198, "step": 13445 }, { "epoch": 130.3831325301205, "grad_norm": 18.43821144104004, "learning_rate": 3.472815533980583e-05, "loss": 0.2095, "step": 13446 }, { "epoch": 130.39277108433734, "grad_norm": 46.401126861572266, "learning_rate": 3.472330097087379e-05, "loss": 0.2786, "step": 13447 }, { "epoch": 130.40240963855422, "grad_norm": 19.215784072875977, "learning_rate": 3.471844660194175e-05, "loss": 0.2951, "step": 13448 }, { "epoch": 130.4120481927711, "grad_norm": 11.43815803527832, "learning_rate": 3.4713592233009715e-05, "loss": 0.2317, "step": 13449 }, { "epoch": 130.42168674698794, "grad_norm": 6.216983318328857, "learning_rate": 3.470873786407767e-05, "loss": 0.2892, "step": 13450 }, { "epoch": 130.43132530120482, "grad_norm": 42.27164840698242, "learning_rate": 3.470388349514564e-05, "loss": 0.1953, "step": 13451 }, { "epoch": 130.4409638554217, "grad_norm": 38.626869201660156, "learning_rate": 3.4699029126213595e-05, "loss": 0.2513, "step": 13452 }, { "epoch": 130.45060240963855, "grad_norm": 28.807144165039062, "learning_rate": 3.469417475728155e-05, "loss": 0.2073, "step": 13453 }, { "epoch": 130.46024096385543, "grad_norm": 2.4145658016204834, "learning_rate": 3.468932038834952e-05, "loss": 0.2763, "step": 13454 }, { "epoch": 130.46987951807228, "grad_norm": 30.210744857788086, "learning_rate": 3.4684466019417475e-05, "loss": 0.2128, "step": 13455 }, { "epoch": 130.47951807228915, "grad_norm": 9.417274475097656, "learning_rate": 3.467961165048544e-05, "loss": 0.3384, "step": 13456 }, { "epoch": 130.48915662650603, "grad_norm": 5.403164386749268, "learning_rate": 3.46747572815534e-05, "loss": 0.2331, "step": 13457 }, { "epoch": 130.49879518072288, "grad_norm": 7.989384174346924, "learning_rate": 3.466990291262136e-05, "loss": 0.2247, "step": 13458 }, { "epoch": 130.50843373493976, "grad_norm": 135.6951904296875, "learning_rate": 3.466504854368932e-05, "loss": 0.2559, "step": 13459 }, { "epoch": 130.51807228915663, "grad_norm": 15.839335441589355, "learning_rate": 3.4660194174757284e-05, "loss": 0.176, "step": 13460 }, { "epoch": 130.52771084337348, "grad_norm": 5.701145648956299, "learning_rate": 3.465533980582524e-05, "loss": 0.2155, "step": 13461 }, { "epoch": 130.53734939759036, "grad_norm": 55.04115295410156, "learning_rate": 3.4650485436893206e-05, "loss": 0.2107, "step": 13462 }, { "epoch": 130.54698795180724, "grad_norm": 73.1676025390625, "learning_rate": 3.4645631067961164e-05, "loss": 0.2745, "step": 13463 }, { "epoch": 130.5566265060241, "grad_norm": 27.29429817199707, "learning_rate": 3.464077669902913e-05, "loss": 0.2158, "step": 13464 }, { "epoch": 130.56626506024097, "grad_norm": 20.13620376586914, "learning_rate": 3.4635922330097086e-05, "loss": 0.2101, "step": 13465 }, { "epoch": 130.57590361445784, "grad_norm": 10.143754005432129, "learning_rate": 3.463106796116505e-05, "loss": 0.2144, "step": 13466 }, { "epoch": 130.5855421686747, "grad_norm": 22.371440887451172, "learning_rate": 3.462621359223301e-05, "loss": 0.1912, "step": 13467 }, { "epoch": 130.59518072289157, "grad_norm": 33.65977478027344, "learning_rate": 3.462135922330097e-05, "loss": 0.3249, "step": 13468 }, { "epoch": 130.60481927710845, "grad_norm": 23.813987731933594, "learning_rate": 3.461650485436894e-05, "loss": 0.3663, "step": 13469 }, { "epoch": 130.6144578313253, "grad_norm": 108.91932678222656, "learning_rate": 3.4611650485436895e-05, "loss": 0.2413, "step": 13470 }, { "epoch": 130.62409638554217, "grad_norm": 7.875975608825684, "learning_rate": 3.460679611650486e-05, "loss": 0.1562, "step": 13471 }, { "epoch": 130.63373493975902, "grad_norm": 17.120338439941406, "learning_rate": 3.460194174757282e-05, "loss": 0.1563, "step": 13472 }, { "epoch": 130.6433734939759, "grad_norm": 8.635640144348145, "learning_rate": 3.459708737864078e-05, "loss": 0.2832, "step": 13473 }, { "epoch": 130.65301204819278, "grad_norm": 23.931604385375977, "learning_rate": 3.459223300970874e-05, "loss": 0.209, "step": 13474 }, { "epoch": 130.66265060240963, "grad_norm": 2.7839744091033936, "learning_rate": 3.4587378640776704e-05, "loss": 0.1767, "step": 13475 }, { "epoch": 130.6722891566265, "grad_norm": 16.203495025634766, "learning_rate": 3.458252427184466e-05, "loss": 0.4259, "step": 13476 }, { "epoch": 130.68192771084338, "grad_norm": 10.845311164855957, "learning_rate": 3.457766990291263e-05, "loss": 0.152, "step": 13477 }, { "epoch": 130.69156626506023, "grad_norm": 4.482211589813232, "learning_rate": 3.4572815533980585e-05, "loss": 0.1574, "step": 13478 }, { "epoch": 130.7012048192771, "grad_norm": 5.943918228149414, "learning_rate": 3.456796116504855e-05, "loss": 0.4474, "step": 13479 }, { "epoch": 130.710843373494, "grad_norm": 49.19792556762695, "learning_rate": 3.456310679611651e-05, "loss": 0.5237, "step": 13480 }, { "epoch": 130.72048192771084, "grad_norm": 8.180340766906738, "learning_rate": 3.4558252427184465e-05, "loss": 0.2198, "step": 13481 }, { "epoch": 130.7301204819277, "grad_norm": 149.86807250976562, "learning_rate": 3.455339805825243e-05, "loss": 0.3975, "step": 13482 }, { "epoch": 130.7397590361446, "grad_norm": 18.406883239746094, "learning_rate": 3.454854368932039e-05, "loss": 0.3475, "step": 13483 }, { "epoch": 130.74939759036144, "grad_norm": 5.746841907501221, "learning_rate": 3.454368932038835e-05, "loss": 0.214, "step": 13484 }, { "epoch": 130.75903614457832, "grad_norm": 7.9307098388671875, "learning_rate": 3.453883495145631e-05, "loss": 0.1778, "step": 13485 }, { "epoch": 130.76867469879517, "grad_norm": 16.277711868286133, "learning_rate": 3.4533980582524274e-05, "loss": 0.2148, "step": 13486 }, { "epoch": 130.77831325301204, "grad_norm": 23.827518463134766, "learning_rate": 3.452912621359223e-05, "loss": 0.517, "step": 13487 }, { "epoch": 130.78795180722892, "grad_norm": 3.0242555141448975, "learning_rate": 3.4524271844660196e-05, "loss": 0.0746, "step": 13488 }, { "epoch": 130.79759036144577, "grad_norm": 24.27935028076172, "learning_rate": 3.4519417475728154e-05, "loss": 0.2139, "step": 13489 }, { "epoch": 130.80722891566265, "grad_norm": 32.21763610839844, "learning_rate": 3.451456310679612e-05, "loss": 0.6133, "step": 13490 }, { "epoch": 130.81686746987953, "grad_norm": 26.05832862854004, "learning_rate": 3.4509708737864076e-05, "loss": 0.2703, "step": 13491 }, { "epoch": 130.82650602409637, "grad_norm": 15.375628471374512, "learning_rate": 3.450485436893204e-05, "loss": 0.1933, "step": 13492 }, { "epoch": 130.83614457831325, "grad_norm": 16.393768310546875, "learning_rate": 3.45e-05, "loss": 0.2985, "step": 13493 }, { "epoch": 130.84578313253013, "grad_norm": 8.761487007141113, "learning_rate": 3.449514563106796e-05, "loss": 0.3114, "step": 13494 }, { "epoch": 130.85542168674698, "grad_norm": 14.245564460754395, "learning_rate": 3.449029126213593e-05, "loss": 0.3094, "step": 13495 }, { "epoch": 130.86506024096386, "grad_norm": 24.539791107177734, "learning_rate": 3.4485436893203885e-05, "loss": 0.2881, "step": 13496 }, { "epoch": 130.87469879518073, "grad_norm": 9.092790603637695, "learning_rate": 3.448058252427185e-05, "loss": 0.1616, "step": 13497 }, { "epoch": 130.88433734939758, "grad_norm": 15.00432300567627, "learning_rate": 3.447572815533981e-05, "loss": 0.2681, "step": 13498 }, { "epoch": 130.89397590361446, "grad_norm": 25.62540054321289, "learning_rate": 3.447087378640777e-05, "loss": 0.4103, "step": 13499 }, { "epoch": 130.90361445783134, "grad_norm": 61.33705520629883, "learning_rate": 3.446601941747573e-05, "loss": 0.2765, "step": 13500 }, { "epoch": 130.9132530120482, "grad_norm": 6.346161365509033, "learning_rate": 3.4461165048543694e-05, "loss": 0.1972, "step": 13501 }, { "epoch": 130.92289156626506, "grad_norm": 16.59469223022461, "learning_rate": 3.445631067961165e-05, "loss": 0.4476, "step": 13502 }, { "epoch": 130.9325301204819, "grad_norm": 23.433183670043945, "learning_rate": 3.4451456310679617e-05, "loss": 0.2499, "step": 13503 }, { "epoch": 130.9421686746988, "grad_norm": 3.714153289794922, "learning_rate": 3.4446601941747574e-05, "loss": 0.1189, "step": 13504 }, { "epoch": 130.95180722891567, "grad_norm": 6.241298675537109, "learning_rate": 3.444174757281554e-05, "loss": 0.2579, "step": 13505 }, { "epoch": 130.96144578313252, "grad_norm": 34.60914611816406, "learning_rate": 3.44368932038835e-05, "loss": 0.183, "step": 13506 }, { "epoch": 130.9710843373494, "grad_norm": 19.98959732055664, "learning_rate": 3.443203883495146e-05, "loss": 0.2523, "step": 13507 }, { "epoch": 130.98072289156627, "grad_norm": 32.37008285522461, "learning_rate": 3.442718446601942e-05, "loss": 0.2634, "step": 13508 }, { "epoch": 130.99036144578312, "grad_norm": 7.28651762008667, "learning_rate": 3.442233009708738e-05, "loss": 0.3581, "step": 13509 }, { "epoch": 131.00602409638554, "grad_norm": 2.140589952468872, "learning_rate": 3.441747572815534e-05, "loss": 0.0653, "step": 13510 }, { "epoch": 131.01566265060242, "grad_norm": 34.076114654541016, "learning_rate": 3.44126213592233e-05, "loss": 0.3336, "step": 13511 }, { "epoch": 131.02530120481927, "grad_norm": 3.3207342624664307, "learning_rate": 3.4407766990291263e-05, "loss": 0.1566, "step": 13512 }, { "epoch": 131.03493975903615, "grad_norm": 23.56751823425293, "learning_rate": 3.440291262135922e-05, "loss": 0.3645, "step": 13513 }, { "epoch": 131.044578313253, "grad_norm": 22.043609619140625, "learning_rate": 3.4398058252427186e-05, "loss": 0.2684, "step": 13514 }, { "epoch": 131.05421686746988, "grad_norm": 9.776117324829102, "learning_rate": 3.4393203883495144e-05, "loss": 0.1283, "step": 13515 }, { "epoch": 131.06385542168675, "grad_norm": 37.00178146362305, "learning_rate": 3.438834951456311e-05, "loss": 0.2241, "step": 13516 }, { "epoch": 131.0734939759036, "grad_norm": 5.45142126083374, "learning_rate": 3.4383495145631066e-05, "loss": 0.2572, "step": 13517 }, { "epoch": 131.08313253012048, "grad_norm": 7.008188247680664, "learning_rate": 3.437864077669903e-05, "loss": 0.247, "step": 13518 }, { "epoch": 131.09277108433736, "grad_norm": 6.273308277130127, "learning_rate": 3.4373786407766995e-05, "loss": 0.1844, "step": 13519 }, { "epoch": 131.1024096385542, "grad_norm": 14.064638137817383, "learning_rate": 3.436893203883495e-05, "loss": 0.2463, "step": 13520 }, { "epoch": 131.11204819277108, "grad_norm": 19.956375122070312, "learning_rate": 3.436407766990292e-05, "loss": 0.252, "step": 13521 }, { "epoch": 131.12168674698796, "grad_norm": 17.330520629882812, "learning_rate": 3.4359223300970875e-05, "loss": 0.1321, "step": 13522 }, { "epoch": 131.1313253012048, "grad_norm": 16.142269134521484, "learning_rate": 3.435436893203884e-05, "loss": 0.4375, "step": 13523 }, { "epoch": 131.1409638554217, "grad_norm": 29.24187660217285, "learning_rate": 3.43495145631068e-05, "loss": 0.3357, "step": 13524 }, { "epoch": 131.15060240963857, "grad_norm": 20.942052841186523, "learning_rate": 3.434466019417476e-05, "loss": 0.1927, "step": 13525 }, { "epoch": 131.16024096385541, "grad_norm": 44.088253021240234, "learning_rate": 3.433980582524272e-05, "loss": 0.4445, "step": 13526 }, { "epoch": 131.1698795180723, "grad_norm": 7.501912593841553, "learning_rate": 3.4334951456310684e-05, "loss": 0.1634, "step": 13527 }, { "epoch": 131.17951807228917, "grad_norm": 21.099693298339844, "learning_rate": 3.433009708737864e-05, "loss": 0.3627, "step": 13528 }, { "epoch": 131.18915662650602, "grad_norm": 7.322117328643799, "learning_rate": 3.4325242718446606e-05, "loss": 0.238, "step": 13529 }, { "epoch": 131.1987951807229, "grad_norm": 17.823589324951172, "learning_rate": 3.4320388349514564e-05, "loss": 0.2357, "step": 13530 }, { "epoch": 131.20843373493975, "grad_norm": 36.718101501464844, "learning_rate": 3.431553398058253e-05, "loss": 0.1753, "step": 13531 }, { "epoch": 131.21807228915662, "grad_norm": 9.441776275634766, "learning_rate": 3.4310679611650486e-05, "loss": 0.1799, "step": 13532 }, { "epoch": 131.2277108433735, "grad_norm": 22.90874671936035, "learning_rate": 3.430582524271845e-05, "loss": 0.3819, "step": 13533 }, { "epoch": 131.23734939759035, "grad_norm": 38.85888671875, "learning_rate": 3.430097087378641e-05, "loss": 0.34, "step": 13534 }, { "epoch": 131.24698795180723, "grad_norm": 197.1866912841797, "learning_rate": 3.4296116504854366e-05, "loss": 0.2146, "step": 13535 }, { "epoch": 131.2566265060241, "grad_norm": 59.796478271484375, "learning_rate": 3.429126213592233e-05, "loss": 0.4391, "step": 13536 }, { "epoch": 131.26626506024095, "grad_norm": 40.37529754638672, "learning_rate": 3.428640776699029e-05, "loss": 0.3265, "step": 13537 }, { "epoch": 131.27590361445783, "grad_norm": 35.75714874267578, "learning_rate": 3.428155339805825e-05, "loss": 0.2634, "step": 13538 }, { "epoch": 131.2855421686747, "grad_norm": 4.4577813148498535, "learning_rate": 3.427669902912621e-05, "loss": 0.2456, "step": 13539 }, { "epoch": 131.29518072289156, "grad_norm": 62.724853515625, "learning_rate": 3.4271844660194176e-05, "loss": 0.4876, "step": 13540 }, { "epoch": 131.30481927710844, "grad_norm": 6.316790580749512, "learning_rate": 3.426699029126213e-05, "loss": 0.1392, "step": 13541 }, { "epoch": 131.3144578313253, "grad_norm": 8.210838317871094, "learning_rate": 3.42621359223301e-05, "loss": 0.2851, "step": 13542 }, { "epoch": 131.32409638554216, "grad_norm": 4.407022476196289, "learning_rate": 3.4257281553398056e-05, "loss": 0.1879, "step": 13543 }, { "epoch": 131.33373493975904, "grad_norm": 8.872472763061523, "learning_rate": 3.425242718446603e-05, "loss": 0.1307, "step": 13544 }, { "epoch": 131.34337349397592, "grad_norm": 11.206933975219727, "learning_rate": 3.4247572815533985e-05, "loss": 0.3401, "step": 13545 }, { "epoch": 131.35301204819277, "grad_norm": 138.3861083984375, "learning_rate": 3.424271844660194e-05, "loss": 0.305, "step": 13546 }, { "epoch": 131.36265060240964, "grad_norm": 5.040293216705322, "learning_rate": 3.423786407766991e-05, "loss": 0.1048, "step": 13547 }, { "epoch": 131.3722891566265, "grad_norm": 27.677114486694336, "learning_rate": 3.4233009708737865e-05, "loss": 0.3674, "step": 13548 }, { "epoch": 131.38192771084337, "grad_norm": 16.589092254638672, "learning_rate": 3.422815533980583e-05, "loss": 0.2606, "step": 13549 }, { "epoch": 131.39156626506025, "grad_norm": 3.237571954727173, "learning_rate": 3.422330097087379e-05, "loss": 0.1952, "step": 13550 }, { "epoch": 131.4012048192771, "grad_norm": 30.5733699798584, "learning_rate": 3.421844660194175e-05, "loss": 0.3552, "step": 13551 }, { "epoch": 131.41084337349398, "grad_norm": 35.77387237548828, "learning_rate": 3.421359223300971e-05, "loss": 0.3777, "step": 13552 }, { "epoch": 131.42048192771085, "grad_norm": 16.18299674987793, "learning_rate": 3.4208737864077674e-05, "loss": 0.3668, "step": 13553 }, { "epoch": 131.4301204819277, "grad_norm": 18.588428497314453, "learning_rate": 3.420388349514563e-05, "loss": 0.4042, "step": 13554 }, { "epoch": 131.43975903614458, "grad_norm": 10.764716148376465, "learning_rate": 3.4199029126213596e-05, "loss": 0.4639, "step": 13555 }, { "epoch": 131.44939759036146, "grad_norm": 10.772717475891113, "learning_rate": 3.4194174757281554e-05, "loss": 0.3273, "step": 13556 }, { "epoch": 131.4590361445783, "grad_norm": 20.410457611083984, "learning_rate": 3.418932038834952e-05, "loss": 0.413, "step": 13557 }, { "epoch": 131.46867469879518, "grad_norm": 8.455412864685059, "learning_rate": 3.4184466019417476e-05, "loss": 0.2652, "step": 13558 }, { "epoch": 131.47831325301206, "grad_norm": 16.855995178222656, "learning_rate": 3.417961165048544e-05, "loss": 0.2092, "step": 13559 }, { "epoch": 131.4879518072289, "grad_norm": 8.205124855041504, "learning_rate": 3.41747572815534e-05, "loss": 0.5084, "step": 13560 }, { "epoch": 131.4975903614458, "grad_norm": 37.673667907714844, "learning_rate": 3.416990291262136e-05, "loss": 0.3459, "step": 13561 }, { "epoch": 131.50722891566264, "grad_norm": 13.346020698547363, "learning_rate": 3.416504854368932e-05, "loss": 0.2415, "step": 13562 }, { "epoch": 131.51686746987951, "grad_norm": 19.017301559448242, "learning_rate": 3.416019417475728e-05, "loss": 0.1562, "step": 13563 }, { "epoch": 131.5265060240964, "grad_norm": 14.82485294342041, "learning_rate": 3.415533980582524e-05, "loss": 0.2734, "step": 13564 }, { "epoch": 131.53614457831324, "grad_norm": 80.09866333007812, "learning_rate": 3.41504854368932e-05, "loss": 0.3428, "step": 13565 }, { "epoch": 131.54578313253012, "grad_norm": 8.941435813903809, "learning_rate": 3.4145631067961165e-05, "loss": 0.2708, "step": 13566 }, { "epoch": 131.555421686747, "grad_norm": 18.775527954101562, "learning_rate": 3.414077669902912e-05, "loss": 0.2278, "step": 13567 }, { "epoch": 131.56506024096385, "grad_norm": 15.414318084716797, "learning_rate": 3.413592233009709e-05, "loss": 0.1973, "step": 13568 }, { "epoch": 131.57469879518072, "grad_norm": 32.21857833862305, "learning_rate": 3.4131067961165045e-05, "loss": 0.2762, "step": 13569 }, { "epoch": 131.5843373493976, "grad_norm": 4.376458644866943, "learning_rate": 3.412621359223302e-05, "loss": 0.2102, "step": 13570 }, { "epoch": 131.59397590361445, "grad_norm": 9.87242317199707, "learning_rate": 3.4121359223300974e-05, "loss": 0.21, "step": 13571 }, { "epoch": 131.60361445783133, "grad_norm": 32.738685607910156, "learning_rate": 3.411650485436894e-05, "loss": 0.3187, "step": 13572 }, { "epoch": 131.6132530120482, "grad_norm": 7.961114406585693, "learning_rate": 3.41116504854369e-05, "loss": 0.1368, "step": 13573 }, { "epoch": 131.62289156626505, "grad_norm": 63.707115173339844, "learning_rate": 3.4106796116504855e-05, "loss": 0.3982, "step": 13574 }, { "epoch": 131.63253012048193, "grad_norm": 24.103172302246094, "learning_rate": 3.410194174757282e-05, "loss": 0.126, "step": 13575 }, { "epoch": 131.6421686746988, "grad_norm": 4.351383686065674, "learning_rate": 3.409708737864078e-05, "loss": 0.2046, "step": 13576 }, { "epoch": 131.65180722891566, "grad_norm": 4.200872898101807, "learning_rate": 3.409223300970874e-05, "loss": 0.281, "step": 13577 }, { "epoch": 131.66144578313254, "grad_norm": 36.3872184753418, "learning_rate": 3.40873786407767e-05, "loss": 0.3756, "step": 13578 }, { "epoch": 131.67108433734938, "grad_norm": 8.559136390686035, "learning_rate": 3.4082524271844664e-05, "loss": 0.2607, "step": 13579 }, { "epoch": 131.68072289156626, "grad_norm": 9.253739356994629, "learning_rate": 3.407766990291262e-05, "loss": 0.3499, "step": 13580 }, { "epoch": 131.69036144578314, "grad_norm": 13.76729965209961, "learning_rate": 3.4072815533980586e-05, "loss": 0.2717, "step": 13581 }, { "epoch": 131.7, "grad_norm": 3.3382821083068848, "learning_rate": 3.4067961165048544e-05, "loss": 0.2173, "step": 13582 }, { "epoch": 131.70963855421687, "grad_norm": 25.57236671447754, "learning_rate": 3.406310679611651e-05, "loss": 0.4233, "step": 13583 }, { "epoch": 131.71927710843374, "grad_norm": 20.74578094482422, "learning_rate": 3.4058252427184466e-05, "loss": 0.131, "step": 13584 }, { "epoch": 131.7289156626506, "grad_norm": 12.131851196289062, "learning_rate": 3.405339805825243e-05, "loss": 0.4845, "step": 13585 }, { "epoch": 131.73855421686747, "grad_norm": 29.587661743164062, "learning_rate": 3.404854368932039e-05, "loss": 0.2524, "step": 13586 }, { "epoch": 131.74819277108435, "grad_norm": 14.307578086853027, "learning_rate": 3.404368932038835e-05, "loss": 0.3706, "step": 13587 }, { "epoch": 131.7578313253012, "grad_norm": 14.275187492370605, "learning_rate": 3.403883495145631e-05, "loss": 0.3709, "step": 13588 }, { "epoch": 131.76746987951807, "grad_norm": 35.02558517456055, "learning_rate": 3.4033980582524275e-05, "loss": 0.3249, "step": 13589 }, { "epoch": 131.77710843373495, "grad_norm": 7.386441707611084, "learning_rate": 3.402912621359223e-05, "loss": 0.2165, "step": 13590 }, { "epoch": 131.7867469879518, "grad_norm": 43.36249542236328, "learning_rate": 3.402427184466019e-05, "loss": 0.2283, "step": 13591 }, { "epoch": 131.79638554216868, "grad_norm": 24.133867263793945, "learning_rate": 3.4019417475728155e-05, "loss": 0.3838, "step": 13592 }, { "epoch": 131.80602409638556, "grad_norm": 43.090877532958984, "learning_rate": 3.401456310679611e-05, "loss": 0.158, "step": 13593 }, { "epoch": 131.8156626506024, "grad_norm": 97.85327911376953, "learning_rate": 3.400970873786408e-05, "loss": 0.2192, "step": 13594 }, { "epoch": 131.82530120481928, "grad_norm": 7.4276933670043945, "learning_rate": 3.400485436893204e-05, "loss": 0.2652, "step": 13595 }, { "epoch": 131.83493975903613, "grad_norm": 54.65967559814453, "learning_rate": 3.4000000000000007e-05, "loss": 0.2017, "step": 13596 }, { "epoch": 131.844578313253, "grad_norm": 5.1592326164245605, "learning_rate": 3.3995145631067964e-05, "loss": 0.3225, "step": 13597 }, { "epoch": 131.8542168674699, "grad_norm": 17.369895935058594, "learning_rate": 3.399029126213593e-05, "loss": 0.2981, "step": 13598 }, { "epoch": 131.86385542168674, "grad_norm": 61.73101806640625, "learning_rate": 3.3985436893203887e-05, "loss": 0.1301, "step": 13599 }, { "epoch": 131.8734939759036, "grad_norm": 15.491436004638672, "learning_rate": 3.398058252427185e-05, "loss": 0.1945, "step": 13600 }, { "epoch": 131.8831325301205, "grad_norm": 20.60275650024414, "learning_rate": 3.397572815533981e-05, "loss": 0.2373, "step": 13601 }, { "epoch": 131.89277108433734, "grad_norm": 5.198672294616699, "learning_rate": 3.3970873786407767e-05, "loss": 0.2323, "step": 13602 }, { "epoch": 131.90240963855422, "grad_norm": 6.055113792419434, "learning_rate": 3.396601941747573e-05, "loss": 0.104, "step": 13603 }, { "epoch": 131.9120481927711, "grad_norm": 16.95500373840332, "learning_rate": 3.396116504854369e-05, "loss": 0.2201, "step": 13604 }, { "epoch": 131.92168674698794, "grad_norm": 27.687644958496094, "learning_rate": 3.3956310679611653e-05, "loss": 0.2517, "step": 13605 }, { "epoch": 131.93132530120482, "grad_norm": 9.707576751708984, "learning_rate": 3.395145631067961e-05, "loss": 0.2359, "step": 13606 }, { "epoch": 131.9409638554217, "grad_norm": 8.903449058532715, "learning_rate": 3.3946601941747576e-05, "loss": 0.1185, "step": 13607 }, { "epoch": 131.95060240963855, "grad_norm": 40.94776153564453, "learning_rate": 3.3941747572815533e-05, "loss": 0.2341, "step": 13608 }, { "epoch": 131.96024096385543, "grad_norm": 19.259010314941406, "learning_rate": 3.39368932038835e-05, "loss": 0.3256, "step": 13609 }, { "epoch": 131.96987951807228, "grad_norm": 16.61104393005371, "learning_rate": 3.3932038834951456e-05, "loss": 0.414, "step": 13610 }, { "epoch": 131.97951807228915, "grad_norm": 25.86032485961914, "learning_rate": 3.392718446601942e-05, "loss": 0.3192, "step": 13611 }, { "epoch": 131.98915662650603, "grad_norm": 16.678678512573242, "learning_rate": 3.392233009708738e-05, "loss": 0.1959, "step": 13612 }, { "epoch": 132.00481927710842, "grad_norm": 25.719341278076172, "learning_rate": 3.391747572815534e-05, "loss": 0.3171, "step": 13613 }, { "epoch": 132.0144578313253, "grad_norm": 19.69392204284668, "learning_rate": 3.39126213592233e-05, "loss": 0.4556, "step": 13614 }, { "epoch": 132.02409638554218, "grad_norm": 5.8239336013793945, "learning_rate": 3.3907766990291265e-05, "loss": 0.1969, "step": 13615 }, { "epoch": 132.03373493975903, "grad_norm": 9.206792831420898, "learning_rate": 3.390291262135922e-05, "loss": 0.3469, "step": 13616 }, { "epoch": 132.0433734939759, "grad_norm": 11.707589149475098, "learning_rate": 3.389805825242719e-05, "loss": 0.1488, "step": 13617 }, { "epoch": 132.05301204819278, "grad_norm": 55.48564529418945, "learning_rate": 3.3893203883495145e-05, "loss": 0.4838, "step": 13618 }, { "epoch": 132.06265060240963, "grad_norm": 13.588260650634766, "learning_rate": 3.38883495145631e-05, "loss": 0.289, "step": 13619 }, { "epoch": 132.0722891566265, "grad_norm": 11.73864459991455, "learning_rate": 3.3883495145631074e-05, "loss": 0.3728, "step": 13620 }, { "epoch": 132.08192771084336, "grad_norm": 5.3724141120910645, "learning_rate": 3.387864077669903e-05, "loss": 0.1066, "step": 13621 }, { "epoch": 132.09156626506024, "grad_norm": 8.942276000976562, "learning_rate": 3.3873786407766996e-05, "loss": 0.2953, "step": 13622 }, { "epoch": 132.10120481927711, "grad_norm": 20.784658432006836, "learning_rate": 3.3868932038834954e-05, "loss": 0.3236, "step": 13623 }, { "epoch": 132.11084337349396, "grad_norm": 30.086536407470703, "learning_rate": 3.386407766990292e-05, "loss": 0.2263, "step": 13624 }, { "epoch": 132.12048192771084, "grad_norm": 21.97323989868164, "learning_rate": 3.3859223300970876e-05, "loss": 0.2173, "step": 13625 }, { "epoch": 132.13012048192772, "grad_norm": 24.97233009338379, "learning_rate": 3.385436893203884e-05, "loss": 0.3966, "step": 13626 }, { "epoch": 132.13975903614457, "grad_norm": 8.044540405273438, "learning_rate": 3.38495145631068e-05, "loss": 0.1558, "step": 13627 }, { "epoch": 132.14939759036145, "grad_norm": 14.910054206848145, "learning_rate": 3.384466019417476e-05, "loss": 0.2006, "step": 13628 }, { "epoch": 132.15903614457832, "grad_norm": 35.61503601074219, "learning_rate": 3.383980582524272e-05, "loss": 0.2451, "step": 13629 }, { "epoch": 132.16867469879517, "grad_norm": 35.868896484375, "learning_rate": 3.383495145631068e-05, "loss": 0.1388, "step": 13630 }, { "epoch": 132.17831325301205, "grad_norm": 4.984102249145508, "learning_rate": 3.383009708737864e-05, "loss": 0.0664, "step": 13631 }, { "epoch": 132.18795180722893, "grad_norm": 24.535503387451172, "learning_rate": 3.38252427184466e-05, "loss": 0.2375, "step": 13632 }, { "epoch": 132.19759036144578, "grad_norm": 15.381063461303711, "learning_rate": 3.3820388349514566e-05, "loss": 0.1384, "step": 13633 }, { "epoch": 132.20722891566265, "grad_norm": 14.378352165222168, "learning_rate": 3.381553398058252e-05, "loss": 0.3855, "step": 13634 }, { "epoch": 132.21686746987953, "grad_norm": 13.258000373840332, "learning_rate": 3.381067961165049e-05, "loss": 0.1199, "step": 13635 }, { "epoch": 132.22650602409638, "grad_norm": 5.9701104164123535, "learning_rate": 3.3805825242718446e-05, "loss": 0.1151, "step": 13636 }, { "epoch": 132.23614457831326, "grad_norm": 33.63702392578125, "learning_rate": 3.380097087378641e-05, "loss": 0.3023, "step": 13637 }, { "epoch": 132.2457831325301, "grad_norm": 33.15889358520508, "learning_rate": 3.379611650485437e-05, "loss": 0.3042, "step": 13638 }, { "epoch": 132.25542168674698, "grad_norm": 20.55403709411621, "learning_rate": 3.379126213592233e-05, "loss": 0.2464, "step": 13639 }, { "epoch": 132.26506024096386, "grad_norm": 4.253203868865967, "learning_rate": 3.378640776699029e-05, "loss": 0.2552, "step": 13640 }, { "epoch": 132.2746987951807, "grad_norm": 13.98932933807373, "learning_rate": 3.3781553398058255e-05, "loss": 0.36, "step": 13641 }, { "epoch": 132.2843373493976, "grad_norm": 22.816572189331055, "learning_rate": 3.377669902912621e-05, "loss": 0.4349, "step": 13642 }, { "epoch": 132.29397590361447, "grad_norm": 5.165993690490723, "learning_rate": 3.377184466019418e-05, "loss": 0.1679, "step": 13643 }, { "epoch": 132.30361445783132, "grad_norm": 25.09433364868164, "learning_rate": 3.3766990291262135e-05, "loss": 0.287, "step": 13644 }, { "epoch": 132.3132530120482, "grad_norm": 17.93970489501953, "learning_rate": 3.37621359223301e-05, "loss": 0.4044, "step": 13645 }, { "epoch": 132.32289156626507, "grad_norm": 28.46797752380371, "learning_rate": 3.3757281553398064e-05, "loss": 0.2147, "step": 13646 }, { "epoch": 132.33253012048192, "grad_norm": 7.029428482055664, "learning_rate": 3.375242718446602e-05, "loss": 0.1824, "step": 13647 }, { "epoch": 132.3421686746988, "grad_norm": 69.58094787597656, "learning_rate": 3.3747572815533986e-05, "loss": 0.1763, "step": 13648 }, { "epoch": 132.35180722891567, "grad_norm": 16.750871658325195, "learning_rate": 3.3742718446601944e-05, "loss": 0.2632, "step": 13649 }, { "epoch": 132.36144578313252, "grad_norm": 7.740984916687012, "learning_rate": 3.373786407766991e-05, "loss": 0.181, "step": 13650 }, { "epoch": 132.3710843373494, "grad_norm": 5.779701232910156, "learning_rate": 3.3733009708737866e-05, "loss": 0.1515, "step": 13651 }, { "epoch": 132.38072289156628, "grad_norm": 10.885688781738281, "learning_rate": 3.372815533980583e-05, "loss": 0.2845, "step": 13652 }, { "epoch": 132.39036144578313, "grad_norm": 6.511817932128906, "learning_rate": 3.372330097087379e-05, "loss": 0.1582, "step": 13653 }, { "epoch": 132.4, "grad_norm": 14.735854148864746, "learning_rate": 3.371844660194175e-05, "loss": 0.3728, "step": 13654 }, { "epoch": 132.40963855421685, "grad_norm": 5.777738094329834, "learning_rate": 3.371359223300971e-05, "loss": 0.2612, "step": 13655 }, { "epoch": 132.41927710843373, "grad_norm": 12.985976219177246, "learning_rate": 3.3708737864077675e-05, "loss": 0.2388, "step": 13656 }, { "epoch": 132.4289156626506, "grad_norm": 49.96842575073242, "learning_rate": 3.370388349514563e-05, "loss": 0.2434, "step": 13657 }, { "epoch": 132.43855421686746, "grad_norm": 391.26458740234375, "learning_rate": 3.369902912621359e-05, "loss": 0.2625, "step": 13658 }, { "epoch": 132.44819277108434, "grad_norm": 8.678410530090332, "learning_rate": 3.3694174757281555e-05, "loss": 0.2216, "step": 13659 }, { "epoch": 132.4578313253012, "grad_norm": 5.965939044952393, "learning_rate": 3.368932038834951e-05, "loss": 0.1488, "step": 13660 }, { "epoch": 132.46746987951806, "grad_norm": 8.201305389404297, "learning_rate": 3.368446601941748e-05, "loss": 0.4137, "step": 13661 }, { "epoch": 132.47710843373494, "grad_norm": 54.59120559692383, "learning_rate": 3.3679611650485435e-05, "loss": 0.2477, "step": 13662 }, { "epoch": 132.48674698795182, "grad_norm": 91.4515151977539, "learning_rate": 3.36747572815534e-05, "loss": 0.4371, "step": 13663 }, { "epoch": 132.49638554216867, "grad_norm": 13.406990051269531, "learning_rate": 3.366990291262136e-05, "loss": 0.2348, "step": 13664 }, { "epoch": 132.50602409638554, "grad_norm": 14.1790771484375, "learning_rate": 3.366504854368932e-05, "loss": 0.2676, "step": 13665 }, { "epoch": 132.51566265060242, "grad_norm": 5.680050373077393, "learning_rate": 3.366019417475728e-05, "loss": 0.2467, "step": 13666 }, { "epoch": 132.52530120481927, "grad_norm": 30.096235275268555, "learning_rate": 3.3655339805825244e-05, "loss": 0.12, "step": 13667 }, { "epoch": 132.53493975903615, "grad_norm": 6.649334907531738, "learning_rate": 3.36504854368932e-05, "loss": 0.2188, "step": 13668 }, { "epoch": 132.544578313253, "grad_norm": 15.386519432067871, "learning_rate": 3.364563106796117e-05, "loss": 0.4441, "step": 13669 }, { "epoch": 132.55421686746988, "grad_norm": 23.556224822998047, "learning_rate": 3.3640776699029125e-05, "loss": 0.3042, "step": 13670 }, { "epoch": 132.56385542168675, "grad_norm": 10.679888725280762, "learning_rate": 3.363592233009709e-05, "loss": 0.6456, "step": 13671 }, { "epoch": 132.5734939759036, "grad_norm": 8.849860191345215, "learning_rate": 3.3631067961165054e-05, "loss": 0.2978, "step": 13672 }, { "epoch": 132.58313253012048, "grad_norm": 5.623312950134277, "learning_rate": 3.362621359223301e-05, "loss": 0.1674, "step": 13673 }, { "epoch": 132.59277108433736, "grad_norm": 28.428014755249023, "learning_rate": 3.3621359223300976e-05, "loss": 0.3297, "step": 13674 }, { "epoch": 132.6024096385542, "grad_norm": 15.144929885864258, "learning_rate": 3.3616504854368934e-05, "loss": 0.0931, "step": 13675 }, { "epoch": 132.61204819277108, "grad_norm": 14.389167785644531, "learning_rate": 3.36116504854369e-05, "loss": 0.2615, "step": 13676 }, { "epoch": 132.62168674698796, "grad_norm": 4.932102203369141, "learning_rate": 3.3606796116504856e-05, "loss": 0.2097, "step": 13677 }, { "epoch": 132.6313253012048, "grad_norm": 21.622529983520508, "learning_rate": 3.360194174757282e-05, "loss": 0.3155, "step": 13678 }, { "epoch": 132.6409638554217, "grad_norm": 22.72714614868164, "learning_rate": 3.359708737864078e-05, "loss": 0.1855, "step": 13679 }, { "epoch": 132.65060240963857, "grad_norm": 10.15408706665039, "learning_rate": 3.359223300970874e-05, "loss": 0.3701, "step": 13680 }, { "epoch": 132.66024096385541, "grad_norm": 6.412547588348389, "learning_rate": 3.35873786407767e-05, "loss": 0.2718, "step": 13681 }, { "epoch": 132.6698795180723, "grad_norm": 9.237106323242188, "learning_rate": 3.3582524271844665e-05, "loss": 0.2287, "step": 13682 }, { "epoch": 132.67951807228917, "grad_norm": 23.313003540039062, "learning_rate": 3.357766990291262e-05, "loss": 0.1865, "step": 13683 }, { "epoch": 132.68915662650602, "grad_norm": 33.05116271972656, "learning_rate": 3.357281553398058e-05, "loss": 0.4189, "step": 13684 }, { "epoch": 132.6987951807229, "grad_norm": 15.28545093536377, "learning_rate": 3.3567961165048545e-05, "loss": 0.5339, "step": 13685 }, { "epoch": 132.70843373493975, "grad_norm": 25.60013771057129, "learning_rate": 3.35631067961165e-05, "loss": 0.3931, "step": 13686 }, { "epoch": 132.71807228915662, "grad_norm": 27.232398986816406, "learning_rate": 3.355825242718447e-05, "loss": 0.3636, "step": 13687 }, { "epoch": 132.7277108433735, "grad_norm": 8.337570190429688, "learning_rate": 3.3553398058252425e-05, "loss": 0.4017, "step": 13688 }, { "epoch": 132.73734939759035, "grad_norm": 5.70578670501709, "learning_rate": 3.354854368932039e-05, "loss": 0.2436, "step": 13689 }, { "epoch": 132.74698795180723, "grad_norm": 8.014358520507812, "learning_rate": 3.354368932038835e-05, "loss": 0.2685, "step": 13690 }, { "epoch": 132.7566265060241, "grad_norm": 26.63762664794922, "learning_rate": 3.353883495145631e-05, "loss": 0.3656, "step": 13691 }, { "epoch": 132.76626506024095, "grad_norm": 44.12220764160156, "learning_rate": 3.353398058252427e-05, "loss": 0.2251, "step": 13692 }, { "epoch": 132.77590361445783, "grad_norm": 18.7071475982666, "learning_rate": 3.3529126213592234e-05, "loss": 0.1919, "step": 13693 }, { "epoch": 132.7855421686747, "grad_norm": 4.895333766937256, "learning_rate": 3.352427184466019e-05, "loss": 0.1483, "step": 13694 }, { "epoch": 132.79518072289156, "grad_norm": 15.834417343139648, "learning_rate": 3.3519417475728157e-05, "loss": 0.2368, "step": 13695 }, { "epoch": 132.80481927710844, "grad_norm": 21.907136917114258, "learning_rate": 3.351456310679612e-05, "loss": 0.3194, "step": 13696 }, { "epoch": 132.8144578313253, "grad_norm": 14.103486061096191, "learning_rate": 3.350970873786408e-05, "loss": 0.293, "step": 13697 }, { "epoch": 132.82409638554216, "grad_norm": 21.26139259338379, "learning_rate": 3.350485436893204e-05, "loss": 0.4435, "step": 13698 }, { "epoch": 132.83373493975904, "grad_norm": 43.1587028503418, "learning_rate": 3.35e-05, "loss": 0.2741, "step": 13699 }, { "epoch": 132.84337349397592, "grad_norm": 37.86965560913086, "learning_rate": 3.3495145631067966e-05, "loss": 0.4653, "step": 13700 }, { "epoch": 132.85301204819277, "grad_norm": 6.677762031555176, "learning_rate": 3.3490291262135923e-05, "loss": 0.1041, "step": 13701 }, { "epoch": 132.86265060240964, "grad_norm": 26.41208267211914, "learning_rate": 3.348543689320389e-05, "loss": 0.3865, "step": 13702 }, { "epoch": 132.8722891566265, "grad_norm": 12.904714584350586, "learning_rate": 3.3480582524271846e-05, "loss": 0.2354, "step": 13703 }, { "epoch": 132.88192771084337, "grad_norm": 2.9021658897399902, "learning_rate": 3.347572815533981e-05, "loss": 0.0861, "step": 13704 }, { "epoch": 132.89156626506025, "grad_norm": 8.862812042236328, "learning_rate": 3.347087378640777e-05, "loss": 0.3647, "step": 13705 }, { "epoch": 132.9012048192771, "grad_norm": 11.0310640335083, "learning_rate": 3.346601941747573e-05, "loss": 0.3068, "step": 13706 }, { "epoch": 132.91084337349398, "grad_norm": 6.528685569763184, "learning_rate": 3.346116504854369e-05, "loss": 0.267, "step": 13707 }, { "epoch": 132.92048192771085, "grad_norm": 11.110827445983887, "learning_rate": 3.3456310679611655e-05, "loss": 0.3935, "step": 13708 }, { "epoch": 132.9301204819277, "grad_norm": 24.73628044128418, "learning_rate": 3.345145631067961e-05, "loss": 0.3485, "step": 13709 }, { "epoch": 132.93975903614458, "grad_norm": 5.049406051635742, "learning_rate": 3.344660194174758e-05, "loss": 0.2057, "step": 13710 }, { "epoch": 132.94939759036146, "grad_norm": 21.242477416992188, "learning_rate": 3.3441747572815535e-05, "loss": 0.1378, "step": 13711 }, { "epoch": 132.9590361445783, "grad_norm": 20.11415672302246, "learning_rate": 3.343689320388349e-05, "loss": 0.1843, "step": 13712 }, { "epoch": 132.96867469879518, "grad_norm": 10.112561225891113, "learning_rate": 3.343203883495146e-05, "loss": 0.2056, "step": 13713 }, { "epoch": 132.97831325301206, "grad_norm": 13.88673210144043, "learning_rate": 3.3427184466019415e-05, "loss": 0.1734, "step": 13714 }, { "epoch": 132.9879518072289, "grad_norm": 9.744302749633789, "learning_rate": 3.342233009708738e-05, "loss": 0.2175, "step": 13715 }, { "epoch": 133.00361445783133, "grad_norm": 7.761867046356201, "learning_rate": 3.341747572815534e-05, "loss": 0.1708, "step": 13716 }, { "epoch": 133.01325301204818, "grad_norm": 3.5420241355895996, "learning_rate": 3.34126213592233e-05, "loss": 0.247, "step": 13717 }, { "epoch": 133.02289156626506, "grad_norm": 28.746578216552734, "learning_rate": 3.340776699029126e-05, "loss": 0.2604, "step": 13718 }, { "epoch": 133.03253012048194, "grad_norm": 3.692765951156616, "learning_rate": 3.3402912621359224e-05, "loss": 0.1684, "step": 13719 }, { "epoch": 133.0421686746988, "grad_norm": 6.0341997146606445, "learning_rate": 3.339805825242718e-05, "loss": 0.2068, "step": 13720 }, { "epoch": 133.05180722891566, "grad_norm": 7.454262733459473, "learning_rate": 3.339320388349515e-05, "loss": 0.2343, "step": 13721 }, { "epoch": 133.06144578313254, "grad_norm": 11.69851016998291, "learning_rate": 3.338834951456311e-05, "loss": 0.2011, "step": 13722 }, { "epoch": 133.0710843373494, "grad_norm": 32.489845275878906, "learning_rate": 3.338349514563107e-05, "loss": 0.6012, "step": 13723 }, { "epoch": 133.08072289156627, "grad_norm": 13.440385818481445, "learning_rate": 3.337864077669903e-05, "loss": 0.3584, "step": 13724 }, { "epoch": 133.09036144578315, "grad_norm": 21.423391342163086, "learning_rate": 3.337378640776699e-05, "loss": 0.4125, "step": 13725 }, { "epoch": 133.1, "grad_norm": 7.325484752655029, "learning_rate": 3.3368932038834955e-05, "loss": 0.1404, "step": 13726 }, { "epoch": 133.10963855421687, "grad_norm": 8.209809303283691, "learning_rate": 3.336407766990291e-05, "loss": 0.2335, "step": 13727 }, { "epoch": 133.11927710843372, "grad_norm": 15.765312194824219, "learning_rate": 3.335922330097088e-05, "loss": 0.1866, "step": 13728 }, { "epoch": 133.1289156626506, "grad_norm": 10.854698181152344, "learning_rate": 3.3354368932038836e-05, "loss": 0.2177, "step": 13729 }, { "epoch": 133.13855421686748, "grad_norm": 5.521624565124512, "learning_rate": 3.33495145631068e-05, "loss": 0.1125, "step": 13730 }, { "epoch": 133.14819277108433, "grad_norm": 6.45716667175293, "learning_rate": 3.334466019417476e-05, "loss": 0.1917, "step": 13731 }, { "epoch": 133.1578313253012, "grad_norm": 18.34492301940918, "learning_rate": 3.333980582524272e-05, "loss": 0.5327, "step": 13732 }, { "epoch": 133.16746987951808, "grad_norm": 15.895339012145996, "learning_rate": 3.333495145631068e-05, "loss": 0.1463, "step": 13733 }, { "epoch": 133.17710843373493, "grad_norm": 5.427181243896484, "learning_rate": 3.3330097087378645e-05, "loss": 0.2063, "step": 13734 }, { "epoch": 133.1867469879518, "grad_norm": 41.104061126708984, "learning_rate": 3.33252427184466e-05, "loss": 0.3327, "step": 13735 }, { "epoch": 133.19638554216868, "grad_norm": 8.415070533752441, "learning_rate": 3.332038834951457e-05, "loss": 0.1434, "step": 13736 }, { "epoch": 133.20602409638553, "grad_norm": 9.268352508544922, "learning_rate": 3.3315533980582525e-05, "loss": 0.1777, "step": 13737 }, { "epoch": 133.2156626506024, "grad_norm": 20.606842041015625, "learning_rate": 3.331067961165049e-05, "loss": 0.3317, "step": 13738 }, { "epoch": 133.2253012048193, "grad_norm": 8.724493980407715, "learning_rate": 3.330582524271845e-05, "loss": 0.228, "step": 13739 }, { "epoch": 133.23493975903614, "grad_norm": 32.69675827026367, "learning_rate": 3.3300970873786405e-05, "loss": 0.2537, "step": 13740 }, { "epoch": 133.24457831325302, "grad_norm": 35.278079986572266, "learning_rate": 3.329611650485437e-05, "loss": 0.4746, "step": 13741 }, { "epoch": 133.2542168674699, "grad_norm": 5.599747657775879, "learning_rate": 3.329126213592233e-05, "loss": 0.2015, "step": 13742 }, { "epoch": 133.26385542168674, "grad_norm": 4.716024398803711, "learning_rate": 3.328640776699029e-05, "loss": 0.201, "step": 13743 }, { "epoch": 133.27349397590362, "grad_norm": 8.752278327941895, "learning_rate": 3.328155339805825e-05, "loss": 0.1739, "step": 13744 }, { "epoch": 133.28313253012047, "grad_norm": 10.841675758361816, "learning_rate": 3.3276699029126214e-05, "loss": 0.2646, "step": 13745 }, { "epoch": 133.29277108433735, "grad_norm": 8.367605209350586, "learning_rate": 3.327184466019418e-05, "loss": 0.1512, "step": 13746 }, { "epoch": 133.30240963855422, "grad_norm": 42.72789001464844, "learning_rate": 3.326699029126214e-05, "loss": 0.3034, "step": 13747 }, { "epoch": 133.31204819277107, "grad_norm": 11.413039207458496, "learning_rate": 3.32621359223301e-05, "loss": 0.2576, "step": 13748 }, { "epoch": 133.32168674698795, "grad_norm": 4.483572006225586, "learning_rate": 3.3257281553398065e-05, "loss": 0.078, "step": 13749 }, { "epoch": 133.33132530120483, "grad_norm": 8.08145809173584, "learning_rate": 3.325242718446602e-05, "loss": 0.2505, "step": 13750 }, { "epoch": 133.34096385542168, "grad_norm": 19.24178695678711, "learning_rate": 3.324757281553398e-05, "loss": 0.2136, "step": 13751 }, { "epoch": 133.35060240963855, "grad_norm": 11.994884490966797, "learning_rate": 3.3242718446601945e-05, "loss": 0.3135, "step": 13752 }, { "epoch": 133.36024096385543, "grad_norm": 14.220369338989258, "learning_rate": 3.32378640776699e-05, "loss": 0.304, "step": 13753 }, { "epoch": 133.36987951807228, "grad_norm": 13.414021492004395, "learning_rate": 3.323300970873787e-05, "loss": 0.1436, "step": 13754 }, { "epoch": 133.37951807228916, "grad_norm": 8.993660926818848, "learning_rate": 3.3228155339805825e-05, "loss": 0.3208, "step": 13755 }, { "epoch": 133.38915662650604, "grad_norm": 47.00698471069336, "learning_rate": 3.322330097087379e-05, "loss": 0.4155, "step": 13756 }, { "epoch": 133.39879518072289, "grad_norm": 15.483074188232422, "learning_rate": 3.321844660194175e-05, "loss": 0.0879, "step": 13757 }, { "epoch": 133.40843373493976, "grad_norm": 8.497507095336914, "learning_rate": 3.321359223300971e-05, "loss": 0.3756, "step": 13758 }, { "epoch": 133.41807228915664, "grad_norm": 5.409975528717041, "learning_rate": 3.320873786407767e-05, "loss": 0.1799, "step": 13759 }, { "epoch": 133.4277108433735, "grad_norm": 3.7519469261169434, "learning_rate": 3.3203883495145634e-05, "loss": 0.1085, "step": 13760 }, { "epoch": 133.43734939759037, "grad_norm": 15.912065505981445, "learning_rate": 3.319902912621359e-05, "loss": 0.1545, "step": 13761 }, { "epoch": 133.44698795180722, "grad_norm": 28.30109405517578, "learning_rate": 3.319417475728156e-05, "loss": 0.3413, "step": 13762 }, { "epoch": 133.4566265060241, "grad_norm": 10.306010246276855, "learning_rate": 3.3189320388349514e-05, "loss": 0.2352, "step": 13763 }, { "epoch": 133.46626506024097, "grad_norm": 4.748488903045654, "learning_rate": 3.318446601941748e-05, "loss": 0.0891, "step": 13764 }, { "epoch": 133.47590361445782, "grad_norm": 12.831518173217773, "learning_rate": 3.317961165048544e-05, "loss": 0.2443, "step": 13765 }, { "epoch": 133.4855421686747, "grad_norm": 4.924576759338379, "learning_rate": 3.31747572815534e-05, "loss": 0.3081, "step": 13766 }, { "epoch": 133.49518072289158, "grad_norm": 7.363048076629639, "learning_rate": 3.316990291262136e-05, "loss": 0.2967, "step": 13767 }, { "epoch": 133.50481927710842, "grad_norm": 8.878419876098633, "learning_rate": 3.316504854368932e-05, "loss": 0.1309, "step": 13768 }, { "epoch": 133.5144578313253, "grad_norm": 15.44411849975586, "learning_rate": 3.316019417475728e-05, "loss": 0.2177, "step": 13769 }, { "epoch": 133.52409638554218, "grad_norm": 14.184847831726074, "learning_rate": 3.315533980582524e-05, "loss": 0.3023, "step": 13770 }, { "epoch": 133.53373493975903, "grad_norm": 9.274717330932617, "learning_rate": 3.3150485436893204e-05, "loss": 0.3888, "step": 13771 }, { "epoch": 133.5433734939759, "grad_norm": 22.124765396118164, "learning_rate": 3.314563106796117e-05, "loss": 0.3334, "step": 13772 }, { "epoch": 133.55301204819278, "grad_norm": 4.561752796173096, "learning_rate": 3.314077669902913e-05, "loss": 0.1255, "step": 13773 }, { "epoch": 133.56265060240963, "grad_norm": 15.00654125213623, "learning_rate": 3.313592233009709e-05, "loss": 0.3641, "step": 13774 }, { "epoch": 133.5722891566265, "grad_norm": 6.474641799926758, "learning_rate": 3.3131067961165055e-05, "loss": 0.3115, "step": 13775 }, { "epoch": 133.58192771084336, "grad_norm": 10.73024845123291, "learning_rate": 3.312621359223301e-05, "loss": 0.3196, "step": 13776 }, { "epoch": 133.59156626506024, "grad_norm": 10.16966438293457, "learning_rate": 3.312135922330098e-05, "loss": 0.3024, "step": 13777 }, { "epoch": 133.60120481927711, "grad_norm": 18.962539672851562, "learning_rate": 3.3116504854368935e-05, "loss": 0.197, "step": 13778 }, { "epoch": 133.61084337349396, "grad_norm": 15.037787437438965, "learning_rate": 3.311165048543689e-05, "loss": 0.435, "step": 13779 }, { "epoch": 133.62048192771084, "grad_norm": 6.595386505126953, "learning_rate": 3.310679611650486e-05, "loss": 0.1632, "step": 13780 }, { "epoch": 133.63012048192772, "grad_norm": 13.671719551086426, "learning_rate": 3.3101941747572815e-05, "loss": 0.1163, "step": 13781 }, { "epoch": 133.63975903614457, "grad_norm": 12.03559398651123, "learning_rate": 3.309708737864078e-05, "loss": 0.2031, "step": 13782 }, { "epoch": 133.64939759036145, "grad_norm": 14.107834815979004, "learning_rate": 3.309223300970874e-05, "loss": 0.2109, "step": 13783 }, { "epoch": 133.65903614457832, "grad_norm": 15.429682731628418, "learning_rate": 3.30873786407767e-05, "loss": 0.3684, "step": 13784 }, { "epoch": 133.66867469879517, "grad_norm": 48.960304260253906, "learning_rate": 3.308252427184466e-05, "loss": 0.5423, "step": 13785 }, { "epoch": 133.67831325301205, "grad_norm": 9.06633472442627, "learning_rate": 3.3077669902912624e-05, "loss": 0.4157, "step": 13786 }, { "epoch": 133.68795180722893, "grad_norm": 8.01599407196045, "learning_rate": 3.307281553398058e-05, "loss": 0.1806, "step": 13787 }, { "epoch": 133.69759036144578, "grad_norm": 17.42967987060547, "learning_rate": 3.3067961165048546e-05, "loss": 0.3532, "step": 13788 }, { "epoch": 133.70722891566265, "grad_norm": 8.237302780151367, "learning_rate": 3.3063106796116504e-05, "loss": 0.3448, "step": 13789 }, { "epoch": 133.71686746987953, "grad_norm": 34.95002365112305, "learning_rate": 3.305825242718447e-05, "loss": 0.2179, "step": 13790 }, { "epoch": 133.72650602409638, "grad_norm": 3.7193374633789062, "learning_rate": 3.3053398058252427e-05, "loss": 0.3081, "step": 13791 }, { "epoch": 133.73614457831326, "grad_norm": 3.6868183612823486, "learning_rate": 3.304854368932039e-05, "loss": 0.0645, "step": 13792 }, { "epoch": 133.7457831325301, "grad_norm": 18.955419540405273, "learning_rate": 3.304368932038835e-05, "loss": 0.346, "step": 13793 }, { "epoch": 133.75542168674698, "grad_norm": 2.3822696208953857, "learning_rate": 3.303883495145631e-05, "loss": 0.0509, "step": 13794 }, { "epoch": 133.76506024096386, "grad_norm": 14.402128219604492, "learning_rate": 3.303398058252427e-05, "loss": 0.2807, "step": 13795 }, { "epoch": 133.7746987951807, "grad_norm": 14.4298095703125, "learning_rate": 3.302912621359223e-05, "loss": 0.3215, "step": 13796 }, { "epoch": 133.7843373493976, "grad_norm": 5.391659259796143, "learning_rate": 3.30242718446602e-05, "loss": 0.2838, "step": 13797 }, { "epoch": 133.79397590361447, "grad_norm": 39.376644134521484, "learning_rate": 3.301941747572816e-05, "loss": 0.4198, "step": 13798 }, { "epoch": 133.80361445783132, "grad_norm": 9.2643404006958, "learning_rate": 3.301456310679612e-05, "loss": 0.2215, "step": 13799 }, { "epoch": 133.8132530120482, "grad_norm": 23.747188568115234, "learning_rate": 3.300970873786408e-05, "loss": 0.448, "step": 13800 }, { "epoch": 133.82289156626507, "grad_norm": 5.185963153839111, "learning_rate": 3.3004854368932045e-05, "loss": 0.1855, "step": 13801 }, { "epoch": 133.83253012048192, "grad_norm": 18.24466323852539, "learning_rate": 3.3e-05, "loss": 0.2161, "step": 13802 }, { "epoch": 133.8421686746988, "grad_norm": 6.974105358123779, "learning_rate": 3.299514563106797e-05, "loss": 0.2404, "step": 13803 }, { "epoch": 133.85180722891567, "grad_norm": 10.104551315307617, "learning_rate": 3.2990291262135925e-05, "loss": 0.3508, "step": 13804 }, { "epoch": 133.86144578313252, "grad_norm": 6.357942581176758, "learning_rate": 3.298543689320389e-05, "loss": 0.2654, "step": 13805 }, { "epoch": 133.8710843373494, "grad_norm": 8.539295196533203, "learning_rate": 3.298058252427185e-05, "loss": 0.243, "step": 13806 }, { "epoch": 133.88072289156628, "grad_norm": 25.543704986572266, "learning_rate": 3.2975728155339805e-05, "loss": 0.2548, "step": 13807 }, { "epoch": 133.89036144578313, "grad_norm": 12.159149169921875, "learning_rate": 3.297087378640777e-05, "loss": 0.2258, "step": 13808 }, { "epoch": 133.9, "grad_norm": 39.45794677734375, "learning_rate": 3.296601941747573e-05, "loss": 0.2048, "step": 13809 }, { "epoch": 133.90963855421685, "grad_norm": 8.859378814697266, "learning_rate": 3.296116504854369e-05, "loss": 0.24, "step": 13810 }, { "epoch": 133.91927710843373, "grad_norm": 25.41267204284668, "learning_rate": 3.295631067961165e-05, "loss": 0.3876, "step": 13811 }, { "epoch": 133.9289156626506, "grad_norm": 9.424633026123047, "learning_rate": 3.2951456310679614e-05, "loss": 0.4277, "step": 13812 }, { "epoch": 133.93855421686746, "grad_norm": 80.93363189697266, "learning_rate": 3.294660194174757e-05, "loss": 0.4122, "step": 13813 }, { "epoch": 133.94819277108434, "grad_norm": 9.011750221252441, "learning_rate": 3.2941747572815536e-05, "loss": 0.2099, "step": 13814 }, { "epoch": 133.9578313253012, "grad_norm": 13.733734130859375, "learning_rate": 3.2936893203883494e-05, "loss": 0.3629, "step": 13815 }, { "epoch": 133.96746987951806, "grad_norm": 35.46942138671875, "learning_rate": 3.293203883495146e-05, "loss": 0.2452, "step": 13816 }, { "epoch": 133.97710843373494, "grad_norm": 7.094212055206299, "learning_rate": 3.2927184466019416e-05, "loss": 0.3271, "step": 13817 }, { "epoch": 133.98674698795182, "grad_norm": 39.77775192260742, "learning_rate": 3.292233009708738e-05, "loss": 0.291, "step": 13818 }, { "epoch": 134.0024096385542, "grad_norm": 22.979909896850586, "learning_rate": 3.291747572815534e-05, "loss": 0.4095, "step": 13819 }, { "epoch": 134.0120481927711, "grad_norm": 9.263836860656738, "learning_rate": 3.29126213592233e-05, "loss": 0.2223, "step": 13820 }, { "epoch": 134.02168674698794, "grad_norm": 5.380708694458008, "learning_rate": 3.290776699029126e-05, "loss": 0.2605, "step": 13821 }, { "epoch": 134.03132530120482, "grad_norm": 72.34098815917969, "learning_rate": 3.2902912621359225e-05, "loss": 0.2229, "step": 13822 }, { "epoch": 134.0409638554217, "grad_norm": 16.6566219329834, "learning_rate": 3.289805825242719e-05, "loss": 0.2519, "step": 13823 }, { "epoch": 134.05060240963854, "grad_norm": 28.770692825317383, "learning_rate": 3.289320388349515e-05, "loss": 0.2138, "step": 13824 }, { "epoch": 134.06024096385542, "grad_norm": 6.809057235717773, "learning_rate": 3.288834951456311e-05, "loss": 0.3131, "step": 13825 }, { "epoch": 134.0698795180723, "grad_norm": 23.57328987121582, "learning_rate": 3.288349514563107e-05, "loss": 0.2745, "step": 13826 }, { "epoch": 134.07951807228915, "grad_norm": 12.925342559814453, "learning_rate": 3.2878640776699035e-05, "loss": 0.2597, "step": 13827 }, { "epoch": 134.08915662650602, "grad_norm": 9.420825004577637, "learning_rate": 3.287378640776699e-05, "loss": 0.3057, "step": 13828 }, { "epoch": 134.0987951807229, "grad_norm": 8.811455726623535, "learning_rate": 3.286893203883496e-05, "loss": 0.2102, "step": 13829 }, { "epoch": 134.10843373493975, "grad_norm": 46.20577621459961, "learning_rate": 3.2864077669902915e-05, "loss": 0.2041, "step": 13830 }, { "epoch": 134.11807228915663, "grad_norm": 51.130348205566406, "learning_rate": 3.285922330097088e-05, "loss": 0.3433, "step": 13831 }, { "epoch": 134.1277108433735, "grad_norm": 10.87099552154541, "learning_rate": 3.285436893203884e-05, "loss": 0.1935, "step": 13832 }, { "epoch": 134.13734939759036, "grad_norm": 9.788514137268066, "learning_rate": 3.2849514563106795e-05, "loss": 0.1879, "step": 13833 }, { "epoch": 134.14698795180723, "grad_norm": 15.053106307983398, "learning_rate": 3.284466019417476e-05, "loss": 0.4622, "step": 13834 }, { "epoch": 134.15662650602408, "grad_norm": 25.895313262939453, "learning_rate": 3.283980582524272e-05, "loss": 0.2475, "step": 13835 }, { "epoch": 134.16626506024096, "grad_norm": 20.075605392456055, "learning_rate": 3.283495145631068e-05, "loss": 0.4126, "step": 13836 }, { "epoch": 134.17590361445784, "grad_norm": 5.374949932098389, "learning_rate": 3.283009708737864e-05, "loss": 0.1988, "step": 13837 }, { "epoch": 134.1855421686747, "grad_norm": 6.440910816192627, "learning_rate": 3.2825242718446604e-05, "loss": 0.2749, "step": 13838 }, { "epoch": 134.19518072289156, "grad_norm": 8.26274299621582, "learning_rate": 3.282038834951456e-05, "loss": 0.3049, "step": 13839 }, { "epoch": 134.20481927710844, "grad_norm": 8.799696922302246, "learning_rate": 3.2815533980582526e-05, "loss": 0.2358, "step": 13840 }, { "epoch": 134.2144578313253, "grad_norm": 3.570174217224121, "learning_rate": 3.2810679611650484e-05, "loss": 0.0838, "step": 13841 }, { "epoch": 134.22409638554217, "grad_norm": 30.416269302368164, "learning_rate": 3.280582524271845e-05, "loss": 0.3787, "step": 13842 }, { "epoch": 134.23373493975905, "grad_norm": 14.325778007507324, "learning_rate": 3.2800970873786406e-05, "loss": 0.2218, "step": 13843 }, { "epoch": 134.2433734939759, "grad_norm": 62.58586120605469, "learning_rate": 3.279611650485437e-05, "loss": 0.2669, "step": 13844 }, { "epoch": 134.25301204819277, "grad_norm": 110.79974365234375, "learning_rate": 3.279126213592233e-05, "loss": 0.5191, "step": 13845 }, { "epoch": 134.26265060240965, "grad_norm": 14.223814010620117, "learning_rate": 3.278640776699029e-05, "loss": 0.2533, "step": 13846 }, { "epoch": 134.2722891566265, "grad_norm": 8.782703399658203, "learning_rate": 3.278155339805826e-05, "loss": 0.2676, "step": 13847 }, { "epoch": 134.28192771084338, "grad_norm": 10.071982383728027, "learning_rate": 3.2776699029126215e-05, "loss": 0.2951, "step": 13848 }, { "epoch": 134.29156626506025, "grad_norm": 11.046741485595703, "learning_rate": 3.277184466019418e-05, "loss": 0.2805, "step": 13849 }, { "epoch": 134.3012048192771, "grad_norm": 6.871456146240234, "learning_rate": 3.276699029126214e-05, "loss": 0.2236, "step": 13850 }, { "epoch": 134.31084337349398, "grad_norm": 7.369706630706787, "learning_rate": 3.27621359223301e-05, "loss": 0.2187, "step": 13851 }, { "epoch": 134.32048192771083, "grad_norm": 8.340028762817383, "learning_rate": 3.275728155339806e-05, "loss": 0.3894, "step": 13852 }, { "epoch": 134.3301204819277, "grad_norm": 9.175713539123535, "learning_rate": 3.2752427184466024e-05, "loss": 0.3015, "step": 13853 }, { "epoch": 134.33975903614459, "grad_norm": 5.447513580322266, "learning_rate": 3.274757281553398e-05, "loss": 0.1576, "step": 13854 }, { "epoch": 134.34939759036143, "grad_norm": 11.88592529296875, "learning_rate": 3.274271844660195e-05, "loss": 0.2831, "step": 13855 }, { "epoch": 134.3590361445783, "grad_norm": 20.14812469482422, "learning_rate": 3.2737864077669904e-05, "loss": 0.1648, "step": 13856 }, { "epoch": 134.3686746987952, "grad_norm": 12.038992881774902, "learning_rate": 3.273300970873787e-05, "loss": 0.1232, "step": 13857 }, { "epoch": 134.37831325301204, "grad_norm": 7.543919086456299, "learning_rate": 3.272815533980583e-05, "loss": 0.1446, "step": 13858 }, { "epoch": 134.38795180722892, "grad_norm": 10.044886589050293, "learning_rate": 3.272330097087379e-05, "loss": 0.0898, "step": 13859 }, { "epoch": 134.3975903614458, "grad_norm": 8.261584281921387, "learning_rate": 3.271844660194175e-05, "loss": 0.1942, "step": 13860 }, { "epoch": 134.40722891566264, "grad_norm": 6.287014007568359, "learning_rate": 3.271359223300971e-05, "loss": 0.1407, "step": 13861 }, { "epoch": 134.41686746987952, "grad_norm": 15.28999137878418, "learning_rate": 3.270873786407767e-05, "loss": 0.3032, "step": 13862 }, { "epoch": 134.4265060240964, "grad_norm": 5.69724178314209, "learning_rate": 3.270388349514563e-05, "loss": 0.425, "step": 13863 }, { "epoch": 134.43614457831325, "grad_norm": 13.451590538024902, "learning_rate": 3.2699029126213594e-05, "loss": 0.3438, "step": 13864 }, { "epoch": 134.44578313253012, "grad_norm": 8.388564109802246, "learning_rate": 3.269417475728155e-05, "loss": 0.2865, "step": 13865 }, { "epoch": 134.455421686747, "grad_norm": 6.885880470275879, "learning_rate": 3.2689320388349516e-05, "loss": 0.1033, "step": 13866 }, { "epoch": 134.46506024096385, "grad_norm": 5.552962779998779, "learning_rate": 3.2684466019417474e-05, "loss": 0.1657, "step": 13867 }, { "epoch": 134.47469879518073, "grad_norm": 6.294094562530518, "learning_rate": 3.267961165048544e-05, "loss": 0.26, "step": 13868 }, { "epoch": 134.48433734939758, "grad_norm": 21.98668670654297, "learning_rate": 3.2674757281553396e-05, "loss": 0.3159, "step": 13869 }, { "epoch": 134.49397590361446, "grad_norm": 16.94541358947754, "learning_rate": 3.266990291262136e-05, "loss": 0.1981, "step": 13870 }, { "epoch": 134.50361445783133, "grad_norm": 10.137677192687988, "learning_rate": 3.266504854368932e-05, "loss": 0.2823, "step": 13871 }, { "epoch": 134.51325301204818, "grad_norm": 6.486586570739746, "learning_rate": 3.266019417475728e-05, "loss": 0.2158, "step": 13872 }, { "epoch": 134.52289156626506, "grad_norm": 12.456603050231934, "learning_rate": 3.265533980582525e-05, "loss": 0.3869, "step": 13873 }, { "epoch": 134.53253012048194, "grad_norm": 23.13381576538086, "learning_rate": 3.2650485436893205e-05, "loss": 0.2008, "step": 13874 }, { "epoch": 134.5421686746988, "grad_norm": 12.719049453735352, "learning_rate": 3.264563106796117e-05, "loss": 0.4659, "step": 13875 }, { "epoch": 134.55180722891566, "grad_norm": 9.27180290222168, "learning_rate": 3.264077669902913e-05, "loss": 0.2598, "step": 13876 }, { "epoch": 134.56144578313254, "grad_norm": 10.50050163269043, "learning_rate": 3.263592233009709e-05, "loss": 0.1822, "step": 13877 }, { "epoch": 134.5710843373494, "grad_norm": 22.778404235839844, "learning_rate": 3.263106796116505e-05, "loss": 0.2966, "step": 13878 }, { "epoch": 134.58072289156627, "grad_norm": 18.012781143188477, "learning_rate": 3.2626213592233014e-05, "loss": 0.4196, "step": 13879 }, { "epoch": 134.59036144578315, "grad_norm": 9.577914237976074, "learning_rate": 3.262135922330097e-05, "loss": 0.1918, "step": 13880 }, { "epoch": 134.6, "grad_norm": 3.0142717361450195, "learning_rate": 3.2616504854368936e-05, "loss": 0.0991, "step": 13881 }, { "epoch": 134.60963855421687, "grad_norm": 5.131934642791748, "learning_rate": 3.2611650485436894e-05, "loss": 0.1246, "step": 13882 }, { "epoch": 134.61927710843372, "grad_norm": 12.359321594238281, "learning_rate": 3.260679611650486e-05, "loss": 0.3036, "step": 13883 }, { "epoch": 134.6289156626506, "grad_norm": 5.458915710449219, "learning_rate": 3.2601941747572816e-05, "loss": 0.1536, "step": 13884 }, { "epoch": 134.63855421686748, "grad_norm": 6.8914055824279785, "learning_rate": 3.259708737864078e-05, "loss": 0.2869, "step": 13885 }, { "epoch": 134.64819277108433, "grad_norm": 37.60691833496094, "learning_rate": 3.259223300970874e-05, "loss": 0.4314, "step": 13886 }, { "epoch": 134.6578313253012, "grad_norm": 8.805107116699219, "learning_rate": 3.25873786407767e-05, "loss": 0.259, "step": 13887 }, { "epoch": 134.66746987951808, "grad_norm": 4.882838249206543, "learning_rate": 3.258252427184466e-05, "loss": 0.2092, "step": 13888 }, { "epoch": 134.67710843373493, "grad_norm": 34.83646011352539, "learning_rate": 3.257766990291262e-05, "loss": 0.3263, "step": 13889 }, { "epoch": 134.6867469879518, "grad_norm": 7.420343399047852, "learning_rate": 3.257281553398058e-05, "loss": 0.4302, "step": 13890 }, { "epoch": 134.69638554216868, "grad_norm": 14.40124225616455, "learning_rate": 3.256796116504854e-05, "loss": 0.1112, "step": 13891 }, { "epoch": 134.70602409638553, "grad_norm": 11.803290367126465, "learning_rate": 3.2563106796116506e-05, "loss": 0.4973, "step": 13892 }, { "epoch": 134.7156626506024, "grad_norm": 13.058704376220703, "learning_rate": 3.2558252427184463e-05, "loss": 0.1742, "step": 13893 }, { "epoch": 134.7253012048193, "grad_norm": 15.91738510131836, "learning_rate": 3.255339805825243e-05, "loss": 0.3269, "step": 13894 }, { "epoch": 134.73493975903614, "grad_norm": 18.965896606445312, "learning_rate": 3.2548543689320386e-05, "loss": 0.2656, "step": 13895 }, { "epoch": 134.74457831325302, "grad_norm": 5.569526195526123, "learning_rate": 3.254368932038835e-05, "loss": 0.2005, "step": 13896 }, { "epoch": 134.7542168674699, "grad_norm": 13.79041862487793, "learning_rate": 3.253883495145631e-05, "loss": 0.332, "step": 13897 }, { "epoch": 134.76385542168674, "grad_norm": 8.765859603881836, "learning_rate": 3.253398058252428e-05, "loss": 0.1972, "step": 13898 }, { "epoch": 134.77349397590362, "grad_norm": 8.84151554107666, "learning_rate": 3.252912621359224e-05, "loss": 0.1479, "step": 13899 }, { "epoch": 134.78313253012047, "grad_norm": 5.522930145263672, "learning_rate": 3.2524271844660195e-05, "loss": 0.2462, "step": 13900 }, { "epoch": 134.79277108433735, "grad_norm": 12.989212989807129, "learning_rate": 3.251941747572816e-05, "loss": 0.1873, "step": 13901 }, { "epoch": 134.80240963855422, "grad_norm": 7.933618068695068, "learning_rate": 3.251456310679612e-05, "loss": 0.1194, "step": 13902 }, { "epoch": 134.81204819277107, "grad_norm": 15.87862777709961, "learning_rate": 3.250970873786408e-05, "loss": 0.258, "step": 13903 }, { "epoch": 134.82168674698795, "grad_norm": 13.644547462463379, "learning_rate": 3.250485436893204e-05, "loss": 0.3267, "step": 13904 }, { "epoch": 134.83132530120483, "grad_norm": 10.510477066040039, "learning_rate": 3.2500000000000004e-05, "loss": 0.2753, "step": 13905 }, { "epoch": 134.84096385542168, "grad_norm": 13.573375701904297, "learning_rate": 3.249514563106796e-05, "loss": 0.1781, "step": 13906 }, { "epoch": 134.85060240963855, "grad_norm": 3.301840305328369, "learning_rate": 3.2490291262135926e-05, "loss": 0.1147, "step": 13907 }, { "epoch": 134.86024096385543, "grad_norm": 69.54247283935547, "learning_rate": 3.2485436893203884e-05, "loss": 0.3627, "step": 13908 }, { "epoch": 134.86987951807228, "grad_norm": 10.873772621154785, "learning_rate": 3.248058252427185e-05, "loss": 0.3185, "step": 13909 }, { "epoch": 134.87951807228916, "grad_norm": 3.313800096511841, "learning_rate": 3.2475728155339806e-05, "loss": 0.1078, "step": 13910 }, { "epoch": 134.88915662650604, "grad_norm": 34.47492599487305, "learning_rate": 3.247087378640777e-05, "loss": 0.2924, "step": 13911 }, { "epoch": 134.89879518072289, "grad_norm": 12.70819091796875, "learning_rate": 3.246601941747573e-05, "loss": 0.3772, "step": 13912 }, { "epoch": 134.90843373493976, "grad_norm": 5.571049213409424, "learning_rate": 3.246116504854369e-05, "loss": 0.158, "step": 13913 }, { "epoch": 134.91807228915664, "grad_norm": 14.510115623474121, "learning_rate": 3.245631067961165e-05, "loss": 0.2557, "step": 13914 }, { "epoch": 134.9277108433735, "grad_norm": 15.974505424499512, "learning_rate": 3.2451456310679615e-05, "loss": 0.501, "step": 13915 }, { "epoch": 134.93734939759037, "grad_norm": 4.8983259201049805, "learning_rate": 3.244660194174757e-05, "loss": 0.1843, "step": 13916 }, { "epoch": 134.94698795180722, "grad_norm": 30.44439125061035, "learning_rate": 3.244174757281553e-05, "loss": 0.4302, "step": 13917 }, { "epoch": 134.9566265060241, "grad_norm": 17.73951530456543, "learning_rate": 3.2436893203883495e-05, "loss": 0.3827, "step": 13918 }, { "epoch": 134.96626506024097, "grad_norm": 23.45370101928711, "learning_rate": 3.243203883495145e-05, "loss": 0.4354, "step": 13919 }, { "epoch": 134.97590361445782, "grad_norm": 8.684083938598633, "learning_rate": 3.242718446601942e-05, "loss": 0.4275, "step": 13920 }, { "epoch": 134.9855421686747, "grad_norm": 12.925365447998047, "learning_rate": 3.2422330097087375e-05, "loss": 0.2611, "step": 13921 }, { "epoch": 135.00120481927712, "grad_norm": 10.833080291748047, "learning_rate": 3.241747572815534e-05, "loss": 0.4158, "step": 13922 }, { "epoch": 135.01084337349397, "grad_norm": 3.9793341159820557, "learning_rate": 3.2412621359223305e-05, "loss": 0.0763, "step": 13923 }, { "epoch": 135.02048192771085, "grad_norm": 6.996010780334473, "learning_rate": 3.240776699029127e-05, "loss": 0.3144, "step": 13924 }, { "epoch": 135.03012048192772, "grad_norm": 12.664405822753906, "learning_rate": 3.240291262135923e-05, "loss": 0.3033, "step": 13925 }, { "epoch": 135.03975903614457, "grad_norm": 25.4429874420166, "learning_rate": 3.239805825242719e-05, "loss": 0.1914, "step": 13926 }, { "epoch": 135.04939759036145, "grad_norm": 13.19361400604248, "learning_rate": 3.239320388349515e-05, "loss": 0.2772, "step": 13927 }, { "epoch": 135.0590361445783, "grad_norm": 13.583498001098633, "learning_rate": 3.238834951456311e-05, "loss": 0.2966, "step": 13928 }, { "epoch": 135.06867469879518, "grad_norm": 5.270559310913086, "learning_rate": 3.238349514563107e-05, "loss": 0.0779, "step": 13929 }, { "epoch": 135.07831325301206, "grad_norm": 13.669820785522461, "learning_rate": 3.237864077669903e-05, "loss": 0.1169, "step": 13930 }, { "epoch": 135.0879518072289, "grad_norm": 10.804645538330078, "learning_rate": 3.2373786407766994e-05, "loss": 0.2742, "step": 13931 }, { "epoch": 135.09759036144578, "grad_norm": 13.198628425598145, "learning_rate": 3.236893203883495e-05, "loss": 0.1697, "step": 13932 }, { "epoch": 135.10722891566266, "grad_norm": 12.0918550491333, "learning_rate": 3.2364077669902916e-05, "loss": 0.2566, "step": 13933 }, { "epoch": 135.1168674698795, "grad_norm": 23.9401798248291, "learning_rate": 3.2359223300970874e-05, "loss": 0.5001, "step": 13934 }, { "epoch": 135.1265060240964, "grad_norm": 28.099578857421875, "learning_rate": 3.235436893203884e-05, "loss": 0.2437, "step": 13935 }, { "epoch": 135.13614457831326, "grad_norm": 9.651599884033203, "learning_rate": 3.2349514563106796e-05, "loss": 0.299, "step": 13936 }, { "epoch": 135.1457831325301, "grad_norm": 5.463251113891602, "learning_rate": 3.234466019417476e-05, "loss": 0.0978, "step": 13937 }, { "epoch": 135.155421686747, "grad_norm": 3.0244407653808594, "learning_rate": 3.233980582524272e-05, "loss": 0.1689, "step": 13938 }, { "epoch": 135.16506024096387, "grad_norm": 21.662073135375977, "learning_rate": 3.233495145631068e-05, "loss": 0.2066, "step": 13939 }, { "epoch": 135.17469879518072, "grad_norm": 13.773075103759766, "learning_rate": 3.233009708737864e-05, "loss": 0.3175, "step": 13940 }, { "epoch": 135.1843373493976, "grad_norm": 31.116525650024414, "learning_rate": 3.2325242718446605e-05, "loss": 0.4487, "step": 13941 }, { "epoch": 135.19397590361444, "grad_norm": 35.32596969604492, "learning_rate": 3.232038834951456e-05, "loss": 0.194, "step": 13942 }, { "epoch": 135.20361445783132, "grad_norm": 7.926493167877197, "learning_rate": 3.231553398058253e-05, "loss": 0.1065, "step": 13943 }, { "epoch": 135.2132530120482, "grad_norm": 9.046496391296387, "learning_rate": 3.2310679611650485e-05, "loss": 0.2633, "step": 13944 }, { "epoch": 135.22289156626505, "grad_norm": 5.566128253936768, "learning_rate": 3.230582524271844e-05, "loss": 0.0756, "step": 13945 }, { "epoch": 135.23253012048193, "grad_norm": 11.31774616241455, "learning_rate": 3.230097087378641e-05, "loss": 0.2107, "step": 13946 }, { "epoch": 135.2421686746988, "grad_norm": 26.493541717529297, "learning_rate": 3.2296116504854365e-05, "loss": 0.6699, "step": 13947 }, { "epoch": 135.25180722891565, "grad_norm": 9.96194076538086, "learning_rate": 3.2291262135922337e-05, "loss": 0.3102, "step": 13948 }, { "epoch": 135.26144578313253, "grad_norm": 10.627510070800781, "learning_rate": 3.2286407766990294e-05, "loss": 0.337, "step": 13949 }, { "epoch": 135.2710843373494, "grad_norm": 9.47526741027832, "learning_rate": 3.228155339805826e-05, "loss": 0.2586, "step": 13950 }, { "epoch": 135.28072289156626, "grad_norm": 29.141887664794922, "learning_rate": 3.227669902912622e-05, "loss": 0.2537, "step": 13951 }, { "epoch": 135.29036144578313, "grad_norm": 8.360867500305176, "learning_rate": 3.227184466019418e-05, "loss": 0.1672, "step": 13952 }, { "epoch": 135.3, "grad_norm": 22.95161247253418, "learning_rate": 3.226699029126214e-05, "loss": 0.3722, "step": 13953 }, { "epoch": 135.30963855421686, "grad_norm": 6.860239505767822, "learning_rate": 3.2262135922330103e-05, "loss": 0.091, "step": 13954 }, { "epoch": 135.31927710843374, "grad_norm": 16.63726234436035, "learning_rate": 3.225728155339806e-05, "loss": 0.1065, "step": 13955 }, { "epoch": 135.32891566265062, "grad_norm": 19.38883399963379, "learning_rate": 3.225242718446602e-05, "loss": 0.2775, "step": 13956 }, { "epoch": 135.33855421686746, "grad_norm": 13.236927032470703, "learning_rate": 3.2247572815533984e-05, "loss": 0.5153, "step": 13957 }, { "epoch": 135.34819277108434, "grad_norm": 7.9932403564453125, "learning_rate": 3.224271844660194e-05, "loss": 0.3149, "step": 13958 }, { "epoch": 135.3578313253012, "grad_norm": 8.60775375366211, "learning_rate": 3.2237864077669906e-05, "loss": 0.2496, "step": 13959 }, { "epoch": 135.36746987951807, "grad_norm": 19.340736389160156, "learning_rate": 3.2233009708737864e-05, "loss": 0.3936, "step": 13960 }, { "epoch": 135.37710843373495, "grad_norm": 5.814827919006348, "learning_rate": 3.222815533980583e-05, "loss": 0.2095, "step": 13961 }, { "epoch": 135.3867469879518, "grad_norm": 8.342453956604004, "learning_rate": 3.2223300970873786e-05, "loss": 0.1401, "step": 13962 }, { "epoch": 135.39638554216867, "grad_norm": 5.280797481536865, "learning_rate": 3.221844660194175e-05, "loss": 0.0795, "step": 13963 }, { "epoch": 135.40602409638555, "grad_norm": 5.247028350830078, "learning_rate": 3.221359223300971e-05, "loss": 0.1592, "step": 13964 }, { "epoch": 135.4156626506024, "grad_norm": 3.829440116882324, "learning_rate": 3.220873786407767e-05, "loss": 0.1449, "step": 13965 }, { "epoch": 135.42530120481928, "grad_norm": 27.351146697998047, "learning_rate": 3.220388349514563e-05, "loss": 0.3458, "step": 13966 }, { "epoch": 135.43493975903615, "grad_norm": 15.352827072143555, "learning_rate": 3.2199029126213595e-05, "loss": 0.2633, "step": 13967 }, { "epoch": 135.444578313253, "grad_norm": 41.458248138427734, "learning_rate": 3.219417475728155e-05, "loss": 0.5258, "step": 13968 }, { "epoch": 135.45421686746988, "grad_norm": 14.880428314208984, "learning_rate": 3.218932038834952e-05, "loss": 0.3708, "step": 13969 }, { "epoch": 135.46385542168676, "grad_norm": 25.87647247314453, "learning_rate": 3.2184466019417475e-05, "loss": 0.2406, "step": 13970 }, { "epoch": 135.4734939759036, "grad_norm": 14.302626609802246, "learning_rate": 3.217961165048543e-05, "loss": 0.378, "step": 13971 }, { "epoch": 135.48313253012049, "grad_norm": 5.434226036071777, "learning_rate": 3.21747572815534e-05, "loss": 0.098, "step": 13972 }, { "epoch": 135.49277108433736, "grad_norm": 10.696218490600586, "learning_rate": 3.2169902912621355e-05, "loss": 0.397, "step": 13973 }, { "epoch": 135.5024096385542, "grad_norm": 6.919604778289795, "learning_rate": 3.2165048543689326e-05, "loss": 0.0934, "step": 13974 }, { "epoch": 135.5120481927711, "grad_norm": 34.52944564819336, "learning_rate": 3.2160194174757284e-05, "loss": 0.2888, "step": 13975 }, { "epoch": 135.52168674698794, "grad_norm": 18.518009185791016, "learning_rate": 3.215533980582525e-05, "loss": 0.3417, "step": 13976 }, { "epoch": 135.53132530120482, "grad_norm": 8.685145378112793, "learning_rate": 3.2150485436893206e-05, "loss": 0.1798, "step": 13977 }, { "epoch": 135.5409638554217, "grad_norm": 10.827672958374023, "learning_rate": 3.214563106796117e-05, "loss": 0.2389, "step": 13978 }, { "epoch": 135.55060240963854, "grad_norm": 14.97678279876709, "learning_rate": 3.214077669902913e-05, "loss": 0.3467, "step": 13979 }, { "epoch": 135.56024096385542, "grad_norm": 16.593420028686523, "learning_rate": 3.213592233009709e-05, "loss": 0.2405, "step": 13980 }, { "epoch": 135.5698795180723, "grad_norm": 19.715665817260742, "learning_rate": 3.213106796116505e-05, "loss": 0.2593, "step": 13981 }, { "epoch": 135.57951807228915, "grad_norm": 13.730531692504883, "learning_rate": 3.212621359223301e-05, "loss": 0.3066, "step": 13982 }, { "epoch": 135.58915662650602, "grad_norm": 8.737629890441895, "learning_rate": 3.212135922330097e-05, "loss": 0.385, "step": 13983 }, { "epoch": 135.5987951807229, "grad_norm": 8.934087753295898, "learning_rate": 3.211650485436893e-05, "loss": 0.327, "step": 13984 }, { "epoch": 135.60843373493975, "grad_norm": 35.50057601928711, "learning_rate": 3.2111650485436896e-05, "loss": 0.3587, "step": 13985 }, { "epoch": 135.61807228915663, "grad_norm": 9.616994857788086, "learning_rate": 3.210679611650485e-05, "loss": 0.3825, "step": 13986 }, { "epoch": 135.6277108433735, "grad_norm": 19.40028190612793, "learning_rate": 3.210194174757282e-05, "loss": 0.077, "step": 13987 }, { "epoch": 135.63734939759036, "grad_norm": 4.754098415374756, "learning_rate": 3.2097087378640776e-05, "loss": 0.1433, "step": 13988 }, { "epoch": 135.64698795180723, "grad_norm": 20.104122161865234, "learning_rate": 3.209223300970874e-05, "loss": 0.484, "step": 13989 }, { "epoch": 135.65662650602408, "grad_norm": 6.320792198181152, "learning_rate": 3.20873786407767e-05, "loss": 0.2873, "step": 13990 }, { "epoch": 135.66626506024096, "grad_norm": 26.373567581176758, "learning_rate": 3.208252427184466e-05, "loss": 0.2439, "step": 13991 }, { "epoch": 135.67590361445784, "grad_norm": 12.413901329040527, "learning_rate": 3.207766990291262e-05, "loss": 0.3395, "step": 13992 }, { "epoch": 135.6855421686747, "grad_norm": 15.357332229614258, "learning_rate": 3.2072815533980585e-05, "loss": 0.2548, "step": 13993 }, { "epoch": 135.69518072289156, "grad_norm": 57.505149841308594, "learning_rate": 3.206796116504854e-05, "loss": 0.2953, "step": 13994 }, { "epoch": 135.70481927710844, "grad_norm": 23.756763458251953, "learning_rate": 3.206310679611651e-05, "loss": 0.4059, "step": 13995 }, { "epoch": 135.7144578313253, "grad_norm": 20.346206665039062, "learning_rate": 3.2058252427184465e-05, "loss": 0.1768, "step": 13996 }, { "epoch": 135.72409638554217, "grad_norm": 40.49728012084961, "learning_rate": 3.205339805825243e-05, "loss": 0.1756, "step": 13997 }, { "epoch": 135.73373493975905, "grad_norm": 10.14546012878418, "learning_rate": 3.204854368932039e-05, "loss": 0.15, "step": 13998 }, { "epoch": 135.7433734939759, "grad_norm": 7.860157489776611, "learning_rate": 3.204368932038835e-05, "loss": 0.1709, "step": 13999 }, { "epoch": 135.75301204819277, "grad_norm": 55.45783233642578, "learning_rate": 3.2038834951456316e-05, "loss": 0.3925, "step": 14000 }, { "epoch": 135.76265060240965, "grad_norm": 22.545499801635742, "learning_rate": 3.2033980582524274e-05, "loss": 0.3627, "step": 14001 }, { "epoch": 135.7722891566265, "grad_norm": 27.523775100708008, "learning_rate": 3.202912621359224e-05, "loss": 0.162, "step": 14002 }, { "epoch": 135.78192771084338, "grad_norm": 17.386919021606445, "learning_rate": 3.2024271844660196e-05, "loss": 0.4572, "step": 14003 }, { "epoch": 135.79156626506025, "grad_norm": 7.35521936416626, "learning_rate": 3.201941747572816e-05, "loss": 0.2301, "step": 14004 }, { "epoch": 135.8012048192771, "grad_norm": 9.651665687561035, "learning_rate": 3.201456310679612e-05, "loss": 0.2814, "step": 14005 }, { "epoch": 135.81084337349398, "grad_norm": 23.865888595581055, "learning_rate": 3.200970873786408e-05, "loss": 0.1948, "step": 14006 }, { "epoch": 135.82048192771083, "grad_norm": 9.014381408691406, "learning_rate": 3.200485436893204e-05, "loss": 0.2045, "step": 14007 }, { "epoch": 135.8301204819277, "grad_norm": 8.93531608581543, "learning_rate": 3.2000000000000005e-05, "loss": 0.4722, "step": 14008 }, { "epoch": 135.83975903614459, "grad_norm": 8.849663734436035, "learning_rate": 3.199514563106796e-05, "loss": 0.2932, "step": 14009 }, { "epoch": 135.84939759036143, "grad_norm": 3.843111276626587, "learning_rate": 3.199029126213592e-05, "loss": 0.1126, "step": 14010 }, { "epoch": 135.8590361445783, "grad_norm": 6.373157978057861, "learning_rate": 3.1985436893203885e-05, "loss": 0.1741, "step": 14011 }, { "epoch": 135.8686746987952, "grad_norm": 5.514218330383301, "learning_rate": 3.198058252427184e-05, "loss": 0.1835, "step": 14012 }, { "epoch": 135.87831325301204, "grad_norm": 9.097932815551758, "learning_rate": 3.197572815533981e-05, "loss": 0.4489, "step": 14013 }, { "epoch": 135.88795180722892, "grad_norm": 21.1464786529541, "learning_rate": 3.1970873786407765e-05, "loss": 0.231, "step": 14014 }, { "epoch": 135.8975903614458, "grad_norm": 9.361170768737793, "learning_rate": 3.196601941747573e-05, "loss": 0.2364, "step": 14015 }, { "epoch": 135.90722891566264, "grad_norm": 23.581342697143555, "learning_rate": 3.196116504854369e-05, "loss": 0.142, "step": 14016 }, { "epoch": 135.91686746987952, "grad_norm": 34.34714126586914, "learning_rate": 3.195631067961165e-05, "loss": 0.3231, "step": 14017 }, { "epoch": 135.9265060240964, "grad_norm": 11.42997932434082, "learning_rate": 3.195145631067961e-05, "loss": 0.3753, "step": 14018 }, { "epoch": 135.93614457831325, "grad_norm": 8.098250389099121, "learning_rate": 3.1946601941747575e-05, "loss": 0.2364, "step": 14019 }, { "epoch": 135.94578313253012, "grad_norm": 9.123903274536133, "learning_rate": 3.194174757281553e-05, "loss": 0.3113, "step": 14020 }, { "epoch": 135.955421686747, "grad_norm": 9.137889862060547, "learning_rate": 3.19368932038835e-05, "loss": 0.2317, "step": 14021 }, { "epoch": 135.96506024096385, "grad_norm": 10.091235160827637, "learning_rate": 3.1932038834951455e-05, "loss": 0.2074, "step": 14022 }, { "epoch": 135.97469879518073, "grad_norm": 8.867045402526855, "learning_rate": 3.192718446601942e-05, "loss": 0.3003, "step": 14023 }, { "epoch": 135.98433734939758, "grad_norm": 6.074838638305664, "learning_rate": 3.1922330097087384e-05, "loss": 0.1432, "step": 14024 }, { "epoch": 135.99397590361446, "grad_norm": 9.158530235290527, "learning_rate": 3.191747572815534e-05, "loss": 0.1454, "step": 14025 }, { "epoch": 136.00963855421688, "grad_norm": 7.951092720031738, "learning_rate": 3.1912621359223306e-05, "loss": 0.0969, "step": 14026 }, { "epoch": 136.01927710843373, "grad_norm": 6.464725971221924, "learning_rate": 3.1907766990291264e-05, "loss": 0.1753, "step": 14027 }, { "epoch": 136.0289156626506, "grad_norm": 14.4631929397583, "learning_rate": 3.190291262135923e-05, "loss": 0.2742, "step": 14028 }, { "epoch": 136.03855421686748, "grad_norm": 9.04403305053711, "learning_rate": 3.1898058252427186e-05, "loss": 0.2438, "step": 14029 }, { "epoch": 136.04819277108433, "grad_norm": 9.30322265625, "learning_rate": 3.189320388349515e-05, "loss": 0.266, "step": 14030 }, { "epoch": 136.0578313253012, "grad_norm": 8.770872116088867, "learning_rate": 3.188834951456311e-05, "loss": 0.3019, "step": 14031 }, { "epoch": 136.0674698795181, "grad_norm": 14.159990310668945, "learning_rate": 3.188349514563107e-05, "loss": 0.1662, "step": 14032 }, { "epoch": 136.07710843373494, "grad_norm": 16.987150192260742, "learning_rate": 3.187864077669903e-05, "loss": 0.4238, "step": 14033 }, { "epoch": 136.0867469879518, "grad_norm": 18.729188919067383, "learning_rate": 3.1873786407766995e-05, "loss": 0.3639, "step": 14034 }, { "epoch": 136.09638554216866, "grad_norm": 7.94149112701416, "learning_rate": 3.186893203883495e-05, "loss": 0.3097, "step": 14035 }, { "epoch": 136.10602409638554, "grad_norm": 15.840968132019043, "learning_rate": 3.186407766990292e-05, "loss": 0.3408, "step": 14036 }, { "epoch": 136.11566265060242, "grad_norm": 9.717700004577637, "learning_rate": 3.1859223300970875e-05, "loss": 0.2548, "step": 14037 }, { "epoch": 136.12530120481927, "grad_norm": 9.23764419555664, "learning_rate": 3.185436893203883e-05, "loss": 0.3324, "step": 14038 }, { "epoch": 136.13493975903614, "grad_norm": 6.771967887878418, "learning_rate": 3.18495145631068e-05, "loss": 0.2561, "step": 14039 }, { "epoch": 136.14457831325302, "grad_norm": 32.69035339355469, "learning_rate": 3.1844660194174755e-05, "loss": 0.2557, "step": 14040 }, { "epoch": 136.15421686746987, "grad_norm": 15.19310188293457, "learning_rate": 3.183980582524272e-05, "loss": 0.1743, "step": 14041 }, { "epoch": 136.16385542168675, "grad_norm": 5.431326866149902, "learning_rate": 3.183495145631068e-05, "loss": 0.2446, "step": 14042 }, { "epoch": 136.17349397590363, "grad_norm": 19.212175369262695, "learning_rate": 3.183009708737864e-05, "loss": 0.3619, "step": 14043 }, { "epoch": 136.18313253012047, "grad_norm": 21.99773597717285, "learning_rate": 3.18252427184466e-05, "loss": 0.3621, "step": 14044 }, { "epoch": 136.19277108433735, "grad_norm": 11.471305847167969, "learning_rate": 3.1820388349514564e-05, "loss": 0.2041, "step": 14045 }, { "epoch": 136.20240963855423, "grad_norm": 25.357065200805664, "learning_rate": 3.181553398058252e-05, "loss": 0.419, "step": 14046 }, { "epoch": 136.21204819277108, "grad_norm": 13.670795440673828, "learning_rate": 3.181067961165049e-05, "loss": 0.3616, "step": 14047 }, { "epoch": 136.22168674698796, "grad_norm": 9.434911727905273, "learning_rate": 3.1805825242718444e-05, "loss": 0.2274, "step": 14048 }, { "epoch": 136.2313253012048, "grad_norm": 17.736120223999023, "learning_rate": 3.180097087378641e-05, "loss": 0.2936, "step": 14049 }, { "epoch": 136.24096385542168, "grad_norm": 6.656794548034668, "learning_rate": 3.1796116504854373e-05, "loss": 0.1986, "step": 14050 }, { "epoch": 136.25060240963856, "grad_norm": 9.408454895019531, "learning_rate": 3.179126213592233e-05, "loss": 0.2688, "step": 14051 }, { "epoch": 136.2602409638554, "grad_norm": 30.436880111694336, "learning_rate": 3.1786407766990296e-05, "loss": 0.4216, "step": 14052 }, { "epoch": 136.2698795180723, "grad_norm": 30.086286544799805, "learning_rate": 3.1781553398058253e-05, "loss": 0.2817, "step": 14053 }, { "epoch": 136.27951807228916, "grad_norm": 7.8614912033081055, "learning_rate": 3.177669902912622e-05, "loss": 0.193, "step": 14054 }, { "epoch": 136.289156626506, "grad_norm": 23.380413055419922, "learning_rate": 3.1771844660194176e-05, "loss": 0.279, "step": 14055 }, { "epoch": 136.2987951807229, "grad_norm": 8.79722785949707, "learning_rate": 3.176699029126214e-05, "loss": 0.2422, "step": 14056 }, { "epoch": 136.30843373493977, "grad_norm": 8.54921817779541, "learning_rate": 3.17621359223301e-05, "loss": 0.2372, "step": 14057 }, { "epoch": 136.31807228915662, "grad_norm": 13.808377265930176, "learning_rate": 3.175728155339806e-05, "loss": 0.1913, "step": 14058 }, { "epoch": 136.3277108433735, "grad_norm": 8.260148048400879, "learning_rate": 3.175242718446602e-05, "loss": 0.1736, "step": 14059 }, { "epoch": 136.33734939759037, "grad_norm": 8.009198188781738, "learning_rate": 3.1747572815533985e-05, "loss": 0.2945, "step": 14060 }, { "epoch": 136.34698795180722, "grad_norm": 6.7906107902526855, "learning_rate": 3.174271844660194e-05, "loss": 0.2555, "step": 14061 }, { "epoch": 136.3566265060241, "grad_norm": 14.196502685546875, "learning_rate": 3.173786407766991e-05, "loss": 0.2152, "step": 14062 }, { "epoch": 136.36626506024098, "grad_norm": 7.329922676086426, "learning_rate": 3.1733009708737865e-05, "loss": 0.0966, "step": 14063 }, { "epoch": 136.37590361445783, "grad_norm": 8.840312004089355, "learning_rate": 3.172815533980583e-05, "loss": 0.3035, "step": 14064 }, { "epoch": 136.3855421686747, "grad_norm": 17.484771728515625, "learning_rate": 3.172330097087379e-05, "loss": 0.5706, "step": 14065 }, { "epoch": 136.39518072289155, "grad_norm": 8.277290344238281, "learning_rate": 3.1718446601941745e-05, "loss": 0.2711, "step": 14066 }, { "epoch": 136.40481927710843, "grad_norm": 5.9388933181762695, "learning_rate": 3.171359223300971e-05, "loss": 0.2142, "step": 14067 }, { "epoch": 136.4144578313253, "grad_norm": 11.508228302001953, "learning_rate": 3.170873786407767e-05, "loss": 0.2268, "step": 14068 }, { "epoch": 136.42409638554216, "grad_norm": 44.44853210449219, "learning_rate": 3.170388349514563e-05, "loss": 0.1345, "step": 14069 }, { "epoch": 136.43373493975903, "grad_norm": 6.600715160369873, "learning_rate": 3.169902912621359e-05, "loss": 0.2265, "step": 14070 }, { "epoch": 136.4433734939759, "grad_norm": 10.068482398986816, "learning_rate": 3.1694174757281554e-05, "loss": 0.1457, "step": 14071 }, { "epoch": 136.45301204819276, "grad_norm": 12.50975513458252, "learning_rate": 3.168932038834951e-05, "loss": 0.277, "step": 14072 }, { "epoch": 136.46265060240964, "grad_norm": 15.424742698669434, "learning_rate": 3.1684466019417476e-05, "loss": 0.3533, "step": 14073 }, { "epoch": 136.47228915662652, "grad_norm": 16.527664184570312, "learning_rate": 3.1679611650485434e-05, "loss": 0.2912, "step": 14074 }, { "epoch": 136.48192771084337, "grad_norm": 6.975654602050781, "learning_rate": 3.1674757281553405e-05, "loss": 0.1868, "step": 14075 }, { "epoch": 136.49156626506024, "grad_norm": 6.901871204376221, "learning_rate": 3.166990291262136e-05, "loss": 0.2009, "step": 14076 }, { "epoch": 136.50120481927712, "grad_norm": 9.36000919342041, "learning_rate": 3.166504854368932e-05, "loss": 0.3201, "step": 14077 }, { "epoch": 136.51084337349397, "grad_norm": 11.422310829162598, "learning_rate": 3.1660194174757286e-05, "loss": 0.2113, "step": 14078 }, { "epoch": 136.52048192771085, "grad_norm": 29.283283233642578, "learning_rate": 3.165533980582524e-05, "loss": 0.1664, "step": 14079 }, { "epoch": 136.53012048192772, "grad_norm": 14.965311050415039, "learning_rate": 3.165048543689321e-05, "loss": 0.3104, "step": 14080 }, { "epoch": 136.53975903614457, "grad_norm": 23.54930877685547, "learning_rate": 3.1645631067961166e-05, "loss": 0.3787, "step": 14081 }, { "epoch": 136.54939759036145, "grad_norm": 15.317506790161133, "learning_rate": 3.164077669902913e-05, "loss": 0.2184, "step": 14082 }, { "epoch": 136.5590361445783, "grad_norm": 9.790881156921387, "learning_rate": 3.163592233009709e-05, "loss": 0.3288, "step": 14083 }, { "epoch": 136.56867469879518, "grad_norm": 8.912912368774414, "learning_rate": 3.163106796116505e-05, "loss": 0.2775, "step": 14084 }, { "epoch": 136.57831325301206, "grad_norm": 25.651134490966797, "learning_rate": 3.162621359223301e-05, "loss": 0.3005, "step": 14085 }, { "epoch": 136.5879518072289, "grad_norm": 3.905816078186035, "learning_rate": 3.1621359223300975e-05, "loss": 0.0956, "step": 14086 }, { "epoch": 136.59759036144578, "grad_norm": 6.047783851623535, "learning_rate": 3.161650485436893e-05, "loss": 0.1837, "step": 14087 }, { "epoch": 136.60722891566266, "grad_norm": 10.120538711547852, "learning_rate": 3.16116504854369e-05, "loss": 0.3542, "step": 14088 }, { "epoch": 136.6168674698795, "grad_norm": 13.513354301452637, "learning_rate": 3.1606796116504855e-05, "loss": 0.476, "step": 14089 }, { "epoch": 136.6265060240964, "grad_norm": 16.2481746673584, "learning_rate": 3.160194174757282e-05, "loss": 0.3982, "step": 14090 }, { "epoch": 136.63614457831326, "grad_norm": 6.857170104980469, "learning_rate": 3.159708737864078e-05, "loss": 0.2047, "step": 14091 }, { "epoch": 136.6457831325301, "grad_norm": 31.342058181762695, "learning_rate": 3.159223300970874e-05, "loss": 0.1981, "step": 14092 }, { "epoch": 136.655421686747, "grad_norm": 7.919424533843994, "learning_rate": 3.15873786407767e-05, "loss": 0.2459, "step": 14093 }, { "epoch": 136.66506024096387, "grad_norm": 32.20677947998047, "learning_rate": 3.158252427184466e-05, "loss": 0.1689, "step": 14094 }, { "epoch": 136.67469879518072, "grad_norm": 5.196857452392578, "learning_rate": 3.157766990291262e-05, "loss": 0.0714, "step": 14095 }, { "epoch": 136.6843373493976, "grad_norm": 8.016406059265137, "learning_rate": 3.157281553398058e-05, "loss": 0.2662, "step": 14096 }, { "epoch": 136.69397590361444, "grad_norm": 14.028977394104004, "learning_rate": 3.1567961165048544e-05, "loss": 0.3734, "step": 14097 }, { "epoch": 136.70361445783132, "grad_norm": 7.425695896148682, "learning_rate": 3.15631067961165e-05, "loss": 0.3207, "step": 14098 }, { "epoch": 136.7132530120482, "grad_norm": 6.109051704406738, "learning_rate": 3.1558252427184466e-05, "loss": 0.1795, "step": 14099 }, { "epoch": 136.72289156626505, "grad_norm": 14.499951362609863, "learning_rate": 3.155339805825243e-05, "loss": 0.2521, "step": 14100 }, { "epoch": 136.73253012048193, "grad_norm": 15.626562118530273, "learning_rate": 3.1548543689320395e-05, "loss": 0.2236, "step": 14101 }, { "epoch": 136.7421686746988, "grad_norm": 13.311361312866211, "learning_rate": 3.154368932038835e-05, "loss": 0.4782, "step": 14102 }, { "epoch": 136.75180722891565, "grad_norm": 6.6439690589904785, "learning_rate": 3.153883495145631e-05, "loss": 0.1706, "step": 14103 }, { "epoch": 136.76144578313253, "grad_norm": 10.866474151611328, "learning_rate": 3.1533980582524275e-05, "loss": 0.4531, "step": 14104 }, { "epoch": 136.7710843373494, "grad_norm": 13.18845272064209, "learning_rate": 3.152912621359223e-05, "loss": 0.2934, "step": 14105 }, { "epoch": 136.78072289156626, "grad_norm": 13.691428184509277, "learning_rate": 3.15242718446602e-05, "loss": 0.3366, "step": 14106 }, { "epoch": 136.79036144578313, "grad_norm": 21.04502296447754, "learning_rate": 3.1519417475728155e-05, "loss": 0.3065, "step": 14107 }, { "epoch": 136.8, "grad_norm": 4.733554840087891, "learning_rate": 3.151456310679612e-05, "loss": 0.2194, "step": 14108 }, { "epoch": 136.80963855421686, "grad_norm": 5.928338527679443, "learning_rate": 3.150970873786408e-05, "loss": 0.1844, "step": 14109 }, { "epoch": 136.81927710843374, "grad_norm": 16.90734100341797, "learning_rate": 3.150485436893204e-05, "loss": 0.2038, "step": 14110 }, { "epoch": 136.82891566265062, "grad_norm": 13.757559776306152, "learning_rate": 3.15e-05, "loss": 0.5134, "step": 14111 }, { "epoch": 136.83855421686746, "grad_norm": 15.27991771697998, "learning_rate": 3.1495145631067964e-05, "loss": 0.2821, "step": 14112 }, { "epoch": 136.84819277108434, "grad_norm": 11.567431449890137, "learning_rate": 3.149029126213592e-05, "loss": 0.2528, "step": 14113 }, { "epoch": 136.8578313253012, "grad_norm": 13.967081069946289, "learning_rate": 3.148543689320389e-05, "loss": 0.241, "step": 14114 }, { "epoch": 136.86746987951807, "grad_norm": 19.830810546875, "learning_rate": 3.1480582524271845e-05, "loss": 0.2487, "step": 14115 }, { "epoch": 136.87710843373495, "grad_norm": 9.8163423538208, "learning_rate": 3.147572815533981e-05, "loss": 0.213, "step": 14116 }, { "epoch": 136.8867469879518, "grad_norm": 10.013928413391113, "learning_rate": 3.147087378640777e-05, "loss": 0.1842, "step": 14117 }, { "epoch": 136.89638554216867, "grad_norm": 13.256781578063965, "learning_rate": 3.146601941747573e-05, "loss": 0.2927, "step": 14118 }, { "epoch": 136.90602409638555, "grad_norm": 9.637051582336426, "learning_rate": 3.146116504854369e-05, "loss": 0.1836, "step": 14119 }, { "epoch": 136.9156626506024, "grad_norm": 8.473210334777832, "learning_rate": 3.145631067961165e-05, "loss": 0.3347, "step": 14120 }, { "epoch": 136.92530120481928, "grad_norm": 10.903780937194824, "learning_rate": 3.145145631067961e-05, "loss": 0.3486, "step": 14121 }, { "epoch": 136.93493975903615, "grad_norm": 6.074684143066406, "learning_rate": 3.144660194174757e-05, "loss": 0.1233, "step": 14122 }, { "epoch": 136.944578313253, "grad_norm": 23.1123046875, "learning_rate": 3.1441747572815534e-05, "loss": 0.2392, "step": 14123 }, { "epoch": 136.95421686746988, "grad_norm": 5.71067476272583, "learning_rate": 3.143689320388349e-05, "loss": 0.1932, "step": 14124 }, { "epoch": 136.96385542168676, "grad_norm": 11.41590404510498, "learning_rate": 3.143203883495146e-05, "loss": 0.1538, "step": 14125 }, { "epoch": 136.9734939759036, "grad_norm": 7.961533546447754, "learning_rate": 3.142718446601942e-05, "loss": 0.2038, "step": 14126 }, { "epoch": 136.98313253012049, "grad_norm": 24.678606033325195, "learning_rate": 3.1422330097087385e-05, "loss": 0.3592, "step": 14127 }, { "epoch": 136.99277108433736, "grad_norm": 11.33835506439209, "learning_rate": 3.141747572815534e-05, "loss": 0.275, "step": 14128 }, { "epoch": 137.00843373493976, "grad_norm": 21.924497604370117, "learning_rate": 3.141262135922331e-05, "loss": 0.4561, "step": 14129 }, { "epoch": 137.01807228915663, "grad_norm": 4.759283542633057, "learning_rate": 3.1407766990291265e-05, "loss": 0.307, "step": 14130 }, { "epoch": 137.02771084337348, "grad_norm": 10.540589332580566, "learning_rate": 3.140291262135922e-05, "loss": 0.3503, "step": 14131 }, { "epoch": 137.03734939759036, "grad_norm": 8.538447380065918, "learning_rate": 3.139805825242719e-05, "loss": 0.2681, "step": 14132 }, { "epoch": 137.04698795180724, "grad_norm": 14.75898551940918, "learning_rate": 3.1393203883495145e-05, "loss": 0.2401, "step": 14133 }, { "epoch": 137.0566265060241, "grad_norm": 3.1720130443573, "learning_rate": 3.138834951456311e-05, "loss": 0.3862, "step": 14134 }, { "epoch": 137.06626506024097, "grad_norm": 5.782515525817871, "learning_rate": 3.138349514563107e-05, "loss": 0.3195, "step": 14135 }, { "epoch": 137.07590361445784, "grad_norm": 58.84169006347656, "learning_rate": 3.137864077669903e-05, "loss": 0.2194, "step": 14136 }, { "epoch": 137.0855421686747, "grad_norm": 108.98571014404297, "learning_rate": 3.137378640776699e-05, "loss": 0.4173, "step": 14137 }, { "epoch": 137.09518072289157, "grad_norm": 65.83109283447266, "learning_rate": 3.1368932038834954e-05, "loss": 0.3949, "step": 14138 }, { "epoch": 137.10481927710845, "grad_norm": 5.829030990600586, "learning_rate": 3.136407766990291e-05, "loss": 0.3, "step": 14139 }, { "epoch": 137.1144578313253, "grad_norm": 29.440879821777344, "learning_rate": 3.1359223300970877e-05, "loss": 0.217, "step": 14140 }, { "epoch": 137.12409638554217, "grad_norm": 205.42385864257812, "learning_rate": 3.1354368932038834e-05, "loss": 0.1389, "step": 14141 }, { "epoch": 137.13373493975902, "grad_norm": 22.300504684448242, "learning_rate": 3.13495145631068e-05, "loss": 0.3147, "step": 14142 }, { "epoch": 137.1433734939759, "grad_norm": 55.9588508605957, "learning_rate": 3.134466019417476e-05, "loss": 0.4084, "step": 14143 }, { "epoch": 137.15301204819278, "grad_norm": 19.79173469543457, "learning_rate": 3.133980582524272e-05, "loss": 0.4058, "step": 14144 }, { "epoch": 137.16265060240963, "grad_norm": 18.358905792236328, "learning_rate": 3.133495145631068e-05, "loss": 0.3065, "step": 14145 }, { "epoch": 137.1722891566265, "grad_norm": 23.869932174682617, "learning_rate": 3.1330097087378643e-05, "loss": 0.4302, "step": 14146 }, { "epoch": 137.18192771084338, "grad_norm": 11.386300086975098, "learning_rate": 3.13252427184466e-05, "loss": 0.2148, "step": 14147 }, { "epoch": 137.19156626506023, "grad_norm": 11.388569831848145, "learning_rate": 3.132038834951456e-05, "loss": 0.347, "step": 14148 }, { "epoch": 137.2012048192771, "grad_norm": 15.78199577331543, "learning_rate": 3.1315533980582523e-05, "loss": 0.2382, "step": 14149 }, { "epoch": 137.210843373494, "grad_norm": 5.209604740142822, "learning_rate": 3.131067961165049e-05, "loss": 0.1266, "step": 14150 }, { "epoch": 137.22048192771084, "grad_norm": 41.02534484863281, "learning_rate": 3.130582524271845e-05, "loss": 0.22, "step": 14151 }, { "epoch": 137.2301204819277, "grad_norm": 25.740222930908203, "learning_rate": 3.130097087378641e-05, "loss": 0.3101, "step": 14152 }, { "epoch": 137.2397590361446, "grad_norm": 4.776670932769775, "learning_rate": 3.1296116504854375e-05, "loss": 0.2382, "step": 14153 }, { "epoch": 137.24939759036144, "grad_norm": 14.16864013671875, "learning_rate": 3.129126213592233e-05, "loss": 0.2887, "step": 14154 }, { "epoch": 137.25903614457832, "grad_norm": 6.375239372253418, "learning_rate": 3.12864077669903e-05, "loss": 0.3281, "step": 14155 }, { "epoch": 137.2686746987952, "grad_norm": 13.303893089294434, "learning_rate": 3.1281553398058255e-05, "loss": 0.4425, "step": 14156 }, { "epoch": 137.27831325301204, "grad_norm": 11.524847030639648, "learning_rate": 3.127669902912622e-05, "loss": 0.2063, "step": 14157 }, { "epoch": 137.28795180722892, "grad_norm": 19.286983489990234, "learning_rate": 3.127184466019418e-05, "loss": 0.2696, "step": 14158 }, { "epoch": 137.29759036144577, "grad_norm": 27.66741943359375, "learning_rate": 3.1266990291262135e-05, "loss": 0.4296, "step": 14159 }, { "epoch": 137.30722891566265, "grad_norm": 10.338199615478516, "learning_rate": 3.12621359223301e-05, "loss": 0.1625, "step": 14160 }, { "epoch": 137.31686746987953, "grad_norm": 34.515342712402344, "learning_rate": 3.125728155339806e-05, "loss": 0.3506, "step": 14161 }, { "epoch": 137.32650602409637, "grad_norm": 3.8919665813446045, "learning_rate": 3.125242718446602e-05, "loss": 0.1808, "step": 14162 }, { "epoch": 137.33614457831325, "grad_norm": 67.33869934082031, "learning_rate": 3.124757281553398e-05, "loss": 0.4292, "step": 14163 }, { "epoch": 137.34578313253013, "grad_norm": 10.776440620422363, "learning_rate": 3.1242718446601944e-05, "loss": 0.1352, "step": 14164 }, { "epoch": 137.35542168674698, "grad_norm": 31.247791290283203, "learning_rate": 3.12378640776699e-05, "loss": 0.393, "step": 14165 }, { "epoch": 137.36506024096386, "grad_norm": 2.341949462890625, "learning_rate": 3.1233009708737866e-05, "loss": 0.227, "step": 14166 }, { "epoch": 137.37469879518073, "grad_norm": 42.838565826416016, "learning_rate": 3.1228155339805824e-05, "loss": 0.295, "step": 14167 }, { "epoch": 137.38433734939758, "grad_norm": 18.559139251708984, "learning_rate": 3.122330097087379e-05, "loss": 0.1459, "step": 14168 }, { "epoch": 137.39397590361446, "grad_norm": 31.914743423461914, "learning_rate": 3.1218446601941746e-05, "loss": 0.3774, "step": 14169 }, { "epoch": 137.40361445783134, "grad_norm": 2.440028190612793, "learning_rate": 3.121359223300971e-05, "loss": 0.2293, "step": 14170 }, { "epoch": 137.4132530120482, "grad_norm": 2.106637477874756, "learning_rate": 3.120873786407767e-05, "loss": 0.1385, "step": 14171 }, { "epoch": 137.42289156626506, "grad_norm": 12.902379989624023, "learning_rate": 3.120388349514563e-05, "loss": 0.2734, "step": 14172 }, { "epoch": 137.4325301204819, "grad_norm": 17.377662658691406, "learning_rate": 3.119902912621359e-05, "loss": 0.5165, "step": 14173 }, { "epoch": 137.4421686746988, "grad_norm": 50.65327453613281, "learning_rate": 3.1194174757281556e-05, "loss": 0.2935, "step": 14174 }, { "epoch": 137.45180722891567, "grad_norm": 2.3839006423950195, "learning_rate": 3.118932038834951e-05, "loss": 0.104, "step": 14175 }, { "epoch": 137.46144578313252, "grad_norm": 24.21894645690918, "learning_rate": 3.118446601941748e-05, "loss": 0.3835, "step": 14176 }, { "epoch": 137.4710843373494, "grad_norm": 8.782649993896484, "learning_rate": 3.117961165048544e-05, "loss": 0.2204, "step": 14177 }, { "epoch": 137.48072289156627, "grad_norm": 7.60492467880249, "learning_rate": 3.11747572815534e-05, "loss": 0.1214, "step": 14178 }, { "epoch": 137.49036144578312, "grad_norm": 27.726707458496094, "learning_rate": 3.1169902912621365e-05, "loss": 0.4155, "step": 14179 }, { "epoch": 137.5, "grad_norm": 25.881603240966797, "learning_rate": 3.116504854368932e-05, "loss": 0.2575, "step": 14180 }, { "epoch": 137.50963855421688, "grad_norm": 50.624576568603516, "learning_rate": 3.116019417475729e-05, "loss": 0.3936, "step": 14181 }, { "epoch": 137.51927710843373, "grad_norm": 8.043099403381348, "learning_rate": 3.1155339805825245e-05, "loss": 0.3736, "step": 14182 }, { "epoch": 137.5289156626506, "grad_norm": 20.14263153076172, "learning_rate": 3.115048543689321e-05, "loss": 0.2547, "step": 14183 }, { "epoch": 137.53855421686748, "grad_norm": 3.783414363861084, "learning_rate": 3.114563106796117e-05, "loss": 0.1076, "step": 14184 }, { "epoch": 137.54819277108433, "grad_norm": 62.07735061645508, "learning_rate": 3.114077669902913e-05, "loss": 0.3207, "step": 14185 }, { "epoch": 137.5578313253012, "grad_norm": 25.917325973510742, "learning_rate": 3.113592233009709e-05, "loss": 0.3643, "step": 14186 }, { "epoch": 137.5674698795181, "grad_norm": 8.723918914794922, "learning_rate": 3.113106796116505e-05, "loss": 0.2946, "step": 14187 }, { "epoch": 137.57710843373494, "grad_norm": 98.79393768310547, "learning_rate": 3.112621359223301e-05, "loss": 0.3613, "step": 14188 }, { "epoch": 137.5867469879518, "grad_norm": 41.71028518676758, "learning_rate": 3.112135922330097e-05, "loss": 0.1792, "step": 14189 }, { "epoch": 137.59638554216866, "grad_norm": 3.543881416320801, "learning_rate": 3.1116504854368934e-05, "loss": 0.227, "step": 14190 }, { "epoch": 137.60602409638554, "grad_norm": 2.7445034980773926, "learning_rate": 3.111165048543689e-05, "loss": 0.322, "step": 14191 }, { "epoch": 137.61566265060242, "grad_norm": 20.035926818847656, "learning_rate": 3.1106796116504856e-05, "loss": 0.2656, "step": 14192 }, { "epoch": 137.62530120481927, "grad_norm": 21.95245361328125, "learning_rate": 3.1101941747572814e-05, "loss": 0.1915, "step": 14193 }, { "epoch": 137.63493975903614, "grad_norm": 4.92338752746582, "learning_rate": 3.109708737864078e-05, "loss": 0.2769, "step": 14194 }, { "epoch": 137.64457831325302, "grad_norm": 40.181373596191406, "learning_rate": 3.1092233009708736e-05, "loss": 0.3622, "step": 14195 }, { "epoch": 137.65421686746987, "grad_norm": 11.438673973083496, "learning_rate": 3.10873786407767e-05, "loss": 0.3732, "step": 14196 }, { "epoch": 137.66385542168675, "grad_norm": 55.79188919067383, "learning_rate": 3.108252427184466e-05, "loss": 0.3184, "step": 14197 }, { "epoch": 137.67349397590363, "grad_norm": 7.732857704162598, "learning_rate": 3.107766990291262e-05, "loss": 0.3516, "step": 14198 }, { "epoch": 137.68313253012047, "grad_norm": 4.288959980010986, "learning_rate": 3.107281553398058e-05, "loss": 0.2254, "step": 14199 }, { "epoch": 137.69277108433735, "grad_norm": 80.99840545654297, "learning_rate": 3.1067961165048545e-05, "loss": 0.2525, "step": 14200 }, { "epoch": 137.70240963855423, "grad_norm": 18.02139663696289, "learning_rate": 3.106310679611651e-05, "loss": 0.2984, "step": 14201 }, { "epoch": 137.71204819277108, "grad_norm": 370.533447265625, "learning_rate": 3.105825242718447e-05, "loss": 0.3298, "step": 14202 }, { "epoch": 137.72168674698796, "grad_norm": 16.011735916137695, "learning_rate": 3.105339805825243e-05, "loss": 0.3565, "step": 14203 }, { "epoch": 137.73132530120483, "grad_norm": 4.000720977783203, "learning_rate": 3.104854368932039e-05, "loss": 0.1504, "step": 14204 }, { "epoch": 137.74096385542168, "grad_norm": 26.084440231323242, "learning_rate": 3.1043689320388354e-05, "loss": 0.3409, "step": 14205 }, { "epoch": 137.75060240963856, "grad_norm": 6.135622024536133, "learning_rate": 3.103883495145631e-05, "loss": 0.1379, "step": 14206 }, { "epoch": 137.7602409638554, "grad_norm": 41.11201477050781, "learning_rate": 3.103398058252428e-05, "loss": 0.1908, "step": 14207 }, { "epoch": 137.7698795180723, "grad_norm": 20.655303955078125, "learning_rate": 3.1029126213592234e-05, "loss": 0.2106, "step": 14208 }, { "epoch": 137.77951807228916, "grad_norm": 66.88492584228516, "learning_rate": 3.10242718446602e-05, "loss": 0.2923, "step": 14209 }, { "epoch": 137.789156626506, "grad_norm": 24.69613265991211, "learning_rate": 3.101941747572816e-05, "loss": 0.2005, "step": 14210 }, { "epoch": 137.7987951807229, "grad_norm": 22.989866256713867, "learning_rate": 3.101456310679612e-05, "loss": 0.1943, "step": 14211 }, { "epoch": 137.80843373493977, "grad_norm": 80.14226531982422, "learning_rate": 3.100970873786408e-05, "loss": 0.1927, "step": 14212 }, { "epoch": 137.81807228915662, "grad_norm": 9.023488998413086, "learning_rate": 3.1004854368932044e-05, "loss": 0.1474, "step": 14213 }, { "epoch": 137.8277108433735, "grad_norm": 5.086927890777588, "learning_rate": 3.1e-05, "loss": 0.2775, "step": 14214 }, { "epoch": 137.83734939759037, "grad_norm": 26.646724700927734, "learning_rate": 3.099514563106796e-05, "loss": 0.2636, "step": 14215 }, { "epoch": 137.84698795180722, "grad_norm": 46.312225341796875, "learning_rate": 3.0990291262135924e-05, "loss": 0.1792, "step": 14216 }, { "epoch": 137.8566265060241, "grad_norm": 3.5057058334350586, "learning_rate": 3.098543689320388e-05, "loss": 0.2968, "step": 14217 }, { "epoch": 137.86626506024098, "grad_norm": 22.992900848388672, "learning_rate": 3.0980582524271846e-05, "loss": 0.3872, "step": 14218 }, { "epoch": 137.87590361445783, "grad_norm": 14.6094388961792, "learning_rate": 3.0975728155339804e-05, "loss": 0.3945, "step": 14219 }, { "epoch": 137.8855421686747, "grad_norm": 1.504604697227478, "learning_rate": 3.097087378640777e-05, "loss": 0.2165, "step": 14220 }, { "epoch": 137.89518072289155, "grad_norm": 10.647858619689941, "learning_rate": 3.0966019417475726e-05, "loss": 0.3684, "step": 14221 }, { "epoch": 137.90481927710843, "grad_norm": 18.54288101196289, "learning_rate": 3.096116504854369e-05, "loss": 0.29, "step": 14222 }, { "epoch": 137.9144578313253, "grad_norm": 7.619987964630127, "learning_rate": 3.095631067961165e-05, "loss": 0.1868, "step": 14223 }, { "epoch": 137.92409638554216, "grad_norm": 16.11631202697754, "learning_rate": 3.095145631067961e-05, "loss": 0.3935, "step": 14224 }, { "epoch": 137.93373493975903, "grad_norm": 15.026836395263672, "learning_rate": 3.094660194174757e-05, "loss": 0.2845, "step": 14225 }, { "epoch": 137.9433734939759, "grad_norm": 18.674762725830078, "learning_rate": 3.0941747572815535e-05, "loss": 0.5113, "step": 14226 }, { "epoch": 137.95301204819276, "grad_norm": 12.701818466186523, "learning_rate": 3.09368932038835e-05, "loss": 0.3338, "step": 14227 }, { "epoch": 137.96265060240964, "grad_norm": 70.2710952758789, "learning_rate": 3.093203883495146e-05, "loss": 0.2756, "step": 14228 }, { "epoch": 137.97228915662652, "grad_norm": 14.461431503295898, "learning_rate": 3.092718446601942e-05, "loss": 0.194, "step": 14229 }, { "epoch": 137.98192771084337, "grad_norm": 5.094772815704346, "learning_rate": 3.092233009708738e-05, "loss": 0.4336, "step": 14230 }, { "epoch": 137.99156626506024, "grad_norm": 33.65760040283203, "learning_rate": 3.0917475728155344e-05, "loss": 0.2225, "step": 14231 }, { "epoch": 138.00722891566264, "grad_norm": 3.340616226196289, "learning_rate": 3.09126213592233e-05, "loss": 0.1171, "step": 14232 }, { "epoch": 138.01686746987951, "grad_norm": 10.910863876342773, "learning_rate": 3.0907766990291267e-05, "loss": 0.1902, "step": 14233 }, { "epoch": 138.0265060240964, "grad_norm": 11.879875183105469, "learning_rate": 3.0902912621359224e-05, "loss": 0.2191, "step": 14234 }, { "epoch": 138.03614457831324, "grad_norm": 10.708977699279785, "learning_rate": 3.089805825242719e-05, "loss": 0.1855, "step": 14235 }, { "epoch": 138.04578313253012, "grad_norm": 8.292436599731445, "learning_rate": 3.0893203883495147e-05, "loss": 0.2766, "step": 14236 }, { "epoch": 138.055421686747, "grad_norm": 29.898073196411133, "learning_rate": 3.088834951456311e-05, "loss": 0.1034, "step": 14237 }, { "epoch": 138.06506024096385, "grad_norm": 7.045713424682617, "learning_rate": 3.088349514563107e-05, "loss": 0.1413, "step": 14238 }, { "epoch": 138.07469879518072, "grad_norm": 9.986627578735352, "learning_rate": 3.087864077669903e-05, "loss": 0.2587, "step": 14239 }, { "epoch": 138.0843373493976, "grad_norm": 46.29998779296875, "learning_rate": 3.087378640776699e-05, "loss": 0.4888, "step": 14240 }, { "epoch": 138.09397590361445, "grad_norm": 27.861473083496094, "learning_rate": 3.0868932038834956e-05, "loss": 0.3097, "step": 14241 }, { "epoch": 138.10361445783133, "grad_norm": 3.644798755645752, "learning_rate": 3.0864077669902913e-05, "loss": 0.1957, "step": 14242 }, { "epoch": 138.1132530120482, "grad_norm": 25.169145584106445, "learning_rate": 3.085922330097087e-05, "loss": 0.2878, "step": 14243 }, { "epoch": 138.12289156626505, "grad_norm": 102.54017639160156, "learning_rate": 3.0854368932038836e-05, "loss": 0.2049, "step": 14244 }, { "epoch": 138.13253012048193, "grad_norm": 22.65525245666504, "learning_rate": 3.0849514563106793e-05, "loss": 0.2986, "step": 14245 }, { "epoch": 138.1421686746988, "grad_norm": 22.700523376464844, "learning_rate": 3.084466019417476e-05, "loss": 0.2719, "step": 14246 }, { "epoch": 138.15180722891566, "grad_norm": 4.340054512023926, "learning_rate": 3.0839805825242716e-05, "loss": 0.2555, "step": 14247 }, { "epoch": 138.16144578313254, "grad_norm": 32.638004302978516, "learning_rate": 3.083495145631068e-05, "loss": 0.3937, "step": 14248 }, { "epoch": 138.17108433734938, "grad_norm": 55.45148849487305, "learning_rate": 3.083009708737864e-05, "loss": 0.4457, "step": 14249 }, { "epoch": 138.18072289156626, "grad_norm": 15.947124481201172, "learning_rate": 3.08252427184466e-05, "loss": 0.2987, "step": 14250 }, { "epoch": 138.19036144578314, "grad_norm": 39.15653991699219, "learning_rate": 3.082038834951457e-05, "loss": 0.2277, "step": 14251 }, { "epoch": 138.2, "grad_norm": 7.512650966644287, "learning_rate": 3.0815533980582525e-05, "loss": 0.2263, "step": 14252 }, { "epoch": 138.20963855421687, "grad_norm": 5.8278632164001465, "learning_rate": 3.081067961165049e-05, "loss": 0.2218, "step": 14253 }, { "epoch": 138.21927710843374, "grad_norm": 15.463374137878418, "learning_rate": 3.080582524271845e-05, "loss": 0.3152, "step": 14254 }, { "epoch": 138.2289156626506, "grad_norm": 12.054211616516113, "learning_rate": 3.080097087378641e-05, "loss": 0.232, "step": 14255 }, { "epoch": 138.23855421686747, "grad_norm": 6.173369407653809, "learning_rate": 3.079611650485437e-05, "loss": 0.207, "step": 14256 }, { "epoch": 138.24819277108435, "grad_norm": 15.070467948913574, "learning_rate": 3.0791262135922334e-05, "loss": 0.2483, "step": 14257 }, { "epoch": 138.2578313253012, "grad_norm": 61.978843688964844, "learning_rate": 3.078640776699029e-05, "loss": 0.1873, "step": 14258 }, { "epoch": 138.26746987951807, "grad_norm": 15.731871604919434, "learning_rate": 3.0781553398058256e-05, "loss": 0.1247, "step": 14259 }, { "epoch": 138.27710843373495, "grad_norm": 29.2666072845459, "learning_rate": 3.0776699029126214e-05, "loss": 0.2487, "step": 14260 }, { "epoch": 138.2867469879518, "grad_norm": 57.493778228759766, "learning_rate": 3.077184466019418e-05, "loss": 0.2961, "step": 14261 }, { "epoch": 138.29638554216868, "grad_norm": 5.001396179199219, "learning_rate": 3.0766990291262136e-05, "loss": 0.2475, "step": 14262 }, { "epoch": 138.30602409638556, "grad_norm": 3.4609570503234863, "learning_rate": 3.07621359223301e-05, "loss": 0.3129, "step": 14263 }, { "epoch": 138.3156626506024, "grad_norm": 32.248104095458984, "learning_rate": 3.075728155339806e-05, "loss": 0.2466, "step": 14264 }, { "epoch": 138.32530120481928, "grad_norm": 15.290575981140137, "learning_rate": 3.075242718446602e-05, "loss": 0.2895, "step": 14265 }, { "epoch": 138.33493975903613, "grad_norm": 71.320556640625, "learning_rate": 3.074757281553398e-05, "loss": 0.3081, "step": 14266 }, { "epoch": 138.344578313253, "grad_norm": 3.391702651977539, "learning_rate": 3.0742718446601945e-05, "loss": 0.1226, "step": 14267 }, { "epoch": 138.3542168674699, "grad_norm": 8.23955249786377, "learning_rate": 3.07378640776699e-05, "loss": 0.2483, "step": 14268 }, { "epoch": 138.36385542168674, "grad_norm": 27.072507858276367, "learning_rate": 3.073300970873786e-05, "loss": 0.2936, "step": 14269 }, { "epoch": 138.3734939759036, "grad_norm": 61.05912780761719, "learning_rate": 3.0728155339805826e-05, "loss": 0.3643, "step": 14270 }, { "epoch": 138.3831325301205, "grad_norm": 31.400257110595703, "learning_rate": 3.072330097087378e-05, "loss": 0.2989, "step": 14271 }, { "epoch": 138.39277108433734, "grad_norm": 4.338119029998779, "learning_rate": 3.071844660194175e-05, "loss": 0.2278, "step": 14272 }, { "epoch": 138.40240963855422, "grad_norm": 16.232688903808594, "learning_rate": 3.0713592233009706e-05, "loss": 0.2681, "step": 14273 }, { "epoch": 138.4120481927711, "grad_norm": 35.69384002685547, "learning_rate": 3.070873786407767e-05, "loss": 0.2918, "step": 14274 }, { "epoch": 138.42168674698794, "grad_norm": 11.588090896606445, "learning_rate": 3.070388349514563e-05, "loss": 0.4689, "step": 14275 }, { "epoch": 138.43132530120482, "grad_norm": 21.8523006439209, "learning_rate": 3.069902912621359e-05, "loss": 0.3429, "step": 14276 }, { "epoch": 138.4409638554217, "grad_norm": 18.359304428100586, "learning_rate": 3.069417475728156e-05, "loss": 0.3357, "step": 14277 }, { "epoch": 138.45060240963855, "grad_norm": 30.434650421142578, "learning_rate": 3.068932038834952e-05, "loss": 0.4125, "step": 14278 }, { "epoch": 138.46024096385543, "grad_norm": 2.617056131362915, "learning_rate": 3.068446601941748e-05, "loss": 0.251, "step": 14279 }, { "epoch": 138.46987951807228, "grad_norm": 9.551302909851074, "learning_rate": 3.067961165048544e-05, "loss": 0.2162, "step": 14280 }, { "epoch": 138.47951807228915, "grad_norm": 8.434064865112305, "learning_rate": 3.06747572815534e-05, "loss": 0.107, "step": 14281 }, { "epoch": 138.48915662650603, "grad_norm": 7.626849174499512, "learning_rate": 3.066990291262136e-05, "loss": 0.2677, "step": 14282 }, { "epoch": 138.49879518072288, "grad_norm": 10.624517440795898, "learning_rate": 3.0665048543689324e-05, "loss": 0.1584, "step": 14283 }, { "epoch": 138.50843373493976, "grad_norm": 16.90264892578125, "learning_rate": 3.066019417475728e-05, "loss": 0.3575, "step": 14284 }, { "epoch": 138.51807228915663, "grad_norm": 4.709434509277344, "learning_rate": 3.0655339805825246e-05, "loss": 0.2143, "step": 14285 }, { "epoch": 138.52771084337348, "grad_norm": 19.503713607788086, "learning_rate": 3.0650485436893204e-05, "loss": 0.1913, "step": 14286 }, { "epoch": 138.53734939759036, "grad_norm": 32.294708251953125, "learning_rate": 3.064563106796117e-05, "loss": 0.2829, "step": 14287 }, { "epoch": 138.54698795180724, "grad_norm": 9.658363342285156, "learning_rate": 3.0640776699029126e-05, "loss": 0.3115, "step": 14288 }, { "epoch": 138.5566265060241, "grad_norm": 12.770373344421387, "learning_rate": 3.063592233009709e-05, "loss": 0.3136, "step": 14289 }, { "epoch": 138.56626506024097, "grad_norm": 29.694093704223633, "learning_rate": 3.063106796116505e-05, "loss": 0.4055, "step": 14290 }, { "epoch": 138.57590361445784, "grad_norm": 9.728772163391113, "learning_rate": 3.062621359223301e-05, "loss": 0.2592, "step": 14291 }, { "epoch": 138.5855421686747, "grad_norm": 17.551603317260742, "learning_rate": 3.062135922330097e-05, "loss": 0.2449, "step": 14292 }, { "epoch": 138.59518072289157, "grad_norm": 5.495551109313965, "learning_rate": 3.0616504854368935e-05, "loss": 0.2719, "step": 14293 }, { "epoch": 138.60481927710845, "grad_norm": 53.77186965942383, "learning_rate": 3.061165048543689e-05, "loss": 0.3814, "step": 14294 }, { "epoch": 138.6144578313253, "grad_norm": 3.359257936477661, "learning_rate": 3.060679611650486e-05, "loss": 0.3027, "step": 14295 }, { "epoch": 138.62409638554217, "grad_norm": 8.128496170043945, "learning_rate": 3.0601941747572815e-05, "loss": 0.2839, "step": 14296 }, { "epoch": 138.63373493975902, "grad_norm": 5.896812915802002, "learning_rate": 3.059708737864077e-05, "loss": 0.3714, "step": 14297 }, { "epoch": 138.6433734939759, "grad_norm": 24.63585090637207, "learning_rate": 3.059223300970874e-05, "loss": 0.2796, "step": 14298 }, { "epoch": 138.65301204819278, "grad_norm": 3.1639404296875, "learning_rate": 3.0587378640776695e-05, "loss": 0.3726, "step": 14299 }, { "epoch": 138.66265060240963, "grad_norm": 10.589858055114746, "learning_rate": 3.058252427184466e-05, "loss": 0.5193, "step": 14300 }, { "epoch": 138.6722891566265, "grad_norm": 14.190409660339355, "learning_rate": 3.057766990291262e-05, "loss": 0.3028, "step": 14301 }, { "epoch": 138.68192771084338, "grad_norm": 26.52685546875, "learning_rate": 3.057281553398059e-05, "loss": 0.2811, "step": 14302 }, { "epoch": 138.69156626506023, "grad_norm": 4.489255905151367, "learning_rate": 3.056796116504855e-05, "loss": 0.2869, "step": 14303 }, { "epoch": 138.7012048192771, "grad_norm": 8.211912155151367, "learning_rate": 3.056310679611651e-05, "loss": 0.155, "step": 14304 }, { "epoch": 138.710843373494, "grad_norm": 4.972629547119141, "learning_rate": 3.055825242718447e-05, "loss": 0.1415, "step": 14305 }, { "epoch": 138.72048192771084, "grad_norm": 7.295995712280273, "learning_rate": 3.0553398058252434e-05, "loss": 0.3213, "step": 14306 }, { "epoch": 138.7301204819277, "grad_norm": 6.3409271240234375, "learning_rate": 3.054854368932039e-05, "loss": 0.1078, "step": 14307 }, { "epoch": 138.7397590361446, "grad_norm": 5.78265905380249, "learning_rate": 3.054368932038835e-05, "loss": 0.3835, "step": 14308 }, { "epoch": 138.74939759036144, "grad_norm": 27.296504974365234, "learning_rate": 3.0538834951456314e-05, "loss": 0.3026, "step": 14309 }, { "epoch": 138.75903614457832, "grad_norm": 2.69596791267395, "learning_rate": 3.053398058252427e-05, "loss": 0.113, "step": 14310 }, { "epoch": 138.76867469879517, "grad_norm": 34.67015838623047, "learning_rate": 3.0529126213592236e-05, "loss": 0.4035, "step": 14311 }, { "epoch": 138.77831325301204, "grad_norm": 9.12798023223877, "learning_rate": 3.0524271844660194e-05, "loss": 0.2447, "step": 14312 }, { "epoch": 138.78795180722892, "grad_norm": 40.36968994140625, "learning_rate": 3.051941747572816e-05, "loss": 0.3086, "step": 14313 }, { "epoch": 138.79759036144577, "grad_norm": 30.897106170654297, "learning_rate": 3.051456310679612e-05, "loss": 0.2119, "step": 14314 }, { "epoch": 138.80722891566265, "grad_norm": 2.6798174381256104, "learning_rate": 3.050970873786408e-05, "loss": 0.2333, "step": 14315 }, { "epoch": 138.81686746987953, "grad_norm": 25.88935089111328, "learning_rate": 3.0504854368932038e-05, "loss": 0.2942, "step": 14316 }, { "epoch": 138.82650602409637, "grad_norm": 4.587162017822266, "learning_rate": 3.05e-05, "loss": 0.0938, "step": 14317 }, { "epoch": 138.83614457831325, "grad_norm": 15.205927848815918, "learning_rate": 3.049514563106796e-05, "loss": 0.2571, "step": 14318 }, { "epoch": 138.84578313253013, "grad_norm": 14.861349105834961, "learning_rate": 3.049029126213592e-05, "loss": 0.3408, "step": 14319 }, { "epoch": 138.85542168674698, "grad_norm": 22.095245361328125, "learning_rate": 3.0485436893203883e-05, "loss": 0.256, "step": 14320 }, { "epoch": 138.86506024096386, "grad_norm": 54.74208450317383, "learning_rate": 3.0480582524271844e-05, "loss": 0.3379, "step": 14321 }, { "epoch": 138.87469879518073, "grad_norm": 38.02848815917969, "learning_rate": 3.0475728155339805e-05, "loss": 0.5742, "step": 14322 }, { "epoch": 138.88433734939758, "grad_norm": 24.653738021850586, "learning_rate": 3.0470873786407766e-05, "loss": 0.2268, "step": 14323 }, { "epoch": 138.89397590361446, "grad_norm": 4.239695072174072, "learning_rate": 3.0466019417475727e-05, "loss": 0.194, "step": 14324 }, { "epoch": 138.90361445783134, "grad_norm": 24.62096405029297, "learning_rate": 3.046116504854369e-05, "loss": 0.4041, "step": 14325 }, { "epoch": 138.9132530120482, "grad_norm": 19.974851608276367, "learning_rate": 3.045631067961165e-05, "loss": 0.3823, "step": 14326 }, { "epoch": 138.92289156626506, "grad_norm": 2.439681053161621, "learning_rate": 3.0451456310679614e-05, "loss": 0.1037, "step": 14327 }, { "epoch": 138.9325301204819, "grad_norm": 35.728126525878906, "learning_rate": 3.0446601941747575e-05, "loss": 0.2925, "step": 14328 }, { "epoch": 138.9421686746988, "grad_norm": 21.53826141357422, "learning_rate": 3.0441747572815536e-05, "loss": 0.3234, "step": 14329 }, { "epoch": 138.95180722891567, "grad_norm": 11.433955192565918, "learning_rate": 3.0436893203883498e-05, "loss": 0.1573, "step": 14330 }, { "epoch": 138.96144578313252, "grad_norm": 5.397688388824463, "learning_rate": 3.043203883495146e-05, "loss": 0.2435, "step": 14331 }, { "epoch": 138.9710843373494, "grad_norm": 5.70395565032959, "learning_rate": 3.042718446601942e-05, "loss": 0.1491, "step": 14332 }, { "epoch": 138.98072289156627, "grad_norm": 31.455068588256836, "learning_rate": 3.042233009708738e-05, "loss": 0.311, "step": 14333 }, { "epoch": 138.99036144578312, "grad_norm": 34.96354675292969, "learning_rate": 3.0417475728155342e-05, "loss": 0.4117, "step": 14334 }, { "epoch": 139.00602409638554, "grad_norm": 12.94675064086914, "learning_rate": 3.0412621359223303e-05, "loss": 0.2408, "step": 14335 }, { "epoch": 139.01566265060242, "grad_norm": 29.95856475830078, "learning_rate": 3.0407766990291265e-05, "loss": 0.2333, "step": 14336 }, { "epoch": 139.02530120481927, "grad_norm": 2.5068280696868896, "learning_rate": 3.0402912621359226e-05, "loss": 0.217, "step": 14337 }, { "epoch": 139.03493975903615, "grad_norm": 15.456953048706055, "learning_rate": 3.0398058252427187e-05, "loss": 0.2747, "step": 14338 }, { "epoch": 139.044578313253, "grad_norm": 9.523652076721191, "learning_rate": 3.0393203883495148e-05, "loss": 0.2313, "step": 14339 }, { "epoch": 139.05421686746988, "grad_norm": 9.267417907714844, "learning_rate": 3.038834951456311e-05, "loss": 0.3031, "step": 14340 }, { "epoch": 139.06385542168675, "grad_norm": 36.974552154541016, "learning_rate": 3.038349514563107e-05, "loss": 0.476, "step": 14341 }, { "epoch": 139.0734939759036, "grad_norm": 9.989754676818848, "learning_rate": 3.037864077669903e-05, "loss": 0.3273, "step": 14342 }, { "epoch": 139.08313253012048, "grad_norm": 23.629148483276367, "learning_rate": 3.0373786407766993e-05, "loss": 0.3847, "step": 14343 }, { "epoch": 139.09277108433736, "grad_norm": 5.970698356628418, "learning_rate": 3.036893203883495e-05, "loss": 0.1221, "step": 14344 }, { "epoch": 139.1024096385542, "grad_norm": 8.849733352661133, "learning_rate": 3.036407766990291e-05, "loss": 0.4295, "step": 14345 }, { "epoch": 139.11204819277108, "grad_norm": 4.764082431793213, "learning_rate": 3.0359223300970873e-05, "loss": 0.2941, "step": 14346 }, { "epoch": 139.12168674698796, "grad_norm": 11.182151794433594, "learning_rate": 3.0354368932038834e-05, "loss": 0.3329, "step": 14347 }, { "epoch": 139.1313253012048, "grad_norm": 13.291661262512207, "learning_rate": 3.0349514563106795e-05, "loss": 0.3791, "step": 14348 }, { "epoch": 139.1409638554217, "grad_norm": 6.144449710845947, "learning_rate": 3.0344660194174756e-05, "loss": 0.2541, "step": 14349 }, { "epoch": 139.15060240963857, "grad_norm": 50.81297302246094, "learning_rate": 3.0339805825242717e-05, "loss": 0.3015, "step": 14350 }, { "epoch": 139.16024096385541, "grad_norm": 8.244401931762695, "learning_rate": 3.033495145631068e-05, "loss": 0.2702, "step": 14351 }, { "epoch": 139.1698795180723, "grad_norm": 8.769792556762695, "learning_rate": 3.0330097087378646e-05, "loss": 0.3652, "step": 14352 }, { "epoch": 139.17951807228917, "grad_norm": 4.5189008712768555, "learning_rate": 3.0325242718446607e-05, "loss": 0.1128, "step": 14353 }, { "epoch": 139.18915662650602, "grad_norm": 46.25679397583008, "learning_rate": 3.032038834951457e-05, "loss": 0.2668, "step": 14354 }, { "epoch": 139.1987951807229, "grad_norm": 2.082273483276367, "learning_rate": 3.0315533980582526e-05, "loss": 0.1461, "step": 14355 }, { "epoch": 139.20843373493975, "grad_norm": 28.245283126831055, "learning_rate": 3.0310679611650487e-05, "loss": 0.3072, "step": 14356 }, { "epoch": 139.21807228915662, "grad_norm": 52.588253021240234, "learning_rate": 3.030582524271845e-05, "loss": 0.1078, "step": 14357 }, { "epoch": 139.2277108433735, "grad_norm": 15.875017166137695, "learning_rate": 3.030097087378641e-05, "loss": 0.3472, "step": 14358 }, { "epoch": 139.23734939759035, "grad_norm": 11.806509971618652, "learning_rate": 3.029611650485437e-05, "loss": 0.2268, "step": 14359 }, { "epoch": 139.24698795180723, "grad_norm": 8.29914665222168, "learning_rate": 3.0291262135922332e-05, "loss": 0.2921, "step": 14360 }, { "epoch": 139.2566265060241, "grad_norm": 17.297897338867188, "learning_rate": 3.0286407766990293e-05, "loss": 0.3394, "step": 14361 }, { "epoch": 139.26626506024095, "grad_norm": 13.461536407470703, "learning_rate": 3.0281553398058254e-05, "loss": 0.4538, "step": 14362 }, { "epoch": 139.27590361445783, "grad_norm": 7.68893575668335, "learning_rate": 3.0276699029126215e-05, "loss": 0.4326, "step": 14363 }, { "epoch": 139.2855421686747, "grad_norm": 5.047132968902588, "learning_rate": 3.0271844660194177e-05, "loss": 0.1649, "step": 14364 }, { "epoch": 139.29518072289156, "grad_norm": 8.356964111328125, "learning_rate": 3.0266990291262138e-05, "loss": 0.1418, "step": 14365 }, { "epoch": 139.30481927710844, "grad_norm": 21.26068687438965, "learning_rate": 3.02621359223301e-05, "loss": 0.483, "step": 14366 }, { "epoch": 139.3144578313253, "grad_norm": 14.575533866882324, "learning_rate": 3.025728155339806e-05, "loss": 0.2878, "step": 14367 }, { "epoch": 139.32409638554216, "grad_norm": 8.623673439025879, "learning_rate": 3.025242718446602e-05, "loss": 0.2242, "step": 14368 }, { "epoch": 139.33373493975904, "grad_norm": 16.167932510375977, "learning_rate": 3.0247572815533982e-05, "loss": 0.2291, "step": 14369 }, { "epoch": 139.34337349397592, "grad_norm": 8.151817321777344, "learning_rate": 3.0242718446601943e-05, "loss": 0.2469, "step": 14370 }, { "epoch": 139.35301204819277, "grad_norm": 19.771244049072266, "learning_rate": 3.02378640776699e-05, "loss": 0.2779, "step": 14371 }, { "epoch": 139.36265060240964, "grad_norm": 19.51400375366211, "learning_rate": 3.0233009708737862e-05, "loss": 0.277, "step": 14372 }, { "epoch": 139.3722891566265, "grad_norm": 6.025661945343018, "learning_rate": 3.0228155339805824e-05, "loss": 0.2292, "step": 14373 }, { "epoch": 139.38192771084337, "grad_norm": 6.804450035095215, "learning_rate": 3.0223300970873785e-05, "loss": 0.2814, "step": 14374 }, { "epoch": 139.39156626506025, "grad_norm": 6.128966808319092, "learning_rate": 3.0218446601941746e-05, "loss": 0.2633, "step": 14375 }, { "epoch": 139.4012048192771, "grad_norm": 8.590731620788574, "learning_rate": 3.0213592233009707e-05, "loss": 0.1523, "step": 14376 }, { "epoch": 139.41084337349398, "grad_norm": 36.36268615722656, "learning_rate": 3.0208737864077668e-05, "loss": 0.3191, "step": 14377 }, { "epoch": 139.42048192771085, "grad_norm": 4.45163631439209, "learning_rate": 3.0203883495145636e-05, "loss": 0.1157, "step": 14378 }, { "epoch": 139.4301204819277, "grad_norm": 5.512685775756836, "learning_rate": 3.0199029126213597e-05, "loss": 0.3069, "step": 14379 }, { "epoch": 139.43975903614458, "grad_norm": 7.450693130493164, "learning_rate": 3.019417475728156e-05, "loss": 0.2839, "step": 14380 }, { "epoch": 139.44939759036146, "grad_norm": 9.220272064208984, "learning_rate": 3.018932038834952e-05, "loss": 0.207, "step": 14381 }, { "epoch": 139.4590361445783, "grad_norm": 5.675910472869873, "learning_rate": 3.0184466019417477e-05, "loss": 0.5253, "step": 14382 }, { "epoch": 139.46867469879518, "grad_norm": 7.262548446655273, "learning_rate": 3.017961165048544e-05, "loss": 0.2209, "step": 14383 }, { "epoch": 139.47831325301206, "grad_norm": 3.161858081817627, "learning_rate": 3.01747572815534e-05, "loss": 0.1486, "step": 14384 }, { "epoch": 139.4879518072289, "grad_norm": 11.25513744354248, "learning_rate": 3.016990291262136e-05, "loss": 0.0849, "step": 14385 }, { "epoch": 139.4975903614458, "grad_norm": 18.7281494140625, "learning_rate": 3.0165048543689322e-05, "loss": 0.3593, "step": 14386 }, { "epoch": 139.50722891566264, "grad_norm": 1.9492594003677368, "learning_rate": 3.0160194174757283e-05, "loss": 0.075, "step": 14387 }, { "epoch": 139.51686746987951, "grad_norm": 18.65642738342285, "learning_rate": 3.0155339805825244e-05, "loss": 0.2879, "step": 14388 }, { "epoch": 139.5265060240964, "grad_norm": 19.827594757080078, "learning_rate": 3.0150485436893205e-05, "loss": 0.2875, "step": 14389 }, { "epoch": 139.53614457831324, "grad_norm": 19.08393669128418, "learning_rate": 3.0145631067961166e-05, "loss": 0.2761, "step": 14390 }, { "epoch": 139.54578313253012, "grad_norm": 6.138540267944336, "learning_rate": 3.0140776699029128e-05, "loss": 0.1028, "step": 14391 }, { "epoch": 139.555421686747, "grad_norm": 81.91327667236328, "learning_rate": 3.013592233009709e-05, "loss": 0.2744, "step": 14392 }, { "epoch": 139.56506024096385, "grad_norm": 20.10558319091797, "learning_rate": 3.013106796116505e-05, "loss": 0.2735, "step": 14393 }, { "epoch": 139.57469879518072, "grad_norm": 17.532230377197266, "learning_rate": 3.012621359223301e-05, "loss": 0.1426, "step": 14394 }, { "epoch": 139.5843373493976, "grad_norm": 42.59558868408203, "learning_rate": 3.0121359223300972e-05, "loss": 0.2988, "step": 14395 }, { "epoch": 139.59397590361445, "grad_norm": 9.179059982299805, "learning_rate": 3.0116504854368933e-05, "loss": 0.13, "step": 14396 }, { "epoch": 139.60361445783133, "grad_norm": 27.744667053222656, "learning_rate": 3.0111650485436894e-05, "loss": 0.2458, "step": 14397 }, { "epoch": 139.6132530120482, "grad_norm": 3.6020565032958984, "learning_rate": 3.0106796116504856e-05, "loss": 0.1064, "step": 14398 }, { "epoch": 139.62289156626505, "grad_norm": 11.424844741821289, "learning_rate": 3.0101941747572813e-05, "loss": 0.2389, "step": 14399 }, { "epoch": 139.63253012048193, "grad_norm": 14.628501892089844, "learning_rate": 3.0097087378640774e-05, "loss": 0.1149, "step": 14400 }, { "epoch": 139.6421686746988, "grad_norm": 5.351166725158691, "learning_rate": 3.0092233009708736e-05, "loss": 0.3321, "step": 14401 }, { "epoch": 139.65180722891566, "grad_norm": 11.350137710571289, "learning_rate": 3.0087378640776697e-05, "loss": 0.4017, "step": 14402 }, { "epoch": 139.66144578313254, "grad_norm": 16.235410690307617, "learning_rate": 3.0082524271844665e-05, "loss": 0.2485, "step": 14403 }, { "epoch": 139.67108433734938, "grad_norm": 12.606432914733887, "learning_rate": 3.0077669902912626e-05, "loss": 0.4782, "step": 14404 }, { "epoch": 139.68072289156626, "grad_norm": 20.474180221557617, "learning_rate": 3.0072815533980587e-05, "loss": 0.2329, "step": 14405 }, { "epoch": 139.69036144578314, "grad_norm": 9.481765747070312, "learning_rate": 3.0067961165048548e-05, "loss": 0.3591, "step": 14406 }, { "epoch": 139.7, "grad_norm": 53.692623138427734, "learning_rate": 3.006310679611651e-05, "loss": 0.4688, "step": 14407 }, { "epoch": 139.70963855421687, "grad_norm": 9.730127334594727, "learning_rate": 3.005825242718447e-05, "loss": 0.3338, "step": 14408 }, { "epoch": 139.71927710843374, "grad_norm": 13.672257423400879, "learning_rate": 3.005339805825243e-05, "loss": 0.4158, "step": 14409 }, { "epoch": 139.7289156626506, "grad_norm": 16.206985473632812, "learning_rate": 3.004854368932039e-05, "loss": 0.1448, "step": 14410 }, { "epoch": 139.73855421686747, "grad_norm": 19.950727462768555, "learning_rate": 3.004368932038835e-05, "loss": 0.2887, "step": 14411 }, { "epoch": 139.74819277108435, "grad_norm": 49.05877685546875, "learning_rate": 3.003883495145631e-05, "loss": 0.2584, "step": 14412 }, { "epoch": 139.7578313253012, "grad_norm": 10.705999374389648, "learning_rate": 3.0033980582524273e-05, "loss": 0.2872, "step": 14413 }, { "epoch": 139.76746987951807, "grad_norm": 6.932000160217285, "learning_rate": 3.0029126213592234e-05, "loss": 0.3204, "step": 14414 }, { "epoch": 139.77710843373495, "grad_norm": 4.074577808380127, "learning_rate": 3.0024271844660195e-05, "loss": 0.3254, "step": 14415 }, { "epoch": 139.7867469879518, "grad_norm": 33.12164306640625, "learning_rate": 3.0019417475728156e-05, "loss": 0.41, "step": 14416 }, { "epoch": 139.79638554216868, "grad_norm": 19.769573211669922, "learning_rate": 3.0014563106796117e-05, "loss": 0.4122, "step": 14417 }, { "epoch": 139.80602409638556, "grad_norm": 38.43991470336914, "learning_rate": 3.000970873786408e-05, "loss": 0.2354, "step": 14418 }, { "epoch": 139.8156626506024, "grad_norm": 8.020301818847656, "learning_rate": 3.000485436893204e-05, "loss": 0.2879, "step": 14419 }, { "epoch": 139.82530120481928, "grad_norm": 15.526362419128418, "learning_rate": 3e-05, "loss": 0.4575, "step": 14420 }, { "epoch": 139.83493975903613, "grad_norm": 13.835356712341309, "learning_rate": 2.9995145631067962e-05, "loss": 0.1701, "step": 14421 }, { "epoch": 139.844578313253, "grad_norm": 54.670738220214844, "learning_rate": 2.9990291262135923e-05, "loss": 0.1914, "step": 14422 }, { "epoch": 139.8542168674699, "grad_norm": 7.150982856750488, "learning_rate": 2.9985436893203884e-05, "loss": 0.1777, "step": 14423 }, { "epoch": 139.86385542168674, "grad_norm": 12.188894271850586, "learning_rate": 2.9980582524271845e-05, "loss": 0.3135, "step": 14424 }, { "epoch": 139.8734939759036, "grad_norm": 6.45560884475708, "learning_rate": 2.9975728155339806e-05, "loss": 0.2086, "step": 14425 }, { "epoch": 139.8831325301205, "grad_norm": 5.087088108062744, "learning_rate": 2.9970873786407768e-05, "loss": 0.1554, "step": 14426 }, { "epoch": 139.89277108433734, "grad_norm": 15.58498477935791, "learning_rate": 2.9966019417475725e-05, "loss": 0.1773, "step": 14427 }, { "epoch": 139.90240963855422, "grad_norm": 24.723865509033203, "learning_rate": 2.9961165048543693e-05, "loss": 0.3196, "step": 14428 }, { "epoch": 139.9120481927711, "grad_norm": 10.054789543151855, "learning_rate": 2.9956310679611654e-05, "loss": 0.3888, "step": 14429 }, { "epoch": 139.92168674698794, "grad_norm": 37.30450439453125, "learning_rate": 2.9951456310679616e-05, "loss": 0.1823, "step": 14430 }, { "epoch": 139.93132530120482, "grad_norm": 29.290157318115234, "learning_rate": 2.9946601941747577e-05, "loss": 0.2713, "step": 14431 }, { "epoch": 139.9409638554217, "grad_norm": 12.568475723266602, "learning_rate": 2.9941747572815538e-05, "loss": 0.3337, "step": 14432 }, { "epoch": 139.95060240963855, "grad_norm": 22.906057357788086, "learning_rate": 2.99368932038835e-05, "loss": 0.3694, "step": 14433 }, { "epoch": 139.96024096385543, "grad_norm": 22.623201370239258, "learning_rate": 2.993203883495146e-05, "loss": 0.4201, "step": 14434 }, { "epoch": 139.96987951807228, "grad_norm": 17.12993049621582, "learning_rate": 2.992718446601942e-05, "loss": 0.2866, "step": 14435 }, { "epoch": 139.97951807228915, "grad_norm": 8.664969444274902, "learning_rate": 2.9922330097087382e-05, "loss": 0.2578, "step": 14436 }, { "epoch": 139.98915662650603, "grad_norm": 49.313743591308594, "learning_rate": 2.9917475728155344e-05, "loss": 0.1279, "step": 14437 }, { "epoch": 140.00481927710842, "grad_norm": 9.537872314453125, "learning_rate": 2.99126213592233e-05, "loss": 0.4506, "step": 14438 }, { "epoch": 140.0144578313253, "grad_norm": 14.323986053466797, "learning_rate": 2.9907766990291263e-05, "loss": 0.2723, "step": 14439 }, { "epoch": 140.02409638554218, "grad_norm": 18.38979148864746, "learning_rate": 2.9902912621359224e-05, "loss": 0.1204, "step": 14440 }, { "epoch": 140.03373493975903, "grad_norm": 4.633047103881836, "learning_rate": 2.9898058252427185e-05, "loss": 0.3831, "step": 14441 }, { "epoch": 140.0433734939759, "grad_norm": 6.999035835266113, "learning_rate": 2.9893203883495146e-05, "loss": 0.1285, "step": 14442 }, { "epoch": 140.05301204819278, "grad_norm": 2.3765475749969482, "learning_rate": 2.9888349514563107e-05, "loss": 0.0912, "step": 14443 }, { "epoch": 140.06265060240963, "grad_norm": 12.092436790466309, "learning_rate": 2.9883495145631068e-05, "loss": 0.1416, "step": 14444 }, { "epoch": 140.0722891566265, "grad_norm": 10.90138053894043, "learning_rate": 2.987864077669903e-05, "loss": 0.3534, "step": 14445 }, { "epoch": 140.08192771084336, "grad_norm": 7.3075079917907715, "learning_rate": 2.987378640776699e-05, "loss": 0.3519, "step": 14446 }, { "epoch": 140.09156626506024, "grad_norm": 13.01922607421875, "learning_rate": 2.9868932038834952e-05, "loss": 0.5383, "step": 14447 }, { "epoch": 140.10120481927711, "grad_norm": 171.499755859375, "learning_rate": 2.9864077669902913e-05, "loss": 0.2429, "step": 14448 }, { "epoch": 140.11084337349396, "grad_norm": 20.443716049194336, "learning_rate": 2.9859223300970874e-05, "loss": 0.2164, "step": 14449 }, { "epoch": 140.12048192771084, "grad_norm": 14.178404808044434, "learning_rate": 2.9854368932038835e-05, "loss": 0.2951, "step": 14450 }, { "epoch": 140.13012048192772, "grad_norm": 9.228598594665527, "learning_rate": 2.9849514563106796e-05, "loss": 0.2538, "step": 14451 }, { "epoch": 140.13975903614457, "grad_norm": 9.658448219299316, "learning_rate": 2.9844660194174757e-05, "loss": 0.2155, "step": 14452 }, { "epoch": 140.14939759036145, "grad_norm": 10.403534889221191, "learning_rate": 2.9839805825242722e-05, "loss": 0.4613, "step": 14453 }, { "epoch": 140.15903614457832, "grad_norm": 6.306214332580566, "learning_rate": 2.9834951456310683e-05, "loss": 0.195, "step": 14454 }, { "epoch": 140.16867469879517, "grad_norm": 37.94664001464844, "learning_rate": 2.9830097087378644e-05, "loss": 0.3642, "step": 14455 }, { "epoch": 140.17831325301205, "grad_norm": 5.9659552574157715, "learning_rate": 2.9825242718446605e-05, "loss": 0.1715, "step": 14456 }, { "epoch": 140.18795180722893, "grad_norm": 8.2260103225708, "learning_rate": 2.9820388349514567e-05, "loss": 0.1084, "step": 14457 }, { "epoch": 140.19759036144578, "grad_norm": 24.837417602539062, "learning_rate": 2.9815533980582528e-05, "loss": 0.3623, "step": 14458 }, { "epoch": 140.20722891566265, "grad_norm": 18.080604553222656, "learning_rate": 2.981067961165049e-05, "loss": 0.3483, "step": 14459 }, { "epoch": 140.21686746987953, "grad_norm": 14.399162292480469, "learning_rate": 2.980582524271845e-05, "loss": 0.2957, "step": 14460 }, { "epoch": 140.22650602409638, "grad_norm": 6.174508094787598, "learning_rate": 2.980097087378641e-05, "loss": 0.2572, "step": 14461 }, { "epoch": 140.23614457831326, "grad_norm": 5.970833778381348, "learning_rate": 2.9796116504854372e-05, "loss": 0.2561, "step": 14462 }, { "epoch": 140.2457831325301, "grad_norm": 11.07824993133545, "learning_rate": 2.9791262135922333e-05, "loss": 0.327, "step": 14463 }, { "epoch": 140.25542168674698, "grad_norm": 103.04547882080078, "learning_rate": 2.9786407766990295e-05, "loss": 0.4576, "step": 14464 }, { "epoch": 140.26506024096386, "grad_norm": 33.040748596191406, "learning_rate": 2.9781553398058252e-05, "loss": 0.3984, "step": 14465 }, { "epoch": 140.2746987951807, "grad_norm": 21.799718856811523, "learning_rate": 2.9776699029126213e-05, "loss": 0.3493, "step": 14466 }, { "epoch": 140.2843373493976, "grad_norm": 14.69987678527832, "learning_rate": 2.9771844660194175e-05, "loss": 0.2735, "step": 14467 }, { "epoch": 140.29397590361447, "grad_norm": 5.897439002990723, "learning_rate": 2.9766990291262136e-05, "loss": 0.1562, "step": 14468 }, { "epoch": 140.30361445783132, "grad_norm": 18.09585952758789, "learning_rate": 2.9762135922330097e-05, "loss": 0.3561, "step": 14469 }, { "epoch": 140.3132530120482, "grad_norm": 32.014286041259766, "learning_rate": 2.9757281553398058e-05, "loss": 0.2861, "step": 14470 }, { "epoch": 140.32289156626507, "grad_norm": 3.866456985473633, "learning_rate": 2.975242718446602e-05, "loss": 0.1632, "step": 14471 }, { "epoch": 140.33253012048192, "grad_norm": 69.89692687988281, "learning_rate": 2.974757281553398e-05, "loss": 0.2321, "step": 14472 }, { "epoch": 140.3421686746988, "grad_norm": 10.246537208557129, "learning_rate": 2.974271844660194e-05, "loss": 0.3367, "step": 14473 }, { "epoch": 140.35180722891567, "grad_norm": 20.676389694213867, "learning_rate": 2.9737864077669903e-05, "loss": 0.1854, "step": 14474 }, { "epoch": 140.36144578313252, "grad_norm": 14.133759498596191, "learning_rate": 2.9733009708737864e-05, "loss": 0.3285, "step": 14475 }, { "epoch": 140.3710843373494, "grad_norm": 20.428327560424805, "learning_rate": 2.9728155339805825e-05, "loss": 0.156, "step": 14476 }, { "epoch": 140.38072289156628, "grad_norm": 5.83751916885376, "learning_rate": 2.9723300970873786e-05, "loss": 0.2811, "step": 14477 }, { "epoch": 140.39036144578313, "grad_norm": 14.665396690368652, "learning_rate": 2.9718446601941747e-05, "loss": 0.3437, "step": 14478 }, { "epoch": 140.4, "grad_norm": 4.782721042633057, "learning_rate": 2.9713592233009712e-05, "loss": 0.2541, "step": 14479 }, { "epoch": 140.40963855421685, "grad_norm": 12.43993854522705, "learning_rate": 2.9708737864077673e-05, "loss": 0.3146, "step": 14480 }, { "epoch": 140.41927710843373, "grad_norm": 8.48719596862793, "learning_rate": 2.9703883495145634e-05, "loss": 0.1539, "step": 14481 }, { "epoch": 140.4289156626506, "grad_norm": 7.925205707550049, "learning_rate": 2.9699029126213595e-05, "loss": 0.1869, "step": 14482 }, { "epoch": 140.43855421686746, "grad_norm": 24.03400230407715, "learning_rate": 2.9694174757281556e-05, "loss": 0.2367, "step": 14483 }, { "epoch": 140.44819277108434, "grad_norm": 22.45306396484375, "learning_rate": 2.9689320388349517e-05, "loss": 0.1775, "step": 14484 }, { "epoch": 140.4578313253012, "grad_norm": 5.93035888671875, "learning_rate": 2.968446601941748e-05, "loss": 0.2647, "step": 14485 }, { "epoch": 140.46746987951806, "grad_norm": 8.567410469055176, "learning_rate": 2.967961165048544e-05, "loss": 0.1789, "step": 14486 }, { "epoch": 140.47710843373494, "grad_norm": 15.70443058013916, "learning_rate": 2.96747572815534e-05, "loss": 0.2547, "step": 14487 }, { "epoch": 140.48674698795182, "grad_norm": 17.789011001586914, "learning_rate": 2.9669902912621362e-05, "loss": 0.2733, "step": 14488 }, { "epoch": 140.49638554216867, "grad_norm": 10.282691955566406, "learning_rate": 2.9665048543689323e-05, "loss": 0.3421, "step": 14489 }, { "epoch": 140.50602409638554, "grad_norm": 28.838064193725586, "learning_rate": 2.9660194174757284e-05, "loss": 0.2914, "step": 14490 }, { "epoch": 140.51566265060242, "grad_norm": 9.754887580871582, "learning_rate": 2.9655339805825245e-05, "loss": 0.2696, "step": 14491 }, { "epoch": 140.52530120481927, "grad_norm": 13.129993438720703, "learning_rate": 2.9650485436893207e-05, "loss": 0.3251, "step": 14492 }, { "epoch": 140.53493975903615, "grad_norm": 93.29779815673828, "learning_rate": 2.9645631067961164e-05, "loss": 0.2856, "step": 14493 }, { "epoch": 140.544578313253, "grad_norm": 31.699533462524414, "learning_rate": 2.9640776699029126e-05, "loss": 0.4006, "step": 14494 }, { "epoch": 140.55421686746988, "grad_norm": 20.604801177978516, "learning_rate": 2.9635922330097087e-05, "loss": 0.204, "step": 14495 }, { "epoch": 140.56385542168675, "grad_norm": 16.716033935546875, "learning_rate": 2.9631067961165048e-05, "loss": 0.3412, "step": 14496 }, { "epoch": 140.5734939759036, "grad_norm": 10.563735961914062, "learning_rate": 2.962621359223301e-05, "loss": 0.278, "step": 14497 }, { "epoch": 140.58313253012048, "grad_norm": 6.094501972198486, "learning_rate": 2.962135922330097e-05, "loss": 0.4124, "step": 14498 }, { "epoch": 140.59277108433736, "grad_norm": 4.844130039215088, "learning_rate": 2.961650485436893e-05, "loss": 0.1492, "step": 14499 }, { "epoch": 140.6024096385542, "grad_norm": 8.58050537109375, "learning_rate": 2.9611650485436892e-05, "loss": 0.2811, "step": 14500 }, { "epoch": 140.61204819277108, "grad_norm": 27.861522674560547, "learning_rate": 2.9606796116504854e-05, "loss": 0.3131, "step": 14501 }, { "epoch": 140.62168674698796, "grad_norm": 26.223920822143555, "learning_rate": 2.9601941747572815e-05, "loss": 0.1392, "step": 14502 }, { "epoch": 140.6313253012048, "grad_norm": 14.606232643127441, "learning_rate": 2.9597087378640776e-05, "loss": 0.2861, "step": 14503 }, { "epoch": 140.6409638554217, "grad_norm": 8.47751522064209, "learning_rate": 2.959223300970874e-05, "loss": 0.1642, "step": 14504 }, { "epoch": 140.65060240963857, "grad_norm": 8.853891372680664, "learning_rate": 2.95873786407767e-05, "loss": 0.3418, "step": 14505 }, { "epoch": 140.66024096385541, "grad_norm": 7.27133846282959, "learning_rate": 2.9582524271844663e-05, "loss": 0.1816, "step": 14506 }, { "epoch": 140.6698795180723, "grad_norm": 20.567230224609375, "learning_rate": 2.9577669902912624e-05, "loss": 0.258, "step": 14507 }, { "epoch": 140.67951807228917, "grad_norm": 23.8400821685791, "learning_rate": 2.9572815533980585e-05, "loss": 0.2486, "step": 14508 }, { "epoch": 140.68915662650602, "grad_norm": 10.78034496307373, "learning_rate": 2.9567961165048546e-05, "loss": 0.2843, "step": 14509 }, { "epoch": 140.6987951807229, "grad_norm": 18.946487426757812, "learning_rate": 2.9563106796116507e-05, "loss": 0.2742, "step": 14510 }, { "epoch": 140.70843373493975, "grad_norm": 8.430658340454102, "learning_rate": 2.955825242718447e-05, "loss": 0.2749, "step": 14511 }, { "epoch": 140.71807228915662, "grad_norm": 22.382123947143555, "learning_rate": 2.955339805825243e-05, "loss": 0.167, "step": 14512 }, { "epoch": 140.7277108433735, "grad_norm": 24.712644577026367, "learning_rate": 2.954854368932039e-05, "loss": 0.4505, "step": 14513 }, { "epoch": 140.73734939759035, "grad_norm": 8.937066078186035, "learning_rate": 2.9543689320388352e-05, "loss": 0.3961, "step": 14514 }, { "epoch": 140.74698795180723, "grad_norm": 11.231918334960938, "learning_rate": 2.9538834951456313e-05, "loss": 0.3341, "step": 14515 }, { "epoch": 140.7566265060241, "grad_norm": 7.331258773803711, "learning_rate": 2.9533980582524274e-05, "loss": 0.283, "step": 14516 }, { "epoch": 140.76626506024095, "grad_norm": 8.847391128540039, "learning_rate": 2.9529126213592235e-05, "loss": 0.2078, "step": 14517 }, { "epoch": 140.77590361445783, "grad_norm": 19.309728622436523, "learning_rate": 2.9524271844660196e-05, "loss": 0.2128, "step": 14518 }, { "epoch": 140.7855421686747, "grad_norm": 10.66330623626709, "learning_rate": 2.9519417475728158e-05, "loss": 0.296, "step": 14519 }, { "epoch": 140.79518072289156, "grad_norm": 28.574508666992188, "learning_rate": 2.9514563106796115e-05, "loss": 0.4167, "step": 14520 }, { "epoch": 140.80481927710844, "grad_norm": 23.187030792236328, "learning_rate": 2.9509708737864076e-05, "loss": 0.2166, "step": 14521 }, { "epoch": 140.8144578313253, "grad_norm": 7.333532810211182, "learning_rate": 2.9504854368932038e-05, "loss": 0.1763, "step": 14522 }, { "epoch": 140.82409638554216, "grad_norm": 3.828115940093994, "learning_rate": 2.95e-05, "loss": 0.4084, "step": 14523 }, { "epoch": 140.83373493975904, "grad_norm": 8.336254119873047, "learning_rate": 2.949514563106796e-05, "loss": 0.2622, "step": 14524 }, { "epoch": 140.84337349397592, "grad_norm": 9.388557434082031, "learning_rate": 2.949029126213592e-05, "loss": 0.1677, "step": 14525 }, { "epoch": 140.85301204819277, "grad_norm": 6.47004508972168, "learning_rate": 2.9485436893203882e-05, "loss": 0.1594, "step": 14526 }, { "epoch": 140.86265060240964, "grad_norm": 9.318249702453613, "learning_rate": 2.9480582524271843e-05, "loss": 0.2123, "step": 14527 }, { "epoch": 140.8722891566265, "grad_norm": 9.578512191772461, "learning_rate": 2.9475728155339804e-05, "loss": 0.2428, "step": 14528 }, { "epoch": 140.88192771084337, "grad_norm": 13.505982398986816, "learning_rate": 2.9470873786407772e-05, "loss": 0.462, "step": 14529 }, { "epoch": 140.89156626506025, "grad_norm": 5.245932102203369, "learning_rate": 2.9466019417475734e-05, "loss": 0.3168, "step": 14530 }, { "epoch": 140.9012048192771, "grad_norm": 17.524192810058594, "learning_rate": 2.946116504854369e-05, "loss": 0.2115, "step": 14531 }, { "epoch": 140.91084337349398, "grad_norm": 33.089847564697266, "learning_rate": 2.9456310679611652e-05, "loss": 0.3045, "step": 14532 }, { "epoch": 140.92048192771085, "grad_norm": 46.310123443603516, "learning_rate": 2.9451456310679614e-05, "loss": 0.3572, "step": 14533 }, { "epoch": 140.9301204819277, "grad_norm": 1.9332648515701294, "learning_rate": 2.9446601941747575e-05, "loss": 0.1109, "step": 14534 }, { "epoch": 140.93975903614458, "grad_norm": 3.4352781772613525, "learning_rate": 2.9441747572815536e-05, "loss": 0.1734, "step": 14535 }, { "epoch": 140.94939759036146, "grad_norm": 6.1031293869018555, "learning_rate": 2.9436893203883497e-05, "loss": 0.1233, "step": 14536 }, { "epoch": 140.9590361445783, "grad_norm": 14.774539947509766, "learning_rate": 2.9432038834951458e-05, "loss": 0.2121, "step": 14537 }, { "epoch": 140.96867469879518, "grad_norm": 14.270528793334961, "learning_rate": 2.942718446601942e-05, "loss": 0.3515, "step": 14538 }, { "epoch": 140.97831325301206, "grad_norm": 21.170190811157227, "learning_rate": 2.942233009708738e-05, "loss": 0.3652, "step": 14539 }, { "epoch": 140.9879518072289, "grad_norm": 22.50996208190918, "learning_rate": 2.941747572815534e-05, "loss": 0.3416, "step": 14540 }, { "epoch": 141.00361445783133, "grad_norm": 23.1771240234375, "learning_rate": 2.9412621359223303e-05, "loss": 0.2065, "step": 14541 }, { "epoch": 141.01325301204818, "grad_norm": 40.77794647216797, "learning_rate": 2.9407766990291264e-05, "loss": 0.3406, "step": 14542 }, { "epoch": 141.02289156626506, "grad_norm": 13.772806167602539, "learning_rate": 2.9402912621359225e-05, "loss": 0.3531, "step": 14543 }, { "epoch": 141.03253012048194, "grad_norm": 27.715471267700195, "learning_rate": 2.9398058252427186e-05, "loss": 0.3416, "step": 14544 }, { "epoch": 141.0421686746988, "grad_norm": 11.525025367736816, "learning_rate": 2.9393203883495147e-05, "loss": 0.3256, "step": 14545 }, { "epoch": 141.05180722891566, "grad_norm": 33.43312072753906, "learning_rate": 2.938834951456311e-05, "loss": 0.3914, "step": 14546 }, { "epoch": 141.06144578313254, "grad_norm": 37.67526626586914, "learning_rate": 2.938349514563107e-05, "loss": 0.4157, "step": 14547 }, { "epoch": 141.0710843373494, "grad_norm": 12.78353214263916, "learning_rate": 2.9378640776699027e-05, "loss": 0.2821, "step": 14548 }, { "epoch": 141.08072289156627, "grad_norm": 12.256091117858887, "learning_rate": 2.937378640776699e-05, "loss": 0.2382, "step": 14549 }, { "epoch": 141.09036144578315, "grad_norm": 9.120211601257324, "learning_rate": 2.936893203883495e-05, "loss": 0.1958, "step": 14550 }, { "epoch": 141.1, "grad_norm": 10.783539772033691, "learning_rate": 2.936407766990291e-05, "loss": 0.1815, "step": 14551 }, { "epoch": 141.10963855421687, "grad_norm": 18.281883239746094, "learning_rate": 2.9359223300970872e-05, "loss": 0.2437, "step": 14552 }, { "epoch": 141.11927710843372, "grad_norm": 24.30508041381836, "learning_rate": 2.9354368932038833e-05, "loss": 0.1447, "step": 14553 }, { "epoch": 141.1289156626506, "grad_norm": 7.250682353973389, "learning_rate": 2.93495145631068e-05, "loss": 0.1246, "step": 14554 }, { "epoch": 141.13855421686748, "grad_norm": 19.706260681152344, "learning_rate": 2.9344660194174762e-05, "loss": 0.2187, "step": 14555 }, { "epoch": 141.14819277108433, "grad_norm": 26.2664737701416, "learning_rate": 2.9339805825242723e-05, "loss": 0.1212, "step": 14556 }, { "epoch": 141.1578313253012, "grad_norm": 24.721092224121094, "learning_rate": 2.9334951456310685e-05, "loss": 0.3423, "step": 14557 }, { "epoch": 141.16746987951808, "grad_norm": 17.051151275634766, "learning_rate": 2.9330097087378646e-05, "loss": 0.3586, "step": 14558 }, { "epoch": 141.17710843373493, "grad_norm": 16.31395721435547, "learning_rate": 2.9325242718446603e-05, "loss": 0.3935, "step": 14559 }, { "epoch": 141.1867469879518, "grad_norm": 5.059254169464111, "learning_rate": 2.9320388349514565e-05, "loss": 0.1745, "step": 14560 }, { "epoch": 141.19638554216868, "grad_norm": 110.2907943725586, "learning_rate": 2.9315533980582526e-05, "loss": 0.2166, "step": 14561 }, { "epoch": 141.20602409638553, "grad_norm": 12.146026611328125, "learning_rate": 2.9310679611650487e-05, "loss": 0.2096, "step": 14562 }, { "epoch": 141.2156626506024, "grad_norm": 65.09769439697266, "learning_rate": 2.9305825242718448e-05, "loss": 0.3884, "step": 14563 }, { "epoch": 141.2253012048193, "grad_norm": 7.582588195800781, "learning_rate": 2.930097087378641e-05, "loss": 0.2057, "step": 14564 }, { "epoch": 141.23493975903614, "grad_norm": 3.4572830200195312, "learning_rate": 2.929611650485437e-05, "loss": 0.2557, "step": 14565 }, { "epoch": 141.24457831325302, "grad_norm": 10.13568115234375, "learning_rate": 2.929126213592233e-05, "loss": 0.288, "step": 14566 }, { "epoch": 141.2542168674699, "grad_norm": 10.955063819885254, "learning_rate": 2.9286407766990293e-05, "loss": 0.2879, "step": 14567 }, { "epoch": 141.26385542168674, "grad_norm": 8.869239807128906, "learning_rate": 2.9281553398058254e-05, "loss": 0.3951, "step": 14568 }, { "epoch": 141.27349397590362, "grad_norm": 8.657123565673828, "learning_rate": 2.9276699029126215e-05, "loss": 0.124, "step": 14569 }, { "epoch": 141.28313253012047, "grad_norm": 12.460515975952148, "learning_rate": 2.9271844660194176e-05, "loss": 0.1891, "step": 14570 }, { "epoch": 141.29277108433735, "grad_norm": 18.56460952758789, "learning_rate": 2.9266990291262137e-05, "loss": 0.2982, "step": 14571 }, { "epoch": 141.30240963855422, "grad_norm": 8.232108116149902, "learning_rate": 2.9262135922330098e-05, "loss": 0.3557, "step": 14572 }, { "epoch": 141.31204819277107, "grad_norm": 12.303061485290527, "learning_rate": 2.925728155339806e-05, "loss": 0.1438, "step": 14573 }, { "epoch": 141.32168674698795, "grad_norm": 32.75699996948242, "learning_rate": 2.925242718446602e-05, "loss": 0.3624, "step": 14574 }, { "epoch": 141.33132530120483, "grad_norm": 8.798298835754395, "learning_rate": 2.9247572815533982e-05, "loss": 0.2829, "step": 14575 }, { "epoch": 141.34096385542168, "grad_norm": 15.110627174377441, "learning_rate": 2.924271844660194e-05, "loss": 0.4077, "step": 14576 }, { "epoch": 141.35060240963855, "grad_norm": 5.18504524230957, "learning_rate": 2.92378640776699e-05, "loss": 0.2781, "step": 14577 }, { "epoch": 141.36024096385543, "grad_norm": 13.433280944824219, "learning_rate": 2.9233009708737862e-05, "loss": 0.2281, "step": 14578 }, { "epoch": 141.36987951807228, "grad_norm": 31.23396110534668, "learning_rate": 2.9228155339805823e-05, "loss": 0.49, "step": 14579 }, { "epoch": 141.37951807228916, "grad_norm": 9.266019821166992, "learning_rate": 2.922330097087379e-05, "loss": 0.2989, "step": 14580 }, { "epoch": 141.38915662650604, "grad_norm": 33.658905029296875, "learning_rate": 2.9218446601941752e-05, "loss": 0.4572, "step": 14581 }, { "epoch": 141.39879518072289, "grad_norm": 11.02899169921875, "learning_rate": 2.9213592233009713e-05, "loss": 0.3382, "step": 14582 }, { "epoch": 141.40843373493976, "grad_norm": 36.13365173339844, "learning_rate": 2.9208737864077674e-05, "loss": 0.4395, "step": 14583 }, { "epoch": 141.41807228915664, "grad_norm": 7.27309513092041, "learning_rate": 2.9203883495145635e-05, "loss": 0.1001, "step": 14584 }, { "epoch": 141.4277108433735, "grad_norm": 6.112865447998047, "learning_rate": 2.9199029126213597e-05, "loss": 0.1461, "step": 14585 }, { "epoch": 141.43734939759037, "grad_norm": 16.81076431274414, "learning_rate": 2.9194174757281558e-05, "loss": 0.4337, "step": 14586 }, { "epoch": 141.44698795180722, "grad_norm": 9.743170738220215, "learning_rate": 2.9189320388349515e-05, "loss": 0.2, "step": 14587 }, { "epoch": 141.4566265060241, "grad_norm": 18.331151962280273, "learning_rate": 2.9184466019417477e-05, "loss": 0.3252, "step": 14588 }, { "epoch": 141.46626506024097, "grad_norm": 15.139010429382324, "learning_rate": 2.9179611650485438e-05, "loss": 0.2271, "step": 14589 }, { "epoch": 141.47590361445782, "grad_norm": 23.122573852539062, "learning_rate": 2.91747572815534e-05, "loss": 0.1754, "step": 14590 }, { "epoch": 141.4855421686747, "grad_norm": 13.866931915283203, "learning_rate": 2.916990291262136e-05, "loss": 0.2519, "step": 14591 }, { "epoch": 141.49518072289158, "grad_norm": 13.47522258758545, "learning_rate": 2.916504854368932e-05, "loss": 0.459, "step": 14592 }, { "epoch": 141.50481927710842, "grad_norm": 6.701035022735596, "learning_rate": 2.9160194174757282e-05, "loss": 0.257, "step": 14593 }, { "epoch": 141.5144578313253, "grad_norm": 9.575214385986328, "learning_rate": 2.9155339805825244e-05, "loss": 0.1995, "step": 14594 }, { "epoch": 141.52409638554218, "grad_norm": 8.033658027648926, "learning_rate": 2.9150485436893205e-05, "loss": 0.1909, "step": 14595 }, { "epoch": 141.53373493975903, "grad_norm": 8.540454864501953, "learning_rate": 2.9145631067961166e-05, "loss": 0.2508, "step": 14596 }, { "epoch": 141.5433734939759, "grad_norm": 11.192680358886719, "learning_rate": 2.9140776699029127e-05, "loss": 0.2805, "step": 14597 }, { "epoch": 141.55301204819278, "grad_norm": 9.767085075378418, "learning_rate": 2.9135922330097088e-05, "loss": 0.3648, "step": 14598 }, { "epoch": 141.56265060240963, "grad_norm": 12.97353458404541, "learning_rate": 2.913106796116505e-05, "loss": 0.2008, "step": 14599 }, { "epoch": 141.5722891566265, "grad_norm": 67.38580322265625, "learning_rate": 2.912621359223301e-05, "loss": 0.1205, "step": 14600 }, { "epoch": 141.58192771084336, "grad_norm": 10.736946105957031, "learning_rate": 2.912135922330097e-05, "loss": 0.192, "step": 14601 }, { "epoch": 141.59156626506024, "grad_norm": 33.236202239990234, "learning_rate": 2.9116504854368933e-05, "loss": 0.2079, "step": 14602 }, { "epoch": 141.60120481927711, "grad_norm": 24.263940811157227, "learning_rate": 2.911165048543689e-05, "loss": 0.3544, "step": 14603 }, { "epoch": 141.61084337349396, "grad_norm": 13.39451789855957, "learning_rate": 2.910679611650485e-05, "loss": 0.1517, "step": 14604 }, { "epoch": 141.62048192771084, "grad_norm": 8.222567558288574, "learning_rate": 2.910194174757282e-05, "loss": 0.2347, "step": 14605 }, { "epoch": 141.63012048192772, "grad_norm": 10.71751880645752, "learning_rate": 2.909708737864078e-05, "loss": 0.5035, "step": 14606 }, { "epoch": 141.63975903614457, "grad_norm": 27.32879066467285, "learning_rate": 2.9092233009708742e-05, "loss": 0.2971, "step": 14607 }, { "epoch": 141.64939759036145, "grad_norm": 17.249155044555664, "learning_rate": 2.9087378640776703e-05, "loss": 0.2287, "step": 14608 }, { "epoch": 141.65903614457832, "grad_norm": 9.696069717407227, "learning_rate": 2.9082524271844664e-05, "loss": 0.2429, "step": 14609 }, { "epoch": 141.66867469879517, "grad_norm": 20.622962951660156, "learning_rate": 2.9077669902912625e-05, "loss": 0.3816, "step": 14610 }, { "epoch": 141.67831325301205, "grad_norm": 29.77079963684082, "learning_rate": 2.9072815533980586e-05, "loss": 0.3891, "step": 14611 }, { "epoch": 141.68795180722893, "grad_norm": 25.828426361083984, "learning_rate": 2.9067961165048548e-05, "loss": 0.271, "step": 14612 }, { "epoch": 141.69759036144578, "grad_norm": 5.57780647277832, "learning_rate": 2.906310679611651e-05, "loss": 0.2262, "step": 14613 }, { "epoch": 141.70722891566265, "grad_norm": 2.9955904483795166, "learning_rate": 2.9058252427184466e-05, "loss": 0.1245, "step": 14614 }, { "epoch": 141.71686746987953, "grad_norm": 10.033088684082031, "learning_rate": 2.9053398058252428e-05, "loss": 0.1892, "step": 14615 }, { "epoch": 141.72650602409638, "grad_norm": 14.994040489196777, "learning_rate": 2.904854368932039e-05, "loss": 0.2329, "step": 14616 }, { "epoch": 141.73614457831326, "grad_norm": 38.96287155151367, "learning_rate": 2.904368932038835e-05, "loss": 0.1829, "step": 14617 }, { "epoch": 141.7457831325301, "grad_norm": 8.968355178833008, "learning_rate": 2.903883495145631e-05, "loss": 0.1797, "step": 14618 }, { "epoch": 141.75542168674698, "grad_norm": 13.46786880493164, "learning_rate": 2.9033980582524272e-05, "loss": 0.506, "step": 14619 }, { "epoch": 141.76506024096386, "grad_norm": 32.32229995727539, "learning_rate": 2.9029126213592233e-05, "loss": 0.3171, "step": 14620 }, { "epoch": 141.7746987951807, "grad_norm": 19.664030075073242, "learning_rate": 2.9024271844660194e-05, "loss": 0.342, "step": 14621 }, { "epoch": 141.7843373493976, "grad_norm": 8.98819351196289, "learning_rate": 2.9019417475728156e-05, "loss": 0.3511, "step": 14622 }, { "epoch": 141.79397590361447, "grad_norm": 2.2037620544433594, "learning_rate": 2.9014563106796117e-05, "loss": 0.2503, "step": 14623 }, { "epoch": 141.80361445783132, "grad_norm": 12.599377632141113, "learning_rate": 2.9009708737864078e-05, "loss": 0.2774, "step": 14624 }, { "epoch": 141.8132530120482, "grad_norm": 10.050141334533691, "learning_rate": 2.900485436893204e-05, "loss": 0.2328, "step": 14625 }, { "epoch": 141.82289156626507, "grad_norm": 6.205036163330078, "learning_rate": 2.9e-05, "loss": 0.3252, "step": 14626 }, { "epoch": 141.83253012048192, "grad_norm": 7.9841837882995605, "learning_rate": 2.899514563106796e-05, "loss": 0.1851, "step": 14627 }, { "epoch": 141.8421686746988, "grad_norm": 13.39744758605957, "learning_rate": 2.8990291262135922e-05, "loss": 0.3322, "step": 14628 }, { "epoch": 141.85180722891567, "grad_norm": 9.448142051696777, "learning_rate": 2.8985436893203884e-05, "loss": 0.3656, "step": 14629 }, { "epoch": 141.86144578313252, "grad_norm": 14.923601150512695, "learning_rate": 2.8980582524271848e-05, "loss": 0.3537, "step": 14630 }, { "epoch": 141.8710843373494, "grad_norm": 10.49311351776123, "learning_rate": 2.897572815533981e-05, "loss": 0.3138, "step": 14631 }, { "epoch": 141.88072289156628, "grad_norm": 18.54485321044922, "learning_rate": 2.897087378640777e-05, "loss": 0.4252, "step": 14632 }, { "epoch": 141.89036144578313, "grad_norm": 22.70565414428711, "learning_rate": 2.896601941747573e-05, "loss": 0.1998, "step": 14633 }, { "epoch": 141.9, "grad_norm": 12.995394706726074, "learning_rate": 2.8961165048543693e-05, "loss": 0.2166, "step": 14634 }, { "epoch": 141.90963855421685, "grad_norm": 5.248199462890625, "learning_rate": 2.8956310679611654e-05, "loss": 0.3178, "step": 14635 }, { "epoch": 141.91927710843373, "grad_norm": 7.953581809997559, "learning_rate": 2.8951456310679615e-05, "loss": 0.2406, "step": 14636 }, { "epoch": 141.9289156626506, "grad_norm": 7.577978610992432, "learning_rate": 2.8946601941747576e-05, "loss": 0.2213, "step": 14637 }, { "epoch": 141.93855421686746, "grad_norm": 9.636711120605469, "learning_rate": 2.8941747572815537e-05, "loss": 0.0819, "step": 14638 }, { "epoch": 141.94819277108434, "grad_norm": 9.653789520263672, "learning_rate": 2.89368932038835e-05, "loss": 0.2946, "step": 14639 }, { "epoch": 141.9578313253012, "grad_norm": 9.78467845916748, "learning_rate": 2.893203883495146e-05, "loss": 0.2579, "step": 14640 }, { "epoch": 141.96746987951806, "grad_norm": 66.97826385498047, "learning_rate": 2.892718446601942e-05, "loss": 0.2048, "step": 14641 }, { "epoch": 141.97710843373494, "grad_norm": 11.42627239227295, "learning_rate": 2.892233009708738e-05, "loss": 0.317, "step": 14642 }, { "epoch": 141.98674698795182, "grad_norm": 17.465776443481445, "learning_rate": 2.891747572815534e-05, "loss": 0.2756, "step": 14643 }, { "epoch": 142.0024096385542, "grad_norm": 10.274478912353516, "learning_rate": 2.89126213592233e-05, "loss": 0.4565, "step": 14644 }, { "epoch": 142.0120481927711, "grad_norm": 16.553430557250977, "learning_rate": 2.8907766990291262e-05, "loss": 0.2832, "step": 14645 }, { "epoch": 142.02168674698794, "grad_norm": 3.8001670837402344, "learning_rate": 2.8902912621359223e-05, "loss": 0.2005, "step": 14646 }, { "epoch": 142.03132530120482, "grad_norm": 6.17253303527832, "learning_rate": 2.8898058252427184e-05, "loss": 0.2693, "step": 14647 }, { "epoch": 142.0409638554217, "grad_norm": 39.90498352050781, "learning_rate": 2.8893203883495145e-05, "loss": 0.2365, "step": 14648 }, { "epoch": 142.05060240963854, "grad_norm": 40.06373977661133, "learning_rate": 2.8888349514563107e-05, "loss": 0.1761, "step": 14649 }, { "epoch": 142.06024096385542, "grad_norm": 19.738895416259766, "learning_rate": 2.8883495145631068e-05, "loss": 0.3931, "step": 14650 }, { "epoch": 142.0698795180723, "grad_norm": 13.272218704223633, "learning_rate": 2.887864077669903e-05, "loss": 0.1994, "step": 14651 }, { "epoch": 142.07951807228915, "grad_norm": 10.37540340423584, "learning_rate": 2.887378640776699e-05, "loss": 0.1373, "step": 14652 }, { "epoch": 142.08915662650602, "grad_norm": 12.93868637084961, "learning_rate": 2.886893203883495e-05, "loss": 0.3137, "step": 14653 }, { "epoch": 142.0987951807229, "grad_norm": 9.968056678771973, "learning_rate": 2.8864077669902912e-05, "loss": 0.3865, "step": 14654 }, { "epoch": 142.10843373493975, "grad_norm": 8.004781723022461, "learning_rate": 2.8859223300970877e-05, "loss": 0.2221, "step": 14655 }, { "epoch": 142.11807228915663, "grad_norm": 37.8279914855957, "learning_rate": 2.8854368932038838e-05, "loss": 0.4299, "step": 14656 }, { "epoch": 142.1277108433735, "grad_norm": 9.39794635772705, "learning_rate": 2.88495145631068e-05, "loss": 0.1443, "step": 14657 }, { "epoch": 142.13734939759036, "grad_norm": 8.015959739685059, "learning_rate": 2.884466019417476e-05, "loss": 0.3186, "step": 14658 }, { "epoch": 142.14698795180723, "grad_norm": 11.339173316955566, "learning_rate": 2.883980582524272e-05, "loss": 0.4206, "step": 14659 }, { "epoch": 142.15662650602408, "grad_norm": 11.148087501525879, "learning_rate": 2.8834951456310683e-05, "loss": 0.4612, "step": 14660 }, { "epoch": 142.16626506024096, "grad_norm": 9.32888412475586, "learning_rate": 2.8830097087378644e-05, "loss": 0.2739, "step": 14661 }, { "epoch": 142.17590361445784, "grad_norm": 16.122655868530273, "learning_rate": 2.8825242718446605e-05, "loss": 0.2159, "step": 14662 }, { "epoch": 142.1855421686747, "grad_norm": 7.188044548034668, "learning_rate": 2.8820388349514566e-05, "loss": 0.2169, "step": 14663 }, { "epoch": 142.19518072289156, "grad_norm": 14.145928382873535, "learning_rate": 2.8815533980582527e-05, "loss": 0.1281, "step": 14664 }, { "epoch": 142.20481927710844, "grad_norm": 18.840513229370117, "learning_rate": 2.8810679611650488e-05, "loss": 0.4614, "step": 14665 }, { "epoch": 142.2144578313253, "grad_norm": 4.211716175079346, "learning_rate": 2.880582524271845e-05, "loss": 0.2279, "step": 14666 }, { "epoch": 142.22409638554217, "grad_norm": 8.281523704528809, "learning_rate": 2.880097087378641e-05, "loss": 0.161, "step": 14667 }, { "epoch": 142.23373493975905, "grad_norm": 4.7886457443237305, "learning_rate": 2.879611650485437e-05, "loss": 0.317, "step": 14668 }, { "epoch": 142.2433734939759, "grad_norm": 12.2748384475708, "learning_rate": 2.879126213592233e-05, "loss": 0.2057, "step": 14669 }, { "epoch": 142.25301204819277, "grad_norm": 20.50712013244629, "learning_rate": 2.878640776699029e-05, "loss": 0.3728, "step": 14670 }, { "epoch": 142.26265060240965, "grad_norm": 4.82244348526001, "learning_rate": 2.8781553398058252e-05, "loss": 0.3392, "step": 14671 }, { "epoch": 142.2722891566265, "grad_norm": 14.437767028808594, "learning_rate": 2.8776699029126213e-05, "loss": 0.2376, "step": 14672 }, { "epoch": 142.28192771084338, "grad_norm": 22.122995376586914, "learning_rate": 2.8771844660194174e-05, "loss": 0.251, "step": 14673 }, { "epoch": 142.29156626506025, "grad_norm": 15.018282890319824, "learning_rate": 2.8766990291262135e-05, "loss": 0.305, "step": 14674 }, { "epoch": 142.3012048192771, "grad_norm": 11.78139591217041, "learning_rate": 2.8762135922330096e-05, "loss": 0.2833, "step": 14675 }, { "epoch": 142.31084337349398, "grad_norm": 12.256669998168945, "learning_rate": 2.8757281553398057e-05, "loss": 0.3324, "step": 14676 }, { "epoch": 142.32048192771083, "grad_norm": 5.050704479217529, "learning_rate": 2.875242718446602e-05, "loss": 0.1733, "step": 14677 }, { "epoch": 142.3301204819277, "grad_norm": 37.31298828125, "learning_rate": 2.874757281553398e-05, "loss": 0.1994, "step": 14678 }, { "epoch": 142.33975903614459, "grad_norm": 14.508121490478516, "learning_rate": 2.874271844660194e-05, "loss": 0.1978, "step": 14679 }, { "epoch": 142.34939759036143, "grad_norm": 10.355368614196777, "learning_rate": 2.8737864077669902e-05, "loss": 0.3697, "step": 14680 }, { "epoch": 142.3590361445783, "grad_norm": 11.543469429016113, "learning_rate": 2.8733009708737867e-05, "loss": 0.2042, "step": 14681 }, { "epoch": 142.3686746987952, "grad_norm": 8.352606773376465, "learning_rate": 2.8728155339805828e-05, "loss": 0.2215, "step": 14682 }, { "epoch": 142.37831325301204, "grad_norm": 22.02928352355957, "learning_rate": 2.872330097087379e-05, "loss": 0.4254, "step": 14683 }, { "epoch": 142.38795180722892, "grad_norm": 15.550063133239746, "learning_rate": 2.871844660194175e-05, "loss": 0.3864, "step": 14684 }, { "epoch": 142.3975903614458, "grad_norm": 3.1310243606567383, "learning_rate": 2.871359223300971e-05, "loss": 0.1261, "step": 14685 }, { "epoch": 142.40722891566264, "grad_norm": 13.946268081665039, "learning_rate": 2.8708737864077672e-05, "loss": 0.2284, "step": 14686 }, { "epoch": 142.41686746987952, "grad_norm": 8.530259132385254, "learning_rate": 2.8703883495145633e-05, "loss": 0.2138, "step": 14687 }, { "epoch": 142.4265060240964, "grad_norm": 4.98291015625, "learning_rate": 2.8699029126213595e-05, "loss": 0.4344, "step": 14688 }, { "epoch": 142.43614457831325, "grad_norm": 12.6509370803833, "learning_rate": 2.8694174757281556e-05, "loss": 0.3267, "step": 14689 }, { "epoch": 142.44578313253012, "grad_norm": 8.334228515625, "learning_rate": 2.8689320388349517e-05, "loss": 0.1217, "step": 14690 }, { "epoch": 142.455421686747, "grad_norm": 8.173662185668945, "learning_rate": 2.8684466019417478e-05, "loss": 0.1657, "step": 14691 }, { "epoch": 142.46506024096385, "grad_norm": 3.3618345260620117, "learning_rate": 2.867961165048544e-05, "loss": 0.1119, "step": 14692 }, { "epoch": 142.47469879518073, "grad_norm": 12.032353401184082, "learning_rate": 2.86747572815534e-05, "loss": 0.2658, "step": 14693 }, { "epoch": 142.48433734939758, "grad_norm": 22.557527542114258, "learning_rate": 2.866990291262136e-05, "loss": 0.2975, "step": 14694 }, { "epoch": 142.49397590361446, "grad_norm": 18.41954231262207, "learning_rate": 2.8665048543689323e-05, "loss": 0.2303, "step": 14695 }, { "epoch": 142.50361445783133, "grad_norm": 16.573110580444336, "learning_rate": 2.8660194174757284e-05, "loss": 0.2268, "step": 14696 }, { "epoch": 142.51325301204818, "grad_norm": 15.673212051391602, "learning_rate": 2.865533980582524e-05, "loss": 0.3903, "step": 14697 }, { "epoch": 142.52289156626506, "grad_norm": 19.494447708129883, "learning_rate": 2.8650485436893203e-05, "loss": 0.3516, "step": 14698 }, { "epoch": 142.53253012048194, "grad_norm": 3.182950973510742, "learning_rate": 2.8645631067961164e-05, "loss": 0.1942, "step": 14699 }, { "epoch": 142.5421686746988, "grad_norm": 35.15137481689453, "learning_rate": 2.8640776699029125e-05, "loss": 0.3193, "step": 14700 }, { "epoch": 142.55180722891566, "grad_norm": 24.48059844970703, "learning_rate": 2.8635922330097086e-05, "loss": 0.3605, "step": 14701 }, { "epoch": 142.56144578313254, "grad_norm": 18.56749153137207, "learning_rate": 2.8631067961165047e-05, "loss": 0.4072, "step": 14702 }, { "epoch": 142.5710843373494, "grad_norm": 20.50682830810547, "learning_rate": 2.862621359223301e-05, "loss": 0.3721, "step": 14703 }, { "epoch": 142.58072289156627, "grad_norm": 7.5098490715026855, "learning_rate": 2.862135922330097e-05, "loss": 0.1991, "step": 14704 }, { "epoch": 142.59036144578315, "grad_norm": 11.284284591674805, "learning_rate": 2.861650485436893e-05, "loss": 0.1232, "step": 14705 }, { "epoch": 142.6, "grad_norm": 8.75335693359375, "learning_rate": 2.86116504854369e-05, "loss": 0.1981, "step": 14706 }, { "epoch": 142.60963855421687, "grad_norm": 24.254947662353516, "learning_rate": 2.860679611650486e-05, "loss": 0.239, "step": 14707 }, { "epoch": 142.61927710843372, "grad_norm": 24.966110229492188, "learning_rate": 2.8601941747572818e-05, "loss": 0.2457, "step": 14708 }, { "epoch": 142.6289156626506, "grad_norm": 5.341765403747559, "learning_rate": 2.859708737864078e-05, "loss": 0.2643, "step": 14709 }, { "epoch": 142.63855421686748, "grad_norm": 8.557881355285645, "learning_rate": 2.859223300970874e-05, "loss": 0.2434, "step": 14710 }, { "epoch": 142.64819277108433, "grad_norm": 8.052094459533691, "learning_rate": 2.85873786407767e-05, "loss": 0.2876, "step": 14711 }, { "epoch": 142.6578313253012, "grad_norm": 17.547056198120117, "learning_rate": 2.8582524271844662e-05, "loss": 0.3886, "step": 14712 }, { "epoch": 142.66746987951808, "grad_norm": 9.37123966217041, "learning_rate": 2.8577669902912623e-05, "loss": 0.5031, "step": 14713 }, { "epoch": 142.67710843373493, "grad_norm": 10.985239028930664, "learning_rate": 2.8572815533980584e-05, "loss": 0.167, "step": 14714 }, { "epoch": 142.6867469879518, "grad_norm": 12.620338439941406, "learning_rate": 2.8567961165048546e-05, "loss": 0.2466, "step": 14715 }, { "epoch": 142.69638554216868, "grad_norm": 7.117542743682861, "learning_rate": 2.8563106796116507e-05, "loss": 0.2477, "step": 14716 }, { "epoch": 142.70602409638553, "grad_norm": 15.263289451599121, "learning_rate": 2.8558252427184468e-05, "loss": 0.3922, "step": 14717 }, { "epoch": 142.7156626506024, "grad_norm": 5.408596992492676, "learning_rate": 2.855339805825243e-05, "loss": 0.0886, "step": 14718 }, { "epoch": 142.7253012048193, "grad_norm": 26.413698196411133, "learning_rate": 2.854854368932039e-05, "loss": 0.3786, "step": 14719 }, { "epoch": 142.73493975903614, "grad_norm": 12.483874320983887, "learning_rate": 2.854368932038835e-05, "loss": 0.2944, "step": 14720 }, { "epoch": 142.74457831325302, "grad_norm": 17.16220474243164, "learning_rate": 2.8538834951456312e-05, "loss": 0.2638, "step": 14721 }, { "epoch": 142.7542168674699, "grad_norm": 8.939002990722656, "learning_rate": 2.8533980582524274e-05, "loss": 0.1864, "step": 14722 }, { "epoch": 142.76385542168674, "grad_norm": 17.213293075561523, "learning_rate": 2.8529126213592235e-05, "loss": 0.2393, "step": 14723 }, { "epoch": 142.77349397590362, "grad_norm": 4.240337371826172, "learning_rate": 2.8524271844660196e-05, "loss": 0.1291, "step": 14724 }, { "epoch": 142.78313253012047, "grad_norm": 14.154228210449219, "learning_rate": 2.8519417475728154e-05, "loss": 0.3625, "step": 14725 }, { "epoch": 142.79277108433735, "grad_norm": 9.043649673461914, "learning_rate": 2.8514563106796115e-05, "loss": 0.2685, "step": 14726 }, { "epoch": 142.80240963855422, "grad_norm": 3.5556578636169434, "learning_rate": 2.8509708737864076e-05, "loss": 0.1137, "step": 14727 }, { "epoch": 142.81204819277107, "grad_norm": 10.815320014953613, "learning_rate": 2.8504854368932037e-05, "loss": 0.2701, "step": 14728 }, { "epoch": 142.82168674698795, "grad_norm": 16.49425506591797, "learning_rate": 2.8499999999999998e-05, "loss": 0.058, "step": 14729 }, { "epoch": 142.83132530120483, "grad_norm": 10.115519523620605, "learning_rate": 2.849514563106796e-05, "loss": 0.2391, "step": 14730 }, { "epoch": 142.84096385542168, "grad_norm": 5.5268235206604, "learning_rate": 2.8490291262135927e-05, "loss": 0.2653, "step": 14731 }, { "epoch": 142.85060240963855, "grad_norm": 51.04685592651367, "learning_rate": 2.848543689320389e-05, "loss": 0.1904, "step": 14732 }, { "epoch": 142.86024096385543, "grad_norm": 26.774999618530273, "learning_rate": 2.848058252427185e-05, "loss": 0.1327, "step": 14733 }, { "epoch": 142.86987951807228, "grad_norm": 31.988014221191406, "learning_rate": 2.847572815533981e-05, "loss": 0.2325, "step": 14734 }, { "epoch": 142.87951807228916, "grad_norm": 9.93351936340332, "learning_rate": 2.8470873786407772e-05, "loss": 0.2733, "step": 14735 }, { "epoch": 142.88915662650604, "grad_norm": 15.50308895111084, "learning_rate": 2.846601941747573e-05, "loss": 0.1345, "step": 14736 }, { "epoch": 142.89879518072289, "grad_norm": 11.147505760192871, "learning_rate": 2.846116504854369e-05, "loss": 0.2286, "step": 14737 }, { "epoch": 142.90843373493976, "grad_norm": 27.37766456604004, "learning_rate": 2.8456310679611652e-05, "loss": 0.411, "step": 14738 }, { "epoch": 142.91807228915664, "grad_norm": 13.927596092224121, "learning_rate": 2.8451456310679613e-05, "loss": 0.3855, "step": 14739 }, { "epoch": 142.9277108433735, "grad_norm": 6.915804862976074, "learning_rate": 2.8446601941747574e-05, "loss": 0.3055, "step": 14740 }, { "epoch": 142.93734939759037, "grad_norm": 12.258910179138184, "learning_rate": 2.8441747572815535e-05, "loss": 0.2535, "step": 14741 }, { "epoch": 142.94698795180722, "grad_norm": 19.454174041748047, "learning_rate": 2.8436893203883496e-05, "loss": 0.4172, "step": 14742 }, { "epoch": 142.9566265060241, "grad_norm": 9.355405807495117, "learning_rate": 2.8432038834951458e-05, "loss": 0.1615, "step": 14743 }, { "epoch": 142.96626506024097, "grad_norm": 11.992511749267578, "learning_rate": 2.842718446601942e-05, "loss": 0.3413, "step": 14744 }, { "epoch": 142.97590361445782, "grad_norm": 12.337446212768555, "learning_rate": 2.842233009708738e-05, "loss": 0.1921, "step": 14745 }, { "epoch": 142.9855421686747, "grad_norm": 18.035337448120117, "learning_rate": 2.841747572815534e-05, "loss": 0.3174, "step": 14746 }, { "epoch": 143.00120481927712, "grad_norm": 11.411433219909668, "learning_rate": 2.8412621359223302e-05, "loss": 0.2028, "step": 14747 }, { "epoch": 143.01084337349397, "grad_norm": 12.226619720458984, "learning_rate": 2.8407766990291263e-05, "loss": 0.2314, "step": 14748 }, { "epoch": 143.02048192771085, "grad_norm": 24.74335479736328, "learning_rate": 2.8402912621359224e-05, "loss": 0.3487, "step": 14749 }, { "epoch": 143.03012048192772, "grad_norm": 9.398396492004395, "learning_rate": 2.8398058252427186e-05, "loss": 0.2026, "step": 14750 }, { "epoch": 143.03975903614457, "grad_norm": 12.108773231506348, "learning_rate": 2.8393203883495147e-05, "loss": 0.1703, "step": 14751 }, { "epoch": 143.04939759036145, "grad_norm": 5.844992160797119, "learning_rate": 2.8388349514563105e-05, "loss": 0.1819, "step": 14752 }, { "epoch": 143.0590361445783, "grad_norm": 7.67632532119751, "learning_rate": 2.8383495145631066e-05, "loss": 0.3124, "step": 14753 }, { "epoch": 143.06867469879518, "grad_norm": 10.48837947845459, "learning_rate": 2.8378640776699027e-05, "loss": 0.0793, "step": 14754 }, { "epoch": 143.07831325301206, "grad_norm": 11.994542121887207, "learning_rate": 2.8373786407766988e-05, "loss": 0.2879, "step": 14755 }, { "epoch": 143.0879518072289, "grad_norm": 10.6730375289917, "learning_rate": 2.8368932038834956e-05, "loss": 0.2114, "step": 14756 }, { "epoch": 143.09759036144578, "grad_norm": 9.914703369140625, "learning_rate": 2.8364077669902917e-05, "loss": 0.0972, "step": 14757 }, { "epoch": 143.10722891566266, "grad_norm": 9.06778335571289, "learning_rate": 2.8359223300970878e-05, "loss": 0.2552, "step": 14758 }, { "epoch": 143.1168674698795, "grad_norm": 8.762157440185547, "learning_rate": 2.835436893203884e-05, "loss": 0.1995, "step": 14759 }, { "epoch": 143.1265060240964, "grad_norm": 8.863994598388672, "learning_rate": 2.83495145631068e-05, "loss": 0.1724, "step": 14760 }, { "epoch": 143.13614457831326, "grad_norm": 8.044168472290039, "learning_rate": 2.834466019417476e-05, "loss": 0.3329, "step": 14761 }, { "epoch": 143.1457831325301, "grad_norm": 15.777074813842773, "learning_rate": 2.8339805825242723e-05, "loss": 0.3478, "step": 14762 }, { "epoch": 143.155421686747, "grad_norm": 8.16347885131836, "learning_rate": 2.833495145631068e-05, "loss": 0.2288, "step": 14763 }, { "epoch": 143.16506024096387, "grad_norm": 11.884389877319336, "learning_rate": 2.833009708737864e-05, "loss": 0.1749, "step": 14764 }, { "epoch": 143.17469879518072, "grad_norm": 11.593278884887695, "learning_rate": 2.8325242718446603e-05, "loss": 0.2275, "step": 14765 }, { "epoch": 143.1843373493976, "grad_norm": 12.79481315612793, "learning_rate": 2.8320388349514564e-05, "loss": 0.2267, "step": 14766 }, { "epoch": 143.19397590361444, "grad_norm": 19.38704490661621, "learning_rate": 2.8315533980582525e-05, "loss": 0.4014, "step": 14767 }, { "epoch": 143.20361445783132, "grad_norm": 9.728660583496094, "learning_rate": 2.8310679611650486e-05, "loss": 0.4846, "step": 14768 }, { "epoch": 143.2132530120482, "grad_norm": 11.047502517700195, "learning_rate": 2.8305825242718447e-05, "loss": 0.3395, "step": 14769 }, { "epoch": 143.22289156626505, "grad_norm": 9.90019416809082, "learning_rate": 2.830097087378641e-05, "loss": 0.2058, "step": 14770 }, { "epoch": 143.23253012048193, "grad_norm": 11.427998542785645, "learning_rate": 2.829611650485437e-05, "loss": 0.3915, "step": 14771 }, { "epoch": 143.2421686746988, "grad_norm": 7.821991443634033, "learning_rate": 2.829126213592233e-05, "loss": 0.2362, "step": 14772 }, { "epoch": 143.25180722891565, "grad_norm": 8.389548301696777, "learning_rate": 2.8286407766990292e-05, "loss": 0.4404, "step": 14773 }, { "epoch": 143.26144578313253, "grad_norm": 7.916322708129883, "learning_rate": 2.8281553398058253e-05, "loss": 0.3594, "step": 14774 }, { "epoch": 143.2710843373494, "grad_norm": 8.537790298461914, "learning_rate": 2.8276699029126214e-05, "loss": 0.2064, "step": 14775 }, { "epoch": 143.28072289156626, "grad_norm": 8.849549293518066, "learning_rate": 2.8271844660194175e-05, "loss": 0.2948, "step": 14776 }, { "epoch": 143.29036144578313, "grad_norm": 7.265552520751953, "learning_rate": 2.8266990291262137e-05, "loss": 0.2179, "step": 14777 }, { "epoch": 143.3, "grad_norm": 44.35464096069336, "learning_rate": 2.8262135922330098e-05, "loss": 0.4604, "step": 14778 }, { "epoch": 143.30963855421686, "grad_norm": 29.445541381835938, "learning_rate": 2.825728155339806e-05, "loss": 0.3396, "step": 14779 }, { "epoch": 143.31927710843374, "grad_norm": 5.417160987854004, "learning_rate": 2.8252427184466017e-05, "loss": 0.1646, "step": 14780 }, { "epoch": 143.32891566265062, "grad_norm": 14.31623649597168, "learning_rate": 2.8247572815533978e-05, "loss": 0.2482, "step": 14781 }, { "epoch": 143.33855421686746, "grad_norm": 13.885478973388672, "learning_rate": 2.8242718446601946e-05, "loss": 0.2808, "step": 14782 }, { "epoch": 143.34819277108434, "grad_norm": 11.693696975708008, "learning_rate": 2.8237864077669907e-05, "loss": 0.2244, "step": 14783 }, { "epoch": 143.3578313253012, "grad_norm": 13.383441925048828, "learning_rate": 2.8233009708737868e-05, "loss": 0.3199, "step": 14784 }, { "epoch": 143.36746987951807, "grad_norm": 7.204343318939209, "learning_rate": 2.822815533980583e-05, "loss": 0.2253, "step": 14785 }, { "epoch": 143.37710843373495, "grad_norm": 7.153209686279297, "learning_rate": 2.822330097087379e-05, "loss": 0.3111, "step": 14786 }, { "epoch": 143.3867469879518, "grad_norm": 5.5692362785339355, "learning_rate": 2.821844660194175e-05, "loss": 0.2359, "step": 14787 }, { "epoch": 143.39638554216867, "grad_norm": 20.18194580078125, "learning_rate": 2.8213592233009713e-05, "loss": 0.3677, "step": 14788 }, { "epoch": 143.40602409638555, "grad_norm": 10.772192001342773, "learning_rate": 2.8208737864077674e-05, "loss": 0.3496, "step": 14789 }, { "epoch": 143.4156626506024, "grad_norm": 15.701167106628418, "learning_rate": 2.8203883495145635e-05, "loss": 0.4449, "step": 14790 }, { "epoch": 143.42530120481928, "grad_norm": 9.804099082946777, "learning_rate": 2.8199029126213593e-05, "loss": 0.3445, "step": 14791 }, { "epoch": 143.43493975903615, "grad_norm": 9.28146743774414, "learning_rate": 2.8194174757281554e-05, "loss": 0.3318, "step": 14792 }, { "epoch": 143.444578313253, "grad_norm": 9.151460647583008, "learning_rate": 2.8189320388349515e-05, "loss": 0.4635, "step": 14793 }, { "epoch": 143.45421686746988, "grad_norm": 8.942721366882324, "learning_rate": 2.8184466019417476e-05, "loss": 0.3686, "step": 14794 }, { "epoch": 143.46385542168676, "grad_norm": 13.911243438720703, "learning_rate": 2.8179611650485437e-05, "loss": 0.2695, "step": 14795 }, { "epoch": 143.4734939759036, "grad_norm": 11.07198429107666, "learning_rate": 2.81747572815534e-05, "loss": 0.4432, "step": 14796 }, { "epoch": 143.48313253012049, "grad_norm": 4.773599624633789, "learning_rate": 2.816990291262136e-05, "loss": 0.2629, "step": 14797 }, { "epoch": 143.49277108433736, "grad_norm": 14.906808853149414, "learning_rate": 2.816504854368932e-05, "loss": 0.3364, "step": 14798 }, { "epoch": 143.5024096385542, "grad_norm": 5.78049898147583, "learning_rate": 2.8160194174757282e-05, "loss": 0.2312, "step": 14799 }, { "epoch": 143.5120481927711, "grad_norm": 3.4801745414733887, "learning_rate": 2.8155339805825243e-05, "loss": 0.1013, "step": 14800 }, { "epoch": 143.52168674698794, "grad_norm": 6.18624210357666, "learning_rate": 2.8150485436893204e-05, "loss": 0.164, "step": 14801 }, { "epoch": 143.53132530120482, "grad_norm": 22.449962615966797, "learning_rate": 2.8145631067961165e-05, "loss": 0.3512, "step": 14802 }, { "epoch": 143.5409638554217, "grad_norm": 48.145416259765625, "learning_rate": 2.8140776699029126e-05, "loss": 0.3273, "step": 14803 }, { "epoch": 143.55060240963854, "grad_norm": 12.086315155029297, "learning_rate": 2.8135922330097087e-05, "loss": 0.3363, "step": 14804 }, { "epoch": 143.56024096385542, "grad_norm": 11.745302200317383, "learning_rate": 2.813106796116505e-05, "loss": 0.1891, "step": 14805 }, { "epoch": 143.5698795180723, "grad_norm": 14.902016639709473, "learning_rate": 2.812621359223301e-05, "loss": 0.3231, "step": 14806 }, { "epoch": 143.57951807228915, "grad_norm": 17.55702018737793, "learning_rate": 2.8121359223300974e-05, "loss": 0.4055, "step": 14807 }, { "epoch": 143.58915662650602, "grad_norm": 10.015786170959473, "learning_rate": 2.8116504854368935e-05, "loss": 0.2896, "step": 14808 }, { "epoch": 143.5987951807229, "grad_norm": 7.564792633056641, "learning_rate": 2.8111650485436897e-05, "loss": 0.2738, "step": 14809 }, { "epoch": 143.60843373493975, "grad_norm": 7.008524417877197, "learning_rate": 2.8106796116504858e-05, "loss": 0.139, "step": 14810 }, { "epoch": 143.61807228915663, "grad_norm": 9.878936767578125, "learning_rate": 2.810194174757282e-05, "loss": 0.2057, "step": 14811 }, { "epoch": 143.6277108433735, "grad_norm": 9.009979248046875, "learning_rate": 2.809708737864078e-05, "loss": 0.2918, "step": 14812 }, { "epoch": 143.63734939759036, "grad_norm": 8.67799186706543, "learning_rate": 2.809223300970874e-05, "loss": 0.1766, "step": 14813 }, { "epoch": 143.64698795180723, "grad_norm": 12.739233016967773, "learning_rate": 2.8087378640776702e-05, "loss": 0.3314, "step": 14814 }, { "epoch": 143.65662650602408, "grad_norm": 16.930238723754883, "learning_rate": 2.8082524271844663e-05, "loss": 0.2707, "step": 14815 }, { "epoch": 143.66626506024096, "grad_norm": 13.024776458740234, "learning_rate": 2.8077669902912625e-05, "loss": 0.2539, "step": 14816 }, { "epoch": 143.67590361445784, "grad_norm": 11.178071975708008, "learning_rate": 2.8072815533980586e-05, "loss": 0.4602, "step": 14817 }, { "epoch": 143.6855421686747, "grad_norm": 7.330697536468506, "learning_rate": 2.8067961165048544e-05, "loss": 0.265, "step": 14818 }, { "epoch": 143.69518072289156, "grad_norm": 11.504839897155762, "learning_rate": 2.8063106796116505e-05, "loss": 0.1495, "step": 14819 }, { "epoch": 143.70481927710844, "grad_norm": 11.361786842346191, "learning_rate": 2.8058252427184466e-05, "loss": 0.3328, "step": 14820 }, { "epoch": 143.7144578313253, "grad_norm": 5.9222612380981445, "learning_rate": 2.8053398058252427e-05, "loss": 0.1337, "step": 14821 }, { "epoch": 143.72409638554217, "grad_norm": 10.316856384277344, "learning_rate": 2.8048543689320388e-05, "loss": 0.3566, "step": 14822 }, { "epoch": 143.73373493975905, "grad_norm": 9.579297065734863, "learning_rate": 2.804368932038835e-05, "loss": 0.214, "step": 14823 }, { "epoch": 143.7433734939759, "grad_norm": 10.040253639221191, "learning_rate": 2.803883495145631e-05, "loss": 0.2488, "step": 14824 }, { "epoch": 143.75301204819277, "grad_norm": 36.73566818237305, "learning_rate": 2.803398058252427e-05, "loss": 0.4385, "step": 14825 }, { "epoch": 143.76265060240965, "grad_norm": 28.149255752563477, "learning_rate": 2.8029126213592233e-05, "loss": 0.2652, "step": 14826 }, { "epoch": 143.7722891566265, "grad_norm": 12.874698638916016, "learning_rate": 2.8024271844660194e-05, "loss": 0.431, "step": 14827 }, { "epoch": 143.78192771084338, "grad_norm": 11.289759635925293, "learning_rate": 2.8019417475728155e-05, "loss": 0.2931, "step": 14828 }, { "epoch": 143.79156626506025, "grad_norm": 21.43059539794922, "learning_rate": 2.8014563106796116e-05, "loss": 0.1774, "step": 14829 }, { "epoch": 143.8012048192771, "grad_norm": 9.216381072998047, "learning_rate": 2.8009708737864077e-05, "loss": 0.3484, "step": 14830 }, { "epoch": 143.81084337349398, "grad_norm": 14.988665580749512, "learning_rate": 2.800485436893204e-05, "loss": 0.3191, "step": 14831 }, { "epoch": 143.82048192771083, "grad_norm": 6.856958389282227, "learning_rate": 2.8000000000000003e-05, "loss": 0.1355, "step": 14832 }, { "epoch": 143.8301204819277, "grad_norm": 13.119081497192383, "learning_rate": 2.7995145631067964e-05, "loss": 0.2898, "step": 14833 }, { "epoch": 143.83975903614459, "grad_norm": 19.319520950317383, "learning_rate": 2.7990291262135925e-05, "loss": 0.3184, "step": 14834 }, { "epoch": 143.84939759036143, "grad_norm": 10.064790725708008, "learning_rate": 2.7985436893203886e-05, "loss": 0.2231, "step": 14835 }, { "epoch": 143.8590361445783, "grad_norm": 19.423259735107422, "learning_rate": 2.7980582524271848e-05, "loss": 0.2632, "step": 14836 }, { "epoch": 143.8686746987952, "grad_norm": 48.80384063720703, "learning_rate": 2.797572815533981e-05, "loss": 0.3913, "step": 14837 }, { "epoch": 143.87831325301204, "grad_norm": 7.075827121734619, "learning_rate": 2.797087378640777e-05, "loss": 0.1578, "step": 14838 }, { "epoch": 143.88795180722892, "grad_norm": 8.262703895568848, "learning_rate": 2.796601941747573e-05, "loss": 0.1988, "step": 14839 }, { "epoch": 143.8975903614458, "grad_norm": 7.66096305847168, "learning_rate": 2.7961165048543692e-05, "loss": 0.2265, "step": 14840 }, { "epoch": 143.90722891566264, "grad_norm": 12.340601921081543, "learning_rate": 2.7956310679611653e-05, "loss": 0.3025, "step": 14841 }, { "epoch": 143.91686746987952, "grad_norm": 13.832282066345215, "learning_rate": 2.7951456310679614e-05, "loss": 0.7171, "step": 14842 }, { "epoch": 143.9265060240964, "grad_norm": 10.046806335449219, "learning_rate": 2.7946601941747576e-05, "loss": 0.2209, "step": 14843 }, { "epoch": 143.93614457831325, "grad_norm": 24.0622615814209, "learning_rate": 2.7941747572815537e-05, "loss": 0.1337, "step": 14844 }, { "epoch": 143.94578313253012, "grad_norm": 9.642755508422852, "learning_rate": 2.7936893203883498e-05, "loss": 0.3821, "step": 14845 }, { "epoch": 143.955421686747, "grad_norm": 9.760741233825684, "learning_rate": 2.7932038834951456e-05, "loss": 0.2571, "step": 14846 }, { "epoch": 143.96506024096385, "grad_norm": 5.597349166870117, "learning_rate": 2.7927184466019417e-05, "loss": 0.1334, "step": 14847 }, { "epoch": 143.97469879518073, "grad_norm": 10.275288581848145, "learning_rate": 2.7922330097087378e-05, "loss": 0.1497, "step": 14848 }, { "epoch": 143.98433734939758, "grad_norm": 8.577414512634277, "learning_rate": 2.791747572815534e-05, "loss": 0.2382, "step": 14849 }, { "epoch": 143.99397590361446, "grad_norm": 15.441156387329102, "learning_rate": 2.79126213592233e-05, "loss": 0.3922, "step": 14850 }, { "epoch": 144.00963855421688, "grad_norm": 5.338624000549316, "learning_rate": 2.790776699029126e-05, "loss": 0.1471, "step": 14851 }, { "epoch": 144.01927710843373, "grad_norm": 13.758708000183105, "learning_rate": 2.7902912621359222e-05, "loss": 0.1452, "step": 14852 }, { "epoch": 144.0289156626506, "grad_norm": 8.728509902954102, "learning_rate": 2.7898058252427184e-05, "loss": 0.2824, "step": 14853 }, { "epoch": 144.03855421686748, "grad_norm": 11.531515121459961, "learning_rate": 2.7893203883495145e-05, "loss": 0.2847, "step": 14854 }, { "epoch": 144.04819277108433, "grad_norm": 27.314350128173828, "learning_rate": 2.7888349514563106e-05, "loss": 0.4159, "step": 14855 }, { "epoch": 144.0578313253012, "grad_norm": 10.673354148864746, "learning_rate": 2.7883495145631067e-05, "loss": 0.1841, "step": 14856 }, { "epoch": 144.0674698795181, "grad_norm": 6.368699073791504, "learning_rate": 2.787864077669903e-05, "loss": 0.2029, "step": 14857 }, { "epoch": 144.07710843373494, "grad_norm": 6.273125171661377, "learning_rate": 2.7873786407766993e-05, "loss": 0.1282, "step": 14858 }, { "epoch": 144.0867469879518, "grad_norm": 10.299687385559082, "learning_rate": 2.7868932038834954e-05, "loss": 0.2894, "step": 14859 }, { "epoch": 144.09638554216866, "grad_norm": 4.485106468200684, "learning_rate": 2.7864077669902915e-05, "loss": 0.0963, "step": 14860 }, { "epoch": 144.10602409638554, "grad_norm": 6.986329555511475, "learning_rate": 2.7859223300970876e-05, "loss": 0.1337, "step": 14861 }, { "epoch": 144.11566265060242, "grad_norm": 14.361699104309082, "learning_rate": 2.7854368932038837e-05, "loss": 0.3105, "step": 14862 }, { "epoch": 144.12530120481927, "grad_norm": 13.935672760009766, "learning_rate": 2.78495145631068e-05, "loss": 0.2816, "step": 14863 }, { "epoch": 144.13493975903614, "grad_norm": 7.721480846405029, "learning_rate": 2.784466019417476e-05, "loss": 0.283, "step": 14864 }, { "epoch": 144.14457831325302, "grad_norm": 4.902124881744385, "learning_rate": 2.783980582524272e-05, "loss": 0.1008, "step": 14865 }, { "epoch": 144.15421686746987, "grad_norm": 10.941447257995605, "learning_rate": 2.7834951456310682e-05, "loss": 0.2102, "step": 14866 }, { "epoch": 144.16385542168675, "grad_norm": 8.112863540649414, "learning_rate": 2.7830097087378643e-05, "loss": 0.5745, "step": 14867 }, { "epoch": 144.17349397590363, "grad_norm": 11.099955558776855, "learning_rate": 2.7825242718446604e-05, "loss": 0.3285, "step": 14868 }, { "epoch": 144.18313253012047, "grad_norm": 112.01397705078125, "learning_rate": 2.7820388349514565e-05, "loss": 0.2671, "step": 14869 }, { "epoch": 144.19277108433735, "grad_norm": 12.12370491027832, "learning_rate": 2.7815533980582527e-05, "loss": 0.2421, "step": 14870 }, { "epoch": 144.20240963855423, "grad_norm": 21.26100730895996, "learning_rate": 2.7810679611650488e-05, "loss": 0.4142, "step": 14871 }, { "epoch": 144.21204819277108, "grad_norm": 6.651111602783203, "learning_rate": 2.780582524271845e-05, "loss": 0.1942, "step": 14872 }, { "epoch": 144.22168674698796, "grad_norm": 10.377091407775879, "learning_rate": 2.780097087378641e-05, "loss": 0.3552, "step": 14873 }, { "epoch": 144.2313253012048, "grad_norm": 13.351173400878906, "learning_rate": 2.7796116504854368e-05, "loss": 0.3777, "step": 14874 }, { "epoch": 144.24096385542168, "grad_norm": 14.597513198852539, "learning_rate": 2.779126213592233e-05, "loss": 0.3484, "step": 14875 }, { "epoch": 144.25060240963856, "grad_norm": 11.793137550354004, "learning_rate": 2.778640776699029e-05, "loss": 0.3777, "step": 14876 }, { "epoch": 144.2602409638554, "grad_norm": 24.404144287109375, "learning_rate": 2.778155339805825e-05, "loss": 0.1767, "step": 14877 }, { "epoch": 144.2698795180723, "grad_norm": 24.637659072875977, "learning_rate": 2.7776699029126212e-05, "loss": 0.316, "step": 14878 }, { "epoch": 144.27951807228916, "grad_norm": 11.450074195861816, "learning_rate": 2.7771844660194173e-05, "loss": 0.2182, "step": 14879 }, { "epoch": 144.289156626506, "grad_norm": 6.022333145141602, "learning_rate": 2.7766990291262135e-05, "loss": 0.1849, "step": 14880 }, { "epoch": 144.2987951807229, "grad_norm": 17.337718963623047, "learning_rate": 2.7762135922330096e-05, "loss": 0.3029, "step": 14881 }, { "epoch": 144.30843373493977, "grad_norm": 7.984795570373535, "learning_rate": 2.7757281553398064e-05, "loss": 0.1649, "step": 14882 }, { "epoch": 144.31807228915662, "grad_norm": 140.3655242919922, "learning_rate": 2.7752427184466025e-05, "loss": 0.3011, "step": 14883 }, { "epoch": 144.3277108433735, "grad_norm": 23.61284637451172, "learning_rate": 2.7747572815533986e-05, "loss": 0.4535, "step": 14884 }, { "epoch": 144.33734939759037, "grad_norm": 7.945851802825928, "learning_rate": 2.7742718446601944e-05, "loss": 0.1407, "step": 14885 }, { "epoch": 144.34698795180722, "grad_norm": 5.5318684577941895, "learning_rate": 2.7737864077669905e-05, "loss": 0.291, "step": 14886 }, { "epoch": 144.3566265060241, "grad_norm": 19.112773895263672, "learning_rate": 2.7733009708737866e-05, "loss": 0.4647, "step": 14887 }, { "epoch": 144.36626506024098, "grad_norm": 13.308605194091797, "learning_rate": 2.7728155339805827e-05, "loss": 0.2979, "step": 14888 }, { "epoch": 144.37590361445783, "grad_norm": 19.597681045532227, "learning_rate": 2.7723300970873788e-05, "loss": 0.3657, "step": 14889 }, { "epoch": 144.3855421686747, "grad_norm": 7.9715375900268555, "learning_rate": 2.771844660194175e-05, "loss": 0.4061, "step": 14890 }, { "epoch": 144.39518072289155, "grad_norm": 10.725386619567871, "learning_rate": 2.771359223300971e-05, "loss": 0.2846, "step": 14891 }, { "epoch": 144.40481927710843, "grad_norm": 8.811982154846191, "learning_rate": 2.7708737864077672e-05, "loss": 0.3616, "step": 14892 }, { "epoch": 144.4144578313253, "grad_norm": 23.35317611694336, "learning_rate": 2.7703883495145633e-05, "loss": 0.2523, "step": 14893 }, { "epoch": 144.42409638554216, "grad_norm": 11.520476341247559, "learning_rate": 2.7699029126213594e-05, "loss": 0.1825, "step": 14894 }, { "epoch": 144.43373493975903, "grad_norm": 23.57246208190918, "learning_rate": 2.7694174757281555e-05, "loss": 0.321, "step": 14895 }, { "epoch": 144.4433734939759, "grad_norm": 15.443455696105957, "learning_rate": 2.7689320388349516e-05, "loss": 0.1453, "step": 14896 }, { "epoch": 144.45301204819276, "grad_norm": 8.045636177062988, "learning_rate": 2.7684466019417477e-05, "loss": 0.2401, "step": 14897 }, { "epoch": 144.46265060240964, "grad_norm": 11.581850051879883, "learning_rate": 2.767961165048544e-05, "loss": 0.3295, "step": 14898 }, { "epoch": 144.47228915662652, "grad_norm": 11.664548873901367, "learning_rate": 2.76747572815534e-05, "loss": 0.437, "step": 14899 }, { "epoch": 144.48192771084337, "grad_norm": 26.72568130493164, "learning_rate": 2.766990291262136e-05, "loss": 0.5261, "step": 14900 }, { "epoch": 144.49156626506024, "grad_norm": 22.141090393066406, "learning_rate": 2.766504854368932e-05, "loss": 0.4521, "step": 14901 }, { "epoch": 144.50120481927712, "grad_norm": 5.806400299072266, "learning_rate": 2.766019417475728e-05, "loss": 0.2368, "step": 14902 }, { "epoch": 144.51084337349397, "grad_norm": 10.264435768127441, "learning_rate": 2.765533980582524e-05, "loss": 0.1652, "step": 14903 }, { "epoch": 144.52048192771085, "grad_norm": 20.920711517333984, "learning_rate": 2.7650485436893202e-05, "loss": 0.2545, "step": 14904 }, { "epoch": 144.53012048192772, "grad_norm": 8.7725191116333, "learning_rate": 2.7645631067961163e-05, "loss": 0.2471, "step": 14905 }, { "epoch": 144.53975903614457, "grad_norm": 13.793095588684082, "learning_rate": 2.7640776699029124e-05, "loss": 0.3512, "step": 14906 }, { "epoch": 144.54939759036145, "grad_norm": 8.707630157470703, "learning_rate": 2.7635922330097086e-05, "loss": 0.2369, "step": 14907 }, { "epoch": 144.5590361445783, "grad_norm": 10.115696907043457, "learning_rate": 2.7631067961165053e-05, "loss": 0.1722, "step": 14908 }, { "epoch": 144.56867469879518, "grad_norm": 10.5291109085083, "learning_rate": 2.7626213592233015e-05, "loss": 0.39, "step": 14909 }, { "epoch": 144.57831325301206, "grad_norm": 9.681763648986816, "learning_rate": 2.7621359223300976e-05, "loss": 0.1984, "step": 14910 }, { "epoch": 144.5879518072289, "grad_norm": 11.543315887451172, "learning_rate": 2.7616504854368937e-05, "loss": 0.4558, "step": 14911 }, { "epoch": 144.59759036144578, "grad_norm": 14.76685905456543, "learning_rate": 2.7611650485436895e-05, "loss": 0.2966, "step": 14912 }, { "epoch": 144.60722891566266, "grad_norm": 13.285505294799805, "learning_rate": 2.7606796116504856e-05, "loss": 0.4141, "step": 14913 }, { "epoch": 144.6168674698795, "grad_norm": 8.957999229431152, "learning_rate": 2.7601941747572817e-05, "loss": 0.1978, "step": 14914 }, { "epoch": 144.6265060240964, "grad_norm": 11.862312316894531, "learning_rate": 2.7597087378640778e-05, "loss": 0.3001, "step": 14915 }, { "epoch": 144.63614457831326, "grad_norm": 16.007701873779297, "learning_rate": 2.759223300970874e-05, "loss": 0.3502, "step": 14916 }, { "epoch": 144.6457831325301, "grad_norm": 16.527162551879883, "learning_rate": 2.75873786407767e-05, "loss": 0.4889, "step": 14917 }, { "epoch": 144.655421686747, "grad_norm": 10.347540855407715, "learning_rate": 2.758252427184466e-05, "loss": 0.221, "step": 14918 }, { "epoch": 144.66506024096387, "grad_norm": 11.968158721923828, "learning_rate": 2.7577669902912623e-05, "loss": 0.3749, "step": 14919 }, { "epoch": 144.67469879518072, "grad_norm": 13.25279712677002, "learning_rate": 2.7572815533980584e-05, "loss": 0.2661, "step": 14920 }, { "epoch": 144.6843373493976, "grad_norm": 11.513815879821777, "learning_rate": 2.7567961165048545e-05, "loss": 0.2921, "step": 14921 }, { "epoch": 144.69397590361444, "grad_norm": 14.295860290527344, "learning_rate": 2.7563106796116506e-05, "loss": 0.1695, "step": 14922 }, { "epoch": 144.70361445783132, "grad_norm": 8.383825302124023, "learning_rate": 2.7558252427184467e-05, "loss": 0.34, "step": 14923 }, { "epoch": 144.7132530120482, "grad_norm": 6.045267105102539, "learning_rate": 2.755339805825243e-05, "loss": 0.1926, "step": 14924 }, { "epoch": 144.72289156626505, "grad_norm": 4.450978755950928, "learning_rate": 2.754854368932039e-05, "loss": 0.1944, "step": 14925 }, { "epoch": 144.73253012048193, "grad_norm": 19.98192024230957, "learning_rate": 2.754368932038835e-05, "loss": 0.3335, "step": 14926 }, { "epoch": 144.7421686746988, "grad_norm": 7.873019218444824, "learning_rate": 2.7538834951456312e-05, "loss": 0.2747, "step": 14927 }, { "epoch": 144.75180722891565, "grad_norm": 20.82904815673828, "learning_rate": 2.7533980582524273e-05, "loss": 0.3241, "step": 14928 }, { "epoch": 144.76144578313253, "grad_norm": 12.964637756347656, "learning_rate": 2.752912621359223e-05, "loss": 0.4963, "step": 14929 }, { "epoch": 144.7710843373494, "grad_norm": 36.73844528198242, "learning_rate": 2.7524271844660192e-05, "loss": 0.3432, "step": 14930 }, { "epoch": 144.78072289156626, "grad_norm": 6.096310615539551, "learning_rate": 2.7519417475728153e-05, "loss": 0.1923, "step": 14931 }, { "epoch": 144.79036144578313, "grad_norm": 19.58173370361328, "learning_rate": 2.7514563106796114e-05, "loss": 0.3938, "step": 14932 }, { "epoch": 144.8, "grad_norm": 5.261050701141357, "learning_rate": 2.7509708737864082e-05, "loss": 0.1441, "step": 14933 }, { "epoch": 144.80963855421686, "grad_norm": 21.830677032470703, "learning_rate": 2.7504854368932043e-05, "loss": 0.293, "step": 14934 }, { "epoch": 144.81927710843374, "grad_norm": 5.742024898529053, "learning_rate": 2.7500000000000004e-05, "loss": 0.3063, "step": 14935 }, { "epoch": 144.82891566265062, "grad_norm": 9.326760292053223, "learning_rate": 2.7495145631067966e-05, "loss": 0.2376, "step": 14936 }, { "epoch": 144.83855421686746, "grad_norm": 7.503039836883545, "learning_rate": 2.7490291262135927e-05, "loss": 0.374, "step": 14937 }, { "epoch": 144.84819277108434, "grad_norm": 6.002324104309082, "learning_rate": 2.7485436893203888e-05, "loss": 0.1294, "step": 14938 }, { "epoch": 144.8578313253012, "grad_norm": 5.897863388061523, "learning_rate": 2.748058252427185e-05, "loss": 0.1379, "step": 14939 }, { "epoch": 144.86746987951807, "grad_norm": 10.121088981628418, "learning_rate": 2.7475728155339807e-05, "loss": 0.2087, "step": 14940 }, { "epoch": 144.87710843373495, "grad_norm": 12.971344947814941, "learning_rate": 2.7470873786407768e-05, "loss": 0.3959, "step": 14941 }, { "epoch": 144.8867469879518, "grad_norm": 6.019290924072266, "learning_rate": 2.746601941747573e-05, "loss": 0.1407, "step": 14942 }, { "epoch": 144.89638554216867, "grad_norm": 17.85394287109375, "learning_rate": 2.746116504854369e-05, "loss": 0.3775, "step": 14943 }, { "epoch": 144.90602409638555, "grad_norm": 8.873075485229492, "learning_rate": 2.745631067961165e-05, "loss": 0.2132, "step": 14944 }, { "epoch": 144.9156626506024, "grad_norm": 8.370711326599121, "learning_rate": 2.7451456310679612e-05, "loss": 0.3361, "step": 14945 }, { "epoch": 144.92530120481928, "grad_norm": 9.785998344421387, "learning_rate": 2.7446601941747574e-05, "loss": 0.3222, "step": 14946 }, { "epoch": 144.93493975903615, "grad_norm": 14.225855827331543, "learning_rate": 2.7441747572815535e-05, "loss": 0.2767, "step": 14947 }, { "epoch": 144.944578313253, "grad_norm": 5.430484771728516, "learning_rate": 2.7436893203883496e-05, "loss": 0.2323, "step": 14948 }, { "epoch": 144.95421686746988, "grad_norm": 14.867111206054688, "learning_rate": 2.7432038834951457e-05, "loss": 0.2936, "step": 14949 }, { "epoch": 144.96385542168676, "grad_norm": 18.145851135253906, "learning_rate": 2.7427184466019418e-05, "loss": 0.3093, "step": 14950 }, { "epoch": 144.9734939759036, "grad_norm": 16.154756546020508, "learning_rate": 2.742233009708738e-05, "loss": 0.2278, "step": 14951 }, { "epoch": 144.98313253012049, "grad_norm": 19.162019729614258, "learning_rate": 2.741747572815534e-05, "loss": 0.1568, "step": 14952 }, { "epoch": 144.99277108433736, "grad_norm": 21.9689884185791, "learning_rate": 2.74126213592233e-05, "loss": 0.2565, "step": 14953 }, { "epoch": 145.00843373493976, "grad_norm": 14.438902854919434, "learning_rate": 2.7407766990291263e-05, "loss": 0.2496, "step": 14954 }, { "epoch": 145.01807228915663, "grad_norm": 13.118935585021973, "learning_rate": 2.7402912621359224e-05, "loss": 0.4157, "step": 14955 }, { "epoch": 145.02771084337348, "grad_norm": 12.234044075012207, "learning_rate": 2.7398058252427185e-05, "loss": 0.2365, "step": 14956 }, { "epoch": 145.03734939759036, "grad_norm": 25.953338623046875, "learning_rate": 2.7393203883495143e-05, "loss": 0.1904, "step": 14957 }, { "epoch": 145.04698795180724, "grad_norm": 4.3102521896362305, "learning_rate": 2.738834951456311e-05, "loss": 0.2737, "step": 14958 }, { "epoch": 145.0566265060241, "grad_norm": 31.138593673706055, "learning_rate": 2.7383495145631072e-05, "loss": 0.2559, "step": 14959 }, { "epoch": 145.06626506024097, "grad_norm": 2.7447562217712402, "learning_rate": 2.7378640776699033e-05, "loss": 0.2955, "step": 14960 }, { "epoch": 145.07590361445784, "grad_norm": 47.30321502685547, "learning_rate": 2.7373786407766994e-05, "loss": 0.1924, "step": 14961 }, { "epoch": 145.0855421686747, "grad_norm": 5.42222261428833, "learning_rate": 2.7368932038834955e-05, "loss": 0.1665, "step": 14962 }, { "epoch": 145.09518072289157, "grad_norm": 7.83751106262207, "learning_rate": 2.7364077669902916e-05, "loss": 0.2855, "step": 14963 }, { "epoch": 145.10481927710845, "grad_norm": 5.111355781555176, "learning_rate": 2.7359223300970878e-05, "loss": 0.1691, "step": 14964 }, { "epoch": 145.1144578313253, "grad_norm": 28.634767532348633, "learning_rate": 2.735436893203884e-05, "loss": 0.1477, "step": 14965 }, { "epoch": 145.12409638554217, "grad_norm": 37.844512939453125, "learning_rate": 2.73495145631068e-05, "loss": 0.4092, "step": 14966 }, { "epoch": 145.13373493975902, "grad_norm": 14.078699111938477, "learning_rate": 2.7344660194174758e-05, "loss": 0.1871, "step": 14967 }, { "epoch": 145.1433734939759, "grad_norm": 45.75930404663086, "learning_rate": 2.733980582524272e-05, "loss": 0.3431, "step": 14968 }, { "epoch": 145.15301204819278, "grad_norm": 38.64972686767578, "learning_rate": 2.733495145631068e-05, "loss": 0.3806, "step": 14969 }, { "epoch": 145.16265060240963, "grad_norm": 77.2573471069336, "learning_rate": 2.733009708737864e-05, "loss": 0.1238, "step": 14970 }, { "epoch": 145.1722891566265, "grad_norm": 16.876386642456055, "learning_rate": 2.7325242718446602e-05, "loss": 0.2681, "step": 14971 }, { "epoch": 145.18192771084338, "grad_norm": 14.962249755859375, "learning_rate": 2.7320388349514563e-05, "loss": 0.1492, "step": 14972 }, { "epoch": 145.19156626506023, "grad_norm": 11.675040245056152, "learning_rate": 2.7315533980582525e-05, "loss": 0.231, "step": 14973 }, { "epoch": 145.2012048192771, "grad_norm": 11.063437461853027, "learning_rate": 2.7310679611650486e-05, "loss": 0.1759, "step": 14974 }, { "epoch": 145.210843373494, "grad_norm": 27.775867462158203, "learning_rate": 2.7305825242718447e-05, "loss": 0.2291, "step": 14975 }, { "epoch": 145.22048192771084, "grad_norm": 5.604377269744873, "learning_rate": 2.7300970873786408e-05, "loss": 0.0956, "step": 14976 }, { "epoch": 145.2301204819277, "grad_norm": 10.890874862670898, "learning_rate": 2.729611650485437e-05, "loss": 0.1084, "step": 14977 }, { "epoch": 145.2397590361446, "grad_norm": 0.9817727208137512, "learning_rate": 2.729126213592233e-05, "loss": 0.1694, "step": 14978 }, { "epoch": 145.24939759036144, "grad_norm": 23.79451560974121, "learning_rate": 2.728640776699029e-05, "loss": 0.0717, "step": 14979 }, { "epoch": 145.25903614457832, "grad_norm": 2.1665985584259033, "learning_rate": 2.7281553398058253e-05, "loss": 0.3035, "step": 14980 }, { "epoch": 145.2686746987952, "grad_norm": 40.63383865356445, "learning_rate": 2.7276699029126214e-05, "loss": 0.2639, "step": 14981 }, { "epoch": 145.27831325301204, "grad_norm": 21.736738204956055, "learning_rate": 2.7271844660194175e-05, "loss": 0.2129, "step": 14982 }, { "epoch": 145.28795180722892, "grad_norm": 5.108791351318359, "learning_rate": 2.726699029126214e-05, "loss": 0.2433, "step": 14983 }, { "epoch": 145.29759036144577, "grad_norm": 15.336740493774414, "learning_rate": 2.72621359223301e-05, "loss": 0.3065, "step": 14984 }, { "epoch": 145.30722891566265, "grad_norm": 18.44607925415039, "learning_rate": 2.725728155339806e-05, "loss": 0.2591, "step": 14985 }, { "epoch": 145.31686746987953, "grad_norm": 2.859116792678833, "learning_rate": 2.7252427184466023e-05, "loss": 0.2235, "step": 14986 }, { "epoch": 145.32650602409637, "grad_norm": 17.320850372314453, "learning_rate": 2.7247572815533984e-05, "loss": 0.3999, "step": 14987 }, { "epoch": 145.33614457831325, "grad_norm": 14.947772026062012, "learning_rate": 2.7242718446601945e-05, "loss": 0.2855, "step": 14988 }, { "epoch": 145.34578313253013, "grad_norm": 1.5448766946792603, "learning_rate": 2.7237864077669906e-05, "loss": 0.2126, "step": 14989 }, { "epoch": 145.35542168674698, "grad_norm": 12.696026802062988, "learning_rate": 2.7233009708737867e-05, "loss": 0.2158, "step": 14990 }, { "epoch": 145.36506024096386, "grad_norm": 21.513164520263672, "learning_rate": 2.722815533980583e-05, "loss": 0.19, "step": 14991 }, { "epoch": 145.37469879518073, "grad_norm": 29.44028091430664, "learning_rate": 2.722330097087379e-05, "loss": 0.3173, "step": 14992 }, { "epoch": 145.38433734939758, "grad_norm": 11.459086418151855, "learning_rate": 2.721844660194175e-05, "loss": 0.3756, "step": 14993 }, { "epoch": 145.39397590361446, "grad_norm": 37.69289016723633, "learning_rate": 2.7213592233009712e-05, "loss": 0.2765, "step": 14994 }, { "epoch": 145.40361445783134, "grad_norm": 24.816381454467773, "learning_rate": 2.720873786407767e-05, "loss": 0.1366, "step": 14995 }, { "epoch": 145.4132530120482, "grad_norm": 8.728103637695312, "learning_rate": 2.720388349514563e-05, "loss": 0.2242, "step": 14996 }, { "epoch": 145.42289156626506, "grad_norm": 10.969090461730957, "learning_rate": 2.7199029126213592e-05, "loss": 0.4352, "step": 14997 }, { "epoch": 145.4325301204819, "grad_norm": 17.944110870361328, "learning_rate": 2.7194174757281553e-05, "loss": 0.2982, "step": 14998 }, { "epoch": 145.4421686746988, "grad_norm": 10.622283935546875, "learning_rate": 2.7189320388349514e-05, "loss": 0.3746, "step": 14999 }, { "epoch": 145.45180722891567, "grad_norm": 9.87585163116455, "learning_rate": 2.7184466019417475e-05, "loss": 0.2441, "step": 15000 }, { "epoch": 145.46144578313252, "grad_norm": 20.305561065673828, "learning_rate": 2.7179611650485437e-05, "loss": 0.1698, "step": 15001 }, { "epoch": 145.4710843373494, "grad_norm": 14.14031982421875, "learning_rate": 2.7174757281553398e-05, "loss": 0.2223, "step": 15002 }, { "epoch": 145.48072289156627, "grad_norm": 24.241031646728516, "learning_rate": 2.716990291262136e-05, "loss": 0.1707, "step": 15003 }, { "epoch": 145.49036144578312, "grad_norm": 61.95704650878906, "learning_rate": 2.716504854368932e-05, "loss": 0.2878, "step": 15004 }, { "epoch": 145.5, "grad_norm": 7.089842796325684, "learning_rate": 2.716019417475728e-05, "loss": 0.2175, "step": 15005 }, { "epoch": 145.50963855421688, "grad_norm": 29.037425994873047, "learning_rate": 2.7155339805825242e-05, "loss": 0.3049, "step": 15006 }, { "epoch": 145.51927710843373, "grad_norm": 26.628381729125977, "learning_rate": 2.7150485436893203e-05, "loss": 0.3085, "step": 15007 }, { "epoch": 145.5289156626506, "grad_norm": 29.419265747070312, "learning_rate": 2.7145631067961165e-05, "loss": 0.3845, "step": 15008 }, { "epoch": 145.53855421686748, "grad_norm": 7.0698394775390625, "learning_rate": 2.714077669902913e-05, "loss": 0.0687, "step": 15009 }, { "epoch": 145.54819277108433, "grad_norm": 14.788382530212402, "learning_rate": 2.713592233009709e-05, "loss": 0.374, "step": 15010 }, { "epoch": 145.5578313253012, "grad_norm": 3.467224359512329, "learning_rate": 2.713106796116505e-05, "loss": 0.2996, "step": 15011 }, { "epoch": 145.5674698795181, "grad_norm": 42.5430793762207, "learning_rate": 2.7126213592233013e-05, "loss": 0.2149, "step": 15012 }, { "epoch": 145.57710843373494, "grad_norm": 16.448759078979492, "learning_rate": 2.7121359223300974e-05, "loss": 0.3707, "step": 15013 }, { "epoch": 145.5867469879518, "grad_norm": 32.44291305541992, "learning_rate": 2.7116504854368935e-05, "loss": 0.3637, "step": 15014 }, { "epoch": 145.59638554216866, "grad_norm": 13.491840362548828, "learning_rate": 2.7111650485436896e-05, "loss": 0.1971, "step": 15015 }, { "epoch": 145.60602409638554, "grad_norm": 8.491942405700684, "learning_rate": 2.7106796116504857e-05, "loss": 0.2248, "step": 15016 }, { "epoch": 145.61566265060242, "grad_norm": 42.08584976196289, "learning_rate": 2.710194174757282e-05, "loss": 0.3545, "step": 15017 }, { "epoch": 145.62530120481927, "grad_norm": 2.8348164558410645, "learning_rate": 2.709708737864078e-05, "loss": 0.1489, "step": 15018 }, { "epoch": 145.63493975903614, "grad_norm": 15.780314445495605, "learning_rate": 2.709223300970874e-05, "loss": 0.2269, "step": 15019 }, { "epoch": 145.64457831325302, "grad_norm": 50.792781829833984, "learning_rate": 2.7087378640776702e-05, "loss": 0.2692, "step": 15020 }, { "epoch": 145.65421686746987, "grad_norm": 7.539266586303711, "learning_rate": 2.7082524271844663e-05, "loss": 0.1653, "step": 15021 }, { "epoch": 145.66385542168675, "grad_norm": 9.278846740722656, "learning_rate": 2.7077669902912624e-05, "loss": 0.0747, "step": 15022 }, { "epoch": 145.67349397590363, "grad_norm": 14.439570426940918, "learning_rate": 2.7072815533980582e-05, "loss": 0.3004, "step": 15023 }, { "epoch": 145.68313253012047, "grad_norm": 15.080349922180176, "learning_rate": 2.7067961165048543e-05, "loss": 0.2507, "step": 15024 }, { "epoch": 145.69277108433735, "grad_norm": 5.925704479217529, "learning_rate": 2.7063106796116504e-05, "loss": 0.2648, "step": 15025 }, { "epoch": 145.70240963855423, "grad_norm": 13.559796333312988, "learning_rate": 2.7058252427184465e-05, "loss": 0.4346, "step": 15026 }, { "epoch": 145.71204819277108, "grad_norm": 8.707927703857422, "learning_rate": 2.7053398058252426e-05, "loss": 0.2805, "step": 15027 }, { "epoch": 145.72168674698796, "grad_norm": 4.241437911987305, "learning_rate": 2.7048543689320388e-05, "loss": 0.1266, "step": 15028 }, { "epoch": 145.73132530120483, "grad_norm": 73.916015625, "learning_rate": 2.704368932038835e-05, "loss": 0.2844, "step": 15029 }, { "epoch": 145.74096385542168, "grad_norm": 17.899580001831055, "learning_rate": 2.703883495145631e-05, "loss": 0.2389, "step": 15030 }, { "epoch": 145.75060240963856, "grad_norm": 2.763669967651367, "learning_rate": 2.703398058252427e-05, "loss": 0.2864, "step": 15031 }, { "epoch": 145.7602409638554, "grad_norm": 5.67909049987793, "learning_rate": 2.7029126213592232e-05, "loss": 0.1487, "step": 15032 }, { "epoch": 145.7698795180723, "grad_norm": 17.330564498901367, "learning_rate": 2.7024271844660193e-05, "loss": 0.4033, "step": 15033 }, { "epoch": 145.77951807228916, "grad_norm": 22.759868621826172, "learning_rate": 2.7019417475728158e-05, "loss": 0.4404, "step": 15034 }, { "epoch": 145.789156626506, "grad_norm": 7.825183868408203, "learning_rate": 2.701456310679612e-05, "loss": 0.2191, "step": 15035 }, { "epoch": 145.7987951807229, "grad_norm": 22.7204647064209, "learning_rate": 2.700970873786408e-05, "loss": 0.2169, "step": 15036 }, { "epoch": 145.80843373493977, "grad_norm": 7.653266429901123, "learning_rate": 2.700485436893204e-05, "loss": 0.3411, "step": 15037 }, { "epoch": 145.81807228915662, "grad_norm": 11.311047554016113, "learning_rate": 2.7000000000000002e-05, "loss": 0.2768, "step": 15038 }, { "epoch": 145.8277108433735, "grad_norm": 5.541877746582031, "learning_rate": 2.6995145631067964e-05, "loss": 0.3166, "step": 15039 }, { "epoch": 145.83734939759037, "grad_norm": 1.9473986625671387, "learning_rate": 2.6990291262135925e-05, "loss": 0.1377, "step": 15040 }, { "epoch": 145.84698795180722, "grad_norm": 29.757225036621094, "learning_rate": 2.6985436893203886e-05, "loss": 0.2303, "step": 15041 }, { "epoch": 145.8566265060241, "grad_norm": 19.21626091003418, "learning_rate": 2.6980582524271847e-05, "loss": 0.3562, "step": 15042 }, { "epoch": 145.86626506024098, "grad_norm": 58.46262741088867, "learning_rate": 2.6975728155339808e-05, "loss": 0.4359, "step": 15043 }, { "epoch": 145.87590361445783, "grad_norm": 53.81080627441406, "learning_rate": 2.697087378640777e-05, "loss": 0.213, "step": 15044 }, { "epoch": 145.8855421686747, "grad_norm": 25.236961364746094, "learning_rate": 2.696601941747573e-05, "loss": 0.1825, "step": 15045 }, { "epoch": 145.89518072289155, "grad_norm": 13.088836669921875, "learning_rate": 2.696116504854369e-05, "loss": 0.2121, "step": 15046 }, { "epoch": 145.90481927710843, "grad_norm": 9.0810546875, "learning_rate": 2.6956310679611653e-05, "loss": 0.4002, "step": 15047 }, { "epoch": 145.9144578313253, "grad_norm": 78.96417999267578, "learning_rate": 2.6951456310679614e-05, "loss": 0.4316, "step": 15048 }, { "epoch": 145.92409638554216, "grad_norm": 60.92538833618164, "learning_rate": 2.6946601941747575e-05, "loss": 0.2395, "step": 15049 }, { "epoch": 145.93373493975903, "grad_norm": 1.3521833419799805, "learning_rate": 2.6941747572815533e-05, "loss": 0.1495, "step": 15050 }, { "epoch": 145.9433734939759, "grad_norm": 49.58253860473633, "learning_rate": 2.6936893203883494e-05, "loss": 0.2667, "step": 15051 }, { "epoch": 145.95301204819276, "grad_norm": 52.42780685424805, "learning_rate": 2.6932038834951455e-05, "loss": 0.0715, "step": 15052 }, { "epoch": 145.96265060240964, "grad_norm": 49.42189025878906, "learning_rate": 2.6927184466019416e-05, "loss": 0.2543, "step": 15053 }, { "epoch": 145.97228915662652, "grad_norm": 47.09099578857422, "learning_rate": 2.6922330097087377e-05, "loss": 0.3731, "step": 15054 }, { "epoch": 145.98192771084337, "grad_norm": 63.06123352050781, "learning_rate": 2.691747572815534e-05, "loss": 0.2162, "step": 15055 }, { "epoch": 145.99156626506024, "grad_norm": 17.02931785583496, "learning_rate": 2.69126213592233e-05, "loss": 0.0915, "step": 15056 }, { "epoch": 146.00722891566264, "grad_norm": 29.952817916870117, "learning_rate": 2.690776699029126e-05, "loss": 0.2958, "step": 15057 }, { "epoch": 146.01686746987951, "grad_norm": 5.231495380401611, "learning_rate": 2.6902912621359222e-05, "loss": 0.2914, "step": 15058 }, { "epoch": 146.0265060240964, "grad_norm": 13.639781951904297, "learning_rate": 2.689805825242719e-05, "loss": 0.4742, "step": 15059 }, { "epoch": 146.03614457831324, "grad_norm": 16.01126480102539, "learning_rate": 2.689320388349515e-05, "loss": 0.4178, "step": 15060 }, { "epoch": 146.04578313253012, "grad_norm": 23.36020851135254, "learning_rate": 2.688834951456311e-05, "loss": 0.2478, "step": 15061 }, { "epoch": 146.055421686747, "grad_norm": 40.2196159362793, "learning_rate": 2.688349514563107e-05, "loss": 0.2976, "step": 15062 }, { "epoch": 146.06506024096385, "grad_norm": 25.862478256225586, "learning_rate": 2.687864077669903e-05, "loss": 0.3159, "step": 15063 }, { "epoch": 146.07469879518072, "grad_norm": 11.821768760681152, "learning_rate": 2.6873786407766992e-05, "loss": 0.1943, "step": 15064 }, { "epoch": 146.0843373493976, "grad_norm": 31.87714958190918, "learning_rate": 2.6868932038834953e-05, "loss": 0.3643, "step": 15065 }, { "epoch": 146.09397590361445, "grad_norm": 13.300101280212402, "learning_rate": 2.6864077669902914e-05, "loss": 0.1435, "step": 15066 }, { "epoch": 146.10361445783133, "grad_norm": 9.714526176452637, "learning_rate": 2.6859223300970876e-05, "loss": 0.3442, "step": 15067 }, { "epoch": 146.1132530120482, "grad_norm": 59.52511978149414, "learning_rate": 2.6854368932038837e-05, "loss": 0.3397, "step": 15068 }, { "epoch": 146.12289156626505, "grad_norm": 51.73287582397461, "learning_rate": 2.6849514563106798e-05, "loss": 0.1494, "step": 15069 }, { "epoch": 146.13253012048193, "grad_norm": 12.690328598022461, "learning_rate": 2.684466019417476e-05, "loss": 0.2582, "step": 15070 }, { "epoch": 146.1421686746988, "grad_norm": 2.1759045124053955, "learning_rate": 2.683980582524272e-05, "loss": 0.2841, "step": 15071 }, { "epoch": 146.15180722891566, "grad_norm": 82.15741729736328, "learning_rate": 2.683495145631068e-05, "loss": 0.4297, "step": 15072 }, { "epoch": 146.16144578313254, "grad_norm": 20.554279327392578, "learning_rate": 2.6830097087378642e-05, "loss": 0.6833, "step": 15073 }, { "epoch": 146.17108433734938, "grad_norm": 27.416728973388672, "learning_rate": 2.6825242718446604e-05, "loss": 0.2724, "step": 15074 }, { "epoch": 146.18072289156626, "grad_norm": 13.30562973022461, "learning_rate": 2.6820388349514565e-05, "loss": 0.3097, "step": 15075 }, { "epoch": 146.19036144578314, "grad_norm": 26.934608459472656, "learning_rate": 2.6815533980582526e-05, "loss": 0.0905, "step": 15076 }, { "epoch": 146.2, "grad_norm": 4.64522647857666, "learning_rate": 2.6810679611650487e-05, "loss": 0.2863, "step": 15077 }, { "epoch": 146.20963855421687, "grad_norm": 40.91659164428711, "learning_rate": 2.6805825242718445e-05, "loss": 0.3144, "step": 15078 }, { "epoch": 146.21927710843374, "grad_norm": 27.52704620361328, "learning_rate": 2.6800970873786406e-05, "loss": 0.3286, "step": 15079 }, { "epoch": 146.2289156626506, "grad_norm": 28.72953224182129, "learning_rate": 2.6796116504854367e-05, "loss": 0.3932, "step": 15080 }, { "epoch": 146.23855421686747, "grad_norm": 13.486839294433594, "learning_rate": 2.6791262135922328e-05, "loss": 0.1894, "step": 15081 }, { "epoch": 146.24819277108435, "grad_norm": 23.020009994506836, "learning_rate": 2.678640776699029e-05, "loss": 0.3227, "step": 15082 }, { "epoch": 146.2578313253012, "grad_norm": 17.745349884033203, "learning_rate": 2.678155339805825e-05, "loss": 0.2959, "step": 15083 }, { "epoch": 146.26746987951807, "grad_norm": 42.43164825439453, "learning_rate": 2.677669902912622e-05, "loss": 0.2811, "step": 15084 }, { "epoch": 146.27710843373495, "grad_norm": 6.3764119148254395, "learning_rate": 2.677184466019418e-05, "loss": 0.2432, "step": 15085 }, { "epoch": 146.2867469879518, "grad_norm": 23.69597053527832, "learning_rate": 2.676699029126214e-05, "loss": 0.2841, "step": 15086 }, { "epoch": 146.29638554216868, "grad_norm": 16.012365341186523, "learning_rate": 2.6762135922330102e-05, "loss": 0.1514, "step": 15087 }, { "epoch": 146.30602409638556, "grad_norm": 17.556785583496094, "learning_rate": 2.6757281553398063e-05, "loss": 0.3218, "step": 15088 }, { "epoch": 146.3156626506024, "grad_norm": 52.31234359741211, "learning_rate": 2.675242718446602e-05, "loss": 0.3211, "step": 15089 }, { "epoch": 146.32530120481928, "grad_norm": 10.25722599029541, "learning_rate": 2.6747572815533982e-05, "loss": 0.2411, "step": 15090 }, { "epoch": 146.33493975903613, "grad_norm": 7.240609169006348, "learning_rate": 2.6742718446601943e-05, "loss": 0.3399, "step": 15091 }, { "epoch": 146.344578313253, "grad_norm": 12.468892097473145, "learning_rate": 2.6737864077669904e-05, "loss": 0.3312, "step": 15092 }, { "epoch": 146.3542168674699, "grad_norm": 48.97119903564453, "learning_rate": 2.6733009708737865e-05, "loss": 0.3359, "step": 15093 }, { "epoch": 146.36385542168674, "grad_norm": 9.72222900390625, "learning_rate": 2.6728155339805827e-05, "loss": 0.4048, "step": 15094 }, { "epoch": 146.3734939759036, "grad_norm": 1.9778013229370117, "learning_rate": 2.6723300970873788e-05, "loss": 0.1385, "step": 15095 }, { "epoch": 146.3831325301205, "grad_norm": 7.189787864685059, "learning_rate": 2.671844660194175e-05, "loss": 0.1553, "step": 15096 }, { "epoch": 146.39277108433734, "grad_norm": 23.256216049194336, "learning_rate": 2.671359223300971e-05, "loss": 0.1878, "step": 15097 }, { "epoch": 146.40240963855422, "grad_norm": 4.303686618804932, "learning_rate": 2.670873786407767e-05, "loss": 0.1912, "step": 15098 }, { "epoch": 146.4120481927711, "grad_norm": 10.978555679321289, "learning_rate": 2.6703883495145632e-05, "loss": 0.1306, "step": 15099 }, { "epoch": 146.42168674698794, "grad_norm": 54.32220458984375, "learning_rate": 2.6699029126213593e-05, "loss": 0.3422, "step": 15100 }, { "epoch": 146.43132530120482, "grad_norm": 22.786834716796875, "learning_rate": 2.6694174757281555e-05, "loss": 0.251, "step": 15101 }, { "epoch": 146.4409638554217, "grad_norm": 37.18766403198242, "learning_rate": 2.6689320388349516e-05, "loss": 0.4261, "step": 15102 }, { "epoch": 146.45060240963855, "grad_norm": 21.889774322509766, "learning_rate": 2.6684466019417477e-05, "loss": 0.2826, "step": 15103 }, { "epoch": 146.46024096385543, "grad_norm": 64.33012390136719, "learning_rate": 2.6679611650485438e-05, "loss": 0.4276, "step": 15104 }, { "epoch": 146.46987951807228, "grad_norm": 4.8169097900390625, "learning_rate": 2.66747572815534e-05, "loss": 0.1664, "step": 15105 }, { "epoch": 146.47951807228915, "grad_norm": 55.09489440917969, "learning_rate": 2.6669902912621357e-05, "loss": 0.629, "step": 15106 }, { "epoch": 146.48915662650603, "grad_norm": 24.64453887939453, "learning_rate": 2.6665048543689318e-05, "loss": 0.2082, "step": 15107 }, { "epoch": 146.49879518072288, "grad_norm": 18.051259994506836, "learning_rate": 2.666019417475728e-05, "loss": 0.0955, "step": 15108 }, { "epoch": 146.50843373493976, "grad_norm": 46.27605056762695, "learning_rate": 2.665533980582524e-05, "loss": 0.245, "step": 15109 }, { "epoch": 146.51807228915663, "grad_norm": 5.904536247253418, "learning_rate": 2.6650485436893208e-05, "loss": 0.1, "step": 15110 }, { "epoch": 146.52771084337348, "grad_norm": 3.7844653129577637, "learning_rate": 2.664563106796117e-05, "loss": 0.222, "step": 15111 }, { "epoch": 146.53734939759036, "grad_norm": 2.721863269805908, "learning_rate": 2.664077669902913e-05, "loss": 0.1178, "step": 15112 }, { "epoch": 146.54698795180724, "grad_norm": 8.286144256591797, "learning_rate": 2.6635922330097092e-05, "loss": 0.2443, "step": 15113 }, { "epoch": 146.5566265060241, "grad_norm": 7.548091411590576, "learning_rate": 2.6631067961165053e-05, "loss": 0.143, "step": 15114 }, { "epoch": 146.56626506024097, "grad_norm": 21.355484008789062, "learning_rate": 2.6626213592233014e-05, "loss": 0.3075, "step": 15115 }, { "epoch": 146.57590361445784, "grad_norm": 18.50648307800293, "learning_rate": 2.6621359223300972e-05, "loss": 0.1889, "step": 15116 }, { "epoch": 146.5855421686747, "grad_norm": 2.64536714553833, "learning_rate": 2.6616504854368933e-05, "loss": 0.1898, "step": 15117 }, { "epoch": 146.59518072289157, "grad_norm": 5.663579940795898, "learning_rate": 2.6611650485436894e-05, "loss": 0.2326, "step": 15118 }, { "epoch": 146.60481927710845, "grad_norm": 13.688603401184082, "learning_rate": 2.6606796116504855e-05, "loss": 0.2187, "step": 15119 }, { "epoch": 146.6144578313253, "grad_norm": 10.574538230895996, "learning_rate": 2.6601941747572816e-05, "loss": 0.2002, "step": 15120 }, { "epoch": 146.62409638554217, "grad_norm": 4.605038642883301, "learning_rate": 2.6597087378640777e-05, "loss": 0.1836, "step": 15121 }, { "epoch": 146.63373493975902, "grad_norm": 3.951129674911499, "learning_rate": 2.659223300970874e-05, "loss": 0.3253, "step": 15122 }, { "epoch": 146.6433734939759, "grad_norm": 9.805501937866211, "learning_rate": 2.65873786407767e-05, "loss": 0.2124, "step": 15123 }, { "epoch": 146.65301204819278, "grad_norm": 33.008323669433594, "learning_rate": 2.658252427184466e-05, "loss": 0.5452, "step": 15124 }, { "epoch": 146.66265060240963, "grad_norm": 5.182271480560303, "learning_rate": 2.6577669902912622e-05, "loss": 0.2816, "step": 15125 }, { "epoch": 146.6722891566265, "grad_norm": 27.12452507019043, "learning_rate": 2.6572815533980583e-05, "loss": 0.618, "step": 15126 }, { "epoch": 146.68192771084338, "grad_norm": 63.451087951660156, "learning_rate": 2.6567961165048544e-05, "loss": 0.2227, "step": 15127 }, { "epoch": 146.69156626506023, "grad_norm": 19.310726165771484, "learning_rate": 2.6563106796116505e-05, "loss": 0.37, "step": 15128 }, { "epoch": 146.7012048192771, "grad_norm": 20.678972244262695, "learning_rate": 2.6558252427184467e-05, "loss": 0.2716, "step": 15129 }, { "epoch": 146.710843373494, "grad_norm": 29.126798629760742, "learning_rate": 2.6553398058252428e-05, "loss": 0.2122, "step": 15130 }, { "epoch": 146.72048192771084, "grad_norm": 19.20490074157715, "learning_rate": 2.654854368932039e-05, "loss": 0.384, "step": 15131 }, { "epoch": 146.7301204819277, "grad_norm": 24.046979904174805, "learning_rate": 2.654368932038835e-05, "loss": 0.3361, "step": 15132 }, { "epoch": 146.7397590361446, "grad_norm": 6.644261837005615, "learning_rate": 2.6538834951456308e-05, "loss": 0.211, "step": 15133 }, { "epoch": 146.74939759036144, "grad_norm": 9.900809288024902, "learning_rate": 2.653398058252427e-05, "loss": 0.4671, "step": 15134 }, { "epoch": 146.75903614457832, "grad_norm": 10.354270935058594, "learning_rate": 2.6529126213592237e-05, "loss": 0.169, "step": 15135 }, { "epoch": 146.76867469879517, "grad_norm": 12.241429328918457, "learning_rate": 2.6524271844660198e-05, "loss": 0.2664, "step": 15136 }, { "epoch": 146.77831325301204, "grad_norm": 14.432880401611328, "learning_rate": 2.651941747572816e-05, "loss": 0.2161, "step": 15137 }, { "epoch": 146.78795180722892, "grad_norm": 17.457176208496094, "learning_rate": 2.651456310679612e-05, "loss": 0.106, "step": 15138 }, { "epoch": 146.79759036144577, "grad_norm": 28.040325164794922, "learning_rate": 2.650970873786408e-05, "loss": 0.3038, "step": 15139 }, { "epoch": 146.80722891566265, "grad_norm": 5.128582000732422, "learning_rate": 2.6504854368932043e-05, "loss": 0.2654, "step": 15140 }, { "epoch": 146.81686746987953, "grad_norm": 4.617157459259033, "learning_rate": 2.6500000000000004e-05, "loss": 0.2849, "step": 15141 }, { "epoch": 146.82650602409637, "grad_norm": 24.957355499267578, "learning_rate": 2.6495145631067965e-05, "loss": 0.0835, "step": 15142 }, { "epoch": 146.83614457831325, "grad_norm": 12.477834701538086, "learning_rate": 2.6490291262135926e-05, "loss": 0.2272, "step": 15143 }, { "epoch": 146.84578313253013, "grad_norm": 23.170305252075195, "learning_rate": 2.6485436893203884e-05, "loss": 0.1597, "step": 15144 }, { "epoch": 146.85542168674698, "grad_norm": 7.915955543518066, "learning_rate": 2.6480582524271845e-05, "loss": 0.3932, "step": 15145 }, { "epoch": 146.86506024096386, "grad_norm": 5.576157569885254, "learning_rate": 2.6475728155339806e-05, "loss": 0.1388, "step": 15146 }, { "epoch": 146.87469879518073, "grad_norm": 23.5979061126709, "learning_rate": 2.6470873786407767e-05, "loss": 0.1959, "step": 15147 }, { "epoch": 146.88433734939758, "grad_norm": 12.353256225585938, "learning_rate": 2.646601941747573e-05, "loss": 0.1235, "step": 15148 }, { "epoch": 146.89397590361446, "grad_norm": 34.80788040161133, "learning_rate": 2.646116504854369e-05, "loss": 0.4474, "step": 15149 }, { "epoch": 146.90361445783134, "grad_norm": 58.88296890258789, "learning_rate": 2.645631067961165e-05, "loss": 0.2912, "step": 15150 }, { "epoch": 146.9132530120482, "grad_norm": 17.67636489868164, "learning_rate": 2.6451456310679612e-05, "loss": 0.1991, "step": 15151 }, { "epoch": 146.92289156626506, "grad_norm": 4.953907489776611, "learning_rate": 2.6446601941747573e-05, "loss": 0.244, "step": 15152 }, { "epoch": 146.9325301204819, "grad_norm": 3.3957722187042236, "learning_rate": 2.6441747572815534e-05, "loss": 0.1977, "step": 15153 }, { "epoch": 146.9421686746988, "grad_norm": 1.8525911569595337, "learning_rate": 2.6436893203883495e-05, "loss": 0.2477, "step": 15154 }, { "epoch": 146.95180722891567, "grad_norm": 25.397308349609375, "learning_rate": 2.6432038834951456e-05, "loss": 0.2348, "step": 15155 }, { "epoch": 146.96144578313252, "grad_norm": 8.516313552856445, "learning_rate": 2.6427184466019418e-05, "loss": 0.2498, "step": 15156 }, { "epoch": 146.9710843373494, "grad_norm": 21.72237777709961, "learning_rate": 2.642233009708738e-05, "loss": 0.1349, "step": 15157 }, { "epoch": 146.98072289156627, "grad_norm": 15.621766090393066, "learning_rate": 2.641747572815534e-05, "loss": 0.3349, "step": 15158 }, { "epoch": 146.99036144578312, "grad_norm": 61.14155578613281, "learning_rate": 2.64126213592233e-05, "loss": 0.2662, "step": 15159 }, { "epoch": 147.00602409638554, "grad_norm": 22.868051528930664, "learning_rate": 2.6407766990291266e-05, "loss": 0.2177, "step": 15160 }, { "epoch": 147.01566265060242, "grad_norm": 8.553295135498047, "learning_rate": 2.6402912621359227e-05, "loss": 0.2462, "step": 15161 }, { "epoch": 147.02530120481927, "grad_norm": 28.429410934448242, "learning_rate": 2.6398058252427188e-05, "loss": 0.3169, "step": 15162 }, { "epoch": 147.03493975903615, "grad_norm": 37.356040954589844, "learning_rate": 2.639320388349515e-05, "loss": 0.31, "step": 15163 }, { "epoch": 147.044578313253, "grad_norm": 14.833170890808105, "learning_rate": 2.638834951456311e-05, "loss": 0.2805, "step": 15164 }, { "epoch": 147.05421686746988, "grad_norm": 11.210765838623047, "learning_rate": 2.638349514563107e-05, "loss": 0.3727, "step": 15165 }, { "epoch": 147.06385542168675, "grad_norm": 21.371856689453125, "learning_rate": 2.6378640776699032e-05, "loss": 0.2859, "step": 15166 }, { "epoch": 147.0734939759036, "grad_norm": 8.728198051452637, "learning_rate": 2.6373786407766994e-05, "loss": 0.2905, "step": 15167 }, { "epoch": 147.08313253012048, "grad_norm": 8.719414710998535, "learning_rate": 2.6368932038834955e-05, "loss": 0.1455, "step": 15168 }, { "epoch": 147.09277108433736, "grad_norm": 32.3114013671875, "learning_rate": 2.6364077669902916e-05, "loss": 0.3455, "step": 15169 }, { "epoch": 147.1024096385542, "grad_norm": 57.91802978515625, "learning_rate": 2.6359223300970877e-05, "loss": 0.2868, "step": 15170 }, { "epoch": 147.11204819277108, "grad_norm": 19.771785736083984, "learning_rate": 2.6354368932038838e-05, "loss": 0.135, "step": 15171 }, { "epoch": 147.12168674698796, "grad_norm": 9.973594665527344, "learning_rate": 2.6349514563106796e-05, "loss": 0.3972, "step": 15172 }, { "epoch": 147.1313253012048, "grad_norm": 28.87187957763672, "learning_rate": 2.6344660194174757e-05, "loss": 0.4237, "step": 15173 }, { "epoch": 147.1409638554217, "grad_norm": 24.4217529296875, "learning_rate": 2.6339805825242718e-05, "loss": 0.2016, "step": 15174 }, { "epoch": 147.15060240963857, "grad_norm": 22.14289665222168, "learning_rate": 2.633495145631068e-05, "loss": 0.1742, "step": 15175 }, { "epoch": 147.16024096385541, "grad_norm": 11.08005428314209, "learning_rate": 2.633009708737864e-05, "loss": 0.1415, "step": 15176 }, { "epoch": 147.1698795180723, "grad_norm": 9.55984115600586, "learning_rate": 2.63252427184466e-05, "loss": 0.2358, "step": 15177 }, { "epoch": 147.17951807228917, "grad_norm": 12.673470497131348, "learning_rate": 2.6320388349514563e-05, "loss": 0.2242, "step": 15178 }, { "epoch": 147.18915662650602, "grad_norm": 39.59548568725586, "learning_rate": 2.6315533980582524e-05, "loss": 0.2119, "step": 15179 }, { "epoch": 147.1987951807229, "grad_norm": 24.705968856811523, "learning_rate": 2.6310679611650485e-05, "loss": 0.2912, "step": 15180 }, { "epoch": 147.20843373493975, "grad_norm": 26.258319854736328, "learning_rate": 2.6305825242718446e-05, "loss": 0.3982, "step": 15181 }, { "epoch": 147.21807228915662, "grad_norm": 19.750364303588867, "learning_rate": 2.6300970873786407e-05, "loss": 0.1759, "step": 15182 }, { "epoch": 147.2277108433735, "grad_norm": 30.024192810058594, "learning_rate": 2.629611650485437e-05, "loss": 0.2705, "step": 15183 }, { "epoch": 147.23734939759035, "grad_norm": 15.395146369934082, "learning_rate": 2.629126213592233e-05, "loss": 0.1915, "step": 15184 }, { "epoch": 147.24698795180723, "grad_norm": 7.909849166870117, "learning_rate": 2.6286407766990294e-05, "loss": 0.1943, "step": 15185 }, { "epoch": 147.2566265060241, "grad_norm": 9.928447723388672, "learning_rate": 2.6281553398058255e-05, "loss": 0.288, "step": 15186 }, { "epoch": 147.26626506024095, "grad_norm": 6.263176441192627, "learning_rate": 2.6276699029126216e-05, "loss": 0.2056, "step": 15187 }, { "epoch": 147.27590361445783, "grad_norm": 44.571964263916016, "learning_rate": 2.6271844660194178e-05, "loss": 0.4404, "step": 15188 }, { "epoch": 147.2855421686747, "grad_norm": 2.095494508743286, "learning_rate": 2.626699029126214e-05, "loss": 0.2087, "step": 15189 }, { "epoch": 147.29518072289156, "grad_norm": 15.545675277709961, "learning_rate": 2.62621359223301e-05, "loss": 0.3106, "step": 15190 }, { "epoch": 147.30481927710844, "grad_norm": 32.024044036865234, "learning_rate": 2.625728155339806e-05, "loss": 0.2762, "step": 15191 }, { "epoch": 147.3144578313253, "grad_norm": 55.89515686035156, "learning_rate": 2.6252427184466022e-05, "loss": 0.2373, "step": 15192 }, { "epoch": 147.32409638554216, "grad_norm": 8.522149085998535, "learning_rate": 2.6247572815533983e-05, "loss": 0.1666, "step": 15193 }, { "epoch": 147.33373493975904, "grad_norm": 3.970202684402466, "learning_rate": 2.6242718446601945e-05, "loss": 0.1072, "step": 15194 }, { "epoch": 147.34337349397592, "grad_norm": 5.705228805541992, "learning_rate": 2.6237864077669906e-05, "loss": 0.2997, "step": 15195 }, { "epoch": 147.35301204819277, "grad_norm": 22.139081954956055, "learning_rate": 2.6233009708737867e-05, "loss": 0.3551, "step": 15196 }, { "epoch": 147.36265060240964, "grad_norm": 110.96804809570312, "learning_rate": 2.6228155339805828e-05, "loss": 0.6612, "step": 15197 }, { "epoch": 147.3722891566265, "grad_norm": 17.776813507080078, "learning_rate": 2.622330097087379e-05, "loss": 0.2681, "step": 15198 }, { "epoch": 147.38192771084337, "grad_norm": 15.192818641662598, "learning_rate": 2.6218446601941747e-05, "loss": 0.3288, "step": 15199 }, { "epoch": 147.39156626506025, "grad_norm": 10.99912166595459, "learning_rate": 2.6213592233009708e-05, "loss": 0.4772, "step": 15200 }, { "epoch": 147.4012048192771, "grad_norm": 14.305418014526367, "learning_rate": 2.620873786407767e-05, "loss": 0.2025, "step": 15201 }, { "epoch": 147.41084337349398, "grad_norm": 8.2688570022583, "learning_rate": 2.620388349514563e-05, "loss": 0.2592, "step": 15202 }, { "epoch": 147.42048192771085, "grad_norm": 13.14148235321045, "learning_rate": 2.619902912621359e-05, "loss": 0.341, "step": 15203 }, { "epoch": 147.4301204819277, "grad_norm": 11.59022331237793, "learning_rate": 2.6194174757281553e-05, "loss": 0.3726, "step": 15204 }, { "epoch": 147.43975903614458, "grad_norm": 21.186134338378906, "learning_rate": 2.6189320388349514e-05, "loss": 0.3169, "step": 15205 }, { "epoch": 147.44939759036146, "grad_norm": 4.384810447692871, "learning_rate": 2.6184466019417475e-05, "loss": 0.1214, "step": 15206 }, { "epoch": 147.4590361445783, "grad_norm": 6.774295330047607, "learning_rate": 2.6179611650485436e-05, "loss": 0.1759, "step": 15207 }, { "epoch": 147.46867469879518, "grad_norm": 4.567718505859375, "learning_rate": 2.6174757281553397e-05, "loss": 0.197, "step": 15208 }, { "epoch": 147.47831325301206, "grad_norm": 8.383198738098145, "learning_rate": 2.6169902912621358e-05, "loss": 0.1604, "step": 15209 }, { "epoch": 147.4879518072289, "grad_norm": 88.88235473632812, "learning_rate": 2.616504854368932e-05, "loss": 0.2256, "step": 15210 }, { "epoch": 147.4975903614458, "grad_norm": 21.780441284179688, "learning_rate": 2.6160194174757284e-05, "loss": 0.3274, "step": 15211 }, { "epoch": 147.50722891566264, "grad_norm": 87.83056640625, "learning_rate": 2.6155339805825245e-05, "loss": 0.5315, "step": 15212 }, { "epoch": 147.51686746987951, "grad_norm": 18.902015686035156, "learning_rate": 2.6150485436893206e-05, "loss": 0.1086, "step": 15213 }, { "epoch": 147.5265060240964, "grad_norm": 10.213658332824707, "learning_rate": 2.6145631067961167e-05, "loss": 0.1739, "step": 15214 }, { "epoch": 147.53614457831324, "grad_norm": 12.770517349243164, "learning_rate": 2.614077669902913e-05, "loss": 0.267, "step": 15215 }, { "epoch": 147.54578313253012, "grad_norm": 46.70633316040039, "learning_rate": 2.613592233009709e-05, "loss": 0.2701, "step": 15216 }, { "epoch": 147.555421686747, "grad_norm": 23.867229461669922, "learning_rate": 2.613106796116505e-05, "loss": 0.4242, "step": 15217 }, { "epoch": 147.56506024096385, "grad_norm": 13.883170127868652, "learning_rate": 2.6126213592233012e-05, "loss": 0.17, "step": 15218 }, { "epoch": 147.57469879518072, "grad_norm": 23.89102554321289, "learning_rate": 2.6121359223300973e-05, "loss": 0.2985, "step": 15219 }, { "epoch": 147.5843373493976, "grad_norm": 17.859024047851562, "learning_rate": 2.6116504854368934e-05, "loss": 0.2911, "step": 15220 }, { "epoch": 147.59397590361445, "grad_norm": 23.104188919067383, "learning_rate": 2.6111650485436895e-05, "loss": 0.2637, "step": 15221 }, { "epoch": 147.60361445783133, "grad_norm": 8.661846160888672, "learning_rate": 2.6106796116504857e-05, "loss": 0.212, "step": 15222 }, { "epoch": 147.6132530120482, "grad_norm": 20.68992042541504, "learning_rate": 2.6101941747572818e-05, "loss": 0.2872, "step": 15223 }, { "epoch": 147.62289156626505, "grad_norm": 7.730920791625977, "learning_rate": 2.609708737864078e-05, "loss": 0.2753, "step": 15224 }, { "epoch": 147.63253012048193, "grad_norm": 21.478708267211914, "learning_rate": 2.609223300970874e-05, "loss": 0.2127, "step": 15225 }, { "epoch": 147.6421686746988, "grad_norm": 20.452850341796875, "learning_rate": 2.60873786407767e-05, "loss": 0.2378, "step": 15226 }, { "epoch": 147.65180722891566, "grad_norm": 5.8891191482543945, "learning_rate": 2.608252427184466e-05, "loss": 0.3139, "step": 15227 }, { "epoch": 147.66144578313254, "grad_norm": 57.561580657958984, "learning_rate": 2.607766990291262e-05, "loss": 0.2332, "step": 15228 }, { "epoch": 147.67108433734938, "grad_norm": 38.132713317871094, "learning_rate": 2.607281553398058e-05, "loss": 0.4953, "step": 15229 }, { "epoch": 147.68072289156626, "grad_norm": 16.641613006591797, "learning_rate": 2.6067961165048542e-05, "loss": 0.2761, "step": 15230 }, { "epoch": 147.69036144578314, "grad_norm": 8.617431640625, "learning_rate": 2.6063106796116504e-05, "loss": 0.2794, "step": 15231 }, { "epoch": 147.7, "grad_norm": 24.805978775024414, "learning_rate": 2.6058252427184465e-05, "loss": 0.3036, "step": 15232 }, { "epoch": 147.70963855421687, "grad_norm": 48.10213851928711, "learning_rate": 2.6053398058252426e-05, "loss": 0.1942, "step": 15233 }, { "epoch": 147.71927710843374, "grad_norm": 12.161723136901855, "learning_rate": 2.6048543689320387e-05, "loss": 0.2924, "step": 15234 }, { "epoch": 147.7289156626506, "grad_norm": 28.39802360534668, "learning_rate": 2.6043689320388348e-05, "loss": 0.2036, "step": 15235 }, { "epoch": 147.73855421686747, "grad_norm": 31.450925827026367, "learning_rate": 2.6038834951456316e-05, "loss": 0.2879, "step": 15236 }, { "epoch": 147.74819277108435, "grad_norm": 16.375595092773438, "learning_rate": 2.6033980582524277e-05, "loss": 0.1953, "step": 15237 }, { "epoch": 147.7578313253012, "grad_norm": 5.03467321395874, "learning_rate": 2.6029126213592235e-05, "loss": 0.0965, "step": 15238 }, { "epoch": 147.76746987951807, "grad_norm": 41.71901321411133, "learning_rate": 2.6024271844660196e-05, "loss": 0.444, "step": 15239 }, { "epoch": 147.77710843373495, "grad_norm": 13.166916847229004, "learning_rate": 2.6019417475728157e-05, "loss": 0.2619, "step": 15240 }, { "epoch": 147.7867469879518, "grad_norm": 8.341665267944336, "learning_rate": 2.601456310679612e-05, "loss": 0.2187, "step": 15241 }, { "epoch": 147.79638554216868, "grad_norm": 13.672168731689453, "learning_rate": 2.600970873786408e-05, "loss": 0.362, "step": 15242 }, { "epoch": 147.80602409638556, "grad_norm": 14.746539115905762, "learning_rate": 2.600485436893204e-05, "loss": 0.255, "step": 15243 }, { "epoch": 147.8156626506024, "grad_norm": 7.073148727416992, "learning_rate": 2.6000000000000002e-05, "loss": 0.204, "step": 15244 }, { "epoch": 147.82530120481928, "grad_norm": 16.097354888916016, "learning_rate": 2.5995145631067963e-05, "loss": 0.1842, "step": 15245 }, { "epoch": 147.83493975903613, "grad_norm": 41.380393981933594, "learning_rate": 2.5990291262135924e-05, "loss": 0.2026, "step": 15246 }, { "epoch": 147.844578313253, "grad_norm": 15.578185081481934, "learning_rate": 2.5985436893203885e-05, "loss": 0.2709, "step": 15247 }, { "epoch": 147.8542168674699, "grad_norm": 52.62128448486328, "learning_rate": 2.5980582524271846e-05, "loss": 0.2467, "step": 15248 }, { "epoch": 147.86385542168674, "grad_norm": 9.230212211608887, "learning_rate": 2.5975728155339808e-05, "loss": 0.1666, "step": 15249 }, { "epoch": 147.8734939759036, "grad_norm": 4.996236324310303, "learning_rate": 2.597087378640777e-05, "loss": 0.2023, "step": 15250 }, { "epoch": 147.8831325301205, "grad_norm": 19.35103416442871, "learning_rate": 2.596601941747573e-05, "loss": 0.2623, "step": 15251 }, { "epoch": 147.89277108433734, "grad_norm": 5.738787651062012, "learning_rate": 2.596116504854369e-05, "loss": 0.323, "step": 15252 }, { "epoch": 147.90240963855422, "grad_norm": 20.22294807434082, "learning_rate": 2.5956310679611652e-05, "loss": 0.2208, "step": 15253 }, { "epoch": 147.9120481927711, "grad_norm": 8.06196117401123, "learning_rate": 2.5951456310679613e-05, "loss": 0.2132, "step": 15254 }, { "epoch": 147.92168674698794, "grad_norm": 7.964545726776123, "learning_rate": 2.594660194174757e-05, "loss": 0.3877, "step": 15255 }, { "epoch": 147.93132530120482, "grad_norm": 11.547810554504395, "learning_rate": 2.5941747572815532e-05, "loss": 0.332, "step": 15256 }, { "epoch": 147.9409638554217, "grad_norm": 19.7912654876709, "learning_rate": 2.5936893203883493e-05, "loss": 0.1399, "step": 15257 }, { "epoch": 147.95060240963855, "grad_norm": 11.26048469543457, "learning_rate": 2.5932038834951454e-05, "loss": 0.2578, "step": 15258 }, { "epoch": 147.96024096385543, "grad_norm": 23.2171573638916, "learning_rate": 2.5927184466019416e-05, "loss": 0.3285, "step": 15259 }, { "epoch": 147.96987951807228, "grad_norm": 8.696111679077148, "learning_rate": 2.5922330097087377e-05, "loss": 0.1048, "step": 15260 }, { "epoch": 147.97951807228915, "grad_norm": 17.54869842529297, "learning_rate": 2.5917475728155345e-05, "loss": 0.4037, "step": 15261 }, { "epoch": 147.98915662650603, "grad_norm": 4.320497512817383, "learning_rate": 2.5912621359223306e-05, "loss": 0.1525, "step": 15262 }, { "epoch": 148.00481927710842, "grad_norm": 27.37523651123047, "learning_rate": 2.5907766990291267e-05, "loss": 0.2771, "step": 15263 }, { "epoch": 148.0144578313253, "grad_norm": 9.825711250305176, "learning_rate": 2.5902912621359228e-05, "loss": 0.2144, "step": 15264 }, { "epoch": 148.02409638554218, "grad_norm": 10.457043647766113, "learning_rate": 2.5898058252427186e-05, "loss": 0.1784, "step": 15265 }, { "epoch": 148.03373493975903, "grad_norm": 10.38471794128418, "learning_rate": 2.5893203883495147e-05, "loss": 0.2059, "step": 15266 }, { "epoch": 148.0433734939759, "grad_norm": 6.113032817840576, "learning_rate": 2.5888349514563108e-05, "loss": 0.2193, "step": 15267 }, { "epoch": 148.05301204819278, "grad_norm": 8.445361137390137, "learning_rate": 2.588349514563107e-05, "loss": 0.1795, "step": 15268 }, { "epoch": 148.06265060240963, "grad_norm": 7.750030994415283, "learning_rate": 2.587864077669903e-05, "loss": 0.1745, "step": 15269 }, { "epoch": 148.0722891566265, "grad_norm": 14.708884239196777, "learning_rate": 2.587378640776699e-05, "loss": 0.2113, "step": 15270 }, { "epoch": 148.08192771084336, "grad_norm": 31.913820266723633, "learning_rate": 2.5868932038834953e-05, "loss": 0.2684, "step": 15271 }, { "epoch": 148.09156626506024, "grad_norm": 10.13917350769043, "learning_rate": 2.5864077669902914e-05, "loss": 0.3167, "step": 15272 }, { "epoch": 148.10120481927711, "grad_norm": 17.278175354003906, "learning_rate": 2.5859223300970875e-05, "loss": 0.2527, "step": 15273 }, { "epoch": 148.11084337349396, "grad_norm": 16.06193733215332, "learning_rate": 2.5854368932038836e-05, "loss": 0.2331, "step": 15274 }, { "epoch": 148.12048192771084, "grad_norm": 3.540365219116211, "learning_rate": 2.5849514563106797e-05, "loss": 0.2286, "step": 15275 }, { "epoch": 148.13012048192772, "grad_norm": 14.267242431640625, "learning_rate": 2.584466019417476e-05, "loss": 0.2872, "step": 15276 }, { "epoch": 148.13975903614457, "grad_norm": 9.407334327697754, "learning_rate": 2.583980582524272e-05, "loss": 0.2725, "step": 15277 }, { "epoch": 148.14939759036145, "grad_norm": 1.6923044919967651, "learning_rate": 2.583495145631068e-05, "loss": 0.1702, "step": 15278 }, { "epoch": 148.15903614457832, "grad_norm": 9.784318923950195, "learning_rate": 2.5830097087378642e-05, "loss": 0.2504, "step": 15279 }, { "epoch": 148.16867469879517, "grad_norm": 6.985170364379883, "learning_rate": 2.5825242718446603e-05, "loss": 0.264, "step": 15280 }, { "epoch": 148.17831325301205, "grad_norm": 19.440114974975586, "learning_rate": 2.5820388349514564e-05, "loss": 0.201, "step": 15281 }, { "epoch": 148.18795180722893, "grad_norm": 25.825359344482422, "learning_rate": 2.5815533980582522e-05, "loss": 0.2784, "step": 15282 }, { "epoch": 148.19759036144578, "grad_norm": 20.74983787536621, "learning_rate": 2.5810679611650483e-05, "loss": 0.4423, "step": 15283 }, { "epoch": 148.20722891566265, "grad_norm": 17.468299865722656, "learning_rate": 2.5805825242718444e-05, "loss": 0.247, "step": 15284 }, { "epoch": 148.21686746987953, "grad_norm": 21.242950439453125, "learning_rate": 2.5800970873786405e-05, "loss": 0.229, "step": 15285 }, { "epoch": 148.22650602409638, "grad_norm": 17.793113708496094, "learning_rate": 2.5796116504854373e-05, "loss": 0.2141, "step": 15286 }, { "epoch": 148.23614457831326, "grad_norm": 34.48897933959961, "learning_rate": 2.5791262135922334e-05, "loss": 0.2366, "step": 15287 }, { "epoch": 148.2457831325301, "grad_norm": 5.654397010803223, "learning_rate": 2.5786407766990296e-05, "loss": 0.3898, "step": 15288 }, { "epoch": 148.25542168674698, "grad_norm": 10.046404838562012, "learning_rate": 2.5781553398058257e-05, "loss": 0.1938, "step": 15289 }, { "epoch": 148.26506024096386, "grad_norm": 21.873979568481445, "learning_rate": 2.5776699029126218e-05, "loss": 0.2948, "step": 15290 }, { "epoch": 148.2746987951807, "grad_norm": 11.931325912475586, "learning_rate": 2.577184466019418e-05, "loss": 0.2833, "step": 15291 }, { "epoch": 148.2843373493976, "grad_norm": 36.9046745300293, "learning_rate": 2.576699029126214e-05, "loss": 0.3426, "step": 15292 }, { "epoch": 148.29397590361447, "grad_norm": 10.651142120361328, "learning_rate": 2.5762135922330098e-05, "loss": 0.4287, "step": 15293 }, { "epoch": 148.30361445783132, "grad_norm": 35.39094543457031, "learning_rate": 2.575728155339806e-05, "loss": 0.4361, "step": 15294 }, { "epoch": 148.3132530120482, "grad_norm": 9.936187744140625, "learning_rate": 2.575242718446602e-05, "loss": 0.2862, "step": 15295 }, { "epoch": 148.32289156626507, "grad_norm": 85.79817199707031, "learning_rate": 2.574757281553398e-05, "loss": 0.349, "step": 15296 }, { "epoch": 148.33253012048192, "grad_norm": 9.572747230529785, "learning_rate": 2.5742718446601943e-05, "loss": 0.2472, "step": 15297 }, { "epoch": 148.3421686746988, "grad_norm": 20.95342445373535, "learning_rate": 2.5737864077669904e-05, "loss": 0.3348, "step": 15298 }, { "epoch": 148.35180722891567, "grad_norm": 3.9015491008758545, "learning_rate": 2.5733009708737865e-05, "loss": 0.1531, "step": 15299 }, { "epoch": 148.36144578313252, "grad_norm": 5.391621112823486, "learning_rate": 2.5728155339805826e-05, "loss": 0.1193, "step": 15300 }, { "epoch": 148.3710843373494, "grad_norm": 15.355708122253418, "learning_rate": 2.5723300970873787e-05, "loss": 0.2651, "step": 15301 }, { "epoch": 148.38072289156628, "grad_norm": 6.428369522094727, "learning_rate": 2.5718446601941748e-05, "loss": 0.2682, "step": 15302 }, { "epoch": 148.39036144578313, "grad_norm": 23.829118728637695, "learning_rate": 2.571359223300971e-05, "loss": 0.3777, "step": 15303 }, { "epoch": 148.4, "grad_norm": 11.849191665649414, "learning_rate": 2.570873786407767e-05, "loss": 0.3811, "step": 15304 }, { "epoch": 148.40963855421685, "grad_norm": 12.408865928649902, "learning_rate": 2.570388349514563e-05, "loss": 0.3213, "step": 15305 }, { "epoch": 148.41927710843373, "grad_norm": 7.830078601837158, "learning_rate": 2.5699029126213593e-05, "loss": 0.2822, "step": 15306 }, { "epoch": 148.4289156626506, "grad_norm": 18.30324363708496, "learning_rate": 2.5694174757281554e-05, "loss": 0.4819, "step": 15307 }, { "epoch": 148.43855421686746, "grad_norm": 5.856470108032227, "learning_rate": 2.5689320388349515e-05, "loss": 0.5012, "step": 15308 }, { "epoch": 148.44819277108434, "grad_norm": 4.866467475891113, "learning_rate": 2.5684466019417476e-05, "loss": 0.2176, "step": 15309 }, { "epoch": 148.4578313253012, "grad_norm": 5.262667655944824, "learning_rate": 2.5679611650485434e-05, "loss": 0.1137, "step": 15310 }, { "epoch": 148.46746987951806, "grad_norm": 26.759130477905273, "learning_rate": 2.5674757281553395e-05, "loss": 0.4441, "step": 15311 }, { "epoch": 148.47710843373494, "grad_norm": 8.076314926147461, "learning_rate": 2.5669902912621363e-05, "loss": 0.2628, "step": 15312 }, { "epoch": 148.48674698795182, "grad_norm": 17.401004791259766, "learning_rate": 2.5665048543689324e-05, "loss": 0.3041, "step": 15313 }, { "epoch": 148.49638554216867, "grad_norm": 12.308216094970703, "learning_rate": 2.5660194174757285e-05, "loss": 0.2154, "step": 15314 }, { "epoch": 148.50602409638554, "grad_norm": 42.6856803894043, "learning_rate": 2.5655339805825247e-05, "loss": 0.3172, "step": 15315 }, { "epoch": 148.51566265060242, "grad_norm": 22.369705200195312, "learning_rate": 2.5650485436893208e-05, "loss": 0.2091, "step": 15316 }, { "epoch": 148.52530120481927, "grad_norm": 38.99583435058594, "learning_rate": 2.564563106796117e-05, "loss": 0.2703, "step": 15317 }, { "epoch": 148.53493975903615, "grad_norm": 29.01560401916504, "learning_rate": 2.564077669902913e-05, "loss": 0.2618, "step": 15318 }, { "epoch": 148.544578313253, "grad_norm": 16.974451065063477, "learning_rate": 2.563592233009709e-05, "loss": 0.0843, "step": 15319 }, { "epoch": 148.55421686746988, "grad_norm": 34.837581634521484, "learning_rate": 2.5631067961165052e-05, "loss": 0.2572, "step": 15320 }, { "epoch": 148.56385542168675, "grad_norm": 14.811802864074707, "learning_rate": 2.562621359223301e-05, "loss": 0.3299, "step": 15321 }, { "epoch": 148.5734939759036, "grad_norm": 28.045602798461914, "learning_rate": 2.562135922330097e-05, "loss": 0.352, "step": 15322 }, { "epoch": 148.58313253012048, "grad_norm": 31.166492462158203, "learning_rate": 2.5616504854368932e-05, "loss": 0.2233, "step": 15323 }, { "epoch": 148.59277108433736, "grad_norm": 7.789793491363525, "learning_rate": 2.5611650485436893e-05, "loss": 0.267, "step": 15324 }, { "epoch": 148.6024096385542, "grad_norm": 11.092650413513184, "learning_rate": 2.5606796116504855e-05, "loss": 0.3219, "step": 15325 }, { "epoch": 148.61204819277108, "grad_norm": 10.746382713317871, "learning_rate": 2.5601941747572816e-05, "loss": 0.5118, "step": 15326 }, { "epoch": 148.62168674698796, "grad_norm": 15.021106719970703, "learning_rate": 2.5597087378640777e-05, "loss": 0.2549, "step": 15327 }, { "epoch": 148.6313253012048, "grad_norm": 6.323270320892334, "learning_rate": 2.5592233009708738e-05, "loss": 0.2561, "step": 15328 }, { "epoch": 148.6409638554217, "grad_norm": 4.437155246734619, "learning_rate": 2.55873786407767e-05, "loss": 0.2168, "step": 15329 }, { "epoch": 148.65060240963857, "grad_norm": 26.460142135620117, "learning_rate": 2.558252427184466e-05, "loss": 0.2466, "step": 15330 }, { "epoch": 148.66024096385541, "grad_norm": 33.81858444213867, "learning_rate": 2.557766990291262e-05, "loss": 0.2202, "step": 15331 }, { "epoch": 148.6698795180723, "grad_norm": 22.474956512451172, "learning_rate": 2.5572815533980583e-05, "loss": 0.2407, "step": 15332 }, { "epoch": 148.67951807228917, "grad_norm": 27.69304847717285, "learning_rate": 2.5567961165048544e-05, "loss": 0.3851, "step": 15333 }, { "epoch": 148.68915662650602, "grad_norm": 5.983154773712158, "learning_rate": 2.5563106796116505e-05, "loss": 0.3005, "step": 15334 }, { "epoch": 148.6987951807229, "grad_norm": 20.975421905517578, "learning_rate": 2.5558252427184466e-05, "loss": 0.2616, "step": 15335 }, { "epoch": 148.70843373493975, "grad_norm": 23.380538940429688, "learning_rate": 2.5553398058252427e-05, "loss": 0.2898, "step": 15336 }, { "epoch": 148.71807228915662, "grad_norm": 15.0075101852417, "learning_rate": 2.5548543689320392e-05, "loss": 0.2115, "step": 15337 }, { "epoch": 148.7277108433735, "grad_norm": 25.02602195739746, "learning_rate": 2.5543689320388353e-05, "loss": 0.2022, "step": 15338 }, { "epoch": 148.73734939759035, "grad_norm": 11.634347915649414, "learning_rate": 2.5538834951456314e-05, "loss": 0.3059, "step": 15339 }, { "epoch": 148.74698795180723, "grad_norm": 12.074411392211914, "learning_rate": 2.5533980582524275e-05, "loss": 0.2293, "step": 15340 }, { "epoch": 148.7566265060241, "grad_norm": 12.447251319885254, "learning_rate": 2.5529126213592236e-05, "loss": 0.212, "step": 15341 }, { "epoch": 148.76626506024095, "grad_norm": 17.333953857421875, "learning_rate": 2.5524271844660197e-05, "loss": 0.5857, "step": 15342 }, { "epoch": 148.77590361445783, "grad_norm": 4.501555919647217, "learning_rate": 2.551941747572816e-05, "loss": 0.2109, "step": 15343 }, { "epoch": 148.7855421686747, "grad_norm": 9.087417602539062, "learning_rate": 2.551456310679612e-05, "loss": 0.1229, "step": 15344 }, { "epoch": 148.79518072289156, "grad_norm": 45.13418197631836, "learning_rate": 2.550970873786408e-05, "loss": 0.3435, "step": 15345 }, { "epoch": 148.80481927710844, "grad_norm": 22.760311126708984, "learning_rate": 2.5504854368932042e-05, "loss": 0.4178, "step": 15346 }, { "epoch": 148.8144578313253, "grad_norm": 28.176584243774414, "learning_rate": 2.5500000000000003e-05, "loss": 0.3689, "step": 15347 }, { "epoch": 148.82409638554216, "grad_norm": 75.3061294555664, "learning_rate": 2.549514563106796e-05, "loss": 0.2842, "step": 15348 }, { "epoch": 148.83373493975904, "grad_norm": 23.1414852142334, "learning_rate": 2.5490291262135922e-05, "loss": 0.3294, "step": 15349 }, { "epoch": 148.84337349397592, "grad_norm": 13.320740699768066, "learning_rate": 2.5485436893203883e-05, "loss": 0.274, "step": 15350 }, { "epoch": 148.85301204819277, "grad_norm": 14.0547513961792, "learning_rate": 2.5480582524271844e-05, "loss": 0.4358, "step": 15351 }, { "epoch": 148.86265060240964, "grad_norm": 59.323760986328125, "learning_rate": 2.5475728155339806e-05, "loss": 0.4156, "step": 15352 }, { "epoch": 148.8722891566265, "grad_norm": 2.589944839477539, "learning_rate": 2.5470873786407767e-05, "loss": 0.2683, "step": 15353 }, { "epoch": 148.88192771084337, "grad_norm": 5.816844940185547, "learning_rate": 2.5466019417475728e-05, "loss": 0.2874, "step": 15354 }, { "epoch": 148.89156626506025, "grad_norm": 7.898524761199951, "learning_rate": 2.546116504854369e-05, "loss": 0.26, "step": 15355 }, { "epoch": 148.9012048192771, "grad_norm": 12.9630708694458, "learning_rate": 2.545631067961165e-05, "loss": 0.2887, "step": 15356 }, { "epoch": 148.91084337349398, "grad_norm": 19.43589973449707, "learning_rate": 2.545145631067961e-05, "loss": 0.1316, "step": 15357 }, { "epoch": 148.92048192771085, "grad_norm": 3.1229307651519775, "learning_rate": 2.5446601941747572e-05, "loss": 0.2408, "step": 15358 }, { "epoch": 148.9301204819277, "grad_norm": 7.520977020263672, "learning_rate": 2.5441747572815534e-05, "loss": 0.1629, "step": 15359 }, { "epoch": 148.93975903614458, "grad_norm": 15.293854713439941, "learning_rate": 2.5436893203883495e-05, "loss": 0.1405, "step": 15360 }, { "epoch": 148.94939759036146, "grad_norm": 6.5751543045043945, "learning_rate": 2.5432038834951456e-05, "loss": 0.2908, "step": 15361 }, { "epoch": 148.9590361445783, "grad_norm": 11.850476264953613, "learning_rate": 2.542718446601942e-05, "loss": 0.3127, "step": 15362 }, { "epoch": 148.96867469879518, "grad_norm": 15.318129539489746, "learning_rate": 2.542233009708738e-05, "loss": 0.1648, "step": 15363 }, { "epoch": 148.97831325301206, "grad_norm": 16.813247680664062, "learning_rate": 2.5417475728155343e-05, "loss": 0.2525, "step": 15364 }, { "epoch": 148.9879518072289, "grad_norm": 20.68073081970215, "learning_rate": 2.5412621359223304e-05, "loss": 0.1835, "step": 15365 }, { "epoch": 149.00361445783133, "grad_norm": 34.4488639831543, "learning_rate": 2.5407766990291265e-05, "loss": 0.3075, "step": 15366 }, { "epoch": 149.01325301204818, "grad_norm": 17.835098266601562, "learning_rate": 2.5402912621359226e-05, "loss": 0.3195, "step": 15367 }, { "epoch": 149.02289156626506, "grad_norm": 6.091256141662598, "learning_rate": 2.5398058252427187e-05, "loss": 0.2023, "step": 15368 }, { "epoch": 149.03253012048194, "grad_norm": 18.73868751525879, "learning_rate": 2.539320388349515e-05, "loss": 0.2249, "step": 15369 }, { "epoch": 149.0421686746988, "grad_norm": 17.25912094116211, "learning_rate": 2.538834951456311e-05, "loss": 0.1712, "step": 15370 }, { "epoch": 149.05180722891566, "grad_norm": 11.117020606994629, "learning_rate": 2.538349514563107e-05, "loss": 0.2247, "step": 15371 }, { "epoch": 149.06144578313254, "grad_norm": 12.585320472717285, "learning_rate": 2.5378640776699032e-05, "loss": 0.1697, "step": 15372 }, { "epoch": 149.0710843373494, "grad_norm": 6.6477861404418945, "learning_rate": 2.5373786407766993e-05, "loss": 0.2789, "step": 15373 }, { "epoch": 149.08072289156627, "grad_norm": 11.052391052246094, "learning_rate": 2.5368932038834954e-05, "loss": 0.2128, "step": 15374 }, { "epoch": 149.09036144578315, "grad_norm": 11.561795234680176, "learning_rate": 2.5364077669902915e-05, "loss": 0.3816, "step": 15375 }, { "epoch": 149.1, "grad_norm": 30.09121322631836, "learning_rate": 2.5359223300970873e-05, "loss": 0.212, "step": 15376 }, { "epoch": 149.10963855421687, "grad_norm": 22.099790573120117, "learning_rate": 2.5354368932038834e-05, "loss": 0.165, "step": 15377 }, { "epoch": 149.11927710843372, "grad_norm": 5.284379005432129, "learning_rate": 2.5349514563106795e-05, "loss": 0.1729, "step": 15378 }, { "epoch": 149.1289156626506, "grad_norm": 6.544186592102051, "learning_rate": 2.5344660194174756e-05, "loss": 0.181, "step": 15379 }, { "epoch": 149.13855421686748, "grad_norm": 17.28357696533203, "learning_rate": 2.5339805825242718e-05, "loss": 0.4695, "step": 15380 }, { "epoch": 149.14819277108433, "grad_norm": 14.712794303894043, "learning_rate": 2.533495145631068e-05, "loss": 0.4349, "step": 15381 }, { "epoch": 149.1578313253012, "grad_norm": 11.780733108520508, "learning_rate": 2.533009708737864e-05, "loss": 0.2301, "step": 15382 }, { "epoch": 149.16746987951808, "grad_norm": 10.766195297241211, "learning_rate": 2.53252427184466e-05, "loss": 0.269, "step": 15383 }, { "epoch": 149.17710843373493, "grad_norm": 17.171640396118164, "learning_rate": 2.5320388349514562e-05, "loss": 0.2004, "step": 15384 }, { "epoch": 149.1867469879518, "grad_norm": 7.5592265129089355, "learning_rate": 2.5315533980582523e-05, "loss": 0.2383, "step": 15385 }, { "epoch": 149.19638554216868, "grad_norm": 43.186073303222656, "learning_rate": 2.5310679611650484e-05, "loss": 0.3041, "step": 15386 }, { "epoch": 149.20602409638553, "grad_norm": 9.355886459350586, "learning_rate": 2.530582524271845e-05, "loss": 0.3267, "step": 15387 }, { "epoch": 149.2156626506024, "grad_norm": 17.607463836669922, "learning_rate": 2.530097087378641e-05, "loss": 0.2985, "step": 15388 }, { "epoch": 149.2253012048193, "grad_norm": 27.352079391479492, "learning_rate": 2.529611650485437e-05, "loss": 0.2369, "step": 15389 }, { "epoch": 149.23493975903614, "grad_norm": 9.762240409851074, "learning_rate": 2.5291262135922332e-05, "loss": 0.3859, "step": 15390 }, { "epoch": 149.24457831325302, "grad_norm": 6.712109565734863, "learning_rate": 2.5286407766990294e-05, "loss": 0.3414, "step": 15391 }, { "epoch": 149.2542168674699, "grad_norm": 23.981657028198242, "learning_rate": 2.5281553398058255e-05, "loss": 0.3039, "step": 15392 }, { "epoch": 149.26385542168674, "grad_norm": 30.781740188598633, "learning_rate": 2.5276699029126216e-05, "loss": 0.2522, "step": 15393 }, { "epoch": 149.27349397590362, "grad_norm": 4.393776893615723, "learning_rate": 2.5271844660194177e-05, "loss": 0.1118, "step": 15394 }, { "epoch": 149.28313253012047, "grad_norm": 6.970836162567139, "learning_rate": 2.5266990291262138e-05, "loss": 0.2491, "step": 15395 }, { "epoch": 149.29277108433735, "grad_norm": 34.36678695678711, "learning_rate": 2.52621359223301e-05, "loss": 0.2103, "step": 15396 }, { "epoch": 149.30240963855422, "grad_norm": 27.430002212524414, "learning_rate": 2.525728155339806e-05, "loss": 0.1697, "step": 15397 }, { "epoch": 149.31204819277107, "grad_norm": 7.095077991485596, "learning_rate": 2.525242718446602e-05, "loss": 0.2387, "step": 15398 }, { "epoch": 149.32168674698795, "grad_norm": 17.233266830444336, "learning_rate": 2.5247572815533983e-05, "loss": 0.3731, "step": 15399 }, { "epoch": 149.33132530120483, "grad_norm": 6.576153755187988, "learning_rate": 2.5242718446601944e-05, "loss": 0.1939, "step": 15400 }, { "epoch": 149.34096385542168, "grad_norm": 41.22308349609375, "learning_rate": 2.5237864077669905e-05, "loss": 0.4195, "step": 15401 }, { "epoch": 149.35060240963855, "grad_norm": 15.187723159790039, "learning_rate": 2.5233009708737866e-05, "loss": 0.4412, "step": 15402 }, { "epoch": 149.36024096385543, "grad_norm": 35.31398391723633, "learning_rate": 2.5228155339805827e-05, "loss": 0.1724, "step": 15403 }, { "epoch": 149.36987951807228, "grad_norm": 12.624879837036133, "learning_rate": 2.5223300970873785e-05, "loss": 0.196, "step": 15404 }, { "epoch": 149.37951807228916, "grad_norm": 17.896047592163086, "learning_rate": 2.5218446601941746e-05, "loss": 0.2607, "step": 15405 }, { "epoch": 149.38915662650604, "grad_norm": 4.892092704772949, "learning_rate": 2.5213592233009707e-05, "loss": 0.2032, "step": 15406 }, { "epoch": 149.39879518072289, "grad_norm": 11.767800331115723, "learning_rate": 2.520873786407767e-05, "loss": 0.2467, "step": 15407 }, { "epoch": 149.40843373493976, "grad_norm": 12.65473461151123, "learning_rate": 2.520388349514563e-05, "loss": 0.3882, "step": 15408 }, { "epoch": 149.41807228915664, "grad_norm": 10.36780834197998, "learning_rate": 2.519902912621359e-05, "loss": 0.1899, "step": 15409 }, { "epoch": 149.4277108433735, "grad_norm": 45.37297821044922, "learning_rate": 2.5194174757281552e-05, "loss": 0.5266, "step": 15410 }, { "epoch": 149.43734939759037, "grad_norm": 8.133523941040039, "learning_rate": 2.5189320388349513e-05, "loss": 0.1574, "step": 15411 }, { "epoch": 149.44698795180722, "grad_norm": 14.022737503051758, "learning_rate": 2.5184466019417474e-05, "loss": 0.2667, "step": 15412 }, { "epoch": 149.4566265060241, "grad_norm": 5.214569568634033, "learning_rate": 2.5179611650485442e-05, "loss": 0.2172, "step": 15413 }, { "epoch": 149.46626506024097, "grad_norm": 14.175561904907227, "learning_rate": 2.51747572815534e-05, "loss": 0.1895, "step": 15414 }, { "epoch": 149.47590361445782, "grad_norm": 21.65380096435547, "learning_rate": 2.516990291262136e-05, "loss": 0.329, "step": 15415 }, { "epoch": 149.4855421686747, "grad_norm": 4.503164291381836, "learning_rate": 2.5165048543689322e-05, "loss": 0.2912, "step": 15416 }, { "epoch": 149.49518072289158, "grad_norm": 6.586244583129883, "learning_rate": 2.5160194174757283e-05, "loss": 0.1902, "step": 15417 }, { "epoch": 149.50481927710842, "grad_norm": 16.68088150024414, "learning_rate": 2.5155339805825245e-05, "loss": 0.2894, "step": 15418 }, { "epoch": 149.5144578313253, "grad_norm": 11.038813591003418, "learning_rate": 2.5150485436893206e-05, "loss": 0.3126, "step": 15419 }, { "epoch": 149.52409638554218, "grad_norm": 17.42481231689453, "learning_rate": 2.5145631067961167e-05, "loss": 0.2947, "step": 15420 }, { "epoch": 149.53373493975903, "grad_norm": 9.440446853637695, "learning_rate": 2.5140776699029128e-05, "loss": 0.3522, "step": 15421 }, { "epoch": 149.5433734939759, "grad_norm": 7.984069347381592, "learning_rate": 2.513592233009709e-05, "loss": 0.2478, "step": 15422 }, { "epoch": 149.55301204819278, "grad_norm": 7.481829643249512, "learning_rate": 2.513106796116505e-05, "loss": 0.2082, "step": 15423 }, { "epoch": 149.56265060240963, "grad_norm": 9.389983177185059, "learning_rate": 2.512621359223301e-05, "loss": 0.2586, "step": 15424 }, { "epoch": 149.5722891566265, "grad_norm": 28.38123893737793, "learning_rate": 2.5121359223300973e-05, "loss": 0.0853, "step": 15425 }, { "epoch": 149.58192771084336, "grad_norm": 12.871728897094727, "learning_rate": 2.5116504854368934e-05, "loss": 0.2436, "step": 15426 }, { "epoch": 149.59156626506024, "grad_norm": 14.988411903381348, "learning_rate": 2.5111650485436895e-05, "loss": 0.4632, "step": 15427 }, { "epoch": 149.60120481927711, "grad_norm": 29.069997787475586, "learning_rate": 2.5106796116504856e-05, "loss": 0.1472, "step": 15428 }, { "epoch": 149.61084337349396, "grad_norm": 5.522287368774414, "learning_rate": 2.5101941747572817e-05, "loss": 0.2862, "step": 15429 }, { "epoch": 149.62048192771084, "grad_norm": 7.1152143478393555, "learning_rate": 2.5097087378640778e-05, "loss": 0.4297, "step": 15430 }, { "epoch": 149.63012048192772, "grad_norm": 8.649794578552246, "learning_rate": 2.5092233009708736e-05, "loss": 0.3581, "step": 15431 }, { "epoch": 149.63975903614457, "grad_norm": 5.872002601623535, "learning_rate": 2.5087378640776697e-05, "loss": 0.173, "step": 15432 }, { "epoch": 149.64939759036145, "grad_norm": 14.275562286376953, "learning_rate": 2.508252427184466e-05, "loss": 0.3736, "step": 15433 }, { "epoch": 149.65903614457832, "grad_norm": 7.338568210601807, "learning_rate": 2.507766990291262e-05, "loss": 0.099, "step": 15434 }, { "epoch": 149.66867469879517, "grad_norm": 25.920909881591797, "learning_rate": 2.507281553398058e-05, "loss": 0.6247, "step": 15435 }, { "epoch": 149.67831325301205, "grad_norm": 19.958887100219727, "learning_rate": 2.5067961165048542e-05, "loss": 0.2121, "step": 15436 }, { "epoch": 149.68795180722893, "grad_norm": 14.706489562988281, "learning_rate": 2.5063106796116503e-05, "loss": 0.2799, "step": 15437 }, { "epoch": 149.69759036144578, "grad_norm": 90.06537628173828, "learning_rate": 2.505825242718447e-05, "loss": 0.2247, "step": 15438 }, { "epoch": 149.70722891566265, "grad_norm": 40.05977249145508, "learning_rate": 2.5053398058252432e-05, "loss": 0.2256, "step": 15439 }, { "epoch": 149.71686746987953, "grad_norm": 7.402284622192383, "learning_rate": 2.5048543689320393e-05, "loss": 0.227, "step": 15440 }, { "epoch": 149.72650602409638, "grad_norm": 10.888375282287598, "learning_rate": 2.5043689320388354e-05, "loss": 0.1895, "step": 15441 }, { "epoch": 149.73614457831326, "grad_norm": 8.440855979919434, "learning_rate": 2.5038834951456312e-05, "loss": 0.3365, "step": 15442 }, { "epoch": 149.7457831325301, "grad_norm": 25.072532653808594, "learning_rate": 2.5033980582524273e-05, "loss": 0.2368, "step": 15443 }, { "epoch": 149.75542168674698, "grad_norm": 12.952613830566406, "learning_rate": 2.5029126213592234e-05, "loss": 0.1865, "step": 15444 }, { "epoch": 149.76506024096386, "grad_norm": 4.238292217254639, "learning_rate": 2.5024271844660195e-05, "loss": 0.2014, "step": 15445 }, { "epoch": 149.7746987951807, "grad_norm": 5.059203624725342, "learning_rate": 2.5019417475728157e-05, "loss": 0.1437, "step": 15446 }, { "epoch": 149.7843373493976, "grad_norm": 20.11317253112793, "learning_rate": 2.5014563106796118e-05, "loss": 0.3865, "step": 15447 }, { "epoch": 149.79397590361447, "grad_norm": 49.84792709350586, "learning_rate": 2.500970873786408e-05, "loss": 0.1285, "step": 15448 }, { "epoch": 149.80361445783132, "grad_norm": 18.236839294433594, "learning_rate": 2.500485436893204e-05, "loss": 0.4915, "step": 15449 }, { "epoch": 149.8132530120482, "grad_norm": 6.151547431945801, "learning_rate": 2.5e-05, "loss": 0.2788, "step": 15450 }, { "epoch": 149.82289156626507, "grad_norm": 7.726411819458008, "learning_rate": 2.4995145631067962e-05, "loss": 0.1557, "step": 15451 }, { "epoch": 149.83253012048192, "grad_norm": 25.163496017456055, "learning_rate": 2.4990291262135923e-05, "loss": 0.2726, "step": 15452 }, { "epoch": 149.8421686746988, "grad_norm": 17.945146560668945, "learning_rate": 2.4985436893203885e-05, "loss": 0.1558, "step": 15453 }, { "epoch": 149.85180722891567, "grad_norm": 41.792327880859375, "learning_rate": 2.4980582524271846e-05, "loss": 0.248, "step": 15454 }, { "epoch": 149.86144578313252, "grad_norm": 155.35206604003906, "learning_rate": 2.4975728155339807e-05, "loss": 0.2423, "step": 15455 }, { "epoch": 149.8710843373494, "grad_norm": 20.156383514404297, "learning_rate": 2.4970873786407768e-05, "loss": 0.5273, "step": 15456 }, { "epoch": 149.88072289156628, "grad_norm": 8.485408782958984, "learning_rate": 2.496601941747573e-05, "loss": 0.252, "step": 15457 }, { "epoch": 149.89036144578313, "grad_norm": 10.996192932128906, "learning_rate": 2.496116504854369e-05, "loss": 0.2683, "step": 15458 }, { "epoch": 149.9, "grad_norm": 6.303364276885986, "learning_rate": 2.495631067961165e-05, "loss": 0.2295, "step": 15459 }, { "epoch": 149.90963855421685, "grad_norm": 18.918472290039062, "learning_rate": 2.4951456310679613e-05, "loss": 0.2207, "step": 15460 }, { "epoch": 149.91927710843373, "grad_norm": 5.8763227462768555, "learning_rate": 2.4946601941747574e-05, "loss": 0.3968, "step": 15461 }, { "epoch": 149.9289156626506, "grad_norm": 10.708406448364258, "learning_rate": 2.4941747572815535e-05, "loss": 0.1775, "step": 15462 }, { "epoch": 149.93855421686746, "grad_norm": 27.77614402770996, "learning_rate": 2.4936893203883496e-05, "loss": 0.2547, "step": 15463 }, { "epoch": 149.94819277108434, "grad_norm": 64.65288543701172, "learning_rate": 2.4932038834951457e-05, "loss": 0.2709, "step": 15464 }, { "epoch": 149.9578313253012, "grad_norm": 10.237765312194824, "learning_rate": 2.492718446601942e-05, "loss": 0.2567, "step": 15465 }, { "epoch": 149.96746987951806, "grad_norm": 14.330127716064453, "learning_rate": 2.492233009708738e-05, "loss": 0.2882, "step": 15466 }, { "epoch": 149.97710843373494, "grad_norm": 11.048579216003418, "learning_rate": 2.491747572815534e-05, "loss": 0.3538, "step": 15467 }, { "epoch": 149.98674698795182, "grad_norm": 23.889923095703125, "learning_rate": 2.4912621359223302e-05, "loss": 0.2334, "step": 15468 }, { "epoch": 150.0024096385542, "grad_norm": 38.86013412475586, "learning_rate": 2.4907766990291266e-05, "loss": 0.2364, "step": 15469 }, { "epoch": 150.0120481927711, "grad_norm": 11.337013244628906, "learning_rate": 2.4902912621359224e-05, "loss": 0.2905, "step": 15470 }, { "epoch": 150.02168674698794, "grad_norm": 14.966612815856934, "learning_rate": 2.4898058252427185e-05, "loss": 0.4628, "step": 15471 }, { "epoch": 150.03132530120482, "grad_norm": 19.043697357177734, "learning_rate": 2.4893203883495146e-05, "loss": 0.2082, "step": 15472 }, { "epoch": 150.0409638554217, "grad_norm": 10.11125659942627, "learning_rate": 2.4888349514563108e-05, "loss": 0.3051, "step": 15473 }, { "epoch": 150.05060240963854, "grad_norm": 14.743233680725098, "learning_rate": 2.488349514563107e-05, "loss": 0.289, "step": 15474 }, { "epoch": 150.06024096385542, "grad_norm": 34.24222946166992, "learning_rate": 2.487864077669903e-05, "loss": 0.2193, "step": 15475 }, { "epoch": 150.0698795180723, "grad_norm": 4.611527919769287, "learning_rate": 2.487378640776699e-05, "loss": 0.2277, "step": 15476 }, { "epoch": 150.07951807228915, "grad_norm": 12.166413307189941, "learning_rate": 2.4868932038834952e-05, "loss": 0.3786, "step": 15477 }, { "epoch": 150.08915662650602, "grad_norm": 6.543128490447998, "learning_rate": 2.4864077669902913e-05, "loss": 0.3025, "step": 15478 }, { "epoch": 150.0987951807229, "grad_norm": 16.661651611328125, "learning_rate": 2.4859223300970874e-05, "loss": 0.2153, "step": 15479 }, { "epoch": 150.10843373493975, "grad_norm": 10.27934741973877, "learning_rate": 2.4854368932038836e-05, "loss": 0.3579, "step": 15480 }, { "epoch": 150.11807228915663, "grad_norm": 11.591361045837402, "learning_rate": 2.4849514563106797e-05, "loss": 0.2634, "step": 15481 }, { "epoch": 150.1277108433735, "grad_norm": 28.19374656677246, "learning_rate": 2.484466019417476e-05, "loss": 0.2126, "step": 15482 }, { "epoch": 150.13734939759036, "grad_norm": 30.84600067138672, "learning_rate": 2.4839805825242722e-05, "loss": 0.2925, "step": 15483 }, { "epoch": 150.14698795180723, "grad_norm": 5.157144546508789, "learning_rate": 2.483495145631068e-05, "loss": 0.2052, "step": 15484 }, { "epoch": 150.15662650602408, "grad_norm": 5.998959541320801, "learning_rate": 2.483009708737864e-05, "loss": 0.1749, "step": 15485 }, { "epoch": 150.16626506024096, "grad_norm": 54.8795166015625, "learning_rate": 2.4825242718446602e-05, "loss": 0.3239, "step": 15486 }, { "epoch": 150.17590361445784, "grad_norm": 7.056516170501709, "learning_rate": 2.4820388349514564e-05, "loss": 0.2487, "step": 15487 }, { "epoch": 150.1855421686747, "grad_norm": 14.366945266723633, "learning_rate": 2.4815533980582525e-05, "loss": 0.2261, "step": 15488 }, { "epoch": 150.19518072289156, "grad_norm": 14.721281051635742, "learning_rate": 2.4810679611650486e-05, "loss": 0.2022, "step": 15489 }, { "epoch": 150.20481927710844, "grad_norm": 4.449299335479736, "learning_rate": 2.4805825242718447e-05, "loss": 0.1505, "step": 15490 }, { "epoch": 150.2144578313253, "grad_norm": 20.22700309753418, "learning_rate": 2.4800970873786408e-05, "loss": 0.2682, "step": 15491 }, { "epoch": 150.22409638554217, "grad_norm": 6.850521087646484, "learning_rate": 2.479611650485437e-05, "loss": 0.1023, "step": 15492 }, { "epoch": 150.23373493975905, "grad_norm": 27.640226364135742, "learning_rate": 2.479126213592233e-05, "loss": 0.3029, "step": 15493 }, { "epoch": 150.2433734939759, "grad_norm": 7.684940338134766, "learning_rate": 2.478640776699029e-05, "loss": 0.2755, "step": 15494 }, { "epoch": 150.25301204819277, "grad_norm": 1.009073257446289, "learning_rate": 2.4781553398058256e-05, "loss": 0.0711, "step": 15495 }, { "epoch": 150.26265060240965, "grad_norm": 8.66852855682373, "learning_rate": 2.4776699029126217e-05, "loss": 0.1971, "step": 15496 }, { "epoch": 150.2722891566265, "grad_norm": 16.160133361816406, "learning_rate": 2.4771844660194175e-05, "loss": 0.1425, "step": 15497 }, { "epoch": 150.28192771084338, "grad_norm": 16.323715209960938, "learning_rate": 2.4766990291262136e-05, "loss": 0.1617, "step": 15498 }, { "epoch": 150.29156626506025, "grad_norm": 6.077288627624512, "learning_rate": 2.4762135922330097e-05, "loss": 0.2433, "step": 15499 }, { "epoch": 150.3012048192771, "grad_norm": 11.71872615814209, "learning_rate": 2.475728155339806e-05, "loss": 0.3078, "step": 15500 }, { "epoch": 150.31084337349398, "grad_norm": 14.686464309692383, "learning_rate": 2.475242718446602e-05, "loss": 0.1316, "step": 15501 }, { "epoch": 150.32048192771083, "grad_norm": 3.0335724353790283, "learning_rate": 2.474757281553398e-05, "loss": 0.1371, "step": 15502 }, { "epoch": 150.3301204819277, "grad_norm": 36.15855407714844, "learning_rate": 2.4742718446601942e-05, "loss": 0.3393, "step": 15503 }, { "epoch": 150.33975903614459, "grad_norm": 12.734237670898438, "learning_rate": 2.4737864077669903e-05, "loss": 0.2639, "step": 15504 }, { "epoch": 150.34939759036143, "grad_norm": 14.227609634399414, "learning_rate": 2.4733009708737864e-05, "loss": 0.4807, "step": 15505 }, { "epoch": 150.3590361445783, "grad_norm": 11.835795402526855, "learning_rate": 2.4728155339805825e-05, "loss": 0.2439, "step": 15506 }, { "epoch": 150.3686746987952, "grad_norm": 79.19133758544922, "learning_rate": 2.472330097087379e-05, "loss": 0.3707, "step": 15507 }, { "epoch": 150.37831325301204, "grad_norm": 8.270841598510742, "learning_rate": 2.471844660194175e-05, "loss": 0.276, "step": 15508 }, { "epoch": 150.38795180722892, "grad_norm": 9.480039596557617, "learning_rate": 2.4713592233009712e-05, "loss": 0.2025, "step": 15509 }, { "epoch": 150.3975903614458, "grad_norm": 14.108437538146973, "learning_rate": 2.4708737864077673e-05, "loss": 0.345, "step": 15510 }, { "epoch": 150.40722891566264, "grad_norm": 9.733097076416016, "learning_rate": 2.470388349514563e-05, "loss": 0.1944, "step": 15511 }, { "epoch": 150.41686746987952, "grad_norm": 24.220745086669922, "learning_rate": 2.4699029126213592e-05, "loss": 0.2115, "step": 15512 }, { "epoch": 150.4265060240964, "grad_norm": 6.59837532043457, "learning_rate": 2.4694174757281553e-05, "loss": 0.1766, "step": 15513 }, { "epoch": 150.43614457831325, "grad_norm": 6.977766990661621, "learning_rate": 2.4689320388349515e-05, "loss": 0.0935, "step": 15514 }, { "epoch": 150.44578313253012, "grad_norm": 11.801860809326172, "learning_rate": 2.4684466019417476e-05, "loss": 0.2487, "step": 15515 }, { "epoch": 150.455421686747, "grad_norm": 21.367738723754883, "learning_rate": 2.4679611650485437e-05, "loss": 0.1085, "step": 15516 }, { "epoch": 150.46506024096385, "grad_norm": 13.972026824951172, "learning_rate": 2.4674757281553398e-05, "loss": 0.2513, "step": 15517 }, { "epoch": 150.47469879518073, "grad_norm": 9.545294761657715, "learning_rate": 2.466990291262136e-05, "loss": 0.2174, "step": 15518 }, { "epoch": 150.48433734939758, "grad_norm": 15.35562515258789, "learning_rate": 2.466504854368932e-05, "loss": 0.2811, "step": 15519 }, { "epoch": 150.49397590361446, "grad_norm": 5.543145179748535, "learning_rate": 2.4660194174757285e-05, "loss": 0.4885, "step": 15520 }, { "epoch": 150.50361445783133, "grad_norm": 13.835529327392578, "learning_rate": 2.4655339805825246e-05, "loss": 0.2665, "step": 15521 }, { "epoch": 150.51325301204818, "grad_norm": 10.108048439025879, "learning_rate": 2.4650485436893207e-05, "loss": 0.2831, "step": 15522 }, { "epoch": 150.52289156626506, "grad_norm": 8.924578666687012, "learning_rate": 2.4645631067961168e-05, "loss": 0.203, "step": 15523 }, { "epoch": 150.53253012048194, "grad_norm": 7.107568740844727, "learning_rate": 2.464077669902913e-05, "loss": 0.2099, "step": 15524 }, { "epoch": 150.5421686746988, "grad_norm": 85.14092254638672, "learning_rate": 2.4635922330097087e-05, "loss": 0.3929, "step": 15525 }, { "epoch": 150.55180722891566, "grad_norm": 5.467709541320801, "learning_rate": 2.4631067961165048e-05, "loss": 0.1145, "step": 15526 }, { "epoch": 150.56144578313254, "grad_norm": 6.767571926116943, "learning_rate": 2.462621359223301e-05, "loss": 0.1121, "step": 15527 }, { "epoch": 150.5710843373494, "grad_norm": 7.670719623565674, "learning_rate": 2.462135922330097e-05, "loss": 0.2604, "step": 15528 }, { "epoch": 150.58072289156627, "grad_norm": 26.452787399291992, "learning_rate": 2.4616504854368932e-05, "loss": 0.2541, "step": 15529 }, { "epoch": 150.59036144578315, "grad_norm": 23.540773391723633, "learning_rate": 2.4611650485436893e-05, "loss": 0.5846, "step": 15530 }, { "epoch": 150.6, "grad_norm": 19.54747200012207, "learning_rate": 2.4606796116504854e-05, "loss": 0.5176, "step": 15531 }, { "epoch": 150.60963855421687, "grad_norm": 9.383055686950684, "learning_rate": 2.4601941747572815e-05, "loss": 0.1685, "step": 15532 }, { "epoch": 150.61927710843372, "grad_norm": 13.85556697845459, "learning_rate": 2.459708737864078e-05, "loss": 0.3877, "step": 15533 }, { "epoch": 150.6289156626506, "grad_norm": 14.337550163269043, "learning_rate": 2.459223300970874e-05, "loss": 0.346, "step": 15534 }, { "epoch": 150.63855421686748, "grad_norm": 27.33216094970703, "learning_rate": 2.4587378640776702e-05, "loss": 0.4017, "step": 15535 }, { "epoch": 150.64819277108433, "grad_norm": 15.952474594116211, "learning_rate": 2.4582524271844663e-05, "loss": 0.325, "step": 15536 }, { "epoch": 150.6578313253012, "grad_norm": 10.973945617675781, "learning_rate": 2.4577669902912624e-05, "loss": 0.5258, "step": 15537 }, { "epoch": 150.66746987951808, "grad_norm": 34.29513931274414, "learning_rate": 2.4572815533980585e-05, "loss": 0.4969, "step": 15538 }, { "epoch": 150.67710843373493, "grad_norm": 8.467442512512207, "learning_rate": 2.4567961165048543e-05, "loss": 0.2683, "step": 15539 }, { "epoch": 150.6867469879518, "grad_norm": 5.762918472290039, "learning_rate": 2.4563106796116504e-05, "loss": 0.2442, "step": 15540 }, { "epoch": 150.69638554216868, "grad_norm": 6.169698715209961, "learning_rate": 2.4558252427184465e-05, "loss": 0.3167, "step": 15541 }, { "epoch": 150.70602409638553, "grad_norm": 8.797184944152832, "learning_rate": 2.4553398058252427e-05, "loss": 0.171, "step": 15542 }, { "epoch": 150.7156626506024, "grad_norm": 12.68108081817627, "learning_rate": 2.4548543689320388e-05, "loss": 0.2426, "step": 15543 }, { "epoch": 150.7253012048193, "grad_norm": 10.720410346984863, "learning_rate": 2.454368932038835e-05, "loss": 0.283, "step": 15544 }, { "epoch": 150.73493975903614, "grad_norm": 9.215028762817383, "learning_rate": 2.4538834951456313e-05, "loss": 0.2429, "step": 15545 }, { "epoch": 150.74457831325302, "grad_norm": 7.615420341491699, "learning_rate": 2.4533980582524275e-05, "loss": 0.2412, "step": 15546 }, { "epoch": 150.7542168674699, "grad_norm": 46.184486389160156, "learning_rate": 2.4529126213592236e-05, "loss": 0.5209, "step": 15547 }, { "epoch": 150.76385542168674, "grad_norm": 11.948328971862793, "learning_rate": 2.4524271844660197e-05, "loss": 0.258, "step": 15548 }, { "epoch": 150.77349397590362, "grad_norm": 8.181312561035156, "learning_rate": 2.4519417475728158e-05, "loss": 0.0936, "step": 15549 }, { "epoch": 150.78313253012047, "grad_norm": 8.874537467956543, "learning_rate": 2.451456310679612e-05, "loss": 0.2903, "step": 15550 }, { "epoch": 150.79277108433735, "grad_norm": 18.80944061279297, "learning_rate": 2.450970873786408e-05, "loss": 0.3358, "step": 15551 }, { "epoch": 150.80240963855422, "grad_norm": 16.12405014038086, "learning_rate": 2.450485436893204e-05, "loss": 0.2452, "step": 15552 }, { "epoch": 150.81204819277107, "grad_norm": 5.081758499145508, "learning_rate": 2.45e-05, "loss": 0.0944, "step": 15553 }, { "epoch": 150.82168674698795, "grad_norm": 20.666959762573242, "learning_rate": 2.449514563106796e-05, "loss": 0.3577, "step": 15554 }, { "epoch": 150.83132530120483, "grad_norm": 9.060982704162598, "learning_rate": 2.449029126213592e-05, "loss": 0.1772, "step": 15555 }, { "epoch": 150.84096385542168, "grad_norm": 8.835237503051758, "learning_rate": 2.4485436893203883e-05, "loss": 0.2414, "step": 15556 }, { "epoch": 150.85060240963855, "grad_norm": 7.21766471862793, "learning_rate": 2.4480582524271844e-05, "loss": 0.1112, "step": 15557 }, { "epoch": 150.86024096385543, "grad_norm": 19.24690055847168, "learning_rate": 2.447572815533981e-05, "loss": 0.2056, "step": 15558 }, { "epoch": 150.86987951807228, "grad_norm": 5.805281162261963, "learning_rate": 2.447087378640777e-05, "loss": 0.2281, "step": 15559 }, { "epoch": 150.87951807228916, "grad_norm": 9.679850578308105, "learning_rate": 2.446601941747573e-05, "loss": 0.1612, "step": 15560 }, { "epoch": 150.88915662650604, "grad_norm": 10.130315780639648, "learning_rate": 2.4461165048543692e-05, "loss": 0.2995, "step": 15561 }, { "epoch": 150.89879518072289, "grad_norm": 7.632774353027344, "learning_rate": 2.4456310679611653e-05, "loss": 0.2037, "step": 15562 }, { "epoch": 150.90843373493976, "grad_norm": 6.423584938049316, "learning_rate": 2.4451456310679614e-05, "loss": 0.274, "step": 15563 }, { "epoch": 150.91807228915664, "grad_norm": 87.9754409790039, "learning_rate": 2.4446601941747575e-05, "loss": 0.3161, "step": 15564 }, { "epoch": 150.9277108433735, "grad_norm": 7.198225498199463, "learning_rate": 2.4441747572815536e-05, "loss": 0.1305, "step": 15565 }, { "epoch": 150.93734939759037, "grad_norm": 19.899452209472656, "learning_rate": 2.4436893203883494e-05, "loss": 0.2639, "step": 15566 }, { "epoch": 150.94698795180722, "grad_norm": 4.612907409667969, "learning_rate": 2.4432038834951455e-05, "loss": 0.1095, "step": 15567 }, { "epoch": 150.9566265060241, "grad_norm": 9.892801284790039, "learning_rate": 2.4427184466019416e-05, "loss": 0.2837, "step": 15568 }, { "epoch": 150.96626506024097, "grad_norm": 15.449461936950684, "learning_rate": 2.4422330097087378e-05, "loss": 0.2381, "step": 15569 }, { "epoch": 150.97590361445782, "grad_norm": 43.435997009277344, "learning_rate": 2.4417475728155342e-05, "loss": 0.189, "step": 15570 }, { "epoch": 150.9855421686747, "grad_norm": 14.503945350646973, "learning_rate": 2.4412621359223303e-05, "loss": 0.3642, "step": 15571 }, { "epoch": 151.00120481927712, "grad_norm": 12.429895401000977, "learning_rate": 2.4407766990291264e-05, "loss": 0.4243, "step": 15572 }, { "epoch": 151.01084337349397, "grad_norm": 9.78697395324707, "learning_rate": 2.4402912621359226e-05, "loss": 0.205, "step": 15573 }, { "epoch": 151.02048192771085, "grad_norm": 4.886343479156494, "learning_rate": 2.4398058252427187e-05, "loss": 0.1097, "step": 15574 }, { "epoch": 151.03012048192772, "grad_norm": 30.102556228637695, "learning_rate": 2.4393203883495148e-05, "loss": 0.4034, "step": 15575 }, { "epoch": 151.03975903614457, "grad_norm": 11.198774337768555, "learning_rate": 2.438834951456311e-05, "loss": 0.2144, "step": 15576 }, { "epoch": 151.04939759036145, "grad_norm": 6.49247932434082, "learning_rate": 2.438349514563107e-05, "loss": 0.271, "step": 15577 }, { "epoch": 151.0590361445783, "grad_norm": 10.925028800964355, "learning_rate": 2.437864077669903e-05, "loss": 0.4718, "step": 15578 }, { "epoch": 151.06867469879518, "grad_norm": 11.382109642028809, "learning_rate": 2.4373786407766992e-05, "loss": 0.5122, "step": 15579 }, { "epoch": 151.07831325301206, "grad_norm": 10.196621894836426, "learning_rate": 2.436893203883495e-05, "loss": 0.2369, "step": 15580 }, { "epoch": 151.0879518072289, "grad_norm": 3.775761604309082, "learning_rate": 2.436407766990291e-05, "loss": 0.1436, "step": 15581 }, { "epoch": 151.09759036144578, "grad_norm": 9.66108512878418, "learning_rate": 2.4359223300970872e-05, "loss": 0.1615, "step": 15582 }, { "epoch": 151.10722891566266, "grad_norm": 18.363710403442383, "learning_rate": 2.4354368932038837e-05, "loss": 0.4013, "step": 15583 }, { "epoch": 151.1168674698795, "grad_norm": 15.533038139343262, "learning_rate": 2.4349514563106798e-05, "loss": 0.4021, "step": 15584 }, { "epoch": 151.1265060240964, "grad_norm": 12.135465621948242, "learning_rate": 2.434466019417476e-05, "loss": 0.3058, "step": 15585 }, { "epoch": 151.13614457831326, "grad_norm": 11.915771484375, "learning_rate": 2.433980582524272e-05, "loss": 0.1711, "step": 15586 }, { "epoch": 151.1457831325301, "grad_norm": 13.73631477355957, "learning_rate": 2.433495145631068e-05, "loss": 0.223, "step": 15587 }, { "epoch": 151.155421686747, "grad_norm": 19.541873931884766, "learning_rate": 2.4330097087378643e-05, "loss": 0.4531, "step": 15588 }, { "epoch": 151.16506024096387, "grad_norm": 21.110633850097656, "learning_rate": 2.4325242718446604e-05, "loss": 0.2726, "step": 15589 }, { "epoch": 151.17469879518072, "grad_norm": 14.462854385375977, "learning_rate": 2.4320388349514565e-05, "loss": 0.3071, "step": 15590 }, { "epoch": 151.1843373493976, "grad_norm": 7.755674362182617, "learning_rate": 2.4315533980582526e-05, "loss": 0.1237, "step": 15591 }, { "epoch": 151.19397590361444, "grad_norm": 25.08656120300293, "learning_rate": 2.4310679611650487e-05, "loss": 0.3491, "step": 15592 }, { "epoch": 151.20361445783132, "grad_norm": 7.158430099487305, "learning_rate": 2.430582524271845e-05, "loss": 0.3053, "step": 15593 }, { "epoch": 151.2132530120482, "grad_norm": 16.550188064575195, "learning_rate": 2.4300970873786406e-05, "loss": 0.3687, "step": 15594 }, { "epoch": 151.22289156626505, "grad_norm": 8.88772201538086, "learning_rate": 2.4296116504854367e-05, "loss": 0.2722, "step": 15595 }, { "epoch": 151.23253012048193, "grad_norm": 17.469608306884766, "learning_rate": 2.4291262135922332e-05, "loss": 0.2691, "step": 15596 }, { "epoch": 151.2421686746988, "grad_norm": 9.576460838317871, "learning_rate": 2.4286407766990293e-05, "loss": 0.2551, "step": 15597 }, { "epoch": 151.25180722891565, "grad_norm": 18.06476593017578, "learning_rate": 2.4281553398058254e-05, "loss": 0.3752, "step": 15598 }, { "epoch": 151.26144578313253, "grad_norm": 16.326135635375977, "learning_rate": 2.4276699029126215e-05, "loss": 0.1332, "step": 15599 }, { "epoch": 151.2710843373494, "grad_norm": 13.209221839904785, "learning_rate": 2.4271844660194176e-05, "loss": 0.204, "step": 15600 }, { "epoch": 151.28072289156626, "grad_norm": 13.08402156829834, "learning_rate": 2.4266990291262138e-05, "loss": 0.2305, "step": 15601 }, { "epoch": 151.29036144578313, "grad_norm": 13.685346603393555, "learning_rate": 2.42621359223301e-05, "loss": 0.2693, "step": 15602 }, { "epoch": 151.3, "grad_norm": 10.012948036193848, "learning_rate": 2.425728155339806e-05, "loss": 0.1491, "step": 15603 }, { "epoch": 151.30963855421686, "grad_norm": 5.612847328186035, "learning_rate": 2.425242718446602e-05, "loss": 0.0833, "step": 15604 }, { "epoch": 151.31927710843374, "grad_norm": 9.892938613891602, "learning_rate": 2.4247572815533982e-05, "loss": 0.1826, "step": 15605 }, { "epoch": 151.32891566265062, "grad_norm": 7.56825590133667, "learning_rate": 2.4242718446601943e-05, "loss": 0.1363, "step": 15606 }, { "epoch": 151.33855421686746, "grad_norm": 12.026525497436523, "learning_rate": 2.4237864077669904e-05, "loss": 0.2658, "step": 15607 }, { "epoch": 151.34819277108434, "grad_norm": 16.023576736450195, "learning_rate": 2.4233009708737866e-05, "loss": 0.1914, "step": 15608 }, { "epoch": 151.3578313253012, "grad_norm": 6.1681084632873535, "learning_rate": 2.4228155339805827e-05, "loss": 0.2048, "step": 15609 }, { "epoch": 151.36746987951807, "grad_norm": 17.834026336669922, "learning_rate": 2.4223300970873788e-05, "loss": 0.2328, "step": 15610 }, { "epoch": 151.37710843373495, "grad_norm": 15.218522071838379, "learning_rate": 2.421844660194175e-05, "loss": 0.2532, "step": 15611 }, { "epoch": 151.3867469879518, "grad_norm": 8.473451614379883, "learning_rate": 2.421359223300971e-05, "loss": 0.1924, "step": 15612 }, { "epoch": 151.39638554216867, "grad_norm": 9.474018096923828, "learning_rate": 2.420873786407767e-05, "loss": 0.1432, "step": 15613 }, { "epoch": 151.40602409638555, "grad_norm": 7.883155345916748, "learning_rate": 2.4203883495145632e-05, "loss": 0.1678, "step": 15614 }, { "epoch": 151.4156626506024, "grad_norm": 8.94921588897705, "learning_rate": 2.4199029126213594e-05, "loss": 0.1378, "step": 15615 }, { "epoch": 151.42530120481928, "grad_norm": 11.414704322814941, "learning_rate": 2.4194174757281555e-05, "loss": 0.2383, "step": 15616 }, { "epoch": 151.43493975903615, "grad_norm": 6.516156196594238, "learning_rate": 2.4189320388349516e-05, "loss": 0.2398, "step": 15617 }, { "epoch": 151.444578313253, "grad_norm": 14.354697227478027, "learning_rate": 2.4184466019417477e-05, "loss": 0.2769, "step": 15618 }, { "epoch": 151.45421686746988, "grad_norm": 13.000885009765625, "learning_rate": 2.4179611650485438e-05, "loss": 0.1586, "step": 15619 }, { "epoch": 151.46385542168676, "grad_norm": 20.35477638244629, "learning_rate": 2.41747572815534e-05, "loss": 0.3691, "step": 15620 }, { "epoch": 151.4734939759036, "grad_norm": 9.759140014648438, "learning_rate": 2.416990291262136e-05, "loss": 0.1513, "step": 15621 }, { "epoch": 151.48313253012049, "grad_norm": 26.883024215698242, "learning_rate": 2.416504854368932e-05, "loss": 0.2567, "step": 15622 }, { "epoch": 151.49277108433736, "grad_norm": 16.703853607177734, "learning_rate": 2.4160194174757283e-05, "loss": 0.353, "step": 15623 }, { "epoch": 151.5024096385542, "grad_norm": 9.39669132232666, "learning_rate": 2.4155339805825244e-05, "loss": 0.1676, "step": 15624 }, { "epoch": 151.5120481927711, "grad_norm": 3.29106068611145, "learning_rate": 2.4150485436893205e-05, "loss": 0.0816, "step": 15625 }, { "epoch": 151.52168674698794, "grad_norm": 14.968116760253906, "learning_rate": 2.4145631067961166e-05, "loss": 0.1242, "step": 15626 }, { "epoch": 151.53132530120482, "grad_norm": 21.257966995239258, "learning_rate": 2.4140776699029127e-05, "loss": 0.2731, "step": 15627 }, { "epoch": 151.5409638554217, "grad_norm": 17.614410400390625, "learning_rate": 2.413592233009709e-05, "loss": 0.144, "step": 15628 }, { "epoch": 151.55060240963854, "grad_norm": 5.407657623291016, "learning_rate": 2.413106796116505e-05, "loss": 0.2021, "step": 15629 }, { "epoch": 151.56024096385542, "grad_norm": 10.100082397460938, "learning_rate": 2.412621359223301e-05, "loss": 0.2051, "step": 15630 }, { "epoch": 151.5698795180723, "grad_norm": 18.574262619018555, "learning_rate": 2.4121359223300972e-05, "loss": 0.2318, "step": 15631 }, { "epoch": 151.57951807228915, "grad_norm": 8.177897453308105, "learning_rate": 2.4116504854368933e-05, "loss": 0.1139, "step": 15632 }, { "epoch": 151.58915662650602, "grad_norm": 4.692479133605957, "learning_rate": 2.4111650485436894e-05, "loss": 0.0984, "step": 15633 }, { "epoch": 151.5987951807229, "grad_norm": 9.643921852111816, "learning_rate": 2.4106796116504855e-05, "loss": 0.2984, "step": 15634 }, { "epoch": 151.60843373493975, "grad_norm": 20.185270309448242, "learning_rate": 2.4101941747572817e-05, "loss": 0.5207, "step": 15635 }, { "epoch": 151.61807228915663, "grad_norm": 7.8916192054748535, "learning_rate": 2.4097087378640778e-05, "loss": 0.1685, "step": 15636 }, { "epoch": 151.6277108433735, "grad_norm": 5.462875843048096, "learning_rate": 2.409223300970874e-05, "loss": 0.1255, "step": 15637 }, { "epoch": 151.63734939759036, "grad_norm": 4.675834655761719, "learning_rate": 2.40873786407767e-05, "loss": 0.2315, "step": 15638 }, { "epoch": 151.64698795180723, "grad_norm": 13.475359916687012, "learning_rate": 2.408252427184466e-05, "loss": 0.1539, "step": 15639 }, { "epoch": 151.65662650602408, "grad_norm": 22.859342575073242, "learning_rate": 2.4077669902912622e-05, "loss": 0.1813, "step": 15640 }, { "epoch": 151.66626506024096, "grad_norm": 14.937881469726562, "learning_rate": 2.4072815533980583e-05, "loss": 0.495, "step": 15641 }, { "epoch": 151.67590361445784, "grad_norm": 7.2305145263671875, "learning_rate": 2.4067961165048545e-05, "loss": 0.3274, "step": 15642 }, { "epoch": 151.6855421686747, "grad_norm": 9.099285125732422, "learning_rate": 2.4063106796116506e-05, "loss": 0.3257, "step": 15643 }, { "epoch": 151.69518072289156, "grad_norm": 6.593160629272461, "learning_rate": 2.4058252427184467e-05, "loss": 0.2807, "step": 15644 }, { "epoch": 151.70481927710844, "grad_norm": 11.479762077331543, "learning_rate": 2.4053398058252428e-05, "loss": 0.2662, "step": 15645 }, { "epoch": 151.7144578313253, "grad_norm": 7.614596843719482, "learning_rate": 2.404854368932039e-05, "loss": 0.3196, "step": 15646 }, { "epoch": 151.72409638554217, "grad_norm": 14.351439476013184, "learning_rate": 2.404368932038835e-05, "loss": 0.3176, "step": 15647 }, { "epoch": 151.73373493975905, "grad_norm": 12.813193321228027, "learning_rate": 2.403883495145631e-05, "loss": 0.3571, "step": 15648 }, { "epoch": 151.7433734939759, "grad_norm": 8.281915664672852, "learning_rate": 2.4033980582524273e-05, "loss": 0.2664, "step": 15649 }, { "epoch": 151.75301204819277, "grad_norm": 18.552879333496094, "learning_rate": 2.4029126213592234e-05, "loss": 0.234, "step": 15650 }, { "epoch": 151.76265060240965, "grad_norm": 21.43924903869629, "learning_rate": 2.4024271844660195e-05, "loss": 0.3243, "step": 15651 }, { "epoch": 151.7722891566265, "grad_norm": 8.77122688293457, "learning_rate": 2.4019417475728156e-05, "loss": 0.1698, "step": 15652 }, { "epoch": 151.78192771084338, "grad_norm": 8.700735092163086, "learning_rate": 2.4014563106796117e-05, "loss": 0.2831, "step": 15653 }, { "epoch": 151.79156626506025, "grad_norm": 4.921889305114746, "learning_rate": 2.400970873786408e-05, "loss": 0.1858, "step": 15654 }, { "epoch": 151.8012048192771, "grad_norm": 6.1760945320129395, "learning_rate": 2.400485436893204e-05, "loss": 0.4153, "step": 15655 }, { "epoch": 151.81084337349398, "grad_norm": 16.546369552612305, "learning_rate": 2.4e-05, "loss": 0.2585, "step": 15656 }, { "epoch": 151.82048192771083, "grad_norm": 36.695255279541016, "learning_rate": 2.3995145631067962e-05, "loss": 0.1521, "step": 15657 }, { "epoch": 151.8301204819277, "grad_norm": 19.311729431152344, "learning_rate": 2.3990291262135923e-05, "loss": 0.2609, "step": 15658 }, { "epoch": 151.83975903614459, "grad_norm": 20.209915161132812, "learning_rate": 2.3985436893203887e-05, "loss": 0.3114, "step": 15659 }, { "epoch": 151.84939759036143, "grad_norm": 22.470684051513672, "learning_rate": 2.3980582524271845e-05, "loss": 0.1476, "step": 15660 }, { "epoch": 151.8590361445783, "grad_norm": 10.6867036819458, "learning_rate": 2.3975728155339806e-05, "loss": 0.273, "step": 15661 }, { "epoch": 151.8686746987952, "grad_norm": 10.707245826721191, "learning_rate": 2.3970873786407767e-05, "loss": 0.323, "step": 15662 }, { "epoch": 151.87831325301204, "grad_norm": 37.08637619018555, "learning_rate": 2.396601941747573e-05, "loss": 0.1857, "step": 15663 }, { "epoch": 151.88795180722892, "grad_norm": 9.186062812805176, "learning_rate": 2.396116504854369e-05, "loss": 0.228, "step": 15664 }, { "epoch": 151.8975903614458, "grad_norm": 7.742168426513672, "learning_rate": 2.395631067961165e-05, "loss": 0.1162, "step": 15665 }, { "epoch": 151.90722891566264, "grad_norm": 5.675647258758545, "learning_rate": 2.3951456310679612e-05, "loss": 0.2322, "step": 15666 }, { "epoch": 151.91686746987952, "grad_norm": 5.267970561981201, "learning_rate": 2.3946601941747573e-05, "loss": 0.1135, "step": 15667 }, { "epoch": 151.9265060240964, "grad_norm": 15.561640739440918, "learning_rate": 2.3941747572815534e-05, "loss": 0.2238, "step": 15668 }, { "epoch": 151.93614457831325, "grad_norm": 18.239608764648438, "learning_rate": 2.3936893203883496e-05, "loss": 0.2234, "step": 15669 }, { "epoch": 151.94578313253012, "grad_norm": 9.923569679260254, "learning_rate": 2.3932038834951457e-05, "loss": 0.2962, "step": 15670 }, { "epoch": 151.955421686747, "grad_norm": 9.150599479675293, "learning_rate": 2.392718446601942e-05, "loss": 0.2161, "step": 15671 }, { "epoch": 151.96506024096385, "grad_norm": 18.3769474029541, "learning_rate": 2.3922330097087382e-05, "loss": 0.4517, "step": 15672 }, { "epoch": 151.97469879518073, "grad_norm": 10.42453670501709, "learning_rate": 2.3917475728155343e-05, "loss": 0.2732, "step": 15673 }, { "epoch": 151.98433734939758, "grad_norm": 8.19084644317627, "learning_rate": 2.39126213592233e-05, "loss": 0.2463, "step": 15674 }, { "epoch": 151.99397590361446, "grad_norm": 9.952309608459473, "learning_rate": 2.3907766990291262e-05, "loss": 0.1073, "step": 15675 }, { "epoch": 152.00963855421688, "grad_norm": 13.003567695617676, "learning_rate": 2.3902912621359224e-05, "loss": 0.3804, "step": 15676 }, { "epoch": 152.01927710843373, "grad_norm": 15.54121208190918, "learning_rate": 2.3898058252427185e-05, "loss": 0.352, "step": 15677 }, { "epoch": 152.0289156626506, "grad_norm": 9.397550582885742, "learning_rate": 2.3893203883495146e-05, "loss": 0.2483, "step": 15678 }, { "epoch": 152.03855421686748, "grad_norm": 8.892698287963867, "learning_rate": 2.3888349514563107e-05, "loss": 0.1001, "step": 15679 }, { "epoch": 152.04819277108433, "grad_norm": 8.471406936645508, "learning_rate": 2.3883495145631068e-05, "loss": 0.2486, "step": 15680 }, { "epoch": 152.0578313253012, "grad_norm": 18.90174102783203, "learning_rate": 2.387864077669903e-05, "loss": 0.3408, "step": 15681 }, { "epoch": 152.0674698795181, "grad_norm": 11.302727699279785, "learning_rate": 2.387378640776699e-05, "loss": 0.2019, "step": 15682 }, { "epoch": 152.07710843373494, "grad_norm": 12.60985279083252, "learning_rate": 2.386893203883495e-05, "loss": 0.1896, "step": 15683 }, { "epoch": 152.0867469879518, "grad_norm": 6.636071681976318, "learning_rate": 2.3864077669902916e-05, "loss": 0.2577, "step": 15684 }, { "epoch": 152.09638554216866, "grad_norm": 10.787261962890625, "learning_rate": 2.3859223300970877e-05, "loss": 0.316, "step": 15685 }, { "epoch": 152.10602409638554, "grad_norm": 7.886296272277832, "learning_rate": 2.385436893203884e-05, "loss": 0.259, "step": 15686 }, { "epoch": 152.11566265060242, "grad_norm": 8.874512672424316, "learning_rate": 2.38495145631068e-05, "loss": 0.1594, "step": 15687 }, { "epoch": 152.12530120481927, "grad_norm": 23.323707580566406, "learning_rate": 2.3844660194174757e-05, "loss": 0.2717, "step": 15688 }, { "epoch": 152.13493975903614, "grad_norm": 10.659453392028809, "learning_rate": 2.383980582524272e-05, "loss": 0.2544, "step": 15689 }, { "epoch": 152.14457831325302, "grad_norm": 9.800143241882324, "learning_rate": 2.383495145631068e-05, "loss": 0.3422, "step": 15690 }, { "epoch": 152.15421686746987, "grad_norm": 11.83284854888916, "learning_rate": 2.383009708737864e-05, "loss": 0.1053, "step": 15691 }, { "epoch": 152.16385542168675, "grad_norm": 6.154016971588135, "learning_rate": 2.3825242718446602e-05, "loss": 0.1452, "step": 15692 }, { "epoch": 152.17349397590363, "grad_norm": 7.514791488647461, "learning_rate": 2.3820388349514563e-05, "loss": 0.1892, "step": 15693 }, { "epoch": 152.18313253012047, "grad_norm": 23.45330238342285, "learning_rate": 2.3815533980582524e-05, "loss": 0.2897, "step": 15694 }, { "epoch": 152.19277108433735, "grad_norm": 14.127752304077148, "learning_rate": 2.3810679611650485e-05, "loss": 0.3655, "step": 15695 }, { "epoch": 152.20240963855423, "grad_norm": 15.03244400024414, "learning_rate": 2.3805825242718446e-05, "loss": 0.1861, "step": 15696 }, { "epoch": 152.21204819277108, "grad_norm": 5.5774760246276855, "learning_rate": 2.380097087378641e-05, "loss": 0.1217, "step": 15697 }, { "epoch": 152.22168674698796, "grad_norm": 10.474286079406738, "learning_rate": 2.3796116504854372e-05, "loss": 0.5078, "step": 15698 }, { "epoch": 152.2313253012048, "grad_norm": 34.545841217041016, "learning_rate": 2.3791262135922333e-05, "loss": 0.2849, "step": 15699 }, { "epoch": 152.24096385542168, "grad_norm": 18.600311279296875, "learning_rate": 2.3786407766990294e-05, "loss": 0.2041, "step": 15700 }, { "epoch": 152.25060240963856, "grad_norm": 15.29812240600586, "learning_rate": 2.3781553398058256e-05, "loss": 0.2558, "step": 15701 }, { "epoch": 152.2602409638554, "grad_norm": 18.83392333984375, "learning_rate": 2.3776699029126213e-05, "loss": 0.3315, "step": 15702 }, { "epoch": 152.2698795180723, "grad_norm": 12.996827125549316, "learning_rate": 2.3771844660194174e-05, "loss": 0.1705, "step": 15703 }, { "epoch": 152.27951807228916, "grad_norm": 16.390974044799805, "learning_rate": 2.3766990291262136e-05, "loss": 0.5542, "step": 15704 }, { "epoch": 152.289156626506, "grad_norm": 22.97637176513672, "learning_rate": 2.3762135922330097e-05, "loss": 0.5371, "step": 15705 }, { "epoch": 152.2987951807229, "grad_norm": 7.200843811035156, "learning_rate": 2.3757281553398058e-05, "loss": 0.1485, "step": 15706 }, { "epoch": 152.30843373493977, "grad_norm": 12.923524856567383, "learning_rate": 2.375242718446602e-05, "loss": 0.2801, "step": 15707 }, { "epoch": 152.31807228915662, "grad_norm": 10.76358413696289, "learning_rate": 2.374757281553398e-05, "loss": 0.123, "step": 15708 }, { "epoch": 152.3277108433735, "grad_norm": 9.014708518981934, "learning_rate": 2.3742718446601945e-05, "loss": 0.2332, "step": 15709 }, { "epoch": 152.33734939759037, "grad_norm": 10.557053565979004, "learning_rate": 2.3737864077669906e-05, "loss": 0.3426, "step": 15710 }, { "epoch": 152.34698795180722, "grad_norm": 5.574372291564941, "learning_rate": 2.3733009708737867e-05, "loss": 0.23, "step": 15711 }, { "epoch": 152.3566265060241, "grad_norm": 7.263434410095215, "learning_rate": 2.3728155339805828e-05, "loss": 0.1701, "step": 15712 }, { "epoch": 152.36626506024098, "grad_norm": 17.449317932128906, "learning_rate": 2.372330097087379e-05, "loss": 0.1749, "step": 15713 }, { "epoch": 152.37590361445783, "grad_norm": 13.3939208984375, "learning_rate": 2.371844660194175e-05, "loss": 0.2887, "step": 15714 }, { "epoch": 152.3855421686747, "grad_norm": 12.13333797454834, "learning_rate": 2.3713592233009708e-05, "loss": 0.2378, "step": 15715 }, { "epoch": 152.39518072289155, "grad_norm": 19.04747200012207, "learning_rate": 2.370873786407767e-05, "loss": 0.3203, "step": 15716 }, { "epoch": 152.40481927710843, "grad_norm": 16.10572052001953, "learning_rate": 2.370388349514563e-05, "loss": 0.2377, "step": 15717 }, { "epoch": 152.4144578313253, "grad_norm": 5.5509796142578125, "learning_rate": 2.369902912621359e-05, "loss": 0.3797, "step": 15718 }, { "epoch": 152.42409638554216, "grad_norm": 7.53316068649292, "learning_rate": 2.3694174757281553e-05, "loss": 0.1757, "step": 15719 }, { "epoch": 152.43373493975903, "grad_norm": 6.140067100524902, "learning_rate": 2.3689320388349514e-05, "loss": 0.2431, "step": 15720 }, { "epoch": 152.4433734939759, "grad_norm": 14.0064697265625, "learning_rate": 2.3684466019417475e-05, "loss": 0.2391, "step": 15721 }, { "epoch": 152.45301204819276, "grad_norm": 6.238461971282959, "learning_rate": 2.367961165048544e-05, "loss": 0.3117, "step": 15722 }, { "epoch": 152.46265060240964, "grad_norm": 54.72893142700195, "learning_rate": 2.36747572815534e-05, "loss": 0.6373, "step": 15723 }, { "epoch": 152.47228915662652, "grad_norm": 9.474616050720215, "learning_rate": 2.3669902912621362e-05, "loss": 0.3058, "step": 15724 }, { "epoch": 152.48192771084337, "grad_norm": 19.78740119934082, "learning_rate": 2.3665048543689323e-05, "loss": 0.2595, "step": 15725 }, { "epoch": 152.49156626506024, "grad_norm": 20.775056838989258, "learning_rate": 2.3660194174757284e-05, "loss": 0.1975, "step": 15726 }, { "epoch": 152.50120481927712, "grad_norm": 10.116788864135742, "learning_rate": 2.3655339805825245e-05, "loss": 0.2465, "step": 15727 }, { "epoch": 152.51084337349397, "grad_norm": 8.232768058776855, "learning_rate": 2.3650485436893206e-05, "loss": 0.1674, "step": 15728 }, { "epoch": 152.52048192771085, "grad_norm": 5.813659191131592, "learning_rate": 2.3645631067961164e-05, "loss": 0.2043, "step": 15729 }, { "epoch": 152.53012048192772, "grad_norm": 22.153114318847656, "learning_rate": 2.3640776699029125e-05, "loss": 0.4178, "step": 15730 }, { "epoch": 152.53975903614457, "grad_norm": 6.855096817016602, "learning_rate": 2.3635922330097087e-05, "loss": 0.2495, "step": 15731 }, { "epoch": 152.54939759036145, "grad_norm": 9.885231018066406, "learning_rate": 2.3631067961165048e-05, "loss": 0.1784, "step": 15732 }, { "epoch": 152.5590361445783, "grad_norm": 4.814488887786865, "learning_rate": 2.362621359223301e-05, "loss": 0.1125, "step": 15733 }, { "epoch": 152.56867469879518, "grad_norm": 13.624587059020996, "learning_rate": 2.362135922330097e-05, "loss": 0.137, "step": 15734 }, { "epoch": 152.57831325301206, "grad_norm": 14.695757865905762, "learning_rate": 2.3616504854368935e-05, "loss": 0.3662, "step": 15735 }, { "epoch": 152.5879518072289, "grad_norm": 14.48632526397705, "learning_rate": 2.3611650485436896e-05, "loss": 0.339, "step": 15736 }, { "epoch": 152.59759036144578, "grad_norm": 6.478667259216309, "learning_rate": 2.3606796116504857e-05, "loss": 0.2749, "step": 15737 }, { "epoch": 152.60722891566266, "grad_norm": 6.294833183288574, "learning_rate": 2.3601941747572818e-05, "loss": 0.1913, "step": 15738 }, { "epoch": 152.6168674698795, "grad_norm": 4.514859676361084, "learning_rate": 2.359708737864078e-05, "loss": 0.1317, "step": 15739 }, { "epoch": 152.6265060240964, "grad_norm": 4.2168731689453125, "learning_rate": 2.359223300970874e-05, "loss": 0.1372, "step": 15740 }, { "epoch": 152.63614457831326, "grad_norm": 12.348633766174316, "learning_rate": 2.35873786407767e-05, "loss": 0.278, "step": 15741 }, { "epoch": 152.6457831325301, "grad_norm": 13.770306587219238, "learning_rate": 2.3582524271844663e-05, "loss": 0.3377, "step": 15742 }, { "epoch": 152.655421686747, "grad_norm": 19.635940551757812, "learning_rate": 2.357766990291262e-05, "loss": 0.1871, "step": 15743 }, { "epoch": 152.66506024096387, "grad_norm": 21.194169998168945, "learning_rate": 2.357281553398058e-05, "loss": 0.3297, "step": 15744 }, { "epoch": 152.67469879518072, "grad_norm": 9.703231811523438, "learning_rate": 2.3567961165048543e-05, "loss": 0.2269, "step": 15745 }, { "epoch": 152.6843373493976, "grad_norm": 14.375800132751465, "learning_rate": 2.3563106796116504e-05, "loss": 0.473, "step": 15746 }, { "epoch": 152.69397590361444, "grad_norm": 16.711790084838867, "learning_rate": 2.3558252427184468e-05, "loss": 0.4934, "step": 15747 }, { "epoch": 152.70361445783132, "grad_norm": 14.626588821411133, "learning_rate": 2.355339805825243e-05, "loss": 0.2316, "step": 15748 }, { "epoch": 152.7132530120482, "grad_norm": 12.762533187866211, "learning_rate": 2.354854368932039e-05, "loss": 0.1028, "step": 15749 }, { "epoch": 152.72289156626505, "grad_norm": 18.240022659301758, "learning_rate": 2.3543689320388352e-05, "loss": 0.4311, "step": 15750 }, { "epoch": 152.73253012048193, "grad_norm": 18.98862075805664, "learning_rate": 2.3538834951456313e-05, "loss": 0.2697, "step": 15751 }, { "epoch": 152.7421686746988, "grad_norm": 12.580343246459961, "learning_rate": 2.3533980582524274e-05, "loss": 0.2306, "step": 15752 }, { "epoch": 152.75180722891565, "grad_norm": 16.284337997436523, "learning_rate": 2.3529126213592235e-05, "loss": 0.3109, "step": 15753 }, { "epoch": 152.76144578313253, "grad_norm": 8.37214469909668, "learning_rate": 2.3524271844660196e-05, "loss": 0.307, "step": 15754 }, { "epoch": 152.7710843373494, "grad_norm": 13.568672180175781, "learning_rate": 2.3519417475728157e-05, "loss": 0.3219, "step": 15755 }, { "epoch": 152.78072289156626, "grad_norm": 9.736194610595703, "learning_rate": 2.351456310679612e-05, "loss": 0.3313, "step": 15756 }, { "epoch": 152.79036144578313, "grad_norm": 6.562644004821777, "learning_rate": 2.3509708737864076e-05, "loss": 0.2373, "step": 15757 }, { "epoch": 152.8, "grad_norm": 11.600902557373047, "learning_rate": 2.3504854368932037e-05, "loss": 0.232, "step": 15758 }, { "epoch": 152.80963855421686, "grad_norm": 9.449248313903809, "learning_rate": 2.35e-05, "loss": 0.1314, "step": 15759 }, { "epoch": 152.81927710843374, "grad_norm": 15.326133728027344, "learning_rate": 2.3495145631067963e-05, "loss": 0.2464, "step": 15760 }, { "epoch": 152.82891566265062, "grad_norm": 22.97336196899414, "learning_rate": 2.3490291262135924e-05, "loss": 0.5099, "step": 15761 }, { "epoch": 152.83855421686746, "grad_norm": 13.88300895690918, "learning_rate": 2.3485436893203885e-05, "loss": 0.2759, "step": 15762 }, { "epoch": 152.84819277108434, "grad_norm": 28.629497528076172, "learning_rate": 2.3480582524271847e-05, "loss": 0.1143, "step": 15763 }, { "epoch": 152.8578313253012, "grad_norm": 8.42236042022705, "learning_rate": 2.3475728155339808e-05, "loss": 0.2043, "step": 15764 }, { "epoch": 152.86746987951807, "grad_norm": 6.090963363647461, "learning_rate": 2.347087378640777e-05, "loss": 0.1877, "step": 15765 }, { "epoch": 152.87710843373495, "grad_norm": 13.69938850402832, "learning_rate": 2.346601941747573e-05, "loss": 0.4331, "step": 15766 }, { "epoch": 152.8867469879518, "grad_norm": 8.226057052612305, "learning_rate": 2.346116504854369e-05, "loss": 0.1525, "step": 15767 }, { "epoch": 152.89638554216867, "grad_norm": 23.399816513061523, "learning_rate": 2.3456310679611652e-05, "loss": 0.3965, "step": 15768 }, { "epoch": 152.90602409638555, "grad_norm": 10.590227127075195, "learning_rate": 2.3451456310679613e-05, "loss": 0.3057, "step": 15769 }, { "epoch": 152.9156626506024, "grad_norm": 9.05672550201416, "learning_rate": 2.3446601941747575e-05, "loss": 0.3146, "step": 15770 }, { "epoch": 152.92530120481928, "grad_norm": 10.8690824508667, "learning_rate": 2.3441747572815532e-05, "loss": 0.2816, "step": 15771 }, { "epoch": 152.93493975903615, "grad_norm": 23.38714027404785, "learning_rate": 2.3436893203883497e-05, "loss": 0.2243, "step": 15772 }, { "epoch": 152.944578313253, "grad_norm": 12.755461692810059, "learning_rate": 2.3432038834951458e-05, "loss": 0.2316, "step": 15773 }, { "epoch": 152.95421686746988, "grad_norm": 14.916550636291504, "learning_rate": 2.342718446601942e-05, "loss": 0.1656, "step": 15774 }, { "epoch": 152.96385542168676, "grad_norm": 9.050002098083496, "learning_rate": 2.342233009708738e-05, "loss": 0.1392, "step": 15775 }, { "epoch": 152.9734939759036, "grad_norm": 24.004823684692383, "learning_rate": 2.341747572815534e-05, "loss": 0.3102, "step": 15776 }, { "epoch": 152.98313253012049, "grad_norm": 2.020477056503296, "learning_rate": 2.3412621359223303e-05, "loss": 0.0595, "step": 15777 }, { "epoch": 152.99277108433736, "grad_norm": 7.1348958015441895, "learning_rate": 2.3407766990291264e-05, "loss": 0.1334, "step": 15778 }, { "epoch": 153.00843373493976, "grad_norm": 6.809925556182861, "learning_rate": 2.3402912621359225e-05, "loss": 0.2988, "step": 15779 }, { "epoch": 153.01807228915663, "grad_norm": 3.284914970397949, "learning_rate": 2.3398058252427186e-05, "loss": 0.2776, "step": 15780 }, { "epoch": 153.02771084337348, "grad_norm": 21.14887237548828, "learning_rate": 2.3393203883495147e-05, "loss": 0.2737, "step": 15781 }, { "epoch": 153.03734939759036, "grad_norm": 19.915136337280273, "learning_rate": 2.338834951456311e-05, "loss": 0.3454, "step": 15782 }, { "epoch": 153.04698795180724, "grad_norm": 28.26780128479004, "learning_rate": 2.338349514563107e-05, "loss": 0.3436, "step": 15783 }, { "epoch": 153.0566265060241, "grad_norm": 17.786487579345703, "learning_rate": 2.337864077669903e-05, "loss": 0.0868, "step": 15784 }, { "epoch": 153.06626506024097, "grad_norm": 32.83147430419922, "learning_rate": 2.3373786407766992e-05, "loss": 0.2808, "step": 15785 }, { "epoch": 153.07590361445784, "grad_norm": 9.361542701721191, "learning_rate": 2.3368932038834953e-05, "loss": 0.3, "step": 15786 }, { "epoch": 153.0855421686747, "grad_norm": 23.914976119995117, "learning_rate": 2.3364077669902914e-05, "loss": 0.3976, "step": 15787 }, { "epoch": 153.09518072289157, "grad_norm": 27.58321189880371, "learning_rate": 2.3359223300970875e-05, "loss": 0.4804, "step": 15788 }, { "epoch": 153.10481927710845, "grad_norm": 3.0748302936553955, "learning_rate": 2.3354368932038836e-05, "loss": 0.2432, "step": 15789 }, { "epoch": 153.1144578313253, "grad_norm": 12.728017807006836, "learning_rate": 2.3349514563106798e-05, "loss": 0.2065, "step": 15790 }, { "epoch": 153.12409638554217, "grad_norm": 70.74726104736328, "learning_rate": 2.334466019417476e-05, "loss": 0.1849, "step": 15791 }, { "epoch": 153.13373493975902, "grad_norm": 24.261157989501953, "learning_rate": 2.333980582524272e-05, "loss": 0.1844, "step": 15792 }, { "epoch": 153.1433734939759, "grad_norm": 16.201982498168945, "learning_rate": 2.333495145631068e-05, "loss": 0.2163, "step": 15793 }, { "epoch": 153.15301204819278, "grad_norm": 15.12890911102295, "learning_rate": 2.3330097087378642e-05, "loss": 0.1929, "step": 15794 }, { "epoch": 153.16265060240963, "grad_norm": 19.957178115844727, "learning_rate": 2.3325242718446603e-05, "loss": 0.2608, "step": 15795 }, { "epoch": 153.1722891566265, "grad_norm": 7.901336193084717, "learning_rate": 2.3320388349514564e-05, "loss": 0.3605, "step": 15796 }, { "epoch": 153.18192771084338, "grad_norm": 14.147404670715332, "learning_rate": 2.3315533980582526e-05, "loss": 0.604, "step": 15797 }, { "epoch": 153.19156626506023, "grad_norm": 11.513361930847168, "learning_rate": 2.3310679611650487e-05, "loss": 0.398, "step": 15798 }, { "epoch": 153.2012048192771, "grad_norm": 73.46852111816406, "learning_rate": 2.3305825242718448e-05, "loss": 0.3046, "step": 15799 }, { "epoch": 153.210843373494, "grad_norm": 48.67055892944336, "learning_rate": 2.330097087378641e-05, "loss": 0.2783, "step": 15800 }, { "epoch": 153.22048192771084, "grad_norm": 18.33557891845703, "learning_rate": 2.329611650485437e-05, "loss": 0.4403, "step": 15801 }, { "epoch": 153.2301204819277, "grad_norm": 33.1168212890625, "learning_rate": 2.329126213592233e-05, "loss": 0.3249, "step": 15802 }, { "epoch": 153.2397590361446, "grad_norm": 21.080097198486328, "learning_rate": 2.3286407766990292e-05, "loss": 0.334, "step": 15803 }, { "epoch": 153.24939759036144, "grad_norm": 5.348907947540283, "learning_rate": 2.3281553398058254e-05, "loss": 0.1369, "step": 15804 }, { "epoch": 153.25903614457832, "grad_norm": 71.50617980957031, "learning_rate": 2.3276699029126215e-05, "loss": 0.3327, "step": 15805 }, { "epoch": 153.2686746987952, "grad_norm": 57.5863151550293, "learning_rate": 2.3271844660194176e-05, "loss": 0.2529, "step": 15806 }, { "epoch": 153.27831325301204, "grad_norm": 79.96080017089844, "learning_rate": 2.3266990291262137e-05, "loss": 0.1782, "step": 15807 }, { "epoch": 153.28795180722892, "grad_norm": 20.632970809936523, "learning_rate": 2.3262135922330098e-05, "loss": 0.3671, "step": 15808 }, { "epoch": 153.29759036144577, "grad_norm": 35.24350357055664, "learning_rate": 2.325728155339806e-05, "loss": 0.2857, "step": 15809 }, { "epoch": 153.30722891566265, "grad_norm": 2.445557117462158, "learning_rate": 2.325242718446602e-05, "loss": 0.2493, "step": 15810 }, { "epoch": 153.31686746987953, "grad_norm": 19.441083908081055, "learning_rate": 2.324757281553398e-05, "loss": 0.2436, "step": 15811 }, { "epoch": 153.32650602409637, "grad_norm": 3.2498350143432617, "learning_rate": 2.3242718446601943e-05, "loss": 0.2078, "step": 15812 }, { "epoch": 153.33614457831325, "grad_norm": 37.61874008178711, "learning_rate": 2.3237864077669904e-05, "loss": 0.2376, "step": 15813 }, { "epoch": 153.34578313253013, "grad_norm": 89.10972595214844, "learning_rate": 2.3233009708737865e-05, "loss": 0.1818, "step": 15814 }, { "epoch": 153.35542168674698, "grad_norm": 12.334773063659668, "learning_rate": 2.3228155339805826e-05, "loss": 0.178, "step": 15815 }, { "epoch": 153.36506024096386, "grad_norm": 3.497774600982666, "learning_rate": 2.3223300970873787e-05, "loss": 0.1028, "step": 15816 }, { "epoch": 153.37469879518073, "grad_norm": 29.994083404541016, "learning_rate": 2.321844660194175e-05, "loss": 0.2028, "step": 15817 }, { "epoch": 153.38433734939758, "grad_norm": 5.753145217895508, "learning_rate": 2.321359223300971e-05, "loss": 0.3217, "step": 15818 }, { "epoch": 153.39397590361446, "grad_norm": 9.117119789123535, "learning_rate": 2.320873786407767e-05, "loss": 0.1977, "step": 15819 }, { "epoch": 153.40361445783134, "grad_norm": 18.901784896850586, "learning_rate": 2.3203883495145632e-05, "loss": 0.1436, "step": 15820 }, { "epoch": 153.4132530120482, "grad_norm": 14.546086311340332, "learning_rate": 2.3199029126213593e-05, "loss": 0.1926, "step": 15821 }, { "epoch": 153.42289156626506, "grad_norm": 1.9418275356292725, "learning_rate": 2.3194174757281554e-05, "loss": 0.1351, "step": 15822 }, { "epoch": 153.4325301204819, "grad_norm": 22.480531692504883, "learning_rate": 2.3189320388349515e-05, "loss": 0.2699, "step": 15823 }, { "epoch": 153.4421686746988, "grad_norm": 14.168465614318848, "learning_rate": 2.3184466019417476e-05, "loss": 0.2475, "step": 15824 }, { "epoch": 153.45180722891567, "grad_norm": 3.093512535095215, "learning_rate": 2.3179611650485438e-05, "loss": 0.3049, "step": 15825 }, { "epoch": 153.46144578313252, "grad_norm": 21.27958106994629, "learning_rate": 2.31747572815534e-05, "loss": 0.2512, "step": 15826 }, { "epoch": 153.4710843373494, "grad_norm": 25.948238372802734, "learning_rate": 2.316990291262136e-05, "loss": 0.2845, "step": 15827 }, { "epoch": 153.48072289156627, "grad_norm": 58.295082092285156, "learning_rate": 2.316504854368932e-05, "loss": 0.3192, "step": 15828 }, { "epoch": 153.49036144578312, "grad_norm": 6.5285797119140625, "learning_rate": 2.3160194174757282e-05, "loss": 0.1868, "step": 15829 }, { "epoch": 153.5, "grad_norm": 24.362825393676758, "learning_rate": 2.3155339805825243e-05, "loss": 0.3171, "step": 15830 }, { "epoch": 153.50963855421688, "grad_norm": 29.442636489868164, "learning_rate": 2.3150485436893205e-05, "loss": 0.3275, "step": 15831 }, { "epoch": 153.51927710843373, "grad_norm": 10.777398109436035, "learning_rate": 2.3145631067961166e-05, "loss": 0.191, "step": 15832 }, { "epoch": 153.5289156626506, "grad_norm": 2.829751968383789, "learning_rate": 2.3140776699029127e-05, "loss": 0.2733, "step": 15833 }, { "epoch": 153.53855421686748, "grad_norm": 18.583097457885742, "learning_rate": 2.3135922330097088e-05, "loss": 0.1862, "step": 15834 }, { "epoch": 153.54819277108433, "grad_norm": 5.506213665008545, "learning_rate": 2.3131067961165052e-05, "loss": 0.3222, "step": 15835 }, { "epoch": 153.5578313253012, "grad_norm": 18.478933334350586, "learning_rate": 2.3126213592233014e-05, "loss": 0.264, "step": 15836 }, { "epoch": 153.5674698795181, "grad_norm": 29.643335342407227, "learning_rate": 2.312135922330097e-05, "loss": 0.2232, "step": 15837 }, { "epoch": 153.57710843373494, "grad_norm": 19.604843139648438, "learning_rate": 2.3116504854368933e-05, "loss": 0.2617, "step": 15838 }, { "epoch": 153.5867469879518, "grad_norm": 3.382657051086426, "learning_rate": 2.3111650485436894e-05, "loss": 0.2681, "step": 15839 }, { "epoch": 153.59638554216866, "grad_norm": 6.317617893218994, "learning_rate": 2.3106796116504855e-05, "loss": 0.2964, "step": 15840 }, { "epoch": 153.60602409638554, "grad_norm": 29.288816452026367, "learning_rate": 2.3101941747572816e-05, "loss": 0.1676, "step": 15841 }, { "epoch": 153.61566265060242, "grad_norm": 11.147612571716309, "learning_rate": 2.3097087378640777e-05, "loss": 0.0887, "step": 15842 }, { "epoch": 153.62530120481927, "grad_norm": 6.831663608551025, "learning_rate": 2.3092233009708738e-05, "loss": 0.1357, "step": 15843 }, { "epoch": 153.63493975903614, "grad_norm": 3.191732168197632, "learning_rate": 2.30873786407767e-05, "loss": 0.1907, "step": 15844 }, { "epoch": 153.64457831325302, "grad_norm": 17.388620376586914, "learning_rate": 2.308252427184466e-05, "loss": 0.2896, "step": 15845 }, { "epoch": 153.65421686746987, "grad_norm": 25.793405532836914, "learning_rate": 2.307766990291262e-05, "loss": 0.1563, "step": 15846 }, { "epoch": 153.66385542168675, "grad_norm": 3.4452552795410156, "learning_rate": 2.3072815533980583e-05, "loss": 0.3571, "step": 15847 }, { "epoch": 153.67349397590363, "grad_norm": 5.79640531539917, "learning_rate": 2.3067961165048547e-05, "loss": 0.3198, "step": 15848 }, { "epoch": 153.68313253012047, "grad_norm": 2.4189646244049072, "learning_rate": 2.306310679611651e-05, "loss": 0.1931, "step": 15849 }, { "epoch": 153.69277108433735, "grad_norm": 17.18594741821289, "learning_rate": 2.305825242718447e-05, "loss": 0.3261, "step": 15850 }, { "epoch": 153.70240963855423, "grad_norm": 3.840162992477417, "learning_rate": 2.3053398058252427e-05, "loss": 0.1884, "step": 15851 }, { "epoch": 153.71204819277108, "grad_norm": 7.590452671051025, "learning_rate": 2.304854368932039e-05, "loss": 0.3982, "step": 15852 }, { "epoch": 153.72168674698796, "grad_norm": 27.352270126342773, "learning_rate": 2.304368932038835e-05, "loss": 0.3706, "step": 15853 }, { "epoch": 153.73132530120483, "grad_norm": 30.934303283691406, "learning_rate": 2.303883495145631e-05, "loss": 0.3966, "step": 15854 }, { "epoch": 153.74096385542168, "grad_norm": 44.953819274902344, "learning_rate": 2.3033980582524272e-05, "loss": 0.4017, "step": 15855 }, { "epoch": 153.75060240963856, "grad_norm": 15.501802444458008, "learning_rate": 2.3029126213592233e-05, "loss": 0.2788, "step": 15856 }, { "epoch": 153.7602409638554, "grad_norm": 12.871708869934082, "learning_rate": 2.3024271844660194e-05, "loss": 0.247, "step": 15857 }, { "epoch": 153.7698795180723, "grad_norm": 5.958487033843994, "learning_rate": 2.3019417475728155e-05, "loss": 0.2419, "step": 15858 }, { "epoch": 153.77951807228916, "grad_norm": 21.562339782714844, "learning_rate": 2.3014563106796117e-05, "loss": 0.2933, "step": 15859 }, { "epoch": 153.789156626506, "grad_norm": 84.7486801147461, "learning_rate": 2.3009708737864078e-05, "loss": 0.3984, "step": 15860 }, { "epoch": 153.7987951807229, "grad_norm": 83.3314437866211, "learning_rate": 2.3004854368932042e-05, "loss": 0.2218, "step": 15861 }, { "epoch": 153.80843373493977, "grad_norm": 9.530939102172852, "learning_rate": 2.3000000000000003e-05, "loss": 0.1685, "step": 15862 }, { "epoch": 153.81807228915662, "grad_norm": 36.163753509521484, "learning_rate": 2.2995145631067965e-05, "loss": 0.3603, "step": 15863 }, { "epoch": 153.8277108433735, "grad_norm": 14.645005226135254, "learning_rate": 2.2990291262135922e-05, "loss": 0.2099, "step": 15864 }, { "epoch": 153.83734939759037, "grad_norm": 4.8820366859436035, "learning_rate": 2.2985436893203883e-05, "loss": 0.2842, "step": 15865 }, { "epoch": 153.84698795180722, "grad_norm": 4.239602088928223, "learning_rate": 2.2980582524271845e-05, "loss": 0.1549, "step": 15866 }, { "epoch": 153.8566265060241, "grad_norm": 12.297701835632324, "learning_rate": 2.2975728155339806e-05, "loss": 0.388, "step": 15867 }, { "epoch": 153.86626506024098, "grad_norm": 30.10553550720215, "learning_rate": 2.2970873786407767e-05, "loss": 0.2524, "step": 15868 }, { "epoch": 153.87590361445783, "grad_norm": 8.961601257324219, "learning_rate": 2.2966019417475728e-05, "loss": 0.2623, "step": 15869 }, { "epoch": 153.8855421686747, "grad_norm": 6.9474968910217285, "learning_rate": 2.296116504854369e-05, "loss": 0.2721, "step": 15870 }, { "epoch": 153.89518072289155, "grad_norm": 18.394479751586914, "learning_rate": 2.295631067961165e-05, "loss": 0.2493, "step": 15871 }, { "epoch": 153.90481927710843, "grad_norm": 1.0739667415618896, "learning_rate": 2.295145631067961e-05, "loss": 0.1589, "step": 15872 }, { "epoch": 153.9144578313253, "grad_norm": 42.09039306640625, "learning_rate": 2.2946601941747576e-05, "loss": 0.1812, "step": 15873 }, { "epoch": 153.92409638554216, "grad_norm": 9.476064682006836, "learning_rate": 2.2941747572815537e-05, "loss": 0.1357, "step": 15874 }, { "epoch": 153.93373493975903, "grad_norm": 19.342777252197266, "learning_rate": 2.2936893203883498e-05, "loss": 0.1848, "step": 15875 }, { "epoch": 153.9433734939759, "grad_norm": 3.347742795944214, "learning_rate": 2.293203883495146e-05, "loss": 0.4838, "step": 15876 }, { "epoch": 153.95301204819276, "grad_norm": 15.513620376586914, "learning_rate": 2.292718446601942e-05, "loss": 0.2178, "step": 15877 }, { "epoch": 153.96265060240964, "grad_norm": 10.748238563537598, "learning_rate": 2.292233009708738e-05, "loss": 0.2983, "step": 15878 }, { "epoch": 153.97228915662652, "grad_norm": 2.4413883686065674, "learning_rate": 2.291747572815534e-05, "loss": 0.3147, "step": 15879 }, { "epoch": 153.98192771084337, "grad_norm": 3.3860955238342285, "learning_rate": 2.29126213592233e-05, "loss": 0.0385, "step": 15880 }, { "epoch": 153.99156626506024, "grad_norm": 11.133444786071777, "learning_rate": 2.2907766990291262e-05, "loss": 0.1811, "step": 15881 }, { "epoch": 154.00722891566264, "grad_norm": 117.73343658447266, "learning_rate": 2.2902912621359223e-05, "loss": 0.3387, "step": 15882 }, { "epoch": 154.01686746987951, "grad_norm": 20.815929412841797, "learning_rate": 2.2898058252427184e-05, "loss": 0.2388, "step": 15883 }, { "epoch": 154.0265060240964, "grad_norm": 16.76983070373535, "learning_rate": 2.2893203883495145e-05, "loss": 0.2775, "step": 15884 }, { "epoch": 154.03614457831324, "grad_norm": 31.853832244873047, "learning_rate": 2.2888349514563106e-05, "loss": 0.1882, "step": 15885 }, { "epoch": 154.04578313253012, "grad_norm": 29.303991317749023, "learning_rate": 2.288349514563107e-05, "loss": 0.4009, "step": 15886 }, { "epoch": 154.055421686747, "grad_norm": 4.301364898681641, "learning_rate": 2.2878640776699032e-05, "loss": 0.2061, "step": 15887 }, { "epoch": 154.06506024096385, "grad_norm": 18.456438064575195, "learning_rate": 2.2873786407766993e-05, "loss": 0.1778, "step": 15888 }, { "epoch": 154.07469879518072, "grad_norm": 3.428697347640991, "learning_rate": 2.2868932038834954e-05, "loss": 0.2165, "step": 15889 }, { "epoch": 154.0843373493976, "grad_norm": 158.1898956298828, "learning_rate": 2.2864077669902915e-05, "loss": 0.354, "step": 15890 }, { "epoch": 154.09397590361445, "grad_norm": 11.567625999450684, "learning_rate": 2.2859223300970877e-05, "loss": 0.3921, "step": 15891 }, { "epoch": 154.10361445783133, "grad_norm": 32.835792541503906, "learning_rate": 2.2854368932038834e-05, "loss": 0.3311, "step": 15892 }, { "epoch": 154.1132530120482, "grad_norm": 8.514859199523926, "learning_rate": 2.2849514563106796e-05, "loss": 0.2255, "step": 15893 }, { "epoch": 154.12289156626505, "grad_norm": 3.624809503555298, "learning_rate": 2.2844660194174757e-05, "loss": 0.2464, "step": 15894 }, { "epoch": 154.13253012048193, "grad_norm": 10.9647798538208, "learning_rate": 2.2839805825242718e-05, "loss": 0.3758, "step": 15895 }, { "epoch": 154.1421686746988, "grad_norm": 4.7348175048828125, "learning_rate": 2.283495145631068e-05, "loss": 0.1053, "step": 15896 }, { "epoch": 154.15180722891566, "grad_norm": 43.872886657714844, "learning_rate": 2.283009708737864e-05, "loss": 0.3531, "step": 15897 }, { "epoch": 154.16144578313254, "grad_norm": 22.494216918945312, "learning_rate": 2.28252427184466e-05, "loss": 0.2615, "step": 15898 }, { "epoch": 154.17108433734938, "grad_norm": 13.176042556762695, "learning_rate": 2.2820388349514566e-05, "loss": 0.2066, "step": 15899 }, { "epoch": 154.18072289156626, "grad_norm": 2.722275972366333, "learning_rate": 2.2815533980582527e-05, "loss": 0.1253, "step": 15900 }, { "epoch": 154.19036144578314, "grad_norm": 16.355384826660156, "learning_rate": 2.2810679611650488e-05, "loss": 0.1487, "step": 15901 }, { "epoch": 154.2, "grad_norm": 7.893939018249512, "learning_rate": 2.280582524271845e-05, "loss": 0.1733, "step": 15902 }, { "epoch": 154.20963855421687, "grad_norm": 16.141490936279297, "learning_rate": 2.280097087378641e-05, "loss": 0.2849, "step": 15903 }, { "epoch": 154.21927710843374, "grad_norm": 28.32240867614746, "learning_rate": 2.279611650485437e-05, "loss": 0.199, "step": 15904 }, { "epoch": 154.2289156626506, "grad_norm": 19.988658905029297, "learning_rate": 2.2791262135922333e-05, "loss": 0.1364, "step": 15905 }, { "epoch": 154.23855421686747, "grad_norm": 3.039745569229126, "learning_rate": 2.278640776699029e-05, "loss": 0.2711, "step": 15906 }, { "epoch": 154.24819277108435, "grad_norm": 84.40877532958984, "learning_rate": 2.278155339805825e-05, "loss": 0.5082, "step": 15907 }, { "epoch": 154.2578313253012, "grad_norm": 54.93001937866211, "learning_rate": 2.2776699029126213e-05, "loss": 0.2169, "step": 15908 }, { "epoch": 154.26746987951807, "grad_norm": 8.896692276000977, "learning_rate": 2.2771844660194174e-05, "loss": 0.171, "step": 15909 }, { "epoch": 154.27710843373495, "grad_norm": 32.680511474609375, "learning_rate": 2.2766990291262135e-05, "loss": 0.4627, "step": 15910 }, { "epoch": 154.2867469879518, "grad_norm": 10.202248573303223, "learning_rate": 2.27621359223301e-05, "loss": 0.3613, "step": 15911 }, { "epoch": 154.29638554216868, "grad_norm": 14.334744453430176, "learning_rate": 2.275728155339806e-05, "loss": 0.1321, "step": 15912 }, { "epoch": 154.30602409638556, "grad_norm": 48.0291748046875, "learning_rate": 2.2752427184466022e-05, "loss": 0.1297, "step": 15913 }, { "epoch": 154.3156626506024, "grad_norm": 13.231820106506348, "learning_rate": 2.2747572815533983e-05, "loss": 0.2896, "step": 15914 }, { "epoch": 154.32530120481928, "grad_norm": 14.745336532592773, "learning_rate": 2.2742718446601944e-05, "loss": 0.497, "step": 15915 }, { "epoch": 154.33493975903613, "grad_norm": 12.393516540527344, "learning_rate": 2.2737864077669905e-05, "loss": 0.322, "step": 15916 }, { "epoch": 154.344578313253, "grad_norm": 15.474944114685059, "learning_rate": 2.2733009708737866e-05, "loss": 0.3825, "step": 15917 }, { "epoch": 154.3542168674699, "grad_norm": 13.005426406860352, "learning_rate": 2.2728155339805828e-05, "loss": 0.1409, "step": 15918 }, { "epoch": 154.36385542168674, "grad_norm": 18.804895401000977, "learning_rate": 2.272330097087379e-05, "loss": 0.1226, "step": 15919 }, { "epoch": 154.3734939759036, "grad_norm": 16.29558563232422, "learning_rate": 2.2718446601941746e-05, "loss": 0.2806, "step": 15920 }, { "epoch": 154.3831325301205, "grad_norm": 13.776588439941406, "learning_rate": 2.2713592233009708e-05, "loss": 0.1322, "step": 15921 }, { "epoch": 154.39277108433734, "grad_norm": 65.11807250976562, "learning_rate": 2.270873786407767e-05, "loss": 0.2584, "step": 15922 }, { "epoch": 154.40240963855422, "grad_norm": 16.666379928588867, "learning_rate": 2.270388349514563e-05, "loss": 0.2156, "step": 15923 }, { "epoch": 154.4120481927711, "grad_norm": 20.71424102783203, "learning_rate": 2.2699029126213594e-05, "loss": 0.3537, "step": 15924 }, { "epoch": 154.42168674698794, "grad_norm": 6.958265781402588, "learning_rate": 2.2694174757281556e-05, "loss": 0.2997, "step": 15925 }, { "epoch": 154.43132530120482, "grad_norm": 17.085865020751953, "learning_rate": 2.2689320388349517e-05, "loss": 0.3461, "step": 15926 }, { "epoch": 154.4409638554217, "grad_norm": 100.1315689086914, "learning_rate": 2.2684466019417478e-05, "loss": 0.1383, "step": 15927 }, { "epoch": 154.45060240963855, "grad_norm": 9.487367630004883, "learning_rate": 2.267961165048544e-05, "loss": 0.3219, "step": 15928 }, { "epoch": 154.46024096385543, "grad_norm": 1.7983012199401855, "learning_rate": 2.26747572815534e-05, "loss": 0.1658, "step": 15929 }, { "epoch": 154.46987951807228, "grad_norm": 23.010738372802734, "learning_rate": 2.266990291262136e-05, "loss": 0.3737, "step": 15930 }, { "epoch": 154.47951807228915, "grad_norm": 5.360426425933838, "learning_rate": 2.2665048543689322e-05, "loss": 0.2255, "step": 15931 }, { "epoch": 154.48915662650603, "grad_norm": 18.832752227783203, "learning_rate": 2.2660194174757284e-05, "loss": 0.3388, "step": 15932 }, { "epoch": 154.49879518072288, "grad_norm": 25.216827392578125, "learning_rate": 2.2655339805825245e-05, "loss": 0.2647, "step": 15933 }, { "epoch": 154.50843373493976, "grad_norm": 10.030233383178711, "learning_rate": 2.2650485436893203e-05, "loss": 0.3469, "step": 15934 }, { "epoch": 154.51807228915663, "grad_norm": 13.090449333190918, "learning_rate": 2.2645631067961164e-05, "loss": 0.159, "step": 15935 }, { "epoch": 154.52771084337348, "grad_norm": 15.456991195678711, "learning_rate": 2.2640776699029128e-05, "loss": 0.2603, "step": 15936 }, { "epoch": 154.53734939759036, "grad_norm": 20.664278030395508, "learning_rate": 2.263592233009709e-05, "loss": 0.1977, "step": 15937 }, { "epoch": 154.54698795180724, "grad_norm": 17.55908203125, "learning_rate": 2.263106796116505e-05, "loss": 0.096, "step": 15938 }, { "epoch": 154.5566265060241, "grad_norm": 0.9518771767616272, "learning_rate": 2.262621359223301e-05, "loss": 0.1878, "step": 15939 }, { "epoch": 154.56626506024097, "grad_norm": 34.64438247680664, "learning_rate": 2.2621359223300973e-05, "loss": 0.32, "step": 15940 }, { "epoch": 154.57590361445784, "grad_norm": 4.6129984855651855, "learning_rate": 2.2616504854368934e-05, "loss": 0.0961, "step": 15941 }, { "epoch": 154.5855421686747, "grad_norm": 4.956485748291016, "learning_rate": 2.2611650485436895e-05, "loss": 0.3646, "step": 15942 }, { "epoch": 154.59518072289157, "grad_norm": 34.3304443359375, "learning_rate": 2.2606796116504856e-05, "loss": 0.1942, "step": 15943 }, { "epoch": 154.60481927710845, "grad_norm": 4.358916759490967, "learning_rate": 2.2601941747572817e-05, "loss": 0.1807, "step": 15944 }, { "epoch": 154.6144578313253, "grad_norm": 34.44036865234375, "learning_rate": 2.259708737864078e-05, "loss": 0.1937, "step": 15945 }, { "epoch": 154.62409638554217, "grad_norm": 5.362836837768555, "learning_rate": 2.259223300970874e-05, "loss": 0.1846, "step": 15946 }, { "epoch": 154.63373493975902, "grad_norm": 3.324282646179199, "learning_rate": 2.2587378640776697e-05, "loss": 0.1301, "step": 15947 }, { "epoch": 154.6433734939759, "grad_norm": 27.145109176635742, "learning_rate": 2.258252427184466e-05, "loss": 0.2039, "step": 15948 }, { "epoch": 154.65301204819278, "grad_norm": 3.81429123878479, "learning_rate": 2.2577669902912623e-05, "loss": 0.2877, "step": 15949 }, { "epoch": 154.66265060240963, "grad_norm": 12.030644416809082, "learning_rate": 2.2572815533980584e-05, "loss": 0.3156, "step": 15950 }, { "epoch": 154.6722891566265, "grad_norm": 19.715848922729492, "learning_rate": 2.2567961165048545e-05, "loss": 0.2737, "step": 15951 }, { "epoch": 154.68192771084338, "grad_norm": 3.184889793395996, "learning_rate": 2.2563106796116507e-05, "loss": 0.172, "step": 15952 }, { "epoch": 154.69156626506023, "grad_norm": 21.671533584594727, "learning_rate": 2.2558252427184468e-05, "loss": 0.1915, "step": 15953 }, { "epoch": 154.7012048192771, "grad_norm": 15.862788200378418, "learning_rate": 2.255339805825243e-05, "loss": 0.174, "step": 15954 }, { "epoch": 154.710843373494, "grad_norm": 8.40114688873291, "learning_rate": 2.254854368932039e-05, "loss": 0.253, "step": 15955 }, { "epoch": 154.72048192771084, "grad_norm": 10.138675689697266, "learning_rate": 2.254368932038835e-05, "loss": 0.0785, "step": 15956 }, { "epoch": 154.7301204819277, "grad_norm": 21.52662467956543, "learning_rate": 2.2538834951456312e-05, "loss": 0.2302, "step": 15957 }, { "epoch": 154.7397590361446, "grad_norm": 5.777371406555176, "learning_rate": 2.2533980582524273e-05, "loss": 0.4136, "step": 15958 }, { "epoch": 154.74939759036144, "grad_norm": 5.009439468383789, "learning_rate": 2.2529126213592235e-05, "loss": 0.2586, "step": 15959 }, { "epoch": 154.75903614457832, "grad_norm": 38.69585037231445, "learning_rate": 2.2524271844660196e-05, "loss": 0.3428, "step": 15960 }, { "epoch": 154.76867469879517, "grad_norm": 28.540109634399414, "learning_rate": 2.2519417475728153e-05, "loss": 0.1569, "step": 15961 }, { "epoch": 154.77831325301204, "grad_norm": 482.4873962402344, "learning_rate": 2.2514563106796118e-05, "loss": 0.2486, "step": 15962 }, { "epoch": 154.78795180722892, "grad_norm": 14.162381172180176, "learning_rate": 2.250970873786408e-05, "loss": 0.1628, "step": 15963 }, { "epoch": 154.79759036144577, "grad_norm": 12.317463874816895, "learning_rate": 2.250485436893204e-05, "loss": 0.1612, "step": 15964 }, { "epoch": 154.80722891566265, "grad_norm": 2.753384590148926, "learning_rate": 2.25e-05, "loss": 0.2098, "step": 15965 }, { "epoch": 154.81686746987953, "grad_norm": 83.5277099609375, "learning_rate": 2.2495145631067963e-05, "loss": 0.2107, "step": 15966 }, { "epoch": 154.82650602409637, "grad_norm": 9.413572311401367, "learning_rate": 2.2490291262135924e-05, "loss": 0.2299, "step": 15967 }, { "epoch": 154.83614457831325, "grad_norm": 4.028763771057129, "learning_rate": 2.2485436893203885e-05, "loss": 0.3138, "step": 15968 }, { "epoch": 154.84578313253013, "grad_norm": 16.16790008544922, "learning_rate": 2.2480582524271846e-05, "loss": 0.4995, "step": 15969 }, { "epoch": 154.85542168674698, "grad_norm": 5.741204261779785, "learning_rate": 2.2475728155339807e-05, "loss": 0.301, "step": 15970 }, { "epoch": 154.86506024096386, "grad_norm": 7.784586429595947, "learning_rate": 2.2470873786407768e-05, "loss": 0.258, "step": 15971 }, { "epoch": 154.87469879518073, "grad_norm": 4.707902431488037, "learning_rate": 2.246601941747573e-05, "loss": 0.3056, "step": 15972 }, { "epoch": 154.88433734939758, "grad_norm": 8.845993995666504, "learning_rate": 2.246116504854369e-05, "loss": 0.1561, "step": 15973 }, { "epoch": 154.89397590361446, "grad_norm": 6.10231351852417, "learning_rate": 2.2456310679611652e-05, "loss": 0.2481, "step": 15974 }, { "epoch": 154.90361445783134, "grad_norm": 29.518203735351562, "learning_rate": 2.2451456310679613e-05, "loss": 0.2981, "step": 15975 }, { "epoch": 154.9132530120482, "grad_norm": 6.970771789550781, "learning_rate": 2.2446601941747574e-05, "loss": 0.2539, "step": 15976 }, { "epoch": 154.92289156626506, "grad_norm": 25.250354766845703, "learning_rate": 2.2441747572815535e-05, "loss": 0.5156, "step": 15977 }, { "epoch": 154.9325301204819, "grad_norm": 6.357021808624268, "learning_rate": 2.2436893203883496e-05, "loss": 0.2608, "step": 15978 }, { "epoch": 154.9421686746988, "grad_norm": 9.246259689331055, "learning_rate": 2.2432038834951457e-05, "loss": 0.25, "step": 15979 }, { "epoch": 154.95180722891567, "grad_norm": 4.452922344207764, "learning_rate": 2.242718446601942e-05, "loss": 0.195, "step": 15980 }, { "epoch": 154.96144578313252, "grad_norm": 36.399173736572266, "learning_rate": 2.242233009708738e-05, "loss": 0.2132, "step": 15981 }, { "epoch": 154.9710843373494, "grad_norm": 20.83085823059082, "learning_rate": 2.241747572815534e-05, "loss": 0.2691, "step": 15982 }, { "epoch": 154.98072289156627, "grad_norm": 39.62914276123047, "learning_rate": 2.2412621359223302e-05, "loss": 0.2772, "step": 15983 }, { "epoch": 154.99036144578312, "grad_norm": 1.6070910692214966, "learning_rate": 2.2407766990291263e-05, "loss": 0.2253, "step": 15984 }, { "epoch": 155.00602409638554, "grad_norm": 13.808568954467773, "learning_rate": 2.2402912621359224e-05, "loss": 0.3378, "step": 15985 }, { "epoch": 155.01566265060242, "grad_norm": 17.806903839111328, "learning_rate": 2.2398058252427185e-05, "loss": 0.3051, "step": 15986 }, { "epoch": 155.02530120481927, "grad_norm": 17.898731231689453, "learning_rate": 2.2393203883495147e-05, "loss": 0.3789, "step": 15987 }, { "epoch": 155.03493975903615, "grad_norm": 18.65511131286621, "learning_rate": 2.2388349514563108e-05, "loss": 0.3513, "step": 15988 }, { "epoch": 155.044578313253, "grad_norm": 4.475820064544678, "learning_rate": 2.238349514563107e-05, "loss": 0.1528, "step": 15989 }, { "epoch": 155.05421686746988, "grad_norm": 13.381897926330566, "learning_rate": 2.237864077669903e-05, "loss": 0.2361, "step": 15990 }, { "epoch": 155.06385542168675, "grad_norm": 16.439563751220703, "learning_rate": 2.237378640776699e-05, "loss": 0.1788, "step": 15991 }, { "epoch": 155.0734939759036, "grad_norm": 5.2349748611450195, "learning_rate": 2.2368932038834952e-05, "loss": 0.1726, "step": 15992 }, { "epoch": 155.08313253012048, "grad_norm": 2.941473960876465, "learning_rate": 2.2364077669902913e-05, "loss": 0.1645, "step": 15993 }, { "epoch": 155.09277108433736, "grad_norm": 9.413360595703125, "learning_rate": 2.2359223300970875e-05, "loss": 0.2352, "step": 15994 }, { "epoch": 155.1024096385542, "grad_norm": 20.31441307067871, "learning_rate": 2.2354368932038836e-05, "loss": 0.2739, "step": 15995 }, { "epoch": 155.11204819277108, "grad_norm": 1.8874915838241577, "learning_rate": 2.2349514563106797e-05, "loss": 0.0613, "step": 15996 }, { "epoch": 155.12168674698796, "grad_norm": 16.483388900756836, "learning_rate": 2.2344660194174758e-05, "loss": 0.1426, "step": 15997 }, { "epoch": 155.1313253012048, "grad_norm": 29.91376495361328, "learning_rate": 2.233980582524272e-05, "loss": 0.2559, "step": 15998 }, { "epoch": 155.1409638554217, "grad_norm": 7.50529146194458, "learning_rate": 2.233495145631068e-05, "loss": 0.2118, "step": 15999 }, { "epoch": 155.15060240963857, "grad_norm": 36.71673583984375, "learning_rate": 2.233009708737864e-05, "loss": 0.2641, "step": 16000 }, { "epoch": 155.16024096385541, "grad_norm": 7.195031642913818, "learning_rate": 2.2325242718446603e-05, "loss": 0.1723, "step": 16001 }, { "epoch": 155.1698795180723, "grad_norm": 18.8610897064209, "learning_rate": 2.2320388349514564e-05, "loss": 0.3106, "step": 16002 }, { "epoch": 155.17951807228917, "grad_norm": 7.855356216430664, "learning_rate": 2.2315533980582525e-05, "loss": 0.172, "step": 16003 }, { "epoch": 155.18915662650602, "grad_norm": 16.18365478515625, "learning_rate": 2.2310679611650486e-05, "loss": 0.2889, "step": 16004 }, { "epoch": 155.1987951807229, "grad_norm": 34.72475814819336, "learning_rate": 2.2305825242718447e-05, "loss": 0.2299, "step": 16005 }, { "epoch": 155.20843373493975, "grad_norm": 33.245723724365234, "learning_rate": 2.230097087378641e-05, "loss": 0.3758, "step": 16006 }, { "epoch": 155.21807228915662, "grad_norm": 10.045482635498047, "learning_rate": 2.229611650485437e-05, "loss": 0.1728, "step": 16007 }, { "epoch": 155.2277108433735, "grad_norm": 84.0745849609375, "learning_rate": 2.229126213592233e-05, "loss": 0.2869, "step": 16008 }, { "epoch": 155.23734939759035, "grad_norm": 23.225915908813477, "learning_rate": 2.2286407766990292e-05, "loss": 0.2475, "step": 16009 }, { "epoch": 155.24698795180723, "grad_norm": 6.114185333251953, "learning_rate": 2.2281553398058253e-05, "loss": 0.1481, "step": 16010 }, { "epoch": 155.2566265060241, "grad_norm": 53.674339294433594, "learning_rate": 2.2276699029126214e-05, "loss": 0.2517, "step": 16011 }, { "epoch": 155.26626506024095, "grad_norm": 7.636082172393799, "learning_rate": 2.227184466019418e-05, "loss": 0.2821, "step": 16012 }, { "epoch": 155.27590361445783, "grad_norm": 30.958127975463867, "learning_rate": 2.2266990291262136e-05, "loss": 0.3045, "step": 16013 }, { "epoch": 155.2855421686747, "grad_norm": 16.694643020629883, "learning_rate": 2.2262135922330098e-05, "loss": 0.451, "step": 16014 }, { "epoch": 155.29518072289156, "grad_norm": 11.965057373046875, "learning_rate": 2.225728155339806e-05, "loss": 0.1658, "step": 16015 }, { "epoch": 155.30481927710844, "grad_norm": 14.120831489562988, "learning_rate": 2.225242718446602e-05, "loss": 0.278, "step": 16016 }, { "epoch": 155.3144578313253, "grad_norm": 5.373762130737305, "learning_rate": 2.224757281553398e-05, "loss": 0.2521, "step": 16017 }, { "epoch": 155.32409638554216, "grad_norm": 4.755550861358643, "learning_rate": 2.2242718446601942e-05, "loss": 0.215, "step": 16018 }, { "epoch": 155.33373493975904, "grad_norm": 23.524124145507812, "learning_rate": 2.2237864077669903e-05, "loss": 0.3834, "step": 16019 }, { "epoch": 155.34337349397592, "grad_norm": 7.957056999206543, "learning_rate": 2.2233009708737864e-05, "loss": 0.3217, "step": 16020 }, { "epoch": 155.35301204819277, "grad_norm": 5.90304708480835, "learning_rate": 2.2228155339805826e-05, "loss": 0.1759, "step": 16021 }, { "epoch": 155.36265060240964, "grad_norm": 7.583195209503174, "learning_rate": 2.2223300970873787e-05, "loss": 0.1092, "step": 16022 }, { "epoch": 155.3722891566265, "grad_norm": 81.95763397216797, "learning_rate": 2.2218446601941748e-05, "loss": 0.0847, "step": 16023 }, { "epoch": 155.38192771084337, "grad_norm": 28.799903869628906, "learning_rate": 2.221359223300971e-05, "loss": 0.2627, "step": 16024 }, { "epoch": 155.39156626506025, "grad_norm": 11.308953285217285, "learning_rate": 2.2208737864077674e-05, "loss": 0.1368, "step": 16025 }, { "epoch": 155.4012048192771, "grad_norm": 5.982016563415527, "learning_rate": 2.2203883495145635e-05, "loss": 0.1827, "step": 16026 }, { "epoch": 155.41084337349398, "grad_norm": 10.90679931640625, "learning_rate": 2.2199029126213592e-05, "loss": 0.2303, "step": 16027 }, { "epoch": 155.42048192771085, "grad_norm": 12.64112377166748, "learning_rate": 2.2194174757281554e-05, "loss": 0.2973, "step": 16028 }, { "epoch": 155.4301204819277, "grad_norm": 94.01908874511719, "learning_rate": 2.2189320388349515e-05, "loss": 0.1784, "step": 16029 }, { "epoch": 155.43975903614458, "grad_norm": 5.030055046081543, "learning_rate": 2.2184466019417476e-05, "loss": 0.2478, "step": 16030 }, { "epoch": 155.44939759036146, "grad_norm": 55.8952751159668, "learning_rate": 2.2179611650485437e-05, "loss": 0.3359, "step": 16031 }, { "epoch": 155.4590361445783, "grad_norm": 55.061031341552734, "learning_rate": 2.2174757281553398e-05, "loss": 0.3876, "step": 16032 }, { "epoch": 155.46867469879518, "grad_norm": 14.406708717346191, "learning_rate": 2.216990291262136e-05, "loss": 0.2941, "step": 16033 }, { "epoch": 155.47831325301206, "grad_norm": 9.22161865234375, "learning_rate": 2.216504854368932e-05, "loss": 0.2294, "step": 16034 }, { "epoch": 155.4879518072289, "grad_norm": 12.175968170166016, "learning_rate": 2.216019417475728e-05, "loss": 0.2577, "step": 16035 }, { "epoch": 155.4975903614458, "grad_norm": 5.210244178771973, "learning_rate": 2.2155339805825243e-05, "loss": 0.1684, "step": 16036 }, { "epoch": 155.50722891566264, "grad_norm": 14.26012897491455, "learning_rate": 2.2150485436893207e-05, "loss": 0.4267, "step": 16037 }, { "epoch": 155.51686746987951, "grad_norm": 10.897314071655273, "learning_rate": 2.214563106796117e-05, "loss": 0.1272, "step": 16038 }, { "epoch": 155.5265060240964, "grad_norm": 7.674515724182129, "learning_rate": 2.214077669902913e-05, "loss": 0.2987, "step": 16039 }, { "epoch": 155.53614457831324, "grad_norm": 33.15388488769531, "learning_rate": 2.213592233009709e-05, "loss": 0.3276, "step": 16040 }, { "epoch": 155.54578313253012, "grad_norm": 39.82363510131836, "learning_rate": 2.213106796116505e-05, "loss": 0.1678, "step": 16041 }, { "epoch": 155.555421686747, "grad_norm": 8.697087287902832, "learning_rate": 2.212621359223301e-05, "loss": 0.2866, "step": 16042 }, { "epoch": 155.56506024096385, "grad_norm": 8.094355583190918, "learning_rate": 2.212135922330097e-05, "loss": 0.4684, "step": 16043 }, { "epoch": 155.57469879518072, "grad_norm": 9.545694351196289, "learning_rate": 2.2116504854368932e-05, "loss": 0.1671, "step": 16044 }, { "epoch": 155.5843373493976, "grad_norm": 57.225257873535156, "learning_rate": 2.2111650485436893e-05, "loss": 0.2802, "step": 16045 }, { "epoch": 155.59397590361445, "grad_norm": 15.945084571838379, "learning_rate": 2.2106796116504854e-05, "loss": 0.233, "step": 16046 }, { "epoch": 155.60361445783133, "grad_norm": 10.654186248779297, "learning_rate": 2.2101941747572815e-05, "loss": 0.1441, "step": 16047 }, { "epoch": 155.6132530120482, "grad_norm": 25.385324478149414, "learning_rate": 2.2097087378640777e-05, "loss": 0.1503, "step": 16048 }, { "epoch": 155.62289156626505, "grad_norm": 32.91648864746094, "learning_rate": 2.2092233009708738e-05, "loss": 0.2079, "step": 16049 }, { "epoch": 155.63253012048193, "grad_norm": 12.031149864196777, "learning_rate": 2.2087378640776702e-05, "loss": 0.1427, "step": 16050 }, { "epoch": 155.6421686746988, "grad_norm": 9.842061042785645, "learning_rate": 2.2082524271844663e-05, "loss": 0.4034, "step": 16051 }, { "epoch": 155.65180722891566, "grad_norm": 11.822829246520996, "learning_rate": 2.2077669902912624e-05, "loss": 0.3325, "step": 16052 }, { "epoch": 155.66144578313254, "grad_norm": 8.686294555664062, "learning_rate": 2.2072815533980586e-05, "loss": 0.1976, "step": 16053 }, { "epoch": 155.67108433734938, "grad_norm": 9.223370552062988, "learning_rate": 2.2067961165048547e-05, "loss": 0.3488, "step": 16054 }, { "epoch": 155.68072289156626, "grad_norm": 12.495956420898438, "learning_rate": 2.2063106796116505e-05, "loss": 0.3119, "step": 16055 }, { "epoch": 155.69036144578314, "grad_norm": 40.85175704956055, "learning_rate": 2.2058252427184466e-05, "loss": 0.2823, "step": 16056 }, { "epoch": 155.7, "grad_norm": 9.901436805725098, "learning_rate": 2.2053398058252427e-05, "loss": 0.2092, "step": 16057 }, { "epoch": 155.70963855421687, "grad_norm": 9.989123344421387, "learning_rate": 2.2048543689320388e-05, "loss": 0.3977, "step": 16058 }, { "epoch": 155.71927710843374, "grad_norm": 13.17243480682373, "learning_rate": 2.204368932038835e-05, "loss": 0.3299, "step": 16059 }, { "epoch": 155.7289156626506, "grad_norm": 11.460533142089844, "learning_rate": 2.203883495145631e-05, "loss": 0.2033, "step": 16060 }, { "epoch": 155.73855421686747, "grad_norm": 47.68974304199219, "learning_rate": 2.203398058252427e-05, "loss": 0.3918, "step": 16061 }, { "epoch": 155.74819277108435, "grad_norm": 11.081401824951172, "learning_rate": 2.2029126213592233e-05, "loss": 0.2613, "step": 16062 }, { "epoch": 155.7578313253012, "grad_norm": 8.90697956085205, "learning_rate": 2.2024271844660197e-05, "loss": 0.2537, "step": 16063 }, { "epoch": 155.76746987951807, "grad_norm": 13.985843658447266, "learning_rate": 2.2019417475728158e-05, "loss": 0.3464, "step": 16064 }, { "epoch": 155.77710843373495, "grad_norm": 5.278184413909912, "learning_rate": 2.201456310679612e-05, "loss": 0.1553, "step": 16065 }, { "epoch": 155.7867469879518, "grad_norm": 30.263994216918945, "learning_rate": 2.200970873786408e-05, "loss": 0.3988, "step": 16066 }, { "epoch": 155.79638554216868, "grad_norm": 13.919747352600098, "learning_rate": 2.200485436893204e-05, "loss": 0.3836, "step": 16067 }, { "epoch": 155.80602409638556, "grad_norm": 8.131640434265137, "learning_rate": 2.2000000000000003e-05, "loss": 0.1085, "step": 16068 }, { "epoch": 155.8156626506024, "grad_norm": 5.7971673011779785, "learning_rate": 2.199514563106796e-05, "loss": 0.3821, "step": 16069 }, { "epoch": 155.82530120481928, "grad_norm": 49.77770233154297, "learning_rate": 2.1990291262135922e-05, "loss": 0.2711, "step": 16070 }, { "epoch": 155.83493975903613, "grad_norm": 4.7737135887146, "learning_rate": 2.1985436893203883e-05, "loss": 0.2329, "step": 16071 }, { "epoch": 155.844578313253, "grad_norm": 32.52824020385742, "learning_rate": 2.1980582524271844e-05, "loss": 0.1322, "step": 16072 }, { "epoch": 155.8542168674699, "grad_norm": 45.7713737487793, "learning_rate": 2.1975728155339805e-05, "loss": 0.2428, "step": 16073 }, { "epoch": 155.86385542168674, "grad_norm": 29.368362426757812, "learning_rate": 2.1970873786407766e-05, "loss": 0.1565, "step": 16074 }, { "epoch": 155.8734939759036, "grad_norm": 11.729947090148926, "learning_rate": 2.196601941747573e-05, "loss": 0.4116, "step": 16075 }, { "epoch": 155.8831325301205, "grad_norm": 16.618453979492188, "learning_rate": 2.1961165048543692e-05, "loss": 0.3304, "step": 16076 }, { "epoch": 155.89277108433734, "grad_norm": 35.24692916870117, "learning_rate": 2.1956310679611653e-05, "loss": 0.1724, "step": 16077 }, { "epoch": 155.90240963855422, "grad_norm": 8.103324890136719, "learning_rate": 2.1951456310679614e-05, "loss": 0.2313, "step": 16078 }, { "epoch": 155.9120481927711, "grad_norm": 27.610490798950195, "learning_rate": 2.1946601941747575e-05, "loss": 0.2089, "step": 16079 }, { "epoch": 155.92168674698794, "grad_norm": 15.520000457763672, "learning_rate": 2.1941747572815537e-05, "loss": 0.1326, "step": 16080 }, { "epoch": 155.93132530120482, "grad_norm": 27.175832748413086, "learning_rate": 2.1936893203883498e-05, "loss": 0.3726, "step": 16081 }, { "epoch": 155.9409638554217, "grad_norm": 9.144906997680664, "learning_rate": 2.193203883495146e-05, "loss": 0.1285, "step": 16082 }, { "epoch": 155.95060240963855, "grad_norm": 95.14610290527344, "learning_rate": 2.1927184466019417e-05, "loss": 0.4107, "step": 16083 }, { "epoch": 155.96024096385543, "grad_norm": 8.085227012634277, "learning_rate": 2.1922330097087378e-05, "loss": 0.2516, "step": 16084 }, { "epoch": 155.96987951807228, "grad_norm": 7.393160820007324, "learning_rate": 2.191747572815534e-05, "loss": 0.0807, "step": 16085 }, { "epoch": 155.97951807228915, "grad_norm": 10.075603485107422, "learning_rate": 2.19126213592233e-05, "loss": 0.3066, "step": 16086 }, { "epoch": 155.98915662650603, "grad_norm": 23.021718978881836, "learning_rate": 2.190776699029126e-05, "loss": 0.2701, "step": 16087 }, { "epoch": 156.00481927710842, "grad_norm": 8.679656982421875, "learning_rate": 2.1902912621359226e-05, "loss": 0.0841, "step": 16088 }, { "epoch": 156.0144578313253, "grad_norm": 6.868359088897705, "learning_rate": 2.1898058252427187e-05, "loss": 0.2823, "step": 16089 }, { "epoch": 156.02409638554218, "grad_norm": 3.278427839279175, "learning_rate": 2.1893203883495148e-05, "loss": 0.1403, "step": 16090 }, { "epoch": 156.03373493975903, "grad_norm": 18.383068084716797, "learning_rate": 2.188834951456311e-05, "loss": 0.2362, "step": 16091 }, { "epoch": 156.0433734939759, "grad_norm": 13.34965705871582, "learning_rate": 2.188349514563107e-05, "loss": 0.1866, "step": 16092 }, { "epoch": 156.05301204819278, "grad_norm": 26.52058982849121, "learning_rate": 2.187864077669903e-05, "loss": 0.4301, "step": 16093 }, { "epoch": 156.06265060240963, "grad_norm": 11.98659610748291, "learning_rate": 2.1873786407766993e-05, "loss": 0.3322, "step": 16094 }, { "epoch": 156.0722891566265, "grad_norm": 41.41917037963867, "learning_rate": 2.1868932038834954e-05, "loss": 0.4489, "step": 16095 }, { "epoch": 156.08192771084336, "grad_norm": 31.525060653686523, "learning_rate": 2.186407766990291e-05, "loss": 0.5031, "step": 16096 }, { "epoch": 156.09156626506024, "grad_norm": 22.607948303222656, "learning_rate": 2.1859223300970873e-05, "loss": 0.2287, "step": 16097 }, { "epoch": 156.10120481927711, "grad_norm": 22.923147201538086, "learning_rate": 2.1854368932038834e-05, "loss": 0.1986, "step": 16098 }, { "epoch": 156.11084337349396, "grad_norm": 24.632925033569336, "learning_rate": 2.1849514563106795e-05, "loss": 0.3722, "step": 16099 }, { "epoch": 156.12048192771084, "grad_norm": 4.0992112159729, "learning_rate": 2.1844660194174756e-05, "loss": 0.2914, "step": 16100 }, { "epoch": 156.13012048192772, "grad_norm": 10.662991523742676, "learning_rate": 2.183980582524272e-05, "loss": 0.257, "step": 16101 }, { "epoch": 156.13975903614457, "grad_norm": 11.18189525604248, "learning_rate": 2.1834951456310682e-05, "loss": 0.4819, "step": 16102 }, { "epoch": 156.14939759036145, "grad_norm": 16.15448760986328, "learning_rate": 2.1830097087378643e-05, "loss": 0.2147, "step": 16103 }, { "epoch": 156.15903614457832, "grad_norm": 22.009775161743164, "learning_rate": 2.1825242718446604e-05, "loss": 0.3538, "step": 16104 }, { "epoch": 156.16867469879517, "grad_norm": 15.872733116149902, "learning_rate": 2.1820388349514565e-05, "loss": 0.1995, "step": 16105 }, { "epoch": 156.17831325301205, "grad_norm": 15.64659595489502, "learning_rate": 2.1815533980582526e-05, "loss": 0.2383, "step": 16106 }, { "epoch": 156.18795180722893, "grad_norm": 40.90347671508789, "learning_rate": 2.1810679611650487e-05, "loss": 0.2852, "step": 16107 }, { "epoch": 156.19759036144578, "grad_norm": 4.548042297363281, "learning_rate": 2.180582524271845e-05, "loss": 0.1179, "step": 16108 }, { "epoch": 156.20722891566265, "grad_norm": 13.058464050292969, "learning_rate": 2.180097087378641e-05, "loss": 0.1918, "step": 16109 }, { "epoch": 156.21686746987953, "grad_norm": 10.130793571472168, "learning_rate": 2.1796116504854368e-05, "loss": 0.5596, "step": 16110 }, { "epoch": 156.22650602409638, "grad_norm": 61.912933349609375, "learning_rate": 2.179126213592233e-05, "loss": 0.2944, "step": 16111 }, { "epoch": 156.23614457831326, "grad_norm": 5.2639617919921875, "learning_rate": 2.178640776699029e-05, "loss": 0.2359, "step": 16112 }, { "epoch": 156.2457831325301, "grad_norm": 8.765646934509277, "learning_rate": 2.1781553398058254e-05, "loss": 0.1117, "step": 16113 }, { "epoch": 156.25542168674698, "grad_norm": 10.459554672241211, "learning_rate": 2.1776699029126216e-05, "loss": 0.3116, "step": 16114 }, { "epoch": 156.26506024096386, "grad_norm": 18.875732421875, "learning_rate": 2.1771844660194177e-05, "loss": 0.1554, "step": 16115 }, { "epoch": 156.2746987951807, "grad_norm": 22.533367156982422, "learning_rate": 2.1766990291262138e-05, "loss": 0.481, "step": 16116 }, { "epoch": 156.2843373493976, "grad_norm": 70.02394104003906, "learning_rate": 2.17621359223301e-05, "loss": 0.1529, "step": 16117 }, { "epoch": 156.29397590361447, "grad_norm": 19.49202537536621, "learning_rate": 2.175728155339806e-05, "loss": 0.2802, "step": 16118 }, { "epoch": 156.30361445783132, "grad_norm": 5.35228157043457, "learning_rate": 2.175242718446602e-05, "loss": 0.3148, "step": 16119 }, { "epoch": 156.3132530120482, "grad_norm": 14.028156280517578, "learning_rate": 2.1747572815533982e-05, "loss": 0.3651, "step": 16120 }, { "epoch": 156.32289156626507, "grad_norm": 59.76432418823242, "learning_rate": 2.1742718446601944e-05, "loss": 0.1886, "step": 16121 }, { "epoch": 156.33253012048192, "grad_norm": 7.630683422088623, "learning_rate": 2.1737864077669905e-05, "loss": 0.1738, "step": 16122 }, { "epoch": 156.3421686746988, "grad_norm": 19.579092025756836, "learning_rate": 2.1733009708737866e-05, "loss": 0.215, "step": 16123 }, { "epoch": 156.35180722891567, "grad_norm": 10.747307777404785, "learning_rate": 2.1728155339805824e-05, "loss": 0.1798, "step": 16124 }, { "epoch": 156.36144578313252, "grad_norm": 22.777690887451172, "learning_rate": 2.1723300970873785e-05, "loss": 0.278, "step": 16125 }, { "epoch": 156.3710843373494, "grad_norm": 9.841218948364258, "learning_rate": 2.171844660194175e-05, "loss": 0.2715, "step": 16126 }, { "epoch": 156.38072289156628, "grad_norm": 59.68397903442383, "learning_rate": 2.171359223300971e-05, "loss": 0.2566, "step": 16127 }, { "epoch": 156.39036144578313, "grad_norm": 50.783164978027344, "learning_rate": 2.170873786407767e-05, "loss": 0.1912, "step": 16128 }, { "epoch": 156.4, "grad_norm": 43.11649703979492, "learning_rate": 2.1703883495145633e-05, "loss": 0.4778, "step": 16129 }, { "epoch": 156.40963855421685, "grad_norm": 3.3367161750793457, "learning_rate": 2.1699029126213594e-05, "loss": 0.0417, "step": 16130 }, { "epoch": 156.41927710843373, "grad_norm": 13.396255493164062, "learning_rate": 2.1694174757281555e-05, "loss": 0.275, "step": 16131 }, { "epoch": 156.4289156626506, "grad_norm": 12.034276008605957, "learning_rate": 2.1689320388349516e-05, "loss": 0.176, "step": 16132 }, { "epoch": 156.43855421686746, "grad_norm": 9.448689460754395, "learning_rate": 2.1684466019417477e-05, "loss": 0.3252, "step": 16133 }, { "epoch": 156.44819277108434, "grad_norm": 11.10759162902832, "learning_rate": 2.167961165048544e-05, "loss": 0.1677, "step": 16134 }, { "epoch": 156.4578313253012, "grad_norm": 7.165306568145752, "learning_rate": 2.16747572815534e-05, "loss": 0.3638, "step": 16135 }, { "epoch": 156.46746987951806, "grad_norm": 36.837890625, "learning_rate": 2.166990291262136e-05, "loss": 0.1556, "step": 16136 }, { "epoch": 156.47710843373494, "grad_norm": 3.390892744064331, "learning_rate": 2.1665048543689322e-05, "loss": 0.1995, "step": 16137 }, { "epoch": 156.48674698795182, "grad_norm": 6.066041946411133, "learning_rate": 2.1660194174757283e-05, "loss": 0.1528, "step": 16138 }, { "epoch": 156.49638554216867, "grad_norm": 10.791607856750488, "learning_rate": 2.1655339805825244e-05, "loss": 0.361, "step": 16139 }, { "epoch": 156.50602409638554, "grad_norm": 11.66588306427002, "learning_rate": 2.1650485436893205e-05, "loss": 0.2457, "step": 16140 }, { "epoch": 156.51566265060242, "grad_norm": 21.593992233276367, "learning_rate": 2.1645631067961166e-05, "loss": 0.4324, "step": 16141 }, { "epoch": 156.52530120481927, "grad_norm": 34.60203170776367, "learning_rate": 2.1640776699029128e-05, "loss": 0.1579, "step": 16142 }, { "epoch": 156.53493975903615, "grad_norm": 22.11075782775879, "learning_rate": 2.163592233009709e-05, "loss": 0.1865, "step": 16143 }, { "epoch": 156.544578313253, "grad_norm": 14.373056411743164, "learning_rate": 2.163106796116505e-05, "loss": 0.2505, "step": 16144 }, { "epoch": 156.55421686746988, "grad_norm": 21.958999633789062, "learning_rate": 2.162621359223301e-05, "loss": 0.1706, "step": 16145 }, { "epoch": 156.56385542168675, "grad_norm": 11.019993782043457, "learning_rate": 2.1621359223300972e-05, "loss": 0.2064, "step": 16146 }, { "epoch": 156.5734939759036, "grad_norm": 26.13070297241211, "learning_rate": 2.1616504854368933e-05, "loss": 0.249, "step": 16147 }, { "epoch": 156.58313253012048, "grad_norm": 12.359053611755371, "learning_rate": 2.1611650485436894e-05, "loss": 0.2263, "step": 16148 }, { "epoch": 156.59277108433736, "grad_norm": 16.921178817749023, "learning_rate": 2.1606796116504856e-05, "loss": 0.4232, "step": 16149 }, { "epoch": 156.6024096385542, "grad_norm": 12.027274131774902, "learning_rate": 2.1601941747572817e-05, "loss": 0.5045, "step": 16150 }, { "epoch": 156.61204819277108, "grad_norm": 12.33978271484375, "learning_rate": 2.1597087378640778e-05, "loss": 0.1775, "step": 16151 }, { "epoch": 156.62168674698796, "grad_norm": 2.672255754470825, "learning_rate": 2.159223300970874e-05, "loss": 0.134, "step": 16152 }, { "epoch": 156.6313253012048, "grad_norm": 9.393978118896484, "learning_rate": 2.15873786407767e-05, "loss": 0.3608, "step": 16153 }, { "epoch": 156.6409638554217, "grad_norm": 6.872413635253906, "learning_rate": 2.158252427184466e-05, "loss": 0.1733, "step": 16154 }, { "epoch": 156.65060240963857, "grad_norm": 4.440461158752441, "learning_rate": 2.1577669902912622e-05, "loss": 0.2906, "step": 16155 }, { "epoch": 156.66024096385541, "grad_norm": 8.480254173278809, "learning_rate": 2.1572815533980584e-05, "loss": 0.1368, "step": 16156 }, { "epoch": 156.6698795180723, "grad_norm": 28.090932846069336, "learning_rate": 2.1567961165048545e-05, "loss": 0.2275, "step": 16157 }, { "epoch": 156.67951807228917, "grad_norm": 6.7428202629089355, "learning_rate": 2.1563106796116506e-05, "loss": 0.3076, "step": 16158 }, { "epoch": 156.68915662650602, "grad_norm": 10.625663757324219, "learning_rate": 2.1558252427184467e-05, "loss": 0.317, "step": 16159 }, { "epoch": 156.6987951807229, "grad_norm": 38.005611419677734, "learning_rate": 2.1553398058252428e-05, "loss": 0.3626, "step": 16160 }, { "epoch": 156.70843373493975, "grad_norm": 7.540511131286621, "learning_rate": 2.154854368932039e-05, "loss": 0.2568, "step": 16161 }, { "epoch": 156.71807228915662, "grad_norm": 18.785436630249023, "learning_rate": 2.154368932038835e-05, "loss": 0.1278, "step": 16162 }, { "epoch": 156.7277108433735, "grad_norm": 54.210140228271484, "learning_rate": 2.153883495145631e-05, "loss": 0.6004, "step": 16163 }, { "epoch": 156.73734939759035, "grad_norm": 42.25011444091797, "learning_rate": 2.1533980582524273e-05, "loss": 0.2231, "step": 16164 }, { "epoch": 156.74698795180723, "grad_norm": 28.87569808959961, "learning_rate": 2.1529126213592234e-05, "loss": 0.3106, "step": 16165 }, { "epoch": 156.7566265060241, "grad_norm": 11.387360572814941, "learning_rate": 2.1524271844660195e-05, "loss": 0.2451, "step": 16166 }, { "epoch": 156.76626506024095, "grad_norm": 4.8804216384887695, "learning_rate": 2.1519417475728156e-05, "loss": 0.1145, "step": 16167 }, { "epoch": 156.77590361445783, "grad_norm": 14.860797882080078, "learning_rate": 2.1514563106796117e-05, "loss": 0.2499, "step": 16168 }, { "epoch": 156.7855421686747, "grad_norm": 9.841903686523438, "learning_rate": 2.150970873786408e-05, "loss": 0.1215, "step": 16169 }, { "epoch": 156.79518072289156, "grad_norm": 5.753096580505371, "learning_rate": 2.150485436893204e-05, "loss": 0.2444, "step": 16170 }, { "epoch": 156.80481927710844, "grad_norm": 9.085471153259277, "learning_rate": 2.15e-05, "loss": 0.1384, "step": 16171 }, { "epoch": 156.8144578313253, "grad_norm": 18.77327537536621, "learning_rate": 2.1495145631067962e-05, "loss": 0.4392, "step": 16172 }, { "epoch": 156.82409638554216, "grad_norm": 4.8878397941589355, "learning_rate": 2.1490291262135923e-05, "loss": 0.2319, "step": 16173 }, { "epoch": 156.83373493975904, "grad_norm": 34.312660217285156, "learning_rate": 2.1485436893203884e-05, "loss": 0.2306, "step": 16174 }, { "epoch": 156.84337349397592, "grad_norm": 24.76851463317871, "learning_rate": 2.1480582524271845e-05, "loss": 0.36, "step": 16175 }, { "epoch": 156.85301204819277, "grad_norm": 10.256804466247559, "learning_rate": 2.1475728155339807e-05, "loss": 0.3811, "step": 16176 }, { "epoch": 156.86265060240964, "grad_norm": 12.194602966308594, "learning_rate": 2.1470873786407768e-05, "loss": 0.3307, "step": 16177 }, { "epoch": 156.8722891566265, "grad_norm": 6.884934902191162, "learning_rate": 2.146601941747573e-05, "loss": 0.1777, "step": 16178 }, { "epoch": 156.88192771084337, "grad_norm": 11.120501518249512, "learning_rate": 2.146116504854369e-05, "loss": 0.3364, "step": 16179 }, { "epoch": 156.89156626506025, "grad_norm": 6.844150066375732, "learning_rate": 2.145631067961165e-05, "loss": 0.2125, "step": 16180 }, { "epoch": 156.9012048192771, "grad_norm": 9.6993408203125, "learning_rate": 2.1451456310679612e-05, "loss": 0.1647, "step": 16181 }, { "epoch": 156.91084337349398, "grad_norm": 12.35807991027832, "learning_rate": 2.1446601941747573e-05, "loss": 0.3103, "step": 16182 }, { "epoch": 156.92048192771085, "grad_norm": 14.257269859313965, "learning_rate": 2.1441747572815535e-05, "loss": 0.2006, "step": 16183 }, { "epoch": 156.9301204819277, "grad_norm": 13.702394485473633, "learning_rate": 2.1436893203883496e-05, "loss": 0.2471, "step": 16184 }, { "epoch": 156.93975903614458, "grad_norm": 38.103851318359375, "learning_rate": 2.1432038834951457e-05, "loss": 0.266, "step": 16185 }, { "epoch": 156.94939759036146, "grad_norm": 13.147793769836426, "learning_rate": 2.1427184466019418e-05, "loss": 0.2949, "step": 16186 }, { "epoch": 156.9590361445783, "grad_norm": 7.972166061401367, "learning_rate": 2.142233009708738e-05, "loss": 0.4754, "step": 16187 }, { "epoch": 156.96867469879518, "grad_norm": 31.700029373168945, "learning_rate": 2.141747572815534e-05, "loss": 0.3279, "step": 16188 }, { "epoch": 156.97831325301206, "grad_norm": 22.270383834838867, "learning_rate": 2.1412621359223305e-05, "loss": 0.283, "step": 16189 }, { "epoch": 156.9879518072289, "grad_norm": 8.918168067932129, "learning_rate": 2.1407766990291263e-05, "loss": 0.2407, "step": 16190 }, { "epoch": 157.00361445783133, "grad_norm": 31.14456558227539, "learning_rate": 2.1402912621359224e-05, "loss": 0.4324, "step": 16191 }, { "epoch": 157.01325301204818, "grad_norm": 13.001754760742188, "learning_rate": 2.1398058252427185e-05, "loss": 0.1851, "step": 16192 }, { "epoch": 157.02289156626506, "grad_norm": 4.692123889923096, "learning_rate": 2.1393203883495146e-05, "loss": 0.2251, "step": 16193 }, { "epoch": 157.03253012048194, "grad_norm": 18.027944564819336, "learning_rate": 2.1388349514563107e-05, "loss": 0.1882, "step": 16194 }, { "epoch": 157.0421686746988, "grad_norm": 12.898406028747559, "learning_rate": 2.138349514563107e-05, "loss": 0.2246, "step": 16195 }, { "epoch": 157.05180722891566, "grad_norm": 21.406816482543945, "learning_rate": 2.137864077669903e-05, "loss": 0.3548, "step": 16196 }, { "epoch": 157.06144578313254, "grad_norm": 9.691915512084961, "learning_rate": 2.137378640776699e-05, "loss": 0.2182, "step": 16197 }, { "epoch": 157.0710843373494, "grad_norm": 15.363600730895996, "learning_rate": 2.1368932038834952e-05, "loss": 0.257, "step": 16198 }, { "epoch": 157.08072289156627, "grad_norm": 21.273157119750977, "learning_rate": 2.1364077669902913e-05, "loss": 0.1626, "step": 16199 }, { "epoch": 157.09036144578315, "grad_norm": 4.076598167419434, "learning_rate": 2.1359223300970874e-05, "loss": 0.2577, "step": 16200 }, { "epoch": 157.1, "grad_norm": 14.522225379943848, "learning_rate": 2.1354368932038835e-05, "loss": 0.3908, "step": 16201 }, { "epoch": 157.10963855421687, "grad_norm": 10.540736198425293, "learning_rate": 2.13495145631068e-05, "loss": 0.2135, "step": 16202 }, { "epoch": 157.11927710843372, "grad_norm": 8.761310577392578, "learning_rate": 2.134466019417476e-05, "loss": 0.2166, "step": 16203 }, { "epoch": 157.1289156626506, "grad_norm": 5.678889751434326, "learning_rate": 2.133980582524272e-05, "loss": 0.1472, "step": 16204 }, { "epoch": 157.13855421686748, "grad_norm": 16.775623321533203, "learning_rate": 2.133495145631068e-05, "loss": 0.319, "step": 16205 }, { "epoch": 157.14819277108433, "grad_norm": 6.8121747970581055, "learning_rate": 2.133009708737864e-05, "loss": 0.2448, "step": 16206 }, { "epoch": 157.1578313253012, "grad_norm": 27.003374099731445, "learning_rate": 2.1325242718446602e-05, "loss": 0.348, "step": 16207 }, { "epoch": 157.16746987951808, "grad_norm": 9.95833969116211, "learning_rate": 2.1320388349514563e-05, "loss": 0.1799, "step": 16208 }, { "epoch": 157.17710843373493, "grad_norm": 34.389854431152344, "learning_rate": 2.1315533980582524e-05, "loss": 0.5083, "step": 16209 }, { "epoch": 157.1867469879518, "grad_norm": 10.337785720825195, "learning_rate": 2.1310679611650486e-05, "loss": 0.2869, "step": 16210 }, { "epoch": 157.19638554216868, "grad_norm": 14.730609893798828, "learning_rate": 2.1305825242718447e-05, "loss": 0.1862, "step": 16211 }, { "epoch": 157.20602409638553, "grad_norm": 7.953160285949707, "learning_rate": 2.1300970873786408e-05, "loss": 0.2629, "step": 16212 }, { "epoch": 157.2156626506024, "grad_norm": 5.388169765472412, "learning_rate": 2.129611650485437e-05, "loss": 0.1301, "step": 16213 }, { "epoch": 157.2253012048193, "grad_norm": 28.73394775390625, "learning_rate": 2.1291262135922333e-05, "loss": 0.2908, "step": 16214 }, { "epoch": 157.23493975903614, "grad_norm": 41.46286392211914, "learning_rate": 2.1286407766990295e-05, "loss": 0.1913, "step": 16215 }, { "epoch": 157.24457831325302, "grad_norm": 25.415992736816406, "learning_rate": 2.1281553398058256e-05, "loss": 0.3094, "step": 16216 }, { "epoch": 157.2542168674699, "grad_norm": 14.769413948059082, "learning_rate": 2.1276699029126217e-05, "loss": 0.1593, "step": 16217 }, { "epoch": 157.26385542168674, "grad_norm": 9.043758392333984, "learning_rate": 2.1271844660194175e-05, "loss": 0.1287, "step": 16218 }, { "epoch": 157.27349397590362, "grad_norm": 8.056327819824219, "learning_rate": 2.1266990291262136e-05, "loss": 0.1864, "step": 16219 }, { "epoch": 157.28313253012047, "grad_norm": 12.099493026733398, "learning_rate": 2.1262135922330097e-05, "loss": 0.227, "step": 16220 }, { "epoch": 157.29277108433735, "grad_norm": 12.51867961883545, "learning_rate": 2.1257281553398058e-05, "loss": 0.2878, "step": 16221 }, { "epoch": 157.30240963855422, "grad_norm": 22.155498504638672, "learning_rate": 2.125242718446602e-05, "loss": 0.3029, "step": 16222 }, { "epoch": 157.31204819277107, "grad_norm": 10.977629661560059, "learning_rate": 2.124757281553398e-05, "loss": 0.3344, "step": 16223 }, { "epoch": 157.32168674698795, "grad_norm": 31.879823684692383, "learning_rate": 2.124271844660194e-05, "loss": 0.2023, "step": 16224 }, { "epoch": 157.33132530120483, "grad_norm": 4.395477771759033, "learning_rate": 2.1237864077669903e-05, "loss": 0.0793, "step": 16225 }, { "epoch": 157.34096385542168, "grad_norm": 4.590161323547363, "learning_rate": 2.1233009708737864e-05, "loss": 0.0834, "step": 16226 }, { "epoch": 157.35060240963855, "grad_norm": 7.7107648849487305, "learning_rate": 2.122815533980583e-05, "loss": 0.3742, "step": 16227 }, { "epoch": 157.36024096385543, "grad_norm": 7.364321708679199, "learning_rate": 2.122330097087379e-05, "loss": 0.2745, "step": 16228 }, { "epoch": 157.36987951807228, "grad_norm": 37.106689453125, "learning_rate": 2.121844660194175e-05, "loss": 0.1315, "step": 16229 }, { "epoch": 157.37951807228916, "grad_norm": 23.276782989501953, "learning_rate": 2.1213592233009712e-05, "loss": 0.4035, "step": 16230 }, { "epoch": 157.38915662650604, "grad_norm": 8.328641891479492, "learning_rate": 2.1208737864077673e-05, "loss": 0.2513, "step": 16231 }, { "epoch": 157.39879518072289, "grad_norm": 16.550819396972656, "learning_rate": 2.120388349514563e-05, "loss": 0.3415, "step": 16232 }, { "epoch": 157.40843373493976, "grad_norm": 9.325145721435547, "learning_rate": 2.1199029126213592e-05, "loss": 0.2476, "step": 16233 }, { "epoch": 157.41807228915664, "grad_norm": 13.444832801818848, "learning_rate": 2.1194174757281553e-05, "loss": 0.3239, "step": 16234 }, { "epoch": 157.4277108433735, "grad_norm": 13.584829330444336, "learning_rate": 2.1189320388349514e-05, "loss": 0.5328, "step": 16235 }, { "epoch": 157.43734939759037, "grad_norm": 10.24951171875, "learning_rate": 2.1184466019417475e-05, "loss": 0.3399, "step": 16236 }, { "epoch": 157.44698795180722, "grad_norm": 2.360506296157837, "learning_rate": 2.1179611650485436e-05, "loss": 0.3516, "step": 16237 }, { "epoch": 157.4566265060241, "grad_norm": 12.804445266723633, "learning_rate": 2.1174757281553398e-05, "loss": 0.1922, "step": 16238 }, { "epoch": 157.46626506024097, "grad_norm": 3.9238100051879883, "learning_rate": 2.1169902912621362e-05, "loss": 0.1961, "step": 16239 }, { "epoch": 157.47590361445782, "grad_norm": 19.696622848510742, "learning_rate": 2.1165048543689323e-05, "loss": 0.2485, "step": 16240 }, { "epoch": 157.4855421686747, "grad_norm": 106.75839233398438, "learning_rate": 2.1160194174757284e-05, "loss": 0.2019, "step": 16241 }, { "epoch": 157.49518072289158, "grad_norm": 11.720973014831543, "learning_rate": 2.1155339805825246e-05, "loss": 0.2051, "step": 16242 }, { "epoch": 157.50481927710842, "grad_norm": 55.591217041015625, "learning_rate": 2.1150485436893207e-05, "loss": 0.2186, "step": 16243 }, { "epoch": 157.5144578313253, "grad_norm": 24.0345458984375, "learning_rate": 2.1145631067961168e-05, "loss": 0.2613, "step": 16244 }, { "epoch": 157.52409638554218, "grad_norm": 10.338069915771484, "learning_rate": 2.1140776699029126e-05, "loss": 0.1863, "step": 16245 }, { "epoch": 157.53373493975903, "grad_norm": 7.512214183807373, "learning_rate": 2.1135922330097087e-05, "loss": 0.2751, "step": 16246 }, { "epoch": 157.5433734939759, "grad_norm": 7.644897937774658, "learning_rate": 2.1131067961165048e-05, "loss": 0.2008, "step": 16247 }, { "epoch": 157.55301204819278, "grad_norm": 10.043253898620605, "learning_rate": 2.112621359223301e-05, "loss": 0.3704, "step": 16248 }, { "epoch": 157.56265060240963, "grad_norm": 8.19226360321045, "learning_rate": 2.112135922330097e-05, "loss": 0.4651, "step": 16249 }, { "epoch": 157.5722891566265, "grad_norm": 8.28219223022461, "learning_rate": 2.111650485436893e-05, "loss": 0.1583, "step": 16250 }, { "epoch": 157.58192771084336, "grad_norm": 17.225669860839844, "learning_rate": 2.1111650485436892e-05, "loss": 0.3269, "step": 16251 }, { "epoch": 157.59156626506024, "grad_norm": 18.31282615661621, "learning_rate": 2.1106796116504857e-05, "loss": 0.2143, "step": 16252 }, { "epoch": 157.60120481927711, "grad_norm": 8.13243579864502, "learning_rate": 2.1101941747572818e-05, "loss": 0.0902, "step": 16253 }, { "epoch": 157.61084337349396, "grad_norm": 40.95182418823242, "learning_rate": 2.109708737864078e-05, "loss": 0.2226, "step": 16254 }, { "epoch": 157.62048192771084, "grad_norm": 26.351726531982422, "learning_rate": 2.109223300970874e-05, "loss": 0.1913, "step": 16255 }, { "epoch": 157.63012048192772, "grad_norm": 16.746917724609375, "learning_rate": 2.10873786407767e-05, "loss": 0.2495, "step": 16256 }, { "epoch": 157.63975903614457, "grad_norm": 8.25638198852539, "learning_rate": 2.1082524271844663e-05, "loss": 0.178, "step": 16257 }, { "epoch": 157.64939759036145, "grad_norm": 4.267422676086426, "learning_rate": 2.1077669902912624e-05, "loss": 0.1201, "step": 16258 }, { "epoch": 157.65903614457832, "grad_norm": 31.289350509643555, "learning_rate": 2.107281553398058e-05, "loss": 0.4967, "step": 16259 }, { "epoch": 157.66867469879517, "grad_norm": 21.318531036376953, "learning_rate": 2.1067961165048543e-05, "loss": 0.2189, "step": 16260 }, { "epoch": 157.67831325301205, "grad_norm": 14.666097640991211, "learning_rate": 2.1063106796116504e-05, "loss": 0.2632, "step": 16261 }, { "epoch": 157.68795180722893, "grad_norm": 9.577181816101074, "learning_rate": 2.1058252427184465e-05, "loss": 0.2305, "step": 16262 }, { "epoch": 157.69759036144578, "grad_norm": 7.721747875213623, "learning_rate": 2.1053398058252426e-05, "loss": 0.2697, "step": 16263 }, { "epoch": 157.70722891566265, "grad_norm": 20.112520217895508, "learning_rate": 2.1048543689320387e-05, "loss": 0.1877, "step": 16264 }, { "epoch": 157.71686746987953, "grad_norm": 26.213529586791992, "learning_rate": 2.1043689320388352e-05, "loss": 0.2806, "step": 16265 }, { "epoch": 157.72650602409638, "grad_norm": 4.767030715942383, "learning_rate": 2.1038834951456313e-05, "loss": 0.276, "step": 16266 }, { "epoch": 157.73614457831326, "grad_norm": 14.09249496459961, "learning_rate": 2.1033980582524274e-05, "loss": 0.1601, "step": 16267 }, { "epoch": 157.7457831325301, "grad_norm": 14.768869400024414, "learning_rate": 2.1029126213592235e-05, "loss": 0.3909, "step": 16268 }, { "epoch": 157.75542168674698, "grad_norm": 10.93905258178711, "learning_rate": 2.1024271844660196e-05, "loss": 0.1861, "step": 16269 }, { "epoch": 157.76506024096386, "grad_norm": 5.3677802085876465, "learning_rate": 2.1019417475728158e-05, "loss": 0.2159, "step": 16270 }, { "epoch": 157.7746987951807, "grad_norm": 4.491023540496826, "learning_rate": 2.101456310679612e-05, "loss": 0.1513, "step": 16271 }, { "epoch": 157.7843373493976, "grad_norm": 8.044408798217773, "learning_rate": 2.100970873786408e-05, "loss": 0.2813, "step": 16272 }, { "epoch": 157.79397590361447, "grad_norm": 12.96938419342041, "learning_rate": 2.1004854368932038e-05, "loss": 0.2594, "step": 16273 }, { "epoch": 157.80361445783132, "grad_norm": 52.08350372314453, "learning_rate": 2.1e-05, "loss": 0.2793, "step": 16274 }, { "epoch": 157.8132530120482, "grad_norm": 23.665565490722656, "learning_rate": 2.099514563106796e-05, "loss": 0.5146, "step": 16275 }, { "epoch": 157.82289156626507, "grad_norm": 97.94366455078125, "learning_rate": 2.099029126213592e-05, "loss": 0.2712, "step": 16276 }, { "epoch": 157.83253012048192, "grad_norm": 54.99656295776367, "learning_rate": 2.0985436893203886e-05, "loss": 0.1933, "step": 16277 }, { "epoch": 157.8421686746988, "grad_norm": 18.672351837158203, "learning_rate": 2.0980582524271847e-05, "loss": 0.413, "step": 16278 }, { "epoch": 157.85180722891567, "grad_norm": 22.138147354125977, "learning_rate": 2.0975728155339808e-05, "loss": 0.1805, "step": 16279 }, { "epoch": 157.86144578313252, "grad_norm": 15.142902374267578, "learning_rate": 2.097087378640777e-05, "loss": 0.2579, "step": 16280 }, { "epoch": 157.8710843373494, "grad_norm": 53.8527717590332, "learning_rate": 2.096601941747573e-05, "loss": 0.3577, "step": 16281 }, { "epoch": 157.88072289156628, "grad_norm": 18.21816062927246, "learning_rate": 2.096116504854369e-05, "loss": 0.3656, "step": 16282 }, { "epoch": 157.89036144578313, "grad_norm": 6.287253379821777, "learning_rate": 2.0956310679611653e-05, "loss": 0.3851, "step": 16283 }, { "epoch": 157.9, "grad_norm": 29.921781539916992, "learning_rate": 2.0951456310679614e-05, "loss": 0.4754, "step": 16284 }, { "epoch": 157.90963855421685, "grad_norm": 10.06120491027832, "learning_rate": 2.0946601941747575e-05, "loss": 0.1532, "step": 16285 }, { "epoch": 157.91927710843373, "grad_norm": 8.5956449508667, "learning_rate": 2.0941747572815536e-05, "loss": 0.1504, "step": 16286 }, { "epoch": 157.9289156626506, "grad_norm": 11.158105850219727, "learning_rate": 2.0936893203883494e-05, "loss": 0.182, "step": 16287 }, { "epoch": 157.93855421686746, "grad_norm": 7.010778903961182, "learning_rate": 2.0932038834951455e-05, "loss": 0.2414, "step": 16288 }, { "epoch": 157.94819277108434, "grad_norm": 41.01157760620117, "learning_rate": 2.0927184466019416e-05, "loss": 0.4074, "step": 16289 }, { "epoch": 157.9578313253012, "grad_norm": 3.2791571617126465, "learning_rate": 2.092233009708738e-05, "loss": 0.1722, "step": 16290 }, { "epoch": 157.96746987951806, "grad_norm": 9.79003620147705, "learning_rate": 2.0917475728155342e-05, "loss": 0.2803, "step": 16291 }, { "epoch": 157.97710843373494, "grad_norm": 12.96339225769043, "learning_rate": 2.0912621359223303e-05, "loss": 0.2218, "step": 16292 }, { "epoch": 157.98674698795182, "grad_norm": 8.361814498901367, "learning_rate": 2.0907766990291264e-05, "loss": 0.2199, "step": 16293 }, { "epoch": 158.0024096385542, "grad_norm": 5.005619049072266, "learning_rate": 2.0902912621359225e-05, "loss": 0.1308, "step": 16294 }, { "epoch": 158.0120481927711, "grad_norm": 3.7675516605377197, "learning_rate": 2.0898058252427186e-05, "loss": 0.1001, "step": 16295 }, { "epoch": 158.02168674698794, "grad_norm": 16.37602424621582, "learning_rate": 2.0893203883495147e-05, "loss": 0.2958, "step": 16296 }, { "epoch": 158.03132530120482, "grad_norm": 3.0749688148498535, "learning_rate": 2.088834951456311e-05, "loss": 0.1109, "step": 16297 }, { "epoch": 158.0409638554217, "grad_norm": 109.36663055419922, "learning_rate": 2.088349514563107e-05, "loss": 0.1725, "step": 16298 }, { "epoch": 158.05060240963854, "grad_norm": 11.457850456237793, "learning_rate": 2.087864077669903e-05, "loss": 0.1531, "step": 16299 }, { "epoch": 158.06024096385542, "grad_norm": 23.202369689941406, "learning_rate": 2.0873786407766992e-05, "loss": 0.2382, "step": 16300 }, { "epoch": 158.0698795180723, "grad_norm": 27.91718864440918, "learning_rate": 2.086893203883495e-05, "loss": 0.4532, "step": 16301 }, { "epoch": 158.07951807228915, "grad_norm": 9.492712020874023, "learning_rate": 2.086407766990291e-05, "loss": 0.2562, "step": 16302 }, { "epoch": 158.08915662650602, "grad_norm": 7.420114994049072, "learning_rate": 2.0859223300970875e-05, "loss": 0.2366, "step": 16303 }, { "epoch": 158.0987951807229, "grad_norm": 9.358012199401855, "learning_rate": 2.0854368932038837e-05, "loss": 0.2674, "step": 16304 }, { "epoch": 158.10843373493975, "grad_norm": 3.172888994216919, "learning_rate": 2.0849514563106798e-05, "loss": 0.1537, "step": 16305 }, { "epoch": 158.11807228915663, "grad_norm": 11.59347915649414, "learning_rate": 2.084466019417476e-05, "loss": 0.3986, "step": 16306 }, { "epoch": 158.1277108433735, "grad_norm": 4.1788649559021, "learning_rate": 2.083980582524272e-05, "loss": 0.1042, "step": 16307 }, { "epoch": 158.13734939759036, "grad_norm": 7.603248119354248, "learning_rate": 2.083495145631068e-05, "loss": 0.2401, "step": 16308 }, { "epoch": 158.14698795180723, "grad_norm": 26.108728408813477, "learning_rate": 2.0830097087378642e-05, "loss": 0.2324, "step": 16309 }, { "epoch": 158.15662650602408, "grad_norm": 8.020170211791992, "learning_rate": 2.0825242718446603e-05, "loss": 0.232, "step": 16310 }, { "epoch": 158.16626506024096, "grad_norm": 18.211091995239258, "learning_rate": 2.0820388349514565e-05, "loss": 0.1803, "step": 16311 }, { "epoch": 158.17590361445784, "grad_norm": 5.249748229980469, "learning_rate": 2.0815533980582526e-05, "loss": 0.1978, "step": 16312 }, { "epoch": 158.1855421686747, "grad_norm": 4.107503414154053, "learning_rate": 2.0810679611650487e-05, "loss": 0.0873, "step": 16313 }, { "epoch": 158.19518072289156, "grad_norm": 14.20644760131836, "learning_rate": 2.0805825242718445e-05, "loss": 0.1623, "step": 16314 }, { "epoch": 158.20481927710844, "grad_norm": 9.181504249572754, "learning_rate": 2.080097087378641e-05, "loss": 0.2507, "step": 16315 }, { "epoch": 158.2144578313253, "grad_norm": 16.332427978515625, "learning_rate": 2.079611650485437e-05, "loss": 0.2516, "step": 16316 }, { "epoch": 158.22409638554217, "grad_norm": 11.052250862121582, "learning_rate": 2.079126213592233e-05, "loss": 0.2808, "step": 16317 }, { "epoch": 158.23373493975905, "grad_norm": 10.042255401611328, "learning_rate": 2.0786407766990293e-05, "loss": 0.4287, "step": 16318 }, { "epoch": 158.2433734939759, "grad_norm": 27.321529388427734, "learning_rate": 2.0781553398058254e-05, "loss": 0.3001, "step": 16319 }, { "epoch": 158.25301204819277, "grad_norm": 10.470516204833984, "learning_rate": 2.0776699029126215e-05, "loss": 0.2102, "step": 16320 }, { "epoch": 158.26265060240965, "grad_norm": 11.367616653442383, "learning_rate": 2.0771844660194176e-05, "loss": 0.2312, "step": 16321 }, { "epoch": 158.2722891566265, "grad_norm": 24.985694885253906, "learning_rate": 2.0766990291262137e-05, "loss": 0.1668, "step": 16322 }, { "epoch": 158.28192771084338, "grad_norm": 10.707063674926758, "learning_rate": 2.07621359223301e-05, "loss": 0.2644, "step": 16323 }, { "epoch": 158.29156626506025, "grad_norm": 7.6180949211120605, "learning_rate": 2.075728155339806e-05, "loss": 0.1939, "step": 16324 }, { "epoch": 158.3012048192771, "grad_norm": 18.127548217773438, "learning_rate": 2.075242718446602e-05, "loss": 0.2269, "step": 16325 }, { "epoch": 158.31084337349398, "grad_norm": 17.11715316772461, "learning_rate": 2.0747572815533982e-05, "loss": 0.2075, "step": 16326 }, { "epoch": 158.32048192771083, "grad_norm": 11.783583641052246, "learning_rate": 2.0742718446601943e-05, "loss": 0.185, "step": 16327 }, { "epoch": 158.3301204819277, "grad_norm": 5.465297698974609, "learning_rate": 2.0737864077669904e-05, "loss": 0.1428, "step": 16328 }, { "epoch": 158.33975903614459, "grad_norm": 11.4555025100708, "learning_rate": 2.0733009708737865e-05, "loss": 0.2899, "step": 16329 }, { "epoch": 158.34939759036143, "grad_norm": 13.241793632507324, "learning_rate": 2.0728155339805826e-05, "loss": 0.2637, "step": 16330 }, { "epoch": 158.3590361445783, "grad_norm": 14.74860668182373, "learning_rate": 2.0723300970873788e-05, "loss": 0.2795, "step": 16331 }, { "epoch": 158.3686746987952, "grad_norm": 25.699918746948242, "learning_rate": 2.071844660194175e-05, "loss": 0.2111, "step": 16332 }, { "epoch": 158.37831325301204, "grad_norm": 7.133108139038086, "learning_rate": 2.071359223300971e-05, "loss": 0.2236, "step": 16333 }, { "epoch": 158.38795180722892, "grad_norm": 34.61408996582031, "learning_rate": 2.070873786407767e-05, "loss": 0.1726, "step": 16334 }, { "epoch": 158.3975903614458, "grad_norm": 6.647228240966797, "learning_rate": 2.0703883495145632e-05, "loss": 0.295, "step": 16335 }, { "epoch": 158.40722891566264, "grad_norm": 7.598083972930908, "learning_rate": 2.0699029126213593e-05, "loss": 0.2109, "step": 16336 }, { "epoch": 158.41686746987952, "grad_norm": 40.8172492980957, "learning_rate": 2.0694174757281554e-05, "loss": 0.237, "step": 16337 }, { "epoch": 158.4265060240964, "grad_norm": 14.34946346282959, "learning_rate": 2.0689320388349516e-05, "loss": 0.2998, "step": 16338 }, { "epoch": 158.43614457831325, "grad_norm": 5.705816745758057, "learning_rate": 2.0684466019417477e-05, "loss": 0.1144, "step": 16339 }, { "epoch": 158.44578313253012, "grad_norm": 6.817687034606934, "learning_rate": 2.0679611650485438e-05, "loss": 0.1453, "step": 16340 }, { "epoch": 158.455421686747, "grad_norm": 32.326263427734375, "learning_rate": 2.06747572815534e-05, "loss": 0.2642, "step": 16341 }, { "epoch": 158.46506024096385, "grad_norm": 31.377492904663086, "learning_rate": 2.066990291262136e-05, "loss": 0.2732, "step": 16342 }, { "epoch": 158.47469879518073, "grad_norm": 38.22553634643555, "learning_rate": 2.066504854368932e-05, "loss": 0.2125, "step": 16343 }, { "epoch": 158.48433734939758, "grad_norm": 9.233524322509766, "learning_rate": 2.0660194174757282e-05, "loss": 0.1782, "step": 16344 }, { "epoch": 158.49397590361446, "grad_norm": 27.002267837524414, "learning_rate": 2.0655339805825244e-05, "loss": 0.4014, "step": 16345 }, { "epoch": 158.50361445783133, "grad_norm": 10.372014045715332, "learning_rate": 2.0650485436893205e-05, "loss": 0.339, "step": 16346 }, { "epoch": 158.51325301204818, "grad_norm": 3.7595574855804443, "learning_rate": 2.0645631067961166e-05, "loss": 0.2394, "step": 16347 }, { "epoch": 158.52289156626506, "grad_norm": 8.224611282348633, "learning_rate": 2.0640776699029127e-05, "loss": 0.2627, "step": 16348 }, { "epoch": 158.53253012048194, "grad_norm": 18.288267135620117, "learning_rate": 2.0635922330097088e-05, "loss": 0.3197, "step": 16349 }, { "epoch": 158.5421686746988, "grad_norm": 4.547739505767822, "learning_rate": 2.063106796116505e-05, "loss": 0.1481, "step": 16350 }, { "epoch": 158.55180722891566, "grad_norm": 3.030216693878174, "learning_rate": 2.062621359223301e-05, "loss": 0.0959, "step": 16351 }, { "epoch": 158.56144578313254, "grad_norm": 12.769187927246094, "learning_rate": 2.062135922330097e-05, "loss": 0.2511, "step": 16352 }, { "epoch": 158.5710843373494, "grad_norm": 11.279073715209961, "learning_rate": 2.0616504854368933e-05, "loss": 0.3489, "step": 16353 }, { "epoch": 158.58072289156627, "grad_norm": 12.27257251739502, "learning_rate": 2.0611650485436894e-05, "loss": 0.3236, "step": 16354 }, { "epoch": 158.59036144578315, "grad_norm": 7.399475574493408, "learning_rate": 2.0606796116504855e-05, "loss": 0.3451, "step": 16355 }, { "epoch": 158.6, "grad_norm": 16.99320411682129, "learning_rate": 2.0601941747572816e-05, "loss": 0.2055, "step": 16356 }, { "epoch": 158.60963855421687, "grad_norm": 6.704705238342285, "learning_rate": 2.0597087378640777e-05, "loss": 0.1625, "step": 16357 }, { "epoch": 158.61927710843372, "grad_norm": 10.294645309448242, "learning_rate": 2.059223300970874e-05, "loss": 0.2839, "step": 16358 }, { "epoch": 158.6289156626506, "grad_norm": 19.94723129272461, "learning_rate": 2.05873786407767e-05, "loss": 0.2702, "step": 16359 }, { "epoch": 158.63855421686748, "grad_norm": 9.364786148071289, "learning_rate": 2.058252427184466e-05, "loss": 0.3007, "step": 16360 }, { "epoch": 158.64819277108433, "grad_norm": 6.534628391265869, "learning_rate": 2.0577669902912622e-05, "loss": 0.2803, "step": 16361 }, { "epoch": 158.6578313253012, "grad_norm": 11.303772926330566, "learning_rate": 2.0572815533980583e-05, "loss": 0.3692, "step": 16362 }, { "epoch": 158.66746987951808, "grad_norm": 5.016048908233643, "learning_rate": 2.0567961165048544e-05, "loss": 0.1451, "step": 16363 }, { "epoch": 158.67710843373493, "grad_norm": 7.265547275543213, "learning_rate": 2.0563106796116505e-05, "loss": 0.1832, "step": 16364 }, { "epoch": 158.6867469879518, "grad_norm": 8.597602844238281, "learning_rate": 2.0558252427184466e-05, "loss": 0.3221, "step": 16365 }, { "epoch": 158.69638554216868, "grad_norm": 22.533668518066406, "learning_rate": 2.055339805825243e-05, "loss": 0.3739, "step": 16366 }, { "epoch": 158.70602409638553, "grad_norm": 20.888015747070312, "learning_rate": 2.054854368932039e-05, "loss": 0.2509, "step": 16367 }, { "epoch": 158.7156626506024, "grad_norm": 7.631852626800537, "learning_rate": 2.054368932038835e-05, "loss": 0.251, "step": 16368 }, { "epoch": 158.7253012048193, "grad_norm": 6.038566589355469, "learning_rate": 2.053883495145631e-05, "loss": 0.2597, "step": 16369 }, { "epoch": 158.73493975903614, "grad_norm": 18.827960968017578, "learning_rate": 2.0533980582524272e-05, "loss": 0.2118, "step": 16370 }, { "epoch": 158.74457831325302, "grad_norm": 9.257575988769531, "learning_rate": 2.0529126213592233e-05, "loss": 0.2549, "step": 16371 }, { "epoch": 158.7542168674699, "grad_norm": 9.467496871948242, "learning_rate": 2.0524271844660195e-05, "loss": 0.1665, "step": 16372 }, { "epoch": 158.76385542168674, "grad_norm": 9.374820709228516, "learning_rate": 2.0519417475728156e-05, "loss": 0.2217, "step": 16373 }, { "epoch": 158.77349397590362, "grad_norm": 30.579837799072266, "learning_rate": 2.0514563106796117e-05, "loss": 0.2781, "step": 16374 }, { "epoch": 158.78313253012047, "grad_norm": 6.249331474304199, "learning_rate": 2.0509708737864078e-05, "loss": 0.3404, "step": 16375 }, { "epoch": 158.79277108433735, "grad_norm": 11.048259735107422, "learning_rate": 2.050485436893204e-05, "loss": 0.2568, "step": 16376 }, { "epoch": 158.80240963855422, "grad_norm": 28.776290893554688, "learning_rate": 2.05e-05, "loss": 0.2974, "step": 16377 }, { "epoch": 158.81204819277107, "grad_norm": 10.20068359375, "learning_rate": 2.0495145631067965e-05, "loss": 0.2508, "step": 16378 }, { "epoch": 158.82168674698795, "grad_norm": 7.389577865600586, "learning_rate": 2.0490291262135926e-05, "loss": 0.1586, "step": 16379 }, { "epoch": 158.83132530120483, "grad_norm": 33.05095291137695, "learning_rate": 2.0485436893203887e-05, "loss": 0.2635, "step": 16380 }, { "epoch": 158.84096385542168, "grad_norm": 6.509715557098389, "learning_rate": 2.0480582524271845e-05, "loss": 0.2978, "step": 16381 }, { "epoch": 158.85060240963855, "grad_norm": 7.589731693267822, "learning_rate": 2.0475728155339806e-05, "loss": 0.3764, "step": 16382 }, { "epoch": 158.86024096385543, "grad_norm": 12.120126724243164, "learning_rate": 2.0470873786407767e-05, "loss": 0.2574, "step": 16383 }, { "epoch": 158.86987951807228, "grad_norm": 13.35584545135498, "learning_rate": 2.0466019417475728e-05, "loss": 0.2073, "step": 16384 }, { "epoch": 158.87951807228916, "grad_norm": 5.87522029876709, "learning_rate": 2.046116504854369e-05, "loss": 0.1655, "step": 16385 }, { "epoch": 158.88915662650604, "grad_norm": 8.895035743713379, "learning_rate": 2.045631067961165e-05, "loss": 0.0842, "step": 16386 }, { "epoch": 158.89879518072289, "grad_norm": 12.310179710388184, "learning_rate": 2.0451456310679612e-05, "loss": 0.3671, "step": 16387 }, { "epoch": 158.90843373493976, "grad_norm": 12.77810001373291, "learning_rate": 2.0446601941747573e-05, "loss": 0.2759, "step": 16388 }, { "epoch": 158.91807228915664, "grad_norm": 7.2198405265808105, "learning_rate": 2.0441747572815534e-05, "loss": 0.2745, "step": 16389 }, { "epoch": 158.9277108433735, "grad_norm": 13.553335189819336, "learning_rate": 2.0436893203883495e-05, "loss": 0.4531, "step": 16390 }, { "epoch": 158.93734939759037, "grad_norm": 10.386476516723633, "learning_rate": 2.043203883495146e-05, "loss": 0.2899, "step": 16391 }, { "epoch": 158.94698795180722, "grad_norm": 14.176239967346191, "learning_rate": 2.042718446601942e-05, "loss": 0.1913, "step": 16392 }, { "epoch": 158.9566265060241, "grad_norm": 9.639891624450684, "learning_rate": 2.0422330097087382e-05, "loss": 0.0749, "step": 16393 }, { "epoch": 158.96626506024097, "grad_norm": 9.090744972229004, "learning_rate": 2.041747572815534e-05, "loss": 0.1264, "step": 16394 }, { "epoch": 158.97590361445782, "grad_norm": 9.008540153503418, "learning_rate": 2.04126213592233e-05, "loss": 0.2427, "step": 16395 }, { "epoch": 158.9855421686747, "grad_norm": 18.025171279907227, "learning_rate": 2.0407766990291262e-05, "loss": 0.2758, "step": 16396 }, { "epoch": 159.00120481927712, "grad_norm": 8.096168518066406, "learning_rate": 2.0402912621359223e-05, "loss": 0.1835, "step": 16397 }, { "epoch": 159.01084337349397, "grad_norm": 15.109601974487305, "learning_rate": 2.0398058252427184e-05, "loss": 0.1615, "step": 16398 }, { "epoch": 159.02048192771085, "grad_norm": 12.075766563415527, "learning_rate": 2.0393203883495145e-05, "loss": 0.347, "step": 16399 }, { "epoch": 159.03012048192772, "grad_norm": 10.673365592956543, "learning_rate": 2.0388349514563107e-05, "loss": 0.2718, "step": 16400 }, { "epoch": 159.03975903614457, "grad_norm": 10.727974891662598, "learning_rate": 2.0383495145631068e-05, "loss": 0.2294, "step": 16401 }, { "epoch": 159.04939759036145, "grad_norm": 31.352569580078125, "learning_rate": 2.037864077669903e-05, "loss": 0.2975, "step": 16402 }, { "epoch": 159.0590361445783, "grad_norm": 11.056821823120117, "learning_rate": 2.037378640776699e-05, "loss": 0.2721, "step": 16403 }, { "epoch": 159.06867469879518, "grad_norm": 15.918426513671875, "learning_rate": 2.0368932038834955e-05, "loss": 0.5555, "step": 16404 }, { "epoch": 159.07831325301206, "grad_norm": 20.169130325317383, "learning_rate": 2.0364077669902916e-05, "loss": 0.1867, "step": 16405 }, { "epoch": 159.0879518072289, "grad_norm": 8.85303020477295, "learning_rate": 2.0359223300970877e-05, "loss": 0.2385, "step": 16406 }, { "epoch": 159.09759036144578, "grad_norm": 5.115702152252197, "learning_rate": 2.0354368932038838e-05, "loss": 0.1106, "step": 16407 }, { "epoch": 159.10722891566266, "grad_norm": 10.72641658782959, "learning_rate": 2.0349514563106796e-05, "loss": 0.1601, "step": 16408 }, { "epoch": 159.1168674698795, "grad_norm": 10.790843963623047, "learning_rate": 2.0344660194174757e-05, "loss": 0.1673, "step": 16409 }, { "epoch": 159.1265060240964, "grad_norm": 10.757442474365234, "learning_rate": 2.0339805825242718e-05, "loss": 0.2432, "step": 16410 }, { "epoch": 159.13614457831326, "grad_norm": 10.407312393188477, "learning_rate": 2.033495145631068e-05, "loss": 0.2406, "step": 16411 }, { "epoch": 159.1457831325301, "grad_norm": 4.952287197113037, "learning_rate": 2.033009708737864e-05, "loss": 0.1846, "step": 16412 }, { "epoch": 159.155421686747, "grad_norm": 5.6442131996154785, "learning_rate": 2.03252427184466e-05, "loss": 0.0844, "step": 16413 }, { "epoch": 159.16506024096387, "grad_norm": 8.503198623657227, "learning_rate": 2.0320388349514563e-05, "loss": 0.2475, "step": 16414 }, { "epoch": 159.17469879518072, "grad_norm": 6.418105602264404, "learning_rate": 2.0315533980582524e-05, "loss": 0.1762, "step": 16415 }, { "epoch": 159.1843373493976, "grad_norm": 12.418229103088379, "learning_rate": 2.031067961165049e-05, "loss": 0.3134, "step": 16416 }, { "epoch": 159.19397590361444, "grad_norm": 9.598873138427734, "learning_rate": 2.030582524271845e-05, "loss": 0.3284, "step": 16417 }, { "epoch": 159.20361445783132, "grad_norm": 7.916131019592285, "learning_rate": 2.030097087378641e-05, "loss": 0.258, "step": 16418 }, { "epoch": 159.2132530120482, "grad_norm": 8.55912971496582, "learning_rate": 2.0296116504854372e-05, "loss": 0.1753, "step": 16419 }, { "epoch": 159.22289156626505, "grad_norm": 14.304137229919434, "learning_rate": 2.0291262135922333e-05, "loss": 0.3545, "step": 16420 }, { "epoch": 159.23253012048193, "grad_norm": 13.465691566467285, "learning_rate": 2.0286407766990294e-05, "loss": 0.4511, "step": 16421 }, { "epoch": 159.2421686746988, "grad_norm": 13.882719993591309, "learning_rate": 2.0281553398058252e-05, "loss": 0.3352, "step": 16422 }, { "epoch": 159.25180722891565, "grad_norm": 11.280506134033203, "learning_rate": 2.0276699029126213e-05, "loss": 0.3082, "step": 16423 }, { "epoch": 159.26144578313253, "grad_norm": 6.764254093170166, "learning_rate": 2.0271844660194174e-05, "loss": 0.3121, "step": 16424 }, { "epoch": 159.2710843373494, "grad_norm": 29.424842834472656, "learning_rate": 2.0266990291262135e-05, "loss": 0.1618, "step": 16425 }, { "epoch": 159.28072289156626, "grad_norm": 29.065868377685547, "learning_rate": 2.0262135922330096e-05, "loss": 0.3345, "step": 16426 }, { "epoch": 159.29036144578313, "grad_norm": 13.359341621398926, "learning_rate": 2.0257281553398058e-05, "loss": 0.4086, "step": 16427 }, { "epoch": 159.3, "grad_norm": 6.2075114250183105, "learning_rate": 2.025242718446602e-05, "loss": 0.4761, "step": 16428 }, { "epoch": 159.30963855421686, "grad_norm": 10.3302001953125, "learning_rate": 2.0247572815533983e-05, "loss": 0.2946, "step": 16429 }, { "epoch": 159.31927710843374, "grad_norm": 9.492302894592285, "learning_rate": 2.0242718446601944e-05, "loss": 0.11, "step": 16430 }, { "epoch": 159.32891566265062, "grad_norm": 12.629077911376953, "learning_rate": 2.0237864077669905e-05, "loss": 0.3434, "step": 16431 }, { "epoch": 159.33855421686746, "grad_norm": 10.939889907836914, "learning_rate": 2.0233009708737867e-05, "loss": 0.2627, "step": 16432 }, { "epoch": 159.34819277108434, "grad_norm": 6.454556941986084, "learning_rate": 2.0228155339805828e-05, "loss": 0.31, "step": 16433 }, { "epoch": 159.3578313253012, "grad_norm": 10.32909870147705, "learning_rate": 2.022330097087379e-05, "loss": 0.2308, "step": 16434 }, { "epoch": 159.36746987951807, "grad_norm": 4.700069427490234, "learning_rate": 2.021844660194175e-05, "loss": 0.1665, "step": 16435 }, { "epoch": 159.37710843373495, "grad_norm": 22.69224739074707, "learning_rate": 2.0213592233009708e-05, "loss": 0.4466, "step": 16436 }, { "epoch": 159.3867469879518, "grad_norm": 7.027539253234863, "learning_rate": 2.020873786407767e-05, "loss": 0.2332, "step": 16437 }, { "epoch": 159.39638554216867, "grad_norm": 7.5946526527404785, "learning_rate": 2.020388349514563e-05, "loss": 0.2682, "step": 16438 }, { "epoch": 159.40602409638555, "grad_norm": 11.771535873413086, "learning_rate": 2.019902912621359e-05, "loss": 0.2965, "step": 16439 }, { "epoch": 159.4156626506024, "grad_norm": 5.861014366149902, "learning_rate": 2.0194174757281552e-05, "loss": 0.3425, "step": 16440 }, { "epoch": 159.42530120481928, "grad_norm": 14.65344524383545, "learning_rate": 2.0189320388349517e-05, "loss": 0.5043, "step": 16441 }, { "epoch": 159.43493975903615, "grad_norm": 7.482609272003174, "learning_rate": 2.0184466019417478e-05, "loss": 0.2334, "step": 16442 }, { "epoch": 159.444578313253, "grad_norm": 8.114073753356934, "learning_rate": 2.017961165048544e-05, "loss": 0.4294, "step": 16443 }, { "epoch": 159.45421686746988, "grad_norm": 4.0882487297058105, "learning_rate": 2.01747572815534e-05, "loss": 0.1806, "step": 16444 }, { "epoch": 159.46385542168676, "grad_norm": 35.58586502075195, "learning_rate": 2.016990291262136e-05, "loss": 0.1847, "step": 16445 }, { "epoch": 159.4734939759036, "grad_norm": 15.138330459594727, "learning_rate": 2.0165048543689323e-05, "loss": 0.3303, "step": 16446 }, { "epoch": 159.48313253012049, "grad_norm": 8.346988677978516, "learning_rate": 2.0160194174757284e-05, "loss": 0.3331, "step": 16447 }, { "epoch": 159.49277108433736, "grad_norm": 15.9085693359375, "learning_rate": 2.0155339805825245e-05, "loss": 0.2096, "step": 16448 }, { "epoch": 159.5024096385542, "grad_norm": 7.307018756866455, "learning_rate": 2.0150485436893206e-05, "loss": 0.3041, "step": 16449 }, { "epoch": 159.5120481927711, "grad_norm": 7.014260768890381, "learning_rate": 2.0145631067961164e-05, "loss": 0.198, "step": 16450 }, { "epoch": 159.52168674698794, "grad_norm": 18.873403549194336, "learning_rate": 2.0140776699029125e-05, "loss": 0.2343, "step": 16451 }, { "epoch": 159.53132530120482, "grad_norm": 13.799026489257812, "learning_rate": 2.0135922330097086e-05, "loss": 0.1971, "step": 16452 }, { "epoch": 159.5409638554217, "grad_norm": 14.602514266967773, "learning_rate": 2.0131067961165047e-05, "loss": 0.149, "step": 16453 }, { "epoch": 159.55060240963854, "grad_norm": 6.902522087097168, "learning_rate": 2.0126213592233012e-05, "loss": 0.1616, "step": 16454 }, { "epoch": 159.56024096385542, "grad_norm": 33.4901008605957, "learning_rate": 2.0121359223300973e-05, "loss": 0.2705, "step": 16455 }, { "epoch": 159.5698795180723, "grad_norm": 15.313247680664062, "learning_rate": 2.0116504854368934e-05, "loss": 0.3265, "step": 16456 }, { "epoch": 159.57951807228915, "grad_norm": 5.26675271987915, "learning_rate": 2.0111650485436895e-05, "loss": 0.1866, "step": 16457 }, { "epoch": 159.58915662650602, "grad_norm": 13.8962984085083, "learning_rate": 2.0106796116504856e-05, "loss": 0.2064, "step": 16458 }, { "epoch": 159.5987951807229, "grad_norm": 9.630101203918457, "learning_rate": 2.0101941747572818e-05, "loss": 0.1941, "step": 16459 }, { "epoch": 159.60843373493975, "grad_norm": 10.367568016052246, "learning_rate": 2.009708737864078e-05, "loss": 0.4458, "step": 16460 }, { "epoch": 159.61807228915663, "grad_norm": 8.516068458557129, "learning_rate": 2.009223300970874e-05, "loss": 0.3031, "step": 16461 }, { "epoch": 159.6277108433735, "grad_norm": 7.1857757568359375, "learning_rate": 2.00873786407767e-05, "loss": 0.2246, "step": 16462 }, { "epoch": 159.63734939759036, "grad_norm": 12.419366836547852, "learning_rate": 2.008252427184466e-05, "loss": 0.2804, "step": 16463 }, { "epoch": 159.64698795180723, "grad_norm": 24.88658905029297, "learning_rate": 2.007766990291262e-05, "loss": 0.305, "step": 16464 }, { "epoch": 159.65662650602408, "grad_norm": 13.847405433654785, "learning_rate": 2.007281553398058e-05, "loss": 0.2842, "step": 16465 }, { "epoch": 159.66626506024096, "grad_norm": 7.522781848907471, "learning_rate": 2.0067961165048542e-05, "loss": 0.1937, "step": 16466 }, { "epoch": 159.67590361445784, "grad_norm": 14.993497848510742, "learning_rate": 2.0063106796116507e-05, "loss": 0.2146, "step": 16467 }, { "epoch": 159.6855421686747, "grad_norm": 3.3734567165374756, "learning_rate": 2.0058252427184468e-05, "loss": 0.078, "step": 16468 }, { "epoch": 159.69518072289156, "grad_norm": 24.647422790527344, "learning_rate": 2.005339805825243e-05, "loss": 0.2367, "step": 16469 }, { "epoch": 159.70481927710844, "grad_norm": 5.822171688079834, "learning_rate": 2.004854368932039e-05, "loss": 0.1749, "step": 16470 }, { "epoch": 159.7144578313253, "grad_norm": 11.883954048156738, "learning_rate": 2.004368932038835e-05, "loss": 0.3364, "step": 16471 }, { "epoch": 159.72409638554217, "grad_norm": 11.705580711364746, "learning_rate": 2.0038834951456312e-05, "loss": 0.204, "step": 16472 }, { "epoch": 159.73373493975905, "grad_norm": 7.956658840179443, "learning_rate": 2.0033980582524274e-05, "loss": 0.1243, "step": 16473 }, { "epoch": 159.7433734939759, "grad_norm": 4.493961334228516, "learning_rate": 2.0029126213592235e-05, "loss": 0.1119, "step": 16474 }, { "epoch": 159.75301204819277, "grad_norm": 9.124812126159668, "learning_rate": 2.0024271844660196e-05, "loss": 0.4661, "step": 16475 }, { "epoch": 159.76265060240965, "grad_norm": 30.7790584564209, "learning_rate": 2.0019417475728157e-05, "loss": 0.4125, "step": 16476 }, { "epoch": 159.7722891566265, "grad_norm": 15.01382827758789, "learning_rate": 2.0014563106796115e-05, "loss": 0.2568, "step": 16477 }, { "epoch": 159.78192771084338, "grad_norm": 15.505809783935547, "learning_rate": 2.0009708737864076e-05, "loss": 0.3775, "step": 16478 }, { "epoch": 159.79156626506025, "grad_norm": 8.20367431640625, "learning_rate": 2.000485436893204e-05, "loss": 0.2841, "step": 16479 }, { "epoch": 159.8012048192771, "grad_norm": 23.521642684936523, "learning_rate": 2e-05, "loss": 0.4092, "step": 16480 }, { "epoch": 159.81084337349398, "grad_norm": 6.650183200836182, "learning_rate": 1.9995145631067963e-05, "loss": 0.1196, "step": 16481 }, { "epoch": 159.82048192771083, "grad_norm": 15.885872840881348, "learning_rate": 1.9990291262135924e-05, "loss": 0.3823, "step": 16482 }, { "epoch": 159.8301204819277, "grad_norm": 8.89291000366211, "learning_rate": 1.9985436893203885e-05, "loss": 0.2671, "step": 16483 }, { "epoch": 159.83975903614459, "grad_norm": 10.160270690917969, "learning_rate": 1.9980582524271846e-05, "loss": 0.2699, "step": 16484 }, { "epoch": 159.84939759036143, "grad_norm": 30.621809005737305, "learning_rate": 1.9975728155339807e-05, "loss": 0.299, "step": 16485 }, { "epoch": 159.8590361445783, "grad_norm": 18.870119094848633, "learning_rate": 1.997087378640777e-05, "loss": 0.2878, "step": 16486 }, { "epoch": 159.8686746987952, "grad_norm": 20.684358596801758, "learning_rate": 1.996601941747573e-05, "loss": 0.4203, "step": 16487 }, { "epoch": 159.87831325301204, "grad_norm": 4.729025363922119, "learning_rate": 1.996116504854369e-05, "loss": 0.0713, "step": 16488 }, { "epoch": 159.88795180722892, "grad_norm": 14.676139831542969, "learning_rate": 1.9956310679611652e-05, "loss": 0.3589, "step": 16489 }, { "epoch": 159.8975903614458, "grad_norm": 9.479826927185059, "learning_rate": 1.9951456310679613e-05, "loss": 0.2729, "step": 16490 }, { "epoch": 159.90722891566264, "grad_norm": 9.832087516784668, "learning_rate": 1.994660194174757e-05, "loss": 0.1259, "step": 16491 }, { "epoch": 159.91686746987952, "grad_norm": 9.536602020263672, "learning_rate": 1.9941747572815535e-05, "loss": 0.2705, "step": 16492 }, { "epoch": 159.9265060240964, "grad_norm": 7.5639328956604, "learning_rate": 1.9936893203883497e-05, "loss": 0.1217, "step": 16493 }, { "epoch": 159.93614457831325, "grad_norm": 5.85150146484375, "learning_rate": 1.9932038834951458e-05, "loss": 0.226, "step": 16494 }, { "epoch": 159.94578313253012, "grad_norm": 12.037491798400879, "learning_rate": 1.992718446601942e-05, "loss": 0.1217, "step": 16495 }, { "epoch": 159.955421686747, "grad_norm": 9.146646499633789, "learning_rate": 1.992233009708738e-05, "loss": 0.3738, "step": 16496 }, { "epoch": 159.96506024096385, "grad_norm": 51.72085189819336, "learning_rate": 1.991747572815534e-05, "loss": 0.3329, "step": 16497 }, { "epoch": 159.97469879518073, "grad_norm": 11.81376838684082, "learning_rate": 1.9912621359223302e-05, "loss": 0.1798, "step": 16498 }, { "epoch": 159.98433734939758, "grad_norm": 9.432657241821289, "learning_rate": 1.9907766990291263e-05, "loss": 0.2163, "step": 16499 }, { "epoch": 159.99397590361446, "grad_norm": 11.926761627197266, "learning_rate": 1.9902912621359225e-05, "loss": 0.2654, "step": 16500 }, { "epoch": 160.00963855421688, "grad_norm": 7.443201065063477, "learning_rate": 1.9898058252427186e-05, "loss": 0.3158, "step": 16501 }, { "epoch": 160.01927710843373, "grad_norm": 17.15595817565918, "learning_rate": 1.9893203883495147e-05, "loss": 0.1339, "step": 16502 }, { "epoch": 160.0289156626506, "grad_norm": 7.1501359939575195, "learning_rate": 1.9888349514563108e-05, "loss": 0.2734, "step": 16503 }, { "epoch": 160.03855421686748, "grad_norm": 16.799589157104492, "learning_rate": 1.988349514563107e-05, "loss": 0.2427, "step": 16504 }, { "epoch": 160.04819277108433, "grad_norm": 14.4208402633667, "learning_rate": 1.987864077669903e-05, "loss": 0.2693, "step": 16505 }, { "epoch": 160.0578313253012, "grad_norm": 11.37161922454834, "learning_rate": 1.987378640776699e-05, "loss": 0.3101, "step": 16506 }, { "epoch": 160.0674698795181, "grad_norm": 14.85262393951416, "learning_rate": 1.9868932038834953e-05, "loss": 0.333, "step": 16507 }, { "epoch": 160.07710843373494, "grad_norm": 9.28533935546875, "learning_rate": 1.9864077669902914e-05, "loss": 0.2963, "step": 16508 }, { "epoch": 160.0867469879518, "grad_norm": 15.21810531616211, "learning_rate": 1.9859223300970875e-05, "loss": 0.3296, "step": 16509 }, { "epoch": 160.09638554216866, "grad_norm": 15.889763832092285, "learning_rate": 1.9854368932038836e-05, "loss": 0.3327, "step": 16510 }, { "epoch": 160.10602409638554, "grad_norm": 11.858291625976562, "learning_rate": 1.9849514563106797e-05, "loss": 0.2836, "step": 16511 }, { "epoch": 160.11566265060242, "grad_norm": 4.835329055786133, "learning_rate": 1.9844660194174758e-05, "loss": 0.0736, "step": 16512 }, { "epoch": 160.12530120481927, "grad_norm": 5.510050296783447, "learning_rate": 1.983980582524272e-05, "loss": 0.1507, "step": 16513 }, { "epoch": 160.13493975903614, "grad_norm": 16.483720779418945, "learning_rate": 1.983495145631068e-05, "loss": 0.3007, "step": 16514 }, { "epoch": 160.14457831325302, "grad_norm": 17.258106231689453, "learning_rate": 1.9830097087378642e-05, "loss": 0.4079, "step": 16515 }, { "epoch": 160.15421686746987, "grad_norm": 15.19589614868164, "learning_rate": 1.9825242718446603e-05, "loss": 0.2322, "step": 16516 }, { "epoch": 160.16385542168675, "grad_norm": 17.715282440185547, "learning_rate": 1.9820388349514564e-05, "loss": 0.463, "step": 16517 }, { "epoch": 160.17349397590363, "grad_norm": 7.861403942108154, "learning_rate": 1.9815533980582525e-05, "loss": 0.204, "step": 16518 }, { "epoch": 160.18313253012047, "grad_norm": 19.442968368530273, "learning_rate": 1.9810679611650486e-05, "loss": 0.2834, "step": 16519 }, { "epoch": 160.19277108433735, "grad_norm": 11.579737663269043, "learning_rate": 1.9805825242718447e-05, "loss": 0.2802, "step": 16520 }, { "epoch": 160.20240963855423, "grad_norm": 11.314361572265625, "learning_rate": 1.980097087378641e-05, "loss": 0.2596, "step": 16521 }, { "epoch": 160.21204819277108, "grad_norm": 7.658018589019775, "learning_rate": 1.979611650485437e-05, "loss": 0.1631, "step": 16522 }, { "epoch": 160.22168674698796, "grad_norm": 20.816598892211914, "learning_rate": 1.979126213592233e-05, "loss": 0.3998, "step": 16523 }, { "epoch": 160.2313253012048, "grad_norm": 7.189706802368164, "learning_rate": 1.9786407766990292e-05, "loss": 0.1818, "step": 16524 }, { "epoch": 160.24096385542168, "grad_norm": 9.193593978881836, "learning_rate": 1.9781553398058253e-05, "loss": 0.3556, "step": 16525 }, { "epoch": 160.25060240963856, "grad_norm": 11.801203727722168, "learning_rate": 1.9776699029126214e-05, "loss": 0.2078, "step": 16526 }, { "epoch": 160.2602409638554, "grad_norm": 4.38261079788208, "learning_rate": 1.9771844660194175e-05, "loss": 0.0904, "step": 16527 }, { "epoch": 160.2698795180723, "grad_norm": 20.25022315979004, "learning_rate": 1.9766990291262137e-05, "loss": 0.2259, "step": 16528 }, { "epoch": 160.27951807228916, "grad_norm": 11.488933563232422, "learning_rate": 1.9762135922330098e-05, "loss": 0.2868, "step": 16529 }, { "epoch": 160.289156626506, "grad_norm": 7.7390666007995605, "learning_rate": 1.975728155339806e-05, "loss": 0.2452, "step": 16530 }, { "epoch": 160.2987951807229, "grad_norm": 8.065471649169922, "learning_rate": 1.975242718446602e-05, "loss": 0.1901, "step": 16531 }, { "epoch": 160.30843373493977, "grad_norm": 6.7865424156188965, "learning_rate": 1.974757281553398e-05, "loss": 0.1999, "step": 16532 }, { "epoch": 160.31807228915662, "grad_norm": 5.373866558074951, "learning_rate": 1.9742718446601942e-05, "loss": 0.11, "step": 16533 }, { "epoch": 160.3277108433735, "grad_norm": 15.408490180969238, "learning_rate": 1.9737864077669904e-05, "loss": 0.1957, "step": 16534 }, { "epoch": 160.33734939759037, "grad_norm": 14.270885467529297, "learning_rate": 1.9733009708737865e-05, "loss": 0.5266, "step": 16535 }, { "epoch": 160.34698795180722, "grad_norm": 9.858158111572266, "learning_rate": 1.9728155339805826e-05, "loss": 0.4577, "step": 16536 }, { "epoch": 160.3566265060241, "grad_norm": 4.467617034912109, "learning_rate": 1.9723300970873787e-05, "loss": 0.0882, "step": 16537 }, { "epoch": 160.36626506024098, "grad_norm": 9.29275894165039, "learning_rate": 1.9718446601941748e-05, "loss": 0.2933, "step": 16538 }, { "epoch": 160.37590361445783, "grad_norm": 12.973647117614746, "learning_rate": 1.971359223300971e-05, "loss": 0.2405, "step": 16539 }, { "epoch": 160.3855421686747, "grad_norm": 4.241943836212158, "learning_rate": 1.970873786407767e-05, "loss": 0.1964, "step": 16540 }, { "epoch": 160.39518072289155, "grad_norm": 9.489755630493164, "learning_rate": 1.970388349514563e-05, "loss": 0.2688, "step": 16541 }, { "epoch": 160.40481927710843, "grad_norm": 6.153846263885498, "learning_rate": 1.9699029126213596e-05, "loss": 0.1145, "step": 16542 }, { "epoch": 160.4144578313253, "grad_norm": 9.4947509765625, "learning_rate": 1.9694174757281554e-05, "loss": 0.3453, "step": 16543 }, { "epoch": 160.42409638554216, "grad_norm": 9.65468692779541, "learning_rate": 1.9689320388349515e-05, "loss": 0.1699, "step": 16544 }, { "epoch": 160.43373493975903, "grad_norm": 7.707568168640137, "learning_rate": 1.9684466019417476e-05, "loss": 0.167, "step": 16545 }, { "epoch": 160.4433734939759, "grad_norm": 4.595914363861084, "learning_rate": 1.9679611650485437e-05, "loss": 0.1842, "step": 16546 }, { "epoch": 160.45301204819276, "grad_norm": 18.181184768676758, "learning_rate": 1.96747572815534e-05, "loss": 0.1708, "step": 16547 }, { "epoch": 160.46265060240964, "grad_norm": 9.087239265441895, "learning_rate": 1.966990291262136e-05, "loss": 0.1815, "step": 16548 }, { "epoch": 160.47228915662652, "grad_norm": 10.62000846862793, "learning_rate": 1.966504854368932e-05, "loss": 0.3295, "step": 16549 }, { "epoch": 160.48192771084337, "grad_norm": 6.833110332489014, "learning_rate": 1.9660194174757282e-05, "loss": 0.1677, "step": 16550 }, { "epoch": 160.49156626506024, "grad_norm": 8.514235496520996, "learning_rate": 1.9655339805825243e-05, "loss": 0.3255, "step": 16551 }, { "epoch": 160.50120481927712, "grad_norm": 7.49924373626709, "learning_rate": 1.9650485436893204e-05, "loss": 0.3654, "step": 16552 }, { "epoch": 160.51084337349397, "grad_norm": 7.102427005767822, "learning_rate": 1.9645631067961165e-05, "loss": 0.254, "step": 16553 }, { "epoch": 160.52048192771085, "grad_norm": 11.92078685760498, "learning_rate": 1.9640776699029126e-05, "loss": 0.2996, "step": 16554 }, { "epoch": 160.53012048192772, "grad_norm": 6.094902515411377, "learning_rate": 1.963592233009709e-05, "loss": 0.1789, "step": 16555 }, { "epoch": 160.53975903614457, "grad_norm": 10.923195838928223, "learning_rate": 1.9631067961165052e-05, "loss": 0.3539, "step": 16556 }, { "epoch": 160.54939759036145, "grad_norm": 7.064004421234131, "learning_rate": 1.962621359223301e-05, "loss": 0.2089, "step": 16557 }, { "epoch": 160.5590361445783, "grad_norm": 46.598026275634766, "learning_rate": 1.962135922330097e-05, "loss": 0.3043, "step": 16558 }, { "epoch": 160.56867469879518, "grad_norm": 10.743476867675781, "learning_rate": 1.9616504854368932e-05, "loss": 0.2549, "step": 16559 }, { "epoch": 160.57831325301206, "grad_norm": 9.126954078674316, "learning_rate": 1.9611650485436893e-05, "loss": 0.1985, "step": 16560 }, { "epoch": 160.5879518072289, "grad_norm": 5.750671863555908, "learning_rate": 1.9606796116504854e-05, "loss": 0.2841, "step": 16561 }, { "epoch": 160.59759036144578, "grad_norm": 7.988302707672119, "learning_rate": 1.9601941747572816e-05, "loss": 0.2746, "step": 16562 }, { "epoch": 160.60722891566266, "grad_norm": 9.449671745300293, "learning_rate": 1.9597087378640777e-05, "loss": 0.2013, "step": 16563 }, { "epoch": 160.6168674698795, "grad_norm": 5.2176194190979, "learning_rate": 1.9592233009708738e-05, "loss": 0.0871, "step": 16564 }, { "epoch": 160.6265060240964, "grad_norm": 11.633851051330566, "learning_rate": 1.95873786407767e-05, "loss": 0.2452, "step": 16565 }, { "epoch": 160.63614457831326, "grad_norm": 6.930350303649902, "learning_rate": 1.958252427184466e-05, "loss": 0.2567, "step": 16566 }, { "epoch": 160.6457831325301, "grad_norm": 8.598071098327637, "learning_rate": 1.957766990291262e-05, "loss": 0.3855, "step": 16567 }, { "epoch": 160.655421686747, "grad_norm": 21.00942611694336, "learning_rate": 1.9572815533980586e-05, "loss": 0.2401, "step": 16568 }, { "epoch": 160.66506024096387, "grad_norm": 6.5068511962890625, "learning_rate": 1.9567961165048547e-05, "loss": 0.2483, "step": 16569 }, { "epoch": 160.67469879518072, "grad_norm": 20.352191925048828, "learning_rate": 1.9563106796116508e-05, "loss": 0.348, "step": 16570 }, { "epoch": 160.6843373493976, "grad_norm": 5.533830165863037, "learning_rate": 1.9558252427184466e-05, "loss": 0.138, "step": 16571 }, { "epoch": 160.69397590361444, "grad_norm": 8.266706466674805, "learning_rate": 1.9553398058252427e-05, "loss": 0.2882, "step": 16572 }, { "epoch": 160.70361445783132, "grad_norm": 10.31589126586914, "learning_rate": 1.9548543689320388e-05, "loss": 0.3862, "step": 16573 }, { "epoch": 160.7132530120482, "grad_norm": 12.976422309875488, "learning_rate": 1.954368932038835e-05, "loss": 0.3331, "step": 16574 }, { "epoch": 160.72289156626505, "grad_norm": 5.2947187423706055, "learning_rate": 1.953883495145631e-05, "loss": 0.1043, "step": 16575 }, { "epoch": 160.73253012048193, "grad_norm": 7.992588520050049, "learning_rate": 1.953398058252427e-05, "loss": 0.1597, "step": 16576 }, { "epoch": 160.7421686746988, "grad_norm": 12.951248168945312, "learning_rate": 1.9529126213592233e-05, "loss": 0.1957, "step": 16577 }, { "epoch": 160.75180722891565, "grad_norm": 7.083374500274658, "learning_rate": 1.9524271844660194e-05, "loss": 0.1932, "step": 16578 }, { "epoch": 160.76144578313253, "grad_norm": 16.488040924072266, "learning_rate": 1.9519417475728155e-05, "loss": 0.4418, "step": 16579 }, { "epoch": 160.7710843373494, "grad_norm": 7.077486515045166, "learning_rate": 1.951456310679612e-05, "loss": 0.1301, "step": 16580 }, { "epoch": 160.78072289156626, "grad_norm": 7.396275520324707, "learning_rate": 1.950970873786408e-05, "loss": 0.2651, "step": 16581 }, { "epoch": 160.79036144578313, "grad_norm": 5.706928730010986, "learning_rate": 1.9504854368932042e-05, "loss": 0.1918, "step": 16582 }, { "epoch": 160.8, "grad_norm": 27.441680908203125, "learning_rate": 1.9500000000000003e-05, "loss": 0.4155, "step": 16583 }, { "epoch": 160.80963855421686, "grad_norm": 8.516054153442383, "learning_rate": 1.9495145631067964e-05, "loss": 0.1419, "step": 16584 }, { "epoch": 160.81927710843374, "grad_norm": 8.30373477935791, "learning_rate": 1.9490291262135922e-05, "loss": 0.1698, "step": 16585 }, { "epoch": 160.82891566265062, "grad_norm": 18.914405822753906, "learning_rate": 1.9485436893203883e-05, "loss": 0.093, "step": 16586 }, { "epoch": 160.83855421686746, "grad_norm": 5.347169876098633, "learning_rate": 1.9480582524271844e-05, "loss": 0.2664, "step": 16587 }, { "epoch": 160.84819277108434, "grad_norm": 6.6147894859313965, "learning_rate": 1.9475728155339805e-05, "loss": 0.2095, "step": 16588 }, { "epoch": 160.8578313253012, "grad_norm": 8.223942756652832, "learning_rate": 1.9470873786407767e-05, "loss": 0.2799, "step": 16589 }, { "epoch": 160.86746987951807, "grad_norm": 10.297938346862793, "learning_rate": 1.9466019417475728e-05, "loss": 0.1933, "step": 16590 }, { "epoch": 160.87710843373495, "grad_norm": 11.30601692199707, "learning_rate": 1.946116504854369e-05, "loss": 0.1363, "step": 16591 }, { "epoch": 160.8867469879518, "grad_norm": 7.573680877685547, "learning_rate": 1.945631067961165e-05, "loss": 0.3522, "step": 16592 }, { "epoch": 160.89638554216867, "grad_norm": 7.514276504516602, "learning_rate": 1.9451456310679614e-05, "loss": 0.1383, "step": 16593 }, { "epoch": 160.90602409638555, "grad_norm": 20.838459014892578, "learning_rate": 1.9446601941747576e-05, "loss": 0.2653, "step": 16594 }, { "epoch": 160.9156626506024, "grad_norm": 7.735123634338379, "learning_rate": 1.9441747572815537e-05, "loss": 0.1424, "step": 16595 }, { "epoch": 160.92530120481928, "grad_norm": 6.7204909324646, "learning_rate": 1.9436893203883498e-05, "loss": 0.1341, "step": 16596 }, { "epoch": 160.93493975903615, "grad_norm": 8.472796440124512, "learning_rate": 1.943203883495146e-05, "loss": 0.1354, "step": 16597 }, { "epoch": 160.944578313253, "grad_norm": 8.518939018249512, "learning_rate": 1.942718446601942e-05, "loss": 0.298, "step": 16598 }, { "epoch": 160.95421686746988, "grad_norm": 9.895663261413574, "learning_rate": 1.9422330097087378e-05, "loss": 0.2417, "step": 16599 }, { "epoch": 160.96385542168676, "grad_norm": 3.955075979232788, "learning_rate": 1.941747572815534e-05, "loss": 0.1953, "step": 16600 }, { "epoch": 160.9734939759036, "grad_norm": 21.04835319519043, "learning_rate": 1.94126213592233e-05, "loss": 0.4567, "step": 16601 }, { "epoch": 160.98313253012049, "grad_norm": 6.4680914878845215, "learning_rate": 1.940776699029126e-05, "loss": 0.1532, "step": 16602 }, { "epoch": 160.99277108433736, "grad_norm": 7.06659460067749, "learning_rate": 1.9402912621359223e-05, "loss": 0.2012, "step": 16603 }, { "epoch": 161.00843373493976, "grad_norm": 4.332350730895996, "learning_rate": 1.9398058252427184e-05, "loss": 0.1801, "step": 16604 }, { "epoch": 161.01807228915663, "grad_norm": 16.453357696533203, "learning_rate": 1.9393203883495148e-05, "loss": 0.3076, "step": 16605 }, { "epoch": 161.02771084337348, "grad_norm": 17.765274047851562, "learning_rate": 1.938834951456311e-05, "loss": 0.3595, "step": 16606 }, { "epoch": 161.03734939759036, "grad_norm": 7.334201335906982, "learning_rate": 1.938349514563107e-05, "loss": 0.1817, "step": 16607 }, { "epoch": 161.04698795180724, "grad_norm": 28.620010375976562, "learning_rate": 1.937864077669903e-05, "loss": 0.4895, "step": 16608 }, { "epoch": 161.0566265060241, "grad_norm": 7.253019332885742, "learning_rate": 1.9373786407766993e-05, "loss": 0.3425, "step": 16609 }, { "epoch": 161.06626506024097, "grad_norm": 3.6047425270080566, "learning_rate": 1.9368932038834954e-05, "loss": 0.212, "step": 16610 }, { "epoch": 161.07590361445784, "grad_norm": 21.693870544433594, "learning_rate": 1.9364077669902915e-05, "loss": 0.2852, "step": 16611 }, { "epoch": 161.0855421686747, "grad_norm": 25.631528854370117, "learning_rate": 1.9359223300970873e-05, "loss": 0.2114, "step": 16612 }, { "epoch": 161.09518072289157, "grad_norm": 16.762115478515625, "learning_rate": 1.9354368932038834e-05, "loss": 0.4066, "step": 16613 }, { "epoch": 161.10481927710845, "grad_norm": 64.15398406982422, "learning_rate": 1.9349514563106795e-05, "loss": 0.0921, "step": 16614 }, { "epoch": 161.1144578313253, "grad_norm": 17.807491302490234, "learning_rate": 1.9344660194174756e-05, "loss": 0.286, "step": 16615 }, { "epoch": 161.12409638554217, "grad_norm": 23.594806671142578, "learning_rate": 1.9339805825242717e-05, "loss": 0.2488, "step": 16616 }, { "epoch": 161.13373493975902, "grad_norm": 32.144283294677734, "learning_rate": 1.933495145631068e-05, "loss": 0.3522, "step": 16617 }, { "epoch": 161.1433734939759, "grad_norm": 9.74411392211914, "learning_rate": 1.9330097087378643e-05, "loss": 0.2541, "step": 16618 }, { "epoch": 161.15301204819278, "grad_norm": 9.493521690368652, "learning_rate": 1.9325242718446604e-05, "loss": 0.3719, "step": 16619 }, { "epoch": 161.16265060240963, "grad_norm": 2.973663806915283, "learning_rate": 1.9320388349514565e-05, "loss": 0.1289, "step": 16620 }, { "epoch": 161.1722891566265, "grad_norm": 2.260971784591675, "learning_rate": 1.9315533980582527e-05, "loss": 0.1306, "step": 16621 }, { "epoch": 161.18192771084338, "grad_norm": 8.811697959899902, "learning_rate": 1.9310679611650488e-05, "loss": 0.3225, "step": 16622 }, { "epoch": 161.19156626506023, "grad_norm": 2.9868295192718506, "learning_rate": 1.930582524271845e-05, "loss": 0.2824, "step": 16623 }, { "epoch": 161.2012048192771, "grad_norm": 10.536012649536133, "learning_rate": 1.930097087378641e-05, "loss": 0.2157, "step": 16624 }, { "epoch": 161.210843373494, "grad_norm": 14.271889686584473, "learning_rate": 1.929611650485437e-05, "loss": 0.1278, "step": 16625 }, { "epoch": 161.22048192771084, "grad_norm": 6.49812126159668, "learning_rate": 1.929126213592233e-05, "loss": 0.2395, "step": 16626 }, { "epoch": 161.2301204819277, "grad_norm": 3.915081262588501, "learning_rate": 1.928640776699029e-05, "loss": 0.3064, "step": 16627 }, { "epoch": 161.2397590361446, "grad_norm": 33.617584228515625, "learning_rate": 1.928155339805825e-05, "loss": 0.269, "step": 16628 }, { "epoch": 161.24939759036144, "grad_norm": 14.012372016906738, "learning_rate": 1.9276699029126212e-05, "loss": 0.3196, "step": 16629 }, { "epoch": 161.25903614457832, "grad_norm": 14.670452117919922, "learning_rate": 1.9271844660194173e-05, "loss": 0.2628, "step": 16630 }, { "epoch": 161.2686746987952, "grad_norm": 9.048766136169434, "learning_rate": 1.9266990291262138e-05, "loss": 0.2773, "step": 16631 }, { "epoch": 161.27831325301204, "grad_norm": 7.2635908126831055, "learning_rate": 1.92621359223301e-05, "loss": 0.2543, "step": 16632 }, { "epoch": 161.28795180722892, "grad_norm": 2.017051935195923, "learning_rate": 1.925728155339806e-05, "loss": 0.1122, "step": 16633 }, { "epoch": 161.29759036144577, "grad_norm": 15.954995155334473, "learning_rate": 1.925242718446602e-05, "loss": 0.1903, "step": 16634 }, { "epoch": 161.30722891566265, "grad_norm": 18.424840927124023, "learning_rate": 1.9247572815533983e-05, "loss": 0.3753, "step": 16635 }, { "epoch": 161.31686746987953, "grad_norm": 14.80783748626709, "learning_rate": 1.9242718446601944e-05, "loss": 0.1365, "step": 16636 }, { "epoch": 161.32650602409637, "grad_norm": 4.9785661697387695, "learning_rate": 1.9237864077669905e-05, "loss": 0.2824, "step": 16637 }, { "epoch": 161.33614457831325, "grad_norm": 12.358409881591797, "learning_rate": 1.9233009708737866e-05, "loss": 0.221, "step": 16638 }, { "epoch": 161.34578313253013, "grad_norm": 14.379444122314453, "learning_rate": 1.9228155339805827e-05, "loss": 0.2284, "step": 16639 }, { "epoch": 161.35542168674698, "grad_norm": 0.7472456693649292, "learning_rate": 1.9223300970873785e-05, "loss": 0.138, "step": 16640 }, { "epoch": 161.36506024096386, "grad_norm": 11.129718780517578, "learning_rate": 1.9218446601941746e-05, "loss": 0.2391, "step": 16641 }, { "epoch": 161.37469879518073, "grad_norm": 56.4840202331543, "learning_rate": 1.9213592233009707e-05, "loss": 0.3854, "step": 16642 }, { "epoch": 161.38433734939758, "grad_norm": 10.493587493896484, "learning_rate": 1.9208737864077672e-05, "loss": 0.3227, "step": 16643 }, { "epoch": 161.39397590361446, "grad_norm": 4.139673233032227, "learning_rate": 1.9203883495145633e-05, "loss": 0.1194, "step": 16644 }, { "epoch": 161.40361445783134, "grad_norm": 4.232799053192139, "learning_rate": 1.9199029126213594e-05, "loss": 0.1142, "step": 16645 }, { "epoch": 161.4132530120482, "grad_norm": 10.08650016784668, "learning_rate": 1.9194174757281555e-05, "loss": 0.3739, "step": 16646 }, { "epoch": 161.42289156626506, "grad_norm": 53.795196533203125, "learning_rate": 1.9189320388349516e-05, "loss": 0.5225, "step": 16647 }, { "epoch": 161.4325301204819, "grad_norm": 48.01018524169922, "learning_rate": 1.9184466019417478e-05, "loss": 0.0856, "step": 16648 }, { "epoch": 161.4421686746988, "grad_norm": 15.673611640930176, "learning_rate": 1.917961165048544e-05, "loss": 0.2841, "step": 16649 }, { "epoch": 161.45180722891567, "grad_norm": 18.027246475219727, "learning_rate": 1.91747572815534e-05, "loss": 0.1762, "step": 16650 }, { "epoch": 161.46144578313252, "grad_norm": 22.448345184326172, "learning_rate": 1.916990291262136e-05, "loss": 0.2045, "step": 16651 }, { "epoch": 161.4710843373494, "grad_norm": 26.501081466674805, "learning_rate": 1.9165048543689322e-05, "loss": 0.3282, "step": 16652 }, { "epoch": 161.48072289156627, "grad_norm": 13.314690589904785, "learning_rate": 1.9160194174757283e-05, "loss": 0.2087, "step": 16653 }, { "epoch": 161.49036144578312, "grad_norm": 17.49687385559082, "learning_rate": 1.915533980582524e-05, "loss": 0.2294, "step": 16654 }, { "epoch": 161.5, "grad_norm": 9.420089721679688, "learning_rate": 1.9150485436893202e-05, "loss": 0.4195, "step": 16655 }, { "epoch": 161.50963855421688, "grad_norm": 15.866645812988281, "learning_rate": 1.9145631067961167e-05, "loss": 0.1674, "step": 16656 }, { "epoch": 161.51927710843373, "grad_norm": 1.8377726078033447, "learning_rate": 1.9140776699029128e-05, "loss": 0.1561, "step": 16657 }, { "epoch": 161.5289156626506, "grad_norm": 14.745696067810059, "learning_rate": 1.913592233009709e-05, "loss": 0.1698, "step": 16658 }, { "epoch": 161.53855421686748, "grad_norm": 7.447325229644775, "learning_rate": 1.913106796116505e-05, "loss": 0.1003, "step": 16659 }, { "epoch": 161.54819277108433, "grad_norm": 11.048685073852539, "learning_rate": 1.912621359223301e-05, "loss": 0.1063, "step": 16660 }, { "epoch": 161.5578313253012, "grad_norm": 20.400711059570312, "learning_rate": 1.9121359223300972e-05, "loss": 0.2129, "step": 16661 }, { "epoch": 161.5674698795181, "grad_norm": 6.386499404907227, "learning_rate": 1.9116504854368934e-05, "loss": 0.1425, "step": 16662 }, { "epoch": 161.57710843373494, "grad_norm": 40.068241119384766, "learning_rate": 1.9111650485436895e-05, "loss": 0.2139, "step": 16663 }, { "epoch": 161.5867469879518, "grad_norm": 3.9758636951446533, "learning_rate": 1.9106796116504856e-05, "loss": 0.223, "step": 16664 }, { "epoch": 161.59638554216866, "grad_norm": 16.334548950195312, "learning_rate": 1.9101941747572817e-05, "loss": 0.2911, "step": 16665 }, { "epoch": 161.60602409638554, "grad_norm": 20.07172393798828, "learning_rate": 1.9097087378640778e-05, "loss": 0.2456, "step": 16666 }, { "epoch": 161.61566265060242, "grad_norm": 2.1962502002716064, "learning_rate": 1.909223300970874e-05, "loss": 0.2149, "step": 16667 }, { "epoch": 161.62530120481927, "grad_norm": 9.633088111877441, "learning_rate": 1.9087378640776697e-05, "loss": 0.4648, "step": 16668 }, { "epoch": 161.63493975903614, "grad_norm": 21.67195701599121, "learning_rate": 1.908252427184466e-05, "loss": 0.2792, "step": 16669 }, { "epoch": 161.64457831325302, "grad_norm": 8.152989387512207, "learning_rate": 1.9077669902912623e-05, "loss": 0.1334, "step": 16670 }, { "epoch": 161.65421686746987, "grad_norm": 5.386829853057861, "learning_rate": 1.9072815533980584e-05, "loss": 0.2174, "step": 16671 }, { "epoch": 161.66385542168675, "grad_norm": 27.446651458740234, "learning_rate": 1.9067961165048545e-05, "loss": 0.3658, "step": 16672 }, { "epoch": 161.67349397590363, "grad_norm": 25.377788543701172, "learning_rate": 1.9063106796116506e-05, "loss": 0.3272, "step": 16673 }, { "epoch": 161.68313253012047, "grad_norm": 11.212488174438477, "learning_rate": 1.9058252427184467e-05, "loss": 0.3785, "step": 16674 }, { "epoch": 161.69277108433735, "grad_norm": 24.529932022094727, "learning_rate": 1.905339805825243e-05, "loss": 0.2969, "step": 16675 }, { "epoch": 161.70240963855423, "grad_norm": 3.6123878955841064, "learning_rate": 1.904854368932039e-05, "loss": 0.2901, "step": 16676 }, { "epoch": 161.71204819277108, "grad_norm": 79.14580535888672, "learning_rate": 1.904368932038835e-05, "loss": 0.3549, "step": 16677 }, { "epoch": 161.72168674698796, "grad_norm": 21.728334426879883, "learning_rate": 1.9038834951456312e-05, "loss": 0.3149, "step": 16678 }, { "epoch": 161.73132530120483, "grad_norm": 6.044976234436035, "learning_rate": 1.9033980582524273e-05, "loss": 0.2118, "step": 16679 }, { "epoch": 161.74096385542168, "grad_norm": 7.638438701629639, "learning_rate": 1.9029126213592234e-05, "loss": 0.2824, "step": 16680 }, { "epoch": 161.75060240963856, "grad_norm": 6.1853790283203125, "learning_rate": 1.9024271844660195e-05, "loss": 0.1253, "step": 16681 }, { "epoch": 161.7602409638554, "grad_norm": 17.461524963378906, "learning_rate": 1.9019417475728156e-05, "loss": 0.2614, "step": 16682 }, { "epoch": 161.7698795180723, "grad_norm": 22.556921005249023, "learning_rate": 1.9014563106796118e-05, "loss": 0.2957, "step": 16683 }, { "epoch": 161.77951807228916, "grad_norm": 21.381288528442383, "learning_rate": 1.900970873786408e-05, "loss": 0.1539, "step": 16684 }, { "epoch": 161.789156626506, "grad_norm": 14.76807689666748, "learning_rate": 1.900485436893204e-05, "loss": 0.4266, "step": 16685 }, { "epoch": 161.7987951807229, "grad_norm": 13.148946762084961, "learning_rate": 1.9e-05, "loss": 0.5542, "step": 16686 }, { "epoch": 161.80843373493977, "grad_norm": 7.396385669708252, "learning_rate": 1.8995145631067962e-05, "loss": 0.1938, "step": 16687 }, { "epoch": 161.81807228915662, "grad_norm": 102.0250015258789, "learning_rate": 1.8990291262135923e-05, "loss": 0.2871, "step": 16688 }, { "epoch": 161.8277108433735, "grad_norm": 33.56145095825195, "learning_rate": 1.8985436893203884e-05, "loss": 0.1492, "step": 16689 }, { "epoch": 161.83734939759037, "grad_norm": 10.291183471679688, "learning_rate": 1.8980582524271846e-05, "loss": 0.1895, "step": 16690 }, { "epoch": 161.84698795180722, "grad_norm": 8.105032920837402, "learning_rate": 1.8975728155339807e-05, "loss": 0.1714, "step": 16691 }, { "epoch": 161.8566265060241, "grad_norm": 5.180678367614746, "learning_rate": 1.8970873786407768e-05, "loss": 0.196, "step": 16692 }, { "epoch": 161.86626506024098, "grad_norm": 36.98563003540039, "learning_rate": 1.896601941747573e-05, "loss": 0.257, "step": 16693 }, { "epoch": 161.87590361445783, "grad_norm": 28.171871185302734, "learning_rate": 1.896116504854369e-05, "loss": 0.1442, "step": 16694 }, { "epoch": 161.8855421686747, "grad_norm": 27.75128173828125, "learning_rate": 1.895631067961165e-05, "loss": 0.4548, "step": 16695 }, { "epoch": 161.89518072289155, "grad_norm": 17.896625518798828, "learning_rate": 1.8951456310679613e-05, "loss": 0.1501, "step": 16696 }, { "epoch": 161.90481927710843, "grad_norm": 4.392050743103027, "learning_rate": 1.8946601941747574e-05, "loss": 0.2758, "step": 16697 }, { "epoch": 161.9144578313253, "grad_norm": 4.6874847412109375, "learning_rate": 1.8941747572815535e-05, "loss": 0.221, "step": 16698 }, { "epoch": 161.92409638554216, "grad_norm": 15.405430793762207, "learning_rate": 1.8936893203883496e-05, "loss": 0.2206, "step": 16699 }, { "epoch": 161.93373493975903, "grad_norm": 87.94181060791016, "learning_rate": 1.8932038834951457e-05, "loss": 0.3959, "step": 16700 }, { "epoch": 161.9433734939759, "grad_norm": 8.547576904296875, "learning_rate": 1.8927184466019418e-05, "loss": 0.3879, "step": 16701 }, { "epoch": 161.95301204819276, "grad_norm": 19.620859146118164, "learning_rate": 1.892233009708738e-05, "loss": 0.2732, "step": 16702 }, { "epoch": 161.96265060240964, "grad_norm": 14.574870109558105, "learning_rate": 1.891747572815534e-05, "loss": 0.3416, "step": 16703 }, { "epoch": 161.97228915662652, "grad_norm": 13.658102035522461, "learning_rate": 1.89126213592233e-05, "loss": 0.3375, "step": 16704 }, { "epoch": 161.98192771084337, "grad_norm": 8.517168998718262, "learning_rate": 1.8907766990291263e-05, "loss": 0.2682, "step": 16705 }, { "epoch": 161.99156626506024, "grad_norm": 16.945341110229492, "learning_rate": 1.8902912621359224e-05, "loss": 0.2512, "step": 16706 }, { "epoch": 162.00722891566264, "grad_norm": 35.629356384277344, "learning_rate": 1.8898058252427185e-05, "loss": 0.3031, "step": 16707 }, { "epoch": 162.01686746987951, "grad_norm": 27.95984649658203, "learning_rate": 1.8893203883495146e-05, "loss": 0.2744, "step": 16708 }, { "epoch": 162.0265060240964, "grad_norm": 5.97838830947876, "learning_rate": 1.8888349514563107e-05, "loss": 0.0972, "step": 16709 }, { "epoch": 162.03614457831324, "grad_norm": 48.204288482666016, "learning_rate": 1.888349514563107e-05, "loss": 0.2487, "step": 16710 }, { "epoch": 162.04578313253012, "grad_norm": 4.44800329208374, "learning_rate": 1.887864077669903e-05, "loss": 0.1947, "step": 16711 }, { "epoch": 162.055421686747, "grad_norm": 10.982306480407715, "learning_rate": 1.887378640776699e-05, "loss": 0.2928, "step": 16712 }, { "epoch": 162.06506024096385, "grad_norm": 25.30946922302246, "learning_rate": 1.8868932038834952e-05, "loss": 0.3388, "step": 16713 }, { "epoch": 162.07469879518072, "grad_norm": 5.306406497955322, "learning_rate": 1.8864077669902913e-05, "loss": 0.2621, "step": 16714 }, { "epoch": 162.0843373493976, "grad_norm": 22.5278377532959, "learning_rate": 1.8859223300970874e-05, "loss": 0.3416, "step": 16715 }, { "epoch": 162.09397590361445, "grad_norm": 16.73182487487793, "learning_rate": 1.8854368932038835e-05, "loss": 0.2988, "step": 16716 }, { "epoch": 162.10361445783133, "grad_norm": 16.619905471801758, "learning_rate": 1.8849514563106797e-05, "loss": 0.2441, "step": 16717 }, { "epoch": 162.1132530120482, "grad_norm": 1.671379566192627, "learning_rate": 1.8844660194174758e-05, "loss": 0.1212, "step": 16718 }, { "epoch": 162.12289156626505, "grad_norm": 13.850605964660645, "learning_rate": 1.8839805825242722e-05, "loss": 0.1952, "step": 16719 }, { "epoch": 162.13253012048193, "grad_norm": 31.1409969329834, "learning_rate": 1.883495145631068e-05, "loss": 0.2023, "step": 16720 }, { "epoch": 162.1421686746988, "grad_norm": 19.84817886352539, "learning_rate": 1.883009708737864e-05, "loss": 0.4248, "step": 16721 }, { "epoch": 162.15180722891566, "grad_norm": 4.107056617736816, "learning_rate": 1.8825242718446602e-05, "loss": 0.2239, "step": 16722 }, { "epoch": 162.16144578313254, "grad_norm": 3.1032276153564453, "learning_rate": 1.8820388349514563e-05, "loss": 0.108, "step": 16723 }, { "epoch": 162.17108433734938, "grad_norm": 44.4076042175293, "learning_rate": 1.8815533980582525e-05, "loss": 0.2403, "step": 16724 }, { "epoch": 162.18072289156626, "grad_norm": 39.061256408691406, "learning_rate": 1.8810679611650486e-05, "loss": 0.2139, "step": 16725 }, { "epoch": 162.19036144578314, "grad_norm": 18.897287368774414, "learning_rate": 1.8805825242718447e-05, "loss": 0.322, "step": 16726 }, { "epoch": 162.2, "grad_norm": 3.5374412536621094, "learning_rate": 1.8800970873786408e-05, "loss": 0.217, "step": 16727 }, { "epoch": 162.20963855421687, "grad_norm": 19.305389404296875, "learning_rate": 1.879611650485437e-05, "loss": 0.1909, "step": 16728 }, { "epoch": 162.21927710843374, "grad_norm": 39.6058235168457, "learning_rate": 1.879126213592233e-05, "loss": 0.2245, "step": 16729 }, { "epoch": 162.2289156626506, "grad_norm": 3.2702505588531494, "learning_rate": 1.878640776699029e-05, "loss": 0.2272, "step": 16730 }, { "epoch": 162.23855421686747, "grad_norm": 23.990917205810547, "learning_rate": 1.8781553398058253e-05, "loss": 0.2441, "step": 16731 }, { "epoch": 162.24819277108435, "grad_norm": 76.6637954711914, "learning_rate": 1.8776699029126217e-05, "loss": 0.3794, "step": 16732 }, { "epoch": 162.2578313253012, "grad_norm": 14.112210273742676, "learning_rate": 1.8771844660194178e-05, "loss": 0.176, "step": 16733 }, { "epoch": 162.26746987951807, "grad_norm": 7.3453497886657715, "learning_rate": 1.8766990291262136e-05, "loss": 0.2935, "step": 16734 }, { "epoch": 162.27710843373495, "grad_norm": 3.0646824836730957, "learning_rate": 1.8762135922330097e-05, "loss": 0.1135, "step": 16735 }, { "epoch": 162.2867469879518, "grad_norm": 64.97764587402344, "learning_rate": 1.875728155339806e-05, "loss": 0.3933, "step": 16736 }, { "epoch": 162.29638554216868, "grad_norm": 3.962329626083374, "learning_rate": 1.875242718446602e-05, "loss": 0.1624, "step": 16737 }, { "epoch": 162.30602409638556, "grad_norm": 13.444849014282227, "learning_rate": 1.874757281553398e-05, "loss": 0.3274, "step": 16738 }, { "epoch": 162.3156626506024, "grad_norm": 38.26493453979492, "learning_rate": 1.8742718446601942e-05, "loss": 0.2896, "step": 16739 }, { "epoch": 162.32530120481928, "grad_norm": 5.095602512359619, "learning_rate": 1.8737864077669903e-05, "loss": 0.1022, "step": 16740 }, { "epoch": 162.33493975903613, "grad_norm": 27.471107482910156, "learning_rate": 1.8733009708737864e-05, "loss": 0.143, "step": 16741 }, { "epoch": 162.344578313253, "grad_norm": 3.271226644515991, "learning_rate": 1.8728155339805825e-05, "loss": 0.1289, "step": 16742 }, { "epoch": 162.3542168674699, "grad_norm": 15.614852905273438, "learning_rate": 1.8723300970873786e-05, "loss": 0.373, "step": 16743 }, { "epoch": 162.36385542168674, "grad_norm": 37.16290283203125, "learning_rate": 1.871844660194175e-05, "loss": 0.2621, "step": 16744 }, { "epoch": 162.3734939759036, "grad_norm": 44.15142059326172, "learning_rate": 1.8713592233009712e-05, "loss": 0.2665, "step": 16745 }, { "epoch": 162.3831325301205, "grad_norm": 6.039505481719971, "learning_rate": 1.8708737864077673e-05, "loss": 0.2804, "step": 16746 }, { "epoch": 162.39277108433734, "grad_norm": 4.493310451507568, "learning_rate": 1.8703883495145634e-05, "loss": 0.278, "step": 16747 }, { "epoch": 162.40240963855422, "grad_norm": 7.502641677856445, "learning_rate": 1.8699029126213592e-05, "loss": 0.0793, "step": 16748 }, { "epoch": 162.4120481927711, "grad_norm": 5.50105619430542, "learning_rate": 1.8694174757281553e-05, "loss": 0.1923, "step": 16749 }, { "epoch": 162.42168674698794, "grad_norm": 17.343751907348633, "learning_rate": 1.8689320388349514e-05, "loss": 0.3481, "step": 16750 }, { "epoch": 162.43132530120482, "grad_norm": 5.255319118499756, "learning_rate": 1.8684466019417476e-05, "loss": 0.109, "step": 16751 }, { "epoch": 162.4409638554217, "grad_norm": 4.325448989868164, "learning_rate": 1.8679611650485437e-05, "loss": 0.2993, "step": 16752 }, { "epoch": 162.45060240963855, "grad_norm": 28.761011123657227, "learning_rate": 1.8674757281553398e-05, "loss": 0.2801, "step": 16753 }, { "epoch": 162.46024096385543, "grad_norm": 28.821054458618164, "learning_rate": 1.866990291262136e-05, "loss": 0.4191, "step": 16754 }, { "epoch": 162.46987951807228, "grad_norm": 2.8780131340026855, "learning_rate": 1.866504854368932e-05, "loss": 0.1013, "step": 16755 }, { "epoch": 162.47951807228915, "grad_norm": 22.260557174682617, "learning_rate": 1.866019417475728e-05, "loss": 0.4364, "step": 16756 }, { "epoch": 162.48915662650603, "grad_norm": 10.597652435302734, "learning_rate": 1.8655339805825246e-05, "loss": 0.1278, "step": 16757 }, { "epoch": 162.49879518072288, "grad_norm": 28.682361602783203, "learning_rate": 1.8650485436893207e-05, "loss": 0.3115, "step": 16758 }, { "epoch": 162.50843373493976, "grad_norm": 12.911110877990723, "learning_rate": 1.8645631067961168e-05, "loss": 0.1808, "step": 16759 }, { "epoch": 162.51807228915663, "grad_norm": 62.094993591308594, "learning_rate": 1.864077669902913e-05, "loss": 0.3059, "step": 16760 }, { "epoch": 162.52771084337348, "grad_norm": 19.190317153930664, "learning_rate": 1.8635922330097087e-05, "loss": 0.2246, "step": 16761 }, { "epoch": 162.53734939759036, "grad_norm": 31.409503936767578, "learning_rate": 1.8631067961165048e-05, "loss": 0.2613, "step": 16762 }, { "epoch": 162.54698795180724, "grad_norm": 51.711212158203125, "learning_rate": 1.862621359223301e-05, "loss": 0.257, "step": 16763 }, { "epoch": 162.5566265060241, "grad_norm": 2.721360683441162, "learning_rate": 1.862135922330097e-05, "loss": 0.2657, "step": 16764 }, { "epoch": 162.56626506024097, "grad_norm": 10.104514122009277, "learning_rate": 1.861650485436893e-05, "loss": 0.1581, "step": 16765 }, { "epoch": 162.57590361445784, "grad_norm": 10.60589599609375, "learning_rate": 1.8611650485436893e-05, "loss": 0.0806, "step": 16766 }, { "epoch": 162.5855421686747, "grad_norm": 21.141250610351562, "learning_rate": 1.8606796116504854e-05, "loss": 0.3915, "step": 16767 }, { "epoch": 162.59518072289157, "grad_norm": 44.0502815246582, "learning_rate": 1.8601941747572815e-05, "loss": 0.3096, "step": 16768 }, { "epoch": 162.60481927710845, "grad_norm": 3.262559652328491, "learning_rate": 1.8597087378640776e-05, "loss": 0.1053, "step": 16769 }, { "epoch": 162.6144578313253, "grad_norm": 33.30974578857422, "learning_rate": 1.859223300970874e-05, "loss": 0.3871, "step": 16770 }, { "epoch": 162.62409638554217, "grad_norm": 5.5649518966674805, "learning_rate": 1.8587378640776702e-05, "loss": 0.4408, "step": 16771 }, { "epoch": 162.63373493975902, "grad_norm": 15.422868728637695, "learning_rate": 1.8582524271844663e-05, "loss": 0.1951, "step": 16772 }, { "epoch": 162.6433734939759, "grad_norm": 33.65092849731445, "learning_rate": 1.8577669902912624e-05, "loss": 0.3411, "step": 16773 }, { "epoch": 162.65301204819278, "grad_norm": 40.32637405395508, "learning_rate": 1.8572815533980585e-05, "loss": 0.263, "step": 16774 }, { "epoch": 162.66265060240963, "grad_norm": 3.8352649211883545, "learning_rate": 1.8567961165048543e-05, "loss": 0.3268, "step": 16775 }, { "epoch": 162.6722891566265, "grad_norm": 33.37020492553711, "learning_rate": 1.8563106796116504e-05, "loss": 0.2555, "step": 16776 }, { "epoch": 162.68192771084338, "grad_norm": 6.445910930633545, "learning_rate": 1.8558252427184465e-05, "loss": 0.2479, "step": 16777 }, { "epoch": 162.69156626506023, "grad_norm": 9.835650444030762, "learning_rate": 1.8553398058252426e-05, "loss": 0.1139, "step": 16778 }, { "epoch": 162.7012048192771, "grad_norm": 8.4033842086792, "learning_rate": 1.8548543689320388e-05, "loss": 0.1299, "step": 16779 }, { "epoch": 162.710843373494, "grad_norm": 12.156495094299316, "learning_rate": 1.854368932038835e-05, "loss": 0.2925, "step": 16780 }, { "epoch": 162.72048192771084, "grad_norm": 22.76418685913086, "learning_rate": 1.853883495145631e-05, "loss": 0.19, "step": 16781 }, { "epoch": 162.7301204819277, "grad_norm": 6.375209331512451, "learning_rate": 1.8533980582524274e-05, "loss": 0.0974, "step": 16782 }, { "epoch": 162.7397590361446, "grad_norm": 12.162818908691406, "learning_rate": 1.8529126213592236e-05, "loss": 0.2694, "step": 16783 }, { "epoch": 162.74939759036144, "grad_norm": 36.56980895996094, "learning_rate": 1.8524271844660197e-05, "loss": 0.2667, "step": 16784 }, { "epoch": 162.75903614457832, "grad_norm": 21.497041702270508, "learning_rate": 1.8519417475728158e-05, "loss": 0.3568, "step": 16785 }, { "epoch": 162.76867469879517, "grad_norm": 34.33867645263672, "learning_rate": 1.851456310679612e-05, "loss": 0.4384, "step": 16786 }, { "epoch": 162.77831325301204, "grad_norm": 47.6843147277832, "learning_rate": 1.850970873786408e-05, "loss": 0.2492, "step": 16787 }, { "epoch": 162.78795180722892, "grad_norm": 13.051590919494629, "learning_rate": 1.850485436893204e-05, "loss": 0.2998, "step": 16788 }, { "epoch": 162.79759036144577, "grad_norm": 9.403839111328125, "learning_rate": 1.85e-05, "loss": 0.3463, "step": 16789 }, { "epoch": 162.80722891566265, "grad_norm": 12.868977546691895, "learning_rate": 1.849514563106796e-05, "loss": 0.3356, "step": 16790 }, { "epoch": 162.81686746987953, "grad_norm": 15.741955757141113, "learning_rate": 1.849029126213592e-05, "loss": 0.2434, "step": 16791 }, { "epoch": 162.82650602409637, "grad_norm": 23.57705307006836, "learning_rate": 1.8485436893203882e-05, "loss": 0.1966, "step": 16792 }, { "epoch": 162.83614457831325, "grad_norm": 23.03622817993164, "learning_rate": 1.8480582524271844e-05, "loss": 0.4186, "step": 16793 }, { "epoch": 162.84578313253013, "grad_norm": 20.85418128967285, "learning_rate": 1.8475728155339805e-05, "loss": 0.22, "step": 16794 }, { "epoch": 162.85542168674698, "grad_norm": 29.933610916137695, "learning_rate": 1.847087378640777e-05, "loss": 0.2391, "step": 16795 }, { "epoch": 162.86506024096386, "grad_norm": 49.37569046020508, "learning_rate": 1.846601941747573e-05, "loss": 0.2091, "step": 16796 }, { "epoch": 162.87469879518073, "grad_norm": 8.275412559509277, "learning_rate": 1.846116504854369e-05, "loss": 0.2355, "step": 16797 }, { "epoch": 162.88433734939758, "grad_norm": 5.918142318725586, "learning_rate": 1.8456310679611653e-05, "loss": 0.3913, "step": 16798 }, { "epoch": 162.89397590361446, "grad_norm": 15.83591079711914, "learning_rate": 1.8451456310679614e-05, "loss": 0.2275, "step": 16799 }, { "epoch": 162.90361445783134, "grad_norm": 15.791193962097168, "learning_rate": 1.8446601941747575e-05, "loss": 0.2815, "step": 16800 }, { "epoch": 162.9132530120482, "grad_norm": 111.04393005371094, "learning_rate": 1.8441747572815536e-05, "loss": 0.2385, "step": 16801 }, { "epoch": 162.92289156626506, "grad_norm": 4.056354522705078, "learning_rate": 1.8436893203883497e-05, "loss": 0.335, "step": 16802 }, { "epoch": 162.9325301204819, "grad_norm": 11.628646850585938, "learning_rate": 1.8432038834951455e-05, "loss": 0.1641, "step": 16803 }, { "epoch": 162.9421686746988, "grad_norm": 9.412939071655273, "learning_rate": 1.8427184466019416e-05, "loss": 0.3901, "step": 16804 }, { "epoch": 162.95180722891567, "grad_norm": 13.525108337402344, "learning_rate": 1.8422330097087377e-05, "loss": 0.1315, "step": 16805 }, { "epoch": 162.96144578313252, "grad_norm": 2.022623062133789, "learning_rate": 1.841747572815534e-05, "loss": 0.2439, "step": 16806 }, { "epoch": 162.9710843373494, "grad_norm": 28.001192092895508, "learning_rate": 1.8412621359223303e-05, "loss": 0.1845, "step": 16807 }, { "epoch": 162.98072289156627, "grad_norm": 22.86680793762207, "learning_rate": 1.8407766990291264e-05, "loss": 0.3007, "step": 16808 }, { "epoch": 162.99036144578312, "grad_norm": 23.638689041137695, "learning_rate": 1.8402912621359225e-05, "loss": 0.199, "step": 16809 }, { "epoch": 163.00602409638554, "grad_norm": 8.013517379760742, "learning_rate": 1.8398058252427187e-05, "loss": 0.2636, "step": 16810 }, { "epoch": 163.01566265060242, "grad_norm": 23.670024871826172, "learning_rate": 1.8393203883495148e-05, "loss": 0.1953, "step": 16811 }, { "epoch": 163.02530120481927, "grad_norm": 31.477346420288086, "learning_rate": 1.838834951456311e-05, "loss": 0.1391, "step": 16812 }, { "epoch": 163.03493975903615, "grad_norm": 9.527496337890625, "learning_rate": 1.838349514563107e-05, "loss": 0.2755, "step": 16813 }, { "epoch": 163.044578313253, "grad_norm": 87.9305191040039, "learning_rate": 1.837864077669903e-05, "loss": 0.2836, "step": 16814 }, { "epoch": 163.05421686746988, "grad_norm": 3.7390449047088623, "learning_rate": 1.8373786407766992e-05, "loss": 0.1278, "step": 16815 }, { "epoch": 163.06385542168675, "grad_norm": 18.83791732788086, "learning_rate": 1.8368932038834953e-05, "loss": 0.173, "step": 16816 }, { "epoch": 163.0734939759036, "grad_norm": 38.16965103149414, "learning_rate": 1.836407766990291e-05, "loss": 0.349, "step": 16817 }, { "epoch": 163.08313253012048, "grad_norm": 23.099178314208984, "learning_rate": 1.8359223300970872e-05, "loss": 0.4439, "step": 16818 }, { "epoch": 163.09277108433736, "grad_norm": 14.345422744750977, "learning_rate": 1.8354368932038833e-05, "loss": 0.2452, "step": 16819 }, { "epoch": 163.1024096385542, "grad_norm": 13.875359535217285, "learning_rate": 1.8349514563106798e-05, "loss": 0.1553, "step": 16820 }, { "epoch": 163.11204819277108, "grad_norm": 26.41163444519043, "learning_rate": 1.834466019417476e-05, "loss": 0.313, "step": 16821 }, { "epoch": 163.12168674698796, "grad_norm": 9.234823226928711, "learning_rate": 1.833980582524272e-05, "loss": 0.324, "step": 16822 }, { "epoch": 163.1313253012048, "grad_norm": 13.156211853027344, "learning_rate": 1.833495145631068e-05, "loss": 0.3847, "step": 16823 }, { "epoch": 163.1409638554217, "grad_norm": 9.390875816345215, "learning_rate": 1.8330097087378643e-05, "loss": 0.2737, "step": 16824 }, { "epoch": 163.15060240963857, "grad_norm": 38.83802032470703, "learning_rate": 1.8325242718446604e-05, "loss": 0.5488, "step": 16825 }, { "epoch": 163.16024096385541, "grad_norm": 5.816518306732178, "learning_rate": 1.8320388349514565e-05, "loss": 0.1762, "step": 16826 }, { "epoch": 163.1698795180723, "grad_norm": 37.96920394897461, "learning_rate": 1.8315533980582526e-05, "loss": 0.2731, "step": 16827 }, { "epoch": 163.17951807228917, "grad_norm": 8.375473976135254, "learning_rate": 1.8310679611650487e-05, "loss": 0.1435, "step": 16828 }, { "epoch": 163.18915662650602, "grad_norm": 42.452816009521484, "learning_rate": 1.8305825242718448e-05, "loss": 0.2714, "step": 16829 }, { "epoch": 163.1987951807229, "grad_norm": 10.054167747497559, "learning_rate": 1.830097087378641e-05, "loss": 0.3745, "step": 16830 }, { "epoch": 163.20843373493975, "grad_norm": 12.20307731628418, "learning_rate": 1.8296116504854367e-05, "loss": 0.1674, "step": 16831 }, { "epoch": 163.21807228915662, "grad_norm": 8.48831558227539, "learning_rate": 1.829126213592233e-05, "loss": 0.2221, "step": 16832 }, { "epoch": 163.2277108433735, "grad_norm": 30.58859634399414, "learning_rate": 1.8286407766990293e-05, "loss": 0.2183, "step": 16833 }, { "epoch": 163.23734939759035, "grad_norm": 7.296892166137695, "learning_rate": 1.8281553398058254e-05, "loss": 0.1711, "step": 16834 }, { "epoch": 163.24698795180723, "grad_norm": 12.629841804504395, "learning_rate": 1.8276699029126215e-05, "loss": 0.2551, "step": 16835 }, { "epoch": 163.2566265060241, "grad_norm": 17.790180206298828, "learning_rate": 1.8271844660194176e-05, "loss": 0.2635, "step": 16836 }, { "epoch": 163.26626506024095, "grad_norm": 5.9383745193481445, "learning_rate": 1.8266990291262137e-05, "loss": 0.3095, "step": 16837 }, { "epoch": 163.27590361445783, "grad_norm": 10.156723022460938, "learning_rate": 1.82621359223301e-05, "loss": 0.1909, "step": 16838 }, { "epoch": 163.2855421686747, "grad_norm": 37.99349594116211, "learning_rate": 1.825728155339806e-05, "loss": 0.2106, "step": 16839 }, { "epoch": 163.29518072289156, "grad_norm": 15.52386474609375, "learning_rate": 1.825242718446602e-05, "loss": 0.2842, "step": 16840 }, { "epoch": 163.30481927710844, "grad_norm": 10.43819522857666, "learning_rate": 1.8247572815533982e-05, "loss": 0.2439, "step": 16841 }, { "epoch": 163.3144578313253, "grad_norm": 9.491585731506348, "learning_rate": 1.8242718446601943e-05, "loss": 0.2574, "step": 16842 }, { "epoch": 163.32409638554216, "grad_norm": 29.036087036132812, "learning_rate": 1.8237864077669904e-05, "loss": 0.1803, "step": 16843 }, { "epoch": 163.33373493975904, "grad_norm": 19.393774032592773, "learning_rate": 1.8233009708737862e-05, "loss": 0.2734, "step": 16844 }, { "epoch": 163.34337349397592, "grad_norm": 18.48889923095703, "learning_rate": 1.8228155339805827e-05, "loss": 0.3704, "step": 16845 }, { "epoch": 163.35301204819277, "grad_norm": 4.94991397857666, "learning_rate": 1.8223300970873788e-05, "loss": 0.1569, "step": 16846 }, { "epoch": 163.36265060240964, "grad_norm": 22.645111083984375, "learning_rate": 1.821844660194175e-05, "loss": 0.3645, "step": 16847 }, { "epoch": 163.3722891566265, "grad_norm": 7.907504081726074, "learning_rate": 1.821359223300971e-05, "loss": 0.1496, "step": 16848 }, { "epoch": 163.38192771084337, "grad_norm": 48.267940521240234, "learning_rate": 1.820873786407767e-05, "loss": 0.1707, "step": 16849 }, { "epoch": 163.39156626506025, "grad_norm": 5.214818477630615, "learning_rate": 1.8203883495145632e-05, "loss": 0.3607, "step": 16850 }, { "epoch": 163.4012048192771, "grad_norm": 18.63065528869629, "learning_rate": 1.8199029126213593e-05, "loss": 0.2098, "step": 16851 }, { "epoch": 163.41084337349398, "grad_norm": 9.043057441711426, "learning_rate": 1.8194174757281555e-05, "loss": 0.3075, "step": 16852 }, { "epoch": 163.42048192771085, "grad_norm": 16.99470329284668, "learning_rate": 1.8189320388349516e-05, "loss": 0.3732, "step": 16853 }, { "epoch": 163.4301204819277, "grad_norm": 30.598421096801758, "learning_rate": 1.8184466019417477e-05, "loss": 0.3181, "step": 16854 }, { "epoch": 163.43975903614458, "grad_norm": 8.992722511291504, "learning_rate": 1.8179611650485438e-05, "loss": 0.1064, "step": 16855 }, { "epoch": 163.44939759036146, "grad_norm": 11.195326805114746, "learning_rate": 1.81747572815534e-05, "loss": 0.2194, "step": 16856 }, { "epoch": 163.4590361445783, "grad_norm": 5.6178507804870605, "learning_rate": 1.816990291262136e-05, "loss": 0.3479, "step": 16857 }, { "epoch": 163.46867469879518, "grad_norm": 4.3224310874938965, "learning_rate": 1.816504854368932e-05, "loss": 0.2165, "step": 16858 }, { "epoch": 163.47831325301206, "grad_norm": 28.07956314086914, "learning_rate": 1.8160194174757283e-05, "loss": 0.3384, "step": 16859 }, { "epoch": 163.4879518072289, "grad_norm": 11.428893089294434, "learning_rate": 1.8155339805825244e-05, "loss": 0.3446, "step": 16860 }, { "epoch": 163.4975903614458, "grad_norm": 21.280941009521484, "learning_rate": 1.8150485436893205e-05, "loss": 0.2413, "step": 16861 }, { "epoch": 163.50722891566264, "grad_norm": 19.808868408203125, "learning_rate": 1.8145631067961166e-05, "loss": 0.1598, "step": 16862 }, { "epoch": 163.51686746987951, "grad_norm": 7.861799716949463, "learning_rate": 1.8140776699029127e-05, "loss": 0.2217, "step": 16863 }, { "epoch": 163.5265060240964, "grad_norm": 20.616863250732422, "learning_rate": 1.813592233009709e-05, "loss": 0.3566, "step": 16864 }, { "epoch": 163.53614457831324, "grad_norm": 21.18230628967285, "learning_rate": 1.813106796116505e-05, "loss": 0.2216, "step": 16865 }, { "epoch": 163.54578313253012, "grad_norm": 7.803322792053223, "learning_rate": 1.812621359223301e-05, "loss": 0.354, "step": 16866 }, { "epoch": 163.555421686747, "grad_norm": 27.675350189208984, "learning_rate": 1.8121359223300972e-05, "loss": 0.3558, "step": 16867 }, { "epoch": 163.56506024096385, "grad_norm": 30.582027435302734, "learning_rate": 1.8116504854368933e-05, "loss": 0.3303, "step": 16868 }, { "epoch": 163.57469879518072, "grad_norm": 25.91695213317871, "learning_rate": 1.8111650485436894e-05, "loss": 0.2051, "step": 16869 }, { "epoch": 163.5843373493976, "grad_norm": 12.86754035949707, "learning_rate": 1.8106796116504855e-05, "loss": 0.4095, "step": 16870 }, { "epoch": 163.59397590361445, "grad_norm": 2.420055866241455, "learning_rate": 1.8101941747572816e-05, "loss": 0.0957, "step": 16871 }, { "epoch": 163.60361445783133, "grad_norm": 21.703018188476562, "learning_rate": 1.8097087378640778e-05, "loss": 0.28, "step": 16872 }, { "epoch": 163.6132530120482, "grad_norm": 7.539157867431641, "learning_rate": 1.809223300970874e-05, "loss": 0.2349, "step": 16873 }, { "epoch": 163.62289156626505, "grad_norm": 1.682423710823059, "learning_rate": 1.80873786407767e-05, "loss": 0.2342, "step": 16874 }, { "epoch": 163.63253012048193, "grad_norm": 4.567487716674805, "learning_rate": 1.808252427184466e-05, "loss": 0.1566, "step": 16875 }, { "epoch": 163.6421686746988, "grad_norm": 3.7359495162963867, "learning_rate": 1.8077669902912622e-05, "loss": 0.0939, "step": 16876 }, { "epoch": 163.65180722891566, "grad_norm": 37.66117477416992, "learning_rate": 1.8072815533980583e-05, "loss": 0.3143, "step": 16877 }, { "epoch": 163.66144578313254, "grad_norm": 23.27448081970215, "learning_rate": 1.8067961165048544e-05, "loss": 0.2818, "step": 16878 }, { "epoch": 163.67108433734938, "grad_norm": 61.94942855834961, "learning_rate": 1.8063106796116506e-05, "loss": 0.2169, "step": 16879 }, { "epoch": 163.68072289156626, "grad_norm": 10.727075576782227, "learning_rate": 1.8058252427184467e-05, "loss": 0.3971, "step": 16880 }, { "epoch": 163.69036144578314, "grad_norm": 12.43814754486084, "learning_rate": 1.8053398058252428e-05, "loss": 0.1537, "step": 16881 }, { "epoch": 163.7, "grad_norm": 10.389368057250977, "learning_rate": 1.804854368932039e-05, "loss": 0.3299, "step": 16882 }, { "epoch": 163.70963855421687, "grad_norm": 3.4046883583068848, "learning_rate": 1.804368932038835e-05, "loss": 0.1963, "step": 16883 }, { "epoch": 163.71927710843374, "grad_norm": 60.64973449707031, "learning_rate": 1.803883495145631e-05, "loss": 0.229, "step": 16884 }, { "epoch": 163.7289156626506, "grad_norm": 6.138768672943115, "learning_rate": 1.8033980582524272e-05, "loss": 0.2939, "step": 16885 }, { "epoch": 163.73855421686747, "grad_norm": 11.240045547485352, "learning_rate": 1.8029126213592234e-05, "loss": 0.4016, "step": 16886 }, { "epoch": 163.74819277108435, "grad_norm": 20.035173416137695, "learning_rate": 1.8024271844660195e-05, "loss": 0.1712, "step": 16887 }, { "epoch": 163.7578313253012, "grad_norm": 2.8697457313537598, "learning_rate": 1.8019417475728156e-05, "loss": 0.3144, "step": 16888 }, { "epoch": 163.76746987951807, "grad_norm": 15.590394973754883, "learning_rate": 1.8014563106796117e-05, "loss": 0.0468, "step": 16889 }, { "epoch": 163.77710843373495, "grad_norm": 10.242039680480957, "learning_rate": 1.8009708737864078e-05, "loss": 0.2483, "step": 16890 }, { "epoch": 163.7867469879518, "grad_norm": 12.080666542053223, "learning_rate": 1.800485436893204e-05, "loss": 0.1316, "step": 16891 }, { "epoch": 163.79638554216868, "grad_norm": 23.413705825805664, "learning_rate": 1.8e-05, "loss": 0.3014, "step": 16892 }, { "epoch": 163.80602409638556, "grad_norm": 11.104670524597168, "learning_rate": 1.799514563106796e-05, "loss": 0.222, "step": 16893 }, { "epoch": 163.8156626506024, "grad_norm": 65.29872131347656, "learning_rate": 1.7990291262135923e-05, "loss": 0.1394, "step": 16894 }, { "epoch": 163.82530120481928, "grad_norm": 6.795324325561523, "learning_rate": 1.7985436893203884e-05, "loss": 0.2933, "step": 16895 }, { "epoch": 163.83493975903613, "grad_norm": 28.42508888244629, "learning_rate": 1.798058252427185e-05, "loss": 0.3508, "step": 16896 }, { "epoch": 163.844578313253, "grad_norm": 4.0430588722229, "learning_rate": 1.7975728155339806e-05, "loss": 0.1328, "step": 16897 }, { "epoch": 163.8542168674699, "grad_norm": 5.321552276611328, "learning_rate": 1.7970873786407767e-05, "loss": 0.3001, "step": 16898 }, { "epoch": 163.86385542168674, "grad_norm": 32.33648681640625, "learning_rate": 1.796601941747573e-05, "loss": 0.3485, "step": 16899 }, { "epoch": 163.8734939759036, "grad_norm": 4.332087516784668, "learning_rate": 1.796116504854369e-05, "loss": 0.3221, "step": 16900 }, { "epoch": 163.8831325301205, "grad_norm": 6.610622882843018, "learning_rate": 1.795631067961165e-05, "loss": 0.1694, "step": 16901 }, { "epoch": 163.89277108433734, "grad_norm": 6.870333194732666, "learning_rate": 1.7951456310679612e-05, "loss": 0.2352, "step": 16902 }, { "epoch": 163.90240963855422, "grad_norm": 3.534167528152466, "learning_rate": 1.7946601941747573e-05, "loss": 0.0961, "step": 16903 }, { "epoch": 163.9120481927711, "grad_norm": 42.021541595458984, "learning_rate": 1.7941747572815534e-05, "loss": 0.222, "step": 16904 }, { "epoch": 163.92168674698794, "grad_norm": 5.688285827636719, "learning_rate": 1.7936893203883495e-05, "loss": 0.1801, "step": 16905 }, { "epoch": 163.93132530120482, "grad_norm": 3.61586856842041, "learning_rate": 1.7932038834951456e-05, "loss": 0.0587, "step": 16906 }, { "epoch": 163.9409638554217, "grad_norm": 22.827075958251953, "learning_rate": 1.7927184466019418e-05, "loss": 0.1614, "step": 16907 }, { "epoch": 163.95060240963855, "grad_norm": 14.759624481201172, "learning_rate": 1.7922330097087382e-05, "loss": 0.2306, "step": 16908 }, { "epoch": 163.96024096385543, "grad_norm": 23.215728759765625, "learning_rate": 1.7917475728155343e-05, "loss": 0.2717, "step": 16909 }, { "epoch": 163.96987951807228, "grad_norm": 40.40121841430664, "learning_rate": 1.79126213592233e-05, "loss": 0.4179, "step": 16910 }, { "epoch": 163.97951807228915, "grad_norm": 5.103194236755371, "learning_rate": 1.7907766990291262e-05, "loss": 0.2453, "step": 16911 }, { "epoch": 163.98915662650603, "grad_norm": 14.075106620788574, "learning_rate": 1.7902912621359223e-05, "loss": 0.1576, "step": 16912 }, { "epoch": 164.00481927710842, "grad_norm": 6.690237045288086, "learning_rate": 1.7898058252427185e-05, "loss": 0.2582, "step": 16913 }, { "epoch": 164.0144578313253, "grad_norm": 16.4896297454834, "learning_rate": 1.7893203883495146e-05, "loss": 0.2062, "step": 16914 }, { "epoch": 164.02409638554218, "grad_norm": 13.710380554199219, "learning_rate": 1.7888349514563107e-05, "loss": 0.2454, "step": 16915 }, { "epoch": 164.03373493975903, "grad_norm": 42.717002868652344, "learning_rate": 1.7883495145631068e-05, "loss": 0.4338, "step": 16916 }, { "epoch": 164.0433734939759, "grad_norm": 6.049154281616211, "learning_rate": 1.787864077669903e-05, "loss": 0.1704, "step": 16917 }, { "epoch": 164.05301204819278, "grad_norm": 4.677532196044922, "learning_rate": 1.787378640776699e-05, "loss": 0.2583, "step": 16918 }, { "epoch": 164.06265060240963, "grad_norm": 5.697997570037842, "learning_rate": 1.786893203883495e-05, "loss": 0.3048, "step": 16919 }, { "epoch": 164.0722891566265, "grad_norm": 24.469186782836914, "learning_rate": 1.7864077669902913e-05, "loss": 0.2193, "step": 16920 }, { "epoch": 164.08192771084336, "grad_norm": 10.701101303100586, "learning_rate": 1.7859223300970877e-05, "loss": 0.3408, "step": 16921 }, { "epoch": 164.09156626506024, "grad_norm": 17.10112190246582, "learning_rate": 1.7854368932038838e-05, "loss": 0.4148, "step": 16922 }, { "epoch": 164.10120481927711, "grad_norm": 7.896427631378174, "learning_rate": 1.78495145631068e-05, "loss": 0.277, "step": 16923 }, { "epoch": 164.11084337349396, "grad_norm": 5.742314338684082, "learning_rate": 1.7844660194174757e-05, "loss": 0.3025, "step": 16924 }, { "epoch": 164.12048192771084, "grad_norm": 41.339630126953125, "learning_rate": 1.7839805825242718e-05, "loss": 0.4373, "step": 16925 }, { "epoch": 164.13012048192772, "grad_norm": 3.6977996826171875, "learning_rate": 1.783495145631068e-05, "loss": 0.2264, "step": 16926 }, { "epoch": 164.13975903614457, "grad_norm": 14.568506240844727, "learning_rate": 1.783009708737864e-05, "loss": 0.2513, "step": 16927 }, { "epoch": 164.14939759036145, "grad_norm": 51.01414489746094, "learning_rate": 1.7825242718446602e-05, "loss": 0.3374, "step": 16928 }, { "epoch": 164.15903614457832, "grad_norm": 20.964977264404297, "learning_rate": 1.7820388349514563e-05, "loss": 0.26, "step": 16929 }, { "epoch": 164.16867469879517, "grad_norm": 7.2135186195373535, "learning_rate": 1.7815533980582524e-05, "loss": 0.3541, "step": 16930 }, { "epoch": 164.17831325301205, "grad_norm": 62.863128662109375, "learning_rate": 1.7810679611650485e-05, "loss": 0.2362, "step": 16931 }, { "epoch": 164.18795180722893, "grad_norm": 25.68766212463379, "learning_rate": 1.7805825242718446e-05, "loss": 0.3268, "step": 16932 }, { "epoch": 164.19759036144578, "grad_norm": 10.943194389343262, "learning_rate": 1.7800970873786407e-05, "loss": 0.1463, "step": 16933 }, { "epoch": 164.20722891566265, "grad_norm": 39.365840911865234, "learning_rate": 1.7796116504854372e-05, "loss": 0.2521, "step": 16934 }, { "epoch": 164.21686746987953, "grad_norm": 18.02819061279297, "learning_rate": 1.7791262135922333e-05, "loss": 0.0955, "step": 16935 }, { "epoch": 164.22650602409638, "grad_norm": 23.83289909362793, "learning_rate": 1.7786407766990294e-05, "loss": 0.3866, "step": 16936 }, { "epoch": 164.23614457831326, "grad_norm": 15.356937408447266, "learning_rate": 1.7781553398058255e-05, "loss": 0.174, "step": 16937 }, { "epoch": 164.2457831325301, "grad_norm": 8.55949592590332, "learning_rate": 1.7776699029126213e-05, "loss": 0.5042, "step": 16938 }, { "epoch": 164.25542168674698, "grad_norm": 17.063152313232422, "learning_rate": 1.7771844660194174e-05, "loss": 0.152, "step": 16939 }, { "epoch": 164.26506024096386, "grad_norm": 24.172731399536133, "learning_rate": 1.7766990291262135e-05, "loss": 0.1376, "step": 16940 }, { "epoch": 164.2746987951807, "grad_norm": 21.767988204956055, "learning_rate": 1.7762135922330097e-05, "loss": 0.3355, "step": 16941 }, { "epoch": 164.2843373493976, "grad_norm": 3.4951934814453125, "learning_rate": 1.7757281553398058e-05, "loss": 0.0831, "step": 16942 }, { "epoch": 164.29397590361447, "grad_norm": 26.597837448120117, "learning_rate": 1.775242718446602e-05, "loss": 0.2907, "step": 16943 }, { "epoch": 164.30361445783132, "grad_norm": 10.708125114440918, "learning_rate": 1.774757281553398e-05, "loss": 0.3193, "step": 16944 }, { "epoch": 164.3132530120482, "grad_norm": 5.8784894943237305, "learning_rate": 1.774271844660194e-05, "loss": 0.1594, "step": 16945 }, { "epoch": 164.32289156626507, "grad_norm": 14.77631950378418, "learning_rate": 1.7737864077669906e-05, "loss": 0.4474, "step": 16946 }, { "epoch": 164.33253012048192, "grad_norm": 12.18572998046875, "learning_rate": 1.7733009708737867e-05, "loss": 0.2756, "step": 16947 }, { "epoch": 164.3421686746988, "grad_norm": 11.857987403869629, "learning_rate": 1.7728155339805828e-05, "loss": 0.2809, "step": 16948 }, { "epoch": 164.35180722891567, "grad_norm": 31.57854652404785, "learning_rate": 1.772330097087379e-05, "loss": 0.1421, "step": 16949 }, { "epoch": 164.36144578313252, "grad_norm": 4.123337745666504, "learning_rate": 1.771844660194175e-05, "loss": 0.065, "step": 16950 }, { "epoch": 164.3710843373494, "grad_norm": 7.3574934005737305, "learning_rate": 1.771359223300971e-05, "loss": 0.1485, "step": 16951 }, { "epoch": 164.38072289156628, "grad_norm": 5.629595756530762, "learning_rate": 1.770873786407767e-05, "loss": 0.219, "step": 16952 }, { "epoch": 164.39036144578313, "grad_norm": 6.726696014404297, "learning_rate": 1.770388349514563e-05, "loss": 0.1683, "step": 16953 }, { "epoch": 164.4, "grad_norm": 34.29933166503906, "learning_rate": 1.769902912621359e-05, "loss": 0.2831, "step": 16954 }, { "epoch": 164.40963855421685, "grad_norm": 6.539140224456787, "learning_rate": 1.7694174757281553e-05, "loss": 0.1433, "step": 16955 }, { "epoch": 164.41927710843373, "grad_norm": 10.918834686279297, "learning_rate": 1.7689320388349514e-05, "loss": 0.3302, "step": 16956 }, { "epoch": 164.4289156626506, "grad_norm": 26.30484962463379, "learning_rate": 1.7684466019417475e-05, "loss": 0.3645, "step": 16957 }, { "epoch": 164.43855421686746, "grad_norm": 11.780969619750977, "learning_rate": 1.7679611650485436e-05, "loss": 0.2166, "step": 16958 }, { "epoch": 164.44819277108434, "grad_norm": 6.967004299163818, "learning_rate": 1.76747572815534e-05, "loss": 0.2144, "step": 16959 }, { "epoch": 164.4578313253012, "grad_norm": 11.345917701721191, "learning_rate": 1.7669902912621362e-05, "loss": 0.2426, "step": 16960 }, { "epoch": 164.46746987951806, "grad_norm": 23.19559097290039, "learning_rate": 1.7665048543689323e-05, "loss": 0.1384, "step": 16961 }, { "epoch": 164.47710843373494, "grad_norm": 6.924272060394287, "learning_rate": 1.7660194174757284e-05, "loss": 0.3651, "step": 16962 }, { "epoch": 164.48674698795182, "grad_norm": 18.095680236816406, "learning_rate": 1.7655339805825245e-05, "loss": 0.3696, "step": 16963 }, { "epoch": 164.49638554216867, "grad_norm": 15.118659973144531, "learning_rate": 1.7650485436893206e-05, "loss": 0.3823, "step": 16964 }, { "epoch": 164.50602409638554, "grad_norm": 20.73439598083496, "learning_rate": 1.7645631067961167e-05, "loss": 0.1663, "step": 16965 }, { "epoch": 164.51566265060242, "grad_norm": 18.628963470458984, "learning_rate": 1.7640776699029125e-05, "loss": 0.3033, "step": 16966 }, { "epoch": 164.52530120481927, "grad_norm": 4.69384241104126, "learning_rate": 1.7635922330097086e-05, "loss": 0.1182, "step": 16967 }, { "epoch": 164.53493975903615, "grad_norm": 7.929694652557373, "learning_rate": 1.7631067961165048e-05, "loss": 0.282, "step": 16968 }, { "epoch": 164.544578313253, "grad_norm": 11.634053230285645, "learning_rate": 1.762621359223301e-05, "loss": 0.3493, "step": 16969 }, { "epoch": 164.55421686746988, "grad_norm": 22.991666793823242, "learning_rate": 1.762135922330097e-05, "loss": 0.1164, "step": 16970 }, { "epoch": 164.56385542168675, "grad_norm": 13.066780090332031, "learning_rate": 1.761650485436893e-05, "loss": 0.2988, "step": 16971 }, { "epoch": 164.5734939759036, "grad_norm": 11.802522659301758, "learning_rate": 1.7611650485436896e-05, "loss": 0.178, "step": 16972 }, { "epoch": 164.58313253012048, "grad_norm": 12.283687591552734, "learning_rate": 1.7606796116504857e-05, "loss": 0.2338, "step": 16973 }, { "epoch": 164.59277108433736, "grad_norm": 14.216714859008789, "learning_rate": 1.7601941747572818e-05, "loss": 0.3096, "step": 16974 }, { "epoch": 164.6024096385542, "grad_norm": 5.12709379196167, "learning_rate": 1.759708737864078e-05, "loss": 0.1434, "step": 16975 }, { "epoch": 164.61204819277108, "grad_norm": 10.66263198852539, "learning_rate": 1.759223300970874e-05, "loss": 0.383, "step": 16976 }, { "epoch": 164.62168674698796, "grad_norm": 42.47065353393555, "learning_rate": 1.75873786407767e-05, "loss": 0.1551, "step": 16977 }, { "epoch": 164.6313253012048, "grad_norm": 22.98694610595703, "learning_rate": 1.7582524271844662e-05, "loss": 0.327, "step": 16978 }, { "epoch": 164.6409638554217, "grad_norm": 9.055916786193848, "learning_rate": 1.7577669902912624e-05, "loss": 0.124, "step": 16979 }, { "epoch": 164.65060240963857, "grad_norm": 8.78701400756836, "learning_rate": 1.757281553398058e-05, "loss": 0.1812, "step": 16980 }, { "epoch": 164.66024096385541, "grad_norm": 3.6650736331939697, "learning_rate": 1.7567961165048542e-05, "loss": 0.1497, "step": 16981 }, { "epoch": 164.6698795180723, "grad_norm": 12.971280097961426, "learning_rate": 1.7563106796116504e-05, "loss": 0.1879, "step": 16982 }, { "epoch": 164.67951807228917, "grad_norm": 43.8811149597168, "learning_rate": 1.7558252427184465e-05, "loss": 0.1872, "step": 16983 }, { "epoch": 164.68915662650602, "grad_norm": 12.991866111755371, "learning_rate": 1.755339805825243e-05, "loss": 0.274, "step": 16984 }, { "epoch": 164.6987951807229, "grad_norm": 4.553498268127441, "learning_rate": 1.754854368932039e-05, "loss": 0.0846, "step": 16985 }, { "epoch": 164.70843373493975, "grad_norm": 12.527621269226074, "learning_rate": 1.754368932038835e-05, "loss": 0.1809, "step": 16986 }, { "epoch": 164.71807228915662, "grad_norm": 10.857348442077637, "learning_rate": 1.7538834951456313e-05, "loss": 0.3051, "step": 16987 }, { "epoch": 164.7277108433735, "grad_norm": 14.006196022033691, "learning_rate": 1.7533980582524274e-05, "loss": 0.311, "step": 16988 }, { "epoch": 164.73734939759035, "grad_norm": 42.93828201293945, "learning_rate": 1.7529126213592235e-05, "loss": 0.3373, "step": 16989 }, { "epoch": 164.74698795180723, "grad_norm": 19.175161361694336, "learning_rate": 1.7524271844660196e-05, "loss": 0.2063, "step": 16990 }, { "epoch": 164.7566265060241, "grad_norm": 12.018815994262695, "learning_rate": 1.7519417475728157e-05, "loss": 0.3594, "step": 16991 }, { "epoch": 164.76626506024095, "grad_norm": 6.6565399169921875, "learning_rate": 1.751456310679612e-05, "loss": 0.2398, "step": 16992 }, { "epoch": 164.77590361445783, "grad_norm": 19.003244400024414, "learning_rate": 1.7509708737864076e-05, "loss": 0.2774, "step": 16993 }, { "epoch": 164.7855421686747, "grad_norm": 13.205500602722168, "learning_rate": 1.7504854368932037e-05, "loss": 0.2237, "step": 16994 }, { "epoch": 164.79518072289156, "grad_norm": 7.777691841125488, "learning_rate": 1.75e-05, "loss": 0.1179, "step": 16995 }, { "epoch": 164.80481927710844, "grad_norm": 24.276844024658203, "learning_rate": 1.749514563106796e-05, "loss": 0.1543, "step": 16996 }, { "epoch": 164.8144578313253, "grad_norm": 10.611364364624023, "learning_rate": 1.7490291262135924e-05, "loss": 0.2949, "step": 16997 }, { "epoch": 164.82409638554216, "grad_norm": 6.4392900466918945, "learning_rate": 1.7485436893203885e-05, "loss": 0.176, "step": 16998 }, { "epoch": 164.83373493975904, "grad_norm": 15.300008773803711, "learning_rate": 1.7480582524271846e-05, "loss": 0.2568, "step": 16999 }, { "epoch": 164.84337349397592, "grad_norm": 16.999221801757812, "learning_rate": 1.7475728155339808e-05, "loss": 0.3307, "step": 17000 }, { "epoch": 164.85301204819277, "grad_norm": 14.567344665527344, "learning_rate": 1.747087378640777e-05, "loss": 0.2341, "step": 17001 }, { "epoch": 164.86265060240964, "grad_norm": 4.6853556632995605, "learning_rate": 1.746601941747573e-05, "loss": 0.2077, "step": 17002 }, { "epoch": 164.8722891566265, "grad_norm": 11.149925231933594, "learning_rate": 1.746116504854369e-05, "loss": 0.3258, "step": 17003 }, { "epoch": 164.88192771084337, "grad_norm": 32.6441650390625, "learning_rate": 1.7456310679611652e-05, "loss": 0.2708, "step": 17004 }, { "epoch": 164.89156626506025, "grad_norm": 12.145599365234375, "learning_rate": 1.7451456310679613e-05, "loss": 0.2021, "step": 17005 }, { "epoch": 164.9012048192771, "grad_norm": 7.979155540466309, "learning_rate": 1.7446601941747574e-05, "loss": 0.2013, "step": 17006 }, { "epoch": 164.91084337349398, "grad_norm": 7.538825511932373, "learning_rate": 1.7441747572815532e-05, "loss": 0.1153, "step": 17007 }, { "epoch": 164.92048192771085, "grad_norm": 9.103312492370605, "learning_rate": 1.7436893203883493e-05, "loss": 0.2548, "step": 17008 }, { "epoch": 164.9301204819277, "grad_norm": 69.44871520996094, "learning_rate": 1.7432038834951458e-05, "loss": 0.3079, "step": 17009 }, { "epoch": 164.93975903614458, "grad_norm": 27.00832748413086, "learning_rate": 1.742718446601942e-05, "loss": 0.2915, "step": 17010 }, { "epoch": 164.94939759036146, "grad_norm": 34.37921905517578, "learning_rate": 1.742233009708738e-05, "loss": 0.2702, "step": 17011 }, { "epoch": 164.9590361445783, "grad_norm": 8.346539497375488, "learning_rate": 1.741747572815534e-05, "loss": 0.2776, "step": 17012 }, { "epoch": 164.96867469879518, "grad_norm": 39.714508056640625, "learning_rate": 1.7412621359223302e-05, "loss": 0.3814, "step": 17013 }, { "epoch": 164.97831325301206, "grad_norm": 6.4240570068359375, "learning_rate": 1.7407766990291264e-05, "loss": 0.1983, "step": 17014 }, { "epoch": 164.9879518072289, "grad_norm": 19.96578025817871, "learning_rate": 1.7402912621359225e-05, "loss": 0.4655, "step": 17015 }, { "epoch": 165.00361445783133, "grad_norm": 17.283164978027344, "learning_rate": 1.7398058252427186e-05, "loss": 0.1834, "step": 17016 }, { "epoch": 165.01325301204818, "grad_norm": 19.74701499938965, "learning_rate": 1.7393203883495147e-05, "loss": 0.1361, "step": 17017 }, { "epoch": 165.02289156626506, "grad_norm": 4.990596771240234, "learning_rate": 1.7388349514563108e-05, "loss": 0.2351, "step": 17018 }, { "epoch": 165.03253012048194, "grad_norm": 6.641876697540283, "learning_rate": 1.738349514563107e-05, "loss": 0.1401, "step": 17019 }, { "epoch": 165.0421686746988, "grad_norm": 19.649404525756836, "learning_rate": 1.737864077669903e-05, "loss": 0.2781, "step": 17020 }, { "epoch": 165.05180722891566, "grad_norm": 6.9723029136657715, "learning_rate": 1.7373786407766988e-05, "loss": 0.1908, "step": 17021 }, { "epoch": 165.06144578313254, "grad_norm": 7.760613441467285, "learning_rate": 1.7368932038834953e-05, "loss": 0.2526, "step": 17022 }, { "epoch": 165.0710843373494, "grad_norm": 20.92886734008789, "learning_rate": 1.7364077669902914e-05, "loss": 0.3607, "step": 17023 }, { "epoch": 165.08072289156627, "grad_norm": 17.060468673706055, "learning_rate": 1.7359223300970875e-05, "loss": 0.3295, "step": 17024 }, { "epoch": 165.09036144578315, "grad_norm": 17.538715362548828, "learning_rate": 1.7354368932038836e-05, "loss": 0.2418, "step": 17025 }, { "epoch": 165.1, "grad_norm": 11.339903831481934, "learning_rate": 1.7349514563106797e-05, "loss": 0.2501, "step": 17026 }, { "epoch": 165.10963855421687, "grad_norm": 6.4115986824035645, "learning_rate": 1.734466019417476e-05, "loss": 0.1751, "step": 17027 }, { "epoch": 165.11927710843372, "grad_norm": 19.135663986206055, "learning_rate": 1.733980582524272e-05, "loss": 0.1794, "step": 17028 }, { "epoch": 165.1289156626506, "grad_norm": 12.055861473083496, "learning_rate": 1.733495145631068e-05, "loss": 0.3434, "step": 17029 }, { "epoch": 165.13855421686748, "grad_norm": 3.265780448913574, "learning_rate": 1.7330097087378642e-05, "loss": 0.0694, "step": 17030 }, { "epoch": 165.14819277108433, "grad_norm": 4.0439934730529785, "learning_rate": 1.7325242718446603e-05, "loss": 0.0587, "step": 17031 }, { "epoch": 165.1578313253012, "grad_norm": 20.947298049926758, "learning_rate": 1.7320388349514564e-05, "loss": 0.2639, "step": 17032 }, { "epoch": 165.16746987951808, "grad_norm": 15.326926231384277, "learning_rate": 1.7315533980582525e-05, "loss": 0.4597, "step": 17033 }, { "epoch": 165.17710843373493, "grad_norm": 8.4259033203125, "learning_rate": 1.7310679611650487e-05, "loss": 0.1221, "step": 17034 }, { "epoch": 165.1867469879518, "grad_norm": 33.41478729248047, "learning_rate": 1.7305825242718448e-05, "loss": 0.3118, "step": 17035 }, { "epoch": 165.19638554216868, "grad_norm": 15.793548583984375, "learning_rate": 1.730097087378641e-05, "loss": 0.2946, "step": 17036 }, { "epoch": 165.20602409638553, "grad_norm": 5.727783203125, "learning_rate": 1.729611650485437e-05, "loss": 0.1334, "step": 17037 }, { "epoch": 165.2156626506024, "grad_norm": 22.076581954956055, "learning_rate": 1.729126213592233e-05, "loss": 0.4737, "step": 17038 }, { "epoch": 165.2253012048193, "grad_norm": 10.622539520263672, "learning_rate": 1.7286407766990292e-05, "loss": 0.2296, "step": 17039 }, { "epoch": 165.23493975903614, "grad_norm": 6.012299537658691, "learning_rate": 1.7281553398058253e-05, "loss": 0.2158, "step": 17040 }, { "epoch": 165.24457831325302, "grad_norm": 20.340295791625977, "learning_rate": 1.7276699029126215e-05, "loss": 0.2088, "step": 17041 }, { "epoch": 165.2542168674699, "grad_norm": 31.702064514160156, "learning_rate": 1.7271844660194176e-05, "loss": 0.427, "step": 17042 }, { "epoch": 165.26385542168674, "grad_norm": 11.501044273376465, "learning_rate": 1.7266990291262137e-05, "loss": 0.2865, "step": 17043 }, { "epoch": 165.27349397590362, "grad_norm": 17.192771911621094, "learning_rate": 1.7262135922330098e-05, "loss": 0.4641, "step": 17044 }, { "epoch": 165.28313253012047, "grad_norm": 8.538010597229004, "learning_rate": 1.725728155339806e-05, "loss": 0.3168, "step": 17045 }, { "epoch": 165.29277108433735, "grad_norm": 10.931157112121582, "learning_rate": 1.725242718446602e-05, "loss": 0.2018, "step": 17046 }, { "epoch": 165.30240963855422, "grad_norm": 18.152206420898438, "learning_rate": 1.724757281553398e-05, "loss": 0.3634, "step": 17047 }, { "epoch": 165.31204819277107, "grad_norm": 7.876725196838379, "learning_rate": 1.7242718446601943e-05, "loss": 0.19, "step": 17048 }, { "epoch": 165.32168674698795, "grad_norm": 16.519107818603516, "learning_rate": 1.7237864077669904e-05, "loss": 0.3014, "step": 17049 }, { "epoch": 165.33132530120483, "grad_norm": 7.449148178100586, "learning_rate": 1.7233009708737865e-05, "loss": 0.1038, "step": 17050 }, { "epoch": 165.34096385542168, "grad_norm": 23.202003479003906, "learning_rate": 1.7228155339805826e-05, "loss": 0.2009, "step": 17051 }, { "epoch": 165.35060240963855, "grad_norm": 7.929396629333496, "learning_rate": 1.7223300970873787e-05, "loss": 0.314, "step": 17052 }, { "epoch": 165.36024096385543, "grad_norm": 3.809725046157837, "learning_rate": 1.721844660194175e-05, "loss": 0.1066, "step": 17053 }, { "epoch": 165.36987951807228, "grad_norm": 12.85799789428711, "learning_rate": 1.721359223300971e-05, "loss": 0.1335, "step": 17054 }, { "epoch": 165.37951807228916, "grad_norm": 8.667706489562988, "learning_rate": 1.720873786407767e-05, "loss": 0.2401, "step": 17055 }, { "epoch": 165.38915662650604, "grad_norm": 4.096334934234619, "learning_rate": 1.7203883495145632e-05, "loss": 0.3241, "step": 17056 }, { "epoch": 165.39879518072289, "grad_norm": 6.65045166015625, "learning_rate": 1.7199029126213593e-05, "loss": 0.2553, "step": 17057 }, { "epoch": 165.40843373493976, "grad_norm": 8.083460807800293, "learning_rate": 1.7194174757281554e-05, "loss": 0.2415, "step": 17058 }, { "epoch": 165.41807228915664, "grad_norm": 17.94443130493164, "learning_rate": 1.7189320388349515e-05, "loss": 0.2446, "step": 17059 }, { "epoch": 165.4277108433735, "grad_norm": 15.53283977508545, "learning_rate": 1.7184466019417476e-05, "loss": 0.3278, "step": 17060 }, { "epoch": 165.43734939759037, "grad_norm": 11.688508987426758, "learning_rate": 1.7179611650485437e-05, "loss": 0.5314, "step": 17061 }, { "epoch": 165.44698795180722, "grad_norm": 26.384538650512695, "learning_rate": 1.71747572815534e-05, "loss": 0.4811, "step": 17062 }, { "epoch": 165.4566265060241, "grad_norm": 11.156020164489746, "learning_rate": 1.716990291262136e-05, "loss": 0.2364, "step": 17063 }, { "epoch": 165.46626506024097, "grad_norm": 7.366277694702148, "learning_rate": 1.716504854368932e-05, "loss": 0.2135, "step": 17064 }, { "epoch": 165.47590361445782, "grad_norm": 4.927035808563232, "learning_rate": 1.7160194174757282e-05, "loss": 0.1999, "step": 17065 }, { "epoch": 165.4855421686747, "grad_norm": 17.5922908782959, "learning_rate": 1.7155339805825243e-05, "loss": 0.2816, "step": 17066 }, { "epoch": 165.49518072289158, "grad_norm": 11.9299955368042, "learning_rate": 1.7150485436893204e-05, "loss": 0.1349, "step": 17067 }, { "epoch": 165.50481927710842, "grad_norm": 6.942302703857422, "learning_rate": 1.7145631067961165e-05, "loss": 0.3613, "step": 17068 }, { "epoch": 165.5144578313253, "grad_norm": 23.806293487548828, "learning_rate": 1.7140776699029127e-05, "loss": 0.3706, "step": 17069 }, { "epoch": 165.52409638554218, "grad_norm": 12.825868606567383, "learning_rate": 1.7135922330097088e-05, "loss": 0.3487, "step": 17070 }, { "epoch": 165.53373493975903, "grad_norm": 16.630908966064453, "learning_rate": 1.713106796116505e-05, "loss": 0.3697, "step": 17071 }, { "epoch": 165.5433734939759, "grad_norm": 5.417672634124756, "learning_rate": 1.7126213592233013e-05, "loss": 0.4362, "step": 17072 }, { "epoch": 165.55301204819278, "grad_norm": 7.115645408630371, "learning_rate": 1.712135922330097e-05, "loss": 0.2624, "step": 17073 }, { "epoch": 165.56265060240963, "grad_norm": 14.503318786621094, "learning_rate": 1.7116504854368932e-05, "loss": 0.2042, "step": 17074 }, { "epoch": 165.5722891566265, "grad_norm": 20.629568099975586, "learning_rate": 1.7111650485436894e-05, "loss": 0.1991, "step": 17075 }, { "epoch": 165.58192771084336, "grad_norm": 8.94167709350586, "learning_rate": 1.7106796116504855e-05, "loss": 0.2558, "step": 17076 }, { "epoch": 165.59156626506024, "grad_norm": 7.6561384201049805, "learning_rate": 1.7101941747572816e-05, "loss": 0.1978, "step": 17077 }, { "epoch": 165.60120481927711, "grad_norm": 14.993860244750977, "learning_rate": 1.7097087378640777e-05, "loss": 0.29, "step": 17078 }, { "epoch": 165.61084337349396, "grad_norm": 8.670059204101562, "learning_rate": 1.7092233009708738e-05, "loss": 0.2058, "step": 17079 }, { "epoch": 165.62048192771084, "grad_norm": 20.98904800415039, "learning_rate": 1.70873786407767e-05, "loss": 0.3017, "step": 17080 }, { "epoch": 165.63012048192772, "grad_norm": 17.482555389404297, "learning_rate": 1.708252427184466e-05, "loss": 0.288, "step": 17081 }, { "epoch": 165.63975903614457, "grad_norm": 16.539894104003906, "learning_rate": 1.707766990291262e-05, "loss": 0.1283, "step": 17082 }, { "epoch": 165.64939759036145, "grad_norm": 23.08771324157715, "learning_rate": 1.7072815533980583e-05, "loss": 0.305, "step": 17083 }, { "epoch": 165.65903614457832, "grad_norm": 8.26163387298584, "learning_rate": 1.7067961165048544e-05, "loss": 0.0635, "step": 17084 }, { "epoch": 165.66867469879517, "grad_norm": 4.047384738922119, "learning_rate": 1.706310679611651e-05, "loss": 0.1261, "step": 17085 }, { "epoch": 165.67831325301205, "grad_norm": 6.545888423919678, "learning_rate": 1.705825242718447e-05, "loss": 0.1529, "step": 17086 }, { "epoch": 165.68795180722893, "grad_norm": 9.287796020507812, "learning_rate": 1.7053398058252427e-05, "loss": 0.3111, "step": 17087 }, { "epoch": 165.69759036144578, "grad_norm": 27.463764190673828, "learning_rate": 1.704854368932039e-05, "loss": 0.247, "step": 17088 }, { "epoch": 165.70722891566265, "grad_norm": 4.428330898284912, "learning_rate": 1.704368932038835e-05, "loss": 0.1409, "step": 17089 }, { "epoch": 165.71686746987953, "grad_norm": 6.510654926300049, "learning_rate": 1.703883495145631e-05, "loss": 0.1742, "step": 17090 }, { "epoch": 165.72650602409638, "grad_norm": 27.88448143005371, "learning_rate": 1.7033980582524272e-05, "loss": 0.3028, "step": 17091 }, { "epoch": 165.73614457831326, "grad_norm": 6.528670310974121, "learning_rate": 1.7029126213592233e-05, "loss": 0.2204, "step": 17092 }, { "epoch": 165.7457831325301, "grad_norm": 14.11075496673584, "learning_rate": 1.7024271844660194e-05, "loss": 0.3337, "step": 17093 }, { "epoch": 165.75542168674698, "grad_norm": 26.969425201416016, "learning_rate": 1.7019417475728155e-05, "loss": 0.3586, "step": 17094 }, { "epoch": 165.76506024096386, "grad_norm": 3.8282551765441895, "learning_rate": 1.7014563106796116e-05, "loss": 0.0874, "step": 17095 }, { "epoch": 165.7746987951807, "grad_norm": 4.475448131561279, "learning_rate": 1.7009708737864078e-05, "loss": 0.087, "step": 17096 }, { "epoch": 165.7843373493976, "grad_norm": 16.629179000854492, "learning_rate": 1.700485436893204e-05, "loss": 0.1813, "step": 17097 }, { "epoch": 165.79397590361447, "grad_norm": 20.2169189453125, "learning_rate": 1.7000000000000003e-05, "loss": 0.193, "step": 17098 }, { "epoch": 165.80361445783132, "grad_norm": 20.56793212890625, "learning_rate": 1.6995145631067964e-05, "loss": 0.3266, "step": 17099 }, { "epoch": 165.8132530120482, "grad_norm": 12.7813081741333, "learning_rate": 1.6990291262135926e-05, "loss": 0.266, "step": 17100 }, { "epoch": 165.82289156626507, "grad_norm": 19.051372528076172, "learning_rate": 1.6985436893203883e-05, "loss": 0.2471, "step": 17101 }, { "epoch": 165.83253012048192, "grad_norm": 6.609585285186768, "learning_rate": 1.6980582524271844e-05, "loss": 0.1586, "step": 17102 }, { "epoch": 165.8421686746988, "grad_norm": 7.902173042297363, "learning_rate": 1.6975728155339806e-05, "loss": 0.2445, "step": 17103 }, { "epoch": 165.85180722891567, "grad_norm": 16.364309310913086, "learning_rate": 1.6970873786407767e-05, "loss": 0.2877, "step": 17104 }, { "epoch": 165.86144578313252, "grad_norm": 39.213539123535156, "learning_rate": 1.6966019417475728e-05, "loss": 0.1316, "step": 17105 }, { "epoch": 165.8710843373494, "grad_norm": 15.619492530822754, "learning_rate": 1.696116504854369e-05, "loss": 0.3468, "step": 17106 }, { "epoch": 165.88072289156628, "grad_norm": 20.469160079956055, "learning_rate": 1.695631067961165e-05, "loss": 0.2971, "step": 17107 }, { "epoch": 165.89036144578313, "grad_norm": 3.7672324180603027, "learning_rate": 1.695145631067961e-05, "loss": 0.1794, "step": 17108 }, { "epoch": 165.9, "grad_norm": 15.052447319030762, "learning_rate": 1.6946601941747572e-05, "loss": 0.3165, "step": 17109 }, { "epoch": 165.90963855421685, "grad_norm": 13.06093692779541, "learning_rate": 1.6941747572815537e-05, "loss": 0.3945, "step": 17110 }, { "epoch": 165.91927710843373, "grad_norm": 7.725008487701416, "learning_rate": 1.6936893203883498e-05, "loss": 0.1851, "step": 17111 }, { "epoch": 165.9289156626506, "grad_norm": 13.100247383117676, "learning_rate": 1.693203883495146e-05, "loss": 0.3275, "step": 17112 }, { "epoch": 165.93855421686746, "grad_norm": 12.557066917419434, "learning_rate": 1.692718446601942e-05, "loss": 0.166, "step": 17113 }, { "epoch": 165.94819277108434, "grad_norm": 8.517034530639648, "learning_rate": 1.692233009708738e-05, "loss": 0.2813, "step": 17114 }, { "epoch": 165.9578313253012, "grad_norm": 29.229639053344727, "learning_rate": 1.691747572815534e-05, "loss": 0.1641, "step": 17115 }, { "epoch": 165.96746987951806, "grad_norm": 47.241676330566406, "learning_rate": 1.69126213592233e-05, "loss": 0.3047, "step": 17116 }, { "epoch": 165.97710843373494, "grad_norm": 32.573333740234375, "learning_rate": 1.690776699029126e-05, "loss": 0.288, "step": 17117 }, { "epoch": 165.98674698795182, "grad_norm": 25.709941864013672, "learning_rate": 1.6902912621359223e-05, "loss": 0.2489, "step": 17118 }, { "epoch": 166.0024096385542, "grad_norm": 8.518110275268555, "learning_rate": 1.6898058252427184e-05, "loss": 0.2087, "step": 17119 }, { "epoch": 166.0120481927711, "grad_norm": 15.689135551452637, "learning_rate": 1.6893203883495145e-05, "loss": 0.2202, "step": 17120 }, { "epoch": 166.02168674698794, "grad_norm": 7.220540523529053, "learning_rate": 1.6888349514563106e-05, "loss": 0.1298, "step": 17121 }, { "epoch": 166.03132530120482, "grad_norm": 14.096612930297852, "learning_rate": 1.6883495145631067e-05, "loss": 0.5238, "step": 17122 }, { "epoch": 166.0409638554217, "grad_norm": 24.043514251708984, "learning_rate": 1.6878640776699032e-05, "loss": 0.2155, "step": 17123 }, { "epoch": 166.05060240963854, "grad_norm": 16.067337036132812, "learning_rate": 1.6873786407766993e-05, "loss": 0.4436, "step": 17124 }, { "epoch": 166.06024096385542, "grad_norm": 12.91761589050293, "learning_rate": 1.6868932038834954e-05, "loss": 0.335, "step": 17125 }, { "epoch": 166.0698795180723, "grad_norm": 8.893909454345703, "learning_rate": 1.6864077669902915e-05, "loss": 0.2632, "step": 17126 }, { "epoch": 166.07951807228915, "grad_norm": 10.680176734924316, "learning_rate": 1.6859223300970876e-05, "loss": 0.2742, "step": 17127 }, { "epoch": 166.08915662650602, "grad_norm": 11.05196475982666, "learning_rate": 1.6854368932038838e-05, "loss": 0.4168, "step": 17128 }, { "epoch": 166.0987951807229, "grad_norm": 3.3755545616149902, "learning_rate": 1.6849514563106795e-05, "loss": 0.129, "step": 17129 }, { "epoch": 166.10843373493975, "grad_norm": 22.508825302124023, "learning_rate": 1.6844660194174757e-05, "loss": 0.2998, "step": 17130 }, { "epoch": 166.11807228915663, "grad_norm": 20.52454376220703, "learning_rate": 1.6839805825242718e-05, "loss": 0.2726, "step": 17131 }, { "epoch": 166.1277108433735, "grad_norm": 7.4451751708984375, "learning_rate": 1.683495145631068e-05, "loss": 0.2489, "step": 17132 }, { "epoch": 166.13734939759036, "grad_norm": 18.02910614013672, "learning_rate": 1.683009708737864e-05, "loss": 0.4328, "step": 17133 }, { "epoch": 166.14698795180723, "grad_norm": 14.86246109008789, "learning_rate": 1.68252427184466e-05, "loss": 0.377, "step": 17134 }, { "epoch": 166.15662650602408, "grad_norm": 2.4167001247406006, "learning_rate": 1.6820388349514562e-05, "loss": 0.06, "step": 17135 }, { "epoch": 166.16626506024096, "grad_norm": 24.052457809448242, "learning_rate": 1.6815533980582527e-05, "loss": 0.3078, "step": 17136 }, { "epoch": 166.17590361445784, "grad_norm": 37.88314437866211, "learning_rate": 1.6810679611650488e-05, "loss": 0.1996, "step": 17137 }, { "epoch": 166.1855421686747, "grad_norm": 7.085541725158691, "learning_rate": 1.680582524271845e-05, "loss": 0.163, "step": 17138 }, { "epoch": 166.19518072289156, "grad_norm": 7.5973615646362305, "learning_rate": 1.680097087378641e-05, "loss": 0.2466, "step": 17139 }, { "epoch": 166.20481927710844, "grad_norm": 10.36884593963623, "learning_rate": 1.679611650485437e-05, "loss": 0.2625, "step": 17140 }, { "epoch": 166.2144578313253, "grad_norm": 4.021109580993652, "learning_rate": 1.6791262135922333e-05, "loss": 0.0786, "step": 17141 }, { "epoch": 166.22409638554217, "grad_norm": 3.250117063522339, "learning_rate": 1.678640776699029e-05, "loss": 0.0818, "step": 17142 }, { "epoch": 166.23373493975905, "grad_norm": 7.235009670257568, "learning_rate": 1.678155339805825e-05, "loss": 0.1858, "step": 17143 }, { "epoch": 166.2433734939759, "grad_norm": 20.28384017944336, "learning_rate": 1.6776699029126213e-05, "loss": 0.3384, "step": 17144 }, { "epoch": 166.25301204819277, "grad_norm": 16.908750534057617, "learning_rate": 1.6771844660194174e-05, "loss": 0.2139, "step": 17145 }, { "epoch": 166.26265060240965, "grad_norm": 6.105428218841553, "learning_rate": 1.6766990291262135e-05, "loss": 0.1948, "step": 17146 }, { "epoch": 166.2722891566265, "grad_norm": 8.543171882629395, "learning_rate": 1.6762135922330096e-05, "loss": 0.386, "step": 17147 }, { "epoch": 166.28192771084338, "grad_norm": 24.903554916381836, "learning_rate": 1.675728155339806e-05, "loss": 0.3137, "step": 17148 }, { "epoch": 166.29156626506025, "grad_norm": 12.176636695861816, "learning_rate": 1.675242718446602e-05, "loss": 0.2108, "step": 17149 }, { "epoch": 166.3012048192771, "grad_norm": 6.149503707885742, "learning_rate": 1.6747572815533983e-05, "loss": 0.1387, "step": 17150 }, { "epoch": 166.31084337349398, "grad_norm": 23.287757873535156, "learning_rate": 1.6742718446601944e-05, "loss": 0.3829, "step": 17151 }, { "epoch": 166.32048192771083, "grad_norm": 9.272951126098633, "learning_rate": 1.6737864077669905e-05, "loss": 0.1873, "step": 17152 }, { "epoch": 166.3301204819277, "grad_norm": 6.093706130981445, "learning_rate": 1.6733009708737866e-05, "loss": 0.1334, "step": 17153 }, { "epoch": 166.33975903614459, "grad_norm": 12.106916427612305, "learning_rate": 1.6728155339805827e-05, "loss": 0.2351, "step": 17154 }, { "epoch": 166.34939759036143, "grad_norm": 5.646626949310303, "learning_rate": 1.672330097087379e-05, "loss": 0.1936, "step": 17155 }, { "epoch": 166.3590361445783, "grad_norm": 6.517972469329834, "learning_rate": 1.6718446601941746e-05, "loss": 0.2608, "step": 17156 }, { "epoch": 166.3686746987952, "grad_norm": 25.060598373413086, "learning_rate": 1.6713592233009707e-05, "loss": 0.3431, "step": 17157 }, { "epoch": 166.37831325301204, "grad_norm": 8.40280818939209, "learning_rate": 1.670873786407767e-05, "loss": 0.2251, "step": 17158 }, { "epoch": 166.38795180722892, "grad_norm": 14.051453590393066, "learning_rate": 1.670388349514563e-05, "loss": 0.3665, "step": 17159 }, { "epoch": 166.3975903614458, "grad_norm": 14.443265914916992, "learning_rate": 1.669902912621359e-05, "loss": 0.2699, "step": 17160 }, { "epoch": 166.40722891566264, "grad_norm": 8.006438255310059, "learning_rate": 1.6694174757281555e-05, "loss": 0.2561, "step": 17161 }, { "epoch": 166.41686746987952, "grad_norm": 12.485121726989746, "learning_rate": 1.6689320388349517e-05, "loss": 0.2829, "step": 17162 }, { "epoch": 166.4265060240964, "grad_norm": 4.9082441329956055, "learning_rate": 1.6684466019417478e-05, "loss": 0.1521, "step": 17163 }, { "epoch": 166.43614457831325, "grad_norm": 8.766718864440918, "learning_rate": 1.667961165048544e-05, "loss": 0.2598, "step": 17164 }, { "epoch": 166.44578313253012, "grad_norm": 13.430476188659668, "learning_rate": 1.66747572815534e-05, "loss": 0.2677, "step": 17165 }, { "epoch": 166.455421686747, "grad_norm": 12.61070442199707, "learning_rate": 1.666990291262136e-05, "loss": 0.3623, "step": 17166 }, { "epoch": 166.46506024096385, "grad_norm": 63.78435134887695, "learning_rate": 1.6665048543689322e-05, "loss": 0.2199, "step": 17167 }, { "epoch": 166.47469879518073, "grad_norm": 2.676962375640869, "learning_rate": 1.6660194174757283e-05, "loss": 0.1801, "step": 17168 }, { "epoch": 166.48433734939758, "grad_norm": 16.903112411499023, "learning_rate": 1.6655339805825245e-05, "loss": 0.2298, "step": 17169 }, { "epoch": 166.49397590361446, "grad_norm": 15.494197845458984, "learning_rate": 1.6650485436893202e-05, "loss": 0.3563, "step": 17170 }, { "epoch": 166.50361445783133, "grad_norm": 8.150042533874512, "learning_rate": 1.6645631067961164e-05, "loss": 0.1115, "step": 17171 }, { "epoch": 166.51325301204818, "grad_norm": 12.206354141235352, "learning_rate": 1.6640776699029125e-05, "loss": 0.1968, "step": 17172 }, { "epoch": 166.52289156626506, "grad_norm": 21.9156494140625, "learning_rate": 1.663592233009709e-05, "loss": 0.228, "step": 17173 }, { "epoch": 166.53253012048194, "grad_norm": 21.973888397216797, "learning_rate": 1.663106796116505e-05, "loss": 0.4926, "step": 17174 }, { "epoch": 166.5421686746988, "grad_norm": 8.718043327331543, "learning_rate": 1.662621359223301e-05, "loss": 0.1666, "step": 17175 }, { "epoch": 166.55180722891566, "grad_norm": 5.90787410736084, "learning_rate": 1.6621359223300973e-05, "loss": 0.2928, "step": 17176 }, { "epoch": 166.56144578313254, "grad_norm": 7.1161723136901855, "learning_rate": 1.6616504854368934e-05, "loss": 0.1316, "step": 17177 }, { "epoch": 166.5710843373494, "grad_norm": 9.19204044342041, "learning_rate": 1.6611650485436895e-05, "loss": 0.1411, "step": 17178 }, { "epoch": 166.58072289156627, "grad_norm": 20.14609718322754, "learning_rate": 1.6606796116504856e-05, "loss": 0.2978, "step": 17179 }, { "epoch": 166.59036144578315, "grad_norm": 12.069332122802734, "learning_rate": 1.6601941747572817e-05, "loss": 0.3337, "step": 17180 }, { "epoch": 166.6, "grad_norm": 11.77859115600586, "learning_rate": 1.659708737864078e-05, "loss": 0.305, "step": 17181 }, { "epoch": 166.60963855421687, "grad_norm": 49.19785690307617, "learning_rate": 1.659223300970874e-05, "loss": 0.1633, "step": 17182 }, { "epoch": 166.61927710843372, "grad_norm": 8.626887321472168, "learning_rate": 1.65873786407767e-05, "loss": 0.2629, "step": 17183 }, { "epoch": 166.6289156626506, "grad_norm": 23.653608322143555, "learning_rate": 1.658252427184466e-05, "loss": 0.2214, "step": 17184 }, { "epoch": 166.63855421686748, "grad_norm": 5.397153854370117, "learning_rate": 1.657766990291262e-05, "loss": 0.1052, "step": 17185 }, { "epoch": 166.64819277108433, "grad_norm": 7.190786838531494, "learning_rate": 1.6572815533980584e-05, "loss": 0.2962, "step": 17186 }, { "epoch": 166.6578313253012, "grad_norm": 5.467103481292725, "learning_rate": 1.6567961165048545e-05, "loss": 0.2004, "step": 17187 }, { "epoch": 166.66746987951808, "grad_norm": 16.288570404052734, "learning_rate": 1.6563106796116506e-05, "loss": 0.2953, "step": 17188 }, { "epoch": 166.67710843373493, "grad_norm": 15.889790534973145, "learning_rate": 1.6558252427184468e-05, "loss": 0.2993, "step": 17189 }, { "epoch": 166.6867469879518, "grad_norm": 13.732921600341797, "learning_rate": 1.655339805825243e-05, "loss": 0.3607, "step": 17190 }, { "epoch": 166.69638554216868, "grad_norm": 6.2911200523376465, "learning_rate": 1.654854368932039e-05, "loss": 0.1605, "step": 17191 }, { "epoch": 166.70602409638553, "grad_norm": 9.226794242858887, "learning_rate": 1.654368932038835e-05, "loss": 0.1765, "step": 17192 }, { "epoch": 166.7156626506024, "grad_norm": 3.4519917964935303, "learning_rate": 1.6538834951456312e-05, "loss": 0.1873, "step": 17193 }, { "epoch": 166.7253012048193, "grad_norm": 4.980487823486328, "learning_rate": 1.6533980582524273e-05, "loss": 0.1891, "step": 17194 }, { "epoch": 166.73493975903614, "grad_norm": 6.9213690757751465, "learning_rate": 1.6529126213592234e-05, "loss": 0.2511, "step": 17195 }, { "epoch": 166.74457831325302, "grad_norm": 7.454799652099609, "learning_rate": 1.6524271844660196e-05, "loss": 0.1487, "step": 17196 }, { "epoch": 166.7542168674699, "grad_norm": 9.603608131408691, "learning_rate": 1.6519417475728157e-05, "loss": 0.1672, "step": 17197 }, { "epoch": 166.76385542168674, "grad_norm": 9.32251262664795, "learning_rate": 1.6514563106796114e-05, "loss": 0.2262, "step": 17198 }, { "epoch": 166.77349397590362, "grad_norm": 41.48992919921875, "learning_rate": 1.650970873786408e-05, "loss": 0.4776, "step": 17199 }, { "epoch": 166.78313253012047, "grad_norm": 10.796630859375, "learning_rate": 1.650485436893204e-05, "loss": 0.3387, "step": 17200 }, { "epoch": 166.79277108433735, "grad_norm": 4.454041004180908, "learning_rate": 1.65e-05, "loss": 0.2982, "step": 17201 }, { "epoch": 166.80240963855422, "grad_norm": 3.9542815685272217, "learning_rate": 1.6495145631067962e-05, "loss": 0.1356, "step": 17202 }, { "epoch": 166.81204819277107, "grad_norm": 20.5218505859375, "learning_rate": 1.6490291262135924e-05, "loss": 0.2855, "step": 17203 }, { "epoch": 166.82168674698795, "grad_norm": 9.476465225219727, "learning_rate": 1.6485436893203885e-05, "loss": 0.1539, "step": 17204 }, { "epoch": 166.83132530120483, "grad_norm": 8.534408569335938, "learning_rate": 1.6480582524271846e-05, "loss": 0.272, "step": 17205 }, { "epoch": 166.84096385542168, "grad_norm": 8.247655868530273, "learning_rate": 1.6475728155339807e-05, "loss": 0.1952, "step": 17206 }, { "epoch": 166.85060240963855, "grad_norm": 9.369516372680664, "learning_rate": 1.6470873786407768e-05, "loss": 0.2853, "step": 17207 }, { "epoch": 166.86024096385543, "grad_norm": 5.805045127868652, "learning_rate": 1.646601941747573e-05, "loss": 0.151, "step": 17208 }, { "epoch": 166.86987951807228, "grad_norm": 3.4071624279022217, "learning_rate": 1.646116504854369e-05, "loss": 0.1174, "step": 17209 }, { "epoch": 166.87951807228916, "grad_norm": 26.556880950927734, "learning_rate": 1.645631067961165e-05, "loss": 0.3558, "step": 17210 }, { "epoch": 166.88915662650604, "grad_norm": 7.22654914855957, "learning_rate": 1.6451456310679613e-05, "loss": 0.257, "step": 17211 }, { "epoch": 166.89879518072289, "grad_norm": 9.375659942626953, "learning_rate": 1.6446601941747574e-05, "loss": 0.2708, "step": 17212 }, { "epoch": 166.90843373493976, "grad_norm": 18.873497009277344, "learning_rate": 1.6441747572815535e-05, "loss": 0.2513, "step": 17213 }, { "epoch": 166.91807228915664, "grad_norm": 5.743538856506348, "learning_rate": 1.6436893203883496e-05, "loss": 0.1951, "step": 17214 }, { "epoch": 166.9277108433735, "grad_norm": 16.624462127685547, "learning_rate": 1.6432038834951457e-05, "loss": 0.2197, "step": 17215 }, { "epoch": 166.93734939759037, "grad_norm": 17.573936462402344, "learning_rate": 1.642718446601942e-05, "loss": 0.3537, "step": 17216 }, { "epoch": 166.94698795180722, "grad_norm": 10.119935035705566, "learning_rate": 1.642233009708738e-05, "loss": 0.2989, "step": 17217 }, { "epoch": 166.9566265060241, "grad_norm": 6.757495880126953, "learning_rate": 1.641747572815534e-05, "loss": 0.3069, "step": 17218 }, { "epoch": 166.96626506024097, "grad_norm": 7.080716133117676, "learning_rate": 1.6412621359223302e-05, "loss": 0.1233, "step": 17219 }, { "epoch": 166.97590361445782, "grad_norm": 5.953092098236084, "learning_rate": 1.6407766990291263e-05, "loss": 0.1158, "step": 17220 }, { "epoch": 166.9855421686747, "grad_norm": 21.687423706054688, "learning_rate": 1.6402912621359224e-05, "loss": 0.2111, "step": 17221 }, { "epoch": 167.00120481927712, "grad_norm": 6.70122766494751, "learning_rate": 1.6398058252427185e-05, "loss": 0.4375, "step": 17222 }, { "epoch": 167.01084337349397, "grad_norm": 6.305485725402832, "learning_rate": 1.6393203883495146e-05, "loss": 0.3206, "step": 17223 }, { "epoch": 167.02048192771085, "grad_norm": 14.10708999633789, "learning_rate": 1.6388349514563108e-05, "loss": 0.2935, "step": 17224 }, { "epoch": 167.03012048192772, "grad_norm": 5.640683174133301, "learning_rate": 1.638349514563107e-05, "loss": 0.1742, "step": 17225 }, { "epoch": 167.03975903614457, "grad_norm": 6.973128318786621, "learning_rate": 1.637864077669903e-05, "loss": 0.1543, "step": 17226 }, { "epoch": 167.04939759036145, "grad_norm": 3.6080081462860107, "learning_rate": 1.637378640776699e-05, "loss": 0.1663, "step": 17227 }, { "epoch": 167.0590361445783, "grad_norm": 9.16196060180664, "learning_rate": 1.6368932038834952e-05, "loss": 0.2669, "step": 17228 }, { "epoch": 167.06867469879518, "grad_norm": 9.031816482543945, "learning_rate": 1.6364077669902913e-05, "loss": 0.2491, "step": 17229 }, { "epoch": 167.07831325301206, "grad_norm": 13.872692108154297, "learning_rate": 1.6359223300970874e-05, "loss": 0.2593, "step": 17230 }, { "epoch": 167.0879518072289, "grad_norm": 9.18254280090332, "learning_rate": 1.6354368932038836e-05, "loss": 0.276, "step": 17231 }, { "epoch": 167.09759036144578, "grad_norm": 3.3029353618621826, "learning_rate": 1.6349514563106797e-05, "loss": 0.0737, "step": 17232 }, { "epoch": 167.10722891566266, "grad_norm": 15.792094230651855, "learning_rate": 1.6344660194174758e-05, "loss": 0.3396, "step": 17233 }, { "epoch": 167.1168674698795, "grad_norm": 28.174007415771484, "learning_rate": 1.633980582524272e-05, "loss": 0.3001, "step": 17234 }, { "epoch": 167.1265060240964, "grad_norm": 6.1903204917907715, "learning_rate": 1.633495145631068e-05, "loss": 0.1458, "step": 17235 }, { "epoch": 167.13614457831326, "grad_norm": 4.575298309326172, "learning_rate": 1.633009708737864e-05, "loss": 0.1975, "step": 17236 }, { "epoch": 167.1457831325301, "grad_norm": 10.125720024108887, "learning_rate": 1.6325242718446603e-05, "loss": 0.5455, "step": 17237 }, { "epoch": 167.155421686747, "grad_norm": 22.652481079101562, "learning_rate": 1.6320388349514564e-05, "loss": 0.3091, "step": 17238 }, { "epoch": 167.16506024096387, "grad_norm": 7.854633331298828, "learning_rate": 1.6315533980582525e-05, "loss": 0.1754, "step": 17239 }, { "epoch": 167.17469879518072, "grad_norm": 4.93293571472168, "learning_rate": 1.6310679611650486e-05, "loss": 0.0894, "step": 17240 }, { "epoch": 167.1843373493976, "grad_norm": 7.9288129806518555, "learning_rate": 1.6305825242718447e-05, "loss": 0.2928, "step": 17241 }, { "epoch": 167.19397590361444, "grad_norm": 8.074642181396484, "learning_rate": 1.6300970873786408e-05, "loss": 0.173, "step": 17242 }, { "epoch": 167.20361445783132, "grad_norm": 11.030224800109863, "learning_rate": 1.629611650485437e-05, "loss": 0.1662, "step": 17243 }, { "epoch": 167.2132530120482, "grad_norm": 7.422823429107666, "learning_rate": 1.629126213592233e-05, "loss": 0.2025, "step": 17244 }, { "epoch": 167.22289156626505, "grad_norm": 11.368736267089844, "learning_rate": 1.628640776699029e-05, "loss": 0.1936, "step": 17245 }, { "epoch": 167.23253012048193, "grad_norm": 29.505434036254883, "learning_rate": 1.6281553398058253e-05, "loss": 0.218, "step": 17246 }, { "epoch": 167.2421686746988, "grad_norm": 8.752609252929688, "learning_rate": 1.6276699029126214e-05, "loss": 0.4128, "step": 17247 }, { "epoch": 167.25180722891565, "grad_norm": 6.673880100250244, "learning_rate": 1.6271844660194175e-05, "loss": 0.24, "step": 17248 }, { "epoch": 167.26144578313253, "grad_norm": 7.473745346069336, "learning_rate": 1.626699029126214e-05, "loss": 0.2066, "step": 17249 }, { "epoch": 167.2710843373494, "grad_norm": 9.513120651245117, "learning_rate": 1.6262135922330097e-05, "loss": 0.1476, "step": 17250 }, { "epoch": 167.28072289156626, "grad_norm": 9.258613586425781, "learning_rate": 1.625728155339806e-05, "loss": 0.2045, "step": 17251 }, { "epoch": 167.29036144578313, "grad_norm": 14.656987190246582, "learning_rate": 1.625242718446602e-05, "loss": 0.2504, "step": 17252 }, { "epoch": 167.3, "grad_norm": 8.350074768066406, "learning_rate": 1.624757281553398e-05, "loss": 0.1087, "step": 17253 }, { "epoch": 167.30963855421686, "grad_norm": 5.211190700531006, "learning_rate": 1.6242718446601942e-05, "loss": 0.092, "step": 17254 }, { "epoch": 167.31927710843374, "grad_norm": 5.008193016052246, "learning_rate": 1.6237864077669903e-05, "loss": 0.2956, "step": 17255 }, { "epoch": 167.32891566265062, "grad_norm": 9.622546195983887, "learning_rate": 1.6233009708737864e-05, "loss": 0.1891, "step": 17256 }, { "epoch": 167.33855421686746, "grad_norm": 12.217633247375488, "learning_rate": 1.6228155339805825e-05, "loss": 0.3497, "step": 17257 }, { "epoch": 167.34819277108434, "grad_norm": 10.336780548095703, "learning_rate": 1.6223300970873787e-05, "loss": 0.3912, "step": 17258 }, { "epoch": 167.3578313253012, "grad_norm": 7.4733052253723145, "learning_rate": 1.6218446601941748e-05, "loss": 0.2702, "step": 17259 }, { "epoch": 167.36746987951807, "grad_norm": 13.300050735473633, "learning_rate": 1.621359223300971e-05, "loss": 0.2038, "step": 17260 }, { "epoch": 167.37710843373495, "grad_norm": 14.441556930541992, "learning_rate": 1.620873786407767e-05, "loss": 0.3125, "step": 17261 }, { "epoch": 167.3867469879518, "grad_norm": 6.521315097808838, "learning_rate": 1.6203883495145635e-05, "loss": 0.3248, "step": 17262 }, { "epoch": 167.39638554216867, "grad_norm": 4.28140115737915, "learning_rate": 1.6199029126213596e-05, "loss": 0.0784, "step": 17263 }, { "epoch": 167.40602409638555, "grad_norm": 18.80849838256836, "learning_rate": 1.6194174757281553e-05, "loss": 0.2157, "step": 17264 }, { "epoch": 167.4156626506024, "grad_norm": 8.258681297302246, "learning_rate": 1.6189320388349515e-05, "loss": 0.1816, "step": 17265 }, { "epoch": 167.42530120481928, "grad_norm": 10.720514297485352, "learning_rate": 1.6184466019417476e-05, "loss": 0.3349, "step": 17266 }, { "epoch": 167.43493975903615, "grad_norm": 5.780106544494629, "learning_rate": 1.6179611650485437e-05, "loss": 0.2127, "step": 17267 }, { "epoch": 167.444578313253, "grad_norm": 24.842844009399414, "learning_rate": 1.6174757281553398e-05, "loss": 0.283, "step": 17268 }, { "epoch": 167.45421686746988, "grad_norm": 14.521574974060059, "learning_rate": 1.616990291262136e-05, "loss": 0.4406, "step": 17269 }, { "epoch": 167.46385542168676, "grad_norm": 9.063725471496582, "learning_rate": 1.616504854368932e-05, "loss": 0.2212, "step": 17270 }, { "epoch": 167.4734939759036, "grad_norm": 13.16413688659668, "learning_rate": 1.616019417475728e-05, "loss": 0.1697, "step": 17271 }, { "epoch": 167.48313253012049, "grad_norm": 9.426595687866211, "learning_rate": 1.6155339805825243e-05, "loss": 0.3055, "step": 17272 }, { "epoch": 167.49277108433736, "grad_norm": 9.682779312133789, "learning_rate": 1.6150485436893204e-05, "loss": 0.2648, "step": 17273 }, { "epoch": 167.5024096385542, "grad_norm": 13.01990795135498, "learning_rate": 1.6145631067961168e-05, "loss": 0.1994, "step": 17274 }, { "epoch": 167.5120481927711, "grad_norm": 6.860156536102295, "learning_rate": 1.614077669902913e-05, "loss": 0.1332, "step": 17275 }, { "epoch": 167.52168674698794, "grad_norm": 11.168176651000977, "learning_rate": 1.613592233009709e-05, "loss": 0.1784, "step": 17276 }, { "epoch": 167.53132530120482, "grad_norm": 11.69958209991455, "learning_rate": 1.6131067961165052e-05, "loss": 0.1404, "step": 17277 }, { "epoch": 167.5409638554217, "grad_norm": 10.761241912841797, "learning_rate": 1.612621359223301e-05, "loss": 0.358, "step": 17278 }, { "epoch": 167.55060240963854, "grad_norm": 14.616942405700684, "learning_rate": 1.612135922330097e-05, "loss": 0.3935, "step": 17279 }, { "epoch": 167.56024096385542, "grad_norm": 3.5959055423736572, "learning_rate": 1.6116504854368932e-05, "loss": 0.1069, "step": 17280 }, { "epoch": 167.5698795180723, "grad_norm": 36.830413818359375, "learning_rate": 1.6111650485436893e-05, "loss": 0.3784, "step": 17281 }, { "epoch": 167.57951807228915, "grad_norm": 11.57894515991211, "learning_rate": 1.6106796116504854e-05, "loss": 0.205, "step": 17282 }, { "epoch": 167.58915662650602, "grad_norm": 7.69023323059082, "learning_rate": 1.6101941747572815e-05, "loss": 0.1138, "step": 17283 }, { "epoch": 167.5987951807229, "grad_norm": 21.7699031829834, "learning_rate": 1.6097087378640776e-05, "loss": 0.3254, "step": 17284 }, { "epoch": 167.60843373493975, "grad_norm": 8.703187942504883, "learning_rate": 1.6092233009708738e-05, "loss": 0.2656, "step": 17285 }, { "epoch": 167.61807228915663, "grad_norm": 9.89088249206543, "learning_rate": 1.60873786407767e-05, "loss": 0.2073, "step": 17286 }, { "epoch": 167.6277108433735, "grad_norm": 15.14786148071289, "learning_rate": 1.6082524271844663e-05, "loss": 0.3939, "step": 17287 }, { "epoch": 167.63734939759036, "grad_norm": 10.370009422302246, "learning_rate": 1.6077669902912624e-05, "loss": 0.2356, "step": 17288 }, { "epoch": 167.64698795180723, "grad_norm": 31.447114944458008, "learning_rate": 1.6072815533980585e-05, "loss": 0.1728, "step": 17289 }, { "epoch": 167.65662650602408, "grad_norm": 12.966504096984863, "learning_rate": 1.6067961165048547e-05, "loss": 0.4383, "step": 17290 }, { "epoch": 167.66626506024096, "grad_norm": 6.747313499450684, "learning_rate": 1.6063106796116504e-05, "loss": 0.3937, "step": 17291 }, { "epoch": 167.67590361445784, "grad_norm": 20.435768127441406, "learning_rate": 1.6058252427184466e-05, "loss": 0.4684, "step": 17292 }, { "epoch": 167.6855421686747, "grad_norm": 7.146467208862305, "learning_rate": 1.6053398058252427e-05, "loss": 0.1674, "step": 17293 }, { "epoch": 167.69518072289156, "grad_norm": 10.516118049621582, "learning_rate": 1.6048543689320388e-05, "loss": 0.3548, "step": 17294 }, { "epoch": 167.70481927710844, "grad_norm": 18.16854476928711, "learning_rate": 1.604368932038835e-05, "loss": 0.3567, "step": 17295 }, { "epoch": 167.7144578313253, "grad_norm": 11.27432632446289, "learning_rate": 1.603883495145631e-05, "loss": 0.3104, "step": 17296 }, { "epoch": 167.72409638554217, "grad_norm": 8.43920612335205, "learning_rate": 1.603398058252427e-05, "loss": 0.0784, "step": 17297 }, { "epoch": 167.73373493975905, "grad_norm": 18.13174819946289, "learning_rate": 1.6029126213592232e-05, "loss": 0.3222, "step": 17298 }, { "epoch": 167.7433734939759, "grad_norm": 12.235583305358887, "learning_rate": 1.6024271844660194e-05, "loss": 0.4258, "step": 17299 }, { "epoch": 167.75301204819277, "grad_norm": 13.490067481994629, "learning_rate": 1.6019417475728158e-05, "loss": 0.3305, "step": 17300 }, { "epoch": 167.76265060240965, "grad_norm": 8.017946243286133, "learning_rate": 1.601456310679612e-05, "loss": 0.2241, "step": 17301 }, { "epoch": 167.7722891566265, "grad_norm": 13.142027854919434, "learning_rate": 1.600970873786408e-05, "loss": 0.2924, "step": 17302 }, { "epoch": 167.78192771084338, "grad_norm": 4.534438133239746, "learning_rate": 1.600485436893204e-05, "loss": 0.1102, "step": 17303 }, { "epoch": 167.79156626506025, "grad_norm": 5.051743984222412, "learning_rate": 1.6000000000000003e-05, "loss": 0.2236, "step": 17304 }, { "epoch": 167.8012048192771, "grad_norm": 8.84562873840332, "learning_rate": 1.599514563106796e-05, "loss": 0.4015, "step": 17305 }, { "epoch": 167.81084337349398, "grad_norm": 14.547760963439941, "learning_rate": 1.599029126213592e-05, "loss": 0.4437, "step": 17306 }, { "epoch": 167.82048192771083, "grad_norm": 15.872587203979492, "learning_rate": 1.5985436893203883e-05, "loss": 0.0545, "step": 17307 }, { "epoch": 167.8301204819277, "grad_norm": 27.68018341064453, "learning_rate": 1.5980582524271844e-05, "loss": 0.2663, "step": 17308 }, { "epoch": 167.83975903614459, "grad_norm": 17.206470489501953, "learning_rate": 1.5975728155339805e-05, "loss": 0.332, "step": 17309 }, { "epoch": 167.84939759036143, "grad_norm": 12.349311828613281, "learning_rate": 1.5970873786407766e-05, "loss": 0.2183, "step": 17310 }, { "epoch": 167.8590361445783, "grad_norm": 3.488104820251465, "learning_rate": 1.5966019417475727e-05, "loss": 0.1399, "step": 17311 }, { "epoch": 167.8686746987952, "grad_norm": 45.26012420654297, "learning_rate": 1.5961165048543692e-05, "loss": 0.3569, "step": 17312 }, { "epoch": 167.87831325301204, "grad_norm": 5.823330402374268, "learning_rate": 1.5956310679611653e-05, "loss": 0.1419, "step": 17313 }, { "epoch": 167.88795180722892, "grad_norm": 15.382569313049316, "learning_rate": 1.5951456310679614e-05, "loss": 0.4858, "step": 17314 }, { "epoch": 167.8975903614458, "grad_norm": 5.272405624389648, "learning_rate": 1.5946601941747575e-05, "loss": 0.07, "step": 17315 }, { "epoch": 167.90722891566264, "grad_norm": 11.741775512695312, "learning_rate": 1.5941747572815536e-05, "loss": 0.2985, "step": 17316 }, { "epoch": 167.91686746987952, "grad_norm": 22.65959930419922, "learning_rate": 1.5936893203883498e-05, "loss": 0.354, "step": 17317 }, { "epoch": 167.9265060240964, "grad_norm": 15.722450256347656, "learning_rate": 1.593203883495146e-05, "loss": 0.4005, "step": 17318 }, { "epoch": 167.93614457831325, "grad_norm": 33.236297607421875, "learning_rate": 1.5927184466019416e-05, "loss": 0.2424, "step": 17319 }, { "epoch": 167.94578313253012, "grad_norm": 3.7372705936431885, "learning_rate": 1.5922330097087378e-05, "loss": 0.098, "step": 17320 }, { "epoch": 167.955421686747, "grad_norm": 15.117687225341797, "learning_rate": 1.591747572815534e-05, "loss": 0.2833, "step": 17321 }, { "epoch": 167.96506024096385, "grad_norm": 13.26105785369873, "learning_rate": 1.59126213592233e-05, "loss": 0.2856, "step": 17322 }, { "epoch": 167.97469879518073, "grad_norm": 7.457457065582275, "learning_rate": 1.590776699029126e-05, "loss": 0.178, "step": 17323 }, { "epoch": 167.98433734939758, "grad_norm": 14.819429397583008, "learning_rate": 1.5902912621359222e-05, "loss": 0.387, "step": 17324 }, { "epoch": 167.99397590361446, "grad_norm": 11.821440696716309, "learning_rate": 1.5898058252427187e-05, "loss": 0.3628, "step": 17325 }, { "epoch": 168.00963855421688, "grad_norm": 12.643454551696777, "learning_rate": 1.5893203883495148e-05, "loss": 0.283, "step": 17326 }, { "epoch": 168.01927710843373, "grad_norm": 8.191234588623047, "learning_rate": 1.588834951456311e-05, "loss": 0.1468, "step": 17327 }, { "epoch": 168.0289156626506, "grad_norm": 7.599942684173584, "learning_rate": 1.588349514563107e-05, "loss": 0.2042, "step": 17328 }, { "epoch": 168.03855421686748, "grad_norm": 10.120060920715332, "learning_rate": 1.587864077669903e-05, "loss": 0.2122, "step": 17329 }, { "epoch": 168.04819277108433, "grad_norm": 10.649652481079102, "learning_rate": 1.5873786407766992e-05, "loss": 0.2848, "step": 17330 }, { "epoch": 168.0578313253012, "grad_norm": 13.279922485351562, "learning_rate": 1.5868932038834954e-05, "loss": 0.3232, "step": 17331 }, { "epoch": 168.0674698795181, "grad_norm": 7.322567462921143, "learning_rate": 1.5864077669902915e-05, "loss": 0.1659, "step": 17332 }, { "epoch": 168.07710843373494, "grad_norm": 12.100068092346191, "learning_rate": 1.5859223300970873e-05, "loss": 0.3764, "step": 17333 }, { "epoch": 168.0867469879518, "grad_norm": 9.100424766540527, "learning_rate": 1.5854368932038834e-05, "loss": 0.1485, "step": 17334 }, { "epoch": 168.09638554216866, "grad_norm": 9.126492500305176, "learning_rate": 1.5849514563106795e-05, "loss": 0.275, "step": 17335 }, { "epoch": 168.10602409638554, "grad_norm": 12.975918769836426, "learning_rate": 1.5844660194174756e-05, "loss": 0.3297, "step": 17336 }, { "epoch": 168.11566265060242, "grad_norm": 28.083711624145508, "learning_rate": 1.5839805825242717e-05, "loss": 0.1895, "step": 17337 }, { "epoch": 168.12530120481927, "grad_norm": 12.997955322265625, "learning_rate": 1.583495145631068e-05, "loss": 0.1961, "step": 17338 }, { "epoch": 168.13493975903614, "grad_norm": 10.282422065734863, "learning_rate": 1.5830097087378643e-05, "loss": 0.2276, "step": 17339 }, { "epoch": 168.14457831325302, "grad_norm": 5.199143409729004, "learning_rate": 1.5825242718446604e-05, "loss": 0.0914, "step": 17340 }, { "epoch": 168.15421686746987, "grad_norm": 8.376659393310547, "learning_rate": 1.5820388349514565e-05, "loss": 0.2443, "step": 17341 }, { "epoch": 168.16385542168675, "grad_norm": 9.628080368041992, "learning_rate": 1.5815533980582526e-05, "loss": 0.1905, "step": 17342 }, { "epoch": 168.17349397590363, "grad_norm": 6.409850597381592, "learning_rate": 1.5810679611650487e-05, "loss": 0.1844, "step": 17343 }, { "epoch": 168.18313253012047, "grad_norm": 3.4367897510528564, "learning_rate": 1.580582524271845e-05, "loss": 0.07, "step": 17344 }, { "epoch": 168.19277108433735, "grad_norm": 15.183862686157227, "learning_rate": 1.580097087378641e-05, "loss": 0.1522, "step": 17345 }, { "epoch": 168.20240963855423, "grad_norm": 7.8945207595825195, "learning_rate": 1.579611650485437e-05, "loss": 0.2303, "step": 17346 }, { "epoch": 168.21204819277108, "grad_norm": 10.693119049072266, "learning_rate": 1.579126213592233e-05, "loss": 0.3799, "step": 17347 }, { "epoch": 168.22168674698796, "grad_norm": 11.7240571975708, "learning_rate": 1.578640776699029e-05, "loss": 0.2038, "step": 17348 }, { "epoch": 168.2313253012048, "grad_norm": 65.79851531982422, "learning_rate": 1.578155339805825e-05, "loss": 0.457, "step": 17349 }, { "epoch": 168.24096385542168, "grad_norm": 19.779071807861328, "learning_rate": 1.5776699029126215e-05, "loss": 0.1981, "step": 17350 }, { "epoch": 168.25060240963856, "grad_norm": 6.2644572257995605, "learning_rate": 1.5771844660194177e-05, "loss": 0.1298, "step": 17351 }, { "epoch": 168.2602409638554, "grad_norm": 5.628178596496582, "learning_rate": 1.5766990291262138e-05, "loss": 0.3045, "step": 17352 }, { "epoch": 168.2698795180723, "grad_norm": 13.415122032165527, "learning_rate": 1.57621359223301e-05, "loss": 0.383, "step": 17353 }, { "epoch": 168.27951807228916, "grad_norm": 14.172009468078613, "learning_rate": 1.575728155339806e-05, "loss": 0.2566, "step": 17354 }, { "epoch": 168.289156626506, "grad_norm": 13.914602279663086, "learning_rate": 1.575242718446602e-05, "loss": 0.2646, "step": 17355 }, { "epoch": 168.2987951807229, "grad_norm": 9.492887496948242, "learning_rate": 1.5747572815533982e-05, "loss": 0.2842, "step": 17356 }, { "epoch": 168.30843373493977, "grad_norm": 10.3165283203125, "learning_rate": 1.5742718446601943e-05, "loss": 0.1309, "step": 17357 }, { "epoch": 168.31807228915662, "grad_norm": 5.676204681396484, "learning_rate": 1.5737864077669905e-05, "loss": 0.1424, "step": 17358 }, { "epoch": 168.3277108433735, "grad_norm": 16.353389739990234, "learning_rate": 1.5733009708737866e-05, "loss": 0.3787, "step": 17359 }, { "epoch": 168.33734939759037, "grad_norm": 14.311125755310059, "learning_rate": 1.5728155339805823e-05, "loss": 0.2726, "step": 17360 }, { "epoch": 168.34698795180722, "grad_norm": 15.697380065917969, "learning_rate": 1.5723300970873785e-05, "loss": 0.3566, "step": 17361 }, { "epoch": 168.3566265060241, "grad_norm": 5.404127597808838, "learning_rate": 1.5718446601941746e-05, "loss": 0.1332, "step": 17362 }, { "epoch": 168.36626506024098, "grad_norm": 6.529258728027344, "learning_rate": 1.571359223300971e-05, "loss": 0.1252, "step": 17363 }, { "epoch": 168.37590361445783, "grad_norm": 13.049782752990723, "learning_rate": 1.570873786407767e-05, "loss": 0.1298, "step": 17364 }, { "epoch": 168.3855421686747, "grad_norm": 10.72072696685791, "learning_rate": 1.5703883495145633e-05, "loss": 0.2224, "step": 17365 }, { "epoch": 168.39518072289155, "grad_norm": 11.52907657623291, "learning_rate": 1.5699029126213594e-05, "loss": 0.1167, "step": 17366 }, { "epoch": 168.40481927710843, "grad_norm": 7.511816501617432, "learning_rate": 1.5694174757281555e-05, "loss": 0.2139, "step": 17367 }, { "epoch": 168.4144578313253, "grad_norm": 5.022531032562256, "learning_rate": 1.5689320388349516e-05, "loss": 0.16, "step": 17368 }, { "epoch": 168.42409638554216, "grad_norm": 8.667278289794922, "learning_rate": 1.5684466019417477e-05, "loss": 0.2613, "step": 17369 }, { "epoch": 168.43373493975903, "grad_norm": 7.829475402832031, "learning_rate": 1.5679611650485438e-05, "loss": 0.2226, "step": 17370 }, { "epoch": 168.4433734939759, "grad_norm": 7.608861446380615, "learning_rate": 1.56747572815534e-05, "loss": 0.2232, "step": 17371 }, { "epoch": 168.45301204819276, "grad_norm": 3.786129951477051, "learning_rate": 1.566990291262136e-05, "loss": 0.1569, "step": 17372 }, { "epoch": 168.46265060240964, "grad_norm": 15.661040306091309, "learning_rate": 1.5665048543689322e-05, "loss": 0.1906, "step": 17373 }, { "epoch": 168.47228915662652, "grad_norm": 9.526203155517578, "learning_rate": 1.566019417475728e-05, "loss": 0.2098, "step": 17374 }, { "epoch": 168.48192771084337, "grad_norm": 18.67253303527832, "learning_rate": 1.5655339805825244e-05, "loss": 0.3062, "step": 17375 }, { "epoch": 168.49156626506024, "grad_norm": 15.596994400024414, "learning_rate": 1.5650485436893205e-05, "loss": 0.2169, "step": 17376 }, { "epoch": 168.50120481927712, "grad_norm": 2.9180963039398193, "learning_rate": 1.5645631067961166e-05, "loss": 0.0551, "step": 17377 }, { "epoch": 168.51084337349397, "grad_norm": 12.935698509216309, "learning_rate": 1.5640776699029127e-05, "loss": 0.1621, "step": 17378 }, { "epoch": 168.52048192771085, "grad_norm": 5.8189496994018555, "learning_rate": 1.563592233009709e-05, "loss": 0.1694, "step": 17379 }, { "epoch": 168.53012048192772, "grad_norm": 7.437264919281006, "learning_rate": 1.563106796116505e-05, "loss": 0.281, "step": 17380 }, { "epoch": 168.53975903614457, "grad_norm": 6.139033317565918, "learning_rate": 1.562621359223301e-05, "loss": 0.2397, "step": 17381 }, { "epoch": 168.54939759036145, "grad_norm": 28.112695693969727, "learning_rate": 1.5621359223300972e-05, "loss": 0.2045, "step": 17382 }, { "epoch": 168.5590361445783, "grad_norm": 9.321730613708496, "learning_rate": 1.5616504854368933e-05, "loss": 0.2995, "step": 17383 }, { "epoch": 168.56867469879518, "grad_norm": 13.10554027557373, "learning_rate": 1.5611650485436894e-05, "loss": 0.229, "step": 17384 }, { "epoch": 168.57831325301206, "grad_norm": 8.017291069030762, "learning_rate": 1.5606796116504855e-05, "loss": 0.2303, "step": 17385 }, { "epoch": 168.5879518072289, "grad_norm": 8.796552658081055, "learning_rate": 1.5601941747572817e-05, "loss": 0.2274, "step": 17386 }, { "epoch": 168.59759036144578, "grad_norm": 7.075427055358887, "learning_rate": 1.5597087378640778e-05, "loss": 0.3115, "step": 17387 }, { "epoch": 168.60722891566266, "grad_norm": 7.328737735748291, "learning_rate": 1.559223300970874e-05, "loss": 0.1722, "step": 17388 }, { "epoch": 168.6168674698795, "grad_norm": 12.591527938842773, "learning_rate": 1.55873786407767e-05, "loss": 0.1483, "step": 17389 }, { "epoch": 168.6265060240964, "grad_norm": 21.276424407958984, "learning_rate": 1.558252427184466e-05, "loss": 0.7271, "step": 17390 }, { "epoch": 168.63614457831326, "grad_norm": 7.363386631011963, "learning_rate": 1.5577669902912622e-05, "loss": 0.2107, "step": 17391 }, { "epoch": 168.6457831325301, "grad_norm": 10.573288917541504, "learning_rate": 1.5572815533980583e-05, "loss": 0.2176, "step": 17392 }, { "epoch": 168.655421686747, "grad_norm": 7.378829002380371, "learning_rate": 1.5567961165048545e-05, "loss": 0.2811, "step": 17393 }, { "epoch": 168.66506024096387, "grad_norm": 63.69400405883789, "learning_rate": 1.5563106796116506e-05, "loss": 0.3032, "step": 17394 }, { "epoch": 168.67469879518072, "grad_norm": 14.691051483154297, "learning_rate": 1.5558252427184467e-05, "loss": 0.3449, "step": 17395 }, { "epoch": 168.6843373493976, "grad_norm": 36.515296936035156, "learning_rate": 1.5553398058252428e-05, "loss": 0.3252, "step": 17396 }, { "epoch": 168.69397590361444, "grad_norm": 10.252126693725586, "learning_rate": 1.554854368932039e-05, "loss": 0.2853, "step": 17397 }, { "epoch": 168.70361445783132, "grad_norm": 28.141254425048828, "learning_rate": 1.554368932038835e-05, "loss": 0.3801, "step": 17398 }, { "epoch": 168.7132530120482, "grad_norm": 14.846113204956055, "learning_rate": 1.553883495145631e-05, "loss": 0.1892, "step": 17399 }, { "epoch": 168.72289156626505, "grad_norm": 14.2885160446167, "learning_rate": 1.5533980582524273e-05, "loss": 0.1594, "step": 17400 }, { "epoch": 168.73253012048193, "grad_norm": 16.839731216430664, "learning_rate": 1.5529126213592234e-05, "loss": 0.2979, "step": 17401 }, { "epoch": 168.7421686746988, "grad_norm": 13.776455879211426, "learning_rate": 1.5524271844660195e-05, "loss": 0.1655, "step": 17402 }, { "epoch": 168.75180722891565, "grad_norm": 18.32915496826172, "learning_rate": 1.5519417475728156e-05, "loss": 0.3534, "step": 17403 }, { "epoch": 168.76144578313253, "grad_norm": 11.615242958068848, "learning_rate": 1.5514563106796117e-05, "loss": 0.2887, "step": 17404 }, { "epoch": 168.7710843373494, "grad_norm": 14.826050758361816, "learning_rate": 1.550970873786408e-05, "loss": 0.2374, "step": 17405 }, { "epoch": 168.78072289156626, "grad_norm": 11.810484886169434, "learning_rate": 1.550485436893204e-05, "loss": 0.3344, "step": 17406 }, { "epoch": 168.79036144578313, "grad_norm": 6.511032581329346, "learning_rate": 1.55e-05, "loss": 0.2624, "step": 17407 }, { "epoch": 168.8, "grad_norm": 9.219816207885742, "learning_rate": 1.5495145631067962e-05, "loss": 0.2703, "step": 17408 }, { "epoch": 168.80963855421686, "grad_norm": 11.26721477508545, "learning_rate": 1.5490291262135923e-05, "loss": 0.4164, "step": 17409 }, { "epoch": 168.81927710843374, "grad_norm": 14.840712547302246, "learning_rate": 1.5485436893203884e-05, "loss": 0.1882, "step": 17410 }, { "epoch": 168.82891566265062, "grad_norm": 6.157181739807129, "learning_rate": 1.5480582524271845e-05, "loss": 0.1527, "step": 17411 }, { "epoch": 168.83855421686746, "grad_norm": 5.889811992645264, "learning_rate": 1.5475728155339806e-05, "loss": 0.1905, "step": 17412 }, { "epoch": 168.84819277108434, "grad_norm": 29.27631378173828, "learning_rate": 1.5470873786407768e-05, "loss": 0.0863, "step": 17413 }, { "epoch": 168.8578313253012, "grad_norm": 8.721063613891602, "learning_rate": 1.546601941747573e-05, "loss": 0.2248, "step": 17414 }, { "epoch": 168.86746987951807, "grad_norm": 14.274102210998535, "learning_rate": 1.546116504854369e-05, "loss": 0.2488, "step": 17415 }, { "epoch": 168.87710843373495, "grad_norm": 8.616128921508789, "learning_rate": 1.545631067961165e-05, "loss": 0.2293, "step": 17416 }, { "epoch": 168.8867469879518, "grad_norm": 34.77491760253906, "learning_rate": 1.5451456310679612e-05, "loss": 0.334, "step": 17417 }, { "epoch": 168.89638554216867, "grad_norm": 13.147012710571289, "learning_rate": 1.5446601941747573e-05, "loss": 0.422, "step": 17418 }, { "epoch": 168.90602409638555, "grad_norm": 17.755531311035156, "learning_rate": 1.5441747572815534e-05, "loss": 0.4648, "step": 17419 }, { "epoch": 168.9156626506024, "grad_norm": 36.70430374145508, "learning_rate": 1.5436893203883496e-05, "loss": 0.2039, "step": 17420 }, { "epoch": 168.92530120481928, "grad_norm": 9.384970664978027, "learning_rate": 1.5432038834951457e-05, "loss": 0.3265, "step": 17421 }, { "epoch": 168.93493975903615, "grad_norm": 8.511691093444824, "learning_rate": 1.5427184466019418e-05, "loss": 0.1795, "step": 17422 }, { "epoch": 168.944578313253, "grad_norm": 5.846603870391846, "learning_rate": 1.542233009708738e-05, "loss": 0.301, "step": 17423 }, { "epoch": 168.95421686746988, "grad_norm": 7.322436809539795, "learning_rate": 1.541747572815534e-05, "loss": 0.1643, "step": 17424 }, { "epoch": 168.96385542168676, "grad_norm": 5.417943954467773, "learning_rate": 1.54126213592233e-05, "loss": 0.1216, "step": 17425 }, { "epoch": 168.9734939759036, "grad_norm": 8.935710906982422, "learning_rate": 1.5407766990291262e-05, "loss": 0.1421, "step": 17426 }, { "epoch": 168.98313253012049, "grad_norm": 23.530292510986328, "learning_rate": 1.5402912621359224e-05, "loss": 0.3371, "step": 17427 }, { "epoch": 168.99277108433736, "grad_norm": 12.361328125, "learning_rate": 1.5398058252427185e-05, "loss": 0.2649, "step": 17428 }, { "epoch": 169.00843373493976, "grad_norm": 9.853617668151855, "learning_rate": 1.5393203883495146e-05, "loss": 0.2673, "step": 17429 }, { "epoch": 169.01807228915663, "grad_norm": 29.47420310974121, "learning_rate": 1.5388349514563107e-05, "loss": 0.2989, "step": 17430 }, { "epoch": 169.02771084337348, "grad_norm": 23.063522338867188, "learning_rate": 1.5383495145631068e-05, "loss": 0.1765, "step": 17431 }, { "epoch": 169.03734939759036, "grad_norm": 13.002592086791992, "learning_rate": 1.537864077669903e-05, "loss": 0.2632, "step": 17432 }, { "epoch": 169.04698795180724, "grad_norm": 6.101226806640625, "learning_rate": 1.537378640776699e-05, "loss": 0.2592, "step": 17433 }, { "epoch": 169.0566265060241, "grad_norm": 41.6173095703125, "learning_rate": 1.536893203883495e-05, "loss": 0.3474, "step": 17434 }, { "epoch": 169.06626506024097, "grad_norm": 15.478041648864746, "learning_rate": 1.5364077669902913e-05, "loss": 0.3255, "step": 17435 }, { "epoch": 169.07590361445784, "grad_norm": 12.529186248779297, "learning_rate": 1.5359223300970874e-05, "loss": 0.2287, "step": 17436 }, { "epoch": 169.0855421686747, "grad_norm": 4.866376876831055, "learning_rate": 1.5354368932038835e-05, "loss": 0.2205, "step": 17437 }, { "epoch": 169.09518072289157, "grad_norm": 9.343948364257812, "learning_rate": 1.5349514563106796e-05, "loss": 0.2762, "step": 17438 }, { "epoch": 169.10481927710845, "grad_norm": 4.055784702301025, "learning_rate": 1.534466019417476e-05, "loss": 0.1563, "step": 17439 }, { "epoch": 169.1144578313253, "grad_norm": 4.134244918823242, "learning_rate": 1.533980582524272e-05, "loss": 0.1328, "step": 17440 }, { "epoch": 169.12409638554217, "grad_norm": 43.14756393432617, "learning_rate": 1.533495145631068e-05, "loss": 0.1495, "step": 17441 }, { "epoch": 169.13373493975902, "grad_norm": 4.016088008880615, "learning_rate": 1.533009708737864e-05, "loss": 0.4378, "step": 17442 }, { "epoch": 169.1433734939759, "grad_norm": 18.113792419433594, "learning_rate": 1.5325242718446602e-05, "loss": 0.2975, "step": 17443 }, { "epoch": 169.15301204819278, "grad_norm": 18.39775848388672, "learning_rate": 1.5320388349514563e-05, "loss": 0.2564, "step": 17444 }, { "epoch": 169.16265060240963, "grad_norm": 38.8370475769043, "learning_rate": 1.5315533980582524e-05, "loss": 0.265, "step": 17445 }, { "epoch": 169.1722891566265, "grad_norm": 73.92423248291016, "learning_rate": 1.5310679611650485e-05, "loss": 0.3083, "step": 17446 }, { "epoch": 169.18192771084338, "grad_norm": 20.884675979614258, "learning_rate": 1.5305825242718447e-05, "loss": 0.2108, "step": 17447 }, { "epoch": 169.19156626506023, "grad_norm": 29.692655563354492, "learning_rate": 1.5300970873786408e-05, "loss": 0.3234, "step": 17448 }, { "epoch": 169.2012048192771, "grad_norm": 95.84358978271484, "learning_rate": 1.529611650485437e-05, "loss": 0.4033, "step": 17449 }, { "epoch": 169.210843373494, "grad_norm": 9.134208679199219, "learning_rate": 1.529126213592233e-05, "loss": 0.3608, "step": 17450 }, { "epoch": 169.22048192771084, "grad_norm": 41.709957122802734, "learning_rate": 1.5286407766990294e-05, "loss": 0.1175, "step": 17451 }, { "epoch": 169.2301204819277, "grad_norm": 40.20358657836914, "learning_rate": 1.5281553398058256e-05, "loss": 0.1535, "step": 17452 }, { "epoch": 169.2397590361446, "grad_norm": 7.33770751953125, "learning_rate": 1.5276699029126217e-05, "loss": 0.2055, "step": 17453 }, { "epoch": 169.24939759036144, "grad_norm": 20.313560485839844, "learning_rate": 1.5271844660194175e-05, "loss": 0.2355, "step": 17454 }, { "epoch": 169.25903614457832, "grad_norm": 4.089105606079102, "learning_rate": 1.5266990291262136e-05, "loss": 0.2766, "step": 17455 }, { "epoch": 169.2686746987952, "grad_norm": 23.606319427490234, "learning_rate": 1.5262135922330097e-05, "loss": 0.3592, "step": 17456 }, { "epoch": 169.27831325301204, "grad_norm": 27.251632690429688, "learning_rate": 1.525728155339806e-05, "loss": 0.234, "step": 17457 }, { "epoch": 169.28795180722892, "grad_norm": 31.845712661743164, "learning_rate": 1.5252427184466019e-05, "loss": 0.2514, "step": 17458 }, { "epoch": 169.29759036144577, "grad_norm": 4.650908470153809, "learning_rate": 1.524757281553398e-05, "loss": 0.1849, "step": 17459 }, { "epoch": 169.30722891566265, "grad_norm": 6.790900707244873, "learning_rate": 1.5242718446601941e-05, "loss": 0.3011, "step": 17460 }, { "epoch": 169.31686746987953, "grad_norm": 22.998897552490234, "learning_rate": 1.5237864077669903e-05, "loss": 0.1314, "step": 17461 }, { "epoch": 169.32650602409637, "grad_norm": 2.2118566036224365, "learning_rate": 1.5233009708737864e-05, "loss": 0.1837, "step": 17462 }, { "epoch": 169.33614457831325, "grad_norm": 28.619762420654297, "learning_rate": 1.5228155339805825e-05, "loss": 0.2231, "step": 17463 }, { "epoch": 169.34578313253013, "grad_norm": 1.5183353424072266, "learning_rate": 1.5223300970873788e-05, "loss": 0.1865, "step": 17464 }, { "epoch": 169.35542168674698, "grad_norm": 3.443532705307007, "learning_rate": 1.5218446601941749e-05, "loss": 0.221, "step": 17465 }, { "epoch": 169.36506024096386, "grad_norm": 18.345928192138672, "learning_rate": 1.521359223300971e-05, "loss": 0.1719, "step": 17466 }, { "epoch": 169.37469879518073, "grad_norm": 21.258012771606445, "learning_rate": 1.5208737864077671e-05, "loss": 0.374, "step": 17467 }, { "epoch": 169.38433734939758, "grad_norm": 52.93589401245117, "learning_rate": 1.5203883495145632e-05, "loss": 0.2767, "step": 17468 }, { "epoch": 169.39397590361446, "grad_norm": 3.96161150932312, "learning_rate": 1.5199029126213593e-05, "loss": 0.3358, "step": 17469 }, { "epoch": 169.40361445783134, "grad_norm": 7.475805759429932, "learning_rate": 1.5194174757281555e-05, "loss": 0.2067, "step": 17470 }, { "epoch": 169.4132530120482, "grad_norm": 5.335136413574219, "learning_rate": 1.5189320388349516e-05, "loss": 0.1621, "step": 17471 }, { "epoch": 169.42289156626506, "grad_norm": 3.1439764499664307, "learning_rate": 1.5184466019417475e-05, "loss": 0.3095, "step": 17472 }, { "epoch": 169.4325301204819, "grad_norm": 51.17869567871094, "learning_rate": 1.5179611650485436e-05, "loss": 0.2051, "step": 17473 }, { "epoch": 169.4421686746988, "grad_norm": 4.567363262176514, "learning_rate": 1.5174757281553397e-05, "loss": 0.2682, "step": 17474 }, { "epoch": 169.45180722891567, "grad_norm": 58.04289245605469, "learning_rate": 1.5169902912621359e-05, "loss": 0.3093, "step": 17475 }, { "epoch": 169.46144578313252, "grad_norm": 15.408805847167969, "learning_rate": 1.5165048543689323e-05, "loss": 0.0706, "step": 17476 }, { "epoch": 169.4710843373494, "grad_norm": 21.676864624023438, "learning_rate": 1.5160194174757284e-05, "loss": 0.1608, "step": 17477 }, { "epoch": 169.48072289156627, "grad_norm": 46.690486907958984, "learning_rate": 1.5155339805825244e-05, "loss": 0.2672, "step": 17478 }, { "epoch": 169.49036144578312, "grad_norm": 43.43992614746094, "learning_rate": 1.5150485436893205e-05, "loss": 0.2947, "step": 17479 }, { "epoch": 169.5, "grad_norm": 7.350569248199463, "learning_rate": 1.5145631067961166e-05, "loss": 0.0952, "step": 17480 }, { "epoch": 169.50963855421688, "grad_norm": 25.192447662353516, "learning_rate": 1.5140776699029127e-05, "loss": 0.2356, "step": 17481 }, { "epoch": 169.51927710843373, "grad_norm": 15.553492546081543, "learning_rate": 1.5135922330097088e-05, "loss": 0.1707, "step": 17482 }, { "epoch": 169.5289156626506, "grad_norm": 18.47991371154785, "learning_rate": 1.513106796116505e-05, "loss": 0.3081, "step": 17483 }, { "epoch": 169.53855421686748, "grad_norm": 6.726265907287598, "learning_rate": 1.512621359223301e-05, "loss": 0.1276, "step": 17484 }, { "epoch": 169.54819277108433, "grad_norm": 26.850177764892578, "learning_rate": 1.5121359223300972e-05, "loss": 0.1611, "step": 17485 }, { "epoch": 169.5578313253012, "grad_norm": 26.760610580444336, "learning_rate": 1.5116504854368931e-05, "loss": 0.1538, "step": 17486 }, { "epoch": 169.5674698795181, "grad_norm": 2.4542782306671143, "learning_rate": 1.5111650485436892e-05, "loss": 0.1497, "step": 17487 }, { "epoch": 169.57710843373494, "grad_norm": 25.3306827545166, "learning_rate": 1.5106796116504853e-05, "loss": 0.2034, "step": 17488 }, { "epoch": 169.5867469879518, "grad_norm": 6.541639804840088, "learning_rate": 1.5101941747572818e-05, "loss": 0.085, "step": 17489 }, { "epoch": 169.59638554216866, "grad_norm": 5.107230186462402, "learning_rate": 1.509708737864078e-05, "loss": 0.2249, "step": 17490 }, { "epoch": 169.60602409638554, "grad_norm": 6.702577590942383, "learning_rate": 1.5092233009708739e-05, "loss": 0.1666, "step": 17491 }, { "epoch": 169.61566265060242, "grad_norm": 8.594819068908691, "learning_rate": 1.50873786407767e-05, "loss": 0.1546, "step": 17492 }, { "epoch": 169.62530120481927, "grad_norm": 14.652993202209473, "learning_rate": 1.5082524271844661e-05, "loss": 0.2787, "step": 17493 }, { "epoch": 169.63493975903614, "grad_norm": 9.30240535736084, "learning_rate": 1.5077669902912622e-05, "loss": 0.1656, "step": 17494 }, { "epoch": 169.64457831325302, "grad_norm": 15.247933387756348, "learning_rate": 1.5072815533980583e-05, "loss": 0.2699, "step": 17495 }, { "epoch": 169.65421686746987, "grad_norm": 27.040815353393555, "learning_rate": 1.5067961165048544e-05, "loss": 0.2746, "step": 17496 }, { "epoch": 169.66385542168675, "grad_norm": 3.7844223976135254, "learning_rate": 1.5063106796116505e-05, "loss": 0.0892, "step": 17497 }, { "epoch": 169.67349397590363, "grad_norm": 2.263279438018799, "learning_rate": 1.5058252427184467e-05, "loss": 0.3203, "step": 17498 }, { "epoch": 169.68313253012047, "grad_norm": 12.881912231445312, "learning_rate": 1.5053398058252428e-05, "loss": 0.2281, "step": 17499 }, { "epoch": 169.69277108433735, "grad_norm": 24.61432456970215, "learning_rate": 1.5048543689320387e-05, "loss": 0.1751, "step": 17500 }, { "epoch": 169.70240963855423, "grad_norm": 5.1300950050354, "learning_rate": 1.5043689320388348e-05, "loss": 0.1476, "step": 17501 }, { "epoch": 169.71204819277108, "grad_norm": 1.6404520273208618, "learning_rate": 1.5038834951456313e-05, "loss": 0.295, "step": 17502 }, { "epoch": 169.72168674698796, "grad_norm": 5.83312463760376, "learning_rate": 1.5033980582524274e-05, "loss": 0.1717, "step": 17503 }, { "epoch": 169.73132530120483, "grad_norm": 34.06910705566406, "learning_rate": 1.5029126213592235e-05, "loss": 0.2338, "step": 17504 }, { "epoch": 169.74096385542168, "grad_norm": 11.05910587310791, "learning_rate": 1.5024271844660195e-05, "loss": 0.2109, "step": 17505 }, { "epoch": 169.75060240963856, "grad_norm": 4.8217926025390625, "learning_rate": 1.5019417475728156e-05, "loss": 0.1601, "step": 17506 }, { "epoch": 169.7602409638554, "grad_norm": 11.02816390991211, "learning_rate": 1.5014563106796117e-05, "loss": 0.3379, "step": 17507 }, { "epoch": 169.7698795180723, "grad_norm": 4.679788112640381, "learning_rate": 1.5009708737864078e-05, "loss": 0.22, "step": 17508 }, { "epoch": 169.77951807228916, "grad_norm": 3.8962602615356445, "learning_rate": 1.500485436893204e-05, "loss": 0.1356, "step": 17509 }, { "epoch": 169.789156626506, "grad_norm": 29.131576538085938, "learning_rate": 1.5e-05, "loss": 0.5397, "step": 17510 }, { "epoch": 169.7987951807229, "grad_norm": 1.819550633430481, "learning_rate": 1.4995145631067962e-05, "loss": 0.3952, "step": 17511 }, { "epoch": 169.80843373493977, "grad_norm": 14.660855293273926, "learning_rate": 1.4990291262135923e-05, "loss": 0.4523, "step": 17512 }, { "epoch": 169.81807228915662, "grad_norm": 17.258447647094727, "learning_rate": 1.4985436893203884e-05, "loss": 0.3614, "step": 17513 }, { "epoch": 169.8277108433735, "grad_norm": 12.976313591003418, "learning_rate": 1.4980582524271847e-05, "loss": 0.3097, "step": 17514 }, { "epoch": 169.83734939759037, "grad_norm": 27.450992584228516, "learning_rate": 1.4975728155339808e-05, "loss": 0.3866, "step": 17515 }, { "epoch": 169.84698795180722, "grad_norm": 72.06396484375, "learning_rate": 1.4970873786407769e-05, "loss": 0.3407, "step": 17516 }, { "epoch": 169.8566265060241, "grad_norm": 8.297052383422852, "learning_rate": 1.496601941747573e-05, "loss": 0.4055, "step": 17517 }, { "epoch": 169.86626506024098, "grad_norm": 5.06147575378418, "learning_rate": 1.4961165048543691e-05, "loss": 0.2173, "step": 17518 }, { "epoch": 169.87590361445783, "grad_norm": 46.00672149658203, "learning_rate": 1.495631067961165e-05, "loss": 0.4527, "step": 17519 }, { "epoch": 169.8855421686747, "grad_norm": 32.41291427612305, "learning_rate": 1.4951456310679612e-05, "loss": 0.1229, "step": 17520 }, { "epoch": 169.89518072289155, "grad_norm": 32.256046295166016, "learning_rate": 1.4946601941747573e-05, "loss": 0.212, "step": 17521 }, { "epoch": 169.90481927710843, "grad_norm": 45.698646545410156, "learning_rate": 1.4941747572815534e-05, "loss": 0.2966, "step": 17522 }, { "epoch": 169.9144578313253, "grad_norm": 13.860774040222168, "learning_rate": 1.4936893203883495e-05, "loss": 0.2086, "step": 17523 }, { "epoch": 169.92409638554216, "grad_norm": 7.1019182205200195, "learning_rate": 1.4932038834951456e-05, "loss": 0.1965, "step": 17524 }, { "epoch": 169.93373493975903, "grad_norm": 40.47434616088867, "learning_rate": 1.4927184466019418e-05, "loss": 0.3098, "step": 17525 }, { "epoch": 169.9433734939759, "grad_norm": 4.69584321975708, "learning_rate": 1.4922330097087379e-05, "loss": 0.1142, "step": 17526 }, { "epoch": 169.95301204819276, "grad_norm": 30.75456428527832, "learning_rate": 1.4917475728155342e-05, "loss": 0.1514, "step": 17527 }, { "epoch": 169.96265060240964, "grad_norm": 7.698364734649658, "learning_rate": 1.4912621359223303e-05, "loss": 0.1762, "step": 17528 }, { "epoch": 169.97228915662652, "grad_norm": 8.208451271057129, "learning_rate": 1.4907766990291264e-05, "loss": 0.2538, "step": 17529 }, { "epoch": 169.98192771084337, "grad_norm": 3.454967498779297, "learning_rate": 1.4902912621359225e-05, "loss": 0.2302, "step": 17530 }, { "epoch": 169.99156626506024, "grad_norm": 60.45987319946289, "learning_rate": 1.4898058252427186e-05, "loss": 0.2159, "step": 17531 }, { "epoch": 170.00722891566264, "grad_norm": 20.00916862487793, "learning_rate": 1.4893203883495147e-05, "loss": 0.2503, "step": 17532 }, { "epoch": 170.01686746987951, "grad_norm": 6.116373062133789, "learning_rate": 1.4888349514563107e-05, "loss": 0.4149, "step": 17533 }, { "epoch": 170.0265060240964, "grad_norm": 15.357522964477539, "learning_rate": 1.4883495145631068e-05, "loss": 0.2413, "step": 17534 }, { "epoch": 170.03614457831324, "grad_norm": 20.774818420410156, "learning_rate": 1.4878640776699029e-05, "loss": 0.28, "step": 17535 }, { "epoch": 170.04578313253012, "grad_norm": 7.049211502075195, "learning_rate": 1.487378640776699e-05, "loss": 0.5237, "step": 17536 }, { "epoch": 170.055421686747, "grad_norm": 14.896397590637207, "learning_rate": 1.4868932038834951e-05, "loss": 0.2694, "step": 17537 }, { "epoch": 170.06506024096385, "grad_norm": 20.139678955078125, "learning_rate": 1.4864077669902912e-05, "loss": 0.2137, "step": 17538 }, { "epoch": 170.07469879518072, "grad_norm": 32.93962860107422, "learning_rate": 1.4859223300970874e-05, "loss": 0.2603, "step": 17539 }, { "epoch": 170.0843373493976, "grad_norm": 38.72905349731445, "learning_rate": 1.4854368932038836e-05, "loss": 0.2652, "step": 17540 }, { "epoch": 170.09397590361445, "grad_norm": 35.68345260620117, "learning_rate": 1.4849514563106798e-05, "loss": 0.3832, "step": 17541 }, { "epoch": 170.10361445783133, "grad_norm": 6.200328826904297, "learning_rate": 1.4844660194174759e-05, "loss": 0.0862, "step": 17542 }, { "epoch": 170.1132530120482, "grad_norm": 26.787790298461914, "learning_rate": 1.483980582524272e-05, "loss": 0.3746, "step": 17543 }, { "epoch": 170.12289156626505, "grad_norm": 16.082483291625977, "learning_rate": 1.4834951456310681e-05, "loss": 0.1429, "step": 17544 }, { "epoch": 170.13253012048193, "grad_norm": 12.241867065429688, "learning_rate": 1.4830097087378642e-05, "loss": 0.2352, "step": 17545 }, { "epoch": 170.1421686746988, "grad_norm": 17.096628189086914, "learning_rate": 1.4825242718446603e-05, "loss": 0.1703, "step": 17546 }, { "epoch": 170.15180722891566, "grad_norm": 7.0609235763549805, "learning_rate": 1.4820388349514563e-05, "loss": 0.2908, "step": 17547 }, { "epoch": 170.16144578313254, "grad_norm": 4.41424560546875, "learning_rate": 1.4815533980582524e-05, "loss": 0.2728, "step": 17548 }, { "epoch": 170.17108433734938, "grad_norm": 14.102532386779785, "learning_rate": 1.4810679611650485e-05, "loss": 0.4433, "step": 17549 }, { "epoch": 170.18072289156626, "grad_norm": 13.532159805297852, "learning_rate": 1.4805825242718446e-05, "loss": 0.2297, "step": 17550 }, { "epoch": 170.19036144578314, "grad_norm": 44.79470443725586, "learning_rate": 1.4800970873786407e-05, "loss": 0.184, "step": 17551 }, { "epoch": 170.2, "grad_norm": 19.02023696899414, "learning_rate": 1.479611650485437e-05, "loss": 0.2221, "step": 17552 }, { "epoch": 170.20963855421687, "grad_norm": 11.78947639465332, "learning_rate": 1.4791262135922331e-05, "loss": 0.2524, "step": 17553 }, { "epoch": 170.21927710843374, "grad_norm": 10.288832664489746, "learning_rate": 1.4786407766990292e-05, "loss": 0.1481, "step": 17554 }, { "epoch": 170.2289156626506, "grad_norm": 14.17890739440918, "learning_rate": 1.4781553398058254e-05, "loss": 0.2208, "step": 17555 }, { "epoch": 170.23855421686747, "grad_norm": 14.81292724609375, "learning_rate": 1.4776699029126215e-05, "loss": 0.3396, "step": 17556 }, { "epoch": 170.24819277108435, "grad_norm": 3.7082579135894775, "learning_rate": 1.4771844660194176e-05, "loss": 0.4241, "step": 17557 }, { "epoch": 170.2578313253012, "grad_norm": 8.29366683959961, "learning_rate": 1.4766990291262137e-05, "loss": 0.3597, "step": 17558 }, { "epoch": 170.26746987951807, "grad_norm": 7.635535717010498, "learning_rate": 1.4762135922330098e-05, "loss": 0.26, "step": 17559 }, { "epoch": 170.27710843373495, "grad_norm": 35.599857330322266, "learning_rate": 1.4757281553398058e-05, "loss": 0.2541, "step": 17560 }, { "epoch": 170.2867469879518, "grad_norm": 25.57312774658203, "learning_rate": 1.4752427184466019e-05, "loss": 0.3319, "step": 17561 }, { "epoch": 170.29638554216868, "grad_norm": 12.61276912689209, "learning_rate": 1.474757281553398e-05, "loss": 0.3753, "step": 17562 }, { "epoch": 170.30602409638556, "grad_norm": 29.061330795288086, "learning_rate": 1.4742718446601941e-05, "loss": 0.1106, "step": 17563 }, { "epoch": 170.3156626506024, "grad_norm": 9.544159889221191, "learning_rate": 1.4737864077669902e-05, "loss": 0.144, "step": 17564 }, { "epoch": 170.32530120481928, "grad_norm": 19.203372955322266, "learning_rate": 1.4733009708737867e-05, "loss": 0.1178, "step": 17565 }, { "epoch": 170.33493975903613, "grad_norm": 12.851895332336426, "learning_rate": 1.4728155339805826e-05, "loss": 0.2923, "step": 17566 }, { "epoch": 170.344578313253, "grad_norm": 10.373446464538574, "learning_rate": 1.4723300970873787e-05, "loss": 0.2249, "step": 17567 }, { "epoch": 170.3542168674699, "grad_norm": 20.58281135559082, "learning_rate": 1.4718446601941749e-05, "loss": 0.2554, "step": 17568 }, { "epoch": 170.36385542168674, "grad_norm": 16.186845779418945, "learning_rate": 1.471359223300971e-05, "loss": 0.2265, "step": 17569 }, { "epoch": 170.3734939759036, "grad_norm": 1.945249080657959, "learning_rate": 1.470873786407767e-05, "loss": 0.0934, "step": 17570 }, { "epoch": 170.3831325301205, "grad_norm": 15.314664840698242, "learning_rate": 1.4703883495145632e-05, "loss": 0.2562, "step": 17571 }, { "epoch": 170.39277108433734, "grad_norm": 11.486319541931152, "learning_rate": 1.4699029126213593e-05, "loss": 0.103, "step": 17572 }, { "epoch": 170.40240963855422, "grad_norm": 10.111398696899414, "learning_rate": 1.4694174757281554e-05, "loss": 0.3073, "step": 17573 }, { "epoch": 170.4120481927711, "grad_norm": 33.27023696899414, "learning_rate": 1.4689320388349514e-05, "loss": 0.1948, "step": 17574 }, { "epoch": 170.42168674698794, "grad_norm": 4.598313808441162, "learning_rate": 1.4684466019417475e-05, "loss": 0.178, "step": 17575 }, { "epoch": 170.43132530120482, "grad_norm": 33.755157470703125, "learning_rate": 1.4679611650485436e-05, "loss": 0.2822, "step": 17576 }, { "epoch": 170.4409638554217, "grad_norm": 2.3824357986450195, "learning_rate": 1.46747572815534e-05, "loss": 0.1506, "step": 17577 }, { "epoch": 170.45060240963855, "grad_norm": 17.530567169189453, "learning_rate": 1.4669902912621362e-05, "loss": 0.3631, "step": 17578 }, { "epoch": 170.46024096385543, "grad_norm": 26.860782623291016, "learning_rate": 1.4665048543689323e-05, "loss": 0.1824, "step": 17579 }, { "epoch": 170.46987951807228, "grad_norm": 18.973697662353516, "learning_rate": 1.4660194174757282e-05, "loss": 0.3308, "step": 17580 }, { "epoch": 170.47951807228915, "grad_norm": 14.135689735412598, "learning_rate": 1.4655339805825243e-05, "loss": 0.2152, "step": 17581 }, { "epoch": 170.48915662650603, "grad_norm": 30.306718826293945, "learning_rate": 1.4650485436893205e-05, "loss": 0.4268, "step": 17582 }, { "epoch": 170.49879518072288, "grad_norm": 8.226619720458984, "learning_rate": 1.4645631067961166e-05, "loss": 0.3266, "step": 17583 }, { "epoch": 170.50843373493976, "grad_norm": 10.076857566833496, "learning_rate": 1.4640776699029127e-05, "loss": 0.2349, "step": 17584 }, { "epoch": 170.51807228915663, "grad_norm": 14.622603416442871, "learning_rate": 1.4635922330097088e-05, "loss": 0.1517, "step": 17585 }, { "epoch": 170.52771084337348, "grad_norm": 61.506919860839844, "learning_rate": 1.4631067961165049e-05, "loss": 0.2235, "step": 17586 }, { "epoch": 170.53734939759036, "grad_norm": 4.242624282836914, "learning_rate": 1.462621359223301e-05, "loss": 0.2668, "step": 17587 }, { "epoch": 170.54698795180724, "grad_norm": 5.107911586761475, "learning_rate": 1.462135922330097e-05, "loss": 0.1695, "step": 17588 }, { "epoch": 170.5566265060241, "grad_norm": 16.251100540161133, "learning_rate": 1.4616504854368931e-05, "loss": 0.2273, "step": 17589 }, { "epoch": 170.56626506024097, "grad_norm": 27.175743103027344, "learning_rate": 1.4611650485436895e-05, "loss": 0.6852, "step": 17590 }, { "epoch": 170.57590361445784, "grad_norm": 24.459402084350586, "learning_rate": 1.4606796116504857e-05, "loss": 0.1126, "step": 17591 }, { "epoch": 170.5855421686747, "grad_norm": 25.355756759643555, "learning_rate": 1.4601941747572818e-05, "loss": 0.375, "step": 17592 }, { "epoch": 170.59518072289157, "grad_norm": 17.570823669433594, "learning_rate": 1.4597087378640779e-05, "loss": 0.1269, "step": 17593 }, { "epoch": 170.60481927710845, "grad_norm": 14.090863227844238, "learning_rate": 1.4592233009708738e-05, "loss": 0.2467, "step": 17594 }, { "epoch": 170.6144578313253, "grad_norm": 4.297793388366699, "learning_rate": 1.45873786407767e-05, "loss": 0.2408, "step": 17595 }, { "epoch": 170.62409638554217, "grad_norm": 21.75481414794922, "learning_rate": 1.458252427184466e-05, "loss": 0.1743, "step": 17596 }, { "epoch": 170.63373493975902, "grad_norm": 5.5409746170043945, "learning_rate": 1.4577669902912622e-05, "loss": 0.2624, "step": 17597 }, { "epoch": 170.6433734939759, "grad_norm": 3.417517900466919, "learning_rate": 1.4572815533980583e-05, "loss": 0.1237, "step": 17598 }, { "epoch": 170.65301204819278, "grad_norm": 8.347728729248047, "learning_rate": 1.4567961165048544e-05, "loss": 0.1899, "step": 17599 }, { "epoch": 170.66265060240963, "grad_norm": 2.035836935043335, "learning_rate": 1.4563106796116505e-05, "loss": 0.1787, "step": 17600 }, { "epoch": 170.6722891566265, "grad_norm": 4.311812400817871, "learning_rate": 1.4558252427184466e-05, "loss": 0.1852, "step": 17601 }, { "epoch": 170.68192771084338, "grad_norm": 25.319473266601562, "learning_rate": 1.4553398058252426e-05, "loss": 0.3912, "step": 17602 }, { "epoch": 170.69156626506023, "grad_norm": 20.268714904785156, "learning_rate": 1.454854368932039e-05, "loss": 0.2533, "step": 17603 }, { "epoch": 170.7012048192771, "grad_norm": 3.9752087593078613, "learning_rate": 1.4543689320388351e-05, "loss": 0.2159, "step": 17604 }, { "epoch": 170.710843373494, "grad_norm": 1.8194711208343506, "learning_rate": 1.4538834951456313e-05, "loss": 0.2609, "step": 17605 }, { "epoch": 170.72048192771084, "grad_norm": 13.085565567016602, "learning_rate": 1.4533980582524274e-05, "loss": 0.331, "step": 17606 }, { "epoch": 170.7301204819277, "grad_norm": 21.86210060119629, "learning_rate": 1.4529126213592233e-05, "loss": 0.1751, "step": 17607 }, { "epoch": 170.7397590361446, "grad_norm": 30.173234939575195, "learning_rate": 1.4524271844660194e-05, "loss": 0.3891, "step": 17608 }, { "epoch": 170.74939759036144, "grad_norm": 20.944974899291992, "learning_rate": 1.4519417475728156e-05, "loss": 0.3898, "step": 17609 }, { "epoch": 170.75903614457832, "grad_norm": 15.473711967468262, "learning_rate": 1.4514563106796117e-05, "loss": 0.3254, "step": 17610 }, { "epoch": 170.76867469879517, "grad_norm": 8.178641319274902, "learning_rate": 1.4509708737864078e-05, "loss": 0.29, "step": 17611 }, { "epoch": 170.77831325301204, "grad_norm": 3.775438070297241, "learning_rate": 1.4504854368932039e-05, "loss": 0.1091, "step": 17612 }, { "epoch": 170.78795180722892, "grad_norm": 7.827235221862793, "learning_rate": 1.45e-05, "loss": 0.2144, "step": 17613 }, { "epoch": 170.79759036144577, "grad_norm": 4.374369144439697, "learning_rate": 1.4495145631067961e-05, "loss": 0.1432, "step": 17614 }, { "epoch": 170.80722891566265, "grad_norm": 26.359254837036133, "learning_rate": 1.4490291262135924e-05, "loss": 0.2179, "step": 17615 }, { "epoch": 170.81686746987953, "grad_norm": 105.15934753417969, "learning_rate": 1.4485436893203885e-05, "loss": 0.389, "step": 17616 }, { "epoch": 170.82650602409637, "grad_norm": 9.502156257629395, "learning_rate": 1.4480582524271846e-05, "loss": 0.1236, "step": 17617 }, { "epoch": 170.83614457831325, "grad_norm": 23.84909439086914, "learning_rate": 1.4475728155339808e-05, "loss": 0.234, "step": 17618 }, { "epoch": 170.84578313253013, "grad_norm": 30.29529571533203, "learning_rate": 1.4470873786407769e-05, "loss": 0.2176, "step": 17619 }, { "epoch": 170.85542168674698, "grad_norm": 16.29684829711914, "learning_rate": 1.446601941747573e-05, "loss": 0.1554, "step": 17620 }, { "epoch": 170.86506024096386, "grad_norm": 22.829092025756836, "learning_rate": 1.446116504854369e-05, "loss": 0.191, "step": 17621 }, { "epoch": 170.87469879518073, "grad_norm": 12.493043899536133, "learning_rate": 1.445631067961165e-05, "loss": 0.2455, "step": 17622 }, { "epoch": 170.88433734939758, "grad_norm": 17.05196762084961, "learning_rate": 1.4451456310679612e-05, "loss": 0.1997, "step": 17623 }, { "epoch": 170.89397590361446, "grad_norm": 13.546090126037598, "learning_rate": 1.4446601941747573e-05, "loss": 0.3329, "step": 17624 }, { "epoch": 170.90361445783134, "grad_norm": 42.026790618896484, "learning_rate": 1.4441747572815534e-05, "loss": 0.2973, "step": 17625 }, { "epoch": 170.9132530120482, "grad_norm": 9.516012191772461, "learning_rate": 1.4436893203883495e-05, "loss": 0.3061, "step": 17626 }, { "epoch": 170.92289156626506, "grad_norm": 5.745837211608887, "learning_rate": 1.4432038834951456e-05, "loss": 0.1868, "step": 17627 }, { "epoch": 170.9325301204819, "grad_norm": 6.224660396575928, "learning_rate": 1.4427184466019419e-05, "loss": 0.2555, "step": 17628 }, { "epoch": 170.9421686746988, "grad_norm": 22.27789306640625, "learning_rate": 1.442233009708738e-05, "loss": 0.2487, "step": 17629 }, { "epoch": 170.95180722891567, "grad_norm": 14.864327430725098, "learning_rate": 1.4417475728155341e-05, "loss": 0.2961, "step": 17630 }, { "epoch": 170.96144578313252, "grad_norm": 15.119646072387695, "learning_rate": 1.4412621359223302e-05, "loss": 0.227, "step": 17631 }, { "epoch": 170.9710843373494, "grad_norm": 13.410346031188965, "learning_rate": 1.4407766990291264e-05, "loss": 0.1116, "step": 17632 }, { "epoch": 170.98072289156627, "grad_norm": 6.271625995635986, "learning_rate": 1.4402912621359225e-05, "loss": 0.1647, "step": 17633 }, { "epoch": 170.99036144578312, "grad_norm": 16.519691467285156, "learning_rate": 1.4398058252427186e-05, "loss": 0.3375, "step": 17634 }, { "epoch": 171.00602409638554, "grad_norm": 8.74892520904541, "learning_rate": 1.4393203883495145e-05, "loss": 0.1263, "step": 17635 }, { "epoch": 171.01566265060242, "grad_norm": 5.684459209442139, "learning_rate": 1.4388349514563106e-05, "loss": 0.2624, "step": 17636 }, { "epoch": 171.02530120481927, "grad_norm": 12.619385719299316, "learning_rate": 1.4383495145631068e-05, "loss": 0.1474, "step": 17637 }, { "epoch": 171.03493975903615, "grad_norm": 9.135810852050781, "learning_rate": 1.4378640776699029e-05, "loss": 0.1927, "step": 17638 }, { "epoch": 171.044578313253, "grad_norm": 68.31782531738281, "learning_rate": 1.437378640776699e-05, "loss": 0.225, "step": 17639 }, { "epoch": 171.05421686746988, "grad_norm": 38.08085250854492, "learning_rate": 1.4368932038834951e-05, "loss": 0.3029, "step": 17640 }, { "epoch": 171.06385542168675, "grad_norm": 25.921052932739258, "learning_rate": 1.4364077669902914e-05, "loss": 0.2033, "step": 17641 }, { "epoch": 171.0734939759036, "grad_norm": 18.069408416748047, "learning_rate": 1.4359223300970875e-05, "loss": 0.2207, "step": 17642 }, { "epoch": 171.08313253012048, "grad_norm": 7.225724220275879, "learning_rate": 1.4354368932038836e-05, "loss": 0.2345, "step": 17643 }, { "epoch": 171.09277108433736, "grad_norm": 3.2091526985168457, "learning_rate": 1.4349514563106797e-05, "loss": 0.1829, "step": 17644 }, { "epoch": 171.1024096385542, "grad_norm": 60.150550842285156, "learning_rate": 1.4344660194174758e-05, "loss": 0.2005, "step": 17645 }, { "epoch": 171.11204819277108, "grad_norm": 6.104107856750488, "learning_rate": 1.433980582524272e-05, "loss": 0.1455, "step": 17646 }, { "epoch": 171.12168674698796, "grad_norm": 32.659751892089844, "learning_rate": 1.433495145631068e-05, "loss": 0.2556, "step": 17647 }, { "epoch": 171.1313253012048, "grad_norm": 14.004858016967773, "learning_rate": 1.4330097087378642e-05, "loss": 0.1848, "step": 17648 }, { "epoch": 171.1409638554217, "grad_norm": 8.227553367614746, "learning_rate": 1.4325242718446601e-05, "loss": 0.2401, "step": 17649 }, { "epoch": 171.15060240963857, "grad_norm": 29.260019302368164, "learning_rate": 1.4320388349514562e-05, "loss": 0.2371, "step": 17650 }, { "epoch": 171.16024096385541, "grad_norm": 6.964827537536621, "learning_rate": 1.4315533980582524e-05, "loss": 0.1321, "step": 17651 }, { "epoch": 171.1698795180723, "grad_norm": 17.06523895263672, "learning_rate": 1.4310679611650485e-05, "loss": 0.2642, "step": 17652 }, { "epoch": 171.17951807228917, "grad_norm": 28.862558364868164, "learning_rate": 1.430582524271845e-05, "loss": 0.2488, "step": 17653 }, { "epoch": 171.18915662650602, "grad_norm": 19.262968063354492, "learning_rate": 1.4300970873786409e-05, "loss": 0.1676, "step": 17654 }, { "epoch": 171.1987951807229, "grad_norm": 13.82264518737793, "learning_rate": 1.429611650485437e-05, "loss": 0.3002, "step": 17655 }, { "epoch": 171.20843373493975, "grad_norm": 22.75909996032715, "learning_rate": 1.4291262135922331e-05, "loss": 0.2391, "step": 17656 }, { "epoch": 171.21807228915662, "grad_norm": 21.634756088256836, "learning_rate": 1.4286407766990292e-05, "loss": 0.4845, "step": 17657 }, { "epoch": 171.2277108433735, "grad_norm": 9.164633750915527, "learning_rate": 1.4281553398058253e-05, "loss": 0.3021, "step": 17658 }, { "epoch": 171.23734939759035, "grad_norm": 17.85363006591797, "learning_rate": 1.4276699029126214e-05, "loss": 0.2349, "step": 17659 }, { "epoch": 171.24698795180723, "grad_norm": 8.807955741882324, "learning_rate": 1.4271844660194176e-05, "loss": 0.2404, "step": 17660 }, { "epoch": 171.2566265060241, "grad_norm": 14.875003814697266, "learning_rate": 1.4266990291262137e-05, "loss": 0.2426, "step": 17661 }, { "epoch": 171.26626506024095, "grad_norm": 20.73715591430664, "learning_rate": 1.4262135922330098e-05, "loss": 0.3421, "step": 17662 }, { "epoch": 171.27590361445783, "grad_norm": 13.481966972351074, "learning_rate": 1.4257281553398057e-05, "loss": 0.0909, "step": 17663 }, { "epoch": 171.2855421686747, "grad_norm": 10.4901704788208, "learning_rate": 1.4252427184466019e-05, "loss": 0.2404, "step": 17664 }, { "epoch": 171.29518072289156, "grad_norm": 8.689409255981445, "learning_rate": 1.424757281553398e-05, "loss": 0.1421, "step": 17665 }, { "epoch": 171.30481927710844, "grad_norm": 15.497360229492188, "learning_rate": 1.4242718446601944e-05, "loss": 0.2125, "step": 17666 }, { "epoch": 171.3144578313253, "grad_norm": 6.994988918304443, "learning_rate": 1.4237864077669905e-05, "loss": 0.3989, "step": 17667 }, { "epoch": 171.32409638554216, "grad_norm": 29.79239273071289, "learning_rate": 1.4233009708737865e-05, "loss": 0.3236, "step": 17668 }, { "epoch": 171.33373493975904, "grad_norm": 7.655212879180908, "learning_rate": 1.4228155339805826e-05, "loss": 0.236, "step": 17669 }, { "epoch": 171.34337349397592, "grad_norm": 17.257604598999023, "learning_rate": 1.4223300970873787e-05, "loss": 0.1942, "step": 17670 }, { "epoch": 171.35301204819277, "grad_norm": 37.48347473144531, "learning_rate": 1.4218446601941748e-05, "loss": 0.2974, "step": 17671 }, { "epoch": 171.36265060240964, "grad_norm": 4.904196739196777, "learning_rate": 1.421359223300971e-05, "loss": 0.3268, "step": 17672 }, { "epoch": 171.3722891566265, "grad_norm": 12.414505004882812, "learning_rate": 1.420873786407767e-05, "loss": 0.3852, "step": 17673 }, { "epoch": 171.38192771084337, "grad_norm": 15.99860668182373, "learning_rate": 1.4203883495145632e-05, "loss": 0.3067, "step": 17674 }, { "epoch": 171.39156626506025, "grad_norm": 28.289962768554688, "learning_rate": 1.4199029126213593e-05, "loss": 0.3456, "step": 17675 }, { "epoch": 171.4012048192771, "grad_norm": 3.492307662963867, "learning_rate": 1.4194174757281552e-05, "loss": 0.1994, "step": 17676 }, { "epoch": 171.41084337349398, "grad_norm": 15.848274230957031, "learning_rate": 1.4189320388349513e-05, "loss": 0.176, "step": 17677 }, { "epoch": 171.42048192771085, "grad_norm": 54.57526397705078, "learning_rate": 1.4184466019417478e-05, "loss": 0.4041, "step": 17678 }, { "epoch": 171.4301204819277, "grad_norm": 26.264196395874023, "learning_rate": 1.4179611650485439e-05, "loss": 0.1748, "step": 17679 }, { "epoch": 171.43975903614458, "grad_norm": 11.774928092956543, "learning_rate": 1.41747572815534e-05, "loss": 0.2786, "step": 17680 }, { "epoch": 171.44939759036146, "grad_norm": 11.194076538085938, "learning_rate": 1.4169902912621361e-05, "loss": 0.3245, "step": 17681 }, { "epoch": 171.4590361445783, "grad_norm": 25.091344833374023, "learning_rate": 1.416504854368932e-05, "loss": 0.4991, "step": 17682 }, { "epoch": 171.46867469879518, "grad_norm": 8.29108715057373, "learning_rate": 1.4160194174757282e-05, "loss": 0.3289, "step": 17683 }, { "epoch": 171.47831325301206, "grad_norm": 25.03887939453125, "learning_rate": 1.4155339805825243e-05, "loss": 0.4514, "step": 17684 }, { "epoch": 171.4879518072289, "grad_norm": 48.41114044189453, "learning_rate": 1.4150485436893204e-05, "loss": 0.185, "step": 17685 }, { "epoch": 171.4975903614458, "grad_norm": 27.558223724365234, "learning_rate": 1.4145631067961165e-05, "loss": 0.4577, "step": 17686 }, { "epoch": 171.50722891566264, "grad_norm": 48.09950256347656, "learning_rate": 1.4140776699029127e-05, "loss": 0.4557, "step": 17687 }, { "epoch": 171.51686746987951, "grad_norm": 4.451396465301514, "learning_rate": 1.4135922330097088e-05, "loss": 0.0774, "step": 17688 }, { "epoch": 171.5265060240964, "grad_norm": 32.412803649902344, "learning_rate": 1.4131067961165049e-05, "loss": 0.4382, "step": 17689 }, { "epoch": 171.53614457831324, "grad_norm": 30.176652908325195, "learning_rate": 1.4126213592233008e-05, "loss": 0.4083, "step": 17690 }, { "epoch": 171.54578313253012, "grad_norm": 7.229525089263916, "learning_rate": 1.4121359223300973e-05, "loss": 0.1234, "step": 17691 }, { "epoch": 171.555421686747, "grad_norm": 6.165441036224365, "learning_rate": 1.4116504854368934e-05, "loss": 0.3404, "step": 17692 }, { "epoch": 171.56506024096385, "grad_norm": 7.9654741287231445, "learning_rate": 1.4111650485436895e-05, "loss": 0.1194, "step": 17693 }, { "epoch": 171.57469879518072, "grad_norm": 30.171091079711914, "learning_rate": 1.4106796116504856e-05, "loss": 0.2742, "step": 17694 }, { "epoch": 171.5843373493976, "grad_norm": 8.854000091552734, "learning_rate": 1.4101941747572817e-05, "loss": 0.2181, "step": 17695 }, { "epoch": 171.59397590361445, "grad_norm": 2.6582865715026855, "learning_rate": 1.4097087378640777e-05, "loss": 0.137, "step": 17696 }, { "epoch": 171.60361445783133, "grad_norm": 25.59973907470703, "learning_rate": 1.4092233009708738e-05, "loss": 0.2444, "step": 17697 }, { "epoch": 171.6132530120482, "grad_norm": 6.153897762298584, "learning_rate": 1.40873786407767e-05, "loss": 0.2573, "step": 17698 }, { "epoch": 171.62289156626505, "grad_norm": 16.259117126464844, "learning_rate": 1.408252427184466e-05, "loss": 0.352, "step": 17699 }, { "epoch": 171.63253012048193, "grad_norm": 7.875236988067627, "learning_rate": 1.4077669902912621e-05, "loss": 0.2932, "step": 17700 }, { "epoch": 171.6421686746988, "grad_norm": 5.195522785186768, "learning_rate": 1.4072815533980583e-05, "loss": 0.2873, "step": 17701 }, { "epoch": 171.65180722891566, "grad_norm": 28.533430099487305, "learning_rate": 1.4067961165048544e-05, "loss": 0.2035, "step": 17702 }, { "epoch": 171.66144578313254, "grad_norm": 23.93749237060547, "learning_rate": 1.4063106796116505e-05, "loss": 0.2058, "step": 17703 }, { "epoch": 171.67108433734938, "grad_norm": 27.311962127685547, "learning_rate": 1.4058252427184468e-05, "loss": 0.0861, "step": 17704 }, { "epoch": 171.68072289156626, "grad_norm": 7.120009899139404, "learning_rate": 1.4053398058252429e-05, "loss": 0.1529, "step": 17705 }, { "epoch": 171.69036144578314, "grad_norm": 59.682613372802734, "learning_rate": 1.404854368932039e-05, "loss": 0.1853, "step": 17706 }, { "epoch": 171.7, "grad_norm": 15.850255012512207, "learning_rate": 1.4043689320388351e-05, "loss": 0.2681, "step": 17707 }, { "epoch": 171.70963855421687, "grad_norm": 69.73804473876953, "learning_rate": 1.4038834951456312e-05, "loss": 0.2579, "step": 17708 }, { "epoch": 171.71927710843374, "grad_norm": 10.452548027038574, "learning_rate": 1.4033980582524272e-05, "loss": 0.0854, "step": 17709 }, { "epoch": 171.7289156626506, "grad_norm": 13.210596084594727, "learning_rate": 1.4029126213592233e-05, "loss": 0.1288, "step": 17710 }, { "epoch": 171.73855421686747, "grad_norm": 14.67188835144043, "learning_rate": 1.4024271844660194e-05, "loss": 0.247, "step": 17711 }, { "epoch": 171.74819277108435, "grad_norm": 29.289411544799805, "learning_rate": 1.4019417475728155e-05, "loss": 0.38, "step": 17712 }, { "epoch": 171.7578313253012, "grad_norm": 19.990829467773438, "learning_rate": 1.4014563106796116e-05, "loss": 0.2299, "step": 17713 }, { "epoch": 171.76746987951807, "grad_norm": 23.683149337768555, "learning_rate": 1.4009708737864077e-05, "loss": 0.4022, "step": 17714 }, { "epoch": 171.77710843373495, "grad_norm": 5.896004676818848, "learning_rate": 1.4004854368932039e-05, "loss": 0.1903, "step": 17715 }, { "epoch": 171.7867469879518, "grad_norm": 6.8828582763671875, "learning_rate": 1.4000000000000001e-05, "loss": 0.2824, "step": 17716 }, { "epoch": 171.79638554216868, "grad_norm": 8.995288848876953, "learning_rate": 1.3995145631067963e-05, "loss": 0.2206, "step": 17717 }, { "epoch": 171.80602409638556, "grad_norm": 27.83450698852539, "learning_rate": 1.3990291262135924e-05, "loss": 0.2961, "step": 17718 }, { "epoch": 171.8156626506024, "grad_norm": 18.657785415649414, "learning_rate": 1.3985436893203885e-05, "loss": 0.2925, "step": 17719 }, { "epoch": 171.82530120481928, "grad_norm": 14.856956481933594, "learning_rate": 1.3980582524271846e-05, "loss": 0.1661, "step": 17720 }, { "epoch": 171.83493975903613, "grad_norm": 8.808192253112793, "learning_rate": 1.3975728155339807e-05, "loss": 0.1544, "step": 17721 }, { "epoch": 171.844578313253, "grad_norm": 18.609294891357422, "learning_rate": 1.3970873786407768e-05, "loss": 0.2318, "step": 17722 }, { "epoch": 171.8542168674699, "grad_norm": 9.091334342956543, "learning_rate": 1.3966019417475728e-05, "loss": 0.3451, "step": 17723 }, { "epoch": 171.86385542168674, "grad_norm": 6.086052417755127, "learning_rate": 1.3961165048543689e-05, "loss": 0.236, "step": 17724 }, { "epoch": 171.8734939759036, "grad_norm": 14.386574745178223, "learning_rate": 1.395631067961165e-05, "loss": 0.3021, "step": 17725 }, { "epoch": 171.8831325301205, "grad_norm": 21.9515438079834, "learning_rate": 1.3951456310679611e-05, "loss": 0.3242, "step": 17726 }, { "epoch": 171.89277108433734, "grad_norm": 20.172948837280273, "learning_rate": 1.3946601941747572e-05, "loss": 0.4772, "step": 17727 }, { "epoch": 171.90240963855422, "grad_norm": 5.177521228790283, "learning_rate": 1.3941747572815534e-05, "loss": 0.1016, "step": 17728 }, { "epoch": 171.9120481927711, "grad_norm": 6.149126052856445, "learning_rate": 1.3936893203883496e-05, "loss": 0.1027, "step": 17729 }, { "epoch": 171.92168674698794, "grad_norm": 10.493465423583984, "learning_rate": 1.3932038834951458e-05, "loss": 0.2288, "step": 17730 }, { "epoch": 171.93132530120482, "grad_norm": 20.05131721496582, "learning_rate": 1.3927184466019419e-05, "loss": 0.2847, "step": 17731 }, { "epoch": 171.9409638554217, "grad_norm": 6.8987884521484375, "learning_rate": 1.392233009708738e-05, "loss": 0.0999, "step": 17732 }, { "epoch": 171.95060240963855, "grad_norm": 6.37957239151001, "learning_rate": 1.3917475728155341e-05, "loss": 0.0713, "step": 17733 }, { "epoch": 171.96024096385543, "grad_norm": 13.386092185974121, "learning_rate": 1.3912621359223302e-05, "loss": 0.2537, "step": 17734 }, { "epoch": 171.96987951807228, "grad_norm": 18.122512817382812, "learning_rate": 1.3907766990291263e-05, "loss": 0.1852, "step": 17735 }, { "epoch": 171.97951807228915, "grad_norm": 15.396363258361816, "learning_rate": 1.3902912621359224e-05, "loss": 0.4108, "step": 17736 }, { "epoch": 171.98915662650603, "grad_norm": 16.768211364746094, "learning_rate": 1.3898058252427184e-05, "loss": 0.3095, "step": 17737 }, { "epoch": 172.00481927710842, "grad_norm": 9.400959968566895, "learning_rate": 1.3893203883495145e-05, "loss": 0.2349, "step": 17738 }, { "epoch": 172.0144578313253, "grad_norm": 9.927162170410156, "learning_rate": 1.3888349514563106e-05, "loss": 0.303, "step": 17739 }, { "epoch": 172.02409638554218, "grad_norm": 55.90419387817383, "learning_rate": 1.3883495145631067e-05, "loss": 0.1886, "step": 17740 }, { "epoch": 172.03373493975903, "grad_norm": 17.168270111083984, "learning_rate": 1.3878640776699032e-05, "loss": 0.1172, "step": 17741 }, { "epoch": 172.0433734939759, "grad_norm": 7.440634727478027, "learning_rate": 1.3873786407766993e-05, "loss": 0.2036, "step": 17742 }, { "epoch": 172.05301204819278, "grad_norm": 20.037677764892578, "learning_rate": 1.3868932038834952e-05, "loss": 0.1082, "step": 17743 }, { "epoch": 172.06265060240963, "grad_norm": 21.998559951782227, "learning_rate": 1.3864077669902914e-05, "loss": 0.2636, "step": 17744 }, { "epoch": 172.0722891566265, "grad_norm": 4.426100254058838, "learning_rate": 1.3859223300970875e-05, "loss": 0.2125, "step": 17745 }, { "epoch": 172.08192771084336, "grad_norm": 57.429237365722656, "learning_rate": 1.3854368932038836e-05, "loss": 0.3538, "step": 17746 }, { "epoch": 172.09156626506024, "grad_norm": 21.358877182006836, "learning_rate": 1.3849514563106797e-05, "loss": 0.3157, "step": 17747 }, { "epoch": 172.10120481927711, "grad_norm": 9.52188491821289, "learning_rate": 1.3844660194174758e-05, "loss": 0.185, "step": 17748 }, { "epoch": 172.11084337349396, "grad_norm": 7.022245407104492, "learning_rate": 1.383980582524272e-05, "loss": 0.2297, "step": 17749 }, { "epoch": 172.12048192771084, "grad_norm": 7.623751640319824, "learning_rate": 1.383495145631068e-05, "loss": 0.2315, "step": 17750 }, { "epoch": 172.13012048192772, "grad_norm": 4.109760284423828, "learning_rate": 1.383009708737864e-05, "loss": 0.2072, "step": 17751 }, { "epoch": 172.13975903614457, "grad_norm": 17.836828231811523, "learning_rate": 1.3825242718446601e-05, "loss": 0.0556, "step": 17752 }, { "epoch": 172.14939759036145, "grad_norm": 20.09939956665039, "learning_rate": 1.3820388349514562e-05, "loss": 0.3073, "step": 17753 }, { "epoch": 172.15903614457832, "grad_norm": 45.21654510498047, "learning_rate": 1.3815533980582527e-05, "loss": 0.2821, "step": 17754 }, { "epoch": 172.16867469879517, "grad_norm": 7.741591453552246, "learning_rate": 1.3810679611650488e-05, "loss": 0.1703, "step": 17755 }, { "epoch": 172.17831325301205, "grad_norm": 17.554405212402344, "learning_rate": 1.3805825242718447e-05, "loss": 0.1725, "step": 17756 }, { "epoch": 172.18795180722893, "grad_norm": 35.50510787963867, "learning_rate": 1.3800970873786408e-05, "loss": 0.1876, "step": 17757 }, { "epoch": 172.19759036144578, "grad_norm": 11.191900253295898, "learning_rate": 1.379611650485437e-05, "loss": 0.3265, "step": 17758 }, { "epoch": 172.20722891566265, "grad_norm": 16.95639419555664, "learning_rate": 1.379126213592233e-05, "loss": 0.243, "step": 17759 }, { "epoch": 172.21686746987953, "grad_norm": 5.391584873199463, "learning_rate": 1.3786407766990292e-05, "loss": 0.1188, "step": 17760 }, { "epoch": 172.22650602409638, "grad_norm": 4.060215473175049, "learning_rate": 1.3781553398058253e-05, "loss": 0.1933, "step": 17761 }, { "epoch": 172.23614457831326, "grad_norm": 6.403021812438965, "learning_rate": 1.3776699029126214e-05, "loss": 0.2165, "step": 17762 }, { "epoch": 172.2457831325301, "grad_norm": 5.780937671661377, "learning_rate": 1.3771844660194175e-05, "loss": 0.2074, "step": 17763 }, { "epoch": 172.25542168674698, "grad_norm": 3.410109758377075, "learning_rate": 1.3766990291262136e-05, "loss": 0.1999, "step": 17764 }, { "epoch": 172.26506024096386, "grad_norm": 10.779540061950684, "learning_rate": 1.3762135922330096e-05, "loss": 0.2596, "step": 17765 }, { "epoch": 172.2746987951807, "grad_norm": 31.315937042236328, "learning_rate": 1.3757281553398057e-05, "loss": 0.4005, "step": 17766 }, { "epoch": 172.2843373493976, "grad_norm": 48.374080657958984, "learning_rate": 1.3752427184466022e-05, "loss": 0.2762, "step": 17767 }, { "epoch": 172.29397590361447, "grad_norm": 7.295719146728516, "learning_rate": 1.3747572815533983e-05, "loss": 0.3004, "step": 17768 }, { "epoch": 172.30361445783132, "grad_norm": 13.595786094665527, "learning_rate": 1.3742718446601944e-05, "loss": 0.3299, "step": 17769 }, { "epoch": 172.3132530120482, "grad_norm": 6.124202251434326, "learning_rate": 1.3737864077669903e-05, "loss": 0.2129, "step": 17770 }, { "epoch": 172.32289156626507, "grad_norm": 18.313337326049805, "learning_rate": 1.3733009708737864e-05, "loss": 0.4108, "step": 17771 }, { "epoch": 172.33253012048192, "grad_norm": 13.754572868347168, "learning_rate": 1.3728155339805826e-05, "loss": 0.1812, "step": 17772 }, { "epoch": 172.3421686746988, "grad_norm": 2.968362808227539, "learning_rate": 1.3723300970873787e-05, "loss": 0.1861, "step": 17773 }, { "epoch": 172.35180722891567, "grad_norm": 6.4787116050720215, "learning_rate": 1.3718446601941748e-05, "loss": 0.0804, "step": 17774 }, { "epoch": 172.36144578313252, "grad_norm": 24.045900344848633, "learning_rate": 1.3713592233009709e-05, "loss": 0.2982, "step": 17775 }, { "epoch": 172.3710843373494, "grad_norm": 7.182285308837891, "learning_rate": 1.370873786407767e-05, "loss": 0.1641, "step": 17776 }, { "epoch": 172.38072289156628, "grad_norm": 6.461848735809326, "learning_rate": 1.3703883495145631e-05, "loss": 0.2349, "step": 17777 }, { "epoch": 172.39036144578313, "grad_norm": 23.100271224975586, "learning_rate": 1.3699029126213593e-05, "loss": 0.4127, "step": 17778 }, { "epoch": 172.4, "grad_norm": 20.22389030456543, "learning_rate": 1.3694174757281555e-05, "loss": 0.206, "step": 17779 }, { "epoch": 172.40963855421685, "grad_norm": 18.15546989440918, "learning_rate": 1.3689320388349517e-05, "loss": 0.3607, "step": 17780 }, { "epoch": 172.41927710843373, "grad_norm": 42.93854522705078, "learning_rate": 1.3684466019417478e-05, "loss": 0.2947, "step": 17781 }, { "epoch": 172.4289156626506, "grad_norm": 6.3181023597717285, "learning_rate": 1.3679611650485439e-05, "loss": 0.1064, "step": 17782 }, { "epoch": 172.43855421686746, "grad_norm": 8.3383150100708, "learning_rate": 1.36747572815534e-05, "loss": 0.2808, "step": 17783 }, { "epoch": 172.44819277108434, "grad_norm": 8.941906929016113, "learning_rate": 1.366990291262136e-05, "loss": 0.1109, "step": 17784 }, { "epoch": 172.4578313253012, "grad_norm": 9.084086418151855, "learning_rate": 1.366504854368932e-05, "loss": 0.1765, "step": 17785 }, { "epoch": 172.46746987951806, "grad_norm": 8.414311408996582, "learning_rate": 1.3660194174757282e-05, "loss": 0.3106, "step": 17786 }, { "epoch": 172.47710843373494, "grad_norm": 14.50710678100586, "learning_rate": 1.3655339805825243e-05, "loss": 0.3317, "step": 17787 }, { "epoch": 172.48674698795182, "grad_norm": 27.96516227722168, "learning_rate": 1.3650485436893204e-05, "loss": 0.3345, "step": 17788 }, { "epoch": 172.49638554216867, "grad_norm": 22.429147720336914, "learning_rate": 1.3645631067961165e-05, "loss": 0.2479, "step": 17789 }, { "epoch": 172.50602409638554, "grad_norm": 27.27630043029785, "learning_rate": 1.3640776699029126e-05, "loss": 0.2264, "step": 17790 }, { "epoch": 172.51566265060242, "grad_norm": 11.604365348815918, "learning_rate": 1.3635922330097087e-05, "loss": 0.2379, "step": 17791 }, { "epoch": 172.52530120481927, "grad_norm": 18.099271774291992, "learning_rate": 1.363106796116505e-05, "loss": 0.3107, "step": 17792 }, { "epoch": 172.53493975903615, "grad_norm": 8.751197814941406, "learning_rate": 1.3626213592233011e-05, "loss": 0.3134, "step": 17793 }, { "epoch": 172.544578313253, "grad_norm": 7.765126705169678, "learning_rate": 1.3621359223300973e-05, "loss": 0.4463, "step": 17794 }, { "epoch": 172.55421686746988, "grad_norm": 5.2714643478393555, "learning_rate": 1.3616504854368934e-05, "loss": 0.3101, "step": 17795 }, { "epoch": 172.56385542168675, "grad_norm": 30.552017211914062, "learning_rate": 1.3611650485436895e-05, "loss": 0.3074, "step": 17796 }, { "epoch": 172.5734939759036, "grad_norm": 10.158405303955078, "learning_rate": 1.3606796116504856e-05, "loss": 0.4337, "step": 17797 }, { "epoch": 172.58313253012048, "grad_norm": 7.499297142028809, "learning_rate": 1.3601941747572815e-05, "loss": 0.4489, "step": 17798 }, { "epoch": 172.59277108433736, "grad_norm": 6.089661598205566, "learning_rate": 1.3597087378640777e-05, "loss": 0.2476, "step": 17799 }, { "epoch": 172.6024096385542, "grad_norm": 18.192726135253906, "learning_rate": 1.3592233009708738e-05, "loss": 0.4409, "step": 17800 }, { "epoch": 172.61204819277108, "grad_norm": 11.341794967651367, "learning_rate": 1.3587378640776699e-05, "loss": 0.1768, "step": 17801 }, { "epoch": 172.62168674698796, "grad_norm": 9.14179801940918, "learning_rate": 1.358252427184466e-05, "loss": 0.1746, "step": 17802 }, { "epoch": 172.6313253012048, "grad_norm": 7.530649662017822, "learning_rate": 1.3577669902912621e-05, "loss": 0.1663, "step": 17803 }, { "epoch": 172.6409638554217, "grad_norm": 5.637279987335205, "learning_rate": 1.3572815533980582e-05, "loss": 0.1802, "step": 17804 }, { "epoch": 172.65060240963857, "grad_norm": 7.406775951385498, "learning_rate": 1.3567961165048545e-05, "loss": 0.388, "step": 17805 }, { "epoch": 172.66024096385541, "grad_norm": 5.235142707824707, "learning_rate": 1.3563106796116506e-05, "loss": 0.1722, "step": 17806 }, { "epoch": 172.6698795180723, "grad_norm": 21.936298370361328, "learning_rate": 1.3558252427184467e-05, "loss": 0.2187, "step": 17807 }, { "epoch": 172.67951807228917, "grad_norm": 12.087688446044922, "learning_rate": 1.3553398058252429e-05, "loss": 0.2847, "step": 17808 }, { "epoch": 172.68915662650602, "grad_norm": 18.05548095703125, "learning_rate": 1.354854368932039e-05, "loss": 0.2379, "step": 17809 }, { "epoch": 172.6987951807229, "grad_norm": 9.34593677520752, "learning_rate": 1.3543689320388351e-05, "loss": 0.2635, "step": 17810 }, { "epoch": 172.70843373493975, "grad_norm": 7.276193141937256, "learning_rate": 1.3538834951456312e-05, "loss": 0.3058, "step": 17811 }, { "epoch": 172.71807228915662, "grad_norm": 9.447429656982422, "learning_rate": 1.3533980582524271e-05, "loss": 0.2356, "step": 17812 }, { "epoch": 172.7277108433735, "grad_norm": 11.921619415283203, "learning_rate": 1.3529126213592233e-05, "loss": 0.2819, "step": 17813 }, { "epoch": 172.73734939759035, "grad_norm": 3.4904837608337402, "learning_rate": 1.3524271844660194e-05, "loss": 0.26, "step": 17814 }, { "epoch": 172.74698795180723, "grad_norm": 22.385915756225586, "learning_rate": 1.3519417475728155e-05, "loss": 0.1936, "step": 17815 }, { "epoch": 172.7566265060241, "grad_norm": 17.096588134765625, "learning_rate": 1.3514563106796116e-05, "loss": 0.2732, "step": 17816 }, { "epoch": 172.76626506024095, "grad_norm": 26.5130672454834, "learning_rate": 1.3509708737864079e-05, "loss": 0.1374, "step": 17817 }, { "epoch": 172.77590361445783, "grad_norm": 6.833312511444092, "learning_rate": 1.350485436893204e-05, "loss": 0.2619, "step": 17818 }, { "epoch": 172.7855421686747, "grad_norm": 3.3525102138519287, "learning_rate": 1.3500000000000001e-05, "loss": 0.145, "step": 17819 }, { "epoch": 172.79518072289156, "grad_norm": 25.684432983398438, "learning_rate": 1.3495145631067962e-05, "loss": 0.4127, "step": 17820 }, { "epoch": 172.80481927710844, "grad_norm": 7.02208948135376, "learning_rate": 1.3490291262135923e-05, "loss": 0.2337, "step": 17821 }, { "epoch": 172.8144578313253, "grad_norm": 8.890639305114746, "learning_rate": 1.3485436893203885e-05, "loss": 0.3691, "step": 17822 }, { "epoch": 172.82409638554216, "grad_norm": 15.240386962890625, "learning_rate": 1.3480582524271846e-05, "loss": 0.1366, "step": 17823 }, { "epoch": 172.83373493975904, "grad_norm": 11.439313888549805, "learning_rate": 1.3475728155339807e-05, "loss": 0.5014, "step": 17824 }, { "epoch": 172.84337349397592, "grad_norm": 35.550655364990234, "learning_rate": 1.3470873786407766e-05, "loss": 0.1952, "step": 17825 }, { "epoch": 172.85301204819277, "grad_norm": 20.54656982421875, "learning_rate": 1.3466019417475728e-05, "loss": 0.2059, "step": 17826 }, { "epoch": 172.86265060240964, "grad_norm": 5.445298671722412, "learning_rate": 1.3461165048543689e-05, "loss": 0.2099, "step": 17827 }, { "epoch": 172.8722891566265, "grad_norm": 5.866983413696289, "learning_rate": 1.345631067961165e-05, "loss": 0.2092, "step": 17828 }, { "epoch": 172.88192771084337, "grad_norm": 14.656501770019531, "learning_rate": 1.3451456310679611e-05, "loss": 0.4011, "step": 17829 }, { "epoch": 172.89156626506025, "grad_norm": 13.970696449279785, "learning_rate": 1.3446601941747575e-05, "loss": 0.1391, "step": 17830 }, { "epoch": 172.9012048192771, "grad_norm": 12.380468368530273, "learning_rate": 1.3441747572815535e-05, "loss": 0.5381, "step": 17831 }, { "epoch": 172.91084337349398, "grad_norm": 66.34992980957031, "learning_rate": 1.3436893203883496e-05, "loss": 0.4147, "step": 17832 }, { "epoch": 172.92048192771085, "grad_norm": 9.238911628723145, "learning_rate": 1.3432038834951457e-05, "loss": 0.3005, "step": 17833 }, { "epoch": 172.9301204819277, "grad_norm": 34.16688537597656, "learning_rate": 1.3427184466019418e-05, "loss": 0.3376, "step": 17834 }, { "epoch": 172.93975903614458, "grad_norm": 27.072301864624023, "learning_rate": 1.342233009708738e-05, "loss": 0.3415, "step": 17835 }, { "epoch": 172.94939759036146, "grad_norm": 11.338053703308105, "learning_rate": 1.341747572815534e-05, "loss": 0.1917, "step": 17836 }, { "epoch": 172.9590361445783, "grad_norm": 7.934049129486084, "learning_rate": 1.3412621359223302e-05, "loss": 0.3129, "step": 17837 }, { "epoch": 172.96867469879518, "grad_norm": 11.659889221191406, "learning_rate": 1.3407766990291263e-05, "loss": 0.2022, "step": 17838 }, { "epoch": 172.97831325301206, "grad_norm": 15.856400489807129, "learning_rate": 1.3402912621359222e-05, "loss": 0.1689, "step": 17839 }, { "epoch": 172.9879518072289, "grad_norm": 5.601223468780518, "learning_rate": 1.3398058252427184e-05, "loss": 0.1852, "step": 17840 }, { "epoch": 173.00361445783133, "grad_norm": 14.193659782409668, "learning_rate": 1.3393203883495145e-05, "loss": 0.3071, "step": 17841 }, { "epoch": 173.01325301204818, "grad_norm": 9.27668571472168, "learning_rate": 1.338834951456311e-05, "loss": 0.2472, "step": 17842 }, { "epoch": 173.02289156626506, "grad_norm": 10.902156829833984, "learning_rate": 1.338349514563107e-05, "loss": 0.4156, "step": 17843 }, { "epoch": 173.03253012048194, "grad_norm": 33.49900817871094, "learning_rate": 1.3378640776699032e-05, "loss": 0.2394, "step": 17844 }, { "epoch": 173.0421686746988, "grad_norm": 19.500938415527344, "learning_rate": 1.3373786407766991e-05, "loss": 0.4103, "step": 17845 }, { "epoch": 173.05180722891566, "grad_norm": 14.408550262451172, "learning_rate": 1.3368932038834952e-05, "loss": 0.2404, "step": 17846 }, { "epoch": 173.06144578313254, "grad_norm": 6.196181297302246, "learning_rate": 1.3364077669902913e-05, "loss": 0.1295, "step": 17847 }, { "epoch": 173.0710843373494, "grad_norm": 22.123794555664062, "learning_rate": 1.3359223300970874e-05, "loss": 0.2776, "step": 17848 }, { "epoch": 173.08072289156627, "grad_norm": 13.682899475097656, "learning_rate": 1.3354368932038836e-05, "loss": 0.1899, "step": 17849 }, { "epoch": 173.09036144578315, "grad_norm": 19.20608139038086, "learning_rate": 1.3349514563106797e-05, "loss": 0.1841, "step": 17850 }, { "epoch": 173.1, "grad_norm": 7.585305213928223, "learning_rate": 1.3344660194174758e-05, "loss": 0.3247, "step": 17851 }, { "epoch": 173.10963855421687, "grad_norm": 7.0699028968811035, "learning_rate": 1.3339805825242719e-05, "loss": 0.2001, "step": 17852 }, { "epoch": 173.11927710843372, "grad_norm": 9.70513916015625, "learning_rate": 1.3334951456310678e-05, "loss": 0.2521, "step": 17853 }, { "epoch": 173.1289156626506, "grad_norm": 15.051156997680664, "learning_rate": 1.333009708737864e-05, "loss": 0.1452, "step": 17854 }, { "epoch": 173.13855421686748, "grad_norm": 24.58254051208496, "learning_rate": 1.3325242718446604e-05, "loss": 0.2917, "step": 17855 }, { "epoch": 173.14819277108433, "grad_norm": 457.9873962402344, "learning_rate": 1.3320388349514565e-05, "loss": 0.21, "step": 17856 }, { "epoch": 173.1578313253012, "grad_norm": 2.726487636566162, "learning_rate": 1.3315533980582526e-05, "loss": 0.0745, "step": 17857 }, { "epoch": 173.16746987951808, "grad_norm": 24.762563705444336, "learning_rate": 1.3310679611650486e-05, "loss": 0.4655, "step": 17858 }, { "epoch": 173.17710843373493, "grad_norm": 5.00171422958374, "learning_rate": 1.3305825242718447e-05, "loss": 0.2141, "step": 17859 }, { "epoch": 173.1867469879518, "grad_norm": 5.776614189147949, "learning_rate": 1.3300970873786408e-05, "loss": 0.1992, "step": 17860 }, { "epoch": 173.19638554216868, "grad_norm": 4.769732475280762, "learning_rate": 1.329611650485437e-05, "loss": 0.1222, "step": 17861 }, { "epoch": 173.20602409638553, "grad_norm": 4.251065731048584, "learning_rate": 1.329126213592233e-05, "loss": 0.1348, "step": 17862 }, { "epoch": 173.2156626506024, "grad_norm": 8.275531768798828, "learning_rate": 1.3286407766990292e-05, "loss": 0.0812, "step": 17863 }, { "epoch": 173.2253012048193, "grad_norm": 15.17910385131836, "learning_rate": 1.3281553398058253e-05, "loss": 0.1575, "step": 17864 }, { "epoch": 173.23493975903614, "grad_norm": 3.9887354373931885, "learning_rate": 1.3276699029126214e-05, "loss": 0.2688, "step": 17865 }, { "epoch": 173.24457831325302, "grad_norm": 20.76451873779297, "learning_rate": 1.3271844660194175e-05, "loss": 0.4073, "step": 17866 }, { "epoch": 173.2542168674699, "grad_norm": 7.724508285522461, "learning_rate": 1.3266990291262134e-05, "loss": 0.1059, "step": 17867 }, { "epoch": 173.26385542168674, "grad_norm": 37.46684646606445, "learning_rate": 1.3262135922330099e-05, "loss": 0.1641, "step": 17868 }, { "epoch": 173.27349397590362, "grad_norm": 6.058568000793457, "learning_rate": 1.325728155339806e-05, "loss": 0.1994, "step": 17869 }, { "epoch": 173.28313253012047, "grad_norm": 5.127651214599609, "learning_rate": 1.3252427184466021e-05, "loss": 0.2109, "step": 17870 }, { "epoch": 173.29277108433735, "grad_norm": 3.453496217727661, "learning_rate": 1.3247572815533982e-05, "loss": 0.1014, "step": 17871 }, { "epoch": 173.30240963855422, "grad_norm": 8.022183418273926, "learning_rate": 1.3242718446601942e-05, "loss": 0.1936, "step": 17872 }, { "epoch": 173.31204819277107, "grad_norm": 27.659151077270508, "learning_rate": 1.3237864077669903e-05, "loss": 0.2419, "step": 17873 }, { "epoch": 173.32168674698795, "grad_norm": 4.9440999031066895, "learning_rate": 1.3233009708737864e-05, "loss": 0.1521, "step": 17874 }, { "epoch": 173.33132530120483, "grad_norm": 9.622671127319336, "learning_rate": 1.3228155339805825e-05, "loss": 0.3613, "step": 17875 }, { "epoch": 173.34096385542168, "grad_norm": 10.684356689453125, "learning_rate": 1.3223300970873786e-05, "loss": 0.4701, "step": 17876 }, { "epoch": 173.35060240963855, "grad_norm": 14.493239402770996, "learning_rate": 1.3218446601941748e-05, "loss": 0.291, "step": 17877 }, { "epoch": 173.36024096385543, "grad_norm": 16.50040626525879, "learning_rate": 1.3213592233009709e-05, "loss": 0.3631, "step": 17878 }, { "epoch": 173.36987951807228, "grad_norm": 4.8145012855529785, "learning_rate": 1.320873786407767e-05, "loss": 0.1944, "step": 17879 }, { "epoch": 173.37951807228916, "grad_norm": 14.772974014282227, "learning_rate": 1.3203883495145633e-05, "loss": 0.3948, "step": 17880 }, { "epoch": 173.38915662650604, "grad_norm": 13.141339302062988, "learning_rate": 1.3199029126213594e-05, "loss": 0.2727, "step": 17881 }, { "epoch": 173.39879518072289, "grad_norm": 6.306957721710205, "learning_rate": 1.3194174757281555e-05, "loss": 0.0831, "step": 17882 }, { "epoch": 173.40843373493976, "grad_norm": 5.9698686599731445, "learning_rate": 1.3189320388349516e-05, "loss": 0.2822, "step": 17883 }, { "epoch": 173.41807228915664, "grad_norm": 28.404111862182617, "learning_rate": 1.3184466019417477e-05, "loss": 0.2167, "step": 17884 }, { "epoch": 173.4277108433735, "grad_norm": 50.189090728759766, "learning_rate": 1.3179611650485439e-05, "loss": 0.4417, "step": 17885 }, { "epoch": 173.43734939759037, "grad_norm": 6.909938335418701, "learning_rate": 1.3174757281553398e-05, "loss": 0.2056, "step": 17886 }, { "epoch": 173.44698795180722, "grad_norm": 8.90160846710205, "learning_rate": 1.3169902912621359e-05, "loss": 0.3364, "step": 17887 }, { "epoch": 173.4566265060241, "grad_norm": 4.562882900238037, "learning_rate": 1.316504854368932e-05, "loss": 0.0947, "step": 17888 }, { "epoch": 173.46626506024097, "grad_norm": 14.989001274108887, "learning_rate": 1.3160194174757281e-05, "loss": 0.4217, "step": 17889 }, { "epoch": 173.47590361445782, "grad_norm": 22.900531768798828, "learning_rate": 1.3155339805825243e-05, "loss": 0.1967, "step": 17890 }, { "epoch": 173.4855421686747, "grad_norm": 5.376640796661377, "learning_rate": 1.3150485436893204e-05, "loss": 0.2599, "step": 17891 }, { "epoch": 173.49518072289158, "grad_norm": 13.953044891357422, "learning_rate": 1.3145631067961165e-05, "loss": 0.3214, "step": 17892 }, { "epoch": 173.50481927710842, "grad_norm": 7.025280475616455, "learning_rate": 1.3140776699029128e-05, "loss": 0.215, "step": 17893 }, { "epoch": 173.5144578313253, "grad_norm": 18.822690963745117, "learning_rate": 1.3135922330097089e-05, "loss": 0.2712, "step": 17894 }, { "epoch": 173.52409638554218, "grad_norm": 11.168899536132812, "learning_rate": 1.313106796116505e-05, "loss": 0.357, "step": 17895 }, { "epoch": 173.53373493975903, "grad_norm": 10.768271446228027, "learning_rate": 1.3126213592233011e-05, "loss": 0.4016, "step": 17896 }, { "epoch": 173.5433734939759, "grad_norm": 5.150656700134277, "learning_rate": 1.3121359223300972e-05, "loss": 0.2276, "step": 17897 }, { "epoch": 173.55301204819278, "grad_norm": 35.83484649658203, "learning_rate": 1.3116504854368933e-05, "loss": 0.3443, "step": 17898 }, { "epoch": 173.56265060240963, "grad_norm": 4.2617011070251465, "learning_rate": 1.3111650485436895e-05, "loss": 0.1995, "step": 17899 }, { "epoch": 173.5722891566265, "grad_norm": 45.47285461425781, "learning_rate": 1.3106796116504854e-05, "loss": 0.2186, "step": 17900 }, { "epoch": 173.58192771084336, "grad_norm": 10.426735877990723, "learning_rate": 1.3101941747572815e-05, "loss": 0.4184, "step": 17901 }, { "epoch": 173.59156626506024, "grad_norm": 97.0506591796875, "learning_rate": 1.3097087378640776e-05, "loss": 0.1915, "step": 17902 }, { "epoch": 173.60120481927711, "grad_norm": 7.740695476531982, "learning_rate": 1.3092233009708737e-05, "loss": 0.2104, "step": 17903 }, { "epoch": 173.61084337349396, "grad_norm": 9.490601539611816, "learning_rate": 1.3087378640776699e-05, "loss": 0.2528, "step": 17904 }, { "epoch": 173.62048192771084, "grad_norm": 27.062273025512695, "learning_rate": 1.308252427184466e-05, "loss": 0.2527, "step": 17905 }, { "epoch": 173.63012048192772, "grad_norm": 7.947254657745361, "learning_rate": 1.3077669902912623e-05, "loss": 0.3496, "step": 17906 }, { "epoch": 173.63975903614457, "grad_norm": 4.789379119873047, "learning_rate": 1.3072815533980584e-05, "loss": 0.1421, "step": 17907 }, { "epoch": 173.64939759036145, "grad_norm": 10.998276710510254, "learning_rate": 1.3067961165048545e-05, "loss": 0.2462, "step": 17908 }, { "epoch": 173.65903614457832, "grad_norm": 8.827983856201172, "learning_rate": 1.3063106796116506e-05, "loss": 0.3355, "step": 17909 }, { "epoch": 173.66867469879517, "grad_norm": 21.769145965576172, "learning_rate": 1.3058252427184467e-05, "loss": 0.2183, "step": 17910 }, { "epoch": 173.67831325301205, "grad_norm": 7.408142566680908, "learning_rate": 1.3053398058252428e-05, "loss": 0.0936, "step": 17911 }, { "epoch": 173.68795180722893, "grad_norm": 21.094709396362305, "learning_rate": 1.304854368932039e-05, "loss": 0.1744, "step": 17912 }, { "epoch": 173.69759036144578, "grad_norm": 1.7588940858840942, "learning_rate": 1.304368932038835e-05, "loss": 0.2135, "step": 17913 }, { "epoch": 173.70722891566265, "grad_norm": 37.39645767211914, "learning_rate": 1.303883495145631e-05, "loss": 0.1968, "step": 17914 }, { "epoch": 173.71686746987953, "grad_norm": 11.313486099243164, "learning_rate": 1.3033980582524271e-05, "loss": 0.2972, "step": 17915 }, { "epoch": 173.72650602409638, "grad_norm": 7.894609451293945, "learning_rate": 1.3029126213592232e-05, "loss": 0.3229, "step": 17916 }, { "epoch": 173.73614457831326, "grad_norm": 13.731192588806152, "learning_rate": 1.3024271844660193e-05, "loss": 0.086, "step": 17917 }, { "epoch": 173.7457831325301, "grad_norm": 3.8804636001586914, "learning_rate": 1.3019417475728158e-05, "loss": 0.1506, "step": 17918 }, { "epoch": 173.75542168674698, "grad_norm": 28.678882598876953, "learning_rate": 1.3014563106796117e-05, "loss": 0.2065, "step": 17919 }, { "epoch": 173.76506024096386, "grad_norm": 11.991791725158691, "learning_rate": 1.3009708737864079e-05, "loss": 0.1595, "step": 17920 }, { "epoch": 173.7746987951807, "grad_norm": 9.339641571044922, "learning_rate": 1.300485436893204e-05, "loss": 0.2254, "step": 17921 }, { "epoch": 173.7843373493976, "grad_norm": 4.631350040435791, "learning_rate": 1.3000000000000001e-05, "loss": 0.1422, "step": 17922 }, { "epoch": 173.79397590361447, "grad_norm": 15.370230674743652, "learning_rate": 1.2995145631067962e-05, "loss": 0.3885, "step": 17923 }, { "epoch": 173.80361445783132, "grad_norm": 5.560338020324707, "learning_rate": 1.2990291262135923e-05, "loss": 0.1698, "step": 17924 }, { "epoch": 173.8132530120482, "grad_norm": 24.97849464416504, "learning_rate": 1.2985436893203884e-05, "loss": 0.2648, "step": 17925 }, { "epoch": 173.82289156626507, "grad_norm": 7.211634159088135, "learning_rate": 1.2980582524271845e-05, "loss": 0.3805, "step": 17926 }, { "epoch": 173.83253012048192, "grad_norm": 4.795867919921875, "learning_rate": 1.2975728155339807e-05, "loss": 0.1142, "step": 17927 }, { "epoch": 173.8421686746988, "grad_norm": 11.154159545898438, "learning_rate": 1.2970873786407766e-05, "loss": 0.274, "step": 17928 }, { "epoch": 173.85180722891567, "grad_norm": 24.383296966552734, "learning_rate": 1.2966019417475727e-05, "loss": 0.3059, "step": 17929 }, { "epoch": 173.86144578313252, "grad_norm": 5.179154872894287, "learning_rate": 1.2961165048543688e-05, "loss": 0.1813, "step": 17930 }, { "epoch": 173.8710843373494, "grad_norm": 9.070854187011719, "learning_rate": 1.2956310679611653e-05, "loss": 0.1588, "step": 17931 }, { "epoch": 173.88072289156628, "grad_norm": 5.432500839233398, "learning_rate": 1.2951456310679614e-05, "loss": 0.3916, "step": 17932 }, { "epoch": 173.89036144578313, "grad_norm": 5.272070407867432, "learning_rate": 1.2946601941747573e-05, "loss": 0.145, "step": 17933 }, { "epoch": 173.9, "grad_norm": 10.89224624633789, "learning_rate": 1.2941747572815535e-05, "loss": 0.194, "step": 17934 }, { "epoch": 173.90963855421685, "grad_norm": 17.686025619506836, "learning_rate": 1.2936893203883496e-05, "loss": 0.3839, "step": 17935 }, { "epoch": 173.91927710843373, "grad_norm": 11.80301570892334, "learning_rate": 1.2932038834951457e-05, "loss": 0.3537, "step": 17936 }, { "epoch": 173.9289156626506, "grad_norm": 27.170305252075195, "learning_rate": 1.2927184466019418e-05, "loss": 0.2452, "step": 17937 }, { "epoch": 173.93855421686746, "grad_norm": 5.566369533538818, "learning_rate": 1.292233009708738e-05, "loss": 0.1498, "step": 17938 }, { "epoch": 173.94819277108434, "grad_norm": 15.77231502532959, "learning_rate": 1.291747572815534e-05, "loss": 0.2924, "step": 17939 }, { "epoch": 173.9578313253012, "grad_norm": 13.155149459838867, "learning_rate": 1.2912621359223302e-05, "loss": 0.2563, "step": 17940 }, { "epoch": 173.96746987951806, "grad_norm": 82.32823181152344, "learning_rate": 1.2907766990291261e-05, "loss": 0.3974, "step": 17941 }, { "epoch": 173.97710843373494, "grad_norm": 5.580760955810547, "learning_rate": 1.2902912621359222e-05, "loss": 0.1778, "step": 17942 }, { "epoch": 173.98674698795182, "grad_norm": 12.018584251403809, "learning_rate": 1.2898058252427187e-05, "loss": 0.2644, "step": 17943 }, { "epoch": 174.0024096385542, "grad_norm": 15.728822708129883, "learning_rate": 1.2893203883495148e-05, "loss": 0.2241, "step": 17944 }, { "epoch": 174.0120481927711, "grad_norm": 7.881922245025635, "learning_rate": 1.2888349514563109e-05, "loss": 0.2292, "step": 17945 }, { "epoch": 174.02168674698794, "grad_norm": 27.76477813720703, "learning_rate": 1.288349514563107e-05, "loss": 0.2541, "step": 17946 }, { "epoch": 174.03132530120482, "grad_norm": 5.724689483642578, "learning_rate": 1.287864077669903e-05, "loss": 0.2565, "step": 17947 }, { "epoch": 174.0409638554217, "grad_norm": 20.564800262451172, "learning_rate": 1.287378640776699e-05, "loss": 0.2306, "step": 17948 }, { "epoch": 174.05060240963854, "grad_norm": 46.16583251953125, "learning_rate": 1.2868932038834952e-05, "loss": 0.243, "step": 17949 }, { "epoch": 174.06024096385542, "grad_norm": 12.641566276550293, "learning_rate": 1.2864077669902913e-05, "loss": 0.1772, "step": 17950 }, { "epoch": 174.0698795180723, "grad_norm": 4.9377641677856445, "learning_rate": 1.2859223300970874e-05, "loss": 0.2266, "step": 17951 }, { "epoch": 174.07951807228915, "grad_norm": 24.4403076171875, "learning_rate": 1.2854368932038835e-05, "loss": 0.3763, "step": 17952 }, { "epoch": 174.08915662650602, "grad_norm": 9.643389701843262, "learning_rate": 1.2849514563106796e-05, "loss": 0.1914, "step": 17953 }, { "epoch": 174.0987951807229, "grad_norm": 22.219669342041016, "learning_rate": 1.2844660194174758e-05, "loss": 0.1692, "step": 17954 }, { "epoch": 174.10843373493975, "grad_norm": 6.495830535888672, "learning_rate": 1.2839805825242717e-05, "loss": 0.2875, "step": 17955 }, { "epoch": 174.11807228915663, "grad_norm": 29.274375915527344, "learning_rate": 1.2834951456310682e-05, "loss": 0.2357, "step": 17956 }, { "epoch": 174.1277108433735, "grad_norm": 14.839136123657227, "learning_rate": 1.2830097087378643e-05, "loss": 0.1974, "step": 17957 }, { "epoch": 174.13734939759036, "grad_norm": 5.250455856323242, "learning_rate": 1.2825242718446604e-05, "loss": 0.1791, "step": 17958 }, { "epoch": 174.14698795180723, "grad_norm": 23.1143856048584, "learning_rate": 1.2820388349514565e-05, "loss": 0.3547, "step": 17959 }, { "epoch": 174.15662650602408, "grad_norm": 20.52884292602539, "learning_rate": 1.2815533980582526e-05, "loss": 0.4355, "step": 17960 }, { "epoch": 174.16626506024096, "grad_norm": 14.389717102050781, "learning_rate": 1.2810679611650486e-05, "loss": 0.1985, "step": 17961 }, { "epoch": 174.17590361445784, "grad_norm": 13.19421672821045, "learning_rate": 1.2805825242718447e-05, "loss": 0.4796, "step": 17962 }, { "epoch": 174.1855421686747, "grad_norm": 11.719839096069336, "learning_rate": 1.2800970873786408e-05, "loss": 0.2246, "step": 17963 }, { "epoch": 174.19518072289156, "grad_norm": 8.210062980651855, "learning_rate": 1.2796116504854369e-05, "loss": 0.4658, "step": 17964 }, { "epoch": 174.20481927710844, "grad_norm": 8.963656425476074, "learning_rate": 1.279126213592233e-05, "loss": 0.2222, "step": 17965 }, { "epoch": 174.2144578313253, "grad_norm": 13.091683387756348, "learning_rate": 1.2786407766990291e-05, "loss": 0.3094, "step": 17966 }, { "epoch": 174.22409638554217, "grad_norm": 7.1094160079956055, "learning_rate": 1.2781553398058252e-05, "loss": 0.1333, "step": 17967 }, { "epoch": 174.23373493975905, "grad_norm": 43.91205596923828, "learning_rate": 1.2776699029126214e-05, "loss": 0.237, "step": 17968 }, { "epoch": 174.2433734939759, "grad_norm": 14.115891456604004, "learning_rate": 1.2771844660194176e-05, "loss": 0.4494, "step": 17969 }, { "epoch": 174.25301204819277, "grad_norm": 7.023593425750732, "learning_rate": 1.2766990291262138e-05, "loss": 0.1062, "step": 17970 }, { "epoch": 174.26265060240965, "grad_norm": 10.58759593963623, "learning_rate": 1.2762135922330099e-05, "loss": 0.1732, "step": 17971 }, { "epoch": 174.2722891566265, "grad_norm": 24.493106842041016, "learning_rate": 1.275728155339806e-05, "loss": 0.3275, "step": 17972 }, { "epoch": 174.28192771084338, "grad_norm": 10.454484939575195, "learning_rate": 1.2752427184466021e-05, "loss": 0.2839, "step": 17973 }, { "epoch": 174.29156626506025, "grad_norm": 4.1379899978637695, "learning_rate": 1.274757281553398e-05, "loss": 0.1461, "step": 17974 }, { "epoch": 174.3012048192771, "grad_norm": 12.310968399047852, "learning_rate": 1.2742718446601942e-05, "loss": 0.1803, "step": 17975 }, { "epoch": 174.31084337349398, "grad_norm": 8.762412071228027, "learning_rate": 1.2737864077669903e-05, "loss": 0.1782, "step": 17976 }, { "epoch": 174.32048192771083, "grad_norm": 22.427745819091797, "learning_rate": 1.2733009708737864e-05, "loss": 0.1888, "step": 17977 }, { "epoch": 174.3301204819277, "grad_norm": 41.31830978393555, "learning_rate": 1.2728155339805825e-05, "loss": 0.2277, "step": 17978 }, { "epoch": 174.33975903614459, "grad_norm": 9.781731605529785, "learning_rate": 1.2723300970873786e-05, "loss": 0.2199, "step": 17979 }, { "epoch": 174.34939759036143, "grad_norm": 7.722806453704834, "learning_rate": 1.2718446601941747e-05, "loss": 0.1609, "step": 17980 }, { "epoch": 174.3590361445783, "grad_norm": 8.577685356140137, "learning_rate": 1.271359223300971e-05, "loss": 0.2055, "step": 17981 }, { "epoch": 174.3686746987952, "grad_norm": 17.591657638549805, "learning_rate": 1.2708737864077671e-05, "loss": 0.3957, "step": 17982 }, { "epoch": 174.37831325301204, "grad_norm": 9.546592712402344, "learning_rate": 1.2703883495145632e-05, "loss": 0.2173, "step": 17983 }, { "epoch": 174.38795180722892, "grad_norm": 8.101862907409668, "learning_rate": 1.2699029126213594e-05, "loss": 0.3277, "step": 17984 }, { "epoch": 174.3975903614458, "grad_norm": 24.503246307373047, "learning_rate": 1.2694174757281555e-05, "loss": 0.1965, "step": 17985 }, { "epoch": 174.40722891566264, "grad_norm": 8.44920539855957, "learning_rate": 1.2689320388349516e-05, "loss": 0.2739, "step": 17986 }, { "epoch": 174.41686746987952, "grad_norm": 98.64318084716797, "learning_rate": 1.2684466019417477e-05, "loss": 0.2185, "step": 17987 }, { "epoch": 174.4265060240964, "grad_norm": 12.825115203857422, "learning_rate": 1.2679611650485437e-05, "loss": 0.2075, "step": 17988 }, { "epoch": 174.43614457831325, "grad_norm": 11.532782554626465, "learning_rate": 1.2674757281553398e-05, "loss": 0.4951, "step": 17989 }, { "epoch": 174.44578313253012, "grad_norm": 11.700994491577148, "learning_rate": 1.2669902912621359e-05, "loss": 0.3538, "step": 17990 }, { "epoch": 174.455421686747, "grad_norm": 8.042603492736816, "learning_rate": 1.266504854368932e-05, "loss": 0.2086, "step": 17991 }, { "epoch": 174.46506024096385, "grad_norm": 14.539338111877441, "learning_rate": 1.2660194174757281e-05, "loss": 0.1539, "step": 17992 }, { "epoch": 174.47469879518073, "grad_norm": 7.163982391357422, "learning_rate": 1.2655339805825242e-05, "loss": 0.3295, "step": 17993 }, { "epoch": 174.48433734939758, "grad_norm": 6.84906530380249, "learning_rate": 1.2650485436893205e-05, "loss": 0.3258, "step": 17994 }, { "epoch": 174.49397590361446, "grad_norm": 5.201640605926514, "learning_rate": 1.2645631067961166e-05, "loss": 0.2628, "step": 17995 }, { "epoch": 174.50361445783133, "grad_norm": 7.849639415740967, "learning_rate": 1.2640776699029127e-05, "loss": 0.2631, "step": 17996 }, { "epoch": 174.51325301204818, "grad_norm": 8.751936912536621, "learning_rate": 1.2635922330097089e-05, "loss": 0.1095, "step": 17997 }, { "epoch": 174.52289156626506, "grad_norm": 17.084802627563477, "learning_rate": 1.263106796116505e-05, "loss": 0.2814, "step": 17998 }, { "epoch": 174.53253012048194, "grad_norm": 27.147216796875, "learning_rate": 1.262621359223301e-05, "loss": 0.2045, "step": 17999 }, { "epoch": 174.5421686746988, "grad_norm": 9.07603931427002, "learning_rate": 1.2621359223300972e-05, "loss": 0.3054, "step": 18000 }, { "epoch": 174.55180722891566, "grad_norm": 10.238840103149414, "learning_rate": 1.2616504854368933e-05, "loss": 0.2075, "step": 18001 }, { "epoch": 174.56144578313254, "grad_norm": 9.218368530273438, "learning_rate": 1.2611650485436893e-05, "loss": 0.1576, "step": 18002 }, { "epoch": 174.5710843373494, "grad_norm": 7.526947498321533, "learning_rate": 1.2606796116504854e-05, "loss": 0.1709, "step": 18003 }, { "epoch": 174.58072289156627, "grad_norm": 11.210051536560059, "learning_rate": 1.2601941747572815e-05, "loss": 0.2076, "step": 18004 }, { "epoch": 174.59036144578315, "grad_norm": 10.61954402923584, "learning_rate": 1.2597087378640776e-05, "loss": 0.2645, "step": 18005 }, { "epoch": 174.6, "grad_norm": 8.59218692779541, "learning_rate": 1.2592233009708737e-05, "loss": 0.2819, "step": 18006 }, { "epoch": 174.60963855421687, "grad_norm": 29.211822509765625, "learning_rate": 1.25873786407767e-05, "loss": 0.3276, "step": 18007 }, { "epoch": 174.61927710843372, "grad_norm": 12.91388988494873, "learning_rate": 1.2582524271844661e-05, "loss": 0.2243, "step": 18008 }, { "epoch": 174.6289156626506, "grad_norm": 7.793254375457764, "learning_rate": 1.2577669902912622e-05, "loss": 0.1346, "step": 18009 }, { "epoch": 174.63855421686748, "grad_norm": 4.142112731933594, "learning_rate": 1.2572815533980583e-05, "loss": 0.238, "step": 18010 }, { "epoch": 174.64819277108433, "grad_norm": 26.283292770385742, "learning_rate": 1.2567961165048545e-05, "loss": 0.1906, "step": 18011 }, { "epoch": 174.6578313253012, "grad_norm": 5.789447784423828, "learning_rate": 1.2563106796116506e-05, "loss": 0.1779, "step": 18012 }, { "epoch": 174.66746987951808, "grad_norm": 8.728928565979004, "learning_rate": 1.2558252427184467e-05, "loss": 0.3643, "step": 18013 }, { "epoch": 174.67710843373493, "grad_norm": 30.756074905395508, "learning_rate": 1.2553398058252428e-05, "loss": 0.329, "step": 18014 }, { "epoch": 174.6867469879518, "grad_norm": 7.5392069816589355, "learning_rate": 1.2548543689320389e-05, "loss": 0.2808, "step": 18015 }, { "epoch": 174.69638554216868, "grad_norm": 20.250043869018555, "learning_rate": 1.2543689320388349e-05, "loss": 0.214, "step": 18016 }, { "epoch": 174.70602409638553, "grad_norm": 8.502861976623535, "learning_rate": 1.253883495145631e-05, "loss": 0.2534, "step": 18017 }, { "epoch": 174.7156626506024, "grad_norm": 52.4690055847168, "learning_rate": 1.2533980582524271e-05, "loss": 0.4297, "step": 18018 }, { "epoch": 174.7253012048193, "grad_norm": 11.09762191772461, "learning_rate": 1.2529126213592235e-05, "loss": 0.3305, "step": 18019 }, { "epoch": 174.73493975903614, "grad_norm": 14.740870475769043, "learning_rate": 1.2524271844660197e-05, "loss": 0.2135, "step": 18020 }, { "epoch": 174.74457831325302, "grad_norm": 6.504122257232666, "learning_rate": 1.2519417475728156e-05, "loss": 0.2901, "step": 18021 }, { "epoch": 174.7542168674699, "grad_norm": 6.639754295349121, "learning_rate": 1.2514563106796117e-05, "loss": 0.1572, "step": 18022 }, { "epoch": 174.76385542168674, "grad_norm": 7.878256320953369, "learning_rate": 1.2509708737864078e-05, "loss": 0.1236, "step": 18023 }, { "epoch": 174.77349397590362, "grad_norm": 6.711104869842529, "learning_rate": 1.250485436893204e-05, "loss": 0.1774, "step": 18024 }, { "epoch": 174.78313253012047, "grad_norm": 16.005537033081055, "learning_rate": 1.25e-05, "loss": 0.0889, "step": 18025 }, { "epoch": 174.79277108433735, "grad_norm": 12.772111892700195, "learning_rate": 1.2495145631067962e-05, "loss": 0.3466, "step": 18026 }, { "epoch": 174.80240963855422, "grad_norm": 10.511175155639648, "learning_rate": 1.2490291262135923e-05, "loss": 0.143, "step": 18027 }, { "epoch": 174.81204819277107, "grad_norm": 7.509664535522461, "learning_rate": 1.2485436893203884e-05, "loss": 0.1302, "step": 18028 }, { "epoch": 174.82168674698795, "grad_norm": 6.7737956047058105, "learning_rate": 1.2480582524271845e-05, "loss": 0.2058, "step": 18029 }, { "epoch": 174.83132530120483, "grad_norm": 10.021625518798828, "learning_rate": 1.2475728155339806e-05, "loss": 0.269, "step": 18030 }, { "epoch": 174.84096385542168, "grad_norm": 10.382975578308105, "learning_rate": 1.2470873786407767e-05, "loss": 0.2221, "step": 18031 }, { "epoch": 174.85060240963855, "grad_norm": 19.690574645996094, "learning_rate": 1.2466019417475729e-05, "loss": 0.4136, "step": 18032 }, { "epoch": 174.86024096385543, "grad_norm": 8.776325225830078, "learning_rate": 1.246116504854369e-05, "loss": 0.3513, "step": 18033 }, { "epoch": 174.86987951807228, "grad_norm": 21.64671516418457, "learning_rate": 1.2456310679611651e-05, "loss": 0.2174, "step": 18034 }, { "epoch": 174.87951807228916, "grad_norm": 5.564736843109131, "learning_rate": 1.2451456310679612e-05, "loss": 0.1803, "step": 18035 }, { "epoch": 174.88915662650604, "grad_norm": 6.413139343261719, "learning_rate": 1.2446601941747573e-05, "loss": 0.3052, "step": 18036 }, { "epoch": 174.89879518072289, "grad_norm": 11.751651763916016, "learning_rate": 1.2441747572815534e-05, "loss": 0.2558, "step": 18037 }, { "epoch": 174.90843373493976, "grad_norm": 18.906600952148438, "learning_rate": 1.2436893203883495e-05, "loss": 0.281, "step": 18038 }, { "epoch": 174.91807228915664, "grad_norm": 10.872005462646484, "learning_rate": 1.2432038834951457e-05, "loss": 0.3412, "step": 18039 }, { "epoch": 174.9277108433735, "grad_norm": 13.172380447387695, "learning_rate": 1.2427184466019418e-05, "loss": 0.2275, "step": 18040 }, { "epoch": 174.93734939759037, "grad_norm": 9.570096015930176, "learning_rate": 1.242233009708738e-05, "loss": 0.3664, "step": 18041 }, { "epoch": 174.94698795180722, "grad_norm": 38.10841751098633, "learning_rate": 1.241747572815534e-05, "loss": 0.1386, "step": 18042 }, { "epoch": 174.9566265060241, "grad_norm": 18.000215530395508, "learning_rate": 1.2412621359223301e-05, "loss": 0.255, "step": 18043 }, { "epoch": 174.96626506024097, "grad_norm": 10.696145057678223, "learning_rate": 1.2407766990291262e-05, "loss": 0.2402, "step": 18044 }, { "epoch": 174.97590361445782, "grad_norm": 11.116412162780762, "learning_rate": 1.2402912621359224e-05, "loss": 0.1286, "step": 18045 }, { "epoch": 174.9855421686747, "grad_norm": 53.4156608581543, "learning_rate": 1.2398058252427185e-05, "loss": 0.1707, "step": 18046 }, { "epoch": 175.00120481927712, "grad_norm": 7.4374823570251465, "learning_rate": 1.2393203883495146e-05, "loss": 0.2472, "step": 18047 }, { "epoch": 175.01084337349397, "grad_norm": 19.934844970703125, "learning_rate": 1.2388349514563109e-05, "loss": 0.256, "step": 18048 }, { "epoch": 175.02048192771085, "grad_norm": 12.355429649353027, "learning_rate": 1.2383495145631068e-05, "loss": 0.3648, "step": 18049 }, { "epoch": 175.03012048192772, "grad_norm": 18.297618865966797, "learning_rate": 1.237864077669903e-05, "loss": 0.2455, "step": 18050 }, { "epoch": 175.03975903614457, "grad_norm": 12.580698013305664, "learning_rate": 1.237378640776699e-05, "loss": 0.2709, "step": 18051 }, { "epoch": 175.04939759036145, "grad_norm": 5.532067775726318, "learning_rate": 1.2368932038834952e-05, "loss": 0.1649, "step": 18052 }, { "epoch": 175.0590361445783, "grad_norm": 4.981380939483643, "learning_rate": 1.2364077669902913e-05, "loss": 0.142, "step": 18053 }, { "epoch": 175.06867469879518, "grad_norm": 21.958515167236328, "learning_rate": 1.2359223300970876e-05, "loss": 0.2464, "step": 18054 }, { "epoch": 175.07831325301206, "grad_norm": 11.025948524475098, "learning_rate": 1.2354368932038837e-05, "loss": 0.1738, "step": 18055 }, { "epoch": 175.0879518072289, "grad_norm": 8.090720176696777, "learning_rate": 1.2349514563106796e-05, "loss": 0.3802, "step": 18056 }, { "epoch": 175.09759036144578, "grad_norm": 4.833549976348877, "learning_rate": 1.2344660194174757e-05, "loss": 0.1915, "step": 18057 }, { "epoch": 175.10722891566266, "grad_norm": 8.114574432373047, "learning_rate": 1.2339805825242718e-05, "loss": 0.1444, "step": 18058 }, { "epoch": 175.1168674698795, "grad_norm": 18.27488899230957, "learning_rate": 1.233495145631068e-05, "loss": 0.3167, "step": 18059 }, { "epoch": 175.1265060240964, "grad_norm": 12.272435188293457, "learning_rate": 1.2330097087378642e-05, "loss": 0.3441, "step": 18060 }, { "epoch": 175.13614457831326, "grad_norm": 4.063762187957764, "learning_rate": 1.2325242718446604e-05, "loss": 0.2011, "step": 18061 }, { "epoch": 175.1457831325301, "grad_norm": 27.452669143676758, "learning_rate": 1.2320388349514565e-05, "loss": 0.3587, "step": 18062 }, { "epoch": 175.155421686747, "grad_norm": 16.933944702148438, "learning_rate": 1.2315533980582524e-05, "loss": 0.4398, "step": 18063 }, { "epoch": 175.16506024096387, "grad_norm": 15.753951072692871, "learning_rate": 1.2310679611650485e-05, "loss": 0.2268, "step": 18064 }, { "epoch": 175.17469879518072, "grad_norm": 8.756693840026855, "learning_rate": 1.2305825242718446e-05, "loss": 0.1539, "step": 18065 }, { "epoch": 175.1843373493976, "grad_norm": 10.781805038452148, "learning_rate": 1.2300970873786408e-05, "loss": 0.2007, "step": 18066 }, { "epoch": 175.19397590361444, "grad_norm": 7.080399990081787, "learning_rate": 1.229611650485437e-05, "loss": 0.0977, "step": 18067 }, { "epoch": 175.20361445783132, "grad_norm": 8.771408081054688, "learning_rate": 1.2291262135922332e-05, "loss": 0.2587, "step": 18068 }, { "epoch": 175.2132530120482, "grad_norm": 11.450142860412598, "learning_rate": 1.2286407766990293e-05, "loss": 0.3081, "step": 18069 }, { "epoch": 175.22289156626505, "grad_norm": 3.567471504211426, "learning_rate": 1.2281553398058252e-05, "loss": 0.0614, "step": 18070 }, { "epoch": 175.23253012048193, "grad_norm": 12.569472312927246, "learning_rate": 1.2276699029126213e-05, "loss": 0.3556, "step": 18071 }, { "epoch": 175.2421686746988, "grad_norm": 26.93488883972168, "learning_rate": 1.2271844660194174e-05, "loss": 0.2223, "step": 18072 }, { "epoch": 175.25180722891565, "grad_norm": 25.965768814086914, "learning_rate": 1.2266990291262137e-05, "loss": 0.2003, "step": 18073 }, { "epoch": 175.26144578313253, "grad_norm": 6.427850246429443, "learning_rate": 1.2262135922330098e-05, "loss": 0.1794, "step": 18074 }, { "epoch": 175.2710843373494, "grad_norm": 13.752391815185547, "learning_rate": 1.225728155339806e-05, "loss": 0.2743, "step": 18075 }, { "epoch": 175.28072289156626, "grad_norm": 10.338509559631348, "learning_rate": 1.225242718446602e-05, "loss": 0.2413, "step": 18076 }, { "epoch": 175.29036144578313, "grad_norm": 7.718460559844971, "learning_rate": 1.224757281553398e-05, "loss": 0.3134, "step": 18077 }, { "epoch": 175.3, "grad_norm": 5.317823886871338, "learning_rate": 1.2242718446601941e-05, "loss": 0.1963, "step": 18078 }, { "epoch": 175.30963855421686, "grad_norm": 6.236185073852539, "learning_rate": 1.2237864077669904e-05, "loss": 0.1728, "step": 18079 }, { "epoch": 175.31927710843374, "grad_norm": 6.95915412902832, "learning_rate": 1.2233009708737865e-05, "loss": 0.1613, "step": 18080 }, { "epoch": 175.32891566265062, "grad_norm": 19.74713706970215, "learning_rate": 1.2228155339805826e-05, "loss": 0.2125, "step": 18081 }, { "epoch": 175.33855421686746, "grad_norm": 22.2669734954834, "learning_rate": 1.2223300970873788e-05, "loss": 0.3744, "step": 18082 }, { "epoch": 175.34819277108434, "grad_norm": 4.915258884429932, "learning_rate": 1.2218446601941747e-05, "loss": 0.3077, "step": 18083 }, { "epoch": 175.3578313253012, "grad_norm": 9.64574146270752, "learning_rate": 1.2213592233009708e-05, "loss": 0.3447, "step": 18084 }, { "epoch": 175.36746987951807, "grad_norm": 22.9141788482666, "learning_rate": 1.2208737864077671e-05, "loss": 0.1881, "step": 18085 }, { "epoch": 175.37710843373495, "grad_norm": 13.62447452545166, "learning_rate": 1.2203883495145632e-05, "loss": 0.2021, "step": 18086 }, { "epoch": 175.3867469879518, "grad_norm": 12.55401611328125, "learning_rate": 1.2199029126213593e-05, "loss": 0.2737, "step": 18087 }, { "epoch": 175.39638554216867, "grad_norm": 3.722033977508545, "learning_rate": 1.2194174757281554e-05, "loss": 0.0863, "step": 18088 }, { "epoch": 175.40602409638555, "grad_norm": 6.375304222106934, "learning_rate": 1.2189320388349516e-05, "loss": 0.2702, "step": 18089 }, { "epoch": 175.4156626506024, "grad_norm": 5.485328674316406, "learning_rate": 1.2184466019417475e-05, "loss": 0.1891, "step": 18090 }, { "epoch": 175.42530120481928, "grad_norm": 18.815889358520508, "learning_rate": 1.2179611650485436e-05, "loss": 0.244, "step": 18091 }, { "epoch": 175.43493975903615, "grad_norm": 38.938865661621094, "learning_rate": 1.2174757281553399e-05, "loss": 0.1618, "step": 18092 }, { "epoch": 175.444578313253, "grad_norm": 4.07663106918335, "learning_rate": 1.216990291262136e-05, "loss": 0.0725, "step": 18093 }, { "epoch": 175.45421686746988, "grad_norm": 4.653036117553711, "learning_rate": 1.2165048543689321e-05, "loss": 0.1475, "step": 18094 }, { "epoch": 175.46385542168676, "grad_norm": 11.502801895141602, "learning_rate": 1.2160194174757282e-05, "loss": 0.2701, "step": 18095 }, { "epoch": 175.4734939759036, "grad_norm": 12.119050025939941, "learning_rate": 1.2155339805825244e-05, "loss": 0.3726, "step": 18096 }, { "epoch": 175.48313253012049, "grad_norm": 10.616097450256348, "learning_rate": 1.2150485436893203e-05, "loss": 0.2737, "step": 18097 }, { "epoch": 175.49277108433736, "grad_norm": 9.786413192749023, "learning_rate": 1.2145631067961166e-05, "loss": 0.2881, "step": 18098 }, { "epoch": 175.5024096385542, "grad_norm": 11.643272399902344, "learning_rate": 1.2140776699029127e-05, "loss": 0.2259, "step": 18099 }, { "epoch": 175.5120481927711, "grad_norm": 7.619131088256836, "learning_rate": 1.2135922330097088e-05, "loss": 0.156, "step": 18100 }, { "epoch": 175.52168674698794, "grad_norm": 26.20686149597168, "learning_rate": 1.213106796116505e-05, "loss": 0.2942, "step": 18101 }, { "epoch": 175.53132530120482, "grad_norm": 7.390076160430908, "learning_rate": 1.212621359223301e-05, "loss": 0.3449, "step": 18102 }, { "epoch": 175.5409638554217, "grad_norm": 8.562698364257812, "learning_rate": 1.2121359223300972e-05, "loss": 0.1968, "step": 18103 }, { "epoch": 175.55060240963854, "grad_norm": 13.14511489868164, "learning_rate": 1.2116504854368933e-05, "loss": 0.1868, "step": 18104 }, { "epoch": 175.56024096385542, "grad_norm": 15.268683433532715, "learning_rate": 1.2111650485436894e-05, "loss": 0.1812, "step": 18105 }, { "epoch": 175.5698795180723, "grad_norm": 7.5364227294921875, "learning_rate": 1.2106796116504855e-05, "loss": 0.2611, "step": 18106 }, { "epoch": 175.57951807228915, "grad_norm": 18.440465927124023, "learning_rate": 1.2101941747572816e-05, "loss": 0.1783, "step": 18107 }, { "epoch": 175.58915662650602, "grad_norm": 31.39453887939453, "learning_rate": 1.2097087378640777e-05, "loss": 0.3653, "step": 18108 }, { "epoch": 175.5987951807229, "grad_norm": 15.987449645996094, "learning_rate": 1.2092233009708739e-05, "loss": 0.2947, "step": 18109 }, { "epoch": 175.60843373493975, "grad_norm": 14.491399765014648, "learning_rate": 1.20873786407767e-05, "loss": 0.3417, "step": 18110 }, { "epoch": 175.61807228915663, "grad_norm": 14.872856140136719, "learning_rate": 1.208252427184466e-05, "loss": 0.2593, "step": 18111 }, { "epoch": 175.6277108433735, "grad_norm": 15.305326461791992, "learning_rate": 1.2077669902912622e-05, "loss": 0.2788, "step": 18112 }, { "epoch": 175.63734939759036, "grad_norm": 3.369706153869629, "learning_rate": 1.2072815533980583e-05, "loss": 0.0869, "step": 18113 }, { "epoch": 175.64698795180723, "grad_norm": 10.490243911743164, "learning_rate": 1.2067961165048544e-05, "loss": 0.1496, "step": 18114 }, { "epoch": 175.65662650602408, "grad_norm": 12.604949951171875, "learning_rate": 1.2063106796116505e-05, "loss": 0.4603, "step": 18115 }, { "epoch": 175.66626506024096, "grad_norm": 10.327322006225586, "learning_rate": 1.2058252427184467e-05, "loss": 0.1181, "step": 18116 }, { "epoch": 175.67590361445784, "grad_norm": 14.466951370239258, "learning_rate": 1.2053398058252428e-05, "loss": 0.3597, "step": 18117 }, { "epoch": 175.6855421686747, "grad_norm": 10.920855522155762, "learning_rate": 1.2048543689320389e-05, "loss": 0.161, "step": 18118 }, { "epoch": 175.69518072289156, "grad_norm": 12.916111946105957, "learning_rate": 1.204368932038835e-05, "loss": 0.2618, "step": 18119 }, { "epoch": 175.70481927710844, "grad_norm": 12.668294906616211, "learning_rate": 1.2038834951456311e-05, "loss": 0.2402, "step": 18120 }, { "epoch": 175.7144578313253, "grad_norm": 8.1838960647583, "learning_rate": 1.2033980582524272e-05, "loss": 0.2367, "step": 18121 }, { "epoch": 175.72409638554217, "grad_norm": 8.826282501220703, "learning_rate": 1.2029126213592233e-05, "loss": 0.2436, "step": 18122 }, { "epoch": 175.73373493975905, "grad_norm": 17.53232765197754, "learning_rate": 1.2024271844660195e-05, "loss": 0.4933, "step": 18123 }, { "epoch": 175.7433734939759, "grad_norm": 9.341836929321289, "learning_rate": 1.2019417475728156e-05, "loss": 0.4435, "step": 18124 }, { "epoch": 175.75301204819277, "grad_norm": 13.953376770019531, "learning_rate": 1.2014563106796117e-05, "loss": 0.2941, "step": 18125 }, { "epoch": 175.76265060240965, "grad_norm": 7.388225555419922, "learning_rate": 1.2009708737864078e-05, "loss": 0.2494, "step": 18126 }, { "epoch": 175.7722891566265, "grad_norm": 10.818621635437012, "learning_rate": 1.200485436893204e-05, "loss": 0.176, "step": 18127 }, { "epoch": 175.78192771084338, "grad_norm": 12.080755233764648, "learning_rate": 1.2e-05, "loss": 0.3704, "step": 18128 }, { "epoch": 175.79156626506025, "grad_norm": 13.065383911132812, "learning_rate": 1.1995145631067961e-05, "loss": 0.2314, "step": 18129 }, { "epoch": 175.8012048192771, "grad_norm": 32.67148971557617, "learning_rate": 1.1990291262135923e-05, "loss": 0.2428, "step": 18130 }, { "epoch": 175.81084337349398, "grad_norm": 7.775418758392334, "learning_rate": 1.1985436893203884e-05, "loss": 0.2003, "step": 18131 }, { "epoch": 175.82048192771083, "grad_norm": 7.854374885559082, "learning_rate": 1.1980582524271845e-05, "loss": 0.2623, "step": 18132 }, { "epoch": 175.8301204819277, "grad_norm": 10.910530090332031, "learning_rate": 1.1975728155339806e-05, "loss": 0.4174, "step": 18133 }, { "epoch": 175.83975903614459, "grad_norm": 33.49671936035156, "learning_rate": 1.1970873786407767e-05, "loss": 0.4642, "step": 18134 }, { "epoch": 175.84939759036143, "grad_norm": 12.399606704711914, "learning_rate": 1.1966019417475728e-05, "loss": 0.2655, "step": 18135 }, { "epoch": 175.8590361445783, "grad_norm": 10.36471939086914, "learning_rate": 1.1961165048543691e-05, "loss": 0.2936, "step": 18136 }, { "epoch": 175.8686746987952, "grad_norm": 7.398384094238281, "learning_rate": 1.195631067961165e-05, "loss": 0.3148, "step": 18137 }, { "epoch": 175.87831325301204, "grad_norm": 31.286794662475586, "learning_rate": 1.1951456310679612e-05, "loss": 0.2804, "step": 18138 }, { "epoch": 175.88795180722892, "grad_norm": 22.19078826904297, "learning_rate": 1.1946601941747573e-05, "loss": 0.2387, "step": 18139 }, { "epoch": 175.8975903614458, "grad_norm": 33.283058166503906, "learning_rate": 1.1941747572815534e-05, "loss": 0.2178, "step": 18140 }, { "epoch": 175.90722891566264, "grad_norm": 3.117434501647949, "learning_rate": 1.1936893203883495e-05, "loss": 0.0552, "step": 18141 }, { "epoch": 175.91686746987952, "grad_norm": 15.91683578491211, "learning_rate": 1.1932038834951458e-05, "loss": 0.3442, "step": 18142 }, { "epoch": 175.9265060240964, "grad_norm": 9.137361526489258, "learning_rate": 1.192718446601942e-05, "loss": 0.2097, "step": 18143 }, { "epoch": 175.93614457831325, "grad_norm": 67.23040771484375, "learning_rate": 1.1922330097087379e-05, "loss": 0.1903, "step": 18144 }, { "epoch": 175.94578313253012, "grad_norm": 9.297663688659668, "learning_rate": 1.191747572815534e-05, "loss": 0.206, "step": 18145 }, { "epoch": 175.955421686747, "grad_norm": 10.17493724822998, "learning_rate": 1.1912621359223301e-05, "loss": 0.2701, "step": 18146 }, { "epoch": 175.96506024096385, "grad_norm": 4.960190296173096, "learning_rate": 1.1907766990291262e-05, "loss": 0.143, "step": 18147 }, { "epoch": 175.97469879518073, "grad_norm": 9.636920928955078, "learning_rate": 1.1902912621359223e-05, "loss": 0.409, "step": 18148 }, { "epoch": 175.98433734939758, "grad_norm": 15.367450714111328, "learning_rate": 1.1898058252427186e-05, "loss": 0.6264, "step": 18149 }, { "epoch": 175.99397590361446, "grad_norm": 20.38266944885254, "learning_rate": 1.1893203883495147e-05, "loss": 0.3324, "step": 18150 }, { "epoch": 176.00963855421688, "grad_norm": 6.562195777893066, "learning_rate": 1.1888349514563107e-05, "loss": 0.2379, "step": 18151 }, { "epoch": 176.01927710843373, "grad_norm": 12.103843688964844, "learning_rate": 1.1883495145631068e-05, "loss": 0.2326, "step": 18152 }, { "epoch": 176.0289156626506, "grad_norm": 9.474990844726562, "learning_rate": 1.1878640776699029e-05, "loss": 0.3296, "step": 18153 }, { "epoch": 176.03855421686748, "grad_norm": 7.10615873336792, "learning_rate": 1.187378640776699e-05, "loss": 0.1843, "step": 18154 }, { "epoch": 176.04819277108433, "grad_norm": 13.21213150024414, "learning_rate": 1.1868932038834953e-05, "loss": 0.4588, "step": 18155 }, { "epoch": 176.0578313253012, "grad_norm": 10.127254486083984, "learning_rate": 1.1864077669902914e-05, "loss": 0.2307, "step": 18156 }, { "epoch": 176.0674698795181, "grad_norm": 7.135554313659668, "learning_rate": 1.1859223300970875e-05, "loss": 0.3259, "step": 18157 }, { "epoch": 176.07710843373494, "grad_norm": 11.555299758911133, "learning_rate": 1.1854368932038835e-05, "loss": 0.2781, "step": 18158 }, { "epoch": 176.0867469879518, "grad_norm": 16.689374923706055, "learning_rate": 1.1849514563106796e-05, "loss": 0.4205, "step": 18159 }, { "epoch": 176.09638554216866, "grad_norm": 12.127108573913574, "learning_rate": 1.1844660194174757e-05, "loss": 0.2281, "step": 18160 }, { "epoch": 176.10602409638554, "grad_norm": 11.509618759155273, "learning_rate": 1.183980582524272e-05, "loss": 0.2291, "step": 18161 }, { "epoch": 176.11566265060242, "grad_norm": 12.66808032989502, "learning_rate": 1.1834951456310681e-05, "loss": 0.3111, "step": 18162 }, { "epoch": 176.12530120481927, "grad_norm": 14.039935111999512, "learning_rate": 1.1830097087378642e-05, "loss": 0.341, "step": 18163 }, { "epoch": 176.13493975903614, "grad_norm": 15.839251518249512, "learning_rate": 1.1825242718446603e-05, "loss": 0.1531, "step": 18164 }, { "epoch": 176.14457831325302, "grad_norm": 8.8616943359375, "learning_rate": 1.1820388349514563e-05, "loss": 0.299, "step": 18165 }, { "epoch": 176.15421686746987, "grad_norm": 10.886855125427246, "learning_rate": 1.1815533980582524e-05, "loss": 0.1201, "step": 18166 }, { "epoch": 176.16385542168675, "grad_norm": 7.159409999847412, "learning_rate": 1.1810679611650485e-05, "loss": 0.1971, "step": 18167 }, { "epoch": 176.17349397590363, "grad_norm": 13.37326717376709, "learning_rate": 1.1805825242718448e-05, "loss": 0.2261, "step": 18168 }, { "epoch": 176.18313253012047, "grad_norm": 10.85305118560791, "learning_rate": 1.1800970873786409e-05, "loss": 0.2692, "step": 18169 }, { "epoch": 176.19277108433735, "grad_norm": 9.011480331420898, "learning_rate": 1.179611650485437e-05, "loss": 0.2747, "step": 18170 }, { "epoch": 176.20240963855423, "grad_norm": 12.607632637023926, "learning_rate": 1.1791262135922331e-05, "loss": 0.2755, "step": 18171 }, { "epoch": 176.21204819277108, "grad_norm": 16.645811080932617, "learning_rate": 1.178640776699029e-05, "loss": 0.3158, "step": 18172 }, { "epoch": 176.22168674698796, "grad_norm": 13.552628517150879, "learning_rate": 1.1781553398058252e-05, "loss": 0.3253, "step": 18173 }, { "epoch": 176.2313253012048, "grad_norm": 9.877299308776855, "learning_rate": 1.1776699029126215e-05, "loss": 0.1418, "step": 18174 }, { "epoch": 176.24096385542168, "grad_norm": 57.36064147949219, "learning_rate": 1.1771844660194176e-05, "loss": 0.1117, "step": 18175 }, { "epoch": 176.25060240963856, "grad_norm": 14.249866485595703, "learning_rate": 1.1766990291262137e-05, "loss": 0.3062, "step": 18176 }, { "epoch": 176.2602409638554, "grad_norm": 15.074628829956055, "learning_rate": 1.1762135922330098e-05, "loss": 0.2961, "step": 18177 }, { "epoch": 176.2698795180723, "grad_norm": 7.262155055999756, "learning_rate": 1.175728155339806e-05, "loss": 0.1663, "step": 18178 }, { "epoch": 176.27951807228916, "grad_norm": 17.775043487548828, "learning_rate": 1.1752427184466019e-05, "loss": 0.491, "step": 18179 }, { "epoch": 176.289156626506, "grad_norm": 14.743439674377441, "learning_rate": 1.1747572815533982e-05, "loss": 0.3292, "step": 18180 }, { "epoch": 176.2987951807229, "grad_norm": 7.578374862670898, "learning_rate": 1.1742718446601943e-05, "loss": 0.177, "step": 18181 }, { "epoch": 176.30843373493977, "grad_norm": 16.638999938964844, "learning_rate": 1.1737864077669904e-05, "loss": 0.2271, "step": 18182 }, { "epoch": 176.31807228915662, "grad_norm": 23.059036254882812, "learning_rate": 1.1733009708737865e-05, "loss": 0.4748, "step": 18183 }, { "epoch": 176.3277108433735, "grad_norm": 13.538347244262695, "learning_rate": 1.1728155339805826e-05, "loss": 0.2095, "step": 18184 }, { "epoch": 176.33734939759037, "grad_norm": 8.073587417602539, "learning_rate": 1.1723300970873787e-05, "loss": 0.0739, "step": 18185 }, { "epoch": 176.34698795180722, "grad_norm": 6.245174407958984, "learning_rate": 1.1718446601941748e-05, "loss": 0.0933, "step": 18186 }, { "epoch": 176.3566265060241, "grad_norm": 10.520028114318848, "learning_rate": 1.171359223300971e-05, "loss": 0.2045, "step": 18187 }, { "epoch": 176.36626506024098, "grad_norm": 13.380317687988281, "learning_rate": 1.170873786407767e-05, "loss": 0.2935, "step": 18188 }, { "epoch": 176.37590361445783, "grad_norm": 16.946496963500977, "learning_rate": 1.1703883495145632e-05, "loss": 0.3985, "step": 18189 }, { "epoch": 176.3855421686747, "grad_norm": 10.178681373596191, "learning_rate": 1.1699029126213593e-05, "loss": 0.3025, "step": 18190 }, { "epoch": 176.39518072289155, "grad_norm": 17.03704071044922, "learning_rate": 1.1694174757281554e-05, "loss": 0.3267, "step": 18191 }, { "epoch": 176.40481927710843, "grad_norm": 18.592544555664062, "learning_rate": 1.1689320388349515e-05, "loss": 0.1591, "step": 18192 }, { "epoch": 176.4144578313253, "grad_norm": 12.86037540435791, "learning_rate": 1.1684466019417476e-05, "loss": 0.2633, "step": 18193 }, { "epoch": 176.42409638554216, "grad_norm": 5.934028148651123, "learning_rate": 1.1679611650485438e-05, "loss": 0.2662, "step": 18194 }, { "epoch": 176.43373493975903, "grad_norm": 47.04694366455078, "learning_rate": 1.1674757281553399e-05, "loss": 0.4086, "step": 18195 }, { "epoch": 176.4433734939759, "grad_norm": 7.930058002471924, "learning_rate": 1.166990291262136e-05, "loss": 0.2747, "step": 18196 }, { "epoch": 176.45301204819276, "grad_norm": 10.231979370117188, "learning_rate": 1.1665048543689321e-05, "loss": 0.1863, "step": 18197 }, { "epoch": 176.46265060240964, "grad_norm": 13.87958812713623, "learning_rate": 1.1660194174757282e-05, "loss": 0.3185, "step": 18198 }, { "epoch": 176.47228915662652, "grad_norm": 8.797310829162598, "learning_rate": 1.1655339805825243e-05, "loss": 0.1212, "step": 18199 }, { "epoch": 176.48192771084337, "grad_norm": 11.571954727172852, "learning_rate": 1.1650485436893204e-05, "loss": 0.2432, "step": 18200 }, { "epoch": 176.49156626506024, "grad_norm": 10.923028945922852, "learning_rate": 1.1645631067961166e-05, "loss": 0.3701, "step": 18201 }, { "epoch": 176.50120481927712, "grad_norm": 9.078173637390137, "learning_rate": 1.1640776699029127e-05, "loss": 0.2704, "step": 18202 }, { "epoch": 176.51084337349397, "grad_norm": 5.765650272369385, "learning_rate": 1.1635922330097088e-05, "loss": 0.1574, "step": 18203 }, { "epoch": 176.52048192771085, "grad_norm": 12.840892791748047, "learning_rate": 1.1631067961165049e-05, "loss": 0.3555, "step": 18204 }, { "epoch": 176.53012048192772, "grad_norm": 27.392854690551758, "learning_rate": 1.162621359223301e-05, "loss": 0.2963, "step": 18205 }, { "epoch": 176.53975903614457, "grad_norm": 8.373364448547363, "learning_rate": 1.1621359223300971e-05, "loss": 0.0786, "step": 18206 }, { "epoch": 176.54939759036145, "grad_norm": 8.231325149536133, "learning_rate": 1.1616504854368933e-05, "loss": 0.3045, "step": 18207 }, { "epoch": 176.5590361445783, "grad_norm": 10.759428024291992, "learning_rate": 1.1611650485436894e-05, "loss": 0.197, "step": 18208 }, { "epoch": 176.56867469879518, "grad_norm": 8.293376922607422, "learning_rate": 1.1606796116504855e-05, "loss": 0.117, "step": 18209 }, { "epoch": 176.57831325301206, "grad_norm": 4.6319780349731445, "learning_rate": 1.1601941747572816e-05, "loss": 0.1387, "step": 18210 }, { "epoch": 176.5879518072289, "grad_norm": 5.4157562255859375, "learning_rate": 1.1597087378640777e-05, "loss": 0.0785, "step": 18211 }, { "epoch": 176.59759036144578, "grad_norm": 6.726620674133301, "learning_rate": 1.1592233009708738e-05, "loss": 0.3045, "step": 18212 }, { "epoch": 176.60722891566266, "grad_norm": 5.840961933135986, "learning_rate": 1.15873786407767e-05, "loss": 0.1501, "step": 18213 }, { "epoch": 176.6168674698795, "grad_norm": 16.2879638671875, "learning_rate": 1.158252427184466e-05, "loss": 0.1152, "step": 18214 }, { "epoch": 176.6265060240964, "grad_norm": 9.337423324584961, "learning_rate": 1.1577669902912622e-05, "loss": 0.1832, "step": 18215 }, { "epoch": 176.63614457831326, "grad_norm": 8.702720642089844, "learning_rate": 1.1572815533980583e-05, "loss": 0.21, "step": 18216 }, { "epoch": 176.6457831325301, "grad_norm": 17.102657318115234, "learning_rate": 1.1567961165048544e-05, "loss": 0.4225, "step": 18217 }, { "epoch": 176.655421686747, "grad_norm": 9.67423152923584, "learning_rate": 1.1563106796116507e-05, "loss": 0.3802, "step": 18218 }, { "epoch": 176.66506024096387, "grad_norm": 9.956989288330078, "learning_rate": 1.1558252427184466e-05, "loss": 0.2704, "step": 18219 }, { "epoch": 176.67469879518072, "grad_norm": 27.34269905090332, "learning_rate": 1.1553398058252427e-05, "loss": 0.3122, "step": 18220 }, { "epoch": 176.6843373493976, "grad_norm": 5.976651668548584, "learning_rate": 1.1548543689320389e-05, "loss": 0.1743, "step": 18221 }, { "epoch": 176.69397590361444, "grad_norm": 7.525356769561768, "learning_rate": 1.154368932038835e-05, "loss": 0.2483, "step": 18222 }, { "epoch": 176.70361445783132, "grad_norm": 8.828043937683105, "learning_rate": 1.153883495145631e-05, "loss": 0.4316, "step": 18223 }, { "epoch": 176.7132530120482, "grad_norm": 11.034725189208984, "learning_rate": 1.1533980582524274e-05, "loss": 0.2399, "step": 18224 }, { "epoch": 176.72289156626505, "grad_norm": 11.099446296691895, "learning_rate": 1.1529126213592235e-05, "loss": 0.4894, "step": 18225 }, { "epoch": 176.73253012048193, "grad_norm": 5.436712265014648, "learning_rate": 1.1524271844660194e-05, "loss": 0.1536, "step": 18226 }, { "epoch": 176.7421686746988, "grad_norm": 9.394115447998047, "learning_rate": 1.1519417475728155e-05, "loss": 0.1707, "step": 18227 }, { "epoch": 176.75180722891565, "grad_norm": 7.226836204528809, "learning_rate": 1.1514563106796117e-05, "loss": 0.2688, "step": 18228 }, { "epoch": 176.76144578313253, "grad_norm": 7.728806972503662, "learning_rate": 1.1509708737864078e-05, "loss": 0.2125, "step": 18229 }, { "epoch": 176.7710843373494, "grad_norm": 12.145700454711914, "learning_rate": 1.1504854368932039e-05, "loss": 0.4019, "step": 18230 }, { "epoch": 176.78072289156626, "grad_norm": 14.327701568603516, "learning_rate": 1.1500000000000002e-05, "loss": 0.2101, "step": 18231 }, { "epoch": 176.79036144578313, "grad_norm": 5.912349224090576, "learning_rate": 1.1495145631067961e-05, "loss": 0.2771, "step": 18232 }, { "epoch": 176.8, "grad_norm": 14.298460006713867, "learning_rate": 1.1490291262135922e-05, "loss": 0.224, "step": 18233 }, { "epoch": 176.80963855421686, "grad_norm": 7.728166580200195, "learning_rate": 1.1485436893203883e-05, "loss": 0.2621, "step": 18234 }, { "epoch": 176.81927710843374, "grad_norm": 13.814085006713867, "learning_rate": 1.1480582524271845e-05, "loss": 0.3432, "step": 18235 }, { "epoch": 176.82891566265062, "grad_norm": 20.170007705688477, "learning_rate": 1.1475728155339806e-05, "loss": 0.2585, "step": 18236 }, { "epoch": 176.83855421686746, "grad_norm": 12.736166000366211, "learning_rate": 1.1470873786407769e-05, "loss": 0.3787, "step": 18237 }, { "epoch": 176.84819277108434, "grad_norm": 7.230924606323242, "learning_rate": 1.146601941747573e-05, "loss": 0.2621, "step": 18238 }, { "epoch": 176.8578313253012, "grad_norm": 12.699268341064453, "learning_rate": 1.146116504854369e-05, "loss": 0.2012, "step": 18239 }, { "epoch": 176.86746987951807, "grad_norm": 13.809453010559082, "learning_rate": 1.145631067961165e-05, "loss": 0.4545, "step": 18240 }, { "epoch": 176.87710843373495, "grad_norm": 10.410402297973633, "learning_rate": 1.1451456310679611e-05, "loss": 0.1407, "step": 18241 }, { "epoch": 176.8867469879518, "grad_norm": 9.737778663635254, "learning_rate": 1.1446601941747573e-05, "loss": 0.245, "step": 18242 }, { "epoch": 176.89638554216867, "grad_norm": 10.779682159423828, "learning_rate": 1.1441747572815535e-05, "loss": 0.39, "step": 18243 }, { "epoch": 176.90602409638555, "grad_norm": 15.466655731201172, "learning_rate": 1.1436893203883497e-05, "loss": 0.3098, "step": 18244 }, { "epoch": 176.9156626506024, "grad_norm": 9.335443496704102, "learning_rate": 1.1432038834951458e-05, "loss": 0.237, "step": 18245 }, { "epoch": 176.92530120481928, "grad_norm": 14.783263206481934, "learning_rate": 1.1427184466019417e-05, "loss": 0.2774, "step": 18246 }, { "epoch": 176.93493975903615, "grad_norm": 9.587711334228516, "learning_rate": 1.1422330097087378e-05, "loss": 0.2223, "step": 18247 }, { "epoch": 176.944578313253, "grad_norm": 10.165375709533691, "learning_rate": 1.141747572815534e-05, "loss": 0.1622, "step": 18248 }, { "epoch": 176.95421686746988, "grad_norm": 9.106081008911133, "learning_rate": 1.14126213592233e-05, "loss": 0.2121, "step": 18249 }, { "epoch": 176.96385542168676, "grad_norm": 11.659732818603516, "learning_rate": 1.1407766990291263e-05, "loss": 0.3615, "step": 18250 }, { "epoch": 176.9734939759036, "grad_norm": 6.312510967254639, "learning_rate": 1.1402912621359225e-05, "loss": 0.1627, "step": 18251 }, { "epoch": 176.98313253012049, "grad_norm": 8.518218994140625, "learning_rate": 1.1398058252427186e-05, "loss": 0.2265, "step": 18252 }, { "epoch": 176.99277108433736, "grad_norm": 15.624192237854004, "learning_rate": 1.1393203883495145e-05, "loss": 0.1364, "step": 18253 }, { "epoch": 177.00843373493976, "grad_norm": 11.603181838989258, "learning_rate": 1.1388349514563106e-05, "loss": 0.2892, "step": 18254 }, { "epoch": 177.01807228915663, "grad_norm": 11.666072845458984, "learning_rate": 1.1383495145631068e-05, "loss": 0.3626, "step": 18255 }, { "epoch": 177.02771084337348, "grad_norm": 20.637041091918945, "learning_rate": 1.137864077669903e-05, "loss": 0.1932, "step": 18256 }, { "epoch": 177.03734939759036, "grad_norm": 19.82520866394043, "learning_rate": 1.1373786407766991e-05, "loss": 0.3411, "step": 18257 }, { "epoch": 177.04698795180724, "grad_norm": 16.781185150146484, "learning_rate": 1.1368932038834953e-05, "loss": 0.2714, "step": 18258 }, { "epoch": 177.0566265060241, "grad_norm": 3.5603413581848145, "learning_rate": 1.1364077669902914e-05, "loss": 0.0493, "step": 18259 }, { "epoch": 177.06626506024097, "grad_norm": 4.046047210693359, "learning_rate": 1.1359223300970873e-05, "loss": 0.161, "step": 18260 }, { "epoch": 177.07590361445784, "grad_norm": 26.877437591552734, "learning_rate": 1.1354368932038834e-05, "loss": 0.2145, "step": 18261 }, { "epoch": 177.0855421686747, "grad_norm": 32.20992660522461, "learning_rate": 1.1349514563106797e-05, "loss": 0.1864, "step": 18262 }, { "epoch": 177.09518072289157, "grad_norm": 7.098305702209473, "learning_rate": 1.1344660194174758e-05, "loss": 0.156, "step": 18263 }, { "epoch": 177.10481927710845, "grad_norm": 18.786165237426758, "learning_rate": 1.133980582524272e-05, "loss": 0.347, "step": 18264 }, { "epoch": 177.1144578313253, "grad_norm": 23.809404373168945, "learning_rate": 1.133495145631068e-05, "loss": 0.3394, "step": 18265 }, { "epoch": 177.12409638554217, "grad_norm": 5.003843784332275, "learning_rate": 1.1330097087378642e-05, "loss": 0.1268, "step": 18266 }, { "epoch": 177.13373493975902, "grad_norm": 2.025139570236206, "learning_rate": 1.1325242718446601e-05, "loss": 0.3218, "step": 18267 }, { "epoch": 177.1433734939759, "grad_norm": 6.848067283630371, "learning_rate": 1.1320388349514564e-05, "loss": 0.1912, "step": 18268 }, { "epoch": 177.15301204819278, "grad_norm": 15.329055786132812, "learning_rate": 1.1315533980582525e-05, "loss": 0.2583, "step": 18269 }, { "epoch": 177.16265060240963, "grad_norm": 18.100582122802734, "learning_rate": 1.1310679611650486e-05, "loss": 0.3269, "step": 18270 }, { "epoch": 177.1722891566265, "grad_norm": 5.6784491539001465, "learning_rate": 1.1305825242718448e-05, "loss": 0.2153, "step": 18271 }, { "epoch": 177.18192771084338, "grad_norm": 24.032243728637695, "learning_rate": 1.1300970873786409e-05, "loss": 0.1381, "step": 18272 }, { "epoch": 177.19156626506023, "grad_norm": 18.596445083618164, "learning_rate": 1.129611650485437e-05, "loss": 0.1194, "step": 18273 }, { "epoch": 177.2012048192771, "grad_norm": 3.8787057399749756, "learning_rate": 1.129126213592233e-05, "loss": 0.1706, "step": 18274 }, { "epoch": 177.210843373494, "grad_norm": 10.944972038269043, "learning_rate": 1.1286407766990292e-05, "loss": 0.184, "step": 18275 }, { "epoch": 177.22048192771084, "grad_norm": 1.9742194414138794, "learning_rate": 1.1281553398058253e-05, "loss": 0.1504, "step": 18276 }, { "epoch": 177.2301204819277, "grad_norm": 22.75920295715332, "learning_rate": 1.1276699029126214e-05, "loss": 0.3379, "step": 18277 }, { "epoch": 177.2397590361446, "grad_norm": 11.389951705932617, "learning_rate": 1.1271844660194176e-05, "loss": 0.2379, "step": 18278 }, { "epoch": 177.24939759036144, "grad_norm": 20.21289825439453, "learning_rate": 1.1266990291262137e-05, "loss": 0.1466, "step": 18279 }, { "epoch": 177.25903614457832, "grad_norm": 16.908891677856445, "learning_rate": 1.1262135922330098e-05, "loss": 0.227, "step": 18280 }, { "epoch": 177.2686746987952, "grad_norm": 4.996585369110107, "learning_rate": 1.1257281553398059e-05, "loss": 0.2044, "step": 18281 }, { "epoch": 177.27831325301204, "grad_norm": 4.048089981079102, "learning_rate": 1.125242718446602e-05, "loss": 0.267, "step": 18282 }, { "epoch": 177.28795180722892, "grad_norm": 13.244439125061035, "learning_rate": 1.1247572815533981e-05, "loss": 0.2946, "step": 18283 }, { "epoch": 177.29759036144577, "grad_norm": 19.213863372802734, "learning_rate": 1.1242718446601942e-05, "loss": 0.3385, "step": 18284 }, { "epoch": 177.30722891566265, "grad_norm": 3.9914498329162598, "learning_rate": 1.1237864077669904e-05, "loss": 0.1491, "step": 18285 }, { "epoch": 177.31686746987953, "grad_norm": 18.201122283935547, "learning_rate": 1.1233009708737865e-05, "loss": 0.1468, "step": 18286 }, { "epoch": 177.32650602409637, "grad_norm": 4.3637776374816895, "learning_rate": 1.1228155339805826e-05, "loss": 0.0902, "step": 18287 }, { "epoch": 177.33614457831325, "grad_norm": 51.4050407409668, "learning_rate": 1.1223300970873787e-05, "loss": 0.2864, "step": 18288 }, { "epoch": 177.34578313253013, "grad_norm": 19.2901611328125, "learning_rate": 1.1218446601941748e-05, "loss": 0.376, "step": 18289 }, { "epoch": 177.35542168674698, "grad_norm": 33.61515808105469, "learning_rate": 1.121359223300971e-05, "loss": 0.3359, "step": 18290 }, { "epoch": 177.36506024096386, "grad_norm": 20.98360824584961, "learning_rate": 1.120873786407767e-05, "loss": 0.2183, "step": 18291 }, { "epoch": 177.37469879518073, "grad_norm": 23.50493812561035, "learning_rate": 1.1203883495145632e-05, "loss": 0.2051, "step": 18292 }, { "epoch": 177.38433734939758, "grad_norm": 25.401235580444336, "learning_rate": 1.1199029126213593e-05, "loss": 0.2932, "step": 18293 }, { "epoch": 177.39397590361446, "grad_norm": 28.898101806640625, "learning_rate": 1.1194174757281554e-05, "loss": 0.2988, "step": 18294 }, { "epoch": 177.40361445783134, "grad_norm": 11.090408325195312, "learning_rate": 1.1189320388349515e-05, "loss": 0.4348, "step": 18295 }, { "epoch": 177.4132530120482, "grad_norm": 21.72368049621582, "learning_rate": 1.1184466019417476e-05, "loss": 0.3333, "step": 18296 }, { "epoch": 177.42289156626506, "grad_norm": 7.8738322257995605, "learning_rate": 1.1179611650485437e-05, "loss": 0.2485, "step": 18297 }, { "epoch": 177.4325301204819, "grad_norm": 9.345586776733398, "learning_rate": 1.1174757281553398e-05, "loss": 0.3019, "step": 18298 }, { "epoch": 177.4421686746988, "grad_norm": 24.253995895385742, "learning_rate": 1.116990291262136e-05, "loss": 0.3569, "step": 18299 }, { "epoch": 177.45180722891567, "grad_norm": 13.346090316772461, "learning_rate": 1.116504854368932e-05, "loss": 0.283, "step": 18300 }, { "epoch": 177.46144578313252, "grad_norm": 27.31892204284668, "learning_rate": 1.1160194174757282e-05, "loss": 0.2447, "step": 18301 }, { "epoch": 177.4710843373494, "grad_norm": 6.6125969886779785, "learning_rate": 1.1155339805825243e-05, "loss": 0.5181, "step": 18302 }, { "epoch": 177.48072289156627, "grad_norm": 18.954288482666016, "learning_rate": 1.1150485436893204e-05, "loss": 0.2521, "step": 18303 }, { "epoch": 177.49036144578312, "grad_norm": 28.40973472595215, "learning_rate": 1.1145631067961165e-05, "loss": 0.439, "step": 18304 }, { "epoch": 177.5, "grad_norm": 3.6042890548706055, "learning_rate": 1.1140776699029126e-05, "loss": 0.1186, "step": 18305 }, { "epoch": 177.50963855421688, "grad_norm": 6.116580009460449, "learning_rate": 1.113592233009709e-05, "loss": 0.259, "step": 18306 }, { "epoch": 177.51927710843373, "grad_norm": 5.631155967712402, "learning_rate": 1.1131067961165049e-05, "loss": 0.4312, "step": 18307 }, { "epoch": 177.5289156626506, "grad_norm": 7.253868103027344, "learning_rate": 1.112621359223301e-05, "loss": 0.1904, "step": 18308 }, { "epoch": 177.53855421686748, "grad_norm": 3.6584999561309814, "learning_rate": 1.1121359223300971e-05, "loss": 0.1102, "step": 18309 }, { "epoch": 177.54819277108433, "grad_norm": 30.829113006591797, "learning_rate": 1.1116504854368932e-05, "loss": 0.425, "step": 18310 }, { "epoch": 177.5578313253012, "grad_norm": 4.945383071899414, "learning_rate": 1.1111650485436893e-05, "loss": 0.2182, "step": 18311 }, { "epoch": 177.5674698795181, "grad_norm": 32.5109748840332, "learning_rate": 1.1106796116504855e-05, "loss": 0.28, "step": 18312 }, { "epoch": 177.57710843373494, "grad_norm": 8.321294784545898, "learning_rate": 1.1101941747572817e-05, "loss": 0.27, "step": 18313 }, { "epoch": 177.5867469879518, "grad_norm": 5.603499412536621, "learning_rate": 1.1097087378640777e-05, "loss": 0.2488, "step": 18314 }, { "epoch": 177.59638554216866, "grad_norm": 72.43185424804688, "learning_rate": 1.1092233009708738e-05, "loss": 0.2883, "step": 18315 }, { "epoch": 177.60602409638554, "grad_norm": 70.79692840576172, "learning_rate": 1.1087378640776699e-05, "loss": 0.2129, "step": 18316 }, { "epoch": 177.61566265060242, "grad_norm": 6.806614398956299, "learning_rate": 1.108252427184466e-05, "loss": 0.2203, "step": 18317 }, { "epoch": 177.62530120481927, "grad_norm": 21.655166625976562, "learning_rate": 1.1077669902912621e-05, "loss": 0.2874, "step": 18318 }, { "epoch": 177.63493975903614, "grad_norm": 19.665807723999023, "learning_rate": 1.1072815533980584e-05, "loss": 0.2281, "step": 18319 }, { "epoch": 177.64457831325302, "grad_norm": 2.019343852996826, "learning_rate": 1.1067961165048545e-05, "loss": 0.1726, "step": 18320 }, { "epoch": 177.65421686746987, "grad_norm": 10.123568534851074, "learning_rate": 1.1063106796116505e-05, "loss": 0.2348, "step": 18321 }, { "epoch": 177.66385542168675, "grad_norm": 50.75279235839844, "learning_rate": 1.1058252427184466e-05, "loss": 0.4613, "step": 18322 }, { "epoch": 177.67349397590363, "grad_norm": 14.092398643493652, "learning_rate": 1.1053398058252427e-05, "loss": 0.1985, "step": 18323 }, { "epoch": 177.68313253012047, "grad_norm": 5.254735946655273, "learning_rate": 1.1048543689320388e-05, "loss": 0.1257, "step": 18324 }, { "epoch": 177.69277108433735, "grad_norm": 11.379866600036621, "learning_rate": 1.1043689320388351e-05, "loss": 0.2491, "step": 18325 }, { "epoch": 177.70240963855423, "grad_norm": 2.3997480869293213, "learning_rate": 1.1038834951456312e-05, "loss": 0.0938, "step": 18326 }, { "epoch": 177.71204819277108, "grad_norm": 5.801321506500244, "learning_rate": 1.1033980582524273e-05, "loss": 0.2843, "step": 18327 }, { "epoch": 177.72168674698796, "grad_norm": 30.96162223815918, "learning_rate": 1.1029126213592233e-05, "loss": 0.2826, "step": 18328 }, { "epoch": 177.73132530120483, "grad_norm": 7.242746353149414, "learning_rate": 1.1024271844660194e-05, "loss": 0.2066, "step": 18329 }, { "epoch": 177.74096385542168, "grad_norm": 30.027250289916992, "learning_rate": 1.1019417475728155e-05, "loss": 0.2937, "step": 18330 }, { "epoch": 177.75060240963856, "grad_norm": 17.201995849609375, "learning_rate": 1.1014563106796116e-05, "loss": 0.3254, "step": 18331 }, { "epoch": 177.7602409638554, "grad_norm": 4.5796027183532715, "learning_rate": 1.1009708737864079e-05, "loss": 0.1192, "step": 18332 }, { "epoch": 177.7698795180723, "grad_norm": 8.608306884765625, "learning_rate": 1.100485436893204e-05, "loss": 0.1383, "step": 18333 }, { "epoch": 177.77951807228916, "grad_norm": 7.320058822631836, "learning_rate": 1.1000000000000001e-05, "loss": 0.1174, "step": 18334 }, { "epoch": 177.789156626506, "grad_norm": 5.182530879974365, "learning_rate": 1.0995145631067961e-05, "loss": 0.2588, "step": 18335 }, { "epoch": 177.7987951807229, "grad_norm": 28.136619567871094, "learning_rate": 1.0990291262135922e-05, "loss": 0.3879, "step": 18336 }, { "epoch": 177.80843373493977, "grad_norm": 34.865814208984375, "learning_rate": 1.0985436893203883e-05, "loss": 0.3817, "step": 18337 }, { "epoch": 177.81807228915662, "grad_norm": 11.292692184448242, "learning_rate": 1.0980582524271846e-05, "loss": 0.4914, "step": 18338 }, { "epoch": 177.8277108433735, "grad_norm": 3.9717612266540527, "learning_rate": 1.0975728155339807e-05, "loss": 0.3648, "step": 18339 }, { "epoch": 177.83734939759037, "grad_norm": 88.3266830444336, "learning_rate": 1.0970873786407768e-05, "loss": 0.2441, "step": 18340 }, { "epoch": 177.84698795180722, "grad_norm": 24.7982234954834, "learning_rate": 1.096601941747573e-05, "loss": 0.2125, "step": 18341 }, { "epoch": 177.8566265060241, "grad_norm": 11.207266807556152, "learning_rate": 1.0961165048543689e-05, "loss": 0.2647, "step": 18342 }, { "epoch": 177.86626506024098, "grad_norm": 17.986886978149414, "learning_rate": 1.095631067961165e-05, "loss": 0.2996, "step": 18343 }, { "epoch": 177.87590361445783, "grad_norm": 9.624980926513672, "learning_rate": 1.0951456310679613e-05, "loss": 0.1518, "step": 18344 }, { "epoch": 177.8855421686747, "grad_norm": 19.56616973876953, "learning_rate": 1.0946601941747574e-05, "loss": 0.1786, "step": 18345 }, { "epoch": 177.89518072289155, "grad_norm": 30.4239501953125, "learning_rate": 1.0941747572815535e-05, "loss": 0.3045, "step": 18346 }, { "epoch": 177.90481927710843, "grad_norm": 18.172555923461914, "learning_rate": 1.0936893203883496e-05, "loss": 0.296, "step": 18347 }, { "epoch": 177.9144578313253, "grad_norm": 3.8960940837860107, "learning_rate": 1.0932038834951456e-05, "loss": 0.278, "step": 18348 }, { "epoch": 177.92409638554216, "grad_norm": 40.45717239379883, "learning_rate": 1.0927184466019417e-05, "loss": 0.2663, "step": 18349 }, { "epoch": 177.93373493975903, "grad_norm": 2.117488384246826, "learning_rate": 1.0922330097087378e-05, "loss": 0.2146, "step": 18350 }, { "epoch": 177.9433734939759, "grad_norm": 10.044347763061523, "learning_rate": 1.0917475728155341e-05, "loss": 0.1107, "step": 18351 }, { "epoch": 177.95301204819276, "grad_norm": 38.18488311767578, "learning_rate": 1.0912621359223302e-05, "loss": 0.2977, "step": 18352 }, { "epoch": 177.96265060240964, "grad_norm": 31.153827667236328, "learning_rate": 1.0907766990291263e-05, "loss": 0.3937, "step": 18353 }, { "epoch": 177.97228915662652, "grad_norm": 6.977631092071533, "learning_rate": 1.0902912621359224e-05, "loss": 0.3176, "step": 18354 }, { "epoch": 177.98192771084337, "grad_norm": 9.28207778930664, "learning_rate": 1.0898058252427184e-05, "loss": 0.2829, "step": 18355 }, { "epoch": 177.99156626506024, "grad_norm": 16.839378356933594, "learning_rate": 1.0893203883495145e-05, "loss": 0.2407, "step": 18356 }, { "epoch": 178.00722891566264, "grad_norm": 10.089237213134766, "learning_rate": 1.0888349514563108e-05, "loss": 0.1755, "step": 18357 }, { "epoch": 178.01686746987951, "grad_norm": 13.338333129882812, "learning_rate": 1.0883495145631069e-05, "loss": 0.31, "step": 18358 }, { "epoch": 178.0265060240964, "grad_norm": 80.52385711669922, "learning_rate": 1.087864077669903e-05, "loss": 0.2986, "step": 18359 }, { "epoch": 178.03614457831324, "grad_norm": 8.965596199035645, "learning_rate": 1.0873786407766991e-05, "loss": 0.2622, "step": 18360 }, { "epoch": 178.04578313253012, "grad_norm": 10.577548027038574, "learning_rate": 1.0868932038834952e-05, "loss": 0.2902, "step": 18361 }, { "epoch": 178.055421686747, "grad_norm": 12.278730392456055, "learning_rate": 1.0864077669902912e-05, "loss": 0.2598, "step": 18362 }, { "epoch": 178.06506024096385, "grad_norm": 10.423871040344238, "learning_rate": 1.0859223300970875e-05, "loss": 0.1125, "step": 18363 }, { "epoch": 178.07469879518072, "grad_norm": 6.4683518409729, "learning_rate": 1.0854368932038836e-05, "loss": 0.3024, "step": 18364 }, { "epoch": 178.0843373493976, "grad_norm": 17.308488845825195, "learning_rate": 1.0849514563106797e-05, "loss": 0.2651, "step": 18365 }, { "epoch": 178.09397590361445, "grad_norm": 65.34349060058594, "learning_rate": 1.0844660194174758e-05, "loss": 0.2292, "step": 18366 }, { "epoch": 178.10361445783133, "grad_norm": 5.16770076751709, "learning_rate": 1.083980582524272e-05, "loss": 0.1535, "step": 18367 }, { "epoch": 178.1132530120482, "grad_norm": 37.075069427490234, "learning_rate": 1.083495145631068e-05, "loss": 0.1657, "step": 18368 }, { "epoch": 178.12289156626505, "grad_norm": 15.597580909729004, "learning_rate": 1.0830097087378642e-05, "loss": 0.1684, "step": 18369 }, { "epoch": 178.13253012048193, "grad_norm": 111.85942840576172, "learning_rate": 1.0825242718446603e-05, "loss": 0.3483, "step": 18370 }, { "epoch": 178.1421686746988, "grad_norm": 10.681689262390137, "learning_rate": 1.0820388349514564e-05, "loss": 0.1804, "step": 18371 }, { "epoch": 178.15180722891566, "grad_norm": 26.843971252441406, "learning_rate": 1.0815533980582525e-05, "loss": 0.1616, "step": 18372 }, { "epoch": 178.16144578313254, "grad_norm": 7.312065124511719, "learning_rate": 1.0810679611650486e-05, "loss": 0.322, "step": 18373 }, { "epoch": 178.17108433734938, "grad_norm": 2.601895332336426, "learning_rate": 1.0805825242718447e-05, "loss": 0.145, "step": 18374 }, { "epoch": 178.18072289156626, "grad_norm": 13.435084342956543, "learning_rate": 1.0800970873786408e-05, "loss": 0.2128, "step": 18375 }, { "epoch": 178.19036144578314, "grad_norm": 10.954198837280273, "learning_rate": 1.079611650485437e-05, "loss": 0.182, "step": 18376 }, { "epoch": 178.2, "grad_norm": 5.853315830230713, "learning_rate": 1.079126213592233e-05, "loss": 0.1918, "step": 18377 }, { "epoch": 178.20963855421687, "grad_norm": 14.061147689819336, "learning_rate": 1.0786407766990292e-05, "loss": 0.3502, "step": 18378 }, { "epoch": 178.21927710843374, "grad_norm": 11.026175498962402, "learning_rate": 1.0781553398058253e-05, "loss": 0.2662, "step": 18379 }, { "epoch": 178.2289156626506, "grad_norm": 6.432183742523193, "learning_rate": 1.0776699029126214e-05, "loss": 0.1499, "step": 18380 }, { "epoch": 178.23855421686747, "grad_norm": 37.49346160888672, "learning_rate": 1.0771844660194175e-05, "loss": 0.2008, "step": 18381 }, { "epoch": 178.24819277108435, "grad_norm": 15.977960586547852, "learning_rate": 1.0766990291262136e-05, "loss": 0.1244, "step": 18382 }, { "epoch": 178.2578313253012, "grad_norm": 15.264603614807129, "learning_rate": 1.0762135922330098e-05, "loss": 0.2768, "step": 18383 }, { "epoch": 178.26746987951807, "grad_norm": 39.92583465576172, "learning_rate": 1.0757281553398059e-05, "loss": 0.2288, "step": 18384 }, { "epoch": 178.27710843373495, "grad_norm": 11.632485389709473, "learning_rate": 1.075242718446602e-05, "loss": 0.1999, "step": 18385 }, { "epoch": 178.2867469879518, "grad_norm": 17.115081787109375, "learning_rate": 1.0747572815533981e-05, "loss": 0.226, "step": 18386 }, { "epoch": 178.29638554216868, "grad_norm": 6.978555679321289, "learning_rate": 1.0742718446601942e-05, "loss": 0.1772, "step": 18387 }, { "epoch": 178.30602409638556, "grad_norm": 3.4859673976898193, "learning_rate": 1.0737864077669903e-05, "loss": 0.2188, "step": 18388 }, { "epoch": 178.3156626506024, "grad_norm": 4.437595844268799, "learning_rate": 1.0733009708737864e-05, "loss": 0.2411, "step": 18389 }, { "epoch": 178.32530120481928, "grad_norm": 13.358569145202637, "learning_rate": 1.0728155339805826e-05, "loss": 0.1314, "step": 18390 }, { "epoch": 178.33493975903613, "grad_norm": 17.095277786254883, "learning_rate": 1.0723300970873787e-05, "loss": 0.2804, "step": 18391 }, { "epoch": 178.344578313253, "grad_norm": 6.418543338775635, "learning_rate": 1.0718446601941748e-05, "loss": 0.2513, "step": 18392 }, { "epoch": 178.3542168674699, "grad_norm": 28.077913284301758, "learning_rate": 1.0713592233009709e-05, "loss": 0.1881, "step": 18393 }, { "epoch": 178.36385542168674, "grad_norm": 30.489242553710938, "learning_rate": 1.070873786407767e-05, "loss": 0.3347, "step": 18394 }, { "epoch": 178.3734939759036, "grad_norm": 12.17304801940918, "learning_rate": 1.0703883495145631e-05, "loss": 0.2863, "step": 18395 }, { "epoch": 178.3831325301205, "grad_norm": 10.156965255737305, "learning_rate": 1.0699029126213592e-05, "loss": 0.2703, "step": 18396 }, { "epoch": 178.39277108433734, "grad_norm": 24.940412521362305, "learning_rate": 1.0694174757281554e-05, "loss": 0.2724, "step": 18397 }, { "epoch": 178.40240963855422, "grad_norm": 10.123764038085938, "learning_rate": 1.0689320388349515e-05, "loss": 0.3655, "step": 18398 }, { "epoch": 178.4120481927711, "grad_norm": 5.297074317932129, "learning_rate": 1.0684466019417476e-05, "loss": 0.218, "step": 18399 }, { "epoch": 178.42168674698794, "grad_norm": 17.740076065063477, "learning_rate": 1.0679611650485437e-05, "loss": 0.2084, "step": 18400 }, { "epoch": 178.43132530120482, "grad_norm": 4.094085216522217, "learning_rate": 1.06747572815534e-05, "loss": 0.2098, "step": 18401 }, { "epoch": 178.4409638554217, "grad_norm": 9.955385208129883, "learning_rate": 1.066990291262136e-05, "loss": 0.2867, "step": 18402 }, { "epoch": 178.45060240963855, "grad_norm": 5.892817497253418, "learning_rate": 1.066504854368932e-05, "loss": 0.294, "step": 18403 }, { "epoch": 178.46024096385543, "grad_norm": 22.96963119506836, "learning_rate": 1.0660194174757282e-05, "loss": 0.2009, "step": 18404 }, { "epoch": 178.46987951807228, "grad_norm": 26.601743698120117, "learning_rate": 1.0655339805825243e-05, "loss": 0.332, "step": 18405 }, { "epoch": 178.47951807228915, "grad_norm": 7.542873382568359, "learning_rate": 1.0650485436893204e-05, "loss": 0.3176, "step": 18406 }, { "epoch": 178.48915662650603, "grad_norm": 13.901658058166504, "learning_rate": 1.0645631067961167e-05, "loss": 0.0959, "step": 18407 }, { "epoch": 178.49879518072288, "grad_norm": 8.189131736755371, "learning_rate": 1.0640776699029128e-05, "loss": 0.3009, "step": 18408 }, { "epoch": 178.50843373493976, "grad_norm": 3.0336928367614746, "learning_rate": 1.0635922330097087e-05, "loss": 0.0892, "step": 18409 }, { "epoch": 178.51807228915663, "grad_norm": 22.76482582092285, "learning_rate": 1.0631067961165048e-05, "loss": 0.2504, "step": 18410 }, { "epoch": 178.52771084337348, "grad_norm": 32.55710983276367, "learning_rate": 1.062621359223301e-05, "loss": 0.1325, "step": 18411 }, { "epoch": 178.53734939759036, "grad_norm": 2.0967154502868652, "learning_rate": 1.062135922330097e-05, "loss": 0.2022, "step": 18412 }, { "epoch": 178.54698795180724, "grad_norm": 18.8737850189209, "learning_rate": 1.0616504854368932e-05, "loss": 0.3353, "step": 18413 }, { "epoch": 178.5566265060241, "grad_norm": 21.7515926361084, "learning_rate": 1.0611650485436895e-05, "loss": 0.1008, "step": 18414 }, { "epoch": 178.56626506024097, "grad_norm": 8.726319313049316, "learning_rate": 1.0606796116504856e-05, "loss": 0.2275, "step": 18415 }, { "epoch": 178.57590361445784, "grad_norm": 26.731346130371094, "learning_rate": 1.0601941747572815e-05, "loss": 0.2226, "step": 18416 }, { "epoch": 178.5855421686747, "grad_norm": 10.441251754760742, "learning_rate": 1.0597087378640777e-05, "loss": 0.3518, "step": 18417 }, { "epoch": 178.59518072289157, "grad_norm": 12.283153533935547, "learning_rate": 1.0592233009708738e-05, "loss": 0.1592, "step": 18418 }, { "epoch": 178.60481927710845, "grad_norm": 4.750656604766846, "learning_rate": 1.0587378640776699e-05, "loss": 0.1222, "step": 18419 }, { "epoch": 178.6144578313253, "grad_norm": 26.512191772460938, "learning_rate": 1.0582524271844662e-05, "loss": 0.4205, "step": 18420 }, { "epoch": 178.62409638554217, "grad_norm": 9.736953735351562, "learning_rate": 1.0577669902912623e-05, "loss": 0.2562, "step": 18421 }, { "epoch": 178.63373493975902, "grad_norm": 23.73917007446289, "learning_rate": 1.0572815533980584e-05, "loss": 0.3077, "step": 18422 }, { "epoch": 178.6433734939759, "grad_norm": 57.2884521484375, "learning_rate": 1.0567961165048543e-05, "loss": 0.2566, "step": 18423 }, { "epoch": 178.65301204819278, "grad_norm": 7.0104780197143555, "learning_rate": 1.0563106796116505e-05, "loss": 0.1651, "step": 18424 }, { "epoch": 178.66265060240963, "grad_norm": 12.842963218688965, "learning_rate": 1.0558252427184466e-05, "loss": 0.4115, "step": 18425 }, { "epoch": 178.6722891566265, "grad_norm": 4.558962345123291, "learning_rate": 1.0553398058252429e-05, "loss": 0.0979, "step": 18426 }, { "epoch": 178.68192771084338, "grad_norm": 3.4517745971679688, "learning_rate": 1.054854368932039e-05, "loss": 0.2624, "step": 18427 }, { "epoch": 178.69156626506023, "grad_norm": 17.378379821777344, "learning_rate": 1.054368932038835e-05, "loss": 0.1798, "step": 18428 }, { "epoch": 178.7012048192771, "grad_norm": 42.787017822265625, "learning_rate": 1.0538834951456312e-05, "loss": 0.4756, "step": 18429 }, { "epoch": 178.710843373494, "grad_norm": 4.628509044647217, "learning_rate": 1.0533980582524271e-05, "loss": 0.1198, "step": 18430 }, { "epoch": 178.72048192771084, "grad_norm": 3.7366979122161865, "learning_rate": 1.0529126213592233e-05, "loss": 0.2336, "step": 18431 }, { "epoch": 178.7301204819277, "grad_norm": 53.15837478637695, "learning_rate": 1.0524271844660194e-05, "loss": 0.3075, "step": 18432 }, { "epoch": 178.7397590361446, "grad_norm": 15.611054420471191, "learning_rate": 1.0519417475728157e-05, "loss": 0.1391, "step": 18433 }, { "epoch": 178.74939759036144, "grad_norm": 7.580196857452393, "learning_rate": 1.0514563106796118e-05, "loss": 0.1216, "step": 18434 }, { "epoch": 178.75903614457832, "grad_norm": 7.393687725067139, "learning_rate": 1.0509708737864079e-05, "loss": 0.235, "step": 18435 }, { "epoch": 178.76867469879517, "grad_norm": 31.208545684814453, "learning_rate": 1.050485436893204e-05, "loss": 0.1646, "step": 18436 }, { "epoch": 178.77831325301204, "grad_norm": 26.42801856994629, "learning_rate": 1.05e-05, "loss": 0.1036, "step": 18437 }, { "epoch": 178.78795180722892, "grad_norm": 95.42304992675781, "learning_rate": 1.049514563106796e-05, "loss": 0.2023, "step": 18438 }, { "epoch": 178.79759036144577, "grad_norm": 9.846586227416992, "learning_rate": 1.0490291262135923e-05, "loss": 0.2295, "step": 18439 }, { "epoch": 178.80722891566265, "grad_norm": 5.191041946411133, "learning_rate": 1.0485436893203885e-05, "loss": 0.2776, "step": 18440 }, { "epoch": 178.81686746987953, "grad_norm": 39.009056091308594, "learning_rate": 1.0480582524271846e-05, "loss": 0.3329, "step": 18441 }, { "epoch": 178.82650602409637, "grad_norm": 91.04637908935547, "learning_rate": 1.0475728155339807e-05, "loss": 0.3889, "step": 18442 }, { "epoch": 178.83614457831325, "grad_norm": 23.688199996948242, "learning_rate": 1.0470873786407768e-05, "loss": 0.1223, "step": 18443 }, { "epoch": 178.84578313253013, "grad_norm": 15.432411193847656, "learning_rate": 1.0466019417475727e-05, "loss": 0.2841, "step": 18444 }, { "epoch": 178.85542168674698, "grad_norm": 3.4358320236206055, "learning_rate": 1.046116504854369e-05, "loss": 0.1912, "step": 18445 }, { "epoch": 178.86506024096386, "grad_norm": 2.4331488609313965, "learning_rate": 1.0456310679611651e-05, "loss": 0.1575, "step": 18446 }, { "epoch": 178.87469879518073, "grad_norm": 6.506209850311279, "learning_rate": 1.0451456310679613e-05, "loss": 0.3279, "step": 18447 }, { "epoch": 178.88433734939758, "grad_norm": 7.467947959899902, "learning_rate": 1.0446601941747574e-05, "loss": 0.3398, "step": 18448 }, { "epoch": 178.89397590361446, "grad_norm": 15.19837760925293, "learning_rate": 1.0441747572815535e-05, "loss": 0.27, "step": 18449 }, { "epoch": 178.90361445783134, "grad_norm": 25.573476791381836, "learning_rate": 1.0436893203883496e-05, "loss": 0.2998, "step": 18450 }, { "epoch": 178.9132530120482, "grad_norm": 6.62333345413208, "learning_rate": 1.0432038834951455e-05, "loss": 0.2326, "step": 18451 }, { "epoch": 178.92289156626506, "grad_norm": 37.949005126953125, "learning_rate": 1.0427184466019418e-05, "loss": 0.1001, "step": 18452 }, { "epoch": 178.9325301204819, "grad_norm": 1.9132599830627441, "learning_rate": 1.042233009708738e-05, "loss": 0.1359, "step": 18453 }, { "epoch": 178.9421686746988, "grad_norm": 12.885191917419434, "learning_rate": 1.041747572815534e-05, "loss": 0.5665, "step": 18454 }, { "epoch": 178.95180722891567, "grad_norm": 14.78382682800293, "learning_rate": 1.0412621359223302e-05, "loss": 0.382, "step": 18455 }, { "epoch": 178.96144578313252, "grad_norm": 41.204734802246094, "learning_rate": 1.0407766990291263e-05, "loss": 0.2668, "step": 18456 }, { "epoch": 178.9710843373494, "grad_norm": 12.734891891479492, "learning_rate": 1.0402912621359222e-05, "loss": 0.3121, "step": 18457 }, { "epoch": 178.98072289156627, "grad_norm": 2.7015938758850098, "learning_rate": 1.0398058252427185e-05, "loss": 0.0984, "step": 18458 }, { "epoch": 178.99036144578312, "grad_norm": 42.60439682006836, "learning_rate": 1.0393203883495146e-05, "loss": 0.1151, "step": 18459 }, { "epoch": 179.00602409638554, "grad_norm": 61.48369216918945, "learning_rate": 1.0388349514563107e-05, "loss": 0.3674, "step": 18460 }, { "epoch": 179.01566265060242, "grad_norm": 12.775084495544434, "learning_rate": 1.0383495145631069e-05, "loss": 0.3948, "step": 18461 }, { "epoch": 179.02530120481927, "grad_norm": 10.72099494934082, "learning_rate": 1.037864077669903e-05, "loss": 0.1886, "step": 18462 }, { "epoch": 179.03493975903615, "grad_norm": 3.5150978565216064, "learning_rate": 1.0373786407766991e-05, "loss": 0.0739, "step": 18463 }, { "epoch": 179.044578313253, "grad_norm": 12.239623069763184, "learning_rate": 1.0368932038834952e-05, "loss": 0.2161, "step": 18464 }, { "epoch": 179.05421686746988, "grad_norm": 26.3699893951416, "learning_rate": 1.0364077669902913e-05, "loss": 0.2024, "step": 18465 }, { "epoch": 179.06385542168675, "grad_norm": 36.954193115234375, "learning_rate": 1.0359223300970874e-05, "loss": 0.1599, "step": 18466 }, { "epoch": 179.0734939759036, "grad_norm": 5.008548736572266, "learning_rate": 1.0354368932038835e-05, "loss": 0.2224, "step": 18467 }, { "epoch": 179.08313253012048, "grad_norm": 26.80190086364746, "learning_rate": 1.0349514563106797e-05, "loss": 0.3109, "step": 18468 }, { "epoch": 179.09277108433736, "grad_norm": 20.11642074584961, "learning_rate": 1.0344660194174758e-05, "loss": 0.1205, "step": 18469 }, { "epoch": 179.1024096385542, "grad_norm": 21.918325424194336, "learning_rate": 1.0339805825242719e-05, "loss": 0.2136, "step": 18470 }, { "epoch": 179.11204819277108, "grad_norm": 9.258039474487305, "learning_rate": 1.033495145631068e-05, "loss": 0.22, "step": 18471 }, { "epoch": 179.12168674698796, "grad_norm": 51.64160919189453, "learning_rate": 1.0330097087378641e-05, "loss": 0.2315, "step": 18472 }, { "epoch": 179.1313253012048, "grad_norm": 3.8586723804473877, "learning_rate": 1.0325242718446602e-05, "loss": 0.0734, "step": 18473 }, { "epoch": 179.1409638554217, "grad_norm": 9.69964599609375, "learning_rate": 1.0320388349514564e-05, "loss": 0.1375, "step": 18474 }, { "epoch": 179.15060240963857, "grad_norm": 63.2308235168457, "learning_rate": 1.0315533980582525e-05, "loss": 0.1835, "step": 18475 }, { "epoch": 179.16024096385541, "grad_norm": 10.234274864196777, "learning_rate": 1.0310679611650486e-05, "loss": 0.1175, "step": 18476 }, { "epoch": 179.1698795180723, "grad_norm": 10.687715530395508, "learning_rate": 1.0305825242718447e-05, "loss": 0.1375, "step": 18477 }, { "epoch": 179.17951807228917, "grad_norm": 18.36526107788086, "learning_rate": 1.0300970873786408e-05, "loss": 0.3441, "step": 18478 }, { "epoch": 179.18915662650602, "grad_norm": 7.934099197387695, "learning_rate": 1.029611650485437e-05, "loss": 0.2781, "step": 18479 }, { "epoch": 179.1987951807229, "grad_norm": 28.25807762145996, "learning_rate": 1.029126213592233e-05, "loss": 0.32, "step": 18480 }, { "epoch": 179.20843373493975, "grad_norm": 16.556930541992188, "learning_rate": 1.0286407766990292e-05, "loss": 0.2546, "step": 18481 }, { "epoch": 179.21807228915662, "grad_norm": 21.702728271484375, "learning_rate": 1.0281553398058253e-05, "loss": 0.2814, "step": 18482 }, { "epoch": 179.2277108433735, "grad_norm": 2.3549795150756836, "learning_rate": 1.0276699029126216e-05, "loss": 0.1559, "step": 18483 }, { "epoch": 179.23734939759035, "grad_norm": 36.61272430419922, "learning_rate": 1.0271844660194175e-05, "loss": 0.1476, "step": 18484 }, { "epoch": 179.24698795180723, "grad_norm": 5.073049068450928, "learning_rate": 1.0266990291262136e-05, "loss": 0.166, "step": 18485 }, { "epoch": 179.2566265060241, "grad_norm": 20.507091522216797, "learning_rate": 1.0262135922330097e-05, "loss": 0.191, "step": 18486 }, { "epoch": 179.26626506024095, "grad_norm": 13.64787769317627, "learning_rate": 1.0257281553398058e-05, "loss": 0.2391, "step": 18487 }, { "epoch": 179.27590361445783, "grad_norm": 10.668330192565918, "learning_rate": 1.025242718446602e-05, "loss": 0.2317, "step": 18488 }, { "epoch": 179.2855421686747, "grad_norm": 4.985360622406006, "learning_rate": 1.0247572815533982e-05, "loss": 0.1895, "step": 18489 }, { "epoch": 179.29518072289156, "grad_norm": 5.723756313323975, "learning_rate": 1.0242718446601944e-05, "loss": 0.218, "step": 18490 }, { "epoch": 179.30481927710844, "grad_norm": 36.12792205810547, "learning_rate": 1.0237864077669903e-05, "loss": 0.2047, "step": 18491 }, { "epoch": 179.3144578313253, "grad_norm": 18.454191207885742, "learning_rate": 1.0233009708737864e-05, "loss": 0.2228, "step": 18492 }, { "epoch": 179.32409638554216, "grad_norm": 9.606369972229004, "learning_rate": 1.0228155339805825e-05, "loss": 0.0823, "step": 18493 }, { "epoch": 179.33373493975904, "grad_norm": 7.434162616729736, "learning_rate": 1.0223300970873786e-05, "loss": 0.3203, "step": 18494 }, { "epoch": 179.34337349397592, "grad_norm": 10.270637512207031, "learning_rate": 1.0218446601941748e-05, "loss": 0.297, "step": 18495 }, { "epoch": 179.35301204819277, "grad_norm": 14.639063835144043, "learning_rate": 1.021359223300971e-05, "loss": 0.3083, "step": 18496 }, { "epoch": 179.36265060240964, "grad_norm": 6.47455358505249, "learning_rate": 1.020873786407767e-05, "loss": 0.231, "step": 18497 }, { "epoch": 179.3722891566265, "grad_norm": 5.596277236938477, "learning_rate": 1.0203883495145631e-05, "loss": 0.0992, "step": 18498 }, { "epoch": 179.38192771084337, "grad_norm": 36.70833969116211, "learning_rate": 1.0199029126213592e-05, "loss": 0.2589, "step": 18499 }, { "epoch": 179.39156626506025, "grad_norm": 27.37784194946289, "learning_rate": 1.0194174757281553e-05, "loss": 0.2976, "step": 18500 }, { "epoch": 179.4012048192771, "grad_norm": 6.199925899505615, "learning_rate": 1.0189320388349514e-05, "loss": 0.2377, "step": 18501 }, { "epoch": 179.41084337349398, "grad_norm": 14.034918785095215, "learning_rate": 1.0184466019417477e-05, "loss": 0.1623, "step": 18502 }, { "epoch": 179.42048192771085, "grad_norm": 7.701962471008301, "learning_rate": 1.0179611650485438e-05, "loss": 0.201, "step": 18503 }, { "epoch": 179.4301204819277, "grad_norm": 6.1380720138549805, "learning_rate": 1.0174757281553398e-05, "loss": 0.2806, "step": 18504 }, { "epoch": 179.43975903614458, "grad_norm": 9.743576049804688, "learning_rate": 1.0169902912621359e-05, "loss": 0.4587, "step": 18505 }, { "epoch": 179.44939759036146, "grad_norm": 6.643368244171143, "learning_rate": 1.016504854368932e-05, "loss": 0.2513, "step": 18506 }, { "epoch": 179.4590361445783, "grad_norm": 19.454967498779297, "learning_rate": 1.0160194174757281e-05, "loss": 0.2873, "step": 18507 }, { "epoch": 179.46867469879518, "grad_norm": 17.18379783630371, "learning_rate": 1.0155339805825244e-05, "loss": 0.3421, "step": 18508 }, { "epoch": 179.47831325301206, "grad_norm": 1.026760220527649, "learning_rate": 1.0150485436893205e-05, "loss": 0.1199, "step": 18509 }, { "epoch": 179.4879518072289, "grad_norm": 21.784730911254883, "learning_rate": 1.0145631067961166e-05, "loss": 0.2686, "step": 18510 }, { "epoch": 179.4975903614458, "grad_norm": 17.857776641845703, "learning_rate": 1.0140776699029126e-05, "loss": 0.31, "step": 18511 }, { "epoch": 179.50722891566264, "grad_norm": 17.19721221923828, "learning_rate": 1.0135922330097087e-05, "loss": 0.4146, "step": 18512 }, { "epoch": 179.51686746987951, "grad_norm": 24.383031845092773, "learning_rate": 1.0131067961165048e-05, "loss": 0.2083, "step": 18513 }, { "epoch": 179.5265060240964, "grad_norm": 24.79542350769043, "learning_rate": 1.012621359223301e-05, "loss": 0.3718, "step": 18514 }, { "epoch": 179.53614457831324, "grad_norm": 12.507889747619629, "learning_rate": 1.0121359223300972e-05, "loss": 0.1867, "step": 18515 }, { "epoch": 179.54578313253012, "grad_norm": 25.518003463745117, "learning_rate": 1.0116504854368933e-05, "loss": 0.4621, "step": 18516 }, { "epoch": 179.555421686747, "grad_norm": 8.030550956726074, "learning_rate": 1.0111650485436894e-05, "loss": 0.2581, "step": 18517 }, { "epoch": 179.56506024096385, "grad_norm": 7.681085586547852, "learning_rate": 1.0106796116504854e-05, "loss": 0.3346, "step": 18518 }, { "epoch": 179.57469879518072, "grad_norm": 9.353669166564941, "learning_rate": 1.0101941747572815e-05, "loss": 0.289, "step": 18519 }, { "epoch": 179.5843373493976, "grad_norm": 17.539093017578125, "learning_rate": 1.0097087378640776e-05, "loss": 0.19, "step": 18520 }, { "epoch": 179.59397590361445, "grad_norm": 179.07992553710938, "learning_rate": 1.0092233009708739e-05, "loss": 0.3657, "step": 18521 }, { "epoch": 179.60361445783133, "grad_norm": 19.83434295654297, "learning_rate": 1.00873786407767e-05, "loss": 0.2526, "step": 18522 }, { "epoch": 179.6132530120482, "grad_norm": 9.578575134277344, "learning_rate": 1.0082524271844661e-05, "loss": 0.0836, "step": 18523 }, { "epoch": 179.62289156626505, "grad_norm": 9.584442138671875, "learning_rate": 1.0077669902912622e-05, "loss": 0.2679, "step": 18524 }, { "epoch": 179.63253012048193, "grad_norm": 16.090116500854492, "learning_rate": 1.0072815533980582e-05, "loss": 0.3506, "step": 18525 }, { "epoch": 179.6421686746988, "grad_norm": 3.0198869705200195, "learning_rate": 1.0067961165048543e-05, "loss": 0.0938, "step": 18526 }, { "epoch": 179.65180722891566, "grad_norm": 11.190184593200684, "learning_rate": 1.0063106796116506e-05, "loss": 0.3447, "step": 18527 }, { "epoch": 179.66144578313254, "grad_norm": 9.864253044128418, "learning_rate": 1.0058252427184467e-05, "loss": 0.2847, "step": 18528 }, { "epoch": 179.67108433734938, "grad_norm": 26.771034240722656, "learning_rate": 1.0053398058252428e-05, "loss": 0.258, "step": 18529 }, { "epoch": 179.68072289156626, "grad_norm": 30.514402389526367, "learning_rate": 1.004854368932039e-05, "loss": 0.3342, "step": 18530 }, { "epoch": 179.69036144578314, "grad_norm": 9.6215181350708, "learning_rate": 1.004368932038835e-05, "loss": 0.1329, "step": 18531 }, { "epoch": 179.7, "grad_norm": 137.89649963378906, "learning_rate": 1.003883495145631e-05, "loss": 0.3166, "step": 18532 }, { "epoch": 179.70963855421687, "grad_norm": 21.3695125579834, "learning_rate": 1.0033980582524271e-05, "loss": 0.3664, "step": 18533 }, { "epoch": 179.71927710843374, "grad_norm": 33.13188552856445, "learning_rate": 1.0029126213592234e-05, "loss": 0.219, "step": 18534 }, { "epoch": 179.7289156626506, "grad_norm": 14.909712791442871, "learning_rate": 1.0024271844660195e-05, "loss": 0.3461, "step": 18535 }, { "epoch": 179.73855421686747, "grad_norm": 36.737648010253906, "learning_rate": 1.0019417475728156e-05, "loss": 0.3193, "step": 18536 }, { "epoch": 179.74819277108435, "grad_norm": 6.9280548095703125, "learning_rate": 1.0014563106796117e-05, "loss": 0.2319, "step": 18537 }, { "epoch": 179.7578313253012, "grad_norm": 11.245404243469238, "learning_rate": 1.0009708737864079e-05, "loss": 0.1533, "step": 18538 }, { "epoch": 179.76746987951807, "grad_norm": 5.999026775360107, "learning_rate": 1.0004854368932038e-05, "loss": 0.3203, "step": 18539 }, { "epoch": 179.77710843373495, "grad_norm": 13.249760627746582, "learning_rate": 1e-05, "loss": 0.2623, "step": 18540 }, { "epoch": 179.7867469879518, "grad_norm": 20.772382736206055, "learning_rate": 9.995145631067962e-06, "loss": 0.4793, "step": 18541 }, { "epoch": 179.79638554216868, "grad_norm": 26.30289077758789, "learning_rate": 9.990291262135923e-06, "loss": 0.1866, "step": 18542 }, { "epoch": 179.80602409638556, "grad_norm": 37.9716796875, "learning_rate": 9.985436893203884e-06, "loss": 0.1791, "step": 18543 }, { "epoch": 179.8156626506024, "grad_norm": 17.35796356201172, "learning_rate": 9.980582524271845e-06, "loss": 0.3121, "step": 18544 }, { "epoch": 179.82530120481928, "grad_norm": 19.72623062133789, "learning_rate": 9.975728155339807e-06, "loss": 0.3907, "step": 18545 }, { "epoch": 179.83493975903613, "grad_norm": 7.5821733474731445, "learning_rate": 9.970873786407768e-06, "loss": 0.1392, "step": 18546 }, { "epoch": 179.844578313253, "grad_norm": 21.82701301574707, "learning_rate": 9.966019417475729e-06, "loss": 0.4541, "step": 18547 }, { "epoch": 179.8542168674699, "grad_norm": 5.253597259521484, "learning_rate": 9.96116504854369e-06, "loss": 0.1513, "step": 18548 }, { "epoch": 179.86385542168674, "grad_norm": 52.890289306640625, "learning_rate": 9.956310679611651e-06, "loss": 0.1549, "step": 18549 }, { "epoch": 179.8734939759036, "grad_norm": 10.487251281738281, "learning_rate": 9.951456310679612e-06, "loss": 0.5159, "step": 18550 }, { "epoch": 179.8831325301205, "grad_norm": 6.793137073516846, "learning_rate": 9.946601941747573e-06, "loss": 0.2254, "step": 18551 }, { "epoch": 179.89277108433734, "grad_norm": 8.855109214782715, "learning_rate": 9.941747572815535e-06, "loss": 0.2658, "step": 18552 }, { "epoch": 179.90240963855422, "grad_norm": 8.5743408203125, "learning_rate": 9.936893203883496e-06, "loss": 0.2883, "step": 18553 }, { "epoch": 179.9120481927711, "grad_norm": 43.26372146606445, "learning_rate": 9.932038834951457e-06, "loss": 0.2534, "step": 18554 }, { "epoch": 179.92168674698794, "grad_norm": 9.523361206054688, "learning_rate": 9.927184466019418e-06, "loss": 0.2521, "step": 18555 }, { "epoch": 179.93132530120482, "grad_norm": 25.13005828857422, "learning_rate": 9.922330097087379e-06, "loss": 0.1604, "step": 18556 }, { "epoch": 179.9409638554217, "grad_norm": 6.194262504577637, "learning_rate": 9.91747572815534e-06, "loss": 0.2198, "step": 18557 }, { "epoch": 179.95060240963855, "grad_norm": 19.38936424255371, "learning_rate": 9.912621359223301e-06, "loss": 0.2584, "step": 18558 }, { "epoch": 179.96024096385543, "grad_norm": 9.192513465881348, "learning_rate": 9.907766990291263e-06, "loss": 0.2072, "step": 18559 }, { "epoch": 179.96987951807228, "grad_norm": 9.356246948242188, "learning_rate": 9.902912621359224e-06, "loss": 0.3376, "step": 18560 }, { "epoch": 179.97951807228915, "grad_norm": 2.7276687622070312, "learning_rate": 9.898058252427185e-06, "loss": 0.2266, "step": 18561 }, { "epoch": 179.98915662650603, "grad_norm": 2.8361318111419678, "learning_rate": 9.893203883495146e-06, "loss": 0.11, "step": 18562 }, { "epoch": 180.00481927710842, "grad_norm": 11.301438331604004, "learning_rate": 9.888349514563107e-06, "loss": 0.2242, "step": 18563 }, { "epoch": 180.0144578313253, "grad_norm": 24.558691024780273, "learning_rate": 9.883495145631068e-06, "loss": 0.2321, "step": 18564 }, { "epoch": 180.02409638554218, "grad_norm": 9.721349716186523, "learning_rate": 9.87864077669903e-06, "loss": 0.3209, "step": 18565 }, { "epoch": 180.03373493975903, "grad_norm": 30.828853607177734, "learning_rate": 9.87378640776699e-06, "loss": 0.2411, "step": 18566 }, { "epoch": 180.0433734939759, "grad_norm": 3.83026385307312, "learning_rate": 9.868932038834952e-06, "loss": 0.3387, "step": 18567 }, { "epoch": 180.05301204819278, "grad_norm": 15.900476455688477, "learning_rate": 9.864077669902913e-06, "loss": 0.3142, "step": 18568 }, { "epoch": 180.06265060240963, "grad_norm": 4.748703479766846, "learning_rate": 9.859223300970874e-06, "loss": 0.2215, "step": 18569 }, { "epoch": 180.0722891566265, "grad_norm": 20.878093719482422, "learning_rate": 9.854368932038835e-06, "loss": 0.1365, "step": 18570 }, { "epoch": 180.08192771084336, "grad_norm": 29.697490692138672, "learning_rate": 9.849514563106798e-06, "loss": 0.6198, "step": 18571 }, { "epoch": 180.09156626506024, "grad_norm": 18.754207611083984, "learning_rate": 9.844660194174757e-06, "loss": 0.3103, "step": 18572 }, { "epoch": 180.10120481927711, "grad_norm": 8.720255851745605, "learning_rate": 9.839805825242719e-06, "loss": 0.195, "step": 18573 }, { "epoch": 180.11084337349396, "grad_norm": 9.980570793151855, "learning_rate": 9.83495145631068e-06, "loss": 0.2704, "step": 18574 }, { "epoch": 180.12048192771084, "grad_norm": 19.19199562072754, "learning_rate": 9.830097087378641e-06, "loss": 0.2014, "step": 18575 }, { "epoch": 180.13012048192772, "grad_norm": 7.537036895751953, "learning_rate": 9.825242718446602e-06, "loss": 0.1189, "step": 18576 }, { "epoch": 180.13975903614457, "grad_norm": 8.562453269958496, "learning_rate": 9.820388349514563e-06, "loss": 0.3037, "step": 18577 }, { "epoch": 180.14939759036145, "grad_norm": 7.394400596618652, "learning_rate": 9.815533980582526e-06, "loss": 0.4204, "step": 18578 }, { "epoch": 180.15903614457832, "grad_norm": 17.03636360168457, "learning_rate": 9.810679611650486e-06, "loss": 0.3419, "step": 18579 }, { "epoch": 180.16867469879517, "grad_norm": 8.931117057800293, "learning_rate": 9.805825242718447e-06, "loss": 0.1734, "step": 18580 }, { "epoch": 180.17831325301205, "grad_norm": 22.797191619873047, "learning_rate": 9.800970873786408e-06, "loss": 0.1996, "step": 18581 }, { "epoch": 180.18795180722893, "grad_norm": 4.8445587158203125, "learning_rate": 9.796116504854369e-06, "loss": 0.0808, "step": 18582 }, { "epoch": 180.19759036144578, "grad_norm": 2.797454595565796, "learning_rate": 9.79126213592233e-06, "loss": 0.1247, "step": 18583 }, { "epoch": 180.20722891566265, "grad_norm": 4.597141742706299, "learning_rate": 9.786407766990293e-06, "loss": 0.1617, "step": 18584 }, { "epoch": 180.21686746987953, "grad_norm": 1.3204525709152222, "learning_rate": 9.781553398058254e-06, "loss": 0.2714, "step": 18585 }, { "epoch": 180.22650602409638, "grad_norm": 16.333839416503906, "learning_rate": 9.776699029126214e-06, "loss": 0.3459, "step": 18586 }, { "epoch": 180.23614457831326, "grad_norm": 3.5836849212646484, "learning_rate": 9.771844660194175e-06, "loss": 0.2144, "step": 18587 }, { "epoch": 180.2457831325301, "grad_norm": 15.958367347717285, "learning_rate": 9.766990291262136e-06, "loss": 0.2237, "step": 18588 }, { "epoch": 180.25542168674698, "grad_norm": 8.769890785217285, "learning_rate": 9.762135922330097e-06, "loss": 0.3224, "step": 18589 }, { "epoch": 180.26506024096386, "grad_norm": 6.3523383140563965, "learning_rate": 9.75728155339806e-06, "loss": 0.1593, "step": 18590 }, { "epoch": 180.2746987951807, "grad_norm": 11.569320678710938, "learning_rate": 9.752427184466021e-06, "loss": 0.2558, "step": 18591 }, { "epoch": 180.2843373493976, "grad_norm": 24.478252410888672, "learning_rate": 9.747572815533982e-06, "loss": 0.2028, "step": 18592 }, { "epoch": 180.29397590361447, "grad_norm": 20.03166961669922, "learning_rate": 9.742718446601942e-06, "loss": 0.3402, "step": 18593 }, { "epoch": 180.30361445783132, "grad_norm": 15.301301002502441, "learning_rate": 9.737864077669903e-06, "loss": 0.2042, "step": 18594 }, { "epoch": 180.3132530120482, "grad_norm": 8.430931091308594, "learning_rate": 9.733009708737864e-06, "loss": 0.3154, "step": 18595 }, { "epoch": 180.32289156626507, "grad_norm": 15.503039360046387, "learning_rate": 9.728155339805825e-06, "loss": 0.2192, "step": 18596 }, { "epoch": 180.33253012048192, "grad_norm": 35.27892303466797, "learning_rate": 9.723300970873788e-06, "loss": 0.2887, "step": 18597 }, { "epoch": 180.3421686746988, "grad_norm": 6.444969177246094, "learning_rate": 9.718446601941749e-06, "loss": 0.1991, "step": 18598 }, { "epoch": 180.35180722891567, "grad_norm": 29.226394653320312, "learning_rate": 9.71359223300971e-06, "loss": 0.2958, "step": 18599 }, { "epoch": 180.36144578313252, "grad_norm": 61.1327018737793, "learning_rate": 9.70873786407767e-06, "loss": 0.2547, "step": 18600 }, { "epoch": 180.3710843373494, "grad_norm": 5.638563632965088, "learning_rate": 9.70388349514563e-06, "loss": 0.2386, "step": 18601 }, { "epoch": 180.38072289156628, "grad_norm": 13.590807914733887, "learning_rate": 9.699029126213592e-06, "loss": 0.3977, "step": 18602 }, { "epoch": 180.39036144578313, "grad_norm": 29.85239601135254, "learning_rate": 9.694174757281555e-06, "loss": 0.1931, "step": 18603 }, { "epoch": 180.4, "grad_norm": 13.382946014404297, "learning_rate": 9.689320388349516e-06, "loss": 0.3691, "step": 18604 }, { "epoch": 180.40963855421685, "grad_norm": 24.372013092041016, "learning_rate": 9.684466019417477e-06, "loss": 0.3201, "step": 18605 }, { "epoch": 180.41927710843373, "grad_norm": 29.18608856201172, "learning_rate": 9.679611650485436e-06, "loss": 0.2855, "step": 18606 }, { "epoch": 180.4289156626506, "grad_norm": 6.147380828857422, "learning_rate": 9.674757281553398e-06, "loss": 0.1664, "step": 18607 }, { "epoch": 180.43855421686746, "grad_norm": 8.349071502685547, "learning_rate": 9.669902912621359e-06, "loss": 0.2578, "step": 18608 }, { "epoch": 180.44819277108434, "grad_norm": 7.083560466766357, "learning_rate": 9.665048543689322e-06, "loss": 0.2942, "step": 18609 }, { "epoch": 180.4578313253012, "grad_norm": 17.72197151184082, "learning_rate": 9.660194174757283e-06, "loss": 0.409, "step": 18610 }, { "epoch": 180.46746987951806, "grad_norm": 9.684890747070312, "learning_rate": 9.655339805825244e-06, "loss": 0.2303, "step": 18611 }, { "epoch": 180.47710843373494, "grad_norm": 17.123456954956055, "learning_rate": 9.650485436893205e-06, "loss": 0.3505, "step": 18612 }, { "epoch": 180.48674698795182, "grad_norm": 30.910009384155273, "learning_rate": 9.645631067961164e-06, "loss": 0.2196, "step": 18613 }, { "epoch": 180.49638554216867, "grad_norm": 14.265904426574707, "learning_rate": 9.640776699029126e-06, "loss": 0.2635, "step": 18614 }, { "epoch": 180.50602409638554, "grad_norm": 2.9064712524414062, "learning_rate": 9.635922330097087e-06, "loss": 0.0478, "step": 18615 }, { "epoch": 180.51566265060242, "grad_norm": 6.196063041687012, "learning_rate": 9.63106796116505e-06, "loss": 0.2643, "step": 18616 }, { "epoch": 180.52530120481927, "grad_norm": 4.872769832611084, "learning_rate": 9.62621359223301e-06, "loss": 0.2376, "step": 18617 }, { "epoch": 180.53493975903615, "grad_norm": 3.2126212120056152, "learning_rate": 9.621359223300972e-06, "loss": 0.236, "step": 18618 }, { "epoch": 180.544578313253, "grad_norm": 21.289583206176758, "learning_rate": 9.616504854368933e-06, "loss": 0.2225, "step": 18619 }, { "epoch": 180.55421686746988, "grad_norm": 7.933207035064697, "learning_rate": 9.611650485436892e-06, "loss": 0.1485, "step": 18620 }, { "epoch": 180.56385542168675, "grad_norm": 30.76276969909668, "learning_rate": 9.606796116504854e-06, "loss": 0.311, "step": 18621 }, { "epoch": 180.5734939759036, "grad_norm": 17.81125259399414, "learning_rate": 9.601941747572816e-06, "loss": 0.3085, "step": 18622 }, { "epoch": 180.58313253012048, "grad_norm": 2.167513608932495, "learning_rate": 9.597087378640778e-06, "loss": 0.2195, "step": 18623 }, { "epoch": 180.59277108433736, "grad_norm": 6.54464864730835, "learning_rate": 9.592233009708739e-06, "loss": 0.203, "step": 18624 }, { "epoch": 180.6024096385542, "grad_norm": 10.747062683105469, "learning_rate": 9.5873786407767e-06, "loss": 0.4474, "step": 18625 }, { "epoch": 180.61204819277108, "grad_norm": 6.562156677246094, "learning_rate": 9.582524271844661e-06, "loss": 0.1535, "step": 18626 }, { "epoch": 180.62168674698796, "grad_norm": 16.474464416503906, "learning_rate": 9.57766990291262e-06, "loss": 0.3534, "step": 18627 }, { "epoch": 180.6313253012048, "grad_norm": 8.35489559173584, "learning_rate": 9.572815533980583e-06, "loss": 0.2199, "step": 18628 }, { "epoch": 180.6409638554217, "grad_norm": 8.230011940002441, "learning_rate": 9.567961165048544e-06, "loss": 0.1158, "step": 18629 }, { "epoch": 180.65060240963857, "grad_norm": 17.313751220703125, "learning_rate": 9.563106796116506e-06, "loss": 0.2473, "step": 18630 }, { "epoch": 180.66024096385541, "grad_norm": 4.857816219329834, "learning_rate": 9.558252427184467e-06, "loss": 0.1573, "step": 18631 }, { "epoch": 180.6698795180723, "grad_norm": 19.009567260742188, "learning_rate": 9.553398058252428e-06, "loss": 0.3422, "step": 18632 }, { "epoch": 180.67951807228917, "grad_norm": 46.24049758911133, "learning_rate": 9.548543689320389e-06, "loss": 0.3609, "step": 18633 }, { "epoch": 180.68915662650602, "grad_norm": 26.021224975585938, "learning_rate": 9.543689320388349e-06, "loss": 0.3955, "step": 18634 }, { "epoch": 180.6987951807229, "grad_norm": 4.446841239929199, "learning_rate": 9.538834951456311e-06, "loss": 0.0696, "step": 18635 }, { "epoch": 180.70843373493975, "grad_norm": 6.540520191192627, "learning_rate": 9.533980582524273e-06, "loss": 0.16, "step": 18636 }, { "epoch": 180.71807228915662, "grad_norm": 11.892359733581543, "learning_rate": 9.529126213592234e-06, "loss": 0.2234, "step": 18637 }, { "epoch": 180.7277108433735, "grad_norm": 19.658798217773438, "learning_rate": 9.524271844660195e-06, "loss": 0.1197, "step": 18638 }, { "epoch": 180.73734939759035, "grad_norm": 13.753497123718262, "learning_rate": 9.519417475728156e-06, "loss": 0.1891, "step": 18639 }, { "epoch": 180.74698795180723, "grad_norm": 5.499739170074463, "learning_rate": 9.514563106796117e-06, "loss": 0.1691, "step": 18640 }, { "epoch": 180.7566265060241, "grad_norm": 4.976439952850342, "learning_rate": 9.509708737864078e-06, "loss": 0.3383, "step": 18641 }, { "epoch": 180.76626506024095, "grad_norm": 22.798372268676758, "learning_rate": 9.50485436893204e-06, "loss": 0.3203, "step": 18642 }, { "epoch": 180.77590361445783, "grad_norm": 9.446725845336914, "learning_rate": 9.5e-06, "loss": 0.1711, "step": 18643 }, { "epoch": 180.7855421686747, "grad_norm": 13.979097366333008, "learning_rate": 9.495145631067962e-06, "loss": 0.2248, "step": 18644 }, { "epoch": 180.79518072289156, "grad_norm": 13.354363441467285, "learning_rate": 9.490291262135923e-06, "loss": 0.1552, "step": 18645 }, { "epoch": 180.80481927710844, "grad_norm": 15.259182929992676, "learning_rate": 9.485436893203884e-06, "loss": 0.241, "step": 18646 }, { "epoch": 180.8144578313253, "grad_norm": 22.153501510620117, "learning_rate": 9.480582524271845e-06, "loss": 0.1679, "step": 18647 }, { "epoch": 180.82409638554216, "grad_norm": 22.276723861694336, "learning_rate": 9.475728155339806e-06, "loss": 0.2435, "step": 18648 }, { "epoch": 180.83373493975904, "grad_norm": 15.566329956054688, "learning_rate": 9.470873786407767e-06, "loss": 0.1735, "step": 18649 }, { "epoch": 180.84337349397592, "grad_norm": 45.63450241088867, "learning_rate": 9.466019417475729e-06, "loss": 0.3299, "step": 18650 }, { "epoch": 180.85301204819277, "grad_norm": 16.920284271240234, "learning_rate": 9.46116504854369e-06, "loss": 0.1683, "step": 18651 }, { "epoch": 180.86265060240964, "grad_norm": 14.438650131225586, "learning_rate": 9.45631067961165e-06, "loss": 0.3341, "step": 18652 }, { "epoch": 180.8722891566265, "grad_norm": 17.360565185546875, "learning_rate": 9.451456310679612e-06, "loss": 0.1956, "step": 18653 }, { "epoch": 180.88192771084337, "grad_norm": 21.07743263244629, "learning_rate": 9.446601941747573e-06, "loss": 0.3966, "step": 18654 }, { "epoch": 180.89156626506025, "grad_norm": 7.723860740661621, "learning_rate": 9.441747572815534e-06, "loss": 0.1533, "step": 18655 }, { "epoch": 180.9012048192771, "grad_norm": 20.157209396362305, "learning_rate": 9.436893203883495e-06, "loss": 0.3355, "step": 18656 }, { "epoch": 180.91084337349398, "grad_norm": 4.517000198364258, "learning_rate": 9.432038834951457e-06, "loss": 0.1846, "step": 18657 }, { "epoch": 180.92048192771085, "grad_norm": 13.184849739074707, "learning_rate": 9.427184466019418e-06, "loss": 0.1349, "step": 18658 }, { "epoch": 180.9301204819277, "grad_norm": 9.129998207092285, "learning_rate": 9.422330097087379e-06, "loss": 0.366, "step": 18659 }, { "epoch": 180.93975903614458, "grad_norm": 31.977643966674805, "learning_rate": 9.41747572815534e-06, "loss": 0.2258, "step": 18660 }, { "epoch": 180.94939759036146, "grad_norm": 39.9202766418457, "learning_rate": 9.412621359223301e-06, "loss": 0.1879, "step": 18661 }, { "epoch": 180.9590361445783, "grad_norm": 14.403388977050781, "learning_rate": 9.407766990291262e-06, "loss": 0.109, "step": 18662 }, { "epoch": 180.96867469879518, "grad_norm": 9.284538269042969, "learning_rate": 9.402912621359223e-06, "loss": 0.1618, "step": 18663 }, { "epoch": 180.97831325301206, "grad_norm": 29.280929565429688, "learning_rate": 9.398058252427185e-06, "loss": 0.3797, "step": 18664 }, { "epoch": 180.9879518072289, "grad_norm": 15.651211738586426, "learning_rate": 9.393203883495146e-06, "loss": 0.2092, "step": 18665 }, { "epoch": 181.00361445783133, "grad_norm": 7.172341823577881, "learning_rate": 9.388349514563109e-06, "loss": 0.1223, "step": 18666 }, { "epoch": 181.01325301204818, "grad_norm": 7.09367036819458, "learning_rate": 9.383495145631068e-06, "loss": 0.125, "step": 18667 }, { "epoch": 181.02289156626506, "grad_norm": 32.32891082763672, "learning_rate": 9.37864077669903e-06, "loss": 0.322, "step": 18668 }, { "epoch": 181.03253012048194, "grad_norm": 7.55044412612915, "learning_rate": 9.37378640776699e-06, "loss": 0.2816, "step": 18669 }, { "epoch": 181.0421686746988, "grad_norm": 20.378995895385742, "learning_rate": 9.368932038834951e-06, "loss": 0.2893, "step": 18670 }, { "epoch": 181.05180722891566, "grad_norm": 32.184547424316406, "learning_rate": 9.364077669902913e-06, "loss": 0.2275, "step": 18671 }, { "epoch": 181.06144578313254, "grad_norm": 11.857213020324707, "learning_rate": 9.359223300970875e-06, "loss": 0.2893, "step": 18672 }, { "epoch": 181.0710843373494, "grad_norm": 18.48870849609375, "learning_rate": 9.354368932038837e-06, "loss": 0.2632, "step": 18673 }, { "epoch": 181.08072289156627, "grad_norm": 11.090295791625977, "learning_rate": 9.349514563106796e-06, "loss": 0.249, "step": 18674 }, { "epoch": 181.09036144578315, "grad_norm": 46.58004379272461, "learning_rate": 9.344660194174757e-06, "loss": 0.5436, "step": 18675 }, { "epoch": 181.1, "grad_norm": 27.396854400634766, "learning_rate": 9.339805825242718e-06, "loss": 0.3186, "step": 18676 }, { "epoch": 181.10963855421687, "grad_norm": 9.649399757385254, "learning_rate": 9.33495145631068e-06, "loss": 0.1696, "step": 18677 }, { "epoch": 181.11927710843372, "grad_norm": 37.00663375854492, "learning_rate": 9.33009708737864e-06, "loss": 0.3111, "step": 18678 }, { "epoch": 181.1289156626506, "grad_norm": 5.895479679107666, "learning_rate": 9.325242718446603e-06, "loss": 0.1444, "step": 18679 }, { "epoch": 181.13855421686748, "grad_norm": 22.132829666137695, "learning_rate": 9.320388349514565e-06, "loss": 0.3509, "step": 18680 }, { "epoch": 181.14819277108433, "grad_norm": 5.925255298614502, "learning_rate": 9.315533980582524e-06, "loss": 0.1723, "step": 18681 }, { "epoch": 181.1578313253012, "grad_norm": 3.2909774780273438, "learning_rate": 9.310679611650485e-06, "loss": 0.1976, "step": 18682 }, { "epoch": 181.16746987951808, "grad_norm": 34.02693176269531, "learning_rate": 9.305825242718446e-06, "loss": 0.3069, "step": 18683 }, { "epoch": 181.17710843373493, "grad_norm": 11.16002368927002, "learning_rate": 9.300970873786407e-06, "loss": 0.4069, "step": 18684 }, { "epoch": 181.1867469879518, "grad_norm": 9.372673034667969, "learning_rate": 9.29611650485437e-06, "loss": 0.3143, "step": 18685 }, { "epoch": 181.19638554216868, "grad_norm": 19.535747528076172, "learning_rate": 9.291262135922331e-06, "loss": 0.1445, "step": 18686 }, { "epoch": 181.20602409638553, "grad_norm": 12.76167106628418, "learning_rate": 9.286407766990293e-06, "loss": 0.2098, "step": 18687 }, { "epoch": 181.2156626506024, "grad_norm": 9.299942970275879, "learning_rate": 9.281553398058252e-06, "loss": 0.3822, "step": 18688 }, { "epoch": 181.2253012048193, "grad_norm": 10.390141487121582, "learning_rate": 9.276699029126213e-06, "loss": 0.2116, "step": 18689 }, { "epoch": 181.23493975903614, "grad_norm": 13.514717102050781, "learning_rate": 9.271844660194174e-06, "loss": 0.3659, "step": 18690 }, { "epoch": 181.24457831325302, "grad_norm": 8.950489044189453, "learning_rate": 9.266990291262137e-06, "loss": 0.2968, "step": 18691 }, { "epoch": 181.2542168674699, "grad_norm": 8.334274291992188, "learning_rate": 9.262135922330098e-06, "loss": 0.1824, "step": 18692 }, { "epoch": 181.26385542168674, "grad_norm": 39.151851654052734, "learning_rate": 9.25728155339806e-06, "loss": 0.276, "step": 18693 }, { "epoch": 181.27349397590362, "grad_norm": 18.67176055908203, "learning_rate": 9.25242718446602e-06, "loss": 0.1658, "step": 18694 }, { "epoch": 181.28313253012047, "grad_norm": 9.94568157196045, "learning_rate": 9.24757281553398e-06, "loss": 0.4324, "step": 18695 }, { "epoch": 181.29277108433735, "grad_norm": 8.641440391540527, "learning_rate": 9.242718446601941e-06, "loss": 0.2705, "step": 18696 }, { "epoch": 181.30240963855422, "grad_norm": 9.438618659973145, "learning_rate": 9.237864077669902e-06, "loss": 0.169, "step": 18697 }, { "epoch": 181.31204819277107, "grad_norm": 19.182361602783203, "learning_rate": 9.233009708737865e-06, "loss": 0.3341, "step": 18698 }, { "epoch": 181.32168674698795, "grad_norm": 15.913389205932617, "learning_rate": 9.228155339805826e-06, "loss": 0.41, "step": 18699 }, { "epoch": 181.33132530120483, "grad_norm": 5.08009672164917, "learning_rate": 9.223300970873788e-06, "loss": 0.2326, "step": 18700 }, { "epoch": 181.34096385542168, "grad_norm": 3.8233542442321777, "learning_rate": 9.218446601941749e-06, "loss": 0.1208, "step": 18701 }, { "epoch": 181.35060240963855, "grad_norm": 21.297548294067383, "learning_rate": 9.213592233009708e-06, "loss": 0.3441, "step": 18702 }, { "epoch": 181.36024096385543, "grad_norm": 10.102442741394043, "learning_rate": 9.20873786407767e-06, "loss": 0.1877, "step": 18703 }, { "epoch": 181.36987951807228, "grad_norm": 6.9737749099731445, "learning_rate": 9.203883495145632e-06, "loss": 0.2341, "step": 18704 }, { "epoch": 181.37951807228916, "grad_norm": 22.539953231811523, "learning_rate": 9.199029126213593e-06, "loss": 0.3082, "step": 18705 }, { "epoch": 181.38915662650604, "grad_norm": 4.039760112762451, "learning_rate": 9.194174757281554e-06, "loss": 0.2785, "step": 18706 }, { "epoch": 181.39879518072289, "grad_norm": 7.869402885437012, "learning_rate": 9.189320388349516e-06, "loss": 0.139, "step": 18707 }, { "epoch": 181.40843373493976, "grad_norm": 5.918426036834717, "learning_rate": 9.184466019417477e-06, "loss": 0.2815, "step": 18708 }, { "epoch": 181.41807228915664, "grad_norm": 13.214360237121582, "learning_rate": 9.179611650485436e-06, "loss": 0.2365, "step": 18709 }, { "epoch": 181.4277108433735, "grad_norm": 3.2889339923858643, "learning_rate": 9.174757281553399e-06, "loss": 0.0324, "step": 18710 }, { "epoch": 181.43734939759037, "grad_norm": 21.845565795898438, "learning_rate": 9.16990291262136e-06, "loss": 0.2251, "step": 18711 }, { "epoch": 181.44698795180722, "grad_norm": 12.80783748626709, "learning_rate": 9.165048543689321e-06, "loss": 0.2752, "step": 18712 }, { "epoch": 181.4566265060241, "grad_norm": 13.262042999267578, "learning_rate": 9.160194174757282e-06, "loss": 0.2211, "step": 18713 }, { "epoch": 181.46626506024097, "grad_norm": 48.22141647338867, "learning_rate": 9.155339805825244e-06, "loss": 0.2807, "step": 18714 }, { "epoch": 181.47590361445782, "grad_norm": 8.072985649108887, "learning_rate": 9.150485436893205e-06, "loss": 0.1878, "step": 18715 }, { "epoch": 181.4855421686747, "grad_norm": 10.990462303161621, "learning_rate": 9.145631067961164e-06, "loss": 0.1923, "step": 18716 }, { "epoch": 181.49518072289158, "grad_norm": 5.896963596343994, "learning_rate": 9.140776699029127e-06, "loss": 0.1052, "step": 18717 }, { "epoch": 181.50481927710842, "grad_norm": 61.795616149902344, "learning_rate": 9.135922330097088e-06, "loss": 0.5502, "step": 18718 }, { "epoch": 181.5144578313253, "grad_norm": 44.35731506347656, "learning_rate": 9.13106796116505e-06, "loss": 0.3938, "step": 18719 }, { "epoch": 181.52409638554218, "grad_norm": 24.124277114868164, "learning_rate": 9.12621359223301e-06, "loss": 0.1399, "step": 18720 }, { "epoch": 181.53373493975903, "grad_norm": 13.407713890075684, "learning_rate": 9.121359223300972e-06, "loss": 0.286, "step": 18721 }, { "epoch": 181.5433734939759, "grad_norm": 11.03622817993164, "learning_rate": 9.116504854368931e-06, "loss": 0.2534, "step": 18722 }, { "epoch": 181.55301204819278, "grad_norm": 21.924039840698242, "learning_rate": 9.111650485436894e-06, "loss": 0.4451, "step": 18723 }, { "epoch": 181.56265060240963, "grad_norm": 4.8570356369018555, "learning_rate": 9.106796116504855e-06, "loss": 0.2026, "step": 18724 }, { "epoch": 181.5722891566265, "grad_norm": 6.232463359832764, "learning_rate": 9.101941747572816e-06, "loss": 0.3149, "step": 18725 }, { "epoch": 181.58192771084336, "grad_norm": 13.3086519241333, "learning_rate": 9.097087378640777e-06, "loss": 0.2505, "step": 18726 }, { "epoch": 181.59156626506024, "grad_norm": 6.593478202819824, "learning_rate": 9.092233009708738e-06, "loss": 0.1574, "step": 18727 }, { "epoch": 181.60120481927711, "grad_norm": 11.886857032775879, "learning_rate": 9.0873786407767e-06, "loss": 0.2477, "step": 18728 }, { "epoch": 181.61084337349396, "grad_norm": 7.931495666503906, "learning_rate": 9.08252427184466e-06, "loss": 0.1853, "step": 18729 }, { "epoch": 181.62048192771084, "grad_norm": 10.930299758911133, "learning_rate": 9.077669902912622e-06, "loss": 0.3451, "step": 18730 }, { "epoch": 181.63012048192772, "grad_norm": 6.215481281280518, "learning_rate": 9.072815533980583e-06, "loss": 0.2327, "step": 18731 }, { "epoch": 181.63975903614457, "grad_norm": 8.177528381347656, "learning_rate": 9.067961165048544e-06, "loss": 0.2499, "step": 18732 }, { "epoch": 181.64939759036145, "grad_norm": 9.31049633026123, "learning_rate": 9.063106796116505e-06, "loss": 0.3762, "step": 18733 }, { "epoch": 181.65903614457832, "grad_norm": 181.23388671875, "learning_rate": 9.058252427184466e-06, "loss": 0.2956, "step": 18734 }, { "epoch": 181.66867469879517, "grad_norm": 22.78037452697754, "learning_rate": 9.053398058252428e-06, "loss": 0.3643, "step": 18735 }, { "epoch": 181.67831325301205, "grad_norm": 11.152669906616211, "learning_rate": 9.048543689320389e-06, "loss": 0.2323, "step": 18736 }, { "epoch": 181.68795180722893, "grad_norm": 4.477017879486084, "learning_rate": 9.04368932038835e-06, "loss": 0.1097, "step": 18737 }, { "epoch": 181.69759036144578, "grad_norm": 8.708544731140137, "learning_rate": 9.038834951456311e-06, "loss": 0.244, "step": 18738 }, { "epoch": 181.70722891566265, "grad_norm": 2.30368971824646, "learning_rate": 9.033980582524272e-06, "loss": 0.0636, "step": 18739 }, { "epoch": 181.71686746987953, "grad_norm": 20.966489791870117, "learning_rate": 9.029126213592233e-06, "loss": 0.3826, "step": 18740 }, { "epoch": 181.72650602409638, "grad_norm": 11.887116432189941, "learning_rate": 9.024271844660194e-06, "loss": 0.3169, "step": 18741 }, { "epoch": 181.73614457831326, "grad_norm": 40.26298522949219, "learning_rate": 9.019417475728156e-06, "loss": 0.3165, "step": 18742 }, { "epoch": 181.7457831325301, "grad_norm": 13.041234970092773, "learning_rate": 9.014563106796117e-06, "loss": 0.1499, "step": 18743 }, { "epoch": 181.75542168674698, "grad_norm": 8.476872444152832, "learning_rate": 9.009708737864078e-06, "loss": 0.2648, "step": 18744 }, { "epoch": 181.76506024096386, "grad_norm": 25.465463638305664, "learning_rate": 9.004854368932039e-06, "loss": 0.3338, "step": 18745 }, { "epoch": 181.7746987951807, "grad_norm": 21.611835479736328, "learning_rate": 9e-06, "loss": 0.3565, "step": 18746 }, { "epoch": 181.7843373493976, "grad_norm": 8.356358528137207, "learning_rate": 8.995145631067961e-06, "loss": 0.159, "step": 18747 }, { "epoch": 181.79397590361447, "grad_norm": 17.54587173461914, "learning_rate": 8.990291262135924e-06, "loss": 0.3764, "step": 18748 }, { "epoch": 181.80361445783132, "grad_norm": 19.634380340576172, "learning_rate": 8.985436893203884e-06, "loss": 0.1635, "step": 18749 }, { "epoch": 181.8132530120482, "grad_norm": 12.597139358520508, "learning_rate": 8.980582524271845e-06, "loss": 0.2283, "step": 18750 }, { "epoch": 181.82289156626507, "grad_norm": 3.715416193008423, "learning_rate": 8.975728155339806e-06, "loss": 0.1213, "step": 18751 }, { "epoch": 181.83253012048192, "grad_norm": 39.051509857177734, "learning_rate": 8.970873786407767e-06, "loss": 0.318, "step": 18752 }, { "epoch": 181.8421686746988, "grad_norm": 5.907983303070068, "learning_rate": 8.966019417475728e-06, "loss": 0.2112, "step": 18753 }, { "epoch": 181.85180722891567, "grad_norm": 12.323227882385254, "learning_rate": 8.961165048543691e-06, "loss": 0.3008, "step": 18754 }, { "epoch": 181.86144578313252, "grad_norm": 9.280801773071289, "learning_rate": 8.95631067961165e-06, "loss": 0.3065, "step": 18755 }, { "epoch": 181.8710843373494, "grad_norm": 12.616058349609375, "learning_rate": 8.951456310679612e-06, "loss": 0.3228, "step": 18756 }, { "epoch": 181.88072289156628, "grad_norm": 22.05304718017578, "learning_rate": 8.946601941747573e-06, "loss": 0.2013, "step": 18757 }, { "epoch": 181.89036144578313, "grad_norm": 15.332228660583496, "learning_rate": 8.941747572815534e-06, "loss": 0.3014, "step": 18758 }, { "epoch": 181.9, "grad_norm": 7.319157600402832, "learning_rate": 8.936893203883495e-06, "loss": 0.3784, "step": 18759 }, { "epoch": 181.90963855421685, "grad_norm": 12.614095687866211, "learning_rate": 8.932038834951456e-06, "loss": 0.3472, "step": 18760 }, { "epoch": 181.91927710843373, "grad_norm": 23.278451919555664, "learning_rate": 8.927184466019419e-06, "loss": 0.2947, "step": 18761 }, { "epoch": 181.9289156626506, "grad_norm": 34.08592224121094, "learning_rate": 8.922330097087379e-06, "loss": 0.2621, "step": 18762 }, { "epoch": 181.93855421686746, "grad_norm": 22.56415367126465, "learning_rate": 8.91747572815534e-06, "loss": 0.3668, "step": 18763 }, { "epoch": 181.94819277108434, "grad_norm": 6.631275653839111, "learning_rate": 8.912621359223301e-06, "loss": 0.4191, "step": 18764 }, { "epoch": 181.9578313253012, "grad_norm": 9.98651123046875, "learning_rate": 8.907766990291262e-06, "loss": 0.433, "step": 18765 }, { "epoch": 181.96746987951806, "grad_norm": 4.261731147766113, "learning_rate": 8.902912621359223e-06, "loss": 0.2644, "step": 18766 }, { "epoch": 181.97710843373494, "grad_norm": 4.898043632507324, "learning_rate": 8.898058252427186e-06, "loss": 0.2547, "step": 18767 }, { "epoch": 181.98674698795182, "grad_norm": 13.19605541229248, "learning_rate": 8.893203883495147e-06, "loss": 0.3625, "step": 18768 }, { "epoch": 182.0024096385542, "grad_norm": 10.273082733154297, "learning_rate": 8.888349514563107e-06, "loss": 0.2475, "step": 18769 }, { "epoch": 182.0120481927711, "grad_norm": 11.459214210510254, "learning_rate": 8.883495145631068e-06, "loss": 0.224, "step": 18770 }, { "epoch": 182.02168674698794, "grad_norm": 9.503101348876953, "learning_rate": 8.878640776699029e-06, "loss": 0.2588, "step": 18771 }, { "epoch": 182.03132530120482, "grad_norm": 8.322175025939941, "learning_rate": 8.87378640776699e-06, "loss": 0.1981, "step": 18772 }, { "epoch": 182.0409638554217, "grad_norm": 6.308482646942139, "learning_rate": 8.868932038834953e-06, "loss": 0.2067, "step": 18773 }, { "epoch": 182.05060240963854, "grad_norm": 27.589447021484375, "learning_rate": 8.864077669902914e-06, "loss": 0.2886, "step": 18774 }, { "epoch": 182.06024096385542, "grad_norm": 15.746674537658691, "learning_rate": 8.859223300970875e-06, "loss": 0.2506, "step": 18775 }, { "epoch": 182.0698795180723, "grad_norm": 17.70029067993164, "learning_rate": 8.854368932038835e-06, "loss": 0.3414, "step": 18776 }, { "epoch": 182.07951807228915, "grad_norm": 4.896849155426025, "learning_rate": 8.849514563106796e-06, "loss": 0.142, "step": 18777 }, { "epoch": 182.08915662650602, "grad_norm": 15.030107498168945, "learning_rate": 8.844660194174757e-06, "loss": 0.4235, "step": 18778 }, { "epoch": 182.0987951807229, "grad_norm": 12.959471702575684, "learning_rate": 8.839805825242718e-06, "loss": 0.1677, "step": 18779 }, { "epoch": 182.10843373493975, "grad_norm": 14.493712425231934, "learning_rate": 8.834951456310681e-06, "loss": 0.2399, "step": 18780 }, { "epoch": 182.11807228915663, "grad_norm": 5.180410861968994, "learning_rate": 8.830097087378642e-06, "loss": 0.3522, "step": 18781 }, { "epoch": 182.1277108433735, "grad_norm": 9.282764434814453, "learning_rate": 8.825242718446603e-06, "loss": 0.1874, "step": 18782 }, { "epoch": 182.13734939759036, "grad_norm": 10.132699966430664, "learning_rate": 8.820388349514563e-06, "loss": 0.3471, "step": 18783 }, { "epoch": 182.14698795180723, "grad_norm": 14.991299629211426, "learning_rate": 8.815533980582524e-06, "loss": 0.1815, "step": 18784 }, { "epoch": 182.15662650602408, "grad_norm": 19.297714233398438, "learning_rate": 8.810679611650485e-06, "loss": 0.3326, "step": 18785 }, { "epoch": 182.16626506024096, "grad_norm": 9.293869972229004, "learning_rate": 8.805825242718448e-06, "loss": 0.2441, "step": 18786 }, { "epoch": 182.17590361445784, "grad_norm": 10.39515495300293, "learning_rate": 8.800970873786409e-06, "loss": 0.3986, "step": 18787 }, { "epoch": 182.1855421686747, "grad_norm": 7.060579776763916, "learning_rate": 8.79611650485437e-06, "loss": 0.253, "step": 18788 }, { "epoch": 182.19518072289156, "grad_norm": 8.299264907836914, "learning_rate": 8.791262135922331e-06, "loss": 0.2105, "step": 18789 }, { "epoch": 182.20481927710844, "grad_norm": 9.364713668823242, "learning_rate": 8.78640776699029e-06, "loss": 0.2355, "step": 18790 }, { "epoch": 182.2144578313253, "grad_norm": 6.481810092926025, "learning_rate": 8.781553398058252e-06, "loss": 0.1345, "step": 18791 }, { "epoch": 182.22409638554217, "grad_norm": 19.617183685302734, "learning_rate": 8.776699029126215e-06, "loss": 0.4292, "step": 18792 }, { "epoch": 182.23373493975905, "grad_norm": 8.276909828186035, "learning_rate": 8.771844660194176e-06, "loss": 0.1896, "step": 18793 }, { "epoch": 182.2433734939759, "grad_norm": 6.250236511230469, "learning_rate": 8.766990291262137e-06, "loss": 0.2189, "step": 18794 }, { "epoch": 182.25301204819277, "grad_norm": 8.118412017822266, "learning_rate": 8.762135922330098e-06, "loss": 0.2845, "step": 18795 }, { "epoch": 182.26265060240965, "grad_norm": 11.449546813964844, "learning_rate": 8.75728155339806e-06, "loss": 0.2245, "step": 18796 }, { "epoch": 182.2722891566265, "grad_norm": 12.924400329589844, "learning_rate": 8.752427184466019e-06, "loss": 0.3387, "step": 18797 }, { "epoch": 182.28192771084338, "grad_norm": 4.022368431091309, "learning_rate": 8.74757281553398e-06, "loss": 0.1178, "step": 18798 }, { "epoch": 182.29156626506025, "grad_norm": 9.646177291870117, "learning_rate": 8.742718446601943e-06, "loss": 0.1522, "step": 18799 }, { "epoch": 182.3012048192771, "grad_norm": 13.265027046203613, "learning_rate": 8.737864077669904e-06, "loss": 0.2252, "step": 18800 }, { "epoch": 182.31084337349398, "grad_norm": 6.299848556518555, "learning_rate": 8.733009708737865e-06, "loss": 0.1548, "step": 18801 }, { "epoch": 182.32048192771083, "grad_norm": 31.75343132019043, "learning_rate": 8.728155339805826e-06, "loss": 0.1851, "step": 18802 }, { "epoch": 182.3301204819277, "grad_norm": 10.950489044189453, "learning_rate": 8.723300970873787e-06, "loss": 0.1826, "step": 18803 }, { "epoch": 182.33975903614459, "grad_norm": 11.211990356445312, "learning_rate": 8.718446601941747e-06, "loss": 0.3008, "step": 18804 }, { "epoch": 182.34939759036143, "grad_norm": 16.68915557861328, "learning_rate": 8.71359223300971e-06, "loss": 0.2346, "step": 18805 }, { "epoch": 182.3590361445783, "grad_norm": 9.51127815246582, "learning_rate": 8.70873786407767e-06, "loss": 0.1849, "step": 18806 }, { "epoch": 182.3686746987952, "grad_norm": 8.51069164276123, "learning_rate": 8.703883495145632e-06, "loss": 0.2571, "step": 18807 }, { "epoch": 182.37831325301204, "grad_norm": 13.4780912399292, "learning_rate": 8.699029126213593e-06, "loss": 0.229, "step": 18808 }, { "epoch": 182.38795180722892, "grad_norm": 6.59748649597168, "learning_rate": 8.694174757281554e-06, "loss": 0.1507, "step": 18809 }, { "epoch": 182.3975903614458, "grad_norm": 7.941317081451416, "learning_rate": 8.689320388349515e-06, "loss": 0.2594, "step": 18810 }, { "epoch": 182.40722891566264, "grad_norm": 9.91206169128418, "learning_rate": 8.684466019417476e-06, "loss": 0.1188, "step": 18811 }, { "epoch": 182.41686746987952, "grad_norm": 4.056039810180664, "learning_rate": 8.679611650485438e-06, "loss": 0.1482, "step": 18812 }, { "epoch": 182.4265060240964, "grad_norm": 8.327935218811035, "learning_rate": 8.674757281553399e-06, "loss": 0.1915, "step": 18813 }, { "epoch": 182.43614457831325, "grad_norm": 4.2293572425842285, "learning_rate": 8.66990291262136e-06, "loss": 0.1285, "step": 18814 }, { "epoch": 182.44578313253012, "grad_norm": 23.791711807250977, "learning_rate": 8.665048543689321e-06, "loss": 0.4154, "step": 18815 }, { "epoch": 182.455421686747, "grad_norm": 15.210363388061523, "learning_rate": 8.660194174757282e-06, "loss": 0.323, "step": 18816 }, { "epoch": 182.46506024096385, "grad_norm": 14.705950736999512, "learning_rate": 8.655339805825243e-06, "loss": 0.2243, "step": 18817 }, { "epoch": 182.47469879518073, "grad_norm": 4.658071041107178, "learning_rate": 8.650485436893204e-06, "loss": 0.1571, "step": 18818 }, { "epoch": 182.48433734939758, "grad_norm": 9.75888729095459, "learning_rate": 8.645631067961166e-06, "loss": 0.2054, "step": 18819 }, { "epoch": 182.49397590361446, "grad_norm": 15.445496559143066, "learning_rate": 8.640776699029127e-06, "loss": 0.2118, "step": 18820 }, { "epoch": 182.50361445783133, "grad_norm": 7.83036470413208, "learning_rate": 8.635922330097088e-06, "loss": 0.1528, "step": 18821 }, { "epoch": 182.51325301204818, "grad_norm": 13.821105003356934, "learning_rate": 8.631067961165049e-06, "loss": 0.2519, "step": 18822 }, { "epoch": 182.52289156626506, "grad_norm": 5.627357482910156, "learning_rate": 8.62621359223301e-06, "loss": 0.089, "step": 18823 }, { "epoch": 182.53253012048194, "grad_norm": 7.257628440856934, "learning_rate": 8.621359223300971e-06, "loss": 0.0901, "step": 18824 }, { "epoch": 182.5421686746988, "grad_norm": 7.591935634613037, "learning_rate": 8.616504854368932e-06, "loss": 0.3494, "step": 18825 }, { "epoch": 182.55180722891566, "grad_norm": 15.995912551879883, "learning_rate": 8.611650485436894e-06, "loss": 0.3642, "step": 18826 }, { "epoch": 182.56144578313254, "grad_norm": 4.186905384063721, "learning_rate": 8.606796116504855e-06, "loss": 0.2464, "step": 18827 }, { "epoch": 182.5710843373494, "grad_norm": 11.05070686340332, "learning_rate": 8.601941747572816e-06, "loss": 0.3762, "step": 18828 }, { "epoch": 182.58072289156627, "grad_norm": 16.136993408203125, "learning_rate": 8.597087378640777e-06, "loss": 0.2101, "step": 18829 }, { "epoch": 182.59036144578315, "grad_norm": 6.790390968322754, "learning_rate": 8.592233009708738e-06, "loss": 0.1826, "step": 18830 }, { "epoch": 182.6, "grad_norm": 5.521005630493164, "learning_rate": 8.5873786407767e-06, "loss": 0.389, "step": 18831 }, { "epoch": 182.60963855421687, "grad_norm": 8.827617645263672, "learning_rate": 8.58252427184466e-06, "loss": 0.228, "step": 18832 }, { "epoch": 182.61927710843372, "grad_norm": 16.186962127685547, "learning_rate": 8.577669902912622e-06, "loss": 0.4381, "step": 18833 }, { "epoch": 182.6289156626506, "grad_norm": 6.8486504554748535, "learning_rate": 8.572815533980583e-06, "loss": 0.1567, "step": 18834 }, { "epoch": 182.63855421686748, "grad_norm": 8.905655860900879, "learning_rate": 8.567961165048544e-06, "loss": 0.2403, "step": 18835 }, { "epoch": 182.64819277108433, "grad_norm": 6.407862186431885, "learning_rate": 8.563106796116507e-06, "loss": 0.1832, "step": 18836 }, { "epoch": 182.6578313253012, "grad_norm": 3.4369540214538574, "learning_rate": 8.558252427184466e-06, "loss": 0.1231, "step": 18837 }, { "epoch": 182.66746987951808, "grad_norm": 10.602879524230957, "learning_rate": 8.553398058252427e-06, "loss": 0.2105, "step": 18838 }, { "epoch": 182.67710843373493, "grad_norm": 26.272647857666016, "learning_rate": 8.548543689320388e-06, "loss": 0.3964, "step": 18839 }, { "epoch": 182.6867469879518, "grad_norm": 5.213664531707764, "learning_rate": 8.54368932038835e-06, "loss": 0.2413, "step": 18840 }, { "epoch": 182.69638554216868, "grad_norm": 12.547745704650879, "learning_rate": 8.53883495145631e-06, "loss": 0.2, "step": 18841 }, { "epoch": 182.70602409638553, "grad_norm": 10.23331356048584, "learning_rate": 8.533980582524272e-06, "loss": 0.2983, "step": 18842 }, { "epoch": 182.7156626506024, "grad_norm": 13.134578704833984, "learning_rate": 8.529126213592235e-06, "loss": 0.3122, "step": 18843 }, { "epoch": 182.7253012048193, "grad_norm": 5.864835262298584, "learning_rate": 8.524271844660194e-06, "loss": 0.1995, "step": 18844 }, { "epoch": 182.73493975903614, "grad_norm": 5.579058647155762, "learning_rate": 8.519417475728155e-06, "loss": 0.1325, "step": 18845 }, { "epoch": 182.74457831325302, "grad_norm": 12.561653137207031, "learning_rate": 8.514563106796116e-06, "loss": 0.3794, "step": 18846 }, { "epoch": 182.7542168674699, "grad_norm": 12.12475299835205, "learning_rate": 8.509708737864078e-06, "loss": 0.1743, "step": 18847 }, { "epoch": 182.76385542168674, "grad_norm": 9.545208930969238, "learning_rate": 8.504854368932039e-06, "loss": 0.1657, "step": 18848 }, { "epoch": 182.77349397590362, "grad_norm": 9.502201080322266, "learning_rate": 8.500000000000002e-06, "loss": 0.1332, "step": 18849 }, { "epoch": 182.78313253012047, "grad_norm": 3.4073307514190674, "learning_rate": 8.495145631067963e-06, "loss": 0.0599, "step": 18850 }, { "epoch": 182.79277108433735, "grad_norm": 23.511268615722656, "learning_rate": 8.490291262135922e-06, "loss": 0.1747, "step": 18851 }, { "epoch": 182.80240963855422, "grad_norm": 8.634635925292969, "learning_rate": 8.485436893203883e-06, "loss": 0.2191, "step": 18852 }, { "epoch": 182.81204819277107, "grad_norm": 25.634748458862305, "learning_rate": 8.480582524271845e-06, "loss": 0.154, "step": 18853 }, { "epoch": 182.82168674698795, "grad_norm": 25.678504943847656, "learning_rate": 8.475728155339806e-06, "loss": 0.299, "step": 18854 }, { "epoch": 182.83132530120483, "grad_norm": 6.607341766357422, "learning_rate": 8.470873786407769e-06, "loss": 0.2809, "step": 18855 }, { "epoch": 182.84096385542168, "grad_norm": 8.34634780883789, "learning_rate": 8.46601941747573e-06, "loss": 0.1717, "step": 18856 }, { "epoch": 182.85060240963855, "grad_norm": 7.9375691413879395, "learning_rate": 8.46116504854369e-06, "loss": 0.1524, "step": 18857 }, { "epoch": 182.86024096385543, "grad_norm": 5.467922687530518, "learning_rate": 8.45631067961165e-06, "loss": 0.2313, "step": 18858 }, { "epoch": 182.86987951807228, "grad_norm": 7.287078380584717, "learning_rate": 8.451456310679611e-06, "loss": 0.2108, "step": 18859 }, { "epoch": 182.87951807228916, "grad_norm": 5.204339027404785, "learning_rate": 8.446601941747573e-06, "loss": 0.1574, "step": 18860 }, { "epoch": 182.88915662650604, "grad_norm": 15.348773002624512, "learning_rate": 8.441747572815534e-06, "loss": 0.3806, "step": 18861 }, { "epoch": 182.89879518072289, "grad_norm": 16.410541534423828, "learning_rate": 8.436893203883497e-06, "loss": 0.4247, "step": 18862 }, { "epoch": 182.90843373493976, "grad_norm": 2.4668219089508057, "learning_rate": 8.432038834951458e-06, "loss": 0.086, "step": 18863 }, { "epoch": 182.91807228915664, "grad_norm": 3.309443473815918, "learning_rate": 8.427184466019419e-06, "loss": 0.0296, "step": 18864 }, { "epoch": 182.9277108433735, "grad_norm": 7.328464508056641, "learning_rate": 8.422330097087378e-06, "loss": 0.2674, "step": 18865 }, { "epoch": 182.93734939759037, "grad_norm": 16.79216194152832, "learning_rate": 8.41747572815534e-06, "loss": 0.1823, "step": 18866 }, { "epoch": 182.94698795180722, "grad_norm": 8.703697204589844, "learning_rate": 8.4126213592233e-06, "loss": 0.2585, "step": 18867 }, { "epoch": 182.9566265060241, "grad_norm": 5.6930108070373535, "learning_rate": 8.407766990291263e-06, "loss": 0.1705, "step": 18868 }, { "epoch": 182.96626506024097, "grad_norm": 31.49251937866211, "learning_rate": 8.402912621359225e-06, "loss": 0.2319, "step": 18869 }, { "epoch": 182.97590361445782, "grad_norm": 8.918808937072754, "learning_rate": 8.398058252427186e-06, "loss": 0.282, "step": 18870 }, { "epoch": 182.9855421686747, "grad_norm": 27.19293212890625, "learning_rate": 8.393203883495145e-06, "loss": 0.4382, "step": 18871 }, { "epoch": 183.00120481927712, "grad_norm": 14.854110717773438, "learning_rate": 8.388349514563106e-06, "loss": 0.2765, "step": 18872 }, { "epoch": 183.01084337349397, "grad_norm": 5.688510417938232, "learning_rate": 8.383495145631067e-06, "loss": 0.1296, "step": 18873 }, { "epoch": 183.02048192771085, "grad_norm": 20.25754165649414, "learning_rate": 8.37864077669903e-06, "loss": 0.3911, "step": 18874 }, { "epoch": 183.03012048192772, "grad_norm": 19.917247772216797, "learning_rate": 8.373786407766991e-06, "loss": 0.2847, "step": 18875 }, { "epoch": 183.03975903614457, "grad_norm": 7.7294111251831055, "learning_rate": 8.368932038834953e-06, "loss": 0.2944, "step": 18876 }, { "epoch": 183.04939759036145, "grad_norm": 19.162220001220703, "learning_rate": 8.364077669902914e-06, "loss": 0.2232, "step": 18877 }, { "epoch": 183.0590361445783, "grad_norm": 14.990655899047852, "learning_rate": 8.359223300970873e-06, "loss": 0.4611, "step": 18878 }, { "epoch": 183.06867469879518, "grad_norm": 7.976815700531006, "learning_rate": 8.354368932038834e-06, "loss": 0.1541, "step": 18879 }, { "epoch": 183.07831325301206, "grad_norm": 3.0435938835144043, "learning_rate": 8.349514563106795e-06, "loss": 0.1993, "step": 18880 }, { "epoch": 183.0879518072289, "grad_norm": 16.10199737548828, "learning_rate": 8.344660194174758e-06, "loss": 0.2866, "step": 18881 }, { "epoch": 183.09759036144578, "grad_norm": 5.469327926635742, "learning_rate": 8.33980582524272e-06, "loss": 0.0901, "step": 18882 }, { "epoch": 183.10722891566266, "grad_norm": 11.990055084228516, "learning_rate": 8.33495145631068e-06, "loss": 0.271, "step": 18883 }, { "epoch": 183.1168674698795, "grad_norm": 10.916125297546387, "learning_rate": 8.330097087378642e-06, "loss": 0.188, "step": 18884 }, { "epoch": 183.1265060240964, "grad_norm": 10.309931755065918, "learning_rate": 8.325242718446601e-06, "loss": 0.3838, "step": 18885 }, { "epoch": 183.13614457831326, "grad_norm": 7.97373104095459, "learning_rate": 8.320388349514562e-06, "loss": 0.2908, "step": 18886 }, { "epoch": 183.1457831325301, "grad_norm": 11.105753898620605, "learning_rate": 8.315533980582525e-06, "loss": 0.247, "step": 18887 }, { "epoch": 183.155421686747, "grad_norm": 31.384634017944336, "learning_rate": 8.310679611650486e-06, "loss": 0.1641, "step": 18888 }, { "epoch": 183.16506024096387, "grad_norm": 6.840803623199463, "learning_rate": 8.305825242718447e-06, "loss": 0.1452, "step": 18889 }, { "epoch": 183.17469879518072, "grad_norm": 6.901218891143799, "learning_rate": 8.300970873786409e-06, "loss": 0.2627, "step": 18890 }, { "epoch": 183.1843373493976, "grad_norm": 10.030752182006836, "learning_rate": 8.29611650485437e-06, "loss": 0.295, "step": 18891 }, { "epoch": 183.19397590361444, "grad_norm": 13.722616195678711, "learning_rate": 8.29126213592233e-06, "loss": 0.1237, "step": 18892 }, { "epoch": 183.20361445783132, "grad_norm": 6.543969631195068, "learning_rate": 8.286407766990292e-06, "loss": 0.3894, "step": 18893 }, { "epoch": 183.2132530120482, "grad_norm": 8.190903663635254, "learning_rate": 8.281553398058253e-06, "loss": 0.2119, "step": 18894 }, { "epoch": 183.22289156626505, "grad_norm": 30.044496536254883, "learning_rate": 8.276699029126214e-06, "loss": 0.4029, "step": 18895 }, { "epoch": 183.23253012048193, "grad_norm": 22.52300453186035, "learning_rate": 8.271844660194175e-06, "loss": 0.2699, "step": 18896 }, { "epoch": 183.2421686746988, "grad_norm": 10.112302780151367, "learning_rate": 8.266990291262137e-06, "loss": 0.219, "step": 18897 }, { "epoch": 183.25180722891565, "grad_norm": 18.987159729003906, "learning_rate": 8.262135922330098e-06, "loss": 0.1375, "step": 18898 }, { "epoch": 183.26144578313253, "grad_norm": 36.458702087402344, "learning_rate": 8.257281553398057e-06, "loss": 0.6097, "step": 18899 }, { "epoch": 183.2710843373494, "grad_norm": 12.042035102844238, "learning_rate": 8.25242718446602e-06, "loss": 0.2579, "step": 18900 }, { "epoch": 183.28072289156626, "grad_norm": 9.263982772827148, "learning_rate": 8.247572815533981e-06, "loss": 0.2038, "step": 18901 }, { "epoch": 183.29036144578313, "grad_norm": 8.176279067993164, "learning_rate": 8.242718446601942e-06, "loss": 0.241, "step": 18902 }, { "epoch": 183.3, "grad_norm": 9.826202392578125, "learning_rate": 8.237864077669903e-06, "loss": 0.2414, "step": 18903 }, { "epoch": 183.30963855421686, "grad_norm": 7.250631332397461, "learning_rate": 8.233009708737865e-06, "loss": 0.2946, "step": 18904 }, { "epoch": 183.31927710843374, "grad_norm": 17.454771041870117, "learning_rate": 8.228155339805826e-06, "loss": 0.3916, "step": 18905 }, { "epoch": 183.32891566265062, "grad_norm": 7.921542644500732, "learning_rate": 8.223300970873787e-06, "loss": 0.1304, "step": 18906 }, { "epoch": 183.33855421686746, "grad_norm": 2.781440258026123, "learning_rate": 8.218446601941748e-06, "loss": 0.1801, "step": 18907 }, { "epoch": 183.34819277108434, "grad_norm": 8.543781280517578, "learning_rate": 8.21359223300971e-06, "loss": 0.2128, "step": 18908 }, { "epoch": 183.3578313253012, "grad_norm": 6.296426296234131, "learning_rate": 8.20873786407767e-06, "loss": 0.2107, "step": 18909 }, { "epoch": 183.36746987951807, "grad_norm": 9.501778602600098, "learning_rate": 8.203883495145632e-06, "loss": 0.2345, "step": 18910 }, { "epoch": 183.37710843373495, "grad_norm": 11.851414680480957, "learning_rate": 8.199029126213593e-06, "loss": 0.2854, "step": 18911 }, { "epoch": 183.3867469879518, "grad_norm": 5.145180702209473, "learning_rate": 8.194174757281554e-06, "loss": 0.0963, "step": 18912 }, { "epoch": 183.39638554216867, "grad_norm": 9.897038459777832, "learning_rate": 8.189320388349515e-06, "loss": 0.2646, "step": 18913 }, { "epoch": 183.40602409638555, "grad_norm": 14.147385597229004, "learning_rate": 8.184466019417476e-06, "loss": 0.1617, "step": 18914 }, { "epoch": 183.4156626506024, "grad_norm": 4.782212257385254, "learning_rate": 8.179611650485437e-06, "loss": 0.0505, "step": 18915 }, { "epoch": 183.42530120481928, "grad_norm": 12.416828155517578, "learning_rate": 8.174757281553398e-06, "loss": 0.2652, "step": 18916 }, { "epoch": 183.43493975903615, "grad_norm": 6.732258319854736, "learning_rate": 8.16990291262136e-06, "loss": 0.1571, "step": 18917 }, { "epoch": 183.444578313253, "grad_norm": 9.110502243041992, "learning_rate": 8.16504854368932e-06, "loss": 0.3815, "step": 18918 }, { "epoch": 183.45421686746988, "grad_norm": 10.397866249084473, "learning_rate": 8.160194174757282e-06, "loss": 0.297, "step": 18919 }, { "epoch": 183.46385542168676, "grad_norm": 15.809244155883789, "learning_rate": 8.155339805825243e-06, "loss": 0.3259, "step": 18920 }, { "epoch": 183.4734939759036, "grad_norm": 28.364225387573242, "learning_rate": 8.150485436893204e-06, "loss": 0.3733, "step": 18921 }, { "epoch": 183.48313253012049, "grad_norm": 10.925552368164062, "learning_rate": 8.145631067961165e-06, "loss": 0.3872, "step": 18922 }, { "epoch": 183.49277108433736, "grad_norm": 7.183149814605713, "learning_rate": 8.140776699029126e-06, "loss": 0.2119, "step": 18923 }, { "epoch": 183.5024096385542, "grad_norm": 13.933158874511719, "learning_rate": 8.135922330097088e-06, "loss": 0.3249, "step": 18924 }, { "epoch": 183.5120481927711, "grad_norm": 24.695125579833984, "learning_rate": 8.131067961165049e-06, "loss": 0.4348, "step": 18925 }, { "epoch": 183.52168674698794, "grad_norm": 9.324999809265137, "learning_rate": 8.12621359223301e-06, "loss": 0.15, "step": 18926 }, { "epoch": 183.53132530120482, "grad_norm": 8.47249984741211, "learning_rate": 8.121359223300971e-06, "loss": 0.2444, "step": 18927 }, { "epoch": 183.5409638554217, "grad_norm": 7.075451374053955, "learning_rate": 8.116504854368932e-06, "loss": 0.1994, "step": 18928 }, { "epoch": 183.55060240963854, "grad_norm": 6.97044038772583, "learning_rate": 8.111650485436893e-06, "loss": 0.1954, "step": 18929 }, { "epoch": 183.56024096385542, "grad_norm": 13.226178169250488, "learning_rate": 8.106796116504854e-06, "loss": 0.3507, "step": 18930 }, { "epoch": 183.5698795180723, "grad_norm": 12.483511924743652, "learning_rate": 8.101941747572817e-06, "loss": 0.2165, "step": 18931 }, { "epoch": 183.57951807228915, "grad_norm": 17.35428237915039, "learning_rate": 8.097087378640777e-06, "loss": 0.3471, "step": 18932 }, { "epoch": 183.58915662650602, "grad_norm": 9.398381233215332, "learning_rate": 8.092233009708738e-06, "loss": 0.1668, "step": 18933 }, { "epoch": 183.5987951807229, "grad_norm": 15.109716415405273, "learning_rate": 8.087378640776699e-06, "loss": 0.1091, "step": 18934 }, { "epoch": 183.60843373493975, "grad_norm": 25.625900268554688, "learning_rate": 8.08252427184466e-06, "loss": 0.4342, "step": 18935 }, { "epoch": 183.61807228915663, "grad_norm": 9.711897850036621, "learning_rate": 8.077669902912621e-06, "loss": 0.2098, "step": 18936 }, { "epoch": 183.6277108433735, "grad_norm": 6.630906105041504, "learning_rate": 8.072815533980584e-06, "loss": 0.1151, "step": 18937 }, { "epoch": 183.63734939759036, "grad_norm": 8.78534984588623, "learning_rate": 8.067961165048545e-06, "loss": 0.2408, "step": 18938 }, { "epoch": 183.64698795180723, "grad_norm": 9.235122680664062, "learning_rate": 8.063106796116505e-06, "loss": 0.1584, "step": 18939 }, { "epoch": 183.65662650602408, "grad_norm": 16.074386596679688, "learning_rate": 8.058252427184466e-06, "loss": 0.1626, "step": 18940 }, { "epoch": 183.66626506024096, "grad_norm": 5.814251899719238, "learning_rate": 8.053398058252427e-06, "loss": 0.2069, "step": 18941 }, { "epoch": 183.67590361445784, "grad_norm": 7.951888084411621, "learning_rate": 8.048543689320388e-06, "loss": 0.213, "step": 18942 }, { "epoch": 183.6855421686747, "grad_norm": 4.82267951965332, "learning_rate": 8.04368932038835e-06, "loss": 0.1446, "step": 18943 }, { "epoch": 183.69518072289156, "grad_norm": 13.42355728149414, "learning_rate": 8.038834951456312e-06, "loss": 0.3116, "step": 18944 }, { "epoch": 183.70481927710844, "grad_norm": 7.797633647918701, "learning_rate": 8.033980582524273e-06, "loss": 0.2464, "step": 18945 }, { "epoch": 183.7144578313253, "grad_norm": 32.15522003173828, "learning_rate": 8.029126213592233e-06, "loss": 0.2127, "step": 18946 }, { "epoch": 183.72409638554217, "grad_norm": 18.0894775390625, "learning_rate": 8.024271844660194e-06, "loss": 0.1639, "step": 18947 }, { "epoch": 183.73373493975905, "grad_norm": 30.79855728149414, "learning_rate": 8.019417475728155e-06, "loss": 0.2638, "step": 18948 }, { "epoch": 183.7433734939759, "grad_norm": 8.318896293640137, "learning_rate": 8.014563106796116e-06, "loss": 0.1784, "step": 18949 }, { "epoch": 183.75301204819277, "grad_norm": 19.48975372314453, "learning_rate": 8.009708737864079e-06, "loss": 0.2525, "step": 18950 }, { "epoch": 183.76265060240965, "grad_norm": 9.867166519165039, "learning_rate": 8.00485436893204e-06, "loss": 0.2073, "step": 18951 }, { "epoch": 183.7722891566265, "grad_norm": 7.049088001251221, "learning_rate": 8.000000000000001e-06, "loss": 0.1773, "step": 18952 }, { "epoch": 183.78192771084338, "grad_norm": 13.954571723937988, "learning_rate": 7.99514563106796e-06, "loss": 0.2124, "step": 18953 }, { "epoch": 183.79156626506025, "grad_norm": 21.702287673950195, "learning_rate": 7.990291262135922e-06, "loss": 0.402, "step": 18954 }, { "epoch": 183.8012048192771, "grad_norm": 9.136384010314941, "learning_rate": 7.985436893203883e-06, "loss": 0.2347, "step": 18955 }, { "epoch": 183.81084337349398, "grad_norm": 11.122517585754395, "learning_rate": 7.980582524271846e-06, "loss": 0.351, "step": 18956 }, { "epoch": 183.82048192771083, "grad_norm": 19.513107299804688, "learning_rate": 7.975728155339807e-06, "loss": 0.1736, "step": 18957 }, { "epoch": 183.8301204819277, "grad_norm": 10.530064582824707, "learning_rate": 7.970873786407768e-06, "loss": 0.196, "step": 18958 }, { "epoch": 183.83975903614459, "grad_norm": 10.786335945129395, "learning_rate": 7.96601941747573e-06, "loss": 0.1421, "step": 18959 }, { "epoch": 183.84939759036143, "grad_norm": 10.357975006103516, "learning_rate": 7.961165048543689e-06, "loss": 0.296, "step": 18960 }, { "epoch": 183.8590361445783, "grad_norm": 14.403515815734863, "learning_rate": 7.95631067961165e-06, "loss": 0.3187, "step": 18961 }, { "epoch": 183.8686746987952, "grad_norm": 6.989804267883301, "learning_rate": 7.951456310679611e-06, "loss": 0.2108, "step": 18962 }, { "epoch": 183.87831325301204, "grad_norm": 6.803229331970215, "learning_rate": 7.946601941747574e-06, "loss": 0.2115, "step": 18963 }, { "epoch": 183.88795180722892, "grad_norm": 11.581917762756348, "learning_rate": 7.941747572815535e-06, "loss": 0.2626, "step": 18964 }, { "epoch": 183.8975903614458, "grad_norm": 6.21002197265625, "learning_rate": 7.936893203883496e-06, "loss": 0.1809, "step": 18965 }, { "epoch": 183.90722891566264, "grad_norm": 13.469770431518555, "learning_rate": 7.932038834951457e-06, "loss": 0.3119, "step": 18966 }, { "epoch": 183.91686746987952, "grad_norm": 6.7016496658325195, "learning_rate": 7.927184466019417e-06, "loss": 0.2049, "step": 18967 }, { "epoch": 183.9265060240964, "grad_norm": 8.821487426757812, "learning_rate": 7.922330097087378e-06, "loss": 0.2967, "step": 18968 }, { "epoch": 183.93614457831325, "grad_norm": 20.225749969482422, "learning_rate": 7.91747572815534e-06, "loss": 0.2082, "step": 18969 }, { "epoch": 183.94578313253012, "grad_norm": 6.909819602966309, "learning_rate": 7.912621359223302e-06, "loss": 0.2181, "step": 18970 }, { "epoch": 183.955421686747, "grad_norm": 9.916111946105957, "learning_rate": 7.907766990291263e-06, "loss": 0.2834, "step": 18971 }, { "epoch": 183.96506024096385, "grad_norm": 17.796855926513672, "learning_rate": 7.902912621359224e-06, "loss": 0.4575, "step": 18972 }, { "epoch": 183.97469879518073, "grad_norm": 9.951980590820312, "learning_rate": 7.898058252427185e-06, "loss": 0.3346, "step": 18973 }, { "epoch": 183.98433734939758, "grad_norm": 10.58951187133789, "learning_rate": 7.893203883495145e-06, "loss": 0.3313, "step": 18974 }, { "epoch": 183.99397590361446, "grad_norm": 15.669605255126953, "learning_rate": 7.888349514563108e-06, "loss": 0.3305, "step": 18975 }, { "epoch": 184.00963855421688, "grad_norm": 11.616720199584961, "learning_rate": 7.883495145631069e-06, "loss": 0.4352, "step": 18976 }, { "epoch": 184.01927710843373, "grad_norm": 12.577234268188477, "learning_rate": 7.87864077669903e-06, "loss": 0.25, "step": 18977 }, { "epoch": 184.0289156626506, "grad_norm": 10.162398338317871, "learning_rate": 7.873786407766991e-06, "loss": 0.3808, "step": 18978 }, { "epoch": 184.03855421686748, "grad_norm": 11.979772567749023, "learning_rate": 7.868932038834952e-06, "loss": 0.2349, "step": 18979 }, { "epoch": 184.04819277108433, "grad_norm": 13.704495429992676, "learning_rate": 7.864077669902912e-06, "loss": 0.1316, "step": 18980 }, { "epoch": 184.0578313253012, "grad_norm": 9.34498119354248, "learning_rate": 7.859223300970873e-06, "loss": 0.2102, "step": 18981 }, { "epoch": 184.0674698795181, "grad_norm": 17.02644157409668, "learning_rate": 7.854368932038836e-06, "loss": 0.3631, "step": 18982 }, { "epoch": 184.07710843373494, "grad_norm": 18.62993812561035, "learning_rate": 7.849514563106797e-06, "loss": 0.264, "step": 18983 }, { "epoch": 184.0867469879518, "grad_norm": 16.876399993896484, "learning_rate": 7.844660194174758e-06, "loss": 0.3261, "step": 18984 }, { "epoch": 184.09638554216866, "grad_norm": 2.9618349075317383, "learning_rate": 7.839805825242719e-06, "loss": 0.0542, "step": 18985 }, { "epoch": 184.10602409638554, "grad_norm": 11.082291603088379, "learning_rate": 7.83495145631068e-06, "loss": 0.2431, "step": 18986 }, { "epoch": 184.11566265060242, "grad_norm": 16.364362716674805, "learning_rate": 7.83009708737864e-06, "loss": 0.37, "step": 18987 }, { "epoch": 184.12530120481927, "grad_norm": 26.1643123626709, "learning_rate": 7.825242718446603e-06, "loss": 0.2586, "step": 18988 }, { "epoch": 184.13493975903614, "grad_norm": 10.401809692382812, "learning_rate": 7.820388349514564e-06, "loss": 0.2891, "step": 18989 }, { "epoch": 184.14457831325302, "grad_norm": 17.108877182006836, "learning_rate": 7.815533980582525e-06, "loss": 0.2044, "step": 18990 }, { "epoch": 184.15421686746987, "grad_norm": 11.735225677490234, "learning_rate": 7.810679611650486e-06, "loss": 0.243, "step": 18991 }, { "epoch": 184.16385542168675, "grad_norm": 7.683297634124756, "learning_rate": 7.805825242718447e-06, "loss": 0.2398, "step": 18992 }, { "epoch": 184.17349397590363, "grad_norm": 6.103055477142334, "learning_rate": 7.800970873786408e-06, "loss": 0.1049, "step": 18993 }, { "epoch": 184.18313253012047, "grad_norm": 21.855955123901367, "learning_rate": 7.79611650485437e-06, "loss": 0.2252, "step": 18994 }, { "epoch": 184.19277108433735, "grad_norm": 5.633406639099121, "learning_rate": 7.79126213592233e-06, "loss": 0.1312, "step": 18995 }, { "epoch": 184.20240963855423, "grad_norm": 24.45745086669922, "learning_rate": 7.786407766990292e-06, "loss": 0.2413, "step": 18996 }, { "epoch": 184.21204819277108, "grad_norm": 13.222648620605469, "learning_rate": 7.781553398058253e-06, "loss": 0.1197, "step": 18997 }, { "epoch": 184.22168674698796, "grad_norm": 22.27859878540039, "learning_rate": 7.776699029126214e-06, "loss": 0.2357, "step": 18998 }, { "epoch": 184.2313253012048, "grad_norm": 26.446800231933594, "learning_rate": 7.771844660194175e-06, "loss": 0.252, "step": 18999 }, { "epoch": 184.24096385542168, "grad_norm": 6.508040428161621, "learning_rate": 7.766990291262136e-06, "loss": 0.1711, "step": 19000 }, { "epoch": 184.25060240963856, "grad_norm": 7.131103515625, "learning_rate": 7.762135922330097e-06, "loss": 0.2299, "step": 19001 }, { "epoch": 184.2602409638554, "grad_norm": 7.779987812042236, "learning_rate": 7.757281553398059e-06, "loss": 0.1716, "step": 19002 }, { "epoch": 184.2698795180723, "grad_norm": 7.323197841644287, "learning_rate": 7.75242718446602e-06, "loss": 0.1727, "step": 19003 }, { "epoch": 184.27951807228916, "grad_norm": 11.912672996520996, "learning_rate": 7.747572815533981e-06, "loss": 0.2872, "step": 19004 }, { "epoch": 184.289156626506, "grad_norm": 14.691526412963867, "learning_rate": 7.742718446601942e-06, "loss": 0.5067, "step": 19005 }, { "epoch": 184.2987951807229, "grad_norm": 13.85490608215332, "learning_rate": 7.737864077669903e-06, "loss": 0.1239, "step": 19006 }, { "epoch": 184.30843373493977, "grad_norm": 14.708559036254883, "learning_rate": 7.733009708737864e-06, "loss": 0.2633, "step": 19007 }, { "epoch": 184.31807228915662, "grad_norm": 11.821600914001465, "learning_rate": 7.728155339805825e-06, "loss": 0.4506, "step": 19008 }, { "epoch": 184.3277108433735, "grad_norm": 8.876230239868164, "learning_rate": 7.723300970873787e-06, "loss": 0.2232, "step": 19009 }, { "epoch": 184.33734939759037, "grad_norm": 12.593559265136719, "learning_rate": 7.718446601941748e-06, "loss": 0.2285, "step": 19010 }, { "epoch": 184.34698795180722, "grad_norm": 8.699782371520996, "learning_rate": 7.713592233009709e-06, "loss": 0.3179, "step": 19011 }, { "epoch": 184.3566265060241, "grad_norm": 4.5615315437316895, "learning_rate": 7.70873786407767e-06, "loss": 0.1647, "step": 19012 }, { "epoch": 184.36626506024098, "grad_norm": 11.489721298217773, "learning_rate": 7.703883495145631e-06, "loss": 0.2896, "step": 19013 }, { "epoch": 184.37590361445783, "grad_norm": 33.714134216308594, "learning_rate": 7.699029126213592e-06, "loss": 0.4243, "step": 19014 }, { "epoch": 184.3855421686747, "grad_norm": 8.893417358398438, "learning_rate": 7.694174757281554e-06, "loss": 0.2022, "step": 19015 }, { "epoch": 184.39518072289155, "grad_norm": 10.85174560546875, "learning_rate": 7.689320388349515e-06, "loss": 0.2471, "step": 19016 }, { "epoch": 184.40481927710843, "grad_norm": 7.193668365478516, "learning_rate": 7.684466019417476e-06, "loss": 0.0585, "step": 19017 }, { "epoch": 184.4144578313253, "grad_norm": 5.350110054016113, "learning_rate": 7.679611650485437e-06, "loss": 0.1135, "step": 19018 }, { "epoch": 184.42409638554216, "grad_norm": 14.833402633666992, "learning_rate": 7.674757281553398e-06, "loss": 0.3259, "step": 19019 }, { "epoch": 184.43373493975903, "grad_norm": 11.119089126586914, "learning_rate": 7.66990291262136e-06, "loss": 0.4092, "step": 19020 }, { "epoch": 184.4433734939759, "grad_norm": 21.237064361572266, "learning_rate": 7.66504854368932e-06, "loss": 0.3938, "step": 19021 }, { "epoch": 184.45301204819276, "grad_norm": 7.732809066772461, "learning_rate": 7.660194174757282e-06, "loss": 0.2, "step": 19022 }, { "epoch": 184.46265060240964, "grad_norm": 8.187138557434082, "learning_rate": 7.655339805825243e-06, "loss": 0.2164, "step": 19023 }, { "epoch": 184.47228915662652, "grad_norm": 8.807751655578613, "learning_rate": 7.650485436893204e-06, "loss": 0.3402, "step": 19024 }, { "epoch": 184.48192771084337, "grad_norm": 9.886106491088867, "learning_rate": 7.645631067961165e-06, "loss": 0.2447, "step": 19025 }, { "epoch": 184.49156626506024, "grad_norm": 9.552929878234863, "learning_rate": 7.640776699029128e-06, "loss": 0.2708, "step": 19026 }, { "epoch": 184.50120481927712, "grad_norm": 10.712648391723633, "learning_rate": 7.635922330097087e-06, "loss": 0.3389, "step": 19027 }, { "epoch": 184.51084337349397, "grad_norm": 5.752227783203125, "learning_rate": 7.631067961165048e-06, "loss": 0.151, "step": 19028 }, { "epoch": 184.52048192771085, "grad_norm": 22.326465606689453, "learning_rate": 7.6262135922330096e-06, "loss": 0.2875, "step": 19029 }, { "epoch": 184.53012048192772, "grad_norm": 7.275888919830322, "learning_rate": 7.621359223300971e-06, "loss": 0.125, "step": 19030 }, { "epoch": 184.53975903614457, "grad_norm": 3.413816213607788, "learning_rate": 7.616504854368932e-06, "loss": 0.1394, "step": 19031 }, { "epoch": 184.54939759036145, "grad_norm": 13.81822681427002, "learning_rate": 7.611650485436894e-06, "loss": 0.328, "step": 19032 }, { "epoch": 184.5590361445783, "grad_norm": 26.507781982421875, "learning_rate": 7.606796116504855e-06, "loss": 0.2938, "step": 19033 }, { "epoch": 184.56867469879518, "grad_norm": 10.576004028320312, "learning_rate": 7.601941747572816e-06, "loss": 0.2118, "step": 19034 }, { "epoch": 184.57831325301206, "grad_norm": 2.835961103439331, "learning_rate": 7.597087378640777e-06, "loss": 0.115, "step": 19035 }, { "epoch": 184.5879518072289, "grad_norm": 20.647794723510742, "learning_rate": 7.5922330097087376e-06, "loss": 0.215, "step": 19036 }, { "epoch": 184.59759036144578, "grad_norm": 10.592238426208496, "learning_rate": 7.587378640776699e-06, "loss": 0.2544, "step": 19037 }, { "epoch": 184.60722891566266, "grad_norm": 4.733745098114014, "learning_rate": 7.5825242718446616e-06, "loss": 0.1289, "step": 19038 }, { "epoch": 184.6168674698795, "grad_norm": 17.270931243896484, "learning_rate": 7.577669902912622e-06, "loss": 0.161, "step": 19039 }, { "epoch": 184.6265060240964, "grad_norm": 7.663885593414307, "learning_rate": 7.572815533980583e-06, "loss": 0.1938, "step": 19040 }, { "epoch": 184.63614457831326, "grad_norm": 9.24367618560791, "learning_rate": 7.567961165048544e-06, "loss": 0.324, "step": 19041 }, { "epoch": 184.6457831325301, "grad_norm": 10.046470642089844, "learning_rate": 7.563106796116505e-06, "loss": 0.1902, "step": 19042 }, { "epoch": 184.655421686747, "grad_norm": 13.541681289672852, "learning_rate": 7.558252427184466e-06, "loss": 0.2175, "step": 19043 }, { "epoch": 184.66506024096387, "grad_norm": 21.076793670654297, "learning_rate": 7.553398058252427e-06, "loss": 0.4937, "step": 19044 }, { "epoch": 184.67469879518072, "grad_norm": 12.010034561157227, "learning_rate": 7.54854368932039e-06, "loss": 0.2733, "step": 19045 }, { "epoch": 184.6843373493976, "grad_norm": 10.593052864074707, "learning_rate": 7.54368932038835e-06, "loss": 0.309, "step": 19046 }, { "epoch": 184.69397590361444, "grad_norm": 8.741422653198242, "learning_rate": 7.538834951456311e-06, "loss": 0.1905, "step": 19047 }, { "epoch": 184.70361445783132, "grad_norm": 7.4297075271606445, "learning_rate": 7.533980582524272e-06, "loss": 0.1497, "step": 19048 }, { "epoch": 184.7132530120482, "grad_norm": 6.608257293701172, "learning_rate": 7.529126213592233e-06, "loss": 0.087, "step": 19049 }, { "epoch": 184.72289156626505, "grad_norm": 15.76626205444336, "learning_rate": 7.524271844660194e-06, "loss": 0.3317, "step": 19050 }, { "epoch": 184.73253012048193, "grad_norm": 10.058210372924805, "learning_rate": 7.5194174757281565e-06, "loss": 0.2094, "step": 19051 }, { "epoch": 184.7421686746988, "grad_norm": 5.372798919677734, "learning_rate": 7.514563106796118e-06, "loss": 0.2278, "step": 19052 }, { "epoch": 184.75180722891565, "grad_norm": 6.9729695320129395, "learning_rate": 7.509708737864078e-06, "loss": 0.2431, "step": 19053 }, { "epoch": 184.76144578313253, "grad_norm": 16.230056762695312, "learning_rate": 7.504854368932039e-06, "loss": 0.3075, "step": 19054 }, { "epoch": 184.7710843373494, "grad_norm": 11.071069717407227, "learning_rate": 7.5e-06, "loss": 0.2712, "step": 19055 }, { "epoch": 184.78072289156626, "grad_norm": 12.2735595703125, "learning_rate": 7.495145631067961e-06, "loss": 0.2582, "step": 19056 }, { "epoch": 184.79036144578313, "grad_norm": 15.144400596618652, "learning_rate": 7.490291262135923e-06, "loss": 0.2477, "step": 19057 }, { "epoch": 184.8, "grad_norm": 9.589399337768555, "learning_rate": 7.4854368932038845e-06, "loss": 0.3269, "step": 19058 }, { "epoch": 184.80963855421686, "grad_norm": 9.864055633544922, "learning_rate": 7.480582524271846e-06, "loss": 0.2714, "step": 19059 }, { "epoch": 184.81927710843374, "grad_norm": 16.939252853393555, "learning_rate": 7.475728155339806e-06, "loss": 0.4609, "step": 19060 }, { "epoch": 184.82891566265062, "grad_norm": 13.081833839416504, "learning_rate": 7.470873786407767e-06, "loss": 0.1565, "step": 19061 }, { "epoch": 184.83855421686746, "grad_norm": 12.639111518859863, "learning_rate": 7.466019417475728e-06, "loss": 0.3038, "step": 19062 }, { "epoch": 184.84819277108434, "grad_norm": 12.838761329650879, "learning_rate": 7.461165048543689e-06, "loss": 0.1137, "step": 19063 }, { "epoch": 184.8578313253012, "grad_norm": 19.848129272460938, "learning_rate": 7.456310679611651e-06, "loss": 0.3942, "step": 19064 }, { "epoch": 184.86746987951807, "grad_norm": 31.053874969482422, "learning_rate": 7.4514563106796125e-06, "loss": 0.2192, "step": 19065 }, { "epoch": 184.87710843373495, "grad_norm": 5.703768730163574, "learning_rate": 7.446601941747574e-06, "loss": 0.1428, "step": 19066 }, { "epoch": 184.8867469879518, "grad_norm": 9.692972183227539, "learning_rate": 7.441747572815534e-06, "loss": 0.4401, "step": 19067 }, { "epoch": 184.89638554216867, "grad_norm": 18.899721145629883, "learning_rate": 7.436893203883495e-06, "loss": 0.5858, "step": 19068 }, { "epoch": 184.90602409638555, "grad_norm": 22.836040496826172, "learning_rate": 7.432038834951456e-06, "loss": 0.3117, "step": 19069 }, { "epoch": 184.9156626506024, "grad_norm": 29.216400146484375, "learning_rate": 7.427184466019418e-06, "loss": 0.5407, "step": 19070 }, { "epoch": 184.92530120481928, "grad_norm": 10.250273704528809, "learning_rate": 7.422330097087379e-06, "loss": 0.4257, "step": 19071 }, { "epoch": 184.93493975903615, "grad_norm": 7.2894978523254395, "learning_rate": 7.4174757281553405e-06, "loss": 0.1319, "step": 19072 }, { "epoch": 184.944578313253, "grad_norm": 11.360594749450684, "learning_rate": 7.412621359223302e-06, "loss": 0.3361, "step": 19073 }, { "epoch": 184.95421686746988, "grad_norm": 12.545698165893555, "learning_rate": 7.407766990291262e-06, "loss": 0.2067, "step": 19074 }, { "epoch": 184.96385542168676, "grad_norm": 12.798324584960938, "learning_rate": 7.402912621359223e-06, "loss": 0.1579, "step": 19075 }, { "epoch": 184.9734939759036, "grad_norm": 9.266562461853027, "learning_rate": 7.398058252427185e-06, "loss": 0.3019, "step": 19076 }, { "epoch": 184.98313253012049, "grad_norm": 16.34260368347168, "learning_rate": 7.393203883495146e-06, "loss": 0.2327, "step": 19077 }, { "epoch": 184.99277108433736, "grad_norm": 12.482290267944336, "learning_rate": 7.388349514563107e-06, "loss": 0.3233, "step": 19078 }, { "epoch": 185.00843373493976, "grad_norm": 20.019088745117188, "learning_rate": 7.3834951456310685e-06, "loss": 0.3825, "step": 19079 }, { "epoch": 185.01807228915663, "grad_norm": 19.458114624023438, "learning_rate": 7.378640776699029e-06, "loss": 0.1635, "step": 19080 }, { "epoch": 185.02771084337348, "grad_norm": 30.95086669921875, "learning_rate": 7.37378640776699e-06, "loss": 0.2934, "step": 19081 }, { "epoch": 185.03734939759036, "grad_norm": 42.795677185058594, "learning_rate": 7.368932038834951e-06, "loss": 0.3703, "step": 19082 }, { "epoch": 185.04698795180724, "grad_norm": 6.4081220626831055, "learning_rate": 7.364077669902913e-06, "loss": 0.294, "step": 19083 }, { "epoch": 185.0566265060241, "grad_norm": 27.329498291015625, "learning_rate": 7.359223300970874e-06, "loss": 0.1196, "step": 19084 }, { "epoch": 185.06626506024097, "grad_norm": 21.65639305114746, "learning_rate": 7.354368932038835e-06, "loss": 0.2299, "step": 19085 }, { "epoch": 185.07590361445784, "grad_norm": 25.432109832763672, "learning_rate": 7.3495145631067966e-06, "loss": 0.0952, "step": 19086 }, { "epoch": 185.0855421686747, "grad_norm": 12.173652648925781, "learning_rate": 7.344660194174757e-06, "loss": 0.184, "step": 19087 }, { "epoch": 185.09518072289157, "grad_norm": 28.80599594116211, "learning_rate": 7.339805825242718e-06, "loss": 0.4144, "step": 19088 }, { "epoch": 185.10481927710845, "grad_norm": 14.772555351257324, "learning_rate": 7.334951456310681e-06, "loss": 0.2694, "step": 19089 }, { "epoch": 185.1144578313253, "grad_norm": 22.944238662719727, "learning_rate": 7.330097087378641e-06, "loss": 0.1851, "step": 19090 }, { "epoch": 185.12409638554217, "grad_norm": 9.1244535446167, "learning_rate": 7.325242718446602e-06, "loss": 0.1726, "step": 19091 }, { "epoch": 185.13373493975902, "grad_norm": 10.690827369689941, "learning_rate": 7.3203883495145634e-06, "loss": 0.3395, "step": 19092 }, { "epoch": 185.1433734939759, "grad_norm": 2.010714292526245, "learning_rate": 7.3155339805825246e-06, "loss": 0.1542, "step": 19093 }, { "epoch": 185.15301204819278, "grad_norm": 9.137922286987305, "learning_rate": 7.310679611650485e-06, "loss": 0.16, "step": 19094 }, { "epoch": 185.16265060240963, "grad_norm": 5.908997058868408, "learning_rate": 7.305825242718448e-06, "loss": 0.1205, "step": 19095 }, { "epoch": 185.1722891566265, "grad_norm": 20.725313186645508, "learning_rate": 7.300970873786409e-06, "loss": 0.1807, "step": 19096 }, { "epoch": 185.18192771084338, "grad_norm": 3.461483955383301, "learning_rate": 7.296116504854369e-06, "loss": 0.23, "step": 19097 }, { "epoch": 185.19156626506023, "grad_norm": 4.192473411560059, "learning_rate": 7.29126213592233e-06, "loss": 0.1959, "step": 19098 }, { "epoch": 185.2012048192771, "grad_norm": 5.691680908203125, "learning_rate": 7.2864077669902914e-06, "loss": 0.1214, "step": 19099 }, { "epoch": 185.210843373494, "grad_norm": 10.261183738708496, "learning_rate": 7.281553398058253e-06, "loss": 0.2501, "step": 19100 }, { "epoch": 185.22048192771084, "grad_norm": 20.8284912109375, "learning_rate": 7.276699029126213e-06, "loss": 0.1867, "step": 19101 }, { "epoch": 185.2301204819277, "grad_norm": 21.976533889770508, "learning_rate": 7.271844660194176e-06, "loss": 0.1932, "step": 19102 }, { "epoch": 185.2397590361446, "grad_norm": 14.91441535949707, "learning_rate": 7.266990291262137e-06, "loss": 0.3429, "step": 19103 }, { "epoch": 185.24939759036144, "grad_norm": 17.609827041625977, "learning_rate": 7.262135922330097e-06, "loss": 0.2374, "step": 19104 }, { "epoch": 185.25903614457832, "grad_norm": 9.267325401306152, "learning_rate": 7.257281553398058e-06, "loss": 0.1558, "step": 19105 }, { "epoch": 185.2686746987952, "grad_norm": 30.932218551635742, "learning_rate": 7.2524271844660195e-06, "loss": 0.4106, "step": 19106 }, { "epoch": 185.27831325301204, "grad_norm": 9.5095853805542, "learning_rate": 7.247572815533981e-06, "loss": 0.3874, "step": 19107 }, { "epoch": 185.28795180722892, "grad_norm": 22.503122329711914, "learning_rate": 7.242718446601943e-06, "loss": 0.396, "step": 19108 }, { "epoch": 185.29759036144577, "grad_norm": 34.385398864746094, "learning_rate": 7.237864077669904e-06, "loss": 0.3061, "step": 19109 }, { "epoch": 185.30722891566265, "grad_norm": 30.777753829956055, "learning_rate": 7.233009708737865e-06, "loss": 0.2537, "step": 19110 }, { "epoch": 185.31686746987953, "grad_norm": 24.47568130493164, "learning_rate": 7.228155339805825e-06, "loss": 0.2926, "step": 19111 }, { "epoch": 185.32650602409637, "grad_norm": 26.225521087646484, "learning_rate": 7.223300970873786e-06, "loss": 0.1972, "step": 19112 }, { "epoch": 185.33614457831325, "grad_norm": 61.226680755615234, "learning_rate": 7.2184466019417475e-06, "loss": 0.203, "step": 19113 }, { "epoch": 185.34578313253013, "grad_norm": 34.46961975097656, "learning_rate": 7.2135922330097095e-06, "loss": 0.1847, "step": 19114 }, { "epoch": 185.35542168674698, "grad_norm": 33.0224723815918, "learning_rate": 7.208737864077671e-06, "loss": 0.224, "step": 19115 }, { "epoch": 185.36506024096386, "grad_norm": 21.259742736816406, "learning_rate": 7.203883495145632e-06, "loss": 0.4203, "step": 19116 }, { "epoch": 185.37469879518073, "grad_norm": 37.51176834106445, "learning_rate": 7.199029126213593e-06, "loss": 0.1974, "step": 19117 }, { "epoch": 185.38433734939758, "grad_norm": 34.64375305175781, "learning_rate": 7.194174757281553e-06, "loss": 0.2017, "step": 19118 }, { "epoch": 185.39397590361446, "grad_norm": 9.43703842163086, "learning_rate": 7.189320388349514e-06, "loss": 0.1509, "step": 19119 }, { "epoch": 185.40361445783134, "grad_norm": 8.108878135681152, "learning_rate": 7.1844660194174755e-06, "loss": 0.1785, "step": 19120 }, { "epoch": 185.4132530120482, "grad_norm": 11.343451499938965, "learning_rate": 7.1796116504854375e-06, "loss": 0.3821, "step": 19121 }, { "epoch": 185.42289156626506, "grad_norm": 19.449796676635742, "learning_rate": 7.174757281553399e-06, "loss": 0.1574, "step": 19122 }, { "epoch": 185.4325301204819, "grad_norm": 41.91556167602539, "learning_rate": 7.16990291262136e-06, "loss": 0.1767, "step": 19123 }, { "epoch": 185.4421686746988, "grad_norm": 8.66486930847168, "learning_rate": 7.165048543689321e-06, "loss": 0.2396, "step": 19124 }, { "epoch": 185.45180722891567, "grad_norm": 2.908491373062134, "learning_rate": 7.160194174757281e-06, "loss": 0.0515, "step": 19125 }, { "epoch": 185.46144578313252, "grad_norm": 15.909530639648438, "learning_rate": 7.155339805825242e-06, "loss": 0.292, "step": 19126 }, { "epoch": 185.4710843373494, "grad_norm": 28.46489906311035, "learning_rate": 7.150485436893204e-06, "loss": 0.327, "step": 19127 }, { "epoch": 185.48072289156627, "grad_norm": 5.692263603210449, "learning_rate": 7.1456310679611655e-06, "loss": 0.1269, "step": 19128 }, { "epoch": 185.49036144578312, "grad_norm": 6.473318576812744, "learning_rate": 7.140776699029127e-06, "loss": 0.2216, "step": 19129 }, { "epoch": 185.5, "grad_norm": 22.030235290527344, "learning_rate": 7.135922330097088e-06, "loss": 0.3049, "step": 19130 }, { "epoch": 185.50963855421688, "grad_norm": 13.590936660766602, "learning_rate": 7.131067961165049e-06, "loss": 0.1815, "step": 19131 }, { "epoch": 185.51927710843373, "grad_norm": 52.29056167602539, "learning_rate": 7.126213592233009e-06, "loss": 0.3052, "step": 19132 }, { "epoch": 185.5289156626506, "grad_norm": 3.7124927043914795, "learning_rate": 7.121359223300972e-06, "loss": 0.1195, "step": 19133 }, { "epoch": 185.53855421686748, "grad_norm": 14.707208633422852, "learning_rate": 7.116504854368932e-06, "loss": 0.3187, "step": 19134 }, { "epoch": 185.54819277108433, "grad_norm": 16.575634002685547, "learning_rate": 7.1116504854368935e-06, "loss": 0.3903, "step": 19135 }, { "epoch": 185.5578313253012, "grad_norm": 9.687956809997559, "learning_rate": 7.106796116504855e-06, "loss": 0.2457, "step": 19136 }, { "epoch": 185.5674698795181, "grad_norm": 23.301998138427734, "learning_rate": 7.101941747572816e-06, "loss": 0.2118, "step": 19137 }, { "epoch": 185.57710843373494, "grad_norm": 12.380325317382812, "learning_rate": 7.097087378640776e-06, "loss": 0.2701, "step": 19138 }, { "epoch": 185.5867469879518, "grad_norm": 5.453337669372559, "learning_rate": 7.092233009708739e-06, "loss": 0.3657, "step": 19139 }, { "epoch": 185.59638554216866, "grad_norm": 9.879929542541504, "learning_rate": 7.0873786407767e-06, "loss": 0.1395, "step": 19140 }, { "epoch": 185.60602409638554, "grad_norm": 8.755455017089844, "learning_rate": 7.08252427184466e-06, "loss": 0.1957, "step": 19141 }, { "epoch": 185.61566265060242, "grad_norm": 15.847679138183594, "learning_rate": 7.0776699029126216e-06, "loss": 0.1923, "step": 19142 }, { "epoch": 185.62530120481927, "grad_norm": 14.155956268310547, "learning_rate": 7.072815533980583e-06, "loss": 0.1017, "step": 19143 }, { "epoch": 185.63493975903614, "grad_norm": 4.7166643142700195, "learning_rate": 7.067961165048544e-06, "loss": 0.1752, "step": 19144 }, { "epoch": 185.64457831325302, "grad_norm": 8.972028732299805, "learning_rate": 7.063106796116504e-06, "loss": 0.2628, "step": 19145 }, { "epoch": 185.65421686746987, "grad_norm": 16.459890365600586, "learning_rate": 7.058252427184467e-06, "loss": 0.4159, "step": 19146 }, { "epoch": 185.66385542168675, "grad_norm": 1.8779574632644653, "learning_rate": 7.053398058252428e-06, "loss": 0.2156, "step": 19147 }, { "epoch": 185.67349397590363, "grad_norm": 52.40987777709961, "learning_rate": 7.0485436893203884e-06, "loss": 0.2442, "step": 19148 }, { "epoch": 185.68313253012047, "grad_norm": 9.117574691772461, "learning_rate": 7.04368932038835e-06, "loss": 0.4388, "step": 19149 }, { "epoch": 185.69277108433735, "grad_norm": 10.280327796936035, "learning_rate": 7.038834951456311e-06, "loss": 0.1262, "step": 19150 }, { "epoch": 185.70240963855423, "grad_norm": 3.3296682834625244, "learning_rate": 7.033980582524272e-06, "loss": 0.1446, "step": 19151 }, { "epoch": 185.71204819277108, "grad_norm": 36.261104583740234, "learning_rate": 7.029126213592234e-06, "loss": 0.3197, "step": 19152 }, { "epoch": 185.72168674698796, "grad_norm": 16.45635223388672, "learning_rate": 7.024271844660195e-06, "loss": 0.1795, "step": 19153 }, { "epoch": 185.73132530120483, "grad_norm": 29.062837600708008, "learning_rate": 7.019417475728156e-06, "loss": 0.3447, "step": 19154 }, { "epoch": 185.74096385542168, "grad_norm": 21.480541229248047, "learning_rate": 7.0145631067961165e-06, "loss": 0.3241, "step": 19155 }, { "epoch": 185.75060240963856, "grad_norm": 11.08051586151123, "learning_rate": 7.009708737864078e-06, "loss": 0.1341, "step": 19156 }, { "epoch": 185.7602409638554, "grad_norm": 11.932172775268555, "learning_rate": 7.004854368932039e-06, "loss": 0.16, "step": 19157 }, { "epoch": 185.7698795180723, "grad_norm": 25.128265380859375, "learning_rate": 7.000000000000001e-06, "loss": 0.2892, "step": 19158 }, { "epoch": 185.77951807228916, "grad_norm": 8.694640159606934, "learning_rate": 6.995145631067962e-06, "loss": 0.1747, "step": 19159 }, { "epoch": 185.789156626506, "grad_norm": 18.618667602539062, "learning_rate": 6.990291262135923e-06, "loss": 0.3465, "step": 19160 }, { "epoch": 185.7987951807229, "grad_norm": 3.7660019397735596, "learning_rate": 6.985436893203884e-06, "loss": 0.31, "step": 19161 }, { "epoch": 185.80843373493977, "grad_norm": 17.670658111572266, "learning_rate": 6.9805825242718445e-06, "loss": 0.2581, "step": 19162 }, { "epoch": 185.81807228915662, "grad_norm": 21.799671173095703, "learning_rate": 6.975728155339806e-06, "loss": 0.224, "step": 19163 }, { "epoch": 185.8277108433735, "grad_norm": 25.284181594848633, "learning_rate": 6.970873786407767e-06, "loss": 0.3315, "step": 19164 }, { "epoch": 185.83734939759037, "grad_norm": 48.62631607055664, "learning_rate": 6.966019417475729e-06, "loss": 0.2909, "step": 19165 }, { "epoch": 185.84698795180722, "grad_norm": 7.850836753845215, "learning_rate": 6.96116504854369e-06, "loss": 0.3005, "step": 19166 }, { "epoch": 185.8566265060241, "grad_norm": 1.7091596126556396, "learning_rate": 6.956310679611651e-06, "loss": 0.1902, "step": 19167 }, { "epoch": 185.86626506024098, "grad_norm": 41.26972961425781, "learning_rate": 6.951456310679612e-06, "loss": 0.4827, "step": 19168 }, { "epoch": 185.87590361445783, "grad_norm": 3.56268048286438, "learning_rate": 6.9466019417475725e-06, "loss": 0.2003, "step": 19169 }, { "epoch": 185.8855421686747, "grad_norm": 41.41291809082031, "learning_rate": 6.941747572815534e-06, "loss": 0.2739, "step": 19170 }, { "epoch": 185.89518072289155, "grad_norm": 29.189767837524414, "learning_rate": 6.9368932038834965e-06, "loss": 0.4092, "step": 19171 }, { "epoch": 185.90481927710843, "grad_norm": 1.320241928100586, "learning_rate": 6.932038834951457e-06, "loss": 0.2328, "step": 19172 }, { "epoch": 185.9144578313253, "grad_norm": 3.0629255771636963, "learning_rate": 6.927184466019418e-06, "loss": 0.4428, "step": 19173 }, { "epoch": 185.92409638554216, "grad_norm": 41.084346771240234, "learning_rate": 6.922330097087379e-06, "loss": 0.325, "step": 19174 }, { "epoch": 185.93373493975903, "grad_norm": 2.680521249771118, "learning_rate": 6.91747572815534e-06, "loss": 0.375, "step": 19175 }, { "epoch": 185.9433734939759, "grad_norm": 8.776408195495605, "learning_rate": 6.9126213592233005e-06, "loss": 0.1283, "step": 19176 }, { "epoch": 185.95301204819276, "grad_norm": 19.42325782775879, "learning_rate": 6.907766990291263e-06, "loss": 0.283, "step": 19177 }, { "epoch": 185.96265060240964, "grad_norm": 19.026107788085938, "learning_rate": 6.902912621359224e-06, "loss": 0.2068, "step": 19178 }, { "epoch": 185.97228915662652, "grad_norm": 22.782289505004883, "learning_rate": 6.898058252427185e-06, "loss": 0.398, "step": 19179 }, { "epoch": 185.98192771084337, "grad_norm": 32.02743911743164, "learning_rate": 6.893203883495146e-06, "loss": 0.2475, "step": 19180 }, { "epoch": 185.99156626506024, "grad_norm": 37.587528228759766, "learning_rate": 6.888349514563107e-06, "loss": 0.1594, "step": 19181 }, { "epoch": 186.00722891566264, "grad_norm": 16.95940399169922, "learning_rate": 6.883495145631068e-06, "loss": 0.2508, "step": 19182 }, { "epoch": 186.01686746987951, "grad_norm": 18.73155975341797, "learning_rate": 6.8786407766990285e-06, "loss": 0.2166, "step": 19183 }, { "epoch": 186.0265060240964, "grad_norm": 6.856253623962402, "learning_rate": 6.873786407766991e-06, "loss": 0.4255, "step": 19184 }, { "epoch": 186.03614457831324, "grad_norm": 18.597015380859375, "learning_rate": 6.868932038834952e-06, "loss": 0.2269, "step": 19185 }, { "epoch": 186.04578313253012, "grad_norm": 5.428654193878174, "learning_rate": 6.864077669902913e-06, "loss": 0.2496, "step": 19186 }, { "epoch": 186.055421686747, "grad_norm": 2.8676226139068604, "learning_rate": 6.859223300970874e-06, "loss": 0.1932, "step": 19187 }, { "epoch": 186.06506024096385, "grad_norm": 27.20417594909668, "learning_rate": 6.854368932038835e-06, "loss": 0.3897, "step": 19188 }, { "epoch": 186.07469879518072, "grad_norm": 37.78248596191406, "learning_rate": 6.849514563106796e-06, "loss": 0.3267, "step": 19189 }, { "epoch": 186.0843373493976, "grad_norm": 10.653484344482422, "learning_rate": 6.844660194174758e-06, "loss": 0.2052, "step": 19190 }, { "epoch": 186.09397590361445, "grad_norm": 7.207540512084961, "learning_rate": 6.839805825242719e-06, "loss": 0.1258, "step": 19191 }, { "epoch": 186.10361445783133, "grad_norm": 14.466660499572754, "learning_rate": 6.83495145631068e-06, "loss": 0.1999, "step": 19192 }, { "epoch": 186.1132530120482, "grad_norm": 36.85952377319336, "learning_rate": 6.830097087378641e-06, "loss": 0.2973, "step": 19193 }, { "epoch": 186.12289156626505, "grad_norm": 11.817476272583008, "learning_rate": 6.825242718446602e-06, "loss": 0.2095, "step": 19194 }, { "epoch": 186.13253012048193, "grad_norm": 25.238969802856445, "learning_rate": 6.820388349514563e-06, "loss": 0.2267, "step": 19195 }, { "epoch": 186.1421686746988, "grad_norm": 7.415563106536865, "learning_rate": 6.815533980582525e-06, "loss": 0.2147, "step": 19196 }, { "epoch": 186.15180722891566, "grad_norm": 8.63006591796875, "learning_rate": 6.810679611650486e-06, "loss": 0.2403, "step": 19197 }, { "epoch": 186.16144578313254, "grad_norm": 70.52720642089844, "learning_rate": 6.805825242718447e-06, "loss": 0.3748, "step": 19198 }, { "epoch": 186.17108433734938, "grad_norm": 27.53571128845215, "learning_rate": 6.800970873786408e-06, "loss": 0.2672, "step": 19199 }, { "epoch": 186.18072289156626, "grad_norm": 80.90693664550781, "learning_rate": 6.796116504854369e-06, "loss": 0.2768, "step": 19200 }, { "epoch": 186.19036144578314, "grad_norm": 33.09213638305664, "learning_rate": 6.79126213592233e-06, "loss": 0.3597, "step": 19201 }, { "epoch": 186.2, "grad_norm": 35.54882049560547, "learning_rate": 6.786407766990291e-06, "loss": 0.2525, "step": 19202 }, { "epoch": 186.20963855421687, "grad_norm": 7.618610858917236, "learning_rate": 6.781553398058253e-06, "loss": 0.2361, "step": 19203 }, { "epoch": 186.21927710843374, "grad_norm": 34.21809005737305, "learning_rate": 6.776699029126214e-06, "loss": 0.1631, "step": 19204 }, { "epoch": 186.2289156626506, "grad_norm": 9.145566940307617, "learning_rate": 6.7718446601941754e-06, "loss": 0.2918, "step": 19205 }, { "epoch": 186.23855421686747, "grad_norm": 14.336812019348145, "learning_rate": 6.766990291262136e-06, "loss": 0.0975, "step": 19206 }, { "epoch": 186.24819277108435, "grad_norm": 9.494568824768066, "learning_rate": 6.762135922330097e-06, "loss": 0.2343, "step": 19207 }, { "epoch": 186.2578313253012, "grad_norm": 2.509716510772705, "learning_rate": 6.757281553398058e-06, "loss": 0.2219, "step": 19208 }, { "epoch": 186.26746987951807, "grad_norm": 18.427566528320312, "learning_rate": 6.75242718446602e-06, "loss": 0.3244, "step": 19209 }, { "epoch": 186.27710843373495, "grad_norm": 46.9473876953125, "learning_rate": 6.747572815533981e-06, "loss": 0.289, "step": 19210 }, { "epoch": 186.2867469879518, "grad_norm": 24.065086364746094, "learning_rate": 6.742718446601942e-06, "loss": 0.4469, "step": 19211 }, { "epoch": 186.29638554216868, "grad_norm": 9.699234008789062, "learning_rate": 6.7378640776699035e-06, "loss": 0.2429, "step": 19212 }, { "epoch": 186.30602409638556, "grad_norm": 9.627418518066406, "learning_rate": 6.733009708737864e-06, "loss": 0.1626, "step": 19213 }, { "epoch": 186.3156626506024, "grad_norm": 10.640632629394531, "learning_rate": 6.728155339805825e-06, "loss": 0.1794, "step": 19214 }, { "epoch": 186.32530120481928, "grad_norm": 3.867694854736328, "learning_rate": 6.723300970873788e-06, "loss": 0.177, "step": 19215 }, { "epoch": 186.33493975903613, "grad_norm": 10.976552963256836, "learning_rate": 6.718446601941748e-06, "loss": 0.4386, "step": 19216 }, { "epoch": 186.344578313253, "grad_norm": 11.47737979888916, "learning_rate": 6.713592233009709e-06, "loss": 0.318, "step": 19217 }, { "epoch": 186.3542168674699, "grad_norm": 25.372983932495117, "learning_rate": 6.70873786407767e-06, "loss": 0.2167, "step": 19218 }, { "epoch": 186.36385542168674, "grad_norm": 7.090587615966797, "learning_rate": 6.7038834951456315e-06, "loss": 0.0648, "step": 19219 }, { "epoch": 186.3734939759036, "grad_norm": 29.34809112548828, "learning_rate": 6.699029126213592e-06, "loss": 0.3559, "step": 19220 }, { "epoch": 186.3831325301205, "grad_norm": 6.587044715881348, "learning_rate": 6.694174757281555e-06, "loss": 0.1741, "step": 19221 }, { "epoch": 186.39277108433734, "grad_norm": 11.547751426696777, "learning_rate": 6.689320388349516e-06, "loss": 0.1156, "step": 19222 }, { "epoch": 186.40240963855422, "grad_norm": 4.6481523513793945, "learning_rate": 6.684466019417476e-06, "loss": 0.1678, "step": 19223 }, { "epoch": 186.4120481927711, "grad_norm": 8.322466850280762, "learning_rate": 6.679611650485437e-06, "loss": 0.1793, "step": 19224 }, { "epoch": 186.42168674698794, "grad_norm": 17.78087615966797, "learning_rate": 6.674757281553398e-06, "loss": 0.3596, "step": 19225 }, { "epoch": 186.43132530120482, "grad_norm": 10.687745094299316, "learning_rate": 6.6699029126213595e-06, "loss": 0.4924, "step": 19226 }, { "epoch": 186.4409638554217, "grad_norm": 11.20865249633789, "learning_rate": 6.66504854368932e-06, "loss": 0.2252, "step": 19227 }, { "epoch": 186.45060240963855, "grad_norm": 47.20060348510742, "learning_rate": 6.660194174757283e-06, "loss": 0.35, "step": 19228 }, { "epoch": 186.46024096385543, "grad_norm": 4.806543827056885, "learning_rate": 6.655339805825243e-06, "loss": 0.1525, "step": 19229 }, { "epoch": 186.46987951807228, "grad_norm": 7.495877265930176, "learning_rate": 6.650485436893204e-06, "loss": 0.1424, "step": 19230 }, { "epoch": 186.47951807228915, "grad_norm": 0.9051516652107239, "learning_rate": 6.645631067961165e-06, "loss": 0.1782, "step": 19231 }, { "epoch": 186.48915662650603, "grad_norm": 9.032970428466797, "learning_rate": 6.640776699029126e-06, "loss": 0.2602, "step": 19232 }, { "epoch": 186.49879518072288, "grad_norm": 5.258671283721924, "learning_rate": 6.6359223300970875e-06, "loss": 0.2533, "step": 19233 }, { "epoch": 186.50843373493976, "grad_norm": 71.14543914794922, "learning_rate": 6.6310679611650495e-06, "loss": 0.1868, "step": 19234 }, { "epoch": 186.51807228915663, "grad_norm": 10.956896781921387, "learning_rate": 6.626213592233011e-06, "loss": 0.1724, "step": 19235 }, { "epoch": 186.52771084337348, "grad_norm": 11.19184684753418, "learning_rate": 6.621359223300971e-06, "loss": 0.1842, "step": 19236 }, { "epoch": 186.53734939759036, "grad_norm": 53.780208587646484, "learning_rate": 6.616504854368932e-06, "loss": 0.2841, "step": 19237 }, { "epoch": 186.54698795180724, "grad_norm": 48.733428955078125, "learning_rate": 6.611650485436893e-06, "loss": 0.2454, "step": 19238 }, { "epoch": 186.5566265060241, "grad_norm": 11.083046913146973, "learning_rate": 6.606796116504854e-06, "loss": 0.181, "step": 19239 }, { "epoch": 186.56626506024097, "grad_norm": 9.26481819152832, "learning_rate": 6.601941747572816e-06, "loss": 0.291, "step": 19240 }, { "epoch": 186.57590361445784, "grad_norm": 21.02425193786621, "learning_rate": 6.5970873786407775e-06, "loss": 0.2408, "step": 19241 }, { "epoch": 186.5855421686747, "grad_norm": 3.983572244644165, "learning_rate": 6.592233009708739e-06, "loss": 0.129, "step": 19242 }, { "epoch": 186.59518072289157, "grad_norm": 16.091405868530273, "learning_rate": 6.587378640776699e-06, "loss": 0.2321, "step": 19243 }, { "epoch": 186.60481927710845, "grad_norm": 25.573741912841797, "learning_rate": 6.58252427184466e-06, "loss": 0.2564, "step": 19244 }, { "epoch": 186.6144578313253, "grad_norm": 11.661178588867188, "learning_rate": 6.577669902912621e-06, "loss": 0.277, "step": 19245 }, { "epoch": 186.62409638554217, "grad_norm": 3.6772289276123047, "learning_rate": 6.572815533980582e-06, "loss": 0.2787, "step": 19246 }, { "epoch": 186.63373493975902, "grad_norm": 17.533008575439453, "learning_rate": 6.567961165048544e-06, "loss": 0.1188, "step": 19247 }, { "epoch": 186.6433734939759, "grad_norm": 10.398303031921387, "learning_rate": 6.5631067961165056e-06, "loss": 0.2443, "step": 19248 }, { "epoch": 186.65301204819278, "grad_norm": 7.669848918914795, "learning_rate": 6.558252427184467e-06, "loss": 0.1761, "step": 19249 }, { "epoch": 186.66265060240963, "grad_norm": 41.05257034301758, "learning_rate": 6.553398058252427e-06, "loss": 0.1498, "step": 19250 }, { "epoch": 186.6722891566265, "grad_norm": 21.655542373657227, "learning_rate": 6.548543689320388e-06, "loss": 0.3329, "step": 19251 }, { "epoch": 186.68192771084338, "grad_norm": 3.9742114543914795, "learning_rate": 6.543689320388349e-06, "loss": 0.2938, "step": 19252 }, { "epoch": 186.69156626506023, "grad_norm": 3.8208210468292236, "learning_rate": 6.538834951456311e-06, "loss": 0.0996, "step": 19253 }, { "epoch": 186.7012048192771, "grad_norm": 36.740699768066406, "learning_rate": 6.5339805825242724e-06, "loss": 0.2903, "step": 19254 }, { "epoch": 186.710843373494, "grad_norm": 232.30152893066406, "learning_rate": 6.5291262135922336e-06, "loss": 0.2354, "step": 19255 }, { "epoch": 186.72048192771084, "grad_norm": 33.87814712524414, "learning_rate": 6.524271844660195e-06, "loss": 0.1945, "step": 19256 }, { "epoch": 186.7301204819277, "grad_norm": 86.19576263427734, "learning_rate": 6.519417475728155e-06, "loss": 0.237, "step": 19257 }, { "epoch": 186.7397590361446, "grad_norm": 8.719476699829102, "learning_rate": 6.514563106796116e-06, "loss": 0.2033, "step": 19258 }, { "epoch": 186.74939759036144, "grad_norm": 14.947023391723633, "learning_rate": 6.509708737864079e-06, "loss": 0.2357, "step": 19259 }, { "epoch": 186.75903614457832, "grad_norm": 18.262521743774414, "learning_rate": 6.504854368932039e-06, "loss": 0.261, "step": 19260 }, { "epoch": 186.76867469879517, "grad_norm": 9.018228530883789, "learning_rate": 6.5000000000000004e-06, "loss": 0.2288, "step": 19261 }, { "epoch": 186.77831325301204, "grad_norm": 11.316781044006348, "learning_rate": 6.495145631067962e-06, "loss": 0.2355, "step": 19262 }, { "epoch": 186.78795180722892, "grad_norm": 11.012410163879395, "learning_rate": 6.490291262135923e-06, "loss": 0.0683, "step": 19263 }, { "epoch": 186.79759036144577, "grad_norm": 6.790249824523926, "learning_rate": 6.485436893203883e-06, "loss": 0.2628, "step": 19264 }, { "epoch": 186.80722891566265, "grad_norm": 15.802300453186035, "learning_rate": 6.480582524271844e-06, "loss": 0.1883, "step": 19265 }, { "epoch": 186.81686746987953, "grad_norm": 14.897784233093262, "learning_rate": 6.475728155339807e-06, "loss": 0.2826, "step": 19266 }, { "epoch": 186.82650602409637, "grad_norm": 18.67487335205078, "learning_rate": 6.470873786407767e-06, "loss": 0.3305, "step": 19267 }, { "epoch": 186.83614457831325, "grad_norm": 2.2919373512268066, "learning_rate": 6.4660194174757285e-06, "loss": 0.1273, "step": 19268 }, { "epoch": 186.84578313253013, "grad_norm": 24.650209426879883, "learning_rate": 6.46116504854369e-06, "loss": 0.1932, "step": 19269 }, { "epoch": 186.85542168674698, "grad_norm": 7.816844940185547, "learning_rate": 6.456310679611651e-06, "loss": 0.1157, "step": 19270 }, { "epoch": 186.86506024096386, "grad_norm": 24.587976455688477, "learning_rate": 6.451456310679611e-06, "loss": 0.2373, "step": 19271 }, { "epoch": 186.87469879518073, "grad_norm": 15.078775405883789, "learning_rate": 6.446601941747574e-06, "loss": 0.1981, "step": 19272 }, { "epoch": 186.88433734939758, "grad_norm": 15.321355819702148, "learning_rate": 6.441747572815535e-06, "loss": 0.2005, "step": 19273 }, { "epoch": 186.89397590361446, "grad_norm": 5.286825180053711, "learning_rate": 6.436893203883495e-06, "loss": 0.2691, "step": 19274 }, { "epoch": 186.90361445783134, "grad_norm": 67.01105499267578, "learning_rate": 6.4320388349514565e-06, "loss": 0.2675, "step": 19275 }, { "epoch": 186.9132530120482, "grad_norm": 6.512877941131592, "learning_rate": 6.427184466019418e-06, "loss": 0.2256, "step": 19276 }, { "epoch": 186.92289156626506, "grad_norm": 4.704639434814453, "learning_rate": 6.422330097087379e-06, "loss": 0.2196, "step": 19277 }, { "epoch": 186.9325301204819, "grad_norm": 13.371848106384277, "learning_rate": 6.417475728155341e-06, "loss": 0.1813, "step": 19278 }, { "epoch": 186.9421686746988, "grad_norm": 23.555023193359375, "learning_rate": 6.412621359223302e-06, "loss": 0.2318, "step": 19279 }, { "epoch": 186.95180722891567, "grad_norm": 62.687896728515625, "learning_rate": 6.407766990291263e-06, "loss": 0.1897, "step": 19280 }, { "epoch": 186.96144578313252, "grad_norm": 19.843305587768555, "learning_rate": 6.402912621359223e-06, "loss": 0.3291, "step": 19281 }, { "epoch": 186.9710843373494, "grad_norm": 20.025419235229492, "learning_rate": 6.3980582524271845e-06, "loss": 0.2089, "step": 19282 }, { "epoch": 186.98072289156627, "grad_norm": 26.069904327392578, "learning_rate": 6.393203883495146e-06, "loss": 0.146, "step": 19283 }, { "epoch": 186.99036144578312, "grad_norm": 7.633580684661865, "learning_rate": 6.388349514563107e-06, "loss": 0.1709, "step": 19284 }, { "epoch": 187.00602409638554, "grad_norm": 5.907906532287598, "learning_rate": 6.383495145631069e-06, "loss": 0.1353, "step": 19285 }, { "epoch": 187.01566265060242, "grad_norm": 20.180810928344727, "learning_rate": 6.37864077669903e-06, "loss": 0.1946, "step": 19286 }, { "epoch": 187.02530120481927, "grad_norm": 35.159175872802734, "learning_rate": 6.37378640776699e-06, "loss": 0.2613, "step": 19287 }, { "epoch": 187.03493975903615, "grad_norm": 5.718818187713623, "learning_rate": 6.368932038834951e-06, "loss": 0.1146, "step": 19288 }, { "epoch": 187.044578313253, "grad_norm": 22.802621841430664, "learning_rate": 6.3640776699029125e-06, "loss": 0.2504, "step": 19289 }, { "epoch": 187.05421686746988, "grad_norm": 13.184568405151367, "learning_rate": 6.359223300970874e-06, "loss": 0.3204, "step": 19290 }, { "epoch": 187.06385542168675, "grad_norm": 9.281669616699219, "learning_rate": 6.354368932038836e-06, "loss": 0.361, "step": 19291 }, { "epoch": 187.0734939759036, "grad_norm": 8.905134201049805, "learning_rate": 6.349514563106797e-06, "loss": 0.2363, "step": 19292 }, { "epoch": 187.08313253012048, "grad_norm": 4.1002197265625, "learning_rate": 6.344660194174758e-06, "loss": 0.188, "step": 19293 }, { "epoch": 187.09277108433736, "grad_norm": 5.252307415008545, "learning_rate": 6.339805825242718e-06, "loss": 0.2523, "step": 19294 }, { "epoch": 187.1024096385542, "grad_norm": 24.305795669555664, "learning_rate": 6.334951456310679e-06, "loss": 0.2168, "step": 19295 }, { "epoch": 187.11204819277108, "grad_norm": 40.69053649902344, "learning_rate": 6.3300970873786405e-06, "loss": 0.1155, "step": 19296 }, { "epoch": 187.12168674698796, "grad_norm": 3.2619073390960693, "learning_rate": 6.3252427184466025e-06, "loss": 0.2189, "step": 19297 }, { "epoch": 187.1313253012048, "grad_norm": 41.5246696472168, "learning_rate": 6.320388349514564e-06, "loss": 0.3596, "step": 19298 }, { "epoch": 187.1409638554217, "grad_norm": 19.524450302124023, "learning_rate": 6.315533980582525e-06, "loss": 0.1631, "step": 19299 }, { "epoch": 187.15060240963857, "grad_norm": 24.330669403076172, "learning_rate": 6.310679611650486e-06, "loss": 0.2897, "step": 19300 }, { "epoch": 187.16024096385541, "grad_norm": 17.319318771362305, "learning_rate": 6.305825242718446e-06, "loss": 0.2629, "step": 19301 }, { "epoch": 187.1698795180723, "grad_norm": 24.29718780517578, "learning_rate": 6.300970873786407e-06, "loss": 0.3084, "step": 19302 }, { "epoch": 187.17951807228917, "grad_norm": 3.0683300495147705, "learning_rate": 6.2961165048543686e-06, "loss": 0.1971, "step": 19303 }, { "epoch": 187.18915662650602, "grad_norm": 18.93166160583496, "learning_rate": 6.2912621359223306e-06, "loss": 0.2919, "step": 19304 }, { "epoch": 187.1987951807229, "grad_norm": 2.973106622695923, "learning_rate": 6.286407766990292e-06, "loss": 0.1432, "step": 19305 }, { "epoch": 187.20843373493975, "grad_norm": 26.220792770385742, "learning_rate": 6.281553398058253e-06, "loss": 0.2619, "step": 19306 }, { "epoch": 187.21807228915662, "grad_norm": 4.742964744567871, "learning_rate": 6.276699029126214e-06, "loss": 0.1676, "step": 19307 }, { "epoch": 187.2277108433735, "grad_norm": 7.846517086029053, "learning_rate": 6.271844660194174e-06, "loss": 0.1658, "step": 19308 }, { "epoch": 187.23734939759035, "grad_norm": 7.08855676651001, "learning_rate": 6.2669902912621354e-06, "loss": 0.1924, "step": 19309 }, { "epoch": 187.24698795180723, "grad_norm": 33.2656135559082, "learning_rate": 6.262135922330098e-06, "loss": 0.3529, "step": 19310 }, { "epoch": 187.2566265060241, "grad_norm": 30.83928108215332, "learning_rate": 6.257281553398059e-06, "loss": 0.275, "step": 19311 }, { "epoch": 187.26626506024095, "grad_norm": 9.36247444152832, "learning_rate": 6.25242718446602e-06, "loss": 0.2944, "step": 19312 }, { "epoch": 187.27590361445783, "grad_norm": 62.264774322509766, "learning_rate": 6.247572815533981e-06, "loss": 0.3264, "step": 19313 }, { "epoch": 187.2855421686747, "grad_norm": 27.434764862060547, "learning_rate": 6.242718446601942e-06, "loss": 0.1826, "step": 19314 }, { "epoch": 187.29518072289156, "grad_norm": 14.98735523223877, "learning_rate": 6.237864077669903e-06, "loss": 0.1444, "step": 19315 }, { "epoch": 187.30481927710844, "grad_norm": 6.412567615509033, "learning_rate": 6.233009708737864e-06, "loss": 0.1692, "step": 19316 }, { "epoch": 187.3144578313253, "grad_norm": 21.403043746948242, "learning_rate": 6.2281553398058255e-06, "loss": 0.2577, "step": 19317 }, { "epoch": 187.32409638554216, "grad_norm": 23.093523025512695, "learning_rate": 6.223300970873787e-06, "loss": 0.1992, "step": 19318 }, { "epoch": 187.33373493975904, "grad_norm": 2.840660333633423, "learning_rate": 6.218446601941748e-06, "loss": 0.096, "step": 19319 }, { "epoch": 187.34337349397592, "grad_norm": 6.6225175857543945, "learning_rate": 6.213592233009709e-06, "loss": 0.1527, "step": 19320 }, { "epoch": 187.35301204819277, "grad_norm": 10.529970169067383, "learning_rate": 6.20873786407767e-06, "loss": 0.2852, "step": 19321 }, { "epoch": 187.36265060240964, "grad_norm": 5.6913371086120605, "learning_rate": 6.203883495145631e-06, "loss": 0.2764, "step": 19322 }, { "epoch": 187.3722891566265, "grad_norm": 15.234907150268555, "learning_rate": 6.199029126213592e-06, "loss": 0.2577, "step": 19323 }, { "epoch": 187.38192771084337, "grad_norm": 8.237438201904297, "learning_rate": 6.194174757281554e-06, "loss": 0.3271, "step": 19324 }, { "epoch": 187.39156626506025, "grad_norm": 10.290499687194824, "learning_rate": 6.189320388349515e-06, "loss": 0.1305, "step": 19325 }, { "epoch": 187.4012048192771, "grad_norm": 9.919015884399414, "learning_rate": 6.184466019417476e-06, "loss": 0.3182, "step": 19326 }, { "epoch": 187.41084337349398, "grad_norm": 15.247172355651855, "learning_rate": 6.179611650485438e-06, "loss": 0.099, "step": 19327 }, { "epoch": 187.42048192771085, "grad_norm": 30.377883911132812, "learning_rate": 6.174757281553398e-06, "loss": 0.2853, "step": 19328 }, { "epoch": 187.4301204819277, "grad_norm": 16.88422203063965, "learning_rate": 6.169902912621359e-06, "loss": 0.3088, "step": 19329 }, { "epoch": 187.43975903614458, "grad_norm": 4.611423969268799, "learning_rate": 6.165048543689321e-06, "loss": 0.1591, "step": 19330 }, { "epoch": 187.44939759036146, "grad_norm": 31.42401885986328, "learning_rate": 6.160194174757282e-06, "loss": 0.2864, "step": 19331 }, { "epoch": 187.4590361445783, "grad_norm": 25.922971725463867, "learning_rate": 6.155339805825243e-06, "loss": 0.2933, "step": 19332 }, { "epoch": 187.46867469879518, "grad_norm": 19.472583770751953, "learning_rate": 6.150485436893204e-06, "loss": 0.2788, "step": 19333 }, { "epoch": 187.47831325301206, "grad_norm": 27.263612747192383, "learning_rate": 6.145631067961166e-06, "loss": 0.1593, "step": 19334 }, { "epoch": 187.4879518072289, "grad_norm": 16.01263999938965, "learning_rate": 6.140776699029126e-06, "loss": 0.3494, "step": 19335 }, { "epoch": 187.4975903614458, "grad_norm": 46.84018325805664, "learning_rate": 6.135922330097087e-06, "loss": 0.0657, "step": 19336 }, { "epoch": 187.50722891566264, "grad_norm": 50.50551223754883, "learning_rate": 6.131067961165049e-06, "loss": 0.3452, "step": 19337 }, { "epoch": 187.51686746987951, "grad_norm": 5.664009094238281, "learning_rate": 6.12621359223301e-06, "loss": 0.3882, "step": 19338 }, { "epoch": 187.5265060240964, "grad_norm": 30.69504737854004, "learning_rate": 6.121359223300971e-06, "loss": 0.4219, "step": 19339 }, { "epoch": 187.53614457831324, "grad_norm": 56.7852668762207, "learning_rate": 6.116504854368933e-06, "loss": 0.3173, "step": 19340 }, { "epoch": 187.54578313253012, "grad_norm": 25.120267868041992, "learning_rate": 6.111650485436894e-06, "loss": 0.2908, "step": 19341 }, { "epoch": 187.555421686747, "grad_norm": 13.051855087280273, "learning_rate": 6.106796116504854e-06, "loss": 0.1269, "step": 19342 }, { "epoch": 187.56506024096385, "grad_norm": 18.229999542236328, "learning_rate": 6.101941747572816e-06, "loss": 0.183, "step": 19343 }, { "epoch": 187.57469879518072, "grad_norm": 7.710931777954102, "learning_rate": 6.097087378640777e-06, "loss": 0.2452, "step": 19344 }, { "epoch": 187.5843373493976, "grad_norm": 28.382936477661133, "learning_rate": 6.0922330097087375e-06, "loss": 0.2752, "step": 19345 }, { "epoch": 187.59397590361445, "grad_norm": 3.724564790725708, "learning_rate": 6.0873786407766995e-06, "loss": 0.2007, "step": 19346 }, { "epoch": 187.60361445783133, "grad_norm": 23.413530349731445, "learning_rate": 6.082524271844661e-06, "loss": 0.1913, "step": 19347 }, { "epoch": 187.6132530120482, "grad_norm": 4.599964618682861, "learning_rate": 6.077669902912622e-06, "loss": 0.1211, "step": 19348 }, { "epoch": 187.62289156626505, "grad_norm": 10.196496963500977, "learning_rate": 6.072815533980583e-06, "loss": 0.2081, "step": 19349 }, { "epoch": 187.63253012048193, "grad_norm": 2.9412267208099365, "learning_rate": 6.067961165048544e-06, "loss": 0.1487, "step": 19350 }, { "epoch": 187.6421686746988, "grad_norm": 15.68066692352295, "learning_rate": 6.063106796116505e-06, "loss": 0.243, "step": 19351 }, { "epoch": 187.65180722891566, "grad_norm": 28.596471786499023, "learning_rate": 6.058252427184466e-06, "loss": 0.0812, "step": 19352 }, { "epoch": 187.66144578313254, "grad_norm": 9.901033401489258, "learning_rate": 6.0533980582524275e-06, "loss": 0.0991, "step": 19353 }, { "epoch": 187.67108433734938, "grad_norm": 18.757444381713867, "learning_rate": 6.048543689320389e-06, "loss": 0.2836, "step": 19354 }, { "epoch": 187.68072289156626, "grad_norm": 14.80396842956543, "learning_rate": 6.04368932038835e-06, "loss": 0.2038, "step": 19355 }, { "epoch": 187.69036144578314, "grad_norm": 19.008464813232422, "learning_rate": 6.038834951456311e-06, "loss": 0.3306, "step": 19356 }, { "epoch": 187.7, "grad_norm": 10.99099063873291, "learning_rate": 6.033980582524272e-06, "loss": 0.256, "step": 19357 }, { "epoch": 187.70963855421687, "grad_norm": 15.08319091796875, "learning_rate": 6.029126213592233e-06, "loss": 0.3911, "step": 19358 }, { "epoch": 187.71927710843374, "grad_norm": 50.675411224365234, "learning_rate": 6.024271844660194e-06, "loss": 0.3869, "step": 19359 }, { "epoch": 187.7289156626506, "grad_norm": 9.313376426696777, "learning_rate": 6.0194174757281556e-06, "loss": 0.3103, "step": 19360 }, { "epoch": 187.73855421686747, "grad_norm": 38.03263854980469, "learning_rate": 6.014563106796117e-06, "loss": 0.1985, "step": 19361 }, { "epoch": 187.74819277108435, "grad_norm": 29.635215759277344, "learning_rate": 6.009708737864078e-06, "loss": 0.2337, "step": 19362 }, { "epoch": 187.7578313253012, "grad_norm": 5.627108573913574, "learning_rate": 6.004854368932039e-06, "loss": 0.1036, "step": 19363 }, { "epoch": 187.76746987951807, "grad_norm": 16.194725036621094, "learning_rate": 6e-06, "loss": 0.2297, "step": 19364 }, { "epoch": 187.77710843373495, "grad_norm": 6.325870990753174, "learning_rate": 5.995145631067961e-06, "loss": 0.0451, "step": 19365 }, { "epoch": 187.7867469879518, "grad_norm": 3.145735263824463, "learning_rate": 5.9902912621359224e-06, "loss": 0.1369, "step": 19366 }, { "epoch": 187.79638554216868, "grad_norm": 15.996419906616211, "learning_rate": 5.985436893203884e-06, "loss": 0.2708, "step": 19367 }, { "epoch": 187.80602409638556, "grad_norm": 32.61451721191406, "learning_rate": 5.980582524271846e-06, "loss": 0.2293, "step": 19368 }, { "epoch": 187.8156626506024, "grad_norm": 6.481442451477051, "learning_rate": 5.975728155339806e-06, "loss": 0.3456, "step": 19369 }, { "epoch": 187.82530120481928, "grad_norm": 9.215289115905762, "learning_rate": 5.970873786407767e-06, "loss": 0.1733, "step": 19370 }, { "epoch": 187.83493975903613, "grad_norm": 5.251705169677734, "learning_rate": 5.966019417475729e-06, "loss": 0.0922, "step": 19371 }, { "epoch": 187.844578313253, "grad_norm": 13.654648780822754, "learning_rate": 5.961165048543689e-06, "loss": 0.3898, "step": 19372 }, { "epoch": 187.8542168674699, "grad_norm": 67.51235961914062, "learning_rate": 5.9563106796116505e-06, "loss": 0.2375, "step": 19373 }, { "epoch": 187.86385542168674, "grad_norm": 17.190921783447266, "learning_rate": 5.951456310679612e-06, "loss": 0.1505, "step": 19374 }, { "epoch": 187.8734939759036, "grad_norm": 13.938578605651855, "learning_rate": 5.946601941747574e-06, "loss": 0.219, "step": 19375 }, { "epoch": 187.8831325301205, "grad_norm": 5.073235511779785, "learning_rate": 5.941747572815534e-06, "loss": 0.1178, "step": 19376 }, { "epoch": 187.89277108433734, "grad_norm": 21.73090362548828, "learning_rate": 5.936893203883495e-06, "loss": 0.372, "step": 19377 }, { "epoch": 187.90240963855422, "grad_norm": 15.790789604187012, "learning_rate": 5.932038834951457e-06, "loss": 0.3022, "step": 19378 }, { "epoch": 187.9120481927711, "grad_norm": 5.235726833343506, "learning_rate": 5.927184466019417e-06, "loss": 0.0478, "step": 19379 }, { "epoch": 187.92168674698794, "grad_norm": 9.315749168395996, "learning_rate": 5.9223300970873785e-06, "loss": 0.1654, "step": 19380 }, { "epoch": 187.93132530120482, "grad_norm": 35.097713470458984, "learning_rate": 5.9174757281553405e-06, "loss": 0.2725, "step": 19381 }, { "epoch": 187.9409638554217, "grad_norm": 3.8693079948425293, "learning_rate": 5.912621359223302e-06, "loss": 0.1013, "step": 19382 }, { "epoch": 187.95060240963855, "grad_norm": 11.312775611877441, "learning_rate": 5.907766990291262e-06, "loss": 0.2429, "step": 19383 }, { "epoch": 187.96024096385543, "grad_norm": 45.86812210083008, "learning_rate": 5.902912621359224e-06, "loss": 0.4172, "step": 19384 }, { "epoch": 187.96987951807228, "grad_norm": 8.90734577178955, "learning_rate": 5.898058252427185e-06, "loss": 0.1479, "step": 19385 }, { "epoch": 187.97951807228915, "grad_norm": 4.009374618530273, "learning_rate": 5.893203883495145e-06, "loss": 0.1638, "step": 19386 }, { "epoch": 187.98915662650603, "grad_norm": 7.101281642913818, "learning_rate": 5.888349514563107e-06, "loss": 0.1825, "step": 19387 }, { "epoch": 188.00481927710842, "grad_norm": 35.168785095214844, "learning_rate": 5.8834951456310685e-06, "loss": 0.539, "step": 19388 }, { "epoch": 188.0144578313253, "grad_norm": 14.886653900146484, "learning_rate": 5.87864077669903e-06, "loss": 0.2965, "step": 19389 }, { "epoch": 188.02409638554218, "grad_norm": 7.089454174041748, "learning_rate": 5.873786407766991e-06, "loss": 0.1364, "step": 19390 }, { "epoch": 188.03373493975903, "grad_norm": 16.389650344848633, "learning_rate": 5.868932038834952e-06, "loss": 0.4205, "step": 19391 }, { "epoch": 188.0433734939759, "grad_norm": 6.681454181671143, "learning_rate": 5.864077669902913e-06, "loss": 0.1606, "step": 19392 }, { "epoch": 188.05301204819278, "grad_norm": 26.948272705078125, "learning_rate": 5.859223300970874e-06, "loss": 0.3566, "step": 19393 }, { "epoch": 188.06265060240963, "grad_norm": 12.406915664672852, "learning_rate": 5.854368932038835e-06, "loss": 0.3465, "step": 19394 }, { "epoch": 188.0722891566265, "grad_norm": 8.319046974182129, "learning_rate": 5.8495145631067965e-06, "loss": 0.2754, "step": 19395 }, { "epoch": 188.08192771084336, "grad_norm": 7.431244850158691, "learning_rate": 5.844660194174758e-06, "loss": 0.2381, "step": 19396 }, { "epoch": 188.09156626506024, "grad_norm": 10.322868347167969, "learning_rate": 5.839805825242719e-06, "loss": 0.1934, "step": 19397 }, { "epoch": 188.10120481927711, "grad_norm": 24.30179214477539, "learning_rate": 5.83495145631068e-06, "loss": 0.2504, "step": 19398 }, { "epoch": 188.11084337349396, "grad_norm": 8.637632369995117, "learning_rate": 5.830097087378641e-06, "loss": 0.2163, "step": 19399 }, { "epoch": 188.12048192771084, "grad_norm": 9.565218925476074, "learning_rate": 5.825242718446602e-06, "loss": 0.3268, "step": 19400 }, { "epoch": 188.13012048192772, "grad_norm": 8.595006942749023, "learning_rate": 5.820388349514563e-06, "loss": 0.0986, "step": 19401 }, { "epoch": 188.13975903614457, "grad_norm": 90.9242172241211, "learning_rate": 5.8155339805825245e-06, "loss": 0.3274, "step": 19402 }, { "epoch": 188.14939759036145, "grad_norm": 9.613030433654785, "learning_rate": 5.810679611650486e-06, "loss": 0.2266, "step": 19403 }, { "epoch": 188.15903614457832, "grad_norm": 18.75973129272461, "learning_rate": 5.805825242718447e-06, "loss": 0.4491, "step": 19404 }, { "epoch": 188.16867469879517, "grad_norm": 6.01456356048584, "learning_rate": 5.800970873786408e-06, "loss": 0.249, "step": 19405 }, { "epoch": 188.17831325301205, "grad_norm": 13.749741554260254, "learning_rate": 5.796116504854369e-06, "loss": 0.2965, "step": 19406 }, { "epoch": 188.18795180722893, "grad_norm": 8.048840522766113, "learning_rate": 5.79126213592233e-06, "loss": 0.3397, "step": 19407 }, { "epoch": 188.19759036144578, "grad_norm": 32.586116790771484, "learning_rate": 5.786407766990291e-06, "loss": 0.2165, "step": 19408 }, { "epoch": 188.20722891566265, "grad_norm": 18.116107940673828, "learning_rate": 5.781553398058253e-06, "loss": 0.2027, "step": 19409 }, { "epoch": 188.21686746987953, "grad_norm": 6.047267436981201, "learning_rate": 5.776699029126214e-06, "loss": 0.1725, "step": 19410 }, { "epoch": 188.22650602409638, "grad_norm": 23.631254196166992, "learning_rate": 5.771844660194175e-06, "loss": 0.136, "step": 19411 }, { "epoch": 188.23614457831326, "grad_norm": 11.955794334411621, "learning_rate": 5.766990291262137e-06, "loss": 0.3396, "step": 19412 }, { "epoch": 188.2457831325301, "grad_norm": 1.722164273262024, "learning_rate": 5.762135922330097e-06, "loss": 0.144, "step": 19413 }, { "epoch": 188.25542168674698, "grad_norm": 41.336849212646484, "learning_rate": 5.757281553398058e-06, "loss": 0.3168, "step": 19414 }, { "epoch": 188.26506024096386, "grad_norm": 7.747739791870117, "learning_rate": 5.7524271844660194e-06, "loss": 0.2255, "step": 19415 }, { "epoch": 188.2746987951807, "grad_norm": 40.89857864379883, "learning_rate": 5.747572815533981e-06, "loss": 0.0982, "step": 19416 }, { "epoch": 188.2843373493976, "grad_norm": 9.975262641906738, "learning_rate": 5.742718446601942e-06, "loss": 0.1959, "step": 19417 }, { "epoch": 188.29397590361447, "grad_norm": 16.217700958251953, "learning_rate": 5.737864077669903e-06, "loss": 0.2308, "step": 19418 }, { "epoch": 188.30361445783132, "grad_norm": 8.526803016662598, "learning_rate": 5.733009708737865e-06, "loss": 0.4301, "step": 19419 }, { "epoch": 188.3132530120482, "grad_norm": 7.358209133148193, "learning_rate": 5.728155339805825e-06, "loss": 0.1918, "step": 19420 }, { "epoch": 188.32289156626507, "grad_norm": 31.739776611328125, "learning_rate": 5.723300970873786e-06, "loss": 0.2955, "step": 19421 }, { "epoch": 188.33253012048192, "grad_norm": 9.555922508239746, "learning_rate": 5.718446601941748e-06, "loss": 0.3936, "step": 19422 }, { "epoch": 188.3421686746988, "grad_norm": 12.136212348937988, "learning_rate": 5.713592233009709e-06, "loss": 0.4683, "step": 19423 }, { "epoch": 188.35180722891567, "grad_norm": 16.252609252929688, "learning_rate": 5.70873786407767e-06, "loss": 0.2493, "step": 19424 }, { "epoch": 188.36144578313252, "grad_norm": 44.496665954589844, "learning_rate": 5.703883495145632e-06, "loss": 0.2421, "step": 19425 }, { "epoch": 188.3710843373494, "grad_norm": 5.974782943725586, "learning_rate": 5.699029126213593e-06, "loss": 0.1415, "step": 19426 }, { "epoch": 188.38072289156628, "grad_norm": 15.547737121582031, "learning_rate": 5.694174757281553e-06, "loss": 0.2594, "step": 19427 }, { "epoch": 188.39036144578313, "grad_norm": 12.680760383605957, "learning_rate": 5.689320388349515e-06, "loss": 0.352, "step": 19428 }, { "epoch": 188.4, "grad_norm": 39.64679718017578, "learning_rate": 5.684466019417476e-06, "loss": 0.5018, "step": 19429 }, { "epoch": 188.40963855421685, "grad_norm": 58.828887939453125, "learning_rate": 5.679611650485437e-06, "loss": 0.3821, "step": 19430 }, { "epoch": 188.41927710843373, "grad_norm": 41.73382568359375, "learning_rate": 5.674757281553399e-06, "loss": 0.4087, "step": 19431 }, { "epoch": 188.4289156626506, "grad_norm": 9.591926574707031, "learning_rate": 5.66990291262136e-06, "loss": 0.1072, "step": 19432 }, { "epoch": 188.43855421686746, "grad_norm": 15.63100528717041, "learning_rate": 5.665048543689321e-06, "loss": 0.3249, "step": 19433 }, { "epoch": 188.44819277108434, "grad_norm": 19.72583770751953, "learning_rate": 5.660194174757282e-06, "loss": 0.2368, "step": 19434 }, { "epoch": 188.4578313253012, "grad_norm": 43.1203727722168, "learning_rate": 5.655339805825243e-06, "loss": 0.2343, "step": 19435 }, { "epoch": 188.46746987951806, "grad_norm": 11.25355339050293, "learning_rate": 5.650485436893204e-06, "loss": 0.3966, "step": 19436 }, { "epoch": 188.47710843373494, "grad_norm": 4.624638080596924, "learning_rate": 5.645631067961165e-06, "loss": 0.0522, "step": 19437 }, { "epoch": 188.48674698795182, "grad_norm": 10.830960273742676, "learning_rate": 5.640776699029127e-06, "loss": 0.2724, "step": 19438 }, { "epoch": 188.49638554216867, "grad_norm": 8.58358097076416, "learning_rate": 5.635922330097088e-06, "loss": 0.3167, "step": 19439 }, { "epoch": 188.50602409638554, "grad_norm": 39.392887115478516, "learning_rate": 5.631067961165049e-06, "loss": 0.3109, "step": 19440 }, { "epoch": 188.51566265060242, "grad_norm": 9.196938514709473, "learning_rate": 5.62621359223301e-06, "loss": 0.3118, "step": 19441 }, { "epoch": 188.52530120481927, "grad_norm": 18.965091705322266, "learning_rate": 5.621359223300971e-06, "loss": 0.1902, "step": 19442 }, { "epoch": 188.53493975903615, "grad_norm": 39.37907791137695, "learning_rate": 5.616504854368932e-06, "loss": 0.2977, "step": 19443 }, { "epoch": 188.544578313253, "grad_norm": 9.278117179870605, "learning_rate": 5.6116504854368935e-06, "loss": 0.3094, "step": 19444 }, { "epoch": 188.55421686746988, "grad_norm": 26.676280975341797, "learning_rate": 5.606796116504855e-06, "loss": 0.3892, "step": 19445 }, { "epoch": 188.56385542168675, "grad_norm": 11.518214225769043, "learning_rate": 5.601941747572816e-06, "loss": 0.2136, "step": 19446 }, { "epoch": 188.5734939759036, "grad_norm": 11.317737579345703, "learning_rate": 5.597087378640777e-06, "loss": 0.4896, "step": 19447 }, { "epoch": 188.58313253012048, "grad_norm": 10.145767211914062, "learning_rate": 5.592233009708738e-06, "loss": 0.2172, "step": 19448 }, { "epoch": 188.59277108433736, "grad_norm": 11.035655975341797, "learning_rate": 5.587378640776699e-06, "loss": 0.3822, "step": 19449 }, { "epoch": 188.6024096385542, "grad_norm": 49.8834228515625, "learning_rate": 5.58252427184466e-06, "loss": 0.1896, "step": 19450 }, { "epoch": 188.61204819277108, "grad_norm": 55.72194290161133, "learning_rate": 5.5776699029126215e-06, "loss": 0.5928, "step": 19451 }, { "epoch": 188.62168674698796, "grad_norm": 17.940723419189453, "learning_rate": 5.572815533980583e-06, "loss": 0.1571, "step": 19452 }, { "epoch": 188.6313253012048, "grad_norm": 8.357542037963867, "learning_rate": 5.567961165048545e-06, "loss": 0.1465, "step": 19453 }, { "epoch": 188.6409638554217, "grad_norm": 33.97183609008789, "learning_rate": 5.563106796116505e-06, "loss": 0.2659, "step": 19454 }, { "epoch": 188.65060240963857, "grad_norm": 23.09076499938965, "learning_rate": 5.558252427184466e-06, "loss": 0.2065, "step": 19455 }, { "epoch": 188.66024096385541, "grad_norm": 5.351449012756348, "learning_rate": 5.553398058252427e-06, "loss": 0.2532, "step": 19456 }, { "epoch": 188.6698795180723, "grad_norm": 12.266133308410645, "learning_rate": 5.548543689320388e-06, "loss": 0.3082, "step": 19457 }, { "epoch": 188.67951807228917, "grad_norm": 9.242572784423828, "learning_rate": 5.5436893203883495e-06, "loss": 0.2991, "step": 19458 }, { "epoch": 188.68915662650602, "grad_norm": 12.542923927307129, "learning_rate": 5.538834951456311e-06, "loss": 0.2758, "step": 19459 }, { "epoch": 188.6987951807229, "grad_norm": 8.845258712768555, "learning_rate": 5.533980582524273e-06, "loss": 0.1514, "step": 19460 }, { "epoch": 188.70843373493975, "grad_norm": 52.76525115966797, "learning_rate": 5.529126213592233e-06, "loss": 0.2005, "step": 19461 }, { "epoch": 188.71807228915662, "grad_norm": 5.925133228302002, "learning_rate": 5.524271844660194e-06, "loss": 0.2956, "step": 19462 }, { "epoch": 188.7277108433735, "grad_norm": 3.3782856464385986, "learning_rate": 5.519417475728156e-06, "loss": 0.1693, "step": 19463 }, { "epoch": 188.73734939759035, "grad_norm": 3.650343894958496, "learning_rate": 5.514563106796116e-06, "loss": 0.2627, "step": 19464 }, { "epoch": 188.74698795180723, "grad_norm": 21.229278564453125, "learning_rate": 5.5097087378640776e-06, "loss": 0.1364, "step": 19465 }, { "epoch": 188.7566265060241, "grad_norm": 13.84349250793457, "learning_rate": 5.5048543689320396e-06, "loss": 0.2816, "step": 19466 }, { "epoch": 188.76626506024095, "grad_norm": 8.217735290527344, "learning_rate": 5.500000000000001e-06, "loss": 0.2726, "step": 19467 }, { "epoch": 188.77590361445783, "grad_norm": 10.041627883911133, "learning_rate": 5.495145631067961e-06, "loss": 0.274, "step": 19468 }, { "epoch": 188.7855421686747, "grad_norm": 26.847150802612305, "learning_rate": 5.490291262135923e-06, "loss": 0.3994, "step": 19469 }, { "epoch": 188.79518072289156, "grad_norm": 46.41250228881836, "learning_rate": 5.485436893203884e-06, "loss": 0.1876, "step": 19470 }, { "epoch": 188.80481927710844, "grad_norm": 3.9415643215179443, "learning_rate": 5.4805825242718444e-06, "loss": 0.1962, "step": 19471 }, { "epoch": 188.8144578313253, "grad_norm": 3.6542625427246094, "learning_rate": 5.4757281553398064e-06, "loss": 0.3022, "step": 19472 }, { "epoch": 188.82409638554216, "grad_norm": 59.05925369262695, "learning_rate": 5.470873786407768e-06, "loss": 0.1499, "step": 19473 }, { "epoch": 188.83373493975904, "grad_norm": 3.2410473823547363, "learning_rate": 5.466019417475728e-06, "loss": 0.2086, "step": 19474 }, { "epoch": 188.84337349397592, "grad_norm": 14.255228042602539, "learning_rate": 5.461165048543689e-06, "loss": 0.165, "step": 19475 }, { "epoch": 188.85301204819277, "grad_norm": 18.273698806762695, "learning_rate": 5.456310679611651e-06, "loss": 0.2284, "step": 19476 }, { "epoch": 188.86265060240964, "grad_norm": 6.626364707946777, "learning_rate": 5.451456310679612e-06, "loss": 0.1399, "step": 19477 }, { "epoch": 188.8722891566265, "grad_norm": 18.421537399291992, "learning_rate": 5.4466019417475725e-06, "loss": 0.4953, "step": 19478 }, { "epoch": 188.88192771084337, "grad_norm": 29.530324935913086, "learning_rate": 5.4417475728155345e-06, "loss": 0.3732, "step": 19479 }, { "epoch": 188.89156626506025, "grad_norm": 12.064749717712402, "learning_rate": 5.436893203883496e-06, "loss": 0.2541, "step": 19480 }, { "epoch": 188.9012048192771, "grad_norm": 2.869965076446533, "learning_rate": 5.432038834951456e-06, "loss": 0.1641, "step": 19481 }, { "epoch": 188.91084337349398, "grad_norm": 15.246346473693848, "learning_rate": 5.427184466019418e-06, "loss": 0.2794, "step": 19482 }, { "epoch": 188.92048192771085, "grad_norm": 15.255586624145508, "learning_rate": 5.422330097087379e-06, "loss": 0.2539, "step": 19483 }, { "epoch": 188.9301204819277, "grad_norm": 8.305455207824707, "learning_rate": 5.41747572815534e-06, "loss": 0.2505, "step": 19484 }, { "epoch": 188.93975903614458, "grad_norm": 36.39578628540039, "learning_rate": 5.412621359223301e-06, "loss": 0.2004, "step": 19485 }, { "epoch": 188.94939759036146, "grad_norm": 4.462472438812256, "learning_rate": 5.4077669902912625e-06, "loss": 0.1616, "step": 19486 }, { "epoch": 188.9590361445783, "grad_norm": 4.162225246429443, "learning_rate": 5.402912621359224e-06, "loss": 0.1659, "step": 19487 }, { "epoch": 188.96867469879518, "grad_norm": 10.822652816772461, "learning_rate": 5.398058252427185e-06, "loss": 0.1141, "step": 19488 }, { "epoch": 188.97831325301206, "grad_norm": 6.8714280128479, "learning_rate": 5.393203883495146e-06, "loss": 0.1934, "step": 19489 }, { "epoch": 188.9879518072289, "grad_norm": 22.251237869262695, "learning_rate": 5.388349514563107e-06, "loss": 0.1854, "step": 19490 }, { "epoch": 189.00361445783133, "grad_norm": 11.285778999328613, "learning_rate": 5.383495145631068e-06, "loss": 0.2063, "step": 19491 }, { "epoch": 189.01325301204818, "grad_norm": 11.763338088989258, "learning_rate": 5.378640776699029e-06, "loss": 0.3442, "step": 19492 }, { "epoch": 189.02289156626506, "grad_norm": 7.322572231292725, "learning_rate": 5.3737864077669905e-06, "loss": 0.2204, "step": 19493 }, { "epoch": 189.03253012048194, "grad_norm": 81.39193725585938, "learning_rate": 5.368932038834952e-06, "loss": 0.2543, "step": 19494 }, { "epoch": 189.0421686746988, "grad_norm": 4.954444885253906, "learning_rate": 5.364077669902913e-06, "loss": 0.1508, "step": 19495 }, { "epoch": 189.05180722891566, "grad_norm": 11.218199729919434, "learning_rate": 5.359223300970874e-06, "loss": 0.2079, "step": 19496 }, { "epoch": 189.06144578313254, "grad_norm": 6.503544807434082, "learning_rate": 5.354368932038835e-06, "loss": 0.2183, "step": 19497 }, { "epoch": 189.0710843373494, "grad_norm": 6.143670082092285, "learning_rate": 5.349514563106796e-06, "loss": 0.1958, "step": 19498 }, { "epoch": 189.08072289156627, "grad_norm": 16.804170608520508, "learning_rate": 5.344660194174757e-06, "loss": 0.3161, "step": 19499 }, { "epoch": 189.09036144578315, "grad_norm": 14.55984878540039, "learning_rate": 5.3398058252427185e-06, "loss": 0.4776, "step": 19500 }, { "epoch": 189.1, "grad_norm": 15.45033073425293, "learning_rate": 5.33495145631068e-06, "loss": 0.208, "step": 19501 }, { "epoch": 189.10963855421687, "grad_norm": 21.831281661987305, "learning_rate": 5.330097087378641e-06, "loss": 0.1456, "step": 19502 }, { "epoch": 189.11927710843372, "grad_norm": 7.500429153442383, "learning_rate": 5.325242718446602e-06, "loss": 0.2019, "step": 19503 }, { "epoch": 189.1289156626506, "grad_norm": 8.214102745056152, "learning_rate": 5.320388349514564e-06, "loss": 0.3502, "step": 19504 }, { "epoch": 189.13855421686748, "grad_norm": 24.655603408813477, "learning_rate": 5.315533980582524e-06, "loss": 0.254, "step": 19505 }, { "epoch": 189.14819277108433, "grad_norm": 5.121215343475342, "learning_rate": 5.310679611650485e-06, "loss": 0.1471, "step": 19506 }, { "epoch": 189.1578313253012, "grad_norm": 19.62297248840332, "learning_rate": 5.305825242718447e-06, "loss": 0.3865, "step": 19507 }, { "epoch": 189.16746987951808, "grad_norm": 11.299230575561523, "learning_rate": 5.300970873786408e-06, "loss": 0.2041, "step": 19508 }, { "epoch": 189.17710843373493, "grad_norm": 4.040712833404541, "learning_rate": 5.296116504854369e-06, "loss": 0.1894, "step": 19509 }, { "epoch": 189.1867469879518, "grad_norm": 5.810382843017578, "learning_rate": 5.291262135922331e-06, "loss": 0.2193, "step": 19510 }, { "epoch": 189.19638554216868, "grad_norm": 10.898507118225098, "learning_rate": 5.286407766990292e-06, "loss": 0.1718, "step": 19511 }, { "epoch": 189.20602409638553, "grad_norm": 31.02064323425293, "learning_rate": 5.281553398058252e-06, "loss": 0.3969, "step": 19512 }, { "epoch": 189.2156626506024, "grad_norm": 63.99892044067383, "learning_rate": 5.276699029126214e-06, "loss": 0.269, "step": 19513 }, { "epoch": 189.2253012048193, "grad_norm": 23.033100128173828, "learning_rate": 5.271844660194175e-06, "loss": 0.2143, "step": 19514 }, { "epoch": 189.23493975903614, "grad_norm": 20.076175689697266, "learning_rate": 5.266990291262136e-06, "loss": 0.3182, "step": 19515 }, { "epoch": 189.24457831325302, "grad_norm": 37.18571472167969, "learning_rate": 5.262135922330097e-06, "loss": 0.1967, "step": 19516 }, { "epoch": 189.2542168674699, "grad_norm": 4.94203519821167, "learning_rate": 5.257281553398059e-06, "loss": 0.1856, "step": 19517 }, { "epoch": 189.26385542168674, "grad_norm": 5.618020534515381, "learning_rate": 5.25242718446602e-06, "loss": 0.168, "step": 19518 }, { "epoch": 189.27349397590362, "grad_norm": 17.256649017333984, "learning_rate": 5.24757281553398e-06, "loss": 0.3907, "step": 19519 }, { "epoch": 189.28313253012047, "grad_norm": 12.711987495422363, "learning_rate": 5.242718446601942e-06, "loss": 0.2543, "step": 19520 }, { "epoch": 189.29277108433735, "grad_norm": 12.811088562011719, "learning_rate": 5.237864077669903e-06, "loss": 0.2236, "step": 19521 }, { "epoch": 189.30240963855422, "grad_norm": 8.642555236816406, "learning_rate": 5.233009708737864e-06, "loss": 0.178, "step": 19522 }, { "epoch": 189.31204819277107, "grad_norm": 5.325251579284668, "learning_rate": 5.228155339805826e-06, "loss": 0.1393, "step": 19523 }, { "epoch": 189.32168674698795, "grad_norm": 16.432159423828125, "learning_rate": 5.223300970873787e-06, "loss": 0.1632, "step": 19524 }, { "epoch": 189.33132530120483, "grad_norm": 6.630801677703857, "learning_rate": 5.218446601941748e-06, "loss": 0.2356, "step": 19525 }, { "epoch": 189.34096385542168, "grad_norm": 21.432544708251953, "learning_rate": 5.213592233009709e-06, "loss": 0.1682, "step": 19526 }, { "epoch": 189.35060240963855, "grad_norm": 3.4826745986938477, "learning_rate": 5.20873786407767e-06, "loss": 0.1223, "step": 19527 }, { "epoch": 189.36024096385543, "grad_norm": 8.433225631713867, "learning_rate": 5.2038834951456314e-06, "loss": 0.2181, "step": 19528 }, { "epoch": 189.36987951807228, "grad_norm": 25.631328582763672, "learning_rate": 5.199029126213593e-06, "loss": 0.2547, "step": 19529 }, { "epoch": 189.37951807228916, "grad_norm": 7.08709716796875, "learning_rate": 5.194174757281554e-06, "loss": 0.2634, "step": 19530 }, { "epoch": 189.38915662650604, "grad_norm": 14.034481048583984, "learning_rate": 5.189320388349515e-06, "loss": 0.1598, "step": 19531 }, { "epoch": 189.39879518072289, "grad_norm": 5.144032001495361, "learning_rate": 5.184466019417476e-06, "loss": 0.1098, "step": 19532 }, { "epoch": 189.40843373493976, "grad_norm": 17.665382385253906, "learning_rate": 5.179611650485437e-06, "loss": 0.23, "step": 19533 }, { "epoch": 189.41807228915664, "grad_norm": 48.4833869934082, "learning_rate": 5.174757281553398e-06, "loss": 0.1597, "step": 19534 }, { "epoch": 189.4277108433735, "grad_norm": 8.572890281677246, "learning_rate": 5.1699029126213595e-06, "loss": 0.187, "step": 19535 }, { "epoch": 189.43734939759037, "grad_norm": 5.930200099945068, "learning_rate": 5.165048543689321e-06, "loss": 0.3387, "step": 19536 }, { "epoch": 189.44698795180722, "grad_norm": 11.972180366516113, "learning_rate": 5.160194174757282e-06, "loss": 0.3134, "step": 19537 }, { "epoch": 189.4566265060241, "grad_norm": 8.241326332092285, "learning_rate": 5.155339805825243e-06, "loss": 0.2602, "step": 19538 }, { "epoch": 189.46626506024097, "grad_norm": 10.79302978515625, "learning_rate": 5.150485436893204e-06, "loss": 0.2338, "step": 19539 }, { "epoch": 189.47590361445782, "grad_norm": 36.27009582519531, "learning_rate": 5.145631067961165e-06, "loss": 0.2706, "step": 19540 }, { "epoch": 189.4855421686747, "grad_norm": 27.282312393188477, "learning_rate": 5.140776699029126e-06, "loss": 0.1817, "step": 19541 }, { "epoch": 189.49518072289158, "grad_norm": 9.514299392700195, "learning_rate": 5.1359223300970875e-06, "loss": 0.2232, "step": 19542 }, { "epoch": 189.50481927710842, "grad_norm": 13.16293716430664, "learning_rate": 5.131067961165049e-06, "loss": 0.284, "step": 19543 }, { "epoch": 189.5144578313253, "grad_norm": 6.857641220092773, "learning_rate": 5.12621359223301e-06, "loss": 0.1772, "step": 19544 }, { "epoch": 189.52409638554218, "grad_norm": 14.243935585021973, "learning_rate": 5.121359223300972e-06, "loss": 0.2493, "step": 19545 }, { "epoch": 189.53373493975903, "grad_norm": 9.160735130310059, "learning_rate": 5.116504854368932e-06, "loss": 0.155, "step": 19546 }, { "epoch": 189.5433734939759, "grad_norm": 8.565135955810547, "learning_rate": 5.111650485436893e-06, "loss": 0.1638, "step": 19547 }, { "epoch": 189.55301204819278, "grad_norm": 3.361161470413208, "learning_rate": 5.106796116504855e-06, "loss": 0.0873, "step": 19548 }, { "epoch": 189.56265060240963, "grad_norm": 11.850555419921875, "learning_rate": 5.1019417475728155e-06, "loss": 0.262, "step": 19549 }, { "epoch": 189.5722891566265, "grad_norm": 10.457060813903809, "learning_rate": 5.097087378640777e-06, "loss": 0.1562, "step": 19550 }, { "epoch": 189.58192771084336, "grad_norm": 5.070122718811035, "learning_rate": 5.092233009708739e-06, "loss": 0.2686, "step": 19551 }, { "epoch": 189.59156626506024, "grad_norm": 17.388896942138672, "learning_rate": 5.087378640776699e-06, "loss": 0.3493, "step": 19552 }, { "epoch": 189.60120481927711, "grad_norm": 30.180803298950195, "learning_rate": 5.08252427184466e-06, "loss": 0.1802, "step": 19553 }, { "epoch": 189.61084337349396, "grad_norm": 6.040780544281006, "learning_rate": 5.077669902912622e-06, "loss": 0.34, "step": 19554 }, { "epoch": 189.62048192771084, "grad_norm": 28.45606231689453, "learning_rate": 5.072815533980583e-06, "loss": 0.1804, "step": 19555 }, { "epoch": 189.63012048192772, "grad_norm": 3.080415725708008, "learning_rate": 5.0679611650485435e-06, "loss": 0.0346, "step": 19556 }, { "epoch": 189.63975903614457, "grad_norm": 8.634824752807617, "learning_rate": 5.063106796116505e-06, "loss": 0.2313, "step": 19557 }, { "epoch": 189.64939759036145, "grad_norm": 12.999287605285645, "learning_rate": 5.058252427184467e-06, "loss": 0.2106, "step": 19558 }, { "epoch": 189.65903614457832, "grad_norm": 5.585487365722656, "learning_rate": 5.053398058252427e-06, "loss": 0.2969, "step": 19559 }, { "epoch": 189.66867469879517, "grad_norm": 14.528498649597168, "learning_rate": 5.048543689320388e-06, "loss": 0.652, "step": 19560 }, { "epoch": 189.67831325301205, "grad_norm": 18.75196647644043, "learning_rate": 5.04368932038835e-06, "loss": 0.193, "step": 19561 }, { "epoch": 189.68795180722893, "grad_norm": 16.54024887084961, "learning_rate": 5.038834951456311e-06, "loss": 0.2683, "step": 19562 }, { "epoch": 189.69759036144578, "grad_norm": 14.971738815307617, "learning_rate": 5.0339805825242715e-06, "loss": 0.2878, "step": 19563 }, { "epoch": 189.70722891566265, "grad_norm": 6.055410861968994, "learning_rate": 5.0291262135922335e-06, "loss": 0.1785, "step": 19564 }, { "epoch": 189.71686746987953, "grad_norm": 14.201848983764648, "learning_rate": 5.024271844660195e-06, "loss": 0.2966, "step": 19565 }, { "epoch": 189.72650602409638, "grad_norm": 36.072120666503906, "learning_rate": 5.019417475728155e-06, "loss": 0.1528, "step": 19566 }, { "epoch": 189.73614457831326, "grad_norm": 4.109537601470947, "learning_rate": 5.014563106796117e-06, "loss": 0.0604, "step": 19567 }, { "epoch": 189.7457831325301, "grad_norm": 21.75999641418457, "learning_rate": 5.009708737864078e-06, "loss": 0.3856, "step": 19568 }, { "epoch": 189.75542168674698, "grad_norm": 6.6839280128479, "learning_rate": 5.004854368932039e-06, "loss": 0.3101, "step": 19569 }, { "epoch": 189.76506024096386, "grad_norm": 48.271793365478516, "learning_rate": 5e-06, "loss": 0.4547, "step": 19570 }, { "epoch": 189.7746987951807, "grad_norm": 22.099802017211914, "learning_rate": 4.9951456310679616e-06, "loss": 0.2033, "step": 19571 }, { "epoch": 189.7843373493976, "grad_norm": 17.75705909729004, "learning_rate": 4.990291262135923e-06, "loss": 0.3097, "step": 19572 }, { "epoch": 189.79397590361447, "grad_norm": 25.456960678100586, "learning_rate": 4.985436893203884e-06, "loss": 0.1927, "step": 19573 }, { "epoch": 189.80361445783132, "grad_norm": 20.086179733276367, "learning_rate": 4.980582524271845e-06, "loss": 0.3368, "step": 19574 }, { "epoch": 189.8132530120482, "grad_norm": 18.56314468383789, "learning_rate": 4.975728155339806e-06, "loss": 0.4525, "step": 19575 }, { "epoch": 189.82289156626507, "grad_norm": 10.623266220092773, "learning_rate": 4.970873786407767e-06, "loss": 0.2134, "step": 19576 }, { "epoch": 189.83253012048192, "grad_norm": 4.417193412780762, "learning_rate": 4.9660194174757284e-06, "loss": 0.1256, "step": 19577 }, { "epoch": 189.8421686746988, "grad_norm": 20.978065490722656, "learning_rate": 4.9611650485436896e-06, "loss": 0.3736, "step": 19578 }, { "epoch": 189.85180722891567, "grad_norm": 54.3564567565918, "learning_rate": 4.956310679611651e-06, "loss": 0.122, "step": 19579 }, { "epoch": 189.86144578313252, "grad_norm": 28.055301666259766, "learning_rate": 4.951456310679612e-06, "loss": 0.428, "step": 19580 }, { "epoch": 189.8710843373494, "grad_norm": 13.614959716796875, "learning_rate": 4.946601941747573e-06, "loss": 0.3798, "step": 19581 }, { "epoch": 189.88072289156628, "grad_norm": 55.71247482299805, "learning_rate": 4.941747572815534e-06, "loss": 0.216, "step": 19582 }, { "epoch": 189.89036144578313, "grad_norm": 9.537864685058594, "learning_rate": 4.936893203883495e-06, "loss": 0.2102, "step": 19583 }, { "epoch": 189.9, "grad_norm": 21.021902084350586, "learning_rate": 4.9320388349514564e-06, "loss": 0.4554, "step": 19584 }, { "epoch": 189.90963855421685, "grad_norm": 47.19050979614258, "learning_rate": 4.927184466019418e-06, "loss": 0.509, "step": 19585 }, { "epoch": 189.91927710843373, "grad_norm": 9.480090141296387, "learning_rate": 4.922330097087379e-06, "loss": 0.3192, "step": 19586 }, { "epoch": 189.9289156626506, "grad_norm": 19.084896087646484, "learning_rate": 4.91747572815534e-06, "loss": 0.2534, "step": 19587 }, { "epoch": 189.93855421686746, "grad_norm": 17.442214965820312, "learning_rate": 4.912621359223301e-06, "loss": 0.256, "step": 19588 }, { "epoch": 189.94819277108434, "grad_norm": 7.65753698348999, "learning_rate": 4.907766990291263e-06, "loss": 0.316, "step": 19589 }, { "epoch": 189.9578313253012, "grad_norm": 4.594153881072998, "learning_rate": 4.902912621359223e-06, "loss": 0.1426, "step": 19590 }, { "epoch": 189.96746987951806, "grad_norm": 5.911007404327393, "learning_rate": 4.8980582524271845e-06, "loss": 0.233, "step": 19591 }, { "epoch": 189.97710843373494, "grad_norm": 9.054054260253906, "learning_rate": 4.8932038834951465e-06, "loss": 0.3318, "step": 19592 }, { "epoch": 189.98674698795182, "grad_norm": 89.33706665039062, "learning_rate": 4.888349514563107e-06, "loss": 0.3316, "step": 19593 }, { "epoch": 190.0024096385542, "grad_norm": 12.469794273376465, "learning_rate": 4.883495145631068e-06, "loss": 0.1542, "step": 19594 }, { "epoch": 190.0120481927711, "grad_norm": 3.0273261070251465, "learning_rate": 4.87864077669903e-06, "loss": 0.0835, "step": 19595 }, { "epoch": 190.02168674698794, "grad_norm": 11.051825523376465, "learning_rate": 4.873786407766991e-06, "loss": 0.1937, "step": 19596 }, { "epoch": 190.03132530120482, "grad_norm": 9.548336029052734, "learning_rate": 4.868932038834951e-06, "loss": 0.198, "step": 19597 }, { "epoch": 190.0409638554217, "grad_norm": 14.040855407714844, "learning_rate": 4.8640776699029125e-06, "loss": 0.2308, "step": 19598 }, { "epoch": 190.05060240963854, "grad_norm": 5.940856456756592, "learning_rate": 4.8592233009708745e-06, "loss": 0.1875, "step": 19599 }, { "epoch": 190.06024096385542, "grad_norm": 8.769341468811035, "learning_rate": 4.854368932038835e-06, "loss": 0.2133, "step": 19600 }, { "epoch": 190.0698795180723, "grad_norm": 15.135574340820312, "learning_rate": 4.849514563106796e-06, "loss": 0.2806, "step": 19601 }, { "epoch": 190.07951807228915, "grad_norm": 7.837410926818848, "learning_rate": 4.844660194174758e-06, "loss": 0.3182, "step": 19602 }, { "epoch": 190.08915662650602, "grad_norm": 7.720034122467041, "learning_rate": 4.839805825242718e-06, "loss": 0.1811, "step": 19603 }, { "epoch": 190.0987951807229, "grad_norm": 10.494301795959473, "learning_rate": 4.834951456310679e-06, "loss": 0.3398, "step": 19604 }, { "epoch": 190.10843373493975, "grad_norm": 10.256226539611816, "learning_rate": 4.830097087378641e-06, "loss": 0.3684, "step": 19605 }, { "epoch": 190.11807228915663, "grad_norm": 12.250589370727539, "learning_rate": 4.8252427184466025e-06, "loss": 0.2595, "step": 19606 }, { "epoch": 190.1277108433735, "grad_norm": 20.10175132751465, "learning_rate": 4.820388349514563e-06, "loss": 0.2779, "step": 19607 }, { "epoch": 190.13734939759036, "grad_norm": 13.766989707946777, "learning_rate": 4.815533980582525e-06, "loss": 0.2555, "step": 19608 }, { "epoch": 190.14698795180723, "grad_norm": 4.046898365020752, "learning_rate": 4.810679611650486e-06, "loss": 0.292, "step": 19609 }, { "epoch": 190.15662650602408, "grad_norm": 28.04939079284668, "learning_rate": 4.805825242718446e-06, "loss": 0.2207, "step": 19610 }, { "epoch": 190.16626506024096, "grad_norm": 18.50339698791504, "learning_rate": 4.800970873786408e-06, "loss": 0.192, "step": 19611 }, { "epoch": 190.17590361445784, "grad_norm": 8.944395065307617, "learning_rate": 4.796116504854369e-06, "loss": 0.3819, "step": 19612 }, { "epoch": 190.1855421686747, "grad_norm": 6.364335060119629, "learning_rate": 4.7912621359223305e-06, "loss": 0.1041, "step": 19613 }, { "epoch": 190.19518072289156, "grad_norm": 23.74849510192871, "learning_rate": 4.786407766990292e-06, "loss": 0.3826, "step": 19614 }, { "epoch": 190.20481927710844, "grad_norm": 21.66086769104004, "learning_rate": 4.781553398058253e-06, "loss": 0.3648, "step": 19615 }, { "epoch": 190.2144578313253, "grad_norm": 11.27698802947998, "learning_rate": 4.776699029126214e-06, "loss": 0.3529, "step": 19616 }, { "epoch": 190.22409638554217, "grad_norm": 15.958869934082031, "learning_rate": 4.771844660194174e-06, "loss": 0.3889, "step": 19617 }, { "epoch": 190.23373493975905, "grad_norm": 14.041999816894531, "learning_rate": 4.766990291262136e-06, "loss": 0.1873, "step": 19618 }, { "epoch": 190.2433734939759, "grad_norm": 15.674661636352539, "learning_rate": 4.762135922330097e-06, "loss": 0.3274, "step": 19619 }, { "epoch": 190.25301204819277, "grad_norm": 17.875015258789062, "learning_rate": 4.7572815533980585e-06, "loss": 0.0981, "step": 19620 }, { "epoch": 190.26265060240965, "grad_norm": 33.021942138671875, "learning_rate": 4.75242718446602e-06, "loss": 0.2896, "step": 19621 }, { "epoch": 190.2722891566265, "grad_norm": 12.559330940246582, "learning_rate": 4.747572815533981e-06, "loss": 0.2663, "step": 19622 }, { "epoch": 190.28192771084338, "grad_norm": 17.07077980041504, "learning_rate": 4.742718446601942e-06, "loss": 0.2833, "step": 19623 }, { "epoch": 190.29156626506025, "grad_norm": 9.8994722366333, "learning_rate": 4.737864077669903e-06, "loss": 0.2325, "step": 19624 }, { "epoch": 190.3012048192771, "grad_norm": 10.171090126037598, "learning_rate": 4.733009708737864e-06, "loss": 0.3395, "step": 19625 }, { "epoch": 190.31084337349398, "grad_norm": 15.197639465332031, "learning_rate": 4.728155339805825e-06, "loss": 0.3172, "step": 19626 }, { "epoch": 190.32048192771083, "grad_norm": 9.637776374816895, "learning_rate": 4.7233009708737866e-06, "loss": 0.3185, "step": 19627 }, { "epoch": 190.3301204819277, "grad_norm": 19.74883270263672, "learning_rate": 4.718446601941748e-06, "loss": 0.1508, "step": 19628 }, { "epoch": 190.33975903614459, "grad_norm": 11.076652526855469, "learning_rate": 4.713592233009709e-06, "loss": 0.2068, "step": 19629 }, { "epoch": 190.34939759036143, "grad_norm": 32.506893157958984, "learning_rate": 4.70873786407767e-06, "loss": 0.1467, "step": 19630 }, { "epoch": 190.3590361445783, "grad_norm": 17.464675903320312, "learning_rate": 4.703883495145631e-06, "loss": 0.3369, "step": 19631 }, { "epoch": 190.3686746987952, "grad_norm": 14.242315292358398, "learning_rate": 4.699029126213592e-06, "loss": 0.2986, "step": 19632 }, { "epoch": 190.37831325301204, "grad_norm": 11.223112106323242, "learning_rate": 4.694174757281554e-06, "loss": 0.2601, "step": 19633 }, { "epoch": 190.38795180722892, "grad_norm": 8.11742115020752, "learning_rate": 4.689320388349515e-06, "loss": 0.1178, "step": 19634 }, { "epoch": 190.3975903614458, "grad_norm": 7.111746788024902, "learning_rate": 4.684466019417476e-06, "loss": 0.1619, "step": 19635 }, { "epoch": 190.40722891566264, "grad_norm": 11.522008895874023, "learning_rate": 4.679611650485438e-06, "loss": 0.2435, "step": 19636 }, { "epoch": 190.41686746987952, "grad_norm": 6.657716274261475, "learning_rate": 4.674757281553398e-06, "loss": 0.2932, "step": 19637 }, { "epoch": 190.4265060240964, "grad_norm": 7.02471923828125, "learning_rate": 4.669902912621359e-06, "loss": 0.1676, "step": 19638 }, { "epoch": 190.43614457831325, "grad_norm": 5.098782062530518, "learning_rate": 4.66504854368932e-06, "loss": 0.1185, "step": 19639 }, { "epoch": 190.44578313253012, "grad_norm": 17.87139892578125, "learning_rate": 4.660194174757282e-06, "loss": 0.2197, "step": 19640 }, { "epoch": 190.455421686747, "grad_norm": 5.790158748626709, "learning_rate": 4.655339805825243e-06, "loss": 0.2725, "step": 19641 }, { "epoch": 190.46506024096385, "grad_norm": 18.086320877075195, "learning_rate": 4.650485436893204e-06, "loss": 0.5909, "step": 19642 }, { "epoch": 190.47469879518073, "grad_norm": 7.0668487548828125, "learning_rate": 4.645631067961166e-06, "loss": 0.2665, "step": 19643 }, { "epoch": 190.48433734939758, "grad_norm": 19.52154541015625, "learning_rate": 4.640776699029126e-06, "loss": 0.1721, "step": 19644 }, { "epoch": 190.49397590361446, "grad_norm": 14.594107627868652, "learning_rate": 4.635922330097087e-06, "loss": 0.2051, "step": 19645 }, { "epoch": 190.50361445783133, "grad_norm": 15.85614013671875, "learning_rate": 4.631067961165049e-06, "loss": 0.3335, "step": 19646 }, { "epoch": 190.51325301204818, "grad_norm": 10.184294700622559, "learning_rate": 4.62621359223301e-06, "loss": 0.2081, "step": 19647 }, { "epoch": 190.52289156626506, "grad_norm": 10.089242935180664, "learning_rate": 4.621359223300971e-06, "loss": 0.2445, "step": 19648 }, { "epoch": 190.53253012048194, "grad_norm": 10.006924629211426, "learning_rate": 4.616504854368933e-06, "loss": 0.3212, "step": 19649 }, { "epoch": 190.5421686746988, "grad_norm": 10.704833030700684, "learning_rate": 4.611650485436894e-06, "loss": 0.4131, "step": 19650 }, { "epoch": 190.55180722891566, "grad_norm": 9.079778671264648, "learning_rate": 4.606796116504854e-06, "loss": 0.4189, "step": 19651 }, { "epoch": 190.56144578313254, "grad_norm": 5.035790920257568, "learning_rate": 4.601941747572816e-06, "loss": 0.0914, "step": 19652 }, { "epoch": 190.5710843373494, "grad_norm": 9.22657299041748, "learning_rate": 4.597087378640777e-06, "loss": 0.2061, "step": 19653 }, { "epoch": 190.58072289156627, "grad_norm": 5.973226070404053, "learning_rate": 4.592233009708738e-06, "loss": 0.352, "step": 19654 }, { "epoch": 190.59036144578315, "grad_norm": 7.871885776519775, "learning_rate": 4.5873786407766995e-06, "loss": 0.1386, "step": 19655 }, { "epoch": 190.6, "grad_norm": 14.148130416870117, "learning_rate": 4.582524271844661e-06, "loss": 0.3301, "step": 19656 }, { "epoch": 190.60963855421687, "grad_norm": 12.184260368347168, "learning_rate": 4.577669902912622e-06, "loss": 0.1933, "step": 19657 }, { "epoch": 190.61927710843372, "grad_norm": 5.596240043640137, "learning_rate": 4.572815533980582e-06, "loss": 0.2596, "step": 19658 }, { "epoch": 190.6289156626506, "grad_norm": 43.86715316772461, "learning_rate": 4.567961165048544e-06, "loss": 0.2073, "step": 19659 }, { "epoch": 190.63855421686748, "grad_norm": 9.359273910522461, "learning_rate": 4.563106796116505e-06, "loss": 0.4208, "step": 19660 }, { "epoch": 190.64819277108433, "grad_norm": 7.103854656219482, "learning_rate": 4.5582524271844655e-06, "loss": 0.1463, "step": 19661 }, { "epoch": 190.6578313253012, "grad_norm": 16.478132247924805, "learning_rate": 4.5533980582524275e-06, "loss": 0.1776, "step": 19662 }, { "epoch": 190.66746987951808, "grad_norm": 6.351244926452637, "learning_rate": 4.548543689320389e-06, "loss": 0.1156, "step": 19663 }, { "epoch": 190.67710843373493, "grad_norm": 10.14985179901123, "learning_rate": 4.54368932038835e-06, "loss": 0.2059, "step": 19664 }, { "epoch": 190.6867469879518, "grad_norm": 31.43819236755371, "learning_rate": 4.538834951456311e-06, "loss": 0.2731, "step": 19665 }, { "epoch": 190.69638554216868, "grad_norm": 11.663161277770996, "learning_rate": 4.533980582524272e-06, "loss": 0.3008, "step": 19666 }, { "epoch": 190.70602409638553, "grad_norm": 7.903310775756836, "learning_rate": 4.529126213592233e-06, "loss": 0.1619, "step": 19667 }, { "epoch": 190.7156626506024, "grad_norm": 14.406914710998535, "learning_rate": 4.524271844660194e-06, "loss": 0.3262, "step": 19668 }, { "epoch": 190.7253012048193, "grad_norm": 9.722543716430664, "learning_rate": 4.5194174757281555e-06, "loss": 0.2411, "step": 19669 }, { "epoch": 190.73493975903614, "grad_norm": 9.180020332336426, "learning_rate": 4.514563106796117e-06, "loss": 0.1371, "step": 19670 }, { "epoch": 190.74457831325302, "grad_norm": 9.178184509277344, "learning_rate": 4.509708737864078e-06, "loss": 0.4311, "step": 19671 }, { "epoch": 190.7542168674699, "grad_norm": 10.684231758117676, "learning_rate": 4.504854368932039e-06, "loss": 0.2012, "step": 19672 }, { "epoch": 190.76385542168674, "grad_norm": 12.349410057067871, "learning_rate": 4.5e-06, "loss": 0.3142, "step": 19673 }, { "epoch": 190.77349397590362, "grad_norm": 8.36299991607666, "learning_rate": 4.495145631067962e-06, "loss": 0.2061, "step": 19674 }, { "epoch": 190.78313253012047, "grad_norm": 26.80917739868164, "learning_rate": 4.490291262135922e-06, "loss": 0.1898, "step": 19675 }, { "epoch": 190.79277108433735, "grad_norm": 14.237757682800293, "learning_rate": 4.4854368932038836e-06, "loss": 0.3426, "step": 19676 }, { "epoch": 190.80240963855422, "grad_norm": 16.437957763671875, "learning_rate": 4.4805825242718455e-06, "loss": 0.2975, "step": 19677 }, { "epoch": 190.81204819277107, "grad_norm": 29.144962310791016, "learning_rate": 4.475728155339806e-06, "loss": 0.5614, "step": 19678 }, { "epoch": 190.82168674698795, "grad_norm": 18.471731185913086, "learning_rate": 4.470873786407767e-06, "loss": 0.2191, "step": 19679 }, { "epoch": 190.83132530120483, "grad_norm": 11.05218505859375, "learning_rate": 4.466019417475728e-06, "loss": 0.2362, "step": 19680 }, { "epoch": 190.84096385542168, "grad_norm": 31.163650512695312, "learning_rate": 4.461165048543689e-06, "loss": 0.2039, "step": 19681 }, { "epoch": 190.85060240963855, "grad_norm": 4.296408653259277, "learning_rate": 4.4563106796116504e-06, "loss": 0.2117, "step": 19682 }, { "epoch": 190.86024096385543, "grad_norm": 22.572559356689453, "learning_rate": 4.4514563106796116e-06, "loss": 0.3221, "step": 19683 }, { "epoch": 190.86987951807228, "grad_norm": 13.924388885498047, "learning_rate": 4.4466019417475736e-06, "loss": 0.3974, "step": 19684 }, { "epoch": 190.87951807228916, "grad_norm": 19.340456008911133, "learning_rate": 4.441747572815534e-06, "loss": 0.2317, "step": 19685 }, { "epoch": 190.88915662650604, "grad_norm": 8.703386306762695, "learning_rate": 4.436893203883495e-06, "loss": 0.1081, "step": 19686 }, { "epoch": 190.89879518072289, "grad_norm": 11.013641357421875, "learning_rate": 4.432038834951457e-06, "loss": 0.1818, "step": 19687 }, { "epoch": 190.90843373493976, "grad_norm": 5.8611741065979, "learning_rate": 4.427184466019417e-06, "loss": 0.0836, "step": 19688 }, { "epoch": 190.91807228915664, "grad_norm": 8.180835723876953, "learning_rate": 4.4223300970873784e-06, "loss": 0.226, "step": 19689 }, { "epoch": 190.9277108433735, "grad_norm": 7.404837131500244, "learning_rate": 4.4174757281553404e-06, "loss": 0.1966, "step": 19690 }, { "epoch": 190.93734939759037, "grad_norm": 19.341054916381836, "learning_rate": 4.412621359223302e-06, "loss": 0.091, "step": 19691 }, { "epoch": 190.94698795180722, "grad_norm": 9.93282699584961, "learning_rate": 4.407766990291262e-06, "loss": 0.2517, "step": 19692 }, { "epoch": 190.9566265060241, "grad_norm": 9.505485534667969, "learning_rate": 4.402912621359224e-06, "loss": 0.1746, "step": 19693 }, { "epoch": 190.96626506024097, "grad_norm": 27.966875076293945, "learning_rate": 4.398058252427185e-06, "loss": 0.3167, "step": 19694 }, { "epoch": 190.97590361445782, "grad_norm": 8.868237495422363, "learning_rate": 4.393203883495145e-06, "loss": 0.2468, "step": 19695 }, { "epoch": 190.9855421686747, "grad_norm": 11.706402778625488, "learning_rate": 4.388349514563107e-06, "loss": 0.1732, "step": 19696 }, { "epoch": 191.00120481927712, "grad_norm": 8.60560131072998, "learning_rate": 4.3834951456310685e-06, "loss": 0.1397, "step": 19697 }, { "epoch": 191.01084337349397, "grad_norm": 15.581082344055176, "learning_rate": 4.37864077669903e-06, "loss": 0.1461, "step": 19698 }, { "epoch": 191.02048192771085, "grad_norm": 8.194450378417969, "learning_rate": 4.37378640776699e-06, "loss": 0.2295, "step": 19699 }, { "epoch": 191.03012048192772, "grad_norm": 10.014636993408203, "learning_rate": 4.368932038834952e-06, "loss": 0.1717, "step": 19700 }, { "epoch": 191.03975903614457, "grad_norm": 10.917706489562988, "learning_rate": 4.364077669902913e-06, "loss": 0.2989, "step": 19701 }, { "epoch": 191.04939759036145, "grad_norm": 14.336064338684082, "learning_rate": 4.359223300970873e-06, "loss": 0.1758, "step": 19702 }, { "epoch": 191.0590361445783, "grad_norm": 10.716379165649414, "learning_rate": 4.354368932038835e-06, "loss": 0.1843, "step": 19703 }, { "epoch": 191.06867469879518, "grad_norm": 20.58364486694336, "learning_rate": 4.3495145631067965e-06, "loss": 0.2067, "step": 19704 }, { "epoch": 191.07831325301206, "grad_norm": 5.610141277313232, "learning_rate": 4.344660194174758e-06, "loss": 0.2025, "step": 19705 }, { "epoch": 191.0879518072289, "grad_norm": 12.097428321838379, "learning_rate": 4.339805825242719e-06, "loss": 0.2152, "step": 19706 }, { "epoch": 191.09759036144578, "grad_norm": 5.713420391082764, "learning_rate": 4.33495145631068e-06, "loss": 0.2445, "step": 19707 }, { "epoch": 191.10722891566266, "grad_norm": 9.314375877380371, "learning_rate": 4.330097087378641e-06, "loss": 0.1609, "step": 19708 }, { "epoch": 191.1168674698795, "grad_norm": 8.032283782958984, "learning_rate": 4.325242718446602e-06, "loss": 0.2577, "step": 19709 }, { "epoch": 191.1265060240964, "grad_norm": 13.834815979003906, "learning_rate": 4.320388349514563e-06, "loss": 0.2201, "step": 19710 }, { "epoch": 191.13614457831326, "grad_norm": 21.95645523071289, "learning_rate": 4.3155339805825245e-06, "loss": 0.2817, "step": 19711 }, { "epoch": 191.1457831325301, "grad_norm": 11.086427688598633, "learning_rate": 4.310679611650486e-06, "loss": 0.3906, "step": 19712 }, { "epoch": 191.155421686747, "grad_norm": 5.700686454772949, "learning_rate": 4.305825242718447e-06, "loss": 0.1263, "step": 19713 }, { "epoch": 191.16506024096387, "grad_norm": 14.11396598815918, "learning_rate": 4.300970873786408e-06, "loss": 0.2124, "step": 19714 }, { "epoch": 191.17469879518072, "grad_norm": 14.454176902770996, "learning_rate": 4.296116504854369e-06, "loss": 0.1863, "step": 19715 }, { "epoch": 191.1843373493976, "grad_norm": 8.506138801574707, "learning_rate": 4.29126213592233e-06, "loss": 0.2353, "step": 19716 }, { "epoch": 191.19397590361444, "grad_norm": 20.060161590576172, "learning_rate": 4.286407766990291e-06, "loss": 0.3179, "step": 19717 }, { "epoch": 191.20361445783132, "grad_norm": 7.842395305633545, "learning_rate": 4.281553398058253e-06, "loss": 0.221, "step": 19718 }, { "epoch": 191.2132530120482, "grad_norm": 6.698565483093262, "learning_rate": 4.276699029126214e-06, "loss": 0.2515, "step": 19719 }, { "epoch": 191.22289156626505, "grad_norm": 9.51131534576416, "learning_rate": 4.271844660194175e-06, "loss": 0.2173, "step": 19720 }, { "epoch": 191.23253012048193, "grad_norm": 3.4178576469421387, "learning_rate": 4.266990291262136e-06, "loss": 0.0875, "step": 19721 }, { "epoch": 191.2421686746988, "grad_norm": 8.601062774658203, "learning_rate": 4.262135922330097e-06, "loss": 0.3914, "step": 19722 }, { "epoch": 191.25180722891565, "grad_norm": 10.620489120483398, "learning_rate": 4.257281553398058e-06, "loss": 0.1439, "step": 19723 }, { "epoch": 191.26144578313253, "grad_norm": 21.74315643310547, "learning_rate": 4.252427184466019e-06, "loss": 0.4064, "step": 19724 }, { "epoch": 191.2710843373494, "grad_norm": 38.751983642578125, "learning_rate": 4.247572815533981e-06, "loss": 0.4098, "step": 19725 }, { "epoch": 191.28072289156626, "grad_norm": 17.291278839111328, "learning_rate": 4.242718446601942e-06, "loss": 0.1364, "step": 19726 }, { "epoch": 191.29036144578313, "grad_norm": 9.022143363952637, "learning_rate": 4.237864077669903e-06, "loss": 0.3202, "step": 19727 }, { "epoch": 191.3, "grad_norm": 11.386868476867676, "learning_rate": 4.233009708737865e-06, "loss": 0.2235, "step": 19728 }, { "epoch": 191.30963855421686, "grad_norm": 13.02416706085205, "learning_rate": 4.228155339805825e-06, "loss": 0.2857, "step": 19729 }, { "epoch": 191.31927710843374, "grad_norm": 13.612634658813477, "learning_rate": 4.223300970873786e-06, "loss": 0.1835, "step": 19730 }, { "epoch": 191.32891566265062, "grad_norm": 17.931514739990234, "learning_rate": 4.218446601941748e-06, "loss": 0.4287, "step": 19731 }, { "epoch": 191.33855421686746, "grad_norm": 10.610804557800293, "learning_rate": 4.213592233009709e-06, "loss": 0.2657, "step": 19732 }, { "epoch": 191.34819277108434, "grad_norm": 11.711033821105957, "learning_rate": 4.20873786407767e-06, "loss": 0.2771, "step": 19733 }, { "epoch": 191.3578313253012, "grad_norm": 5.750218868255615, "learning_rate": 4.203883495145632e-06, "loss": 0.2574, "step": 19734 }, { "epoch": 191.36746987951807, "grad_norm": 5.068578243255615, "learning_rate": 4.199029126213593e-06, "loss": 0.2624, "step": 19735 }, { "epoch": 191.37710843373495, "grad_norm": 11.687504768371582, "learning_rate": 4.194174757281553e-06, "loss": 0.3294, "step": 19736 }, { "epoch": 191.3867469879518, "grad_norm": 16.241092681884766, "learning_rate": 4.189320388349515e-06, "loss": 0.2808, "step": 19737 }, { "epoch": 191.39638554216867, "grad_norm": 14.958148956298828, "learning_rate": 4.184466019417476e-06, "loss": 0.2269, "step": 19738 }, { "epoch": 191.40602409638555, "grad_norm": 6.785990238189697, "learning_rate": 4.179611650485437e-06, "loss": 0.1164, "step": 19739 }, { "epoch": 191.4156626506024, "grad_norm": 5.824944972991943, "learning_rate": 4.174757281553398e-06, "loss": 0.1618, "step": 19740 }, { "epoch": 191.42530120481928, "grad_norm": 25.466075897216797, "learning_rate": 4.16990291262136e-06, "loss": 0.2408, "step": 19741 }, { "epoch": 191.43493975903615, "grad_norm": 8.851223945617676, "learning_rate": 4.165048543689321e-06, "loss": 0.2788, "step": 19742 }, { "epoch": 191.444578313253, "grad_norm": 14.21731185913086, "learning_rate": 4.160194174757281e-06, "loss": 0.3521, "step": 19743 }, { "epoch": 191.45421686746988, "grad_norm": 7.900482654571533, "learning_rate": 4.155339805825243e-06, "loss": 0.1225, "step": 19744 }, { "epoch": 191.46385542168676, "grad_norm": 12.197418212890625, "learning_rate": 4.150485436893204e-06, "loss": 0.3763, "step": 19745 }, { "epoch": 191.4734939759036, "grad_norm": 22.629642486572266, "learning_rate": 4.145631067961165e-06, "loss": 0.2233, "step": 19746 }, { "epoch": 191.48313253012049, "grad_norm": 2.9662420749664307, "learning_rate": 4.140776699029127e-06, "loss": 0.1296, "step": 19747 }, { "epoch": 191.49277108433736, "grad_norm": 7.965096473693848, "learning_rate": 4.135922330097088e-06, "loss": 0.2178, "step": 19748 }, { "epoch": 191.5024096385542, "grad_norm": 12.646430969238281, "learning_rate": 4.131067961165049e-06, "loss": 0.2159, "step": 19749 }, { "epoch": 191.5120481927711, "grad_norm": 11.801518440246582, "learning_rate": 4.12621359223301e-06, "loss": 0.4094, "step": 19750 }, { "epoch": 191.52168674698794, "grad_norm": 15.326214790344238, "learning_rate": 4.121359223300971e-06, "loss": 0.3323, "step": 19751 }, { "epoch": 191.53132530120482, "grad_norm": 41.10453796386719, "learning_rate": 4.116504854368932e-06, "loss": 0.2199, "step": 19752 }, { "epoch": 191.5409638554217, "grad_norm": 7.06415319442749, "learning_rate": 4.1116504854368935e-06, "loss": 0.2409, "step": 19753 }, { "epoch": 191.55060240963854, "grad_norm": 7.597271919250488, "learning_rate": 4.106796116504855e-06, "loss": 0.2482, "step": 19754 }, { "epoch": 191.56024096385542, "grad_norm": 5.999216079711914, "learning_rate": 4.101941747572816e-06, "loss": 0.2244, "step": 19755 }, { "epoch": 191.5698795180723, "grad_norm": 10.219707489013672, "learning_rate": 4.097087378640777e-06, "loss": 0.2278, "step": 19756 }, { "epoch": 191.57951807228915, "grad_norm": 7.340946197509766, "learning_rate": 4.092233009708738e-06, "loss": 0.2516, "step": 19757 }, { "epoch": 191.58915662650602, "grad_norm": 22.52867889404297, "learning_rate": 4.087378640776699e-06, "loss": 0.2811, "step": 19758 }, { "epoch": 191.5987951807229, "grad_norm": 12.892845153808594, "learning_rate": 4.08252427184466e-06, "loss": 0.1899, "step": 19759 }, { "epoch": 191.60843373493975, "grad_norm": 13.545854568481445, "learning_rate": 4.0776699029126215e-06, "loss": 0.1878, "step": 19760 }, { "epoch": 191.61807228915663, "grad_norm": 8.654553413391113, "learning_rate": 4.072815533980583e-06, "loss": 0.2092, "step": 19761 }, { "epoch": 191.6277108433735, "grad_norm": 16.6429443359375, "learning_rate": 4.067961165048544e-06, "loss": 0.2408, "step": 19762 }, { "epoch": 191.63734939759036, "grad_norm": 10.075438499450684, "learning_rate": 4.063106796116505e-06, "loss": 0.2476, "step": 19763 }, { "epoch": 191.64698795180723, "grad_norm": 36.08232879638672, "learning_rate": 4.058252427184466e-06, "loss": 0.4464, "step": 19764 }, { "epoch": 191.65662650602408, "grad_norm": 13.504632949829102, "learning_rate": 4.053398058252427e-06, "loss": 0.2323, "step": 19765 }, { "epoch": 191.66626506024096, "grad_norm": 15.011550903320312, "learning_rate": 4.048543689320388e-06, "loss": 0.2263, "step": 19766 }, { "epoch": 191.67590361445784, "grad_norm": 13.370892524719238, "learning_rate": 4.0436893203883495e-06, "loss": 0.5387, "step": 19767 }, { "epoch": 191.6855421686747, "grad_norm": 9.329450607299805, "learning_rate": 4.038834951456311e-06, "loss": 0.2663, "step": 19768 }, { "epoch": 191.69518072289156, "grad_norm": 14.917797088623047, "learning_rate": 4.033980582524273e-06, "loss": 0.3248, "step": 19769 }, { "epoch": 191.70481927710844, "grad_norm": 32.91347885131836, "learning_rate": 4.029126213592233e-06, "loss": 0.1774, "step": 19770 }, { "epoch": 191.7144578313253, "grad_norm": 16.124771118164062, "learning_rate": 4.024271844660194e-06, "loss": 0.3096, "step": 19771 }, { "epoch": 191.72409638554217, "grad_norm": 18.479482650756836, "learning_rate": 4.019417475728156e-06, "loss": 0.4085, "step": 19772 }, { "epoch": 191.73373493975905, "grad_norm": 12.792609214782715, "learning_rate": 4.014563106796116e-06, "loss": 0.206, "step": 19773 }, { "epoch": 191.7433734939759, "grad_norm": 6.802131652832031, "learning_rate": 4.0097087378640775e-06, "loss": 0.1216, "step": 19774 }, { "epoch": 191.75301204819277, "grad_norm": 8.600703239440918, "learning_rate": 4.0048543689320395e-06, "loss": 0.1361, "step": 19775 }, { "epoch": 191.76265060240965, "grad_norm": 7.520071029663086, "learning_rate": 4.000000000000001e-06, "loss": 0.1277, "step": 19776 }, { "epoch": 191.7722891566265, "grad_norm": 11.585025787353516, "learning_rate": 3.995145631067961e-06, "loss": 0.2161, "step": 19777 }, { "epoch": 191.78192771084338, "grad_norm": 27.431856155395508, "learning_rate": 3.990291262135923e-06, "loss": 0.1434, "step": 19778 }, { "epoch": 191.79156626506025, "grad_norm": 7.326686382293701, "learning_rate": 3.985436893203884e-06, "loss": 0.1111, "step": 19779 }, { "epoch": 191.8012048192771, "grad_norm": 7.413844108581543, "learning_rate": 3.980582524271844e-06, "loss": 0.1736, "step": 19780 }, { "epoch": 191.81084337349398, "grad_norm": 8.92119312286377, "learning_rate": 3.9757281553398055e-06, "loss": 0.2173, "step": 19781 }, { "epoch": 191.82048192771083, "grad_norm": 5.58105993270874, "learning_rate": 3.9708737864077675e-06, "loss": 0.1549, "step": 19782 }, { "epoch": 191.8301204819277, "grad_norm": 7.562410831451416, "learning_rate": 3.966019417475729e-06, "loss": 0.1866, "step": 19783 }, { "epoch": 191.83975903614459, "grad_norm": 5.336896896362305, "learning_rate": 3.961165048543689e-06, "loss": 0.1686, "step": 19784 }, { "epoch": 191.84939759036143, "grad_norm": 20.620071411132812, "learning_rate": 3.956310679611651e-06, "loss": 0.1786, "step": 19785 }, { "epoch": 191.8590361445783, "grad_norm": 6.199432373046875, "learning_rate": 3.951456310679612e-06, "loss": 0.1074, "step": 19786 }, { "epoch": 191.8686746987952, "grad_norm": 14.83978271484375, "learning_rate": 3.946601941747572e-06, "loss": 0.2478, "step": 19787 }, { "epoch": 191.87831325301204, "grad_norm": 5.444101333618164, "learning_rate": 3.941747572815534e-06, "loss": 0.0947, "step": 19788 }, { "epoch": 191.88795180722892, "grad_norm": 6.659621238708496, "learning_rate": 3.9368932038834956e-06, "loss": 0.1341, "step": 19789 }, { "epoch": 191.8975903614458, "grad_norm": 11.133477210998535, "learning_rate": 3.932038834951456e-06, "loss": 0.2482, "step": 19790 }, { "epoch": 191.90722891566264, "grad_norm": 25.934612274169922, "learning_rate": 3.927184466019418e-06, "loss": 0.436, "step": 19791 }, { "epoch": 191.91686746987952, "grad_norm": 10.526362419128418, "learning_rate": 3.922330097087379e-06, "loss": 0.2053, "step": 19792 }, { "epoch": 191.9265060240964, "grad_norm": 54.45538330078125, "learning_rate": 3.91747572815534e-06, "loss": 0.242, "step": 19793 }, { "epoch": 191.93614457831325, "grad_norm": 25.09046173095703, "learning_rate": 3.912621359223301e-06, "loss": 0.115, "step": 19794 }, { "epoch": 191.94578313253012, "grad_norm": 8.539177894592285, "learning_rate": 3.9077669902912624e-06, "loss": 0.3739, "step": 19795 }, { "epoch": 191.955421686747, "grad_norm": 16.170013427734375, "learning_rate": 3.902912621359224e-06, "loss": 0.3394, "step": 19796 }, { "epoch": 191.96506024096385, "grad_norm": 10.529325485229492, "learning_rate": 3.898058252427185e-06, "loss": 0.1754, "step": 19797 }, { "epoch": 191.97469879518073, "grad_norm": 9.573311805725098, "learning_rate": 3.893203883495146e-06, "loss": 0.2539, "step": 19798 }, { "epoch": 191.98433734939758, "grad_norm": 3.4891862869262695, "learning_rate": 3.888349514563107e-06, "loss": 0.0835, "step": 19799 }, { "epoch": 191.99397590361446, "grad_norm": 11.799059867858887, "learning_rate": 3.883495145631068e-06, "loss": 0.1595, "step": 19800 }, { "epoch": 192.00963855421688, "grad_norm": 8.530496597290039, "learning_rate": 3.878640776699029e-06, "loss": 0.1978, "step": 19801 }, { "epoch": 192.01927710843373, "grad_norm": 6.6641035079956055, "learning_rate": 3.8737864077669905e-06, "loss": 0.3236, "step": 19802 }, { "epoch": 192.0289156626506, "grad_norm": 12.959328651428223, "learning_rate": 3.868932038834952e-06, "loss": 0.2692, "step": 19803 }, { "epoch": 192.03855421686748, "grad_norm": 6.281196117401123, "learning_rate": 3.864077669902913e-06, "loss": 0.1431, "step": 19804 }, { "epoch": 192.04819277108433, "grad_norm": 7.0485358238220215, "learning_rate": 3.859223300970874e-06, "loss": 0.3107, "step": 19805 }, { "epoch": 192.0578313253012, "grad_norm": 9.981892585754395, "learning_rate": 3.854368932038835e-06, "loss": 0.2554, "step": 19806 }, { "epoch": 192.0674698795181, "grad_norm": 11.374152183532715, "learning_rate": 3.849514563106796e-06, "loss": 0.2379, "step": 19807 }, { "epoch": 192.07710843373494, "grad_norm": 61.47528076171875, "learning_rate": 3.844660194174757e-06, "loss": 0.1575, "step": 19808 }, { "epoch": 192.0867469879518, "grad_norm": 8.461282730102539, "learning_rate": 3.8398058252427185e-06, "loss": 0.2316, "step": 19809 }, { "epoch": 192.09638554216866, "grad_norm": 12.442951202392578, "learning_rate": 3.83495145631068e-06, "loss": 0.2809, "step": 19810 }, { "epoch": 192.10602409638554, "grad_norm": 10.968072891235352, "learning_rate": 3.830097087378641e-06, "loss": 0.2476, "step": 19811 }, { "epoch": 192.11566265060242, "grad_norm": 7.471534252166748, "learning_rate": 3.825242718446602e-06, "loss": 0.2373, "step": 19812 }, { "epoch": 192.12530120481927, "grad_norm": 7.316954612731934, "learning_rate": 3.820388349514564e-06, "loss": 0.1897, "step": 19813 }, { "epoch": 192.13493975903614, "grad_norm": 19.89263916015625, "learning_rate": 3.815533980582524e-06, "loss": 0.2083, "step": 19814 }, { "epoch": 192.14457831325302, "grad_norm": 16.721403121948242, "learning_rate": 3.8106796116504853e-06, "loss": 0.2175, "step": 19815 }, { "epoch": 192.15421686746987, "grad_norm": 27.749235153198242, "learning_rate": 3.805825242718447e-06, "loss": 0.142, "step": 19816 }, { "epoch": 192.16385542168675, "grad_norm": 9.283204078674316, "learning_rate": 3.800970873786408e-06, "loss": 0.3212, "step": 19817 }, { "epoch": 192.17349397590363, "grad_norm": 11.615677833557129, "learning_rate": 3.7961165048543688e-06, "loss": 0.2257, "step": 19818 }, { "epoch": 192.18313253012047, "grad_norm": 9.154701232910156, "learning_rate": 3.7912621359223308e-06, "loss": 0.1708, "step": 19819 }, { "epoch": 192.19277108433735, "grad_norm": 12.876127243041992, "learning_rate": 3.7864077669902915e-06, "loss": 0.252, "step": 19820 }, { "epoch": 192.20240963855423, "grad_norm": 6.496553421020508, "learning_rate": 3.7815533980582526e-06, "loss": 0.1724, "step": 19821 }, { "epoch": 192.21204819277108, "grad_norm": 30.158998489379883, "learning_rate": 3.7766990291262134e-06, "loss": 0.3464, "step": 19822 }, { "epoch": 192.22168674698796, "grad_norm": 10.853828430175781, "learning_rate": 3.771844660194175e-06, "loss": 0.2559, "step": 19823 }, { "epoch": 192.2313253012048, "grad_norm": 12.828845977783203, "learning_rate": 3.766990291262136e-06, "loss": 0.1589, "step": 19824 }, { "epoch": 192.24096385542168, "grad_norm": 8.709051132202148, "learning_rate": 3.762135922330097e-06, "loss": 0.1516, "step": 19825 }, { "epoch": 192.25060240963856, "grad_norm": 26.810562133789062, "learning_rate": 3.757281553398059e-06, "loss": 0.4265, "step": 19826 }, { "epoch": 192.2602409638554, "grad_norm": 11.594810485839844, "learning_rate": 3.7524271844660195e-06, "loss": 0.2813, "step": 19827 }, { "epoch": 192.2698795180723, "grad_norm": 12.040166854858398, "learning_rate": 3.7475728155339807e-06, "loss": 0.2923, "step": 19828 }, { "epoch": 192.27951807228916, "grad_norm": 9.127885818481445, "learning_rate": 3.7427184466019422e-06, "loss": 0.2267, "step": 19829 }, { "epoch": 192.289156626506, "grad_norm": 8.116878509521484, "learning_rate": 3.737864077669903e-06, "loss": 0.1941, "step": 19830 }, { "epoch": 192.2987951807229, "grad_norm": 5.434265613555908, "learning_rate": 3.733009708737864e-06, "loss": 0.1246, "step": 19831 }, { "epoch": 192.30843373493977, "grad_norm": 45.29445266723633, "learning_rate": 3.7281553398058257e-06, "loss": 0.2826, "step": 19832 }, { "epoch": 192.31807228915662, "grad_norm": 17.66079330444336, "learning_rate": 3.723300970873787e-06, "loss": 0.128, "step": 19833 }, { "epoch": 192.3277108433735, "grad_norm": 9.410825729370117, "learning_rate": 3.7184466019417475e-06, "loss": 0.2091, "step": 19834 }, { "epoch": 192.33734939759037, "grad_norm": 13.424236297607422, "learning_rate": 3.713592233009709e-06, "loss": 0.1939, "step": 19835 }, { "epoch": 192.34698795180722, "grad_norm": 8.879570007324219, "learning_rate": 3.7087378640776703e-06, "loss": 0.3763, "step": 19836 }, { "epoch": 192.3566265060241, "grad_norm": 9.395161628723145, "learning_rate": 3.703883495145631e-06, "loss": 0.4117, "step": 19837 }, { "epoch": 192.36626506024098, "grad_norm": 6.9041571617126465, "learning_rate": 3.6990291262135925e-06, "loss": 0.1688, "step": 19838 }, { "epoch": 192.37590361445783, "grad_norm": 5.096565246582031, "learning_rate": 3.6941747572815537e-06, "loss": 0.1708, "step": 19839 }, { "epoch": 192.3855421686747, "grad_norm": 7.701168060302734, "learning_rate": 3.6893203883495144e-06, "loss": 0.1099, "step": 19840 }, { "epoch": 192.39518072289155, "grad_norm": 6.507519245147705, "learning_rate": 3.6844660194174756e-06, "loss": 0.2091, "step": 19841 }, { "epoch": 192.40481927710843, "grad_norm": 14.223855972290039, "learning_rate": 3.679611650485437e-06, "loss": 0.2673, "step": 19842 }, { "epoch": 192.4144578313253, "grad_norm": 10.22226619720459, "learning_rate": 3.6747572815533983e-06, "loss": 0.2677, "step": 19843 }, { "epoch": 192.42409638554216, "grad_norm": 8.595595359802246, "learning_rate": 3.669902912621359e-06, "loss": 0.1376, "step": 19844 }, { "epoch": 192.43373493975903, "grad_norm": 7.260865211486816, "learning_rate": 3.6650485436893206e-06, "loss": 0.1041, "step": 19845 }, { "epoch": 192.4433734939759, "grad_norm": 13.143649101257324, "learning_rate": 3.6601941747572817e-06, "loss": 0.4199, "step": 19846 }, { "epoch": 192.45301204819276, "grad_norm": 9.514802932739258, "learning_rate": 3.6553398058252424e-06, "loss": 0.2296, "step": 19847 }, { "epoch": 192.46265060240964, "grad_norm": 11.235267639160156, "learning_rate": 3.6504854368932044e-06, "loss": 0.2788, "step": 19848 }, { "epoch": 192.47228915662652, "grad_norm": 6.374825477600098, "learning_rate": 3.645631067961165e-06, "loss": 0.1759, "step": 19849 }, { "epoch": 192.48192771084337, "grad_norm": 5.222932815551758, "learning_rate": 3.6407766990291263e-06, "loss": 0.2137, "step": 19850 }, { "epoch": 192.49156626506024, "grad_norm": 12.354799270629883, "learning_rate": 3.635922330097088e-06, "loss": 0.3754, "step": 19851 }, { "epoch": 192.50120481927712, "grad_norm": 10.435165405273438, "learning_rate": 3.6310679611650486e-06, "loss": 0.2236, "step": 19852 }, { "epoch": 192.51084337349397, "grad_norm": 14.070378303527832, "learning_rate": 3.6262135922330097e-06, "loss": 0.4389, "step": 19853 }, { "epoch": 192.52048192771085, "grad_norm": 9.591669082641602, "learning_rate": 3.6213592233009713e-06, "loss": 0.3652, "step": 19854 }, { "epoch": 192.53012048192772, "grad_norm": 6.859612941741943, "learning_rate": 3.6165048543689324e-06, "loss": 0.2081, "step": 19855 }, { "epoch": 192.53975903614457, "grad_norm": 10.972738265991211, "learning_rate": 3.611650485436893e-06, "loss": 0.1665, "step": 19856 }, { "epoch": 192.54939759036145, "grad_norm": 12.976487159729004, "learning_rate": 3.6067961165048547e-06, "loss": 0.3842, "step": 19857 }, { "epoch": 192.5590361445783, "grad_norm": 14.442286491394043, "learning_rate": 3.601941747572816e-06, "loss": 0.2518, "step": 19858 }, { "epoch": 192.56867469879518, "grad_norm": 3.3897130489349365, "learning_rate": 3.5970873786407766e-06, "loss": 0.0376, "step": 19859 }, { "epoch": 192.57831325301206, "grad_norm": 20.57148551940918, "learning_rate": 3.5922330097087378e-06, "loss": 0.5047, "step": 19860 }, { "epoch": 192.5879518072289, "grad_norm": 5.039952278137207, "learning_rate": 3.5873786407766993e-06, "loss": 0.2152, "step": 19861 }, { "epoch": 192.59759036144578, "grad_norm": 16.1741886138916, "learning_rate": 3.5825242718446605e-06, "loss": 0.1847, "step": 19862 }, { "epoch": 192.60722891566266, "grad_norm": 11.573238372802734, "learning_rate": 3.577669902912621e-06, "loss": 0.2724, "step": 19863 }, { "epoch": 192.6168674698795, "grad_norm": 12.88215160369873, "learning_rate": 3.5728155339805828e-06, "loss": 0.2428, "step": 19864 }, { "epoch": 192.6265060240964, "grad_norm": 10.877924919128418, "learning_rate": 3.567961165048544e-06, "loss": 0.241, "step": 19865 }, { "epoch": 192.63614457831326, "grad_norm": 3.3309080600738525, "learning_rate": 3.5631067961165046e-06, "loss": 0.0832, "step": 19866 }, { "epoch": 192.6457831325301, "grad_norm": 10.67110824584961, "learning_rate": 3.558252427184466e-06, "loss": 0.2153, "step": 19867 }, { "epoch": 192.655421686747, "grad_norm": 9.941617965698242, "learning_rate": 3.5533980582524273e-06, "loss": 0.2727, "step": 19868 }, { "epoch": 192.66506024096387, "grad_norm": 3.783834457397461, "learning_rate": 3.548543689320388e-06, "loss": 0.0926, "step": 19869 }, { "epoch": 192.67469879518072, "grad_norm": 5.85626220703125, "learning_rate": 3.54368932038835e-06, "loss": 0.1038, "step": 19870 }, { "epoch": 192.6843373493976, "grad_norm": 18.00307846069336, "learning_rate": 3.5388349514563108e-06, "loss": 0.2302, "step": 19871 }, { "epoch": 192.69397590361444, "grad_norm": 11.721177101135254, "learning_rate": 3.533980582524272e-06, "loss": 0.3189, "step": 19872 }, { "epoch": 192.70361445783132, "grad_norm": 6.840135097503662, "learning_rate": 3.5291262135922335e-06, "loss": 0.1631, "step": 19873 }, { "epoch": 192.7132530120482, "grad_norm": 9.296359062194824, "learning_rate": 3.5242718446601942e-06, "loss": 0.3569, "step": 19874 }, { "epoch": 192.72289156626505, "grad_norm": 12.157074928283691, "learning_rate": 3.5194174757281554e-06, "loss": 0.3522, "step": 19875 }, { "epoch": 192.73253012048193, "grad_norm": 14.998205184936523, "learning_rate": 3.514563106796117e-06, "loss": 0.3762, "step": 19876 }, { "epoch": 192.7421686746988, "grad_norm": 7.004604339599609, "learning_rate": 3.509708737864078e-06, "loss": 0.1685, "step": 19877 }, { "epoch": 192.75180722891565, "grad_norm": 7.626594543457031, "learning_rate": 3.504854368932039e-06, "loss": 0.1975, "step": 19878 }, { "epoch": 192.76144578313253, "grad_norm": 8.65560245513916, "learning_rate": 3.5000000000000004e-06, "loss": 0.2418, "step": 19879 }, { "epoch": 192.7710843373494, "grad_norm": 8.003771781921387, "learning_rate": 3.4951456310679615e-06, "loss": 0.0955, "step": 19880 }, { "epoch": 192.78072289156626, "grad_norm": 15.228022575378418, "learning_rate": 3.4902912621359222e-06, "loss": 0.1707, "step": 19881 }, { "epoch": 192.79036144578313, "grad_norm": 8.383556365966797, "learning_rate": 3.4854368932038834e-06, "loss": 0.0998, "step": 19882 }, { "epoch": 192.8, "grad_norm": 14.873260498046875, "learning_rate": 3.480582524271845e-06, "loss": 0.4861, "step": 19883 }, { "epoch": 192.80963855421686, "grad_norm": 6.781433582305908, "learning_rate": 3.475728155339806e-06, "loss": 0.2366, "step": 19884 }, { "epoch": 192.81927710843374, "grad_norm": 16.236833572387695, "learning_rate": 3.470873786407767e-06, "loss": 0.2707, "step": 19885 }, { "epoch": 192.82891566265062, "grad_norm": 17.944351196289062, "learning_rate": 3.4660194174757284e-06, "loss": 0.262, "step": 19886 }, { "epoch": 192.83855421686746, "grad_norm": 24.929088592529297, "learning_rate": 3.4611650485436895e-06, "loss": 0.3478, "step": 19887 }, { "epoch": 192.84819277108434, "grad_norm": 6.709324836730957, "learning_rate": 3.4563106796116503e-06, "loss": 0.1765, "step": 19888 }, { "epoch": 192.8578313253012, "grad_norm": 14.029158592224121, "learning_rate": 3.451456310679612e-06, "loss": 0.2604, "step": 19889 }, { "epoch": 192.86746987951807, "grad_norm": 16.65929412841797, "learning_rate": 3.446601941747573e-06, "loss": 0.2598, "step": 19890 }, { "epoch": 192.87710843373495, "grad_norm": 21.95915985107422, "learning_rate": 3.441747572815534e-06, "loss": 0.2577, "step": 19891 }, { "epoch": 192.8867469879518, "grad_norm": 11.534279823303223, "learning_rate": 3.4368932038834957e-06, "loss": 0.2712, "step": 19892 }, { "epoch": 192.89638554216867, "grad_norm": 20.7202091217041, "learning_rate": 3.4320388349514564e-06, "loss": 0.3163, "step": 19893 }, { "epoch": 192.90602409638555, "grad_norm": 12.681571006774902, "learning_rate": 3.4271844660194176e-06, "loss": 0.4764, "step": 19894 }, { "epoch": 192.9156626506024, "grad_norm": 7.246720314025879, "learning_rate": 3.422330097087379e-06, "loss": 0.2227, "step": 19895 }, { "epoch": 192.92530120481928, "grad_norm": 23.615835189819336, "learning_rate": 3.41747572815534e-06, "loss": 0.2999, "step": 19896 }, { "epoch": 192.93493975903615, "grad_norm": 10.007213592529297, "learning_rate": 3.412621359223301e-06, "loss": 0.1807, "step": 19897 }, { "epoch": 192.944578313253, "grad_norm": 14.100552558898926, "learning_rate": 3.4077669902912626e-06, "loss": 0.2149, "step": 19898 }, { "epoch": 192.95421686746988, "grad_norm": 10.87286376953125, "learning_rate": 3.4029126213592237e-06, "loss": 0.1643, "step": 19899 }, { "epoch": 192.96385542168676, "grad_norm": 3.3961405754089355, "learning_rate": 3.3980582524271844e-06, "loss": 0.0761, "step": 19900 }, { "epoch": 192.9734939759036, "grad_norm": 10.849224090576172, "learning_rate": 3.3932038834951456e-06, "loss": 0.3372, "step": 19901 }, { "epoch": 192.98313253012049, "grad_norm": 12.043728828430176, "learning_rate": 3.388349514563107e-06, "loss": 0.2059, "step": 19902 }, { "epoch": 192.99277108433736, "grad_norm": 13.914925575256348, "learning_rate": 3.383495145631068e-06, "loss": 0.2916, "step": 19903 }, { "epoch": 193.00843373493976, "grad_norm": 8.454968452453613, "learning_rate": 3.378640776699029e-06, "loss": 0.276, "step": 19904 }, { "epoch": 193.01807228915663, "grad_norm": 5.493555545806885, "learning_rate": 3.3737864077669906e-06, "loss": 0.3127, "step": 19905 }, { "epoch": 193.02771084337348, "grad_norm": 36.30319595336914, "learning_rate": 3.3689320388349517e-06, "loss": 0.2664, "step": 19906 }, { "epoch": 193.03734939759036, "grad_norm": 3.178056478500366, "learning_rate": 3.3640776699029125e-06, "loss": 0.1995, "step": 19907 }, { "epoch": 193.04698795180724, "grad_norm": 30.760595321655273, "learning_rate": 3.359223300970874e-06, "loss": 0.2902, "step": 19908 }, { "epoch": 193.0566265060241, "grad_norm": 3.7027177810668945, "learning_rate": 3.354368932038835e-06, "loss": 0.2841, "step": 19909 }, { "epoch": 193.06626506024097, "grad_norm": 34.541229248046875, "learning_rate": 3.349514563106796e-06, "loss": 0.4849, "step": 19910 }, { "epoch": 193.07590361445784, "grad_norm": 13.691951751708984, "learning_rate": 3.344660194174758e-06, "loss": 0.187, "step": 19911 }, { "epoch": 193.0855421686747, "grad_norm": 6.581696033477783, "learning_rate": 3.3398058252427186e-06, "loss": 0.173, "step": 19912 }, { "epoch": 193.09518072289157, "grad_norm": 23.194393157958984, "learning_rate": 3.3349514563106797e-06, "loss": 0.1281, "step": 19913 }, { "epoch": 193.10481927710845, "grad_norm": 73.12779235839844, "learning_rate": 3.3300970873786413e-06, "loss": 0.1061, "step": 19914 }, { "epoch": 193.1144578313253, "grad_norm": 10.006959915161133, "learning_rate": 3.325242718446602e-06, "loss": 0.3761, "step": 19915 }, { "epoch": 193.12409638554217, "grad_norm": 20.470073699951172, "learning_rate": 3.320388349514563e-06, "loss": 0.2546, "step": 19916 }, { "epoch": 193.13373493975902, "grad_norm": 12.55116081237793, "learning_rate": 3.3155339805825248e-06, "loss": 0.3175, "step": 19917 }, { "epoch": 193.1433734939759, "grad_norm": 5.052649974822998, "learning_rate": 3.3106796116504855e-06, "loss": 0.2741, "step": 19918 }, { "epoch": 193.15301204819278, "grad_norm": 27.236326217651367, "learning_rate": 3.3058252427184466e-06, "loss": 0.4197, "step": 19919 }, { "epoch": 193.16265060240963, "grad_norm": 6.919017314910889, "learning_rate": 3.300970873786408e-06, "loss": 0.4759, "step": 19920 }, { "epoch": 193.1722891566265, "grad_norm": 3.250070810317993, "learning_rate": 3.2961165048543693e-06, "loss": 0.2119, "step": 19921 }, { "epoch": 193.18192771084338, "grad_norm": 11.34114933013916, "learning_rate": 3.29126213592233e-06, "loss": 0.2514, "step": 19922 }, { "epoch": 193.19156626506023, "grad_norm": 16.796836853027344, "learning_rate": 3.286407766990291e-06, "loss": 0.2079, "step": 19923 }, { "epoch": 193.2012048192771, "grad_norm": 34.44121551513672, "learning_rate": 3.2815533980582528e-06, "loss": 0.3229, "step": 19924 }, { "epoch": 193.210843373494, "grad_norm": 7.728461265563965, "learning_rate": 3.2766990291262135e-06, "loss": 0.2826, "step": 19925 }, { "epoch": 193.22048192771084, "grad_norm": 13.671236991882324, "learning_rate": 3.2718446601941746e-06, "loss": 0.1605, "step": 19926 }, { "epoch": 193.2301204819277, "grad_norm": 4.471472263336182, "learning_rate": 3.2669902912621362e-06, "loss": 0.236, "step": 19927 }, { "epoch": 193.2397590361446, "grad_norm": 12.38995361328125, "learning_rate": 3.2621359223300974e-06, "loss": 0.2619, "step": 19928 }, { "epoch": 193.24939759036144, "grad_norm": 11.693078994750977, "learning_rate": 3.257281553398058e-06, "loss": 0.215, "step": 19929 }, { "epoch": 193.25903614457832, "grad_norm": 3.2893383502960205, "learning_rate": 3.2524271844660197e-06, "loss": 0.1197, "step": 19930 }, { "epoch": 193.2686746987952, "grad_norm": 25.925966262817383, "learning_rate": 3.247572815533981e-06, "loss": 0.2312, "step": 19931 }, { "epoch": 193.27831325301204, "grad_norm": 11.042470932006836, "learning_rate": 3.2427184466019415e-06, "loss": 0.1512, "step": 19932 }, { "epoch": 193.28795180722892, "grad_norm": 16.144372940063477, "learning_rate": 3.2378640776699035e-06, "loss": 0.2309, "step": 19933 }, { "epoch": 193.29759036144577, "grad_norm": 12.252056121826172, "learning_rate": 3.2330097087378642e-06, "loss": 0.1575, "step": 19934 }, { "epoch": 193.30722891566265, "grad_norm": 97.41959381103516, "learning_rate": 3.2281553398058254e-06, "loss": 0.2614, "step": 19935 }, { "epoch": 193.31686746987953, "grad_norm": 25.48211669921875, "learning_rate": 3.223300970873787e-06, "loss": 0.2331, "step": 19936 }, { "epoch": 193.32650602409637, "grad_norm": 6.9274420738220215, "learning_rate": 3.2184466019417477e-06, "loss": 0.1473, "step": 19937 }, { "epoch": 193.33614457831325, "grad_norm": 76.82955169677734, "learning_rate": 3.213592233009709e-06, "loss": 0.3754, "step": 19938 }, { "epoch": 193.34578313253013, "grad_norm": 44.55927658081055, "learning_rate": 3.2087378640776704e-06, "loss": 0.1069, "step": 19939 }, { "epoch": 193.35542168674698, "grad_norm": 4.946399211883545, "learning_rate": 3.2038834951456315e-06, "loss": 0.2682, "step": 19940 }, { "epoch": 193.36506024096386, "grad_norm": 35.99739456176758, "learning_rate": 3.1990291262135923e-06, "loss": 0.4098, "step": 19941 }, { "epoch": 193.37469879518073, "grad_norm": 49.81039047241211, "learning_rate": 3.1941747572815534e-06, "loss": 0.1183, "step": 19942 }, { "epoch": 193.38433734939758, "grad_norm": 22.199167251586914, "learning_rate": 3.189320388349515e-06, "loss": 0.2574, "step": 19943 }, { "epoch": 193.39397590361446, "grad_norm": 6.840488910675049, "learning_rate": 3.1844660194174757e-06, "loss": 0.2247, "step": 19944 }, { "epoch": 193.40361445783134, "grad_norm": 14.534321784973145, "learning_rate": 3.179611650485437e-06, "loss": 0.2212, "step": 19945 }, { "epoch": 193.4132530120482, "grad_norm": 3.2551262378692627, "learning_rate": 3.1747572815533984e-06, "loss": 0.1393, "step": 19946 }, { "epoch": 193.42289156626506, "grad_norm": 26.726261138916016, "learning_rate": 3.169902912621359e-06, "loss": 0.2082, "step": 19947 }, { "epoch": 193.4325301204819, "grad_norm": 80.31041717529297, "learning_rate": 3.1650485436893203e-06, "loss": 0.1661, "step": 19948 }, { "epoch": 193.4421686746988, "grad_norm": 8.349723815917969, "learning_rate": 3.160194174757282e-06, "loss": 0.1424, "step": 19949 }, { "epoch": 193.45180722891567, "grad_norm": 88.17347717285156, "learning_rate": 3.155339805825243e-06, "loss": 0.351, "step": 19950 }, { "epoch": 193.46144578313252, "grad_norm": 33.78553009033203, "learning_rate": 3.1504854368932037e-06, "loss": 0.361, "step": 19951 }, { "epoch": 193.4710843373494, "grad_norm": 15.566768646240234, "learning_rate": 3.1456310679611653e-06, "loss": 0.1974, "step": 19952 }, { "epoch": 193.48072289156627, "grad_norm": 4.57170295715332, "learning_rate": 3.1407766990291264e-06, "loss": 0.1288, "step": 19953 }, { "epoch": 193.49036144578312, "grad_norm": 36.15656280517578, "learning_rate": 3.135922330097087e-06, "loss": 0.3132, "step": 19954 }, { "epoch": 193.5, "grad_norm": 6.413238048553467, "learning_rate": 3.131067961165049e-06, "loss": 0.3823, "step": 19955 }, { "epoch": 193.50963855421688, "grad_norm": 4.476768970489502, "learning_rate": 3.12621359223301e-06, "loss": 0.1054, "step": 19956 }, { "epoch": 193.51927710843373, "grad_norm": 5.129333019256592, "learning_rate": 3.121359223300971e-06, "loss": 0.0961, "step": 19957 }, { "epoch": 193.5289156626506, "grad_norm": 4.425961971282959, "learning_rate": 3.116504854368932e-06, "loss": 0.1897, "step": 19958 }, { "epoch": 193.53855421686748, "grad_norm": 15.149895668029785, "learning_rate": 3.1116504854368933e-06, "loss": 0.4561, "step": 19959 }, { "epoch": 193.54819277108433, "grad_norm": 43.92034912109375, "learning_rate": 3.1067961165048544e-06, "loss": 0.2762, "step": 19960 }, { "epoch": 193.5578313253012, "grad_norm": 6.600011825561523, "learning_rate": 3.1019417475728156e-06, "loss": 0.2929, "step": 19961 }, { "epoch": 193.5674698795181, "grad_norm": 7.353704929351807, "learning_rate": 3.097087378640777e-06, "loss": 0.2654, "step": 19962 }, { "epoch": 193.57710843373494, "grad_norm": 27.88811683654785, "learning_rate": 3.092233009708738e-06, "loss": 0.1874, "step": 19963 }, { "epoch": 193.5867469879518, "grad_norm": 19.458219528198242, "learning_rate": 3.087378640776699e-06, "loss": 0.1422, "step": 19964 }, { "epoch": 193.59638554216866, "grad_norm": 12.484864234924316, "learning_rate": 3.0825242718446606e-06, "loss": 0.1071, "step": 19965 }, { "epoch": 193.60602409638554, "grad_norm": 9.717238426208496, "learning_rate": 3.0776699029126213e-06, "loss": 0.1468, "step": 19966 }, { "epoch": 193.61566265060242, "grad_norm": 5.222217082977295, "learning_rate": 3.072815533980583e-06, "loss": 0.1747, "step": 19967 }, { "epoch": 193.62530120481927, "grad_norm": 16.798885345458984, "learning_rate": 3.0679611650485436e-06, "loss": 0.3804, "step": 19968 }, { "epoch": 193.63493975903614, "grad_norm": 47.59022903442383, "learning_rate": 3.063106796116505e-06, "loss": 0.295, "step": 19969 }, { "epoch": 193.64457831325302, "grad_norm": 1.2925447225570679, "learning_rate": 3.0582524271844663e-06, "loss": 0.0872, "step": 19970 }, { "epoch": 193.65421686746987, "grad_norm": 3.8247430324554443, "learning_rate": 3.053398058252427e-06, "loss": 0.2662, "step": 19971 }, { "epoch": 193.66385542168675, "grad_norm": 20.900672912597656, "learning_rate": 3.0485436893203886e-06, "loss": 0.1073, "step": 19972 }, { "epoch": 193.67349397590363, "grad_norm": 68.57369232177734, "learning_rate": 3.0436893203883498e-06, "loss": 0.1433, "step": 19973 }, { "epoch": 193.68313253012047, "grad_norm": 11.712677001953125, "learning_rate": 3.038834951456311e-06, "loss": 0.3359, "step": 19974 }, { "epoch": 193.69277108433735, "grad_norm": 15.437050819396973, "learning_rate": 3.033980582524272e-06, "loss": 0.3631, "step": 19975 }, { "epoch": 193.70240963855423, "grad_norm": 3.0507864952087402, "learning_rate": 3.029126213592233e-06, "loss": 0.2055, "step": 19976 }, { "epoch": 193.71204819277108, "grad_norm": 27.45395278930664, "learning_rate": 3.0242718446601943e-06, "loss": 0.2456, "step": 19977 }, { "epoch": 193.72168674698796, "grad_norm": 34.27532958984375, "learning_rate": 3.0194174757281555e-06, "loss": 0.288, "step": 19978 }, { "epoch": 193.73132530120483, "grad_norm": 7.155500411987305, "learning_rate": 3.0145631067961166e-06, "loss": 0.1284, "step": 19979 }, { "epoch": 193.74096385542168, "grad_norm": 33.75149154663086, "learning_rate": 3.0097087378640778e-06, "loss": 0.2125, "step": 19980 }, { "epoch": 193.75060240963856, "grad_norm": 6.755082607269287, "learning_rate": 3.004854368932039e-06, "loss": 0.2125, "step": 19981 }, { "epoch": 193.7602409638554, "grad_norm": 16.252578735351562, "learning_rate": 3e-06, "loss": 0.1781, "step": 19982 }, { "epoch": 193.7698795180723, "grad_norm": 10.915521621704102, "learning_rate": 2.9951456310679612e-06, "loss": 0.3005, "step": 19983 }, { "epoch": 193.77951807228916, "grad_norm": 15.628555297851562, "learning_rate": 2.990291262135923e-06, "loss": 0.3608, "step": 19984 }, { "epoch": 193.789156626506, "grad_norm": 12.440567970275879, "learning_rate": 2.9854368932038835e-06, "loss": 0.1587, "step": 19985 }, { "epoch": 193.7987951807229, "grad_norm": 51.116939544677734, "learning_rate": 2.9805825242718447e-06, "loss": 0.2644, "step": 19986 }, { "epoch": 193.80843373493977, "grad_norm": 30.287935256958008, "learning_rate": 2.975728155339806e-06, "loss": 0.1944, "step": 19987 }, { "epoch": 193.81807228915662, "grad_norm": 5.709720134735107, "learning_rate": 2.970873786407767e-06, "loss": 0.5029, "step": 19988 }, { "epoch": 193.8277108433735, "grad_norm": 5.9226298332214355, "learning_rate": 2.9660194174757285e-06, "loss": 0.3636, "step": 19989 }, { "epoch": 193.83734939759037, "grad_norm": 56.83823013305664, "learning_rate": 2.9611650485436892e-06, "loss": 0.3551, "step": 19990 }, { "epoch": 193.84698795180722, "grad_norm": 14.642549514770508, "learning_rate": 2.956310679611651e-06, "loss": 0.1662, "step": 19991 }, { "epoch": 193.8566265060241, "grad_norm": 5.278178691864014, "learning_rate": 2.951456310679612e-06, "loss": 0.0602, "step": 19992 }, { "epoch": 193.86626506024098, "grad_norm": 3.947901725769043, "learning_rate": 2.9466019417475727e-06, "loss": 0.2663, "step": 19993 }, { "epoch": 193.87590361445783, "grad_norm": 0.8160721659660339, "learning_rate": 2.9417475728155342e-06, "loss": 0.1826, "step": 19994 }, { "epoch": 193.8855421686747, "grad_norm": 6.285401821136475, "learning_rate": 2.9368932038834954e-06, "loss": 0.2045, "step": 19995 }, { "epoch": 193.89518072289155, "grad_norm": 7.596768379211426, "learning_rate": 2.9320388349514565e-06, "loss": 0.0927, "step": 19996 }, { "epoch": 193.90481927710843, "grad_norm": 53.66067123413086, "learning_rate": 2.9271844660194177e-06, "loss": 0.2962, "step": 19997 }, { "epoch": 193.9144578313253, "grad_norm": 20.784271240234375, "learning_rate": 2.922330097087379e-06, "loss": 0.2822, "step": 19998 }, { "epoch": 193.92409638554216, "grad_norm": 16.355937957763672, "learning_rate": 2.91747572815534e-06, "loss": 0.1372, "step": 19999 }, { "epoch": 193.93373493975903, "grad_norm": 81.85647583007812, "learning_rate": 2.912621359223301e-06, "loss": 0.4457, "step": 20000 }, { "epoch": 193.9433734939759, "grad_norm": 65.05027770996094, "learning_rate": 2.9077669902912623e-06, "loss": 0.1941, "step": 20001 }, { "epoch": 193.95301204819276, "grad_norm": 29.452266693115234, "learning_rate": 2.9029126213592234e-06, "loss": 0.2061, "step": 20002 }, { "epoch": 193.96265060240964, "grad_norm": 24.927194595336914, "learning_rate": 2.8980582524271846e-06, "loss": 0.4047, "step": 20003 }, { "epoch": 193.97228915662652, "grad_norm": 15.35805892944336, "learning_rate": 2.8932038834951457e-06, "loss": 0.3237, "step": 20004 }, { "epoch": 193.98192771084337, "grad_norm": 27.48906707763672, "learning_rate": 2.888349514563107e-06, "loss": 0.3536, "step": 20005 }, { "epoch": 193.99156626506024, "grad_norm": 6.748548984527588, "learning_rate": 2.8834951456310684e-06, "loss": 0.2148, "step": 20006 }, { "epoch": 194.00722891566264, "grad_norm": 23.29896354675293, "learning_rate": 2.878640776699029e-06, "loss": 0.1534, "step": 20007 }, { "epoch": 194.01686746987951, "grad_norm": 1.8119852542877197, "learning_rate": 2.8737864077669903e-06, "loss": 0.1836, "step": 20008 }, { "epoch": 194.0265060240964, "grad_norm": 4.740179061889648, "learning_rate": 2.8689320388349514e-06, "loss": 0.1296, "step": 20009 }, { "epoch": 194.03614457831324, "grad_norm": 2.715160608291626, "learning_rate": 2.8640776699029126e-06, "loss": 0.3047, "step": 20010 }, { "epoch": 194.04578313253012, "grad_norm": 4.748963356018066, "learning_rate": 2.859223300970874e-06, "loss": 0.1535, "step": 20011 }, { "epoch": 194.055421686747, "grad_norm": 53.96849060058594, "learning_rate": 2.854368932038835e-06, "loss": 0.2773, "step": 20012 }, { "epoch": 194.06506024096385, "grad_norm": 3.722292423248291, "learning_rate": 2.8495145631067964e-06, "loss": 0.248, "step": 20013 }, { "epoch": 194.07469879518072, "grad_norm": 23.200130462646484, "learning_rate": 2.8446601941747576e-06, "loss": 0.2617, "step": 20014 }, { "epoch": 194.0843373493976, "grad_norm": 29.802379608154297, "learning_rate": 2.8398058252427183e-06, "loss": 0.1498, "step": 20015 }, { "epoch": 194.09397590361445, "grad_norm": 6.482555389404297, "learning_rate": 2.83495145631068e-06, "loss": 0.2041, "step": 20016 }, { "epoch": 194.10361445783133, "grad_norm": 21.073360443115234, "learning_rate": 2.830097087378641e-06, "loss": 0.251, "step": 20017 }, { "epoch": 194.1132530120482, "grad_norm": 38.395591735839844, "learning_rate": 2.825242718446602e-06, "loss": 0.4405, "step": 20018 }, { "epoch": 194.12289156626505, "grad_norm": 28.371946334838867, "learning_rate": 2.8203883495145633e-06, "loss": 0.3293, "step": 20019 }, { "epoch": 194.13253012048193, "grad_norm": 14.877172470092773, "learning_rate": 2.8155339805825245e-06, "loss": 0.2431, "step": 20020 }, { "epoch": 194.1421686746988, "grad_norm": 4.071734428405762, "learning_rate": 2.8106796116504856e-06, "loss": 0.2822, "step": 20021 }, { "epoch": 194.15180722891566, "grad_norm": 17.458515167236328, "learning_rate": 2.8058252427184468e-06, "loss": 0.1809, "step": 20022 }, { "epoch": 194.16144578313254, "grad_norm": 13.987590789794922, "learning_rate": 2.800970873786408e-06, "loss": 0.3209, "step": 20023 }, { "epoch": 194.17108433734938, "grad_norm": 18.315065383911133, "learning_rate": 2.796116504854369e-06, "loss": 0.158, "step": 20024 }, { "epoch": 194.18072289156626, "grad_norm": 32.68824005126953, "learning_rate": 2.79126213592233e-06, "loss": 0.1398, "step": 20025 }, { "epoch": 194.19036144578314, "grad_norm": 8.355802536010742, "learning_rate": 2.7864077669902913e-06, "loss": 0.0819, "step": 20026 }, { "epoch": 194.2, "grad_norm": 10.811864852905273, "learning_rate": 2.7815533980582525e-06, "loss": 0.2359, "step": 20027 }, { "epoch": 194.20963855421687, "grad_norm": 40.501434326171875, "learning_rate": 2.7766990291262136e-06, "loss": 0.2344, "step": 20028 }, { "epoch": 194.21927710843374, "grad_norm": 11.299378395080566, "learning_rate": 2.7718446601941748e-06, "loss": 0.358, "step": 20029 }, { "epoch": 194.2289156626506, "grad_norm": 4.047242641448975, "learning_rate": 2.7669902912621363e-06, "loss": 0.2289, "step": 20030 }, { "epoch": 194.23855421686747, "grad_norm": 8.281394958496094, "learning_rate": 2.762135922330097e-06, "loss": 0.1878, "step": 20031 }, { "epoch": 194.24819277108435, "grad_norm": 70.19935607910156, "learning_rate": 2.757281553398058e-06, "loss": 0.1392, "step": 20032 }, { "epoch": 194.2578313253012, "grad_norm": 39.54213333129883, "learning_rate": 2.7524271844660198e-06, "loss": 0.6646, "step": 20033 }, { "epoch": 194.26746987951807, "grad_norm": 34.477474212646484, "learning_rate": 2.7475728155339805e-06, "loss": 0.2973, "step": 20034 }, { "epoch": 194.27710843373495, "grad_norm": 54.42631912231445, "learning_rate": 2.742718446601942e-06, "loss": 0.1053, "step": 20035 }, { "epoch": 194.2867469879518, "grad_norm": 7.623990535736084, "learning_rate": 2.7378640776699032e-06, "loss": 0.4163, "step": 20036 }, { "epoch": 194.29638554216868, "grad_norm": 26.968711853027344, "learning_rate": 2.733009708737864e-06, "loss": 0.2008, "step": 20037 }, { "epoch": 194.30602409638556, "grad_norm": 45.28232192993164, "learning_rate": 2.7281553398058255e-06, "loss": 0.3919, "step": 20038 }, { "epoch": 194.3156626506024, "grad_norm": 27.23796272277832, "learning_rate": 2.7233009708737862e-06, "loss": 0.4187, "step": 20039 }, { "epoch": 194.32530120481928, "grad_norm": 40.55217742919922, "learning_rate": 2.718446601941748e-06, "loss": 0.3713, "step": 20040 }, { "epoch": 194.33493975903613, "grad_norm": 5.015995979309082, "learning_rate": 2.713592233009709e-06, "loss": 0.1405, "step": 20041 }, { "epoch": 194.344578313253, "grad_norm": 2.003847122192383, "learning_rate": 2.70873786407767e-06, "loss": 0.1, "step": 20042 }, { "epoch": 194.3542168674699, "grad_norm": 1.509896993637085, "learning_rate": 2.7038834951456312e-06, "loss": 0.365, "step": 20043 }, { "epoch": 194.36385542168674, "grad_norm": 62.2078971862793, "learning_rate": 2.6990291262135924e-06, "loss": 0.2582, "step": 20044 }, { "epoch": 194.3734939759036, "grad_norm": 16.267650604248047, "learning_rate": 2.6941747572815535e-06, "loss": 0.3781, "step": 20045 }, { "epoch": 194.3831325301205, "grad_norm": 20.313806533813477, "learning_rate": 2.6893203883495147e-06, "loss": 0.3287, "step": 20046 }, { "epoch": 194.39277108433734, "grad_norm": 6.4005327224731445, "learning_rate": 2.684466019417476e-06, "loss": 0.3977, "step": 20047 }, { "epoch": 194.40240963855422, "grad_norm": 10.684758186340332, "learning_rate": 2.679611650485437e-06, "loss": 0.2969, "step": 20048 }, { "epoch": 194.4120481927711, "grad_norm": 5.10153865814209, "learning_rate": 2.674757281553398e-06, "loss": 0.4773, "step": 20049 }, { "epoch": 194.42168674698794, "grad_norm": 14.123345375061035, "learning_rate": 2.6699029126213593e-06, "loss": 0.2428, "step": 20050 }, { "epoch": 194.43132530120482, "grad_norm": 20.45891571044922, "learning_rate": 2.6650485436893204e-06, "loss": 0.452, "step": 20051 }, { "epoch": 194.4409638554217, "grad_norm": 9.100675582885742, "learning_rate": 2.660194174757282e-06, "loss": 0.139, "step": 20052 }, { "epoch": 194.45060240963855, "grad_norm": 5.753143310546875, "learning_rate": 2.6553398058252427e-06, "loss": 0.1632, "step": 20053 }, { "epoch": 194.46024096385543, "grad_norm": 11.745420455932617, "learning_rate": 2.650485436893204e-06, "loss": 0.0923, "step": 20054 }, { "epoch": 194.46987951807228, "grad_norm": 16.669565200805664, "learning_rate": 2.6456310679611654e-06, "loss": 0.3836, "step": 20055 }, { "epoch": 194.47951807228915, "grad_norm": 18.46263313293457, "learning_rate": 2.640776699029126e-06, "loss": 0.2138, "step": 20056 }, { "epoch": 194.48915662650603, "grad_norm": 24.42369842529297, "learning_rate": 2.6359223300970877e-06, "loss": 0.4028, "step": 20057 }, { "epoch": 194.49879518072288, "grad_norm": 7.322354793548584, "learning_rate": 2.6310679611650484e-06, "loss": 0.1658, "step": 20058 }, { "epoch": 194.50843373493976, "grad_norm": 13.921182632446289, "learning_rate": 2.62621359223301e-06, "loss": 0.1677, "step": 20059 }, { "epoch": 194.51807228915663, "grad_norm": 7.929676055908203, "learning_rate": 2.621359223300971e-06, "loss": 0.1412, "step": 20060 }, { "epoch": 194.52771084337348, "grad_norm": 35.68069076538086, "learning_rate": 2.616504854368932e-06, "loss": 0.5333, "step": 20061 }, { "epoch": 194.53734939759036, "grad_norm": 42.39064407348633, "learning_rate": 2.6116504854368934e-06, "loss": 0.2986, "step": 20062 }, { "epoch": 194.54698795180724, "grad_norm": 1.091833233833313, "learning_rate": 2.6067961165048546e-06, "loss": 0.1003, "step": 20063 }, { "epoch": 194.5566265060241, "grad_norm": 3.1648428440093994, "learning_rate": 2.6019417475728157e-06, "loss": 0.2874, "step": 20064 }, { "epoch": 194.56626506024097, "grad_norm": 20.195621490478516, "learning_rate": 2.597087378640777e-06, "loss": 0.2521, "step": 20065 }, { "epoch": 194.57590361445784, "grad_norm": 13.28279972076416, "learning_rate": 2.592233009708738e-06, "loss": 0.2398, "step": 20066 }, { "epoch": 194.5855421686747, "grad_norm": 30.65127944946289, "learning_rate": 2.587378640776699e-06, "loss": 0.3223, "step": 20067 }, { "epoch": 194.59518072289157, "grad_norm": 20.617250442504883, "learning_rate": 2.5825242718446603e-06, "loss": 0.1678, "step": 20068 }, { "epoch": 194.60481927710845, "grad_norm": 28.979459762573242, "learning_rate": 2.5776699029126214e-06, "loss": 0.2223, "step": 20069 }, { "epoch": 194.6144578313253, "grad_norm": 31.046815872192383, "learning_rate": 2.5728155339805826e-06, "loss": 0.1948, "step": 20070 }, { "epoch": 194.62409638554217, "grad_norm": 33.839141845703125, "learning_rate": 2.5679611650485437e-06, "loss": 0.2225, "step": 20071 }, { "epoch": 194.63373493975902, "grad_norm": 3.3509342670440674, "learning_rate": 2.563106796116505e-06, "loss": 0.0697, "step": 20072 }, { "epoch": 194.6433734939759, "grad_norm": 27.692289352416992, "learning_rate": 2.558252427184466e-06, "loss": 0.263, "step": 20073 }, { "epoch": 194.65301204819278, "grad_norm": 4.001583576202393, "learning_rate": 2.5533980582524276e-06, "loss": 0.1297, "step": 20074 }, { "epoch": 194.66265060240963, "grad_norm": 37.62554931640625, "learning_rate": 2.5485436893203883e-06, "loss": 0.2638, "step": 20075 }, { "epoch": 194.6722891566265, "grad_norm": 11.138190269470215, "learning_rate": 2.5436893203883495e-06, "loss": 0.3657, "step": 20076 }, { "epoch": 194.68192771084338, "grad_norm": 12.406956672668457, "learning_rate": 2.538834951456311e-06, "loss": 0.2186, "step": 20077 }, { "epoch": 194.69156626506023, "grad_norm": 12.506498336791992, "learning_rate": 2.5339805825242718e-06, "loss": 0.2941, "step": 20078 }, { "epoch": 194.7012048192771, "grad_norm": 4.9957051277160645, "learning_rate": 2.5291262135922333e-06, "loss": 0.2123, "step": 20079 }, { "epoch": 194.710843373494, "grad_norm": 59.82468032836914, "learning_rate": 2.524271844660194e-06, "loss": 0.3174, "step": 20080 }, { "epoch": 194.72048192771084, "grad_norm": 3.414307117462158, "learning_rate": 2.5194174757281556e-06, "loss": 0.142, "step": 20081 }, { "epoch": 194.7301204819277, "grad_norm": 2.8525335788726807, "learning_rate": 2.5145631067961168e-06, "loss": 0.1846, "step": 20082 }, { "epoch": 194.7397590361446, "grad_norm": 15.206414222717285, "learning_rate": 2.5097087378640775e-06, "loss": 0.2407, "step": 20083 }, { "epoch": 194.74939759036144, "grad_norm": 25.11567497253418, "learning_rate": 2.504854368932039e-06, "loss": 0.2784, "step": 20084 }, { "epoch": 194.75903614457832, "grad_norm": 10.598575592041016, "learning_rate": 2.5e-06, "loss": 0.2255, "step": 20085 }, { "epoch": 194.76867469879517, "grad_norm": 10.884781837463379, "learning_rate": 2.4951456310679614e-06, "loss": 0.2504, "step": 20086 }, { "epoch": 194.77831325301204, "grad_norm": 45.50873565673828, "learning_rate": 2.4902912621359225e-06, "loss": 0.1857, "step": 20087 }, { "epoch": 194.78795180722892, "grad_norm": 5.761752128601074, "learning_rate": 2.4854368932038836e-06, "loss": 0.1235, "step": 20088 }, { "epoch": 194.79759036144577, "grad_norm": 3.3186697959899902, "learning_rate": 2.4805825242718448e-06, "loss": 0.3184, "step": 20089 }, { "epoch": 194.80722891566265, "grad_norm": 14.450979232788086, "learning_rate": 2.475728155339806e-06, "loss": 0.0812, "step": 20090 }, { "epoch": 194.81686746987953, "grad_norm": 9.416763305664062, "learning_rate": 2.470873786407767e-06, "loss": 0.1159, "step": 20091 }, { "epoch": 194.82650602409637, "grad_norm": 7.122878074645996, "learning_rate": 2.4660194174757282e-06, "loss": 0.4078, "step": 20092 }, { "epoch": 194.83614457831325, "grad_norm": 12.83045482635498, "learning_rate": 2.4611650485436894e-06, "loss": 0.2277, "step": 20093 }, { "epoch": 194.84578313253013, "grad_norm": 19.639928817749023, "learning_rate": 2.4563106796116505e-06, "loss": 0.3762, "step": 20094 }, { "epoch": 194.85542168674698, "grad_norm": 27.474365234375, "learning_rate": 2.4514563106796117e-06, "loss": 0.3854, "step": 20095 }, { "epoch": 194.86506024096386, "grad_norm": 9.991110801696777, "learning_rate": 2.4466019417475732e-06, "loss": 0.1789, "step": 20096 }, { "epoch": 194.87469879518073, "grad_norm": 109.81658935546875, "learning_rate": 2.441747572815534e-06, "loss": 0.3188, "step": 20097 }, { "epoch": 194.88433734939758, "grad_norm": 16.28013038635254, "learning_rate": 2.4368932038834955e-06, "loss": 0.2212, "step": 20098 }, { "epoch": 194.89397590361446, "grad_norm": 5.137087821960449, "learning_rate": 2.4320388349514562e-06, "loss": 0.2412, "step": 20099 }, { "epoch": 194.90361445783134, "grad_norm": 27.823408126831055, "learning_rate": 2.4271844660194174e-06, "loss": 0.2349, "step": 20100 }, { "epoch": 194.9132530120482, "grad_norm": 6.642688751220703, "learning_rate": 2.422330097087379e-06, "loss": 0.1994, "step": 20101 }, { "epoch": 194.92289156626506, "grad_norm": 25.44774055480957, "learning_rate": 2.4174757281553397e-06, "loss": 0.252, "step": 20102 }, { "epoch": 194.9325301204819, "grad_norm": 2.4569711685180664, "learning_rate": 2.4126213592233013e-06, "loss": 0.1945, "step": 20103 }, { "epoch": 194.9421686746988, "grad_norm": 167.01063537597656, "learning_rate": 2.4077669902912624e-06, "loss": 0.2128, "step": 20104 }, { "epoch": 194.95180722891567, "grad_norm": 3.8648993968963623, "learning_rate": 2.402912621359223e-06, "loss": 0.2603, "step": 20105 }, { "epoch": 194.96144578313252, "grad_norm": 17.070966720581055, "learning_rate": 2.3980582524271847e-06, "loss": 0.1913, "step": 20106 }, { "epoch": 194.9710843373494, "grad_norm": 4.666495323181152, "learning_rate": 2.393203883495146e-06, "loss": 0.2376, "step": 20107 }, { "epoch": 194.98072289156627, "grad_norm": 23.813129425048828, "learning_rate": 2.388349514563107e-06, "loss": 0.2157, "step": 20108 }, { "epoch": 194.99036144578312, "grad_norm": 9.747291564941406, "learning_rate": 2.383495145631068e-06, "loss": 0.2433, "step": 20109 }, { "epoch": 195.00602409638554, "grad_norm": 26.508455276489258, "learning_rate": 2.3786407766990293e-06, "loss": 0.1841, "step": 20110 }, { "epoch": 195.01566265060242, "grad_norm": 15.762093544006348, "learning_rate": 2.3737864077669904e-06, "loss": 0.2327, "step": 20111 }, { "epoch": 195.02530120481927, "grad_norm": 29.13924217224121, "learning_rate": 2.3689320388349516e-06, "loss": 0.336, "step": 20112 }, { "epoch": 195.03493975903615, "grad_norm": 12.531272888183594, "learning_rate": 2.3640776699029127e-06, "loss": 0.2218, "step": 20113 }, { "epoch": 195.044578313253, "grad_norm": 12.788217544555664, "learning_rate": 2.359223300970874e-06, "loss": 0.1853, "step": 20114 }, { "epoch": 195.05421686746988, "grad_norm": 13.57178020477295, "learning_rate": 2.354368932038835e-06, "loss": 0.1963, "step": 20115 }, { "epoch": 195.06385542168675, "grad_norm": 2.6695570945739746, "learning_rate": 2.349514563106796e-06, "loss": 0.0749, "step": 20116 }, { "epoch": 195.0734939759036, "grad_norm": 4.769087791442871, "learning_rate": 2.3446601941747573e-06, "loss": 0.2058, "step": 20117 }, { "epoch": 195.08313253012048, "grad_norm": 2.800157308578491, "learning_rate": 2.339805825242719e-06, "loss": 0.3032, "step": 20118 }, { "epoch": 195.09277108433736, "grad_norm": 6.998040676116943, "learning_rate": 2.3349514563106796e-06, "loss": 0.1952, "step": 20119 }, { "epoch": 195.1024096385542, "grad_norm": 8.204352378845215, "learning_rate": 2.330097087378641e-06, "loss": 0.0929, "step": 20120 }, { "epoch": 195.11204819277108, "grad_norm": 5.6757330894470215, "learning_rate": 2.325242718446602e-06, "loss": 0.1796, "step": 20121 }, { "epoch": 195.12168674698796, "grad_norm": 26.96636390686035, "learning_rate": 2.320388349514563e-06, "loss": 0.1937, "step": 20122 }, { "epoch": 195.1313253012048, "grad_norm": 5.70195198059082, "learning_rate": 2.3155339805825246e-06, "loss": 0.1804, "step": 20123 }, { "epoch": 195.1409638554217, "grad_norm": 15.369386672973633, "learning_rate": 2.3106796116504853e-06, "loss": 0.2613, "step": 20124 }, { "epoch": 195.15060240963857, "grad_norm": 6.492893218994141, "learning_rate": 2.305825242718447e-06, "loss": 0.1688, "step": 20125 }, { "epoch": 195.16024096385541, "grad_norm": 4.939613342285156, "learning_rate": 2.300970873786408e-06, "loss": 0.3852, "step": 20126 }, { "epoch": 195.1698795180723, "grad_norm": 37.22685241699219, "learning_rate": 2.296116504854369e-06, "loss": 0.426, "step": 20127 }, { "epoch": 195.17951807228917, "grad_norm": 13.329302787780762, "learning_rate": 2.2912621359223303e-06, "loss": 0.1855, "step": 20128 }, { "epoch": 195.18915662650602, "grad_norm": 3.0760202407836914, "learning_rate": 2.286407766990291e-06, "loss": 0.202, "step": 20129 }, { "epoch": 195.1987951807229, "grad_norm": 27.17254066467285, "learning_rate": 2.2815533980582526e-06, "loss": 0.1426, "step": 20130 }, { "epoch": 195.20843373493975, "grad_norm": 15.493535995483398, "learning_rate": 2.2766990291262138e-06, "loss": 0.4919, "step": 20131 }, { "epoch": 195.21807228915662, "grad_norm": 26.07163429260254, "learning_rate": 2.271844660194175e-06, "loss": 0.3208, "step": 20132 }, { "epoch": 195.2277108433735, "grad_norm": 14.643075942993164, "learning_rate": 2.266990291262136e-06, "loss": 0.3677, "step": 20133 }, { "epoch": 195.23734939759035, "grad_norm": 18.462854385375977, "learning_rate": 2.262135922330097e-06, "loss": 0.4616, "step": 20134 }, { "epoch": 195.24698795180723, "grad_norm": 3.38934588432312, "learning_rate": 2.2572815533980583e-06, "loss": 0.1855, "step": 20135 }, { "epoch": 195.2566265060241, "grad_norm": 3.419395685195923, "learning_rate": 2.2524271844660195e-06, "loss": 0.3314, "step": 20136 }, { "epoch": 195.26626506024095, "grad_norm": 21.90935516357422, "learning_rate": 2.247572815533981e-06, "loss": 0.32, "step": 20137 }, { "epoch": 195.27590361445783, "grad_norm": 10.376776695251465, "learning_rate": 2.2427184466019418e-06, "loss": 0.1334, "step": 20138 }, { "epoch": 195.2855421686747, "grad_norm": 24.11469268798828, "learning_rate": 2.237864077669903e-06, "loss": 0.2396, "step": 20139 }, { "epoch": 195.29518072289156, "grad_norm": 7.456814289093018, "learning_rate": 2.233009708737864e-06, "loss": 0.2368, "step": 20140 }, { "epoch": 195.30481927710844, "grad_norm": 13.751172065734863, "learning_rate": 2.2281553398058252e-06, "loss": 0.4164, "step": 20141 }, { "epoch": 195.3144578313253, "grad_norm": 11.659123420715332, "learning_rate": 2.2233009708737868e-06, "loss": 0.3009, "step": 20142 }, { "epoch": 195.32409638554216, "grad_norm": 2.0525810718536377, "learning_rate": 2.2184466019417475e-06, "loss": 0.1873, "step": 20143 }, { "epoch": 195.33373493975904, "grad_norm": 7.281027317047119, "learning_rate": 2.2135922330097087e-06, "loss": 0.1097, "step": 20144 }, { "epoch": 195.34337349397592, "grad_norm": 8.523512840270996, "learning_rate": 2.2087378640776702e-06, "loss": 0.2715, "step": 20145 }, { "epoch": 195.35301204819277, "grad_norm": 6.125074863433838, "learning_rate": 2.203883495145631e-06, "loss": 0.2306, "step": 20146 }, { "epoch": 195.36265060240964, "grad_norm": 28.453384399414062, "learning_rate": 2.1990291262135925e-06, "loss": 0.1905, "step": 20147 }, { "epoch": 195.3722891566265, "grad_norm": 10.464801788330078, "learning_rate": 2.1941747572815537e-06, "loss": 0.3243, "step": 20148 }, { "epoch": 195.38192771084337, "grad_norm": 45.589599609375, "learning_rate": 2.189320388349515e-06, "loss": 0.2338, "step": 20149 }, { "epoch": 195.39156626506025, "grad_norm": 34.87397003173828, "learning_rate": 2.184466019417476e-06, "loss": 0.1794, "step": 20150 }, { "epoch": 195.4012048192771, "grad_norm": 8.454657554626465, "learning_rate": 2.1796116504854367e-06, "loss": 0.3356, "step": 20151 }, { "epoch": 195.41084337349398, "grad_norm": 13.06851863861084, "learning_rate": 2.1747572815533982e-06, "loss": 0.3363, "step": 20152 }, { "epoch": 195.42048192771085, "grad_norm": 4.557186126708984, "learning_rate": 2.1699029126213594e-06, "loss": 0.2387, "step": 20153 }, { "epoch": 195.4301204819277, "grad_norm": 4.3978590965271, "learning_rate": 2.1650485436893205e-06, "loss": 0.2671, "step": 20154 }, { "epoch": 195.43975903614458, "grad_norm": 28.78054428100586, "learning_rate": 2.1601941747572817e-06, "loss": 0.2986, "step": 20155 }, { "epoch": 195.44939759036146, "grad_norm": 10.674775123596191, "learning_rate": 2.155339805825243e-06, "loss": 0.3396, "step": 20156 }, { "epoch": 195.4590361445783, "grad_norm": 15.67960262298584, "learning_rate": 2.150485436893204e-06, "loss": 0.3291, "step": 20157 }, { "epoch": 195.46867469879518, "grad_norm": 13.729869842529297, "learning_rate": 2.145631067961165e-06, "loss": 0.3848, "step": 20158 }, { "epoch": 195.47831325301206, "grad_norm": 9.277822494506836, "learning_rate": 2.1407766990291267e-06, "loss": 0.2111, "step": 20159 }, { "epoch": 195.4879518072289, "grad_norm": 5.5349273681640625, "learning_rate": 2.1359223300970874e-06, "loss": 0.1809, "step": 20160 }, { "epoch": 195.4975903614458, "grad_norm": 7.713726043701172, "learning_rate": 2.1310679611650486e-06, "loss": 0.1275, "step": 20161 }, { "epoch": 195.50722891566264, "grad_norm": 40.68922424316406, "learning_rate": 2.1262135922330097e-06, "loss": 0.2993, "step": 20162 }, { "epoch": 195.51686746987951, "grad_norm": 133.1302947998047, "learning_rate": 2.121359223300971e-06, "loss": 0.2237, "step": 20163 }, { "epoch": 195.5265060240964, "grad_norm": 73.81166076660156, "learning_rate": 2.1165048543689324e-06, "loss": 0.1632, "step": 20164 }, { "epoch": 195.53614457831324, "grad_norm": 4.714239120483398, "learning_rate": 2.111650485436893e-06, "loss": 0.2578, "step": 20165 }, { "epoch": 195.54578313253012, "grad_norm": 8.461483001708984, "learning_rate": 2.1067961165048547e-06, "loss": 0.1928, "step": 20166 }, { "epoch": 195.555421686747, "grad_norm": 5.299306392669678, "learning_rate": 2.101941747572816e-06, "loss": 0.1, "step": 20167 }, { "epoch": 195.56506024096385, "grad_norm": 28.68057632446289, "learning_rate": 2.0970873786407766e-06, "loss": 0.5455, "step": 20168 }, { "epoch": 195.57469879518072, "grad_norm": 12.410776138305664, "learning_rate": 2.092233009708738e-06, "loss": 0.2993, "step": 20169 }, { "epoch": 195.5843373493976, "grad_norm": 4.173856735229492, "learning_rate": 2.087378640776699e-06, "loss": 0.1749, "step": 20170 }, { "epoch": 195.59397590361445, "grad_norm": 39.8309211730957, "learning_rate": 2.0825242718446604e-06, "loss": 0.1967, "step": 20171 }, { "epoch": 195.60361445783133, "grad_norm": 13.761249542236328, "learning_rate": 2.0776699029126216e-06, "loss": 0.2438, "step": 20172 }, { "epoch": 195.6132530120482, "grad_norm": 12.031877517700195, "learning_rate": 2.0728155339805823e-06, "loss": 0.2804, "step": 20173 }, { "epoch": 195.62289156626505, "grad_norm": 13.290201187133789, "learning_rate": 2.067961165048544e-06, "loss": 0.3375, "step": 20174 }, { "epoch": 195.63253012048193, "grad_norm": 24.746807098388672, "learning_rate": 2.063106796116505e-06, "loss": 0.3773, "step": 20175 }, { "epoch": 195.6421686746988, "grad_norm": 8.404343605041504, "learning_rate": 2.058252427184466e-06, "loss": 0.3112, "step": 20176 }, { "epoch": 195.65180722891566, "grad_norm": 8.847917556762695, "learning_rate": 2.0533980582524273e-06, "loss": 0.1997, "step": 20177 }, { "epoch": 195.66144578313254, "grad_norm": 24.76300048828125, "learning_rate": 2.0485436893203885e-06, "loss": 0.3016, "step": 20178 }, { "epoch": 195.67108433734938, "grad_norm": 8.830366134643555, "learning_rate": 2.0436893203883496e-06, "loss": 0.1754, "step": 20179 }, { "epoch": 195.68072289156626, "grad_norm": 18.00572967529297, "learning_rate": 2.0388349514563107e-06, "loss": 0.24, "step": 20180 }, { "epoch": 195.69036144578314, "grad_norm": 11.456793785095215, "learning_rate": 2.033980582524272e-06, "loss": 0.1251, "step": 20181 }, { "epoch": 195.7, "grad_norm": 4.900350093841553, "learning_rate": 2.029126213592233e-06, "loss": 0.2719, "step": 20182 }, { "epoch": 195.70963855421687, "grad_norm": 14.981196403503418, "learning_rate": 2.024271844660194e-06, "loss": 0.3298, "step": 20183 }, { "epoch": 195.71927710843374, "grad_norm": 28.492015838623047, "learning_rate": 2.0194174757281553e-06, "loss": 0.4616, "step": 20184 }, { "epoch": 195.7289156626506, "grad_norm": 7.394037246704102, "learning_rate": 2.0145631067961165e-06, "loss": 0.178, "step": 20185 }, { "epoch": 195.73855421686747, "grad_norm": 5.099673271179199, "learning_rate": 2.009708737864078e-06, "loss": 0.0546, "step": 20186 }, { "epoch": 195.74819277108435, "grad_norm": 24.45915412902832, "learning_rate": 2.0048543689320388e-06, "loss": 0.3409, "step": 20187 }, { "epoch": 195.7578313253012, "grad_norm": 14.690584182739258, "learning_rate": 2.0000000000000003e-06, "loss": 0.3016, "step": 20188 }, { "epoch": 195.76746987951807, "grad_norm": 9.754364967346191, "learning_rate": 1.9951456310679615e-06, "loss": 0.1204, "step": 20189 }, { "epoch": 195.77710843373495, "grad_norm": 10.359598159790039, "learning_rate": 1.990291262135922e-06, "loss": 0.1336, "step": 20190 }, { "epoch": 195.7867469879518, "grad_norm": 21.909107208251953, "learning_rate": 1.9854368932038838e-06, "loss": 0.3893, "step": 20191 }, { "epoch": 195.79638554216868, "grad_norm": 16.339672088623047, "learning_rate": 1.9805825242718445e-06, "loss": 0.6092, "step": 20192 }, { "epoch": 195.80602409638556, "grad_norm": 9.282222747802734, "learning_rate": 1.975728155339806e-06, "loss": 0.2226, "step": 20193 }, { "epoch": 195.8156626506024, "grad_norm": 13.374844551086426, "learning_rate": 1.970873786407767e-06, "loss": 0.3999, "step": 20194 }, { "epoch": 195.82530120481928, "grad_norm": 23.194530487060547, "learning_rate": 1.966019417475728e-06, "loss": 0.0929, "step": 20195 }, { "epoch": 195.83493975903613, "grad_norm": 9.140671730041504, "learning_rate": 1.9611650485436895e-06, "loss": 0.2914, "step": 20196 }, { "epoch": 195.844578313253, "grad_norm": 22.96380043029785, "learning_rate": 1.9563106796116506e-06, "loss": 0.2639, "step": 20197 }, { "epoch": 195.8542168674699, "grad_norm": 9.03053092956543, "learning_rate": 1.951456310679612e-06, "loss": 0.209, "step": 20198 }, { "epoch": 195.86385542168674, "grad_norm": 7.373568534851074, "learning_rate": 1.946601941747573e-06, "loss": 0.2515, "step": 20199 }, { "epoch": 195.8734939759036, "grad_norm": 10.407747268676758, "learning_rate": 1.941747572815534e-06, "loss": 0.2127, "step": 20200 }, { "epoch": 195.8831325301205, "grad_norm": 7.799955368041992, "learning_rate": 1.9368932038834952e-06, "loss": 0.1577, "step": 20201 }, { "epoch": 195.89277108433734, "grad_norm": 5.497785568237305, "learning_rate": 1.9320388349514564e-06, "loss": 0.0971, "step": 20202 }, { "epoch": 195.90240963855422, "grad_norm": 1.3647825717926025, "learning_rate": 1.9271844660194175e-06, "loss": 0.2264, "step": 20203 }, { "epoch": 195.9120481927711, "grad_norm": 27.900075912475586, "learning_rate": 1.9223300970873787e-06, "loss": 0.4036, "step": 20204 }, { "epoch": 195.92168674698794, "grad_norm": 32.058387756347656, "learning_rate": 1.91747572815534e-06, "loss": 0.1555, "step": 20205 }, { "epoch": 195.93132530120482, "grad_norm": 8.691458702087402, "learning_rate": 1.912621359223301e-06, "loss": 0.1697, "step": 20206 }, { "epoch": 195.9409638554217, "grad_norm": 42.66501998901367, "learning_rate": 1.907766990291262e-06, "loss": 0.2653, "step": 20207 }, { "epoch": 195.95060240963855, "grad_norm": 16.510461807250977, "learning_rate": 1.9029126213592235e-06, "loss": 0.2852, "step": 20208 }, { "epoch": 195.96024096385543, "grad_norm": 14.94073486328125, "learning_rate": 1.8980582524271844e-06, "loss": 0.3174, "step": 20209 }, { "epoch": 195.96987951807228, "grad_norm": 5.832249641418457, "learning_rate": 1.8932038834951458e-06, "loss": 0.1393, "step": 20210 }, { "epoch": 195.97951807228915, "grad_norm": 6.273739814758301, "learning_rate": 1.8883495145631067e-06, "loss": 0.2436, "step": 20211 }, { "epoch": 195.98915662650603, "grad_norm": 7.804363250732422, "learning_rate": 1.883495145631068e-06, "loss": 0.2905, "step": 20212 }, { "epoch": 196.00481927710842, "grad_norm": 6.147828102111816, "learning_rate": 1.8786407766990294e-06, "loss": 0.269, "step": 20213 }, { "epoch": 196.0144578313253, "grad_norm": 17.374568939208984, "learning_rate": 1.8737864077669903e-06, "loss": 0.3328, "step": 20214 }, { "epoch": 196.02409638554218, "grad_norm": 5.234609603881836, "learning_rate": 1.8689320388349515e-06, "loss": 0.1007, "step": 20215 }, { "epoch": 196.03373493975903, "grad_norm": 13.741357803344727, "learning_rate": 1.8640776699029128e-06, "loss": 0.192, "step": 20216 }, { "epoch": 196.0433734939759, "grad_norm": 13.306642532348633, "learning_rate": 1.8592233009708738e-06, "loss": 0.1467, "step": 20217 }, { "epoch": 196.05301204819278, "grad_norm": 8.786718368530273, "learning_rate": 1.8543689320388351e-06, "loss": 0.2063, "step": 20218 }, { "epoch": 196.06265060240963, "grad_norm": 38.262393951416016, "learning_rate": 1.8495145631067963e-06, "loss": 0.3014, "step": 20219 }, { "epoch": 196.0722891566265, "grad_norm": 7.3897294998168945, "learning_rate": 1.8446601941747572e-06, "loss": 0.2814, "step": 20220 }, { "epoch": 196.08192771084336, "grad_norm": 4.911440849304199, "learning_rate": 1.8398058252427186e-06, "loss": 0.0753, "step": 20221 }, { "epoch": 196.09156626506024, "grad_norm": 11.028624534606934, "learning_rate": 1.8349514563106795e-06, "loss": 0.3583, "step": 20222 }, { "epoch": 196.10120481927711, "grad_norm": 1.6434956789016724, "learning_rate": 1.8300970873786409e-06, "loss": 0.1618, "step": 20223 }, { "epoch": 196.11084337349396, "grad_norm": 12.409689903259277, "learning_rate": 1.8252427184466022e-06, "loss": 0.1784, "step": 20224 }, { "epoch": 196.12048192771084, "grad_norm": 11.143086433410645, "learning_rate": 1.8203883495145631e-06, "loss": 0.2587, "step": 20225 }, { "epoch": 196.13012048192772, "grad_norm": 19.69713020324707, "learning_rate": 1.8155339805825243e-06, "loss": 0.318, "step": 20226 }, { "epoch": 196.13975903614457, "grad_norm": 11.205492973327637, "learning_rate": 1.8106796116504857e-06, "loss": 0.2028, "step": 20227 }, { "epoch": 196.14939759036145, "grad_norm": 5.349958896636963, "learning_rate": 1.8058252427184466e-06, "loss": 0.111, "step": 20228 }, { "epoch": 196.15903614457832, "grad_norm": 18.660812377929688, "learning_rate": 1.800970873786408e-06, "loss": 0.2733, "step": 20229 }, { "epoch": 196.16867469879517, "grad_norm": 8.224407196044922, "learning_rate": 1.7961165048543689e-06, "loss": 0.1939, "step": 20230 }, { "epoch": 196.17831325301205, "grad_norm": 3.761215925216675, "learning_rate": 1.7912621359223302e-06, "loss": 0.1777, "step": 20231 }, { "epoch": 196.18795180722893, "grad_norm": 4.647323131561279, "learning_rate": 1.7864077669902914e-06, "loss": 0.0935, "step": 20232 }, { "epoch": 196.19759036144578, "grad_norm": 20.82181739807129, "learning_rate": 1.7815533980582523e-06, "loss": 0.1769, "step": 20233 }, { "epoch": 196.20722891566265, "grad_norm": 24.81338119506836, "learning_rate": 1.7766990291262137e-06, "loss": 0.2503, "step": 20234 }, { "epoch": 196.21686746987953, "grad_norm": 20.045303344726562, "learning_rate": 1.771844660194175e-06, "loss": 0.1785, "step": 20235 }, { "epoch": 196.22650602409638, "grad_norm": 22.381906509399414, "learning_rate": 1.766990291262136e-06, "loss": 0.3788, "step": 20236 }, { "epoch": 196.23614457831326, "grad_norm": 229.0798797607422, "learning_rate": 1.7621359223300971e-06, "loss": 0.159, "step": 20237 }, { "epoch": 196.2457831325301, "grad_norm": 19.90760040283203, "learning_rate": 1.7572815533980585e-06, "loss": 0.2286, "step": 20238 }, { "epoch": 196.25542168674698, "grad_norm": 30.89577865600586, "learning_rate": 1.7524271844660194e-06, "loss": 0.4419, "step": 20239 }, { "epoch": 196.26506024096386, "grad_norm": 44.091209411621094, "learning_rate": 1.7475728155339808e-06, "loss": 0.3598, "step": 20240 }, { "epoch": 196.2746987951807, "grad_norm": 14.092370986938477, "learning_rate": 1.7427184466019417e-06, "loss": 0.5433, "step": 20241 }, { "epoch": 196.2843373493976, "grad_norm": 21.262460708618164, "learning_rate": 1.737864077669903e-06, "loss": 0.2435, "step": 20242 }, { "epoch": 196.29397590361447, "grad_norm": 6.131903648376465, "learning_rate": 1.7330097087378642e-06, "loss": 0.2272, "step": 20243 }, { "epoch": 196.30361445783132, "grad_norm": 11.565910339355469, "learning_rate": 1.7281553398058251e-06, "loss": 0.3233, "step": 20244 }, { "epoch": 196.3132530120482, "grad_norm": 4.681595802307129, "learning_rate": 1.7233009708737865e-06, "loss": 0.1116, "step": 20245 }, { "epoch": 196.32289156626507, "grad_norm": 10.988563537597656, "learning_rate": 1.7184466019417478e-06, "loss": 0.1893, "step": 20246 }, { "epoch": 196.33253012048192, "grad_norm": 2.6045820713043213, "learning_rate": 1.7135922330097088e-06, "loss": 0.0593, "step": 20247 }, { "epoch": 196.3421686746988, "grad_norm": 27.43682098388672, "learning_rate": 1.70873786407767e-06, "loss": 0.4527, "step": 20248 }, { "epoch": 196.35180722891567, "grad_norm": 9.086945533752441, "learning_rate": 1.7038834951456313e-06, "loss": 0.206, "step": 20249 }, { "epoch": 196.36144578313252, "grad_norm": 19.802600860595703, "learning_rate": 1.6990291262135922e-06, "loss": 0.4008, "step": 20250 }, { "epoch": 196.3710843373494, "grad_norm": 14.1509370803833, "learning_rate": 1.6941747572815536e-06, "loss": 0.3411, "step": 20251 }, { "epoch": 196.38072289156628, "grad_norm": 4.956682205200195, "learning_rate": 1.6893203883495145e-06, "loss": 0.2996, "step": 20252 }, { "epoch": 196.39036144578313, "grad_norm": 13.403027534484863, "learning_rate": 1.6844660194174759e-06, "loss": 0.345, "step": 20253 }, { "epoch": 196.4, "grad_norm": 3.1368398666381836, "learning_rate": 1.679611650485437e-06, "loss": 0.1957, "step": 20254 }, { "epoch": 196.40963855421685, "grad_norm": 4.296817779541016, "learning_rate": 1.674757281553398e-06, "loss": 0.1859, "step": 20255 }, { "epoch": 196.41927710843373, "grad_norm": 10.941267967224121, "learning_rate": 1.6699029126213593e-06, "loss": 0.146, "step": 20256 }, { "epoch": 196.4289156626506, "grad_norm": 34.77444076538086, "learning_rate": 1.6650485436893207e-06, "loss": 0.1445, "step": 20257 }, { "epoch": 196.43855421686746, "grad_norm": 4.892519474029541, "learning_rate": 1.6601941747572816e-06, "loss": 0.1869, "step": 20258 }, { "epoch": 196.44819277108434, "grad_norm": 16.814455032348633, "learning_rate": 1.6553398058252427e-06, "loss": 0.2506, "step": 20259 }, { "epoch": 196.4578313253012, "grad_norm": 12.54500961303711, "learning_rate": 1.650485436893204e-06, "loss": 0.2632, "step": 20260 }, { "epoch": 196.46746987951806, "grad_norm": 7.593749523162842, "learning_rate": 1.645631067961165e-06, "loss": 0.1213, "step": 20261 }, { "epoch": 196.47710843373494, "grad_norm": 11.50998306274414, "learning_rate": 1.6407766990291264e-06, "loss": 0.2326, "step": 20262 }, { "epoch": 196.48674698795182, "grad_norm": 8.648585319519043, "learning_rate": 1.6359223300970873e-06, "loss": 0.195, "step": 20263 }, { "epoch": 196.49638554216867, "grad_norm": 7.683798313140869, "learning_rate": 1.6310679611650487e-06, "loss": 0.2159, "step": 20264 }, { "epoch": 196.50602409638554, "grad_norm": 9.051285743713379, "learning_rate": 1.6262135922330098e-06, "loss": 0.2383, "step": 20265 }, { "epoch": 196.51566265060242, "grad_norm": 19.084701538085938, "learning_rate": 1.6213592233009708e-06, "loss": 0.3008, "step": 20266 }, { "epoch": 196.52530120481927, "grad_norm": 4.77796745300293, "learning_rate": 1.6165048543689321e-06, "loss": 0.1054, "step": 20267 }, { "epoch": 196.53493975903615, "grad_norm": 4.827019214630127, "learning_rate": 1.6116504854368935e-06, "loss": 0.2299, "step": 20268 }, { "epoch": 196.544578313253, "grad_norm": 12.485587120056152, "learning_rate": 1.6067961165048544e-06, "loss": 0.2217, "step": 20269 }, { "epoch": 196.55421686746988, "grad_norm": 4.925459861755371, "learning_rate": 1.6019417475728158e-06, "loss": 0.4331, "step": 20270 }, { "epoch": 196.56385542168675, "grad_norm": 26.033336639404297, "learning_rate": 1.5970873786407767e-06, "loss": 0.2913, "step": 20271 }, { "epoch": 196.5734939759036, "grad_norm": 20.043357849121094, "learning_rate": 1.5922330097087378e-06, "loss": 0.3633, "step": 20272 }, { "epoch": 196.58313253012048, "grad_norm": 24.44892120361328, "learning_rate": 1.5873786407766992e-06, "loss": 0.1956, "step": 20273 }, { "epoch": 196.59277108433736, "grad_norm": 23.133541107177734, "learning_rate": 1.5825242718446601e-06, "loss": 0.2572, "step": 20274 }, { "epoch": 196.6024096385542, "grad_norm": 16.539113998413086, "learning_rate": 1.5776699029126215e-06, "loss": 0.2119, "step": 20275 }, { "epoch": 196.61204819277108, "grad_norm": 23.361408233642578, "learning_rate": 1.5728155339805826e-06, "loss": 0.1472, "step": 20276 }, { "epoch": 196.62168674698796, "grad_norm": 48.96040725708008, "learning_rate": 1.5679611650485436e-06, "loss": 0.5465, "step": 20277 }, { "epoch": 196.6313253012048, "grad_norm": 5.693526744842529, "learning_rate": 1.563106796116505e-06, "loss": 0.1865, "step": 20278 }, { "epoch": 196.6409638554217, "grad_norm": 15.59093189239502, "learning_rate": 1.558252427184466e-06, "loss": 0.261, "step": 20279 }, { "epoch": 196.65060240963857, "grad_norm": 24.286476135253906, "learning_rate": 1.5533980582524272e-06, "loss": 0.2916, "step": 20280 }, { "epoch": 196.66024096385541, "grad_norm": 3.144426107406616, "learning_rate": 1.5485436893203886e-06, "loss": 0.1061, "step": 20281 }, { "epoch": 196.6698795180723, "grad_norm": 15.291264533996582, "learning_rate": 1.5436893203883495e-06, "loss": 0.2897, "step": 20282 }, { "epoch": 196.67951807228917, "grad_norm": 29.077051162719727, "learning_rate": 1.5388349514563107e-06, "loss": 0.4283, "step": 20283 }, { "epoch": 196.68915662650602, "grad_norm": 26.21852684020996, "learning_rate": 1.5339805825242718e-06, "loss": 0.3856, "step": 20284 }, { "epoch": 196.6987951807229, "grad_norm": 4.492645740509033, "learning_rate": 1.5291262135922332e-06, "loss": 0.1862, "step": 20285 }, { "epoch": 196.70843373493975, "grad_norm": 19.150442123413086, "learning_rate": 1.5242718446601943e-06, "loss": 0.2482, "step": 20286 }, { "epoch": 196.71807228915662, "grad_norm": 6.613364219665527, "learning_rate": 1.5194174757281555e-06, "loss": 0.2854, "step": 20287 }, { "epoch": 196.7277108433735, "grad_norm": 18.308828353881836, "learning_rate": 1.5145631067961166e-06, "loss": 0.0761, "step": 20288 }, { "epoch": 196.73734939759035, "grad_norm": 20.48996925354004, "learning_rate": 1.5097087378640777e-06, "loss": 0.2585, "step": 20289 }, { "epoch": 196.74698795180723, "grad_norm": 22.378738403320312, "learning_rate": 1.5048543689320389e-06, "loss": 0.3948, "step": 20290 }, { "epoch": 196.7566265060241, "grad_norm": 5.878901958465576, "learning_rate": 1.5e-06, "loss": 0.1607, "step": 20291 }, { "epoch": 196.76626506024095, "grad_norm": 23.455612182617188, "learning_rate": 1.4951456310679614e-06, "loss": 0.2805, "step": 20292 }, { "epoch": 196.77590361445783, "grad_norm": 13.250185012817383, "learning_rate": 1.4902912621359223e-06, "loss": 0.241, "step": 20293 }, { "epoch": 196.7855421686747, "grad_norm": 6.087864398956299, "learning_rate": 1.4854368932038835e-06, "loss": 0.1764, "step": 20294 }, { "epoch": 196.79518072289156, "grad_norm": 18.555931091308594, "learning_rate": 1.4805825242718446e-06, "loss": 0.1764, "step": 20295 }, { "epoch": 196.80481927710844, "grad_norm": 6.436002731323242, "learning_rate": 1.475728155339806e-06, "loss": 0.2888, "step": 20296 }, { "epoch": 196.8144578313253, "grad_norm": 23.583385467529297, "learning_rate": 1.4708737864077671e-06, "loss": 0.2768, "step": 20297 }, { "epoch": 196.82409638554216, "grad_norm": 2.790982246398926, "learning_rate": 1.4660194174757283e-06, "loss": 0.151, "step": 20298 }, { "epoch": 196.83373493975904, "grad_norm": 8.239583015441895, "learning_rate": 1.4611650485436894e-06, "loss": 0.2441, "step": 20299 }, { "epoch": 196.84337349397592, "grad_norm": 23.26832389831543, "learning_rate": 1.4563106796116506e-06, "loss": 0.2804, "step": 20300 }, { "epoch": 196.85301204819277, "grad_norm": 12.552424430847168, "learning_rate": 1.4514563106796117e-06, "loss": 0.0552, "step": 20301 }, { "epoch": 196.86265060240964, "grad_norm": 29.08009910583496, "learning_rate": 1.4466019417475729e-06, "loss": 0.2429, "step": 20302 }, { "epoch": 196.8722891566265, "grad_norm": 15.387149810791016, "learning_rate": 1.4417475728155342e-06, "loss": 0.3528, "step": 20303 }, { "epoch": 196.88192771084337, "grad_norm": 28.4330997467041, "learning_rate": 1.4368932038834951e-06, "loss": 0.2966, "step": 20304 }, { "epoch": 196.89156626506025, "grad_norm": 21.962509155273438, "learning_rate": 1.4320388349514563e-06, "loss": 0.54, "step": 20305 }, { "epoch": 196.9012048192771, "grad_norm": 18.61345100402832, "learning_rate": 1.4271844660194174e-06, "loss": 0.2013, "step": 20306 }, { "epoch": 196.91084337349398, "grad_norm": 44.79917907714844, "learning_rate": 1.4223300970873788e-06, "loss": 0.1271, "step": 20307 }, { "epoch": 196.92048192771085, "grad_norm": 17.95446014404297, "learning_rate": 1.41747572815534e-06, "loss": 0.2565, "step": 20308 }, { "epoch": 196.9301204819277, "grad_norm": 10.469099998474121, "learning_rate": 1.412621359223301e-06, "loss": 0.2324, "step": 20309 }, { "epoch": 196.93975903614458, "grad_norm": 17.9055233001709, "learning_rate": 1.4077669902912622e-06, "loss": 0.1549, "step": 20310 }, { "epoch": 196.94939759036146, "grad_norm": 4.974667072296143, "learning_rate": 1.4029126213592234e-06, "loss": 0.2056, "step": 20311 }, { "epoch": 196.9590361445783, "grad_norm": 15.560912132263184, "learning_rate": 1.3980582524271845e-06, "loss": 0.2854, "step": 20312 }, { "epoch": 196.96867469879518, "grad_norm": 4.748813629150391, "learning_rate": 1.3932038834951457e-06, "loss": 0.1574, "step": 20313 }, { "epoch": 196.97831325301206, "grad_norm": 16.073511123657227, "learning_rate": 1.3883495145631068e-06, "loss": 0.3422, "step": 20314 }, { "epoch": 196.9879518072289, "grad_norm": 16.367225646972656, "learning_rate": 1.3834951456310682e-06, "loss": 0.21, "step": 20315 }, { "epoch": 197.00361445783133, "grad_norm": 9.663505554199219, "learning_rate": 1.378640776699029e-06, "loss": 0.233, "step": 20316 }, { "epoch": 197.01325301204818, "grad_norm": 31.379446029663086, "learning_rate": 1.3737864077669903e-06, "loss": 0.4406, "step": 20317 }, { "epoch": 197.02289156626506, "grad_norm": 10.133044242858887, "learning_rate": 1.3689320388349516e-06, "loss": 0.2788, "step": 20318 }, { "epoch": 197.03253012048194, "grad_norm": 8.545489311218262, "learning_rate": 1.3640776699029128e-06, "loss": 0.2899, "step": 20319 }, { "epoch": 197.0421686746988, "grad_norm": 4.836940765380859, "learning_rate": 1.359223300970874e-06, "loss": 0.2997, "step": 20320 }, { "epoch": 197.05180722891566, "grad_norm": 13.413064002990723, "learning_rate": 1.354368932038835e-06, "loss": 0.4379, "step": 20321 }, { "epoch": 197.06144578313254, "grad_norm": 26.1882266998291, "learning_rate": 1.3495145631067962e-06, "loss": 0.3201, "step": 20322 }, { "epoch": 197.0710843373494, "grad_norm": 15.277892112731934, "learning_rate": 1.3446601941747573e-06, "loss": 0.2637, "step": 20323 }, { "epoch": 197.08072289156627, "grad_norm": 7.434031963348389, "learning_rate": 1.3398058252427185e-06, "loss": 0.2042, "step": 20324 }, { "epoch": 197.09036144578315, "grad_norm": 27.817522048950195, "learning_rate": 1.3349514563106796e-06, "loss": 0.2701, "step": 20325 }, { "epoch": 197.1, "grad_norm": 9.490818977355957, "learning_rate": 1.330097087378641e-06, "loss": 0.222, "step": 20326 }, { "epoch": 197.10963855421687, "grad_norm": 8.408478736877441, "learning_rate": 1.325242718446602e-06, "loss": 0.3639, "step": 20327 }, { "epoch": 197.11927710843372, "grad_norm": 14.7435302734375, "learning_rate": 1.320388349514563e-06, "loss": 0.1527, "step": 20328 }, { "epoch": 197.1289156626506, "grad_norm": 16.229488372802734, "learning_rate": 1.3155339805825242e-06, "loss": 0.0582, "step": 20329 }, { "epoch": 197.13855421686748, "grad_norm": 4.018621444702148, "learning_rate": 1.3106796116504856e-06, "loss": 0.1123, "step": 20330 }, { "epoch": 197.14819277108433, "grad_norm": 14.283169746398926, "learning_rate": 1.3058252427184467e-06, "loss": 0.2775, "step": 20331 }, { "epoch": 197.1578313253012, "grad_norm": 9.367977142333984, "learning_rate": 1.3009708737864079e-06, "loss": 0.1922, "step": 20332 }, { "epoch": 197.16746987951808, "grad_norm": 27.351531982421875, "learning_rate": 1.296116504854369e-06, "loss": 0.3118, "step": 20333 }, { "epoch": 197.17710843373493, "grad_norm": 18.989177703857422, "learning_rate": 1.2912621359223302e-06, "loss": 0.1618, "step": 20334 }, { "epoch": 197.1867469879518, "grad_norm": 23.75333023071289, "learning_rate": 1.2864077669902913e-06, "loss": 0.2719, "step": 20335 }, { "epoch": 197.19638554216868, "grad_norm": 18.16623306274414, "learning_rate": 1.2815533980582524e-06, "loss": 0.3349, "step": 20336 }, { "epoch": 197.20602409638553, "grad_norm": 23.816152572631836, "learning_rate": 1.2766990291262138e-06, "loss": 0.2635, "step": 20337 }, { "epoch": 197.2156626506024, "grad_norm": 3.883120059967041, "learning_rate": 1.2718446601941747e-06, "loss": 0.3453, "step": 20338 }, { "epoch": 197.2253012048193, "grad_norm": 7.806713581085205, "learning_rate": 1.2669902912621359e-06, "loss": 0.0639, "step": 20339 }, { "epoch": 197.23493975903614, "grad_norm": 19.413158416748047, "learning_rate": 1.262135922330097e-06, "loss": 0.3703, "step": 20340 }, { "epoch": 197.24457831325302, "grad_norm": 4.682285308837891, "learning_rate": 1.2572815533980584e-06, "loss": 0.0538, "step": 20341 }, { "epoch": 197.2542168674699, "grad_norm": 3.4869544506073, "learning_rate": 1.2524271844660195e-06, "loss": 0.2343, "step": 20342 }, { "epoch": 197.26385542168674, "grad_norm": 16.46276092529297, "learning_rate": 1.2475728155339807e-06, "loss": 0.1543, "step": 20343 }, { "epoch": 197.27349397590362, "grad_norm": 4.966773509979248, "learning_rate": 1.2427184466019418e-06, "loss": 0.1557, "step": 20344 }, { "epoch": 197.28313253012047, "grad_norm": 40.99409484863281, "learning_rate": 1.237864077669903e-06, "loss": 0.5076, "step": 20345 }, { "epoch": 197.29277108433735, "grad_norm": 9.215176582336426, "learning_rate": 1.2330097087378641e-06, "loss": 0.2191, "step": 20346 }, { "epoch": 197.30240963855422, "grad_norm": 16.781116485595703, "learning_rate": 1.2281553398058253e-06, "loss": 0.212, "step": 20347 }, { "epoch": 197.31204819277107, "grad_norm": 6.631156921386719, "learning_rate": 1.2233009708737866e-06, "loss": 0.3558, "step": 20348 }, { "epoch": 197.32168674698795, "grad_norm": 27.000736236572266, "learning_rate": 1.2184466019417478e-06, "loss": 0.4477, "step": 20349 }, { "epoch": 197.33132530120483, "grad_norm": 15.476207733154297, "learning_rate": 1.2135922330097087e-06, "loss": 0.3498, "step": 20350 }, { "epoch": 197.34096385542168, "grad_norm": 5.029570579528809, "learning_rate": 1.2087378640776698e-06, "loss": 0.2192, "step": 20351 }, { "epoch": 197.35060240963855, "grad_norm": 26.277700424194336, "learning_rate": 1.2038834951456312e-06, "loss": 0.1999, "step": 20352 }, { "epoch": 197.36024096385543, "grad_norm": 8.10944938659668, "learning_rate": 1.1990291262135923e-06, "loss": 0.1695, "step": 20353 }, { "epoch": 197.36987951807228, "grad_norm": 13.152849197387695, "learning_rate": 1.1941747572815535e-06, "loss": 0.2601, "step": 20354 }, { "epoch": 197.37951807228916, "grad_norm": 10.206459045410156, "learning_rate": 1.1893203883495146e-06, "loss": 0.2584, "step": 20355 }, { "epoch": 197.38915662650604, "grad_norm": 4.325229644775391, "learning_rate": 1.1844660194174758e-06, "loss": 0.1371, "step": 20356 }, { "epoch": 197.39879518072289, "grad_norm": 7.13734769821167, "learning_rate": 1.179611650485437e-06, "loss": 0.1631, "step": 20357 }, { "epoch": 197.40843373493976, "grad_norm": 5.939944744110107, "learning_rate": 1.174757281553398e-06, "loss": 0.3463, "step": 20358 }, { "epoch": 197.41807228915664, "grad_norm": 11.8012113571167, "learning_rate": 1.1699029126213594e-06, "loss": 0.2016, "step": 20359 }, { "epoch": 197.4277108433735, "grad_norm": 47.06268310546875, "learning_rate": 1.1650485436893206e-06, "loss": 0.3082, "step": 20360 }, { "epoch": 197.43734939759037, "grad_norm": 7.2106547355651855, "learning_rate": 1.1601941747572815e-06, "loss": 0.1674, "step": 20361 }, { "epoch": 197.44698795180722, "grad_norm": 10.666325569152832, "learning_rate": 1.1553398058252427e-06, "loss": 0.1195, "step": 20362 }, { "epoch": 197.4566265060241, "grad_norm": 8.380398750305176, "learning_rate": 1.150485436893204e-06, "loss": 0.2978, "step": 20363 }, { "epoch": 197.46626506024097, "grad_norm": 19.830930709838867, "learning_rate": 1.1456310679611652e-06, "loss": 0.5947, "step": 20364 }, { "epoch": 197.47590361445782, "grad_norm": 8.238235473632812, "learning_rate": 1.1407766990291263e-06, "loss": 0.3013, "step": 20365 }, { "epoch": 197.4855421686747, "grad_norm": 13.495376586914062, "learning_rate": 1.1359223300970875e-06, "loss": 0.2827, "step": 20366 }, { "epoch": 197.49518072289158, "grad_norm": 22.58602523803711, "learning_rate": 1.1310679611650486e-06, "loss": 0.3257, "step": 20367 }, { "epoch": 197.50481927710842, "grad_norm": 22.170412063598633, "learning_rate": 1.1262135922330097e-06, "loss": 0.2683, "step": 20368 }, { "epoch": 197.5144578313253, "grad_norm": 9.702598571777344, "learning_rate": 1.1213592233009709e-06, "loss": 0.2182, "step": 20369 }, { "epoch": 197.52409638554218, "grad_norm": 7.148112773895264, "learning_rate": 1.116504854368932e-06, "loss": 0.0926, "step": 20370 }, { "epoch": 197.53373493975903, "grad_norm": 8.853304862976074, "learning_rate": 1.1116504854368934e-06, "loss": 0.2665, "step": 20371 }, { "epoch": 197.5433734939759, "grad_norm": 14.87298583984375, "learning_rate": 1.1067961165048543e-06, "loss": 0.146, "step": 20372 }, { "epoch": 197.55301204819278, "grad_norm": 3.5986709594726562, "learning_rate": 1.1019417475728155e-06, "loss": 0.2737, "step": 20373 }, { "epoch": 197.56265060240963, "grad_norm": 9.324963569641113, "learning_rate": 1.0970873786407768e-06, "loss": 0.2117, "step": 20374 }, { "epoch": 197.5722891566265, "grad_norm": 21.642961502075195, "learning_rate": 1.092233009708738e-06, "loss": 0.335, "step": 20375 }, { "epoch": 197.58192771084336, "grad_norm": 11.13514232635498, "learning_rate": 1.0873786407766991e-06, "loss": 0.3558, "step": 20376 }, { "epoch": 197.59156626506024, "grad_norm": 10.937851905822754, "learning_rate": 1.0825242718446603e-06, "loss": 0.2821, "step": 20377 }, { "epoch": 197.60120481927711, "grad_norm": 19.01100730895996, "learning_rate": 1.0776699029126214e-06, "loss": 0.2041, "step": 20378 }, { "epoch": 197.61084337349396, "grad_norm": 6.404417514801025, "learning_rate": 1.0728155339805826e-06, "loss": 0.1159, "step": 20379 }, { "epoch": 197.62048192771084, "grad_norm": 3.0224924087524414, "learning_rate": 1.0679611650485437e-06, "loss": 0.1011, "step": 20380 }, { "epoch": 197.63012048192772, "grad_norm": 5.22365665435791, "learning_rate": 1.0631067961165048e-06, "loss": 0.148, "step": 20381 }, { "epoch": 197.63975903614457, "grad_norm": 5.645364761352539, "learning_rate": 1.0582524271844662e-06, "loss": 0.1096, "step": 20382 }, { "epoch": 197.64939759036145, "grad_norm": 35.69493103027344, "learning_rate": 1.0533980582524274e-06, "loss": 0.2729, "step": 20383 }, { "epoch": 197.65903614457832, "grad_norm": 8.74090576171875, "learning_rate": 1.0485436893203883e-06, "loss": 0.167, "step": 20384 }, { "epoch": 197.66867469879517, "grad_norm": 10.650311470031738, "learning_rate": 1.0436893203883494e-06, "loss": 0.3065, "step": 20385 }, { "epoch": 197.67831325301205, "grad_norm": 6.090388774871826, "learning_rate": 1.0388349514563108e-06, "loss": 0.1863, "step": 20386 }, { "epoch": 197.68795180722893, "grad_norm": 8.015876770019531, "learning_rate": 1.033980582524272e-06, "loss": 0.1867, "step": 20387 }, { "epoch": 197.69759036144578, "grad_norm": 80.45863342285156, "learning_rate": 1.029126213592233e-06, "loss": 0.287, "step": 20388 }, { "epoch": 197.70722891566265, "grad_norm": 6.0970845222473145, "learning_rate": 1.0242718446601942e-06, "loss": 0.3368, "step": 20389 }, { "epoch": 197.71686746987953, "grad_norm": 12.409235954284668, "learning_rate": 1.0194174757281554e-06, "loss": 0.3636, "step": 20390 }, { "epoch": 197.72650602409638, "grad_norm": 8.907100677490234, "learning_rate": 1.0145631067961165e-06, "loss": 0.2334, "step": 20391 }, { "epoch": 197.73614457831326, "grad_norm": 6.263080596923828, "learning_rate": 1.0097087378640777e-06, "loss": 0.2012, "step": 20392 }, { "epoch": 197.7457831325301, "grad_norm": 8.339865684509277, "learning_rate": 1.004854368932039e-06, "loss": 0.3998, "step": 20393 }, { "epoch": 197.75542168674698, "grad_norm": 28.650074005126953, "learning_rate": 1.0000000000000002e-06, "loss": 0.3451, "step": 20394 }, { "epoch": 197.76506024096386, "grad_norm": 7.750759124755859, "learning_rate": 9.95145631067961e-07, "loss": 0.3051, "step": 20395 }, { "epoch": 197.7746987951807, "grad_norm": 4.832184791564941, "learning_rate": 9.902912621359222e-07, "loss": 0.1707, "step": 20396 }, { "epoch": 197.7843373493976, "grad_norm": 21.397451400756836, "learning_rate": 9.854368932038836e-07, "loss": 0.2578, "step": 20397 }, { "epoch": 197.79397590361447, "grad_norm": 13.325520515441895, "learning_rate": 9.805825242718447e-07, "loss": 0.1834, "step": 20398 }, { "epoch": 197.80361445783132, "grad_norm": 10.589550018310547, "learning_rate": 9.75728155339806e-07, "loss": 0.3859, "step": 20399 }, { "epoch": 197.8132530120482, "grad_norm": 14.670184135437012, "learning_rate": 9.70873786407767e-07, "loss": 0.3342, "step": 20400 }, { "epoch": 197.82289156626507, "grad_norm": 16.08097267150879, "learning_rate": 9.660194174757282e-07, "loss": 0.2377, "step": 20401 }, { "epoch": 197.83253012048192, "grad_norm": 9.448556900024414, "learning_rate": 9.611650485436893e-07, "loss": 0.3166, "step": 20402 }, { "epoch": 197.8421686746988, "grad_norm": 7.265958309173584, "learning_rate": 9.563106796116505e-07, "loss": 0.1765, "step": 20403 }, { "epoch": 197.85180722891567, "grad_norm": 28.073795318603516, "learning_rate": 9.514563106796117e-07, "loss": 0.2899, "step": 20404 }, { "epoch": 197.86144578313252, "grad_norm": 6.3717522621154785, "learning_rate": 9.466019417475729e-07, "loss": 0.2583, "step": 20405 }, { "epoch": 197.8710843373494, "grad_norm": 11.289384841918945, "learning_rate": 9.41747572815534e-07, "loss": 0.3082, "step": 20406 }, { "epoch": 197.88072289156628, "grad_norm": 12.868566513061523, "learning_rate": 9.368932038834952e-07, "loss": 0.2854, "step": 20407 }, { "epoch": 197.89036144578313, "grad_norm": 13.73924446105957, "learning_rate": 9.320388349514564e-07, "loss": 0.1063, "step": 20408 }, { "epoch": 197.9, "grad_norm": 63.93102264404297, "learning_rate": 9.271844660194176e-07, "loss": 0.2118, "step": 20409 }, { "epoch": 197.90963855421685, "grad_norm": 5.338530540466309, "learning_rate": 9.223300970873786e-07, "loss": 0.2029, "step": 20410 }, { "epoch": 197.91927710843373, "grad_norm": 14.0101957321167, "learning_rate": 9.174757281553397e-07, "loss": 0.3606, "step": 20411 }, { "epoch": 197.9289156626506, "grad_norm": 29.61530876159668, "learning_rate": 9.126213592233011e-07, "loss": 0.3362, "step": 20412 }, { "epoch": 197.93855421686746, "grad_norm": 8.281998634338379, "learning_rate": 9.077669902912621e-07, "loss": 0.119, "step": 20413 }, { "epoch": 197.94819277108434, "grad_norm": 5.913414001464844, "learning_rate": 9.029126213592233e-07, "loss": 0.177, "step": 20414 }, { "epoch": 197.9578313253012, "grad_norm": 12.055670738220215, "learning_rate": 8.980582524271844e-07, "loss": 0.0926, "step": 20415 }, { "epoch": 197.96746987951806, "grad_norm": 42.77798843383789, "learning_rate": 8.932038834951457e-07, "loss": 0.3614, "step": 20416 }, { "epoch": 197.97710843373494, "grad_norm": 9.829866409301758, "learning_rate": 8.883495145631068e-07, "loss": 0.1604, "step": 20417 }, { "epoch": 197.98674698795182, "grad_norm": 11.462276458740234, "learning_rate": 8.83495145631068e-07, "loss": 0.2887, "step": 20418 }, { "epoch": 198.0024096385542, "grad_norm": 20.826051712036133, "learning_rate": 8.786407766990292e-07, "loss": 0.2126, "step": 20419 }, { "epoch": 198.0120481927711, "grad_norm": 14.03166389465332, "learning_rate": 8.737864077669904e-07, "loss": 0.2375, "step": 20420 }, { "epoch": 198.02168674698794, "grad_norm": 10.859418869018555, "learning_rate": 8.689320388349515e-07, "loss": 0.327, "step": 20421 }, { "epoch": 198.03132530120482, "grad_norm": 17.18272590637207, "learning_rate": 8.640776699029126e-07, "loss": 0.2301, "step": 20422 }, { "epoch": 198.0409638554217, "grad_norm": 4.587817668914795, "learning_rate": 8.592233009708739e-07, "loss": 0.1679, "step": 20423 }, { "epoch": 198.05060240963854, "grad_norm": 21.189199447631836, "learning_rate": 8.54368932038835e-07, "loss": 0.2715, "step": 20424 }, { "epoch": 198.06024096385542, "grad_norm": 19.7801513671875, "learning_rate": 8.495145631067961e-07, "loss": 0.2165, "step": 20425 }, { "epoch": 198.0698795180723, "grad_norm": 2.8581507205963135, "learning_rate": 8.446601941747573e-07, "loss": 0.1196, "step": 20426 }, { "epoch": 198.07951807228915, "grad_norm": 10.167181968688965, "learning_rate": 8.398058252427185e-07, "loss": 0.3657, "step": 20427 }, { "epoch": 198.08915662650602, "grad_norm": 16.46493911743164, "learning_rate": 8.349514563106797e-07, "loss": 0.3932, "step": 20428 }, { "epoch": 198.0987951807229, "grad_norm": 7.235502243041992, "learning_rate": 8.300970873786408e-07, "loss": 0.2055, "step": 20429 }, { "epoch": 198.10843373493975, "grad_norm": 16.327219009399414, "learning_rate": 8.25242718446602e-07, "loss": 0.4163, "step": 20430 }, { "epoch": 198.11807228915663, "grad_norm": 13.961453437805176, "learning_rate": 8.203883495145632e-07, "loss": 0.4092, "step": 20431 }, { "epoch": 198.1277108433735, "grad_norm": 4.16052770614624, "learning_rate": 8.155339805825243e-07, "loss": 0.3248, "step": 20432 }, { "epoch": 198.13734939759036, "grad_norm": 13.647665023803711, "learning_rate": 8.106796116504854e-07, "loss": 0.1319, "step": 20433 }, { "epoch": 198.14698795180723, "grad_norm": 9.458465576171875, "learning_rate": 8.058252427184467e-07, "loss": 0.2838, "step": 20434 }, { "epoch": 198.15662650602408, "grad_norm": 10.742345809936523, "learning_rate": 8.009708737864079e-07, "loss": 0.456, "step": 20435 }, { "epoch": 198.16626506024096, "grad_norm": 7.937085151672363, "learning_rate": 7.961165048543689e-07, "loss": 0.2705, "step": 20436 }, { "epoch": 198.17590361445784, "grad_norm": 27.210474014282227, "learning_rate": 7.912621359223301e-07, "loss": 0.2245, "step": 20437 }, { "epoch": 198.1855421686747, "grad_norm": 11.005067825317383, "learning_rate": 7.864077669902913e-07, "loss": 0.2767, "step": 20438 }, { "epoch": 198.19518072289156, "grad_norm": 19.28226089477539, "learning_rate": 7.815533980582525e-07, "loss": 0.4244, "step": 20439 }, { "epoch": 198.20481927710844, "grad_norm": 28.721487045288086, "learning_rate": 7.766990291262136e-07, "loss": 0.2239, "step": 20440 }, { "epoch": 198.2144578313253, "grad_norm": 27.414640426635742, "learning_rate": 7.718446601941748e-07, "loss": 0.1356, "step": 20441 }, { "epoch": 198.22409638554217, "grad_norm": 3.201918601989746, "learning_rate": 7.669902912621359e-07, "loss": 0.1024, "step": 20442 }, { "epoch": 198.23373493975905, "grad_norm": 23.772682189941406, "learning_rate": 7.621359223300972e-07, "loss": 0.2675, "step": 20443 }, { "epoch": 198.2433734939759, "grad_norm": 14.040449142456055, "learning_rate": 7.572815533980583e-07, "loss": 0.2114, "step": 20444 }, { "epoch": 198.25301204819277, "grad_norm": 1.6151422262191772, "learning_rate": 7.524271844660194e-07, "loss": 0.0767, "step": 20445 }, { "epoch": 198.26265060240965, "grad_norm": 12.191230773925781, "learning_rate": 7.475728155339807e-07, "loss": 0.33, "step": 20446 }, { "epoch": 198.2722891566265, "grad_norm": 9.679521560668945, "learning_rate": 7.427184466019417e-07, "loss": 0.2622, "step": 20447 }, { "epoch": 198.28192771084338, "grad_norm": 3.5008656978607178, "learning_rate": 7.37864077669903e-07, "loss": 0.1152, "step": 20448 }, { "epoch": 198.29156626506025, "grad_norm": 6.312493324279785, "learning_rate": 7.330097087378641e-07, "loss": 0.2166, "step": 20449 }, { "epoch": 198.3012048192771, "grad_norm": 5.670646667480469, "learning_rate": 7.281553398058253e-07, "loss": 0.2352, "step": 20450 }, { "epoch": 198.31084337349398, "grad_norm": 10.492179870605469, "learning_rate": 7.233009708737864e-07, "loss": 0.2645, "step": 20451 }, { "epoch": 198.32048192771083, "grad_norm": 25.72566032409668, "learning_rate": 7.184466019417476e-07, "loss": 0.6069, "step": 20452 }, { "epoch": 198.3301204819277, "grad_norm": 12.420978546142578, "learning_rate": 7.135922330097087e-07, "loss": 0.2982, "step": 20453 }, { "epoch": 198.33975903614459, "grad_norm": 8.393589973449707, "learning_rate": 7.0873786407767e-07, "loss": 0.2052, "step": 20454 }, { "epoch": 198.34939759036143, "grad_norm": 17.777118682861328, "learning_rate": 7.038834951456311e-07, "loss": 0.1613, "step": 20455 }, { "epoch": 198.3590361445783, "grad_norm": 8.21733283996582, "learning_rate": 6.990291262135923e-07, "loss": 0.2965, "step": 20456 }, { "epoch": 198.3686746987952, "grad_norm": 5.858559608459473, "learning_rate": 6.941747572815534e-07, "loss": 0.3156, "step": 20457 }, { "epoch": 198.37831325301204, "grad_norm": 11.257736206054688, "learning_rate": 6.893203883495146e-07, "loss": 0.2797, "step": 20458 }, { "epoch": 198.38795180722892, "grad_norm": 7.938432693481445, "learning_rate": 6.844660194174758e-07, "loss": 0.1725, "step": 20459 }, { "epoch": 198.3975903614458, "grad_norm": 16.459796905517578, "learning_rate": 6.79611650485437e-07, "loss": 0.4864, "step": 20460 }, { "epoch": 198.40722891566264, "grad_norm": 85.3182373046875, "learning_rate": 6.747572815533981e-07, "loss": 0.3843, "step": 20461 }, { "epoch": 198.41686746987952, "grad_norm": 8.190557479858398, "learning_rate": 6.699029126213592e-07, "loss": 0.3784, "step": 20462 }, { "epoch": 198.4265060240964, "grad_norm": 10.840935707092285, "learning_rate": 6.650485436893205e-07, "loss": 0.1577, "step": 20463 }, { "epoch": 198.43614457831325, "grad_norm": 13.23756217956543, "learning_rate": 6.601941747572815e-07, "loss": 0.1917, "step": 20464 }, { "epoch": 198.44578313253012, "grad_norm": 31.474300384521484, "learning_rate": 6.553398058252428e-07, "loss": 0.5221, "step": 20465 }, { "epoch": 198.455421686747, "grad_norm": 14.294294357299805, "learning_rate": 6.504854368932039e-07, "loss": 0.2436, "step": 20466 }, { "epoch": 198.46506024096385, "grad_norm": 6.208407878875732, "learning_rate": 6.456310679611651e-07, "loss": 0.1072, "step": 20467 }, { "epoch": 198.47469879518073, "grad_norm": 7.256781101226807, "learning_rate": 6.407766990291262e-07, "loss": 0.0915, "step": 20468 }, { "epoch": 198.48433734939758, "grad_norm": 16.983592987060547, "learning_rate": 6.359223300970874e-07, "loss": 0.4022, "step": 20469 }, { "epoch": 198.49397590361446, "grad_norm": 29.309045791625977, "learning_rate": 6.310679611650485e-07, "loss": 0.1554, "step": 20470 }, { "epoch": 198.50361445783133, "grad_norm": 8.625900268554688, "learning_rate": 6.262135922330098e-07, "loss": 0.1609, "step": 20471 }, { "epoch": 198.51325301204818, "grad_norm": 7.636291027069092, "learning_rate": 6.213592233009709e-07, "loss": 0.1478, "step": 20472 }, { "epoch": 198.52289156626506, "grad_norm": 14.105759620666504, "learning_rate": 6.165048543689321e-07, "loss": 0.2244, "step": 20473 }, { "epoch": 198.53253012048194, "grad_norm": 11.02506160736084, "learning_rate": 6.116504854368933e-07, "loss": 0.2864, "step": 20474 }, { "epoch": 198.5421686746988, "grad_norm": 6.274532318115234, "learning_rate": 6.067961165048543e-07, "loss": 0.2298, "step": 20475 }, { "epoch": 198.55180722891566, "grad_norm": 8.115767478942871, "learning_rate": 6.019417475728156e-07, "loss": 0.1956, "step": 20476 }, { "epoch": 198.56144578313254, "grad_norm": 27.67302131652832, "learning_rate": 5.970873786407767e-07, "loss": 0.547, "step": 20477 }, { "epoch": 198.5710843373494, "grad_norm": 32.08329391479492, "learning_rate": 5.922330097087379e-07, "loss": 0.2967, "step": 20478 }, { "epoch": 198.58072289156627, "grad_norm": 7.092564582824707, "learning_rate": 5.87378640776699e-07, "loss": 0.3508, "step": 20479 }, { "epoch": 198.59036144578315, "grad_norm": 9.82118034362793, "learning_rate": 5.825242718446603e-07, "loss": 0.2547, "step": 20480 }, { "epoch": 198.6, "grad_norm": 13.250371932983398, "learning_rate": 5.776699029126213e-07, "loss": 0.2745, "step": 20481 }, { "epoch": 198.60963855421687, "grad_norm": 16.05927848815918, "learning_rate": 5.728155339805826e-07, "loss": 0.3588, "step": 20482 }, { "epoch": 198.61927710843372, "grad_norm": 12.522875785827637, "learning_rate": 5.679611650485437e-07, "loss": 0.5104, "step": 20483 }, { "epoch": 198.6289156626506, "grad_norm": 8.962361335754395, "learning_rate": 5.631067961165049e-07, "loss": 0.3133, "step": 20484 }, { "epoch": 198.63855421686748, "grad_norm": 19.095016479492188, "learning_rate": 5.58252427184466e-07, "loss": 0.2673, "step": 20485 }, { "epoch": 198.64819277108433, "grad_norm": 9.126097679138184, "learning_rate": 5.533980582524272e-07, "loss": 0.3695, "step": 20486 }, { "epoch": 198.6578313253012, "grad_norm": 27.395248413085938, "learning_rate": 5.485436893203884e-07, "loss": 0.3171, "step": 20487 }, { "epoch": 198.66746987951808, "grad_norm": 10.310669898986816, "learning_rate": 5.436893203883496e-07, "loss": 0.301, "step": 20488 }, { "epoch": 198.67710843373493, "grad_norm": 2.9537079334259033, "learning_rate": 5.388349514563107e-07, "loss": 0.0958, "step": 20489 }, { "epoch": 198.6867469879518, "grad_norm": 7.123836994171143, "learning_rate": 5.339805825242719e-07, "loss": 0.1938, "step": 20490 }, { "epoch": 198.69638554216868, "grad_norm": 7.158716678619385, "learning_rate": 5.291262135922331e-07, "loss": 0.209, "step": 20491 }, { "epoch": 198.70602409638553, "grad_norm": 15.068617820739746, "learning_rate": 5.242718446601941e-07, "loss": 0.216, "step": 20492 }, { "epoch": 198.7156626506024, "grad_norm": 10.795052528381348, "learning_rate": 5.194174757281554e-07, "loss": 0.2068, "step": 20493 }, { "epoch": 198.7253012048193, "grad_norm": 8.884270668029785, "learning_rate": 5.145631067961165e-07, "loss": 0.1927, "step": 20494 }, { "epoch": 198.73493975903614, "grad_norm": 11.670027732849121, "learning_rate": 5.097087378640777e-07, "loss": 0.2105, "step": 20495 }, { "epoch": 198.74457831325302, "grad_norm": 15.137890815734863, "learning_rate": 5.048543689320388e-07, "loss": 0.2337, "step": 20496 }, { "epoch": 198.7542168674699, "grad_norm": 15.90142822265625, "learning_rate": 5.000000000000001e-07, "loss": 0.0939, "step": 20497 }, { "epoch": 198.76385542168674, "grad_norm": 8.405405044555664, "learning_rate": 4.951456310679611e-07, "loss": 0.2133, "step": 20498 }, { "epoch": 198.77349397590362, "grad_norm": 49.194419860839844, "learning_rate": 4.902912621359224e-07, "loss": 0.1418, "step": 20499 }, { "epoch": 198.78313253012047, "grad_norm": 38.71684646606445, "learning_rate": 4.854368932038835e-07, "loss": 0.4577, "step": 20500 }, { "epoch": 198.79277108433735, "grad_norm": 5.9449286460876465, "learning_rate": 4.805825242718447e-07, "loss": 0.2339, "step": 20501 }, { "epoch": 198.80240963855422, "grad_norm": 9.778194427490234, "learning_rate": 4.7572815533980586e-07, "loss": 0.1431, "step": 20502 }, { "epoch": 198.81204819277107, "grad_norm": 6.245667457580566, "learning_rate": 4.70873786407767e-07, "loss": 0.1098, "step": 20503 }, { "epoch": 198.82168674698795, "grad_norm": 10.601603507995605, "learning_rate": 4.660194174757282e-07, "loss": 0.3414, "step": 20504 }, { "epoch": 198.83132530120483, "grad_norm": 9.315213203430176, "learning_rate": 4.611650485436893e-07, "loss": 0.2882, "step": 20505 }, { "epoch": 198.84096385542168, "grad_norm": 3.082127571105957, "learning_rate": 4.5631067961165055e-07, "loss": 0.106, "step": 20506 }, { "epoch": 198.85060240963855, "grad_norm": 7.291627883911133, "learning_rate": 4.5145631067961165e-07, "loss": 0.2163, "step": 20507 }, { "epoch": 198.86024096385543, "grad_norm": 5.704419136047363, "learning_rate": 4.4660194174757285e-07, "loss": 0.0548, "step": 20508 }, { "epoch": 198.86987951807228, "grad_norm": 5.658013343811035, "learning_rate": 4.41747572815534e-07, "loss": 0.1419, "step": 20509 }, { "epoch": 198.87951807228916, "grad_norm": 6.469446182250977, "learning_rate": 4.368932038834952e-07, "loss": 0.3307, "step": 20510 }, { "epoch": 198.88915662650604, "grad_norm": 8.166939735412598, "learning_rate": 4.320388349514563e-07, "loss": 0.1743, "step": 20511 }, { "epoch": 198.89879518072289, "grad_norm": 161.28347778320312, "learning_rate": 4.271844660194175e-07, "loss": 0.3704, "step": 20512 }, { "epoch": 198.90843373493976, "grad_norm": 9.262096405029297, "learning_rate": 4.2233009708737863e-07, "loss": 0.1584, "step": 20513 }, { "epoch": 198.91807228915664, "grad_norm": 10.217977523803711, "learning_rate": 4.174757281553398e-07, "loss": 0.0977, "step": 20514 }, { "epoch": 198.9277108433735, "grad_norm": 20.308250427246094, "learning_rate": 4.12621359223301e-07, "loss": 0.3464, "step": 20515 }, { "epoch": 198.93734939759037, "grad_norm": 7.567174434661865, "learning_rate": 4.0776699029126217e-07, "loss": 0.3295, "step": 20516 }, { "epoch": 198.94698795180722, "grad_norm": 24.920278549194336, "learning_rate": 4.0291262135922337e-07, "loss": 0.5337, "step": 20517 }, { "epoch": 198.9566265060241, "grad_norm": 11.18128776550293, "learning_rate": 3.9805825242718446e-07, "loss": 0.1757, "step": 20518 }, { "epoch": 198.96626506024097, "grad_norm": 5.450294494628906, "learning_rate": 3.9320388349514566e-07, "loss": 0.1635, "step": 20519 }, { "epoch": 198.97590361445782, "grad_norm": 14.415067672729492, "learning_rate": 3.883495145631068e-07, "loss": 0.3344, "step": 20520 }, { "epoch": 198.9855421686747, "grad_norm": 11.097992897033691, "learning_rate": 3.8349514563106795e-07, "loss": 0.2093, "step": 20521 }, { "epoch": 199.00120481927712, "grad_norm": 6.131253719329834, "learning_rate": 3.7864077669902915e-07, "loss": 0.1555, "step": 20522 }, { "epoch": 199.01084337349397, "grad_norm": 26.914655685424805, "learning_rate": 3.7378640776699035e-07, "loss": 0.1666, "step": 20523 }, { "epoch": 199.02048192771085, "grad_norm": 3.697765827178955, "learning_rate": 3.689320388349515e-07, "loss": 0.1362, "step": 20524 }, { "epoch": 199.03012048192772, "grad_norm": 9.195625305175781, "learning_rate": 3.6407766990291264e-07, "loss": 0.4279, "step": 20525 }, { "epoch": 199.03975903614457, "grad_norm": 7.7525506019592285, "learning_rate": 3.592233009708738e-07, "loss": 0.172, "step": 20526 }, { "epoch": 199.04939759036145, "grad_norm": 10.385431289672852, "learning_rate": 3.54368932038835e-07, "loss": 0.3041, "step": 20527 }, { "epoch": 199.0590361445783, "grad_norm": 14.890335083007812, "learning_rate": 3.4951456310679613e-07, "loss": 0.3837, "step": 20528 }, { "epoch": 199.06867469879518, "grad_norm": 5.008652687072754, "learning_rate": 3.446601941747573e-07, "loss": 0.1208, "step": 20529 }, { "epoch": 199.07831325301206, "grad_norm": 5.867541313171387, "learning_rate": 3.398058252427185e-07, "loss": 0.1926, "step": 20530 }, { "epoch": 199.0879518072289, "grad_norm": 15.984606742858887, "learning_rate": 3.349514563106796e-07, "loss": 0.3507, "step": 20531 }, { "epoch": 199.09759036144578, "grad_norm": 12.347147941589355, "learning_rate": 3.3009708737864077e-07, "loss": 0.2543, "step": 20532 }, { "epoch": 199.10722891566266, "grad_norm": 18.542499542236328, "learning_rate": 3.2524271844660197e-07, "loss": 0.4331, "step": 20533 }, { "epoch": 199.1168674698795, "grad_norm": 6.15609884262085, "learning_rate": 3.203883495145631e-07, "loss": 0.1691, "step": 20534 }, { "epoch": 199.1265060240964, "grad_norm": 9.315641403198242, "learning_rate": 3.1553398058252426e-07, "loss": 0.3386, "step": 20535 }, { "epoch": 199.13614457831326, "grad_norm": 8.124152183532715, "learning_rate": 3.1067961165048546e-07, "loss": 0.2413, "step": 20536 }, { "epoch": 199.1457831325301, "grad_norm": 28.42595672607422, "learning_rate": 3.0582524271844665e-07, "loss": 0.289, "step": 20537 }, { "epoch": 199.155421686747, "grad_norm": 9.784184455871582, "learning_rate": 3.009708737864078e-07, "loss": 0.215, "step": 20538 }, { "epoch": 199.16506024096387, "grad_norm": 14.44860553741455, "learning_rate": 2.9611650485436895e-07, "loss": 0.3092, "step": 20539 }, { "epoch": 199.17469879518072, "grad_norm": 14.792508125305176, "learning_rate": 2.9126213592233014e-07, "loss": 0.3394, "step": 20540 }, { "epoch": 199.1843373493976, "grad_norm": 24.091609954833984, "learning_rate": 2.864077669902913e-07, "loss": 0.2703, "step": 20541 }, { "epoch": 199.19397590361444, "grad_norm": 14.206841468811035, "learning_rate": 2.8155339805825244e-07, "loss": 0.2832, "step": 20542 }, { "epoch": 199.20361445783132, "grad_norm": 18.385879516601562, "learning_rate": 2.766990291262136e-07, "loss": 0.2572, "step": 20543 }, { "epoch": 199.2132530120482, "grad_norm": 12.737919807434082, "learning_rate": 2.718446601941748e-07, "loss": 0.2705, "step": 20544 }, { "epoch": 199.22289156626505, "grad_norm": 10.549520492553711, "learning_rate": 2.669902912621359e-07, "loss": 0.4025, "step": 20545 }, { "epoch": 199.23253012048193, "grad_norm": 10.592602729797363, "learning_rate": 2.6213592233009707e-07, "loss": 0.177, "step": 20546 }, { "epoch": 199.2421686746988, "grad_norm": 23.768966674804688, "learning_rate": 2.5728155339805827e-07, "loss": 0.3388, "step": 20547 }, { "epoch": 199.25180722891565, "grad_norm": 6.550848007202148, "learning_rate": 2.524271844660194e-07, "loss": 0.0663, "step": 20548 }, { "epoch": 199.26144578313253, "grad_norm": 8.930279731750488, "learning_rate": 2.4757281553398056e-07, "loss": 0.1955, "step": 20549 }, { "epoch": 199.2710843373494, "grad_norm": 11.684928894042969, "learning_rate": 2.4271844660194176e-07, "loss": 0.0547, "step": 20550 }, { "epoch": 199.28072289156626, "grad_norm": 6.2185540199279785, "learning_rate": 2.3786407766990293e-07, "loss": 0.1689, "step": 20551 }, { "epoch": 199.29036144578313, "grad_norm": 10.79786491394043, "learning_rate": 2.330097087378641e-07, "loss": 0.2782, "step": 20552 }, { "epoch": 199.3, "grad_norm": 13.582331657409668, "learning_rate": 2.2815533980582528e-07, "loss": 0.1776, "step": 20553 }, { "epoch": 199.30963855421686, "grad_norm": 11.91347599029541, "learning_rate": 2.2330097087378642e-07, "loss": 0.308, "step": 20554 }, { "epoch": 199.31927710843374, "grad_norm": 23.829675674438477, "learning_rate": 2.184466019417476e-07, "loss": 0.1699, "step": 20555 }, { "epoch": 199.32891566265062, "grad_norm": 10.990973472595215, "learning_rate": 2.1359223300970874e-07, "loss": 0.5391, "step": 20556 }, { "epoch": 199.33855421686746, "grad_norm": 25.10235595703125, "learning_rate": 2.087378640776699e-07, "loss": 0.3737, "step": 20557 }, { "epoch": 199.34819277108434, "grad_norm": 12.315135955810547, "learning_rate": 2.0388349514563108e-07, "loss": 0.3643, "step": 20558 }, { "epoch": 199.3578313253012, "grad_norm": 17.406326293945312, "learning_rate": 1.9902912621359223e-07, "loss": 0.4904, "step": 20559 }, { "epoch": 199.36746987951807, "grad_norm": 7.381198883056641, "learning_rate": 1.941747572815534e-07, "loss": 0.1415, "step": 20560 }, { "epoch": 199.37710843373495, "grad_norm": 8.231431007385254, "learning_rate": 1.8932038834951458e-07, "loss": 0.1678, "step": 20561 }, { "epoch": 199.3867469879518, "grad_norm": 14.775771141052246, "learning_rate": 1.8446601941747575e-07, "loss": 0.2567, "step": 20562 }, { "epoch": 199.39638554216867, "grad_norm": 18.49690818786621, "learning_rate": 1.796116504854369e-07, "loss": 0.1802, "step": 20563 }, { "epoch": 199.40602409638555, "grad_norm": 23.467342376708984, "learning_rate": 1.7475728155339807e-07, "loss": 0.244, "step": 20564 }, { "epoch": 199.4156626506024, "grad_norm": 7.626484394073486, "learning_rate": 1.6990291262135924e-07, "loss": 0.1417, "step": 20565 }, { "epoch": 199.42530120481928, "grad_norm": 4.596919059753418, "learning_rate": 1.6504854368932038e-07, "loss": 0.2647, "step": 20566 }, { "epoch": 199.43493975903615, "grad_norm": 19.35647201538086, "learning_rate": 1.6019417475728156e-07, "loss": 0.4593, "step": 20567 }, { "epoch": 199.444578313253, "grad_norm": 12.987382888793945, "learning_rate": 1.5533980582524273e-07, "loss": 0.176, "step": 20568 }, { "epoch": 199.45421686746988, "grad_norm": 6.299477577209473, "learning_rate": 1.504854368932039e-07, "loss": 0.3434, "step": 20569 }, { "epoch": 199.46385542168676, "grad_norm": 9.423845291137695, "learning_rate": 1.4563106796116507e-07, "loss": 0.2607, "step": 20570 }, { "epoch": 199.4734939759036, "grad_norm": 6.0382585525512695, "learning_rate": 1.4077669902912622e-07, "loss": 0.237, "step": 20571 }, { "epoch": 199.48313253012049, "grad_norm": 21.74507713317871, "learning_rate": 1.359223300970874e-07, "loss": 0.1673, "step": 20572 }, { "epoch": 199.49277108433736, "grad_norm": 7.174163341522217, "learning_rate": 1.3106796116504854e-07, "loss": 0.2472, "step": 20573 }, { "epoch": 199.5024096385542, "grad_norm": 8.713092803955078, "learning_rate": 1.262135922330097e-07, "loss": 0.1672, "step": 20574 }, { "epoch": 199.5120481927711, "grad_norm": 6.31298303604126, "learning_rate": 1.2135922330097088e-07, "loss": 0.2578, "step": 20575 }, { "epoch": 199.52168674698794, "grad_norm": 8.963530540466309, "learning_rate": 1.1650485436893205e-07, "loss": 0.2759, "step": 20576 }, { "epoch": 199.53132530120482, "grad_norm": 9.855755805969238, "learning_rate": 1.1165048543689321e-07, "loss": 0.3376, "step": 20577 }, { "epoch": 199.5409638554217, "grad_norm": 7.945336818695068, "learning_rate": 1.0679611650485437e-07, "loss": 0.1637, "step": 20578 }, { "epoch": 199.55060240963854, "grad_norm": 25.625944137573242, "learning_rate": 1.0194174757281554e-07, "loss": 0.3925, "step": 20579 }, { "epoch": 199.56024096385542, "grad_norm": 9.056851387023926, "learning_rate": 9.70873786407767e-08, "loss": 0.2311, "step": 20580 }, { "epoch": 199.5698795180723, "grad_norm": 4.317712306976318, "learning_rate": 9.223300970873787e-08, "loss": 0.0832, "step": 20581 }, { "epoch": 199.57951807228915, "grad_norm": 8.00445556640625, "learning_rate": 8.737864077669903e-08, "loss": 0.1901, "step": 20582 }, { "epoch": 199.58915662650602, "grad_norm": 7.138917922973633, "learning_rate": 8.252427184466019e-08, "loss": 0.2594, "step": 20583 }, { "epoch": 199.5987951807229, "grad_norm": 13.810276985168457, "learning_rate": 7.766990291262136e-08, "loss": 0.295, "step": 20584 }, { "epoch": 199.60843373493975, "grad_norm": 6.455498695373535, "learning_rate": 7.281553398058254e-08, "loss": 0.2659, "step": 20585 }, { "epoch": 199.61807228915663, "grad_norm": 7.769881725311279, "learning_rate": 6.79611650485437e-08, "loss": 0.252, "step": 20586 }, { "epoch": 199.6277108433735, "grad_norm": 12.97164249420166, "learning_rate": 6.310679611650485e-08, "loss": 0.3516, "step": 20587 }, { "epoch": 199.63734939759036, "grad_norm": 19.484344482421875, "learning_rate": 5.8252427184466026e-08, "loss": 0.3565, "step": 20588 }, { "epoch": 199.64698795180723, "grad_norm": 7.474972248077393, "learning_rate": 5.3398058252427185e-08, "loss": 0.3397, "step": 20589 }, { "epoch": 199.65662650602408, "grad_norm": 7.77719783782959, "learning_rate": 4.854368932038835e-08, "loss": 0.1982, "step": 20590 }, { "epoch": 199.66626506024096, "grad_norm": 27.35228157043457, "learning_rate": 4.3689320388349516e-08, "loss": 0.4191, "step": 20591 }, { "epoch": 199.67590361445784, "grad_norm": 11.310966491699219, "learning_rate": 3.883495145631068e-08, "loss": 0.3547, "step": 20592 }, { "epoch": 199.6855421686747, "grad_norm": 6.348087310791016, "learning_rate": 3.398058252427185e-08, "loss": 0.217, "step": 20593 }, { "epoch": 199.69518072289156, "grad_norm": 4.892882823944092, "learning_rate": 2.9126213592233013e-08, "loss": 0.0997, "step": 20594 }, { "epoch": 199.70481927710844, "grad_norm": 12.820352554321289, "learning_rate": 2.4271844660194175e-08, "loss": 0.2557, "step": 20595 }, { "epoch": 199.7144578313253, "grad_norm": 9.515077590942383, "learning_rate": 1.941747572815534e-08, "loss": 0.1298, "step": 20596 }, { "epoch": 199.72409638554217, "grad_norm": 8.731186866760254, "learning_rate": 1.4563106796116507e-08, "loss": 0.1378, "step": 20597 }, { "epoch": 199.73373493975905, "grad_norm": 12.484709739685059, "learning_rate": 9.70873786407767e-09, "loss": 0.3651, "step": 20598 }, { "epoch": 199.7433734939759, "grad_norm": 14.410330772399902, "learning_rate": 4.854368932038835e-09, "loss": 0.2763, "step": 20599 }, { "epoch": 199.75301204819277, "grad_norm": 13.135978698730469, "learning_rate": 0.0, "loss": 0.214, "step": 20600 } ], "logging_steps": 1, "max_steps": 20600, "num_input_tokens_seen": 0, "num_train_epochs": 200, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 5.552213062003458e+19, "train_batch_size": 1, "trial_name": null, "trial_params": null }